Globalization and Firm Dynamics - Lirias@Lessius
Globalization and Firm Dynamics - Lirias@Lessius
Globalization and Firm Dynamics - Lirias@Lessius
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
FACULTEIT ECONOMISCHE EN<br />
TOEGEPASTE ECONOMISCHE<br />
WETENSCHAPPEN<br />
KATHOLIEKE<br />
UNIVERSITEIT<br />
LEUVEN<br />
<strong>Globalization</strong> <strong>and</strong> <strong>Firm</strong> <strong>Dynamics</strong><br />
Proefschrift voorgedragen tot<br />
het behalen van de graad van<br />
Doctor in de Economische<br />
Wetenschappen<br />
door<br />
Ilke VAN BEVEREN<br />
Number 293 2008
Doctoral committee<br />
Advisor:<br />
prof. dr. Jozef Konings Katholieke Universiteit Leuven<br />
Co-advisor:<br />
prof. dr. Ysabel Nauwelaerts Lessius, Antwerpen<br />
Members:<br />
prof. dr. Bruno Cassiman IESE Business School, Barcelona<br />
prof. dr. Beata Smarzynska Javorcik University of Oxford<br />
prof. dr. Jo Swinnen Katholieke Universiteit Leuven<br />
prof. dr. Frank Verboven Katholieke Universiteit Leuven<br />
prof. dr. Gerald Willmann Katholieke Universiteit Leuven<br />
Daar de proefschriften in de reeks van de Faculteit Economische en Toegepaste<br />
Economische Wetenschappen het persoonlijk werk zijn van hun auteurs, zijn<br />
alleen deze laatsten daarvoor verantwoordelijk<br />
i
Acknowledgements<br />
I would like to begin this word of thanks by thanking two people, without<br />
whom I might have never embarked on this journey: Guido Van Rompuy,<br />
professor in economics at Lessius <strong>and</strong> supervisor of my undergraduate thesis<br />
<strong>and</strong> Mathew Tharakan, with whom I have had the pleasure to work for three<br />
years at the University of Antwerp. Guido, thank you for guiding me through<br />
my first steps into quantitative research <strong>and</strong> for stimulating my interest in<br />
economics in general. Mathew, I would like to thank you for your endless<br />
patience, for the countless hours spent discussing our research project, for<br />
your kindness <strong>and</strong> most of all for allowing me to benefit from your enormous<br />
expertise during my years at the University of Antwerp.<br />
This project could not have been realized without the support of Joep<br />
Konings, who has been my supervisor these past four years. Joep, thank<br />
you for always pointing me in the right direction whenever I felt lost. It<br />
continues to amaze me how, every time I was on the verge of giving up on<br />
some research idea because of some difficulty encountered along the way, you<br />
could put me back on track within fifteen minutes. Most of all, I would like<br />
to thank you for providing me with a tool kit that will allow me to continue<br />
to do good scientific research in the coming years. This tool kit includes not<br />
only knowledge, but also a strong sense of self-confidence <strong>and</strong> independence,<br />
which is perhaps even more important.<br />
I would also like to express my gratitude to my co-supervisor, Ysabel<br />
Nauwelaerts. Ysabel, thank you for your support these last few years. Bruno<br />
Cassiman, Beata Smarzynska Javorcik, Jo Swinnen, Frank Verboven <strong>and</strong><br />
Gerald Willmann, who were part of my scientific committee, have been instrumental<br />
in shaping the final version of my Ph.D. Bruno, I would like<br />
iii
iv<br />
thank you particularly for your input on chapter 4. Beata <strong>and</strong> Gerald are<br />
thanked for providing critical comments on all chapters, which have proven<br />
very helpful. Jo, thank you for many critical <strong>and</strong> insightful comments given<br />
at various LICOS seminars, they have greatly improved the end product. Finally,<br />
Frank, thank you for your methodological <strong>and</strong> econometric suggestions,<br />
particularly in relation to chapters 1 <strong>and</strong> 4.<br />
Without my two co-authors, Veerle Slootmaekers <strong>and</strong> Marialuz Moreno<br />
Badia, chapter 2 could not have been written. Veerle <strong>and</strong> Marialuz, I would<br />
like to thank you for a wonderful cooperation <strong>and</strong> learning experience! I very<br />
much hope we can continue to work together in the future.<br />
The firm-level data used in chapter 2 were kindly provided by the Estonian<br />
Business Registry. Specifically, Larissa Merlukova <strong>and</strong> Kadri Rohulaid of the<br />
Centre of Registers <strong>and</strong> Infosystems are thanked for providing the data <strong>and</strong><br />
some clarifications on the database. The trade data used in this chapter were<br />
obtained through the IMF. The Community Innovation Survey data used in<br />
chapter 4 were kindly provided by the Belgian Science Policy (Belspo). I<br />
am especially grateful to Manu Monard, Peter Teirlinck <strong>and</strong> the CFS/STAT<br />
Commission of Belspo for allowing me to access the data for Belgium; for<br />
answering questions related to the data <strong>and</strong> for their hospitality during my<br />
visits there.<br />
All chapters in this thesis have benefited greatly from comments made by<br />
participants of the LICOS International Trade <strong>and</strong> Development Seminars.<br />
Carmine Ornaghi provided useful comments on chapter 1. Chapter 1 has also<br />
benefited greatly from numerous discussions with Veerle Slootmaekers <strong>and</strong><br />
Damiaan Persyn. An anonymous referee is thanked for the suggestions made<br />
on chapter 3. Leo Sleuwaegen, Hylke V<strong>and</strong>enbussche <strong>and</strong> Filip De Beule are<br />
thanked for their insights on chapter 4. Participants at the Spring Meeting<br />
for Young Economists in Hamburg (2007) <strong>and</strong> at the EARIE conference in<br />
Toulouse (2008) provided useful comments on chapters 1 <strong>and</strong> 4 respectively.<br />
The department of business studies at Lessius <strong>and</strong> particularly Paul Verheyen<br />
<strong>and</strong> Flora Carrijn, are gratefully acknowledged for providing me with
the resources to work on my Ph.D. <strong>and</strong> for allowing me to spend most of my<br />
time on my research these past few years. LICOS, Centre for Institutions<br />
<strong>and</strong> Economic Performance, <strong>and</strong> particularly Jo Swinnen, is thanked for providing<br />
me with an office <strong>and</strong> most importantly, for allowing me to benefit<br />
from a highly stimulating research environment.<br />
I would also like to express my deepest gratitude to my friends <strong>and</strong> colleagues<br />
at both Lessius <strong>and</strong> LICOS. To Filip De Beule, for allowing me to<br />
benefit from his expertise both in terms of research <strong>and</strong> teaching <strong>and</strong> for his<br />
comments on my work; but also for his general support. Filip, I hope we<br />
can continue to work together over the coming years. To my fellow graduate<br />
students <strong>and</strong> other colleagues at Lessius <strong>and</strong> particularly to An, Anneleen,<br />
Bert, Carlos, Frederick, Frederiek, Karen, Kelly, Lien, Nico, Raf, Steve, Sofie,<br />
Sophie, Tim, Tom; thank you all very much for your continuous support, but<br />
also for many shared moments of joy, laughter <strong>and</strong> friendship, which have<br />
given me a home away from home. I look forward to continue to be a part<br />
of our growing <strong>and</strong> dynamic team!<br />
I have also greatly benefited from the expertise <strong>and</strong> help of many of my<br />
fellow Ph.D. students at LICOS. Damiaan, I think you must realize that your<br />
Stata expertise has saved more than one graduate student at LICOS from<br />
ultimate disaster, but still: thank you so much for all your help, countless<br />
discussions <strong>and</strong> everlasting patience. Stijn, thank you for introducing me to<br />
Belfirst <strong>and</strong> for sharing research ideas <strong>and</strong> insights. Emilia, thank you for<br />
your support <strong>and</strong> especially for your sense of humor, which has brightened my<br />
day more than once. Ziga, thank you for your critical comments at various<br />
seminars <strong>and</strong> many useful discussions. Veerle, working together with you has<br />
truly exceeded all my expectations! Italo, apart from your insights, I have<br />
tremendously enjoyed your optimistic view of the world <strong>and</strong> your openness<br />
towards people. I wish you all the best in your future careers! Finally, I<br />
would also like to thank Tom Van Ourti, whom I had the pleasure to work<br />
with when I was still at the University of Antwerp, <strong>and</strong> who introduced me<br />
to Stata.<br />
v
vi<br />
Last but certainly not least, a word of gratitude is in order for my friends<br />
<strong>and</strong> family. I would like to thank my parents, for allowing me to make my<br />
own choices in life. Dad, thank you for coming over for my public defense,<br />
that means a lot to me! Special thanks goes to my sister, Jo, who never<br />
forgot to enquire about my progress <strong>and</strong> who was always just a phone call<br />
away. I would also like to thank Michel’s family, whom we can always rely on,<br />
<strong>and</strong> who have always been very supportive throughout these last few years.<br />
My biggest critic throughout this venture has always been Michel, while<br />
at the same time, he has been my greatest fan. Michel, thank you for continuously<br />
reminding me of what really matters in life. Your critical appraisal<br />
of the world in general <strong>and</strong> the sense <strong>and</strong> nonsense of academic research in<br />
particular forces me to always question the relevance to society of everything<br />
I do, which is a good thing. Most of all, I simply want to thank you for being<br />
a part of my life!<br />
Ilke Van Beveren<br />
Antwerp, November 19 2008.
Contents<br />
Doctoral committee i<br />
Acknowledgements iii<br />
Table of contents ix<br />
List of Figures xiii<br />
List of Tables xv<br />
General introduction 1<br />
1 Total Factor Productivity Estimation: A Practical Review 9<br />
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9<br />
1.2 Total factor productivity: Methodological issues . . . . . . . . 12<br />
1.2.1 Some preliminaries on the production function . . . . . 12<br />
1.2.2 Endogeneity of attrition or selection bias . . . . . . . . 14<br />
1.2.3 Endogeneity of input choice or simultaneity bias . . . . 15<br />
1.2.4 Omitted output price bias . . . . . . . . . . . . . . . . 16<br />
1.2.5 Omitted input price bias . . . . . . . . . . . . . . . . . 17<br />
1.2.6 Endogeneity of the product mix (multi-product firms) . 18<br />
1.2.7 Summary of methodological issues . . . . . . . . . . . . 19<br />
1.3 Total factor productivity estimation . . . . . . . . . . . . . . . 21<br />
1.3.1 Fixed effects estimation . . . . . . . . . . . . . . . . . 21<br />
1.3.2 Instrumental variables (IV) <strong>and</strong> GMM . . . . . . . . . 22<br />
1.3.3 Olley-Pakes estimation algorithm . . . . . . . . . . . . 23<br />
1.3.4 Levinsohn-Petrin estimation algorithm . . . . . . . . . 26<br />
1.3.5 Olley-Pakes versus Levinsohn-Petrin . . . . . . . . . . 28<br />
ix
x Contents<br />
1.3.6 Extensions of the Olley-Pakes methodology . . . . . . . 30<br />
1.3.7 Summary of estimation algorithms . . . . . . . . . . . 33<br />
1.4 Empirical application: Food <strong>and</strong> beverages industry in Belgium 34<br />
1.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48<br />
2 <strong>Globalization</strong> Drives Strategic Product Switching 53<br />
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53<br />
2.2 Industry dynamics in Estonia . . . . . . . . . . . . . . . . . . 56<br />
2.3 Determinants of firm dynamics . . . . . . . . . . . . . . . . . 63<br />
2.3.1 <strong>Firm</strong> characteristics . . . . . . . . . . . . . . . . . . . . 64<br />
2.3.2 Product market characteristics: Domestic . . . . . . . 67<br />
2.3.3 Product market characteristics: International . . . . . 68<br />
2.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72<br />
2.4.1 Baseline results . . . . . . . . . . . . . . . . . . . . . . 72<br />
2.4.2 Self-selection into new markets . . . . . . . . . . . . . 77<br />
2.5 Robustness checks . . . . . . . . . . . . . . . . . . . . . . . . . 84<br />
2.5.1 Results by size class . . . . . . . . . . . . . . . . . . . 84<br />
2.5.2 Results by time period . . . . . . . . . . . . . . . . . . 87<br />
2.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88<br />
2.A Data appendix . . . . . . . . . . . . . . . . . . . . . . . . . . 91<br />
2.A.1 Data <strong>and</strong> sample selection . . . . . . . . . . . . . . . . 91<br />
2.A.2 Definitions of variables . . . . . . . . . . . . . . . . . . 93<br />
3 Footloose Multinationals in Belgium? 101<br />
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101<br />
3.2 Literature review . . . . . . . . . . . . . . . . . . . . . . . . . 103<br />
3.3 Data <strong>and</strong> preliminary facts . . . . . . . . . . . . . . . . . . . . 106<br />
3.4 Empirical model . . . . . . . . . . . . . . . . . . . . . . . . . . 114<br />
3.5 Empirical results . . . . . . . . . . . . . . . . . . . . . . . . . 117<br />
3.6 Robustness checks . . . . . . . . . . . . . . . . . . . . . . . . . 120<br />
3.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123<br />
3.A Data appendix . . . . . . . . . . . . . . . . . . . . . . . . . . 125<br />
3.A.1 Sample selection . . . . . . . . . . . . . . . . . . . . . 125<br />
3.A.2 Definition of variables . . . . . . . . . . . . . . . . . . 126
Contents xi<br />
4 Multinational firms, Research Effort <strong>and</strong> Innovative Output:<br />
An Integrated Approach 129<br />
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129<br />
4.2 The innovation production function . . . . . . . . . . . . . . . 133<br />
4.3 Data <strong>and</strong> empirical facts . . . . . . . . . . . . . . . . . . . . . 141<br />
4.4 Empirical results . . . . . . . . . . . . . . . . . . . . . . . . . 146<br />
4.4.1 Preliminary evidence . . . . . . . . . . . . . . . . . . . 147<br />
4.4.2 The innovation production function: baseline results . 149<br />
4.4.3 High-tech versus low-tech sectors . . . . . . . . . . . . 152<br />
4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156<br />
4.A Data appendix . . . . . . . . . . . . . . . . . . . . . . . . . . 159<br />
4.A.1 Sample selection . . . . . . . . . . . . . . . . . . . . . 159<br />
4.A.2 Definitions of variables . . . . . . . . . . . . . . . . . . 159<br />
Bibliography 163<br />
Doctoral Dissertations Faculty of Business <strong>and</strong> Economics 177
List of Figures<br />
1.1 Weighted productivity index: Comparison estimation methods 45<br />
2.1 Sample size distribution . . . . . . . . . . . . . . . . . . . . . 85<br />
3.1 Foreign multinationals by country of origin . . . . . . . . . . . 108<br />
3.2 Kaplan-Meier survival functions by nationality of ownership . 111<br />
xiii
List of Tables<br />
1.1 TFP estimation: Summary of methodological issues . . . . . . 20<br />
1.2 TFP estimation: Summary of estimation algorithms . . . . . . 35<br />
1.3 Summary statistics of key variables . . . . . . . . . . . . . . . 38<br />
1.4 Production function estimates . . . . . . . . . . . . . . . . . . 39<br />
1.5 Production function estimates: Variety-specific dem<strong>and</strong> . . . . 43<br />
1.6 Decomposition aggregate TFP: De Loecker methodology . . . 47<br />
1.7 Comparison of decomposition results . . . . . . . . . . . . . . 49<br />
2.1 Exits <strong>and</strong> industry switches, 1997-2004 . . . . . . . . . . . . . 58<br />
2.2 Sector distribution . . . . . . . . . . . . . . . . . . . . . . . . 59<br />
2.3 Four-digit product switches decomposed . . . . . . . . . . . . 61<br />
2.4 Destination of product switches by technology class . . . . . . 62<br />
2.5 Summary statistics . . . . . . . . . . . . . . . . . . . . . . . . 71<br />
2.6 Baseline specification . . . . . . . . . . . . . . . . . . . . . . . 74<br />
2.7 Product switching versus industry switching . . . . . . . . . . 78<br />
2.8 Industry switching: Manufacturing versus services . . . . . . . 80<br />
2.9 Unit value difference between industry of origin <strong>and</strong> destination 83<br />
2.10 Determinants of firm dynamics across size categories . . . . . 86<br />
2.11 Determinants of firm dynamics across time . . . . . . . . . . . 89<br />
2.A.1Sector classification: Manufacturing . . . . . . . . . . . . . . . 97<br />
2.A.2Sector classification: Services . . . . . . . . . . . . . . . . . . 98<br />
3.1 Sector distribution of firms . . . . . . . . . . . . . . . . . . . . 109<br />
3.2 Summary statistics by ownership type (1996-2001) . . . . . . . 112<br />
3.3 Regression results: Cox proportional hazard model (1996-2001) 118<br />
3.4 Robustness checks . . . . . . . . . . . . . . . . . . . . . . . . . 122<br />
xv
xvi List of Tables<br />
4.1 Sector distribution of firms by firm type . . . . . . . . . . . . 142<br />
4.2 Sample distribution according to global engagement status . . 144<br />
4.3 Summary statistics by firm type . . . . . . . . . . . . . . . . . 145<br />
4.4 Innovation production function: Preliminary evidence . . . . . 148<br />
4.5 Innovation production function: Two-stage estimation . . . . . 150<br />
4.6 Innovation production function: High-technology sectors . . . 153<br />
4.7 Innovation production function: Low-technology sectors . . . . 155<br />
4.A.1Sector classification: High-tech versus low-tech . . . . . . . . . 161
General Introduction<br />
<strong>Globalization</strong> is traditionally associated with increasing integration of the<br />
world economy, both in terms of international trade <strong>and</strong> foreign direct investment.<br />
Technological advances, combined with the continuing liberalization of<br />
capital <strong>and</strong> goods markets, as well as the opening up of a number of “new”<br />
markets such as India or China, have led to increased trade <strong>and</strong> financial<br />
flows, both within the EU <strong>and</strong> the world (EU, 2005).<br />
Several trends can illustrate this evolution. First, in 2004, international<br />
trade flows increased by 9 percent, compared to a rise of 4 percent for world<br />
GDP (WTO, 2005). Second, FDI stocks have, in spite of the slowdown in<br />
flows since 2000, more than doubled since the 1980s <strong>and</strong> accounted for more<br />
than 20 percent of global GDP in 2005 (UNCTAD, 2004). Moreover, between<br />
1990 <strong>and</strong> 2003, sales <strong>and</strong> exports of foreign affiliates have tripled, while employment<br />
has more than doubled. By comparison, GDP has only grown by<br />
about 60 percent overall over the same period. According to UNCTAD estimates,<br />
around 61,000 multinational companies worldwide, together with<br />
their 900,000 foreign affiliates are responsible for this international production.<br />
Global international production is estimated to account for about 10<br />
percent of GDP <strong>and</strong> 30 percent of world exports (UNCTAD, 2004).<br />
International economic integration brings gains to the world economy through<br />
lower prices for consumers <strong>and</strong> firms, increased variety <strong>and</strong> potential efficiency<br />
gains in production (Bernard, Jensen, Redding <strong>and</strong> Schott, 2007c).<br />
According to the EU (2005), about 20 percent of the increase in living st<strong>and</strong>ards<br />
over the past 50 years in the EU is attributable to its deeper integration<br />
in the world economy. These aggregate data mask important heterogeneity<br />
among groups in the economy however, <strong>and</strong> specifically, among workers <strong>and</strong><br />
1
2 General Introduction<br />
firms. As a consequence, while economists generally agree that the overall<br />
impact of globalization on welfare is positive, public concerns largely remain.<br />
While high-skilled labor in developed countries has been generally shown<br />
to benefit from the increased internationalization of economic activity, the<br />
lower-skilled have been faced with lower relative wages (in the United States)<br />
<strong>and</strong> fewer employment opportunities (in Europe) 1 . Similarly, while trade<br />
liberalization has been shown to increase aggregate industry productivity<br />
growth, this masks important differences between (heterogeneous) firms.<br />
Empirical analyses of trade liberalization at the firm level provide evidence<br />
that this aggregate productivity growth is driven by the expansion <strong>and</strong> entry<br />
into export markets of high-productivity firms, as well as by the contraction<br />
<strong>and</strong> exit of low-productivity firms.<br />
This thesis links in with a recent <strong>and</strong> growing literature dealing with what<br />
has been labeled by Greenaway (2004) as firm-level adjustment to globalization.<br />
This literature has been motivated by a wealth of empirical work<br />
documenting firm heterogeneity in international trade, starting out with the<br />
seminal contribution of Bernard <strong>and</strong> Jensen (1995), who were among the<br />
first to note the important differences between exporters <strong>and</strong> non-exporters<br />
in US manufacturing sectors. In response to these empirical findings, several<br />
theoretical contributions have emerged in recent years taking into account<br />
firm-level heterogeneity in open-economy models (Bernard et al., 2007c). Examples<br />
of such models include Melitz (2003), Melitz <strong>and</strong> Ottaviano (2008)<br />
<strong>and</strong> Costantini <strong>and</strong> Melitz (2007). The growing availability of firm- <strong>and</strong><br />
worker-level data sets, as well as the development of more powerful econometric<br />
tools have also been instrumental as driving forces in this growing<br />
str<strong>and</strong> of literature (Greenaway, 2004).<br />
As noted by Greenaway (2004), there are several dimensions to firm-level<br />
adjustment to globalization. A first dimension is concerned with the link between<br />
globalization <strong>and</strong> firm-level performance, i.e. its productivity. Melitz<br />
(2003) demonstrates that, in the presence of imperfect competition <strong>and</strong> het-<br />
1 For a comprehensive survey on the impact of international trade on wages, with a<br />
focus on the US case, I refer to Feenstra (2000).
erogeneous firms, trade liberalization will lead to an increase in a country’s<br />
imports <strong>and</strong> will hence erode domestic firms’ market shares. This will cause<br />
firms at the high end of the productivity distribution to exp<strong>and</strong> into export<br />
markets (by more than their reduction in domestic sales), while low productivity<br />
firms will be forced to contract or exit.<br />
An important issue in the vast number of studies dealing with the impact<br />
of international trade on firm-level performance concerns the measurement<br />
of total factor productivity (TFP) at the firm level. Chapter 1 in this thesis<br />
aims to contribute to this literature by providing a practical overview of the<br />
methodological issues that arise when estimating total factor productivity at<br />
the establishment level, as well as of the existing (parametric <strong>and</strong> semiparametric)<br />
techniques designed to overcome them. Apart from the well-known<br />
simultaneity <strong>and</strong> selection bias; attention is given to methodological issues<br />
that have emerged more recently <strong>and</strong> that are related to the use of deflated<br />
values of inputs <strong>and</strong> outputs (as opposed to quantities) in estimating productivity<br />
at the firm level, as well as to the endogeneity of product choice. Using<br />
data on single-product firms active in the Belgian food <strong>and</strong> beverages sector,<br />
the biases introduced in traditional TFP estimates are illustrated <strong>and</strong> the<br />
performance of a number of alternative estimators that have been proposed<br />
in the literature is discussed.<br />
Recent theoretical <strong>and</strong> empirical work has uncovered a different dimension<br />
of firm-level adjustment to globalization. In two influential papers; Bernard,<br />
Redding <strong>and</strong> Schott (2005, 2006b) point to the importance of taking firms’<br />
product-level decisions explicitly into account when analyzing firm dynamics.<br />
Bernard, Jensen <strong>and</strong> Schott (2006a) <strong>and</strong> Greenaway, Gullstr<strong>and</strong> <strong>and</strong><br />
Kneller (2008) build on this work to further explore the determinants of<br />
firms’ product-level strategies. A central result emerging from this literature<br />
is that, apart from the choice between continuing its production or exiting,<br />
firms have other strategies available to them to respond to increasing globalization<br />
pressures. Chapter 2 of this thesis links in with this small but growing<br />
literature investigating the impact of trade liberalization on firm strategies<br />
<strong>and</strong> specifically, on the choice between continuing its production, switching<br />
products or exiting from the market.<br />
3
4 General Introduction<br />
Specifically, in Chapter 2, which is joint work with Marialuz Moreno Badia<br />
(IMF) <strong>and</strong> Veerle Slootmaekers (OECD), we use firm-level data for Estonia<br />
between 1997 <strong>and</strong> 2005 to analyze the impact of international competition<br />
on firm dynamics, considering both firm closedown <strong>and</strong> product switching.<br />
The analysis contributes to the literature in two important ways: (1) this<br />
is the first paper to study the determinants of exit <strong>and</strong> product switching<br />
in an emerging market; <strong>and</strong> (2) we consider explicitly the role of export<br />
opportunities. Results indicate that globalization does not affect firm exit<br />
significantly while it is an important factor explaining product switching.<br />
Previous studies on industrial countries have shown that product switching<br />
has been a defensive strategy against low-cost imports. In contrast, results<br />
obtained in Chapter 2 suggest that Estonian firms have switched products as<br />
an offensive strategy to take advantage of the export opportunities created<br />
by trade liberalization.<br />
Perhaps the most widely researched str<strong>and</strong> of the literature dealing with<br />
firm-level adjustment to globalization concerns the activities of multinational<br />
firms <strong>and</strong> the consequences of their presence in industries <strong>and</strong> countries.<br />
While most of the literature in this line of research has focused on the identification<br />
of potential spillovers of these firms to the domestic economy, a small<br />
literature has focused on these firms’ potential “footloose” nature. Given the<br />
vast amounts of resources spent by governments worldwide to attract multinational<br />
firms (Greenaway, 2004), this question is particularly relevant from<br />
a policy perspective. Chapter 3 in my thesis links in with this literature.<br />
In Chapter 3, firm-level panel data for the years 1996-2001, covering all<br />
sectors of the economy, is used to estimate the impact of multinational ownership<br />
on the exit decisions of firms located in Belgium. In the analysis, I<br />
clearly distinguish for nationality of ownership, allowing for differences between<br />
firms that are foreign-owned <strong>and</strong> multinationals rooted in the domestic<br />
economy. Controlling for various firm- <strong>and</strong> industry-specific factors, it is<br />
found that while foreign multinationals are more likely to shut down operations<br />
compared to national firms in both manufacturing <strong>and</strong> service sectors,<br />
domestic multinationals only exhibit significantly higher exit rates in the<br />
manufacturing industries. The analysis has important policy implications,
especially in terms of the desirability of the large impact of multinational<br />
firms on employment <strong>and</strong> output generation in Belgium.<br />
While the literature on the relationship between firms’ international activities<br />
<strong>and</strong> their productivity abounds 2 , it remains largely silent on the sources<br />
of the productivity advantages associated with firms’ global integration. One<br />
way for firms to achieve productivity increases, is through the accumulation<br />
of knowledge or innovative output (Castellani <strong>and</strong> Zanfei, 2007). Chapter 4<br />
of this volume links in with this recent body of literature.<br />
Specifically, Chapter 4 investigates whether firms’ innovative output differs<br />
according to their global engagement status. The chapter relies on the innovation<br />
production function framework introduced by Mairesse <strong>and</strong> Mohnen<br />
(2002) <strong>and</strong> takes the endogeneity of research inputs specifically into account<br />
in the innovation output function. As a consequence, I am able to distinguish<br />
between the indirect (through higher research spending) <strong>and</strong> direct effect of<br />
firms’ global integration on its innovative output.<br />
From a policy perspective, the analysis in Chapter 4 yields several important<br />
insights. Overall, the empirical results indicate that (part of) the<br />
innovation premium associated with firms’ global engagement status in Belgium<br />
can be traced back to these firms’ different spending patterns, i.e. their<br />
higher likelihood to engage in continuous R&D spending. However, within<br />
the high technology sectors, firms with exposure on international markets<br />
(whether through exporting or FDI) are found to be more likely to generate<br />
innovative sales, conditional on their R&D spending patterns; while in low<br />
technology sectors, these firms are found to be less likely to be innovative.<br />
This result suggests that government policies aimed at the attraction of<br />
(foreign) multinational firms in order to stimulate innovative output (or inputs)<br />
are more likely to be successful (i.e. will generate a higher rate of<br />
return) when they are specifically targeted at high technology sectors. For<br />
the low technology sectors, the high returns to R&D obtained in these sectors,<br />
2 For reviews of this extensive literature, I refer to Greenaway <strong>and</strong> Kneller (2007) <strong>and</strong><br />
Wagner (2007). Important theoretical contributions in this field include Melitz (2003) <strong>and</strong><br />
Helpman, Melitz <strong>and</strong> Yeaple (2004).<br />
5
6 General Introduction<br />
especially for the selection stage of the model, suggest that policies targeted<br />
towards these sectors might be more successful when aimed at stimulating<br />
research efforts in general, indiscriminate of the global engagement status of<br />
the firm.
Chapter 1<br />
Total Factor Productivity<br />
Estimation: A Practical Review<br />
1.1 Introduction<br />
While the origins of total factor productivity analysis can be traced back<br />
to the seminal paper by Solow (Solow, 1957); recent years have seen a surge<br />
in both theoretical <strong>and</strong> empirical studies on total factor productivity (TFP).<br />
This renewed interest has been driven both by the increasing availability of<br />
firm-level data, allowing for estimation of TFP at the level of the individual<br />
establishment (Bartelsman <strong>and</strong> Doms, 2000); as well as by a number of<br />
methodological improvements that have emerged from the literature since the<br />
mid-1990s (Ackerberg, Benkard, Berry <strong>and</strong> Pakes, 2007, henceforth ABBP).<br />
Typically, establishment-level productivity studies assume output (usually<br />
measured as deflated sales or value added) to be a function of the inputs the<br />
firm employs <strong>and</strong> its productivity (Katayama, Lu <strong>and</strong> Tybout, 2005). The<br />
measure of TFP obtained as the residual in this functional relationship is then<br />
used to evaluate the impact of various policy measures, such as the extent of<br />
foreign ownership (eg. Smarzynska Javorcik, 2004), trade liberalization (eg.<br />
Pavcnik, 2002; Amiti <strong>and</strong> Konings, 2007; De Loecker, 2007) <strong>and</strong> antidumping<br />
protection (eg. Konings, 2008).<br />
∗ Published as LICOS Discussion Paper 182/2007.<br />
9
10 Total factor productivity estimation<br />
However, several methodological issues emerge when TFP is estimated using<br />
traditional methods, i.e. by applying Ordinary Least Squares (OLS) to<br />
a balanced panel of (continuing) firms. First, since productivity <strong>and</strong> input<br />
choices are likely to be correlated, OLS estimation of firm-level production<br />
functions introduces a simultaneity or endogeneity problem. Moreover, by<br />
using a balanced panel, no allowance is made for entry <strong>and</strong> exit, resulting in a<br />
selection bias. Although the simultaneity <strong>and</strong> selection bias are well-known 1 ;<br />
several other methodological issues have emerged more recently. Specifically,<br />
the typical practice of proxying for firm-level prices using industry-level deflators<br />
has been challenged (see for instance Katayama et al., 2005). Moreover,<br />
Bernard, Redding <strong>and</strong> Schott (2005) note that firms’ product choices are<br />
likely to be related to their productivity.<br />
In response to these methodological issues, several (parametric <strong>and</strong> semiparametric)<br />
estimators have been proposed in the literature. However, traditional<br />
estimators used to overcome endogeneity issues (i.e. fixed effects,<br />
instrumental variables <strong>and</strong> Generalized Method of Moments or GMM) have<br />
not proved satisfactory for the case of production functions. Likely causes<br />
for these estimators’ poor performance are related to their underlying assumptions.<br />
Therefore, a number of semiparametric alternatives have been<br />
proposed. Both Olley <strong>and</strong> Pakes (1996, henceforth OP) <strong>and</strong> Levinsohn <strong>and</strong><br />
Petrin (2003, henceforth LP) have developed a semiparametric estimator that<br />
addresses the simultaneity bias (<strong>and</strong> the selection bias in the case of the OP<br />
estimator). Several extensions to their model have already been introduced<br />
(eg. De Loecker, 2007).<br />
The present paper aims to provide empirical researchers with an overview<br />
of the methodological issues that arise when estimating TFP at the establishment<br />
level, as well as of the existing techniques designed to overcome them.<br />
Using data on single-product firms active in the Belgian food <strong>and</strong> beverages<br />
sector, I illustrate the biases introduced in traditional TFP estimates <strong>and</strong> discuss<br />
the performance of a number of alternative estimators that have been<br />
used in the literature. The food <strong>and</strong> beverages industry in Belgium presents<br />
1 They have been documented at least since Marschak <strong>and</strong> Andrews (1944) <strong>and</strong> Wed-<br />
ervang (1965) respectively.
Introduction 11<br />
an interesting case, since the sector underwent significant restructuring at<br />
the end of the 1990s following the dioxin crisis 2 .<br />
The production function coefficients obtained using various estimation<br />
techniques (i.e. OLS, fixed effects, GMM, Olley-Pakes, Levinsohn-Petrin<br />
<strong>and</strong> De Loecker) are generally in line with theoretical predictions. Aggregate<br />
productivity growth in the food <strong>and</strong> beverages industry increases significantly<br />
after 1999, consistent with the period of restructuring <strong>and</strong> increasing<br />
investments in the sector in response to the dioxin sc<strong>and</strong>al (VRWB, 2003).<br />
Decomposing aggregate productivity into a within productivity component<br />
<strong>and</strong> a reallocation share on the basis of firms’ turnover shares shows that<br />
this increase is mainly due to the average firm becoming more productive;<br />
while reallocation of market shares explains only a minor part. Applying<br />
the same decomposition using employment rather than output shares yields<br />
similar results.<br />
The rest of the paper is structured as follows. Section 1.2 provides an<br />
overview of the methodological issues arising when estimating firm-level TFP.<br />
In section 1.3, several estimation methods are reviewed, with specific attention<br />
for their advantages <strong>and</strong> drawbacks. Section 1.4 illustrates the different<br />
methodologies for the Belgian food <strong>and</strong> beverages industry (NACE 15). Section<br />
1.5 concludes.<br />
Given the vast amount of papers that continue to emerge in this field, a<br />
number of choices have to be made at the outset. First, since primary interest<br />
is in consistent estimation of TFP, attention will mostly be limited to recent<br />
papers, i.e. from 1990 onwards. Second, only parametric <strong>and</strong> semiparametric<br />
estimators applied to TFP estimation will be discussed here. Van<br />
Biesebroeck (2007) provides an excellent review of several non-parametric<br />
methods (specifically, index numbers <strong>and</strong> data envelopment analysis) 3 used<br />
2 The Economist (1999). Excessive concentrations of dioxin were found in eggs, chicken,<br />
pork <strong>and</strong> milk, caused by contaminated animal food.<br />
3 Van Biesebroeck (2007) compares the robustness of five commonly used techniques<br />
to estimate TFP: index numbers, data envelopment analysis, stochastic frontiers, GMM<br />
<strong>and</strong> semiparametric estimation; to the presence of measurement error <strong>and</strong> to differences<br />
in production technology.
12 Total factor productivity estimation<br />
to estimate firm-level productivity. Finally, given the multitude of papers<br />
dealing with the impact of some policy measure on TFP, it is beyond the<br />
scope of the present paper to provide a complete review of all the empirical<br />
work done in this area. Selection of which references to include is therefore<br />
based on their methodological <strong>and</strong> econometric contributions to the field.<br />
1.2 Total factor productivity: Methodologi-<br />
cal issues<br />
1.2.1 Some preliminaries on the production function<br />
I start by assuming that production takes the form of a Cobb-Douglas<br />
production function. However, as shown by ABBP (2007); estimation methods<br />
discussed in the next section carry over to other types of production<br />
functions, provided some basic requirements are met 4 . Specifically, the production<br />
function looks as follows:<br />
Yit = AitK βk<br />
it Lβl it Mβm it<br />
(1.1)<br />
where Yit represents physical output of firm i in period t; Kit, Lit <strong>and</strong><br />
Mit are inputs of capital, labor <strong>and</strong> materials respectively <strong>and</strong> Ait is the<br />
Hicksian neutral efficiency level of firm i in period t.<br />
While Yit, Kit, Lit <strong>and</strong> Mit are all observed by the econometrician (although<br />
usually in value terms rather than in quantities, see below), Ait is<br />
unobservable to the researcher. Taking natural logs of (1.1) results in a linear<br />
production function,<br />
yit = β0 + βkkit + βllit + βmmit + εit<br />
where lower-case letters refer to natural logarithms <strong>and</strong><br />
ln (Ait) = β0 + εit<br />
4 Variable inputs need to have positive cross-partials with productivity <strong>and</strong> the value<br />
of the firm has to be increasing in the amount of fixed inputs used (ABBP, 2007).
Total factor productivity: Methodological issues 13<br />
.<br />
While β0 measures the mean efficiency level across firms <strong>and</strong> over time;<br />
εit is the time- <strong>and</strong> producer-specific deviation from that mean, which can<br />
then be further decomposed into an observable (or at least predictable) <strong>and</strong><br />
unobservable component. This results in the following equation, which will<br />
serve as the starting point for the rest of this <strong>and</strong> the next section:<br />
yit = β0 + βkkit + βllit + βmmit + ωit + u q<br />
it<br />
(1.2)<br />
where ωit represents firm-level productivity5 <strong>and</strong> u q<br />
it is an i.i.d. component,<br />
representing unexpected deviations from the mean due to measurement<br />
error, unexpected delays or other external circumstances.<br />
Typically, empirical researchers estimate (1.2) <strong>and</strong> solve for ωit. Estimated<br />
productivity can then be calculated as follows:<br />
ˆωit = yit − ˆ βkkit − ˆ βllit − ˆ βmmit<br />
(1.3)<br />
<strong>and</strong> productivity in levels can be obtained as the exponential of ˆωit, i.e.<br />
ˆΩit = exp (ˆωit). The productivity measure resulting from (1.3) can be used to<br />
evaluate the influence <strong>and</strong> impact of various policy variables directly at the<br />
firm level; or alternatively, firm-level TFP can be aggregated to the industry<br />
level by calculating the share-weighted average of ˆ Ωit.<br />
Weights used to aggregate firm-level TFP can be either firm-level output<br />
shares, as in OP; or employment shares, as in De Loecker <strong>and</strong> Konings<br />
(2006). As will be illustrated in section 1.4, normalized industry productivity<br />
can then be further decomposed into an unweighted average <strong>and</strong> a (crosssectional)<br />
sample covariance term. While differences in the unweighted average<br />
over time refer to within-firm changes in TFP; changes in the sample<br />
covariance term signal reallocation of market shares as the driver of productivity<br />
shifts (Olley <strong>and</strong> Pakes, 1996; De Loecker <strong>and</strong> Konings, 2006).<br />
5 The productivity term is identified through the assumption that ωit is a state variable<br />
in the firm’s decision problem (i.e. it is a determinant of both firm selection <strong>and</strong> input<br />
dem<strong>and</strong> decisions), while u q<br />
it is either measurement error or a non-predictable productivity<br />
shock (Olley <strong>and</strong> Pakes, 1996).
14 Total factor productivity estimation<br />
In what follows, it will be shown that estimating (1.2) using OLS leads to<br />
biased productivity estimates, caused by the endogeneity of input choices <strong>and</strong><br />
selection bias. Moreover, in the presence of imperfect competition in output<br />
<strong>and</strong>/or input markets, an omitted variable bias will arise in st<strong>and</strong>ard TFP<br />
estimation if data on physical inputs <strong>and</strong> output <strong>and</strong> their corresponding<br />
firm-level prices are unavailable. Finally, if firms produce multiple products,<br />
potentially differing in their production technology; failure to estimate the<br />
production function at the appropriate product level, rather than at the firm<br />
level, will also introduce a bias in st<strong>and</strong>ard TFP measures. I will discuss<br />
each of these problems in turn.<br />
1.2.2 Endogeneity of attrition or selection bias<br />
Traditionally, entry <strong>and</strong> exit of firms is accounted for in TFP estimation<br />
by constructing a balanced panel; i.e. by omitting all firms that enter or exit<br />
over the sample period (Olley <strong>and</strong> Pakes, 1996). However, several theoretical<br />
models (eg. Jovanovic, 1982; Hopenhayn, 1992) predict that the growth <strong>and</strong><br />
exit of firms is motivated to a large extent by productivity differences at<br />
the firm level. Empirically, Fariñas <strong>and</strong> Ruano (2005) find, for a sample of<br />
Spanish manufacturing firms, that entry <strong>and</strong> exit decisions are systematically<br />
related to differences in productivity. They show that firms’ exit patterns<br />
reflect initial productivity differences, leading to the prediction that higher<br />
productivity will lower the exit probability at the firm level.<br />
Moreover, Dunne et al. (1988) report exit rates in excess of 30 percent<br />
between two census years in US manufacturing <strong>and</strong> a strong correlation between<br />
entry <strong>and</strong> exit rates in the data. Since low productivity firms have a<br />
stronger tendency to exit than their more productive counterparts, omitting<br />
all firms subject to entry or exit is likely to lead to biased results. If firms<br />
have some knowledge about their productivity level ωit prior to their exit,<br />
this will generate correlation between εit <strong>and</strong> the fixed input capital, conditional<br />
on being in the data set (ABBP, 2007). This correlation has its origin<br />
in the fact that firms with a higher capital supply will (ceteris paribus) be<br />
able to withst<strong>and</strong> lower ωit without exiting.
Total factor productivity: Methodological issues 15<br />
In sum, the selection bias or “endogeneity of attrition”- problem will generate<br />
a negative correlation between εit <strong>and</strong> Kit, causing the capital coefficient<br />
to be biased downwards in a balanced sample (i.e. not taking entry <strong>and</strong> exit<br />
into account). While this selection problem has been discussed in the literature<br />
at least since the work of Wedervang (1965), the estimation algorithm<br />
introduced by Olley <strong>and</strong> Pakes (1996) was the first to take it explicitly 6 into<br />
account.<br />
1.2.3 Endogeneity of input choice or simultaneity bias<br />
Although (1.2) can be estimated using Ordinary Least Squares (OLS),<br />
this method requires that the inputs in the production function are exogenous<br />
or, in other words, determined independently from the firm’s efficiency<br />
level. Marschak <strong>and</strong> Andrews (1944) already noted that inputs in the production<br />
function are not independently chosen, but rather determined by<br />
the characteristics of the firm, including its efficiency. This “endogeneity of<br />
inputs” or simultaneity bias is defined as the correlation between the level of<br />
inputs chosen <strong>and</strong> unobserved productivity shocks (De Loecker, 2007).<br />
If the firm has prior knowledge of ωit at the time input decisions are made,<br />
endogeneity arises since input quantities will be (partly) determined by prior<br />
beliefs about its productivity (Olley <strong>and</strong> Pakes, 1996; ABBP, 2007). Hence,<br />
if there is serial correlation in ωit, a positive productivity shock will lead to<br />
increased variable input usage; i.e. E (xitωit) > 0 , where xit = (lit, mit);<br />
introducing an upward bias in the input coefficients for labor <strong>and</strong> materials<br />
(De Loecker, 2007). In the presence of many inputs <strong>and</strong> simultaneity<br />
issues, it is generally impossible to determine the direction of the bias in the<br />
capital coefficient. Levinsohn <strong>and</strong> Petrin (2003) illustrate, for a two-input<br />
production function where labor is the only freely variable input <strong>and</strong> capital<br />
is quasi-fixed, that the capital coefficient will be biased downward if a positive<br />
correlation exists between labor <strong>and</strong> capital (which is the most likely<br />
setup).<br />
6 It is possible to correct implicitly for the selection bias by using an unbalanced panel of<br />
firms. But, as will be shown in section 1.3, Olley <strong>and</strong> Pakes (1996) introduce an additional<br />
(explicit) correction in their estimation algorithm, i.e. they take the firm-level survival<br />
probability into account.
16 Total factor productivity estimation<br />
Traditional methods to deal with heterogeneity <strong>and</strong> endogeneity issues include<br />
fixed effects <strong>and</strong> instrumental variables estimation (Wooldridge, 2005).<br />
However, as I will discuss below, both alternatives to OLS are plagued by<br />
a number of problems. Both the estimation algorithm introduced by Olley<br />
<strong>and</strong> Pakes (1996) <strong>and</strong> Levinsohn <strong>and</strong> Petrin (2003) provide a more adequate<br />
solution to the simultaneity problem discussed here.<br />
1.2.4 Omitted output price bias<br />
In the absence of information on firm-level prices, which are typically unavailable<br />
to the researcher, industry-level price indices are usually applied<br />
to deflate firm-level sales (<strong>and</strong> hence obtain a measure of the firm’s output)<br />
in traditional production function estimates (De Loecker, 2007). However, if<br />
firm-level price variation is correlated with input choice; this will result in biased<br />
input coefficients. The problem can be illustrated as follows. Replacing<br />
output in quantities by deflated sales in (1.2) results in the following model:<br />
rit = pit + yit − p it<br />
= β0 + βkkit + βllit + βmmit + (pit − p it) + ωit + u q<br />
it<br />
(1.4)<br />
where rit represents deflated sales, pit are firm-level prices <strong>and</strong> p it is<br />
the industry-level price deflator, all in logarithmic form. For now, inputs<br />
are still assumed to be available in quantities. From (1.4) it is clear that<br />
if input choice is correlated with unobserved firm-level price differences, i.e.<br />
E (xit (pit − p it)) = 0, where xit = (lit, mit, kit); a bias is introduced in the<br />
input coefficients.<br />
Assuming inputs <strong>and</strong> output are positively correlated <strong>and</strong> output <strong>and</strong> price<br />
are negatively correlated (as in a st<strong>and</strong>ard dem<strong>and</strong> <strong>and</strong> supply framework);<br />
the correlation between (variable) inputs <strong>and</strong> firm-level prices will be negative;<br />
resulting in a negative bias for the coefficients on labor <strong>and</strong> materials<br />
(De Loecker, 2007). Hence, the bias resulting from using industry-level price<br />
deflators rather than firm-level prices to deflate sales, will generally be opposite<br />
to the bias introduced by simultaneity of input choice <strong>and</strong> productivity<br />
discussed in the previous section.
Total factor productivity: Methodological issues 17<br />
Since the omitted output price bias will only arise if industry-level price<br />
deflators are used <strong>and</strong> if firm-level prices deviate from these deflators (i.e. in<br />
the presence of imperfect competition), it can be avoided by using quantities<br />
of output rather than deflated sales. However, since this requires information<br />
on actual firm level prices, it is a very rare setup. Exceptions include<br />
Dunne <strong>and</strong> Roberts (1992), Eslava, Haltiwanger, Kugler <strong>and</strong> Kugler (2004),<br />
Foster, Haltiwanger <strong>and</strong> Syverson (2008), Jaum<strong>and</strong>reu <strong>and</strong> Mairesse (2004)<br />
<strong>and</strong> Mairesse <strong>and</strong> Jaum<strong>and</strong>reu (2005). Alternatively, it is possible (in the<br />
absence of information on prices) to introduce dem<strong>and</strong> for output into the<br />
system <strong>and</strong> solve for firm-level prices 7 , as suggested by Klette <strong>and</strong> Griliches<br />
(1996), Levinsohn <strong>and</strong> Melitz (2002) <strong>and</strong>, in the context of the Olley-Pakes<br />
semiparametric estimator, De Loecker (2007). The specifics of the latter<br />
estimation algorithm will be discussed in section 1.3.<br />
1.2.5 Omitted input price bias<br />
In the presence of imperfect competition in input markets, input prices<br />
are likely to be firm-specific. However, since input prices (like output prices)<br />
are typically unavailable, quantities of inputs are usually proxied by deflated<br />
values of inputs for capital <strong>and</strong> materials (the amount of labor used tends to<br />
be available in annual accounts data commonly used to estimate production<br />
function relationships). Assuming that quantities of output are given, this<br />
leads to the following relationship:<br />
yit = β0 + βk kit + βllit + βm mit + ωit + u q<br />
it<br />
<br />
kit + p k it − pk <br />
it + βllit<br />
+ βm (mit + p m it − pmit ) + ωit + u q<br />
it<br />
yit = β0 + βk<br />
(1.5)<br />
where kit <strong>and</strong> mit are deflated values of capital <strong>and</strong> material inputs<br />
respectively, pk it <strong>and</strong> pm it represent firm-level prices of these inputs <strong>and</strong> pk it <strong>and</strong><br />
pm it refer to their industry-level price indices. From (1.5) it is clear that in<br />
the presence of unobserved firm-level input price differences, coefficients on<br />
7 Ornaghi (2006) criticizes this approach however, see section 1.3 below.
18 Total factor productivity estimation<br />
kit <strong>and</strong> mit will be biased.<br />
De Loecker (2007) argues that if imperfect output markets are treated explicitly,<br />
this can partly take care of the omitted input price bias, to the extent<br />
that higher input prices are reflected in higher output prices; which in turn<br />
depends on the relevant firm-level mark-up. However, Levinsohn <strong>and</strong> Melitz<br />
(2002) argue that even with a competitive factor market, adjustment costs<br />
will lead to differences in the shadow price of the input index across firms,<br />
induced by differences in current levels of the quasi-fixed factors (capital).<br />
Katayama et al. (2005) similarly argue that factor prices are important to<br />
take into account in TFP estimation procedures.<br />
Similar to the situation of imperfect competition in output markets, a<br />
number of studies are able to exploit information on input prices <strong>and</strong> quantities<br />
to resolve the omitted price bias, examples include Eslava et al. (2004)<br />
<strong>and</strong> Ornaghi (2006). A formal solution for the bias induced by firm-specific<br />
input prices has yet to be introduced.<br />
1.2.6 Endogeneity of the product mix (multi-product<br />
firms)<br />
Bernard, Redding <strong>and</strong> Schott (2005, henceforth BRS) argue that firms’<br />
decisions on which goods to produce, are typically made at a more disaggregated<br />
level than is recorded in manufacturing data sets (either using census<br />
data or annual accounts data). If firms produce multiple products within<br />
the same industry <strong>and</strong> if these products differ in their production technology<br />
or in the dem<strong>and</strong> they face, this will lead to biased TFP estimates, since<br />
the production function assumes identical production techniques <strong>and</strong> final<br />
dem<strong>and</strong> (through the use of common output price deflators) across products<br />
manufactured by a single firm.<br />
BRS (2006b) have examined the pervasiveness <strong>and</strong> determinants of product<br />
switching among US manufacturing firms for the period 1972-1992. They<br />
find that two-thirds of firms alter their mix of five-digit SIC codes every five<br />
years <strong>and</strong> they further demonstrate that product adding <strong>and</strong> dropping by
Total factor productivity: Methodological issues 19<br />
surviving firms accounts for nearly one-third of the aggregate growth in real<br />
US manufacturing output between 1992 <strong>and</strong> 1997.<br />
In principle, consistent estimation of TFP in the presence of multi-product<br />
firms requires information on the product mix, product-level output, inputs,<br />
as well as prices. Given these high requirements in terms of data, BRS (2005)<br />
suggest several (partial) solutions to circumvent the bias introduced by multiproduct<br />
firms. First, in the absence of information on inputs <strong>and</strong> outputs at<br />
the product level, it is possible to sort firms into groups that make a single<br />
product, which will eliminate the bias introduced by endogenous product<br />
choice. Alternatively, if the researcher has knowledge of the number <strong>and</strong><br />
type of products produced by each firm, consistent estimates of productivity<br />
can be obtained by allowing the parameters of the production technology to<br />
vary across firms making different products. De Loecker (2007) is among the<br />
first to take the number of products as well as product-specific dem<strong>and</strong> into<br />
account when estimating TFP for the Belgian textiles sector. However, his<br />
estimation procedure provides only a partial solution to the bias introduced<br />
by endogenous product choice (cfr. section 1.3).<br />
1.2.7 Summary of methodological issues<br />
Traditional productivity estimates, obtained as the residual from a balanced<br />
OLS regression of deflated output on deflated inputs <strong>and</strong> a constant,<br />
are plagued by a number of econometric <strong>and</strong> specification issues. Table 1.1<br />
provides an overview.<br />
First, given the prevalence of entry <strong>and</strong> exit in manufacturing populations,<br />
the use of a balanced panel introduces a selection bias in the sample, causing<br />
the capital coefficient to be biased downward. Second, if firms have some<br />
prior knowledge or expectations concerning their efficiency, current input<br />
choice will be correlated with productivity. Coefficients on variable inputs<br />
will be biased upward as a result of this endogeneity or simultaneity problem,<br />
while the coefficient on capital will be biased downward provided the<br />
correlation between labor <strong>and</strong> capital is positive.
20 Total factor productivity estimation<br />
Table 1.1: TFP estimation: Summary of methodological issues<br />
Multi-product firms Endogenous product choice: undetermined Bernard, Redding, Schott (2005)<br />
Differences in production technologies across Bernard, Redding, Schott (2006b)<br />
products produced by single firm. De Loecker (2007)<br />
Omitted input Imperfect competition in input markets: downward bias in βl Levinsohn <strong>and</strong> Melitz (2002)<br />
price bias Correlation between firm-level deviation of downward bias in βm <br />
Katayama et al. (2005)<br />
input price deflators<br />
<strong>and</strong> inputs xit. upward bias in βk De Loecker (2007)<br />
p k,m<br />
it<br />
− pk,m it<br />
Selection bias Endogeneity of attrition: downward bias in βk Wedervang (1965)<br />
Correlation between εit <strong>and</strong> Kit (the quasi-fixed Olley <strong>and</strong> Pakes (1996)<br />
input), conditional on being in the data set. ABBP (2007)<br />
Simultaneity bias Endogeneity of inputs: upward bias in βl Marschak <strong>and</strong> Andrews (1944)<br />
Correlation between εit <strong>and</strong> inputs xit if firms’ upward bias in βm Olley <strong>and</strong> Pakes (1996)<br />
prior beliefs about εit influence its choice of inputs. downward bias in βk Levinsohn <strong>and</strong> Petrin (2003)<br />
ABBP (2007)<br />
Ackerberg et al. (2006)<br />
Omitted output Imperfect competition in output markets: downward bias in βl Klette <strong>and</strong> Griliches (1996)<br />
price bias Correlation between firm-level deviation of downward bias in βm Levinsohn <strong>and</strong> Melitz (2002)<br />
output price deflator (pit − pit) <strong>and</strong> inputs xit. upward bias in βk De Loecker (2007)<br />
Origin of the bias Definition Direction of the bias References
Total factor productivity estimation 21<br />
Third, in the presence of imperfect competition in input <strong>and</strong>/or output<br />
markets, the failure to take firm-level deviations from the industry-level price<br />
deflator into account will result in an omitted output <strong>and</strong>/or input price bias.<br />
The resulting bias(es) will, in a st<strong>and</strong>ard dem<strong>and</strong>/supply framework, work<br />
in the opposite direction as the simultaneity bias, rendering any prior on the<br />
overall direction of the bias hard. Finally, if firms produce multiple products,<br />
which potentially differ in terms of their production technology <strong>and</strong> dem<strong>and</strong>,<br />
an additional bias will be introduced in traditional TFP estimates. I now<br />
turn to the various estimators that have been introduced in the literature on<br />
consistent estimation of total factor productivity.<br />
1.3 Total factor productivity estimation<br />
1.3.1 Fixed effects estimation<br />
By assuming that ωit is plant-specific, but time-invariant; it is possible to<br />
estimate (1.2) using a fixed effects estimator (Pavcnik, 2002; Levinsohn <strong>and</strong><br />
Petrin, 2003). The estimating equation then becomes:<br />
yit = β0 + βkkit + βllit + βmmit + ωi + u q<br />
it<br />
(1.6)<br />
Equation (1.6) can be estimated in levels using a Least Square Dummy<br />
Variable Estimator (LSDV, i.e. including firm-specific effects) or in first (or<br />
mean) differences. Provided unobserved productivity ωit does not vary over<br />
time, estimation of (1.6) will result in consistent coefficients on labor, capital<br />
<strong>and</strong> materials.<br />
Fixed effects or within estimators have a long tradition in the production<br />
function literature, in fact they were introduced to economics in this context<br />
(Mundlak, 1961; Hoch, 1962). By using only the within-firm variation in the<br />
sample, the fixed effects estimator overcomes the simultaneity bias discussed<br />
in the previous section (ABBP, 2007). Moreover, to the extent that exit<br />
decisions are determined by the time-invariant, firm-specific effects ωi, <strong>and</strong><br />
, the within estimator also eliminates the selection bias, caused by<br />
endogenous exit in the sample. As a result, estimation of (1.6) using either<br />
not by u q<br />
it
22 Total factor productivity estimation<br />
the balanced or unbalanced (i.e. allowing for entry <strong>and</strong> exit) sample should<br />
result in similar estimates for the coefficients.<br />
In spite of the attractive properties of the fixed effects estimator, it does<br />
not perform well in practice (ABBP, 2007). Estimation of (1.6) often leads<br />
to unreasonably low estimates of the capital coefficient. Moreover, Olley <strong>and</strong><br />
Pakes (1996) perform fixed effects on the balanced <strong>and</strong> unbalanced sample<br />
<strong>and</strong> find large differences between the two sets of coefficients, suggesting the<br />
assumptions underlying the model are invalid. The time-invariant nature of<br />
ωi in the fixed effects model has been relaxed by Blundell <strong>and</strong> Bond (1999)<br />
in the context of production functions, by allowing productivity to be decomposed<br />
into a fixed effect <strong>and</strong> an autoregressive AR(1)-component.<br />
1.3.2 Instrumental variables (IV) <strong>and</strong> GMM<br />
An alternative method to achieve consistency of coefficients in the production<br />
function is by instrumenting the independent variables that cause<br />
the endogeneity problems (i.e. the inputs in the production function) by<br />
regressors that are correlated with these inputs, but uncorrelated with unobserved<br />
productivity. To achieve consistency of this IV estimator, three<br />
requirements have to be met (ABBP, 2007). First, instruments need to be<br />
correlated with the endogenous regressors (inputs). Second, the instruments<br />
can not enter the production function directly <strong>and</strong> finally, instruments need<br />
to be uncorrelated with the error term.<br />
Assuming input <strong>and</strong> output markets operate perfectly competitive, input<br />
<strong>and</strong> output prices are natural choices of instruments for the production function<br />
(ABBP, 2007). Other examples of instruments include variables that<br />
shift the dem<strong>and</strong> for output or the supply of inputs. Like the fixed effects<br />
estimator, the IV estimator has not been particularly successful in practice.<br />
One of the obvious shortcomings of the technique is the lack of appropriate<br />
instruments in many data sets. Input <strong>and</strong> output prices are usually not<br />
reported in typical plant or firm level data sets <strong>and</strong> if they are reported, frequently<br />
not enough variation exists in the data in order to identify coefficients<br />
of the production function (ABBP, 2007). Moreover, while estimation using<br />
IV techniques overcomes the simultaneity bias (provided the instruments are
Total factor productivity estimation 23<br />
appropriate), it does not provide a solution for the selection issues. If input<br />
prices are used as instruments for input quantities <strong>and</strong> if exit decisions are<br />
driven (in part) by changes in these input prices, results will remain biased.<br />
In response to these unsatisfactory results, Blundell <strong>and</strong> Bond (1999) propose<br />
an extended GMM estimator. They attribute the bad performance of<br />
st<strong>and</strong>ard IV estimators to the weak instruments used for identification, i.e.<br />
lagged levels of variables are often used as instruments in the estimation<br />
in first differences. They propose an extended GMM estimator using lagged<br />
first-differences of the variables as instruments in the level equations <strong>and</strong> find<br />
that this estimator yields more reasonable parameter estimates. As already<br />
noted above, they also stress the importance of allowing for an autoregressive<br />
component in ωit.<br />
1.3.3 Olley-Pakes estimation algorithm<br />
As an alternative to the methods discussed above; Olley <strong>and</strong> Pakes (1996)<br />
have developed a consistent semiparametric estimator. This estimator solves<br />
the simultaneity problem by using the firm’s investment decision to proxy for<br />
unobserved productivity shocks. Selection issues are addressed by incorporating<br />
an exit rule into the model. In what follows, the proposed methodology<br />
will be discussed in somewhat more detail. It should be noted here however,<br />
that the focus in this section is on the estimation methodology. For the more<br />
technical aspects (<strong>and</strong> related proofs), the interested reader is referred to<br />
Ericson <strong>and</strong> Pakes (1995) <strong>and</strong> Olley <strong>and</strong> Pakes (1996).<br />
Olley <strong>and</strong> Pakes (1996) were the first to introduce an estimation algorithm<br />
that takes both the selection <strong>and</strong> simultaneity problem explicitly into<br />
account. They develop a dynamic model of firm behavior that allows for<br />
idiosyncratic productivity shocks, as well as for entry <strong>and</strong> exit. At the start<br />
of each period, each incumbent firm decides whether to exit or to continue<br />
its operations. If it exits, it receives a particular sell-off value <strong>and</strong> it never<br />
re-enters. If it continues, it chooses an appropriate level of variable inputs<br />
<strong>and</strong> investment. The firm is assumed to maximize the expected discounted<br />
value of net cash flows <strong>and</strong> investment <strong>and</strong> exit decisions will depend on the<br />
firm’s perceptions about the distribution of future market structure, given
24 Total factor productivity estimation<br />
the information currently available. Both the lower bound to productivity<br />
(i.e. the cut-off value below which the firm chooses to exit) <strong>and</strong> the investment<br />
decision are determined as part of a Markov perfect Nash equilibrium<br />
<strong>and</strong> will hence depend on all parameters determining equilibrium behavior.<br />
In order to achieve consistency, a number of assumptions need to be made.<br />
First, the model assumes there is only one unobserved state variable at the<br />
firm level, i.e. its productivity. Second, the model imposes monotonicity<br />
on the investment variable, in order to ensure invertibility of the investment<br />
dem<strong>and</strong> function. This implies that investment has to be increasing in productivity,<br />
conditional on the values of all state variables. As a consequence,<br />
only non-negative values of investment can be used in the analysis. This<br />
condition needs to hold for at least some known subset of the sample (see<br />
below). Finally, if industry-wide price indices are used to deflate inputs <strong>and</strong><br />
output in value terms to proxy for their respective quantities , it is implicitly<br />
assumed that all firms in the industry face common input <strong>and</strong> output prices<br />
(Ackerberg, Benkard, Berry <strong>and</strong> Pakes, 2007).<br />
Starting out from the basic Cobb-Douglas production function 8 given by<br />
(1.2), the estimation procedure can be described as follows. Capital is a state<br />
variable, only affected by current <strong>and</strong> past levels of ωit. Investment can be<br />
calculated as:<br />
Iit = Kit+1 − (1 − δ) Kit<br />
Hence, investment decisions at the firm level can be shown to depend on<br />
capital <strong>and</strong> productivity or iit = it (kit, ωit), where lower-case notation refers<br />
to logarithmic transformation of variables, as above. Provided investment<br />
is strictly increasing in productivity, conditional on capital, this investment<br />
decision can be inverted, allowing us to express unobserved productivity as<br />
a function of observables:<br />
ωit = ht (kit, iit)<br />
where ht (.) = i −1<br />
t (.). Using this information, (1.2) can be rewritten as:<br />
8 The production function in (1.2) differs from that employed by OP in two respects.<br />
First, OP include age as an additional state variable, which is omitted here. Second, OP<br />
start out from a value added production function, i.e. including only labor <strong>and</strong> capital as<br />
production factors.
Total factor productivity estimation 25<br />
yit = β0 + βkkit + βllit + βmmit + ht (kit, iit) + u q<br />
it<br />
Next, define the function ϕ (iit, kit) as follows:<br />
ϕ (iit, kit) = β0 + βkkit + ht (iit, kit)<br />
(1.7)<br />
Estimation of (1.7) proceeds in two steps (OP, 1996). In the first stage of<br />
the estimation algorithm, the following equation is estimated using OLS:<br />
yit = βllit + βmmit + ϕ (iit, kit) + u q<br />
it<br />
(1.8)<br />
where ϕ (iit, kit) is approximated by a higher order polynomial in iit <strong>and</strong><br />
kit (including a constant term). Estimation of (1.8) results in a consistent<br />
estimate of the coefficients on labor <strong>and</strong> materials (the variable factors of<br />
production).<br />
In order to recover the coefficient on the capital variable, it is necessary<br />
to exploit information on firm dynamics. Productivity is assumed to follow<br />
a first order Markov process, i.e. ωit+1 = E (ωit+1|ωit) + ξit+1, where<br />
ξit+1 represents the news component <strong>and</strong> is assumed to be uncorrelated with<br />
productivity <strong>and</strong> capital in period t+1. As noted above, firms will continue<br />
to operate provided their productivity level exceeds the lower bound, i.e.<br />
χit+1 = 1 if ωit+1 ≥ ω it+1, where χit+1 is a survival indicator variable. Since<br />
the news component ξit+1 is correlated with the variable inputs; labor <strong>and</strong><br />
material inputs are subtracted from the log of output. Considering the expectation<br />
of E (yit+1 − βllit+1 − βmmit+1), conditional on the survival of the<br />
firm results in the following expression:<br />
E [yit+1 − βllit+1 − βmmit+1|kit+1, χit+1 = 1]<br />
= β0 + βkkit+1 + E [ωit+1|ωit, χit+1 = 1]<br />
The second stage of the estimation algorithm can then be derived as follows:
26 Total factor productivity estimation<br />
yit+1 −βllit+1 − βmmit+1<br />
= β0 + βkkit+1 + E (ωit+1|ωit, χit+1) + ξit+1 + u q<br />
it+1<br />
= β0 + βkkit+1 + g (Pit, ϕt − βkkit) + ξit+1 + u q<br />
it+1<br />
(1.9)<br />
where E (ωit+1|ωit, χit+1) = g (Pit, ϕit − βkkit) follows from the law of<br />
motion for the productivity shocks <strong>and</strong> Pit is the probability of survival of<br />
firm i in the next period 9 , i.e. Pit = Pr {χit+1 = 1}. A consistent estimate<br />
of the coefficient on capital is obtained by substituting the estimated coefficients<br />
on labor <strong>and</strong> materials from the first stage, as well as the estimated<br />
survival probability in (1.8). As in the first stage of the estimation procedure,<br />
the function g (Pit, ϕit − βkkit) is approximated using a higher order polynomial<br />
expansion in Pit <strong>and</strong> ϕit − βkkit. Finally, this results in the following<br />
estimating equation:<br />
yit+1 −βllit+1 − βmmit+1<br />
<br />
= β0 + βkkit+1 + g Pit, ϕt − <br />
βkkit<br />
+ ξit+1 + u q<br />
it+1<br />
(1.10)<br />
The coefficient on capital can then be obtained by applying Non-Linear<br />
Least Squares on (1.10).<br />
1.3.4 Levinsohn-Petrin estimation algorithm<br />
While Olley <strong>and</strong> Pakes (1996) use the investment decision to proxy for<br />
unobserved productivity; Levinsohn <strong>and</strong> Petrin (2003) rely on intermediate<br />
inputs as a proxy. The monotonicity condition of OP requires that investment<br />
is strictly increasing in productivity. Since this implies that only observations<br />
with positive investment can be used when estimating (1.8) <strong>and</strong> (1.10), this<br />
can result in a significant loss in efficiency, depending on the data at h<strong>and</strong>.<br />
9 An estimate of Pit can be obtained by estimating a probit model, where the dependent<br />
variable is a survival dummy (i.e. dummy equal to one if the firm survives in a particular<br />
period). Left-h<strong>and</strong> side variables are the same polynomial terms used in the first stage<br />
of the estimation procedure, i.e. a higher-order polynomial in investment <strong>and</strong> capital,<br />
including a constant term. Pit can then be obtained as the predicted survival probability<br />
from this regression.
Total factor productivity estimation 27<br />
Moreover, if firms report zero investment in a significant number of cases, this<br />
casts doubt on the validity of the monotonicity condition. Hence, Levinsohn<br />
<strong>and</strong> Petrin (2003) use intermediate inputs rather than investment as a proxy.<br />
Since firms typically report positive use of materials <strong>and</strong> energy in each<br />
year, it is possible to retain most observations; which also implies that the<br />
monotonicity condition is more likely to hold.<br />
Their estimation algorithm differs from that introduced by OP in two important<br />
respects. First, they use intermediate inputs to proxy for unobserved<br />
productivity, rather than investment. This implies that intermediate inputs<br />
(materials in this case) are expressed as a function of capital <strong>and</strong> productivity,<br />
i.e. mit = mt (kit, ωit). Provided the monotonicity condition is met <strong>and</strong><br />
materials inputs are strictly increasing in ωit, this function can be inverted,<br />
again allowing us to express unobserved productivity as a function of observables,<br />
i.e. ωit = st (kit, mit), where st (.) = m −1<br />
t (.). Using this expression, it<br />
is possible to rewrite (1.2), analogous to the OP-approach described above.<br />
yit = β0 + βkkit + βllit + βmmit + st (kit, mit) + u q<br />
it<br />
(1.11)<br />
It should be noted that the coefficient on the proxy variable, i.e. materials;<br />
is now only recovered in the second stage of the estimation algorithm, rather<br />
than in the first as for the OP approach. The second difference between<br />
the approach used by OP <strong>and</strong> LP is in the correction for the selection bias.<br />
While OP allow for both an unbalanced panel as well as the incorporation<br />
of the survival probability in the second stage of the estimation algorithm,<br />
LP do not incorporate the survival probability in the second stage; since the<br />
efficiency gains associated with it in the empirical results presented by OP<br />
were very small provided an unbalanced panel was used. Apart from using<br />
materials instead of investment as a proxy <strong>and</strong> omitting the survival correction<br />
in the second stage 10 , estimation is fully analogous to the approach used<br />
by OP <strong>and</strong> summarized above. Moreover, Petrin, Levinsohn <strong>and</strong> Poi (2003)<br />
have developed a Stata program implementing the LP approach (levpet). For<br />
further details on the LP approach, I refer to LP <strong>and</strong> Petrin et al. (2003).<br />
10 In principle it is possible to implement the explicit correction for firm survival in the<br />
LP estimation algorithm.
28 Total factor productivity estimation<br />
1.3.5 Olley-Pakes versus Levinsohn-Petrin<br />
As was noted above, the OP <strong>and</strong> LP estimation algorithms are analogous<br />
apart from the use of different proxies <strong>and</strong> the in- or exclusion of the survival<br />
probability to correct for the selection bias. How then, is one to choose<br />
among the two estimators? I will briefly discuss some of the results emerging<br />
from the literature here.<br />
It is useful to start with the most obvious shortcoming of the OP estimation<br />
algorithm, i.e. the invertibility condition, which implies that only firms with<br />
positive investment can be included in the analysis. Although consistent<br />
production function coefficients can be obtained by estimating (1.10) for the<br />
subset in the sample with recorded positive investment; this implies a loss<br />
in efficiency <strong>and</strong>, particularly if there are few firms with positive investment<br />
flows in the industry, can cast doubt on the monotonicity condition (see<br />
above).<br />
Moreover, according to Ackerberg, Caves <strong>and</strong> Frazer (2006), collinearity<br />
between labor <strong>and</strong> the non-parametric terms (i.e. the polynomial in materials<br />
<strong>and</strong> capital for LP <strong>and</strong> in investment <strong>and</strong> capital for OP) in the first stage of<br />
the estimation algorithm can cause the labor coefficient to be unidentified.<br />
This collinearity arises from the fact that labor, like materials <strong>and</strong> capital,<br />
needs to be allocated in some way by the firm, at some point in time. While<br />
this problem can arise in the context of the OP <strong>and</strong> LP estimator, it is<br />
particularly problematic for the LP estimator.<br />
For the LP estimator, since labor <strong>and</strong> materials are both chosen simultaneously,<br />
a natural assumption could be that they are allocated in similar<br />
ways. However, this would imply that labor <strong>and</strong> materials are both chosen<br />
as a function of productivity <strong>and</strong> capital:<br />
mit = ft (ωit, kit)<br />
lit = gt (ωit, kit)<br />
Hence, both labor <strong>and</strong> materials depend on the same state variables. Using<br />
the invertibility condition of LP, i.e. ωit = f −1<br />
t (mit, kit), this leads to the
Total factor productivity estimation 29<br />
following result (Ackerberg et al., 2006):<br />
−1<br />
lit = gt ft (mit, kit) , kit = ht (mit, kit)<br />
Since it is not possible to simultaneously estimate a non-parametric function<br />
of ωit <strong>and</strong> kit together with the coefficient on the labor variable, which<br />
is also a function of those same variables; the labor coefficient will not be<br />
identified in the first stage. Hence collinearity between the labor variable <strong>and</strong><br />
the non-parametric function in the first stage can cause the labor coefficient<br />
to be unidentified. Ackerberg, Caves <strong>and</strong> Frazer further investigate to what<br />
extent plausible assumptions can be made about the data generating process<br />
for labor in order to “save” the LP first stage estimation, with little success.<br />
As noted above, this collinearity problem can also arise in the context of the<br />
OP estimation procedure. However, for the OP estimator, identification of<br />
the labor coefficient can be achieved by assuming that labor is not a perfectly<br />
variable input <strong>and</strong> that firms decide on the allocation of labor without perfect<br />
information about their future productivity (i.e. investment <strong>and</strong> labor are<br />
determined by different information sets). If this assumption holds for the<br />
data at h<strong>and</strong>, the labor coefficient can be identified in the first stage of<br />
the estimation algorithm in the case of OP. For LP, this assumption does<br />
not solve the collinearity problem, since choosing labor prior to choosing<br />
material inputs will make the choice of the latter directly dependent on the<br />
choice of labor inputs, again preventing identification of the labor coefficient<br />
in the first stage. This difference between the two estimators stems from<br />
the fact that investment, unlike materials, is not directly linked to period<br />
t outcomes, so that a firm’s allocation of labor will not directly affect its<br />
investment decisions (Ackerberg et al., 2006).<br />
Ackerberg, Caves <strong>and</strong> Frazer suggest an alternative estimation procedure,<br />
where the coefficient on labor (in a value added production function) is no<br />
longer estimated in the first stage of the algorithm. All input coefficients are<br />
obtained in the second stage, while the first stage only serves to net out the<br />
error component in the production function.
30 Total factor productivity estimation<br />
Moreover, in the presence of imperfect competition in input or output<br />
markets, consistency of either the OP or LP estimator is likely to break down,<br />
as an omitted price variable will bias results. Therefore, the OP algorithm has<br />
been augmented to take imperfect competition in output markets explicitly<br />
into account (De Loecker, 2007, see below). For LP however, De Loecker<br />
(2007, Appendix C) shows that imperfect competition in output markets is<br />
likely to invalidate the invertibility condition, while it has no effect on the<br />
monotonicity condition of OP. Therefore, even if the LP estimation algorithm<br />
is augmented with the correction for imperfect competition (discussed below),<br />
coefficients are likely to be biased. Hence, I will focus on the OP algorithm<br />
in what follows.<br />
1.3.6 Extensions of the Olley-Pakes methodology<br />
Many of the extensions <strong>and</strong> alternatives that emerge from the literature are<br />
still work in progress, making it particularly hard to choose among the many<br />
c<strong>and</strong>idates. For a recent technical review of a number of extensions to the OP<br />
methodology, I refer to ABBP 11 (2007). Alternatives to the semiparametric<br />
estimators of OP <strong>and</strong> LP are proposed by (among others) Katayama et al.<br />
(2005). However, a full discussion of these works lies beyond the scope of the<br />
present paper.<br />
As was noted in section 1.2, De Loecker (2007) implements the correction<br />
for the omitted output price bias, introduced by Klette <strong>and</strong> Griliches<br />
(1996) in the OP estimation algorithm. In what follows, the specifics of his<br />
model will be discussed. While De Loecker (2007) also introduces a correction<br />
for multi-product firms, I have elected not to discuss this extension here<br />
for two reasons. First, in the absence of product-level data on inputs <strong>and</strong><br />
outputs, consistent estimation of TFP can only be obtained by either focusing<br />
on single-product firms or by allowing the parameters of the production<br />
technology to vary across firms making different products (BRS, 2005). Although<br />
De Loecker (2007) is able to exploit information on which products a<br />
11 ABBP focus on the assumptions underlying the semiparametric estimators introduced<br />
by OP <strong>and</strong> LP <strong>and</strong> show how to test their validity <strong>and</strong> how to relax some of them; they do<br />
not treat the bias introduced by endogenous product choice or by imperfect competition<br />
in input <strong>and</strong> output markets explicitly.
Total factor productivity estimation 31<br />
firm produces, allowing him to introduce product level dem<strong>and</strong> rather than<br />
industry level dem<strong>and</strong> as well as to control for the number of products a<br />
firm produces; the production technology is still (necessarily) assumed to be<br />
identical across products in an industry.<br />
Moreover, BRS (2006b) find that more than 60 percent of US firms alter<br />
their product mix every five years. This implies that any information on the<br />
product space firms are active in, would have to be dynamic in nature 12 . Since<br />
typical annual accounts data usually provide no or very limited information<br />
at the relevant product level <strong>and</strong> given the remaining biases in the resulting<br />
production function coefficients in the absence of (dynamic) product-level<br />
data on inputs <strong>and</strong> outputs, I will restrict attention to single-product firms.<br />
The relevant model to start from in the presence of imperfect competition<br />
in the output market is given by (1.4). In order to estimate (1.4) consistently<br />
without information on establishment-level prices, it is necessary to impose<br />
some structure on the dem<strong>and</strong> system, which will be used to implicitly solve<br />
for the firm-level prices. Following De Loecker (2007), I start out from a<br />
simple conditional (Dixit-Stiglitz) dem<strong>and</strong> system 13 :<br />
Qit = QJt<br />
η Pit<br />
PJt<br />
exp u d <br />
it<br />
where Qit represents dem<strong>and</strong> for the firm’s product, QJt is industry output<br />
at time t, Pit<br />
PJt<br />
is the price of firm i relative to the average price in industry<br />
J, u d it is an idiosyncratic firm-specific dem<strong>and</strong> shock <strong>and</strong> η is the elasticity of<br />
substitution (dem<strong>and</strong>) between differentiated goods in the industry (−∞ <<br />
η < −1).<br />
Taking natural logarithms results in the following expression for the<br />
dem<strong>and</strong> system.<br />
12 Although De Loecker has very detailed information on which firms are active in which<br />
sectors, the data are only available for 2001. Hence the firm-level product mix is necessarily<br />
assumed to be constant over the sample period in his analysis.<br />
13 The industry is assumed to be characterized by product differentiation. A key characteristic<br />
of Dixit-Stiglitz dem<strong>and</strong> is that the price (substitution) elasticities are constant<br />
over time <strong>and</strong> independent of the number of varieties.
32 Total factor productivity estimation<br />
qit = qJt + ηpit − ηpJt + u d it<br />
(1.12)<br />
It is possible to derive an expression for pit from (1.12) <strong>and</strong> substitute the<br />
result into (1.4).<br />
pit = 1 <br />
qit − qJt − u<br />
η<br />
d <br />
it + pJt<br />
rit = pit + yit − pit = 1 <br />
qit − qJt − u<br />
η<br />
d <br />
it + pJt + yit − pit Using the fact that changes in the industry-wide price index p it can be<br />
considered as a weighted average of the changes in firm-specific prices, i.e.<br />
p it = pJt , results in the following relationship:<br />
rit =<br />
η + 1<br />
η<br />
(β0 + βkkit + βllit + βmmit + ωit + u q 1<br />
it ) −<br />
η qJt − 1<br />
η udit (1.13)<br />
where ωit will be proxied by the investment decision as in section 1.3.3.<br />
Hence, it is clear from (1.13) that consistent estimation in the presence of<br />
imperfectly competitive output markets requires adding a term to the production<br />
function. By putting structure on the dem<strong>and</strong> system, it is possible<br />
to proxy for unobserved firm-level prices by adding industry output as an<br />
additional regressor in the production function 14 . Specifically, the final estimating<br />
equation looks as follows:<br />
rit = α0 + αkkit + αllit + αmmit + ω ′<br />
it + u′ q<br />
it + αηqJt − u ′ d<br />
it<br />
(1.14)<br />
where αh = ((η + 1) /η) βh for h = 0, l, m, k; ω ′<br />
it = ((η + 1) /η) ωit <strong>and</strong><br />
αη = (−1/η). The final production function coefficients can be obtained by<br />
multiplying the coefficients obtained in (1.14) with the relevant mark-up, i.e.<br />
η/ (η + 1). Similarly, firm-level productivity is now obtained as follows:<br />
14 Ornaghi (2006) invalidates the correction suggested by Klette <strong>and</strong> Griliches by confirming<br />
the existence of asymmetric biases among the input coefficients introduced by the<br />
use of deflated values of inputs <strong>and</strong> outputs rather than observed quantities. Given this<br />
asymmetric bias, multiplying all input coefficients with an identical upward correction<br />
term (i.e. the mark-up) as illustrated in (1.15) can not yield unbiased input coefficients.
Total factor productivity estimation 33<br />
ˆωit =<br />
<br />
ˆη<br />
ˆω<br />
ˆη + 1<br />
′<br />
<br />
ˆη<br />
it = (˜rit − ˆαkkit − ˆαllit − ˆαmmit − ˆαηqJt) (1.15)<br />
ˆη + 1<br />
Hence, for the OP estimator including the correction for market power,<br />
productivity as obtained in (1.3) additionally needs to be multiplied by the<br />
relevant mark-up; as shown in (1.15). Although this correction simply implies<br />
a rescaling of firm-level productivity in this particular case, it is straightforward<br />
to interact industry output at a more disaggregated level with sector<br />
dummies at an equal level of aggregation to allow the dem<strong>and</strong> elasticity <strong>and</strong><br />
relevant mark-up to vary across sub-sectors 15 . Allowing the dem<strong>and</strong> elasticity<br />
to vary across sub-sectors in (1.14) leads to the following estimating<br />
equation (De Loecker, 2007):<br />
rit = α0 + αllit + αkkit + αmmit + ω ′<br />
it + u′ q<br />
it +<br />
M<br />
s=1<br />
αηsqJtsIis − u ′ d<br />
it<br />
(1.16)<br />
where s represents the sub-sector <strong>and</strong> M equals the total number of<br />
sub-sectors. Iis is a dummy variable equal to 1 if a firm is active in a given<br />
sub-sector <strong>and</strong> qJts is the relevant industry dem<strong>and</strong> shifter, proxied by output<br />
in the different sub-sectors. The number of estimated elasticities ηs equals the<br />
number of sub-sectors in the industry. Industry output is simply replaced in<br />
the estimation by M<br />
αηsqJtsIis. It should be noted that if dem<strong>and</strong> parameters<br />
i=1<br />
are allowed to vary across sub-sectors; the resulting production coefficients<br />
βh will also be specific to those sub-sectors, since the estimates obtained<br />
from estimating (1.16) have to be transformed using the relevant (sub-sector)<br />
mark-up.<br />
1.3.7 Summary of estimation algorithms<br />
Table 1.2 summarizes the different estimation algorithms discussed in this<br />
section. While fixed effects <strong>and</strong> instrumental variables methods are theoretically<br />
able to solve the simultaneity bias introduced when estimating (1.2)<br />
15 De Loecker additionally includes product dummies in the first stage of the estimation<br />
algorithm to control for unobserved product quality differences.
34 Total factor productivity estimation<br />
using OLS; their application has not been entirely successful. Likely causes<br />
for the failure of both techniques to produce sensible <strong>and</strong> unbiased results<br />
are the lack of time-invariance of ωit in the case of fixed effects <strong>and</strong> the lack<br />
of good instruments in the case of IV estimation. Blundell <strong>and</strong> Bond (1999)<br />
have developed an extended GMM estimator, taking some of these issues<br />
into account.<br />
Both semiparametric estimators (OP <strong>and</strong> LP) are able to resolve simultaneity<br />
issues by using a proxy variable to substitute for unobserved productivity;<br />
assuming a strict monotonicity condition holds <strong>and</strong> ωit is the only<br />
unobserved firm-level variable (i.e. the scalar unobservable). While it is<br />
possible to take selection issues into account by using an unbalanced panel<br />
for both estimators, only the OP estimation algorithm explicitly takes the<br />
survival probability at the firm level into account in the second stage of the<br />
estimation algorithm. Extensions have been developed, mainly in the context<br />
of the OP procedure, to take imperfect competition in output markets,<br />
as well as multi-product firms into account (De Loecker, 2007).<br />
1.4 Empirical application: Food <strong>and</strong> bever-<br />
ages industry in Belgium<br />
In what follows, I will illustrate the different methodologies introduced<br />
in the previous section, using firm-level data on the Belgian food <strong>and</strong> beverages<br />
industry. The data set is constructed on the basis of the Belfirst<br />
database, which groups annual accounts data on the entire population of<br />
limited-liability firms located in Belgium. The database is commercialized<br />
by BvDEP (2006). <strong>Firm</strong>s are uniquely defined by their VAT number <strong>and</strong><br />
data on employment, net value added, total fixed assets etc. are available<br />
for the years 1996-2005. <strong>Firm</strong>s are classified into sectors according to the<br />
NACE-Bel nomenclature, i.e. a five-digit extension of the NACE (Revision<br />
1) classification commonly used for European statistics 16 . Producer price<br />
indices used to deflate firm-level output are available from Eurostat (2007)<br />
16 The NACE Rev. 1 classification can be downloaded from the Eurostat Ramon server:<br />
http://europa.eu.int/comm/eurostat/ramon/.
Empirical application: Food <strong>and</strong> beverages industry in Belgium 35<br />
Estimation algorithm Assumptions Resolved issues References<br />
Fixed effects ωit is plant-specific, but time- Simultaneity Mundlak (1961)<br />
invariant. Selection if ωit = ωi, ∀i Hoch (1962)<br />
ABBP (2007)<br />
Instrumental variables Correlation between instruments <strong>and</strong> Simultaneity Blundell <strong>and</strong> Bond (1999)<br />
& GMM endogenous regressors. Selection (unbalanced panel) ABBP (2007)<br />
No correlation between instruments<br />
<strong>and</strong> error term.<br />
Semiparametric estimator: Invertibility condition: investment Simultaneity Olley <strong>and</strong> Pakes (1996)<br />
Olley & Pakes has to be strictly increasing in ωit. Selection (unbalanced panel) ABBP (2007)<br />
Scalar unobservable assumption: Selection (survival probability) Ackerberg et al. (2006)<br />
ωit is only unobserved state variable.<br />
Semiparametric estimator: Invertibility condition: Simultaneity Levinsohn <strong>and</strong> Petrin (2003)<br />
Levinsohn & Petrin mit has to be strictly increasing in ωit. Selection (unbalanced panel) Petrin et al. (2003)<br />
Scalar unobservable assumption: Ackerberg et al. (2006)<br />
ωit is only unobserved state variable.<br />
OP with imperfect Assumptions OP. Simultaneity Klette <strong>and</strong> Griliches (1996)<br />
competition in output Selection (unbalanced panel) Levinsohn <strong>and</strong> Melitz (2002)<br />
markets Selection (survival probability) De Loecker (2007)<br />
Omitted output price bias<br />
Extended OP Assumptions OP. Simultaneity Klette <strong>and</strong> Griliches (1996)<br />
including correction Common production technology Selection (unbalanced panel) Levinsohn <strong>and</strong> Melitz (2002)<br />
for multi-product firms for all products of a firm. Selection (survival probability) De Loecker (2007)<br />
Dem<strong>and</strong> elasticity is common Omitted output price bias<br />
across products <strong>and</strong> constant. Endogenous product choice<br />
Table 1.2: TFP estimation: Summary of estimation algorithms
36 Total factor productivity estimation<br />
at the three-digit Nace level. Deflators for material inputs <strong>and</strong> investment<br />
were obtained from Belgostat (2007).<br />
Following Mata <strong>and</strong> Portugal (1994); Mata et al. (1995) <strong>and</strong> Van Beveren<br />
(2007b); entry <strong>and</strong> exit in the sample are defined as economic exit <strong>and</strong> entry<br />
17 , implying that exit occurs if a firm’s employment drops to zero in a<br />
particular year <strong>and</strong> entry takes place if there was no previous employment<br />
recorded. <strong>Firm</strong>s exhibiting irregular exit or entry patterns are omitted from<br />
the sample. Similarly, in order to verify that no re-entry occurs after a firm<br />
exits, the last two years in the sample are dropped.<br />
There are several reasons why the evolution of TFP in the food <strong>and</strong> beverages<br />
sector in Belgium is of interest. First, the sector represents a significant<br />
share of industrial employment in Belgium, accounting for 14.2 percent of the<br />
total (CRB, 2004), second only to the metals industry (16 percent). Moreover,<br />
the outbreak of the dioxin crisis in 1999, when excessive concentrations<br />
of dioxin were found in eggs, chickens, milk <strong>and</strong> pork; resulting from contaminated<br />
animal food (The Economist, 1999); led to a period of significant<br />
restructuring <strong>and</strong> increasing investments in the sector; reflected in the sample<br />
by high entry <strong>and</strong> exit rates (see below). Given these preliminaries, it<br />
can be expected that some of these events will be reflected in the industry’s<br />
TFP performance.<br />
Using the Belfirst database, I was able to collect information on all firms<br />
active in the food <strong>and</strong> beverages sector (NACE 15). <strong>Firm</strong>s with no recorded<br />
data on one of the variables used in the empirical analysis are omitted 18 , as<br />
well as firms producing multiple products. To identify multi-product firms,<br />
I rely on the number of five-digit NACE-Bel codes a firm lists, i.e. the most<br />
17 Although the Belfirst database reports firms’ legal status <strong>and</strong> hence also legal exit;<br />
I do not rely on this measure for two reasons. First, inspection of the data reveals that<br />
the official date associated with the legal status in the database often does not concur<br />
with the actual time the firm exits the market. Second, communications with Bureau Van<br />
Dijk made clear that although the legal status is correctly reported whenever available,<br />
many companies fail to report their annual accounts after ending their activities. For the<br />
specifics associated with the exit <strong>and</strong> entry variables, I refer to Van Beveren (2007a).<br />
18 Belgian accounting rules only require firms to report full annual accounts (including<br />
data on turnover) once a certain threshold in terms of employment, total assets or turnover<br />
is reached. Therefore, the sample necessarily excludes smaller firms.
Empirical application: Food <strong>and</strong> beverages industry in Belgium 37<br />
detailed level available in the database. If a firm is active in more than one<br />
five-digit sector, it is omitted from the analysis. Finally, the data are checked<br />
for outliers <strong>and</strong> gaps. <strong>Firm</strong>s exhibiting variable input growth of more than<br />
200 percent (employment <strong>and</strong> materials inputs) in one year or output growth<br />
of more than 500 percent are excluded from the sample.<br />
This results in a final sample of 1,025 firms (5,551 observations). Table 1.3<br />
reports summary statistics for the sample for the period 1996-2003. From the<br />
table it is clear that the average firm in the sample is relatively large (average<br />
employment amounts to 54.61 employees). By comparison, in the full sample<br />
of firms active in sector 15, the average firm employs about 30 people. As<br />
noted above, the period considered here involved significant restructuring in<br />
the sector, translated in high entry <strong>and</strong> exit rates. Specifically, 184 firms<br />
(18 percent) enter the sample between 1996 <strong>and</strong> 2003; while 131 firms (13<br />
percent) exit over the same period.<br />
Table 1.4 reports the production function coefficients obtained using the<br />
different methodologies introduced in section 1.3. All reported estimates<br />
are obtained for the unbalanced panel of firms (allowing for implicit entry<br />
<strong>and</strong> exit); apart from the fixed effects estimator, where I report both the<br />
unbalanced <strong>and</strong> balanced sample result. The first column in the table reports<br />
the number of observations associated with each specific estimator <strong>and</strong> clearly<br />
shows one of the main advantages of the LP estimator compared to OP. Since<br />
material inputs are used to proxy for unobservable productivity; I am able<br />
to retain the full sample of firms in the first estimation stage; while for OP,<br />
only those observations with positive investment can be retained in the first<br />
stage. In the second stage, one year of observations is lost due to the dynamic<br />
nature of the model, both for OP <strong>and</strong> LP.<br />
All estimations reported in table 1.4 are performed in Stata 10. For the<br />
OLS <strong>and</strong> fixed effects estimators, built-in comm<strong>and</strong>s reg <strong>and</strong> xtreg are used.<br />
The GMM estimator is obtained using the xtabond2 comm<strong>and</strong>, due to Roodman<br />
(2006). No built-in or user-developed comm<strong>and</strong> exists to date to implement<br />
the OP estimator 19 ; but Arnold (2005) provides some practical tips,<br />
19 A user-developed comm<strong>and</strong>, opreg, has recently been made available in Stata, due to<br />
Yasar, Raciborski <strong>and</strong> Poi (2008). I have not relied on this comm<strong>and</strong> for the empirical
38 Total factor productivity estimation<br />
Table 1.3: Summary statistics of key variables<br />
Real values are obtained by deflating monetary values using three-digit producer price indices obtained<br />
from Eurostat. Output is defined as turnover of the firm. Employment is measured as the number of<br />
employees (full-time equivalents). The materials variable includes raw materials, consumables, services<br />
<strong>and</strong> other goods. Capital is defined as total fixed tangible assets. Investment is calculated on the basis<br />
of firm-level capital, using a st<strong>and</strong>ard depreciation rate of 15 percent. Data pertain to the Food <strong>and</strong><br />
Beverages sector (NACE 15) in Belgium, for the years 1996 to 2003.<br />
Real output ( Rjt,ex 1,000) 5,551 19,454.79 61,007.18 0.97 950,812.10<br />
Employment (Ljt) 5,551 54.61 181.91 1 3,443.00<br />
Real materials ( Mjt,ex 1,000) 5,551 16,600.85 49,454.16 1.14 807,434.90<br />
Real capital ( Kjt,ex 1,000) 5,551 3,036.16 15,605.24 0.99 447,185.80<br />
Real (pos.) investment ( Ijt,ex 1,000) 3,588 662.46 2,653.91 0.01 61,377.32<br />
St<strong>and</strong>ard<br />
Variable N Mean Deviation Minimum Maximum
Empirical application: Food <strong>and</strong> beverages industry in Belgium 39<br />
Labor Materials Capital<br />
Method N βl SE βm SE β SE<br />
OLS 5,551 0.2113*** [0.0152] 0.7700*** [0.0138] 0.0266*** [0.0072]<br />
Fixed Effects (balanced) 3,568 0.1696*** [0.0192] 0.6474*** [0.0419] 0.0277*** [0.0063]<br />
Fixed Effects (unbalanced) 5,551 0.1685*** [0.0166] 0.6814*** [0.0379] 0.0248*** [0.0052]<br />
GMM 5,551 0.1520*** [0.0368] 0.7890*** [0.0434] 0.0372** [0.0173]<br />
OP (no survival correction) 3,588 0.1925*** [0.0153] 0.7722*** [0.0150] 0.0445** [0.0195]<br />
OP (survival correction) 3,588 0.1925*** [0.0153] 0.7722*** [0.0150] 0.0453*** [0.0167]<br />
Levinsohn-Petrin 5,551 0.2139*** [0.0148] 0.7915*** [0.0802] 0.0484** [0.0205]<br />
De Loecker (1) 3,588 αl = 0.1947*** [0.0153] αm = 0.7686*** [0.0151] αk = 0.0461* [0.0240]<br />
Transformed coefficients DL αq = 0.2926*** 0.2707*** [0.0223] 1.0837*** [0.0426] 0.0654** [0.0338]<br />
[0.0199]<br />
Values are coefficients, st<strong>and</strong>ard errors reported between brackets. (1) The coefficients for the DL estimator are obtained by<br />
multiplying the alpha’s with the relevant mark-up. The elasticity of substitution η equals (−1/αq) or -3.42. The relevant mark-up<br />
therefore equals η/ (η + 1) = 1.41.<br />
Table 1.4: Production function estimates
40 Total factor productivity estimation<br />
particularly on the implementation of the nonlinear second stage. The LP<br />
estimator was implemented using the levpet comm<strong>and</strong>, due to Petrin et al.<br />
(2003).<br />
In order to interpret the estimated coefficients, it is useful to briefly go back<br />
to table 1.1. In the third column of this table, the general direction of the<br />
biases introduced by the different endogeneity issues are given. Theoretically,<br />
the fixed effects estimator corrects for both the simultaneity <strong>and</strong> selection<br />
bias, hence the coefficients on the variable inputs (labor <strong>and</strong> materials) are<br />
expected to be lower compared to the OLS result; while the coefficient on<br />
capital is expected to be higher. While the coefficients on the variable inputs<br />
in table 1.4 are in line with expectations (βl <strong>and</strong> βm are lower compared to<br />
the first row); the capital coefficient is still very low, both for the balanced<br />
<strong>and</strong> unbalanced sample.<br />
Moreover, as was discussed in section 1.3, comparing the results of the<br />
balanced <strong>and</strong> unbalanced sample for the FE estimator enables us to determine<br />
whether the FE estimator adequately corrects for the selection bias;<br />
i.e. whether exit decisions at the firm level are only determined by the timeinvariant,<br />
firm-specific effects ωi. Given the small differences between the<br />
coefficients obtained for the balanced <strong>and</strong> unbalanced sample; results in table<br />
1.4 suggest that the FE estimator is able to correct for the selection bias<br />
in the sample.<br />
Since the GMM estimator is theoretically able to correct for the simultaneity<br />
bias, βl <strong>and</strong> βm in row 4 of table 1.4 are expected to be lower, while<br />
βk should increase compared to their OLS counterparts; similarly to the FE<br />
estimator. Results in row 4 show a lower labor coefficient <strong>and</strong> higher capital<br />
coefficient (in line with expectations); but lower coefficient on materials (not<br />
in line with expectations).<br />
The last four rows in table 1.4 display the production function coefficients<br />
for the semiparametric estimators of OP (both with <strong>and</strong> without explicit correction<br />
for firms’ survival probability), LP <strong>and</strong> De Loecker. Comparing OP<br />
estimations.
Empirical application: Food <strong>and</strong> beverages industry in Belgium 41<br />
estimates to the OLS estimates in the first row, shows that the coefficients<br />
on both labor <strong>and</strong> materials are lower compared to OLS results, while the<br />
capital coefficient is significantly higher; which is in line with expectations.<br />
Including the estimated survival probability in the second stage of the estimation<br />
algorithm has virtually no impact on the capital coefficient. This<br />
result is in line with the findings of OP, who similarly found no significant<br />
improvement in the capital coefficient from the explicit correction for survival<br />
when an unbalanced panel is used. Although the LP coefficient on capital<br />
is higher than its OLS counterpart, the labor <strong>and</strong> materials coefficients are<br />
somewhat higher than the OLS estimates.<br />
The final row of table 1.4 summarizes the results of estimating (1.14) using<br />
the estimation algorithm introduced 20 by De Loecker (2007). Essentially, this<br />
amounts to the inclusion of industry output in the first stage of estimation<br />
<strong>and</strong> subtracting the resulting coefficient times output from the left-h<strong>and</strong>-side<br />
in (1.10). Industry output is calculated at the three-digit level in each year<br />
as the share-weighted average of firm-level outputs, where shares are based<br />
on deflated revenues. This comes from the observation that the industry<br />
price index (which is available at the three-digit level) represents a shareweighted<br />
average of firm-level prices, where weights are output shares (De<br />
Loecker, 2007). For now, the elasticity of dem<strong>and</strong> (substitution) is assumed<br />
to be identical across the different subsectors within the food <strong>and</strong> beverages<br />
industry.<br />
As was shown in section 3, the coefficient on industry output αq relates<br />
to the elasticity of dem<strong>and</strong> in the following way: αq = (−1/η). Moreover,<br />
using the dem<strong>and</strong> elasticity, which amounts to -3.42; it is possible to calculate<br />
the relevant mark-up at the industry level η/ (η + 1), equal to 1.41.<br />
This estimate is somewhat higher than the result found by Konings (2001),<br />
who find a mark-up of 1.30 for the food <strong>and</strong> beverages industry in Belgium<br />
in the period 1992-1996. The last row in table 1.4 further reports both the<br />
20 Although the correction for market power in output markets was originally suggested<br />
by Klette <strong>and</strong> Griliches (1996), De Loecker was the first to implement this correction<br />
into the semiparametric estimation framework introduced by Olley <strong>and</strong> Pakes (1996).<br />
Abraham, Konings <strong>and</strong> Slootmaekers (2007a) report results of the DL estimator as a<br />
robustness check in their paper on FDI spillovers in China.
42 Total factor productivity estimation<br />
estimated coefficients <strong>and</strong> the true production coefficients βh = (η/η + 1) αh.<br />
Consistent with the theoretically predicted biases in table 1.1, the coefficients<br />
on labor <strong>and</strong> materials are significantly higher compared to the OP coefficients<br />
without including industry output. However, the coefficient on capital<br />
is somewhat higher compared to the basic OP results, which is not in line<br />
with expectations.<br />
As was indicated in section 1.3, it is straightforward to allow the dem<strong>and</strong><br />
elasticity to vary over the different three-digit industries by interacting industry<br />
output with the respective industry dummies in (1.16). Since this results<br />
both in different dem<strong>and</strong> elasticities <strong>and</strong> associated mark-ups; production<br />
function coefficients also become specific for each separate three-digit industry<br />
in this case. However, note that while production coefficients become<br />
variety-specific in this case, the production technology is still assumed to be<br />
constant for all three-digit industries within the food <strong>and</strong> beverages sector.<br />
Table 1.5 reports the results of estimating (1.16) for the sample of singleproduct<br />
firms in the food <strong>and</strong> beverages industry. The first row in table 1.5<br />
shows the estimated coefficients αh. Compared to the estimated coefficient<br />
for the constant-elasticity estimator reported in the last row of table 1.4, the<br />
labor <strong>and</strong> materials coefficients are very similar, while the capital coefficient<br />
is somewhat higher. Turning to the industry-specific output coefficients, it<br />
is clear that large variation exists between the different three-digit subsectors<br />
of the food <strong>and</strong> beverages industry. Calculated dem<strong>and</strong> elasticities vary<br />
between -2.8 <strong>and</strong> -3.6; associated mark-ups range between 1.39 <strong>and</strong> 1.56.<br />
These differences point to the importance of allowing the dem<strong>and</strong> (substitution)<br />
elasticity to vary across different sub-sectors of a particular industry. As<br />
a consequence, variety-specific production coefficients also vary considerably<br />
across the different three-digit industries.<br />
Two caveats should be noted here. First, I have continued to assume<br />
throughout that input prices for materials (capital) at the firm level are<br />
adequately captured by the materials (investment) deflator. To the extent<br />
that input price differences are translated into output price deviations, which<br />
are taken into account using industry output, this should partly take care of
Empirical application: Food <strong>and</strong> beverages industry in Belgium 43<br />
Three-digit industry Output Dem<strong>and</strong> Labor Materials Capital<br />
NACE Description Coefficient Elasticity Mark-up Coefficient Coefficient Coefficient<br />
- αh(h = l, m, k) - - - 0.1948*** 0.7685*** 0.0569***<br />
[0.0153] [0.0151] [0.0215]<br />
151 Meat (products) 0.3348*** -2.9869 1.5033 0.2896*** 1.1592*** 0.0863***<br />
[0.0224] [0.0251] [0.0517] [0.0335]<br />
152 Fish(products) 0.3552*** -2.8154 1.5508 0.2981*** 1.1931*** 0.0888***<br />
[0.0239] [0.0265] [0.0577] [0.0347]<br />
153 Fruit <strong>and</strong> vegetables 0.3145*** -3.1799 1.4587 0.2802*** 1.1215*** 0.0835***<br />
[0.0218] [0.0237] [0.0477] [0.0323]<br />
154 Oils <strong>and</strong> fats 0.3587*** -2.7881 1.5593 0.2999*** 1.2003*** 0.0894***<br />
[0.0243] [0.0270] [0.0588] [0.0347]<br />
155 Dairy products 0.2951*** -3.3888 1.4186 0.2728*** 1.0919*** 0.0813***<br />
[0.0197] [0.0227] [0.0416] [0.0314]<br />
156 Grain mill products 0.3026*** -3.3046 1.4339 0.2780*** 1.1126*** 0.0828***<br />
[0.0224] [0.0235] [0.0446] [0.0320]<br />
157 Prepared animal feeds 0.3103*** -3.2226 1.4499 0.2786*** 1.1151*** 0.0830***<br />
[0.0206] [0.0238] [0.0438] [0.0321]<br />
158 Other food products 0.2871*** -3.4831 1.4027 0.2692*** 1.0774*** 0.0802***<br />
[0.0194] [0.0220] [0.0407] [0.0309]<br />
159 Beverages 0.2784*** -3.592 1.3858 0.2656*** 1.0631*** 0.0791***<br />
[0.0184] [0.0216] [0.0377] [0.0304]<br />
Values are coefficients, st<strong>and</strong>ard errors reported between brackets. The variety-specific production function<br />
coefficients are obtained by multiplying the alpha’s (given in the first row) with the relevant mark-up. The<br />
elasticity of substitution (dem<strong>and</strong>) η is obtained as the inverse <strong>and</strong> negative of the output coefficient. The<br />
relevant mark-up equals η/(η + 1).<br />
Table 1.5: Production function estimates: Variety-specific dem<strong>and</strong>
44 Total factor productivity estimation<br />
the omitted input price bias (De Loecker, 2007). However, as was already<br />
noted in section 1.2, a formal solution to this bias (in the absence of firm-level<br />
data on input prices) has yet to be introduced.<br />
Second, the selection of single-product firms in the sample is obtained by<br />
resorting to the NACE-Bel codes reported by firms in their annual accounts,<br />
where the codes typically relate to the latest year available. Hence, the<br />
selection of firms is made in a particular year, whereas it is quite possible<br />
that some of these firms produced multiple products in any of the previous<br />
years.<br />
The production function coefficients obtained in tables 1.4 <strong>and</strong> 1.5 can be<br />
used to calculate firm-level productivity for each of the sample years. By<br />
imposing coefficient stability on the model, it is possible to retain the full<br />
sample of firms for all estimators, even in the absence of positive investment<br />
(as for the OP estimators). <strong>Firm</strong>-level productivity for the OLS, fixed effects,<br />
GMM, OP (with <strong>and</strong> without survival correction) <strong>and</strong> LP estimators<br />
is obtained on the basis of (1.3). For the OP estimator including the correction<br />
for market power (De Loecker, with or without variety-specific dem<strong>and</strong>),<br />
productivity as obtained in (1.3) additionally needs to be multiplied by the<br />
relevant mark-up; as was shown in (1.15).<br />
Finally, using the estimates of firm-level productivity obtained from applying<br />
(1.3) <strong>and</strong> (1.15) to the sample using the production function coefficients<br />
from tables 1.4 <strong>and</strong> 1.5, it is possible to calculate aggregate industry productivity<br />
for each year in the sample as a weighted average of firm-level TFP:<br />
Nt <br />
ˆPJt = sit ˆ Ωit<br />
i=1<br />
(1.17)<br />
where sit is a firm-specific weight, equal to (Sit/ ( <br />
i Sit)) <strong>and</strong> S represents<br />
either turnover or employment (De Loecker <strong>and</strong> Konings, 2006). Normalizing<br />
this index to 1 in 1996 allows us to compare the evolution of aggregate TFP<br />
in the food <strong>and</strong> beverages industry for the different estimators discussed here.
Empirical application: Food <strong>and</strong> beverages industry in Belgium 45<br />
TFP (1996 = 1.00)<br />
.95 1 1.05 1.1 1.15 1.2<br />
1996 1997 1998 1999 2000 2001 2002 2003<br />
year<br />
ols fe_b<br />
fe gmm<br />
lp opbasic<br />
opsurv dl<br />
dl_var<br />
(Weights are turnover shares, 1996 = 1)<br />
Figure 1.1: Weighted productivity index: Comparison estimation methods<br />
Figure 1.1 shows the evolution of industry productivity between 1996 <strong>and</strong><br />
2003, using turnover shares as weights. From the figure, it is clear that TFP<br />
in the food <strong>and</strong> beverages industry exhibits a clear upward trend in the period<br />
following the dioxin crisis of 1999. However, whereas TFP continues to increase<br />
until 2002 when imperfect competition in output markets is not taken<br />
into account; TFP estimated using the DL methodology increases sharply<br />
between 1999 <strong>and</strong> 2000 <strong>and</strong> exhibits a more of less stable pattern after that.<br />
For the DL estimator with variety-specific dem<strong>and</strong>, this pattern is even more<br />
apparent. Moreover, compared to the other estimators shown in figure 1.1,<br />
TFP calculated using the coefficients of table 1.5 declines more sharply prior<br />
to 1999 <strong>and</strong> grows less strongly after 1999. These results suggest that imperfect<br />
competition in output markets, when not taken into account in the<br />
production function estimation, may yield misleading results concerning the<br />
timing <strong>and</strong> magnitude of productivity shocks. The different growth pattern<br />
observed for the DL estimator with <strong>and</strong> without variety-specific dem<strong>and</strong> further<br />
suggests that it is important to take the dem<strong>and</strong> structure into account<br />
at the appropriate level of aggregation 21 .<br />
21 Ideally, this would be at the product level. However, this would require not only information<br />
on aggregate product output, but also on product-level price evolutions (indices).
46 Total factor productivity estimation<br />
To assess whether the evolution of aggregate TFP in the food <strong>and</strong> beverages<br />
industry is due to firm-level improvements in TFP or rather to the<br />
reallocation of market shares between firms, various decompositions can be<br />
used (De Loecker <strong>and</strong> Konings, 2006). I will rely on the decomposition 22 introduced<br />
by Olley <strong>and</strong> Pakes (1996), who decompose aggregate productivity<br />
into a within component <strong>and</strong> a covariance term in the following way:<br />
Nt <br />
ˆPJt =<br />
ˆPJt =<br />
i=1<br />
Nt <br />
ˆPJt = sit ˆ Ωit<br />
i=1<br />
<br />
¯ˆΩt<br />
(¯st + ∆sit) + ∆ˆ <br />
Ωit<br />
<br />
Nt¯st ¯ <br />
Ωt<br />
ˆ +<br />
ˆPJt = ¯ ˆ<br />
Pit +<br />
Nt <br />
i=1<br />
Nt <br />
i=1<br />
<br />
∆sit∆ˆ <br />
Ωit<br />
<br />
∆sit∆ˆ <br />
Ωit<br />
where ¯ Pit<br />
ˆ is the unweighted average of plant-level total factor productivity<br />
<strong>and</strong> Nt <br />
∆sit∆ˆ <br />
Ωit refers to the sample covariance between TFP <strong>and</strong> output<br />
i=1<br />
(or employment) shares. The results of applying this decomposition using either<br />
turnover (left-h<strong>and</strong> side) or employment shares (right-h<strong>and</strong> side) for the<br />
TFP measure of De Loecker allowing for three-digit industry-specific dem<strong>and</strong><br />
elasticities, are displayed in table 1.6. The first column for each type of share<br />
consists of the share-weighted average productivity measured calculated on<br />
the basis of (1.17), normalized to 1 for 1996. The second <strong>and</strong> third column<br />
show the percentage contribution of the within productivity component <strong>and</strong><br />
the reallocation share to aggregate weighted TFP respectively.<br />
One might also argue that in such a case, it is preferable to allow not only the industry<br />
output coefficient, but also the input coefficients to vary across products, i.e. to estimate<br />
a separate production function for each of the products (or sub-sectors in the absence of<br />
product-level information).<br />
22 An alternative to the OP decomposition is provided by Foster et al. (2006) . In<br />
addition to a within firm <strong>and</strong> reallocation term, they allow for a separate net-entry <strong>and</strong><br />
interaction term. Given the complexity of their decomposition, it is beyond the scope of<br />
the present paper to apply it here.
Empirical application: Food <strong>and</strong> beverages industry in Belgium 47<br />
Year Turnover shares Employment shares<br />
Weighted Mean Reallo- Weighted Mean Reallo-<br />
Index TFP cation Index TFP cation<br />
(1996 = 1) (%) (%) (1996 = 1) (%) (%)<br />
1996 1.000 102.71 -2.71 1.000 107.18 -7.18<br />
1997 0.9260 101.95 -1.95 0.9420 104.58 -4.58<br />
1998 0.9166 101.09 -1.09 0.9338 103.54 -3.54<br />
1999 0.935 100.96 -0.96 0.9637 102.22 -2.22<br />
2000 1.0506 101.14 -1.14 1.0791 102.76 -2.76<br />
2001 1.0451 100.84 -0.84 1.0748 102.32 -2.32<br />
2002 1.0297 100.51 -0.51 1.0687 101.06 -1.06<br />
2003 1.0323 99.95 0.05 1.0855 99.19 0.81<br />
Weighted average productivity is calculated according to (1.17),<br />
weights are firm-level turnover or employment shares.<br />
Table 1.6: Decomposition aggregate TFP: De Loecker methodology
48 Total factor productivity estimation<br />
From table 1.6, it is clear that most of the productivity improvements<br />
realized in the food <strong>and</strong> beverages sector since 1996 have been associated<br />
with within firm productivity growth. When employment shares rather than<br />
turnover shares are used (right-h<strong>and</strong> side of the table), the reallocation share<br />
is somewhat larger than for the case of turnover shares. Hence, I conclude<br />
that most of the productivity increases realized in the food <strong>and</strong> beverages<br />
industry in Belgium following the dioxin sc<strong>and</strong>al in 1999 were due to the<br />
average firm becoming more productive, while reallocation of market share<br />
(either in terms of employment or turnover) has only played a minor role.<br />
Reallocation shares are consistently negative throughout the sample period,<br />
both using turnover <strong>and</strong> employment shares, with the exception of 2003,<br />
when it becomes positive in both cases.<br />
For comparison purposes, table 1.7 summarizes the results of the OP decomposition<br />
for each of the different estimators listed in table 1.2. The table<br />
shows, apart from weighted normalized TFP in 2003 for each of the estimators,<br />
the average shares of unweighted average TFP <strong>and</strong> the sample covariance<br />
term in aggregate weighted industry productivity. Values reported are<br />
eight-year averages. Although the within firm growth component dominates<br />
regardless of the estimators applied to calculate industry productivity, there<br />
are some important differences worth noting.<br />
Of the eight decompositions summarized in table 1.7, five yield similar<br />
results. Specifically, for the OLS, GMM, OP <strong>and</strong> De Loecker estimators<br />
the sample covariance terms (both for turnover <strong>and</strong> employment) are small<br />
<strong>and</strong> positive. For both fixed effects estimators however, reallocation shares<br />
are much larger, although still positive. The De Loecker estimator allowing<br />
for variety-specific dem<strong>and</strong>, as well as the LP estimator yield a small but<br />
consistently negative sample covariance term between productivity <strong>and</strong> either<br />
output or employment.<br />
1.5 Conclusions<br />
This paper has reviewed the methodological issues arising when total factor<br />
productivity or TFP is estimated at the establishment level. The traditional
Conclusions 49<br />
Turnover shares Employment shares<br />
Weighted Mean Reallo- Weighted Mean Reallo-<br />
Method Index TFP (2) cation Index TFP (2) cation<br />
(1996 = 1)(1) (%) (%) (2) (1996 = 1)(1) (%) (%) (2)<br />
OLS 1.1393 99.24 0.76 1.1606 99.39 0.61<br />
Fixed Effects (balanced) 1.1619 92.18 7.82 1.1773 84.58 15.42<br />
Fixed Effects (unbalanced) 1.1657 90.42 9.58 1.1787 81.53 18.47<br />
GMM 1.1453 98.52 1.48 1.1686 96.44 3.56<br />
OP (no survival correction) 1.1377 99.45 0.55 1.1582 99.57 0.43<br />
OP (survival correction) 1.1375 99.48 0.52 1.1580 99.64 0.36<br />
Levinsohn-Petrin 1.1293 101.68 -1.68 1.1504 105.08 -5.08<br />
De Loecker 1.1148 97.74 2.26 1.1613 96.62 3.38<br />
DL (variety-specific) 1.0323 101.15 -1.15 1.0855 102.86 -2.86<br />
Weighted average productivity is calculated as in equation 17, weights are firm-level turnover of employment<br />
shares. (1) Weighted normalized TFP in 2003 (1996 = 1). (2) Values reported are eight-year<br />
averages of the shares of unweighted average TFP <strong>and</strong> the sample covariance term.<br />
Table 1.7: Comparison of decomposition results
50 Total factor productivity estimation<br />
biases introduced by the simultaneity of input choice <strong>and</strong> endogeneity of attrition<br />
have been discussed; as well as a number of issues that have emerged<br />
more recently, i.e. related to imperfect competition in input <strong>and</strong>/or output<br />
markets <strong>and</strong> endogeneity of product choice. Various estimators have been<br />
introduced in the literature attempting to overcome some of these issues.<br />
Given the relatively poor performance <strong>and</strong> shortcomings of the more traditional<br />
estimators, i.e. fixed effects <strong>and</strong> GMM; a number of semiparametric<br />
estimators have been introduced, which have been briefly reviewed here. A<br />
recent extension to these estimators taking the omitted output price bias into<br />
account; in addition to dealing adequately with simultaneity <strong>and</strong> selection<br />
issues has also been discussed.<br />
I have illustrated the performance of these estimators using data on the<br />
food <strong>and</strong> beverages industry in Belgium in the period 1996 to 2003, when<br />
the sector was undergoing significant changes <strong>and</strong> restructuring, especially<br />
following the outbreak of the dioxin crisis in 1999. Findings confirm the<br />
theoretically expected biases in traditional production function estimates,<br />
obtained using OLS. Moreover, the evolution of industry TFP over the sample<br />
period shows a clear upward trend in aggregate productivity following the<br />
dioxin sc<strong>and</strong>al in 1999.<br />
Which estimator would researchers ideally want to use then? In light of the<br />
traditionally poor performance of both the GMM <strong>and</strong> fixed effects estimators,<br />
it would seem that the semiparametric estimators are to be preferred, <strong>and</strong><br />
specifically the Olley-Pakes methodology. Moreover, comparing aggregate<br />
industry productivity growth patterns for the different estimators shows that<br />
a failure to take imperfect competition in output markets into account may<br />
yield misleading results concerning the timing <strong>and</strong> magnitude of observed<br />
industry growth patterns, hence favoring the estimator of De Loecker.<br />
However, the choice of which estimator to use will essentially also depend<br />
on the data at h<strong>and</strong>. Reliable industry output measures are not always<br />
available to the researcher. Similarly, positive investment data are not always<br />
available for a sufficiently large sample of firms within an industry or<br />
might not be trustworthy. Data can also be prone to measurement error
Conclusions 51<br />
or production technology may differ widely within an industry, invalidating<br />
some of the parametric methods discussed here.<br />
Van Biesebroeck (2007) compares the sensitivity of different estimators<br />
(index numbers, data envelopment analysis or DEA, stochastic frontiers, IV<br />
(GMM) <strong>and</strong> semiparametric estimation. He finds that the GMM-SYS estimator<br />
is the most robust technique when there is a lot of measurement error<br />
or some technological heterogeneity. However, for the GMM-SYS estimator<br />
to be reliable, at least some of the productivity differences have to be<br />
constant over time. He further notes that the GMM estimator might lead<br />
to downwardly biased input coefficients when measurement error becomes<br />
severe. When measurement error is small, technology is heterogeneous <strong>and</strong><br />
returns to scale are not constant, non-parametric techniques such as DEA or<br />
index numbers should be preferred.<br />
In spite of the multitude of estimators that have been developed in recent<br />
years in order to achieve consistent estimates of total factor productivity, a<br />
number of issues remain to be resolved. In particular, both the lack of a<br />
formal correction for the omitted input price bias in the presence of imperfect<br />
competition in input markets, as well as the implications of endogenous<br />
product choice following from BRS (2005, 2006b) offer ample scope for future<br />
research.
Chapter 2<br />
<strong>Globalization</strong> Drives Strategic<br />
Product Switching<br />
2.1 Introduction<br />
<strong>Globalization</strong> has led to increasingly integrated markets across the world,<br />
changing the competitive environment in which firms operate. In the face of<br />
international competition in domestic <strong>and</strong> foreign markets, the least productive<br />
firms may be forced into bankruptcy while the most productive ones will<br />
take advantage of new business opportunities in foreign markets. Moreover,<br />
incumbent firms may respond by increasing their productivity or, since this<br />
may prove difficult in mature industries, by diversifying into a different industry<br />
or product variety. The importance of firm shutdown <strong>and</strong> changes in<br />
the product mix as a response to pressures from international trade has been<br />
highlighted in recent empirical studies (see, for example, Bernard, Jensen<br />
<strong>and</strong> Schott, 2006a; <strong>and</strong> Greenaway, Gullstr<strong>and</strong> <strong>and</strong> Kneller, 2008). However,<br />
this literature has focused on only one aspect of trade -import competitionignoring<br />
firm dynamics induced by profitable opportunities in export markets.<br />
This second aspect is potentially very important for emerging markets,<br />
particularly in the aftermath of trade liberalization.<br />
∗ This is joint work with Marialuz Moreno Badia (IMF) <strong>and</strong> Veerle Slootmaekers (OECD<br />
<strong>and</strong> KU Leuven, LICOS). Published as LICOS Discussion Paper 209/2008 <strong>and</strong> IMF Working<br />
Paper 08/246. The views expressed herein are those of the authors <strong>and</strong> should not be<br />
held to represent those of the institutions of affiliation.<br />
53
54 <strong>Globalization</strong> drives strategic product switching<br />
The purpose of this paper is to analyze the impact of international trade<br />
on firm dynamics, focusing on how production patterns are adjusted in response<br />
to import competition <strong>and</strong> changing conditions in export markets.<br />
Our focus is Estonian manufacturing firms from 1997 to 2005. Estonia is<br />
a particularly interesting case because of the firm restructuring <strong>and</strong> trade<br />
liberalization that took place in the aftermath of the transition process <strong>and</strong><br />
in the run-up to European Union (EU) membership. Buoyed partly by the<br />
Association Agreement with the EU, Estonian exports of goods increased<br />
by 240 percent 1 . This extraordinary performance was accompanied by an<br />
increase in product variety (K<strong>and</strong>ogan, 2006) <strong>and</strong> a shift toward exports of<br />
higher quality <strong>and</strong> technological intensity (Fabrizio et al., 2007). This seems<br />
to suggest that Estonian firms may not have merely responded defensively to<br />
increased competition from importers, but also reacted offensively by taking<br />
advantage of the opportunities created by trade liberalization.<br />
To identify these effects, we use a longitudinal data set of Estonian manufacturing<br />
firms <strong>and</strong> consider three potential firm strategies: continue its business,<br />
switch products, or close down. To model these strategic alternatives, we<br />
estimate a multinomial logit model in which the firm decision is a function of<br />
firm-level <strong>and</strong> product-market characteristics. Following the previous literature,<br />
we include the value of imports, type of trade (intra- or interindustry),<br />
<strong>and</strong> revealed comparative advantage as measures of trade. To identify the<br />
effect of export opportunities, we also include in our estimation the value<br />
of exports, the degree of competition in export markets, <strong>and</strong> the quality of<br />
exports relative to direct competitors.<br />
Overall, we find that firm exit is mainly determined by firm characteristics,<br />
whereas product switching also depends on conditions in export markets.<br />
In particular, firms are more likely to switch if they are in sectors without<br />
revealed comparative advantage, with less exports, or with lower product<br />
quality relative to export competitors. One interpretation of this result is<br />
that firms are more willing to incur the sunk costs of breaking into a new<br />
1 The Association Agreement with the EU was signed in 1995 <strong>and</strong> entered into force<br />
in 1998. The agreement replaced previous treaties with the EU (an Agreement on Trade<br />
<strong>and</strong> Commercial Cooperation, signed in 1992, which was converted into a Free Trade<br />
Agreement in 1994). For a more detailed description, see Weber <strong>and</strong> Taube (1999).
Introduction 55<br />
product line or industry when the long-term prospects of the current export<br />
market for their products are weak, particularly early in the sample when<br />
trade flows were increasing rapidly. These results are in contrast with previous<br />
studies on industrial countries that find that firms switch products as a<br />
defense against low-cost imports. Interestingly, we find that the conditions<br />
on export markets matter predominantly for switches within the same industry<br />
or what we call product switches 2 . Switches across industries on the other<br />
h<strong>and</strong> are determined by firm-level characteristics. These results suggest that<br />
product diversification was a major strategy of Estonian firms faced with<br />
increasing trade openness. Finally, we find a positive link between a firm’s<br />
capital intensity <strong>and</strong> quality upgrading. However, moving up the quality<br />
ladder is not necessarily related to technology upgrading; it occurs mainly<br />
within the medium-high-tech sector.<br />
The literature on the relationship between globalization <strong>and</strong> firm dynamics<br />
has exp<strong>and</strong>ed rapidly in recent years (see, for example, Melitz, 2003; Bernard,<br />
Eaton, Jensen <strong>and</strong> Kortum, 2003; <strong>and</strong> Helpman, Melitz <strong>and</strong> Yeaple, 2004).<br />
This literature builds on Melitz’s (2003) dynamic industry model with heterogeneous<br />
firms, where sunk costs of market entry result in self-selection<br />
into export markets. More recent theoretical work models how these dynamics<br />
interact with a country’s industry characteristics (Bernard, Redding <strong>and</strong><br />
Schott, 2007b <strong>and</strong> Melitz <strong>and</strong> Ottaviano, 2008). Rigorous empirical work,<br />
triggered by the work of Bernard <strong>and</strong> Jensen (1995), has further nourished<br />
the underst<strong>and</strong>ing of firm adjustment to trade liberalization <strong>and</strong> falling trade<br />
costs 3 . In short, these studies document substantial variation in productivity<br />
across firms, frequent firm entries <strong>and</strong> exits, sizeable sunk costs of entry into<br />
export markets, <strong>and</strong> better performance among exporting firms.<br />
2 Product switches are defined as changes in industry at the four-digit level henceforth.<br />
Although four-digit NACE codes are not true products in the strictest sense of the word,<br />
this is the most detailed classification we have in our data. This notation has also been<br />
used in Bernard et al. (2006a) <strong>and</strong> Greenaway et al. (2008).<br />
3 For an overview of the empirical literature, see Tybout (2003) <strong>and</strong> Bernard, Jensen,<br />
Redding <strong>and</strong> Schott (2007c).
56 <strong>Globalization</strong> drives strategic product switching<br />
Our paper is more closely related in spirit to the work of Bernard et al.<br />
(2006a) <strong>and</strong> Greenaway et al. (2008) 4 . These papers reveal a new dimension<br />
of adjustment to increased international competition by illustrating that<br />
firms are more likely to change their product mix than to shut down in response<br />
to globalization. Controlling for a number of firm <strong>and</strong> industry characteristics,<br />
they find that firms are more likely to switch away from industries<br />
where exposure to low-wage countries is high. Bernard et al. (2006a) note<br />
that U.S. firms shift towards industries facing less competitive pressure, but<br />
with greater capital <strong>and</strong> skill intensity than the industry of origin. Greenaway<br />
et al. (2008) broaden the analysis, <strong>and</strong> consider mergers <strong>and</strong> acquisitions as<br />
a third exit strategy. They report that closure is the least likely exit strategy<br />
in Sweden, as most firms merge with or acquire another firm in response<br />
to higher levels of international competition. A primary contribution of our<br />
paper relative to these studies is that we consider explicitly the role of export<br />
opportunities in determining firm dynamics. Also, ours is the first paper to<br />
examine the impact of globalization on firm dynamics in an emerging market<br />
context 5 . Since countries at different stages of development exhibit large<br />
differences in terms of firm size distribution, efficiency, <strong>and</strong> cost structure<br />
it is important to explore whether enterprises in emerging markets respond<br />
differently to globalization than enterprises in more advanced countries.<br />
The remainder of this paper is organized as follows. Section 2.2 describes<br />
the industry dynamics in Estonia. Section 2.3 outlines the estimation strategy<br />
used to analyze the impact of international trade on firm dynamics in<br />
Estonia. Section 2.4 presents the results. Section 2.5 discusses the robustness<br />
checks. Section 2.6 concludes.<br />
2.2 Industry dynamics in Estonia<br />
The data used in this paper are provided by the Estonian Business Registry<br />
<strong>and</strong> cover the years 1997-2005. The data set is an unbalanced panel<br />
4 These papers build on the work of Bartelsman <strong>and</strong> Doms (2000), Foster, Haltiwanger<br />
<strong>and</strong> Krizan (2006) <strong>and</strong> Bernard <strong>and</strong> Jensen (2007a).<br />
5 Goldberg, Kh<strong>and</strong>elwal, Pavcnik <strong>and</strong> Topalova (2008) analyze the response of Indian<br />
firms to trade liberalization but they focus on product margins rather than looking into<br />
the firm dynamics per se.
Industry dynamics in Estonia 57<br />
containing detailed information on balance sheets <strong>and</strong> income statements of<br />
all registered firms in Estonia. The unit of observation is the firm, which<br />
can be tracked over time using a unique registration code 6 . As all business<br />
entities in Estonia are required to file their annual accounts with the registry,<br />
the data set comprises firms from all size classes, including microenterprises<br />
with less than 10 employees.<br />
Our primary interest lies in identifying the impact of international competition<br />
on firms’ strategic choices. Therefore, we focus on the sector for which we<br />
observe trade flows at a disaggregated level, namely, the manufacturing sector.<br />
Nonetheless, since companies active in sectors other than manufacturing<br />
are equally obliged to report to the registry, we are able to identify not only<br />
industry switches within the manufacturing sector, but also switches to other<br />
sectors of the economy 7 . The sample used in the empirical analysis consists<br />
of 4,844 firms <strong>and</strong> 16,117 observations (see Table 2.1). Entry <strong>and</strong> exit are<br />
observed, <strong>and</strong> the number of manufacturing entities in the registry increased<br />
significantly over the sample period, from 1,196 firms in 1997 to 2,767 firms<br />
in 2004 8 . For each firm we observe its primary sector of activity at the fourdigit<br />
NACE (General Industrial Classification of Economic Activities within<br />
the European Community) level 9 . Unfortunately we do not have information<br />
on the total number of different products per firm, but we do observe changes<br />
in its main product line over time.<br />
<strong>Firm</strong> exit (Exitit+1) is identified using the firm’s official liquidation date,<br />
available from the registry. In addition to exit, we are interested in industry<br />
switches at both the two-digit (Switch2d,it+1) <strong>and</strong> four-digit (Switch4d,it+1)<br />
6 For a detailed description of the data, see section 2.A. More information on the<br />
Estonian Business Registry can be found in Masso et al. (2004)<br />
7 <strong>Firm</strong>s switching to other sectors of the economy are retained until their last year of<br />
activity within the manufacturing sector.<br />
8 Since we cannot observe switching for 2005 (no data for 2006 are available), this year<br />
is omitted from the analysis.<br />
9 The classification used by the Registry is the EMTAK Classification of Economic Activities<br />
of Estonia. EMTAK is a five-digit extension of the four-digit European NACE<br />
classification system, the official statistical classification system of economic activities in<br />
the European Community. The first four digits of the EMTAK codes are therefore equivalent<br />
to the NACE (Rev. 1.1.) codes. The NACE classification system can be downloaded<br />
from the Eurostat Ramon server (http://europa.eu.int/comm/eurostat/ramon/).
58 <strong>Globalization</strong> drives strategic product switching<br />
Number of Industry Product<br />
Year observations switch (1) switch (1) Exit (1)<br />
Panel A: Distribution of the full sample<br />
1997 1,196 14.8 21.7 5.1<br />
1998 1,398 9.5 14.5 4.9<br />
1999 1,621 11.0 16.5 4.4<br />
2000 1,930 8.9 12.8 3.1<br />
2001 2,175 6.6 11.4 3.3<br />
2002 2,396 6.2 9.9 2.2<br />
2003 2,634 6.0 9.3 1.8<br />
2004 2,767 4.8 7.3 0.7<br />
Total (observations) 16,117 1,244 1,912 452<br />
Total (firms) 4,844 1,090 1,566 452<br />
Panel B: Distribution for microenterprises (2)<br />
1997 515 19.6 26.8 6.8<br />
1998 616 12.2 16.6 6.3<br />
1999 765 12.7 17.9 5.2<br />
2000 944 11.2 14.4 3.2<br />
2001 1,100 8.6 13.5 4.5<br />
2002 1,232 8.4 11.5 2.5<br />
2003 1,423 7.6 11.1 2.0<br />
2004 1,553 6.2 8.4 0.8<br />
Total (observations) 8,148 782 1,092 267<br />
Total (firms) 3,214 709 949 267<br />
Notes: (1) Industry (product) switches are identified at the two-digit (four-digit)<br />
level. Values are percentages. (2) <strong>Firm</strong>s with less than 10 employees.<br />
Table 2.1: Exits <strong>and</strong> industry switches, 1997-2004<br />
NACE level. For the remainder of the paper, we refer to two-digit switches<br />
as industry switches <strong>and</strong> to four-digit switches as product switches. Whereas<br />
Bernard et al. (2006a) study switches at the product level, Greenaway et al.<br />
(2008) study switches at the industry level because they want to focus on big<br />
changes in production <strong>and</strong> to minimize the possibility of classification problems.<br />
Nevertheless, the latter analyze the product switches as a robustness<br />
check <strong>and</strong> report substantial differences in the determinants of both types of<br />
switching. In our analysis, we focus on the product-level switches, but we<br />
also discuss the differences with the industry switches.
Industry dynamics in Estonia 59<br />
Observations Industry switches Product switches Exits<br />
NACE Two-digit industry Number Percent Number Percent Number Percent Number Percent<br />
15 Food <strong>and</strong> beverages 1,314 8.2 83 6.3 131 10.0 58 4.4<br />
17 Textiles 689 4.3 44 6.4 61 8.9 21 3.0<br />
18 Clothing 1,396 8.7 59 4.2 111 8.0 48 3.4<br />
19 Leather (products) 270 1.7 8 3.0 10 3.7 6 2.2<br />
20 Wood (products) 3,529 21.9 255 7.2 410 11.6 120 3.4<br />
21 Pulp, paper (products) 198 1.2 13 6.6 16 8.1 3 1.5<br />
22 Publishing <strong>and</strong> printing 1,455 9.0 74 5.1 139 9.6 33 2.3<br />
23 Coke, petroleum products 5 0.0 1 20.0 1 20.0 0 0.0<br />
24 Chemicals 356 2.2 41 11.5 48 13.5 8 2.2<br />
25 Rubber <strong>and</strong> plastic 601 3.7 48 8.0 74 12.3 12 2.0<br />
26 Non-metallic mineral products 509 3.2 43 8.5 63 12.4 13 2.6<br />
27 Basic metals 16 0.1 5 31.3 5 31.3 0 0.0<br />
28 Fabricated metal products 1,819 11.3 150 8.3 273 15.0 41 2.3<br />
29 Machinery <strong>and</strong> equipment 812 5.0 125 15.4 160 19.7 18 2.2<br />
30 Office machinery <strong>and</strong> computers 59 0.4 17 28.8 17 28.8 0 0.0<br />
31 Electrical machinery 340 2.1 40 11.8 51 15.0 9 2.6<br />
32 Communications equipment 290 1.8 45 15.5 53 18.3 5 1.7<br />
33 Medical, precision, optical instruments 413 2.6 38 9.2 44 10.7 6 1.5<br />
34 Motor vehicles 121 0.8 12 9.9 14 11.6 2 1.7<br />
35 Other transport equipment 295 1.8 34 11.5 37 12.5 8 2.7<br />
36 Furniture 1,630 10.1 109 6.7 194 11.9 41 2.5<br />
Total 16,117 100.0 1,244 7.7 1912 11.9 452 2.8<br />
Industry (product) switches are defined at the two-digit (four-digit) NACE level. All industry switches are also observed at the<br />
product level; out of 1,912 product switches in the sample, 1,244 are observed at both the product <strong>and</strong> industry level.<br />
Table 2.2: Sector distribution
60 <strong>Globalization</strong> drives strategic product switching<br />
Table 2.1 reports the distribution of the sample over time. Out of 4,844<br />
firms, 452 firms exited between 1997 <strong>and</strong> 2004, <strong>and</strong> 1,566 firms switched<br />
products. Of the latter group, 1,090 changed to a different industry <strong>and</strong> 476<br />
firms only switched products within the same industry. Industry <strong>and</strong> product<br />
switches were more frequent at the end of the 1990s, but the rates declined<br />
steadily toward the end of the sample period. On average, the industry<br />
switching rate is 7.7 percent, compared with a switching rate of 11.9 percent<br />
at the product level. The product switching rate is only slightly higher than<br />
the figures reported by Bernard et al. (2006a) <strong>and</strong> Greenaway et al. (2008),<br />
who find a product switching rate of 7-8 percent for the U.S. <strong>and</strong> Sweden,<br />
respectively.<br />
Tables 2.2, 2.3 <strong>and</strong> 2.4 give a preliminary indication that there are broad<br />
differences in firm dynamics across sectors. First, it is clear from Table 2.2<br />
that sectors facing a high number of exits also tend to undergo more industry<br />
changes. This is, however, related to the size of the sectors, <strong>and</strong> switching<br />
<strong>and</strong> exit rates are not necessarily correlated. For example, the switching rates<br />
of ”basic metals,” ”office machinery <strong>and</strong> computers,” <strong>and</strong> ”coke, petroleum<br />
products” are very high but the exit rates in these sectors are zero. Rather<br />
than permanently exiting the market, companies in those sectors seem to<br />
look for other opportunities by switching to a different industry. Second,<br />
about 35 percent of all product switches occur within the same two-digit<br />
sector; 23 percent to other two-digit manufacturing sectors; 4 percent to the<br />
primary sector; <strong>and</strong> 38 percent to services (Table 2.3 10 ).<br />
Third, the majority of firms in our sample are active in low-tech manufacturing<br />
(65 percent) <strong>and</strong> most of the product switches take place among firms in<br />
this group (Table 2.4 11 ). However, product switches from medium-high-tech<br />
10 The primary sector comprises NACE 1-14 (agriculture <strong>and</strong> mining activities);<br />
manufacturing sector or secondary sector comprises NACE 15-37; <strong>and</strong> services or the<br />
tertiary sector comprises NACE 40-99. Most firms that switch to services end up in<br />
industries like wholesale trade <strong>and</strong> retail trade. We explore further the nature of these<br />
switches in Section 2.4.<br />
11 Using the Eurostat classification, which is available at<br />
http://europa.eu.int/comm/eurostat/ in the section “Science <strong>and</strong> Technology”,<br />
we classify the manufacturing sectors according to technology intensity <strong>and</strong> services<br />
according to knowledge intensity. Appendix tables 2.A.1 <strong>and</strong> 2.A.2 provides a list of
Product Within same<br />
Switches Industry To To To<br />
NACE Two-digit industry (N) (two-digit) primary manufacturing services<br />
Industry dynamics in Estonia 61<br />
15 Food <strong>and</strong> beverages 131 36.6 14.5 2.3 46.6<br />
17 Textiles 61 27.9 0.0 39.3 32.8<br />
18 Clothing 111 46.9 0.0 23.4 29.7<br />
19 Leather (products) 10 20.0 0.0 50.0 30.0<br />
20 Wood (products) 410 37.8 11.5 15.9 34.9<br />
21 Pulp, paper (products) 16 18.8 0.0 31.3 50.0<br />
22 Publishing <strong>and</strong> printing 139 46.8 1.4 10.1 41.7<br />
23 Coke, petroleum products 1 0.0 0.0 0.0 100.0<br />
24 Chemicals 48 14.6 4.2 22.9 58.3<br />
25 Rubber <strong>and</strong> plastic 74 35.1 0.0 32.4 32.4<br />
26 Non-metallic mineral products 63 31.8 9.5 25.4 33.3<br />
27 Basic metals 5 0.0 0.0 100.0 0.0<br />
28 Fabricated metal products 273 45.1 0.0 24.5 30.4<br />
29 Machinery <strong>and</strong> equipment 160 21.9 1.9 30.0 46.3<br />
30 Office machinery <strong>and</strong> computers 17 0.0 0.0 5.9 94.1<br />
31 Electrical machinery 51 21.6 0.0 29.4 49.0<br />
32 Communications equipment 53 15.1 0.0 39.6 45.3<br />
33 Medical <strong>and</strong> optical instruments 44 13.6 0.0 38.6 47.7<br />
34 Motor vehicles 14 14.3 7.1 28.6 50.0<br />
35 Other transport equipment 37 8.1 0.0 46.0 46.0<br />
36 Furniture 194 43.8 1.6 23.7 30.9<br />
Total 1,912 34.9 4.3 22.7 38.0<br />
The last four columns in the table decompose product switches into four different categories: product switches<br />
that occur within the same two-digit manufacturing sector; to the primary sector; to other two-digit manufacturing<br />
sectors; <strong>and</strong> to services. The row total of these four categories equals 100 percent for each two-digit industry. Values<br />
are percentages, unless otherwise indicated.<br />
Table 2.3: Four-digit product switches decomposed
62 <strong>Globalization</strong> drives strategic product switching<br />
Table 2.4: Destination of product switches by technology class<br />
Notes: Sectors are classified according to technology intensity, based on the classification of industry according to technology<br />
level from Eurostat. The list of NACE (Rev. 1.1) codes assigned to each particular technology class is given in appendix<br />
tables 2.A.1 <strong>and</strong> 2.A.2.<br />
N 16,117 452 1,912 52 122 298 630 174 445 191<br />
High tech 5.1 2.7 6.4 25.2 12.2 4.9 5.7 17.1 32.5 2.4<br />
Medium-high tech 9.8 8.0 13.9 4.5 21.1 20.7 2.3 7.5 32.3 11.7<br />
Medium-low tech 20.0 16.4 23.6 1.3 10.4 47.0 7.5 7.5 15.5 10.6<br />
Low tech 65.0 73.0 56.1 0.3 0.4 2.3 54.4 9.2 23.2 10.2<br />
Total 100.0 100.0 100.0 2.7 6.4 15.6 32.9 9.1 23.3 10.0<br />
Destination industry by technology class<br />
Number Number Number Manufacturing Services<br />
Origin of of of product High Medium- Medium- Low Knowl. Less-knowl.<br />
industry firms exits switches tech high-tech low-tech tech intensive intensive other
Determinants of firm dynamics 63<br />
<strong>and</strong> medium-low-tech industries account for 13.9 percent <strong>and</strong> 23.6 percent of<br />
switches, respectively, which is more than the proportion of observations in<br />
these groups (10 percent <strong>and</strong> 20 percent, respectively). Meanwhile, about<br />
three-fourths of the total number of bankruptcies take place in low-tech<br />
manufacturing sectors. Within the manufacturing sector, firms switch mostly<br />
to products with similar technology content. This is especially the case for<br />
the medium-low-tech to low-tech firms, where half of all firms switch to other<br />
sectors in the same technology class. Table 2.4 also shows that a significant<br />
proportion of manufacturing firms are moving into the less-knowledgeintensive<br />
services sector, independently of the technology intensity of the<br />
origin industry.<br />
2.3 Determinants of firm dynamics<br />
At the end of each observed period, a firm decides whether to continue its<br />
activities in the same product line or to exit. It can exit from a particular<br />
product line <strong>and</strong> enter a new product line or even a new industry in the<br />
next year, or it can exit altogether (i.e., “die”), in which case the firm drops<br />
permanently out of the sample. To model these strategic alternatives, we<br />
estimate a multinomial logit model, as follows (Greene, 2008, p. 844):<br />
Pr Yit+1 = j <br />
Xt ¯ ′<br />
= exp β jXit<br />
<br />
1 +<br />
2<br />
k=1<br />
exp (β ′ kXit)<br />
<br />
(2.1)<br />
where j equals 0 for continuing firms, 1 for firms that switch products<br />
(industries), <strong>and</strong> 2 for closing enterprises. The vector of covariates (Xit)<br />
contains a number of one-year lagged firm- <strong>and</strong> product-level variables, in<br />
addition to a constant, year dummies, <strong>and</strong> two-digit industry dummies:<br />
<strong>Firm</strong> level:<br />
ln(Sizeit), ln(Ageit), ln(Capitalit), ln(Wageit), ln(TFPit), Foreignit<br />
NACE (Rev. 1.1) codes assigned to each particular technology class.
64 <strong>Globalization</strong> drives strategic product switching<br />
Domestic market:<br />
International market:<br />
Sunkjt, Herfjt<br />
ln(Importsjt), IITjt, CAjt, ln(Exportsjt), Herfexjt, ln(UV Rjt)<br />
Subscript i refers to firms, j to products <strong>and</strong> t to time. All productlevel<br />
variables are defined at the four-digit NACE level. For the complete<br />
definitions of these variables, we refer to section 2.A.2.<br />
Whereas the determinants of firm <strong>and</strong> plant death have been an active area<br />
of empirical <strong>and</strong> theoretical research, the literature on product switching is<br />
still in its infancy. Our motivation for deciding which variables to include in<br />
the multinomial logit model is therefore grounded in the relevant literature<br />
<strong>and</strong> complemented by the salient facts emerging from the summary statistics<br />
presented in Table 2.5. Our goals are (1) to identify groups of variables–firm<br />
<strong>and</strong> product level–that may matter for the exit/switching decision; <strong>and</strong> (2)<br />
to check whether they differ according to the exit strategy.<br />
2.3.1 <strong>Firm</strong> characteristics<br />
A common feature of the theoretical models on heterogeneous firms is the<br />
negative relation between failure rates <strong>and</strong> a firm’s age <strong>and</strong> size (Jovanovic,<br />
1982; Hopenhayn, 1992 <strong>and</strong> Ericson <strong>and</strong> Pakes, 1995). This selection effect is<br />
driven by the interaction of economies of scale <strong>and</strong> an idiosyncratic learning<br />
process, <strong>and</strong> is confirmed by a large empirical literature 12 . To capture this<br />
scale effect, we include a firm-size variable–measured by employment in year t<br />
(Sizeit). Additionally, Dunne, Klimek <strong>and</strong> Roberts (2005) demonstrate that<br />
a firm’s past experience positively affects its survival rate. We control for this<br />
experience effect by including the age of the firm (Ageit), which is calculated<br />
12 The existence of economies of scale implies, on the one h<strong>and</strong>, sunk costs for larger<br />
firms, which discourage or delay exit, <strong>and</strong>, on the other h<strong>and</strong>, strong cost disadvantages<br />
for smaller companies. Moreover, larger firms are in general more diversified, which makes<br />
them less vulnerable to negative productivity shocks. See Caves (1998) for an overview of<br />
the empirical literature on firm turnover.
Determinants of firm dynamics 65<br />
as the number of years since its official registration date. A priori, we expect<br />
exit to be negatively related to size <strong>and</strong> age. However, the effect of these<br />
variables on product switching is ambiguous: small <strong>and</strong> young firms have<br />
greater flexibility to switch but larger <strong>and</strong> older firms have more experience<br />
<strong>and</strong> resources to switch 13 .<br />
A further implication of the theoretical models on firm dynamics is the<br />
link between the productivity of a firm <strong>and</strong> its survival. Upon entering the<br />
market, firms pay a sunk cost of entry, after which they discover their true<br />
productivity. If firm productivity is below the zero profitability cutoff, the<br />
firm will immediately exit the market. Hence, from these models, a negative<br />
relationship between firm-level productivity <strong>and</strong> exit is predicted, a finding<br />
which is confirmed by the empirical literature. Fariñas <strong>and</strong> Ruano (2005)<br />
explore the relationships among entry, exit, <strong>and</strong> firm-level productivity for<br />
a sample of Spanish manufacturing firms. They find that entry <strong>and</strong> exit<br />
decisions by firms are systematically related to productivity differences <strong>and</strong><br />
that the productivity distribution of exiting firms is stochastically dominated<br />
by that of continuing firms. We therefore expect firm productivity to be<br />
negatively related to firm death.<br />
Switching products could be seen as a form of exit, where unproductive<br />
companies that cannot face competition turn to a market with a lower degree<br />
of competition. Yet, entering a new industry or product market implies that<br />
a firm has to incur additional adjustment costs, related to the production of a<br />
new good. If these sunk costs of entry in a new product market are substantial,<br />
the same reasoning as above applies, that is, only the more productive<br />
firms will be able to enter this product market <strong>and</strong> start production. We<br />
therefore expect firm-level productivity to be negatively or positively related<br />
to product switching –depending on the importance of product market entry<br />
barriers. Productivity is defined as total factor productivity (TFPit). We<br />
estimate TFP at the two-digit industry level, while allowing firms to switch<br />
industries over time, <strong>and</strong> apply the methodology of Levinsohn <strong>and</strong> Petrin<br />
13 In fact, Greenaway et al. (2008) do not find significant effects of age <strong>and</strong> size on<br />
industry switching. Bernard et al. (2006a), on the other h<strong>and</strong>, find a positive coefficient<br />
for size <strong>and</strong> a negative coefficient for age (both significant).
66 <strong>Globalization</strong> drives strategic product switching<br />
(2003) to account for selectivity <strong>and</strong> simultaneity 14 .<br />
The labor cost variable (Wageit) is defined as total real labor costs per<br />
employee at the firm level. The expected effect of this variable on firm dynamics<br />
is ambiguous. To the extent that higher labor costs reflect higher<br />
skill intensities <strong>and</strong> associated sunk costs at the firm level (related to investments<br />
in firm-specific human capital), higher wages can, ceteris paribus, act<br />
as a barrier to exit from a particular product market. Audretsch <strong>and</strong> Mahmood<br />
(1995) find a negative relationship between wages <strong>and</strong> the propensity<br />
to exit in a study on the performance of 12,000 U.S. manufacturing plants.<br />
However, it is not clear whether this relationship will continue to hold when<br />
productivity at the firm level is taken into account. If firms pay higher wages<br />
for a given level of productivity, this could signal lower competitiveness <strong>and</strong>,<br />
hence, increase their exit probability, ceteris paribus (Konings, 2005).<br />
Capital intensity (Capitalit) is defined as total fixed assets per employee.<br />
<strong>Firm</strong>s with higher capital intensity are expected to face higher sunk costs,<br />
which will act as an exit barrier both at the product <strong>and</strong> firm level. <strong>Firm</strong>s<br />
with larger capital stock can also expect larger future returns for a given level<br />
of current productivity <strong>and</strong>, hence, will continue operating at lower productivity<br />
levels (Olley <strong>and</strong> Pakes, 1996). Therefore, we expect capital intensity to<br />
be negatively related to firm exit. In the case of product switching, the overall<br />
effect is less clear-cut. Even though capital intensity can be interpreted<br />
as a form of sunk cost, an opposite force may be at work. Instead of being<br />
passive or defensive when faced with increasing competitive pressures, a firm<br />
can actively look for the new opportunities offered by globalization. The<br />
production of these new goods requires additional investment to enter the<br />
new product market, which can be more easily incurred by capital-intensive<br />
firms that liquidate or transfer their assets into the new sector. Hence, the<br />
sign of the capital intensity variable for product switching is ambiguous.<br />
Several empirical studies have provided evidence that foreign multinational<br />
enterprises are more footloose than domestic firms, that is, they are more<br />
14 For a detailed description of the methodology employed to estimate TFP using the<br />
current data set, we refer to Moreno Badia <strong>and</strong> Slootmaekers (2008).
Determinants of firm dynamics 67<br />
likely to exit the market than domestic firms of comparable size, productivity,<br />
<strong>and</strong> wages (Görg <strong>and</strong> Strobl, 2003; Bernard <strong>and</strong> Sjöholm, 2003; <strong>and</strong><br />
Van Beveren, 2007a). However, foreign multinationals typically lack in-depth<br />
knowledge of the host market <strong>and</strong> need to overcome substantial disadvantages<br />
when entering foreign markets, causing them to incur higher sunk costs <strong>and</strong><br />
hence reducing their exit probabilities. Moreover, since they usually have<br />
more diversified sources of income, they can withst<strong>and</strong> larger shocks before<br />
being forced to exit the market. Hence, the effect of foreign ownership<br />
(Foreignit) on exit is ambiguous. Nevertheless, since multinationals are, by<br />
their very nature, more flexible than purely domestic firms, they can respond<br />
more quickly to adverse shocks in the host country <strong>and</strong> reinvent themselves<br />
through product switches. Hence, we expect to find a positive relationship<br />
between foreign ownership <strong>and</strong> product switching.<br />
2.3.2 Product market characteristics: Domestic<br />
Besides firm structure, the characteristics of the product market in which<br />
a firm operates also affect its evolution. While international competition<br />
may exert a strong influence on firm dynamics, conditions in the domestic<br />
market can also play an important role. This is especially true in a country<br />
with a history of a planned economic system, <strong>and</strong> where the domestic market<br />
may still be undergoing substantial changes as state monopolies are broken<br />
up <strong>and</strong> a new private sector emerges. Hence, in our empirical model on<br />
exit strategies, we incorporate the product market characteristics of both<br />
the home <strong>and</strong> foreign markets in which Estonian firms are active.<br />
Hopenhayn (1992) illustrates how an increase in sunk entry costs lowers<br />
the entry rate <strong>and</strong>, hence, also the probability of exit since incumbent firms<br />
face less competition through new entry. Intuitively, the argument is as<br />
follows. High initial investment costs to enter a product market will act as a<br />
natural deterrent to entry, since only the most promising firms will be able to<br />
start production. A lower entry rate implies less competition for incumbent<br />
producers <strong>and</strong> will induce fewer firms to exit. In addition, the high initial<br />
investment cost will act as an exit barrier, forcing inefficient firms to stay in<br />
the market to recover at least some of the initial sunk cost. We define sunk<br />
costs (Sunkjt) as the natural logarithm of the median of real sales in each
68 <strong>Globalization</strong> drives strategic product switching<br />
particular four-digit industry j at time t 15 . We expect to find a negative<br />
relationship between sunk costs at the product level <strong>and</strong> both types of exit<br />
in our empirical analysis, since higher sunk costs are associated with both<br />
higher entry <strong>and</strong> exit barriers at the product level.<br />
A central prediction of the stochastic dynamic model in Asplund <strong>and</strong> Nocke<br />
(2006) is that the level of firm turnover is positively related to the size of the<br />
domestic market. The smaller average markup in larger markets, resulting<br />
from tougher competition, implies that the marginal surviving firm has to be<br />
more efficient in larger than in smaller markets. To capture this competition<br />
effect, we include the Herfindahl-Hirschman index for the domestic market<br />
(Herfjt). It is defined as the sum of the squares of the market shares of each<br />
individual firm. As such, it ranges from 0 to 1 as it moves from a very large<br />
amount of very small firms to a single monopolistic producer.<br />
As noted by Görg <strong>and</strong> Strobl (2003), the impact of industry concentration<br />
on firm turnover is ambiguous. On the one h<strong>and</strong>, higher concentration is<br />
associated with wider price-cost margins, which will increase the survival<br />
chances of firms. However, behavior by aggressive rivals in a concentrated<br />
market can actually raise exit probabilities. Görg <strong>and</strong> Strobl (2003) in fact<br />
find a positive relationship between industry concentration <strong>and</strong> exit, using<br />
data on the Irish manufacturing sector between 1973 <strong>and</strong> 1996. Hence, the<br />
impact of concentration on firm exit (whether through firm death or product<br />
switching) can be negative or positive, depending on the behavior of the<br />
firm’s rivals in the domestic market.<br />
2.3.3 Product market characteristics: International<br />
Increased international trade implies higher competitive pressure in the<br />
domestic market. This pressure may force firms to improve their efficiency or<br />
to switch to products in which they have a comparative advantage. However,<br />
greater economic openness also creates new business opportunities in foreign<br />
markets. With improved access to the international markets, firms can raise<br />
15 This measure is known in the literature as the minimum efficiency scale (MES). We<br />
prefer the MES based on sales to a MES based on median employment in the sector, since<br />
the latter is not able to capture the fixed costs of capital-intensive industries adequately.
Determinants of firm dynamics 69<br />
their sales <strong>and</strong> exp<strong>and</strong> their production capacity to benefit from economies of<br />
scale, or explore new product markets with better prospects. To analyze the<br />
impact of increasing globalization, we include a number of variables capturing<br />
various aspects of international trade.<br />
In theory, the impact of trade on exit could be driven by two aspects: import<br />
competition, through smaller markups (Melitz <strong>and</strong> Ottaviano, 2008) <strong>and</strong><br />
export intensity (Melitz, 2003). The empirical literature so far has mainly<br />
focused on the first aspect, import competition, which is associated with<br />
higher exit <strong>and</strong> switching rates (Bernard et al., 2006a; Coucke <strong>and</strong> Sleuwaegen,<br />
2006; <strong>and</strong> Greenaway et al., 2008). A recent study by Colantone <strong>and</strong><br />
Sleuwaegen (2007) draws attention to the significance of the second aspect of<br />
international competition: export intensity. In this case, the most successful<br />
firms self-select into the export market <strong>and</strong> continue to grow by capturing<br />
new market opportunities abroad. This raises the average efficiency level <strong>and</strong><br />
increases the pressure on factor prices in the home market, thereby crowding<br />
out the least efficient firms. At the same time, firms in sectors with promising<br />
exporting markets will be less likely to switch sectors. To capture both aspects<br />
of increasing international competition, we include imports (Importsjt)<br />
<strong>and</strong> exports (Exportsjt), both defined at the four-digit product level, in our<br />
empirical model 16 . We expect imports to have a positive effect on exit <strong>and</strong><br />
product switching. Exports, however, are expected to have a positive effect<br />
on exit but a negative one on product switching.<br />
Rising intra-industry trade due to falling trade costs raises the number of<br />
products supplied in a market. This, in turn, raises competition <strong>and</strong> cuts<br />
into firms’ markups. Only the more productive firms survive this pressure,<br />
while others are forced out of business. To capture this effect, we include<br />
the Grubel-Lloyd intra-industry trade index (IITjt) in our empirical model,<br />
defined as [1 − (|Xjt − Mjt|/(Xjt + Mjt))]. The effect of IIT on exit <strong>and</strong><br />
switching is however ambiguous since this index captures both the effect of<br />
16 Ideally, we would have preferred to weight imports <strong>and</strong> exports at the product level<br />
by domestic production <strong>and</strong> include measures of import penetration <strong>and</strong> export intensity<br />
in the empirical model. However, no reliable data exist on domestic production by sector<br />
(either at the two-digit industry or four-digit product level) for Estonia. Hence, we include<br />
the value of imports <strong>and</strong> exports in our empirical model.
70 <strong>Globalization</strong> drives strategic product switching<br />
import competition <strong>and</strong> export opportunities.<br />
In product markets where Estonia has a revealed comparative advantage<br />
<strong>and</strong>, therefore, good export opportunities, we expect fewer exits irrespective<br />
of its form (either firm closedown or product switch). The comparative<br />
advantage dummy (CAjt) takes the value of one if exports are larger than<br />
imports for a given four-digit product.<br />
Finally, we want to check how firms’ strategies are related to changes in the<br />
quality embedded in Estonian exports. A substantial amount of theoretical<br />
work predicts that quality systematically affects the direction of international<br />
trade, a finding that is confirmed by some recent empirical papers (eg. Hallak,<br />
2006) 17 . As a measure of product quality, we use a composite index of the unit<br />
value of Estonia’s exports in a given geographic market relative to the unit<br />
value of all exporters in that market (UV Rjt). On the premise that a higher<br />
relative price reflects higher quality than direct competitors’, UV Rjt acts as<br />
a proxy for product quality (Hallak <strong>and</strong> Schott, 2008). However, concerns<br />
remain that this measure could be picking up factors other than quality.<br />
This is especially the case if local monopolies exist <strong>and</strong> competition does not<br />
arbitrage away differences in quality-adjusted prices. To control for this effect<br />
of markups on the unit value of exports, we include the Herfindahl-Hirschman<br />
index of the export market (Herfexjt). This measure of competitiveness is<br />
a composite index of the weighted sum of the Herfindahl-Hirschman indices<br />
for each of the relevant geographical export markets.<br />
Table 2.5 gives a first indication of the differences in characteristics across<br />
the exit strategies by summarizing a number of firm- <strong>and</strong> product-level characteristics<br />
for three groups separately: (1) all firms in the sample; (2) firms<br />
undergoing either an industry switch (two-digit) or product switch (fourdigit);<br />
<strong>and</strong> (3) exiting firms.<br />
17 For a theoretical background on trade <strong>and</strong> quality, see, among others, Falvey <strong>and</strong><br />
Kierzkowski (1987); Flam <strong>and</strong> Helpman (1987); Stokey (1991) <strong>and</strong> Murphy <strong>and</strong> Shleifer<br />
(1997). Empirical papers on this topic include Schott (2004); Schott (2004); Hummels <strong>and</strong><br />
Klenow (2005) <strong>and</strong> Hallak <strong>and</strong> Schott (2008).
Determinants of firm dynamics 71<br />
Industry Product<br />
Variable All switch switch Exit<br />
Number of observations 16,117 1,244 1,912 452<br />
(Percentage of total) (100.0 ) (7.7) (11.9 ) (2.8)<br />
Size 28.1 21.6*** 26.9 20.7***<br />
(Number of employees) (100.4) (73.9) (131.4) (46.2)<br />
Age 6.9 6.1*** 6.3*** 5.7***<br />
(Years) (3.5) (3.3) (3.3) (3.1)<br />
Capital 104.7 119.7** 109.7 91.4<br />
(Thous<strong>and</strong>s of Krooni) (270.8) (307.6) (266.6) (455.6)<br />
Wage 60.3 60.3 58.4* 50.7**<br />
(Thous<strong>and</strong>s of Krooni) (66.6) (71.7) (63.0) (112.9)<br />
TFP 61.2 60.4 58.3** 41.3***<br />
(Thous<strong>and</strong>s of Krooni) (99.6) (85.8) (80.8) (130.2)<br />
Foreign 0.1 0.1 0.1** 0.1**<br />
(Ownership dummy) (0.3) (0.3) (0.3) (0.3)<br />
Sunk 14.7 14.7* 14.8 14.7<br />
(Minimum Efficient Scale) (0.7) (0.7) (0.7) (0.6)<br />
Herf 0.1 0.1*** 0.1*** 0.1**<br />
(HHI domestic market) (0.2) (0.2) (0.2) (0.2)<br />
Imports 417,270 489,774*** 448,386** 337,168***<br />
(Imports, thous<strong>and</strong>s of Krooni) (583,623) (707,494) (644,580) (349,766)<br />
IIT 0.5 0.5*** 0.5*** 0.5**<br />
(Intra-industry trade) (0.3) (0.3) (0.3) (0.2)<br />
CA 0.6 0.5*** 0.5*** 0.6***<br />
(Comparative advantage) (0.5) (0.5) (0.5) (0.5)<br />
Exports 766,759 715,993** 692,268*** 807,780<br />
(Exports, thous<strong>and</strong>s of Krooni) (1,046,466) (1,050,674) (1,019,043) (973,382)<br />
Herfex 0.2 0.2* 0.2 0.2<br />
(HHI export market) (0.1) (0.1) (0.1) (0.1)<br />
UVR 1.3 1.2 1.3 1.3<br />
(Relative unit values) (3.3) (2.7) (4.1) (3.8)<br />
Notes: Reported values are means (st<strong>and</strong>ard deviations), with the exception of the first<br />
row. Significance levels (*** p
72 <strong>Globalization</strong> drives strategic product switching<br />
Compared with continuing firms, enterprises that exit or switch industries<br />
are significantly smaller <strong>and</strong> younger. <strong>Firm</strong>s that switch industries are on<br />
average more capital intensive than continuing firms, while both exiting firms<br />
<strong>and</strong> product switching firms have a significantly lower labor cost <strong>and</strong> productivity.<br />
Turning to product characteristics, we find that industry switchers<br />
tend to be active in industries with a higher level of sunk costs, while the opposite<br />
is true for exiting firms. Switchers also tend to come from sectors with<br />
significantly higher market power, as indicated by the Herfindahl-Hirschman<br />
index for the domestic market. Sectors characterized by higher imports, less<br />
intra-industry trade <strong>and</strong> lower exports display a higher rate of industry <strong>and</strong><br />
product switching than other sectors. Prior to switching, these firms tend<br />
to be in sectors with relatively lower revealed comparative advantage. Enterprises<br />
that permanently exit the market, however, are active in sectors<br />
with lower imports <strong>and</strong> higher exports. They also tend to be more present<br />
in industries with revealed comparative advantage than do continuing firms.<br />
2.4 Results<br />
2.4.1 Baseline results<br />
In this section we report the results from a multinomial logit regression in<br />
which we analyze the determinants of three alternative strategies at the firm<br />
level: (1) stay active in the same product market (the baseline category);<br />
(2) change its main product line or (3) exit entirely from the market. Table<br />
6 reports the coefficients <strong>and</strong> st<strong>and</strong>ard errors as well as the marginal effects<br />
of the variables on the probability of exit (whether in the form of a product<br />
switch or a true exit). These marginal effects are calculated at the mean<br />
of the independent variables, to provide some guidance on the magnitude<br />
of the effect (reported in italics in the tables). We pool the observations<br />
across years for all firms in the sample, <strong>and</strong> we include year <strong>and</strong> two-digit<br />
industry fixed effects to control for aggregate variation in industry dynamics.<br />
St<strong>and</strong>ard errors are clustered at the firm level.<br />
The results on the firm characteristics confirm our priors on firm dynamics<br />
as discussed in Section 2.3. Controlling for size at the firm level, exiting firms
Results 73<br />
are on average younger, less productive <strong>and</strong> have lower capital intensity 18 .<br />
These results suggest that the performance of these firms is not good enough<br />
to keep up with the dynamics in the market. The probability of product<br />
switching, on the other h<strong>and</strong>, is significantly decreasing with plant size <strong>and</strong><br />
age, while significantly increasing with firm productivity <strong>and</strong> capital intensity.<br />
In other words, smaller <strong>and</strong> younger firms are more likely to change their<br />
main product line than their older <strong>and</strong> larger counterparts. Moreover, only<br />
the more productive <strong>and</strong> capital-intensive firms switch to different product<br />
markets. This is a first indication that switches are not necessarily driven by<br />
a lack of competitiveness but are rather the outcome of firms’ own choices.<br />
Controlling for the other firm-level characteristics discussed above, foreign<br />
ownership of the firm has no significant impact on either product switching<br />
behavior or exit. A possible explanation for this can be found in the motives<br />
of foreign firms to invest in central <strong>and</strong> eastern European countries. Although<br />
cost advantage plays a role, various studies illustrate the importance of high<br />
market potential as an incentive for foreign firms to enter these markets 19 .<br />
This suggests that these investments are partly strategic <strong>and</strong> forward looking<br />
<strong>and</strong> that foreign firms will not necessarily exit the market more rapidly than<br />
domestic firms when faced with short-run adverse shocks.<br />
With respect to the conditions in the domestic market, the sunk cost variable<br />
is never significant, whereas the coefficient on the Herfindahl-Hirschman<br />
index only has a positive <strong>and</strong> significant sign for product switching. As noted<br />
by Caves (1998), concentration <strong>and</strong> sunk costs of a particular industry are simultaneously<br />
determined, since the forces of exit <strong>and</strong> entry will influence the<br />
equilibrium number of firms in an industry. Hence, including both variables<br />
together as independent variables may be the reason for the insignificant coefficients<br />
on sunk costs. As a robustness check, we ran the empirical model<br />
including either only the Herfindahl-Hirschman index or only the sunk cost<br />
measure in our basic specification, but the results are equivalent to those<br />
18 The marginal probabilities for exit are quite small in magnitude because the likelihood<br />
of exit in any period is relatively small.<br />
19 See, for instance Bevan <strong>and</strong> Estrin (2004) <strong>and</strong> Carstensen <strong>and</strong> Toubal (2004).
74 <strong>Globalization</strong> drives strategic product switching<br />
Variables Product switch Exit<br />
log(Employment) -0.143*** -0.014 -0.036 0.000<br />
[0.026] [0.054]<br />
log(Age) -0.178*** -0.017 -0.300*** -0.005<br />
[0.044] [0.072]<br />
log(TFP) 0.092** 0.010 -0.367*** -0.007<br />
[0.041] [0.066]<br />
log(Capital) 0.051** 0.005 -0.220*** -0.004<br />
[0.020] [0.036]<br />
log(Wages) -0.080 -0.008 0.169 0.003<br />
[0.049] [0.108]<br />
Foreign (d) -0.088 -0.008 0.044 0.001<br />
[0.094] [0.185]<br />
Herfindahl domestic market 0.253* 0.025 -0.230 -0.004<br />
[0.144] [0.334]<br />
Sunk 0.008 0.000 0.139 0.002<br />
[0.057] [0.093]<br />
log(Exports) -0.064* -0.006 0.063 0.001<br />
[0.038] [0.068]<br />
Herfindahl export market 0.007 0.002 -0.739 -0.013<br />
[0.386] [0.679]<br />
log(UVR) -0.102* -0.010 0.07 0.001<br />
[0.055] [0.105]<br />
log(Imports) 0.014 0.001 -0.011 0.000<br />
[0.049] [0.091]<br />
IIT -0.212* -0.020 -0.135 -0.002<br />
[0.128] [0.235]<br />
CA (d) -0.218** -0.021 -0.011 0.000<br />
[0.106] [0.207]<br />
Number of observations: 16,043<br />
Pseudo R-square: 0.054<br />
Notes: This table reports the results from a multinomial logit (0=continuing;<br />
1=switching products; 2=closing). Robust st<strong>and</strong>ard errors are in brackets<br />
below coefficient estimates. The numbers in italics next to the coefficient<br />
estimates represent the marginal probability change at the mean of the independent<br />
variable or the discrete change of a dummy variable (d) from 0 to<br />
1. Though not reported, all regressions include a constant, two-digit industry,<br />
<strong>and</strong> time fixed effects. St<strong>and</strong>ard errors are clustered at the firm level.<br />
Significance levels: *** p
Results 75<br />
shown in Table 2.6 20 . The positive coefficient on the Herfindahl-Hirschman<br />
index for product switching tells us that firms are more likely to switch away<br />
from the more concentrated industries. A marginal increase in average sector<br />
concentration pushes the probability of switching up with 2.5 percentage<br />
points. This result suggests that firms confronted with the aggressive behavior<br />
of their rivals may be persuaded to exploit opportunities in a different<br />
product market.<br />
Contrary to our initial expectations we do not find that international competition<br />
is driving firm exit in Estonia. Conversely, switching is strongly<br />
influenced by the product market evolution, both in the domestic <strong>and</strong> international<br />
scene. As a rule, Estonian firms active in a sector with more exports<br />
should be able to compete in global markets. Export-intensive sectors are<br />
therefore more dynamic <strong>and</strong> promising for companies. Unfortunately, the<br />
data do not allow us to distinguish between exporters <strong>and</strong> nonexporters at<br />
the firm level. However, following Melitz’s (2003) arguments, developments<br />
in the export market will have repercussions on domestic market structure.<br />
In this spirit, we expect enterprises to react to changes in global markets to<br />
remain viable in an increasingly competitive environment. This idea is reflected<br />
in all export variables listed in Table 2.6. The comparative advantage<br />
variable indicates that firms have a 2.1 percentage point lower probability<br />
to switch away from products in which Estonia has a revealed comparative<br />
advantage, that is, products for which exports are larger than imports. On<br />
the other h<strong>and</strong>, the total value of exports decreases the likelihood of product<br />
switching but does not have a significant impact on the likelihood of exit.<br />
This is in contrast with Melitz’s argument that the least productive firms,<br />
confronted with international competition (i.e. exports), will be forced to<br />
close down.<br />
A similar idea is reflected in the result on intra-industry trade for product<br />
switching. Higher rates of intra-industry trade denote that Estonia is<br />
simultaneously exporting <strong>and</strong> importing a particular product, <strong>and</strong> thus that<br />
the number of varieties supplied in that product market is higher. This in-<br />
20 Results are not reported here for brevity but can be obtained from the authors upon<br />
request.
76 <strong>Globalization</strong> drives strategic product switching<br />
crease in intra-industry trade could either represent stiffer competition from<br />
imports or better export opportunities. Whereas inefficient firms could be<br />
driven into bankruptcy due to this increased pressure, the more productive<br />
firms may take advantage of these opportunities. Indeed, controlling for firm<br />
<strong>and</strong> domestic market characteristics, firms are less likely to leave industries<br />
in which export opportunities abound. A rise of intra-industry trade leads<br />
to a decrease in the probability of product switching. It does not generate,<br />
however, a destructive competition effect on domestic firms, as reflected in<br />
the insignificant coefficient for exit. Turning to the unit value variable, we<br />
find that, to the extent that price differences reflect differences in quality,<br />
there is a negative relation between the quality of Estonia’s exports <strong>and</strong> the<br />
probability of switching. In other words, the lower the quality of the products,<br />
the more likely firms will change their products. Thus, firms tend to<br />
exit low-quality exporting sectors.<br />
Existing empirical work, in particular the paper of Bernard et al. (2006a),<br />
focuses on the impact of imports coming from low-wage countries. The authors<br />
provide evidence, using data on the U.S. manufacturing sector, that<br />
a higher degree of import penetration from low-wage countries is associated<br />
with a higher probability of exiting <strong>and</strong> switching products. However, in the<br />
case of Estonia, the origin of imports does not seem to play a role in the<br />
strategic decisions taken by firms 21 .<br />
Overall, we notice that the determinants of product switching are very different<br />
from the determinants of firm death. On the one h<strong>and</strong>, the insignificant<br />
coefficients for all domestic <strong>and</strong> international product market characteristics<br />
for firm exit show that bankruptcy in Estonia is entirely driven by the firm’s<br />
own behavior, rather than by a reaction to external factors. On the other<br />
h<strong>and</strong>, the results for our trade variables suggest that Estonian firms are exploiting<br />
opportunities in global markets. These findings imply that product<br />
switching might be more than just an alternative way of escaping from increasing<br />
competition. Both firm <strong>and</strong> industry characteristics point toward<br />
an active policy of looking for new <strong>and</strong> better opportunities. Rather than<br />
21 Results are not reported here for brevity, but can be obtained from the authors upon<br />
request.
Results 77<br />
switch products out of defense, firms seem to be changing their product lines<br />
out of choice. In the following sections, we will investigate this hypothesis in<br />
more detail.<br />
2.4.2 Self-selection into new markets<br />
Our results suggest that firms systematically self-select into new product<br />
markets on the basis of their performance in <strong>and</strong> knowledge about the<br />
market. This result is a first new insight in firm dynamics <strong>and</strong> an important<br />
contribution to the existing empirical literature which restricted its attention<br />
to the import side of international competition. Whereas these studies find<br />
that firms in industrial countries change their product lines out of defense<br />
against low-cost competition, we find that Estonian firms follow an offensive<br />
strategy by actively exploring the market. In this subsection we want to<br />
dig deeper into this new facet of product switching. We start by exploring<br />
potential differences between product switching <strong>and</strong> industry switching. Afterward,<br />
we look more closely at the characteristics of the product markets or<br />
industries to which firms switch. In particular, we investigate the differences<br />
in export unit value ratio <strong>and</strong> technology intensity between the origin <strong>and</strong><br />
destination industry.<br />
Industry versus product switching<br />
In order to compare the determinants of industry (two-digit) <strong>and</strong> product<br />
(four-digit) switches, we split the group of product switches into (1) product<br />
switches that are not observed at the two-digit level, that is, firms that<br />
change their main four-digit product line but stay within the same two-digit<br />
industry; <strong>and</strong> (2) industry switchers, that is, those firms that switch to a<br />
new product line in a different two-digit industry. Using this distinction,<br />
we estimate the same multinomial logit model as before, except that the<br />
dependent variable now takes on four different values (rather than three):<br />
0 for firms that stay in the same product market, 1 for firms that switch<br />
products but not industries, 2 for firms that change two-digit industries, <strong>and</strong><br />
3 for exit.
78 <strong>Globalization</strong> drives strategic product switching<br />
Variable Product switch Industry switch Exit<br />
log(Employment) 0.070* 0.003 -0.258*** -0.016 -0.036 0.000<br />
[0.040] [0.033] [0.054]<br />
log(Age) -0.201*** -0.006 -0.166*** -0.010 -0.300*** -0.005<br />
[0.071] [0.053] [0.072]<br />
log(TFP) 0.058 0.002 0.104** 0.007 -0.368*** -0.007<br />
[0.063] [0.050] [0.066]<br />
log(Capital) 0.01 0.000 0.070*** 0.005 -0.220*** -0.004<br />
[0.035] [0.024] [0.036]<br />
log(Wages) -0.091 -0.003 -0.069 -0.004 0.169 0.003<br />
[0.078] [0.059] [0.108]<br />
Foreign (d) -0.169 -0.005 -0.045 -0.003 0.045 0.001<br />
[0.151] [0.117] [0.185]<br />
Herfindahl domestic market 0.194 0.005 0.275* 0.017 -0.229 -0.004<br />
[0.251] [0.165] [0.335]<br />
Sunk costs 0.003 0.000 0.003 0.000 0.14 0.002<br />
[0.109] [0.062] [0.094]<br />
log(Exports) -0.156*** -0.005 0.002 0.000 0.064 0.001<br />
[0.058] [0.046] [0.068]<br />
Herfindahl export market -0.738 -0.023 0.426 0.029 -0.743 -0.013<br />
[0.697] [0.436] [0.679]<br />
log(UVR) -0.277** -0.008 -0.009 0.000 0.071 0.001<br />
[0.111] [0.060] [0.105]<br />
log(Imports) -0.026 -0.001 0.042 0.003 -0.011 0.000<br />
[0.074] [0.060] [0.091]<br />
IIT -0.149 -0.004 -0.262 -0.016 -0.136 -0.002<br />
[0.191] [0.160] [0.236]<br />
CA (d) -0.268* -0.008 -0.209 -0.013 -0.011 0.000<br />
[0.162] [0.129] [0.208]<br />
Number of observations 16,043<br />
Pseudo R-square 0.059<br />
Notes: This table reports the results from a multinomial logit regression (0=continuing;<br />
1=product switching within the same industry; 2=industry switching; 3=closing). Robust<br />
st<strong>and</strong>ard errors are in brackets below coefficient estimates. The numbers in italics next<br />
to the coefficient estimates represent the marginal probability change at the mean of the<br />
independent variable or the discrete change of a dummy variable (d) from 0 to 1. Though<br />
not reported, all regressions include a constant, two-digit industry, <strong>and</strong> time fixed effects.<br />
St<strong>and</strong>ard errors are clustered at the firm level. Significance levels: *** p
Results 79<br />
Table 2.7 shows a striking difference between the determinants of industry<br />
switching <strong>and</strong> those of product switching. International trade does not seem<br />
to play a significant role in the dynamics behind industry switching. <strong>Firm</strong>s<br />
change to a different two-digit industry in response to changes in their own<br />
performance <strong>and</strong> the domestic market, but not on account of international<br />
trade aspects. We also find that, while more productive <strong>and</strong> more capitalintensive<br />
firms are more likely to switch industries, the effect of these variables<br />
on product switches is insignificant. This implies that the results obtained<br />
earlier on these variables are driven by industry switches rather than product<br />
switches. The insignificant effect of the trade variables on industry switches,<br />
along with the negative link between exports <strong>and</strong> unit values, on the one<br />
h<strong>and</strong>, <strong>and</strong> product switching, on the other h<strong>and</strong>, further suggests that the<br />
switching pattern in Estonia is determined by product differentiation. Enterprises<br />
observe changes within their sector <strong>and</strong> respond by modifying their<br />
products to take part in the growth process.<br />
As was already noted in section 2.2, firms that switch industries towards<br />
the services sector are retained in the sample until their last year of activity<br />
in the manufacturing sector. This allows us to delve deeper into the<br />
determinants of industry switching by comparing firms that switch within<br />
the manufacturing sector to firms that switch towards services. In our sample,<br />
7 percent of the 1,244 industry switches are to the primary sector, 35<br />
percent stay within the manufacturing sector, <strong>and</strong> 58 percent of industry<br />
switches are to the services sector. Each of these dynamics is likely to be<br />
driven by diverging underlying causes. Given the growing importance of the<br />
services sector in Estonia’s domestic economy, the switches to services are of<br />
particular interest to us.<br />
We therefore define a new dependent variable that equals zero for twodigit<br />
switches within manufacturing <strong>and</strong> 1 for switches to services, <strong>and</strong> run<br />
a logit regression (Table 2.8). In this case, we do not compare stayers with<br />
switchers, but instead explore dissimilarities among the switchers. Similar<br />
to our previous regressions, we include time <strong>and</strong> two-digit industry fixed<br />
effects, <strong>and</strong> cluster the st<strong>and</strong>ard errors at the firm level. This regression<br />
reveals that larger firms <strong>and</strong> foreign firms are more likely to switch within
80 <strong>Globalization</strong> drives strategic product switching<br />
Variable Coefficient Marginal effect<br />
log(Employment) -0.461*** -0.111<br />
[0.063]<br />
log(Age) 0.072 0.017<br />
[0.103]<br />
log(TFP) 0.182** 0.044<br />
[0.083]<br />
log(Capital) -0.021 -0.005<br />
[0.045]<br />
log(Wages) 0.135 0.033<br />
[0.107]<br />
Foreign (d) -0.401* -0.098<br />
[0.208]<br />
Herfindahl domestic market 0.244 0.059<br />
[0.312]<br />
Sunk 0.133 0.032<br />
[0.118]<br />
log(Exports) 0.058 0.014<br />
[0.096]<br />
Herfindahl export market 2.380** 0.572<br />
[1.026]<br />
log(UVR) -0.009 -0.002<br />
[0.124]<br />
log(Imports) -0.221* -0.053<br />
[0.123]<br />
IIT -0.173 -0.042<br />
[0.308]<br />
CA (d) 0.312 0.075<br />
[0.266]<br />
Number of observations: 1,244<br />
Pseudo R-square: 0.059<br />
Notes: Results reported are from a logit estimation comparing (two-digit) industry<br />
switches to other manufacturing sectors (dependent variable equal to 0) with<br />
industry switches to services (dependent variable equals 1). Robust st<strong>and</strong>ard errors<br />
are in brackets below coefficient estimates, the numbers in italics next to the<br />
coefficient estimates represent the marginal probability change at the mean of<br />
the independent variable or the discrete change of a dummy variable (d) from 0<br />
to 1. Though not reported, all regressions include a constant, two-digit industry<br />
dummies, <strong>and</strong> time dummies. St<strong>and</strong>ard errors are clustered at the firm level.<br />
Significance levels: *** p
Results 81<br />
the manufacturing sector. Conversely, the more productive firms tend to<br />
leave the manufacturing sector <strong>and</strong> enter the services sector. A closer look<br />
at the destinations reveals that the majority of the firms enter the wholesale<br />
trade <strong>and</strong> retail trade sectors 22 .<br />
Turning to the trade variables, we see that a higher concentration in exporting<br />
markets drives firms into the services sector while import growth is<br />
associated with industry switches within manufacturing. Further data analysis<br />
reveals that the increase in switching to services is entirely driven by<br />
switches to the distribution sector 23 . These results suggest that firms tend<br />
to move from the production side to the distribution sector if the export market<br />
is highly concentrated. This does not necessarily mean, however, that a<br />
firm stops producing the product. For example, an enterprise that realizes<br />
the potential in the market for its products, could continue production but<br />
shift its attention toward the distribution of its products. Unfortunately, our<br />
data set provides only firms’ main sector of activity, <strong>and</strong> not the global picture<br />
of their activities. We also find that, if imports generate strong pressures<br />
there is less incentive for the firm to move into the distribution sector, given<br />
the lack of competitiveness of its product. To minimize the loss of sunk investments,<br />
it is more efficient to enter into a comparable market where firms<br />
can continue employing their physical <strong>and</strong> human capital without much additional<br />
cost or effort.<br />
Hence, Estonian firms that are switching in response to changes in the<br />
international trade environment either move into the distribution of goods<br />
(some of which they may have produced in the past) or switch toward other<br />
manufacturing industries. Which switching strategy they adopt depends<br />
on whether the changes in the global environment are manifested through<br />
imports or rather driven by the concentration in export markets. These<br />
results suggest that Estonian companies are aware of the prospects in the<br />
global market <strong>and</strong> are trying to exploit these opportunities by proactively<br />
changing their business plans.<br />
22 Out of the 727 industry switches to the services sector observed in the sample, about<br />
54 percent goes to the retail <strong>and</strong> wholesale sectors (NACE codes 50, 51, <strong>and</strong> 52).<br />
23 If we exclude the wholesale <strong>and</strong> retail sectors from the analysis, no significant results<br />
are obtained for the export variables.
82 <strong>Globalization</strong> drives strategic product switching<br />
Quality upgrading versus technology upgrading<br />
In our baseline results, we observed that Estonian companies tend to leave<br />
low-quality exporting sectors (as measured by the relative export unit value).<br />
The question now is to which sectors these firms are moving. Do they switch<br />
to products with an even lower relative unit value because they are not<br />
able to compete within the price quality range of the export market, or do<br />
they switch to sectors with a higher relative unit value because they see<br />
opportunities at the higher end of the quality array? Fabrizio et al. (2007)<br />
document an impressive shift in product quality <strong>and</strong> technology intensity of<br />
exports for Estonia <strong>and</strong> other central <strong>and</strong> eastern European countries over<br />
the past decade. With these facts in our mind, we now explore the direction<br />
of switches <strong>and</strong> accompanying firm characteristics in detail.<br />
Of the total number of product switches within manufacturing, 483 switches<br />
occur to markets with a lower relative export unit value, whereas 429 switches<br />
go in the opposite direction. In other words, almost half of the product<br />
switches result in quality upgrading. In the first column of Table 2.9, we<br />
check the firm characteristics behind this shift up the quality ladder. To<br />
do so, we calculate for each firm that switches within manufacturing the<br />
log difference in the export unit value ratio between its origin <strong>and</strong> destination<br />
industry, at the four-digit level. More specifically, a positive value for<br />
the log difference st<strong>and</strong>s for a switch to a product with a higher unit value,<br />
<strong>and</strong> thus of higher quality. These log differences in export unit value ratios<br />
are then regressed on a number of firm characteristics using ordinary least<br />
squares (OLS), while controlling for market power in the export market, as<br />
well as for industry <strong>and</strong> time fixed effects. The results indicate a positive link<br />
between a firm’s capital intensity <strong>and</strong> quality upgrading. Among the firms<br />
that alter their product line, only the more capital-intensive firms are able<br />
to move into higher-quality product markets. Controlling for other characteristics,<br />
these firms tend to be smaller than the average Estonian switching<br />
firm.<br />
To explore whether this quality upgrading is related to technology upgrading,<br />
we define two dummies to capture the change in technological intensity.<br />
The dummy Technology upgrading equals 1 if a firm moves towards a sector
Results 83<br />
Variables (1) (2) (3)<br />
log(Employment) -0.032* -0.027 0.005<br />
[0.018] [0.018] [0.020]<br />
log(Age) 0.028 0.024 0.001<br />
[0.031] [0.031] [0.034]<br />
log(TFP) 0.052 0.052 -0.044<br />
[0.042] [0.043] [0.029]<br />
log(Capital) 0.048*** 0.048*** 0.043**<br />
[0.018] [0.018] [0.019]<br />
log(Wages) -0.017 -0.025 0.016<br />
[0.046] [0.047] [0.044]<br />
Foreign (d) 0.022 0.007 0.022<br />
[0.070] [0.070] [0.069]<br />
Herfindahl export market -0.453 -0.473 -0.625**<br />
[0.33] [0.330] [0.250]<br />
Technology upgrading ... 0.085 ...<br />
[0.094]<br />
Technology downgrading ... -0.186** ...<br />
[0.079]<br />
Destination High tech ... ... -0.119<br />
[0.150]<br />
Destination Medium-high-tech ... ... 0.202**<br />
[0.091]<br />
Destination Medium-low-tech ... ... 0.069<br />
[0.049]<br />
Industry fixed effects (two-digit) Yes Yes No<br />
Year fixed effects Yes Yes Yes<br />
Number of observations: 1,097 1,097 1,097<br />
R-square 0.097 0.103 0.036<br />
Notes: The dependent variable is the log difference in export unit<br />
value ratio between the origin industry <strong>and</strong> destination industry, at<br />
the four-digit level. The dummy technology up- (down-) grading<br />
equals 1 if the firm moves up (down) one category of technology intensity.<br />
The dummies H-tech, MH-tech, <strong>and</strong> ML-tech destination equal<br />
1 if a firm moves to respectively a high-tech, medium-high-tech or<br />
medium-low-tech sector. The regressions are estimated using OLS,<br />
<strong>and</strong> robust st<strong>and</strong>ard errors are in brackets below the coefficient estimates.<br />
Coefficients for the constant <strong>and</strong> industry <strong>and</strong> year dummies<br />
are suppressed. St<strong>and</strong>ard errors are clustered at the firm level. Significance<br />
levels: *** p
84 <strong>Globalization</strong> drives strategic product switching<br />
with a higher intensity of technology (this is the case for 97 observations),<br />
whereas the dummy Technology downgrading equals 1 if a firm moves down<br />
the technology ladder (123 observations). As can be seen in the second column<br />
of Table 2.9, technology downgrading is negatively related to the log<br />
difference in export unit value ratio, whereas the coefficient on technology<br />
upgrading is positive but statistically insignificant. These results suggest<br />
that quality upgrading is not necessarily related to technology upgrading.<br />
Yet this finding is not completely unexpected, as the majority of the product<br />
changes happen along the same level of technology (877 observations).<br />
To underst<strong>and</strong> at which technological level this quality upgrading is taking<br />
place, we return to our original specification, used in column 1 of Table 2.9,<br />
while adding three dummies identifying the technological intensity of the<br />
industry of destination for product switches. The results in column 3 of<br />
Table 2.9 should be interpreted relative to the base category -the low tech<br />
industry. The positive <strong>and</strong> significant coefficient on the medium-high-tech<br />
industry dummy shows that Estonian companies are moving up the quality<br />
ladder mainly within the medium-high-tech sectors.<br />
2.5 Robustness checks<br />
2.5.1 Results by size class<br />
Because of data constraints, the empirical literature so far has been able to<br />
look only at the switching behavior of relatively large firms. The U.S. Longitudinal<br />
Research Database used by Bernard et al. (2006a) includes only<br />
firms with a minimum of 10 employees, while Greenaway et al. (2008) study<br />
Swedish manufacturing firms with at least 50 employees. An important advantage<br />
of our data is the absence of any size thresholds. This allows us<br />
to draw conclusions for the entire population of manufacturing firms in Estonia<br />
<strong>and</strong> also reveals important insights in the dynamics of the smallest<br />
among them, namely, microenterprises employing fewer than 10 employees.<br />
This feature is particularly important for a transition country: while a small<br />
number of large enterprises dominated the economic l<strong>and</strong>scape of Estonia<br />
during the Soviet era, the transition period has been characterized by the
Robustness checks 85<br />
1−9 10−19<br />
20−49 50−99<br />
100−249 250−499<br />
500 or more<br />
Figure 2.1: Sample size distribution<br />
emergence of many small <strong>and</strong> medium-sized enterprises. Masso et al. (2004)<br />
compare the average size of Estonian enterprises to the Organization for Economic<br />
Cooperation <strong>and</strong> Development (OECD) average <strong>and</strong> find that, while<br />
the mean size is very close to this average, the st<strong>and</strong>ard deviation is much<br />
smaller due to the small number of very large enterprises in Estonia. In fact,<br />
about 50 percent of manufacturing firms in our sample are microenterprises<br />
(Figure 1), <strong>and</strong> more than three fourths of the firms employ fewer than 50<br />
employees. These microbusinesses are also more dynamic (Table 2.1, Panel<br />
B). Compared with the sample average (Table 2.1, Panel A), we notice that<br />
microbusinesses modify their product lines more frequently (13.4 percent in<br />
the micro sample versus 11.9 percent in the full sample) <strong>and</strong> have a higher<br />
exit rate than the average Estonian firm (3.3 percent versus 2.8 percent).<br />
To analyze differences in determinants across size categories, we run our<br />
baseline specification for small <strong>and</strong> big firms separately. Table 2.10 shows<br />
the results for firms with fewer than 10 employees (product switching in<br />
column 1 <strong>and</strong> exiting in column 2) versus the rest of the sample (columns 3<br />
<strong>and</strong> 4). The results for the firm-level characteristics are very similar to our<br />
baseline results in Table 2.6, except for labor cost which turns out to be a
86 <strong>Globalization</strong> drives strategic product switching<br />
Table 2.10: Determinants of firm dynamics across size categories<br />
Notes: This table reports the results from a multinomial logit regression (0=continuing; 1=switching products; 2=closing), for two groups: firms with fewer<br />
than 10 employees (columns 1 <strong>and</strong> 2) <strong>and</strong> firms with 10 employees or more (columns 3 <strong>and</strong> 4). Robust st<strong>and</strong>ard errors are in brackets below coefficient<br />
estimates. The numbers in italics next to the coefficient estimates represent the marginal probability change at the mean of the independent variable or<br />
the discrete change of a dummy variable (d) from 0 to 1. Though not reported, all regressions include a constant, two-digit industry <strong>and</strong> time fixed effects.<br />
St<strong>and</strong>ard errors are clustered at the firm level. Significance levels: *** p
Robustness checks 87<br />
significant determinant of both product switching <strong>and</strong> exiting for firms with<br />
10 employees or more. In particular, larger firms with relatively high wages<br />
are, on the one h<strong>and</strong>, more likely to go bankrupt <strong>and</strong>, on the other h<strong>and</strong>,<br />
less likely to change their product line. Remember from Section 2.3 that<br />
high labor cost can either reflect inefficient use of labor or high skill intensity<br />
of the work force. If the labor costs of a firm are too high compared to its<br />
competitors, for a given level of productivity, the firm will not be able to<br />
compete in the market <strong>and</strong> will face bankruptcy. For firms with a large pool<br />
of employees the efficient use of its labor force becomes more crucial than<br />
for small firms where labor costs are not a determinant of its strategies. Yet,<br />
to the extent that higher wages reflect higher skill intensities, investments<br />
in product-specific human capital become a sunk cost acting as a barrier to<br />
change to a different product market. These sunk costs are substantially<br />
lower for firms with few employees since fewer employees will have to be<br />
re-trained for the production of the new products. This is reflected in the<br />
insignificance of the coefficient on labor cost for the smallest firms, whereas<br />
a marginal change in this variable implies a decrease in the probability of<br />
product switching of 2.4 percentage points for larger firms.<br />
Another major difference between small <strong>and</strong> larger firms is the effect of<br />
international openness on their switching behavior. Product switching among<br />
microenterprises does not seem to be driven by the level of exports per se,<br />
but is fairly sensitive to changes in the relative unit values of Estonia’s export<br />
products, with an elasticity of around 0.13. The opposite is true for product<br />
switching among larger firms, which seems to be affected only by the level of<br />
exports.<br />
2.5.2 Results by time period<br />
Now we consider whether the determinants behind firm dynamics have<br />
changed over time by splitting the sample into two periods, 1997-2000 <strong>and</strong><br />
2001-04 (Table 2.11). Several interesting insights come out of this exercise.<br />
The firm determinants of product switching are roughly alike across both<br />
periods. Remarkably, trade emerges as an important driver of the productswitching<br />
behavior of firms in the first half of the sample period (1997-2000),<br />
while acting solely as a driver of firm death in the second half of the pe-
88 <strong>Globalization</strong> drives strategic product switching<br />
riod. This seems to reflect a Melitz-type effect: as the quality of Estonia’s<br />
exported products increases relative to its competitors in a certain sector,<br />
the most successful firms will probably self-select into the export market.<br />
Over time, this raises the average efficiency level in the sector <strong>and</strong> firms that<br />
cannot cope with the pace of quality upgrading are forced to exit. On the<br />
other h<strong>and</strong>, globalization is no longer encouraging firms to switch products<br />
or industries in the most recent period. Changes in a firm’s product mix<br />
are rather driven by evolutions in the domestic market, as reflected by the<br />
Herfindahl-Hirschman index.<br />
From a pessimistic point of view, one could argue that the driving forces<br />
behind the self-selection process of moving into more promising markets have<br />
tapered off in Estonia. Whereas Estonia significantly exp<strong>and</strong>ed its market<br />
share in the 1990s, this process has decelerated since the beginning of the new<br />
century. This could indicate that some firms might be losing market share<br />
because they cannot withst<strong>and</strong> competition <strong>and</strong> are unable to proactively<br />
search for better opportunities. However, from a positive point of view, this<br />
finding could also be interpreted as a sign that Estonia has reached a steady<br />
state after the wide-ranging restructuring process of the 1990s. Switching<br />
<strong>and</strong> exit rates were substantial at the end of the 1990s before gradually<br />
declining toward the end of the sample period in 2004. Estonia’s transition<br />
from a planned economic system was accompanied by extensive privatization<br />
<strong>and</strong> restructuring, in conjunction with the dismantling of trade barriers <strong>and</strong><br />
the inflow of foreign investment. This catching-up process has now slowed,<br />
<strong>and</strong> changes in the market are mainly serving to keep the industrial sectors<br />
healthy by forcing the least efficient firms to exit.<br />
2.6 Conclusions<br />
This paper provides new evidence on the link between globalization <strong>and</strong><br />
firm dynamics, focusing on the case of Estonia. We contribute to the literature<br />
in two important respects. First, this is the first paper to study the<br />
determinants of exit <strong>and</strong> product switching in an emerging market. Second,<br />
we consider explicitly the effect of export market conditions on firm<br />
dynamics. For that purpose, we include three product-level measures in our
Conclusions 89<br />
1997 - 2000 (N = 6,090) 2001 - 2004 (N = 9,953)<br />
Product switch (N = 979)) Exit (N = 262) Product switch (N = 933)) Exit (N = 190)<br />
log(Employment) -0.130*** -0.017 0.007 0.001 -0.122*** -0.010 -0.032 0.000<br />
[0.035] [0.069] [0.038] [0.080]<br />
log(Age) -0.167** -0.019 -0.471*** -0.013 -0.217*** -0.017 -0.152 -0.001<br />
[0.065] [0.100] [0.055] [0.096]<br />
log(TFP) 0.114** 0.016 -0.308*** -0.009 0.092* 0.008 -0.385*** -0.004<br />
[0.058] [0.102] [0.055] [0.083]<br />
log(Capital) 0.057** 0.008 -0.180*** -0.005 0.040 0.003 -0.271*** -0.003<br />
[0.029] [0.053] [0.027] [0.046]<br />
log(Wage) -0.072 -0.010 0.123 0.004 -0.179*** -0.014 -0.014 0.000<br />
[0.068] [0.156] [0.066] [0.129]<br />
Foreign (d) -0.196 -0.023 -0.192 -0.004 0.019 0.001 0.342 0.004<br />
[0.126] [0.255] [0.126] [0.264]<br />
Herfindahl domestic market -0.132 -0.014 -0.651 -0.018 0.724*** 0.057 0.622 0.006<br />
[0.195] [0.415] [0.210] [0.560]<br />
Sunk 0.108 0.014 0.110 0.003 -0.089 -0.007 0.166 0.002<br />
[0.074] [0.126] [0.080] [0.159]<br />
log(Exports) -0.085* -0.012 0.114 0.004 -0.031 -0.002 0.003 0.000<br />
[0.048] [0.090] [0.058] [0.120]<br />
Herfindahl export market -0.397 -0.053 0.363 0.012 0.595 0.049 -1.918 -0.021<br />
[0.557] [0.856] [0.547] [1.185]<br />
log(UVR) -0.156* -0.019 -0.303* -0.008 -0.071 -0.006 0.241** 0.003<br />
[0.085] [0.169] [0.072] [0.116]<br />
log(Imports) -0.058 -0.007 -0.010 0.000 0.038 0.003 0.000 0.000<br />
[0.065] [0.127] [0.074] [0.147]<br />
IIT -0.233 -0.029 -0.227 -0.006 -0.108 -0.009 0.112 0.001<br />
[0.171] [0.324] [0.191] [0.401]<br />
CA (d) -0.310** -0.039 -0.220 -0.005 -0.216 -0.018 0.100 0.001<br />
[0.155] [0.287] [0.144] [0.322]<br />
Pseudo R-square 0.039 0.035<br />
Notes: This table reports the results from a multinomial logit regression (0=continuing; 1=switching products; 2=closing), for two periods: 1997-2000<br />
(columns 1 <strong>and</strong> 2) <strong>and</strong> 2001-04 (columns 3 <strong>and</strong> 4). Robust st<strong>and</strong>ard errors are in brackets below coefficient estimates. The numbers in italics next to the<br />
coefficient estimates represent the marginal probability change at the mean of the independent variable or the discrete change of a dummy variable (d)<br />
from 0 to 1. Though not reported, all regressions include a constant, <strong>and</strong> two-digit industry fixed effects. St<strong>and</strong>ard errors are clustered at the firm level.<br />
Significance levels: *** p
90 <strong>Globalization</strong> drives strategic product switching<br />
estimation: (1) the value of exports; (2) the degree of competition in export<br />
markets; <strong>and</strong> (3) the quality of exports relative to direct competitors.<br />
Our results indicate that globalization is generally not an important driver<br />
of firm exit, while it emerges as an important factor explaining product<br />
switching. What matters for exit are firm characteristics: younger firms<br />
<strong>and</strong> those with lower productivity <strong>and</strong> capital intensity are more likely to<br />
exit. Meanwhile, product switching is also affected by conditions in export<br />
markets. In particular, firms are more likely to switch if they are in sectors<br />
with relatively small revealed comparative advantage, <strong>and</strong> where the total<br />
value <strong>and</strong> quality of exports are relatively low. However, this effect only<br />
matters for product switches <strong>and</strong> for the early sample period when trade flows<br />
were increasing rapidly. A possible interpretation of our results is that firms<br />
initially moved out of products for which prospects in the export markets,<br />
which were increasingly opening up, were not good. This result is in contrast<br />
with previous studies on industrial countries which have found that firms<br />
change their product line as a defensive strategy against low-cost imports.<br />
Finally, we find that firms switching to relatively higher quality products are<br />
more capital intensive; however, these switches are not related to technology<br />
upgrading.<br />
Our findings raise a number of questions worthy of further research. First,<br />
it would be interesting to know whether the effect of intra-industry trade<br />
on product switching is related to trade in different products (vertical intraindustry<br />
trade) or similar products (horizontal intra-industry trade). Second,<br />
additional theoretical models need to be developed to gain further insights<br />
into the determinants of product switching versus industry switching, occurring<br />
both within the manufacturing sector <strong>and</strong> to services. Finally, it<br />
could be important to explore whether the quality of exports matters for the<br />
product switches irrespective of the exporting status of the firm.
Data appendix 91<br />
2.A Data appendix<br />
2.A.1 Data <strong>and</strong> sample selection<br />
Trade data<br />
The trade data are from the UN Comtrade (commodity trade statistics)<br />
database <strong>and</strong> consist of the trade values <strong>and</strong> quantities of import flows. We<br />
include all goods (not just manufacturing) <strong>and</strong> use import flows because reporting<br />
of imports is generally more reliable than that of exports. This means<br />
that Estonia’s exports are calculated by looking at the imports of its trading<br />
partners. The import data are at the six-digit product level, according<br />
to the Harmonized System (HS) classification 1988/92. For each product,<br />
an observation consists of the reporter, country of origin, time, trade value<br />
in dollars, quantity, <strong>and</strong> units in which the quantity is expressed. In order<br />
to create our sample, we focus on the top geographic destinations of Estonia’s<br />
imports/exports that account for at least 90 percent of imports/exports<br />
during the period 1997-2005.<br />
<strong>Firm</strong> data<br />
The firm-level data used in this paper are provided by the Estonian Business<br />
Registry <strong>and</strong> cover the period 1995-2005. Due to missing information<br />
on employment for the years 1995 <strong>and</strong> 1996, these years are omitted from<br />
the sample. Since the main purpose of this paper is to identify the impact<br />
of international trade on firm dynamics, we can work only with those sectors<br />
for which we observe imports <strong>and</strong> exports. This implies that those firms<br />
that are not active in the manufacturing sector have to be excluded from the<br />
sample 24 . However, firms that switch to the primary or services sector are retained<br />
until their last year of activity within the manufacturing sector. This<br />
allows us to distinguish between switches occurring within the manufacturing<br />
sector <strong>and</strong> those to other sectors of the economy. Several cleaning procedures<br />
are applied to the sample:<br />
24 Trade data from Comtrade are available only for 3 out of 28 two-digit service sectors<br />
(computers <strong>and</strong> related activities, business services, <strong>and</strong> other service activities). Given<br />
this limited coverage of services trade, we exclude these sectors from the analysis <strong>and</strong> focus<br />
on the manufacturing sector.
92 <strong>Globalization</strong> drives strategic product switching<br />
• We construct a longitudinal panel using registration codes <strong>and</strong> apply<br />
several corrections to take into account changes in firms’ registration<br />
codes (i.e., firms’ identification number): (1) firms that change registration<br />
codes because of a transfer from the Enterprise Registry to the<br />
Business Registry are considered the same firm; (2) in case of acquisitions,<br />
the acquiring <strong>and</strong> acquired firms are considered to be a single<br />
entity for the entire sample period; <strong>and</strong> (3) for all other transactions<br />
(e.g., merger, breakup <strong>and</strong> divesture), firms are treated as separate<br />
entities both before <strong>and</strong> after the transaction.<br />
• Observations with extreme values on one of the variables used in the<br />
empirical analysis are dropped, as well as all observations with missing<br />
information on some variables. This leaves us with 38 percent of the<br />
registered firms, accounting on average for about 60 percent of aggregated<br />
value added in the manufacturing sector 25 .<br />
• Because state-owned firms are not necessarily profit-maximizing agents,<br />
it is not certain whether their behavior can be captured accurately by<br />
our empirical model. Hence, we exclude state-owned firms from the<br />
sample (eight firms).<br />
• All observations for which no matching trade data could be obtained<br />
are omitted. While some of these missing trade data are related to<br />
incorrect EMTAK 26 (NACE) codes reported by firms, others are related<br />
to the concordance used to convert the trade data from Harmonized<br />
System (HS, six-digit, 1992) to NACE (Rev.1.1., four-digit). The trade<br />
data are converted using a concordance table provided by Eurostat.<br />
For the majority of the codes, the concordance gives a many-to-one<br />
mapping, that is, multiple HS 6 digit codes are mapped into one NACE<br />
code. However, there are a number of cases where one HS code maps<br />
25 Comparability between the micro <strong>and</strong> macro data is limited, however, owing to<br />
methodological inconsistencies. Value added at the macro level is a broader concept since<br />
it covers not only the activities of enterprises but also of other economic units. According<br />
to the Statistical Office of Estonia, all enterprises registered in Estonia accounted for about<br />
70 percent of aggregate value added in 2005.<br />
26 The industry classification used in the Estonian Business Registry, EMTAK, is a<br />
five-digit extension of the NACE (Rev.1.1.) classification, that is, the official statistical<br />
classification of economic activities in the European Community.
Data appendix 93<br />
into more than one NACE code. To minimize potential errors, we<br />
have omitted these codes from our concordance table <strong>and</strong> dropped the<br />
resulting observations from our sample. <strong>Firm</strong>s active in sectors that do<br />
not appear in our concordance table are also dropped.<br />
2.A.2 Definitions of variables<br />
<strong>Firm</strong>-level variables<br />
All monetary variables are expressed in real terms. Output <strong>and</strong> intermediate<br />
input deflators, as well as the gross capital formation price index, were<br />
obtained from the Statistical Office of Estonia. Deflators are available for 16<br />
sectors corresponding to the International St<strong>and</strong>ards Industrial Classification<br />
(ISIC Rev.3.1) at the one-digit level.<br />
Exitit+1<br />
Dummy variable, equal to 1 in period t if the firm exits in period t+1. <strong>Firm</strong><br />
exit is defined based on the official date of liquidation from the Commercial<br />
Register. If a firm disappears from the data set <strong>and</strong> it has an official liquidation<br />
date, it is considered to exit in the last year of observation. Liquidation<br />
due to a merger, acquisition or reregistration in the registry is not considered<br />
an exit.<br />
Switch2d,it+1<br />
Dummy variable, equal to 1 in period t if the firm switches two-digit NACE<br />
industries in period t + 1. An industry switch is defined as a change in the<br />
firm’s primary sector of activity.<br />
Switch4d,it+1<br />
Dummy variable, equal to 1 in period t if the firm changes four-digit NACE<br />
products in period t+1. A product switch is defined as a change in the firm’s<br />
main product line.<br />
Ageit<br />
Age of the firm in period t, defined as the number of years the firm has been<br />
in the registry (using the registry entry date).
94 <strong>Globalization</strong> drives strategic product switching<br />
Sizeit<br />
<strong>Firm</strong> size, measured by the number of employees in period t.<br />
Wageit<br />
Average real labor cost, defined as total firm-level labor costs divided by the<br />
number of employees.<br />
Capitalit<br />
Capital intensity, measured as real capital per employee. Capital is defined<br />
as the sum of tangible <strong>and</strong> intangible assets, net of goodwill at the firm level.<br />
TFPit<br />
Total factor productivity at the firm level, estimated at the two-digit industry<br />
level using the methodology of Levinsohn <strong>and</strong> Petrin (2003) while taking<br />
into account industry switches over time.<br />
Foreignit<br />
Foreign ownership dummy, equal to 1 if at least 50 percent of the firm’s<br />
shares are foreign owned.<br />
Product-level variables<br />
All product-level variables are defined at the four-digit NACE level.<br />
Sunkjt<br />
Sunk cost variable, defined as the natural logarithm of the median of real<br />
sales in each particular four-digit industry j at time t.<br />
Herfjt<br />
Herfindahl-Hirschman index for the domestic market, defined as the sum of<br />
squared market shares. Market shares are defined as firm-level real sales over<br />
product-level total real sales.<br />
Importsjt
Data appendix 95<br />
Total imports of product j at time t, measured in Estonian Krooni (EEK).<br />
IITjt<br />
Intra-industry trade variable, defined as the Grubel-Lloyd index, that is,<br />
[1 − (|Xjt − Mjt|/(Xjt + Mjt))], where Xjt <strong>and</strong> Mjt are, respectively, exports<br />
<strong>and</strong> imports of product j at time t.<br />
CAjt<br />
Revealed comparative advantage dummy, equal to 1 if Xjt > Mjt, i.e. if<br />
exports are larger than imports for product j at time t.<br />
Exportsjt<br />
Total exports of product j at time t, measured in Estonian Krooni (EEK).<br />
Herfexjt<br />
The Herfindahl-Hirschman index for the export market is calculated at the<br />
Harmonized System (HS) six-digit level. Conversion to NACE Rev. 1.1. at<br />
the four-digit level is achieved using a concordance table provided by Eurostat<br />
(cfr. section 2.A.1). We define a market as a pair consisting of a geographic<br />
destination <strong>and</strong> a product. We calculate the index of concentration in market<br />
m as<br />
H j<br />
m,t = <br />
∀Exporter<br />
where sn,t is the market share of exporter n in market m at period t. Aggregating<br />
across all export markets for product j, we obtain the overall index<br />
of market concentration for exports of product j:<br />
Herfex jt = <br />
∀m<br />
s 2 n,t<br />
H j<br />
m,t ∗ β j<br />
m,t<br />
where β j<br />
m,t is the share of market m in total exports of product j in period t.<br />
UV Rjt<br />
Average relative unit values index for Estonian export products, taking into<br />
account Estonia’s main competitors. In particular, we compute the unit value<br />
for product j in market p by dividing the export value of each exporter by
96 <strong>Globalization</strong> drives strategic product switching<br />
the export quantity. We consider only those markets in which quantities are<br />
expressed in the same unit across the sample of exporters for that market.<br />
Relative unit values for product j in market p are then calculated dividing<br />
the unit value of Estonia by the weighted average of the unit values of its<br />
competitors in that market. The overall relative unit value for product j is<br />
the weighted sum of the relative unit values across all markets, with weights<br />
equal to export shares.
Data appendix 97<br />
Manufacturing<br />
NACE Description<br />
High-technology manufacturing<br />
244 Pharmaceuticals<br />
30 Office machinery - computers<br />
32 Radio, TV, communication equipment<br />
33 Medical, precision, optical instruments<br />
353 Aircraft - spacecraft<br />
Medium-high-technology manufacturing<br />
24 Chemicals, excl. pharmaceuticals<br />
29 Machinery <strong>and</strong> equipment<br />
31 Electrical machinery<br />
34 Motor vehicles<br />
35 Other transport equipment (excl. 351 & 353)<br />
Medium-low-technology manufacturing<br />
23 Coke, refined petroleum products<br />
25 Rubber <strong>and</strong> plastic<br />
26 Nonmetallic mineral products<br />
27 Basic metals<br />
28 Fabricated metal products<br />
351 Building/repairing of ships <strong>and</strong> boats<br />
Low-technology manufacturing<br />
15 Food <strong>and</strong> beverages<br />
16 Tobacco<br />
17 Textiles<br />
18 Clothing<br />
19 Leather (products)<br />
20 Wood (products)<br />
21 Pulp, paper (products)<br />
22 Publishing <strong>and</strong> printing<br />
36 Furniture<br />
37 Recycling<br />
Source: Eurostat<br />
Table 2.A.1: Sector classification: Manufacturing
98 <strong>Globalization</strong> drives strategic product switching<br />
Services<br />
NACE Description<br />
Knowledge-intensive services<br />
61 Water transport<br />
62 Air transport<br />
64 Post <strong>and</strong> telecommunications<br />
65 Financial intermediation<br />
66 Insurance <strong>and</strong> pension funding<br />
67 Ancilliary financial activities<br />
70 Real estate activities<br />
71 Renting activities<br />
72 Computer <strong>and</strong> related activities<br />
73 Research <strong>and</strong> development<br />
74 Other business activities<br />
80 Education<br />
85 Health <strong>and</strong> social work<br />
92 Recreational activities<br />
Less-knowledge-intensive services<br />
50 Wholesale/retail trade of motor vehicles<br />
51 Wholesale trade<br />
52 Retail trade<br />
55 Hotels <strong>and</strong> restaurants<br />
60 L<strong>and</strong> transport<br />
63 Supporting transport activities<br />
75 Public administration, defense<br />
90 Sewage <strong>and</strong> refuse disposal<br />
91 Activities of membership organizations<br />
93 Other service activities<br />
95 Activities of households<br />
99 Extraterritorial organizations <strong>and</strong> bodies<br />
Source: Eurostat<br />
Table 2.A.2: Sector classification: Services
Chapter 3<br />
Footloose Multinationals in<br />
Belgium?<br />
3.1 Introduction<br />
The popular claim that production abroad has a detrimental impact on<br />
employment <strong>and</strong> exports at home is generally not confirmed by the available<br />
empirical evidence (see Lipsey (2002) for a review of this extensive literature).<br />
Within host countries, multinational firms tend to pay higher wages for given<br />
productivity levels <strong>and</strong> they are also generally found to be more efficient than<br />
local firms. Although evidence on the existence of spillovers of these effects to<br />
the domestic economy is mixed, foreign direct investment (FDI) has played a<br />
clear role in the transformation of some economies from exporters of agricultural<br />
goods <strong>and</strong> raw materials to exporters of manufacturing goods (Lipsey,<br />
2002). However, in spite of all the evidence pointing to the beneficial impact<br />
of multinational firms <strong>and</strong> their operations on home <strong>and</strong> host countries’<br />
economic performance, public concerns largely remain.<br />
One reason for these concerns is that multinationals, due to their ability<br />
to set up production abroad in a profitable way, are often considered to be<br />
footloose; causing them to react more swiftly to shocks <strong>and</strong> hence possibly to<br />
relocate production more rapidly than national 1 firms. This naturally raises<br />
∗ Published in Review of World Economics, 143(3), 2007, 483-507.<br />
1 The term national firms will be used to refer to firms without access to a global<br />
network.<br />
101
102 Footloose multinationals in Belgium?<br />
policy concerns, particularly with respect to foreign multinationals, as the<br />
recent turmoil surrounding the takeover of the European steel group Arcelor<br />
by Mittal Steel has illustrated 2 .<br />
The main objective of the present paper is to determine the impact of<br />
multinational ownership on the exit decisions of firms located in Belgium.<br />
The empirical analysis links in with a small <strong>and</strong> recent literature 3 dealing<br />
with the impact of multinational ownership on exit patterns of manufacturing<br />
establishments <strong>and</strong> contributes to this literature in two important respects.<br />
First, to my knowledge, the role of ownership structure in shutdown decisions<br />
at the firm level has thus far only been considered for the manufacturing<br />
sector. Due to the extensive coverage of the database employed, I am able to<br />
extend the analysis to cover all sectors of the economy 4 . Second, this study is<br />
the first to control for the (possibly differing) impact of foreign <strong>and</strong> domestic<br />
multinational ownership on shutdown decisions at the firm level.<br />
From a policy perspective, the analysis could yield several important insights.<br />
The reduction of the corporate tax rate in Belgium in 2003, as well as<br />
the introduction of the ”notional interest deduction” scheme 5 in June 2005,<br />
although both non-discriminatory between domestic <strong>and</strong> foreign companies,<br />
are primarily aimed at the attraction of new, capital-intensive investments<br />
by foreign multinationals. Moreover, V<strong>and</strong>enbussche <strong>and</strong> Tan (2005) find<br />
2 The Economist (2006, February 2 nd : 11-12). Several government officials in France<br />
<strong>and</strong> Luxembourg were strongly opposed to the takeover of Arcelor, Europe’s largest steel<br />
company, by Mittal Steel, in spite of the weak strategic importance of the steel industry<br />
today. Similarly, the French government raised objections against possible plans of<br />
US-based PepsiCo to take over the French dairy group Danone (The Economist, 2005,<br />
September 1 st : 56).<br />
3 Both Bernard <strong>and</strong> Sjöholm (2003) <strong>and</strong> Görg <strong>and</strong> Strobl (2003) provide evidence that<br />
foreign multinationals, in Indonesia <strong>and</strong> Irel<strong>and</strong> respectively, are more footloose than domestic<br />
companies, i.e. they are more likely to exit the market, after controlling for various<br />
firm <strong>and</strong> industry characteristics. Bernard <strong>and</strong> Jensen (2007a) provide similar evidence<br />
for US multinationals in their home market. For a more detailed discussion of the related<br />
literature, I refer to section 3.2.<br />
4 Since the primary sector represents only a small share of the sample total (less than<br />
2 percent of all firms are active in this sector, accounting for 0.66 of total employment<br />
<strong>and</strong> 1.37 percent of net value added in 2001), it is excluded from the analysis. However,<br />
results do not change when firms active in agriculture <strong>and</strong> mining are taken into account<br />
in the empirical estimations.<br />
5 A tax deduction for equity financing, implemented since January 2006.
Literature review 103<br />
evidence of substantial tax discrimination in favor of foreign multinationals<br />
in Belgium compared to domestic companies.<br />
If, as was shown for a number of other countries, multinationals (either<br />
foreign or home-based) are found to be more footloose than national firms,<br />
this information could certainly be of relevance in fine-tuning current policies,<br />
especially given the impact of multinational firms on the Belgian economy,<br />
both in terms of employment <strong>and</strong> sales. Although less than 2 percent of the<br />
population of firms in Belgium are foreign-owned <strong>and</strong> an equal percentage<br />
are domestic multinationals; foreign multinationals accounted for about 25<br />
percent of total employment <strong>and</strong> 31 percent of net value added in 2001.<br />
Domestic multinationals accounted for another 18 percent of employment<br />
<strong>and</strong> 21 percent of total net value added 6 .<br />
The rest of the paper is organized as follows. In section 3.2 I present a<br />
selective review of the related literature. Section 3.3 introduces the data set<br />
<strong>and</strong> presents some summary statistics, while section 3.4 <strong>and</strong> 3.5 discuss the<br />
empirical model <strong>and</strong> results respectively. In section 3.6 the robustness of the<br />
results is verified on the basis of a number of sensitivity analyses applied to<br />
the model. Section 3.7 concludes.<br />
3.2 Literature review<br />
The focus in this section is on the literature investigating the footloose<br />
nature of multinational enterprises (MNEs) <strong>and</strong> specifically dealing with the<br />
impact of multinational ownership on firm turnover. For a more complete review<br />
of the main theoretical <strong>and</strong> empirical contributions in the industrial organization<br />
literature on firm dynamics <strong>and</strong> turnover, I refer to Caves (1998),<br />
although I will come back to some of the relevant issues emerging from this<br />
literature when I discuss the empirical model in section 3.4.<br />
6 Data are calculated by the author on the basis of the Belfirst database (BvDEP,<br />
2004), which groups the annual accounts of virtually all limited-liability companies active<br />
in Belgium. A more detailed description of the database will be given in section refdata
104 Footloose multinationals in Belgium?<br />
Flamm (1984) looks into the footloose image of US multinationals abroad,<br />
by modeling the volatility of US FDI in the semiconductor industry in response<br />
to changes in the host country environment. His findings indicate that<br />
while US FDI is only moderately sensitive to wage levels in the host countries,<br />
adjustment of the investment portfolio in response to adverse shocks<br />
(such as changes in wages or other costs) occurs extremely rapid.<br />
Rodrik (1997) further notes the asymmetry between groups in the economy<br />
that are free to take their resources wherever they are most in dem<strong>and</strong>,<br />
such as multinational firms; <strong>and</strong> those that can not, including semi-skilled<br />
<strong>and</strong> unskilled workers in the home country. He argues that the increased<br />
substitutability of the services of these immobile groups is likely to lead to<br />
a higher elasticity of dem<strong>and</strong> for their services. Konings <strong>and</strong> Murphy (2006)<br />
find support for this hypothesis using firm level data on 1,067 multinational<br />
firms <strong>and</strong> their 2,078 affiliates located in the European Union (EU) <strong>and</strong><br />
Central <strong>and</strong> Eastern European Countries (CEECs). Their results indicate<br />
that some substitution in employment takes place between the parent firm<br />
<strong>and</strong> its North EU affiliates, i.e. the elasticity of parent employment with<br />
respect to affiliate wages is positive <strong>and</strong> statistically significant; while no<br />
substitution effects are found for the CEEC <strong>and</strong> South EU affiliates, contrary<br />
to the popular belief that major reallocation of employment towards the<br />
CEECs is taking place. Stronger substitution effects are found when the<br />
sectors of activity of the parent <strong>and</strong> affiliate are different.<br />
Similarly, V<strong>and</strong>enbussche <strong>and</strong> Tan (2005) argue that since corporate tax<br />
policy is an important instrument for national governments wanting to attract<br />
foreign direct investment <strong>and</strong> given the fact that multinationals’ mobility<br />
increases their bargaining power, it is likely that foreign multinationals 7<br />
pay lower (effective) taxes. Their empirical results, for Belgium, suggest<br />
that there is considerable tax discrimination in favor of foreign firms, again<br />
lending support to the footloose image of MNEs.<br />
7 They do not distinguish between domestic firms with <strong>and</strong> without access to a global<br />
network.
Literature review 105<br />
Focusing on the extensive margin, several studies have taken up the issue<br />
whether multinationals are more footloose. Theoretically, the expected<br />
impact of multinational ownership on firm turnover is ambiguous. On the<br />
one h<strong>and</strong>, it can be argued that MNEs have certain specific characteristics<br />
that enable them to set up production abroad in a profitable way; allowing<br />
them in turn to respond more swiftly to negative shocks than is possible for<br />
national firms. On the other h<strong>and</strong>, given the fact that multinationals are on<br />
average more skill- <strong>and</strong> capital-intensive than incumbent firms, they might<br />
face higher sunk costs when setting up production, causing them to exit the<br />
market less rapidly, all else equal (Lipsey, 2002). Moreover, since domestic<br />
MNEs are likely to be more strongly rooted in their home country; it is not<br />
certain that domestic <strong>and</strong> foreign multinationals will react in the same way<br />
to changes in the economy.<br />
Bernard <strong>and</strong> Sjöholm (2003), Görg <strong>and</strong> Strobl (2003) <strong>and</strong> Gibson <strong>and</strong><br />
Harris (1996) provide some of the first evidence on the footloose nature of<br />
multinationals in the host country; i.e. they assess the effect of foreign ownership<br />
on plant-level exit patterns in the manufacturing sector in Indonesia,<br />
Irel<strong>and</strong> <strong>and</strong> New Zeal<strong>and</strong> respectively. Both Bernard <strong>and</strong> Sjöholm (2003)<br />
<strong>and</strong> Görg <strong>and</strong> Strobl (2003) find that while foreign-owned plants face lower<br />
hazard rates than domestic plants, this difference is caused by the different<br />
characteristics of these plants, rather than by their foreign character as such.<br />
Foreign plants are on average older, larger <strong>and</strong> more productive, causing<br />
them to exhibit lower exit rates. After controlling for these plant-specific<br />
differences, foreign plants become more likely to exit the market. Gibson<br />
<strong>and</strong> Harris (1996) on the other h<strong>and</strong>, find evidence that foreign MNEs are<br />
less likely to exit the market than incumbent firms, after controlling for plant<br />
<strong>and</strong> industry characteristics; although it is possible these results are partly<br />
driven by the increased trade liberalization that was taking place in New<br />
Zeal<strong>and</strong> at the time.<br />
Bernard <strong>and</strong> Jensen (2007a) perform an analysis of US manufacturing<br />
plants’ deaths in order to determine the impact of the multinational character<br />
of US firms on exit patterns in the home country (they are not able to identify<br />
foreign MNEs in the data). Their model also takes into account whether an
106 Footloose multinationals in Belgium?<br />
establishment is part of a single- or multi-plant firm. Their findings indicate<br />
that multinational plants, as well as plants belonging to a larger group, are<br />
more likely to exit the market, after controlling for various other plant- <strong>and</strong><br />
industry-specific variables. Unconditionally, both plants owned by multiplant<br />
firms <strong>and</strong> plants owned by US multinationals exhibit lower exit rates<br />
than incumbent plants.<br />
For Belgium, both Van den Cruyce (1999) <strong>and</strong> Pennings <strong>and</strong> Sleuwaegen<br />
(2000, 2002) have searched for the determinants of different modes of<br />
downsizing at the firm level (exit, partial relocation or downsizing) in the<br />
manufacturing sector. They find a positive impact of multinational ownership<br />
on relocation decisions of firms, but no significant impact on exit<br />
patterns. However, a serious drawback of their multinational ownership variable<br />
is that it does not distinguish for nationality of ownership; foreign <strong>and</strong><br />
domestic multinationals firms are taken together in the empirical analysis.<br />
This paper differs from previous studies in two important respects. First,<br />
due to the extensive coverage of the database employed, I am able to investigate<br />
the impact of multinational ownership on exit patterns separately<br />
for firms active in manufacturing <strong>and</strong> services. Second, by distinguishing<br />
between foreign firms <strong>and</strong> domestic MNEs in the empirical analysis, it is<br />
possible to investigate whether nationality of ownership plays a role in shaping<br />
the exit patterns of multinational firms.<br />
3.3 Data <strong>and</strong> preliminary facts<br />
The data set employed in this paper is constructed on the basis of the<br />
Belfirst database, which groups the annual accounts reported by firms located<br />
in Belgium to the National Bank of Belgium 8 . The database is com-<br />
8 “Most enterprises in which the liability of the shareholders or members is limited to<br />
their contribution to the company, plus some other enterprises, have to file their annual<br />
accounts <strong>and</strong>/or consolidated accounts with the Central Balance Sheet Office at the National<br />
Bank every year” (http://www.nbb.be). This implies that the database consists<br />
of the complete population of (limited-liability) firms in Belgium (318,316 companies).<br />
However, about 160,000 firms in the database are very small, i.e. they have no recorded<br />
employment over the sample period. Since a number of variables used in the empirical<br />
analysis (such as firm size <strong>and</strong> wages) are not available for these companies <strong>and</strong> given
Data <strong>and</strong> preliminary facts 107<br />
mercialized by BvDEP (2004) <strong>and</strong> has been used in a growing number of<br />
academic papers in recent years 9 . <strong>Firm</strong>s in the database are uniquely defined<br />
by their VAT number <strong>and</strong> data on employment, wages, net value added etc.<br />
are available for the years 1996-2003. Sectors in the database are classified<br />
according to the NACE-Bel nomenclature, i.e. a five-digit extension of the<br />
NACE (Revision 1) Classification commonly used for European statistics 10 .<br />
Furthermore, the database includes information on the ownership structure<br />
of firms, in terms of both subsidiaries <strong>and</strong> shareholders, either foreign or<br />
domestic.<br />
To identify entry <strong>and</strong> exit, I follow the procedure used by Mata <strong>and</strong> Portugal<br />
(1994) <strong>and</strong> Mata, Portugal <strong>and</strong> Guimarães (1995). For the specifics<br />
associated with the exit variable <strong>and</strong> the selection of the sample used in the<br />
empirical analysis, I refer to the data appendix (section 3.A). Omitting all<br />
observations that do not fit the definition of exit, as well as all firms for<br />
which data needed in the empirical analysis are incomplete, results in an<br />
unbalanced sample of 26,046 companies 11 .<br />
A firm is considered to be foreign-owned in the data set if it has some<br />
foreign ownership. Likewise, a firm is identified as a domestic multinational<br />
if it has subsidiaries in countries other than Belgium <strong>and</strong> it is not foreignowned.<br />
Ideally, I would have liked to identify multinationals on the basis<br />
of some minimum share of direct ownership, but unfortunately these data<br />
the otherwise limited reporting requirements for these firms, they are necessarily excluded<br />
from the analysis.<br />
9 Examples include V<strong>and</strong>enbussche, Crabbé <strong>and</strong> Janssen (2005) in a study on the determinants<br />
of effective corporate tax rates of large Belgian firms <strong>and</strong> De Loecker (2007)<br />
who estimates the impact of increased trade liberalization on productivity in the Belgian<br />
textile sector.<br />
10 The NACE Rev. 1 classification can be downloaded from the Eurostat Ramon server:<br />
http://europa.eu.int/comm/eurostat/ramon/.<br />
11 Omitting all companies with no recorded employment over the sample period, as well<br />
as firms that exhibit irregular entry <strong>and</strong> exit patterns or were subject to an ownership<br />
change over the sample period <strong>and</strong> observations with missing data for any of the independent<br />
variables results in a sample of 98,100 enterprises. Following Mata et al. (1994, 1995),<br />
only firms that employ 10 employees or more in at least one sample year are included (see<br />
appendix 3.A). Some of these restrictions will be relaxed in section 3.6 however.
108 Footloose multinationals in Belgium?<br />
Number of firms<br />
0 200 400 600 800<br />
864<br />
Netherl<strong>and</strong>s<br />
621<br />
France<br />
252<br />
US<br />
172<br />
Germany<br />
145<br />
UK<br />
95 94 81<br />
Others EU<br />
Sweden<br />
Switzerl<strong>and</strong><br />
52 45 45 37<br />
Figure 3.1: Foreign multinationals by country of origin<br />
are missing for the majority of cases 12 . Another drawback of the ownership<br />
measure is that it is time-invariant; the information refers to the latest year<br />
available. Figure 3.1 displays the country distribution of the foreign firms in<br />
the sample.<br />
As can be seen in figure 3.1, most ownership of foreign firms is concentrated<br />
in the EU (83 percent); the US accounts for about 10 percent <strong>and</strong><br />
only 7 percent of foreign owners originate in other countries of the world.<br />
Germany, the Netherl<strong>and</strong>s, the U.K. <strong>and</strong> France, which are all direct neighbors<br />
of Belgium, together account for over 70 percent of foreign ownership.<br />
Table 3.1 shows the sector distribution of firms in the sample, distinguished<br />
by ownership type.<br />
About 25 percent of all firms in the sample are active in manufacturing<br />
(6,524 firms), while 75 percent or 19,522 firms are active in services. Within<br />
manufacturing, the largest share is taken up by the metal industries (19<br />
percent), followed closely by the food <strong>and</strong> textile industries (shares of 14<br />
12 Similarly, information regarding indirect ownership was missing for the majority of<br />
cases.<br />
Denmark<br />
Luxembourg<br />
Japan<br />
ROW
Data <strong>and</strong> preliminary facts 109<br />
National Foreign Domestic Total<br />
Sector of activity <strong>Firm</strong>s MNEs MNEs N<br />
D. Manufacturing 5,202 704 618 6,524<br />
DA. Food, beverages <strong>and</strong> tobacco 756 75 93 924<br />
DB. Textiles 646 23 64 733<br />
DC. Leather (products) 33 1 2 36<br />
DD. Wood (products) 231 7 13 251<br />
DE. Pulp <strong>and</strong> paper 583 51 59 693<br />
DF. Coke <strong>and</strong> petroleum (products) 7 9 4 20<br />
DG. Chemicals 170 127 58 355<br />
DH. Rubber <strong>and</strong> plastics 201 48 43 292<br />
DI. Other non-metallic products 337 53 30 420<br />
DJ. Metallic products 1,032 111 80 1,223<br />
DK. Machinery <strong>and</strong> equipment n.e.c. 329 72 61 462<br />
DL. Electrical <strong>and</strong> optical equipment 259 73 63 395<br />
DM. Transport equipment 130 32 18 180<br />
DN. Manufacturing n.e.c. 488 22 30 540<br />
E-Q. Services <strong>and</strong> related activities 16,828 1,799 895 19,522<br />
E. Electricity, gas <strong>and</strong> water supply 28 10 4 42<br />
F. Construction 3,670 145 97 3,912<br />
G. Wholesale <strong>and</strong> retail trade 6,212 785 343 7,340<br />
H. Hotels <strong>and</strong> restaurants 1,165 39 13 1,217<br />
I. Transport <strong>and</strong> communication 2,017 196 112 2,325<br />
J. Financial intermediation 269 57 37 363<br />
K. Other business activities 2,504 516 249 3,269<br />
L. Public services 12 0 0 12<br />
M. Education 44 4 4 52<br />
N. Health <strong>and</strong> social work 380 2 7 389<br />
O. Other service activities 524 45 29 598<br />
P. Private household act. 1 0 0 1<br />
Q. Extra-territorial org. 2 0 0 2<br />
Total 22,030 2,503 1,513 26,046<br />
Table 3.1: Sector distribution of firms
110 Footloose multinationals in Belgium?<br />
<strong>and</strong> 11 percent respectively). Compared to the sample average, domestic<br />
multinationals are somewhat over-represented in the manufacturing sector;<br />
41 percent of all domestic MNEs are active in this sector, compared to 25<br />
percent for the overall sample. Within the service sector, most firms are<br />
active in wholesale <strong>and</strong> retail activities (38 percent) <strong>and</strong> in construction (20<br />
percent).<br />
In order to examine whether multinational firms, either foreign or domestic,<br />
are unconditionally more or less likely to exit than national firms,<br />
I calculate (nonparametric) Kaplan-Meier survival estimates separately for<br />
each group in Own (Own = 0 for national firms; 1 for foreign MNEs <strong>and</strong> 2<br />
for domestic MNEs):<br />
ˆS (a) = <br />
j|aj≤a<br />
<br />
nj − dj<br />
nj<br />
(3.1)<br />
where nj indicates the number of plants that have survived up to aj<br />
years of age; dj is the number of plants that die at age aj <strong>and</strong> ˆ S(a) gives the<br />
probability of surviving up to age aj (Greene, 2003, 789-789). The survival<br />
functions corresponding to (3.1) are plotted in figure 3.2. Analysis time<br />
represents firms’ age.<br />
From the graph, it is clear that multinationals (foreign or domestic, Own<br />
equals 1 or 2) have a higher survival rate than national firms (Own equals<br />
0). A log-rank test of equality of the survival functions yields similar results;<br />
the hypothesis that survival functions are equal for multinational firms <strong>and</strong><br />
national firms is rejected decisively (with probabilities of 0.00 <strong>and</strong> associated<br />
Chi-square values higher than 10) for both foreign <strong>and</strong> domestic MNEs. This<br />
result holds when stratifying over two-digit sectors <strong>and</strong> years. The difference<br />
between foreign <strong>and</strong> domestic MNEs is less clear-cut, both graphically <strong>and</strong><br />
mathematically. If survival is not stratified by sector <strong>and</strong> year, their survival<br />
functions are significantly different at the 5 percent level (probability of 0.03).<br />
This result no longer holds when stratifying over sector <strong>and</strong> years, the logrank<br />
test of equality of the survival functions does not reject the hypothesis<br />
of equality in this case (probability of 0.13).
Data <strong>and</strong> preliminary facts 111<br />
Survival probability<br />
0.00 0.25 0.50 0.75 1.00<br />
Kaplan−Meier survival estimates<br />
0 50 100 150<br />
Analysis time<br />
Own = Domestic Own = Foreign MNE<br />
Own = Domestic MNE<br />
Figure 3.2: Kaplan-Meier survival functions by nationality of ownership<br />
The finding that multinationals exhibit lower exit rates can also be seen in<br />
the second row of table 3.2. Between 1996 <strong>and</strong> 2001, about 9 percent of national<br />
firms exit the sample, compared to 6 percent for domestic MNEs <strong>and</strong><br />
7 percent for foreign firms. Hence, it is clear that multinationals, either foreign<br />
or domestic are unconditionally less likely to exit the market compared<br />
to national firms. This finding is consistent with the results obtained by<br />
Bernard <strong>and</strong> Jensen (2007a); Bernard <strong>and</strong> Sjöholm (2003); Görg <strong>and</strong> Strobl<br />
(2003); who also report lower exit rates for multinational firms.<br />
Table 3.2 further reports summary statistics for the pooled sample, distinguished<br />
by ownership type. All reported measures, with the exception of<br />
the number of firms <strong>and</strong> exits, represent averages over the sample period;<br />
st<strong>and</strong>ard deviations are reported in brackets. For a detailed description of<br />
the data <strong>and</strong> variables used, I refer to the data appendix (section 3.A).<br />
The first row in table 3.2 shows the number of firms distinguished by ownership<br />
type. National firms account for about 85 percent of the sample total,<br />
while domestic <strong>and</strong> foreign MNEs account for about 5 <strong>and</strong> 10 percent respectively.<br />
Although the number of multinationals in the sample seems rather
112 Footloose multinationals in Belgium?<br />
Domestic Foreign Domestic<br />
Variable firms MNEs MNEs<br />
Number of firms 22,030 2,503 1,513<br />
(% of total firms) (84.58) (9.61) (5.81)<br />
Number of exits 2,040 187 90<br />
(Total % over sample period) (9.26) (7.47) (5.95)<br />
Age (1) 18.99 22.82 23.3<br />
(years) (14.83) (18.96) (18.98)<br />
Size (1) 29.97 165.22 204.51<br />
(number of employees) (120.83) (577.67) (1,453.59)<br />
Wage per employee (1) 34.14 52.2 45.58<br />
(x 1,000e) (33.95) (28.56) (38.40)<br />
Productivity per worker (1) 54.19 95.08 100.37<br />
(x 1,000e) (91.71) (208.76) (518.70)<br />
Labor costs relative to productivity (2) 0.83 0.89 0.78<br />
(8.79) (4.79) (2.03)<br />
Industry concentration (1) 503.71 811.97 823.07<br />
(Herfindahl index three-digit NACE) (870.59) (1,107.58) (1,165.77)<br />
Notes: (1) Values represent sample means. St<strong>and</strong>ard deviations are reported in<br />
brackets. (2) Wage over productivity. Value represents sample mean, st<strong>and</strong>ard<br />
deviation reported in brackets.<br />
Table 3.2: Summary statistics by ownership type (1996-2001)
Data <strong>and</strong> preliminary facts 113<br />
limited, their importance in terms of employment <strong>and</strong> turnover should not<br />
be underestimated. Multinational firms account for over 50 percent of total<br />
employment <strong>and</strong> more than 60 percent of net value added in the sample. This<br />
is caused by the fact that multinationals, both domestic <strong>and</strong> foreign, are on<br />
average much larger than national firms. The average foreign multinational<br />
employs more than 5 people for each person employed in a national firm (165<br />
versus 30); domestic MNEs employ (on average) more than 6 people for each<br />
employee in a national firm (205 versus 30). The average multinational is<br />
also older than national firms (23 versus 19 years).<br />
Turning to the figures on wages <strong>and</strong> productivity per employee in table 3.2,<br />
it is clear that multinationals, either foreign or domestic, are on average more<br />
productive <strong>and</strong> pay higher wages compared to national firms. Since economic<br />
theory suggests a close link between labor costs <strong>and</strong> productivity, table 3.2<br />
also displays the sample average of labor costs relative to productivity. The<br />
figures indicate that the average foreign firm pays 89 cents per euro of net<br />
value added, compared to 83 cents per euro for national firms <strong>and</strong> only 78<br />
cents per euro for domestic MNEs. If these numbers are interpreted as a<br />
measure of competitiveness at the firm level, they suggest that foreign firms<br />
are, on average, less competitive than their domestic counterparts (either<br />
national or multinational) given that they pay higher wages after controlling<br />
for productivity 13 (Konings, 2005). The last row of table 3.2 shows that<br />
multinationals tend to be active in sectors characterized by a higher industry<br />
concentration, as measured by the Herfindahl concentration index (see<br />
appendix 3.A for the specifics of this variable).<br />
Since the simple summary measures presented here do not allow us to<br />
disentangle the impact of firm- <strong>and</strong> industry-specific effects from that of<br />
multinational ownership (foreign or domestic) on firms’ propensity to exit, I<br />
now turn my attention to the full empirical model.<br />
13 If value added is considered a proxy for total output, the ratio of the total wage bill to<br />
value added gives an indication of the relative competitiveness of a firm, since it measures<br />
to what extent value added of the firm covers the wage bill. Lipsey (2002) sums up a<br />
number of motives for foreign firms to pay higher wages for labor of a given quality.
114 Footloose multinationals in Belgium?<br />
3.4 Empirical model<br />
In order to assess the impact of multinational ownership on exit patterns,<br />
it is necessary to estimate a model that accounts for the differences in size,<br />
productivity <strong>and</strong> wages between multinational <strong>and</strong> national firms. Given the<br />
nature of the data 14 <strong>and</strong> following the related literature (eg. Bernard <strong>and</strong><br />
Sjöholm (2003); Chen (2002); Disney et al. (2003); Görg <strong>and</strong> Strobl (2003);<br />
Mata <strong>and</strong> Portugal (1994); Mata et al. (1995)), I estimate a proportional<br />
hazard model:<br />
λ (a, x, β, λ0) = ϕ (x, β) λ0 (a) (3.2)<br />
where the hazard function λ (.) depends multiplicatively on the vector<br />
of explanatory variables x with unknown coefficients β <strong>and</strong> the baseline hazard<br />
λ0 (a) (corresponding to ϕ (.) equal to 1). For the special case where<br />
ϕ (x, β) = exp (x ′ β), estimation of β does not require specification of the<br />
baseline hazard λ0 (a) (Kiefer, 1988, 666).<br />
Since the fundamental interest is in the effect of the covariates on the<br />
hazard (<strong>and</strong> not in the shape of the baseline hazard), a natural choice is<br />
to normalize the baseline hazard to 1 <strong>and</strong> estimate (3.2) using the partial<br />
likelihood approach first suggested by Cox (1972). Specifically, the empirical<br />
model looks as follows:<br />
h (a) = h0 (a) exp [αOwn + βX]<br />
X = [ln (Sizeit) , ln (Wageit) , ln (Prodit),ln (Herfjt)]<br />
Own = [Fori, Domi]<br />
(3.3)<br />
where h (a) is the hazard rate; i.e. the rate at which plants exit at age a,<br />
conditional upon having survived up to a-1; h0 (a) is baseline hazard, allowed<br />
to vary over two-digit sectors <strong>and</strong> years; X is the vector of control variables,<br />
including firm-level employment (Sizeit), wages (Wageit), labor productivity<br />
(Prodit) <strong>and</strong> industry-level concentration, measured using a Herfindahl<br />
14 The data are characterized by left truncation (delayed entry), i.e. firms may exist for<br />
some time before the sample period starts; <strong>and</strong> right censoring occurs, i.e. it is possible<br />
that a firm has not failed at the end of the sample period. Kiefer (1988) presents an<br />
overview of the different concepts related to duration data. Unlike conventional regression<br />
methods, duration models are specifically designed to take these issues into account.
Empirical model 115<br />
index (Herfjt). Fori <strong>and</strong> Domi are the foreign <strong>and</strong> domestic multinational<br />
ownership dummies respectively.<br />
Estimations of (3.3) are stratified by two-digit sector <strong>and</strong> year, allowing for<br />
equal coefficients across strata, but baseline hazards specific to each sector<br />
<strong>and</strong> year. In what follows, I will briefly discuss the definitions <strong>and</strong> expected<br />
signs of the variables included in (3.3).<br />
An empirical implication of models dealing with firm dynamics <strong>and</strong> allowing<br />
for entry, exit <strong>and</strong> heterogeneity across firms (eg. Hopenhayn (1992);<br />
Ericson <strong>and</strong> Pakes (1995)); is that firms’ hazard rate is decreasing in size,<br />
conditional on age. Dunne et al. (1988, 1989) already established a positive<br />
relationship between firms’ age <strong>and</strong> size <strong>and</strong> their survival probabilities <strong>and</strong><br />
many empirical studies find support for this hypothesis (eg. Audretsch <strong>and</strong><br />
Mahmood, 1995). Consequently, I include firms’ current size 15 in the model,<br />
defined in terms of employment at time t (expressed in logarithm form).<br />
<strong>Firm</strong> size is expected to have a negative effect on the hazard rate (positive<br />
effect on firm survival).<br />
The wage variable is defined as firms’ wages 16 , divided by employment in<br />
year t. To the extent that higher wages reflect higher skill intensities <strong>and</strong><br />
associated sunk costs at the firm level (related to training <strong>and</strong> investment in<br />
firm-specific human capital), higher wages can be expected to have a negative<br />
effect on firm failure. Audretsch <strong>and</strong> Mahmood (1995) find support for the<br />
negative relationship between wages <strong>and</strong> the propensity to exit in their study<br />
on the post-entry performance of 12,000 US manufacturing plants. However,<br />
it is not certain whether this result will hold when taking into account productivity<br />
at the firm level. As was already alluded to in section 3; if firms pay<br />
higher wages for given levels of productivity, this could signal lower competitiveness<br />
<strong>and</strong> hence increase their exit probability, ceteris paribus (Konings,<br />
2005).<br />
15 The effect of age on the hazard rate is incorporated directly into the model, since<br />
duration is a function of the firm’s age. Consequently, multicollinearity issues prevent<br />
inclusion of the firm’s age as a separate regressor in the model.<br />
16 The wage variable covers both remunerations <strong>and</strong> social security costs (including pensions).<br />
For an overview of the definitions of the independent variables, I refer to section 3.A.
116 Footloose multinationals in Belgium?<br />
Several theoretical models (eg. Jovanovic, 1982; Hopenhayn, 1992) predict<br />
that the growth <strong>and</strong> exit of firms is motivated to a large extent by productivity<br />
differences at the firm level. Empirically, Fariñas <strong>and</strong> Ruano (2005) find,<br />
for a sample of Spanish manufacturing firms, that entry <strong>and</strong> exit decisions<br />
are systematically negatively related to differences in productivity at the firm<br />
level. A negative effect of the labor productivity variable on firms’ hazard<br />
rates is therefore expected.<br />
In addition to the firm-specific variables, an industry-specific variable is<br />
introduced in the model. Economic theory predicts a negative relationship<br />
between industry concentration <strong>and</strong> turnover. Higher industry concentration,<br />
as measured by the Herfindahl index, is expected to lead to increased<br />
price-cost margins <strong>and</strong>, through increased profitability, reduce firm turnover<br />
(Mueller, 1991). However, Caves (1998) notes that as concentration is determined<br />
by the extent of entry barriers in an industry, as well as the degree of<br />
cooperation among producers, it is not clear what the effect of including this<br />
variable to explain hazard rates will be. Results obtained in earlier studies<br />
are similarly ambiguous. Görg <strong>and</strong> Strobl (2003) find a significantly positive<br />
impact of concentration on hazard rates at the firm level, while Mata<br />
<strong>and</strong> Portugal (1994) find a negative, though insignificant effect of concentration<br />
on exit patterns. The Herfindahl index (Herfjt) is calculated at the<br />
three-digit industry level using all firms with reported turnover in the Belfirst<br />
database (see section 3.A). The market shares are calculated on the basis of<br />
firm <strong>and</strong> industry turnover in each specific year.<br />
Finally, I come to the two ownership variables, Domi <strong>and</strong> Fori. As was<br />
noted previously, there is no clear theoretical indication about the impact<br />
of multinational ownership on exit patterns at the firm level. It can be<br />
argued that multinationals’ ability to shift production around the world in a<br />
profitable way enables them to respond more swiftly to adverse shocks in the<br />
home or host country, causing them to exit the market more rapidly, all else<br />
equal. However, it is also possible that multinationals face higher sunk costs<br />
when setting up production, due to their higher capital <strong>and</strong> skill intensity<br />
(on average), causing them to exit the market less rapidly, ceteris paribus. In<br />
addition, it is not inconceivable that foreign multinationals react differently
Empirical results 117<br />
to adverse shocks compared to domestic multinationals that are more firmly<br />
rooted in the local economy. I now turn to the discussion of the empirical<br />
results.<br />
3.5 Empirical results<br />
Table 3.3 reports the results of the Cox proportional hazard model applied<br />
to the unbalanced sample of 26,046 firms over 6 years (1996-2001). Correlations<br />
between the independent variables are generally low, with the exception<br />
of ln(Wagesit) <strong>and</strong> ln(Prodit) for which the correlation amounts to 0.70. For<br />
each regression, coefficients <strong>and</strong> associated robust st<strong>and</strong>ard errors, adjusted<br />
for clustering at the firm level, are reported. The first two columns in the<br />
table show the results of the model applied to all sectors (NACE 15-99),<br />
while columns III-IV <strong>and</strong> V-VI display results separately for manufacturing<br />
<strong>and</strong> services respectively.<br />
Columns I, III <strong>and</strong> V in table 3.3 show the results for the basic model in<br />
(3.3). From figure 3.2 it is clear that multinationals (either foreign or domestic)<br />
exhibit lower exit rates than national firms. Estimation of the full model<br />
however, reveals that after controlling for the fact that multinationals are on<br />
average larger, more productive <strong>and</strong> pay higher wages, foreign firms are more<br />
likely to exit the market than national firms; the coefficient on Fori is positive<br />
<strong>and</strong> significant both for the manufacturing <strong>and</strong> service sectors. Results<br />
for the domestic ownership variable (Domi) are less clear-cut. While the domestic<br />
ownership dummy is found to have a significantly positive coefficient<br />
in the manufacturing sector, its coefficient is very small <strong>and</strong> insignificant for<br />
services.<br />
In order to interpret the magnitude of these effects, it is useful to calculate<br />
the hazard ratio corresponding to these coefficients by taking its exponential.<br />
For a dummy variable, the hazard ratio can be interpreted as the increase in<br />
the overall hazard rate facing the firm, corresponding to Fori or Domi equal<br />
to 1. Calculation of the exponential of the coefficients on Fori yields hazard<br />
ratios between 1.27 (for services) <strong>and</strong> 1.70 (for manufacturing), indicating<br />
that the hazard rate is between 1.3 <strong>and</strong> 1.7 times higher for foreign firms than
118 Footloose multinationals in Belgium?<br />
All Manufacturing Services<br />
Sectors (NACE 15-99) (NACE 15-37) (NACE 40-99)<br />
Variables I II III IV V VI<br />
Domestic MNE 0.10 0.09 0.50** 0.50** -0.04 -0.05<br />
[0.12] [0.12] [0.20] [0.20] [0.15] [0.15]<br />
Foreign MNE 0.30*** - 0.53*** 0.24** -<br />
[0.09] [0.18] [0.10]<br />
EU Parent - 0.32*** - 0.45** - 0.30***<br />
[0.09] [0.19] [0.10]<br />
US Parent - 0.35 - 0.78* - 0.16<br />
[0.28] [0.45] [0.38]<br />
ROW Parent - -0.24 - 1.57*** - 0.88<br />
[0.46] [0.50] [0.67]<br />
log(Employment) -0.60*** -0.60*** -0.61*** -0.61*** -0.61*** -0.61***<br />
[0.03] [0.03] [0.06] [0.06] [0.04] [0.04]<br />
log(Wages) 0.11 0.11 0.55*** 0.55*** 0.04 0.05<br />
[0.07] [0.07] [0.19] [0.19] [0.08] [0.08]<br />
log(Productivity) -0.76*** -0.76*** -0.95*** -0.95*** -0.73*** -0.73***<br />
[0.04] [0.04] [0.10] [0.10] [0.05] [0.05]<br />
log(Herfindahl) 0.00 0.00 -0.10 -0.10 0.02 0.02<br />
[0.03] [0.03] [0.08] [0.08] [0.03] [0.03]<br />
Wald test 764.6*** 769.6*** 219.7*** 225.0*** 578.7*** 581.3***<br />
N 143,010 36,721 106,289<br />
Number of firms 26,046 6,524 19,522<br />
Number of failures 2,317 520 1797<br />
% of total 8.9 8.0 9.2<br />
Notes: Robust st<strong>and</strong>ard errors, adjusted for clustering at the plant level (reported<br />
in brackets). Baseline hazard stratified by two-digit sector <strong>and</strong> year. Efron option<br />
for h<strong>and</strong>ling ties. Significance levels: *** p
Empirical results 119<br />
for national firms, depending on their sector of activity. Similarly, taking the<br />
exponential of the coefficient on Domi yields hazard ratios between 0.96 for<br />
services (insignificant) <strong>and</strong> 1.65 for firms active in manufacturing, indicating<br />
that domestic MNEs active in the production sector are 1.65 times as likely<br />
to exit than national firms.<br />
The results obtained for the foreign ownership variable are in line with<br />
the findings of Bernard <strong>and</strong> Sjöholm (2003) <strong>and</strong> Görg <strong>and</strong> Strobl (2003)<br />
for Indonesia <strong>and</strong> Irel<strong>and</strong> respectively. Moreover, the results for the domestic<br />
ownership dummy are similarly in line with those obtained (for the<br />
manufacturing sector) by Bernard <strong>and</strong> Jensen (2007a) for the US. These<br />
findings lend support to the hypothesis that MNEs, either domestic or foreign,<br />
are indeed more footloose than national firms. Furthermore, the fact<br />
that domestic multinationals engaged in service activities do not exhibit significantly<br />
higher exit rates than national firms, after controlling for their<br />
differing characteristics, has important policy implications.<br />
Results obtained on the firm- <strong>and</strong> industry-specific independent variables<br />
in table 3.3 are in accordance with expectations. The coefficient on firm size<br />
is significantly negative, supporting the hypothesis that larger firms tend<br />
to exhibit lower exit rates. Wages have a positive impact on firms’ hazard<br />
rate after controlling for productivity at the firm level, in line with the hypothesis<br />
that firms are relatively less competitive if they pay higher wages<br />
for given productivity levels. However, if productivity is omitted from the<br />
regression (unreported), I find a significantly negative sign for the wage variable,<br />
supporting the hypothesis that higher wages reflect a higher relative<br />
skill intensity, leading to higher sunk costs <strong>and</strong> hence a lower probability of<br />
exit. The coefficient on industry concentration as measured by the Herfindahl<br />
index is never statistically significant.<br />
Columns II, IV <strong>and</strong> VI in table 3.3 present the results of a refinement to<br />
the model. Specifically, I introduce three separate variables representing the<br />
home countries of the foreign firms to replace the foreign ownership dummy.<br />
Discussion of results here is limited to the ownership variables, since all other<br />
variables perform similarly to the basic model. The variable EUi groups all
120 Footloose multinationals in Belgium?<br />
firms with home countries in the European Union (EU), while USi represents<br />
firms originating in the US or Canada. ROWi groups all other home<br />
countries.<br />
As can be seen in the table, a positive <strong>and</strong> significant effect is found for all<br />
three variables in the manufacturing sector, while for services only firms originating<br />
in the EU exhibit significantly higher exit rates than national firms.<br />
For manufacturing, corresponding hazard ratios for the coefficients are 2.18<br />
(for USi) <strong>and</strong> 4.81 (for ROWi), compared to 1.57 for firms in the European<br />
Union. For services, the hazard ratio corresponding to EUi amounts to 1.35;<br />
while for firms originating in the US or the rest of the world, results are insignificant.<br />
These results suggest that manufacturing firms, originating from<br />
outside the EU, are more likely to exit the market, not only compared to<br />
national firms, but also compared to foreign MNEs originating in the EU.<br />
In summary, after controlling for the fact that multinational firms tend to<br />
be larger, more productive <strong>and</strong> pay higher wages than domestic firms (cfr.<br />
table 3.2), both foreign <strong>and</strong> domestic multinationals show a higher propensity<br />
to exit than national firms active in manufacturing. For services, only foreign<br />
multinationals exhibit significantly higher hazard ratios. Furthermore, within<br />
the manufacturing sector, firms originating from outside the EU are found<br />
to exhibit higher hazard ratios, both compared to foreign firms from within<br />
the EU <strong>and</strong> compared to national firms. The magnitude of these effects is in<br />
line with results obtained in earlier studies. Görg <strong>and</strong> Strobl (2003) find that<br />
foreign firms’ hazard rates are 1.4 times as high as domestic companies <strong>and</strong><br />
Bernard <strong>and</strong> Sjöholm (2003) associate foreign ownership with an increase in<br />
the exit probability of 20 percent.<br />
3.6 Robustness checks<br />
In order to verify the robustness of the results presented in the previous<br />
section, a number of sensitivity analyses have been applied to the original<br />
model, both in terms of sample selection <strong>and</strong> methodology used. First, as<br />
is explained in detail in the data appendix (section 3.A), the sample used<br />
to estimate (3.3) in the previous section consists only of firms employing 10
Robustness checks 121<br />
or more people in at least one sample year. However, a common finding in<br />
papers dealing with exit rates at the firm level is that small plants tend to<br />
exhibit (proportionately) higher exit rates than their larger counterparts (eg.<br />
Bernard <strong>and</strong> Jensen, 2007a; Disney et al., 2003). Bernard <strong>and</strong> Sjöholm (2003)<br />
further note that given the larger average size of MNEs, exclusion of small<br />
firms from an empirical analysis dealing with the impact of multinationality<br />
on exit patterns is likely to seriously bias the results .<br />
In order to verify the sensitivity of the results displayed in table 3.3 to the<br />
inclusion of small firms in the sample, I have re-estimated (3.3) using the full<br />
sample of firms with positive employment in at least one sample year (98,100<br />
firms). The results are displayed separately for manufacturing <strong>and</strong> services in<br />
the first two columns of table 3.4. The most notable differences compared to<br />
estimation of the restricted sample include the increase in the size effect for<br />
the full sample (from about 0.60 to more than 1.00); the significantly negative<br />
coefficient for wages; the significantly positive effect of industry concentration<br />
for services <strong>and</strong> the lower coefficient for the productivity variable, which is<br />
only half as large as for the restricted sample. For the ownership variables,<br />
results are very similar to those obtained in table 3.3, although the coefficient<br />
on the foreign ownership variable is somewhat larger in both sectors (0.83 for<br />
manufacturing <strong>and</strong> 0.33 for services, compared to 0.53 <strong>and</strong> 0.24 respectively).<br />
Hence, it has been shown that the results obtained in the previous section<br />
are robust to the inclusion of small firms in the analysis.<br />
Second, although the Belfirst database provides information on whether<br />
firms have been subject to a change in ownership (eg. a merger or takeover),<br />
it is not possible to track these firms following the ownership change. It is<br />
possible that a takeover results in an exit, followed by the start-up of a new<br />
firm; in which case only the exit of the firm can be observed. Since the exit<br />
rate of those firms that have been subject to some form of ownership change<br />
amounts to 85 percent, compared to 8.90 percent for the sample; it seems<br />
more than likely that a number of these exits are in fact misclassified. Moreover,<br />
no information is available on the nationality of ownership before <strong>and</strong><br />
after the change. Therefore, all firms that have been subject to a takeover, a<br />
merger or a scission over the sample period, were initially omitted from the
122 Footloose multinationals in Belgium?<br />
Cox PH Model (1) Cox PH Model (1)<br />
Model Incl. small firms (2) Incl. takeovers (3) Probit (4)<br />
Variables I II III IV V VI<br />
Sector of activity Industry Services Industry Services Industry Services<br />
Constant - - - - -1.07*** -1.19***<br />
[0.35] [0.32]<br />
Domestic MNE 0.59*** -0.02 0.50*** 0.00 0.14* 0.01<br />
[0.22] [0.09] [0.19] [0.14] [0.08] [0.06]<br />
Foreign MNE 0.83*** 0.33*** 0.43** 0.20** 0.25*** 0.11***<br />
[0.19] [0.08] [0.17] [0.09] [0.07] [0.04]<br />
Merger dummy - - 3.09*** 2.61*** - -<br />
[0.20] [0.07]<br />
log(Age) - - - - 0.01 -0.03***<br />
[0.02] [0.01]<br />
log(Employment) -1.12*** -1.26*** -0.57*** -0.55*** -0.25*** -0.23***<br />
[0.04] [0.01] [0.06] [0.04] [0.03] [0.02]<br />
log(Wages) -0.47*** -0.46*** 0.52*** 0.02 0.28*** 0.06***<br />
[0.05] [0.01] [0.19] [0.08] [0.09] [0.04]<br />
log(Productivity) -0.46*** -0.33*** -0.91*** -0.67*** -0.44*** -0.35***<br />
[0.04] [0.01] [0.10] [0.05] [0.05] [0.03]<br />
log(Herfindahl) 0.00 0.06*** -0.08 0.03 -0.04 0.01<br />
[0.04] [0.01] [0.07] [0.03] [0.03] [0.01]<br />
Wald test 1,434.2*** 13,600.6*** 382.4*** 1,807.6*** 509.0*** 1,738.4***<br />
N 71,226 367,747 37,079 107,897 36,721 106,289<br />
Number of firms 14,017 84,083 6,594 19,826 6,524 19,522<br />
Number of failures 1,760 19,172 583 2,049 520 1,797<br />
% of total 12.6 22.8 8.8 10.3 8.0 9.2<br />
Notes: Robust st<strong>and</strong>ard errors, adjusted for clustering at the plant level (reported in brackets).<br />
Significance levels: *** p
Conclusions 123<br />
sample.<br />
However, it is possible to re-estimate 3.3 on the sample including firms<br />
that have been subject to an ownership change, taking their different nature<br />
into account by introducing a dummy variable indicating whether the firm<br />
has been subject to some form of merger or takeover (Mergeri). Results<br />
for this alternative model are displayed in columns III <strong>and</strong> IV of table 3.4.<br />
Again, results are in line with those obtained in table 3.3. Results for the<br />
sample including all firms subject to an ownership change, but excluding the<br />
merger dummy (unreported), are also in line with the results obtained using<br />
the restricted sample.<br />
Finally, in the last two columns of table 3.4 the sensitivity of the results<br />
to the estimation technique used is verified. Following Bernard <strong>and</strong> Jensen<br />
(2007a), a probit model is used to assess the determinants of exit behavior at<br />
the firm level. Focusing on the ownership variables, the associated marginal<br />
effects for the coefficients on Fori <strong>and</strong> Domi are between 0.003 <strong>and</strong> 0.007,<br />
implying that multinational firms are between 20 (for domestic MNEs active<br />
in manufacturing) <strong>and</strong> 50 percent (for foreign MNEs active in manufacturing)<br />
more likely to exit compared to national firms . Again, overall, results of the<br />
probit model are very similar to those obtained using the Cox proportional<br />
hazard model.<br />
3.7 Conclusions<br />
This paper has investigated the hypothesis that multinational companies<br />
in Belgium are more footloose than national firms. In the empirical analysis<br />
I have distinguished for nationality of ownership, using a different ownership<br />
variable for foreign <strong>and</strong> domestic multinationals. Using an unbalanced sample<br />
of 26,046 firms located in Belgium between 1996 <strong>and</strong> 2001, I find that<br />
multinational firms, both foreign <strong>and</strong> domestic, have a lower propensity to<br />
exit unconditionally; that is, they exhibit lower exit rates compared to national<br />
firms. However, multinationals also tend to be larger, more productive<br />
<strong>and</strong> pay higher wages than national firms. Since all these characteristics have<br />
been found to have significant effects on the probability of firm failure, a Cox
124 Footloose multinationals in Belgium?<br />
proportional hazard model is estimated to explicitly test for the determinants<br />
of exit in the sample.<br />
Consistent with previous studies, I find a significantly negative influence<br />
of size <strong>and</strong> productivity at the firm level on the exit probability of firms.<br />
Results further indicate that firms that pay higher wages for given levels of<br />
productivity, are more likely to exit. For the ownership variables, the results<br />
are mixed. The analysis clearly shows that foreign multinationals are more<br />
footloose than national firms of comparable size, age <strong>and</strong> productivity, both<br />
in the manufacturing <strong>and</strong> service industries. Domestic MNEs on the other<br />
h<strong>and</strong>, while more likely to exit in the manufacturing sector, do not exhibit<br />
significantly higher exit rates than national firms in the service sectors, after<br />
controlling for firm- <strong>and</strong> industry-specific variables. I have presented a<br />
number of refinements <strong>and</strong> sensitivity analyses to the model to verify the<br />
robustness of these results.<br />
These findings have clear policy implications, especially in terms of the<br />
desirability of the large impact of multinational firms on employment <strong>and</strong><br />
output generation in Belgium. The fact that domestic multinationals, while<br />
sharing most of the beneficial characteristics of foreign firms in terms of<br />
employment generation <strong>and</strong> average productivity, do not seem to share the<br />
footloose nature of their foreign counterparts in the service sector sheds a<br />
new light on recent government measures in Belgium, primarily aimed at<br />
attracting foreign investments; as well as on the recent liberalization of many<br />
service industries, such as telecommunications <strong>and</strong> electricity. Naturally, a<br />
drawback of the present analysis is that it only focuses on the extensive<br />
margin, i.e. on the shutdown decision of the firm.
Data appendix 125<br />
3.A Data appendix<br />
3.A.1 Sample selection<br />
All data are obtained from the Belfirst database (BvDEP, 2004). Exit 17<br />
occurs if a firm’s employment drops to zero in a particular year. Likewise, a<br />
firm enters the market in a certain year if there was no previous employment<br />
recorded. In order to ensure that a failure to report in a given year or a<br />
sudden drop in employment for some other reason is not misclassified as an<br />
exit, the data are subject to a number of robustness checks. First, to ensure<br />
that an entering plant is indeed “new”, I impose the condition that a plant’s<br />
year of incorporation can not differ by more than two years from the year the<br />
firm enters the market according to the entry variable 18 . Second, to avoid<br />
misclassification of “temporary exits”, all firms that re-enter within two years<br />
after having exited are dropped from the sample 19 . For similar reasons, I omit<br />
the last two years for which data are available (2002 <strong>and</strong> 2003), since I can<br />
not reliably identify entry <strong>and</strong> exit for these years. Third, I omit all firms that<br />
have been subject to takeovers, mergers or acquisitions from the sample 20 .<br />
Finally, since it is more plausible for small 21 firms to reduce employment to<br />
zero without actually exiting the market, I further limit attention to those<br />
firms employing at least 10 employees in any particular sample year.<br />
17 Although the Belfirst database reports firms’ legal status <strong>and</strong> hence also legal exit,<br />
I have chosen not to rely on this measure for two reasons. First, inspection of the data<br />
reveals that the official date associated with the legal status in the database often does<br />
not concur with the actual time the firm exits the market. Second, communications with<br />
Bureau Van Dijk made clear that although the legal status is correctly reported whenever<br />
available; many companies fail to report to the National Bank after ending their activities.<br />
18 I allow for this two year lag, since it is possible that a firm already exists as a legal<br />
entity for some time before actually starting its activities. The official year of incorporation<br />
is replaced by the actual date of entry in these cases.<br />
19 This procedure differs somewhat from that employed by Mata <strong>and</strong> Portugal (1994)<br />
who only omit firms with a temporary exit of at least two years. If a firm exits only one<br />
year <strong>and</strong> then enters again, it is considered alive in their sample, while in this case it is<br />
omitted from the sample.<br />
20 Cases are identified using the legal status variable. The categories omitted are:<br />
“Merger with another company to form a third one”, “Absorption by another company”<br />
<strong>and</strong> “Scission into several companies”.<br />
21 I consider firms employing less than 10 employees to be “small”.
126 Footloose multinationals in Belgium?<br />
3.A.2 Definition of variables<br />
Ageit<br />
Difference between year t <strong>and</strong> the official year of incorporation of the firm<br />
(replaced by the actual date of entry where necessary).<br />
Sizeit<br />
Average number of employees (full-time equivalents, fte) at the firm level in<br />
year t.<br />
Wageit<br />
“Remunerations, social security costs <strong>and</strong> pensions” per employee (fte) of<br />
firm i in year t, measured in thous<strong>and</strong>s of euros. The wage variable is expressed<br />
in real terms, using three-digit producer price indices to deflate the<br />
monetary values.<br />
Prodit<br />
“Net Value Added” per employee (fte) for firm i in year t, measured in<br />
thous<strong>and</strong>s of euros. Labor productivity is expressed in real terms, using<br />
three-digit producer price indices to deflate the monetary values.<br />
Herfjt<br />
Herfindahl concentration index, measured at three-digit NACE level <strong>and</strong> calculated<br />
for the full sample of firms in the Belfirst database with available data<br />
on turnover. Market shares are calculated on the basis of firm <strong>and</strong> industry<br />
turnover.<br />
Fori<br />
Foreign multinational ownership variable. A firm is considered to be foreignowned<br />
if it has some foreign ownership.<br />
Domi<br />
Domestic multinational ownership variable. A firm is identified as a domestic<br />
MNE if it has subsidiaries in countries other than Belgium <strong>and</strong> it is not<br />
foreign-owned.
Chapter 4<br />
Multinational firms, Research<br />
Effort <strong>and</strong> Innovative Output:<br />
An Integrated Approach<br />
4.1 Introduction<br />
While the literature on the relationship between firms’ international activities<br />
<strong>and</strong> their productivity abounds 1 , it remains largely silent on the sources<br />
of the productivity advantages associated with firms’ global integration. One<br />
way for firms to achieve productivity increases, is through the accumulation<br />
of knowledge or innovative output (Castellani <strong>and</strong> Zanfei, 2007).<br />
The central purpose of this paper is to investigate whether firms’ innovative<br />
output differs according to their global engagement status. The analysis is<br />
carried out using cross-section data from the fourth Community Innovation<br />
Survey for Belgium (CIS4), pertaining to the years 2002-2004. To estimate<br />
the firm-level innovation production function, I rely on the methodology<br />
implemented by Mairesse <strong>and</strong> Mohnen (2004).<br />
The present analysis is most closely related to the works of Castellani <strong>and</strong><br />
Zanfei (2006); Criscuolo, Haskel <strong>and</strong> Slaughter (2005) <strong>and</strong> Frenz <strong>and</strong> Ietto-<br />
1 For reviews of this extensive literature, I refer to Greenaway <strong>and</strong> Kneller (2007) <strong>and</strong><br />
Wagner (2007). Important theoretical contributions in this field include Melitz (2003) <strong>and</strong><br />
Helpman et al. (2004).<br />
129
130 Multinationals, research effort <strong>and</strong> innovative output<br />
Gillies (2007). While the first use CIS2 data for Italy, the latter two both<br />
use CIS data for the UK (waves 2 <strong>and</strong> 3). Controlling for firm size <strong>and</strong><br />
sector of activity, these papers generally find a positive correlation between<br />
firms’ global engagement status <strong>and</strong> their innovative output. Criscuolo et<br />
al. (2005) additionally provide evidence of a positive relationship between<br />
firm-level research inputs <strong>and</strong> their innovative output.<br />
As noted by Mairesse <strong>and</strong> Mohnen (2002), analyzing innovative output<br />
is not very different from analyzing firm-level production, since innovative<br />
output is simply a function of innovative inputs (research effort) at the firm<br />
level. Moreover, drawing the analogy with the traditional production function<br />
literature further, simultaneity of research inputs <strong>and</strong> output introduces<br />
endogeneity issues. Mairesse <strong>and</strong> Mohnen (2004) take this endogeneity of research<br />
inputs in the innovation output production function into account by<br />
instrumenting for research inputs. Their approach follows that of Crépon,<br />
Duguet <strong>and</strong> Mairesse (1998), who introduced an empirical model taking into<br />
account both the selection issues inherent in the CIS data, as well as the endogeneity<br />
of research inputs in the innovative output function. Since similar<br />
selection issues apply to both innovative inputs <strong>and</strong> outputs, instrumenting<br />
for research inputs is achieved by once again estimating a Heckman selection<br />
model.<br />
The present analysis contributes to the literature in two important ways.<br />
First, from a methodological point of view, the empirical analysis offers an<br />
integrated approach. Potential endogeneity of research inputs is explicitly<br />
taken into account in the innovation output function. Moreover, by applying<br />
a two-step estimation procedure as in Mairesse <strong>and</strong> Mohnen (2004), I am<br />
able to distinguish between the indirect (through higher R&D spending) <strong>and</strong><br />
direct effect of firms’ global integration on innovative output.<br />
Second, from a policy perspective, the analysis may yield important insights.<br />
The Belgian economy is characterized by a high dependence on multinational<br />
firms, not only in terms of employment <strong>and</strong> sales, but also in terms<br />
of innovative performance. Cincera et al. (2006) find that seventy percent of<br />
all patents invented in Belgium are owned by foreign firms, either directly
Introduction 131<br />
(through foreign assignees) or indirectly (through a Belgian subsidiary of a<br />
foreign firm).<br />
Teirlinck (2005a) further shows that the majority of total R&D spending in<br />
Belgium originated in foreign-owned firms in 2001. He then goes on to show<br />
that this dominance of foreign firms in total R&D spending is related to their<br />
size <strong>and</strong> sector distribution. Furthermore, his findings indicate that, once the<br />
technology intensity of the sector is taken into account, location motives for<br />
R&D are not different for foreign <strong>and</strong> domestic firms. These results suggest<br />
that government policy aimed at stimulating or attracting R&D does not<br />
necessarily need to target foreign firms specifically. Rather, policy should<br />
be aimed at fostering entrepreneurship in certain high-technology sectors.<br />
However, one could argue that what matters to long-term growth is not<br />
research inputs, but rather innovative output; as a driver of total factor<br />
productivity growth.<br />
The empirical results can be summarized as follows. Exporters <strong>and</strong> multinational<br />
firms (both foreign <strong>and</strong> home-based) are found to be more likely<br />
to generate innovative sales than national firms, both unconditionally (not<br />
taking R&D inputs into account) <strong>and</strong> conditional upon their engagement in<br />
R&D investment. However, taking into account that globally engaged firms<br />
exhibit different R&D spending patterns, i.e. conditional on research effort;<br />
only exporters are found to be significantly more likely to generate innovative<br />
sales. Foreign affiliates of multinational firms are not found to be more<br />
innovative than purely domestic (national) firms. Home-based multinationals<br />
are significantly less likely to innovate, after controlling for their higher<br />
average inputs. In terms of innovation intensity (the share of innovative sales<br />
in total sales), globally active firms are not found to behave differently from<br />
national firms, either conditionally or unconditionally. Distinguishing between<br />
high-technology <strong>and</strong> low-technology sectors further shows that there<br />
is an innovation premium associated with firms’ international activities in<br />
high-technology sectors, conditional on their research intensity; while in lowtechnology<br />
sectors, a negative innovation premium emerges.
132 Multinationals, research effort <strong>and</strong> innovative output<br />
A caveat is worth noting here. Aw, Roberts <strong>and</strong> Winston (2007) provide<br />
evidence showing that firms jointly make their decisions to innovate <strong>and</strong> to<br />
engage in international markets. Costantini <strong>and</strong> Melitz (2007) build a dynamic<br />
model that captures the self-selection of more productive firms into<br />
international markets, the joint export <strong>and</strong> innovation decisions <strong>and</strong> the continuing<br />
innovation of firms following entry into global markets. From this<br />
point of view, innovation can be linked to exports <strong>and</strong> productivity in two<br />
ways. <strong>Firm</strong>s can innovate prior to entry on international markets, enabling<br />
them to gain the productivity advantage needed to overcome the sunk cost<br />
associated with global engagement. On the other h<strong>and</strong>, firms’ international<br />
experiences may induce further innovative activities (for instance stimulated<br />
by contacts with foreign buyers), hence reinforcing their productivity advantages.<br />
Since the data used in this paper are cross-sectional in nature, I am not able<br />
to investigate to what extent results obtained here are driven by selection<br />
effects. Several papers investigate the selection hypothesis specifically for<br />
innovative output, with mixed results. Wakelin (1998), Basile (2001) <strong>and</strong><br />
Cassiman <strong>and</strong> Golovko (2007) study the impact of innovative output (product<br />
or process innovation) on both the export status <strong>and</strong> intensity of firms in<br />
the UK, Italy <strong>and</strong> Spain respectively. Their empirical findings support the<br />
selection hypothesis. Damijan, Kostevc <strong>and</strong> Polanec (2008) apply matching<br />
techniques to Slovenian firm-level data to investigate the dual relationship<br />
between firms’ innovation <strong>and</strong> export decisions. Their results lend support<br />
to the learning hypothesis (although only for process innovations), while<br />
they find no evidence of selection effects. Given these mixed results, results<br />
obtained here should be interpreted as correlations <strong>and</strong> can not be interpreted<br />
as causal effects of firms’ global activities on innovation output.<br />
The rest of the paper is organized as follows. Section 4.2 formalizes the<br />
concept of the innovation production function, which will serve as the estimating<br />
framework in the empirical analysis. Section 4.3 discusses the data<br />
<strong>and</strong> relevant empirical facts, while section 4.4 presents the results. Section 4.5<br />
concludes.
The innovation production function 133<br />
4.2 The innovation production function<br />
As noted by Mairesse <strong>and</strong> Mohnen (2002, 2004), analyzing firm-level innovative<br />
output is not very different from analyzing regular production output.<br />
<strong>Firm</strong>-level innovative output, like regular production output, results from<br />
the transformation of certain inputs into output; a relationship that can be<br />
analyzed in the context of a production function. In its most general form,<br />
this production function looks as follows:<br />
Yi = Aif(Xi) (4.1)<br />
where Yi represents physical (innovative) output of firm i <strong>and</strong> Xi are<br />
firm-level (research) inputs, measured in quantities. Ai is the unobservable<br />
firm-level productivity term, labeled “Innovativity” or “Innovativeness” by<br />
Mairesse <strong>and</strong> Mohnen (2002). Since firm-level quantities of inputs <strong>and</strong> outputs<br />
are generally unavailable, estimation of (4.1) in the traditional production<br />
function literature is usually achieved by deflating inputs <strong>and</strong> outputs<br />
using appropriate (industry-level) price indices.<br />
However, estimating equation (4.1) for a balanced panel of firms, using Ordinary<br />
Least Squares (OLS) raises a number of methodological issues, comparable<br />
to the issues arising when estimating traditional production functions 2 .<br />
First, to the extent that less innovative firms are more likely to exit, the<br />
use of a balanced panel of firms introduces a selection bias in the sample.<br />
Second, since firms simultaneously decide on the allocation of inputs <strong>and</strong><br />
production of output, endogeneity issues arise when (4.1) is estimated using<br />
OLS. Third, if input or output markets are characterized by imperfect<br />
competition, the use of industry-wide deflators rather than firm-level prices<br />
will result in an omitted output <strong>and</strong> input price bias. Finally, if firms produce<br />
multiple products (or different products), which potentially differ in<br />
terms of their production technology <strong>and</strong> dem<strong>and</strong>, an additional bias will be<br />
introduced in traditional TFP estimates.<br />
2 For a recent review of the different methodological issues arising when estimating<br />
output production functions, <strong>and</strong> the existing parametric <strong>and</strong> semiparametric techniques<br />
designed to overcome them, I refer to Van Beveren (2007b).
134 Multinationals, research effort <strong>and</strong> innovative output<br />
Several parametric <strong>and</strong> semiparametric estimators have been introduced in<br />
the production function literature to take some of these issues into account.<br />
Apart from the more conventional techniques of Instrumental Variable (IV)<br />
or Generalized Method of Moments (GMM) estimation, several estimators<br />
have been specifically developed for the estimation of production functions.<br />
Olley <strong>and</strong> Pakes (1996) <strong>and</strong> Levinsohn <strong>and</strong> Petrin (2003) introduce a semiparametric<br />
estimator. They are able to resolve the simultaneity bias by<br />
using a proxy variable to substitute for unobserved productivity (investment<br />
in Olley <strong>and</strong> Pakes (1996) <strong>and</strong> materials in Levinsohn <strong>and</strong> Petrin (2003)).<br />
Both estimators are also able to correct for the selection bias, by taking<br />
firms’ survival probability into account in the estimation procedure. De<br />
Loecker (2007) extends the algorithm developed by Olley <strong>and</strong> Pakes to take<br />
imperfect competition in output markets, as well as multi-product firms into<br />
account.<br />
Unfortunately, while data on firm-level output <strong>and</strong> inputs tend to be available<br />
in relatively long time-series, often for the full population of firms in a<br />
particular country; innovation data generally suffer from a number of important<br />
drawbacks, rendering consistent estimation of (4.1) even more problematic.<br />
The discussion here will focus specifically on the Community Innovation<br />
Survey data. While other firm-level data sets on innovation exist 3 ,<br />
the CIS data have the important advantage of being uniform (with some<br />
exceptions) across European countries. Moreover, the CIS data have been<br />
used in a growing number of contributions, some recent examples include<br />
Cassiman <strong>and</strong> Veugelers (2006) <strong>and</strong> Griffith, Huergo, Mairesse <strong>and</strong> Peters<br />
(2006) 4 . The Community Innovation Survey is carried out every two years<br />
in all European Union (EU) member countries, under the supervision of the<br />
authorized national agencies. <strong>Firm</strong>-level data (anonymous or not) can only<br />
be obtained through the various national statistical agencies collecting the<br />
data 5 .<br />
3 See for instance Cassiman <strong>and</strong> Golovko (2007) for Spain.<br />
4 For an overview of the general characteristics of the CIS data <strong>and</strong> its advantages <strong>and</strong><br />
drawbacks, I refer to Smith (2005). Hall <strong>and</strong> Mairesse (2006) provide a (selective) review<br />
of a number of recent empirical papers applying the CIS data.<br />
5 Eurostat only provides the data in micro-aggregated form.
The innovation production function 135<br />
When using the CIS data to estimate (4.1), several difficulties emerge.<br />
First, the CIS data are cross-sectional in nature. While data of different<br />
waves can be merged, this results in limited sample sizes, since sampling is<br />
performed independently for each wave (i.e. firms are not always observed<br />
in each wave). Moreover, while some researchers have been able to obtain<br />
firm-level data for different waves (eg. Frenz <strong>and</strong> Ietto-Gillies, 2007; Criscuolo<br />
et al., 2005), most studies use cross-section data for a particular wave. This<br />
limits the available estimators to those that do not require information on<br />
firm dynamics <strong>and</strong> hence rules out the use of GMM <strong>and</strong> the semi-parametric<br />
estimators of Olley <strong>and</strong> Pakes, Levinsohn <strong>and</strong> Petrin <strong>and</strong> De Loecker.<br />
The use of cross-section data has an important additional drawback. By<br />
definition cross-section data contain no information on firm entry <strong>and</strong> exit.<br />
In fact, in the Belgian case, only firms that continued their activities throughout<br />
the period 2002-2004 were selected for the CIS4 population (Teirlinck,<br />
2005b). Using different waves of the questionnaire would still require additional<br />
external information sources on entry <strong>and</strong> exit of firms to identify firm<br />
dynamics, since firms are not tracked across different waves. Hence, potential<br />
selection effects can generally not be accounted for.<br />
Another issue researchers encounter when estimating (4.1) specifically for<br />
innovative output is the lack of suitable price deflators (or ideally, firm-level<br />
prices) of innovative inputs <strong>and</strong> outputs. On the output side, using industrylevel<br />
output price indices to deflate innovate sales imposes the assumption<br />
that prices of new products evolve in the same way as regular output prices.<br />
On the input side, finding appropriate deflators is even more problematic,<br />
specifically when use is made of the CIS data. <strong>Firm</strong>s in the CIS questionnaire<br />
are asked to report their total internal R&D expenditures in a particular<br />
year. These expenditures include personnel costs, material costs <strong>and</strong> capital<br />
expenditures on materials <strong>and</strong> equipment specifically for R&D. Apart from<br />
the fact that no price indices exist specifically for R&D personnel, material<br />
costs <strong>and</strong> investment; no information is available in the CIS data on the shares<br />
of each of these expenditures in firms’ total research spending, rendering any<br />
weighting of these price indices to obtain a composite index arbitrary.
136 Multinationals, research effort <strong>and</strong> innovative output<br />
Moreover, since firms’ expenditures reported in the CIS questionnaire, include<br />
information on both variable inputs (personnel <strong>and</strong> materials costs)<br />
as well as fixed inputs (capital expenditures), little can be said about the<br />
direction of the biases introduced by the methodological issues listed above.<br />
For instance, Levinsohn <strong>and</strong> Petrin (2003) show, in a general production<br />
function setup, where labor <strong>and</strong> capital are positively correlated, that the<br />
simultaneity bias will lead to an upward bias in the coefficient of the variable<br />
input (labor) <strong>and</strong> a downward bias for the fixed input coefficient (capital).<br />
Since variable <strong>and</strong> fixed inputs are not separable when use is made of the<br />
CIS-variable on internal research spending, it is not possible to make any<br />
predictions on the direction of the biases introduced.<br />
Apart from the drawbacks of the data as compared to regularly available<br />
production function data, the CIS data possess another important characteristic<br />
that needs to be taken into account. Only firms that had ongoing,<br />
successful or ab<strong>and</strong>oned innovation activities during the period considered<br />
are asked to fill out the full questionnaire. Non-innovators on the other h<strong>and</strong>,<br />
only need to answer particular general questions 6 . As a consequence, the CIS<br />
data are characterized by selection issues, unrelated to the selection issues<br />
introduced by the use of a sample of continuing firms.<br />
Given these important data constraints, researchers interested in estimating<br />
(4.1) can essentially only rely on IV estimation to correct for the simultaneity<br />
bias. A natural choice of model to take the selection issues inherent in<br />
the questionnaire into account is the Heckman selection model. Crépon et al.<br />
(1998) were the first to propose such a framework, linking research inputs to<br />
innovative output (<strong>and</strong> innovative output to total factor productivity). Their<br />
approach corrects for simultaneity by instrumenting for R&D inputs in the<br />
innovation output production function <strong>and</strong> tackles selection issues through<br />
estimation of a generalized tobit (Heckman) model.<br />
Many recent studies implement the approach of Crépon et al. (1998), examples<br />
include Benavente (2006); Griffith et al. (2006); Lööf <strong>and</strong> Heshmati<br />
6 For Belgium, these are mostly related to general firm characteristics; such as employment,<br />
turnover <strong>and</strong> group membership.
The innovation production function 137<br />
(2004) <strong>and</strong> Van Leeuwen <strong>and</strong> Klomp (2006). The description of the model<br />
given below follows most closely that of Mairesse <strong>and</strong> Mohnen (2004), who<br />
implement the approach of Crépon et al. (1998) using French CIS3 data.<br />
The basic model proposed by Crépon et al. (1998) <strong>and</strong> implemented by<br />
Mairesse <strong>and</strong> Mohnen (2004) consists of a two-stage model. In the first<br />
stage, firms in the sample decide on whether or not they invest in R&D <strong>and</strong>,<br />
if they invest, on the scale of their investment. Estimation of the first stage is<br />
achieved using a generalized tobit (Heckman) model. In the second stage of<br />
the model, innovative output is related to firms’ research effort (instrumented<br />
from the first stage) <strong>and</strong> other factors. Since the present analysis focuses on<br />
innovative sales to measure innovative output 7 , selection issues apply also to<br />
the second stage of the model. Hence, similar to the first stage, a generalized<br />
tobit model is estimated; allowing for a selection <strong>and</strong> innovation intensity<br />
equation.<br />
Formally, stage 1 of the model can be summarized as follows (Greene, 2008;<br />
Verbeek, 2000):<br />
⎧<br />
⎪⎨<br />
⎪⎩<br />
s ∗ rd,i = x′ 1<br />
rd,i β 1 rd,i + ε1 rd,i<br />
y ∗ rd,i = x′ 2<br />
rd,i β 2 rd,i + ε2 rd,i<br />
y ∗ rd,i = yrd,i; srd,i = 1 if s ∗ rd,i<br />
> 0<br />
yrd,i not observed; srd,i = 0 if s ∗ rd,i<br />
≤ 0<br />
(4.2)<br />
where yrd,i refers to firm i’s R&D intensity, which is only observed for those<br />
firms that continuously engaged in R&D between 2002 <strong>and</strong> 2004 <strong>and</strong> srd,i<br />
is the selection variable, i.e. a dummy variable equal to one for continuous<br />
are their corresponding latent variables, xrd,i<br />
R&D performers. y∗ rd,i <strong>and</strong> s∗rd,i refers to the explanatory variables, βrd,i to the coefficients of the model <strong>and</strong><br />
εrd,i are the (correlated) error terms.<br />
7 The CIS data contain a number of other innovation output indicators, such as whether<br />
a firm has introduced a product or process innovation or has applied for a patent. I have<br />
chosen to rely only on the innovative sales variable here because of its close relation to<br />
regular production function output. Mairesse <strong>and</strong> Mohnen (2004) implement the approach<br />
outlined here for the different output measures available in the CIS data.
138 Multinationals, research effort <strong>and</strong> innovative output<br />
Similarly, stage 2 of the model looks as follows:<br />
⎧<br />
⎪⎨<br />
⎪⎩<br />
s ∗ inn,i = x′ 3<br />
inn,i β 3 inn,i + ε3 inn,i<br />
y ∗ inn,i = x′ 4<br />
inn,i β 4 inn,i + ε4 inn,i<br />
y ∗ inn,i = yinn,i; sinn,i = 1 if s ∗ inn,i<br />
> 0<br />
yinn,i not observed; sinn,i = 0 if s ∗ inn,i<br />
≤ 0<br />
(4.3)<br />
where yinn,i refers to firm i’s innovative sales intensity, which is only observed<br />
for those firms that generated innovative sales between 2002 <strong>and</strong> 2004,<br />
<strong>and</strong> sinn,i is the selection variable, i.e. a dummy variable equal to one for<br />
innovators. y ∗ inn,i <strong>and</strong> s ∗ inn,i are their corresponding latent variables, xinn,i<br />
refers to the explanatory variables, βinn,i to the coefficients of the model <strong>and</strong><br />
εinn,i are the (correlated) error terms.<br />
Since only firms with ongoing, successful or ab<strong>and</strong>oned innovation activities<br />
are required to fill out the full CIS questionnaire, the choice of which<br />
variables to include in x ′ 1<br />
rd,i is limited to those variables available for all firms.<br />
Variables included are firm size (measured as the natural logarithm of employment),<br />
two-digit sector dummies, three region dummies <strong>and</strong> firms’ global<br />
engagement status. Four (exclusive) groups of firms are distinguished: (1)<br />
firms that are only active on the domestic market (the baseline category,<br />
captured in the regression constant), (2) firms that export but are not part<br />
of a multinational firm, (3) foreign affiliates of multinational firms <strong>and</strong> (4)<br />
home-based multinationals.<br />
Since firms with reported R&D expenditures are by definition innovators,<br />
i.e. they are required to fill out the full questionnaire, more variables are available<br />
to explain the scale of investment in research at the firm level. Hence,<br />
x ′ 2<br />
rd,i includes, apart from the sector, region <strong>and</strong> global engagement dummies,<br />
a dummy indicating whether the firm engaged in research cooperation activities<br />
between 2002 <strong>and</strong> 2004, as well as a funding dummy, equal to one if the<br />
firm has acquired regional, national or EU funding during the sample period.<br />
Employment is excluded from x ′ 2<br />
rd,i for identification purposes 8 .<br />
8 If employment is included in the second stage of the model, its coefficient is not<br />
significant. This is not surprising, since research effort is specified relative to total sales.
The innovation production function 139<br />
Similarly to the first stage, x ′ 3<br />
rd,i <strong>and</strong> x ′ 4<br />
rd,i include two-digit sector dummies,<br />
three region dummies <strong>and</strong> firms’ global engagement status. In addition, the<br />
predicted value of the logarithm of firms’ R&D intensity from the first stage,<br />
y∗ rd,i , enters both equations of the generalized tobit model. As in the first<br />
stage, employment only enters the selection equation of the model. No other<br />
explanatory variables are included in the second stage. This implies that the<br />
cooperation <strong>and</strong> funding dummies act as exclusion restrictions in the model<br />
<strong>and</strong> are essentially assumed to affect innovative sales only indirectly through<br />
their effect on research effort (Mairesse <strong>and</strong> Mohnen, 2004).<br />
Whether these are reasonable assumptions is open for debate. Both cooperation<br />
<strong>and</strong> funding can be considered as sources of external finance to<br />
the firm for its research activities. From this point of view, they are likely<br />
to enhance firms’ own research efforts, but will not necessarily contribute<br />
to the success or failure of a particular research project over <strong>and</strong> beyond<br />
their impact on research spending. On the other h<strong>and</strong>, to the extent that<br />
cooperation generates knowledge spillovers or other externalities, its impact<br />
on innovative output may not be limited to its effect on research spending.<br />
Due to the innovation-centric nature of the CIS questionnaire <strong>and</strong> the crosssectional<br />
nature of the data, finding good instruments for research effort is a<br />
difficult task, that certainly merits more attention in future work.<br />
Apart from the global engagement dummies, which are defined as in Castellani<br />
<strong>and</strong> Zanfei (2006) <strong>and</strong> Criscuolo et al. (2005), the specification of the<br />
model outlined above closely follows that of Mairesse <strong>and</strong> Mohnen (2004).<br />
However, a number of important differences with respect to their model<br />
are worth noting here. First, in the R&D selection equation, Mairesse <strong>and</strong><br />
Mohnen (2004) additionally include dem<strong>and</strong> pull <strong>and</strong> cost push indicators.<br />
However, these variables are not available for all firms in the CIS4 data for<br />
Belgium. Mairesse <strong>and</strong> Mohnen (2004) additionally include firms’ weighted<br />
market shares <strong>and</strong> a diversification index, calculated using data from external<br />
sources, in the first stage of the model. They further include a number<br />
of firm-specific knowledge sources, available in the CIS data, in the R&Dintensity<br />
equation.
140 Multinationals, research effort <strong>and</strong> innovative output<br />
Further, the measure of R&D-intensity used by Mairesse <strong>and</strong> Mohnen<br />
(2004) is defined as the natural logarithm of R&D per employee. I have<br />
chosen to define research effort relative to firm turnover. As noted above,<br />
no suitable (industry-level) price indices exist specific to research inputs <strong>and</strong><br />
outputs. By weighting innovative inputs <strong>and</strong> outputs by total turnover at<br />
the firm-level, I am able to take out at least that part of firm-level price<br />
variation that is common to regular output as well as research inputs <strong>and</strong><br />
outputs. Weighting the input variable by the number of employees, while<br />
innovative output is measured relative to sales; as is done in Mairesse <strong>and</strong><br />
Mohnen (2004) implies that this general price variation is taken out on the<br />
output side, but not on the input side, introducing an additional bias in the<br />
estimation results. I therefore rely on firm-level sales to weight both research<br />
inputs <strong>and</strong> innovative outputs in the empirical specification.<br />
The approach of Crépon et al. (1998) <strong>and</strong> Mairesse <strong>and</strong> Mohnen (2004)<br />
offers a number of important advantages over existing work analyzing the<br />
impact of firms’ global activities on their innovative performance (eg. Castellani<br />
<strong>and</strong> Zanfei, 2006; Criscuolo, Haskel <strong>and</strong> Slaughter, 2005 <strong>and</strong> Frenz <strong>and</strong><br />
Ietto-Gillies, 2007). First, from a methodological point of view, the empirical<br />
model offers an integrated approach, taking both the selection issues inherent<br />
in the CIS data, as well as potential endogeneity of research inputs in the<br />
innovation output function into account. While Criscuolo et al. (2005) take<br />
the endogeneity of R&D inputs in the innovative production function into<br />
account 9 , they do not take the selection issues noted above explicitly into account.<br />
Second, by applying a two-step estimation procedure, it is possible to<br />
distinguish between the indirect (through higher R&D spending) <strong>and</strong> direct<br />
effect of firms’ global integration on innovative output.<br />
9 Criscuolo et al. (2005) instrument for R&D personnel (their measure of research inputs)<br />
using the four-digit (NACE) industry averages of R&D personnel (or alternatively,<br />
the share of R&D personnel in the total employee base), constructed using CIS2-data,<br />
excluding those firms that reappeared in CIS3. They obtain similar results as in their<br />
baseline specification when instrumenting for research inputs.
Data <strong>and</strong> empirical facts 141<br />
4.3 Data <strong>and</strong> empirical facts<br />
The innovation data used in the empirical analysis are taken from the<br />
Community Innovation Survey (CIS4) for Belgium <strong>and</strong> are obtained through<br />
the Belgian Science Policy 10 . The CIS data contain limited information on<br />
ownership. The data allow for identification of all firms that are part of a<br />
group <strong>and</strong> also lists the country of origin of the headquarters of that group.<br />
Additionally, the database contains information on firm-level exports in 2004.<br />
This enables me to identify exporters <strong>and</strong> foreign affiliates of multinational<br />
firms. To verify whether firms are part of a home-based multinational (i.e. a<br />
Belgian firm with subsidiaries abroad), I use subsidiary information obtained<br />
from Belfirst 11 (BvDEP, 2006). For a detailed description of the sample selection<br />
<strong>and</strong> definitions of variables, I refer to the data appendix (section 4.A).<br />
The CIS4 questionnaire pertains to the years 2002 to 2004. However, qualitative<br />
questions are asked only once for the entire period, while quantitative<br />
measures apply to the year 2004. Only employment <strong>and</strong> turnover are reported<br />
for 2002 <strong>and</strong> 2004. Hence, all innovation data are cross-sectional in<br />
nature. The sample used in the empirical analysis consists of 2,988 firms.<br />
Table 4.1 gives the sector distribution of the sample by ownership type. The<br />
last column in the table lists total R&D spending by sector. Sectors are<br />
classified according to the NACE (Rev. 1.1) classification 12 .<br />
There are important differences in the sector distribution of firms in the<br />
manufacturing sector according to the degree of internationalization. While<br />
domestic firms <strong>and</strong> exporters dominate in Metallic products <strong>and</strong> in Food, beverages<br />
<strong>and</strong> tobacco, foreign affiliates of multinational firms feature predominantly<br />
in Chemicals <strong>and</strong> Metallic products. Home-based multinationals, like<br />
domestic firms, tend to be active in Food, beverages <strong>and</strong> tobacco, but also in<br />
10 I would like to thank Manu Monard, Peter Teirlinck <strong>and</strong> the CFS-STAT Commission<br />
for allowing me to access the firm-level data; for answering questions related to the data<br />
used <strong>and</strong> for their hospitality during my visits there.<br />
11 The Belfirst database contains ownership <strong>and</strong> accounting information on the population<br />
of Belgian firms. Other recent papers using this database include Abraham, Konings<br />
<strong>and</strong> Vanormelingen (2007b); Konings (2008) <strong>and</strong> Monfort, V<strong>and</strong>enbussche <strong>and</strong> Forlani<br />
(2008).<br />
12 The NACE classification can be downloaded from the Eurostat Ramon server at<br />
http://ec.europa.eu/eurostat/ramon.
Table 4.1: Sector distribution of firms by firm type<br />
142 Multinationals, research effort <strong>and</strong> innovative output<br />
Notes: Four firm types are distinguished: (1) National firms: do not export <strong>and</strong> are not part of a<br />
multinational firm; (2) Exporters: firms that export, but are not part of a multinational; (3) Foreign<br />
affiliates of multinational firms; (4) Home-based (domestic) multinational firms. Total internal R&D<br />
expenditures in 2004 of all continuous R&D performers between 2002-2004. Variables are defined in the<br />
data appendix (section 4.A.2).<br />
National Foreign Domestic Total<br />
<strong>Firm</strong>s Exporters MNEs MNEs Int. R&D<br />
C. Mining <strong>and</strong> quarrying 4 8 6 2 538<br />
CA. Of energy producing materials 2 0 0 0 0<br />
CB. Excluding energy producing materials 2 8 6 2 538<br />
D. Manufacturing 272 546 321 99 731,510<br />
DA. Food, beverages <strong>and</strong> tobacco 62 71 32 13 20,343<br />
DB. Textiles 25 55 13 10 11,878<br />
DC. Leather (products) 1 3 1 0 2,636<br />
DD. Wood (products) 8 26 2 4 589<br />
DE. Pulp <strong>and</strong> paper 28 67 26 9 5,942<br />
DF. Coke <strong>and</strong> petroleum (products) 2 2 2 0 2,400<br />
DG. Chemicals 5 27 48 10 121,069<br />
DH. Rubber <strong>and</strong> plastics 7 23 20 6 43,848<br />
DI. Other non-metallic products 9 30 23 7 9,604<br />
DJ. Metallic products 56 90 51 9 34,587<br />
DK. Machinery <strong>and</strong> equipment n.e.c. 15 44 32 12 67,968<br />
DL. Electrical <strong>and</strong> optical equipment 24 39 28 12 336,433<br />
DM. Transport equipment 11 25 32 4 71,334<br />
DN. Manufacturing n.e.c. 19 44 11 3 2,880<br />
E-Q. Services <strong>and</strong> related activities 786 432 438 74 1,065,743<br />
E. Electricity, gas <strong>and</strong> water supply 5 0 2 0 14,716<br />
F. Construction 141 31 17 11 175<br />
G. Wholesale <strong>and</strong> retail trade 262 191 216 20 32,424<br />
H. Transport, storage <strong>and</strong> communication 162 95 62 9 113,445<br />
J. Financial intermediation 36 12 32 12 52,282<br />
K. Other business activities 180 103 109 22 852,702<br />
Total 1,062 986 765 175 1,797,791
Data <strong>and</strong> empirical facts 143<br />
Machinery <strong>and</strong> Electrical <strong>and</strong> optical equipment. The majority of domestic<br />
multinationals <strong>and</strong> exporters are active in the manufacturing sector, while<br />
foreign firms <strong>and</strong> national firms dominate in services. Within the tertiary<br />
sector, Wholesale <strong>and</strong> retail trade <strong>and</strong> Other business activities are the most<br />
populated sectors.<br />
In terms of R&D spending, the two largest sectors are Other business<br />
activities, which includes the IT sector <strong>and</strong> Electrical <strong>and</strong> optical equipment,<br />
together they account for more than 60 percent of R&D spending in the<br />
sample. Chemicals, Transport <strong>and</strong> communication <strong>and</strong> Transport equipment<br />
are also included in the top five. Finally, from the table, it is clear that<br />
R&D spending is highly concentrated in a few sectors; the top five sectors<br />
together account for more than eighty percent of total research spending in<br />
the sample.<br />
Table 4.2 summarizes the sample distribution of firms according to their<br />
global engagement status. In terms of the number of firms, domestic firms<br />
<strong>and</strong> firms that only export (but are not part of a multinational firm) dominate,<br />
together these firm account for almost 70 percent of the total number<br />
of firms. However, in terms of employment <strong>and</strong> output generation, <strong>and</strong> also<br />
for innovative inputs (internal R&D expenditures) <strong>and</strong> output (innovative<br />
sales), multinational firms clearly dominate. Taken together, foreign <strong>and</strong><br />
home-based multinationals generated 67 percent of total employment, 80<br />
percent of turnover, 62 percent of internal R&D expenditures <strong>and</strong> 87 percent<br />
of total innovative sales in the sample.<br />
From table 4.2 it is clear that globally active firms dominate the sample,<br />
both in terms of total research efforts as well as total innovative sales.<br />
These preliminary facts suggest that the innovation premium attributed to<br />
globally active firms (whether exporters or multinationals) in previous work<br />
(eg. Castellani <strong>and</strong> Zanfei, 2006; Criscuolo et al., 2005; <strong>and</strong> Frenz <strong>and</strong> Ietto-<br />
Gillies, 2007) might be (partly) related to these firms’ different spending patterns<br />
on R&D. By estimating the innovation production function along the<br />
lines suggested by Mairesse <strong>and</strong> Mohnen (2004) <strong>and</strong> described in section 4.2,<br />
I am able to distinguish between the indirect (through R&D spending) <strong>and</strong>
144 Multinationals, research effort <strong>and</strong> innovative output<br />
National Foreign Domestic<br />
<strong>Firm</strong> type firms Exporters MNEs MNEs<br />
Number of firms 1,062 986 765 175<br />
(% of total) (35.54) (33.00) (25.60) (5.86)<br />
Employment (fte) 77,456 51,045 157,316 103,914<br />
(% of total) (19.87) (13.10) (40.37) (26.66)<br />
Turnover (xe1,000) 16,300,000 12,000,000 78,800,000 40,200,000<br />
(% of total) (11.08) (8.15) (53.48) (27.30)<br />
Internal R&D (xe1,000) 577,191 93,899 632,698 494,003<br />
(% of total) (32.11) (5.22) (35.19) (27.48)<br />
Innovative sales (xe1,000) 1,119,372 1,235,825 11,900,000 4,456,683<br />
(% of total) (5.98) (6.60) (63.61) (23.81)<br />
Notes: Four firm types are distinguished: (1) National firms: do not export <strong>and</strong><br />
are not part of a multinational firm; (2) Exporters: firms that export, but are not<br />
part of multinational; (3) Foreign affiliates of multinational firms; (4) Home-based<br />
multinational firms. Variables are defined in the data appendix (section 4.A.2).<br />
All data pertain to the year 2004 <strong>and</strong> represent sample totals (percentage of<br />
total).<br />
Table 4.2: Sample distribution according to global engagement status<br />
direct impact of firms’ global engagement status on innovative output. However,<br />
before proceeding to the empirical results, table 4.3 presents summary<br />
statistics for the key variables used in the empirical analysis, again distinguished<br />
according to firms’ global engagement status.<br />
A number of striking features emerge from table 4.3. Contrary to expectations,<br />
firms that only export are (significantly) smaller than national firms<br />
(on average). Multinational firms on the other h<strong>and</strong>, are on average much<br />
larger than both domestic firms <strong>and</strong> exporters, which also explains their<br />
large impact on total employment <strong>and</strong> output generation in the sample (cfr.<br />
table 4.2). Moreover, both exporters <strong>and</strong> multinational firms (foreign <strong>and</strong><br />
home-based) show a higher propensity to engage continuously in R&D <strong>and</strong><br />
to generate innovative sales, compared to national firms. In relative terms,<br />
home-based multinationals show the highest proportion of R&D spending<br />
firms (43 percent) as well as the highest proportion of firms generating innovative<br />
sales (49 percent).
Data <strong>and</strong> empirical facts 145<br />
National Foreign Domestic<br />
Variables <strong>Firm</strong>s Exporters MNEs MNEs<br />
All firms<br />
N 1,062 986 765 175<br />
Employment 72.93 51.77** 205.64*** 593.79***<br />
Cont. R&D 0.07 0.21*** 0.23*** 0.43***<br />
Innovators 0.13 0.32*** 0.37*** 0.49***<br />
Continuous R&D performers<br />
N (cont. R&D) 75 205 178 75<br />
R&D intensity 0.06 0.07 0.08 0.06<br />
Cooperation 0.48 0.48 0.70 0.68***<br />
Funding 0.29 0.44** 0.38* 0.56***<br />
Innovators<br />
N (innovator) 140 311 283 86<br />
Innovative sales intensity 0.24 0.25 0.23 0.24<br />
Notes: Four firm types are distinguished: (1) National firms: do not export<br />
<strong>and</strong> are not part of a multinational firm; (2) Exporters: firms that export, but<br />
are not part of multinational; (3) Foreign affiliates of multinational firms; (4)<br />
Home-based (domestic) multinational firms. Values are sample means (except<br />
when the number of firms is reported). Variables are defined in the data<br />
appendix (section 4.A.2). Significance levels (* p < 0.10 ; ** p < 0.05 ; *** p<br />
< 0.01) refer to a one-tailed test on the difference with respect to the baseline<br />
category (national firms).<br />
Table 4.3: Summary statistics by firm type
146 Multinationals, research effort <strong>and</strong> innovative output<br />
The second part of table 4.3 summarizes the key variables relevant to the<br />
first stage of the empirical model, i.e. estimation of firms’ research intensity.<br />
In spite of the large contribution of multinationals to total R&D spending<br />
in the sample (cfr. table 4.2), sample averages of the R&D over sales ratio<br />
(R&D-intensity) are not found to be different according to firms’ global engagement<br />
status. This lends support to the findings of Teirlinck (2005a) that<br />
the dominance of foreign firms in R&D spending is at least partly attributable<br />
to their larger size. Domestic multinationals are more likely to engage in cooperation<br />
activities compared to national firms. Finally, exporters, foreign<br />
affiliates <strong>and</strong> home-based multinationals are more likely than domestic firms<br />
to have obtained funding from regional, national or EU authorities between<br />
2002 <strong>and</strong> 2004.<br />
The last part of table 4.3 shows (apart from the number of firms that<br />
generate innovative sales) the average innovative sales intensity for the subsample<br />
of firms with innovative sales, which is the relevant sample in the final<br />
stage of the estimation procedure. Similar to the R&D-intensity variable,<br />
no differences between the firms according to their global activities can be<br />
discerned. Again, this suggests that multinational firms’ dominance in terms<br />
of total innovative inputs <strong>and</strong> output is, at least partly, attributable to their<br />
larger average size.<br />
4.4 Empirical results<br />
Before proceeding to the estimation results of the full empirical model<br />
summarized in (4.2) <strong>and</strong> (4.3), it is useful to investigate whether there is in<br />
fact an innovation premium associated with firms’ global engagement status<br />
in Belgium, taking into account firm size <strong>and</strong> sector distribution. As<br />
was noted above, Castellani <strong>and</strong> Zanfei (2006), Criscuolo et al. (2005) <strong>and</strong><br />
Frenz <strong>and</strong> Ietto-Gillies (2007) provide evidence that firms that are active on<br />
international markets are more innovative than their national counterparts,<br />
controlling for size <strong>and</strong> sector distribution.
Empirical results 147<br />
4.4.1 Preliminary evidence<br />
Table 4.4 presents preliminary evidence on the existence of an innovation<br />
premium for globally engaged firms in Belgium. Both parts of the table<br />
(models I <strong>and</strong> II) estimate an innovation production function. In the first<br />
part of the table (model I), only the global engagement dummies, firm size<br />
(in the selection equation) <strong>and</strong> sector <strong>and</strong> region dummies are taken into<br />
account. The second part of the table (model II) controls for R&D inputs<br />
by including a dummy equal to one for firms that continuously engage in<br />
R&D 13 .<br />
Results obtained in table 4.4 suggest that exporters <strong>and</strong> multinational<br />
firms (whether foreign <strong>and</strong> home-based) are more likely to generate innovative<br />
sales compared to national firms. However, conditional on selection, firms’<br />
global engagement status does not have a significant impact on the scale of<br />
innovative output (i.e. the innovative sales share). Comparing the results<br />
of the two models in table 4.4 further points to the importance of taking<br />
R&D inputs into account in the innovation output function. Controlling for<br />
continuous R&D engagement of the firm lowers the innovation premium of<br />
globally active firms <strong>and</strong> also significantly reduces the coefficient obtained on<br />
firm size.<br />
Overall, these preliminary results support the existence of an innovation<br />
premium associated with firms’ global engagement status. Table 4.4 shows<br />
that globally active firms are more likely to generate innovative sales than<br />
their national counterparts. However, conditional on their higher selection<br />
probability, they do not generate more innovative sales (in relative terms)<br />
than domestically oriented firms.<br />
A final point is worth noting about these preliminary results. The different<br />
results obtained for the selection <strong>and</strong> continuous part of the model, suggest<br />
that it is important to estimate a flexible model, such as the generalized<br />
13 Since the R&D-intensity variable in (4.3) is specified in logarithmic form <strong>and</strong> is hence<br />
undefined for firms with no reported R&D expenditures, it is not possible to implement<br />
this variable here. I therefore rely on a dummy variable indicating firms’ engagement in<br />
research effort.
148 Multinationals, research effort <strong>and</strong> innovative output<br />
(I) (II)<br />
Selection Innovative Selection Innovative<br />
Dependent variable equation sales equation equation sales equation<br />
Exporter (d) 0.16*** 0.16 0.12*** 0.15<br />
[0.02] [0.10] [0.02] [0.10]<br />
Foreign MNE (d) 0.14*** 0.00 0.13*** -0.03<br />
[0.03] [0.11] [0.03] [0.11]<br />
Home-based MNE (d) 0.19*** 0.07 0.11** 0.04<br />
[0.05] [0.14] [0.05] [0.14]<br />
log(Employment) 0.07*** - 0.03*** -<br />
[0.01] [0.01]<br />
R&D (d) - - 0.51*** 0.05<br />
[0.03] [0.08]<br />
Sector dummies yes yes<br />
Region dummies yes yes<br />
N 2,988 2,988<br />
left-censored N 2,168 2,168<br />
Results of heckman selection model. Dependent variable: log of innovative<br />
sales intensity. Reported values represent marginal effects evaluated at the<br />
mean of the independent variable or the discrete change of a dummy variable<br />
(d) from 0 to 1, st<strong>and</strong>ard errors are reported in brackets. For the selection<br />
equation, marginal effects refer to the marginal probability change. For the<br />
intensity equation, marginal effects refer to the change in intensity, conditional<br />
upon being selected. The group of exporters comprises of firms that export<br />
but are not part of a multinational firm. R&D dummy is equal to one for<br />
continuous R&D performers. Significance levels: * p < 0.10 ; ** p < 0.05 ;<br />
*** p < 0.01. Variables are defined in the data appendix (section 4.A.2).<br />
Table 4.4: Innovation production function: Preliminary evidence
Empirical results 149<br />
tobit setting applied here, hence allowing coefficients to differ in the two<br />
estimation stages. Both Castellani <strong>and</strong> Zanfei (2006) <strong>and</strong> Criscuolo et al.<br />
(2005) estimate a tobit model 14 , hence restricting the coefficients of the two<br />
estimation stages to be equal.<br />
4.4.2 The innovation production function: baseline re-<br />
sults<br />
Table 4.5 shows the results of estimating the full empirical model, as specified<br />
in (4.2) <strong>and</strong> (4.3). In the first stage, a generalized tobit model is estimated,<br />
where R&D-intensity (specified in logarithmic form) is the dependent<br />
variable. As was noted in section 4.2, selection issues inherent in the CIS<br />
data limit the choice of variables that can be included in the selection equation<br />
of this model to those questions answered by all firms in the CIS sample.<br />
Hence, while the selection <strong>and</strong> intensity equation of stage 1 both include firm<br />
size, global engagement status, sector <strong>and</strong> region dummies; the R&D intensity<br />
equation further includes two dummies, representing whether the firm<br />
has engaged in cooperation activities (Cooperation) or has obtained funding<br />
from the regional, national or European level during the period 2002-2004<br />
(Funding).<br />
Results for the first stage of the estimation algorithm (firm-level R&D intensity)<br />
suggest that exporters <strong>and</strong> multinational firms (both home-based<br />
<strong>and</strong> foreign) are more likely to engage in R&D investment, controlling for<br />
their size <strong>and</strong> distribution across sectors <strong>and</strong> regions. However, conditional<br />
on their higher selection probability, only domestic multinationals are shown<br />
to spend significantly more on R&D than purely domestic firms. <strong>Firm</strong>s that<br />
have acquired funding from the regional, national or EU level spend significantly<br />
more on internal research effort, conditional on selection. Cooperation<br />
is not found to be a significant determinant of internal R&D intensity.<br />
The right-h<strong>and</strong> side of table 4.5 shows the results of estimating (4.3),<br />
while instrumenting for research effort using the predicted logarithm of R&D-<br />
14 Frenz <strong>and</strong> Ietto-Gillies (2007) investigate innovative output using a probit model,<br />
where the dependent variable is a dummy variable equal to one if the firm has introduced<br />
a (product or process) innovation.
150 Multinationals, research effort <strong>and</strong> innovative output<br />
Stage 1 Stage 2<br />
Selection R&D intensity Selection Inn. Sales<br />
Exporter (d) 0.12*** 0.15 0.08*** 0.07<br />
[0.02] [0.24] [0.03] [0.11]<br />
Foreign MNE (d) 0.05** 0.24 0.04 -0.09<br />
[0.02] [0.24] [0.03] [0.12]<br />
Home-based MNE (d) 0.17*** 0.58** -0.09** -0.21<br />
[0.04] [0.26] [0.04] [0.19]<br />
log(Employment) 0.07*** - 0.14***<br />
[0.01] [0.01]<br />
Cooperation (d) - -0.16 - -<br />
[0.15]<br />
Funding (d) - 0.54*** - -<br />
[0.15]<br />
log(R&D/sales)* (1) 0.41*** 0.42**<br />
[0.05] [0.19]<br />
Sector dummies yes yes<br />
Region dummies yes yes<br />
N 2,988 2,988<br />
left-censored N 2,455 2,168<br />
Results of two-step heckman selection model. Reported values represent<br />
marginal effects evaluated at the mean of the independent variable or the discrete<br />
change of a dummy variable (d) from 0 to 1, st<strong>and</strong>ard errors are reported<br />
in brackets. For the selection equation, marginal effects refer to the marginal<br />
probability change. For the intensity equation, marginal effects refer to the<br />
change in intensity, conditional upon being selected. The group of exporters<br />
comprises of firms that export but are not part of a multinational firm. R&D<br />
dummy is equal to one for continuous R&D performers. Significance levels: * p<br />
< 0.10 ; ** p < 0.05 ; *** p < 0.01. Variables are defined in the data appendix<br />
(section 4.A.2). (1) Predicted R&D intensity obtained from first stage of the<br />
estimation procedure.<br />
Table 4.5: Innovation production function: Two-stage estimation
Empirical results 151<br />
intensity obtained from the first stage. As was noted in section 4.2, the<br />
cooperation <strong>and</strong> funding variable act as exclusion restrictions in the second<br />
stage, i.e. they are assumed to affect the innovative sales share only through<br />
their impact on internal R&D (Mairesse <strong>and</strong> Mohnen, 2004). Results shown<br />
in the right-h<strong>and</strong> panel of table 4.5 show some striking differences compared<br />
to the results in table 4.4.<br />
Conditional on their higher R&D spending patterns, multinational firms<br />
are no longer found to be significantly more likely to generate innovative sales,<br />
compared to their uni-national counterparts. Only exporters are found to be<br />
significantly more likely to be innovative, conditional on their R&D spending,<br />
although the marginal effect is lower than in table 4.4 (0.08 instead of 0.16 in<br />
the left-h<strong>and</strong> side panel of table 4.4). Home-based multinationals are found<br />
to be significantly less likely to generate innovative sales, once their higher<br />
probability to engage in R&D <strong>and</strong> their higher spending patterns are taken<br />
into account.<br />
The marginal effect on the predicted R&D intensity (in logs) in table 4.5<br />
amounts to 0.41 for the selection stage, suggesting that a ten percent increase<br />
in firm-level research effort is associated with an increase of 4.1 percentage<br />
points in the probability that the firm will generate innovative sales. In light<br />
of the overall probability that a firm in the sample generates innovative sales,<br />
which amounts to 27.44 percent 15 ; this suggests that a ten percent increase<br />
in firm-level research effort is associated with a 15 percent increase in the<br />
probability that the firm will generate innovative output 16 . For innovative<br />
sales intensity, a ten percent increase in research spending is associated with<br />
a 4.2 percent increase in innovative sales intensity 17 . The magnitude of these<br />
returns to R&D effects is much lower than the results obtained by Mairesse<br />
<strong>and</strong> Mohnen (2004) for France, who report elasticities up to ten times larger<br />
than those obtained here.<br />
15 Out of the 2,988 firms in the sample, 820 or 27.44 percent generated innovative sales<br />
during the period 2002-2004.<br />
16 (27.44 + 4.1) / 27.44 = 14.94.<br />
17 The marginal effect amounts to 4.2. Since both variables are expressed in logarithmic<br />
form, this marginal effect can be interpreted as an elasticity.
152 Multinationals, research effort <strong>and</strong> innovative output<br />
4.4.3 High-tech versus low-tech sectors<br />
Cassiman <strong>and</strong> Mendi (2008) compare the performance of subsidiaries of foreign<br />
<strong>and</strong> domestic firms in Spain, using a data set comparable to the CIS data<br />
<strong>and</strong> find that foreign firms are less innovative than subsidiaries of Spanish<br />
firms, both in terms of innovative inputs <strong>and</strong> output. However, they further<br />
show that this result is mainly driven by the low-technology manufacturing<br />
<strong>and</strong> service sectors, while the difference in performance between foreign <strong>and</strong><br />
domestic affiliates is insignificant for high- <strong>and</strong> medium-high tech industries.<br />
Furthermore, Mairesse <strong>and</strong> Mohnen (2004) report sizeable differences in the<br />
returns to R&D (both in the selection <strong>and</strong> intensity equation of stage two)<br />
between high-technology <strong>and</strong> low-technology sectors. To investigate to what<br />
extent results obtained in table 4.5 differ according to the technology intensity<br />
of the sector, tables 4.6 <strong>and</strong> 4.7 show the results of estimating (4.2) <strong>and</strong><br />
(4.3) separately for (medium-)high technology <strong>and</strong> (medium-)low technology<br />
sectors respectively.<br />
<strong>Firm</strong>s are classified into high- versus low-tech sectors based on the classification<br />
of industry according to technology from Eurostat. The different<br />
NACE (Rev. 1.1.) codes assigned to each category are listed in appendix<br />
table 4.A.1. Comparing the results of tables 4.6 <strong>and</strong> 4.7, several important<br />
differences are worth noting.<br />
I will begin by discussing the left-h<strong>and</strong> side of tables 4.6 <strong>and</strong> 4.7, i.e. the<br />
R&D intensity part of the model shown in (4.2). Exporters <strong>and</strong> domestic<br />
multinational firms are more likely to engage in continuous R&D spending,<br />
both in high- <strong>and</strong> low-technology sectors. However, the magnitude of the<br />
marginal effects (i.e. the change in the probability to engage in research<br />
effort that is associated with a change from 0 to 1 for the dummy variables) is<br />
much larger in the high- than in the low-technology sectors. Foreign affiliates<br />
of multinational firms are found to be more likely to engage in continuous<br />
research effort in high-tech sectors, while for low-tech sectors, the foreign<br />
ownership dummy has an insignificant marginal effect.<br />
Turning to the results on the R&D intensity equation in the first stage,<br />
both foreign <strong>and</strong> domestic multinational ownership are positively <strong>and</strong> signifi-
Empirical results 153<br />
Stage 1 Stage 2<br />
Selection R&D intensity Selection Inn. Sales<br />
Exporter (d) 0.28*** 0.07 0.23*** 0.12<br />
[0.05] [0.35] [0.05] [0.16]<br />
Foreign MNE (d) 0.13** -0.09 0.15*** 0.23<br />
[0.05] [0.36] [0.05] [0.17]<br />
Home-based MNE (d) 0.34*** 0.26 0.20*** 0.13<br />
[0.08] [0.38] [0.08] [0.25]<br />
log(Employment) 0.10*** - 0.10*** -<br />
[0.01] [0.02]<br />
Cooperation - -0.45** - -<br />
[0.21]<br />
Funding - 0.77*** - -<br />
[0.21]<br />
log(R&D/sales)* (1) - - 0.10 0.32<br />
[0.06] [0.21]<br />
Sector dummies yes yes<br />
Region dummies yes yes<br />
N 2,988 2,988<br />
left-censored N 2,455 2,168<br />
Results of two-step heckman selection model. Reported values represent<br />
marginal effects evaluated at the mean of the independent variable or the discrete<br />
change of a dummy variable (d) from 0 to 1, st<strong>and</strong>ard errors are reported<br />
in brackets. For the selection equation, marginal effects refer to the marginal<br />
probability change. For the intensity equation, marginal effects refer to the<br />
change in intensity, conditional upon being selected. The group of exporters<br />
comprises of firms that export but are not part of a multinational firm. R&D<br />
dummy is equal to one for continuous R&D performers. Significance levels: * p<br />
< 0.10 ; ** p < 0.05 ; *** p < 0.01. Variables are defined in the data appendix<br />
(section 4.A.2). (1) Predicted R&D intensity obtained from first stage of the<br />
estimation procedure.<br />
Table 4.6: Innovation production function: High-technology sectors
154 Multinationals, research effort <strong>and</strong> innovative output<br />
cantly related to firm-level R&D intensity, conditional on selection. However,<br />
this result only holds in the low-technology sectors. In high-technology sectors,<br />
firms’ global engagement status is not significantly related to research<br />
intensity, conditional on selection. Furthermore, firm-level cooperation activities<br />
<strong>and</strong> funding opportunities are only found to be significantly related<br />
to the intensity of research effort in high-technology sectors, with opposite<br />
signs (negative <strong>and</strong> positive respectively).<br />
The right-h<strong>and</strong> sides of tables 4.6 <strong>and</strong> 4.7 list the results of the second<br />
estimation stage (4.3) separately for high-tech <strong>and</strong> low-tech sectors respectively.<br />
Again, there are some important differences between the sectors that<br />
are worth noting. While firms’ global engagement status (whether through<br />
exporting or FDI) is positively <strong>and</strong> significantly related to the probability<br />
that firms generate innovative sales in high-tech sectors, the opposite applies<br />
for low-technology sectors. These results suggest that exporters <strong>and</strong><br />
multinational firms are more (less) likely to innovate in high- (low-) tech<br />
sectors. Also striking is the result on the R&D intensity variable, which is<br />
insignificant for high-technology sectors <strong>and</strong> highly significant for low-tech<br />
activities.<br />
Finally, firms’ global engagement status is not significantly related to the<br />
innovative sales intensity for the high-tech sectors (table 4.6), while multinational<br />
ownership is significantly negatively related to this intensity in the<br />
low-tech sectors. The R&D elasticity (i.e. the marginal effect obtained for<br />
the predicted R&D intensity from the first stage) is insignificant in both<br />
sectors.<br />
These results suggest that globally engaged firms (whether through exporting<br />
or FDI) active in high-technology sectors are able to benefit from<br />
their access to a global network, allowing them to be more innovative, even<br />
when taking into account that these firms tend to spend more (on average)<br />
on R&D compared to national firms. The fact that globally integrated firms<br />
are not found to be more innovative in low-tech sectors, conditional on their<br />
R&D spending patterns might suggest that, within the low-technology sectors,<br />
firms’ international activities serve as a means to achieve economies of
Empirical results 155<br />
Stage 1 Stage 2<br />
Selection R&D intensity Selection Inn. Sales<br />
Exporter (d) 0.07*** 0.15 -0.051* 0.02<br />
[0.02] [0.31] [0.027] [0.15]<br />
Foreign MNE (d) 0.04 0.52* -0.259*** -0.47**<br />
[0.02] [0.31] [0.026] [0.23]<br />
Home-based MNE (d) 0.11** 0.83** -0.216*** -0.63<br />
[0.05] [0.35] [0.012] [0.41]<br />
log(Employment) 0.06*** - 0.279*** -<br />
[0.01] [0.021]<br />
Cooperation - 0.17 - -<br />
[0.22]<br />
Funding - 0.36 - -<br />
[0.23]<br />
log(R&D/sales)* (1) - - 1.02*** 0.55<br />
[0.08] [0.36]<br />
Sector dummies yes yes<br />
Region dummies yes yes<br />
N 2,988 2,988<br />
left-censored N 2,455 2,168<br />
Results of two-step heckman selection model. Reported values represent<br />
marginal effects evaluated at the mean of the independent variable or the discrete<br />
change of a dummy variable (d) from 0 to 1, st<strong>and</strong>ard errors are reported<br />
in brackets. For the selection equation, marginal effects refer to the marginal<br />
probability change. For the intensity equation, marginal effects refer to the<br />
change in intensity, conditional upon being selected. The group of exporters<br />
comprises of firms that export but are not part of a multinational firm. R&D<br />
dummy is equal to one for continuous R&D performers. Significance levels: * p<br />
< 0.10 ; ** p < 0.05 ; *** p < 0.01. Variables are defined in the data appendix<br />
(section 4.A.2). (1) Predicted R&D intensity obtained from first stage of the<br />
estimation procedure.<br />
Table 4.7: Innovation production function: Low-technology sectors
156 Multinationals, research effort <strong>and</strong> innovative output<br />
scale, rather than as a learning opportunity allowing the firm to continuously<br />
improve its existing product lines <strong>and</strong> exp<strong>and</strong> into new ones. However, in<br />
the absence of any dynamic information on firms’ global engagement <strong>and</strong><br />
innovation activities, it might equally be argued that results in the high-tech<br />
sectors are driven by a learning effect (access to the global market as a driver<br />
for innovation), while results in the low-tech sectors can be explained by selection<br />
effects (innovation as a driver of the firms’ international expansion).<br />
Hence, these conclusions should be treated with caution.<br />
4.5 Conclusions<br />
This paper has investigated whether firms’ innovative output differs according<br />
to their global engagement status. Specifically, by relying on the innovation<br />
production function framework introduced by Mairesse <strong>and</strong> Mohnen<br />
(2002) <strong>and</strong> by taking the endogeneity of research inputs specifically into account<br />
in the innovation output function, following the model of Crépon et<br />
al. (1998), I am able to distinguish between the indirect (through higher research<br />
spending) <strong>and</strong> direct effect of firms’ global integration on its innovative<br />
output.<br />
Unconditionally, exporters <strong>and</strong> multinational firms in Belgium are found to<br />
be significantly more likely to generate innovative output than their domestic<br />
counterparts, pointing to the existence of an innovation premium, not unlike<br />
the productivity premium associated with firms’ international activities. In<br />
terms of innovative sales intensity, no significant differences emerge between<br />
the different types of firms.<br />
However, taking into account that internationally engaged firms typically<br />
spend more on R&D than national firms (i.e. conditional on research spending<br />
patterns), only exporters are found to be more likely to generate innovative<br />
sales, while home-based multinationals are significantly less likely to innovate.<br />
Distinguishing between high-technology <strong>and</strong> low-technology sectors<br />
further shows that there is an innovation premium associated with firms’ international<br />
activities in high-technology sectors, conditional on their research<br />
intensity; while in low-technology sectors, a negative innovation premium
Conclusions 157<br />
emerges.<br />
From a policy perspective, the analysis yields several important insights.<br />
Overall, the empirical analysis has shown that (part of) the innovation premium<br />
associated with firms’ global engagement status in Belgium can be<br />
traced back to these firms’ different spending patterns, i.e. their higher likelihood<br />
to engage in continuous R&D spending. However, within the hightechnology<br />
sectors, firms with exposure to international markets (whether<br />
through exporting or FDI) are found to be more likely to generate innovative<br />
sales, conditional on their R&D spending patterns; while in low-technology<br />
sectors, these firms are found to be less likely to be innovative. This result<br />
suggests that government policies aimed at the attraction of (foreign) multinational<br />
firms in order to stimulate innovative output (or inputs) are more<br />
likely to be successful (i.e. will generate a higher rate of return) when they<br />
are specifically targeted at high-technology sectors.<br />
For the low-technology sectors, the high returns to R&D obtained in these<br />
sectors, especially for the selection stage of the model, suggests that policies<br />
targeted towards these sectors might be more successful when aimed at stimulating<br />
research efforts in general, indiscriminate of the global engagement<br />
status of the firm. However, as already noted above, these conclusions should<br />
be treated with caution, since, in the absence of any dynamic information<br />
on the timing of firms’ innovation <strong>and</strong> internationalization decisions, I am<br />
not able to provide any insights into the direction of causality between firms’<br />
international exposure <strong>and</strong> its innovation activities.<br />
From a methodological point of view, the analysis clearly points to the<br />
importance of taking exporters’ <strong>and</strong> multinational firms’ different research<br />
spending patterns into account when investigating their impact on (correlation<br />
with) innovative output. Results clearly show that failure to take these<br />
firms’ higher average innovative effort into account, might lead to misleading<br />
results concerning the innovation premium associated with firms’ international<br />
activities.<br />
Finally, as highlighted in section 4.2, drawing the analogy between the<br />
innovation production function <strong>and</strong> the traditional production function liter-
158 Multinationals, research effort <strong>and</strong> innovative output<br />
ature further, offers numerous challenges to researchers. While the framework<br />
of Mairesse <strong>and</strong> Mohnen (2002), Mairesse <strong>and</strong> Mohnen (2004) <strong>and</strong> Crépon et<br />
al. (1998) serves as a useful starting point, many issues remain unresolved.<br />
This is related to the fact that while many of the existing techniques applied<br />
in the production function literature (eg. GMM or semiparametric<br />
estimation) can easily be carried over to the innovation production function,<br />
choices of estimation techniques are often constrained by the lack of dynamics<br />
in many firm-level innovation data sets.
Data appendix 159<br />
4.A Data appendix<br />
4.A.1 Sample selection<br />
The population of CIS4-firms was selected on the basis of the full population<br />
of Belgian firms, registered at the National Office for Social Security<br />
at the end of 2004 (Teirlinck, 2005b). Of these, all firms with at least ten<br />
employees were selected. After applying a number of corrections 18 , a population<br />
of about 25,000 firms was retained. Sampling was performed on the<br />
population after stratifying according to sector (NACE two-digit, three-digit<br />
in some cases), size (three size classes) <strong>and</strong> region (either at the two-digit<br />
Nuts or provincial level). The final questionnaire was sent to 1,814 firms in<br />
Brussels; 2,118 in the Walloon region <strong>and</strong> 3,000 in Fl<strong>and</strong>ers. The full sample<br />
of CIS4-firms consists of 3,322 firms.<br />
<strong>Firm</strong>s with missing identification number are omitted (one firm), as well<br />
as firms with exports amounting to more than 100 percent of total sales in<br />
2004 (one firm). Similarly, firms reporting unrealistically high R&D to sales<br />
ratios are omitted (14 firms). All firms for which no matching information<br />
could be obtained from Belfirst are similarly omitted (100 firms). Finally,<br />
218 firms were subject to a merger, acquisition or absorption by another<br />
company between 2002 <strong>and</strong> 2005. These firms are not taken into account<br />
in the empirical analysis. The final sample of firms used in the empirical<br />
analysis consists of 2,988 firms <strong>and</strong> accounts for about 65 percent of turnover<br />
<strong>and</strong> employment generation in the full CIS sample.<br />
4.A.2 Definitions of variables<br />
Innovation input <strong>and</strong> output measures<br />
Innovative effort<br />
Intramural R&D over sales ratio. Specific question from CIS-survey: “Please<br />
estimate the amount of expenditure on intramural R&D (including personnel<br />
<strong>and</strong> related costs as well as capital expenditures on buildings <strong>and</strong> equipment<br />
18 <strong>Firm</strong>s that were no longer active were omitted, as well as firms that appeared with<br />
a double identification number. For a detailed overview of the population selection <strong>and</strong><br />
sampling process, I refer to Teirlinck (2005b).
160 Multinationals, research effort <strong>and</strong> innovative output<br />
specifically for R&D) in 2004 only.”.<br />
Share of new products in turnover<br />
Please give the percentage of your total turnover in 2004 from: goods <strong>and</strong><br />
service innovations introduced during 2002-2004 that were new to your market,<br />
or only new to the firm .<br />
Control variables<br />
Employment<br />
Number of employees in 2004, full-time equivalents.<br />
Exporter dummy<br />
Equal to one if the firm has reported positive exports in 2004, but is not part<br />
of a multinational group, i.e. the firm is not part of a foreign- or home-based<br />
multinational firm.<br />
Foreign ownership dummy<br />
Equal to one if the firm is part of a Belgian group with foreign headquarters.<br />
Domestic multinational ownership dummy<br />
Equal to one if the firm is part of a Belgian group <strong>and</strong> has foreign subsidiaries.<br />
Cooperation<br />
Dummy equal to one if the firm has engaged in cooperation activities between<br />
2002 <strong>and</strong> 2004.<br />
Funding<br />
Dummy equal to one if the firm has obtained funding from regional, national<br />
or EU authorities between 2002 <strong>and</strong> 2004.
Data appendix 161<br />
High-technology Low-technology<br />
24 Chemicals 15 Food <strong>and</strong> beverages<br />
29 Machinery <strong>and</strong> equipment 16 Tobacco<br />
30 Office machinery (computers) 17 Textiles<br />
31 Electrical machinery 18 Clothing<br />
32 Radio, TV, communication 19 Leather (products)<br />
33 Medical <strong>and</strong> optical instruments 20 Wood (products)<br />
34 Motor vehicles 21 Pulp <strong>and</strong> paper (products)<br />
35 Other transport equipment (excl. 351) 22 Publishing <strong>and</strong> printing<br />
61 Water transport 23 Petroleum products<br />
62 Air transport 25 Rubber <strong>and</strong> plastics<br />
64 Post <strong>and</strong> telecommunications 26 Nonmetallic mineral products<br />
65 Financial intermediation 27 Basic metals<br />
66 Insurance <strong>and</strong> pension funding 28 Fabricated metals<br />
67 Ancilliary financial activities 351 Shipbuilding <strong>and</strong> repairs<br />
70 Real estate activities 36 Furniture<br />
71 Renting activities 37 Recycling<br />
72 Computer <strong>and</strong> related activities 50 Motor vehicles trade<br />
73 Research <strong>and</strong> development 51 Wholesale trade<br />
74 Other business activities 52 Retail trade<br />
55 Hotels <strong>and</strong> restaurants<br />
60 L<strong>and</strong> transport<br />
63 Supporting transport activities<br />
Table 4.A.1: Sector classification: High-tech versus low-tech
Bibliography<br />
Abraham, Filip, Jozef Konings, <strong>and</strong> Stijn Vanormelingen, “The<br />
effect of globalization on union bargaining <strong>and</strong> price-cost margins of<br />
firms,” Licos Discussion Paper Series, 2007b, 181/2007. 141<br />
, , <strong>and</strong> Veerle Slootmaekers, “FDI spillovers in the Chinese<br />
manufacturing sector: Evidence of firm heterogeneity,” CEPR Discussion<br />
Paper, 2007a, 6573. 41<br />
Ackerberg, Daniel A., Kevin Caves, <strong>and</strong> Garth Frazer, “Structural<br />
identification of production functions,” Unpublished manuscript., 2006.<br />
20, 28, 29, 35<br />
Ackerberg, Daniel, C. Lanier Benkard, Steven Berry, <strong>and</strong> Ariel<br />
Pakes, “Econometric tools for analyzing market outcomes,” in James<br />
Heckman <strong>and</strong> Edward Leamer, eds., H<strong>and</strong>book of Econometrics, Vol.<br />
6(1), Amsterdam: North-Holl<strong>and</strong>, 2007, pp. 4171–4276. 9, 12, 14, 15,<br />
20, 21, 22, 24, 30, 35<br />
Amiti, Mary <strong>and</strong> Jozef Konings, “Trade liberalization, intermediate inputs,<br />
<strong>and</strong> productivity: Evidence from Indonesia,” American Economic<br />
Review, 2007, 97 (5), 1611–1638. 9<br />
Arnold, Jens Matthias, “Productivity estimation at the plant level: A<br />
practical guide,” Unpublished manuscript., 2005. 37<br />
Asplund, Marcus <strong>and</strong> Volker Nocke, “<strong>Firm</strong> turnover in imperfectly competitive<br />
markets,” Review of Economic Studies, 2006, 73 (2), 295–327.<br />
68<br />
Audretsch, David B. <strong>and</strong> Talat Mahmood, “New firm survival: New results<br />
using a hazard function,” The Review of Economics <strong>and</strong> Statistics,<br />
1995, 77 (1), 97–103. 66, 115<br />
163
164 Bibliography<br />
Aw, Bee Yan, Mark J. Roberts, <strong>and</strong> Tor Winston, “Export market<br />
participation, investments in R&D <strong>and</strong> worker training, <strong>and</strong> the evolution<br />
of firm productivity,” The World Economy, 2007, 30 (1), 83–104.<br />
132<br />
Bartelsman, Eric J. <strong>and</strong> Mark Doms, “Underst<strong>and</strong>ing productivity:<br />
Lessons from longitudinal microdata,” Journal of Economic Literature,<br />
2000, 38 (3), 569–594. 9, 56<br />
Basile, Roberto, “Export behaviour of Italian manufacturing firms over the<br />
nineties: the role of innovation,” Research Policy, 2001, 30 (8), 1185–<br />
1201. 132<br />
Belgostat, Materials <strong>and</strong> Investment Price Indices., National Bank of Belgium,<br />
2007. 36<br />
Benavente, Jose Miguel, “The role of research <strong>and</strong> innovation in promoting<br />
productivity in chile,” Economics of Innovation <strong>and</strong> New Technology,<br />
2006, 15 (4), 301 – 315. 136<br />
Bernard, Andrew B. <strong>and</strong> Fredrik Sjöholm, “Foreign owners <strong>and</strong> plant<br />
survival,” National Bureau of Economic Research Working Paper Series,<br />
2003, 10039. 67, 102, 105, 111, 114, 119, 120, 121<br />
<strong>and</strong> J. Bradford Jensen, “Exporters, jobs <strong>and</strong> wages in US<br />
manufacturing: 1976-1987,” Brookings Papers on Economic Activity -<br />
Microeconomics, 1995, pp. 67–119. 2, 55<br />
<strong>and</strong> , “<strong>Firm</strong> structure, multinationals, <strong>and</strong> manufacturing plant<br />
deaths,” Review of Economics <strong>and</strong> Statistics, 2007a, 89 (2), 193–204.<br />
56, 102, 105, 111, 119, 121, 123<br />
, , <strong>and</strong> Peter K. Schott, “Survival of the best fit: Exposure<br />
to low-wage countries <strong>and</strong> the (uneven) growth of U.S. manufacturing<br />
plants,” Journal of International Economics, 2006a, 68 (1), 219–237. 3,<br />
53, 55, 56, 58, 60, 65, 69, 76, 84<br />
, , Stephen J. Redding, <strong>and</strong> Peter K. Schott, “<strong>Firm</strong>s in international<br />
trade,” Journal of Economic Perspectives, 2007c, 2 (3), 105–<br />
130. 1, 2, 55<br />
, Jonathan Eaton, J. Bradford Jensen, <strong>and</strong> Samuel Kortum,<br />
“Plants <strong>and</strong> productivity in international trade,” American Economic<br />
Review, 2003, 93 (4), 1268–1290. 55
Bibliography 165<br />
, Stephen J. Redding, <strong>and</strong> Peter K. Schott, “Products <strong>and</strong> productivity,”<br />
Unpublished manuscript, 2005. 3, 10, 18, 19, 20, 30, 51<br />
, , <strong>and</strong> , “Multi-product firms <strong>and</strong> product switching,” Unpublished<br />
manuscript, 2006b. 3, 18, 20, 31, 51<br />
, , <strong>and</strong> , “Comparative advantage <strong>and</strong> heterogeneous <strong>Firm</strong>s,”<br />
Review of Economic Studies, 2007b, 74 (1), 31–66. 55<br />
Bevan, Alan A. <strong>and</strong> Saul Estrin, “The determinants of foreign direct investment<br />
into European transition economies,” Journal of Comparative<br />
Economics, 2004, 32 (4), 775–787. 73<br />
Blundell, Richard <strong>and</strong> Steve Bond, “GMM estimation with persistent<br />
panel data: An application to production functions,” IFS Working Paper,<br />
1999, W99/4. 22, 23, 34, 35<br />
BvDEP, Belfirst Database, DVD-rom. Bureau van Dijk Electronic Publishing.,<br />
2004. 103, 107, 125<br />
, Belfirst Database, DVD-rom. Bureau van Dijk Electronic Publishing,<br />
2006. 34, 141<br />
Carstensen, Kai <strong>and</strong> Farid Toubal, “Foreign direct investment in Central<br />
<strong>and</strong> Eastern European countries: a dynamic panel analysis,” Journal of<br />
Comparative Economics, 2004, 32 (1), 3–22. 73<br />
Cassiman, Bruno <strong>and</strong> Elena Golovko, “Innovation <strong>and</strong> the exportproductivity<br />
link,” IESE Working Paper Series, 2007, 668. 132, 134<br />
<strong>and</strong> Pedro Mendi, “Multinationals as channels for international<br />
technology diffusion: Empirical evidence from Spain,” Unpublished<br />
manuscript, 2008. 152<br />
<strong>and</strong> Reinhilde Veugelers, “In search of complementarity in innovation<br />
strategy: Internal R&D <strong>and</strong> external knowledge acquisition,” Management<br />
Science, 2006, 52 (1), 68–82. 134<br />
Castellani, Davide <strong>and</strong> Antonello Zanfei, Multinational firms, Innovation<br />
<strong>and</strong> Productivity, Cheltenham: Edward Elgar Publishing, 2006.<br />
129, 139, 140, 143, 146, 149<br />
<strong>and</strong> , “Internationalisation, innovation <strong>and</strong> productivity: How do<br />
firms differ in Italy?,” The World Economy, 2007, 30 (1), 156–176. 5,<br />
129
166 Bibliography<br />
Caves, Richard E., “Industrial organization <strong>and</strong> new findings on the<br />
turnover <strong>and</strong> mobility of firms,” Journal of Economic Literature, 1998,<br />
36 (4), 1947–1982. 64, 73, 103, 116<br />
Chen, M. Y., “Survival duration of plants: Evidence from the US petroleum<br />
refining industry,” International Journal of Industrial Organization,<br />
2002, 20 (4), 517–555. 114<br />
Cincera, Michele, Bruno van Pottelsberghe de la Potterie, <strong>and</strong><br />
Veugelers Reinhilde, “Assessing the foreign control of production of<br />
technology: The case of a small open economy,” Scientometrics, 2006,<br />
66 (3), 493–512. 130<br />
Colantone, Italo <strong>and</strong> Leo Sleuwaegen, “Entry <strong>and</strong> exit of firms in a<br />
global economy: A cross-country <strong>and</strong> industry analysis,” Vlerick Leuven<br />
Working Paper Series, 2007, 2007/36. 69<br />
Costantini, James A. <strong>and</strong> Marc Melitz, “The dynamics of firm-level<br />
adjustment to trade liberalization,” Unpublished manuscript, 2007. 2,<br />
132<br />
Coucke, Kristien <strong>and</strong> Leo Sleuwaegen, “Exit in globalising industries:<br />
The role of international outsourcing,” Vlerick Leuven Working Paper<br />
Series, 2006, 2006/14. 69<br />
Cox, D. R., “Regression models <strong>and</strong> life-tables,” Journal of the Royal Statistical<br />
Society - Series B (Methodological), 1972, 34 (2), 187–220. 114<br />
CRB, Verslag over de economische ontwikkeling in de voedingsindustrie in<br />
2003, Centrale Raad voor het Bedrijfsleven, 2004. 36<br />
Crépon, Bruno, Emmanuel Duguet, <strong>and</strong> Jacques Mairesse, “Research,<br />
innovation <strong>and</strong> productivity: An econometric analysis at the<br />
firm level,” Economics of Innovation & New Technology, 1998, 7 (2),<br />
115. 130, 136, 137, 140, 156, 158<br />
Criscuolo, Chiara, Jonathan E. Haskel, <strong>and</strong> Matthew J. Slaughter,<br />
“Global engagement <strong>and</strong> the innovation activities of firms,” NBER<br />
Working Paper Series, 2005, 11479. 129, 130, 135, 139, 140, 143, 146,<br />
149<br />
Damijan, Joˇze P., Črt Kostevc, <strong>and</strong> Saˇso Polanec, “From innovation<br />
to exporting or vice versa? Causal link between innovation activity <strong>and</strong><br />
exporting in Slovenian microdata,” Licos Discussion Paper Series, 2008,<br />
204/2008. 132
Bibliography 167<br />
De Loecker, Jan, “Product differentiation, multi-product firms <strong>and</strong> estimating<br />
the impact of trade liberalization on productivity,” National<br />
Bureau of Economic Research Working Paper Series, 2007, 13155. 9,<br />
10, 15, 16, 17, 18, 19, 20, 30, 33, 34, 35, 41, 44, 107, 134<br />
<strong>and</strong> Jozef Konings, “Job reallocation <strong>and</strong> productivity growth in a<br />
post-socialist economy: Evidence from Slovenian manufacturing,” European<br />
Journal of Political Economy, 2006, 22 (2), 388–408. 13, 44,<br />
46<br />
Disney, Richard, Jonathan Haskel, <strong>and</strong> Ylva Heden, “Entry, exit <strong>and</strong><br />
establishment survival in UK manufacturing,” The Journal of Industrial<br />
Economics, 2003, 51 (1), 91–112. 114, 121<br />
Dunne, Timothy <strong>and</strong> Mark J. Roberts, “Costs, dem<strong>and</strong> <strong>and</strong> imperfect<br />
competition as determinants of plant level output prices,” CES Discussion<br />
Paper, 1992, 92-5. 17<br />
, , <strong>and</strong> Larry Samuelson, “Patterns of firm entry <strong>and</strong> exit in US<br />
manufacturing industries,” R<strong>and</strong> Journal of Economics, 1988, 19 (4),<br />
495–515. 14, 115<br />
, , <strong>and</strong> , “The growth <strong>and</strong> failure of US manufacturing plants,”<br />
Quarterly Journal of Economics, 1989, 104 (4), 671–698. 115<br />
, Shawn D. Klimek, <strong>and</strong> Mark J. Roberts, “Exit from regional<br />
manufacturing markets: The role of entrant experience,” International<br />
Journal of Industrial Organization, 2005, 23 (5-6), 399–421. 64<br />
Ericson, Richard <strong>and</strong> Ariel Pakes, “Markov-perfect industry dynamics:<br />
A framework for empirical work,” The Review of Economic Studies,<br />
1995, 62 (1), 53–82. 23, 64, 115<br />
Eslava, M., J. Haltiwanger, A. Kugler, <strong>and</strong> M. Kugler, “The effects<br />
of structural reforms on productivity <strong>and</strong> profitability enhancing reallocation:<br />
evidence from Colombia,” Journal of Development Economics,<br />
2004, 75 (2), 333–371. 17, 18<br />
EU, The EU economy 2005 review: Rising international economic integration,<br />
opportunities <strong>and</strong> challenges, Brussels: European Commission,<br />
2005. 1<br />
Eurostat, Producer price indices, 3-digit level, Belgium: Nace 15, 2007. 34
168 Bibliography<br />
Fabrizio, Stefania, Deniz Igan, <strong>and</strong> Ashoka Mody, “The dynamics of<br />
product quality <strong>and</strong> international competitiveness,” IMF Working Paper<br />
Series, 2007, 07/97. 54, 82<br />
Falvey, R. <strong>and</strong> H. Kierzkowski, “Product quality, intra-industry trade<br />
<strong>and</strong> (im)perfect competition,” in H. Kierzkowski, ed., Protection <strong>and</strong><br />
competition in international trade, Oxford: Basil Blackwell, 1987. 70<br />
Fariñas, Jose C. <strong>and</strong> Sonia Ruano, “<strong>Firm</strong> productivity, heterogeneity,<br />
sunk costs <strong>and</strong> market selection,” International Journal of Industrial<br />
Organization, 2005, 23 (7-8), 505–534. 14, 65, 116<br />
Feenstra, Robert C., The impact of international trade on wages, Chicago:<br />
The University of Chicago Press, 2000. 2<br />
Flam, Harry <strong>and</strong> Elhanan Helpman, “Vertical product differentiation<br />
<strong>and</strong> North-South trade,” American Economic Review, 1987, 77 (5), 810–<br />
822. 70<br />
Flamm, Kenneth, “The volatility of offshore investment,” Journal of Development<br />
Economics, 1984, 16 (3), 231–248. 103<br />
Foster, Lucia, John Haltiwanger, <strong>and</strong> C. J. Krizan, “Market selection,<br />
reallocation, <strong>and</strong> restructuring in the US retail trade sector in the 1990s,”<br />
Review of Economics <strong>and</strong> Statistics, 2006, 88 (4), 748–758. 46, 56<br />
, , <strong>and</strong> Chad Syverson, “Reallocation, firm turnover, <strong>and</strong> efficiency:<br />
Selection on productivity or profitability?,” American Economic<br />
Review, 2008, 98 (1), 394–425. 17<br />
Frenz, Marion <strong>and</strong> Grazia Ietto-Gillies, “Does multinationality affect<br />
the propensity to innovate? An analysis of the Third UK Community<br />
Innovation Survey,” International Review of Applied Economics, 2007,<br />
21 (1), 99 – 117. 129, 135, 140, 143, 146, 149<br />
Gibson, J. K. <strong>and</strong> R. I. D. Harris, “Trade liberalisation <strong>and</strong> plant exit<br />
in New Zeal<strong>and</strong> manufacturing,” Review of Economics <strong>and</strong> Statistics,<br />
1996, 78 (3), 521–529. 105<br />
Goldberg, Pinelopi K., Amit Kh<strong>and</strong>elwal, Nina Pavcnik, <strong>and</strong> Petia<br />
Topalova, “Multi-product firms <strong>and</strong> product turnover in the developing<br />
world: Evidence from India,” National Bureau of Economic Research<br />
Working Paper Series, 2008, 14127. 56
Bibliography 169<br />
Görg, Holger <strong>and</strong> Eric Strobl, “Footloose multinationals?,” The Manchester<br />
School, 2003, 71 (1), 1–19. 67, 68, 102, 105, 111, 114, 116, 119,<br />
120<br />
Greenaway, David, “The assessment: <strong>Firm</strong>-level adjustment to globalization,”<br />
Oxford Review of Economic Policy, 2004, 20 (3), 335–342. 2,<br />
4<br />
<strong>and</strong> Richard Kneller, “<strong>Firm</strong> heterogeneity, exporting <strong>and</strong> foreign<br />
direct investment,” The Economic Journal, 2007, 117 (517), F134–F161.<br />
5, 129<br />
, Joakim Gullstr<strong>and</strong>, <strong>and</strong> Richard Kneller, “Surviving globalisation,”<br />
Journal of International Economics, 2008, 74 (2), 264–277. 3, 53,<br />
55, 56, 58, 60, 65, 69, 84<br />
Greene, William H., Econometric Analysis, fifth ed., New Jersey: Prentice<br />
Hall, 2003. 110<br />
, Econometric Analysis, sixth ed., New Jersey: Prentice Hall, 2008. 63,<br />
137<br />
Griffith, Rachel, Elena Huergo, Jacques Mairesse, <strong>and</strong> Bettina Peters,<br />
“Innovation <strong>and</strong> productivity across four European countries,” Oxford<br />
Review of Economic Policy, 2006, 22 (4), 483–498. 134, 136<br />
Hall, Bronwyn H. <strong>and</strong> Jacques Mairesse, “Empirical studies of innovation<br />
in the knowledge-driven economy,” Economics of Innovation <strong>and</strong><br />
New Technology, 2006, 15 (4), 289 – 299. 134<br />
Hallak, Juan C. <strong>and</strong> Peter K. Schott, “Estimating cross-country differences<br />
in product quality,” National Bureau of Economic Research Working<br />
Paper Series, 2008, 13807. 70<br />
Hallak, Juan Carlos, “Product quality <strong>and</strong> the direction of trade,” Journal<br />
of International Economics, 2006, 68 (1), 238–265. 70<br />
Helpman, Elhanan, Marc J. Melitz, <strong>and</strong> Stephen R. Yeaple, “Export<br />
versus FDI with heterogenous firms,” American Economic Review, 2004,<br />
94 (1), 300–316. 5, 55, 129<br />
Hoch, Irving, “Estimation of production function parameters combining<br />
time series <strong>and</strong> cross-section data,” Econometrica, 1962, 30 (1), 34–53.<br />
21, 35
170 Bibliography<br />
Hopenhayn, Hugo A., “Entry, exit <strong>and</strong> firm dynamics in long run equilibrium,”<br />
Econometrica, 1992, 60 (5), 1127–1150. 14, 64, 67, 115, 116<br />
Hummels, David <strong>and</strong> Peter J. Klenow, “The variety <strong>and</strong> quality of a<br />
nation’s exports,” American Economic Review, 2005, 95 (3), 704–723.<br />
70<br />
Jaum<strong>and</strong>reu, Jordi <strong>and</strong> Jacques Mairesse, “Using price <strong>and</strong> dem<strong>and</strong><br />
information to identify production functions,” MPRA Discussion Paper,<br />
2004, 1247. 17<br />
Jovanovic, Boyan, “Selection <strong>and</strong> the evolution of industry,” Econometrica,<br />
1982, 50 (3), 649–670. 14, 64, 116<br />
K<strong>and</strong>ogan, Yener, “Does product Differentiation Explain the Increase in<br />
Exports of Transition Countries?,” Eastern European Economics, 2006,<br />
44 (2), 6–22. 54<br />
Katayama, Hajime, Shihua Lu, <strong>and</strong> James R. Tybout, “<strong>Firm</strong>-level<br />
productivity studies: Illusions <strong>and</strong> a solution,” Unpublished manuscript,<br />
2005. 9, 10, 18, 20, 30<br />
Kiefer, Nicholas M., “Economic duration data <strong>and</strong> hazard functions,”<br />
Journal of Economic Literature, 1988, 26 (2), 646. 114<br />
Klette, Tor Jakob <strong>and</strong> Zvi Griliches, “The inconsistency of common<br />
scale estimators when output prices are unobserved <strong>and</strong> endogenous,”<br />
Journal of Applied Econometrics, 1996, 11 (4), 343–361. 17, 20, 30, 35,<br />
41<br />
Konings, Jozef, “The effects of foreign direct investment on domestic firms:<br />
Evidence from firm-level panel data in emerging economies,” Economics<br />
of Transition, 2001, 9 (3), 619. 41<br />
, “Wage costs <strong>and</strong> industry (re)location in the enlarged European<br />
Union,” Swedish Economic Policy Review, 2005, 12, 57–81. 66, 113,<br />
115<br />
, “The impact of training on productivity <strong>and</strong> wages: Evidence from<br />
Belgian firm level panel data,” Licos Discussion Paper Series, 2008,<br />
(197/2008). 9, 141<br />
<strong>and</strong> Alan P. Murphy, “Do multinational enterprises relocate employment<br />
to low wage regions? Evidence from European multinationals,”<br />
Review of World Economics, 2006, 142 (2), 267–286. 104
Bibliography 171<br />
Levinsohn, J. <strong>and</strong> A. Petrin, “Estimating production functions using<br />
inputs to control for unobservables,” Review of Economic Studies, 2003,<br />
70 (2), 317–341. 10, 15, 16, 20, 21, 26, 27, 35, 65, 134, 136<br />
Levinsohn, Jim <strong>and</strong> Marc J. Melitz, “Productivity in a differentiated<br />
product markets equilibrium,” Unpublished manuscript, 2002. 17, 18,<br />
20, 35<br />
Lipsey, Robert E., “Home <strong>and</strong> host country effects of FDI,” NBER Working<br />
Paper Series, 2002, No. 9293. 101, 105<br />
Lööf, Hans <strong>and</strong> Almas Heshmati, “The impact of public funding on<br />
private R&D investment. New evidence from a firm level innovation<br />
study,” CESIS Working Paper Series, 2004, 06. 136<br />
Mairesse, Jacques <strong>and</strong> Jordi Jaum<strong>and</strong>reu, “Panel-data estimates of the<br />
production function <strong>and</strong> the revenue function: What difference does it<br />
make?,” Sc<strong>and</strong>inavian Journal of Economics, 2005, 107 (4), 651–672. 17<br />
<strong>and</strong> Pierre Mohnen, “Accounting for innovation <strong>and</strong> measuring innovativeness:<br />
An illustrative framework <strong>and</strong> an application,” American<br />
Economic Review, 2002, 92 (2), 226–230. 5, 130, 133, 156, 158<br />
<strong>and</strong> , “The importance of R&D for innovation: A reassessment<br />
using French survey data,” The Journal of Technology Transfer, 2004,<br />
30 (1), 183–197. 129, 130, 133, 137, 139, 140, 143, 151, 152, 158<br />
Marschak, Jacob <strong>and</strong> Jr. Andrews William H., “R<strong>and</strong>om simultaneous<br />
equations <strong>and</strong> the theory of production,” Econometrica, 1944, 12 (3/4),<br />
143–205. 10, 15, 20<br />
Masso, Jaan, Raul Eamets, <strong>and</strong> Kaia Philips, “Where have all the<br />
jobs gone? Gross job flows in Estonia,” University of Tartu Faculty of<br />
Economics <strong>and</strong> Business Administration Working Paper, 2004, 2004/28.<br />
57, 85<br />
Mata, José <strong>and</strong> Pedro Portugal, “Life duration of new firms,” The Journal<br />
of Industrial Economics, 1994, 42 (3), 227–245. 36, 107, 114, 116,<br />
125<br />
, , <strong>and</strong> Paulo Guimarães, “The survival of new plants: Start-up<br />
conditions <strong>and</strong> post-entry evolution,” International Journal of Industrial<br />
Organization, 1995, 13 (4), 459–481. 36, 107, 114
172 Bibliography<br />
Melitz, Marc J., “The impact of trade on intra-industry reallocations <strong>and</strong><br />
aggregate industry productivity,” Econometrica, 2003, 71 (6), 1695–<br />
1725. 2, 5, 55, 69, 129<br />
<strong>and</strong> Gianmarco I. P. Ottaviano, “Market size, trade, <strong>and</strong> productivity,”<br />
Review of Economic Studies, 2008, 75 (1), 295–316. 2, 55, 69<br />
Monfort, Ph., Hylke V<strong>and</strong>enbussche, <strong>and</strong> E. Forlani, “Chinese competition<br />
<strong>and</strong> skill-upgrading in European textiles: <strong>Firm</strong>-level evidence,”<br />
Licos Discussion Paper Series, 2008, 198/2008. 141<br />
Moreno Badia, Marialuz <strong>and</strong> Veerle Slootmaekers, “Productivity<br />
growth <strong>and</strong> financial constraints in Estonia,” Licos Discussion Paper<br />
Series, 2008, 208/2008. 66<br />
Mueller, D.C., “Entry, exit <strong>and</strong> the competitive process,” in P.A. Geroski<br />
<strong>and</strong> J. Schwalbach, eds., Entry <strong>and</strong> Market Contestability: An International<br />
Comparison, Cambridge, Massachussets: Blackwell Publishing,<br />
1991. 116<br />
Mundlak, Yair, “Empirical production function free of management bias,”<br />
Journal of Farm Economics, 1961, 43 (1), 44–56. 21, 35<br />
Murphy, Kevin M. <strong>and</strong> Andrei Shleifer, “Quality <strong>and</strong> trade,” Journal<br />
of Development Economics, 1997, 53 (1), 1–15. 70<br />
Olley, Steven G. <strong>and</strong> Ariel Pakes, “The dynamics of productivity in the<br />
telecommunications equipment industry,” Econometrica, 1996, 64 (6),<br />
1263–1297. 10, 13, 14, 15, 16, 20, 22, 23, 25, 26, 35, 41, 46, 66, 134<br />
Ornaghi, Carmine, “Assessing the effects of measurement errors on the<br />
estimation of production functions,” Journal of Applied Econometrics,<br />
2006, 21 (6), 879–891. 17, 18, 32<br />
Pavcnik, Nina, “Trade liberalization, exit, <strong>and</strong> productivity improvements:<br />
Evidence from Chilean plants,” Review of Economic Studies, 2002, 69<br />
(1), 245–276. 9, 21<br />
Pennings, Enrico <strong>and</strong> Leo Sleuwaegen, “International relocation: firm<br />
<strong>and</strong> industry determinants,” Economics Letters, 2000, 67 (2), 179–186.<br />
106<br />
<strong>and</strong> , “The reorganization decisions of troubled firms: Exit, downscale<br />
or relocate,” Vlerick Leuven Working Paper Series, 2002, 2002/21.<br />
106
Bibliography 173<br />
Petrin, Amil, Jim Levinsohn, <strong>and</strong> Brian P. Poi, “Production function<br />
estimation in Stata using inputs to control for unobservables,” Stata<br />
Journal, 2003, 4 (2), 113–123. 27, 35, 40<br />
Rodrik, D., Has <strong>Globalization</strong> Gone Too Far?, Washington DC: Institute<br />
for International Economics, 1997. 104<br />
Roodman, David, “How to do xtabond2: An introduction to difference<br />
<strong>and</strong> system GMM in Stata,” Center for Global Development Discussion<br />
Paper, 2006, 103. 37<br />
Schott, Peter K., “Across-product versus within-product specialization in<br />
international trade,” Quarterly Journal of Economics, 2004, 119 (2),<br />
647–678. 70<br />
Smarzynska Javorcik, Beata, “Does foreign direct investment increase the<br />
productivity of domestic firms? In search of spillovers through backward<br />
linkages,” American Economic Review, 2004, 94 (3), 605–627. 9<br />
Smith, Keith, “Measuring innovation,” in Jan Fagerberg, David C. Mowery,<br />
<strong>and</strong> Richard R. Nelson, eds., The Oxford H<strong>and</strong>book of Innovation,<br />
Oxford: Oxford University Press, 2005, pp. 148–177. 134<br />
Solow, Robert M., “Technical change <strong>and</strong> the aggregate production function,”<br />
The Review of Economics <strong>and</strong> Statistics, 1957, 39 (3), 312–320.<br />
9<br />
Stokey, Nancy L., “The volume <strong>and</strong> composition of trade between rich <strong>and</strong><br />
poor countries,” The Review of Economic Studies, 1991, 58 (1), 63–80.<br />
70<br />
Teirlinck, Peter, “Location <strong>and</strong> agglomeration of foreign R&D activities in<br />
a small open economy,” in Peter Teirlinck <strong>and</strong> André Spithoven, eds.,<br />
Beyond Borders: Internationalisation of R&D <strong>and</strong> Policy Implications<br />
for Small Open Economies, Amsterdam: Elsevier, 2005a, pp. 207–234.<br />
131, 146<br />
, “Report CIS 4,” Technical Report 2005b. 135, 159<br />
The Economist, “Belgium sinks in deeper,” 1999, June 10th, 1999. 11, 36<br />
, “Europe’s nascent merger boom,” 2005, September 1st, 2005. 102<br />
, “Cross-border mergers: Heavy Mittal,” 2006, February 2nd, 2006. 102
174 Bibliography<br />
Tybout, James R., “Plant <strong>and</strong> firm-level evidence on new trade theories,”<br />
in K Choi <strong>and</strong> James Harrigan, eds., H<strong>and</strong>book of International Trade,<br />
Oxford: Blackwell Publishing, 2003. 55<br />
UNCTAD, World Investment Report 2004: The Shift towards Services, New<br />
York: United Nations Conference on Trade <strong>and</strong> Development, 2004. 1<br />
Van Beveren, Ilke, “Footloose multinationals in Belgium?,” Review of<br />
World Economics, 2007a, 143 (3), 483–507. 36, 67<br />
, “Total factor productivity estimation: A practical review,” Licos Discussion<br />
Paper Series, 2007b, 182/2007. 36, 133<br />
Van Biesebroeck, Johannes, “Robustness of productivity estimates,”<br />
Journal of Industrial Economics, 2007, 55 (3), 529–569. 11, 51<br />
Van den Cruyce, Bart, “The impact of relocation, size, age, capital intensity<br />
<strong>and</strong> multinational group membership,” Federaal Planbureau Working<br />
Paper, 1999, 8-99. 106<br />
Van Leeuwen, George <strong>and</strong> Luuk Klomp, “On the contribution of innovation<br />
to multi-factor productivity growth,” Economics of Innovation<br />
<strong>and</strong> New Technology, 2006, 15 (4), 367 – 390. 137<br />
V<strong>and</strong>enbussche, Hylke <strong>and</strong> Chang Tan, “The taxation of multinationals:<br />
<strong>Firm</strong> level evidence for Belgium,” Licos Discussion Paper Series,<br />
2005, 160/2005. 102, 104<br />
, Karen Crabbé, <strong>and</strong> Boudewijn Janssen, “Is there regional tax<br />
competition? <strong>Firm</strong> level evidence for Belgium,” De Economist, 2005,<br />
153 (3), 257–276. 107<br />
Verbeek, Marno, A Guide to Modern Econometrics, West Sussex: John<br />
Wiley, 2000. 137<br />
VRWB, De voedingsindustrie in Vla<strong>and</strong>eren: Op weg naar 2010, Vlaamse<br />
Raad voor het Wetenschapsbeleid, 2003. 11<br />
Wagner, Joachim, “Exports <strong>and</strong> productivity: A survey of the evidence<br />
from firm-level data,” The World Economy, 2007, 30 (1), 60–82. 5, 129<br />
Wakelin, Katharine, “Innovation <strong>and</strong> export behaviour at the firm level,”<br />
Research Policy, 1998, 26 (7-8), 829–841. 132
Bibliography 175<br />
Wedervang, F., Development of a Population of Industrial <strong>Firm</strong>s, Oslo:<br />
Sc<strong>and</strong>inavian University Books, 1965. 10, 15, 20<br />
Wooldridge, Jeffrey M., “On estimating firm-level production functions<br />
using proxy variables to control for unobservables,” Unpublished<br />
manuscript., 2005. 16<br />
WTO, WTO Annual Report 2005, Geneva: World Trade Organization, 2005.<br />
1<br />
Yasar, Mahmut, Rafal Raciborski, <strong>and</strong> Brian P. Poi, “Production<br />
function estimation in Stata using the Olley <strong>and</strong> Pakes method,” Stata<br />
Journal, 2008, 8 (2), 221–231. 37
Doctoral Dissertations Faculty<br />
of Business <strong>and</strong> Economics<br />
from August 1, 1971<br />
1. GEPTS, Stefaan. Stability <strong>and</strong> efficiency of resource allocation processes<br />
in discrete commodity spaces. Leuven, KUL, 1971. 86 pp.<br />
2. PEETERS, Theo. Determinanten van de internationale h<strong>and</strong>el in fabrikaten.<br />
Leuven, Acco, 1971. 290 pp.<br />
3. VAN LOOY, Wim. Personeelsopleiding : een onderzoek naar investeringsaspekten<br />
van opleiding. Hasselt, Vereniging voor wetenschappelijk<br />
onderzoek in Limburg, 1971. VII, 238 pp.<br />
4. THARAKAN, Mathew. Indian exports to the European community<br />
: problems <strong>and</strong> prospects. Leuven, Faculty of economics <strong>and</strong> applied<br />
economics, 1972. X,343 pp.<br />
5. HERROELEN, Willy. Heuristische programmatie : methodologische<br />
benadering en praktische toepassing op complexe combinatorische problemen.<br />
Leuven, Aurelia scientifica, 1972. X, 367 pp.<br />
6. VANDENBULCKE, Jacques. De studie en de evaluatie van dataorganisatiemethodes<br />
en data-zoekmethodes. Leuven, s.n., 1973. 3 V.<br />
7. PENNYCUICK, Roy A. The economics of the ecological syndrome.<br />
Leuven, Acco, 1973. XII, 177 pp.<br />
8. KAWATA, T. Bualum. Formation du capital d’origine belge, dette<br />
publique et stratégie du développement au Zaire. Leuven, KUL, 1973.<br />
V, 342 pp.<br />
9. DONCKELS, Rik. Doelmatige oriëntering van de sectorale subsidiepolitiek<br />
in België : een theoretisch onderzoek met empirische toetsing.<br />
Leuven, K.U.Leuven, 1974. VII, 156 pp.<br />
177
178 Doctoral dissertations<br />
10. VERHELST, Maurice. Contribution to the analysis of organizational<br />
information systems <strong>and</strong> their financial benefits. Leuven, K.U.Leuven,<br />
1974. 2 V.<br />
11. CLEMEUR, Hugo. Enkele verzekeringstechnische vraagstukken in het<br />
licht van de nutstheorie. Leuven, Aurelia scientifica, 1974. 193 pp.<br />
12. HEYVAERT, Edward. De ontwikkeling van de moderne bank- en<br />
krediettechniek tijdens de zestiende en zeven-tiende eeuw in Europa<br />
en te Amsterdam in het bijzonder. Leuven, K.U.Leuven, 1975. 186 pp.<br />
13. VERTONGHEN, Robert. Investeringscriteria voor publieke investeringen<br />
: het uitwerken van een operationele theorie met een toepassing op<br />
de verkeersinfrastructuur. Leuven, Acco, 1975. 254 pp.<br />
14. Niet toegekend.<br />
15. VANOVERBEKE, Lieven. Microeconomisch onderzoek van de sectoriële<br />
arbeidsmobiliteit. Leuven, Acco, 1975. 205 pp.<br />
16. DAEMS, Herman. The holding company : essays on financial intermediation,<br />
concentration <strong>and</strong> capital market imperfections in the Belgian<br />
economy. Leuven, K.U.Leuven, 1975. XII, 268 pp.<br />
17. VAN ROMPUY, Eric. Groot-Brittannië en de Europese monetaire<br />
integratie : een onderzoek naar de gevolgen van de Britse toetreding<br />
op de gepl<strong>and</strong>e Europese monetaire unie. Leuven, Acco, 1975. XIII,<br />
222 pp.<br />
18. MOESEN, Wim. Het beheer van de staatsschuld en de termijnstructuur<br />
van de intrestvoeten met een toepassing voor België. Leuven,<br />
V<strong>and</strong>er, 1975. XVI, 250 pp.<br />
19. LAMBRECHT, Marc. Capacity constrained multi-facility dynamic lotsize<br />
problem. Leuven, KUL, 1976. 165 pp.<br />
20. RAYMAECKERS, Erik. De mens in de onderneming en de theorie van<br />
het producenten-gedrag : een bijdrage tot transdisciplinaire analyse.<br />
Leuven, Acco, 1976. XIII, 538 pp.<br />
21. TEJANO, Albert. Econometric <strong>and</strong> input-output models in development<br />
planning : the case of the Philippines. Leuven, KUL, 1976. XX,<br />
297 pp.
DOCTORAL DISSERTATIONS 179<br />
22. MARTENS, Bernard. Prijsbeleid en inflatie met een toepassing op<br />
België. Leuven, KUL, 1977. IV, 253 pp.<br />
23. VERHEIRSTRAETEN, Albert. Geld, krediet en intrest in de Belgische<br />
financiële sector. Leuven, Acco, 1977. XXII, 377 pp.<br />
24. GHEYSSENS, Lieven. International diversification through the government<br />
bond market : a risk-return analysis. Leuven, s.n., 1977. 188<br />
pp.<br />
25. LEFEBVRE, Chris. Boekhoudkundige verwerking en financiële verslaggeving<br />
van huurkooptransacties en ver-kopen op afbetaling bij ondernemingen<br />
die aan consumenten verkopen. Leuven, KUL, 1977. 228<br />
pp.<br />
26. KESENNE, Stefan. Tijdsallocatie en vrijetijdsbesteding : een econometrisch<br />
onderzoek. Leuven, s.n., 1978. 163 pp.<br />
27. VAN HERCK, Gustaaf. Aspecten van optimaal bedrijfsbeleid volgens<br />
het marktwaardecriterium : een risico-rendementsanalyse. Leuven,<br />
KUL, 1978. IV, 163 pp.<br />
28. VAN POECK, Andre. World price trends <strong>and</strong> price <strong>and</strong> wage development<br />
in Belgium : an investigation into the relevance of the Sc<strong>and</strong>inavian<br />
model of inflation for Belgium. Leuven, s.n., 1979. XIV, 260<br />
pp.<br />
29. VOS, Herman. De industriële technologieverwerving in Brazilië : een<br />
analyse. Leuven, s.n., 1978. onregelmatig gepagineerd.<br />
30. DOMBRECHT, Michel. Financial markets, employment <strong>and</strong> prices in<br />
open economies. Leuven, KUL, 1979. 182 pp.<br />
31. DE PRIL, Nelson. Bijdrage tot de actuariële studie van het bonusmalussysteem.<br />
Brussel, OAB, 1979. 112 pp.<br />
32. CARRIN, Guy. Economic aspects of social security : a public economics<br />
approach. Leuven, KUL, 1979. Onregelm. gepag.<br />
33. REGIDOR, Baldomero. An empirical investigation of the distribution<br />
of stock-market prices <strong>and</strong> weak-form effi-ciency of the Brussels stock<br />
exchange. Leuven, KUL, 1979. 214 pp.
180 Doctoral dissertations<br />
34. DE GROOT, Roger. Ongelijkheden voor stop loss premies gebaseerd<br />
op E.T. systemen in het kader van de veral-gemeende convexe analyse.<br />
Leuven, KUL, 1979. 155 pp.<br />
35. CEYSSENS, Martin. On the peak load problem in the presence of<br />
rationizing by waiting. Leuven, KUL, 1979. IX, 217 pp.<br />
36. ABDUL RAZK ABDUL. Mixed enterprise in Malaysia : the case study<br />
of joint venture between Malysian public corporations <strong>and</strong> foreign enterprises.<br />
Leuven, KUL, 1979. 324 pp.<br />
37. DE BRUYNE, Guido. Coordination of economic policy : a gametheoretic<br />
approach. Leuven, KUL, 1980. 106 pp.<br />
38. KELLES, Gerard. Dem<strong>and</strong>, supply, price change <strong>and</strong> trading volume<br />
on financial markets of the matching-order type. Vraag, aanbod, koersontwikkeling<br />
en omzet op financiële markten van het Europese type.<br />
Leuven, KUL, 1980. 222 pp.<br />
39. VAN EECKHOUDT, Marc. De invloed van de looptijd, de coupon en<br />
de verwachte inflatie op het opbrengstverloop van vastrentende finaciële<br />
activa. Leuven, KUL, 1980. 294 pp.<br />
40. SERCU, Piet. Mean-variance asset pricing with deviations from purchasing<br />
power parity. Leuven, s.n., 1981. XIV, 273 pp.<br />
41. DEQUAE, Marie-Gemma. Inflatie, belastingsysteem en waarde van de<br />
onderneming. Leuven, KUL, 1981. 436 pp.<br />
42. BRENNAN, John. An empirical investigation of Belgian price regulation<br />
by prior notification : 1975 - 1979 - 1982. Leuven, KUL, 1982.<br />
XIII, 386 pp.<br />
43. COLLA, Annie. Een econometrische analyse van ziekenhuiszorgen.<br />
Leuven, KUL, 1982. 319 pp.<br />
44. Niet toegekend.<br />
45. SCHOKKAERT, Eric. Modelling consumer preference formation. Leuven,<br />
KUL, 1982. VIII, 287 pp.<br />
46. DEGADT, Jan. Specificatie van een econometrisch model voor vervuilingsproblemen<br />
met proeven van toepassing op de waterverontreiniging<br />
in België. Leuven, s.n., 1982. 2 V.
DOCTORAL DISSERTATIONS 181<br />
47. LANJONG, Mohammad Nasir. A study of market efficiency <strong>and</strong> riskreturn<br />
relationships in the Malaysian capital market. s.l., s.n., 1983.<br />
XVI, 287 pp.<br />
48. PROOST, Stef. De allocatie van lokale publieke goederen in een economie<br />
met een centrale overheid en lokale overheden. Leuven, s.n., 1983. onregelmatig.<br />
gepagineerd.<br />
49. VAN HULLE, Cynthia (08/83). Shareholders’ unanimity <strong>and</strong> optimal<br />
corporate decision making in imperfect capital markets. s.l., s.n., 1983.<br />
147 pp. + appendix.<br />
50. VAN WOUWE, Martine (2/12/83). Ordening van risico’s met toepassing<br />
op de berekening van ultieme ruïnekansen. Leuven, s.n., 1983. 109<br />
pp.<br />
51. D’ALCANTARA, Gonzague (15/12/83). SERENA : a macroeconomic<br />
sectoral regional <strong>and</strong> national account econometric model for the Belgian<br />
economy. Leuven, KUL, 1983. 595 pp.<br />
52. D’HAVE, Piet (24/02/84). De vraag naar geld in België. Leuven, KUL,<br />
1984. XI, 318 pp.<br />
53. MAES, Ivo (16/03/84). The contribution of J.R. Hicks to macroeconomic<br />
<strong>and</strong> monetary theory. Leuven, KUL, 1984. V, 224 pp.<br />
54. SUBIANTO, Bambang (13/09/84). A study of the effects of specific<br />
taxes <strong>and</strong> subsidies on a firms’ R&D investment plan. s.l., s.n., 1984.<br />
V, 284 pp.<br />
55. SLEUWAEGEN, Leo (26/10/84). Location <strong>and</strong> investment decisions<br />
by multinational enterprises in Belgium <strong>and</strong> Europe. Leuven, KUL,<br />
1984. XII, 247 pp.<br />
56. GEYSKENS, Erik (27/03/85). Produktietheorie en dualiteit. Leuven,<br />
s.n., 1985. VII, 392 pp.<br />
57. COLE, Frank (26/06/85). Some algorithms for geometric programming.<br />
Leuven, KUL, 1985. 166 pp.<br />
58. STANDAERT, Stan (26/09/86). A study in the economics of repressed<br />
consumption. Leuven, KUL, 1986. X, 380 pp.
182 Doctoral dissertations<br />
59. DELBEKE, Jos (03/11/86). Trendperioden in de geldhoeveelheid van<br />
België 1877-1983 : een theoretische en empirische analyse van de ”Banking<br />
school” hypothese. Leuven, KUL, 1986. XII, 430 pp.<br />
60. VANTHIENEN, Jan (08/12/86). Automatiseringsaspecten van de specificatie,<br />
constructie en manipulatie van beslissings-tabellen. Leuven,<br />
s.n., 1986. XIV, 378 pp.<br />
61. LUYTEN, Robert (30/04/87). A systems-based approach for multiechelon<br />
production/inventory systems. s.l., s.n., 1987. 3V.<br />
62. MERCKEN, Roger (27/04/87). De invloed van de data base benadering<br />
op de interne controle. Leuven, s.n., 1987. XIII, 346 pp.<br />
63. VAN CAYSEELE, Patrick (20/05/87). Regulation <strong>and</strong> international<br />
innovative activities in the pharmaceutical industry. s.l., s.n., 1987.<br />
XI, 169 pp.<br />
64. FRANCOIS, Pierre (21/09/87). De empirische relevantie van de independence<br />
from irrelevant alternatives. Assumptie indis-crete keuzemodellen.<br />
Leuven, s.n., 1987. IX, 379 pp.<br />
65. DECOSTER, André (23/09/88). Family size, welfare <strong>and</strong> public policy.<br />
Leuven, KUL. Faculteit Economische en toegepaste economische<br />
wetenschappen, 1988. XIII, 444 pp.<br />
66. HEIJNEN, Bart (09/09/88). Risicowijziging onder invloed van vrijstellingen<br />
en herverzekeringen : een theoretische ana-lyse van optimaliteit<br />
en premiebepaling. Leuven, KUL. Faculteit Economische en<br />
toegepaste economische wetenschappen, 1988. onregelmatig gepagineerd.<br />
67. GEEROMS, Hans (14/10/88). Belastingvermijding. Theoretische analyse<br />
van de determinanten van de belastingontduiking en de belastingontwijking<br />
met empirische verificaties. Leuven, s.n., 1988. XIII, 409,<br />
5 pp.<br />
68. PUT, Ferdi (19/12/88). Introducing dynamic <strong>and</strong> temporal aspects in<br />
a conceptual (database) schema. Leuven, KUL. Faculteit Economische<br />
en toegepaste economische wetenschappen, 1988. XVIII, 415 pp.<br />
69. VAN ROMPUY, Guido (13/01/89). A supply-side approach to tax<br />
reform programs. Theory <strong>and</strong> empirical evidence for Belgium. Leuven,<br />
KUL. Faculteit Economische en toegepaste economische wetenschappen,<br />
1989. XVI, 189, 6 pp.
DOCTORAL DISSERTATIONS 183<br />
70. PEETERS, Ludo (19/06/89). Een ruimtelijk evenwichtsmodel van de<br />
graanmarkten in de E.G. : empirische specificatie en beleidstoepassingen.<br />
Leuven, K.U.Leuven. Faculteit Economische en toegepaste economische<br />
wetenschappen, 1989. XVI, 412 pp.<br />
71. PACOLET, Jozef (10/11/89). Marktstructuur en operationele efficiëntie<br />
in de Belgische financiële sector. Leuven, K.U.Leuven. Faculteit Economische<br />
en toegepaste economische wetenschappen, 1989. XXII, 547 pp.<br />
72. VANDEBROEK, Martina (13/12/89). Optimalisatie van verzekeringscontracten<br />
en premieberekeningsprincipes. Leuven, K.U.Leuven. Faculteit<br />
Economische en toegepaste economische wetenschappen, 1989. 95 pp.<br />
73. WILLEKENS, Francois. Determinance of government growth in industrialized<br />
countries with applications to Belgium. Leuven, K.U.Leuven.<br />
Faculteit Economische en toegepaste economische wetenschappen, 1990.<br />
VI, 332 pp.<br />
74. VEUGELERS, Reinhilde (02/04/90). Scope decisions of multinational<br />
enterprises. Leuven, K.U.Leuven. Faculteit Economische en toegepaste<br />
economische wetenschappen, 1990. V, 221 pp.<br />
75. KESTELOOT, Katrien (18/06/90). Essays on performance diagnosis<br />
<strong>and</strong> tacit cooperation in international oligopolies. Leuven, K.U.Leuven.<br />
Faculteit Economische en toegepaste economische wetenschappen, 1990.<br />
227 pp.<br />
76. WU, Changqi (23/10/90). Strategic aspects of oligopolistic vertical<br />
integration. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1990. VIII, 222 pp.<br />
77. ZHANG, Zhaoyong (08/07/91). A disequilibrium model of China’s<br />
foreign trade behaviour. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 1991. XII, 256 pp.<br />
78. DHAENE, Jan (25/11/91). Verdelingsfuncties, benaderingen en foutengrenzen<br />
van stochastische grootheden geasso-cieerd aan verzekeringspolissen<br />
en -portefeuilles. Leuven, K.U.Leuven, Faculteit Economische en<br />
toegepaste economische wetenschappen, 1991. 146 pp.<br />
79. BAUWELINCKX, Thierry (07/01/92). Hierarchical credibility techniques.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1992. 130 pp.
184 Doctoral dissertations<br />
80. DEMEULEMEESTER, Erik (23/3/92). Optimal algorithms for various<br />
classes of multiple resource-constrained project scheduling problems.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 1992. 180 pp.<br />
81. STEENACKERS, Anna (1/10/92). Risk analysis with the classical<br />
actuarial risk model : theoretical extensions <strong>and</strong> applications to Reinsurance.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1992. 139 pp.<br />
82. COCKX, Bart (24/09/92). The minimum income guarantee. Some<br />
views from a dynamic perspective. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 1992. XVII,<br />
401 pp.<br />
83. MEYERMANS, Eric (06/11/92). Econometric allocation systems for<br />
the foreign exchange market : Specification, estimation <strong>and</strong> testing of<br />
transmission mechanisms under currency substitution. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 1992.<br />
XVIII, 343 pp.<br />
84. CHEN, Guoqing (04/12/92). Design of fuzzy relational databases based<br />
on fuzzy functional dependency. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 1992. 176 pp.<br />
85. CLAEYS, Christel (18/02/93). Vertical <strong>and</strong> horizontal category structures<br />
in consumer decision making : The nature of product hierarchies<br />
<strong>and</strong> the effect of br<strong>and</strong> typicality. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 1993. 348 pp.<br />
86. CHEN, Shaoxiang (25/03/93). The optimal monitoring policies for<br />
some stochastic <strong>and</strong> dynamic production processes. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 1993.<br />
170 pp.<br />
87. OVERWEG, Dirk (23/04/93). Approximate parametric analysis <strong>and</strong><br />
study of cost capacity management of computer configurations. Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 1993. 270 pp.<br />
88. DEWACHTER, Hans (22/06/93). Nonlinearities in speculative prices :<br />
The existence <strong>and</strong> persistence of nonlinearity in foreign exchange rates.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 1993. 151 pp.
DOCTORAL DISSERTATIONS 185<br />
89. LIN, Liangqi (05/07/93). Economic determinants of voluntary accounting<br />
choices for R & D expenditures in Belgium. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 1993.<br />
192 pp.<br />
90. DHAENE, Geert (09/07/93). Encompassing: formulation, properties<br />
<strong>and</strong> testing. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1993. 117 pp.<br />
91. LAGAE, Wim (20/09/93). Marktconforme verlichting van soevereine<br />
buitenl<strong>and</strong>se schuld door private crediteuren: een neo-institutionele<br />
analyse. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1993. 241 pp.<br />
92. VAN DE GAER, Dirk (27/09/93). Equality of opportunity <strong>and</strong> investment<br />
in human capital. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 1993. 172 pp.<br />
93. SCHROYEN, Alfred (28/02/94). Essays on redistributive taxation<br />
when monitoring is costly. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 1994. 203 pp.+ V.<br />
94. STEURS, Geert (15/07/94). Spillovers <strong>and</strong> cooperation in research<br />
<strong>and</strong> development. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1994. 266 pp.<br />
95. BARAS, Johan (15/09/94). The small sample distribution of the Wald,<br />
Lagrange multiplier <strong>and</strong> likelihood ratio tests for homogeneity <strong>and</strong> symmetry<br />
in dem<strong>and</strong> analysis: a Monte Carlo study. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 1994.<br />
169 pp.<br />
96. GAEREMYNCK, Ann (08/09/94). The use of depreciation in accounting<br />
as a signalling device. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 1994. 232 pp.<br />
97. BETTENDORF, Leon (22/09/94). A dynamic applied general equilibrium<br />
model for a small open economy. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 1994. 149 pp.<br />
98. TEUNEN, Marleen (10/11/94). Evaluation of interest r<strong>and</strong>omness in<br />
actuarial quantities. Leuven, K.U.Leuven, Faculteit Economische en<br />
toegepaste economische wetenschappen, 1994. 214 pp.
186 Doctoral dissertations<br />
99. VAN OOTEGEM, Luc (17/01/95). An economic theory of private<br />
donations. Leuven. K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1995. 236 pp.<br />
100. DE SCHEPPER, Ann (20/03/95). Stochastic interest rates <strong>and</strong> the<br />
probabilistic behaviour of actuarial functions. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 1995.<br />
211 pp.<br />
101. LAUWERS, Luc (13/06/95). Social choice with infinite populations.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 1995. 79 pp.<br />
102. WU, Guang (27/06/95). A systematic approach to object-oriented<br />
business modeling. Leuven, K.U.Leuven, Faculteit Economische en<br />
toegepaste economische wetenschappen, 1995. 248 pp.<br />
103. WU, Xueping (21/08/95). Term structures in the Belgian market :<br />
model estimation <strong>and</strong> pricing error analysis. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 1995.<br />
133pp.<br />
104. PEPERMANS, Guido (30/08/95). Four essays on retirement from the<br />
labor force. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1995. 128pp.<br />
105. ALGOED, Koen (11/09/95). Essays on insurance: a view from a dynamic<br />
perspective. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1995. 136 pp.<br />
106. DEGRYSE, Hans (10/10/95). Essays on financial intermediation, product<br />
differentiation, <strong>and</strong> market structure. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 1995. 218<br />
pp.<br />
107. MEIR, Jos (05/12/95). Het strategisch groepsconcept toegepast op<br />
de Belgische financiële sector. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 1995. 257 pp.<br />
108. WIJAYA, Miryam Lilian (08/01/96). Voluntary reciprocity as an informal<br />
social insurance mechanism: a game theoretic approach. Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische wetenschappen,<br />
1996. 124 pp.
DOCTORAL DISSERTATIONS 187<br />
109. VANDAELE, Nico (12/02/96). The impact of lot sizing on queueing<br />
delays : multi product, multi machine models. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 1996.<br />
243 pp.<br />
110. GIELENS, Geert (27/02/96). Some essays on discrete time target zones<br />
<strong>and</strong> their tails. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1996. 131 pp.<br />
111. GUILLAUME, Dominique (20/03/96). Chaos, r<strong>and</strong>omness <strong>and</strong> order<br />
in the foreign exchange markets. essays on the modelling of the markets.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 1996. 171 pp.<br />
112. DEWIT, Gerda (03/06/96). Essays on export insurance subsidization.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 1996. 186 pp.<br />
113. VAN DEN ACKER, Carine (08/07/96). Belief-function theory <strong>and</strong><br />
its application to the modeling of uncertainty in financial statement<br />
auditing. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1996. 147 pp.<br />
114. IMAM, Mahmood Osman (31/07/96). Choice of IPO Flotation Methods<br />
in Belgium in an Asymmetric Information Framework <strong>and</strong> Pricing<br />
of IPO’s in the Long-Run. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 1996. 221 pp.<br />
115. NICAISE, Ides (06/09/96). Poverty <strong>and</strong> Human Capital. Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische wetenschappen,<br />
1996. 209 pp.<br />
116. EYCKMANS, Johan (18/09/97). On the Incentives of Nations to Join<br />
International Environmental Agreements. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 1997. XV<br />
+ 348 pp.<br />
117. CRISOLOGO-MENDOZA, Lorelei (16/10/97). Essays on Decision<br />
Making in Rural Households: a study of three villages in the Cordillera<br />
Region of the Philippines. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 1997. 256 pp.<br />
118. DE REYCK, Bert (26/01/98). Scheduling Projects with Generalized<br />
Precedence Relations: Exact <strong>and</strong> Heuristic Procedures. Leuven, K.U.Leuven,
188 Doctoral dissertations<br />
Faculteit Economische en toegepaste economische wetenschappen, 1998.<br />
XXIV+337 pp.<br />
119. VANDEMAELE Sigrid (30/04/98). Determinants of Issue Procedure<br />
Choice within the Context of the French IPO Market: Analysis within<br />
an Asymmetric Information Framework. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 1998. 241<br />
pp.<br />
120. VERGAUWEN Filip (30/04/98). <strong>Firm</strong> Efficiency <strong>and</strong> Compensation<br />
Schemes for the Management of Innovative Activities <strong>and</strong> Knowledge<br />
Transfers. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1998. VIII+175 pp.<br />
121. LEEMANS Herlinde (29/05/98). The Two-Class Two-Server Queueing<br />
Model with Nonpreemptive Heterogeneous Priority Structures. Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 1998. 211 pp.<br />
122. GEYSKENS Inge (4/09/98). Trust, Satisfaction, <strong>and</strong> Equity in Marketing<br />
Channel Relationships. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 1998. 202 pp.<br />
123. SWEENEY John (19/10/98). Why Hold a Job ? The Labour Market<br />
Choice of the Low-Skilled. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 1998. 278 pp.<br />
124. GOEDHUYS Micheline (17/03/99). Industrial Organisation in Developing<br />
Countries, Evidence from Côte d’Ivoire. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 1999.<br />
251 pp.<br />
125. POELS Geert (16/04/99). On the Formal Aspects of the Measurement<br />
of Object-Oriented Software Specifications. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 1999.<br />
507 pp.<br />
126. MAYERES Inge (25/05/99). The Control of Transport Externalities:<br />
A General Equilibrium Analysis. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 1999. XIV +<br />
294 pp.<br />
127. LEMAHIEU Wilfried (5/07/99). Improved Navigation <strong>and</strong> Maintenance<br />
through an Object-Oriented Approach to Hypermedia Modelling.
DOCTORAL DISSERTATIONS 189<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 1999. 284 pp.<br />
128. VAN PUYENBROECK Tom (8/07/99). Informational Aspects of Fiscal<br />
Federalism. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1999. 192 pp.<br />
129. VAN DEN POEL Dirk (5/08/99). Response Modeling for Database<br />
Marketing Using Binary Classification. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 1999. 342 pp.<br />
130. GIELENS Katrijn (27/08/99). International Entry Decisions in the<br />
Retailing Industry: Antecedents <strong>and</strong> Performance Consequences. Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 1999. 336 pp.<br />
131. PEETERS Aneleen (16/12/99). Labour Turnover Costs, Employment<br />
<strong>and</strong> Temporary Work. Leuven, K.U.Leuven, Faculteit Economische en<br />
toegepaste economische wetenschappen, 1999. 207 pp.<br />
132. VANHOENACKER Jurgen (17/12/99). Formalizing a Knowledge Management<br />
Architecture Meta-Model for Integrated Business Process Management.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 1999. 252 pp.<br />
133. NUNES Paulo (12/04/2000). Contingent Valuation of the Benefits of<br />
Natural Areas <strong>and</strong> its Warglow Component. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 2000.<br />
XXI + 282 pp.<br />
134. VAN DEN CRUYCE Bart (7/04/2000). Statistische discriminatie van<br />
allochtonen op jobmarkten met rigide lonen. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 2000.<br />
135. REPKINE Alex<strong>and</strong>re (15/03/2000). Industrial Restructuring in Countries<br />
of Central <strong>and</strong> Eastern Europe : combining Branch-, <strong>Firm</strong>- <strong>and</strong><br />
Product-level Data for a better Underst<strong>and</strong>ing of Enterprises’Behaviour<br />
during Transition towards Market Economy. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 2000.<br />
136. AKSOY Yunus (21/06/2000). Essays on international price rigidities<br />
<strong>and</strong> exchange rates. Leuven, K.U.Leuven, Faculteit Economische en<br />
toegepaste economische wetenschappen, 2000. IX + 236 pp.
190 Doctoral dissertations<br />
137. RIYANTO Yohanes Eko (22/06/2000). Essays on the internal <strong>and</strong> external<br />
delegation of authority in firms. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 2000. VIII +<br />
280 pp.<br />
138. HUYGHEBAERT, Nancy (20/12/2000). The Capital Structure of<br />
Business Start-ups. Leuven, K.U.Leuven, Faculteit Economische en<br />
toegepaste economische wetenschappen, 2000, VIII + 332 pp.<br />
139. FRANCKX Laurent (22/01/2001). Ambient Inspections <strong>and</strong> Commitment<br />
in Environmental Enforcement. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 2001, VIII +<br />
286 pp.<br />
140. VANDILLE Guy (16/02/2001). Essays on the Impact of Income Redistribution<br />
on Trade. Leuven, K.U.Leuven, Faculteit Economische en<br />
toegepaste economische wetenschappen, 2001, VIII + 176 pp.<br />
141. MARQUERING Wessel (27/04/2001). Modeling <strong>and</strong> Forecasting Stock<br />
Market Returns <strong>and</strong> Volatility. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2001.<br />
142. FAGGIO Giulia (07/05/2001). Labor Market Adjustment <strong>and</strong> Enterprise<br />
Behavior in Transition. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2001, 150 pp.<br />
143. GOOS Peter (30/05/2001). The Optimal Design of Blocked <strong>and</strong> Splitplot<br />
experiments. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 2001.<br />
144. LABRO Eva (01/06/2001). Total Cost of Ownership Supplier Selection<br />
based on Activity Based Costing <strong>and</strong> Mathematical Programming.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 2001.<br />
145. VANHOUCKE Mario (07/06/2001). Exact Algorithms for various<br />
Types of Project Scheduling Problems. Nonregular Objectives <strong>and</strong><br />
time/cost Trade-offs. Leuven, K.U.Leuven, Faculteit Economische en<br />
toegepaste economische wetenschappen, 2001, 316 pp.<br />
146. BILSEN Valentijn (28/08/2001). Entrepreneurship <strong>and</strong> Private Sector<br />
Development in Central European Transition Countries. Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische wetenschappen,<br />
2001 XVI + 188 pp.
DOCTORAL DISSERTATIONS 191<br />
147. NIJS Vincent (10/08/2001). Essays on the dynamic Category-level Impact<br />
of Price promotions. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2001.<br />
148. CHERCHYE Laurens (24/09/2001). Topics in Non-parametric Production<br />
<strong>and</strong> Efficiency Analysis. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2001 VII + 169 pp.<br />
149. VAN DENDER Kurt (15/10/2001). Aspects of Congestion Pricing<br />
for Urban Transport. Leuven, K.U.Leuven, Faculteit Economische en<br />
toegepaste economische wetenschappen, 2001 VII + 203 pp.<br />
150. CAPEAU Bart (26/10/2001). In defence of the excess dem<strong>and</strong> approach<br />
to poor peasants’economic behaviour. Theory <strong>and</strong> Empirics of<br />
non-recursive agricultural household modeling. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 2001,<br />
XIII + 286 pp.<br />
151. CALTHROP Edward (09/11/2001). Essays in urban transport economics.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 2001.<br />
152. VANDER BAUWHEDE (03/12/2001). Earnings management in an<br />
Non-Anglo-Saxon environment. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2001, 408 pp.<br />
153. DE BACKER Koenraad (22/01/2002). Multinational firms <strong>and</strong> industry<br />
dynamics in host countries : the case of Belgium. Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische wetenschappen,<br />
2002, VII + 165 pp.<br />
154. BOUWEN Jan (08/02/2002). Transactive memory in operational workgroups.<br />
Concept elaboration <strong>and</strong> case study. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 2002,<br />
319 pp. + appendix 102 pp.<br />
155. VAN DEN BRANDE Inge (13/03/2002). The psychological contract<br />
between employer <strong>and</strong> employee : a survey among Flemish employees.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 2002, VIII + 470 pp.<br />
156. VEESTRAETEN Dirk (19/04/2002). Asset Price <strong>Dynamics</strong> under<br />
Announced Policy Switching. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2002, 176 pp.
192 Doctoral dissertations<br />
157. PEETERS Marc (16/05/2002). One dimensional cutting <strong>and</strong> packing<br />
: new problems <strong>and</strong> algorithms. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 2002.<br />
158. SKUDELNY Frauke (21/05/2002). Essays on the economic consequences<br />
of the European monetary union. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 2002.<br />
159. DE WEERDT Joachim (07/06/2002). Social networks, transfers <strong>and</strong><br />
insurance in developing countries. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2002, VI+129 pp.<br />
160. TACK Lieven (25/06/2002). Optimal run orders in design of experiments.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 2002, xxxi+344 pp.<br />
161. POELMANS Stephan (10/07/2002). Making workflow systems work.<br />
An investigation into the importance of task-appropriation fit, end-user<br />
support <strong>and</strong> other technological characteristics. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 2002,<br />
237 pp.<br />
162. JANS Raf (26/09/2002). Capacitated Lot Sizing Problems : New Applications,<br />
Formulations <strong>and</strong> Algorithms. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 2002.<br />
163. VIAENE Stijn (25/10/2002). Learning to Detect Fraud from enriched<br />
Insurance Claims Data (Context, Theory <strong>and</strong> Applications). Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische wetenschappen,<br />
2002, 315 pp.<br />
164. AYALEW Tekabe (08/11/2002). Inequality <strong>and</strong> Capital Investment in<br />
a Subsistence Economy.Leuven, K.U.Leuven, Faculteit Economische en<br />
toegepaste economische wetenschappen, 2002. V + 148 pp.<br />
165. MUES Christophe (12/11/2002). On the Use of Decision Tables <strong>and</strong><br />
Diagrams in Knowledge Modeling <strong>and</strong> Verification. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 2002.<br />
222 pp.<br />
166. BROCK Ellen (13/03/2003). The Impact of International Trade on<br />
European Labour Markets. K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 2002.
DOCTORAL DISSERTATIONS 193<br />
167. VERMEULEN Frederic (29/11/2002). Essays on the collective Approach<br />
to Household Labour Supply. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 2002. XIV +<br />
203 pp.<br />
168. CLUDTS Stephan (11/12/2002). Combining participation in decisionmaking<br />
with financial participation : theoretical <strong>and</strong> empirical perspectives.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 2002. XIV + 247 pp.<br />
169. WARZYNSKI Frederic (09/12/2003). The dynamic effect of competition<br />
on price cost margins <strong>and</strong> innovation. Leuven, K.U.Leuven, Faculteit<br />
Economische en toegepaste economische wetenschappen, 2003.<br />
170. VERWIMP Philip (14/01/2003). Development <strong>and</strong> genocide in Rw<strong>and</strong>a<br />
; a political economy analysis of peasants <strong>and</strong> power under the Habyarimana<br />
regime. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 2003.<br />
171. BIGANO Andrea (25/02/2003). Environmental regulation of the electricity<br />
sector in a European Market Framework. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 2003,<br />
XX + 310 pp.<br />
172. MAES Konstantijn (24/03/2003). Modeling the Term Structure of Interest<br />
Rates Across Countries. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2003, V+246 pp.<br />
173. VINAIMONT Tom (26/02/2003). The performance of One- versus<br />
Two-Factor Models of the Term Structure of Interest Rates. Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische wetenschappen,<br />
2003.<br />
174. OOGHE Erwin (15/04/2003). Essays in multi-dimensional social choice.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 2003, VIII+108 pp.<br />
175. FORRIER Anneleen (25/04/2003). Temporary employment, employability<br />
<strong>and</strong> training. Leuven, K.U.Leuven, Faculteit Economische en<br />
toegepaste economische wetenschappen, 2003.<br />
176. CARDINAELS Eddy (28/04/2003). The role of cost system accuracy<br />
in managerial decision making. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2003. 144 pp.
194 Doctoral dissertations<br />
177. DE GOEIJ Peter (02/07/2003). Modeling Time-Varying Volatility <strong>and</strong><br />
Interest Rates. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 2003. VII+225 pp.<br />
178. LEUS Roel (19/09/2003). The generation of stable project plans.<br />
Complexity <strong>and</strong> exact algorithms. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2003.<br />
179. MARINHEIRO Carlos (23/09/2003). EMU <strong>and</strong> fiscal stabilisation policy<br />
: the case of small countries. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2003.<br />
180. BAESSENS Bart (24/09/2003). Developing intelligent systems for<br />
credit scoring using machine learning techniques. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 2003.<br />
181. KOCZY Laszlo (18/09/2003). Solution concepts <strong>and</strong> outsider behaviour<br />
in coalition formation games. Leuven, K.U.Leuven, Faculteit Economische<br />
en toegepaste economische wetenschappen, 2003.<br />
182. ALTOMONTE Carlo (25/09/2003). Essays on Foreign Direct Investment<br />
in transition countries : learning from the evidence. Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische wetenschappen,<br />
2003.<br />
183. DRIES Liesbeth (10/11/2003). Transition, Globalisation <strong>and</strong> Sectoral<br />
Restructuring: Theory <strong>and</strong> Evidence from the Polish Agri-Food Sector.<br />
Leuven, K.U.Leuven, Faculteit Economische en toegepaste economische<br />
wetenschappen, 2003.<br />
184. DEVOOGHT Kurt (18/11/2003). Essays On Responsibility-Sensitive<br />
Egalitarianism <strong>and</strong> the Measurement of Income Inequality. Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische wetenschappen,<br />
2003.<br />
185. DELEERSNYDER Barbara (28/11/2003). Marketing in Turbulent<br />
Times. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 2003.<br />
186. ALI Daniel (19/12/2003). Essays on Household Consumption <strong>and</strong><br />
Production Decisions under Uncertainty in Rural Ethiopia. Leuven,<br />
K.U.Leuven, Faculteit Economische en toegepaste economische wetenschappen,<br />
2003.
DOCTORAL DISSERTATIONS 195<br />
187. WILLEMS Bert (14/01/2004). Electricity networks <strong>and</strong> generation<br />
market power. Leuven, K.U.Leuven, Faculteit Economische en toegepaste<br />
economische wetenschappen, 2004.<br />
188. JANSSENS Gust (30/01/2004). Advanced Modelling of Conditional<br />
Volatility <strong>and</strong> Correlation in Financial Markets. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 2004.<br />
189. THOEN Vincent (19/01/2004). On the valuation <strong>and</strong> disclosure practices<br />
implemented by venture capital providers. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 2004.<br />
190. MARTENS Jurgen (16/02/2004). A fuzzy set <strong>and</strong> stochastic system<br />
theoretic technique to validate simulation models. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 2004.<br />
191. ALTAVILLA Carlo (21/05/2004). Monetary policy implementation<br />
<strong>and</strong> transmission mechanisms in the Euro area. Leuven, K.U.Leuven,<br />
Faculteit Economische en toegepaste economische wetenschappen, 2004.<br />
192. DE BRUYNE Karolien (07/06/2004). Essays in the location of economic<br />
activity. Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2004.<br />
193. ADEM Jan (25/06/2004). Mathematical programming approaches for<br />
the supervised classification problem. Leuven, K.U.Leuven, Faculteit<br />
Economische en Toegepaste Economische Wetenschappen, 2004.<br />
194. LEROUGE Davy (08/07/2004). Predicting Product Preferences : the<br />
effect of internal <strong>and</strong> external cues. Leuven, K.U.Leuven, Faculteit<br />
Economische en Toegepaste Economische Wetenschappen, 2004.<br />
195. VANDENBROECK Katleen (16/07/2004). Essays on output growth,<br />
social learning <strong>and</strong> l<strong>and</strong> allocation in agriculture : micro-evidence from<br />
Ethiopia <strong>and</strong> Tanzania. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2004.<br />
196. GRIMALDI Maria (03/09/004). The exchange rate, heterogeneity of<br />
agents <strong>and</strong> bounded rationality. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2004.<br />
197. SMEDTS Kristien (26/10/2004). Financial integration in EMU in the<br />
framework of the no-arbitrage theory. Leuven, K.U.Leuven, Faculteit<br />
Economische en Toegepaste Economische Wetenschappen, 2004.
196 Doctoral dissertations<br />
198. KOEVOETS Wim (12/11/2004). Essays on Unions, Wages <strong>and</strong> Employment.<br />
Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2004.<br />
199. CALLENS Marc (22/11/2004). Essays on multilevel logistic Regression.<br />
Leuven, K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2004.<br />
200. RUGGOO Arvind (13/12/2004). Two stage designs robust to model<br />
uncertainty. Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2004.<br />
201. HOORELBEKE Dirk (28/01/2005). Bootstrap <strong>and</strong> Pivoting Techniques<br />
for Testing Multiple Hypotheses. Leuven, K.U.Leuven, Faculteit<br />
Economische en Toegepaste Economische Wetenschappen, 2005.<br />
202. ROUSSEAU S<strong>and</strong>ra (17/02/2005). Selecting Environmental Policy Instruments<br />
in the Presence of Incomplete Compliance. Leuven, K.U.Leuven,<br />
Faculteit Economische en Toegepaste Economische Wetenschappen, 2005.<br />
203. VAN DER MEULEN Sofie (17/02/2005). Quality of Financial Statements<br />
: Impact of the external auditor <strong>and</strong> applied accounting st<strong>and</strong>ards.<br />
Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2005.<br />
204. DIMOVA Ralitza (21/02/2005). Winners <strong>and</strong> Losers during Structural<br />
Reform <strong>and</strong> Crisis : the Bulgarian Labour Market Perspective. Leuven,<br />
K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2005.<br />
205. DARKIEWICZ Grzegorz (28/02/2005). Value-at-risk in Insurance <strong>and</strong><br />
Finance : the Comonotonicity Approach. Leuven, K.U.Leuven, Faculteit<br />
Economische en Toegepaste Economische Wetenschappen, 2005.<br />
206. DE MOOR Lieven (20/05/2005). The Structure of International Stock<br />
Returns : Size, Country <strong>and</strong> Sector Effects in Capital Asset Pricing.<br />
Leuven, K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2005.<br />
207. EVERAERT Greetje (27/06/2005). Soft Budget Constraints <strong>and</strong> Trade<br />
Policies : The Role of Institutional <strong>and</strong> External Constraints. Leuven,<br />
K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2005.
DOCTORAL DISSERTATIONS 197<br />
208. SIMON Steven (06/07/2005). The Modeling <strong>and</strong> Valuation of complex<br />
Derivatives : the Impact of the Choice of the term structure<br />
model. Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2005.<br />
209. MOONEN Linda (23/09/2005). Algorithms for some graph-theoretical<br />
optimization problems. Leuven, K.U.Leuven, Faculteit Economische en<br />
Toegepaste Economische Wetenschappen, 2005.<br />
210. COUCKE Kristien (21/09/2005). <strong>Firm</strong> <strong>and</strong> industry adjustment under<br />
de-industrialisation <strong>and</strong> globalization of the Belgian economy. Leuven,<br />
K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2005.<br />
211. DECAMPS MARC (21/10/2005). Some actuarial <strong>and</strong> financial applications<br />
of generalized diffusion processes. Leuven, K.U.Leuven, Faculteit<br />
Economische en Toegepaste Economische Wetenschappen, 2005.<br />
212. KIM HELENA (29/11/2005). Escalation games: an instrument to analyze<br />
conflicts. The strategic approach to the bargaining problem. Leuven,<br />
K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2005.<br />
213. GERMENJI ETLEVA (06/01/2006). Essays on the economics of emigration<br />
from Albania. Leuven, K.U.Leuven, Faculteit Economische en<br />
Toegepaste Economische Wetenschappen, 2006.<br />
214. BELIEN JEROEN (18/01/2006). Exact <strong>and</strong> heuristic methodologies<br />
for scheduling in hospitals: problems, formulations <strong>and</strong> algorithms.<br />
Leuven, K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2006.<br />
215. JOOSSENS KRISTEL (10/02/2006). Robust discriminant analysis.<br />
Leuven, K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2006.<br />
216. VRANKEN LIESBET (13/02/2006). L<strong>and</strong> markets <strong>and</strong> production efficiency<br />
in transition economies. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2006.<br />
217. VANSTEENKISTE ISABEL (22/02/2006). Essays on non-linear modelling<br />
in international macroeconomics. Leuven, K.U.Leuven, Faculteit<br />
Economische en Toegepaste Economische Wetenschappen, 2006.
198 Doctoral dissertations<br />
218. WUYTS Gunther (31/03/2006). Essays on the liquidity of financial<br />
markets. Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2006.<br />
219. DE BLANDER Rembert (28/04/2006). Essays on endogeneity <strong>and</strong> parameter<br />
heterogeneity in cross-section <strong>and</strong> panel data. Leuven, K.U.Leuven,<br />
Faculteit Economische en Toegepaste Economische Wetenschappen, 2006.<br />
220. DE LOECKER Jan (12/05/2006). Industry dynamics <strong>and</strong> productivity.<br />
Leuven, K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2006.<br />
221. LEMMENS Aurélie (12/05/2006). Advanced classification <strong>and</strong> timeseries<br />
methods in marketing. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2006.<br />
222. VERPOORTEN Marijke (22/05/2006). Conflict <strong>and</strong> survival: an analysis<br />
of shocks, coping strategies <strong>and</strong> economic mobility in Rw<strong>and</strong>a,<br />
1990-2002. Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2006.<br />
223. BOSMANS Kristof (26/05/2006). Measuring economic inequality <strong>and</strong><br />
inequality aversion. Leuven, K.U.Leuven, Faculteit Economische en<br />
Toegepaste Economische Wetenschappen, 2006.<br />
224. BRENKERS R<strong>and</strong>y (29/05/2006). Policy reform in a market with<br />
differentiated products: applications from the car market. Leuven,<br />
K.U.Leuven, Faculteit Economische en Toegepaste Economische Wetenschappen,<br />
2006.<br />
225. BRUYNEEL Sabrina (02/06/2006). Self-econtrol depletion: Mechanisms<br />
<strong>and</strong> its effects on consumer behavior. Leuven, K.U.Leuven, Faculteit<br />
Economische en Toegepaste Economische Wetenschappen, 2006.<br />
226. FAEMS Dries (09/06/2006). Collaboration for innovation: Processes of<br />
governance <strong>and</strong> learning. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2006.<br />
227. BRIERS Barbara (28/06/2006). Countering the scrooge in each of us:<br />
on the marketing of cooperative behavior. Leuven, K.U.Leuven, Faculteit<br />
Economische en Toegepaste Economische Wetenschappen, 2006.<br />
228. ZANONI Patrizia (04/07/2006). Beyond demography: Essays on diversity<br />
in organizations. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2006.
DOCTORAL DISSERTATIONS 199<br />
229. VAN DEN ABBEELE Alex<strong>and</strong>ra (11/09/2006). Management control<br />
of interfirm relations: the role of information. Leuven, K.U.Leuven,<br />
Faculteit Economische en Toegepaste Economische Wetenschappen, 2006.<br />
230. DEWAELHEYNS Nico (18/09/2006). Essays on internal capital markets,<br />
bankruptcy <strong>and</strong> bankruptcy reform. Leuven, K.U.Leuven, Faculteit<br />
Economische en Toegepaste Economische Wetenschappen, 2006.<br />
231. RINALDI Laura (19/09/2006). Essays on card payments <strong>and</strong> household<br />
debt. Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2006.<br />
232. DUTORDOIR Marie (22/09/2006). Determinants <strong>and</strong> stock price effects<br />
of Western European convertible debt offerings: an empirical<br />
analysis. Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2006.<br />
233. LYKOGIANNI Elissavet (20/09/2006). Essays on strategic decisions of<br />
multinational enterprises: R&D decentralization, technology transfers<br />
<strong>and</strong> modes of foreign entry. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2006.<br />
234. ZOU Jianglei (03/10/2006). Inter-firm ties, plant networks, <strong>and</strong> multinational<br />
firms: essays on FDI <strong>and</strong> trade by Japanse firms. Leuven,<br />
K.U.Leuven, Faculteit Economische en Toegepaste Economische Wetenschappen,<br />
2006.<br />
235. GEYSKENS Kelly (12/10/2006). The ironic effects of food temptations<br />
on self-control performance. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2006.<br />
236. BRUYNSEELS Liesbeth (17/10/2006). Client strategic actions, goingconcern<br />
audit opinions <strong>and</strong> audit reporting errors. Leuven, K.U.Leuven,<br />
Faculteit Economische en Toegepaste Economische Wetenschappen, 2006.<br />
237. KESSELS Roselinde (23/10/2006). Optimal designs for the measurement<br />
of consumer preferences. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2006.<br />
238. HUTCHINSON John (25/10/2006). The size distribution <strong>and</strong> growth<br />
of firms in transition countries. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2006.
200 Doctoral dissertations<br />
239. RENDERS Annelies (26/10/2006). Corporate governance in Europe:<br />
The relation with accounting st<strong>and</strong>ards choice, performance <strong>and</strong> benefits<br />
of control. Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2006.<br />
240. DE WINNE Sophie (30/10/2006). Exploring terra incognita: human<br />
resource management <strong>and</strong> firm performance in small <strong>and</strong> medium-sized<br />
businesses. Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2006.<br />
241. KADITI Eleni (10/11/2006). Foreign direct investments in transition<br />
economies. Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2006.<br />
242. ANDRIES Petra (17/11/2006). Technology-based ventures in emerging<br />
industries: the quest for a viable business model. Leuven, K.U.Leuven,<br />
Faculteit Economische en Toegepaste Economische Wetenschappen, 2006.<br />
243. BOUTE Robert (04/12/2006). The impact of replenishment rules<br />
with endogenous lead times on supply chain performance. Leuven,<br />
K.U.Leuven, Faculteit Economische en Toegepaste Economische Wetenschappen,<br />
2006.<br />
244. MAES Johan (20/12/2006). Corporate entrepreneurship: an integrative<br />
analysis of a resource-based model. Evidence from Flemish enterprises.<br />
Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2006.<br />
245. GOOSSENS Dries (20/12/2006). Exact methods for combinatorial<br />
auctions. Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2006.<br />
246. GOETHALS Frank (22/12/2006). Classifying <strong>and</strong> assessing extended<br />
enterprise integration approaches. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2006.<br />
247. VAN DE VONDER Stijn (22/12/2006). Proactive-reactive procedures<br />
for robust project scheduling. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2006.<br />
248. SAVEYN Bert (27/02/2007). Environmental policy in a federal state.<br />
Leuven, K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2007.
DOCTORAL DISSERTATIONS 201<br />
249. CLEEREN Kathleen (13/03/2007). Essays on competitive structure<br />
<strong>and</strong> product-harm crises. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2007.<br />
250. THUYSBAERT Bram (27/04/2007). Econometric essays on the measurement<br />
of poverty. Leuven, K.U.Leuven, Faculteit Economische en<br />
Toegepaste Economische Wetenschappen, 2007.<br />
251. DE BACKER Manu (07/05/2007). The use of Petri net theory for business<br />
process verification. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2007.<br />
252. MILLET Kobe (15/05/2007). Prenatal testosterone, personality, <strong>and</strong><br />
economic behavior. Leuven, K.U.Leuven, Faculteit Economische en<br />
Toegepaste Economische Wetenschappen, 2007.<br />
253. HUYSMANS Johan (13/06/2007). Comprehensible predictive models:<br />
New methods <strong>and</strong> insights. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2007.<br />
254. FRANCKEN Nathalie (26/06/2007). Mass Media, Government Policies<br />
<strong>and</strong> Economic Development: Evidence from Madagascar. Leuven,<br />
K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2007.<br />
255. SCHOUBBEN Frederiek (02/07/2007). The impact of a stock listing<br />
on the determinants of firm performance <strong>and</strong> investment policy. Leuven,<br />
K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2007.<br />
256. DELHAYE Eef (04/07/2007). Economic analysis of traffic safety. Leuven,<br />
K.U.Leuven, Faculteit Economische en Toegepaste Economische<br />
Wetenschappen, 2007.<br />
257. VAN ACHTER Mark (06/07/2007). Essays on the market microstructure<br />
of financial markets. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2007.<br />
258. GOUKENS Caroline (20/08/2007). Desire for variety: underst<strong>and</strong>ing<br />
consumers’ preferences for variety seeking. Leuven, K.U.Leuven, Faculteit<br />
Economische en Toegepaste Economische Wetenschappen, 2007.<br />
259. KELCHTERMANS Stijn (12/09/2007). In pursuit of excellence: essays<br />
on the organization of higher education <strong>and</strong> research. Leuven,
202 Doctoral dissertations<br />
K.U.Leuven, Faculteit Economische en Toegepaste Economische Wetenschappen,<br />
2007.<br />
260. HUSSINGER Katrin (14/09/2007). Essays on internationalization, innovation<br />
<strong>and</strong> firm performance. Leuven, K.U.Leuven, Faculteit Economische<br />
en Toegepaste Economische Wetenschappen, 2007.<br />
261. CUMPS Bjorn (04/10/2007). Business-ICT alignment <strong>and</strong> determinants.<br />
Leuven, K.U.Leuven, Faculteit Economische en Toegepaste<br />
Economische Wetenschappen, 2007.<br />
262. LYRIO Marco (02/11/2007). Modeling the yield curve with macro<br />
factors. Leuven, K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen,<br />
2007.<br />
263. VANPEE Rosanne (16/11/2007). Home bias <strong>and</strong> the implicit costs of<br />
investing abroad. Leuven, K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen,<br />
2007.<br />
264. LAMBRECHTS Olivier (27/11/2007). Robust project scheduling subject<br />
to resource breakdowns. Leuven, K.U.Leuven, Faculteit Economie<br />
en Bedrijfswetenschappen, 2007.<br />
265. DE ROCK Bram (03/12/2007). Collective choice behaviour: non parametric<br />
characterization. Leuven, K.U.Leuven, Faculteit Economie en<br />
Bedrijfswetenschappen, 2007.<br />
266. MARTENS David (08/01/2008). Building acceptable classification<br />
models for financial engineering applications. Leuven, K.U.Leuven,<br />
Faculteit Economie en Bedrijfswetenschappen, 2008.<br />
267. VAN KERCKHOVEN Johan (17/01/2008). Predictive modelling: variable<br />
selection <strong>and</strong> classification efficiencies. Leuven, K.U.Leuven, Faculteit<br />
Economie en Bedrijfswetenschappen, 2008.<br />
268. CIAIAN Pavel (12/02/2008). L<strong>and</strong>, EU accession <strong>and</strong> market imperfections.<br />
Leuven, K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen,<br />
2008.<br />
269. TRUYTS Tom (27/02/2008). Diamonds are a girl’s best friend: five<br />
essays on the economics of social status. Leuven, K.U.Leuven, Faculteit<br />
Economie en Bedrijfswetenschappen, 2008.
DOCTORAL DISSERTATIONS 203<br />
270. LEWIS Vivien (17/03/2008). Applications in dynamic general equilibrium<br />
macroeconomics. Leuven, K.U.Leuven, Faculteit Economie en<br />
Bedrijfswetenschappen, 2008.<br />
271. CAPPELLEN Tineke (04/04/2008). Worldwide coordination in a transnational<br />
environment: An inquiry into the work <strong>and</strong> careers of global<br />
managers. Leuven, K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen,<br />
2008.<br />
272. RODRIGUEZ Victor (18/04/2008). Material transfer agreements: research<br />
agenda choice, co-publication activity <strong>and</strong> visibility in biotechnology.<br />
Leuven, K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen,<br />
2008.<br />
273. QUAN Qi (14/04/2008). Privatization in China: Examining the endogeneity<br />
of the process <strong>and</strong> its implications for the performance of<br />
newly privatized firms. Leuven, K.U.Leuven, Faculteit Economie en<br />
Bedrijfswetenschappen, 2008.<br />
274. DELMOTTE Jeroen (30/04/2008). Evaluating the HR function: Empirical<br />
studies on HRM architecture <strong>and</strong> HRM system strength. Leuven,<br />
K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen, 2008.<br />
275. ORSINI Kristian (05/05/2008). Making work pay: Insights from microsimulation<br />
<strong>and</strong> r<strong>and</strong>om utility models. Leuven, K.U.Leuven, Faculteit<br />
Economie en Bedrijfswetenschappen, 2008.<br />
276. HOUSSA Romain (13/05/2008). Macroeconomic fluctuations in developing<br />
countries. Leuven, K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen,<br />
2008.<br />
277. SCHAUMANS Catherine (20/05/2008). Entry, regulation <strong>and</strong> economic<br />
efficiency: essays on health professionals. Leuven, K.U.Leuven,<br />
Faculteit Economie en Bedrijfswetenschappen, 2008.<br />
278. CRABBE Karen (21/05/2008). Essays on corporate tax competition<br />
in Europe. Leuven, K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen,<br />
2008.<br />
279. GELPER Sarah (30/05/2008). Economic time series analysis: Granger<br />
causality <strong>and</strong> robustness. Leuven, K.U.Leuven, Faculteit Economie en<br />
Bedrijfswetenschappen, 2008.
204 Doctoral dissertations<br />
280. VAN HOVE Jan (20/06/2008). The impact of technological innovation<br />
<strong>and</strong> spillovers on the pattern <strong>and</strong> direction of international trade. Leuven,<br />
K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen, 2008.<br />
281. DE VILLE DE GOYET Cédric (04/07/2008). Hedging with futures<br />
in agricultural commodity markets. Leuven, K.U.Leuven, Faculteit<br />
Economie en Bedrijfswetenschappen, 2008.<br />
282. FRANCK Tom (15/07/2008). Capital structure <strong>and</strong> product market<br />
interactions: evidence from business start-ups <strong>and</strong> private firms. Leuven,<br />
K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen, 2008.<br />
283. ILBAS Pelin (15/09/2008). Optimal monetary policy design in dynamic<br />
macroeconomics. Leuven, K.U.Leuven, Faculteit Economie en<br />
Bedrijfswetenschappen, 2008.<br />
284. GOEDERTIER Stijn (16/09/2008). Declarative techniques for modeling<br />
<strong>and</strong> mining business processes. Leuven, K.U.Leuven, Faculteit<br />
Economie en Bedrijfswetenschappen, 2008.<br />
285. LAMEY Lien (22/09/2008). The private-label nightmare: can national<br />
br<strong>and</strong>s ever wake up? Leuven, K.U.Leuven, Faculteit Economie en<br />
Bedrijfswetenschappen, 2008.<br />
286. VANDEKERCKHOVE Jan (23/09/2008). Essays on research <strong>and</strong> development<br />
with spillovers. Leuven, K.U.Leuven, Faculteit Economie en<br />
Bedrijfswetenschappen, 2008.<br />
287. PERNOT Eli (25/09/2008). Management control system design for<br />
supplier relationships in manufacturing: Case study evidence from<br />
the automotive industry. Leuven, K.U.Leuven, Faculteit Economie en<br />
Bedrijfswetenschappen, 2008.<br />
288. AERTS Kris (16/10/2008). Essays on the economics of evaluation:<br />
public policy <strong>and</strong> corporate strategies in innovation. Leuven, K.U.Leuven,<br />
Faculteit Economie en Bedrijfswetenschappen, 2008.<br />
289. BOUDT Kris (27/11/2008). Estimation of financial risk under nonnormal<br />
distributions. Leuven, K.U.Leuven, Faculteit Economie en<br />
Bedrijfswetenschappen, 2008.<br />
290. MARKIEWICZ Agnieszka (01/12/2008). Essays in exchange rate economics.<br />
Leuven, K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen,<br />
2008.
DOCTORAL DISSERTATIONS 205<br />
291. GLADY Nicolas (08/12/2008). Customer profitability modeling. Leuven,<br />
K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen, 2008<br />
292. MIERZEJEWSKI Fern<strong>and</strong>o (11/12/2008). Essays on liquidity-preference<br />
in markets with borrowing restrictions. Leuven, K.U.Leuven, Faculteit<br />
Economie en Bedrijfswetenschappen, 2008.<br />
293. VAN BEVEREN Ilke (15/12/2008). <strong>Globalization</strong> <strong>and</strong> firm dynamics.<br />
Leuven, K.U.Leuven, Faculteit Economie en Bedrijfswetenschappen,<br />
2008.