讲解 ECON 6600: Econometrics Spring 2024讲解 Python程序

Spring 2024

Department of Economics

ECON 6600: Econometrics

Empirical Project

(Due in Brightspace: 11:59PM, April 22 – late submission results in 10% reduction per day)

Instructions:

. Choose an empirical paper published in one of the following journals:

o American Economic Review, Econometrica, Journal of Political Economy, Quarterly Journal of Economics, Review of Economic Studies, Review of Economics and Statistics.

These journals have online data and code archives so you can replicate empirical estimates of the papers published there. (Some papers use confidential data and the authors cannot circulate such data. In that case, you cannot replicate their results. If this occurs, then please choose another paper whose results you can replicate.)

. Confirm that the paper of your choice runs the OLS (Stata’s regress command) for its

main estimation result(s) under the i.i.d. setting, i.e., the main results are generated by:

regress (with no standard error option);

regress, robust;

regress, vce(hc2);

regress, vce(hc3); or

regress, cluster().

. Some papers drop, truncate, or winsorize outliers (units with large values of Y). Confirm that the paper of your choice does not drop, truncate, or winsorize outliers.

. Replicate the main empirical result(s) of the paper you chose by using the data and code files found in the journal’s online archive. (You do not have to replicate all results – only the main regression estimates for the key results of the paper will suffice.)

. Use the Stata command testout (available from SSC) to test the consistency and root-n asymptotic normality of the main regressions you replicated in the above step.

. Concisely summarize your results in a report. Be clear about which regression you replicated by referring to the equation number, the table number, and the table column number in the paper, as well as the correct bibliographic citation of the paper of your choice. In case of clustering, please indicate the clustering variable. Submit your report along with the data file(s), a code file, and a pdf copy of the original paper in a single zip file to Brightspace. Write all procedures (replication + test) in a single *.do file.

If properly done, you have already done most of the work in Homework #9 and #10. It remains to organize your findings and reports. The project will be graded by the instructor, not TA.

Continued to the next page.

Important Notes:

. You may collaborate with your classmates, but choose a different paper from your collaborators, i.e., each student should work on a unique paper. (Of course, there is a chance that two students in the class happen to choose the identical paper, but the likelihood of this event is very small given the large number of empirical economics papers published in the list of the above journals.)

. A failure to comply with the above instructions results in a reduction of earned grades.

. As the syllabus states, this project accounts for 15% of the total grade. Optionally, you may choose a second paper and may earn an extra credit of 3%. Furthermore, you may choose a third paper and mean earn another extra credit of 2%. In total, you can earn 20% out of 15% if you submit three projects. Again, these additional papers need to be different from those of your classmates. In case you submit multiple projects, I will use the one with the highest score as your first project accounting for 15%. So, working on multiple projects may serve to mitigate grade risks, as well as to earn extra points.

Significance of Your Work:

. Your project contributes to the body of our scientific knowledge. If the testout command rejects the consistency or the root-n asymptotic normality, then your test results imply that the main results reported in that paper are incredible. If you get such a test result for a paper that has large impacts (e.g., papers that are cited a lot or papers written by famous economists), then your discovery can possibly open a new and important research agenda. Specifically, you may later conduct follow-up research and may possibly overturn the results reported by that important paper.