Goldin & Homonoff - DataSpace at Princeton University
Goldin & Homonoff - DataSpace at Princeton University
Goldin & Homonoff - DataSpace at Princeton University
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
variable (y) represents cigarette demand, τ e is the log excise tax r<strong>at</strong>e, τ s is the log sales tax r<strong>at</strong>e, x are<br />
covari<strong>at</strong>es th<strong>at</strong> do not vary between individuals interviewed in the same st<strong>at</strong>e, month, and year, and z<br />
are individual-level covari<strong>at</strong>es. We include st<strong>at</strong>e fixed effects µs to capture unobserved factors th<strong>at</strong> are<br />
correl<strong>at</strong>ed with both st<strong>at</strong>e tax r<strong>at</strong>es and the level of smoking demand. Year fixed effects λt capture time<br />
trends in smoking demand as well as yearly shocks to n<strong>at</strong>ional cigarette consumption, such as a n<strong>at</strong>ional<br />
anti-smoking campaign. Finally, πm is a calendar month effect, which accounts for seasonal or monthly<br />
p<strong>at</strong>terns in cigarette demand.<br />
As is standard in the cigarette demand liter<strong>at</strong>ure, 14 we model the decision of whether an individual<br />
smokes (the extensive margin) separ<strong>at</strong>ely from the decision of how much to smoke, conditional on being<br />
a smoker (the intensive margin). Consequently, in some specific<strong>at</strong>ions y is a binary choice variable<br />
indic<strong>at</strong>ing whether the individual reports being a smoker, and in other specific<strong>at</strong>ions y is the non-zero<br />
count of the number of cigarettes consumed in the last month, where the sample is restricted to self-<br />
reported smokers. This “double-hurdle” model is common in the cigarette demand liter<strong>at</strong>ure because the<br />
decision of whether to smoke may be fundamentally different than the decision of how much to smoke,<br />
and is inform<strong>at</strong>ive as to whether taxes affect consumption by turning smokers into non-smokers or by<br />
inducing current smokers to reduce the number of cigarettes they smoke. 15<br />
Table II presents the results of this analysis. 16 The specific<strong>at</strong>ions in Columns 1 and 4 regress smoking<br />
demand on the two tax r<strong>at</strong>es, individual demographic variables, and st<strong>at</strong>e, year, and calendar month<br />
fixed-effects. Since st<strong>at</strong>e taxes are often increased to meet budgetary shortfalls in bad economic times, it<br />
is likely th<strong>at</strong> tax r<strong>at</strong>e changes are correl<strong>at</strong>ed with st<strong>at</strong>e-level economic variables th<strong>at</strong> are not captured by<br />
st<strong>at</strong>e fixed effects. If cigarette consumption is also correl<strong>at</strong>ed with the business cycle, this omitted variable<br />
could bias our results. To account for this possibility, Columns 2 and 5 include st<strong>at</strong>e-level measures of<br />
real income and unemployment r<strong>at</strong>e. 17<br />
14See Frank J. Chaloupka and Kenneth E. Warner (2000) for a helpful review of the extensive liter<strong>at</strong>ure on estim<strong>at</strong>ing<br />
cigarette demand.<br />
15A drawback of the two-part approach is th<strong>at</strong> estim<strong>at</strong>ion results for the intensive margin may be biased by changes to the<br />
composition of the smoking popul<strong>at</strong>ion. We investig<strong>at</strong>e the robustness of this specific<strong>at</strong>ion in Section II.E.2.<br />
16We estim<strong>at</strong>e demand on the extensive margin with a linear probability model. A Probit model yields similar results.<br />
Because unobserved shocks to smoking demand may be correl<strong>at</strong>ed across time for consumers living in the same st<strong>at</strong>e, all<br />
tables report standard errors th<strong>at</strong> are clustered <strong>at</strong> the st<strong>at</strong>e level.<br />
17Real st<strong>at</strong>e income d<strong>at</strong>a comes from the Bureau of Economic Analysis and the st<strong>at</strong>e unemployment r<strong>at</strong>e d<strong>at</strong>a comes from<br />
the Bureau of Labor St<strong>at</strong>istics. Both variables are measured quarterly.<br />
15