Structural Modelling in Policymaking

2020-03-02 connect learns tool utilization of evidence in decision-making

Structural modelling, that is the use of behavioural models to add a framework to the decision problem of an agent, is a useful yet underused tool in evaluation. This paper provides a general introduction to structural modelling, as well as an overview of other commonly used evaluation techniques in Economics and other social sciences. It then goes on to show with three key case studies, how structural models can be used to enrich the findings from randomised control trials. The case studies cover a wide range of policy questions: examining demand for health products in Kenya, incentivising teachers to attend school in India, and evaluating conditional cash transfers for education in Mexico. The case studies show how structural models add to our understanding of the mechanisms behind a given treatment effect, how the findings may change when the policy is rolled out under different circumstances, as well as allowing for the evaluation of different policies that were not originally trialled. The common pitfalls of structural models are discussed, with guidance provided throughout on how to conduct sensitivity analysis and model validation. It is hoped that this paper will persuade other researchers to use structural models, in conjunction with randomised control trials, that will lead to improved evaluation results, a deeper understanding of important problems and better informed policymaking in the future.