Seminar 1 (week 5)

I. Empirical methods

These questions are based on the methodology readings by Finseraas and Kotsadam, but almost any introduction to applied econometrics could also work.

1. We want to study how income affects savings, i.e. the marginal propensity to consume, in a sample of poor farmers. Discuss whether simply regressing savings on income would yield correct estimates?  Specify what you mean by "correct"
   One solution to the problems discussed above is to use instrumental variables. One suggested rainfall for the income of poor farmers is rainfall during the growing period (Paxson 1992). Explain what instrumental variables is, why rainfall could help, and how you would go forth to estimate the IV.
2. We want to study whether female politicians behave differently when male politicians when it comes to spending on schools. Explain why simply comparing districts controlled by female politicians to districts controlled by male politicians (or studying the share of female politicians) would yield biased results.
   If we use data from e.g. India, politicians are elected by majority voting, i.e. one representative from each voting district. Typically there are only two candidates running, so to get elected is it sufficiently to get exactly above 50 % of the votes. Explain what a regression discontinuity design is and how the threshold at 50 % of the vote can be used in a regression discontinuity design (see Clots-Figueras (2012) for details).

II. Global inequality

The Penn World Tables provide PPP-adjusted GDP numbers for a number of countries. The file pwt contains data on GDP per capita and population in Stata format and as a  tab separated text file. You should download one of these files an place it somewhere in your personal directory.
To analyze inequality I prefer to use inequal7 in Stata available from SSC (type ssc install inequal7 to install), but feel free to use any package that computes Gini coefficients and handles weighted data.

1. Compute inequality between countries using the Gini coefficient for 1990 and 2010. Did inequality go up or down?
2. As discussed in class, there may be arguments for weighing these estimates by population size. Repeat the calculation, weighing by population. What do you find in this case? Compare with your findings in a. What is the role of India and China?
3. Discuss what is missing to say something about global inequality, and discuss approaches to estimating this.