The role of tobacco taxes in starting and quitting smoking: duration analysis of British data

This paper presents new evidence on the determinants of starting and quitting smoking using duration data from the British Health and Lifestyle Survey (HALS). Self-reported data in individual smoking histories coupled with availability of a long time series for the tax rate on cigarettes are used to construct a longitudinal data set in which the tax rate is treated as a time varying covariate. This overcomes the problem of the lack of cross section variation in process that has plagued previous studies of smoking in Britain. The study is the first to identify tax-price elasticities for starting and quitting in Britain. Results for age of starting smoking prior to quitting are reported for Cox, Weibull and gamma models. All of the models are estimated separately for males and females and extensive diagnostic tests are used to guide model specification. A sensitivity analysis is used to assess the robustness of the estimated tax elasticities for starting and quitting. Since the early 1990s successive governments have had a commitment to annual increases in the real level of tobacco taxes, to achieve health policy objectives and encourage people to stop smoking. Our estimated price elasticities directly relate to the impact of above inflation tax rises on the number of years smoked by current smokers. The estimates of the impact of tax on the probability of starting and the age of starting are not encouraging as we do not find a significant effect. However the point estimates of the elasticity of quitting are well defined for males and robust for both males and females. All of our point estimates are in the range -0.40 to -0.63. If the typical number of years smoked is 25 years this implies that the 5% real increase in tobacco duty would lead, on average, to a reduction of between 6 and 9.5 months of smoking for each smoker. Recent estimates suggest that there are around 12.1 million current smokers in the UK (ASH 1999). Thus the potential saving in total number of years smoked across the population is substantial.