We investigate the impacts of food safety on a weakly separable U.S. meat demand system (beef, pork, and poultry) using both the Generalized Almost Ideal Demand System (GAIDS) and the Rotterdam model. To measure food safety, indices are constructed based on the number of meat safety articles reported by the top 50 English language newspapers. The GAIDS permits estimation of food safety parameters in a theoretically consistent framework using the concept of demographic translation. The Rotterdam model offers a comparison of estimates to the GAIDS and a further test of the robustness of the food safety elasticities. We find that inferences with respect to food safety and autocorrelation are fragile to functional form choices. From the models investigated there is mixed evidence as to whether food safety concerns have impacted demand. Evidence from the GAI model indicates that food safety impacts could last for several quarters, whereas evidence from the Rotterdam model fails to reject the hypothesis that food safety variables are statistically different from zero over any period. There is also mixed evidence concerning autocorrelation. In the GAI model the problem of autocorrelation disappears by including food safety variables, which are found to be statistically significant and seemingly rectifying the misspecified model that omits food safety variables. This is not the case for the Rotterdam model where a correction for serial correlation is needed even in the presence of the food safety variables, which themselves are not statistically significant. The fragility of these inferences and estimated economic effects to specification choices, particularly to functional form and how the food safety variables enter the demand functions, make it difficult to draw many definitive conclusions about the magnitude or sign of food safety impacts on demand. Concerns of their statistical significance notwithstanding, the most definitive observation is that they are likely to be very small relative to price and expenditure effects and to other possible factors that may have impacted demand.


Downloads Statistics

Download Full History