ECONOMETRICS IV (ADVANCED ECONOMETRICS)

http://www.nes.ru/~sanatoly/Econometrics4/Econometrics4.htm

1st Module, 2002 / 2003

Instructor: Stanislav Anatolyev

TAs:

This course is a continuation of Econometrics III. This time we will concentrate on estimation and inference in nonlinear models. Beside the Nonlinear Least Squares estimation that we have already studied, the methods will include Maximum Likelihood and Generalized Method of Moments, the two leading contemporary econometric techniques. Also, we will devote a few lectures to learn how to handle panel data.

ORGANIZATION

There will be six weekly homework assignments that account for 20% of the final grade. The assignment will contain both analytical problems and computer exercises. Solutions for computer exercises can be submitted one for a group of 2 or 3 students. The groups should be determined at the beginning and should not change during the module. Suggested solutions will be distributed. The Problems and Solutions manual contains additional problems for independent work and discussion in sections. The final exam, which accounts for 80% of the grade, will have an open-book format.

LITERATURE

· Anatolyev, S. (2002) Intermediate and Advanced Econometrics: Problems and Solutions, Sections 6–9, New Economic School (SA)

· Hayashi, F. (2000) Econometrics, Princeton University Press

· Goldberger, A. (1991) A Course in Econometrics, Harvard University Press (AG)

· Greene, W. (2000) Econometric Analysis, Prentice Hall (WG)

· Hamilton, J. (1994) Time Series Analysis, Princeton University Press (TS)

· Hall, A. (2000) Generalized Method of Moments, in Baltagi, B., ed., Companion to Theoretical Econometrics, Blackwell Publishers (AH)

· Newey, W., McFadden, D. (1994) Large Sample Estimation and Hypothesis Testing, in Handbook of Econometrics, vol. 4. (NM)

· Baltagi, B. (1995, 2001) Econometric Analysis of Panel Data, John Wiley & Sons (BB)

SYLLABUS

I. Extremum Estimators and Maximum Likelihood

1.       Extremum Estimators (SA 6)

·         Extremum estimators and their asymptotic properties. Identification conditions.

·         NLLS and nonlinear IV and 2SLS as extremum estimators.

2.       The Maximum Likelihood (ML) estimator (SA 7; AG 12; WG; TS 5; NM)

·         Likelihood function and likelihood principle. ML as an extremum estimator.

·         Consistency and asymptotic normality of ML estimators.

·         Asymptotic efficiency of ML estimators.

·         Conditional, joint and marginal ML. Concentrated ML.

·         ML asymptotic tests: Wald, Likelihood Ratio, Lagrange Multiplier.

·         Example: binary choice models.

·         ML and time series data.

II. Method of Moments

1.       The Classical Method of Moments (CMM) estimator (WG)

·         Moment restrictions and moment functions.

·         Just-identifying moment restrictions and CMM.

2.       Generalized Method of Moments (GMM) estimator (SA 8; AH; WG; TS 14; NM)

·         Overidentifying moment restrictions. The GMM optimization problem.

·         Asymptotic properties of GMM estimators.

·         Efficient GMM and feasible efficient GMM.

·         GMM asymptotic tests: Wald, Distance Difference, Lagrange Multiplier.

·         The test of overidentifying restrictions (the J-test).

·         Efficient GMM as optimal instrumental variables estimation.

·         OLS, IV, 2SLS and ML as special cases of GMM estimation.

·         GMM and time series data. GMM and Rational Expectations models.

·         Finite sample deficiencies of GMM estimators. Modifications of GMM estimators.

·         Bootstrapping GMM.

III. Panel Data

1.       Panel Data (SA 9; BB)

·         One-way and two-way error component models.

·         Random effects and fixed. Pooled OLS, GLS, LSDV, Within and Between estimators in a one-way error component model.

·         Dynamic panel regression.