What is the syntax to retrieve r-class stored results in multiple equation models, like bivariate poisson, var, or sure?

[Oscar Ozfidan]

I can retrieve coefficient estimates by:
gen Cf = [equationname]_b[varname]
,no problem!
But I am having trouble retrieving r-class results. I tried gen Cf = [equationname]_r[mean], Cf = [equationname]_mean, gen Cf = _r(mean1) but none of them worked.
Also, if not the same, what is the syntax to retrieve e-class stored results in multiple equation models?
Thanks

[Andrew Musau]

You need to be specific about the names of the commands and state the source if these are user-written commands. To access the return results stored in r(), type

return list

For estimation results:

ereturn list

[Oscar Ozfidan]

OK. How do you get the mean of the sample for the first equation (or equation named "FirstVariable") in a vector auto regression which is not user-written. When I type "return list" i see them but how do I retrieve them as a scalar or as a variable? When I say "gen samplemean1 = ?" what do I write in where "?" appears ? I I could not find any documentation on the syntax for multiple equation models.

[Andrew Musau]

Can you copy and paste the output of -return list- and point to what you want?

[Joro Kolev]

What -r(table)- contains is the same as those coefficients and standard errors that you know how to refer to, except that there are some more derived statistics like p-values.

So either refer to the coefficients, or just refer to the element of the matrix that you need.

[Joro Kolev]

. webuse lutkepohl2, clear
(Quarterly SA West German macro data, Bil DM, from Lutkepohl 1993 Table E.1)

. var dln_inv dln_consump

Vector autoregression

Sample:  1960q4 - 1982q4                        Number of obs     =         89
Log likelihood =   439.6021                     AIC               =   -9.65398
FPE            =   2.20e-07                     HQIC              =  -9.541273
Det(Sigma_ml)  =   1.76e-07                     SBIC              =  -9.374358

Equation           Parms      RMSE     R-sq      chi2     P>chi2
----------------------------------------------------------------
dln_inv               5     .043901   0.0995   9.837951   0.0432
dln_consump           5     .010679   0.1010   9.996723   0.0405
----------------------------------------------------------------

------------------------------------------------------------------------------
             |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
dln_inv      |
     dln_inv |
         L1. |  -.2783987   .1093257    -2.55   0.011    -.4926731   -.0641243
         L2. |  -.1251467   .1080765    -1.16   0.247    -.3369727    .0866794
             |
 dln_consump |
         L1. |   .8989791   .4343833     2.07   0.038     .0476034    1.750355
         L2. |   .7896433   .4455018     1.77   0.076    -.0835241    1.662811
             |
       _cons |  -.0080688   .0123968    -0.65   0.515    -.0323662    .0162285
-------------+----------------------------------------------------------------
dln_consump  |
     dln_inv |
         L1. |  -.0017619   .0265929    -0.07   0.947     -.053883    .0503591
         L2. |   .0545457    .026289     2.07   0.038     .0030202    .1060712
             |
 dln_consump |
         L1. |   -.017519   .1056614    -0.17   0.868    -.2246114    .1895735
         L2. |   .1657664   .1083659     1.53   0.126    -.0466268    .3781596
             |
       _cons |   .0148851   .0030155     4.94   0.000     .0089749    .0207953
------------------------------------------------------------------------------

. return list

scalars:
              r(level) =  95

matrices:
              r(table) :  9 x 10

. mat Rtable = r(table)

. sca Brtable = Rtable[1,1]

. sca Bcoeff = [dln_inv]_b[l.dln_inv]

. sca list
    Bcoeff = -.27839867
   Brtable = -.27839867

[Oscar Ozfidan]

Figured it out. "gen rmse1= e(rmse_1)" works. Regardless of the equation name rmse of the first equation is referred to as rmse_1 and second as as rmse_2.

. ereturn list

scalars:
e(N) = 66
e(df_eq) = 5
e(k) = 10
e(obs_1) = 66
e(k_1) = 5
e(rmse_1) = 50.45066250639565
e(r2_1) = .9916945830794686
e(ll_1) = -349.8359022135093
e(df_m1) = 4
e(obs_2) = 66
e(k_2) = 5
e(rmse_2) = .7366700545715245
e(r2_2) = .1508585972749652
e(ll_2) = -70.87957272626753
e(df_m2) = 4
e(chi2_1) = 7880.6209380514
e(chi2_2) = 11.72557054478299
e(ll) = -420.0144702859178
e(rank) = 10
e(sbic) = 13.36250739381969
e(hqic) = 13.16183798747264
e(aic) = 13.03074152381569
e(fpe) = 1564.887340845201
e(mlag) = 2
e(tmax) = 1343
e(tmin) = 1278
e(N_gaps) = 0
e(k_dv) = 2
e(k_eq) = 2
e(detsig) = 1155.117198033127
e(detsig_ml) = 1155.117198033127

macros:
e(cmd) : "var"
e(cmdline) : "var Sira Res"
e(predict) : "var_p"
e(marginsdefault) : "predict(xb equation(Sira)) predict(xb equation(Res))"
e(marginsnotok) : "stdp Residuals"
e(marginsok) : "xb default"
e(lags) : "1 2"
e(endog) : "Sira Res"
e(timevar) : "order1"
e(tsfmt) : "%9.0g"
e(title) : "Vector autoregression"
e(eqnames) : "Sira Res "
e(depvar) : "Sira Res"
e(properties) : "b V"

matrices:
e(b) : 1 x 10
e(V) : 10 x 10
e(bf) : 1 x 10
e(G) : 5 x 5
e(Sigma) : 2 x 2

functions:
e(sample)

[Joro Kolev]

Good that you figured it out, two remarks from the pedants' corner:

1. What you called "the mean of the sample" in #3, such a concept does not exist in statistics/econometrics. Further the root mean squared error is hard to be imagined to be mean of the sample, but is rather what the name says. The square root, of the mean of the squared error.

3. Assigning a scalar (the root mean squared error is a scalar quantity) to a whole variable (=whole data vector of repeated same value) is a bit of a wasteful practice. I dont know what you are about to do with this next, but probably it is better if you assign the root mean scared error to a scalar, rather than a variable.

[Oscar Ozfidan]

Joro, I am new to Stata did not know dep var means not stored that way.

[Joro Kolev]

Joro, I am new to Stata did not know dep var means not stored that way.

If you need the mean of the dependent variable calculated over the estimation sample, this quantity is not in e(rmse).

To get the mean of the dependent variable over the sample you do:

summ y if e(sample)

and then you retrieve the r(mean) from the return list.

Like this:

. sysuse auto, clear
(1978 Automobile Data)

. reg price mpg headroom

      Source |       SS           df       MS      Number of obs   =        74
-------------+----------------------------------   F(2, 71)        =     10.44
       Model |   144280501         2  72140250.4   Prob > F        =    0.0001
    Residual |   490784895        71  6912463.32   R-squared       =    0.2272
-------------+----------------------------------   Adj R-squared   =    0.2054
       Total |   635065396        73  8699525.97   Root MSE        =    2629.2

------------------------------------------------------------------------------
       price |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         mpg |  -259.1057   58.42485    -4.43   0.000    -375.6015   -142.6098
    headroom |  -334.0215   399.5499    -0.84   0.406    -1130.701    462.6585
       _cons |   12683.31   2074.497     6.11   0.000     8546.885    16819.74
------------------------------------------------------------------------------

. summ price if e(sample)

    Variable |        Obs        Mean    Std. Dev.       Min        Max
-------------+---------------------------------------------------------
       price |         74    6165.257    2949.496       3291      15906

. return list

scalars:
                  r(N) =  74
              r(sum_w) =  74
               r(mean) =  6165.256756756757
                r(Var) =  8699525.974268788
                 r(sd) =  2949.495884768919
                r(min) =  3291
                r(max) =  15906
                r(sum) =  456229

. sca MeanPrice = r(mean)


. sca list MeanPrice
 MeanPrice =  6165.2568