There is strong evidence that an increase in the price of tobacco products reduces tobacco use. There are disturbingly few studies that used individual-level data to examine the effect of tobacco prices on tobacco use in low- and middle-income countries (LMICs). In South Africa, studies have predominantly relied on aggregate time-series data and have been unable to examine the effect of price on smoking participation. In this paper, we use longitudinal data to analyse changes in individual smoking behaviour, including smoking participation and intensity.
We compare the results of pooled OLS (POLS), Heckman two-step procedure, random and fixed effect (RE, FE) panel regression. Our results points to the importance of cigarette prices in reducing both smoking prevalence and intensity. The price estimates from the POLS and RE are generally small, but estimates from the Heckman procedure are between -0.52 for economy price and -0.61 for mid-price. Therefore, a 10% increase in cigarette prices reduces tobacco consumption by up to 6%. Fixed effect estimates show no significant effect of price on smoking intensity. This therefore suggest that additional attention be paid to the methodological approach vis-a-vis the nature of price when using panel data to estimate price elasticity. The findings suggest that in the presence of the changing market structure (from near monopoly to a more competitive market) in 2010, tobacco excise tax increases remain a desirable policy tool for reducing cigarette consumption.