Coursework I
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
Coursework I (submission by 15:00 March 8, 2024)
Instructions
1. Submit all your answers, including Stata code and output, in a single PDF file. Append the code and full outputs in an appendix. There is no strict format requirement, but it is impor- tant to make your answers easy to follow and understand. Include everything in a single PDF file.
2. The name of the document should be your candidate number, e.g., AB12345. pdf and noth- ing else.
3. The answer document should display the question text in full and the relevant answer un- derneath.
4. Questions 1 and 2 are each worth 25 points, and question 3 is worth 50 points.
Questions
1. In the paper “Partisan Shocks and Financial Markets: Evidence from Close National Elec- tions” by Daniele Girardi (available on Keats), an RDD design is used to test the effect of left-wing election victories on stock market performance.
Answer the following (max 800 words):
• Explain the design that was used.
• What is the variable that shows clear discontinuity around the threshold?
• What are the underlying assumptions of such a design?
• What could invalidate them in the context of the paper?
2. The Abowd-Kramarz-Margolis (AKM) Model (paper available on Keats) is aseminal model in labor economics, describing the dynamics of wages among workers. It can be represented as follows:
yijt = α + βXijt + ψj + ϕi + ϵijt ,
where:
• yijt is the log of wages for worker i in firm j at time t.
• α is the intercept term.
• Xijt represents observable characteristics of the worker-firm match (e.g., worker’sex- perience, education) with β as the vector of coefficients.
• ψj represents firm fixed effects, capturing unobserved firm-specific wage determi- nants.
• ϕi represents worker fixedeffects, capturing unobserved worker-specificcharacteris- tics that affect wages.
• ϵijt is the idiosyncratic error term, assumed to be normally distributed.
Answer the following (max 800 words):
• How does the application of fixedeffects in the model address the issue of unobserved heterogeneity in the estimation of wage determinants?
• What are the assumptions we are making when estimating simultaneously worker fixedeffects and firm fixedeffects?
• Suppose that βXijt = 0, i.e., that no observable characteristics are available, or that they have no effect on wages. What determines in this model the dynamics of an individual worker?
3. Excerptfromthedatausedinthepaper“Efficientsemiparametricestimationofmulti-valued treatment effectsunder ignorability” by Matias Cattaneo (available on Keats) can be directly imported into Stata using:
use https://www. stata-press. com/data/r17/cattaneo2.
This is a cross-sectional dataset, which includes the birthweights of babies (in the variable bweight) along with multiple characteristics of their mothers and fathers. The description of each variable is included in the dataset when importing it.
Your goal is to estimate two effects: the effect of mother’s alcohol consumption during preg- nancy on a child’s birthweight; the effect of mother’s smoking on a child’s birthweight.
• Describe the empirical design you choose to employ – what are your considerations when choosing this design? What are the limitations of this design?
• Present your results and interpret them (please include your code and output regres- sion tables as an appendix).
• If you chose a design different from OLS, compare your results to the results of OLS regression with appropriate controls. Is there a significant difference? Why or why not?
• Use your original design to test the effect of both consuming alcohol and smoking dur- ing pregnancy on a child’s birthweight. Present your results and interpret them (please include your code and output regression tables as an appendix).
2024-03-02