error 102 "too few variables specified"

[Mateus Ferreira Lima]

Dear statalist members,

I´m running an econometric analysis using the following code: "ivreg2 dv (x z1 z2) w, liml"

The full command I´m typing on the prompt is (in accordance to the sintaxe above):
ivreg2 coop1(Costs Knowledge Market Regulation = IV_POLICIES IV_POLICIES IV_ORGAN_FAILURE IV_MARK_FAILURE IV_PP_FAILURE IV_RD_SOURCES IV_MARKET_SOURCES IV_INST_SOURCES IV_OTHER_SOURCES IV_RD_FOUNDS IV_OTHER_FOUNDS) Dummy_Natural Dummy_Labor Dummy_Scale Firms_Size, liml.

However, It is appearing the following message on my screen too few variables specified r (102) and I don´t know how to proceed from this.

I would greatly appreciate any help on this.

Best regards,
Mateus

[Roman Mostazir]

Show us the exact command you have typed in Stata. Your commands here showing line breaks without continuation forward slashes "///" and if this is the exact command you typed Stata will not pick it up as a complete instruction. On the point, you have incorporated IV_POLICIES twice as instrument. I had no problem running your command on a hypothetical data:

/ / Generate hypothetical data

loc all coop1 Costs Knowledge Market Regulation ///
IV_POLICIES IV_ORGAN_FAILURE ///
IV_MARK_FAILURE IV_PP_FAILURE IV_RD_SOURCES ///
IV_MARKET_SOURCES IV_INST_SOURCES IV_OTHER_SOURCES ///
IV_RD_FOUNDS IV_OTHER_FOUNDS
set obs 10000
set seed 37898
foreach var of local all {
gen `var'= runiform()+5

}
loc x Dummy_Natural Dummy_Labor Dummy_Scale Firms_Size


foreach var of loc x {
gen `var'=runiform()>.5
}
*******************************

//Run the model:

ivreg2 coop1 (Costs Knowledge Market Regulation = IV_POLICIES ///
IV_POLICIES IV_ORGAN_FAILURE IV_MARK_FAILURE IV_PP_FAILURE ///
IV_RD_SOURCES IV_MARKET_SOURCES IV_INST_SOURCES IV_OTHER_SOURCES IV_RD_FOUNDS IV_OTHER_FOUNDS) ///
Dummy_Natural Dummy_Labor Dummy_Scale Firms_Size, liml

//Results:

Warning - duplicate variables detected
Duplicates:         IV_POLICIES

LIML estimation
---------------
k               =1.00004
lambda          =1.00004

Estimates efficient for homoskedasticity only
Statistics consistent for homoskedasticity only

                                                      Number of obs =    10000
                                                      F(  8,  9991) =     0.03
                                                      Prob > F      =   1.0000
Total (centered) SS     =  856.1766542                Centered R2   = -64.9345
Total (uncentered) SS   =  303292.1924                Uncentered R2 =   0.8139
Residual SS             =  56451.60719                Root MSE      =    2.376

-------------------------------------------------------------------------------
        coop1 |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------+----------------------------------------------------------------
        Costs |   3.611278   10.64135     0.34   0.734    -17.24538    24.46793
    Knowledge |  -1.253558   4.241273    -0.30   0.768    -9.566301    7.059184
       Market |  -1.368845    4.97499    -0.28   0.783    -11.11965    8.381955
   Regulation |  -7.178747   20.06962    -0.36   0.721    -46.51449    32.15699
Dummy_Natural |  -.0189444   .0930717    -0.20   0.839    -.2013616    .1634728
  Dummy_Labor |    -.01491   .0818909    -0.18   0.856    -.1754133    .1455933
  Dummy_Scale |  -.0803632    .254717    -0.32   0.752    -.5795994     .418873
   Firms_Size |   .0159943   .0917247     0.17   0.862    -.1637827    .1957713
        _cons |   39.57171   96.60697     0.41   0.682    -149.7745    228.9179
-------------------------------------------------------------------------------
Underidentification test (Anderson canon. corr. LM statistic):           0.560
                                                   Chi-sq(7) P-val =    0.9992
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic):                0.056
Stock-Yogo weak ID test critical values:                       <not available>
------------------------------------------------------------------------------
Sargan statistic (overidentification test of all instruments):           0.431
                                                   Chi-sq(6) P-val =    0.9986
------------------------------------------------------------------------------
Anderson-Rubin statistic (overidentification test of all instruments):   0.431
                                                   Chi-sq(6) P-val =    0.9986
------------------------------------------------------------------------------
Instrumented:         Costs Knowledge Market Regulation
Included instruments: Dummy_Natural Dummy_Labor Dummy_Scale Firms_Size
Excluded instruments: IV_POLICIES IV_ORGAN_FAILURE IV_MARK_FAILURE
                      IV_PP_FAILURE IV_RD_SOURCES IV_MARKET_SOURCES
                      IV_INST_SOURCES IV_OTHER_SOURCES IV_RD_FOUNDS
                      IV_OTHER_FOUNDS
Duplicates:           IV_POLICIES

[William Lisowski]

When I look at help ivreg2 I see the following syntax:

        ivreg2 depvar [varlist1] (varlist2=varlist_iv) [weight] [if exp] [in
              range] [, gmm2s bw(#) kernel(string) dkraay(integer) kiefer liml
              fuller(#) kclass(#) coviv cue b0(matrix) robust cluster(varlist)
              orthog(varlist_ex) endog(varlist_en) redundant(varlist_ex)
              partial(varlist) small noconstant center smatrix(matrix)
              wmatrix(matrix) first ffirst savefirst savefprefix(prefix) sfirst
              savesfirst savesfprefix(prefix) rf saverf saverfprefix(prefix)
              nocollin noid level(#) bvclean noheader nofooter eform(string)
              depname(varname) plus ]

which suggests to me that the variables you include after the parenthesized list of instrumental variables should instead be placed before the list, just after the dependent variable, as in the following.

ivreg2 coop1 Dummy_Natural Dummy_Labor Dummy_Scale Firms_Size (Costs Knowledge Market Regulation = IV_POLICIES IV_POLICIES IV_ORGAN_FAILURE IV_MARK_FAILURE IV_PP_FAILURE IV_RD_SOURCES IV_MARKET_SOURCES IV_INST_SOURCES IV_OTHER_SOURCES IV_RD_FOUNDS IV_OTHER_FOUNDS), liml

But I could be wrong, I'm in no position to test this. But it can't hurt to try!

Crossed with post #2, which shows that the syntax used can work, and points out a more likely source of the problem.