Yang’s remittances study has implications for survey design and empirical analyses

February 1, 2023

Researchers working with measuring remittance data often run into one big problem — misreporting. To provide better guidance for their own and their colleagues’ work, Dean Yang,  professor of public policy and economics, and co-authors Giuseppe De Arcangelis, Alexander Fertig, Yuna Liang, and Peter Srouji dove into the accuracy of remittance data.

“All such studies require microdata on remittances, but it is rare to obtain administrative data directly from remittance companies (money transfer operators, or MTOs),” they write. “Most studies therefore rely on remittance data reported either by senders or recipients, but these self-reports can be difficult to collect reliably. Survey respondents may misreport remittances, intentionally or not, introducing bias to empirical analyses using survey-based remittance data.”

In the article published in the Journal of Development Economics, the authors analyze three different data sources for remittances sent by Filipino migrants in the United Arab Emirates (UAE). The first is administrative records obtained from a popular money transfer operator (MTO) called UAE Exchange. The other two sources are self-reported remittance data — one set obtained through traditional survey methodology, and the other through a new app-based data collection platform called Padalapp, developed to collect high-frequency remittance data.

To find discrepancies between the three sources of data, the researchers compare the data in five different ways: recipient vs. migrant survey data, Padalapp data vs. migrant survey data, Padalapp data vs. recipient survey data, migrant survey data v. UAE Exchange administrative data, and Padalapp data vs. UAE Exchange administrative data.

Yang and his colleagues find that the most accurate source of self-reported data is the self-reported migrant survey data, which captures over 90% of the true remittance amount. Self-reported recipient survey data is next, with under-reporting of remittances coming in at around 25%. Finally, Padalapp users significantly under-report remittances, which the researchers attribute to the length of the study, difficulty of compliance, and fatigue. 

The study provides important guidance for those studying remittances by discovering the most reliable data sources. 

“In instances where researchers only have access to a single source of self-reported data, our findings provide an estimate of the degree of measurement error they can expect,” Yang and his colleagues advise. “In rare cases where researchers have access to multiple sources of self-reports, we provide guidance on how to assess the relative accuracy of these sources.”

Read the entirety of the paper in the Journal of Development Economics.