Despite numerous anti-smoking policy initiatives, smoking remains the leading preventable cause of death and disease in the U.S. (USHHS, 2010).  21.8% of the population currently smokes (MMWR, 2010).  One standard public health policy to reduce smoking is cigarette taxation.  Economic theory predicts that a price increase will lead to a reduction in the quantity of cigarettes demanded.  If tax increases are passed on to consumers, taxes can raise prices (Kenkel, 2005).  A series of empirical studies have evaluated the effectiveness of cigarette taxes (e.g., Ross & Chaploupka, 2003; DiCicca et al, 2002; Cook & Carpenter, 2008).  This line of research provides some evidence that cigarette taxes can reduce smoking outcomes. 

The elderly (65 years+) comprise an increasingly large portion of the U.S. population (Agency on Aging (AOA), 2012).  In 2009 there were 39.6 million elderly (12.9% of the population) and by 2030 there will be 72.1 million (19% of the population).  Understanding health determinants in this growing sector of the population is necessary for effective health policy and improving population health.  Only one published study (DeCicca & McLeod, 2008) has examined the effectiveness of cigarette taxation among the elderly.  This study draws cross-sectional data from the Centers for Disease Control to examine the effect of cigarette taxes on smoking participation.  Findings imply that cigarette taxes can reduce the probability of smoking among the elderly. 

We use the geocoded Health and Retirement Study (HRS) to further investigate the effectiveness of smoking policies on elderly smoking behaviors.  We extend DeCicca & McLeod (DM) in five substantial ways.  First, the HRS allows us to examine a wider range of smoking outcomes than considered in the DM study: any smoking, daily smoking, number of cigarettes smoked in the past 30 days, cessation, and relapse.  Second, because the HRS is a longitudinal survey we are able to include person fixed effect and exploit within person variation in cigarette taxes.   Inclusion of person fixed effects may minimize omitted variable bias that results from difficult-to-observe between state differences.  Third, we rigorously examine the effectiveness of an additional health policy lever, public smoking bans, and its interaction with taxes.  Fourth, because the HRS is a large sample of the elderly we are able to separately examine the near-elderly (51-64), elderly (65-79), and oldest old (80+).  Five, the HRS contains detailed smoking histories and we are able to compare effects between lifetime heavy smokers and non-lifetime heavy smokers. We control for a rich set of personal characteristics, and use state fixed effects, state-level demographics, unemployment rates, and state anti-smoking sentiment to control for between state differences. 

Our findings are consistent with the work of DM: we find evidence that the elderly do respond to cigarette taxes.  We observe heterogeneity in tax responsiveness by outcome, age, gender, and smoking history.  We find weak evidence that smoking bans in public locations may lead to reductions in some smoking outcomes.  Our results shed light on how well standard anti-smoking policy levers may operate as the U.S. population ages.