6.7.4.3.4. statsmodels.sandbox.regression.gmm.spec_hausman

statsmodels.sandbox.regression.gmm.spec_hausman(params_e, params_i, cov_params_e, cov_params_i, dof=None)

Hausmans specification test

Parameters:

params_e : array

efficient and consistent under Null hypothesis, inconsistent under alternative hypothesis

params_i: array

consistent under Null hypothesis, consistent under alternative hypothesis

cov_params_e : array, 2d

covariance matrix of parameter estimates for params_e

cov_params_i : array, 2d

covariance matrix of parameter estimates for params_i

example instrumental variables OLS estimator is `e`, IV estimator is `i`

Notes

Todos,Issues - check dof calculations and verify for linear case - check one-sided hypothesis

References

Greene section 5.5 p.82/83