This paper studies the use of civilian-based mobile crisis response teams (CRTs)—a non-uniformed pair consisting of a mental health worker and a medic—as a component of emergency response to 911 calls. We provide the first evaluation of the longest-running CRT program in the United States, Crisis Assistance Helping Out on the Streets (CAHOOTS) in Eugene, Oregon, which responds to calls involving mental illness, homelessness, and addiction either instead of or in addition to police officers. We use two complementary research designs to understand the effects and possible scope of these programs. First, we find that a series of program expansions into new areas and times reduced the likelihood that a 911 call resulted in an arrest by 24 percentage points. The arrest reduction likely reflects the role of CRTs in de-escalating tense situations and resolving incidents without coercive measures. CRTs are most often dispatched to the same calls as the police, acting as an addition rather than a substitute. Second, we exploit idiosyncratic variation in CRT availability in the post-expansion periods to estimate the effect of additional marginal program expansions. We find that most marginal responses do not affect arrests and would otherwise go unanswered, suggesting that the program has reached a scale where it can respond to the most urgent calls.
@unpublished{davis2026cahoots,
title={Mobile Crisis Response Teams Support Better Policing: Evidence from CAHOOTS},
author={Davis, Jonathan and Norris, Samuel and Schmitt, Jadon and Shem-Tov, Yotam and Strickland, Chelsea},
year={2026},
note={Working Paper}
}We study the long-run consequences of losing a low-wage job using linked employeremployee wage records and household surveys. For full-time workers earning $15 per hour or less, job loss due to an idiosyncratic firm-wide contraction decreases earnings six years later by 13% and cumulative earnings by over $40,000. Long-run losses stem primarily from reductions in employment and hours rather than wage rates and are concentrated among workers displaced from jobs in industries with higher average wages, tenure, unionization rates, and full-time share. By contrast, workers initially earning $15-$30 per hour see comparable long-run losses driven primarily by reductions in hourly wages.
@unpublished{rose2025lowwage,
title={How Replaceable Is a Low-Wage Job?},
author={Rose, Evan K. and Shem-Tov, Yotam},
year={2025},
note={Working Paper}
}This paper develops new approaches for estimating multi-dimensional teacher effects and uses them to understand teachers’ impacts on their students’ future criminal justice contact (CJC). Using a unique data set linking the universe of North Carolina public school data to administrative arrest records, we find a standard deviation of teacher effects on students’ future arrests of 2.7 percentage points (11% of the sample mean). Teachers’ effects on CJC are orthogonal to their effects on academic achievement, implying assignment to a high test score value-added teacher does not reduce future CJC. However, teachers who reduce suspensions and improve attendance substantially reduce future arrests. Similar patterns emerge when allowing teacher impacts to vary by student sex, race, and socio-economic status. The results suggest that the development of non-cognitive skills is central to the returns to education for crime and highlight an important dimension of teachers’ social value not targeted by test score-based quality metrics.
@unpublished{rose2025teachers,
title={The Effects of Teacher Quality on Adult Criminal Justice Contact},
author={Rose, Evan K. and Schellenberg, Jonathan and Shem-Tov, Yotam},
year={2025},
note={Working Paper}
}Researchers using instrumental variables to investigate ordered treatments often recode treatment into an indicator for any exposure. We investigate this estimand under the assumption that the instruments shift compliers from no treatment to some but not from some treatment to more. We show that when there are extensive margin compliers only (EMCO) this estimand captures a weighted average of treatment effects that can be partially unbundled into each complier group’s potential outcome means. We also establish an equivalence between EMCO and a two-factor selection model and apply our results to study treatment heterogeneity in the Oregon Health Insurance Experiment.
@article{rose2024recoding,
title={On Recoding Ordered Treatments as Binary Indicators},
author={Rose, Evan K. and Shem-Tov, Yotam},
journal={Review of Economics and Statistics},
year={2024},
note={Accepted}
}We study the effect of incarceration on wages, self‐employment, and taxes and transfers in North Carolina and Ohio using two quasi‐experimental research designs: discontinuities in sentencing guidelines and random assignment to judges. Across both states, incarceration generates short‐term drops in economic activity while individuals remain in prison. As a result, a year‐long sentence decreases cumulative earnings over five years by 13%. Beyond five years, however, there is no evidence of lower employment, wage earnings, or self‐employment in either state, as well as among defendants with no prior incarceration history. These results suggest that upstream factors, such as other types of criminal justice interactions or pre‐existing labor market detachment, are more likely to be the cause of low earnings among the previously incarcerated, who we estimate would earn just $5000 per year on average if spared a prison sentence.
@article{garin2025incarceration,
title={The Impact of Incarceration on Employment, Earnings, and Tax Filing},
author={Garin, Andrew and Koustas, Dmitri and McPherson, Carl and Norris, Samuel and Pecenco, Matthew and Rose, Evan K. and Shem-Tov, Yotam and Weaver, Jeffrey},
journal={Econometrica},
volume={93},
number={2},
pages={503--538},
year={2025}
}As millions of soldiers deployed to Iraq or Afghanistan between 2001 and 2021, Veteran Affairs Disability Compensation payments quadrupled and the veteran suicide rate rose rapidly. We estimate the causal contribution of combat deployments to declining veteran well-being. Deployments increase injuries, combat deaths, and disability compensation, but we find limited effects on suicide, deaths of despair, financial health, incarceration, or education. Our estimates suggest that deployment cannot explain either the recent rise in disability payments, which is more likely driven by policy changes, or the surge in noncombat deaths, which is better explained by shifts in observable characteristics of soldiers.
@article{bruhn2024deployments,
title={The Effects of Combat Deployments on Veterans' Outcomes},
author={Bruhn, Jesse and Greenberg, Kyle and Gudgeon, Matthew and Rose, Evan K. and Shem-Tov, Yotam},
journal={Journal of Political Economy},
volume={132},
number={8},
pages={2830--2879},
year={2024}
}This paper studies the effect of a restorative justice intervention targeted at 143 youth ages 13 to 17 facing felony charges of medium severity (e.g., burglary, assault). Eligible youths were randomly assigned to participate in the Make-it-Right (MIR) restorative justice program or a control group where they faced standard criminal prosecution. We estimate the effects of MIR on the likelihood that a youth will be rearrested in the four years following randomization. Assignment to MIR reduces the probability of a rearrest within six months by 19 percentage points, a 44 percent reduction relative to the control group. Moreover, the reduction in recidivism persists even four years after randomization. Thus, our estimates show that restorative justice conferencing can reduce recidivism among youth charged with relatively serious offenses and can be an effective alternative to traditional criminal justice practices.
@article{shemtov2024restorative,
title={Can Restorative Justice Conferencing Reduce Recidivism? Evidence From the Make-it-Right Program},
author={Shem-Tov, Yotam and Raphael, Steven and Skog, Alissa},
journal={Econometrica},
volume={92},
number={1},
pages={61--78},
year={2024}
}Most criminal defendants cannot afford to hire an attorney. To provide constitutionally mandated legal services, states commonly use either private court-appointed attorneys or a public defender organization. This paper investigates the relative efficacy of these two modes of indigent defense by comparing outcomes of codefendants assigned to different types of attorneys within the same case. Using data from San Francisco, I show that in multiple defendant cases, public defender assignment is plausibly as good as random. I find that public defenders reduce the probability of any prison sentence by 22% and the length of prison sentences by 10%.
@article{shemtov2022indigent,
title={Make or Buy? The Provision of Indigent Defense Services in the U.S.},
author={Shem-Tov, Yotam},
journal={Review of Economics and Statistics},
volume={104},
number={4},
pages={819--827},
year={2022}
}Despite interest in the role of school discipline in the creation of racial inequality, previous research has been unable to identify how students who receive suspensions in school differ from unsuspended classmates on key young adult outcomes. We utilize novel data to document the links between high school discipline and important young adult outcomes related to criminal justice contact, social safety net program participation, post-secondary education, and the labor market. We show that the link between school discipline and young adult outcomes tends to be stronger for Black students than for White students, and that inequality in exposure to school discipline accounts for approximately 30 percent of the Black-White disparities in young adult criminal justice outcomes and SNAP receipt.
@article{davison2022discipline,
title={School Discipline and Racial Disparities in Early Adulthood},
author={Davison, Miles and Penner, Andrew and Penner, Emily and Pharris-Ciurej, Nikolas and Porter, Sonya R. and Rose, Evan K. and Shem-Tov, Yotam and Yoo, Paul},
journal={Educational Researcher},
volume={51},
number={3},
pages={231--234},
year={2022}
}We study the causal effect of incarceration on reoffending using discontinuities in North Carolina’s sentencing guidelines. A regression discontinuity analysis shows that 1 year of incarceration causes a reduction in the likelihood of being reincarcerated within 3, 5, and 8 years from sentencing by 44%, 29%, and 21%, respectively. To parse the potentially heterogeneous dose response relationship underlying these effects, we develop an econometric model of prison sentences and recidivism. We find that incarceration has meaningful reoffending-reducing average effects that diminish in incarceration length. As a result, budget-neutral reductions in sentence length combined with increases in incarceration rates can decrease recidivism.
@article{rose2021incarceration,
title={How Does Incarceration Affect Reoffending? Estimating the Dose-Response Function},
author={Rose, Evan K. and Shem-Tov, Yotam},
journal={Journal of Political Economy},
volume={129},
number={12},
pages={3302--3356},
year={2021}
}We derive new variance formulas for inference on a general class of estimands of causal average treatment effects in a randomized control trial. We generalize the seminal work of Robins and show that when the researcher’s objective is inference on sample average treatment effect of the treated (SATT), a consistent variance estimator exists. Although this estimand is equal to the sample average treatment effect (SATE) in expectation, potentially large differences in both accuracy and coverage can occur by the change of estimand, even asymptotically. Inference on SATE, even using a conservative confidence interval, provides incorrect coverage of SATT. We demonstrate the applicability of the new theoretical results using an empirical application with hundreds of online experiments with an average sample size of approximately 100 million observations per experiment. An R package, estCI, that implements all the proposed estimation procedures is available. Supplementary materials for this article are available online.
@article{sekhon2021inference,
title={Inference on a New Class of Sample Average Treatment Effects},
author={Sekhon, Jasjeet S. and Shem-Tov, Yotam},
journal={Journal of the American Statistical Association},
volume={116},
number={534},
pages={798--804},
year={2021}
}The gold standard for identifying causal relationships is a randomized controlled experiment. In many applications in the social sciences and medicine, the researcher does not control the assignment mechanism and instead may rely upon natural experiments or matching methods as a substitute to experimental randomization. The standard testable implication of random assignment is covariate balance between the treated and control units. Covariate balance is commonly used to validate the claim of as good as random assignment. We propose a new nonparametric test of covariate balance. Our Classification Permutation Test (CPT) is based on a combination of classification methods (e.g., random forests) with Fisherian permutation inference. We revisit four real data examples and present Monte Carlo power simulations to demonstrate the applicability of the CPT relative to other nonparametric tests of equality of multivariate distributions.
@article{gagnonbartsch2019classification,
title={The Classification Permutation Test: A Flexible Non-Parametric Test for Equality of Multivariate Distributions},
author={Gagnon-Bartsch, Johann and Shem-Tov, Yotam},
journal={Annals of Applied Statistics},
volume={13},
number={3},
pages={1464--1483},
year={2019}
}Since the seminal papers of Fehr and Schmidt (1999) and Bolton and Ockenfels (2000), fairness has become an important discussion point in economics. Is it unfair that different people pay different prices for the same good or service? We provide what we believe to be a novel approach: We let normal everyday consumers play the role of sellers who have access to consumers’ data (and willingness to pay). A strong finding of behaviour in this setup is that subjects charge a fixed percentage (approximately 64%) of the willingness to pay from each of their subjects, leading to a fair, whilst uneven, distribution of prices. Interestingly, this 64% price level does not change when we vary the number of sellers competing in the market.
@article{shemtov2015fairness,
title={A Note on Fairness and Personalised Pricing},
author={Shem-Tov, Yotam and Vulkan, Nir},
journal={Economics Letters},
volume={136},
pages={179--183},
year={2015}
}This study provides first empirical results on entrepreneurs’ negotiation behavior. In a series of negotiation tasks, we compare persuasive behaviors and negotiation outcomes of entrepreneurs and non-entrepreneurs. Our results show that entrepreneurs make extensive use of emotions and arguments as means of persuasion. Due to their assertive behavior, they close fewer deals; however, when they close a deal, they make higher profits than non-entrepreneurs. These results demonstrate the relevance of studying entrepreneurs’ interpersonal interactions as determinants of entrepreneurial success and highlight the role expressed emotions and arguments play in this context.
@article{artinger2015entrepreneurs,
title={Entrepreneurs' Negotiation Behavior},
author={Artinger, Sabrina and Shem-Tov, Yotam and Vulkan, Nir},
journal={Small Business Economics},
volume={44},
pages={737--757},
year={2015}
}We extend the zero tax at the top result obtained in the closed economy case with bounded skill distributions for the case of unbounded skill distributions in the presence of international labor mobility and tax competition. We show that in the equilibrium for the tax competition game, the optimal marginal income tax rate converges to zero as the income level tends to infinity. We further show in simulations that the zero-marginal tax result is not a local property: over a substantial range at the higher end of the income distribution, the optimal tax is approximately given by a lump-sum tax set at its Laffer rate. We further show that the range in which the optimal marginal tax is approximately set to zero is widening as migration costs decrease.
@article{blumkin2015tax,
title={International Tax Competition: Zero Tax Rate at the Top Re-Established},
author={Blumkin, Tomer and Sadka, Efraim and Shem-Tov, Yotam},
journal={International Tax and Public Finance},
volume={22},
pages={760--776},
year={2015}
}In this paper we demonstrate that supplementing the optimal non-linear income tax system with a binding maximum wage rule attains a Pareto improvement, by serving to mitigate the mimicking incentives of the high-skill individuals without entailing distortions.
@article{blumkin2013maximum,
title={A Case for Maximum Wage},
author={Blumkin, Tomer and Sadka, Efraim and Shem-Tov, Yotam},
journal={Economics Letters},
volume={120},
number={3},
pages={374--378},
year={2013}
}