Labor (mis?)measurement in agriculture

Livelihoods are changing rapidly in rural areas. Measuring and categorizing peoples’ labor activities in relation to the agricultural sector is important for understanding income earning opportunities and designing effective policy. Conventional data collection methods ask about individuals’ main work activities over the past year. Descriptions are recorded in the field, postcoded, and eventually categorized. This approach is costly to collect, fatiguing for respondents, and may create distortions. We show that a more direct approach, asking respondents to categorize their major work activities themselves, provides similar resulting data despite some caveats and lessons for best enumeration practices. We compare these main activities to a series of yes/no questions about participation in a set of specific work tasks. We find a 12% incidence of “missing†work, whereby individuals who reported participation in at least one but did not have any recorded major activities. Looking by sector of work, women and youth are disproportionately more likely to have agricultural contributions “missed,†while we find no corresponding bias in undercounting of non-agricultural work. Finally, we test the effect of randomly positioning the task-based questions before the listing of major activities but do not find significant effects on the number or type of activities reported.