5. Single equation methods for estimating structural systems

A. Indirect least squares
B. Two-stage least squares
C. Limited information maximum likelihood

6. Multiple equation methods for estimating structural systems

A. Three-stage least squares
B. Full-information maximum likelihood

7. Strategies for estimating large structural systems

A. Recursive system estimation
B. Block recursive systems

8. Nonlinear methods for estimating structural systems

9. Generalized Method of Moments (GMM) estimation

Estimation exercise:

During this section of the course, a major estimation exercise will be required to give each student experience with each basic structural estimator. Students can benefit from discussing together how to apply the various estimators. However, each student is expected to develop and estimate his/her own structural systems. That is, each student should do his/her own specification and estimation independently. Estimation must be done using Matlab because of specific assignments about calculation and verification of estimators. Since a short course was offered to familiarize students with Matlab, no class time will be spent on learning Matlab software.

Part III-- Kirwan

(Mar 24-Apr 9, 2010)

Office hours: Tuesday and Thursday 2-3pm, Room 2120

Recommended Texts: Cameron, A.C. and P.K. Trivedi. Microeconometrics. Cambridge UP. New York, NY. 2005.

Angrist, J.D. and J-S Pischke. Mostly Harmless Econometrics: An Empiricist’s Companion. Princeton UP. Princeton, NJ. 2009.

Additional Readings: Reading list to be distributed on March 12.

1. Regression and the Conditional Expectation Function

2. Instrumental Variables

A. Wald Estimator
B. Constant Effects
C. Heterogeneous Effects – Local Average Treatment Effects

3. Problems with IV

A. 2SLS Bias
B. Clustering

4. Special Topics: Regression Discontinuity; Matching

5. Panel Data Models

A. Random Effects Models
B. Fixed Effects Models
C. Hausman Test
A. Special Topic: Difference-in-Differences

Part IV--Alberini

(April 12-May 10, 2010)

Suggested Topics

1. Recap of Panel data models (if needed; possible topics include a) Random effects models and GLS estimation; b) Fixed effects models and the ―within‖ estimator, c) the Hausman test, d) the Hausman-Taylor and related estimation procedures, e) dynamic panel data models: difficulties and estimation procedures (instrumental variable plus GLS, bias-corrected LSDV, Arellano-Bond and similar techniques) (based on Cheng Hsiao (2003), Analysis of Panel Data, Cambridge, UK: Cambridge University Press, and other materials that will be distributed to the class)

2. Censored and Truncated Models, including the Tobit Model (Greene, Section 22.3)

3. Binary Data Models: Probit and Logit Models (Greene, Ch. 21).

A. The probit (logit) equations

B. Derivation of the likelihood function, first-order conditions and covariance matrix of the estimates (Greene 21.1—21.4)

C. Asymptotic properties of the ML probit (logit) estimates (Greene 21.1—21.4)

D. Interpretation of coefficients, elasticities and marginal effects, and goodness of fit measures

E. Applications: (a) models of participation in programs, derivation of the estimation curve and of its confidence intervals using the delta method; (b) dichotomous choice contingent valuation: how to recover mean/median WTP and compute its standard error using the delta method (Cameron and James, 1987)

F. The fixed effects logit model and the random effects probit model (Greene, 21.5)

G. The bivariate probit model

4. Discrete/continuous models

A. Heckman two-step estimation and maximum likelihod estimation (lecture notes and Wooldridge, Ch. 17, page 560-571). Use of this model to address self-selection into a sample, an endogenous dummy in the RHS of a regression model.

B. Testing for endogeneity of regressors in a discrete/continuous model (Rivers and Vuong, 1988)

C. Switching regression models with endogenous switching

D. Using the technique of propensity score matching in quasi experiments (Wooldridge, Ch. 18, page 614-642)

5. Models for Count Data (Greene, Ch. 21.9):

A. the Poisson model: ML estimation and properties of the ML estimates

B. the problem of overdispersion, and how to address it using (a) an explicit overdispersion parameter, or (b) the negative binomial model

6. Discrete Choice Models (Greene, section 21.7):

A. The multinomial logit model and the conditional logit model: Assumptions, interpretation of coefficients, elasticities and marginal effects, goodness of fit measures, and a Hausman test for IIA

B. The nested logit model

C. The mixed logit model: comparison of estimated coefficients, predicted probabilities and elasticities from mixed logit and conditional logit (Train, 1998, 1999)

D. Hierarchical Bayes Models (Train, 2003)

E. An alternate approach to heterogeneity: Latent Class Models (Scarpa and Thiene, 2005)

Possible Special Topics if Time Permits

7. Duration Analysis (TBA).

References

Cameron, Trudy A. and Michelle D. James (1987), ―Efficient Estimation Methods for ‗Closed Ended‘ Contingent Valuation Surveys,‖ Review of Economics and Statistics, 69(2) , 269-286.

Rivers, D. and Vuong (1988), ―Limited Information Estimators and Exogeneity Testing for Simultaneous Probit Models,‖ Journal of Econometrics, 39(3), 347-366.

Scarpa, Riccardo and Mara Thiene (2005), ―Destination Choice Models for Rock Climbing in the Northeastern Alps: A Latent-Class Approach Based on Intensity of Preferences,‖ Land Economics, 81(3), 426-444.

Train, Kenneth E. (1998), ―Recreation Demand Models with Taste Difference Over People,‖ Land Economics, 74(2), 230-239.

Train, Kenneth E. (1999), ―Mixed Logit Models for Recreation Demand,‖ Chapter 4 in Joseph A. Herriges and Catherine L. Kling (eds.), Valuing Recreation and the Environment. Revealed Preference Methods in Theory and Practice, Cheltenham, UK: Edward Elgar Publishing.

Train, Kenneth E. (2003), Discrete Choice Methods with Simulation, Cambridge, UK: Cambridge University Press.

Academic Integrity

The University of Maryland, College Park has a nationally recognized Code of Academic Integrity, administered by the Student Honor Council. This Code sets standards for academic integrity at Maryland for all undergraduate and graduate students. As a student you are responsible for upholding these standards for this course. It is very important for you to be aware of the consequences of cheating, fabrication, facilitation, and plagiarism. For more information on the Code of Academic Integrity or the Student Honor Council, please visit http://www.shc.umd.edu .

Honor Pledge

To further exhibit your commitment to academic integrity, remember to sign the Honor Pledge on all examinations and assignments:

"I pledge on my honor that I have not given or received any unauthorized assistance on this examination (assignment)."