Homework 6

Author
Affiliation

Alex Cardazzi

Old Dominion University

All materials can be found at alexcardazzi.github.io.

Completion Requirements: Complete the following questions in RStudio via the homework template. When you are ready, submit your rendered html to Canvas.

Grading Criteria: Full credit will be given to correct, well formatted, and detailed answers. Partial credit will be given if I can follow your work and/or see your thought process via code, comments, and text. Point totals are listed next to each question.

This homework is based on the paper titled “Slum Clearance and Urban Renewal in the United States” by Collins and Shester (2013) published in the American Economic Journal: Applied Economics.

The data used can be found here. It is recommended that you read the paper, which can be found on Canvas, before completing this homework.

Question 1

  1. What is the research question this paper is answering? What is its significance? (1 Point)

  2. Explain the identification strategy used in this paper. (1 Point)

  1. Draw a (simplified) DAG for this setting, or write down the first and second stage regressions used in the IV estimation in the notation used in the notes. (1 Point)

Question 2

Use these data to complete this homework.

  1. Read the data into R as an object called city. (1 Point)

  2. Plot the average per-capita funding by years of potential participation. Does this figure support the authors candidate for an instrument? (1 Point)

Question 3

  1. Estimate the effect of per-capita funds on the different outcomes in Table 3 using naive/endogenous OLS. To make it a bit easier, here are the control variables used in the paper: owner_occ_pct50 + log(med_prop_val50) + dilap_pct50 + units_old_pct50 + no_plumbing_pct50 + crowded_units_pct50 + log(pop50) + nonwhite_pct50 + manuf_pct50 + emp_rate50 + median_school50 + log(med_fam_inc50) + income_2k_pct50. (1 Point)

  2. Tabulate the regression coefficients in a single table using functions from the modelsummary package. Do not display the coefficients for the control variables. (1 Point)

  3. Interpret the primary coefficients of interest as if they were causal. (1 Point)

  4. What argument(s) do the authors make for why we cannot interpret these estimates as causal? (1 Point)

  5. Could the author(s) have estimated a difference-in-differences? Why or why not (aside from the data not being in a panel format)? (1 Point)

Question 4

  1. Estimate the reduced form relationship between the instrument and the previous outcomes (conditional on the control variables from the previous question). These estimates should match the estimates on the first row of Table 2. Present the estimates in a table.

  2. Interpret these coefficients of interest as if they were causal. (1 Point)

  3. Can these estimates be interpreted as causal? If no, why not? If yes, then why can’t we just stop here? (1 Point)

Question 5

  1. Estimate the effect of per-capita funds on the different outcomes in Table 3 using IV. Display the coefficients in a table using modelsummary. (1 Point)

  2. Interpret these coefficients. (1 Point)

  3. How do these coefficients compare to those from the naive regression? (2 Points)

  4. Use fitstat() to display the F-statistic and the Wald-statistic. What are the general takeaways from these tests? (1 Point)

  5. Discuss the assumptions necessary for a causal interpretation of these results and whether you believe they have been satisfied. Use quantitative evidence where possible. (2 Points)