Money and lies: proxy respondents and the mismeasurement of income in surveys

Abstract

When sampled individuals are not found at home, many surveys rely on a proxy respondent: another knowledgeable household member. We study the difference between self- and proxy-reported labor income in Mexico. We use the panel structure of the Mexican labor force survey and compare workers’ income when they report it themselves to their income when another household member does the reporting. We find that the monthly wage of male workers is 6.1% lower when reported by a proxy. For female workers, the reporting gap is minute. We provide evidence that the gap in the reported income of male workers is due to asymmetry of information within the household, in part due to men hiding income from their relatives. Finally, we study the implications of using proxy respondents and find that it can lead to an underestimation of the gender wage gap by 60%.