Please Help - The Wald chi2 model statistic has been reported as missing ...

[Sara Jaa]

Hi,
I use panel regression regression with robust std error but STATA reports the following messages;

1.The Wald chi2 model statistic has been reported as missing.
2.corr(u_i, X) = 0 (assumed) >> Is there any problem with the model?
3.sigma_u and rho are zero
4.when command e(r2_a) following outreg2, STATA reports to check eret list and I found there's no e(r2_a) in the list. What should I do If I want the Adjusted R2

Commands and results are in the attachment.

Many Thanks
Sara

[Joseph Coveney]

What should I do If I want the Adjusted R2

Maybe:

1. Don't use xtreg, and

2. Don't use robust

You can get adjusted R² with regress and entering your (fixed effects) panel variable as a factor variable.

[William Lisowski]

Welcome to Statalist, Sara.

In general, posting Word files is discouraged on Statalist, and a substantial number of members either do not have Microsoft products or do not open Microsoft documents from unknown sources. Below I have posted the contents of your attachment in a form readable by everyone.

For future reference, please review the Statalist FAQ linked to from the top of the page, as well as from the Advice on Posting link on the page you used to create your post. See especially sections 9-12 on how to best pose your question. It's particularly helpful to copy commands and output from your Stata Results window and paste them into your Statalist post using CODE delimiters, as I did below, and to use the dataex command to provide sample data, as described in section 12 of the FAQ.

. xtreg ROA IBac IBac2 lnA Loan Dep OBS TRD NPL NII NOE Prov HC PLT FLD, robust
note: IBac2 omitted because of collinearity
 
Random-effects GLS regression                   Number of obs      =       200
Group variable: Bank1                           Number of groups   =        10
 
R-sq:  within  = 0.3373                         Obs per group: min =        20
       between = 0.9251                                        avg =      20.0
       overall = 0.5745                                        max =        20
 
                                                Wald chi2(9)       =         .
corr(u_i, X)   = 0 (assumed)                    Prob > chi2        =         .
 
                                 (Std. Err. adjusted for 10 clusters in Bank1)
------------------------------------------------------------------------------
             |               Robust
         ROA |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
       IBac2 |  -.6349743   .5325246    -1.19   0.233    -1.678703    .4087547
       IBac2 |          0  (omitted)
         lnA |   .1580466   .0396991     3.98   0.000     .0802377    .2358554
        Loan |  -1.432578   1.196068    -1.20   0.231    -3.776829    .9116731
         Dep |   -.859116   .4353156    -1.97   0.048    -1.712319    -.005913
         OBS |   .1319344   .1428143     0.92   0.356    -.1479765    .4118453
         TRD |  -.0953054   .0340292    -2.80   0.005    -.1620013   -.0286094
         NPL |  -13.36173    4.16206    -3.21   0.001    -21.51922   -5.204238
         NII |   359.0665   42.54074     8.44   0.000     275.6881    442.4448
         NOE |  -.1685176   .0704297    -2.39   0.017    -.3065572    -.030478
        Prov |     -100.4   56.21452    -1.79   0.074    -210.5784    9.778461
          HC |   -.225829     .05348    -4.22   0.000    -.3306478   -.1210101
         PLT |   .0838018   .1519869     0.55   0.581    -.2140871    .3816906
         FLD |  -.1073445   .0949364    -1.13   0.258    -.2934165    .0787275
       _cons |  -.6895448   .7503506    -0.92   0.358    -2.160205    .7811154
-------------+----------------------------------------------------------------
     sigma_u |          0
     sigma_e |  .46590246
         rho |          0   (fraction of variance due to u_i)
------------------------------------------------------------------------------
 
. outreg2 using rescale1.doc, replace ctitle(OLS) addstat (Adjusted R-squared, e(r2_a))
check eret list for the existence of e(r2_a)
invalid syntax
r(198);
 
end of do-file
 
r(198);
 
. eret list
 
scalars:
            e(N_clust) =  10
               e(rank) =  9
               e(df_m) =  9
               e(chi2) =  .
                  e(p) =  .
            e(sigma_u) =  0
            e(sigma_e) =  .4659024606051799
              e(sigma) =  .4659024606051799
                e(rho) =  0
               e(rmse) =  .4859895636219217
                  e(N) =  200
               e(Tbar) =  20
               e(Tcon) =  1
                e(N_g) =  10
              e(g_min) =  20
              e(g_avg) =  20
              e(g_max) =  20
              e(theta) =  0
               e(r2_o) =  .5745486554892864
               e(r2_b) =  .9251134891397371
               e(r2_w) =  .3373263960345694
 
macros:
         e(properties) : "b V"
           e(chi2type) : "Wald"
             e(depvar) : "ROA"
            e(vcetype) : "Robust"
                e(vce) : "robust"
           e(clustvar) : "Bank1"
               e(ivar) : "Bank1"
              e(model) : "re"
            e(predict) : "xtrere_p"
       e(marginsnotok) : "E U UE SCore STDP XBU"
                e(cmd) : "xtreg"
            e(cmdline) : "xtreg ROA IBac IBac2 lnA Loan Dep OBS TRD NPL NII NOE Prov HC PLT FLD, robust"
 
matrices:
                  e(b) :  1 x 15
                  e(V) :  15 x 15
                 e(bf) :  1 x 15
               e(VCEf) :  15 x 15
 
functions:
             e(sample)

[William Lisowski]

Looking at the output posted above, it appears that the xtreg command specified what were supposed to be two distinct independent variables IBac and IBac2. However, Stata did not find IBac and thus treated IBac as an abbreviation for IBac2. This explains why IBac2 appears twice in the coefficients table, the second time dropped due to collinearity.

A first step toward answering questions 1-3 would be to respecify the model correctly - either omitting IBac, or replacing IBac with the correctly-named variable, or ensuring that IBac is in the dataset, and re-running the xtreg.

[Carlo Lazzaro]

Sara:
Joseph and William provided helpul advice.
I would only add that the missing Wald chi2 statsistic is due to the fact that the number of your constraints (coefficients) is greater than the number of your clusters (see -help j_robustsingular- for more details) .

[Sara Jaa]

Maybe:

1. Don't use xtreg, and

2. Don't use robust

You can get adjusted R² with regress and entering your (fixed effects) panel variable as a factor variable.

Dear Joseph,

To get adjusted r2, I have no idea how to use regress command and entering panel variables as a factor variable? I wonder, if I use the plain "regress y x" command to get the plain OLS estimates using the panel data I have, will that regression be correct? Or do I need to instruct Stata with a special command to run the plain OLS using my panel data?

Best regards,
Sara

[Carlo Lazzaro]

Sara:
a missing Wald chi2 model statistic is also a warning about rethinking your regression model.
In your cae, it seems that you're asking too much out of your data inn terms of clusters and coefficients (see -help j_robustsingular- for more details).

[Joseph Coveney]

To get adjusted r2, I have no idea how to use regress command and entering panel variables as a factor variable?

regress ROA /* IBac */ IBac2 lnA Loan Dep OBS TRD NPL NII NOE Prov HC PLT FLD i.Bank1

I'm not sure that the Adjusted R² that you get will be what you want, though, because it includes the indicator variables for your grouping variable (Bank1). Perhaps that's why xtreg doesn't report Adjusted R².

[Andrew Musau]

I'm not sure that the Adjusted R² that you get will be what you want, though, because it includes the indicator variables for your grouping variable (Bank1). Perhaps that's why xtreg doesn't report Adjusted R².

Joseph makes an important point about the adjusted R2 statistic in -xtreg, fe- with the inclusion of the firm dummies available as r2_a after typing ereturn list. It is meaningless (and in a lot of cases negative). However, there are instances where you may have quite a large number of regressors in your model and may want to compare the within R2 to the within adjusted R2 not including the firm dummies. You should use the following formula after -xtreg- to obtain the appropriate adjusted R2:

scalar R2_a = e(r2_w) - ((1- e(r2_w))* (e(df_b) / (e(N)-e(df_b)-1)))
di R2_a

The justification is simple: the within R2 is the R2 statistic from the regression of the within-mean deviated variables. The same regression reports an adjusted R2 which adjusts for addition of variables into the model. The following example using the Grunfeld dataset illustrates:

webuse grunfeld
xtreg invest mvalue kstock, fe
scalar R2_a = e(r2_w) - ((1- e(r2_w))* (e(df_b) / (e(N)-e(df_b)-1)))
di R2_a    

*\\generate within-means of variables
local vars "invest mvalue kstock"
foreach x of local vars{
bys company: egen `x'm= mean(`x')
}

*\\generate mean-deviated variables
foreach x of local vars{
gen `x'd= `x'm-`x'
}

reg investd mvalued kstockd

Output:

. xtreg invest mvalue kstock, fe

Fixed-effects (within) regression               Number of obs     =        200
Group variable: company                         Number of groups  =         10

R-sq:                                           Obs per group:
     within  = 0.7668                                         min =         20
     between = 0.8194                                         avg =       20.0
     overall = 0.8060                                         max =         20

                                                F(2,188)          =     309.01
corr(u_i, Xb)  = -0.1517                        Prob > F          =     0.0000

------------------------------------------------------------------------------
      invest |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      mvalue |   .1101238   .0118567     9.29   0.000     .0867345    .1335131
      kstock |   .3100653   .0173545    17.87   0.000     .2758308    .3442999
       _cons |  -58.74393   12.45369    -4.72   0.000    -83.31086     -34.177
-------------+----------------------------------------------------------------
     sigma_u |  85.732501
     sigma_e |  52.767964
         rho |  .72525012   (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(9, 188) = 49.18                     Prob > F = 0.0000

. 
. scalar R2_a = e(r2_w) - ((1- e(r2_w))* (e(df_b) / (e(N)-e(df_b)-1)))

. 
. di R2_a    
.76438965


. reg investd mvalued kstockd

      Source |       SS           df       MS      Number of obs   =       200
-------------+----------------------------------   F(2, 197)       =    323.81
       Model |   1720874.1         2   860437.05   Prob > F        =    0.0000
    Residual |  523478.127       197  2657.24938   R-squared       =    0.7668
-------------+----------------------------------   Adj R-squared   =    0.7644
       Total |  2244352.23       199  11278.1519   Root MSE        =    51.549

------------------------------------------------------------------------------
     investd |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
     mvalued |   .1101238   .0115827     9.51   0.000     .0872818    .1329658
     kstockd |   .3100653   .0169534    18.29   0.000     .2766318    .3434989
       _cons |  -1.93e-08    3.64503    -0.00   1.000    -7.188288    7.188288
------------------------------------------------------------------------------

[Sara Jaa]

I am really appreciate with your advises, Thank you so much

Best regards,
Sara