This project asks you to write an essay that reviews, replicates, and extends an empirical paper in one of the topics we cover in 14.771. You must choose an empirical paper that has available data. It does not have to be on the reading list. As the risk of stating the obvious, you MAY NOT choose a paper on which you were a research assistant.
Try to choose a paper for which the data source is pretty complete, i.e., it includes not only the processed data to produce the tables in the paper and code to go with them, but also clear documentation.
Your essay should have three components, outlined below.
- What question does the study ask, and why is this of economic interest? What are the most important findings in the paper?
- Describe an ideal research design for the question at hand. Is the work causal or descriptive? If the former, which assumptions support a causal interpretation of the results presented in your chosen paper? Are the econometric techniques used in the study likely to yield estimates with a causal interpretation? Are the results convincing?
- Identify the main findings and use the authors’ data to replicate the published/working paper findings if possible. Note that you need only to replicate the main results of the paper, not all the tables.
- Summarize and compare your replication results to the original results, with original and replication results reported side-by-side in a single table. Highlight any differences. Explain why you think your results differ from the original (if they do).
- Send us the code attached with your report. This should include the construction from the raw data sets to the data set ready for analysis to the extent you have these data available.
Extend the work either by either:
- (a) estimating alternative specifications that may illuminate issues and questions raised by the paper (e.g., specification checks; subsamples of special interest; distributional effect)
- (b) OR: if available, start from the very raw data to produce different variables that the authors may have chosen to include (or a different way to compute some of their main outcomes)
The product of this exercise is a short essay. Start polishing your writing skills now rather than the summer before you go on the job market. Learn to tell stories with numbers by imitating the good work of others. And as is increasingly required in economics as well, please submit a documented code to go with your results. The papers in the journals of the American Economic Association (AEA) have their data available for replication. That could be a good place to look for papers to replicate! Bear in mind that this is only true from 2015 on.
Datahub for Field Experiments in Economics and Public Policy with data for 150 different recent RCTs.