RCTs for better policy? The case of public systems in developing countries

This paper considers the application of randomised controlled trials (RCTs) to improve public systems in developing countries. Arguing that existing critiques of RCTs as to problems with extrapolation and narrowness of scope are especially relevant in this context, I consider the claim that these shortcomings can be ameliorated through better causal explanations. I analyse how theoretical mathematical models are used to construct causal explanations, and argue that it is still difficult to extrapolate or address the subjectivity inherent in the choice of interventions. I illustrate these arguments using two prominent RCTs that have trialled interventions to improve government schools in India.