What did they say? Respondent identity, question framing, and the measurement of employment

Drawing from two labour market experiments in rural India, we offer insights on the influence of survey design on the measurement of employment. The first experiment contrasts self-reported estimates of employment with proxy-reported estimates from spouses. We find that employment estimates based on reports by men underestimate women's employment by six percentage points compared to estimates from women themselves. There are significant differences in the types of employment activities reported by self and proxy. These divergences emanate from asymmetric measurement errors, stemming from gender-based norm disparities and divergent interpretations of employment and information asymmetry between spouses concerning marginal activities. The second experiment investigates if framing of questions and recall period has an impact on reporting labour market outcomes. We find that using multiple questions to probe about weekly employment status results in a 10-percentage-point-higher estimate of employment for women compared to a single question about their major activity.