COURSE OVERVIEW

This applied course offer a rigorous overview of the more advanced technical capabilities currently available in Stata for linear regression analysis. Thus providing participants with a unique hands-on opportunity to acquire the necessary theoretical and applied skills to independently apply advanced linear regression techniques in Stata.

 

In the opening session, the application of Ordinary Least Squares (OLS) techniques for both estimation and inference in linear models in the presence of exogenous regressors is discussed. In the second session, participants address the problems which arise when trying to estimate linear models containing endogenous regressors. Instrumental Variable and Generalized Methods of Moments (GMM) techniques for both estimation and inference, together with a discussion of the available tests for both weak identification and inference procedures under weak identification, therefore form the core of this closing session.

 

COURSE STRUCTURE

Individual sessions are composed of both a theoretical component (in which the techniques and underlying principles behind them are explained), and an applied (hands-on) segment using Stata, during which participants implement the techniques using real data under the watchful eye of the course tutor. Throughout the course, theoretical sessions are reinforced by applied examples, in which the course tutor discusses and highlights potential pitfalls and the advantages of individual techniques. The intuition behind the choice and implementation of a specific technique is of the utmost importance. In this manner, the course leader is able to bridge the “often difficult” gap between abstract theoretical methodologies, and the practical issues one encounters when dealing with real data.

 

COURSE OUTCOME

At the end of the course, participants are expected to be able to autonomously implement the theories and methodologies discussed during the course.

 

TARGET AUDIENCE 

This course is particular interest to researchers in public and private research centres, Master and Ph.D. Students working in the following fields: Agricultural Economics, Economics, Finance, Management, Public Health, and the Political and Social Sciences seeking to acquire the applied and theoretical toolset to enable them to independently apply linear regression techniques in their empirical research.

 

COURSE REQUISITES

It is assumed that course participants have:

  • at some point followed a basic course in econometrics or statistics;
  • a knowledge of Stata or other statistical software, SPSS, SAS.

PROGRAM


SESSION I: LINEAR REGRESSION WITH ALL EXOGENOUS REGRESSORS

  1. The OLS estimator: regress
  2. Categorical variables, dummies, interactions and marginal effects: margins
  3. Testing hypotheses on model coefficients: test, testparm, lincom, nlcom
  4. OLS predicted values: predict, margins
  5. Testing heteroskedasticity: estatimtest, estathettest
  6. Testing autocorrelation: estat dwatson; estat durbinalt; estat bgodfrey; actest (Baum et al., 2007; Baum et al. 2013); abar (Roodman, 2009)
  7. Consistent variance-covariance estimators under:
    • heteroskedasticity: the regress options vce(robust), vce(hc2), vce(hc3)
    • cluster correlation: the regress option vce(cluster clustervar)
    • autocorrelation: newey

 

SESSION II: LINEAR REGRESSION WITH POSSIBLY ENDOGENOUS REGRESSORS

  1. Optimal estimation and inference under i.i.d. errors with the Two-Stage-Least- Square estimator: ivregress 2sls, ivreg2 (Baum et al 2003, 2007 )
  2. Optimal estimation and inference under non-i.i.d. errors with overidentified GMM estimators: ivregress gmm , ivreg2
  3. Consistent variance-covariance estimators under:
    • heteroskedasticity: ivregress…,vce(robust);ivreg2…,robust
    • cluster correlation: ivregress…,vce(cluster clustervar); ivreg2…,cluster(clustervar)
    • twoway cluster correlation: ivreg2…,cluster(varlist)
    • autocorrelation: ivregress…,vce(hac kernel); ivreg2…,bw(#)
  4. Specification tests:
    • Testing heteroskedasticity: ivhettest (Baum et al. 2003)
    • Testingautocorrelation: actest; abar
    • Testing overidentifying restrictions: estat overid
    • Testing subsets of overidentifying restrictions: ivreg2…,orthog(varlist_inst)
    • Testing subsets of regressors for endogeneity: estat endogenous; ivreg2…,orthog(varlist_regr)
    • Tests for weak instruments: ivregress…,first; ivreg2…,first
    • A robust test for weak instruments with one endogenous variable: weakivtest
  5. Inference with weak instruments: ivreg2…,first; condivreg (Mikusheva et al 2006); weakiv
  6. Estimation and inference using heteroskedasticity without instruments: ivreg2h

SUGGESTED READINGS