How to Do xtabond2 David Roodman Research Fellow Center for Global Development
xtabond2 in a nutshell First ado version in 11/03, Mata version in 11/05. Extends built-in xtabond, to do system GMM, Windmeijer correction, revamped syntax Estimators designed for Small-T, large-N panels One dependent variable Dynamic Linear Regressors endogenous and predetermined Fixed individual effects Arbitrary autocorrelation and het. within panels General application
Outline of paper Introduction to linear GMM Motivation and design of difference and system GMM xtabond2 syntax
Black box problem Canned & sophisticated procedure Dangers in hidden sophistication finite sample ≠ asymptotic Users should understand motivation and limits of estimator
Linear GMM in one slide
Linear GMM and 2SLS
Linear GMM in another slide (Holtz-Eakin, Newey, and Rosen 1988)
Difference and system GMM
Problem: Dynamic Panel Bias (Nickell 1981)
Partial solution: OLS in differences
Problem: Other endogeneity
Solution: Instrument with lags (2SLS) (Anderson and Hsiao 1981)
Problem: Inefficiency
Solution: GMM & GMM-style instruments (Holtz-Eakin, Newey, and Rosen 1988)
Problem: Autocorrelation
Solution: Restrict to deeper lags
Arellano-Bond AR() test
Problem: Weak instruments
Solution: Instead of purging fixed effects, find instruments orthogonal to them (Arellano and Bover 1995)
Relationship among moments (Tue Gorgens)
Problem: Two-step errors too small
Problem, cont’d
Solution: finite-sample correction (Windmeijer 2005)
Problem: too many instruments
Solution: consider limiting instruments
xtabond2 syntax Z Y X “GMM-style” Classic
Examples
Examples, cont’d
Run times for bbest (seconds) 700 MHz PC xtabond2 ado 57 xtabond2 Mata, favoring space 14 xtabond2 Mata, favoring speed 11 DPD for Ox 3