TY - JOUR
T1 - Partial identification of local average treatment effects with an invalid instrument
AU - Flores, Carlos A.
AU - Flores-Lagunes, Alfonso
N1 - Funding Information:
Detailed comments from the editor, associate editor, and two anonymous referees greatly improved the article and are gratefully acknowledged. Useful comments were also provided by Alberto Abadie, Joshua Angrist, Xuan Chen, Guido Imbens, Chris Parmeter, and conference/seminar participants at the 7th IZA Conference on Labor Market Policy Evaluation at Harvard University, the 2012 Society of Labor Economists Meeting, the 2011 North American Summer Meeting of the Econometric Society, the 2010 NY Camp Econometrics, SUNY-Binghamton, Syracuse, Cal Poly-San Luis Obispo, and the Office of the Comptroller of the Currency. This work has been partially supported by the NSF under grants SES-0852211 and SES-0852139. Flores also acknowledges funding from the McLam-ore summer research award at the University of Miami. Competent research assistance was provided by Maria Bampasidou. All errors are our own.
Publisher Copyright:
© 2013 American Statistical Association.
PY - 2013
Y1 - 2013
N2 - We derive nonparametric bounds for local average treatment effects (LATE) without imposing the exclusion restriction assumption or requiring an outcome with bounded support. Instead, we employ assumptions requiring weak monotonicity of mean potential and counterfactual outcomes within or across subpopulations defined by the values of the potential treatment status under each value of the instrument. The key element in our derivation is a result relating LATE to a causal mediation effect, which allows us to exploit partial identification results from the causal mediation analysis literature. The bounds are employed to analyze the effect of attaining a GED, high school, or vocational degree on future labor market outcomes using randomization into a training program as an invalid instrument. The resulting bounds are informative, indicating that the local effect when assigned to training for those whose degree attainment is affected by the instrument is at most 12.7 percentage points on employment and $64.4 on weekly earnings.
AB - We derive nonparametric bounds for local average treatment effects (LATE) without imposing the exclusion restriction assumption or requiring an outcome with bounded support. Instead, we employ assumptions requiring weak monotonicity of mean potential and counterfactual outcomes within or across subpopulations defined by the values of the potential treatment status under each value of the instrument. The key element in our derivation is a result relating LATE to a causal mediation effect, which allows us to exploit partial identification results from the causal mediation analysis literature. The bounds are employed to analyze the effect of attaining a GED, high school, or vocational degree on future labor market outcomes using randomization into a training program as an invalid instrument. The resulting bounds are informative, indicating that the local effect when assigned to training for those whose degree attainment is affected by the instrument is at most 12.7 percentage points on employment and $64.4 on weekly earnings.
KW - Causal inference
KW - Instrumental variables
KW - Mediation analysis
KW - Nonparametric bounds
KW - Principal stratification
UR - http://www.scopus.com/inward/record.url?scp=84897727553&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84897727553&partnerID=8YFLogxK
U2 - 10.1080/07350015.2013.822760
DO - 10.1080/07350015.2013.822760
M3 - Article
AN - SCOPUS:84897727553
SN - 0735-0015
VL - 31
SP - 534
EP - 545
JO - Journal of Business and Economic Statistics
JF - Journal of Business and Economic Statistics
IS - 4
ER -