Bertrand & Mullainathan (2004) study race in the labor market by sending fictitious re- sumes to help-wanted ads in Boston and Chicago newspapers. To manipulate perceived race, resumes are randomly assigned African-American- or White-sounding names (resumes are exactly the same except for the name). (a) In the absence of this experiment, would simply comparing means between african- americans and whites have uncovered the impact of race on call-back rates? Why/Why not? (b) Why even after randomly assigning resumes to different help wanted ads one may want to use a regression model. Write down the model that would test the hypothesis that call- back rates between african-americans and whites is the same. (c) What are the main advantages of using linear regression models in this case? (d) What are the main challenges of RCTs ? Are you concerned about these issues in this case? Question 4 Pinotti (2017) studies the effect of immigrant legalization on the crime rate of immigrants in Italy. Author uses fixed quotas of residence permits that are available each year, applications are submitted electronically on specific days, and are processed on a first come, first served basis until the available quotas are exhausted. Crucial to this study, applicants do not know when quotas are exhausted. For example, if there are 4000 permits and 10,000 applicants, all 10,000 applications are ranked in chronological order with the first 4,000 being granted residence permits and the 4001th not receiving it. (a) Do you think that comparing crime rates of immigrants who were granted residence per- mits with those who dont will uncover the causal effects of residence permits on crime? Why? (b) Using a regression disconinuity design, how can you estimate the impact of residence permits on crime? Make sure to write down a model.(Hint: Think about who narrowly gets the permit and who narrowly misses it) (c) If immigrants can manipulate the visa permit process shall we be concerned about internal validity? Assume that immigrants more determined to work and less likely to commit crimes bribe authorities to get visa permits, is our estimator underestimating or overestimating the effect of the policy?

Bertrand \& Mullainathan (2004) study race in the labor market by sending fictitious resumes to help-wanted ads in Boston and Chicago newspapers. To manipulate perceived race, resumes are randomly assigned African-American- or White-sounding names (resumes are exactly the same except for the name). (a) In the absence of this experiment, would simply comparing means between africanamericans and whites have uncovered the impact of race on call-back rates? Why/Why not? (b) Why even after randomly assigning resumes to different help wanted ads one may want to use a regression model. Write down the model that would test the hypothesis that callback rates between african-americans and whites is the same. (c) What are the main advantages of using linear regression models in this case? (d) What are the main challenges of RCT's ? Are you concerned about these issues in this case