Personnel Economics Spring 2024

Please read “Does Working from Home Work?  Evidence from a Chinese Experiment” (Quarterly Journal of Economics,  2015) by Nicholas A. Bloom, James Liang, John Roberts, Zhichun Jenny Ying. Make sure you are using the published QJE version of the paper, in replica- tion files and previous version, some tables and figures are labeled differently. Download the data from Nick Bloom’s Stanford website. In WFH.zip, you will find several Stata dta files and a do file called ’Tables2014.do’, which will help you replicate the findings of this paper. The paper can be replicated by running this file, but you will be asked to make adjustments to the analysis.  Do not try to code the regressions from scratch.

The main goal of the assignment is to get you acquainted with personnel data.  Please don’t be discouraged if there is a part of the assignment which is hard or unclear.  Submit the assignment as a single PDF file.  Sections that need to be turned in are in BOLD and should be numbered based on their step. You will need to upload your Stata do file that is clearly commented and organized.

1.  Table 2 Replication (20 points)

(a)  Replicate Columns 1 and 2 from Table 2 (titled The Performance Inpact of WFH). In 1-2 sentences, how are these estimation strategies different?

(b) Create a variable called pre_experiment_mean that is equal to the mean of perform1 before the experiment starts for each person.   This value will be constant for each person. This is a bit tricky, be careful.

Replicate Column 2 from Table 2 again, but this time, control for the pre-treatment mean of each person by adding it into the regression.  Report the point estimate, rounded 3 digits.

(c)  Briefly comment on the difference between these three approaches, how is this related to Footnote 17?

2.  Table 2 Extenstion (10 points)

(a)  Estimate the diff-in-diff estimator (Column 1) across one measures of heterogeneity. For example, estimate the treatment effect for workers who are above or below the median productivity level as in Figure X. Report the point estimate and standard error, rounded 3 digits.  Explain the form of heterogeneity you are exploring and why it may be interesting.