Moral hazard under discrete information disclosure: Evidence from food-safety inspections

This paper provides evidence on the responses of producers to mandatory disclosure of discrete quality ratings, a type of policy sometimes referred to as “naming and shaming”. The context is a series of four regulatory changes undertaken by the U.S. Department of Agriculture (USDA) over 2006–2015, regarding disclosure of information about the results of tests for Salmonella in chicken carcasses at slaughter establishments. If establishments exceed certain rates of positive samples, they are designated “Category 2” or “Category 3”. Under some policy regimes, Category 2 and 3 establishments have been listed on a public USDA website. I employ carcass-level data on Salmonella test results over 1999–2018 for all federally inspected establishments to explore the effects of public disclosure and other policy changes on Salmonella test outcomes. First, using a regression discontinuity approach, I demonstrate that: (1) When establishments fail to meet categorization thresholds but these failures do not subject them to public disclosure, Salmonella test performance worsens. (2) When establishments fail to meet thresholds and are therefore subjected to public disclosure, there is no statistically significant change in Salmonella test performance. (3) Under one policy regime, establishment operators relaxed efforts after sustained good Salmonella test performance ensured they would avoid public disclosure. Second, I document that when establishments have more leeway with respect to the thresholds, their performance on Salmonella tests worsens. Third, I use a regression discontinuity in time approach to demonstrate the effects of the series of policy changes on average Salmonella test results. I show that the introduction of public disclosure in 2008 reduced the average rate of positive Salmonella samples by about 55 percent. On the other hand, a tightening of standards in 2011 had a bifurcating effect wherein establishments that performed poorly (prior to the policy change) tended to perform even worse and middling establishments tended to improve. There was no statistically significant effect on the best-performing establishments. The net effect of the tightening of standards in 2011 was to more than double average Salmonella rates. My results provide evidence that the public disclosure of discrete information about Salmonella in chicken carcasses results in producers exerting extra effort to avoid disclosure, and less effort when disclosure status is already certain.