Rural Labor and Long Recall Loss

Commonly used data collection practices use annual recall to capture individuals’ labor activities over a year. However, long recall periods are likely to suffer from distortions and loss, particularly when work patterns are seasonal and informal. In a panel of rural households in Malawi, we use a survey experiment to test the effect of using long recall periods on the reported number of labor activities, labor supply, and types of work relative to those resulting from a set of shorter, quarterly interviews. We document large losses from the longer recall window, particularly on the extensive margin of labor supply with reductions of over 20%. These losses are greatest for periods furthest from the last survey round and are especially large among individuals whose labor supply is being reported for them, reaching as high as 50% losses for some outcomes. The composition of households’ primary respondents, predominantly male and older, as well as differential effects by age both suggest that use of long recall may lead to meaningful biases by both age and gender in resulting data.