Measuring Agricultural Wages Amidst Conflict: Evidence From Myanmar

ABSTRACT Violent conflict is globally on the rise, and an increasing body of literature analyses its impacts on agricultural production and food security. Yet the quality of data collected in conflict zones is sometimes questioned and remains underexplored. This paper addresses this gap by examining two types of reporting bias—recall and question order bias—under varying levels of conflict in Myanmar. We focus on agricultural wages as the key variable of interest because wages are frequently measured in surveys and represent an important development indicator, serving as both a key input cost for farmers and a crucial income source for hired farm workers. We match ACLED conflict data with farm survey data collected across three panel rounds, with the third round including recall modules to assess recall bias. We also implement a question order experiment in the second round to assess whether the sequencing of wage questions by year and gender influences reported wages. We find modest recall bias in reported wages, but no statistically significant evidence that recall bias increases with conflict severity and only limited evidence of question order bias. Among the question order biases, only the interaction between conflict severity and asking about the previous year's wage first is statistically significant, and its magnitude is economically small. Overall reporting biases are modest and are driven more by the type of wage being reported than by conflict severity. We conclude that reliable data can be collected in conflict settings—and such data are rare but essential for evidence‐based policies and interventions as conflicts continue to rise globally.