Estimating the Health Effect of Cigarette Smoking Duration in South Africa.

This paper deepens the empirical analysis of the health effects of smoking by using the average treatment effect on the treated (ATET) and regression discontinuity design (RDD) to estimate the impact of smoking duration on health. The paper estimates the effect of cigarette smoking on health, that is, the exogenous increase in the probability of smoking-related ill health when individuals smoke up to a certain number of years. Using the National Income Dynamic survey (NIDS), the study finds that the probability of reporting poor health and/or suffering smoking-related diseases increases with the years of smoking. The magnitude of the effect is higher when smoking-related diseases rather than self-assessed health is considered but varies across time, socioeconomic status, and with different health outcomes. The effects are robust under several different parametric and non-parametric models. Using RDD, the paper also finds evidence of a discrete jump in poor health when individuals smoke up to 30 years. The results suggest that policies that are designed to reduce current levels of cigarette smoking may have a desirable impact and can create both current and future public health benefits.