No News Is Good News: Stochastic Parameters versus Media Coverage Indices in Demand Models after Food Scares

We develop a stochastic parameter approach to model the time‐varying impacts of food scares on consumption, as an alternative to the inclusion of news coverage indices in the demand function. We empirically test the methodology on data from four food scares, the 1982 heptachlor milk contamination in Oahu, Hawaii and the bovine spongiform encephalopathy and two Escherichia coli scares on U.S. meat demand over the period 1993–9. Results show that the inclusion of time‐varying parameters in demand models enables the capturing of the impact of food safety information and provides better short‐term forecasts.