&&ReWrAp:HEADERFOOTER:0:ReWrAp&&

Date of Award

Spring 2012

Degree Type

Thesis

Degree Name

Doctor of Philosophy (PhD)

Department

Economics

Supervisor

Philip DeCicca

Language

English

Committee Member

Jeremiah Hurley and Lonnie Magee

Abstract

Formal economic analysis of tobacco products dates back to the middle of the 20th century. At the beginning, most of the research was done by the tobacco industry itself. Later, research interests switched to the public health perspective following the publication of British and American reports on smoking and health. Many papers consistently demonstrate that smoking significantly damages health and that the cigarette tax is a popular policy tool to reduce smoking. My first two essays focus on the issue of the cigarette excise tax and smoking. In the first essay, we analyze a possible tax avoidance behaviour measured by pack versus carton purchasing behaviour, since carton purchasing is associated with a substantial quantity discount. We find that smokers who intend to quit switch to packs in response to tax increases, while smokers who do not want to quit smoking systematically switch to cartons when taxes increase. In the second essay, we investigate whether the smoking participation of relatively older smokers is sensitive to cigarette taxes. The consensus in the literature is that older smokers are not price-responsive. Medical research, however, suggests quitting smoking even at an older age can still generate substantial health benefit for the old smokers. Using the most recent large tax increases across US, we find that both the older smokers’ desire to quit and actual smoking participation rate are responsive to tax increases. These results run contrary to most of the literature and may suggest that tax increases generate substantial health benefits for older smokers. In the last essay, we extend the model specification test proposed by Fan et al. (2006) to the extent that we also smooth the discrete dependent variable. We derive the null distribution of the test and also show that the test is consistent even when the null hypothesis fails to hold. Finally, a Monte Carlo simulation study shows that by smoothing the categorical dependent variable, our test enjoys substantial power gains.

McMaster University Library