I am running a two-step dynamic difference GMM with a pilot Stata command. This command calculate estimates and standard errors but does not calculate the Hansen j test of overidentifying restrictions. The pilot command builds on xtabond2 in terms of syntax. I am estimating something like this:
Code:
newabond y l.y l.x l.k yeardum*, gmm(y k x, lag(2 6) collapse) iv(yeardum*) noleveleq small two noconstant
Code:
estat overid
I have access to the original data and the following information, do you believe it is possible to calculate the Hansen j statistic and p-value with this information?
HTML Code:
Scalars e(N) Number of complete observations in untransformed data (system GMM) or transformed data (difference GMM) e(N_g) Number of included individuals e(g_min) Lowest number of observations in an included individual e(g_max) Highest number of observations in an included individual e(g_avg) Average number of observations per included individual e(iterations) Number of iterations after iterated GMM estimation e(chi2) Wald statistic (if small not specified) e(chi2p) p-value of Wald statistic (if small not specified) e(F) F-statistic e(Fp) p-value of F-statistic Macros e(ivinstsi) Variables listed in ivstyle group i e(gmminstsi) Variables listed in gmmstyle group e(tvar) Time variable e(ivar) Individual (panel) variable e(depvar) Dependent variable e(xvars) List of regressors e(properties) "system" or "difference" e(small) "small" for small Matrices e(b) Coefficient vector e(V) Variance-covariance matrix of the estimators
If anyone could tell me the steps I need in order to calculate the Hansen j I would be extremely grateful.
I thank you a lot in advance for your help
0 Response to GMM: Calculating Hansen J statistic and p-value by hand
Post a Comment