Download Random Effects Linear Model - Econometric Analysis of Panel Data - Lecture Slides and more Slides Econometrics and Mathematical Economics in PDF only on Docsity! Econometric Analysis of Panel Data 5. Random Effects Linear Model Docsity.com The Random Effects Model The random effects model ci is uncorrelated with xit for all t; E[ci |Xi] = 0 E[εit|Xi,ci]=0 it it i it i i i i i i i i i i i i N i=1 i 1 2 N y = +c +ε , observation for person i at time t = +c + , T observations in group i = + + , note (c ,c ,...,c ) = + + , T observations in the sample c=( , ,... ) , ′ ′= Σ ′ ′ ′ ′ x β y X β i ε X β c ε c y Xβ c ε c c c Ni=1 iT by 1 vectorΣ Docsity.com Notation 2 2 2 2 u u u 2 2 2 2 u u u i i 2 2 2 2 u u u 2 2 u i i 2 2 u i 1 2 N Var[ +u ] = T T = = Var[ | ] + + = + ′+ × ′+ = i i T T ε i I ii I ii Ω Ω 0 0 0 Ω 0 w X 0 0 Ω ε ε ε ε ε σ σ σ σ σ σ σ σ σ σ σ σ σ σ σ σ i (Note these differ only in the dimension T ) Docsity.com Regression Model-Orthogonality iN N i 1 i i 1 i i i iN i 1 i i i i i i i i N i i i 1 i 1 plim #observations 1 1 plim plim ( +u ) T T 1 plim T + Tu T T T T plim f + f u , 0 < f < 1 T T T pli = = = = = ′ ′Σ = Σ = Σ Σ ′ ′ Σ Σ Σ ′ ′ Σ Σ = Σ N N i=1 i i i=1 i i N Ni i i i i=1 i=1 N Ni i i i i=1 i=1 X'w 0 X w X ε i 0 X ε X i X ε X i i i i i i m f + f u T ′ Σ Σ = N Ni i i=1 i=1 X ε x 0 Docsity.com Convergence of Moments N i 1 iN i 1 i N i 1 iN i 1 i N N i 1 i u i 1 i i f a weighted sum of individual moment matrices T T f a weighted sum of individual moment matrices T T = f f T Note asymptoti = = = = = = ′′ = Σ = Σ ′′ = Σ = Σ ′ ′Σ + Σ i i i i i 2 2i i i i X XX X X Ω XX ΩX X X x xεσ σ i i i cs are with respect to N. Each matrix is the T moments for the T observations. Should be 'well behaved' in micro level data. The average of N such matrices should be likewise. T or T is assum ′i iX X ed to be fixed (and small). Docsity.com Estimating the Variance for OLS 1 1 N N N N i 1 i i 1 i i 1 i i 1 i N i 1 iN i 1 i N i 1 iN i 1 i Var[ | ] T T T T f , where = =E[ | ] T T In the spirit of the White estimator, use ˆ ˆ ˆf , T T − − = = = = = = = = ′ ′ ′ = Σ Σ Σ Σ ′′ ′= Σ Σ ′ ′′ = Σ Σ i i i i i i i i i i i i 1 X X X ΩX X Xb X X Ω XX ΩX Ω w w X X w w XX ΩX w = Hypothesis tests are then based on Wald statistics. i iy - X b THIS IS THE 'CLUSTER' ESTIMATOR Docsity.com Mechanics ( )1 1Ni 1 i i ˆ ˆEst.Var[ | ] ˆ = set of T OLS residuals for individual i. = TxK data on exogenous variable for individual i. ˆ = K x 1 vector of products ˆ ˆ( )( ) − − =′ ′ ′ ′= Σ ′ ′ ′ ′ = i i i i i i i i i i i i b X X X X w w X X X w X X w X w w X ( ) ( )( ) ( )( ) ( )( ) N i 1 N N i 1 i 1 KxK matrix (rank 1, outer product) ˆ ˆ = sum of N rank 1 matrices. Rank K. ˆˆ ˆWe could compute this as = . Why not do it that way? = = = ′ ′Σ ≤ ′ ′ ′Σ Σ i i i i i i i i i i i X w w X X w w X X Ω X Docsity.com Cornwell and Rupert Data Cornwell and Rupert Returns to Schooling Data, 595 Individuals, 7 Years Variables in the file are EXP = work experience, EXPSQ = EXP2 WKS = weeks worked OCC = occupation, 1 if blue collar, IND = 1 if manufacturing industry SOUTH = 1 if resides in south SMSA = 1 if resides in a city (SMSA) MS = 1 if married FEM = 1 if female UNION = 1 if wage set by unioin contract ED = years of education BLK = 1 if individual is black LWAGE = log of wage = dependent variable in regressions These data were analyzed in Cornwell, C. and Rupert, P., "Efficient Estimation with Panel Data: An Empirical Comparison of Instrumental Variable Estimators," Journal of Applied Econometrics, 3, 1988, pp. 149-155. See Baltagi, page 122 for further analysis. The data were downloaded from the website for Baltagi's text. Docsity.com Generalized Least Squares i 1 N 1 N i 1 i 1 2 T2 2 2 i u i ˆ=[ ] [ ] =[ ] [ ] 1 I T (note, depends on i only through T ) − − = = ε ε ε ′ ′ ′ ′Σ Σ σ ′= − σ σ + σ -1 -1 -1 -1 i i i i i i -1 i β X Ω X X Ω y X Ω X X Ω y Ω ii Docsity.com Panel Data Algebra (1) 2 2 ε u i i 2 2 2 2 ε u ε 2 2 ε ε 2 ε = σ +σ , depends on 'i' because it is T T = σ [ ], = σ / σ = σ [ ] = σ [ ], = , = . Using (A-66) in Greene (p. 822) 1 1 σ 1+ = ′ × ′ρ ρ ′ ′ρ + ρ ′= ′ i 2 i 2 i -1 -1 -1 -1 i -1 Ω I ii Ω I + ii Ω I + ii A bb A I b i Ω A - A bb A b A b 2 2 u 2 2 2 2 2 ε i ε ε i u σ1 1 1 = σ 1+T σ σ +Tσ ′ ′ρ ρ I - ii I - ii Docsity.com Panel Data Algebra (2) 2 2 2 2 2 2 ε u ε i u ε i u 2 2 ε i u 2 2 2 2 2 ε i u i ε ε i u 2 2 ε i u 2 ε (Based on Wooldridge p. 286) σ +σ σ +Tσ σ +Tσ σ +Tσ ( ) (σ +Tσ )[ ( )], = σ /(σ +Tσ ) (σ +Tσ )[ ] (σ ′ ′ ′= = = = − = + η − η = + η = -1 i i D i D i i D D i i D D Ω I ii I i(i i) i I P I I M P I P P M 2 i u 1 i 1 / 2 i i i i i a a i +Tσ ) (1 / ) (Prove by multiplying. .) 1 (1 / ) , =1- 1 (Note ) − − = + η = = + η = − θ θ η − θ = + η i i i i i i D D D D i i i i D D D i i i D D S S P M P M 0 S P M I P S P M Docsity.com Estimators for the Variances i i it it i OLS T TN 2 N 2 2 i 1 t 1 it i 1 t 1 U 2 2 2 U i LSDV i y u With a consistent estimator of , say , (y ) estimates ( ) Divide by something to estimate = With the LSDV estimates, a and , = = = = ε ε = ′= + ε + ′Σ Σ Σ Σ σ + σ σ σ + σ Σ it it x β β b - x b b i iT TN 2 N 2 1 t 1 it i 1 t 1 2 2 2 2 2 U U (y a ) estimates Divide by something to estimate Estimate with ( ) - .ˆ = = = ε ε ε ε ′Σ Σ Σ σ σ σ σ + σ σ i it- - x b Docsity.com Feasible GLS i 2 2 u TN 2 2 i 1 t 1 it i it LSDV N i 1 i N 2 2 i 1 u Feasible GLS requires (only) consistent estimators of and . Candidates: (y a ) From the robust LSDV estimator: ˆ T K N From the pooled OLS estimator: ε = = ε = = ε σ σ ′Σ Σ − − σ = Σ − − Σ Σ σ + σ = x b i i i T 2 t 1 it OLS it OLS N i 1 i N 2 2 2 i 1 it i MEANS u T 1 TN 2 2 i 1 t 1 s t 1 it is it is i u u N i 1 i (y a ) T K 1 (y a ) From the group means regression: / T N K 1 ˆ ˆw w (Wooldridge) Based on E[w w | ] if t s, ˆ T K = = = ε − = = = + = ′− − Σ − − ′Σ − − σ + σ = − − Σ Σ Σ = σ ≠ σ = Σ − − x b x b X N There are many others. x´ does not contain a constant term in the preceding. Docsity.com Practical Problems with FGLS 2 uAll of the preceding regularly produce negative estimates of . Estimation is made very complicated in unbalanced panels. A bulletproof solution (originally used in TSP, now LIMDEP and others). From th σ i i i i TN 2 2 i 1 t 1 it i it LSDV N i 1 i TN 2 2 2 2i 1 t 1 it OLS it OLS u N i 1 i T TN 2 N 2 i 1 t 1 it OLS it OLS i 1 t 1 i u (y a ) e robust LSDV estimator: ˆ T (y a ) From the pooled OLS estimator: ˆ T (y a ) (y ˆ = = ε = = = ε ε = = = = = ′Σ Σ − − σ = Σ ′Σ Σ − − σ + σ = ≥ σ Σ ′Σ Σ − − − Σ Σ σ = x b x b x b 2t i it LSDV N i 1 i a ) 0 T= ′− − ≥ Σ x b x´ does not contain a constant term in the preceding. Docsity.com Testing for Effects: LM Test 2 2N 2 N 2 i 1 i i 1 i i N N 2 N i 1 i 1 it i 1 i Breusch and Pagan Lagrange Multiplier statistic Assuming normality (and for convenience now, a balanced panel) (Te ) [(Te ) ]NT NT LM= 1 2(T-1) e 2(T-1) Co = = = = = ′Σ Σ − − = ′Σ Σ Σ i i e e e e i N 2 N i 1 i i 1 i nverges to chi-squared[1] under the null hypothesis of no common effects. (For unbalanced panels, the scale in front becomes ( T ) /[2 T (T 1)].)= =Σ Σ − Docsity.com Testing for Effects: Moments ( ) N T-1 T i=1 t=1 s=t+1 it is 2N T-1 T i=1 t=1 s=t+1 it is Wooldridge (page 265) suggests based on the off diagonal elements e e Z= e e which converges to standard normal. ("We are not assuming any particular distribut Σ Σ Σ Σ Σ Σ it 2 ion for the . Instead, we derive a similar test that has the advantage of being valid for any distribution...") It's convenient to examine Z which, by the Slutsky theorem converges (also) to chi- sq ε uared with one degree of freedom. Docsity.com Testing (2) T-1 T t=1 s=t+1 it is T-1 T t=1 s=t+1 it is i e e = 1/2 of the sum of all off diagonal elements of = 1/2 the sum of all the elements minus the diagonal elements. e e =1/2[ ( ) ]. But, = Te . S Σ Σ ′ ′ ′ ′ ′Σ Σ − i i i i i i i e e i e e i e e i e { } T-1 T 2 t=1 s=t+1 it is i 2N 2 N 2 i 1 i2 i 1 N 2 2 N 2 2 i 1 i i 1 i N 2i 1 i i iN 2 ri 1 i o, e e = (1/2)[(Te ) ] [(Te ) ] [ ]2(T 1) Z LM [(Te ) ] NT [(Te ) ] Note, also r N r Z= ,where r (Te ) . sr The claim th = = = = = = ′Σ Σ − ′Σ − ′Σ− = = × ′ ′Σ − Σ − Σ ′= = − Σ i i i i i i i i i i i i e e e e e e e e e e e e i i 1,...,Nat one function of [e , ] is more valid than the other seems a little dubious. =′i ie e Docsity.com Hausman Test for FE vs. RE Estimator Random Effects E[ci|Xi] = 0 Fixed Effects E[ci|Xi] ≠ 0 FGLS (Random Effects) Consistent and Efficient Inconsistent LSDV (Fixed Effects) Consistent Inefficient Consistent Possibly Efficient Docsity.com Hausman Test for Effects -1 d ˆ ˆBasis for the test, ˆ ˆˆ ˆ ˆ ˆWald Criterion: = ; W = [Var( )] A lemma (Hausman (1978)): Under the null hypothesis (RE) ˆ nT[ ] N[ , ] (efficient) ˆ nT[ ′ → FE RE FE RE RE RE FE β - β q β - β q q q β - β 0 V β d] N[ , ] (inefficient) ˆ ˆˆNote: = ( )-( ). The lemma states that in the ˆ ˆjoint limiting distribution of nT[ ] and nT , the limiting covariance, is . But, = → − FE FE RE RE Q,RE Q,RE FE,R - β 0 V q β - β β β β - β q C 0 C C - . Then, Var[ ] = + - - . Using the lemma, = . It follows that Var[ ]= - . Based on the preceding ˆ ˆ ˆ ˆ ˆ ˆH=( ) [Est.Var( ) - Est.Var( )] ( ′ ′ E RE FE RE FE,RE FE,RE FE,RE RE FE RE -1 FE RE FE RE FE RE V q V V C C C V q V V β - β β β β - β ) β does not contain the constant term in the preceding. Docsity.com Computing the Hausman Statistic 1 2 N i 1 i i -1 2 2 N i ui i 1 i i 2 2 i i u 2 2 u 1ˆEst.Var[ ] Iˆ T Tˆ ˆˆEst.Var[ ] I , 0 = 1ˆˆ T Tˆ ˆ ˆAs long as and are consistent, as N , Est.Var[ˆ ˆ − ε = ε = ε ε ′ ′= σ Σ − σγ′ ′= σ Σ − ≤ γ ≤ σ + σ σ σ → ∞ FE i RE i F β X ii X β X ii X β 2 ˆ] Est.Var[ ] will be nonnegative definite. In a finite sample, to ensure this, both must be computed using the same estimate of . The one based on LSDV willˆ generally be the better choice. Note ε − σ E REβ ˆthat columns of zeros will appear in Est.Var[ ] if there are time invariant variables in . FEβ X β does not contain the constant term in the preceding. Docsity.com