[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: New version of -ivreg2- on SSC
Thanks to Kit Baum, a new version (2.1.15) of -ivreg2- by Baum-Schaffer-Stillman is now available from SSC.
The main addition is a new option, -endog-. The -endog- option makes it easier for the user to carry out a test of the endogeneity of an endogenous regressor. It complements the existing -orthog- option, which is used to carry out a C or "distance GMM" test of the exogeneity (orthogonality) of an exogenous regressor or excluded instrument. To illustrate:
1. Test the endogeneity of an endogenous regressor:
ivreg2 lw s expr tenure rns smsa (iq=med kww age mrt), endog(iq)
2. Test the orthogonality of an exogenous regressor:
ivreg2 lw s expr tenure rns smsa (iq=med kww age mrt), orthog(s)
3. Test the orthogonality of an excluded instrument:
ivreg2 lw s expr tenure rns smsa (iq=med kww age mrt), orthog(age)
For those who are curious, the endogeneity test is also a C or distance GMM test, just like that reported for ‑orthog‑, but applied to a different equation. To generate the endogeneity test statistic, ‑ivreg2‑ calls itself recursively with the ‑orthog‑ option. Under conditional homoskedasticity, the endogeneity test statistic is identical to a Durbin-Wu-Hausman test, and as such is also available from ‑ivendog‑; the main advantage of the ‑endog‑ option is that it can produce test statistics that are robust to various violations of homoskedasticity. See the ‑ivreg2‑ help file and the references therein for more discussion.
The new version also incorporates some minor bug fixes, notably for the continuously-updated (CUE) GMM estimator.
Prof. Mark E. Schaffer
Centre for Economic Reform and Transformation
Department of Economics
School of Management & Languages
Edinburgh EH14 4AS UK
* For searches and help try: