5. Good institutions facilitate innovation and economic production, so they are often considered a key determinant of economic development. In Acemoglu et al (2001), as discussed in class, the authors use historical cross-sectional data to examine this question. First, consider a naive OLS regression as follows: Iny =utaR, +yX, +6 where colonies are indexed by i, y, denotes per capita income in 1995, R, represents average protection against expropriation risk in 1985-1995, and X is a vector of country covariates. Iny =utaRi+yXi + 6 (i) Why does R represent institutions? What kind of institutions? Why do we think it would have a positive causal relationship with y,? Elaborate. [2 marks] (ii) Write down the strict exogeneity assumption that is necessary for consistent estimation of the coefficient a. Propose two possible ways in which the above assumption would be violated and, in each case, explain the expected direction of bias. [3 marks] Using settler mortality M as an instrument for Ry, they use an instrumental variable strategy to overcome possible identification issues above. Nonetheless, for this to be valid, M, must satisfy the relevance condition and the exclusion restriction. (iii) Describe two possible reasons why the exclusion restriction may not hold. Make sure you distinguish these reasons from the ones you proposed in part (ii). [2 marks] Table 2 (below) presents results from instrumental variables regressions (Panels A and B) and ordinary least squares regressions (Panel C) in Acemoglu et al (2001). Note: Average protection against expropriation risk is an index that takes values between 0 and 10, where a higher score represents better institutional quality.Table 2 - Institutions and Economic Performance Panel A: Instrumental variables, second stage Dependent variable: In(per capita GDP) (1) (2) (3) Average protection against expropriation risk 1.10 1.16 1.20 [0.22] [0.34] [0.29] Latitude -0.75 -1,10 (1.70] [1.56] British colonial dummy -0.78 -0.80 [0.35] French colonial dummy -0.12 -0.06 [0.05] [0.02] French legal origin dummy D.96 [0.39] Constant term 1.37 1.00 -0.03 [1.39] [1.96] [1.88] Panel B: Instrumental variables, first stage Dependent variable: Average protection against expropriation risk In(settler mortality) -0.53 -0.43 -0.44 [0.14] [0.16] [0.14] Latitude 1.97 2.10 [1.40] [1.30] British colonial dummy 0.63 0.55 [0.32] [0.27] French colonial dummy 0.05 -0.12 [0.02] French legal origin dummy -0.70 [0.32] Constant term 8.75 7.96 3.63 [0.69] [0.88] [0.79] Panel C: Ordinary least squares Dependent variable: In(per capita GDP) Average protection against expropriation risk 0.53 0.47 0.56 [0.19] [0.07] [0.06] Number of observations 64 64 64 Note: Standard errors in parentheses. These results are my own calculations based on the data used by Acemoglu et al (2001). The omitted category for colonial identity is "other", which includes for example Spanish and Portuguese. The omitted category for legal origin is British common law.Refer to Table 2, column (1) to answer parts (iv)-(v). (iv) According to the instrumental variables result, is the effect of institutions on economic performance positive? Explain. [1 mark] (v) The average protection against expropriation risk scores are 7.59 and 9.73, for Indonesia and New Zealand respectively. Hypothetically, how much would Indonesia's per capita GDP increase if it could improve its institutions to the level of New Zealand's? [2 marks] (vi) Drawing empirical evidence from both papers - Swee & Panza (2016) and Acemoglu et al (2001) - can we conclude that "good geography" necessarily leads to "good institutions"? What common lesson(s) can we draw from these papers? Carefully explain your arguments. [3 marks]