The Robustness of Price Elasticity Estimates: A Revisit of Various Methodologies Used To Estimate Demand for Cigarettes

Abstract

Although the price elasticity of cigarettes has been estimated using a variety of models and methods, few studies apply these methods to the same dataset and compare the estimates from alternative approaches. This paper employs the National Longitudinal Survey of Youth 1997 to provide the first comprehensive study that addresses most methodology limitations found in the previous literature and compares cigarette price elasticity estimated from many approaches. We estimate the elasticity by allowing the addictiveness to differ by non-addictive, myopic and rational smoking. We account for the state-level anti-smoking sentiment using its time-invariant and time-variant measures. We control for the heterogeneity of smoking behavior among population using the two-part model in a cross-sectional approach and the two-way error component model in a longitudinal approach. The estimates of elasticity are significant in almost all specifications and strikingly close in magnitude. The price elasticity of smoking participation ranges from -0.08 to -0.24 with an average of -0.16, and the price elasticity of cigarette consumption rages from -0.07 to -0.24 with an average of -0.14. This indicates that when estimated using the same panel dataset with a relative long time span, the elasticity is likely to be estimated consistently regardless of approach.