Legislator and Citizen Tolerance of Corruption: Data from Three Countries

• In Event: Elections, Accountability, and Representation

Abstract

We survey representative samples of adult voters and also the universe of national and first-level subnational legislators in three countries (Italy, Colombia, and Pakistan) to study their tolerance for corruption (via corruption-related vignettes) and their beliefs about the prevalence of corruption and anti-corruption enforcement in their countries.

The survey includes an experimental manipulation to assess respondents' willingness to engage in deception. In addition, we conduct an experimental manipulation with legislators where we provide them aggregate results of our citizen survey to show them what citizens consider corrupt. We use this informational treatment to assess if it realigns legislator preferences with those observed in the electorate.

Response rates of legislators vary. In the two provinces in Pakistan where we survey provincial legislators, we have very high response rates approaching 80 percent. We do not survey national legislators in Pakistan. In Italy, response rates for national and regional legislators are approximately 10 percent. In Colombia, we anticipate high response rates thanks to our partnership with an NGO there but the data have yet to be collected.

Our analysis mainly investigates differences between legislators and citizens in each country, but we also describe the patterns of differences amongst the three countries, which represent vastly different levels of economic development.

Preliminary results show that citizens in all three countries consider corruption highly prevalent, contrary to our pre-registered hypothesis that voters would report higher levels of corruption as the level of economic development fell. Also contrary to our pre-registered expectations, we find evidence that legislators are just as intolerant of corruption as voters.

Authors

• Miriam A. Golden, Stanford University

• Raymond Fisman, Boston University