Can beer taxes affect teen pregnancy? Evidence based on teen abortion rates and birth rates.

• Author: Sen, Bisakha

1. Introduction

   In the United States, teen pregnancy rates continue to be higher than in comparable developed nations like Sweden, Canada, or France (Darroch, Frost, and Singh 2001; Levine 2001). The effects of teen pregnancy are problematic, given the evidence that teen childbearing has dire educational, economic, and perhaps psychological consequences for the teen mothers as well as adverse consequences for their children. While a teen pregnancy may be terminated via abortion to prevent these consequences, there still remains the concern that the very experience of being pregnant and having to undergo the abortion might have adverse psychological and physiological effects for a teen woman. Hence, given the potential individual and social costs associated with teen pregnancy, there has been considerable interest in empirically investigating its determinants for the purpose of effective policy formulation.

   A number of studies have emphasized the positive relationships between alcohol use and outcomes like risky sexual activity and childbearing among youth. This leads to the conjecture that teen alcohol use increases the risk of teen pregnancy via effects on participation in sexual intercourse and contraception use. Therefore, policies that reduce teen alcohol use may also have the positive externality of reducing teen pregnancy. One such possible policy is increased tax rates on alcohol. This study focuses on the effects of state beer taxes on two alternate pregnancy outcomes--state-level abortion rates and birth rates--among women aged 15-19 years. The results suggest that increasing beer taxes can bring about reductions in teen abortion rates, but the reductions are likely to be rather small in magnitude. There is less evidence to suggest that increasing beer taxes will reduce teen birth rates.

2. Background

   The associational relationship between alcohol use and 'pregnancy-risk' behavior is well established. For example, Butcher, Thompson, and O'Neal (1991) report that, among young college students, 57% of women stated that they had sexual intercourse 1-5 times primarily because they were intoxicated. Cooper, Peiree, and Huselid (1994) and Shrier et al. (1996) present evidence that higher alcohol use is associated with more frequent sexual intercourse, sexual intercourse with relative strangers, and noncontracepted sexual intercourse among teens. Additionally, there is evidence that alcohol use is associated with early initiation into sexual activity (Bentler and Newcomb 1986; Mott and Haurin 1988; Staton et al. 1999) and greater likelihood of teen motherhood (Mensch and Kandel 1992). The association between alcohol use and 'risky' sexual activity or teen motherhood does not in itself prove causality because both sets of problem behaviors may be driven by an underlying penchant for risk-taking or rebelliousness, in which case reducing access to alcohol for teens, in itself, may not affect teen pregnancy. While some studies in the economic literature have tried to account for the potential endogeneity between alcohol use and risky sexual behavior and test for evidence of 'causality' using individual-level data, the results are mixed. For instance, Rees, Argys, and Averett (2001) find little evidence of a causal effect. However, Sen (2002) and Hofler, Lee, and Sen (2002) find evidence that there is a causal effect, though it is noticeably smaller than what is suggested by the simple associational relationships.

   An alternate approach is to use reduced-form models and a viable instrument for alcohol use and thereafter directly test the effects of that instrument either on adolescent sexual activity at the individual level or on adolescent pregnancy rates at an aggregated state (or county) level. (1) The latter method may be preferred because individual-level data on adolescent sexual activity is subject to problems of misreporting. One potential instrument is alcohol taxes, provided two conditions are satisfied. First, the demand for alcohol among adolescents must be sensitive to alcohol price as per the law of demand. Second, alcohol taxes must be transferred to consumers so that increases (decreases) in taxes actually translate to increases (decreases) in alcohol price.

   Many studies in the economic literature have documented the sensitivity of youthful drinking to beer prices (beer is widely assumed to be the alcoholic beverage of choice among youth). (2) Price also has an effect on the likelihood of abstention (Moore and Cook 1995). Additionally, the evidence provided by Chaloupka and Wechsler (1996) suggests that, among college students, the price of beer has an impact on female students more so than male students. Because it may be logically assumed that accidental pregnancies often occur when the female partner participates in risky or careless sexual activities while intoxicated to some degree, it may be conjectured that alcohol prices should have some impact on pregnancy rates among the youth.

   The other issue is whether alcohol taxes actually translate into higher alcohol prices. Alcohol excise taxes are paid by wholesale dealers. A report from the Center for Science in the Public Interest (1989) suggests that the taxes are ultimately passed on to consumers in the form of higher retail prices. More recently, Young and Agnieszka (2002) use state-level data and find that beer taxes are actually more than fully passed on to consumers in the form of retail prices within a quarter of a year. (3) Thus, it can be asserted with some confidence that changes in beer taxes translate to changes in beer prices and can thus affect quantity demanded for alcohol.

   Beer taxes have traditionally been a fairly popular instrument in the economic literature, both to test for existence of actual causality between alcohol consumption and various associated 'problem' phenomena, and (perhaps more important) to provide a direct scope for policy formulation by indicating whether the phenomena in question can be reduced via increased alcohol taxes. A number of studies consider the effects of beer taxes (occasionally in addition to other alcohol policy variables) on traffic fatalities. A useful summary of the results of these studies is provided by Bensen, Rasmussen, and Mast (1999). Some authors find evidence that higher taxes significantly reduce fatality rates (Saffer and Grossman 1987; Chaloupka, Saffer, and Grossman 1993; Ruhm 1996). Others find more mixed results (Dee 1999; Mast, Benson, and Rasmussen 1999; Young and Likens 2000), where the tax effects are sensitive to alternate specifications and inclusion of other alcohol-related policies. (4) Various studies also consider the effects of alcohol taxes on violence. Examples are Cook and Moore (1993), who find that increased beer taxes have significantly negative effects on rape and robbery at the state level, though not on homicide; Markowitz and Grossman (2000), who find that increased beer taxes reduce incidents of physical violence toward children; and Markowitz (2001), who finds that increased beer taxes reduce the likelihood of youths engaging in physical fights at school. (5) A recent study by Benson, Rasmussen, and Zimmerman (2001) also confirms the association between alcohol and crime, though it does appear to vary by type of alcohol, with liquor appearing to be more important for some crimes (like rape), and beer being more important for others (like assault). However, these authors caution that some of the apparently significant alcohol policy effects on crime may arise from the correlation between the policy variables and crime-deterrence variables (like arrests per crime committed and size of the police force). (6)

   In spite of extensive evidence of the associational relationship between alcohol use and 'risky' sexual activity, to my knowledge, only Chesson, Harrison, and Kassler (2000) consider the effects of alcohol taxes on a phenomenon related to sexual activity--the rates of sexually transmitted diseases at the state level. They find strong negative effects for the population overall and for adolescent men, with somewhat weaker effects for adolescent women. While Dee (2001) considers the effects of state legal minimum drinking ages on teen childbearing, (7) to my knowledge, no existing study investigates if alcohol taxes have any effects on teen pregnancy--specifically, teen abortion and birth rams. This work attempts to fill that gap.

3. Model, Data, and Specifications

   A relatively straightforward formulation of the underlying structural relationship between teen alcohol use and teen pregnancy is that alcohol use increases the probability of sexual intercourse and noncontracepted intercourse via impairing judgment. Specifically, it may be theorized that engaging in sexual intercourse provides a 'benefit' via imparting physical and psychological pleasure. Noncontracepted intercourse may offer the added benefit of 'spontaneity' and avoidance of physical side effects resulting from some contraceptives. However, such acts come with a 'cost'--namely, the risk of an unwanted pregnancy, which can either result in the psychological and physical trauma associated with an abortion or in premature motherhood--with opportunity costs in terms of forsaken educational, career, and lifestyle opportunities, (8) Being under the influence of alcohol may bring about myopia, whereby the future is sufficiently discounted so that all future costs associated with pregnancy are perceived as negligible. Thus, the probability of engaging in sexual intercourse (noncontracepted intercourse) increases. Additionally, alcohol use among young males may increase the probability of a pregnancy among teen women via an exogenous 'shock' like rape. (9) The effects of this phenomenon might even be larger than what is inferred from official crime data because male alcohol use might lead to incidents of 'date rapes' and other forms of coercive sexual activity that do not necessarily...