Applied Econometrics for Practitioners
World Bank, Kathmandu, Nepal
Lectures: Monday, Wednesday and Fridays: 8-10AM
Instructor: Dean Jolliffe
Teaching Assistant: Ganesh Thapa
This course is designed to teach the tools and techniques of applied statistics and empirical microeconomics. The class focus is on cross-sectional econometric analysis, with some discussion of panel data analysis (there is essentially no coverage of time-series econometrics). The class will focus on the issues related to international development and domestic social issues (e.g. poverty, education, health, agriculture, labor issues) of Nepal, and will rely on exercises and examples from the Nepal Living Standard Survey (NLSS 2011) data. This course covers some advanced, but commonly needed, methods of statistical analysis. The emphasis is on applications, not derivations, of methodology. The course will entail a significant amount of data analysis with the Stata software (version 12). Topics covered include: Review of multivariate regression and violations of model assumptions, complex sample design – stratification and multi-stage selection, proper use of sample weights and population inference from sample statistics, instrumental variables estimation, limited dependent variables (Linear Probability, Logit & Probit models), censored dependent variables (Poisson & Tobit models), quantile regression and robust estimation, introduction to the bootstrap, introduction to fixed effects modeling with cross-sectional and panel data.
Prerequisites: Basic Stata skills, or an aptitude for learning software from manuals or online help and exposure to the statistics/microeconomics.
Texts and Materials: Wooldridge, Jeffrey M. Introductory Econometrics: A Modern Approach, Fifth Edition.
Optional/Supplemental: Deaton, Angus. The Analysis of Household Surveys: A Microeconometric Approach to Development Policy. Baltimore, MD: Johns Hopkins University.
Course Requirements and Rules
Upon successful completion of 4 problem sets and active class participation, we will provide you with a certificate of course completion. The write-up for the problem set should include the essential program code and output when appropriate. Do not hand in entire log files, but do include enough of the Stata code to show the core commands used along with other relevant information (e.g. whether the results are weighted, clustered, stratified). Problem sets are due at the beginning of class on the due date, to be announced at the beginning of the course. Late problem sets are not accepted because we may discuss them in class on the due date.
The First Phase
1. December 10: Course Overview, Introduction and Review of OLS
Review OLS assumptions with a focus on how they affect our
ability to answer policy questions. Discuss bivariate OLS
estimator, the variance of this estimator, and the implications
for assessing data quality.
Wooldridge chapters 2, 3, 4 & 7;
NPR story on anti-smoking campaign. Reply by Steven Landsburg,
Slate (2003). Kristof editorial on infant mortality rates ("Health
Care: Ask Cuba").
2. December 12: A. Brief discussion of NPR story on smoking, Kristof editorial, NYT story on HIV estimate.
B. Begin discussion of Sample Design and Weights
NYT story on HIV overestimate.
Deaton chapters 1 & 2
3. December 14: Sample Design and Weights (cont.)
Discuss sampling, sample frame & inference, sample design &
Mortality before and after the 2003 invasion of Iraq: cluster sample
survey,” Lancet, 2004. Responses from Slate and the Economist. “
Mortality after the 2003 invasion of Iraq: a cross-sectional cluster
sample survey,” Lancet, 2006.
4. December 17: A. Discuss Lancet article and responses.
B. Data Problems: Measurement Error
Iron Law of Econometrics
ME: Wooldridge 9.3 & 9.4,
5. December 19: A. Data Problems: Non-response bias, Missing Observations, Heteroscedasticity
B. Binary Dependent Variable (Linear Probability, Logit,
Wooldridge 8.1, 8.2, 8.3; 7.5 & 17.1
The Second Phase
6. January 7: Binary Dependent Variable (cont.)
7. January 9: Instrumental Variable (IV) estimation & Simultaneous Equations Model (SEM)
Simultaneity bias. Identification and simultaneous equations.
IV: Wooldridge 15.1 – 15.3, 15.5—15.6; SEM: Wooldridge 16.1 –
“The Effect of Cocaine Prices on Crime” (Desimone, 2001).
“Landing on all fours: Communist elites in post-Soviet
Russia” (Geishecker & DeNew, Journal of Comparative Economics,
“Health Economics and Applications in Developing Countries”
(T. Paul Schultz, Journal of Health Economics, 2004,
8. January 11: A. Discuss Desimone & Geishecker papers, comment on Manski & Schultz
B. Sample-selection Bias
Heckman correction: benefits, problems and alternatives.
Wooldridge section 17.5,
Identification Problems in the Social Sciences and Everyday
Life” (Manski, 2003).
“The Ambulance-Homicide Theory” (NYTimes Magazine, 2002).
"Sample Selection in the Estimation of Air Bag and Seat
Belt Effectiveness” (Levitt & Porter, 2001).
9. January 13: A. Discuss Levitt & Porter
B. Poisson & Tobit Estimators
Estimation when the dependent variable is censored at zero.
Wooldridge 17.2 & 17.3
10. January 15: Robust Estimation, Least Absolute Deviations, Bootstrap Estimation
Outliers, Violations of normality
Wooldridge 9.4, Efron and Tibshirani, An Introduction to the
Bootstrap, 1993: Chapters 1, 2, 4 & 6.
11. January 18: Robust Estimation, Least Absolute Deviations,
Bootstrap Estimation (cont.)
12. January 20: Fixed Effects Estimation
Fixed-effects estimation using cross-sectional data. Fixed
effects estimation using panel data. Benefits of panel data.
Compare with independently sampled data from two time
Wooldridge chapter 13
In addition to the lectures, there will be 4 to 5 TA sessions to assist the participants on the problem sets (dates are to be determined in consultation with the participants). Below are tentative topics for the TA sessions:
1. Introduction to Stata – basic tips on using Stata, loading the NLSS data
2. More Stata tips, guidance for problem set 1
3. Review problem set 1, guidance for problem set 2
4. Review problem set 2, guidance for problem set 3
5. Review problem set 3, guidance for problem set 4
Last Updated: Nov 23, 2018