{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

Econometrics-I-14

Ii use mean square ˆ var n n requires to be well

Info iconThis preview shows pages 19–26. Sign up to view the full content.

View Full Document Right Arrow Icon
ii Use Mean Square ˆ Var[ | ]= ? n n Requires  to be   "well behaved" n Either converge to a constant matrix or diverge. Heteroscedasticity case: 1 1 n n Autocorrelatio - = σ ÷ ÷ = ϖ -1 -1 -1 X' X Ω X 0 β X' X Ω X' X Ω x x ' n n 2 i j i 1 j 1 ij n case: 1 1 . n  terms. Convergence is unclear. n n = = = ϖ ∑ ∑ -1 X' X Ω x x '
Background image of page 19

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Part 14: Generalized Regression Asymptotic Normality ™  19/44 - - - - - - - = ÷ ÷ 1 1 1 1 1 1 ' 1 ˆ n( ) n ' n n Converge to normal with a stable variance O(1)? '  a constant matrix? n 1 '  a mean to which we can apply the n                    central limit theorem? Het X X Ω X β β Ω ε X X Ω X Ω ε - = ε ε = σ ÷ ÷ ÷ ÷ ϖ ϖ ϖ ϖ ϖ n 1 2 i i i i i 1 i i i i i i eroscedasticity case? 1 1 '   .  Var ,  is just data. n n   Apply Lindeberg-Feller.  (Or assuming  /  is a draw from a common   distribution with mean and fixed va x x X = Ω ε x riance - some recent treatments.) Autocorrelation case?
Background image of page 20
Part 14: Generalized Regression Asymptotic Normality (Cont.) ™  20/44 n n 1 i j i i i 1 j 1 1 For the autocorrelation case 1 1 '   n n Does the double sum converge?  Uncertain.  Requires elements of   to become small as the distance between i and j increases. (Has to resem - = = - ε ε ∑ ∑ ij X =  x x Ω ε Ω Ω ble the heteroscedasticity case.)
Background image of page 21

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Part 14: Generalized Regression Test Statistics (Assuming Known Ω) p With known Ω , apply all familiar results to the transformed model: p With normality, t and F statistics apply to least squares based on Py and PX p With asymptotic normality, use Wald statistics and the chi-squared distribution, still based on the transformed model. ™  21/44
Background image of page 22
Part 14: Generalized Regression Generalized (Weighted) Least Squares Heteroscedasticity Case ™  22/44 1 2 2 2 n 1 -1/2 2 n 1 n n 1 i i i 1 i 1 i i i i i 2 0 ... 0 0 ... 0 Var[ ] 0 0 ... 0 0 0 ... 1/ 0 ... 0 0 1/ ... 0                 0 0 ... 0 0 0 ... 1/ 1 1 ˆ ( ) ( ) y ˆ y ˆ - - = = ϖ ϖ ε = σ Ω = σ ϖ ϖ ϖ = ϖ = = ÷ ÷ ϖ ϖ - ϖ σ = -1 -1 i i X' X X' y x x x β Ω Ω x β 2 n i 1 n K = ÷ ÷ -
Background image of page 23

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Part 14: Generalized Regression Autocorrelation t = t-1 + ut (‘First order autocorrelation.’ How does this come about?) Assume -1 <  < 1. Why? ut = ‘nonautocorrelated white noise’ t = t-1 + ut (the autoregressive form) = (t-2 + ut-1) + ut = ... (continue to substitute) = ut + ut-1 + 2ut-2 + 3ut-3 + ... = (the moving average form) ™  23/44
Background image of page 24
Part 14: Generalized Regression Autocorrelation ™  24/44 2 t t t 1 t 1 i t i i= 0 2 2i 2 u u 2 i=0 t t 1 t t 1 t 2 t t 1 t t 1 t Var[ ] Var[u u u ...]           =  Var u           = 1 An easier way: Since Var[ ] Var[ ] and  u Var[ ] Var[ ] Var[u ] 2 Cov[ ,u ]            = - - - - - - - ε = + ρ + ρ +
Background image of page 25

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Image of page 26
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}