Demand for cigarettes revisited: an application of the threshold regression model.

Recent estimates of the income elasticity of cigarette demand have pointed to a disturbing result: a nearly zero or sometimes negative income elasticity. In order to explore the nonlinearity embedded in the cigarette demand structure, we employ a four-regime panel model (dynamic fixed effect) to estimate the cigarette demand function in the United States. The results indicate that income elasticity is (i) positively significant for the income level less than 8,568 US$, (ii) positive but statistically insignificant for the income greater than 18,196 US$, and (iii) negatively significant for the income range between 8,568 and 18,196 US$. In addition, we find that the price elasticity assumes the greatest absolute value for the income level in excess of 18,196 US$, but becomes most inelastic for the income level between 11,129 and 18,196 US$. [ABSTRACT FROM AUTHOR]