Constructing synthetic panels for the purpose of studying poverty dynamics: A primer

Abstract

We describe a recently developed approach for constructing synthetic panels from cross-section data and consider how it can be employed to study poverty dynamics. Initially introduced as a means to estimating upper and lower bounds on poverty transitions in the absence of panel data, further refinements to the method aim to permit as well the calculation of point estimates. We describe the assumptions that underpin the basic approach and its extensions and discuss their plausibility. We chart applications of the method in various contexts, with a view to gauging its overall validity and robustness. While proper panel-based analysis of welfare dynamics is clearly preferable, we suggest that the method can be useful in the all-too-common settings where panel data are unavailable or suffer from particularly pressing quality concerns.