Oaxaca-Blinder type Counterfactual Decomposition Methods for Duration Outcomes

Título traducido de la contribución: Tipo Oaxaca-Blinder Métodos de descomposición contrafáctica para los resultados de la duración

Andres Felipe Garcia Suaza

Resultado de la investigación: Documento de Trabajo

Resumen

Los procedimientos de inferencia existentes para realizar la descomposición contrafáctica de la diferencia entre las características de distribución, aplicables cuando los datos se observan plenamente, no son adecuados para los resultados censurados. Esto puede explicar la falta de análisis contrafact-tuales usando variables objetivo relacionadas con los resultados de duración, típicamente observados bajo la censura derecha. Por ejemplo, hay muchos estudios que realizan una descomposición contraproducente de las diferencias salariales entre hombres y mujeres, pero muy pocos sobre las diferencias en la duración del desempleo de género. Proporcionamos un método de descomposición tipo Oaxaca-Blinder de la media para los datos censurados. La estimación consistente de los componentes de descomposición se basa en un estimador previo de la distribución conjunta de la duración y las covariables bajo restricciones adecuadas del mecanismo de censura. Para descomponer otras características distributivas, como la mediana o el coeficiente de Gini, proponemos un método inferencial para la descomposición contrafáctica introduciendo restricciones en la forma funcional de la distribución condicional de la duración de las covariables dadas. Proporcionamos una justificación formal para la inferencia asintótica y estudiamos el rendimiento de la muestra finita a través de los experimentos de Monte Carlo. Finalmente, aplicamos la metodología propuesta al análisis de las brechas de duración del desempleo en España. Este estudio sugiere que factores que van más allá de las características socioeconómicas de los trabajadores juegan un papel relevante a la hora de explicar la diferencia entre varias medidas de distribución de la duración del desempleo, como la media, la probabilidad de estar en paro de larga duración y el coeficiente de Gini.
Idioma originalEnglish (US)
Número de páginas52
EstadoPublished - 2016

Huella dactilar

Decomposition
Inference
Covariates
Unemployment duration
Censoring
Gini coefficient
Justification
Factors
Monte Carlo experiment
Median
Censored data
Gender wage gap
Conditional distribution
Finite sample
Workers
Methodology
Spain
Estimator
Socioeconomic characteristics
Duration of unemployment

Citar esto

@techreport{f1c0a31b8105488e9ff00531c9177c6f,
title = "Oaxaca-Blinder type Counterfactual Decomposition Methods for Duration Outcomes",
abstract = "Existing inference procedures to perform counterfactual decomposition of the difference between distributional features, applicable when data is fully observed, are not suitable for censored outcomes. This may explain the lack of counterfac- tual analyses using target variables related to duration outcomes, typically observed under right censoring. For instance, there are many studies performing counterfac- tual decomposition of the gender wage gaps, but very few on gender unemployment duration gaps. We provide an Oaxaca-Blinder type decomposition method of the mean for censored data. Consistent estimation of the decomposition components is based on a prior estimator of the joint distribution of duration and covariates under suitable restrictions on the censoring mechanism. To decompose other distribu- tional features, such as the median or the Gini coefficient, we propose an inferential method for the counterfactual decomposition by introducing restrictions on the func tional form of the conditional distribution of duration given covariates. We provide formal justification for asymptotic inference and study the finite sample performance through Monte Carlo experiments. Finally, we apply the proposed methodology to the analysis of unemployment duration gaps in Spain. This study suggests that factors beyond the workers' socioeconomic characteristics play a relevant role in explaining the difference between several unemployment duration distribution fea- tures such as the mean, the probability of being long term unemployed and the Gini coefficient.",
author = "{Garcia Suaza}, {Andres Felipe}",
year = "2016",
language = "English (US)",
type = "WorkingPaper",

}

Oaxaca-Blinder type Counterfactual Decomposition Methods for Duration Outcomes. / Garcia Suaza, Andres Felipe.

2016.

Resultado de la investigación: Documento de Trabajo

TY - UNPB

T1 - Oaxaca-Blinder type Counterfactual Decomposition Methods for Duration Outcomes

AU - Garcia Suaza, Andres Felipe

PY - 2016

Y1 - 2016

N2 - Existing inference procedures to perform counterfactual decomposition of the difference between distributional features, applicable when data is fully observed, are not suitable for censored outcomes. This may explain the lack of counterfac- tual analyses using target variables related to duration outcomes, typically observed under right censoring. For instance, there are many studies performing counterfac- tual decomposition of the gender wage gaps, but very few on gender unemployment duration gaps. We provide an Oaxaca-Blinder type decomposition method of the mean for censored data. Consistent estimation of the decomposition components is based on a prior estimator of the joint distribution of duration and covariates under suitable restrictions on the censoring mechanism. To decompose other distribu- tional features, such as the median or the Gini coefficient, we propose an inferential method for the counterfactual decomposition by introducing restrictions on the func tional form of the conditional distribution of duration given covariates. We provide formal justification for asymptotic inference and study the finite sample performance through Monte Carlo experiments. Finally, we apply the proposed methodology to the analysis of unemployment duration gaps in Spain. This study suggests that factors beyond the workers' socioeconomic characteristics play a relevant role in explaining the difference between several unemployment duration distribution fea- tures such as the mean, the probability of being long term unemployed and the Gini coefficient.

AB - Existing inference procedures to perform counterfactual decomposition of the difference between distributional features, applicable when data is fully observed, are not suitable for censored outcomes. This may explain the lack of counterfac- tual analyses using target variables related to duration outcomes, typically observed under right censoring. For instance, there are many studies performing counterfac- tual decomposition of the gender wage gaps, but very few on gender unemployment duration gaps. We provide an Oaxaca-Blinder type decomposition method of the mean for censored data. Consistent estimation of the decomposition components is based on a prior estimator of the joint distribution of duration and covariates under suitable restrictions on the censoring mechanism. To decompose other distribu- tional features, such as the median or the Gini coefficient, we propose an inferential method for the counterfactual decomposition by introducing restrictions on the func tional form of the conditional distribution of duration given covariates. We provide formal justification for asymptotic inference and study the finite sample performance through Monte Carlo experiments. Finally, we apply the proposed methodology to the analysis of unemployment duration gaps in Spain. This study suggests that factors beyond the workers' socioeconomic characteristics play a relevant role in explaining the difference between several unemployment duration distribution fea- tures such as the mean, the probability of being long term unemployed and the Gini coefficient.

UR - https://ideas.repec.org/p/col/000092/014186.ht ml

M3 - Working paper

BT - Oaxaca-Blinder type Counterfactual Decomposition Methods for Duration Outcomes

ER -