Three essays on development economics : public policies and geographical discontinuities[Trois essais en économie du développement : politiques publiques et discontinuités géographiques]

Evidence-based policies are crucial to maximize well-being. In recent years, governments, international development organizations and academia have mainstreamed the need for data and evidence in the design, implementation and evaluation of public policies. The 2019 Nobel Prize in economics awarded to Duflo, Banerjee and Kremer is another clear example of the relevance of policy evaluation to tackle many global and local development issues. While randomized experiments appear to be one of the most precise ways to identify the causal effect of policies, a wide range of government programs are not designed in a way that ensures their ex-post evaluation – as this can be costly and often requires human resources and capacity that tend to be scarce particularly in developing countries and regions. Yet policy makers need evidence to adjust policies, re-allocate budget and adapt actions in order to maximize people's well-being given the available resources.This thesis presents three cases where ex-post policy evaluation is possible either by exploiting geographical discontinuities in implementation of the policy (Chapters 1 and 3) or regional discontinuities created by policy regulations (Chapter 2). Each chapter focuses on different (but highly interrelated) issues of the 2030 Agenda and the Sustainable Development Goals (SDGs). While Chapter 1 assesses the impact of universal healthcare on maternal and perinatal mortality, Chapter 3 explores the effect of decentralization on outcomes related to the global goal to end hunger, notably on food security. Chapter 2 studies how funds stimulate business creation and productivity, which is part of the SDGs for economic prosperity.