ECO 231W

Undergraduate Econometrics


You will replicate the project paper (the same one of question 2 in the Midterm). This means that you will get the data and clean it according to the descriptions in the paper. You can find the data and the codebook in the Download page. Then you will generate the same variables that the authors say they do, and run the same regressions. Finally, you will make sure that your code presents all the values in the table in a format that is very easy to see. You need to reproduce Tables 2, 3 and 4.

The code must be clean and organized. You have a good measure of freedom in deciding how you will organize your code. The point is to use comments to create sections for each different part of the replication. You should also use comments to explain what the commands are doing. The way you should approach this is as someone who is writing a code that will be used by others. Could they run it if they had the same data set? Could they understand it with minimum effort?

As with the homeworks, this assignment is a .do file. One .do file. You are graded on the correctness of what you did, on how close you get to the results, on how clean and elegant is your code, and on how easy it is to understand. Think that you are working with a coauthor, or with colleagues in a company. You don’t want to explain so much that it is tiresome to read, but you must make sure that the reader doesn’t spend time figuring out what you intended with this or that command.

The replication is usually a challenging assignment. The hard truth is that you can very seldom replicate exactly the same results of a paper. Sometimes papers have mistakes, but even when they don’t, the authors may have done something with the data that they didn’t describe in the paper, perhaps by distraction, perhaps intentionally. So, you won’t get the same results that they did. How close can you get? As you try to figure this out you will gain a very deep understanding of the paper.

Usually this process is very time consuming. Don’t play with this, you never replicated a paper before, you don’t know what it is like. You will run into coding problems, you may need to learn a Stata trick, you may have to start again. Even when you get it to run, you may not be getting results that are close enough, what are you doing differently? You will need time to go to office hours and experiment until you figure it out.

Youngmin is the specialist in the replication. The undergrad TAs have done replications of their own when they took this course, so they can help as well. However, on this particular assignment, try to get Youngmin’s help if you can.

The replication is due on Friday 03/31 at 5:00 pm. You must upload the replication file in the upload page. For details on how to upload it, read this. The password is still the same as for the homeworks. The replication is graded out of 10 points, and it is worth 10% of the grade.

If your .do file doesn’t run, the TA will request that you fix it, and give you a deadline (usually until the next day). Each time he has to request a correction for the file to run, you lose 2 points. If you fail to make the deadline of a correction request, your replication won’t be accepted anymore. You choose which email we will use to contact you when you upload the file. Make sure that it is one you check often.

Every hour past the due time, your replication loses 10 points. Here is a breakdown:

Received before 5:00 pm until 5:59: no penalty
Received between 6:00 and 6:59: -2
Received between 7:00 and 7:59: -4

and so on until 10 pm, at which time we will not accept your replication any longer. Read the policy on late assignments here.

NOTICE: the replication is due before or at 5:00 pm. We only start discounting points at 6:00 pm to allow for internet delays, browser reloading or any other unpleasant things that may happen. We will not accept any excuse related to technology. If you can’t rely on your connection, or your browser, then take measures to upload your homework earlier, or even better, the night before.

Think that there are way too many rules to follow in this course? Read this.