[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

From |
Kit Baum <baum@bc.edu> |

To |
statalist@hsphsun2.harvard.edu |

Subject |
st: Test of equality of the coefficients between two samples endogeneity and IVREG2 |

Date |
Wed, 23 Nov 2005 07:10:43 -0500 |

The Chow test, in OLS or IV, can always be expressed in terms of creating a dummy--in your case for the sample, 0=sample A, 1=sample B, and interacting it with the regressors. The regression then is augmented with the dummy (which is a constant term shift for sample B) and the interacted regressors. The joint test that the dummy coefficient and all interactions are zero is the 'Chow test'. Each interaction term's coefficient is a test for equality of that coefficient across samples. You normally have to worry about heteroskedasticity caused by pooling in this context, but since you're using GMM, it will take care of it and 'robustify' the F- and t-statistics.

In the context of instrumental variables, you will of course have to consider (migrated*dummy) to be an additional endogenous regressor and find an instrument for it.

Kit Baum, Boston College Economics

http://ideas.repec.org/e/pba1.html

On Nov 23, 2005, at 2:33 AM, Nasser wrote:

I am using IVREG2 with gmm option to estimate the coefficient of a

dichotomous endogenous variable (migrated=3D1 or 0). I apply the model=

to

two samples. I would like to test for the equality of the two coeffici=

ents

between the two samples. Is there a particular test in IVREG2 (with gm=

m

option) that is equivalent to OLS Chew test?

* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

- Prev by Date:
**RE: st: right justifying strings, left padding strings with zeros** - Next by Date:
**st: Axis's limits when one var has missing when other is max** - Previous by thread:
**st: RE: saving and later retrieving results from an estimated model** - Next by thread:
**st: Axis's limits when one var has missing when other is max** - Index(es):

© Copyright 1996–2015 StataCorp LP | Terms of use | Privacy | Contact us | What's new | Site index |