Many IVs estimation of dynamic panel regression models with measurement error

A-Tier
Journal: Journal of Econometrics
Year: 2017
Volume: 200
Issue: 2
Pages: 251-259

Score contribution per author:

1.341 = (α=2.01 / 3 authors) × 2.0x A-tier

α: calibrated so average coauthorship-adjusted count equals average raw count

Abstract

In this paper, we investigate a dynamic linear panel regression model with measurement error. We consider the panel data estimation whose time dimension (T) is not small and comparable to the cross sectional dimension (N). First, we show that the 2SLS estimator suffers from the bias problem due to many instrumental variables. Using the alternative asymptotics where T3N goes to a constant as N,T→∞, we characterize its asymptotic bias due to many IVs. As a bias reduction method, we investigate the JIVE and derive its limiting distribution under the alternative asymptotics.

Technical Details

RePEc Handle
repec:eee:econom:v:200:y:2017:i:2:p:251-259
Journal Field
Econometrics
Author Count
3
Added to Database
2026-01-25