Diagnostic tests for spatial dependence in Fixed Effects regression

[Kodi Hannon]

Hello everybody

I am kindly asking you some help inherent spatial econometrics.

I have to derive the Lagrange Multipliers values inherent spatial dependence for a fixed effects (xtreg) regression (in practice, to test whether a non spatial model is preferable to a spatial one).

Now, there already exists such a command (spatdiag) but it only works for simple OLS regression.

I was hence wondering if there exists an analogous command but for the xtreg (FE) regression (since I was not able to find anything).

Thank you very much!

Kodi

[Jorge Eduardo Perez Perez]

You could use the least squares dummy variable estimator, i.e. include dummy variables for the fixed effects instead of using xtreg, and use spatdiag afterwards.

[Kodi Hannon]

Hi Jorge, thank you for your comment.

Unfortunately, when I try with the LSDV estimator, spatdiag does not work, since it gives me the following error:

factor variables and time-series operators not allowed
r(101);

[Jorge Eduardo Perez Perez]

Try using the LSDV estimator without using factor variables, by using the xi command instead. You could also try using areg

clear
set matsize 5000
webuse nlswork, clear
keep in 1/100
xtset idcode
* xtreg
xtreg ln_w age  ttl_exp tenure not_smsa south, fe
* LSDV with factor variables
reg ln_w age  ttl_exp tenure not_smsa south i.idcode
* LSDV without factor variables
xi i.idcode
reg ln_w age  ttl_exp tenure not_smsa south _I*
* areg version
areg ln_w age  ttl_exp tenure not_smsa south, absorb(idcode)

[Kodi Hannon]

Thank you very much Jorge! Now it works!

Thank you!!!!!!!!!

[Souryabrata Mohapatra]

I used the above but it said

xtreg $ylist $xlist, fe
reg $ylist $xlist i.censuscode
reg $ylist $xlist _I*

spmat summarize idw_weights
Summary of spatial-weighting object idw_weights
------------------------------------------------
        Matrix |                    Description
---------------+--------------------------------
    Dimensions |                      536 x 536
     Stored as |                      536 x 536
 Values        |
           min |                              0
         min>0 |                       .0164724
          mean |                       .0018657
           max |                              1
------------------------------------------------

spatdiag, weights(idw_weights)
Matrix idw_weights does not exist

Please advise.