4ª Reunión Española de
Usuarios de STATA 2011
Métodos de descomposición en
Economía utilizando STATA
Raúl Ramos
AQR-IREA (Universitat de Barcelona) & IZA
Métodos de descomposición en Economía utilizando STATA
Estructura de la presentación
• Motivación y breve descripción de los principales métodos
de descomposición en economía (laboral).
• Revisión y discusión de los procedimientos existentes en
STATA.
• Algunos ejemplos a partir del análisis de los microdatos de
las Encuestas de Presupuestos Familiares 2006 y 2009.
Ficheros de datos: http://www.ine.es/prodyser/micro_epf2006.htm
Rutinas de STATA: http://www.raulramos.cat/stata2011
Métodos de descomposición en Economía utilizando STATA
Motivación
• ¿Por qué los hombres tienen salarios superiores a los de las
mujeres?
•
¿Qué factores explican el crecimiento a lo largo del tiempo
en la desigualdad de la renta?
• ¿Por qué existen diferencias en el uso de las nuevas
tecnologías entre hombres y mujeres?
Métodos de descomposición en Economía utilizando STATA
• Los economistas intentamos ofrecer respuestas a las
preguntas planteadas a través de la utilización de los
métodos de descomposición.
• Estos métodos se basan en las contribuciones de Oaxaca
(1973) y Blinder (1973), pero han tenido desarrollos
posteriores que han permitido ampliar el abánico de temas
susceptibles de ser analizados desde esta perspectiva.
Fortin, Lemieux, Firpo (2010), http://www.nber.org/papers/w16045
Métodos de descomposición en Economía utilizando STATA
Oaxaca-Blinder
• Queremos explicar el salario (W) que recibe un trabajador/a
en función de su nivel de estudios (S) a través de la
estimación de un modelo de regresión (ecuación de Mincer):
W Hi   H 1   H 2  S Hi  U Hi


W Mi   M 1   M 2  S Mi  U Mi



W M  W H  S Mi  S Hi  ˆH 2  ˆM 1  ˆH 1  S Mi  ˆM 2  ˆH 2

  

           

explic ada
(c arac terí s tic as )
no explic ada
(c oefic ien tes )

Métodos de descomposición en Economía utilizando STATA
. bysort HOMBRE: sum LWAGE ANOSEST if LWAGE!=. & ANOSEST!=.
-> HOMBRE = 0
Variable
Obs
Mean
LWAGE
ANOSEST
9874
9874
6.752326
10.37543
Variable
Obs
Mean
LWAGE
ANOSEST
13746
13746
6.999458
9.967772
Std. Dev.
.4520183
4.111178
Min
Max
6.214608
0
8.006368
17
Min
Max
6.214608
0
8.006368
17
-> HOMBRE = 1
Std. Dev.
.4510732
3.736099
. oaxaca LWAGE ANOSEST, by(HOMBRE) nodetail
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
LWAGE
Coef.
overall
group_1
group_2
difference
endowments
coefficients
interaction
6.752326
6.999458
-.2471321
.0232102
-.2717751
.0014328
Std. Err.
.0045491
.0038474
.0059579
.0029963
.0051093
.0005592
z
1484.32
1819.25
-41.48
7.75
-53.19
2.56
P>|z|
0.000
0.000
0.000
0.000
0.000
0.010
=
=
=
=
23620
linear
9874
13746
[95% Conf. Interval]
6.74341
6.991917
-.2588095
.0173376
-.2817891
.0003367
6.761242
7.006999
-.2354548
.0290828
-.2617611
.0025288
Métodos de descomposición en Economía utilizando STATA
. oaxaca LWAGE ANOSEST, by(HOMBRE) nodetail
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
LWAGE
Coef.
overall
group_1
group_2
difference
endowments
coefficients
interaction
6.752326
6.999458
-.2471321
.0232102
-.2717751
.0014328
Std. Err.
.0045491
.0038474
.0059579
.0029963
.0051093
.0005592
z
1484.32
1819.25
-41.48
7.75
-53.19
2.56
P>|z|
0.000
0.000
0.000
0.000
0.000
0.010
=
=
=
=
23620
linear
9874
13746
[95% Conf. Interval]
6.74341
6.991917
-.2588095
.0173376
-.2817891
.0003367
6.761242
7.006999
-.2354548
.0290828
-.2617611
.0025288
. oaxaca LWAGE ANOSEST EXPPOT EXPPOT2, by(HOMBRE) nodetail
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
LWAGE
Coef.
overall
group_1
group_2
difference
endowments
coefficients
interaction
6.752326
6.999458
-.2471321
-.0031307
-.2567434
.0127419
Std. Err.
.0045494
.0038476
.0059583
.003501
.005005
.0014051
z
1484.23
1819.17
-41.48
-0.89
-51.30
9.07
P>|z|
0.000
0.000
0.000
0.371
0.000
0.000
=
=
=
=
23620
linear
9874
13746
[95% Conf. Interval]
6.743409
6.991917
-.2588101
-.0099924
-.266553
.009988
6.761243
7.006999
-.2354541
.0037311
-.2469337
.0154958
Métodos de descomposición en Economía utilizando STATA
. oaxaca LWAGE ANOSEST EXPPOT EXPPOT2, by(HOMBRE) nodetail
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
LWAGE
Coef.
overall
group_1
group_2
difference
endowments
coefficients
interaction
6.752326
6.999458
-.2471321
-.0031307
-.2567434
.0127419
Std. Err.
.0045494
.0038476
.0059583
.003501
.005005
.0014051
z
1484.23
1819.17
-41.48
-0.89
-51.30
9.07
P>|z|
0.000
0.000
0.000
0.371
0.000
0.000
=
=
=
=
23620
linear
9874
13746
[95% Conf. Interval]
6.743409
6.991917
-.2588101
-.0099924
-.266553
.009988
6.761243
7.006999
-.2354541
.0037311
-.2469337
.0154958
. nldecompose, by(HOMBRE) threefold: regress LWAGE ANOSEST EXPPOT EXPPOT2
Number of obs (A) =
Number of obs (B) =
Results
Coef.
Percentage
Omega = 1
Char
Coef
Int
-.0096113
.2440015
.0127419
-3.889115%
98.7332%
5.155911%
Omega = 0
Char
Coef
Int
.0031307
.2567434
-.0127419
1.266796%
103.8891%
-5.155911%
Raw
.2471321
100%
13746
9874
Métodos de descomposición en Economía utilizando STATA
. oaxaca LWAGE ANOSEST EXPPOT EXPPOT2, by(HOMBRE) weight(0) nodetail
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
LWAGE
Coef.
overall
group_1
group_2
difference
explained
unexplained
6.752326
6.999458
-.2471321
-.0031307
-.2440015
Std. Err.
.0045494
.0038476
.0059583
.003501
.0050192
z
1484.23
1819.17
-41.48
-0.89
-48.61
P>|z|
0.000
0.000
0.000
0.371
0.000
=
=
=
=
23620
linear
9874
13746
[95% Conf. Interval]
6.743409
6.991917
-.2588101
-.0099924
-.253839
6.761243
7.006999
-.2354541
.0037311
-.2341639
.
.
.
. oaxaca LWAGE ANOSEST EXPPOT EXPPOT2, by(HOMBRE) weight(1) nodetail
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
LWAGE
Coef.
overall
group_1
group_2
difference
explained
unexplained
6.752326
6.999458
-.2471321
.0096113
-.2567434
Std. Err.
.0045494
.0038476
.0059583
.0034659
.005005
z
1484.23
1819.17
-41.48
2.77
-51.30
P>|z|
0.000
0.000
0.000
0.006
0.000
=
=
=
=
23620
linear
9874
13746
[95% Conf. Interval]
6.743409
6.991917
-.2588101
.0028181
-.266553
6.761243
7.006999
-.2354541
.0164044
-.2469337
Métodos de descomposición en Economía utilizando STATA
. bysort HOMBRE: sum OCUP
-> HOMBRE = 0
Variable
Obs
Mean
OCUP
28417
.3350459
Variable
Obs
Mean
OCUP
27278
.5063788
Std. Dev.
.4720148
Min
Max
0
1
Min
Max
0
1
-> HOMBRE = 1
Std. Dev.
.4999685
. nldecompose, by(HOMBRE) threefold: logit OCUP NATIVO SOLTERO EDAD NMIEM7
Number of obs (A) =
Number of obs (B) =
Results
Coef.
Percentage
Omega = 1
Char
Coef
Int
-.0082834
.1925912
-.0129749
-4.8347%
112.4076%
-7.572933%
Omega = 0
Char
Coef
Int
-.0212584
.1796163
.0129749
-12.40763%
104.8347%
7.572933%
Raw
.1713328
100%
27278
28417
Métodos de descomposición en Economía utilizando STATA
. fairlie OCUP NATIVO SOLTERO EDAD NMIEM7, by(HOMBRE) reference(1)
Iteration
Iteration
Iteration
Iteration
0:
1:
2:
3:
log
log
log
log
likelihood
likelihood
likelihood
likelihood
=
=
=
=
-18905.449
-16972.542
-16931.588
-16931.428
Logistic regression
Number of obs
LR chi2(4)
Prob > chi2
Pseudo R2
Log likelihood = -16931.428
OCUP
Coef.
NATIVO
SOLTERO
EDAD
NMIEM7
_cons
-.505698
-2.512533
-.0464858
-.477288
3.910227
Std. Err.
.0577876
.0469048
.0011812
.0162986
.0934924
z
-8.75
-53.57
-39.35
-29.28
41.82
P>|z|
Non-linear decomposition by HOMBRE (G)
.
OCUP
Coef.
NATIVO
SOLTERO
EDAD
NMIEM7
.0006068
.0249098
-.0093371
.005099
Std. Err.
.0000883
.0004716
.0003832
.0002741
-.6189595
-2.604464
-.048801
-.5092327
3.726985
6.87
52.82
-24.37
18.60
-.3924364
-2.420601
-.0441706
-.4453433
4.093469
50
100
Number of obs
N of obs G=0
N of obs G=0
Pr(Y!=0|G=0)
Pr(Y!=0|G=1)
Difference
Total explained
z
27278
3948.04
0.0000
0.1044
[95% Conf. Interval]
0.000
0.000
0.000
0.000
0.000
Decomposition replications (100)
1
2
3
4
5
..................................................
..................................................
=
=
=
=
=
55695
=
28417
=
27278
=
.33504592
=
.50637877
= -.17133284
=
.02125835
P>|z|
[95% Conf. Interval]
0.000
0.000
0.000
0.000
.0004337
.0239856
-.010088
.0045618
.0007799
.0258341
-.0085861
.0056363
Métodos de descomposición en Economía utilizando STATA
. oaxaca LWAGE ANOSEST EXPPOT EXPPOT2, by(HOMBRE) weight(0)
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
Std. Err.
z
P>|z|
=
=
=
=
23620
linear
9874
13746
LWAGE
Coef.
[95% Conf. Interval]
overall
group_1
group_2
difference
explained
unexplained
6.752326
6.999458
-.2471321
-.0031307
-.2440015
.0045494
.0038476
.0059583
.003501
.0050192
1484.23
1819.17
-41.48
-0.89
-48.61
0.000
0.000
0.000
0.371
0.000
6.743409
6.991917
-.2588101
-.0099924
-.253839
6.761243
7.006999
-.2354541
.0037311
-.2341639
explained
ANOSEST
EXPPOT
EXPPOT2
.0229593
-.0454813
.0193913
.0029697
.0068799
.0067165
7.73
-6.61
2.89
0.000
0.000
0.004
.0171387
-.0589655
.0062272
.0287799
-.031997
.0325554
unexplained
ANOSEST
EXPPOT
EXPPOT2
_cons
.1086091
-.3237409
.2243474
-.253217
.0159958
.0324256
.0191586
.0256104
6.79
-9.98
11.71
-9.89
0.000
0.000
0.000
0.000
.0772579
-.3872939
.1867972
-.3034125
.1399602
-.2601879
.2618976
-.2030215
. oaxaca LWAGE ANOSEST EXPPOT EXPPOT2, by(HOMBRE) nodetail
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
LWAGE
Coef.
overall
group_1
group_2
difference
6.752326
6.999458
-.2471321
Std. Err.
.0045494
.0038476
.0059583
z
1484.23
1819.17
-41.48
P>|z|
0.000
0.000
0.000
=
=
=
=
23620
linear
9874
13746
[95% Conf. Interval]
6.743409
6.991917
-.2588101
6.761243
7.006999
-.2354541
Métodos de descomposición en Economía utilizando STATA
. oaxaca LWAGE ESTPRIM ESTSEC EXPPOT EXPPOT2, by(HOMBRE) weight(0)
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
Std. Err.
z
P>|z|
=
=
=
=
23620
linear
9874
13746
LWAGE
Coef.
overall
group_1
group_2
difference
explained
unexplained
[95% Conf. Interval]
6.752326
6.999458
-.2471321
.0076865
-.2548186
.0045495
.0038477
.0059584
.003499
.0049956
1484.18
1819.12
-41.48
2.20
-51.01
0.000
0.000
0.000
0.028
0.000
6.743409
6.991917
-.2588105
.0008286
-.2646099
6.761243
7.006999
-.2354538
.0145445
-.2450274
explained
ESTPRIM
ESTSEC
EXPPOT
EXPPOT2
.032433
.0018083
-.0495804
.0230256
.0032515
.001329
.007483
.0079679
9.97
1.36
-6.63
2.89
0.000
0.174
0.000
0.004
.0260602
-.0007966
-.0642468
.0074087
.0388058
.0044131
-.0349139
.0386425
unexplained
ESTPRIM
ESTSEC
EXPPOT
EXPPOT2
_cons
-.0650163
-.0194887
-.250333
.1663749
-.0863557
.0079946
.0028643
.0324129
.0187925
.0178037
-8.13
-6.80
-7.72
8.85
-4.85
0.000
0.000
0.000
0.000
0.000
-.0806854
-.0251025
-.313861
.1295423
-.1212504
-.0493472
-.0138748
-.1868049
.2032075
-.051461
.
Variable
overall
group_1
group_2
difference
explained
unexplained
explained
ESTPRIM
ESTSEC
EXPPOT
EXPPOT2
ESTTER
TER
6.752326
6.9994581
-.24713211
.00768652
-.25481864
.03243298
.00180828
-.04958036
.02302563
PRIM
6.752326
6.9994581
-.24713211
.00768652
-.25481864
-.00164564
-.04958036
.02302563
.03588689
. oaxaca LWAGE ESTSEC ESTTER EXPPOT EXPPOT2, by(HOMBRE) weight(0)
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
Std. Err.
z
P>|z|
=
=
=
=
23620
linear
9874
13746
LWAGE
Coef.
overall
group_1
group_2
difference
explained
unexplained
[95% Conf. Interval]
6.752326
6.999458
-.2471321
.0076865
-.2548186
.0045495
.0038477
.0059584
.003499
.0049956
1484.18
1819.12
-41.48
2.20
-51.01
0.000
0.000
0.000
0.028
0.000
6.743409
6.991917
-.2588105
.0008286
-.2646099
6.761243
7.006999
-.2354538
.0145445
-.2450274
explained
ESTSEC
ESTTER
EXPPOT
EXPPOT2
-.0016456
.0358869
-.0495804
.0230256
.0012091
.0027258
.007483
.0079679
-1.36
13.17
-6.63
2.89
0.173
0.000
0.000
0.004
-.0040154
.0305444
-.0642468
.0074087
.0007241
.0412294
-.0349139
.0386425
unexplained
ESTSEC
ESTTER
EXPPOT
EXPPOT2
_cons
.0003023
.0259728
-.250333
.1663749
-.1971357
.0024749
.003221
.0324129
.0187925
.0181032
0.12
8.06
-7.72
8.85
-10.89
0.903
0.000
0.000
0.000
0.000
-.0045483
.0196598
-.313861
.1295423
-.2326174
.0051529
.0322858
-.1868049
.2032075
-.1616541
unexplained
ESTPRIM
ESTSEC
EXPPOT
EXPPOT2
ESTTER
_cons
-.06501626
-.01948867
-.25033295
.16637491
-.08635567
.0003023
-.25033295
.16637491
.02597285
-.19713574
Métodos de descomposición en Economía utilizando STATA
. oaxaca LWAGE ANOSEST EXPPOT EXPPOT2 normalize(ESTPRIM ESTSEC ESTTER), by(HOMBRE) weight(0)
(normalized: ESTPRIM ESTSEC ESTTER)
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
Std. Err.
z
P>|z|
=
=
=
=
23620
linear
9874
13746
LWAGE
Coef.
[95% Conf. Interval]
overall
group_1
group_2
difference
explained
unexplained
6.752326
6.999458
-.2471321
.0035228
-.2506549
.0045497
.0038478
.0059586
.0035922
.0049645
1484.14
1819.08
-41.47
0.98
-50.49
0.000
0.000
0.000
0.327
0.000
6.743409
6.991917
-.2588108
-.0035179
-.2603851
6.761243
7.007
-.2354535
.0105634
-.2409247
explained
ANOSEST
EXPPOT
EXPPOT2
ESTPRIM
ESTSEC
ESTTER
.0137584
-.0475207
.0207843
.0078755
-7.80e-06
.0086331
.0019162
.0071789
.0071965
.0009461
.000041
.0009322
7.18
-6.62
2.89
8.32
-0.19
9.26
0.000
0.000
0.004
0.000
0.849
0.000
.0100027
-.0615912
.0066795
.0060212
-.0000882
.0068061
.017514
-.0334502
.0348892
.0097298
.0000726
.0104601
unexplained
ANOSEST
EXPPOT
EXPPOT2
ESTPRIM
ESTSEC
ESTTER
_cons
-.0840635
-.2490752
.1628606
-.042871
-.0067038
.025924
-.056726
.0287433
.0323043
.0193317
.0072424
.0015754
.0032397
.037474
-2.92
-7.71
8.42
-5.92
-4.26
8.00
-1.51
0.003
0.000
0.000
0.000
0.000
0.000
0.130
-.1403994
-.3123905
.1249712
-.0570659
-.0097915
.0195744
-.1301737
-.0277276
-.1857598
.20075
-.0286761
-.0036161
.0322736
.0167216
Métodos de descomposición en Economía utilizando STATA
. oaxaca LWAGE ANOSEST EXPPOT EXPPOT2, by(HOMBRE) weight(0) model1(heckman, twostep select (OCUP= NATIVO SOLTERO ED
> (heckman, twostep select (OCUP= NATIVO SOLTERO EDAD NMIEM7))
Blinder-Oaxaca decomposition
Number of obs
Model
N of obs 1
N of obs 2
Group 1: HOMBRE = 0
Group 2: HOMBRE = 1
Std. Err.
z
P>|z|
=
=
=
=
15863
linear
6580
9283
LWAGE
Coef.
[95% Conf. Interval]
overall
group_1
group_2
difference
explained
unexplained
6.791242
7.229495
-.4382532
.011543
-.4497963
.0201357
.0103095
.0226215
.0045549
.0221315
337.27
701.25
-19.37
2.53
-20.32
0.000
0.000
0.000
0.011
0.000
6.751776
7.209289
-.4825906
.0026156
-.4931733
6.830707
7.249701
-.3939159
.0204704
-.4064193
explained
ANOSEST
EXPPOT
EXPPOT2
.0597605
-.0694218
.0212043
.0039885
.0064712
.0035783
14.98
-10.73
5.93
0.000
0.000
0.000
.0519432
-.0821051
.0141911
.0675777
-.0567385
.0282176
unexplained
ANOSEST
EXPPOT
EXPPOT2
_cons
.0335772
-.203953
-.0736814
-.2057391
.0282256
.0403399
.0267852
.0356934
1.19
-5.06
-2.75
-5.76
0.234
0.000
0.006
0.000
-.0217441
-.2830176
-.1261793
-.2756969
.0888984
-.1248883
-.0211834
-.1357813
Métodos de descomposición en Economía utilizando STATA
. nopomatch ANOSEST EXPPOT EXPPOT2, outcome(LWAGE) by(HOMBRE) sd replace
*****************************************************************
Gap in ANOSEST EXPPOT EXPPOT2 decomposition
*****
*****************************************************************
=.03659748
D
D0 =.03749842
DM =-.00001405
DF =.00011499
DX =-.00100188
*****************************************************************
percM =.99636258
percF =.99817685
*****************************************************************
Calculating Standard Deviation
Std.error DO = .00072236
Métodos de descomposición en Economía utilizando STATA
. nopomatch ANOSEST EXPPOT EXPPOT2, outcome(WAGE) by(HOMBRE) sd replace
*****************************************************************
*****
Gap in ANOSEST EXPPOT EXPPOT2 decomposition
*****************************************************************
D
=.27070513
D0 =.28253065
DM =.00024646
DF =.0007391
DX =-.01281108
*****************************************************************
percM =.99636258
percF =.99817685
*****************************************************************
Calculating Standard Deviation
Std.error DO = .00692002
Métodos de descomposición en Economía utilizando STATA
. rqdeco LWAGE ANOSEST EXPPOT EXPPOT2, by(HOMBRE) qlow(0.1) qhigh(0.9) qstep(0.1) vce(boot) reps(2) noprint
Fitting base model
(bootstrapping ..)
Decomposition of differences in distribution using quantile regression
Total number of observations
23619
Number of observations in group 0
9873
Number of observations in group 1
13746
Number of quantile regressions estimated
100
The variance has been estimated by bootstraping the results 2 times
Effects of coefficients (discrimination)
.3
.25
0
.1
.2
Log wage effects
.35
.3
.4
.4
Decomposition of differences in distribution
.2
.4
.6
.8
1
Quantile
Total differential
Effects of coefficients
Effects of characteristics
.2
0
0
.2
.4
.6
Quantile
.8
1
Métodos de descomposición en Economía utilizando STATA
. jmpierce2 est2006M est2006H est2009M est2009H, detail
Decomposition of individual differentials:
Sample 1
Sample 2
raw differential
quantity
effect
residual
gap
-.2471187
-.2531201
.0095614
.0109148
-.2566801
-.2640349
Difference in (components of) differentials:
Total
D
E
U
-.0060013
.0013535
-.0073548
Decomposition of difference in predicted gap:
E
Q
P
QP
Total
.0013535
-.0035803
.0076021
-.0026683
ANOSEST
EXPPOT
EXPPOT2
-.0039431
.0078312
-.0025347
-.0071249
4.64e-06
.0035399
.0043113
.0078278
-.004537
-.0011295
-1.29e-06
-.0015375
Decomposition of diffence in residual gap:
Total
D
E
U
Q
P
QP
=
=
=
=
=
=
U
Q
P
QP
-.0073548
.0050216
-.0104891
-.0018873
difference in differential
difference in predicted gap
difference in residual gap
quantity effect
price effect
interaction Q x P
Difference in (components of) differentials:
Métodos de descomposición en Economía utilizando STATA
. smithwelch est2006M est2006H est2009M est2009H, detail
Decompositions of individual differentials:
dD
dE
dC
dEC
ANOSEST
EXPPOT
EXPPOT2
_cons
.0018548
-.0785843
.020389
.0503392
-.0028438
.0036177
.0016642
0
.0057979
-.0864155
.0229236
.0503392
-.0010993
.0042135
-.0041989
0
Total
-.0060013
.0024382
-.0073548
-.0010847
Decomposition of difference in differentials:
Sample 1
D
E
C
EC
dE
D
E
C
EC
ANOSEST
EXPPOT
EXPPOT2
_cons
.1319002
-.3688332
.2435752
-.2537609
.0229215
-.0454705
.019369
0
.1047036
-.3407522
.2331294
-.2537609
.0042751
.0173894
-.0089233
0
ANOSEST
EXPPOT
EXPPOT2
_cons
-.0028438
.0036177
.0016642
0
-.0060049
7.52e-06
.0065639
0
.0042832
.0036108
-.0036595
0
-.0011221
-5.97e-07
-.0012401
0
Total
-.2471187
-.0031799
-.2566801
.0127413
Total
.0024382
.0005665
.0042345
-.0023628
Sample 2
D
E
C
EC
dC
D
E
C
EC
ANOSEST
EXPPOT
EXPPOT2
_cons
.1337551
-.4474175
.2639641
-.2034218
.0200777
-.0418527
.0210333
0
.1105015
-.4271677
.256053
-.2034218
.0031758
.0216029
-.0131222
0
ANOSEST
EXPPOT
EXPPOT2
_cons
.0057979
-.0864155
.0229236
.0503392
.005077
-.0030433
-2.87e-06
0
.0006876
-.0826341
.0229268
.0503392
.0000333
-.000738
-2.82e-07
0
Total
-.0073548
.0020308
-.0086806
-.000705
Total
-.2531201
-.0007417
-.2640349
.0116566
dEC
D
E
C
EC
ANOSEST
EXPPOT
EXPPOT2
_cons
-.0010993
.0042135
-.0041989
0
-.00112
-2.88e-06
-.003024
0
.0000281
.004217
-.0008775
0
-7.35e-06
-6.97e-07
-.0002974
0
Total
-.0010847
-.0041468
.0033676
-.0003054
Difference in (components of) differentials:
dD
dE
dC
dEC
ANOSEST
EXPPOT
EXPPOT2
_cons
.0018548
-.0785843
.020389
.0503392
-.0028438
.0036177
.0016642
0
.0057979
-.0864155
.0229236
.0503392
-.0010993
.0042135
-.0041989
0
Total
-.0060013
.0024382
-.0073548
-.0010847
Decomposition of difference in differentials:
dE
D
E
C
EC
ANOSEST
EXPPOT
EXPPOT2
_cons
-.0028438
.0036177
.0016642
0
-.0060049
7.52e-06
.0065639
0
.0042832
.0036108
-.0036595
0
-.0011221
-5.97e-07
-.0012401
0
Total
.0024382
.0005665
.0042345
-.0023628
dC
D
E
C
EC
D
E
C
EC
=
=
=
=
differential / difference in component of differential
part of D due to differences in endowments
part of D due to differences in coefficients
interaction E x C
Métodos de descomposición en Economía utilizando STATA
Síntesis
• Naturaleza de la variable: continua vs discreta
• Descomposición agregada vs detallada (identificación)
• Problemas de selección (Heckman vs Matching)
• Descomposición en la media o a lo largo de la distribución
(solo para variables continuas …)
• Comparación de las diferencias entre grupos a lo largo del
tiempo
oaxaca
jmpierce2
nldecompose
smithwelch
fairlie
nopomatch
rqdeco
Métodos de descomposición en Economía utilizando STATA
Otros procedimientos de interés
• decompose, gdecomp, ldecomp
• dfl (di Nardo, Fortin, Lemieux, 1996 – Van Kerm, 2003)
• gfields (Fields, 2002)
• shapley (Shorrocks, 1999)
• search inequality
inequal, rspread, glcurve, descogini, ineqerr, kdensity,
akdensity, changemean, … y mucho más
Descargar

Métodos de descomposición en Economía utilizando STATA