ECONOMICS 8750

ECONOMETRICS

Spring Semester, 2003

Instructor: Mary Beth Walker, 643 RCB, 404-651-3751,

e-mail: mbwalker@gsu.edu

Time and location: Tuesday, Thursday, 11:00-12:15 p.m., Room 233 Aderhold Building.

Office hours: Monday, 10:00-12:00.

Wednesday, 1:00 – 2:30.

Objectives Our aim is to develop familiarity with a wide variety of

linear statistical techniques which are routinely used in the analysis of

economic data. Where appropriate, these techniques will be placed on a

strong theoretical basis. Primary emphasis will be placed on applications,

both in terms of analysis of empirical results and hands-on computer

assignments. As a byproduct, we hope to develop certain theoretical and

computing skills which will facilitate the mastery of additional techniques

in the future.

Textbook: The basic text for the course is Badi Baltagi,

Econometrics, 2nd revised edition, Springer-Verlag, 1999. Excellent

supplements and reference books are Davidson and MacKinnon, estimation

and inference in econometrics, Oxford Press, 1993, and William Greene,

Econometric Analysis, fourth edition, MacMillan, 1998. There may be a fifth edition of Greene by now.

Computer Programs: For problem sets you may use either STATA or SAS or Gauss. Gauss is a matrix programming language.

Grading: There will be problem sets, a midterm exam and a final.

They will count toward the grade as follows.

Assignments: 25 %

Midterm: 35 %

Final: 40 %

Exam (and other) dates: The midterm exam will be on Thursday, February 20.

The final exam is on May 1, 12:30 p.m. This

semester's drop date is March 10. This is the last day you can drop and

possibly receive a ''W.''

All University rules regarding drop dates and hardship withdrawals will be adhered to. New University rules regarding class attendance will require me to give ‘WF’ grades to any student whose name appears on my class roll but who quits coming to class. After the semester midpoint, anyone who quits coming to class, but does not formally drop the course must be given the ‘WF’ grade. In order to ensure that this grade is not given in error, attendance will be taken weekly.

Prerequisites: Working knowledge of calculus, linear algebra, and

mathematical statistics is necessary. You must know matrix algebra to take

this course! Basic computer skills will be needed to complete some problem

sets.

Outline and Reading List:

Note: some supplemental readings have also been listed. Required readings

are starred (*).

Week 1: Basic concepts

Notation

Introduction to least squares

Readings: *Baltagi, 2 (if necessary) and 3

Week 2: Classical linear model

Bivariate regression

Prediction in the bivariate model

Readings: *Baltagi, 3.

Week 3: Classical linear model

Multivariate regression

Algebraic and statistical results

Readings: *Baltagi, 4.1-4.5, 7.1-7.4.

Week 4: Classical linear model

Restricted estimation

Hypothesis testing

Readings: *Baltagi, 4.6,7.6-7.9.

Week 5: Classical linear model

Hypothesis testing

Prediction

Readings: *Baltagi 7.5, 4.7.

Week 6: Asymptotic theory

Convergence in probability

Consistency

Readings: Greene, 4.4.

Week 7: midterm and more asymptotic theory

Convergence in distribution

Maximum likelihood estimation

Maximum likelihood testing

Readings: Greene 4, parts of Chapter 9. White, Asymptotic Theory for

Econometricians, (Academic Press, 1975), 1-4, Davidson and MacKinnon,

4.

When ideal conditions are violated

Week 8: Multicollinearity

Week 9: Stochastic regressors and instrumental variables

Week 10: Nonscalar covariances

Week 11: Heteroskedasticity

Week 12: Serial correlation

Week 13: Misspecification testing

Readings: *Baltagi, chapters 5, 8, and 9. *White, H., ''A

Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test

of Heteroskedasticity,'' Econometrica 48, 1980.

Weeks 14 and 15: Extensions of the classical linear model

Systems of regression equations

Catch up on everything

Readings: *Baltagi, 10.

If we can fit it in: Dichotomous dependent variable models

Linear probability model

Logit models

Probit models

Readings: *Baltagi, 13.1-13.7.