Question 2 (80 marka) The government is worried about gender bias in hiring by private firms in the manufacturing sector. To address this concem, it passes a new law at the beginning of 2010 that requires finns In manufacturing to report any discrimination complaints by employees to the govemment. Prior to the law, employees might file complaints to the firm but firms could choose whether to report these to the government or not. The new taw does not apply to firms in the service sector or to government agencies (public sector employees). The government asks a team of researchers to study the effect of this law on discrimination complaints. a. Researcher A uses an OLS regression using data from 2011. It relates the number of complaints to a treatment dummy for the new lawwhich is equal to 1 for manufacturing firms.. Discuss the conditions under which this regression would reveal the causal effect of the law, and briefly comment on whether these conditions would be satistied. [25 marks] b. Researcher B proposes to study whether this regulation was effective by comparing the number of complaints in the manufacturing sector before and after the regulation. She gathers data on complaints for two years before the law. 2008 and 2009, and two years after, 2011 and 2012. Discuss how she could evaluate whether there is a significant difference in the number of complaints. . Explain whether this approach would yield the causal effect of the new law. [25 marks] c. Her colleague suggests using a difference in differences methodology. exploiting information also from the public sector and the service sector. Write down the kind of regression that you would use for this approach and explain which coefficient, in such a regression, captures the causal effect of the new law. What are the identifying assumptions required for such an analysis? Do you think that this approach is preferable to the approach in part b? Explain why. (30 marks] Question 2 (80 marka) The government is worried about gender bias in hiring by private firms in the manufacturing sector. To address this concem, it passes a new law at the beginning of 2010 that requires finns In manufacturing to report any discrimination complaints by employees to the govemment. Prior to the law, employees might file complaints to the firm but firms could choose whether to report these to the government or not. The new taw does not apply to firms in the service sector or to government agencies (public sector employees). The government asks a team of researchers to study the effect of this law on discrimination complaints. a. Researcher A uses an OLS regression using data from 2011. It relates the number of complaints to a treatment dummy for the new lawwhich is equal to 1 for manufacturing firms.. Discuss the conditions under which this regression would reveal the causal effect of the law, and briefly comment on whether these conditions would be satistied. [25 marks] b. Researcher B proposes to study whether this regulation was effective by comparing the number of complaints in the manufacturing sector before and after the regulation. She gathers data on complaints for two years before the law. 2008 and 2009, and two years after, 2011 and 2012. Discuss how she could evaluate whether there is a significant difference in the number of complaints. . Explain whether this approach would yield the causal effect of the new law. [25 marks] c. Her colleague suggests using a difference in differences methodology. exploiting information also from the public sector and the service sector. Write down the kind of regression that you would use for this approach and explain which coefficient, in such a regression, captures the causal effect of the new law. What are the identifying assumptions required for such an analysis? Do you think that this approach is preferable to the approach in part b? Explain why. (30 marks]