Inference on Causal Effects in a Generalized Regression Kink Design

Card, David; Pei, Zhuan; Lee, David; Wéber, Andrea

doi:10.2139/ssrn.2553874

Cited by 19 publications

(46 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The CCT was estimated using the triangular kernel. The robustness of the result using the CCT estimator is reassuring, since some results have previously been sensitive to the choice of method (e.g., Card, Johnston, Leung, Mas, & Pei, ; Card, Pei, Lee, & Weber, ).…”

Section: Resultsmentioning

confidence: 99%

“…To implement RKD, it is crucial to check for the smoothness conditions of covariates at the kink point (see Corollary 2 in Card, Pei, et al, , p. 2469). Following Card et al (), we study the smoothness of a covariate index, namely the predicted values of a regression of sickness duration on a set of 31 covariates (sex, age, age squared, living in the Helsinki region, higher education, year, diagnosis at the one‐letter level).…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

A kink that makes you sick: The effect of sick pay on absence

Böckerman

Kanninen

Suoniemi

2018

J of Applied Econometrics

View full text Add to dashboard Cite

SummaryWe exploit a regression kink design to estimate the elasticity of the duration of sickness absence with respect to replacement rate. Elasticity is a central parameter in defining the optimal social insurance scheme compensating for lost earnings due to sickness. We use comprehensive administrative data and a kink in the policy rule near the median earnings. We find a statistically significant estimate of the elasticity of the order of one. | INTRODUCTIONAbsenteeism leads to sizeable losses of working time worldwide. In some OECD countries, nearly 10% of annual working days are lost because of sickness absence (DICE Database, 2017; see Treble & Barmby, 2011). The costs of absenteeism are considerable for individuals themselves, employers, co-workers, and health and benefit systems. A sickness insurance system is designed to protect individuals from earnings losses. Among cash benefits, sickness insurance is one of the most important social protection schemes in Europe (Eurostat, 2017). The key policy parameter of the system is the replacement rate, that is, the ratio of sickness insurance benefits to past earnings.We examine the effect of the replacement rate of the Finnish sickness insurance on the duration of sickness absence. We find a substantial and robust behavioral response in a universal insurance scheme. The statistically significant point estimate of the elasticity of the duration of sickness absence with respect to the replacement rate is centered around one. Elasticity refers to the intensive margin, that is, the incentive effect conditional on being sick. The point estimate is lower when we exclude long spells.The elasticity of the duration of sickness absence is a vital parameter of an optimal sickness insurance system, because the replacement rate affects workers' financial incentives to be absent from work through a moral hazard or hidden action effect. In the absence of externalities, an optimal social insurance system balances the marginal costs of more generous payments to the sick, with the welfare gain resulting from the consumption smoothing that the benefits allow, captured in the Baily-Chetty formula (Baily, 1978;Chetty, 2006;Pichler & Ziebarth, 2017).We use a regression kink design (RKD; see Section 3) to identify the causal effect of the replacement rate. The method is similar to regression discontinuity design, where one exploits level jumps in policy rules, but in the RKD the slope of the policy rule changes and we analyze whether the slope of the behavioral response changes as well. Unlike in most other countries (Frick & Malo, 2008, pp. 510-511), the compensation of sickness insurance in Finland is not a fixed fraction of past earnings. Instead, the policy rule follows a piecewise linear scheme with a replacement rate that decreases with earnings. The institutional setting allows us to use RKD, in which the identification of the effect is based on a predetermined, nonlinear benefit function (Card, Lee, Pei, & Weber, 2012;Nielsen, Sørensen, & Taber, 2010). ---------------------...

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

A kink that makes you sick: The effect of sick pay on absence

Böckerman

Kanninen

Suoniemi

2018

J of Applied Econometrics

View full text Add to dashboard Cite

show abstract

“…The assumptions necessary for the consistency of this estimator as discussed by Card et al () are: (1) the direct marginal impact of need on y , g ( need ), is continuously differentiable at 0; (2) the density of need conditional on U is continuously differentiable at the threshold for Title IV loan eligibility; (3) monotonicity at the kink, that is, the change in the slope between TitleIV and need is of the same sign for all individuals. This last assumption is clearly satisfied in our case because borrowing is an increasing function of need .…”

Section: The Rk Frameworkmentioning

confidence: 99%

“…Card et al () impose an additional identifying assumption that the right and left limits of TitleIV ( need ) are equal at the threshold. In theory (and practice), this assumption is satisfied by the need formula.…”

mentioning

confidence: 99%

Life After Debt: Postgraduation Consequences of Federal Student Loans

Gervais¹,

Ziebarth

2019

Economic Inquiry

View full text Add to dashboard Cite

We estimate the causal effect of student loans on postgraduation labor market outcomes exploiting a kink in the formula determining eligibility for need‐based student loans. Using a representative sample of students graduating with a bachelor's degree in 1993, we find that student debt has nonnegative effects on earnings. This result holds with differing levels of statistical significance across a battery of different empirical designs: (1) ordinary least squares, (2) partially linear, and (3) regression kink. We find similar results for the 2008 graduating cohort. (JEL I22, I26, J32, J33)

show abstract

“…The Supporting Information Appendix reports additional methodological discussions and empirical results, and presents an extension of our new randomization‐based methods to the case of “fuzzy” RD designs where compliance is imperfect . See Ganong and Jäger () for permutation‐based inference in “kink” RD designs (Card et al., ; Chiang & Sasaki, ).…”

Section: Introductionmentioning

confidence: 99%

Comparing Inference Approaches for RD Designs: A Reexamination of the Effect of Head Start on Child Mortality

Cattaneo¹,

Titiunik²,

Vázquez-Baré³

2017

J. Pol. Anal. Manage.

137

145

View full text Add to dashboard Cite

The regression discontinuity (RD) design is a popular quasi‐experimental design for causal inference and policy evaluation. The most common inference approaches in RD designs employ “flexible” parametric and nonparametric local polynomial methods, which rely on extrapolation and large‐sample approximations of conditional expectations using observations somewhat near the cutoff that determines treatment assignment. An alternative inference approach employs the idea of local randomization, where the very few units closest to the cutoff are regarded as randomly assigned to treatment and finite‐sample exact inference methods are used. In this paper, we contrast these approaches empirically by re‐analyzing the influential findings of Ludwig and Miller (), who studied the effect of Head Start assistance on child mortality employing parametric RD methods. We first review methods based on approximations of conditional expectations, which are relatively well developed in the literature, and then present new methods based on randomization inference. In particular, we extend the local randomization framework to allow for parametric adjustments of the potential outcomes; our extended framework substantially relaxes strong assumptions in prior literature and better resembles other RD inference methods. We compare all these methods formally, focusing on both estimands and inference properties. In addition, we develop new approaches for randomization‐based sensitivity analysis specifically tailored to RD designs. Applying all these methods to the Head Start data, we find that the original RD treatment effect reported in the literature is quite stable and robust, an empirical finding that enhances the credibility of the original result. All the empirical methods we discuss are readily available in general purpose software in R and Stata; we also provide the dataset and software code needed to replicate all our results.

show abstract

Inference on Causal Effects in a Generalized Regression Kink Design

Cited by 19 publications

References 52 publications

A kink that makes you sick: The effect of sick pay on absence

A kink that makes you sick: The effect of sick pay on absence

Life After Debt: Postgraduation Consequences of Federal Student Loans

Comparing Inference Approaches for RD Designs: A Reexamination of the Effect of Head Start on Child Mortality

Contact Info

Product

Resources

About