Hello everyone
I am currently trying to do an evaluation of the effect of access to technological extension services on the productivity of firms in a developing country. The literature on technological extensionism highlights that access to extension services can potentially be endogenous (unobservables simultaneously affect the firm's decision to seek extension services and its performance).
My database however is quite small and comes from an innovation survey executed by an official agency that collects different characteristics of a sample of firms. Within this database, the number of firms that during the period sought extension services is only 49 out of a total of 1,289 observations in the sample.
So far I have used eteffects in STATA version 16.0 to control for treatment endogeneity and applied the endogeneity test afterwards. The test results show that it is not possible to reject the null hypothesis so there is no endogeneity and I could apply other methods such as IPWRA. However, when reviewing the options I found the extended regression models (ERM) that also allow to calculate the ATE and ATET. So I applied the same model as in eteffects, specifying that the treatment is endogenous. The results in this case prove the treatment is indeed endogenous and, moreover, the estimate for the ATET is totally different than the one calculated by eteffects.
Could someone explain to me why the difference in the results of the two approaches?
Thank you for any guidance
Sam
1. ETEFFECTS
2. EREGRESS
I attach a sample of my database (sorry, the variable labels are in Spanish).
I am currently trying to do an evaluation of the effect of access to technological extension services on the productivity of firms in a developing country. The literature on technological extensionism highlights that access to extension services can potentially be endogenous (unobservables simultaneously affect the firm's decision to seek extension services and its performance).
My database however is quite small and comes from an innovation survey executed by an official agency that collects different characteristics of a sample of firms. Within this database, the number of firms that during the period sought extension services is only 49 out of a total of 1,289 observations in the sample.
So far I have used eteffects in STATA version 16.0 to control for treatment endogeneity and applied the endogeneity test afterwards. The test results show that it is not possible to reject the null hypothesis so there is no endogeneity and I could apply other methods such as IPWRA. However, when reviewing the options I found the extended regression models (ERM) that also allow to calculate the ATE and ATET. So I applied the same model as in eteffects, specifying that the treatment is endogenous. The results in this case prove the treatment is indeed endogenous and, moreover, the estimate for the ATET is totally different than the one calculated by eteffects.
Could someone explain to me why the difference in the results of the two approaches?
Thank you for any guidance
Sam
1. ETEFFECTS
Code:
gl treat lnexport15 lnL15 c.age##c.age skill_15 i.grupo15 i.Qkextrj15 i.Acc i.provcite cuota i.div eteffects (lnDprod lnDkfijo lnDL i.div i.size_2015) (cite $treat i.size_2015, probit), atet control(0) //aequation vce(robust) estat endogenous
Code:
eregress lnDprod lnDkfijo lnDL i.div i.size_2015, entreat(cite = $treat i.size_2015) estat teffects, atet
Code:
* Example generated by -dataex-. To install: ssc install dataex
clear
input float(lnDprod lnDkfijo lnDL size_2015 cite lnexport15 lnL15 age skill_15 grupo15 Qkextrj15 Acc_pub provcite cuota div)
-.021850586 -.33870125 1 3 0 0 4.2195077 13 .29850745 0 1 0 1 .0025692764 23
-1.0519571 0 4 1 0 0 2.1972246 8 .25 0 1 0 1 .00013030018 25
.4869909 1.4994154 0 3 0 0 5.204007 42 .0883978 0 3 0 1 .002718331 71
-.14678574 9.583786 0 2 0 11.78296 2.995732 30 .2105263 0 1 0 1 .00011448725 22
.05129337 .05129147 -1 2 0 16.96421 3.0445225 8 .1 0 1 0 1 .0020406202 24
-1.0468416 -.5463896 7 1 0 0 2.397895 8 0 0 1 0 1 .0014742848 32
-.8403025 -.4627581 0 1 0 0 2.0794415 7 0 0 1 0 1 .00008931365 25
-.5025492 0 0 1 0 0 1.609438 10 0 0 1 0 0 .00025951708 16
.07564926 .27859974 -5 2 0 13.66666 3.713572 46 .5 0 1 0 1 .0025356836 20
-.14143372 0 25 2 1 17.315104 3.0445225 24 .2 0 1 0 1 .0006784001 10
-.07459164 0 0 1 0 0 2.397895 40 0 1 1 0 1 .00005121665 23
.3238068 .5459385 5 3 0 0 4.3307333 14 .2 0 1 0 1 .0036374955 32
-.18734646 .13291168 53 4 0 0 5.70711 36 .033333335 0 1 0 1 .001871514 10
.11271858 .3231201 -18 3 0 12.61154 4.060443 77 .2982456 1 1 0 1 .0040043467 15
.11267185 -1.1793947 3 3 0 0 5.209486 26 .2032967 1 1 0 1 .0005115914 10
.01586914 -1.7421885 2 3 0 13.728864 5.398163 23 .59090906 1 2 0 1 .015181175 20
-.14756012 -1.645667 67 3 0 16.159956 5.043425 44 .11688311 0 1 0 1 .004423147 25
-.11185074 0 0 4 0 18.162052 5.620401 55 .23636363 0 1 0 1 .021334114 24
.503417 .7052422 -2 2 0 0 3.218876 24 .375 0 1 0 1 .005785304 25
.3714514 8.746828 -3 2 0 0 2.6390574 32 .15384616 0 1 0 1 .00023822644 14
-.07223606 -1.0824804 0 2 0 0 2.70805 14 .2142857 0 1 0 1 .0039513726 16
-.03809166 -1.4982705 0 2 0 10.330584 2.772589 14 .3333333 0 1 0 1 .0007267679 28
0 0 0 1 0 13.815512 1.7917595 5 .2 0 1 0 0 .00001892976 10
.10793304 .37629795 25 3 0 14.3313 4.779123 32 .15254237 0 1 0 1 .0024480834 25
.3469362 9.661702 0 2 0 0 3.367296 11 .4285714 0 1 0 0 .00003982736 10
-.3774986 11.99938 2 2 0 0 3.178054 15 .13043478 0 1 0 1 .0021171982 14
-.3845348 .06218529 1 1 0 0 1.3862944 7 .6666667 0 1 0 1 .00002493976 71
-.24039364 .58119583 38 3 0 0 5.225747 23 .07027027 0 1 1 1 .014327654 15
.2751875 2.887886 5 3 0 0 4.7095304 16 .0909091 0 1 0 1 .0005476744 25
.4650097 -1.0190048 13 3 0 0 4.158883 12 .15873016 1 1 0 0 .0007025523 10
-.2226267 7.259292 -3 2 0 0 2.995732 7 .57894737 0 1 0 1 .0004186834 71
-.18076324 -.3808956 -6 3 0 17.189821 4.3694477 10 .25641027 0 1 0 1 .002557078 24
-.3849268 .5747299 0 2 0 14.167748 3.9318256 8 .76 0 3 0 1 .0012052236 71
-.04464245 .4353123 0 3 0 11.59549 4.564348 29 .3157895 0 1 0 1 .0006192101 10
-.8131294 -1.4821582 10 4 0 18.386717 6.05444 25 .02352941 0 1 0 0 .001858902 10
-.18500614 -.3580999 3 4 0 12.266867 5.934894 23 .3050398 0 3 0 1 .02189973 22
-.06857872 -.05429745 257 4 0 18.233652 8.504918 12 .024301337 0 1 0 1 .001652708 10
.25405693 -1.0460367 16 4 0 0 5.752573 39 .19745223 1 2 1 1 .05051396 23
.06768894 3.0409956 1 1 0 14.99056 1.94591 17 .3333333 0 3 0 1 .00031754575 11
.1249981 -.0901394 20 3 0 16.927423 5.361292 29 .10377359 0 1 0 1 .0039022586 14
-.22162056 1.3668618 0 1 0 0 1.7917595 19 .4 0 1 0 0 .0004641471 31
.00795269 -.7731724 225 4 1 18.823112 8.043663 63 .2997109 1 2 0 1 .07408877 10
.008747101 .6601477 -4 3 1 11.356926 4.0943446 19 .5254237 1 3 0 1 .0046413112 20
-.7868834 -3.892228 0 1 0 0 2.397895 7 1 0 3 0 1 .00023225507 71
-1.301571 -.10422134 2 2 0 16.691484 3.367296 12 .3214286 0 1 0 1 .0079781 22
.25770855 .4922538 0 3 0 17.513212 5.463832 14 .021276595 0 3 0 0 .0007827886 10
-.08994293 2.060296 0 1 0 0 2.1972246 14 .25 0 1 0 1 .0003540879 22
-.26158428 -.7679377 -33 3 0 0 5.398163 7 .10454545 0 3 0 1 .02005634 33
-.10461807 .2203636 44 4 0 19.17188 6.860664 57 .09548793 1 2 0 1 .04536956 22
.1892042 -10.48865 0 2 0 0 2.484907 8 0 0 1 0 1 .0017294813 31
-.25763416 1.6497202 1 1 1 12.856174 2.0794415 13 .7142857 0 1 1 1 .00015687557 11
-.20214462 -.03921986 4 2 0 0 3.135494 15 .4545455 0 1 0 1 .0046134954 16
.11091042 -1.780286 0 4 0 17.459373 5.765191 32 .14150943 0 1 0 1 .006076913 22
-.16748047 1.0581417 -56 3 0 0 5.030438 17 .09868421 0 1 0 1 .0013416485 25
. . -6 1 0 0 1.94591 30 0 0 1 0 1 .00010547184 23
-.05818367 .07474613 66 4 0 19.68265 7.592366 77 .1826438 1 1 0 1 .06662825 10
.7532358 -.08581448 -141 4 0 0 5.556828 19 .1007752 0 3 0 1 .006396934 20
-.10923672 .897191 21 3 0 10.088015 4.406719 34 .3827161 0 1 0 1 .0008205485 71
-.8360291 .09221363 0 3 0 18.425793 4.26268 9 .25714287 0 3 0 1 .0019029792 10
.05856609 -1.897893 30 3 0 15.44169 5.525453 19 .1 0 1 0 1 .00572302 22
-1.26694 0 2 1 0 0 1.0986123 5 0 0 1 0 1 .0005050882 27
.17764187 1.8856926 2 3 0 17.090055 4.1108737 21 .3333333 0 1 0 1 .0031127864 20
-2.980569 -11.146204 -5 2 0 0 3.0445225 8 .5 0 1 0 1 .005294049 71
-.360817 -1.4382372 0 2 0 10.953014 3.295837 34 .03846154 0 1 0 1 .00049461547 25
.5469475 .3209362 1 2 0 0 3.912023 8 .06122449 0 1 1 1 .002531606 16
.26773167 .43429375 13 2 0 0 3.78419 9 .9767442 1 1 0 1 .0006287515 71
2.046277 -2.405614 -9 2 0 0 2.70805 31 .3571429 0 1 0 1 .00019434135 31
-.3823385 9.254453 0 1 0 0 1.3862944 18 0 0 1 0 1 .0003113205 27
2.641924 0 -2 3 0 0 4.7004805 14 .11926606 0 1 0 1 .0003143995 33
-.2295313 -.02360916 2 1 0 0 2.0794415 17 0 0 1 0 0 .00013545017 25
.1944046 -3.142589 0 1 0 0 2.1972246 6 .375 0 1 0 1 .0001124937 22
-.20025635 1.1126423 0 1 0 0 2.397895 9 .3 0 1 0 1 .000580177 10
-.28901958 -1.0432911 14 3 0 0 4.564348 24 .05263158 0 1 0 1 .009643066 15
-.05709457 -1.6741934 1 3 0 0 4.553877 50 .25531915 0 1 0 1 .003230231 27
-.0030174255 -1.1208677 -3 4 0 0 5.743003 22 .019292604 0 1 0 1 .003691942 33
.23391533 11.278747 0 1 0 16.34124 2.397895 18 .5 0 1 0 0 .00025765295 10
-.9022388 .5791025 26 3 0 13.557114 4.941642 22 .7410072 0 3 0 1 .004595376 71
.426012 2.7777014 10 2 0 0 2.944439 4 .11111111 0 1 0 1 .0009716832 28
.07714462 -.006279945 -42 4 0 17.784084 7.675082 24 .3525313 0 3 0 1 .02928413 10
-.26942635 .17977715 -14 3 0 12.795702 4.644391 30 .19417475 1 1 0 1 .005755572 28
.022652626 -.11623669 73 4 0 16.64166 6.234411 11 .066797644 0 1 0 1 .0007288266 10
-.5401287 0 0 1 0 0 1.7917595 22 0 0 1 0 1 .0001734117 31
.06868839 .9075851 9 3 0 17.721613 5.062595 37 .05732484 0 1 0 1 .02119965 27
-.22326756 .9552946 3 2 1 12.456236 3.78419 43 .744186 0 1 0 1 .0029437 27
-.0019016266 .8681054 2 3 0 0 4.394449 23 .25 0 1 0 0 .0014063765 25
.008501053 -.31473255 -24 4 0 0 5.673323 19 .9068965 1 3 0 1 .20628814 33
.4296713 .464838 0 2 0 0 2.890372 30 .1764706 0 1 0 1 .0004863654 33
.17627716 -7.434703 -53 3 0 0 4.1271343 13 .06557377 0 1 0 1 .0015428582 14
1.0340643 -1.513542 7 1 0 0 2.3025851 5 .11111111 0 1 0 0 .000011357856 10
-.10409737 -.05284691 -16 3 0 17.327057 4.75359 21 .4347826 0 3 0 1 .016138596 22
.1163559 -.8664169 77 4 0 15.62732 7.109062 109 .14729951 0 2 1 1 .009371053 10
2.399113 6.416732 0 2 0 0 2.6390574 10 .23076923 0 1 0 1 .000021944765 25
2.0331945 0 1 2 0 0 2.564949 4 .3333333 1 1 0 1 .00016544062 71
-.14650536 .003691673 66 4 0 18.384626 7.255591 49 .11095406 1 2 0 0 .00568074 10
-.02972698 -.08861637 6 2 0 0 3.78419 24 .04651163 0 1 0 1 .0018428043 25
.26714134 .2840023 4 3 0 14.290452 3.970292 25 .13461539 1 1 0 0 .007423409 11
.2038088 7.400843 -1 2 0 0 2.564949 11 .08333334 0 1 0 1 .0005515471 14
.2023802 .6268883 -4 3 0 15.758952 5.181784 26 .27118644 0 1 0 1 .011997193 32
-.04707336 -1.752819 7 1 0 0 1.7917595 5 1 1 3 0 1 .0016874084 33
-.8939953 -8.242458 0 2 0 0 2.772589 32 .06666667 0 1 0 1 .0005713536 15
end
label values size_2015 size_2015
label def size_2015 1 "Microempresa = 1", modify
label def size_2015 2 "Pequeña empresa = 1", modify
label def size_2015 3 "Mediana empresa = 1", modify
label def size_2015 4 "Empresa grande = 1", modify
label values cite cite
label def cite 0 "No cliente CITE", modify
label def cite 1 "Cliente CITE", modify
label values Qkextrj15 Qkextrj15
label def Qkextrj15 1 "Capital extranjero 2015 [0%] = 1", modify
label def Qkextrj15 2 "Capital extranjero 2015 <0; 50%] = 1", modify
label def Qkextrj15 3 "Capital extranjero 2015 <50% a más> = 1", modify
label values div div
label def div 10 "Elaboración de productos alimenticios", modify
label def div 11 "Elaboración de bebidas", modify
label def div 14 "Fabricación de prendas de vestir", modify
label def div 15 "Fabricación de productos de cuero y productos conexos", modify
label def div 16 "Producción de madera y fabricación de productos de madera y corcho", modify
label def div 20 "Fabricación de sustancias y productos químicos", modify
label def div 22 "Fabricación de productos de caucho y de plástico", modify
label def div 23 "Fabricación de otros productos minerales no metálicos", modify
label def div 24 "Fabricación de metales comunes", modify
label def div 25 "Fabricación de productos de metal, excepto maquinaria y equipo", modify
label def div 27 "Fabricación de equipo eléctrico", modify
label def div 28 "Fabricación de maquinaria y equipo n.c.p.", modify
label def div 31 "Fabricación de muebles", modify
label def div 32 "Otras industrias manufactureras", modify
label def div 33 "Reparación e instalación de maquinaria y equipo", modify
label def div 71 "Arquitectura e ingeniería; ensayos y análisis técnicos", modify
label var lnDprod "Log. Cambio productividad"
label var lnDkfijo "Log. Cambio K/L"
label var lnDL "Log. Cambio en trabajadores"
label var size_2015 "Estrato tamaño 2015"
label var cite "Cliente CITE"
label var lnexport15 "Log. valor exportado 2015"
label var lnL15 "Log. Total de empleados 2015"
label var age "Años de experiencia"
label var skill_15 "% Personal calificado 2015"
label var grupo15 "Pertenece a grupo económico 2015"
label var Qkextrj15 "Capital extranjero 2015"
label var Acc_pub "Accedió a programa de apoyo público"
label var provcite "CITE en provincia 2015"
label var cuota "Market share 2015"
label var div "Grupo CIIU"
