Replication Exercise #2Promotion Policy and Diff-in-diff Estimation1.Read Ahn, T, J Niven, and A Veilleux (2021), How long have you been waiting? Explaining the role of irrelevant...

1 answer below »









Replication Exercise #2














Promotion Policy and Diff-in-diff Estimation





























1.








Read Ahn, T, J Niven, and A Veilleux (2021), How long have you been waiting? Explaining the role of irrelevant information in the promotion of US Navy officers,

Economics Bulletin


















2.








Reproduce Table 4 in the paper using the Stata dataset “replication_promotion.dta” You are welcome to use R if you wish. I should be able to run your script without errors by only changing the parent directory.

















3.








Additional questions to answer:








a.








Who do the authors just run LPM? What is lost or gained by doing this instead of logit/probit?








b.








In the text, the authors use In-Zone group as the control group since the authors claim they are not impacted by the policy change, yet they write that this is

technically


incorrect. Explain their reasoning for why the IZ group is not exactly a control group and why they treat it as a control group anyway.









c.








What other variables or information would be useful to have to answer the research question more fully?








4.








You are welcome to work with a partner – if you do, only one student should submit the answers with both students’ names on the documents.

















5.








Submit (1) do-file; (2) replication of Table 4 (*.rtf or *.xls). Include your answers to “Additional questions” in the *.rtf or *.doc file, or as a block comment at the top of the do-file. Filenames should have your name, e.g. “Bacolod_replication.do”

















Helpful hints:








·








The file “replication_promotion.dta” has the following variables:











o








mastid: unique index number of anonymized officer








o








promoted: dummy variable, with officer promoted=1 and otherwise=0








o








belowzone: dummy variable, with officer currently BZ=1 and otherwise=0








o








inzone: dummy variable, with officer currently IZ=1 and otherwise=0








o








abovezone: dummy variable, with officer currently AZ=1 and otherwise=0








o








female: dummy variable, with female=1 and male=0








o








nonwhite: dummy variable, with non-white=1 and white=0








o








married: dummy variable, with married=1 and single/divorced/widowed=0








o








morethan3kids: dummy variable, with 4+ kids=1 and 3 or fewer kids=0








o








priorenlisted: dummy variable, with ascension source via enlisted=1 and otherwise=0








o








md_jag: dummy variable, with doctor/lawyer=1 and otherwise=0








o








ma: dummy variable, with a MA/MS degree=1 and otherwise=0








o








LCDR: dummy variable, with current rank at LCDR=1 and otherwise=0








o








CDR: dummy variable, with current rank at CDR=1 and otherwise=0








o








totalsuffixcodes: Total # of suffix codes held (proxy for ability) year: calendar year











Answered 3 days AfterJul 15, 2023

Answer To: Replication Exercise #2Promotion Policy and Diff-in-diff Estimation1.Read Ahn, T, J...

Banasree answered on Jul 18 2023
28 Votes
Page | 2
Q2.Ans.
Q3.a)Ans.
The authors may have chosen to develop a linear probability model (LPM) rather than a logit or probit model for several reasons. The LPM provides a simple interpretation of the parameters as a small
effect on the probability of growth. It also allows easy comparison of coefficients across different specifications and models. However, the use of LPM also has some limitations. A key point is that LPMs assume frequent marginal effects, which may not hold in practice. This can lead to biased estimates and spurious standard errors. Another disadvantage is that the predicted probabilities from LPMs may fall in the distance [0, 1], which violates the probability limit. In contrast, the logit and probit models use a nonlinear variable in the form of a linear predictor to estimate the probability of controlling the increase. This model accounts for the limits of probability and allows for non-constant marginal effects. Logit and probit models also yield fairly standard errors, especially in dealing with two outcomes. The choice between LPM and logit/probit models depends on the specific research question, the nature of the data, and the appropriateness of assumptions for analysis. For binary outcomes such as growth measures, logit and probit models are often used, because they overcome limitations the application of LPM. However, the authors may have chosen LPM because of its simplicity, ease of interpretation and comparability in their particular research setting.
Q3.b)Ans.
The authors acknowledge that treating the In-Zone (IZ) cohort as the control cohort is technically inaccurate because promotions in the Navy are competitive, and fluctuations in promotion rates provide The Below Zone (BZ) and Above Zone (AZ) officers can affect the promotion opportunities available as IZ officers. However, they still chose to treat the IZ group as a control group for several reasons:
1. IZ police authority: IZ police make up most of the promoted officers (about 90 percent). Thus, changes in the growth rates of BZ and AZ adults will have only a minor effect on IZ growth rates.
2. Impact on IZ Promotion: The number of jobs fixed in promotions limits the number of officers who can be promoted. This suggests that changes in BZ and AZ...
SOLUTION.PDF

Answer To This Question Is Available To Download

Related Questions & Answers

More Questions »

Submit New Assignment

Copy and Paste Your Assignment Here