This preview shows pages 1–4. 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 Document
Unformatted text preview: Economics 4261 Introduction to Econometrics Fall 2010 HOMEWORK 4 : REGRESSION INFERENCE KEY Question 1 Consider the regression model y X , with K predictors (excluding intercept). Using our standard assumptions, (a) Show SSR e 1 e 1 M X . (b) Let s 2 SSR n K 1 Using Assumptions A1A4, show that E r s 2  X s 2 . This will require you to use the trace operator mentioned in HW1. You should proceed in two steps: (1) Notice that SSR 1 M X is a quadratic form. Write this out using sum notation and show that E r 1 M X  X s 2 trace r M s . (2) Show that trace r M s n K 1. Note that since V ar p  X q 2 p X 1 X q 1 , regression packages printout y V ar p  X q s 2 p X 1 X q 1 .................................................................................................................. (For this proof, I let M M X and P P X . ) Using the hint, we proceed in two steps. The first step tells us that SSR 2 trace p X q . Then under the second part, we establish the form of trace p X q . Step 1 First note that E r 1 X  X s n i 1 n j 1 m ij E r i j  X s , where the above holds for all i,j , and since M is just a function of X , we can take it outside the expectation. Next, by spherical errors, E r i j  X s 0 for all i j . For i j , we have E r i j  X s 2 . Putting these two facts together yields E r 1 X  X s n i 1 m ii E r 2 i  X s 2 n i 1 m ii 2 trace p M q . Step 2 Next we need to obtain the expression for trace p M q . We note that trace p M q trace p I n q trace p P q . The trace of the identity matrix is n , hence trace p M q n trace p P q . Recall that P X p X 1 X q 1 X 1 so kwilliams@umn.edu http://www.econ.umn.edu/ ~ will3324 Page 1 of 9 Economics 4261 Introduction to Econometrics Fall 2010 trace p P q trace r X p X 1 X q 1 X 1 s trace rp X 1 X q 1 p X 1 X qs trace r I #regression coefficients s . The above work holds from the first homework assignment. The total number of regressors is K . We also have a coefficient for the intercept. Thus, trace r I #regression coefficients s K 1, which implies trace p M q n K 1 . Lets check our work: E r s 2  X s SSR n K 1 SSR 1 n K 1 E r 1 M X  X s 1 n K 1 2 trace r M s 1 n K 1 2 p n K 1 q 1 n K 1 2 . kwilliams@umn.edu http://www.econ.umn.edu/ ~ will3324 Page 2 of 9 Economics 4261 Introduction to Econometrics Fall 2010 Question 2 Consider the linear model y i 1 x 1 i 2 x 2 i i , such that i P 1 ,..., 10. Suppose X 1 X 10 50 5 50 304 34 5 34 5 and SSE 10 i 1 p y i y q 2 . 4987 and MSE . 1752 Perform calculations at the 95% confidence level....
View
Full
Document
This note was uploaded on 02/24/2011 for the course ECON 4261 taught by Professor Staff during the Spring '08 term at Minnesota.
 Spring '08
 Staff
 Econometrics

Click to edit the document details