Significance level for AR(2) in Arellano-Bond test

[Mohan Jangid]

What should be the cutoff significance level for AR(2) in the Arellano-Bond test for AR(2) in the first differences in the two-step system GMM?

I have applied the two-step system GMM and got this:
Arellano-Bond test for AR(1) in first differences: z = -6.62 Pr > z = 0.000
Arellano-Bond test for AR(2) in first differences: z = -1.74 Pr > z = 0.082

is it okay or model is misspecified?

[Sebastian Kripfganz]

There is no universally accepted threshold. If there is serial correlation in the idiosyncratic error term, this typically invalidates the most of the instruments used in the system GMM approach. Hence, this would be a serious violation of the model assumption. A precautionary stance is therefore advisable, which would call for a significance level higher than the commonly applied 5%.

Personally, I would not feel very confident with a p-value of just 0.08.

You might want to double-check your results with alternative serial correlation tests. The xtdpdgmm command has the postestimation command estat serialpm as an alternative to estat serial.

Alternatively, adding further lags of the dependent and or independent variables as regressors to the model often helps to eliminate residual serial correlation.

More on GMM for linear dynamic panel models:

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