计量经济学基础第6章课件).pptVIP

AIC auxiliary regression BIC collinearity F-test irrelevant variables nonsample information omitted variables Keywords omitted variable bias overall significance prediction RESET restricted least squares restricted model restricted SSE SC single and joint null hypothesis testing many parameters unrestricted model unrestricted SSE Key Words Appendices The F-statistic is defined as: We can also show that: For sufficiently large sample: 6A Chi-Square and F-Tests: More Details Eq. 6A.1 Eq. 6A.2 Eq. 6A.3 But we can also show that: From the book’s appendix, we know that: Therefore: Eq. 6A.4 Eq. 6A.5 6A Chi-Square and F-Tests: More Details A little reflection shows that: 6A Chi-Square and F-Tests: More Details When testing in the equation we get 6A Chi-Square and F-Tests: More Details Testing we get 6A Chi-Square and F-Tests: More Details Consider the model: Now suppose we incorrectly omit xi3 and estimate: 6B Omitted-Variable Bias: A Proof Notice the new disturbance term It’s 6B Omitted-Variable Bias: A Proof The estimator for β2 is: where Eq. 6B.1 6B Omitted-Variable Bias: A Proof Substituting for vi yields: where 6B Omitted-Variable Bias: A Proof Hence: 6B Omitted-Variable Bias: A Proof Some points for choosing a model (Continued): Consider various model selection criteria The adequacy of a model can be tested using a general specification test known as RESET 6.3 Model Specification 6.3.3 Choosing the Model There are three main model selection criteria: R2 AIC SC (BIC) 6.3.4 Model Selection Criteria 6.3 Model Specification A common feature of the criteria we describe is that they are suitable only for comparing models with the same dependent variable, not models with different dependent variables like y and ln(y) 6.3 Model Specification 6.3.4 Model Selection Criteria The problem is that R2 can be made large by adding more and more variables, even if the variables added have no justification Algebraically,