Bounds on Average and Quantile Treatment Effects on Duration Outcomes Under Censoring, Selection, and Noncompliance

German Blanco, Xuan Chen, Carlos A. Flores, Alfonso Flores-Lagunes

Research output: Contribution to journalArticle

Abstract

We consider the problem of assessing the effects of a treatment on duration outcomes using data from a randomized evaluation with noncompliance. For such settings, we derive nonparametric sharp bounds for average and quantile treatment effects addressing three pervasive problems simultaneously: self-selection into the spell of interest, endogenous censoring of the duration outcome, and noncompliance with the assigned treatment. Ignoring any of these issues could yield biased estimates of the effects. Notably, the proposed bounds do not impose the independent censoring assumption—which is commonly used to address censoring but is likely to fail in important settings—or exclusion restrictions to address endogeneity of censoring and selection. Instead, they employ monotonicity and stochastic dominance assumptions. To illustrate the use of these bounds we assess the effects of the Job Corps (JC) training program on its participants’ last complete employment spell duration. Our estimated bounds suggest that JC participation may increase the average duration of the last complete employment spell before week 208 after randomization by at least 5.6 log points (5.8%) for individuals who comply with their treatment assignment and experience a complete employment spell whether or not they enrolled in JC. The estimated quantile treatment effects suggest the impacts may be heterogeneous, and strengthen our conclusions based on the estimated average effects.

Original languageEnglish (US)
JournalJournal of Business and Economic Statistics
DOIs
StatePublished - Jan 1 2019

Fingerprint

Noncompliance
Treatment Effects
Censoring
Quantile
Endogeneity
Stochastic Dominance
Sharp Bound
Randomisation
Biased
Monotonicity
Assignment
Likely
Restriction
Quantile treatment effects
Average treatment effect
training program
Evaluation
exclusion
Estimate
participation

Keywords

  • Duration outcomes
  • Independent censoring
  • Job corps
  • Partial identification
  • Principal stratification

ASJC Scopus subject areas

  • Statistics and Probability
  • Social Sciences (miscellaneous)
  • Economics and Econometrics
  • Statistics, Probability and Uncertainty

Cite this

@article{30d8b501b8d845cdbd0b154bcd522c08,
title = "Bounds on Average and Quantile Treatment Effects on Duration Outcomes Under Censoring, Selection, and Noncompliance",
abstract = "We consider the problem of assessing the effects of a treatment on duration outcomes using data from a randomized evaluation with noncompliance. For such settings, we derive nonparametric sharp bounds for average and quantile treatment effects addressing three pervasive problems simultaneously: self-selection into the spell of interest, endogenous censoring of the duration outcome, and noncompliance with the assigned treatment. Ignoring any of these issues could yield biased estimates of the effects. Notably, the proposed bounds do not impose the independent censoring assumption—which is commonly used to address censoring but is likely to fail in important settings—or exclusion restrictions to address endogeneity of censoring and selection. Instead, they employ monotonicity and stochastic dominance assumptions. To illustrate the use of these bounds we assess the effects of the Job Corps (JC) training program on its participants’ last complete employment spell duration. Our estimated bounds suggest that JC participation may increase the average duration of the last complete employment spell before week 208 after randomization by at least 5.6 log points (5.8{\%}) for individuals who comply with their treatment assignment and experience a complete employment spell whether or not they enrolled in JC. The estimated quantile treatment effects suggest the impacts may be heterogeneous, and strengthen our conclusions based on the estimated average effects.",
keywords = "Duration outcomes, Independent censoring, Job corps, Partial identification, Principal stratification",
author = "German Blanco and Xuan Chen and Flores, {Carlos A.} and Alfonso Flores-Lagunes",
year = "2019",
month = "1",
day = "1",
doi = "10.1080/07350015.2019.1609975",
language = "English (US)",
journal = "Journal of Business and Economic Statistics",
issn = "0735-0015",
publisher = "American Statistical Association",

}

TY - JOUR

T1 - Bounds on Average and Quantile Treatment Effects on Duration Outcomes Under Censoring, Selection, and Noncompliance

AU - Blanco, German

AU - Chen, Xuan

AU - Flores, Carlos A.

AU - Flores-Lagunes, Alfonso

PY - 2019/1/1

Y1 - 2019/1/1

N2 - We consider the problem of assessing the effects of a treatment on duration outcomes using data from a randomized evaluation with noncompliance. For such settings, we derive nonparametric sharp bounds for average and quantile treatment effects addressing three pervasive problems simultaneously: self-selection into the spell of interest, endogenous censoring of the duration outcome, and noncompliance with the assigned treatment. Ignoring any of these issues could yield biased estimates of the effects. Notably, the proposed bounds do not impose the independent censoring assumption—which is commonly used to address censoring but is likely to fail in important settings—or exclusion restrictions to address endogeneity of censoring and selection. Instead, they employ monotonicity and stochastic dominance assumptions. To illustrate the use of these bounds we assess the effects of the Job Corps (JC) training program on its participants’ last complete employment spell duration. Our estimated bounds suggest that JC participation may increase the average duration of the last complete employment spell before week 208 after randomization by at least 5.6 log points (5.8%) for individuals who comply with their treatment assignment and experience a complete employment spell whether or not they enrolled in JC. The estimated quantile treatment effects suggest the impacts may be heterogeneous, and strengthen our conclusions based on the estimated average effects.

AB - We consider the problem of assessing the effects of a treatment on duration outcomes using data from a randomized evaluation with noncompliance. For such settings, we derive nonparametric sharp bounds for average and quantile treatment effects addressing three pervasive problems simultaneously: self-selection into the spell of interest, endogenous censoring of the duration outcome, and noncompliance with the assigned treatment. Ignoring any of these issues could yield biased estimates of the effects. Notably, the proposed bounds do not impose the independent censoring assumption—which is commonly used to address censoring but is likely to fail in important settings—or exclusion restrictions to address endogeneity of censoring and selection. Instead, they employ monotonicity and stochastic dominance assumptions. To illustrate the use of these bounds we assess the effects of the Job Corps (JC) training program on its participants’ last complete employment spell duration. Our estimated bounds suggest that JC participation may increase the average duration of the last complete employment spell before week 208 after randomization by at least 5.6 log points (5.8%) for individuals who comply with their treatment assignment and experience a complete employment spell whether or not they enrolled in JC. The estimated quantile treatment effects suggest the impacts may be heterogeneous, and strengthen our conclusions based on the estimated average effects.

KW - Duration outcomes

KW - Independent censoring

KW - Job corps

KW - Partial identification

KW - Principal stratification

UR - http://www.scopus.com/inward/record.url?scp=85068211812&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85068211812&partnerID=8YFLogxK

U2 - 10.1080/07350015.2019.1609975

DO - 10.1080/07350015.2019.1609975

M3 - Article

JO - Journal of Business and Economic Statistics

JF - Journal of Business and Economic Statistics

SN - 0735-0015

ER -