As international migration and migrant remittances have increased in recent years, there is a clear need for improved data on international migration and migrant remittances to understand the effects that various policies can have on migrants and migrant households. In a new paper, we argue that large, multi-purpose data collection efforts present good opportunities to study migration in a cost-effective manner. Many countries now implement nationally representative, multi-topic household surveys à la Living Standards Measurement Study (LSMS) surveys, primarily for the purposes of welfare monitoring and analysis. Although LSMS survey questionnaires are designed to study numerous aspects of household welfare and behavior, collecting detailed migration information has not been a priority for most multi-topic household surveys, resulting in large knowledge gaps on migration. Integrating migration information into these data collection efforts can be an efficient way to collect migration data.
Our primary objective is to provide basic guidelines for collecting migration information as part of a multi-topic household survey. The paper first discusses definitional and measurement issues surrounding migration. We then address key methodological tradeoffs researchers face in designing migration survey modules. Among others, these tradeoffs involve the number of questions for different types of migrants, the recall period, the expected accuracy of answers, and the overall survey length. We focus on guiding analysts and survey practitioners through these tradeoffs, and provide sample modules that researchers can adapt to their specific study context.
Using LSMS-style surveys to study migration offers specific advantages over other types of surveys, such as Labor Force Surveys. Most importantly, multi-topic surveys allow the study a much larger range of issues surrounding migration than other types of surveys. Migration is not a random event, so relationships between migration and other variables measured in surveys cannot automatically be considered causal. But even if causal relationships cannot be statistically identified, multi-topic household surveys can further our understanding of relationships between migration and a diverse set of policy-relevant variables.
A drawback of using LSMS-style surveys to study migration is that migrant populations are both clustered and typically small relative to the entire population. Thus, few migrants are captured within the typical sample used in an LSMS-style survey, with the exception of countries with large migration prevalence. Since migration is consequently a statistical “rare event,” we suggest two alternative sample designs: 1) disproportionate sampling of high migration PSUs, and 2) stratified random sampling within PSUs, or two-phase sampling. For the latter, we advocate using newly constructed listings of selected PSUs, allowing the identification and oversampling of migrant households, which can either be carried out in parallel or sequentially to the main survey.
Research on migration (both internal and international) within the multi-topic structure of LSMS surveys can help establish the determinants of migration, inform the creation of future migration policy, and elucidate the impacts of migration on source households. We hope that the data resulting from the thoughtful incorporation of migration components into LSMS-style surveys will enrich future debate on the complex relationship between migration, welfare and policy