Course Meeting Times
Lectures: 2 sessions / week, 1.5 hours / session
Recitations: 1 session / week, 1.5 hours / session
This course presents micro-econometric models, including large sample theory for estimation and hypothesis testing, generalized method of moments (GMM), estimation of censored and truncated specifications, quantile regression, structural estimation, nonparametric and semiparametric estimation, treatment effects, panel data, bootstrapping, simulation methods, and Bayesian methods. The methods are illustrated with economic applications.
The first half of the course (Part A) is taught by Prof. Chernozhukov, while the second half (Part B) is taught by Prof. Newey.
The prerequisites include 14.382 Econometrics or permission of the instructor.
Please see the readings.
The grades for each part count equally towards the final grade.
Part A – Chernozhukov
Part B – Newey
The following three articles are referenced in the Recitation topics. Articles that pertain to lecture topics can be found in the readings.
Ahn, H., and J. L. Powell. “Semiparametric Estimation of Censored Selection Models with a Nonparametric Selection Mechanism.” Journal of Econometrics 58 (1993): 3-29.
Autor, D., L. F. Katz, and M. S. Kearney. “Rising Wage Inequality: The Role of Composition and Prices.” National Bureau of Economic Research (NBER) Working Paper No. 11628 (August 2005): 1-65.
Chernozhukov, V., and H. Hong. “Three-step Censored Quantile Regression and Extramarital Affairs.” Journal of the American Statistical Association 97, no. 459 (September 2002): 872-882.
The calendar below provides information on the course’s lecture (L) and Recitation (R) sessions. Part A consists of sessions L1-L12, while Part B consists of sessions L13-L25.
|SES #||TOPICS||KEY DATES|
|L1||Methods for nonlinear models: maximum likelihood estimation (MLE), generalized method of moments (GMM), minimum distance, extremum||Problem set A-1 out two days after Ses #L1|
|L2-L3||Large sample theory, asymptotic theory, discrete choice, censoring, and sample selection|
|R1||Extremum estimators, variance estimation, hypothesis tests|
|L4-L5||Large sample theory, asymptotic theory, discrete choice, censoring, and sample selection (cont.)|
|R2||ML computation: probit using ordinary least squares (OLS) command, hypothesis tests; two-step estimation: Heckman correction||
Problem set A-1 due
Problem set A-2 out
|L6||Bootstrap, subsampling, and finite-sample methods|
|L7||Bootstrap, subsampling, and finite-sample methods (cont.)|
|L8||Quantile regression (QR) and distributional methods|
|R4||Quantile regression: integral transformation/Skorohod representation, conditional means vs. conditional quantiles, inference for quantile regression, and high-tech application: wage decompositions (Autor, Katz, and Kearney 2005)||Problem set A-2 due|
|L9||Quantile regression (QR) and distributional methods (cont.)|
|R5||QR applications: 3-step procedure for censored QR (Chernozhukov and Hong 2002); digression: duration models; brief introduction to R, wage decomposition|
|L10-L11||Bayesian and quasi-Bayesian methods (from a classical view)|
|R6||Accept-reject sampling, the Gibbs sampler, and Monte Carlo optimization|
|L12||Bounds and partial identification||
Problem set A-3 out
Midterm exam two days after Ses #L12
|L13-L14||GMM: identification, estimation, testing, bias, selecting moments||Problem set B-1 out on Ses #L14|
|R7||Duration models: main concepts, practical issues; GMM: higher-order bias for two stage least squares (2SLS) estimation, adding moments and efficiency|
|L15||Weak and many instruments|
|L16||Nonparametric estimation||Problem set A-3 due|
|R8||Nonparametric regression: theoretical bias and variance of the Nadaraya-Watson estimator, confidence intervals, bandwidth choice: cross-validation in kernel regression|
|L17||Nonparametric estimation (cont.)||
Problem set B-1 due
Problem set B-2 out one day after Ses #L17
|R9||GMM with condition moment restriction: optimal IV vs. efficient weighting matrix, example from Problem set B-1; nonparametric regression: kernel regression asymptotics, local linear estimation, bandwidth selection, generalized cross-validation|
|L21||Nonlinear models in panel data||Problem set B-2 due|
|R10||Series estimation and discontinuities, an example for the partially linear model: semiparametric selection models (Ahn and Powell 1993); treatment effects: the LaLonde debate||Problem set B-3 out|
|L22||Nonlinear models in panel data (cont.)|
|L23||Economic modeling and econometrics|
|R11||Nonlinear panel data: incidental parameters problem, conditional MLE: Logit case; method of simulated moments (MSM): brief introduction to numerical integration, simulated estimation|
|L24-L25||Economic modeling and econometrics (cont.)||
Problem set B-3 due
Final exam seven days after Ses #L25