The Kiviet estimator is not a very good strategy (his own caveat)
because it assumes that everything other than the lagged dep. var.
(which is included by default) is strictly exogenous. See also:
ftp://ftp.unibocconi.it/pub/RePEc/cri/papers/WP165Bruno.pdf
To run the -xtlsdvc- with -xtreg, fe- estimates, do the following:
/* Begin */
webuse nlswork, clear
xtset idcode year
// ssc install xtlsdvc, replace
g lag_lnw = L.ln_w
// inconsistent fixed effects estimator
xtreg ln_w lag_lnw age ttl_exp tenure not_smsa south, fe
mat my = e(b)
// note that 1000 bs reps could take days
xtlsdvc ln_w age ttl_exp tenure not_smsa south, initial(my) bias(3) vcov(1000)
/* End */
T
On Sat, May 2, 2009 at 1:59 PM, olfa alouini <[email protected]> wrote:
> Dear all,
>
> I have a panel of 15 countries, 40 years of data. I want to estimate the effect of a number of variables on growth rates.
>
> After running xtserial I believe I have an autocorrelation problem. I have pasted the output in case I interpreted this wrongly:
> xtserial GDPgrowth_copr Population Open Output_thkm Delta_inf Deltadef
> Wooldridge test for autocorrelation in panel data
> H0: no first-order autocorrelation
> F( 1, 10) = 141.386
> Prob> F = 0.0000
>
>
>
> To address this issue, I want to run a dynamic. However including a lagged endogenous variable (in this case growth [ i t-1] makes the FE estimator I'd like to use inefficient. I know Kiviet (1995) proposed a way to correct the bias induced by the inclusion of the lagged variable. How can I implement this correction with Stata?
>
> I tried with the -xtlsdvc- command, but the initial estimation has to be either ah, xtabond or xtabond2, not xtreg,fe. I tried to retrieve my results from xtreg, fe and put them in the "my" matrix form but did not succeed. Any help would be greatly appreciated.
>
> In addition to that, I was wondering if I had to change my estimation strategy if I estimated my model for only ten years of data (in this case N>T), or is the bias corrected FE estimator still valid and more efficient than GMM for instance?
>
> Also how many dependent variables can be included in xtlsdvc? and is the lagged variable automatically included?
>
> Thanks in advance,
>
> Olfa
>
> _________________________________________________________________
> Hotmail® has a new way to see what's up with your friends.
> http://windowslive.com/Tutorial/Hotmail/WhatsNew?ocid=TXT_TAGLM_WL_HM_Tutorial_WhatsNew1_052009
>
>
> *
> * For searches and help try:
> * http://www.stata.com/help.cgi?search
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
--
To every ω-consistent recursive class κ of formulae there correspond
recursive class signs r, such that neither v Gen r nor Neg(v Gen r)
belongs to Flg(κ) (where v is the free variable of r).
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
