Test: ¿Biases in Long run coefficients in a Dynamic Panel Data?

[Manuel Almodovar]

Dear stata users:

I am learning the dynamic panel data models to use them in my research (at the moment with the xtabond2 command). To enrich the analysis I would like to include the Long run coefficients (in principle using the nlcom command).

The problem is that I have found a paper (quote at the end) that warns about the bias of the long run coefficients. In this respect my questions are:
1. Are the long run coefficients obtained with an xtabond2 and nlcom reliable?
2. Is it advisable to use the long run coefficients in publications or are they usually not relevant?

I indicate the paper mentioned above: W.R. Reed & M. Zhu (2016) On Estimating Long-Run Effects In Models with Lagged Dependent Variables. WORKING PAPER No. 16/2016

[Manuel Almodovar]

xtabond2 GDP07 L.GDP07 Grad10 Grad18  i.year,  gmm (l.GDP07  Grad10 Grad18  , collapse lag (1    5))    iv    (i.year,    eq(level))    small    twostep    robust    nodiffsargan
Favoring speed over space. To switch, type or click on mata: mata set matafavor space, perm.
2014b.year dropped due to collinearity
2016.year dropped due to collinearity
Warning: Two-step estimated covariance matrix of moments is singular.
Using a generalized inverse to calculate optimal weighting matrix for two-step estimation.