Using Linked Household-level Datasets to Explain Consumer Response to BSE in Canada

Household-level Canadian meat purchases from 2002-2008, household-level egg purchases from 2002-2005 and Food Opinion Survey in 2008 were used to understand how consumers who have different concerns about nutrition react to BSE events and how beef consumption after BSE discoveries were shaped by consumers concerns of food safety and their trust of government and the industry decision makers. Three measures of beef purchased were used to explore consumers’ reaction. A random effects logit model was applied to test whether any beef purchased during a given month. Consumption in terms of unit purchases was measured with a random effects Negative Binomial model and consumption in terms of beef expenditure was measured with a standard random effects model. Consumer behaviors in Alberta differed from Ontario. Consumer reactions to BSE in Alberta were stronger than Ontario. Overall, the more risk consumers attached to BSE, the less beef they purchased in both provinces. Random effects in the three models controlled for unobserved but persistent aspects of households and changed the sign of estimated effects of demographic variables.