Health Information, Taxes and Life-Course Smoking: Evidence from Turkey

In this paper, we analyze the impact of information about health risk of smoking and taxes on an individual’s decisions to start or stop smoking. We use new individual level data from Turkey. First, we construct life-course smoking histories of Turkish males and females using data from the Global Adult Tobacco Survey. We track the smoking behavior from 1925 to 2008. To measure what information these individuals might have seen, we counted the number of articles about health risks of smoking that were published each year in Milliyet, Turkey's major newspaper. We counted all articles published between 1925 and 2008. We use variation in the flow of such information to model the probability that individuals started or stopped smoking.

As a country, Turkey provides an interesting setting for the type of behavior we will analyze because smokers faced rich and varied influences on their smoking decisions. Although the smoking prevalence rate was one of the highest in the world especially among males, as of 2013, it is the first and the only country in the world to have fully implemented all of the World Health Organization's recommended tobacco control policies. Successive generations of Turks experienced dramatically different economic, political, and social conditions. The government changed the level and structure of tobacco taxes. The Turkish government first allowed but then ended the sale of cigarettes through a government-owned monopoly during the period we study. Since the 1990s local and multinational firms compete in a relatively free market; a change event that is associated with large jumps in cigarette consumption. Finally, since the 1980s Turkey has increasingly regulated the sale and consumption of tobacco to reduce tobacco use.

We use retrospective smoking information (age started, age quit) to construct life-course smoking histories of current and former smokers. This maximizes the size of our analysis sample. We track smoking behavior of multiple birth cohorts of men and women in every year of their lives. Consequently, we can exploit the substantial variation in taxes, tobacco control policies, and in the information about health risks of smoking that each birth cohort faced. In addition to the counts of articles published in widely circulated Turkish newspaper, we also consider the articles on smoking health risks that were published in US magazines as a possible determinant of the appearance of articles in the Turkish press. In our analysis, we control for other factors affecting individual’s decision to smoke or to quit, including education, age, location income level as well as the socio-economic development of the country during the long sample period. We also analyze how the effect of information changes by education, age or gender using interaction variables.

Our initial findings indicate that information about health risk of smoking significantly and negatively associated with female’s decision to initiate or quit smoking and male’s decision to quit. This information is found to be affecting all individuals’ decision to quit regardless of their age or education level. However, it affects most the initiation decision of youngest female cohorts.