On Ricardian Equivalence and Twin Divergence: The Spanish ...

On Ricardian Equivalence and Twin Divergence:The Spanish Experience in the 2009 Crisis ∗M. Cardoso a and R. Doménech a,ba BBVA Economic Research Departmentb University of Valencia, SpainPreliminary and incomplete. November 23, 2008.AbstractDuring the current economic crisis in the Spanish economy, households are increasing very quicklytheir saving rate, the government budget has changed from surplus to a huge deficit and the currentaccount deficit is adjusting rapidly. In this paper we explore some possible explanations of thisevidence. Using a DGE model for the Spanish economy, we find that expansionary fiscal policieshave only small negative effects on the current account, even assuming a large proportion of non-Ricardian consumers. Therefore, the current negative correlation between the government balancebudget and the current account, or twin divergence, represents a departure from traditional modelsthat either emphasize the presence of Ricardian equivalence or that completely ignore it. Alternativeexplanations for the strong change in consumption and saving patterns within the economy (outputloss, deleveraging, uncertainty), seem like better candidates to explain the twin divergence.Keywords: Ricardian equivalence, current account, government budget deficit.JEL Classification: F32, F31, E62, F41, C32.1. IntroductionIn 2007 the Spanish economy exhibited a large current account deficit (10% of GDP), apublic budget surplus (2.2% of GDP), and a relatively low households saving rate (6.5% ofGDP), at least compared with their investment rate (almost 10% of GDP). However, withthe current economic crisis, the situation has changed dramatically, as Figures 1, 2 and∗ This paper has been prepared for the ECFIN Workshop on External imbalances and public finances in theEU, Brussels, 27 November 2009. The views expressed herein are the sole responsibility of the authors and do notreflect the opinions of their institutions. R. Doménech also thank the support of CICYT grant ECO2008-04669.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 312.011.010.09.08.07.06.0Households, savings and investment% of GDP13.012.011.010.09.08.07.06.05.04.019871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009 (p)Source: BBVA ERDSavingsInvestment5.04.0Figure 1: Households saving and investment rates (in percentage of GDP), Spain, 1995-2009.Figure 2: General Government Balance (in percentage of GDP), Spain, 1995-2009.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 410.05.0Spain, Net Borrowing and Lending by sectorsAs a percentage of GDPTotal economy (rigth)10.05.00.0­5.0­10.00.0­5.0­15.0­20.0198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009Source: BBVA ERDGeneral GovernmentFinancial InstitutionsNon Financial CorporationHouseholds(p) Forecast­10.0­15.0Figure 3: Current Account (in percentage of GDP), Spain, 1986-2009.motivated by precautionary saving, in response to the greater uncertainty and the highunemployment rate (17,4% in the first quarter of 2009), and the desire of reducing theirlevels of debt, after the dramatic fall in their wealth and the expectations of future higherreal interest rates, as a consequence of the international finance turmoil. Given these explanations,although this evidence is observational equivalent to Ricardian equivalence,it seems that only a fraction of the increase in households saving rate is explained by thelarge current deficit in government’s budget. In fact, our results show that private agentsonly compensate partially the effects of fiscal policies on the saving rate. Therefore, at thecurrent moment, the emphasis on future fiscal consolidation should be based more on thesustainability of government debt than on the effects on external imbalances. Accordingto these results, the current facts of the Spanish economy could be appropriately explainedby the presence of a large negative output shock as suggested by Kim and Roubini (2008)for the US, that has been tried to be counteracted with a large positive fiscal stimulus implementedby the Spanish Governemnt.The structure of this paper is the following, In Section 2 we present the generalequilibrium model for the Spanish economy that we use for economic policy evaluation.In Section 3 we analyse the effects of three different temporary fiscal policies: an increaseof public consumption, an increase in public investment and a reduction of the labourincome tax rate. In this section we discuss also the sensitivity of our results to changes insome crucial parameters. Finally, Section 4 presents the main conclusions.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 52. Theoretical frameworkIn this paper we use we use REMS, a Rational Expectations Model of the Spanish economydesigned by Boscá et al (2007), to analyse the effects of expansionary fiscal policies..REMS is a decentralized, small open economy model, where households, firms, policymakersand the external sector actively interact each period by trading one final good,two financial assets and three production factors 2 . In order to produce gross output, firmsemploy physical capital (public and private), labour and an intermediate input (energy).While private physical capital and energy are exchanged in a context of perfect competition,the labour market is not Walrasian. Households possess the available productionfactors. They also own all the firms operating in the economy. In such a scenario, eachrepresentative household rents physical capital and labour services out to firms, for whichthey are paid income in the form of interest and wages. New jobs are created after investingin search activities. The fact that exchange in the labour market is resource andtime-consuming generates a monopoly rent associated with each job match. It is assumedthat the worker and the firm bargain over these monopoly rents in Nash fashion.Each period the government faces a budget constraint where expenditure items arefinanced by means of public debt and various distortionary taxes. Intertemporal sustainabilityof fiscal balance is ensured by a conventional policy reaction function, whereby alump-sum tax transfer responds to the deviation of the debt to GDP ratio from its longtermtarget level.Monetary policy is geared by the European Central Bank (ECB) by means of interestrates movements, which target EMU inflation in which the weight of the Spanish economyis approximately 10 per cent.Each household is made up of working-age agents who may be active or inactive.In turn, active workers participating in the labour market may either be employed or unemployed.If unemployed, agents are actively searching for a job. Firms’ investment invacant posts is endogenously determined and so are job inflows. Finally, job destructionis taken as exogenous.2.1 HouseholdsFollowing Galí et al. (2007), liquidity-constrained consumers are incorporated into thestandard Keynesian model. There are two types of representative households. One representativehousehold enjoys unlimited access to capital markets, so its members substituteconsumption intertemporally in response to changes in interest rates. We will refer to thesehouseholds as "Ricardian or optimizing consumers". Another representative household2 In this section we will present a brief description of the main characteristics of the model. For a completeaccount of the model, see Boscá et. al (2007).

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 6does not have access to capital markets, so its members can only consume out of currentlabour income. We will refer to these liquidity-constrained consumers as "rule-of-thumb(RoT) consumers".Both types of households maximize intertemporal utility by selecting streams ofconsumption and leisure. Household members may be either employed or unemployed,but are able to fully insure each other against fluctuations in employment, as in Andolfatto(1996) or Merz (1995). For the sake of simplicity, we assume that only optimizingconsumers hold money balances, as well as foreign and domestic bonds. However, taxesare levied on both Ricardian and the liquidity-constrained consumers.Optimizing householdsRicardian households face the following maximization programme:maxc t ,n t ,j t ,k t,bt o,bo,emut ,m o t∞E t ∑ βln t c o t − hc o t−1 + no (T − lt−1 φ 1t ) 1−η1+ (1 − n o t−11 − η)φ (T − l 2t ) 1−η2+ χ1 − η m ln (m o t )t=0subject to r t (1 − τ k t ) + τk t δ k o t−1 + w t 1 − τ l t−(1 + τ c t )co tm o t−11+π c tP c tP t− Pi tP tj o tn o t−1 l 1t + rr t s 1 − n o t−1 l2t + 1 − τ l t g st − trh t ++ 1 + rt−1 n b ot−11+π c + (1 + r emu t−1t1+π c t 1 + φ 2− γ A γ Nm o t + bo t + bo,emu tφbt= 0 j otk o t−1t−1 ) bo,emu(1)(2)γ A γ N k o t = j o t + (1 − δ)k o t−1 (3)γ N n o t = (1 − σ)n o t−1 + ρw t s(1 − no t−1 ) (4)All variables in the maximisation problem above are stationary 3 . In our notation, variablesindexed by r and o respectively denote RoT and optimizing households. Non-indexedvariables apply indistinctly to both types of households. Thus, c o t ,no t−1 and s(1 − no t−1 )represent, consumption, the employment rate and the unemployment rate of Ricardianhouseholds. s is the share of the non-employed searching for a job, which is assumed to beexogenous. 4 T, l 1t and l 2t are the time endowment, hours worked per employee, and hours3 Thus, variables are expressed in efficiency units per persons of a working age.4 For simplicity, we assume that the leisure utility of the unemployed searching for a job is the same as for the

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 7devoted to job searching by the unemployed. Note that, while the household decides overl 1t , the same cannot be said of l 2t : time devoted to job searching is assumed to be a functionof overall economic activity, so that individual households take it as given 5 .Several parameters are presented in the utility function of Ricardian households.Future utility is discounted at a rate of β ∈ (0, 1). The parameter −η 1 measures the negativeof the Frisch elasticity of labour supply. As consumption is subject to habits, the parameterh takes a positive value. In general φ 1 ̸= φ 2, i.e., the subjective value of leisure imputedby workers may vary across employment statuses. For simplicity, we adopt the money-inthe-utilityfunction approach to incorporate money into the model.Maximization of (1) is constrained as follows. First, the budget constraint (2) describesthe various sources and uses of income. The term w t1 − τ l n o t−1 l 1t captures netlabour income earned by the fraction of employed workers, where w t stands for hourlyreal wages. The product rr t w t1 − τ l s 1 − n o t−1l2t measures unemployment benefitsaccruing to the unemployed, where rr t denotes the (exogenous) replacement rate of theunemployment subsidy to the market wage. There are four assets in the economy, namelyprivate physical capital (k o t ), domestic and Euro-zone bonds (bo t and bow t ) and money balances(m o t ). All assets are owned by Ricardian households. Barring money, the remainingassets yield some remuneration. Net return on capital is captured by r t k o t−1 (1 − τk ) +τ k δk o t−1 , where r t represents the gross return on physical capital. Note that depreciation istax-deductible as reflected in τ k δk o t−1. Interest payments on domestic and foreign debt arerespectively captured by rt−1n bt−1o1+π c , and r emu bt−1owt t−1 1+π c , where r n and r emu represent the nominaltinterest rates on domestic and EMU bonds, which may differ because of a risk premium.The other two expenditure categories are lump-sum transfers, trh t , and other governmenttransfers, g st .Total revenues can either be invested in private capital or spent on consumption.The household’s consumption and investment are respectively given by (1 + τ c ) Pc tP tc o t andPti P tjto 1 + φ jt2 k, where τ c is the consumption income tax. Note that total investmentt−1 outlays are affected by the increasing marginal costs of installation jto 1 + φ jt2 k. Also,t−1the presence in the model of the relative prices Pt c/P t and Pt i/P t implies that a distinctionis made between the three deflators of consumption, investment and aggregate output.The remaining constraints faced by Ricardian households concern the laws of motionfor capital and employment. Each period the private capital stock k o t depreciates at theexogenous rate δ and is accumulated through investment, jt o . Thus, it evolves accordingnon-active.5 We assume for empirical reasons that the search effort undertaken by unemployed workers varies along thecycle depending positively on the GDP growth rate.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 9cation of capital over time:q t = 1 + πc t+11 + r n t"r t+1 (1 − τ k t+1 ) + τk t+1 δ + φ 2j 2 t+1k 2 t+ q t+1 (1 − δ)where q t is Tobin’s q, that is obtained from the first order condition for investment:"!#q t = Pi tjto 1 + φP tk o t−1#(9)(10)Deriving the Lagrangian with respect to foreign debt b o,emutwith (7) yieldsand combining with1 + r n t = φ bt(1 + r nwt ) (11)implying that the interest parity condition holds between domestic and EMU bonds to theextent that they are perfect substitutes. Note that (11) slightly differs from the standarduncovered interest parity condition in that there is no risk associated with exchange ratemovements, as both domestic and foreign bonds are expressed in the same currency. Notethat the specification above assumes that Ricardian households incur in a risk premiumwhen buying foreign bonds (φ bt). Specifically, the risk premium is made up of a functionof net foreign asset holdings in the following wayln φ bt= −φ b(exp (b o,emut ) − 1) (12)Finally, it is easy to show that the first order condition for money can be rewrittenas a money demand function by using the current-value shadow price of income (6)m o t = 1γ A γ Nχ m (1 + τ c t) Pc tP t1 + r n tr n t1c o t −hco t−11− βhc o t+1 −hco t (13)Rule-of-thumb householdsRoT households do not benefit from access to capital markets and also do not hold money,so they face the following maximization programme 6 :maxc r t ,nr t∞E t ∑ βln t c r t − hc r t−1 + nr (T − lt−1 φ 1t ) 1−η1+ (1 − n r t−11 − η)φ (T − l 2t ) 1−η21 − ηt=06 Alternatively, Coenen, McAdam and Straub (2008) allow these households to smooth consumption by changingmoney holdings.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 10subject to the law of motion of employment (4) and the specific liquidity constraint wherebyeach period’s consumption expenditure must be equal to current labour income and governmenttransfers, as reflected in:w t1 − τ l tn r t−1 l 1t + rr t s 1 − n r t−1 l2t + gst 1 − τ l t − trh t − (1 + τ c t)c r Ptct = 0 (14)P tγ N n r t = (1 − σ)n r t−1 + ρw t s(1 − nr t−1 ) (15)Note that RoT consumers do not save and, as a result, they do not hold any assets. Thisfeature of RoT consumers considerably simplifies the solution to the optimization programme,which is characterized by the following equations concerning optimal consumption,c r t , and optimal employment, nr t :λ r 1t = 1(P c t /P t)c r t (1 + τc t ) 1c r t − hcr t−1!h− βc r t+1 − hcr t(16)γ N λ r 3t = βE t(φ 1(T−l 1t+1 ) 1−η1−η(T−l− φ 2t ) 1−η2 1−η+ λ1t+1 r w t+1 1 − τ ll+1(1 − σ) − χ1 ρ w t+1+λ r 3t+1(l 1t+1 − rr t+1 sl 2t ))(17)AggregationAggregate consumption and employment can be defined as a weighted average of thecorresponding variables for each household type 7 :c t = (1 − λ r ) c o t + λ r c r t (18)n t = (1 − λ r ) n o t + λ r n r t (19)For the variables that exclusively concern Ricardian households, aggregation is merelyperformed as:k t = (1 − λ r ) k o t (20)7 1 − λ r and λ r denote the constant fractions in the working-age population of Ricardian and RoT consumers.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 11j t = (1 − λ r ) j o t (21)b t = (1 − λ r ) b o t (22)b emut= (1 − λ r ) b oemut (23)m t = (1 − λ r ) m o t (24)2.2 Factor demandsProduction in the economy takes place at two different levels. At the lower level, an infinitenumber of monopolistically competing firms produce differentiated intermediategoods (y i ), which imperfectly substitute each other in the production of the final good.These differentiated goods are then aggregated by competitive retailers into a final domesticgood (y) using a CES aggregator.Intermediate producers solve a two-stage problem. In the first stage, each firm facesa cost minimization problem which results in optimal demands for production factors.When choosing optimal streams of capital, energy, employment and vacancies intermediateproducers set prices by varying the mark-up according to demand conditions. Varietyproducer i ∈ (0, 1) uses three inputs: a composite input of private capital and energy,labour and public capital, so that technological possibilities are given by:y it = z it hak −ρit−1i − 1 1−α ρ+ (1 − a)e−ρit(n it−1 l i1t ) α k p ζit−1(25)where all variables are scaled by the trend component of total factor productivity and z trepresents a transitory technology shock. Each variety producer rents physical capital,k t−1 , and labour services, n t−1 l 1t , from households, and uses public capital services, k p t−1 ,provided by the government. Intermediate energy inputs e t can be either imported fromabroad or produced at home. The technical elasticity of substitution between private capitaland energy is given by1+ρ 1 . α ∈ (0, 1) is a distribution parameter: it determines relativefactor shares in the steady state. Furthermore, it is convenient to denote capital services

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 12by k iet as:k iet =hak −ρit−1i − 1ρ+ (1 − a)e−ρit(26)Factor demands are obtained by solving the cost minimization problem faced byeach variety producer (we drop the industry index i when no confusion arises)subject to∞min E tk t ,n t ,v t ,e t∑ β t λo 1t+1λ o r t k t−1 + w t (1 + τ sc ) n t−1 l 1t + κ v v t + Pe te t (1 + τ e )t=0 1tP t(27)y t = z it hak −ρt−1i − 1 1−α ρ+ (1 − a)e−ρt(n t−1 l 1t ) α k p ζt−1 − κ f (28)γ N n t = (1 − σ)n t−1 + ρ f t v t (29)where, in accordance with the ownership structure of the economy, future profits are discountedat the household relevant rate β. κ v captures recruiting costs per vacancy, κ f is anentry cost which ensures that extraordinary profits vanish in imperfectly-competitive equilibrium,τ sc is the social security tax rate levied on gross wages 8 , and ρ f tis the probabilitythat a vacancy will be filled in any given period t. It is worth noting that the probability offilling a vacant post ρ f tis exogenous from the firm’s perspective. However, from the perspectiveof the overall economy, this probability is endogenously determined according tothe following Cobb-Douglas matching function:ρ w t (1 − n t−1) = ρ f t v t = χ 1 v χ 2t [s (1 − n t−1 ) l 2t ] 1−χ 2(30)Under the assumption of symmetry, the solution to the optimization programmeabove generates the following first order conditions for private capital, employment, energyand the number of vacancies y 1+ρr t+1 = (1 − α)mc t+1 ket+1t+1 a(31)k et+1 k tγ N λ nd λ o 1t+1 yt = βE tλ1to αmc t+1t+1 − wn t+1 (1 + τ sct+1 )l 1t+1 + λ ndt+1 (1 − σ) t(32)8 Note that, in our specification, firms bear the statutory incidence of social security contributions.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 13(1 − α)(1 − a)mc ty tke t−1 kete t 1+ρ= Pe tP t(1 + τ e t) (33)κ v v t = λ ndt χ 1 χ 2 v χ 2t(s(1 − n t−1 )l 2t ) 1−χ 2 (34)where the real marginal cost (mc t ) corresponds to the Lagrange multiplier associated withthe first restriction (28), whereas λ ndt denotes the Lagrange multiplier associated with thesecond restriction (29).2.3 Pricing behavior of intermediate firms: the New Phillips curveIntermediate firms enjoy market power and are, therefore, price setters. Each intermediatefirm produces a variety y i and faces a downward-sloping demand curve that takes theform below:y it = y tPitP t −ε(35)wherePitis the relative price of variety y i , ε can be expressed in terms of the elasticityof substitution between intermediate goods, ς ≥ 0, as ε = (1 + ς) /ς, and y t represents theproduction of the final product as defined byP t Z 1y t =01+ς Z 1y 1/1+ςitdiand P t = P − 1 −ςςitdi(36)0Variety producers act as monopolists and choose prices when allowed. We use thewell-known Calvo hypothesis (Calvo, 1983), thereby assuming some overlapping adjustmentin prices. Those firms that do not reset their prices optimally at a given date adjustthem according to a simple indexation rule to catch up with lagged inflation. Thus, eachperiod a proportion θ of firms simply set P it = (1 + π t−1 ) κ P it−1 (with κ representing thedegree of indexation), while only a measure 1 − θ of firms set their prices, eP it , to maximizethe present value of expected profits. Consequently, 1 − θ represents the probabilityof adjusting prices each period, whereas θ can be interpreted as a measure of price rigidity.Thus, the maximization problem of the representative variety producer generates a firstorder condition given by:eP it =hiε ∑ ∞ j=0 (βθ)j E t ρ it,t+jP ε+1t+j mc it+jy t+j π −εt+jhε − 1∑ ∞ j=0 (βθ)j E t ρ it,t+jPt+j ε y t+jπ (1−ε) i (37)t+j

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 14where ρ t,t+jis a price kernel which captures the marginal utility of an additional unit ofprofits accruing to optimizing households at t + j. The corresponding aggregate priceindex is equal toP t =hθ π κ t−1 P i 1−εt−1 + (1 − θ) eP t1−ε11−εAs is standard in the literature 9 , equation (38) can be used to obtain an expressionfor aggregate inflation in the form below:π t =β1 + κβ E tπ t+1 +(1 − βθ) (1 − θ)cmc t +θ(1 + κβ)(38)κ1 + κβ π t−1 (39)where cmc t measures the deviation of the firm’s marginal cost from the steady state, i.e.,mc t = ε−1ε(1 + cmc t ). Equation (39) is known in the literature as the New Phillips curve. Itparticipates of the conventional Phillips-curve philosophy that inflation is influenced byactivity in the short run, emphasizing real marginal costs as the relevant variable to theinflation process, which is seen as a forward-looking phenomenon. The reduced form ofthe New Phillips curve can be simplified as:π t = β f E t π t+1 + λ cmc t + β b π t−1 (40)2.4 Trade in the labour market: the labour contractThe key departure of search models from the competitive paradigm is that trading in thelabour market is subject to transaction costs. Each period, the unemployed engage insearch activities in order to find vacant posts spread over the economy. Costly search inthe labour market implies that there are simultaneous inflows into and outflows out of thestate of employment, so that an increase (reduction) in the stock of unemployment resultsfrom the predominance of job destruction (creation) over job creation (destruction). Stableunemployment occurs whenever inflows and outflows cancel out one another, i.e.,ρ f t v t = ρ w t (1 − n t−1) = χ 1 v χ 2t [(1 − n t−1 ) l 2t ] 1−χ 2= (1 − σ)n t−1 (41)Because it takes time (for households) and real resources (for firms) to make profitablecontacts, some pure economic rent emerges with each new job, which is equal tothe sum of the expected transaction (search) costs the firm and the worker will further incurif they refuse to match. The emergence of such rent gives rise to a bilateral monopolyframework.9 See, for example, Galí, Gertler and López-Salido (2001).

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 15Once a representative job-seeking worker and vacancy-offering firm match they negotiatea labour contract in hours and wages. We therefore stick to the efficient-bargaininghypothesis instead of the right-to-manage hypothesis (see Trigari, 2006, for further detailsabout the implications of these two different hypotheses).Several wage and hours determination schemes can be applied to a bilateral monopolyframework. In particular, we will assume that firms and workers negotiate by means of aNash bargain, so the outcome of the bargaining process maximizes the productmax λ r λr 3tw t+1 , l 1t+1 λ1tr+ (1 − λ r ) λo λw 3tλ1toλ ndt (1−λ w )(42)where λ w ∈ (0, 1) reflects the worker’s bargaining power. The first term in brackets representsthe representative worker surplus while the second is the firm surplus. More specifically,λ3t o /λo 1t and λr 3t /λr 1t respectively denote the earning premium (in terms of consumption)of employment over unemployment for a Ricardian and a RoT worker. Similarly, λ ndtrepresents the profit premium of a filled over an unfilled vacancy for the representativefirm. Note that this bargaining scheme features the same wage for all workers, irrespectiveof whether they are Ricardian or RoT.Optimal real wage and hours worked (42) satisfy the following conditions (seeBoscá et. al, 2007 for further details):(1 + τ sct )w t l 1t =λ wh1 − (1 − λ w ) rr t s l 2t(1 − λ w ) (1 + τsc+ hit )1 − (1 − λ w ) rr t s l 2t 1 − τ l 64l t1t2"il 1t ˜λ3tλ 1t−1y tαmc t + (1 − σ) κ vn t−1˜λ1tλ 1t−1 [ρ w t− (1 − σ)] −#ρ f t ˜1λ 1t−13 u t75λ 1t−1˜λ1t(43)!˜λ1ty tαmc t =λ 1t−1 n t−1 l 1,t˜1!(1 + τsct )λ 1t−1 1 − τ l φ 1[1 − l 1t ] −η (44)twhere:u t = Pc tc t (1 + τ c (T − lPt)φ 1t+1 ) 1−η1t 1 − η! ˜λ1t=λ 1t−1λ r λ1trλ1t−1r+ (1 − λ r )(T − l− 2t ) 1−η φ 21 − ηλ o 1tλ o 1t−1(45)

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 16! ˜1= λ r 1λ 1t−1 λ1t−1r + (1 − λ r )1λ o 1t−1! ˜λ3t=λ 1t−1λ r λ3trλ1t−1r+ (1 − λ r )λ o 3tλ o 1t−1Unlike the Walrasian outcome, the wage prevailing in the search equilibrium is at somepoint between the marginal rate of substitution of consumption for leisure and the marginalproductivity of labour, depending on worker bargaining power λ w . Put differently,the wage is a weighted average between the highest feasible wage (i.e., the marginal productivityof labour plus the cost of posting a vacancy corrected by the probability that thevacancy will be filled) and the lowest acceptable wage (i.e., the reservation wage as givenby the disutility from work corrected by the probability of finding a job). Consequently,the equilibrium wage depends on a number of policy parameters and institutional variablesdescribing labour market performance. Notice that when λ r = 0, all consumers areRicardian, and, therefore, the solutions for the wage rate and hours simplify to the standardones.2.5 GovernmentEach period the government decides the size and composition of public expenditure andthe mix of taxes and new debt holdings required to finance total outlays. It is assumedthat government purchases of goods and services (g c t ) and public investment (gi t ) followan exogenously given pattern. Conversely, interest payments on government bonds (1 +r t )b t−1 , unemployment benefits g ut (1 − n t−1 ) , and government social transfers g st areassumed to be endogenous. The two latter expenditure categories are given byg ut = rr t w t (46)g st = tr t gdp t (47)whereby g ut and g st are made proportional to the level of real wages, w t ,and activity, gdp t ,through rr t and tr t .Government expenditure is financed by direct taxation, levied on either labour income(personal labour income tax, τ l t, and social security contributions, τsct) or capital income(τ k t ), as well as indirect taxation, represented by consumption (τc t) and energy taxes

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 17(τ e t). Government revenues are therefore given byt t = (τ l t + τ sct )w t (n t−1 l 1t ) + τ k t (r t − δ) k t−1 (48)+τ c Ptct c t + τ e Ptet e t + trh t + τ lP t Ptrrw t (1 − n t−1 )l 2t + τ l tg sttwhere trh t stands for lump-sum transfers as defined below.Each period total receipts and outlays are made consistent by means of the government’sbudget constraintγ A γ N b t = g c t + g i t + g ut (1 − n t−1 ) + g st − t t + (1 + rn t )1 + π tb t−1 (49)Equation (49) reflects that the gap between total receipts and outlays is financed by variationsin lump-sum transfers to households, trh t (which enter the fiscal budget rule throughthe term t t ), and/or the issue of domestic bonds (b t − b t−1 ). As it stands, equation (49) hasan intertemporal dynamic nature. Note that government income from seniorage is nil.Dynamic sustainability of public debt requires the introduction of a debt rule thatmakes one or several fiscal categories an instrument for debt stabilization. In order toenforce the government’s intertemporal budget constraint, the following fiscal policy reactionfunction is imposedtrh t = trh t−1 + ψ 1"b tgdp t− # bbt+ ψgdp 2−b t−1gdp t gdp t−1 where bgdpis the long-run target for the debt-to-GDP ratio and ψ 1> 0 captures thespeed of adjustment from the current ratio towards the desired target. The value of ψ 2> 0will be chosen to ensure a smooth adjustment of actual debt towards its steady-state level.Note that while in the baseline specification debt stabilization is accomplished throughvariations in lump-sum transfers, nothing precludes other receipt or spending categoriesfrom playing this role.Government investment augments public capital, which in turn depreciates at therate δ p and thus follows the law of motion:(50)γ A γ N k p t= g i t + (1 − δ p )k p t−1(51)2.6 Monetary policyMonetary policy is geared by the European Central Bank (ECB), which targets EMU inflationby means of movements in interest rates. More specifically, short-term interest rates

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 18are governed by the following reaction functionln 1 + remu t1 + r emu = ρr ln 1 + remu t−11 + r emu + ρπ (1 − ρ r ) ln(π emut− π emu ) + ρ y (1 − ρ r ) ln ∆ ln y emut (52)where all the variables indexed by emu refer to EMU aggregates. Thus, rtemu and π emut arethe Euro-zone nominal short-term interest rate and consumption price deflator (to whichthe Spanish economy contributes according to its relative size), and ∆ ln y emut measuresthe deviation of GDP growth from its trend. As explained in Woodford (2003), (52) is theoptimal outcome of a rational central bank facing and objective function under generalequilibrium conditions.Finally, the disappearance of national currencies since the inception of the monetaryunion means that the intra-euro-area real exchange rate is simply given by the ratio ofrelative prices between the domestic economy and the remaining EMU members, so realappreciation/depreciation developments are driven by the inflation differential vis-à-visthe euro area:rer t+1rer t= 1 + πemu t+11 + π t+1(53)2.7 The External SectorThe small open economy hypothesis adopted in REMS implies that world prices andworld demand are taken as given. It also means that feedback linkages between the domesticeconomy, EMU and the rest of the world are ignored. Another simplifying assumptionconcerns the nature of final and intermediate goods produced at home, which are allconsidered to be tradable.The allocation of consumption and investment between domestic and foreign produced goodsLet us think of aggregate consumption (investment) as a composite basket of home andforeign produced goods. There is a representative consumption (investment) distributorwhose role is to determine the share of aggregate consumption (investment) to be satisfiedwith home produced goods c h (i h ) and foreign imported goods c f (i f ). This is carried outon the basis of CES technology:c t = (1 − ω ct ) 1σc−1σcσccht+ ω ct1σc σc−1σcc f t! σcσc−1(54)

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 19i t =!σ(1 − ω it ) 1 σ i −11 σ σ i −1 iσ i −1σ i i iht+ ω σiσit i i f t(55)where σ c (σ i ) is the consumption (investment) elasticity of substitution between domesticand foreign goods.Each period, the representative consumption distributor chooses c ht and c f t so as tominimize production costs subject to the technological constraint given by (54). The solutionof this problem provides the optimal allocation of aggregate consumption betweendomestic and foreign goods, c ht and c f t : −σc Ptc ht = (1 − ω c )Ptc c t (56)c f t = ω c PmtP c t −σcc t (57)where P t and Ptm are respectively the prices of home and foreign produced goods. Ptcrepresents both the price of the consumption good borne by households and the shadowcost of production borne by the aggregator.Proceeding in the same manner as with the investment distributor problem, similarexpressions can be obtained regarding the optimal allocation of aggregate investmentbetween domestic and foreign goods, i ht and i f ti ht = (1 − ω i )P tP i t! −σii t (58)i f t = ω iPtm Pti! −σii t (59)Price formationIn the preceding analysis, the price of domestically produced consumption and investmentgoods is equal to the GDP deflator, P t . In order to obtain the consumption pricedeflator, one needs to further incorporate the demand schedules provided by (56) and (57)for home and foreign consumption goods into the cost of producing one unit of aggregateconsumption goods (P t c ht + P m t c f t ). Bearing in mind that the production cost per

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 20unit equates to the price of production, it is straightforward to express the consumption(investment) price deflator as a function of the GDP and import deflatorsP c t =(1 − ω ct )P 1−σ ct+ ω ct P m1−σ 1c 1−σct(60)P i t =(1 − ω it )P 1−σ it 1+ ω it P m1−σ it1−σ i (61)The exogenous world price is a weighted average calculated on the basis of finalgood and intermediate good prices, PFM and P e , both expressed in terms of the domesticcurrency. Given the small open economy assumption, the relevant foreign price is definedas:P m t= eα e P e t + (1 − eα e )PFM t(62)where eα e stands for the ratio of energy imports to overall imports.Let us consider that export prices charged by Spanish firms deviate from pricescharged by competitors in foreign markets, at least temporarily. This well-known pricingto-markethypothesis is consistent with a model of monopolistic competition among firmswhere each firm regards its influence on other firms as negligible. We may define theSpanish export price deflator asP x t= P (1−ptm)tPFM t ptm(63)where P x t is the export price deflator, PFM t is a competitors price index expressed in eurosand the parameter ptm determines the extent to which there is pricing-to-market.Exports and ImportsAggregate imports include two final goods, foreign consumption and investment, and oneintermediate commodity, energy:im t = c f t + i f t + α e e t (64)where α e represents the ratio of energy imports over total energy consumption.Exports demand can be defined in terms of aggregate consumption and investmentfrom abroad, y w t, and the ratio of the export price deflator to the competitors price index

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 21expressed in euros, P x t /PFM t: Pex t = stx x −σxty w t(65)PFM tPlugging (63) into (65) yields the exports demand under the small open economyassumption and the pricing-to-market hypothesis:ex t = s x t PFMtP t (1−ptm)σxy w t(66)Stock-flow interaction between the current account balance and the accumulation of foreign assetsIn the model, the current account balance is defined as the trade balance plus interest ratereceipts/payments from net foreign assets:ca t = Px tex t − Pm tim t + (rtemu − π t ) bt−1 oemu(67)P t P tFollowing standard practice in the literature (see, for example, Obstfeld and Rogoff,1995, 1996), net foreign assets are regarded as a stock variable resulting from the accumulationof current account flows. This is illustrated by the following dynamic equation:γ A γ N b oemutφ bt=(1 + remu1 + π c tt )b oemut−1+ Px tP tex t − Pm tP tim t (68)(68) is obtained by combining the Ricardian households’ budget constraint (assuming azero net supply for domestic bonds and money), the government’s budget constraint andthe economy’s aggregate resource constraint.2.8 Accounting identities in the economyGross output can be defined as the sum of (final) demand components and the (intermediate)consumption of energy:y t = c ht + i ht + g t + Px tP tex t + κ v v t + Pe tP t(1 − α e )e t + κ f (69)whereas value added generated in the economy is given by:gdp t = y t − Pe tP te t − κ f − κ v v t (70)Note that, in accordance with previous definitions, c ht and i ht are equal to overall domestic

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 22consumption and investment minus consumption and investment goods imported fromabroad. Thus, c ht and i ht are consistent with the definitions above for gross output andvalue added.2.9 Model calibrationModel parameters have been fixed using a hybrid approach of calibration and estimation.Some parameter values are taken from different related DGE models. Several other parametersare calculated from the sample average counterpart of long-run conditions. Theremaining parameters have been estimated on the basis of selected model’s equations.Altogether, these parameters produce a baseline solution that accurately resembles the behaviourof the Spanish economy over the last two decades.The data used in the calibration come from the REMSDB database. All series coverthe period 1985:3 2006:4. At the beginning of the sample, the third quarter of 1985 displaysadequate cyclical properties for most of the endogenous variables (see Puch and Licandro,1997, and Boscá et al, 2007b). Several variables included in the model have no directstatistical counterpart from official sources. Such variables include consumption and employmentof RoT and optimizing consumers, Lagrange multipliers, Tobin’s q, compositecapital stock, marginal cost and total factor productivity. In order to sidestep the lack ofdata availability affecting these variables, we use the model’s related behavioural equationsto compute them.Table 1 lists the values of parameters and exogenous variables. The implied steadystate values of the endogenous variables are given in Table 2 10 . Roughly speaking, thecalibration strategy follows a sequence in which one starts by setting the value of a numberof parameters which are subsequently used to obtain a measure of the level of totalfactor productivity. This makes it possible to express all variables in the model in termsof efficiency units. The remaining parameters are then fixed on the basis of the model’sequations with variables measured in efficiency units (see Boscá et. al., 2007 for furtherdetails).3. Fiscal policy and the current account3.1 A temporary increase of public consumptionWe first analyse the effects of a temporary but persistent increase of public consumptionupon the main macroeconomic variables in our model. The initial shock in public con-10 The model has been programmed in relative prices. This means that all prices are relative to the deflatorprice index, P t , and the real exchange rate is defined as rer t = PFM tP t.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 23Table 1 − Parameter Valuesg 0.0905 l 2 0.2183 β b 0.5035 π emu 0.0000 τ c 0.1069 φ 1 2.9483 bgdp2.4000 T 1.3690τ l 0.1107 η 2.0000 rr 0.3832 λ w 0.4274τ k 0.2080 σ c 1.2063 τ e t 0.2000 ω i 0.1381tr 0.1381 φ b 0.0060 ψ 1 0.0100 σ i 0.9305τ sc 0.2215 φ 2 2.2389 ψ 2 0.2000 λ r 0.5000P e 0.8620 ρ 0.9571 σ x 1.3000 h 0.6000β 0.9908 a 0.9985 r emu 0.0158 α e 0.4980α 0.5938 χ m 0.1987 ζ 0.0600 eα e 0.1049δ 0.0141 mc 0.7961 γ A 1.0018 ω c 0.0930φ 5.5000 γ 1.7500 γ N 1.0047 κ f 0.2730κ v 1.4966 s 0.2600 δ p 0.0105 s x 0.0142σ 0.0180 ρ r 0.7500 ptm 0.5767 ρ w 0.7600χ 1 0.6115 λ 0.2006 ω c 0.0930 φ e 25.000χ 2 0.5725 β f 0.4965 y w 7.7855 ρ e 0.8500sumption is equivalent to 1 per cent of GDP (more than 5 per cent in terms of public consumption),similar to the size of the expansionary policies implemented through publicexpenditure in 2008 and 2009 by the Spanish government. 11 Figure 4 presents the impulseresponsefunctions of the main endogenous variables. The response of output is clearlypositive on impact and the short term multiplier is equal to 1.0. As expected, the responseof private consumption is also positive, due to the presence of rule-of-thumb consumer,since the consumption of optimizing households falls, given the negative wealth effects ofhigher government spending. The higher level of output is obtained with a higher levelof total hours worked (labor supply also rises with the negative wealth effect of optimizinghouseholds), which is accompanied with slightly greater real wages, in spite of theirrigidities. On the contrary, private investment decreases, reaching the lowest levels afterthree quarters. The fall of investment is driven by the fall in Tobin’s q.The increase in marginal costs also pushes the inflation rate up. Higher domesticprices as well as higher levels of output and private consumption cause an increase in importson impact. However, this increase of imports has a short life since imports are moresensitive to private investment, which falls from t + 1 to t + 3. The increase in domestic11 In particular, the law of movement of public consumption is given by:g c t = 0.9g c t−1 + 0.1gc + ε g twhere g c is the steady-state level of public consumption and ε g is the shock.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 24b t 1.3951 i ht 0.0987b o t 1.3951 im t 0.1565Table 2 − Steady State˜λ1t+1λ 1t˜λ3t+1λ 1t1.0000 r t 0.03350.2539 r n t 0.0142c t 0.3192 j t 0.1582 λ3t o 0.1997 g st 0.0803c f t 0.0542 jt o 0.3163 λ3t r 1.4439 trh t 0.0067c ht 0.2659 k t 7.5943 λ ndt 0.4648 t t 0.2024c o t 0.3570 ke t 6.3208 λ1t o 2.6355 Ut n -1.258c r t 0.2814 k o t 15.189 λ1t r 3.3428 g ut 0.0600pe t 0.0477 l 1t 0.4487c tp 0.9734 tv t 0.0595rer t 0.8390 m t 2.6116 π t 0.0000 w t 1.2830ex t 0.1431 cmc t 0.0000 gdp t 0.5813 y t 0.9861bt w 0.0000 m o t 5.2232 π c t 0.0000 φ bt 1.0000bt ow 0.0000 n t 0.6102 p t0.9228 k p t 1.1131gt c 0.0905 n o t 0.6102p m tp t0.8267 l 2t 0.2183i t 0.1672 n r ptt 0.6102p 0.9037 tgt i 0.0173 i f1f t 0.0785 λ1t0.3393 q t 1.0285p i tprices also affects negatively foreign competitiveness, therefore inducing a reduction ofexports. As shown in Figure 4 the real exchange rate appreciates and the terms of trade(P x /P m ) increase. In spite of the higher inflation rate, nominal interest rate increases veryslightly, given the small weight of the Spanish economy in EMU inflation (around 10 percent). Finally, as expected, public debt increases, reaching its highest level seven quartersafter the shock, and then falls steadily as a result of the fiscal rule.A very interesting way of analysing the effects of the positive shock in public consumptionupon the external imbalances is given by Figure 5. In this figure we have representedthe absolute deviations of the savings, investment and public consumption, all inrelative terms with respect to GDP, from their steady-state levels (that is, x t − x). As wecan observe, the public consumption rate increases above 0.8 percentage points, and thesaving rate almost falls 0.6pp. In fact, private savings only compensate a 46.6 per cent ofthe increase of public consumption, on average over the first 20 quarters. Therefore, thereis only partial Ricardian equivalence in our economy.However, the fall in the investment rate compensates partially the fall of the savingrate and, therefore, the increase in the current account deficit is very small: on impact isequal to 0.2pp of GDP, smaller than 1/4 of the increase in the public consumption rate. In

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 2520GDP C I10ImportsExports­20 5 10 15 200.50Consumer price inf lationNominal interest rate­0.50 5 10 15 20210­10 5 10 15 2010.5Real wageTotal hours workedPublic debt00 5 10 15 20­10 5 10 15 200.5­0.50Real exchange rateTerms of trade­10 5 10 15 20420Real rental rate of capitalReal marginal cost­20 5 10 15 2010500 5 10 15 20gFigure 4: Impulse-response functions after a temporary shock in public consumption.other words, although there is some evidence of twin deficits, only a small fraction of thegovernment budget deficit translates into a current account deficit. Surprisingly, despitethe large differences between SIGMA and REMS, our results are very similar to those obtainedby Erceg, Guerrieri and Gust (2005), who find that a fiscal deficit has a relativelysmall effect on the U.S. trade balance, irrespective of whether the source is a spendingincrease or tax cut. In their benchmark calibration, where as in REMS non-ricardian consumersrepresent 50 per cent total households, they find that a rise in the fiscal deficit ofone percentage point of GDP induces the current account to deteriorate by less than 0.2percentage point of GDP.3.2 A temporary increase of public investmentAs a second exercise of expansionary fiscal policy we analyse the effects of a temporaryincrease in public investment. As in the case of public soncumption, the initial shock inpublic investment is equivalent to 1 per cent of GDP (more than 5 per cent in terms ofpublic investment) and similar in its magnitude to the Fund for Local Investment imple-

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 261Saving rateInvestment ratePublic consumption0.50­0.50 2 4 6 8 10 12 14 16 18 20Figure 5: The response of saving, investment and public consumption rates after atemporary shock in public consumption. Absolute deviations from their steady-statevalues.mented by the Spanish government in 2009. 12As it can be seen in Figure 6, the response of output is again clearly positive onimpact, although the short-term multiplier is slightly smaller (0.97), and the response ofprivate consumption is also positive (the consumption of optimizing households falls).As in the previous case, the positive shock in public investment induces a higher levelof total hours worked and slightly greater real wages, and a fall in private investment.The rest of variables shows similar responses to thoses observed previously after a publicconsumption shock. The main difference is in the persistence of the shock, since publicinvestment accumulates into public capital, which depreciates slighty.The effects of the positive shock in public investment upon the external imbalancesis analyzed in Figure 7. In this case, the public investment rate increases above 0.9 percentagepoints, whereas the total investment rate only increases 0.6pp on impact, sinceprivate investment falls. The saving rate also increases in this case, because public savingsare higher as a consequence of greater public revenues, and the economy exhibits a very12 As for public consumption, the law of movement of public investment is given by:i g t= 0.9i g t−1 + 0.1ig + ε i twhere i g is the steady-state level of public consumption and ε i is the shock.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 2720­20.50GDP C I0 5 10 15 20­0.50 5 10 15 20210Consumer price inf lationNominal interest rate­10 5 10 15 201.51Real wageTotal hours worked0.5Public debt00 5 10 15 2010.50ImportsExports­0.50 5 10 15 200.5­0.50Real exchange rateTerms of trade­10 5 10 15 20420Real rental rate of capitalReal marginal cost­20 5 10 15 20402000 5 10 15 20igFigure 6: Impulse-response functions after a temporary shock in public investment.small current account deficit (0,12 per cent of GDP) confirming again the previous results.Although the current account deficit is very persistent, its size in terms of GDP is small,given place to a weak relationship between the deficits in the government budget and inthe current account.3.3 A temporary reduction of labor income taxesThe next exercise we consider is a temporary reduction of 10 per cent in the labour incometax rate. The ex-ante effects of this reduction in the tax rate are equivalent to 0.5per cent of GDP, very close in its magnitude to the tax rebate implemented by the Spanishgovernment in 2008 and 2009. 13 Figure 8 shows the impulse-response functions afterthe shock. Output increases on impact (the tax multiplier is equal to -0.61, smaller thanfor public consumption and investment), and remains above its steady-state level during13 The law of movement of the labor income tax rate is given by:τ l t = 0.9τ l t−1 + 0.1τl + ε τ twhere τ l is the steady-state level of the labor income tax rate and ε τ is the shock.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 2810.9Saving rateInvestment ratePublic investment0.80.70.60.50.40.30.20.100 2 4 6 8 10 12 14 16 18 20Figure 7: The response of saving, investment and public investment rates (in percentagepoint of GDP) after a temporary shock in public investment. Absolutedeviations from their steady-state values.many quarters.The increase in private consumption is even higher than for GDP, an it is explainedby the increase in the consumption of non-ricardian households, since optimizing householdsdo not change their consumption levels. The crowding out effects upon privateinvestment are now smaller, as well as the increase in the marginal cost and the rentalrate of capital. The fall in the labor income tax rate induce a reduction in the real wage(before taxes) and, as expected, an increase in total hours worked, which take advantageof lower taxes. The fall in the labor income tax rate translates into a small depreciationof the real exchange rate and into a small fall of the terms of trade, which gives place tohigher exports through the improvement in relative prices. However, imports increasemore than exports, particularly on impact, first through higher private consumption andlater through higher private investment from t + 3 onwards.The effects of the temporary reduction in the labour income tax rate upon the externalimbalances is analyzed in Figure 9. In this case, revenues from labor income taxesfall more than 0.5 percentage points of GDP, the investment rate only decreases 0.12pp onimpact and the saving rate also falls 0.17pp. Given the small responses of the investmentand saving rate, the economy exhibits a very small current account deficit (0,05 per centof GDP), only slightly smaller than the decline in the case of the public consumption and

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 2910.50GDP C I0.50ImportsExports­0.50 5 10 15 200.050Consumer price inf lationNominal interest rate­0.50 5 10 15 200.50Real exchange rateTerms of trade­0.050 5 10 15 2010.50Real wageTotal hours worked­0.50 5 10 15 2010.5Public debt00 5 10 15 20­0.50 5 10 15 200.50Real rental rate of capitalReal marginal cost­0.50 5 10 15 20­50Labor income tax rate­100 5 10 15 20Figure 8: Impulse-response functions after a temporary reduction in the labor income tax rate.investment shocks.3.4 Sensitivity analysisIn order to check the robustness of our results, we have analyzed again the effects of ashock in public consumption (equivalent to 1 per cent of GDP) under different specificationsof our model. The results of these exercises are shown in Table 3. For each exercisewe show the fiscal multiplier, the investment rate, the saving rate and the current accountbalance in terms of GDP. All this variables refer to the same quarter in which the fiscalstimulus is implemented. To facilitate comparisons, the first row just shows the results ofour baseline.In the second row we show the results for λ r = 0.8, that is, when we increase theshare of rule-of-thumb consumers in the economy. As expected, the fiscal multiplier increases,almost a 20 per cent with respect baseline. Additionally, as the share of Ricardianconsumers is smaller, the effect of the fiscal deficit upon the current account deficit is alsogreater, increasing to -0.23 per cent of GDP, almost a quarter of the initial fiscal stimulus.Conversely, when we reduce the share of rule-of-thumb consumers, as in row (3), both thefiscal multiplier and the current account deficit are smaller, since Ricardian equivalence is

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 300.10­0.1­0.2­0.3­0.4­0.5Saving rateInvestment rateLabor taxes­0.60 2 4 6 8 10 12 14 16 18 20Figure 9: The response of saving and investment rates, and revenues from laborincome taxes over GDP after a temporary reduction of the labor income tax rate.Absolute deviations from their steady-state values.greater. These results are similar to the ones also obtained by Erceg, Guerrieri and Gust(2005).In the fourth row we increase the bargaining power of workers (λ w = 0.9). Lookingat the solution for wages (equation (43)), a higher value of λ w increases the sensitivity ofwages to marginal labour productivity, since workers have the power to set wages up tothe highest feasible level. In this case, both the fiscal multiplier and the effects upon thecurrent account balance are higher than in the baseline.In row (5) we set the pricing-to-market parameter to zero, that is, exports prices areexactly equal to domestic prices. Therefore, as the fiscal stimulus increases domestic prices,exports prices also increase by the same amount and relative prices remain constant. Asexpected, the loss in competitiveness induces a smaller fiscal multiplier and a slightlysmaller current account deficit with respect to the baseline.In the sixth row we multiply by 2 the elasticity of imports and exports to relativeprices. In this case the fiscal multiplier is a 20 per cent smaller than in the baseline but thecurrent account balance only increases slightly to -0.20. In row (7), instead of increasingthe sensitivity of exports and imports to relative prices, we reduce the openness of theeconomy, dividing the scale factors in exports and imports equations (s x , ω c and ω i ) by2. Although the fiscal multiplier is not affected by this change in the specification of the

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 31Table 3 − Sensitivity AnalysisMultiplier I/GDP S/GDP CA/GDP(1) Baseline 1.01 23.55 23.38 -0.17(2) λ r = 0.8 1.19 23.44 23.20 -0.23(3) λ r = 0.0 0.82 23.64 23.54 -0.09(4) λ w = 0.9 1.08 26.92 26.67 -0.25(5) ptm = 0.0 0.96 23.33 23.16 -0.16(6) σ x .σ c , σ i × 2 0.81 22.66 22.46 -0.20(7) s x , ω c , ω i × 0.5 1.01 21.68 21.61 -0.08(8) h = 0.0 0.94 23.64 23.49 -0.15(9) ρ w = 0.9 1.05 23.45 23.29 -0.16(10) φ b= 0.06 0.96 23.48 23.35 -0.13(11) λ r = λ w = 0.9, φ b= 6 × 10 −6 1.53 26.67 26.21 -0.46(12) Spending reversal fiscal rule 1.06 23.77 23.56 -0.22(13) Independent central bank 0.60 23.73 23.73 -0.01Notes: The variables in the columns refer to the impact fiscal multiplier, the investment and savingratios, and the current account surplus in terms of GDP, in the same period in which public consumptionincreases by 1% of GDP.model, as expected the effects upon the current account balance are now smaller than inthe baseline, approximately a half of its current account deficit, in line with the results ofCorsetti and Muller (2006).In rows (8) and (9), we show that the baseline results are also robust to changes inthe habit parameter (h = 0) and in the coefficient of real wage inertia (ρ w = 0.9). In bothcases the fiscal multiplier and the effects upon the current account balance do not changevery much. In row (10) we increase the sensitivity of the risk premium to the externaldebt. As expected, this change reduces both the fiscal multiplier and the current accountbalance.Combining those changes with greater effects in the external position, in row (11)we have increased the share of rule-of-thumb consumer up to the (improbable) value of90%, the bargaining power of workers (λ w = 0.9) and we have reduced significantly thecoefficient φ bwhich measures the sensitivity of the risk premium to external debt, thatnow converges to the steady state very slowly. Even in this extreme situation in whichthe fiscal multiplier increases a 50 per cent with respect the baseline, the current accountdeficit goes to 0.46 of GDP, less than half the initial fiscal stimulus.In row (12) we have changed the specification of the fiscal rule. Contrary to thebaseline, in which fiscal consolidation is achieved with lump-sum transfers, in this casepublic spending also reacts to the deviations of the debt to GDP ratio from its steady-state

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 32level:g c t = 0.9g c t−1 + 0.1gc − 0.002" # b t−1 b− − ε ggdp t gdptThe effects of this type of rule over the fiscal multiplier have also been analysed by Corsetti,Meier, and Müller (2009). According to these authors, the effects of fiscal stimulus dependsnot only on short-term tax and spending policies, but also on the expectations byagents about the offsetting measures in the future. They find that accounting for such fiscalrules that make explicit spending reversals brings new Keynesian models in line withthe crowding-in of private consumption. As we can see in row (12), these results are confirmedalthough the effects are relatively small: both the fiscal multiplier and the currentaccount deficit increase with a fiscal rule for public expenditure which is sensitive to thedeviations of the public debt to GDP ratio from its steady-state level.Finally in row (13) we have evaluated the implications of being in a currency areawith no independent monetary policy. In principle, for a relatively small country as Spain,both the fiscal multiplier and the effects of the fiscal stimulus upon the current accountbalance should be greater since monetary policy reacts only slightly increasing interestrates. Davig and Leeper (2009) have found that government spending generates positiveconsumption multipliers and a higher fiscal multiplier for GDP when monetary policyis passive, that is, in a regime in which the monetary authority does not increase realinterest rates in response to expansive fiscal policies. In the limiting case in which the zerobound on nominal interest rates is binding, Christiano, Eichenbaum and Rebelo (2009)have found a similar results. In row (13) we have change the specification of the modelassuming that the exchange rate is flexible (therefore, the uncovered interest parity holds)and that the short-term interest rates are governed by a standard Taylor rule:ln 1 + r t1 + r = 0.75 ln 1 + r t−1+ 1.75(1 − ρ r ) ln(π t − π) (71)1 + rAs expected, the fiscal multiplier is now much smaller and the effects upon the currentaccount balance are almost negligible.In summary, this sensitivity analysis confirms the main results obtained in our baseline.The effects of fiscal policy upon the current account balance are relatively small. In orderto obtain greater effects on the current account in line with the twin deficit hypothesiswe have to impose values of some parameters as, for example, the share of rule-of-thumbconsumers that seems to be very improbable.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 334. ConclusionsIn this paper we have explored the effects of different expansionary fiscal policies uponthe trade deficit in Spain, a small open economy in a currency union and, therefore, withno independent central bank. Using a DGE model for the Spanish economy, we find thatexpansionary fiscal policies have only small negative effects on the current account, evenassuming a large proportion of non-Ricardian consumers. Although Ricardian equivalenceholds only partially, the crowding-out effects upon private investment implied byfiscal policies compensate the behaviour of national savings, thus avoiding large effectsupon the trade deficit.An additional interpretation of these results is that the current negative correlationbetween the large deficit in the government balance budget and the quick correction ofthe current account cannot be explained by the twin divergence hypothesis in the Spanisheconomy, and it may be explained by a large output shock that more than compensatesthe effects of fiscal policy. Additionally, the large increase in the households saving rate inthe short run seems to be motivated by precautionary saving and the desire of reducingtheir levels of debt, as response to the international financial turmoil and the huge increaseof unemployment, rather than by the large fiscal stimulus, as proposed by the Ricardianequivalence hypothesis.From a policy perspective, our results suggest that the emphasis on the future fiscalconsolidation should be based more on the sustainability of government debt than on theeffects on external imbalances.

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 34References(To be completed)Andolfatto, D. (1996): “Business Cycles and labour-Market Search”. American Economic Review, 86(1).Andrés, J. and R. Doménech (2006): ˇAutomatic Stabilizers, Fiscal Rules and Macroeconomic Stabilityı.European Economic Review, 50(6), 1487-1506.Andrés, J., R. Doménech and A. Fatás (2008): ˇThe Stabilizing Role of Government Sizeı. Journal ofEconomic Dynamics and Control, 32, 571–593.Baxter, M. (1995): ˇInternational Trade and Business Cycles,ı in Handbook of International Economics,Vol. 3, eds. G. M. Grossman and K. Rogoff. Amsterdam. North-Holland.Boscá, J. E., Díaz, A., Doménech, R., Ferri J., Pérez, E. and L. Puch (2009): “A Rational ExpectationsModel for Simulation and Policy Evaluation of the Spanish”. Investigaciones Económicas(forthcoming).Boscá, J. E., Doménech, R. and Ferri J.(2009): “Tax Reforms and Labour-Market Performance: AnEvaluation for Spain using REMS”. Moneda y Crédito (forthcoming)Calvo, G. (1983): “Staggered prices in a Utility Maximizing Framework”. Journal of Monetary Economics12, 383-398.Cavallo, M. (2005): “Understanding the Twin Deficits: New Approaches, New Resultsı. FRBSF EconomicLetter, Number 2005-16.Cavallo, M. (2007): “Government Consumption Expenditures and the Current Account.ı Public Financeand Management, 7, January 2007, 1.Christiano, L., M. Eichenbaum and S. Rebelo (2009): “When is the government spending multiplierlarge?,” NBER Working Paper 15394.Corsetti, G. and G. Müller (2007): “Twin Deficits, Openness and the Business Cycleı, Working Paper2007/20, EUI.Corsetti, G., G. J. Müller (2006): “Twin Deficits: Squaring, Theory, Evidence and Common Sense.ıEconomic Policy, 597–638.Corsetti, G., A. Meier, and G. J. Müller (2006): “Fiscal Stimulus with Spending Reversals". CEPR DPNo. 7302.Davig, T. and E. M. Leeper (2009): “Monetary-Fiscal Policy Interactions and Fiscal Stimulus,” NBERWorking Papers 15133.Erceg, C. J., L. Guerrieri, and C. Gust (2005): ˇExpansionary Fiscal Shocks and the Trade Deficit.ıInternational Finance, vol. 8, 363-397.Galí, J. (1999): “Technology, Employment and the Business Cycle. Do Technology Shocks ExplainAggregate Fluctuations?”. American Economic Review, 89, 1, 249-271.Galí, J., M. Gertler and J. D. Lopez-Salido (2001): “European Inflation Dynamics”. European EconomicReview, 45, 1237-1270.Galí, J., Lopez-Salido, J. D. and J. Vallés (2007): “Understanding the Effects of Government Spendingon Consumption” Journal of the European Economic Association, 5(1), 227-270.Kim, S. and N. Roubini (2008): ˇTwin Deficit or Twin Divergence? Fiscal Policy, Current Account,

ON RICARDIAN EQUIVALENCE AND TWIN DIVERGENCE IN SPAIN 35and Real Exchange Rate in the US,ı Journal of International Economics, 74(2), 362-383.Merz, M. (1995): “Search in the labour Market and the Real Business Cycle”. Journal of MonetaryEconomics, 36, 269-300.Obstfeld, M. and K. Rogoff (1995): “Exchange Rate Dynamics Redux”. Journal of Political Economy,103, 624-60.Obstfeld, M. and K. Rogoff (1996): Foundations of International Macroeconomics. MIT Press.Puch, L. and O. Licandro (1997): “Are There Any Special Features In The Spanish Business Cycle?”.Investigaciones Economicas, XXI(2), 361-394.Seater, J. J. (1993): ˇRicardian Equivalence.ı Journal of Economic Literature, 31(1), 142-190.Trigari, A. (2006): “The Role of Search Frictions and Bargaining for Inflation Dynamics”. IGIERWorking paper No. 304.Woodford, M. (2003): Interest and Prices, Foundations of a Theory of Monetary Policy. Princeton UniversityPress.