Estimating the Demand for Lottery Gambling

Abstract

This study examines the role of functional form and the definition of effective price in the estimation of the demand elasticity for Lotto Texas. Using a nonparametric local mean function analysis, we find that the demand restriction imbedded in a semilog specification with effective price defined as net expected revenue per ticket best matches the observed data. In contrast, the restriction of a constant elasticity of sales with respect to net expected revenue—implicitly assumed by the log-linear specification typically adopted in this literature—is strongly rejected. Further, we find that omission of drawings with negative net expected revenue can meaningfully influence coefficient estimation. Our regression analysis finds that the effect of omitting such drawings is an order of magnitude times greater than that of failing to account for the endogeneity of effective price.