Econometrics Harvard Economics（哈佛经济学计量经济学）.pdfVIP

Econometrics Honor’s Exam Review Session Spring 2011 Eunice Han Topics covered in lectures 1.  OLS •  The Assumptions •  Omitted Variable Bias •  Hypothesis Testing •  Confidence Intervals •  Heteroskedasticity •  Nonlinear Regression Models: Polynomials, Logs, and Interaction Terms 2.  Panel Data: •  Fixed Effects •  Clustered HAC SE 3.  Internal Validity and External Validity 4.  Binary Dependent Variables: LPM, Probit and Logit Model 5.  Instrumental Variables 6.  Time Series Data •  Stationarity •  Forecasting Models 2 •  Newey-West HAC SE Least Squares Assumptions 1. (X ,Y ) are iid; random sampling i i 2. Large outliers are rare; If this condition fails, OLS estimator is not consistent. 3. E (u |X) = 0; Conditional Mean Zero assumption. Xs are i i exogenous. This assumption fails if X and u are correalted. 4. No Perfect Multicollinearity Condition: The regressors are said to be perfectly multicollinear if one of the regressors is a perfect linear function of the other regressor(s).   If all the assumptions are satisfied, the OLS estimates are unbiased and consistent. 3 Univariate Regression Model   Do married men make more money than single men? lwage = β + β married + u ; i=1,2,…,n i 0 1 i i lwage= log(monthly wage) married =1 if married, 0 if single Q1. How would you interpret β and β ? 0 1 β0 (intercept) is the average lwage of single men. β (slope) represents the difference in average lwage between single 1 € € and married men.