Econometrics for summative evaluations: an introduction to recent developments

Abstract

There has recently been a rapid expansion of interest in the econometrics of summative program evaluation, both within Australia and around the world. We provide a review of the key issue and recent developments in this field. A central feature of recent developments is the attempt to allow for program impacts that vary across individuals. This contrasts with earlier econometric approaches which implicitly assumed a homogenous treatment effect. We survey alternative non-experimental estimation strategies, and note that they can be characterised by (1) an assumption about how untreated outcomes vary across individuals: this assumption in turn suggests how the counter-factual untreated outcomes of program participants should be estimated, and (2) the way in which the estimator aggregates or weights the program impacts of different individuals in the treatment group.