Are You Being Asked? Impacts of Respondent Selection on Measuring Employment in Malawi

Abstract

Accurate estimates of men's and women's employment are necessary for understanding sources of productivity and growth and designing well-targeted, gender-sensitive labour policies. This paper aims to address a key question—how respondent selection in household and labour force surveys affects these estimates—by leveraging two concurrent national surveys in Malawi that relied on the same questionnaire and field teams but differed in their approach to respondent selection. As compared with direct and private interviews with respondents, the ‘business-as-usual’ approach that allows for proxy reporting when targeted respondents are not available, as well as a mix of other standard survey approaches often used under time and resource constraints, is associated with significantly lower reporting of employment across a range of wage and self-employment activities. Although the effects are seemingly limited in absolute terms, they are quite large in relative terms, vis-à-vis the average participation rates and they tend to be more pronounced for women respondents and concerning questions with longer/12-month recall periods. The analysis also examines how household wealth, proxy reporting and difficulties associated with interpreting questions may be linked to lower reporting in the business-as-usual approach, and which can be examined in future methodological experimentation.