Measuring Self-Employment Income in Household Surveys: Evidence From A Randomised Survey Experiment

This paper investigates how recall periods affect the accuracy of self-employment income data in household surveys conducted in developing countries. Using randomised survey experiments in Cambodia, we compare fishery income estimates obtained through different recall periods – weekly, monthly, seasonal, and yearly – and assess the potential of mobile phone interviews to reduce recall bias. The results reveal substantial recall-induced distortions: longer recall periods systematically lead to overreporting of catch volumes and associated costs, while underreporting the diversity of fish species. Furthermore, recall bias is consistently lower when data are collected via mobile phone interviews compared to traditional face-to-face methods.