Research on the Impact of Higher Cigarette Taxes on Maternal Smoking

» Project Details

Project Summary:

The high rate of low birth weight in the U.S. is well documented. Observational evidence indicates one of the most important identifiable risk factors associated with low birth weight is maternal smoking during pregnancy. In a recently completed working paper, we have examined whether higher cigarette taxes can be used to reduce maternal smoking and improve birth outcomes. Using changes in state excise taxes over the 1989-1992 time period, we examined whether maternal smoking and birth outcomes are responsive to higher cigarette taxes. Data on the outcomes of interest are taken from the Natality Detail files, generating a sample of roughly 10.5 million births. Our estimates are the first to show that taxes so alter the smoking behavior of pregnant women. The results indicate that smoking participation declines when excise taxes are increased and the price elasticity is roughly -.25. More importantly, our results indicate that the tax-induced smoking cessation translates into higher birth weights. What is the heterogeneity in the impact of cigarette taxes across demographic groups? The rate of low birth weight varies tremendously across demographic groups, with young mothers, mothers with low education, African-American women, women with high-parity births, and unmarried women having on average lower birth weight infants. Many of these groups also have high rates of maternal smoking. For cigarette taxes to be an effective policy lever, taxes must alter the smoking rates for these high-risk groups. Given the large sample we will employ, we can analyze the heterogeneity in the cigarette demand elasticities across groups to a greater degree than any previous study. What is the impact of large tax changes on maternal smoking and birth outcomes? In most instances, cigarette tax hikes are relatively small, which requires the use of incredibly large samples to detect the impact of tax hikes on birth outcomes. A statistically significant relationship can be detected with smaller data sets only if there are large changes in tax rates. In the past few years, a number of states have implemented large excise tax increases. By expanding our sample frame, we can examine a number of states in isolation and use these large tax changes as a natural experiment. We will examine changes in smoking rates in the two years before and after tax hikes and compare these with changes in neighboring states with no changes in nominal rates. For example, Michigan instituted a 50-cent cigarette tax hike in 1994. We have identified five states since 1990 (Texas, New York, Massachusetts, Arizona, and Michigan) where the tax changes were large enough, relative to the number of births, to provide sufficient samples to analyze the impact of taxes on birth outcomes. The focus on these natural experiments in isolation will also help us better understand the impact on maternal smoking of massive tax changes such as those being discussed as part of a "global tobacco settlement."