Information, endogeneity, and consumer health behaviour: application to dietary intakes

Abstract

Due to heightened public health interest, a growing number of consumer health behaviour studies are focusing on the effect of health information on the demand for health inputs and outcomes. Many of these studies, however, have overlooked the potential endogeneity of information variables stemming from unobserved individual heterogeneity. Due to simultaneity bias, ignoring endogeneity may lead to inaccurate estimates of informational effects on health behaviour. Using dietary intake data for fat, saturated fat, cholesterol, and fibre, this paper illustrates the pitfalls of treating health information related to these nutrients as exogenous variables in their demand equations. In most of the estimated models, the exogeneity of information is statistically rejected. When the information variables are treated as exogenous variables, their effects on dietary intakes are underestimated. The estimated effects of key intake determinants such as income, education, ethnicity, and race are also different in such a specification compared to a specification which treats information variables as endogenous. Implications for nutrition education policies are discussed.