Institutions and Economic Development - Foundations of Development Policy - Lecture Notes, Study notes of Development Economics

Download Institutions and Economic Development - Foundations of Development Policy - Lecture Notes and more Study notes Development Economics in PDF only on Docsity! 1 14.74 Colonial history, institutions and economic development The paper “The Colonial Origins of Comparative development: An Empirical Investigation,” by Daron Acemoglu, Simon Johnson and James A. Robinson seeks to answer a very important question: do good institutions matter for development? Through the course, we have seen several channels where institutions should matter: • • • • This paper tries to answer a more “reduced form” question: without getting into all these details, if we give Zaire the same institutions as the US, how much richer would Zaire be? The question and the identification problem If we think of the quality of institution as our “treatment” (as in the quality of schooling), we are trying to measure the effect of the treatment on an outcome: log GDP of the country. Mathematically, we could write it in the following way: log yi = µ + αRi + Xiγ + �i where: log yi = log(GDP ) of country i in year 1995 1 Docsity.com Ri = quality of institutions, measured by the “expropriation risk” which we discussed last time. Xi= control variables (e.g. latitude) If we run the OLS regression, corresponding to this regression, we find α = 0.52 standard error 0.06 Look at the graph very strong correlation → What does it indicate? • • What are the problems with this interpretation? – – – All these problems create a correlation between Ri and �i which biases OLS. → 2 Using instrumental variables to solve the identification problem We cannot randomly assign institutions! look for a variable which predicts current → institutions but does not predict directly current GDP per capita. Call this variable Z. How is Z called? Refresher: method to recover an unbiased estimate of α using Z: 1. Run the regression: Ri = λR + βRZi + XiγR + VRi if Zi affects Ri, we expect βR to be significant (otherwise, this is not a good instru­ ment!) What is the name of this regression? Form the variable: R̂i = λ̂R + β̂RZi + Xiγ̂R 2 Docsity.com – Data on current institutions. Results – βs = – βc = (using log(Mi) as an instrument) – βR = (using log(Mi) as an instrument) – Table 4: Main results ∗ First stage: Ri = λR + βR log(Mi) + XiγR + ViR βR = ∗ Second stage equation α = – What are the potential problems with the strategy: ∗ ∗ Conclusion: if we gave Zaire the institutions of the US, Zaire would be 5 times richer! 5 A more specific look at the question: The impact of Colo­ nial Institutions in India Examines the long term impact of institutions in India. Most of India was colonized by the British, but they used different types of technology to collect land taxes (the major source of revenue for the British): 1. The landord, or zamindari system asked a landlord to collect taxes on behalf of the British (have you seen Lagaan?). Landlords could collect as much as they wanted and any extra was theirs to keep. 5 Docsity.com 2. The individual responsibility, or raiyatwari system asked the farmers to arrange be­ tween themselves to collect the taxes. There also existed a village-based system quite similar. There are many reasons why a system put in place during a colonial period may persist long after independence. - - - - The British’s idea about what system was better evolved over time, so that different places in India who were colonized at different times have different institutions, even within current states. See Map. See Figure 4: dates of conquest determines in part time of governance: places colonized between 1820 and 1856 more likely to have a non-landlord system. Still correlated with yield today!! Can look at OLS (compare non-landlord/landlord area) and IV, using as instrument the date of conquest. Conclusion: Even today, yields are higher in non-landlord districts. Very persistent effects of long-run institutions. 6 Docsity.com