Response by Adults to Increases in Cigarette Prices by Sociodemographic Characteristics.

• Author: Farrelly, Matthew C.

Matthew C. Farrelly [*]

Jeremy W. Bray [+]

Terry Pechacek [++]

Trevor Woollery [++]

Cigarette excise taxes are widely viewed by health economists as an effective tool to reduce cigarette consumption. However, those opposed to increasing cigarette excise taxes often state that the taxes unfairly target certain segments of the population, notably the poor and minorities. Some of this opposition may have been fueled by a lack of understanding of how the tax will affect the health and welfare of various demographic groups of interest. This article provides guidance to policy makers by estimating price elasticities among adults by gender, income, age, and race or ethnicity. Women, adults with income at or below the median income, young adults, African-Americans, and Hispanics are most responsive to cigarette price increases. For example, adults with income at or below the median are more than four times as price-responsive as those with income above the median.

1. Introduction

   The prevalence of cigarette smoking among high school seniors increased more than 30% between 1992 and 1997 (Monitoring the Future 1998). In response to this recent increase, federal, state, and local initiatives have been proposed and enacted to curb smoking, especially among teenagers. These initiatives include increasing cigarette excise taxes. Increasing these taxes effectively discourages smoking among both teenagers and adults (Chaloupka and Grossman 1996; CDC 1998; Evans and Farrelly 1998; Evans and Huang 1998). Those opposed to increased cigarette excise taxes argue that these taxes are regressive and hit "hardest those least able to afford them" (Tobacco Institute 1998). In addition, increases in cigarette prices may impose a disproportionate financial burden on racial and ethnic minority groups and the poor. If these same groups were the most responsive to taxes, however, then the regressive effects of the tax would be mitigated by above-average decreases in smoking.

   It is important to understand the sociodemographic characteristics of smokers who are responsive to tax changes and those who are not. Previous studies have found that certain sociodemographic groups are more price-responsive than others. For example, Chaloupka and Grossman (1996), Evans and Farrelly (1998), Evans and Huang (1998), and Tauras and Chaloupka (1999) showed that young adults are more price-responsive than older adults. The results of studies by Evans and Huang (1998) and Chaloupka and Pacula (1999) suggest that, among high school seniors, nonwhites are more price-responsive than whites. It is not known whether cigarette price increases differentially affect adults of different races and ethnicities. Only one study, conducted in Britain, has examined price elasticities by income group (Townsend, Roderick, and Cooper 1994). These authors found that people in lower socioeconomic groups are more price-responsive than those in higher socioeconomic groups. Nor is it known whether the poor (young or old ) are more or less price-responsive than the rich. To answer these questions, we evaluated the effect of cigarette price increases by gender, income, age, and race or ethnicity with a nationally representative sample of more than 350,000 adults. This article also elaborates on work presented in an earlier, brief summary of results from similar models (CDC 1998).

2. Data

   We pooled data from 14 years (1976-1980, 1983, 1985, and 1987-1993) from the National Health Interview Survey (NHIS). The NHIS is a nationally representative multistage probability sample of the civilian, noninstitutionalized U.S. population age 18 and older) The NHIS obtains information about the amount and distribution of illness, disability, and chronic impairments and about the kinds of health services respondents receive. In supplements to the NHIS before 1992, respondents were asked, "Have you smoked 100 cigarettes in your entire life?" and "Do you smoke cigarettes now?" In 1992 and 1993, participants were asked, "Do you now smoke cigarettes every day, some days, or not at all?" We define current smokers as those who reported having smoked at least 100 cigarettes during their lifetimes and who reported currently smoking cigarettes either every day or some days. Current smokers were asked, "On average, how many cigarettes do you smoke per day?" We define cigarette demand as the number of cigarettes smok ed per day, conditional on the respondent's being a current smoker. Information on gender, income, age, race or ethnicity, and other sociodemographic factors was obtained from the core NHIS questionnaire.

   Average annual cigarette prices were obtained for each state from the Tobacco Institute's The Tax Burden on Tobacco (1998). Prices were adjusted for inflation (constant 1982-1984 dollars) and merged into the NHIS by year and state of residence. [2] Combined, the 14 cross-sections of the NHIS consisted of 367,106 respondents; of those, complete sociodemographic and price data were available for 354,228 (approximately 25,000 respondents per year).

   Figure 1 illustrates the consistent differences in the prevalence of cigarette smoking among non-Hispanic African-Americans, Hispanics, and non-Hispanic whites [3] from 1976 through 1993 (data are interpolated for 1981, 1982, 1984, and 1986). The prevalence of smoking dropped more rapidly for nonwhites than for whites (38% vs. 30%). During this period, the real price of cigarettes increased by 48%. Determining differential price effects for African-Americans, Hispanics, and whites may explain some of these differences in smoking prevalence.

   Table 1 presents summary statistics for the entire 1976-1993 NHIS database. During this period, a mean of 29% of the sample smoked, and smokers consumed a mean of 20 cigarettes per day. Ten percent of the sample was African-American, 6% was Hispanic, and 81% was white. The average age of the respondents was 44 years, and almost 75% had at least a high school degree.

3. Methods

   The two-part estimation procedure has been used extensively in health economics to model the demand for medical care (Duan et al. 1982, 1984; Manning et al. 1987), drinking (Manning, Blumburg, and Moulton 1995), and cigarette smoking (Wasserman et al. 1991; Grossman et al. 1993; Chaloupka and Wechsler 1997; Chaloupka, Tauras, and Grossman 1997). Using this two-part framework, we model separately the decision to smoke and the quantity of cigarettes smoked. In the first stage of the two-part...