what does it mean for the hansen test to be less than 0.05 (GMM)?

[raphael rodrigues]

Hello. I am estimating regressions using GMM-DIFFERENCE and GMM-SYSTEM, but I am confused about the interpretation of the hansen test result, that is, what is the effect of this result for this regression (bias in the coefficients?)

ESTIMATION USING GMM-DIFFERENCE
Arellano-Bond test for AR(1) in first differences: z = -1.92 Pr > z = 0.054
Arellano-Bond test for AR(2) in first differences: z = -1.06 Pr > z = 0.290
Sargan test of overid. restrictions: chi2(19) = 37.35 Prob > chi2 = 0.007 (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(19) = 34.77 Prob > chi2 = 0.015 (Robust, but weakened by many instruments.)
Difference-in-Hansen tests of exogeneity of instrument subsets: iv(dummy1011 dummy1213 dummy1415 dummy1617 dummy1819) Hansen test excluding group: chi2(17) = 33.60 Prob > chi2 = 0.009
Difference (null H = exogenous): chi2(2) = 1.17 Prob > chi2 = 0.556

And for the GMM-SYSTEM
Arellano-Bond test for AR(1) in first differences: z = -3.10 Pr > z = 0.002
Arellano-Bond test for AR(2) in first differences: z = -0.88 Pr > z = 0.381
Sargan test of overid. restrictions: chi2(28) = 174.26 Prob > chi2 = 0.000 (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(28) = 42.15 Prob > chi2 = 0.042 (Robust, but weakened by many instruments.)
Difference-in-Hansen tests of exogeneity of instrument subsets:
GMM instruments for levels
Hansen test excluding group: chi2(20) = 30.98 Prob > chi2 = 0.055
Difference (null H = exogenous): chi2(8) = 11.17 Prob > chi2 = 0.192
iv(dummy1011 dummy1213 dummy1415 dummy1617 dummy1819)
Hansen test excluding group: chi2(25) = 38.47 Prob > chi2 = 0.042
Difference (null H = exogenous): chi2(3) = 3.68 Prob > chi2 = 0.298

[Sebastian Kripfganz]

A significant Hansen test suggests that your model is misspecified. One potential source of misspecification could be invalid instruments. As a consequence, your coefficient estimates are generally biased/inconsistent.

More on dynamic panel GMM:

• Kripfganz, S. (2019). Generalized method of moments estimation of linear dynamic panel data models. Proceedings of the 2019 London Stata Conference.

[raphael rodrigues]

Tks, Sebastian. I found your document when i was looking for an answer to my question in other discussions on this forum. I will report the regression, but paying attention on the consequences of the test results