The interdependence of cigarette and liquor demand.

• Author: Goel, Rajeev K.

1. Introduction

   Public policies have focused on cigarettes and liquor as prime targets of excise taxation. Dual objectives of this taxation have been consumption reduction and revenue generation.(1) Given that cigarettes and liquor are habit-forming substances, quite often they are consumed by the same individuals, i.e., a large number of smokers also drink and vice versa. It is then likely these goods are related in consumption. If it follows that the goods are related in consumption, then a public policy for one good cannot be independent of the other, even if its authors intend for it to be in practice. For instance, government agencies such as the Bureau of Alcohol, Tobacco and Firearms (ATF) could better coordinate their policies in light of the information about how cigarettes and alcohol are related in consumption. When cigarettes and liquor are complements, public policy needs to check consumption of only one good to reduce consumption of both. In this paper we present the results of estimating a model that captures interdependencies in the joint consumption of cigarettes and liquor. In determining the cross-price effects between cigarettes and liquor we consider particular aspects of demand for these two commodities such as habit formation and advertising effects as well as the effects of government intervention in tobacco and alcohol markets.

   This research makes at least two contributions to the literature focusing on demand for liquor and cigarettes. First, we establish empirically whether these goods are substitutes or complements in consumption. Second, results from our analysis permit answers to questions about whether states should consider setting or changing tax rates on cigarettes and liquor as a joint decision to maximize tax revenues. Until now, there existed no empirical justification for such a joint decision.

   A pooled time series cross-section data set on U.S. cigarette and liquor consumption and prices spanning nearly three decades provides the foundation for empirical work that appears to be unprecedented in the literature.

   Various demand studies have been performed for these commodities separately [1; 2; 3; 4; 11; 18; 20; 25]. However, there is little consensus in the literature regarding the elasticity estimates. For example, estimates of the price elasticity of liquor demand in the U.S. range from -2.03 to 0.08 (see Ornstein [21] for a survey). Similar dispersion in own-price elasticity estimates is found in the case of cigarettes. One reason for the lack of consensus might be a bias in own-price elasticity estimates because of omitted cross-price effects.

   Knowledge of the cross-price effects would be valuable to legislators looking to bolster tax revenues by increasing tax rates on cigarettes or liquor. To make informed decisions about taxing these commodities, legislators should know whether they are substitutes or complements in consumption, as well as know the effect of changes in tax rates on cigarette and liquor consumption and hence on tax revenues.

   Own-price elasticity estimates are also useful inputs in the development of federal and state tax policy. Taxes for low demand elasticity commodities can be raised without a proportionate decrease in revenues. Commodities like cigarettes, liquor and gasoline have historically been assumed to have low demand elasticities. Therefore, these goods have been the target of frequent tax increases by federal, state, and local governments. For instance, more than 100 state liquor tax increases were passed in the 1980s [23]. In 1983 alone, federal, state and local government revenues from distilled spirits were about $6.9 billion [8]. In addition, the price elasticity is useful to any firm in determining the effect of a new tax or changes in tax rates on its sales or profits.

   The next section outlines the methodology. Data are described in section III. The econometric results are interpreted in section IV and public policy conclusions drawn in the final section.

2. Methodology

   This study analyzes the joint demand for cigarettes and liquor with cigarette and liquor consumption as dependent variables and, as economic theory suggests, income, lagged consumption (to account for a habit-persistence effect), advertising, and prices of related goods as explanatory variables. Commodity-specific regulatory details, such as the 1971 ban on broadcast advertising of cigarettes and the state monopoly in liquor sales, are included because regulatory intervention policies influence the markets for these goods.

   The model includes the following relationships (note that subscript L denotes liquor and C denotes cigarettes):

   CON[S.sub.C] = f([P.sub.C], [P.sub.L], [A.sub.C], Y, [H.sub.C], [R.sub.C]) (1)

   CON[S.sub.L] - g([P.sub.L], [P.sub.C], [A.sub.L], Y, [H.sub.L], [R.sub.L] (2)

   where CONS is per-capita consumption, P is price, A is real advertising expenditure, Y is real per-capita disposable income, H represents a habit-persistence effect (measured by lagged consumption) and R represents regulatory changes.(2)

   Two notable regulatory measures have been undertaken by the U.S. government with respect to cigarette advertising. First, under the Fairness Doctrine, anti-smoking messages were subsidized by the government beginning in 1968. A dummy variable, FAIRNESS, was introduced in the cigarette consumption equation to measure the effectiveness of the Fairness Doctrine. FAIRNESS takes the value of one between 1968 and 1970 and zero Otherwise.(3) Second, the Fairness Doctrine was followed by a complete ban on all broadcast advertising of cigarettes beginning January 1971. We use a dummy variable (BROADCAST) in the cigarette equation to measure the effectiveness of this policy. BROADCAST takes a value of one for the period 1971 to 1982 and zero otherwise.

   In the United States, distilled spirits are marketed through two different channels. On the one hand, there are "control" states in which the state government has a monopoly on sales.(4) On the other hand, in "license" states, the distribution outlets are owned by private firms. To determine the effect of state control of public distribution on liquor consumption, a dummy variable (CONTROL) that takes a value of one for control states and zero otherwise was created. In addition, we use state-specific dummy variables to ascertain fixed effects that vary across states.

   Under the Federal Alcohol Administration Act of 1935, the Bureau of Alcohol, Tobacco and Firearms and the Federal. Trade Commission have jurisdiction...