Large-N and Large-T Properties of Panel Data Estimators and the Hausman Test



4             4     1 T-1 s s         Q-1

4!8 f sup ∣∣Xit∣∣4qj sup T∑∑∑α,(s) q

,              i,T    s=1 p=0 k=0

4∞                    1

sup ∣Xit 4q   X(s +1)2 sup ai (s) q < M,

(43)


i,t                  s=1              i

where the last bound holds because supi ai (s) is of size —3 ^-ɪ (see Assumption
3(i)). By the similar fashion, we can also show that

II,III,IV ≤ 4!8


4

sup∣Xit4q        (

i,t                s=1


s +1)2 sup αi (s) q <M,
i


and we have all the required result. ¥

Part (c)

Let Yit = xh,3,it EFz xh,3,it. Using the arguments similar to those used in
the proof of Part (b), we can show that under Assumption 4,


2M (sup kYitkFzi,4q) X s2 (sup aFzi (s) qq )


a.s.,


(44)


for some constant M. By Assumption 4(i), P= s2 supi aFZ (s) q a.s..
Finally, since the terms sup
i t kYitkF 4q and PS=1 s2 supi αFz. (s) q are Fo-
measurable, we have the desired result by choosing

Mz =2M


sup kYitk4Fzi,4q Xs2 sup aFzi


q1

(s) ~ J . ¥


Part (d)

Again, let Yit = xh,3,it EFzi xh,3,it. From (44) , we have


sup
i,T


Yit


sup E
i,T


Yit


≤ 2ME


(sup kYit Fzi ,4q) X s2 (sup aFzi (s) q—q~ ɔ


≤ 2M E


sup kYitk8Fzi,4q


)]2 E{ x s2('


36


q — 1
sup αFzi (s) q


(45)




More intriguing information

1. Lumpy Investment, Sectoral Propagation, and Business Cycles
2. TRADE NEGOTIATIONS AND THE FUTURE OF AMERICAN AGRICULTURE
3. The name is absent
4. The Triangular Relationship between the Commission, NRAs and National Courts Revisited
5. Expectations, money, and the forecasting of inflation
6. The name is absent
7. Improving Business Cycle Forecasts’ Accuracy - What Can We Learn from Past Errors?
8. THE WELFARE EFFECTS OF CONSUMING A CANCER PREVENTION DIET
9. The Shepherd Sinfonia
10. CONSUMER PERCEPTION ON ALTERNATIVE POULTRY