This preview shows pages 1–8. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: 1 Econometrics Lecture #10: Further Issues in Dummy Variables Outline Combining Ftests and dummy interactions: the Chow Test Dummies for time periods Differenceindifferences Did the Mariel Boatlift lower the wages of Miamis workers? Maybe briefly: Dummy Dependent Variable 2 Chow Test Special case of Ftest Chow test asks if separate models should be estimated for different groups? Ex: men & women Recall from last time: represents the intercept from the excluded group men 1 is how much different the intercept is for women then men, So 1 0 says women have a different intercept i i i i i u female yrsed yrsed female wg + + + + = 3 2 1 ) ln( Chow Test Recall from last time: 2 represents the slope for the excluded group the returns to education for men 3 is how much different the slope is for women then men, So 3 0 says women have a different slope Therefore the hypothesis that women & men have the same wageeducation relationship can be tested with H : 1 = 3 = 0 An Ftest! i i i i i u female yrsed yrsed female wg + + + + = 3 2 1 ) ln( Chow Test Example 1. Estimate fully interacted model = all variables interacted w/female R 2 unrestricted Equivalent to estimating separate m/f models 1. Estimate pooled model R 2 restricted 1. Compute Fstat, decide i i i i i i i i u female etc etc female female yrsed yrsed female wg + + + + + + + + = ...) . ... exp exp ( ) ln( 4 3 3 2 1 i i u yrsed wg + + + = ... exp ) ln( 3 2 Chow Test Example R 2 unrestricted R 2 restricted Regression 1 Regression 2 Regression 3 female0.5940.190*** (0.471) (0.0645) yrsed 0.0799*** 0.0965*** 0.0905*** (0.0258) (0.0173) (0.0175) yrsed_fem 0.0301 (0.0348) Constant 1.603*** 1.383*** 1.362*** (0.345) (0.233) (0.237) Observations 220 220 220 Rsquared 0.147 0.144 0.110 Regressions of Ln(wage) on Education, Female *** p<0.01, ** p<0.05, * p<0.1 Standard errors in parentheses Chow Test What are R 2 unres. , R 2 res , q, N and K? R 2 unres. = 0.147, R 2 res. = 0.110 q = 2, N = 220, K = 3 5% level critical value for F 2,216 3 Fstat = 4.68 > 3 reject H at 5% level Test says wage regressions should be run separately for men and women ( 29 ( 29 ( 29 1 1 2 2 2 = K N R q R R stat F ed unrestrict restricted ed unrestrict ( 29 ( 29 ( 29 68 . 4 1 3 220 147 . 1 2 110 . 147 ....
View
Full
Document
This document was uploaded on 01/31/2011.
 Spring '09
 Econometrics

Click to edit the document details