Re: st: xtabond2 - variables in level

[Tomasz Boniek <tboniek@studbocconi.it>]

Hi Statalisters,

one users already gave me an answer which was not what I was looking for 
because I stated my problem in the wrong way. I will try to explain my 
problem more carefully.

Suppose I want to estimate a basic Cobb Douglas production function. 
Please ignore mispecification of the model or endogeneity issues, since 
here I will keep the model oversimplified on purpose. The model I want 
to estimate is the following:

y_it = B1*y_i(t-1) + B2*k_it + B3*l_it + B4*w_it + u_i + e_it

where y is log of added value, k and l are logs of capital and labor and 
w represents unobservable total factor productivity (TFP) and u_i is the 
fixed effect. Now, if I take first differences, I would obtain:

D(y_it)= B1*D(y_i(t-1)) + B2*D(k_it) + B3*D(l_it) + B4*D(w_it) + D(e_it).

where D(...) means first difference. Being the TFP unobservable, one has 
to proxy for it. I will proxy for it by taking a variable called /rents/ 
lagged one period, which is a firm-level measure of competition. 
However, I want to get a coefficient for the level of rents rather than 
its first difference. The model I would want to estimate should be

D(y_it)= B_t + B1*D(y_i(t-1)) + B2*D(k_it) + B3*D(l_it) + 
B4*rents_i(t-1) + D(e_it).

The reference papers I am using estimated this function using DPD98 for 
Gauss. However, I want to find out whether it is possible to estimate 
that function using xtabond2. In Stata, I would write

xi: xtabond2 y yL1 k l rentsL1 i.year, gmm(yL1) iv(i.year k l rentsL1) 
noleveleq small

where all the variable are in logs, and *L1 means one period lag. If I 
am not wrong, if I do so the coefficient I get for rentsL1 will be on 
its first difference (i.e. rentsL1 - rentsL2). Is there a way to have 
the coefficient for level of rents, keeping all the other variables in 
first difference?

I hope someone is able to help me out!

Tomasz Boniek

Tomasz Boniek wrote:

> Hi,
>
> I have a problem with xtabond2. Suppose I run the following:
>
> xtabond2 ln_tfp L.ln_tfp L.rents yr*, gmm(L.ln_tfp L.rents, lag(2 3))  
> iv(yr*) robust noleveleq small
>
> where yr* are a series of year dummies and the economic meaning of the 
> other variable is irrelevant for my question.
>
> In this case, if I'm not wrong, I will estimate a one-step difference 
> GMM. All the variables in this case will be in first difference. My 
> question is:
> Is there a way to avoid differentiation of the variable L.rents? I 
> would like to use as a regressors (treating it as endogenous) not the 
> first difference lagged one period (i.e. rents(t-1)-rents(t-2)) but 
> just the lagged level (i.e. rents(t-1)). Is there a way to do that in 
> Stata?
>
> Thanks for the help,
>
> Tomasz Boniek
>
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/