Evaluation of Development Programs: Experimental methods
Credits: 4 ECTS
The proponents of public policies and institutional designs always argue that their proposals produce desirable effects. Democratic elections increase the provision of public goods such as education. Compulsory vaccination improves children’s health. Electoral quotas for women change gender attitudes. A big question in studies of development is: How do we know that such claims are true?
Traditional program evaluations rely on observational data, i.e. the study of naturally occurring events. This approach has been increasingly criticized because the types of units that decide to participate in programs are often very different from the units that choose not to participate.
In response to these problems, organizations such as the World Bank of the US Foreign Aid Agency increasingly rely on experimental studies to evaluate programs.
The course has the following main aims:
- To understand the recent criticism to observational evaluation programs and the advantages of experimental research.
- To present the principles of experimental designs with a focus on field experiments. We will also learn how to interpret and analyze experimental data through practical cases.
- To review applications of experimental research in development studies with an emphasis on the effects of health, education, and economic development interventions on the population’s well-being.
- To provide an overview about other applications in the fields of political economy and international relations with an emphasis on the effects of internal public opinion on outcomes such as trade policy and participation in war.
We will draw extensively on the work conducted by the Poverty Action Lab http://www.povertyactionlab.org/, a leading institution in the application of experimental design in development economics as well as the interdisciplinary network Evidence in Governance and Politics (EGAP) http://egap.org/.
In addition, we will use a hands-on approach in data analysis and teach how to analyse experimental data using R, an open source programming language widely used in scientific and professional spheres.
- Attendance and participation: 30%
Students must read the lectures in advance of the class. The bulk of the sessions will consist in discussions of the readings. Students are expected to participate actively in the discussion.
- In class presentations: 30%
In week 5 each student will present an experimental evaluation conducted by the http://www.povertyactionlab.org/. In week 10 they will present an article about an experimental design.
- A final project: 40 %
For the final assignment, the students will design their own experimental evaluation program for a public policy or institution. The project will describe the hypothesis to be tested, the experimental treatment, the case selection, the measurement of outcome variables and the strategy of analysis.