Jesse Rothstein

I investigate medium- and long-term impacts of the Great Recession on post-recession college graduates. Most so-called “scarring” models emphasize effects of initial conditions that attenuate over the first decade of a worker’s career. But early career recessions may also have permanent effects. I decompose the recent cohorts’ experience into transitory time effects, medium-term scarring, and permanent cohort effects. Cohort effects are strongly cyclical. Medium-term scarring explains only half of this cyclicality. The long-run cumulative effect of the recession on graduates’ employment is more than twice as large as the immediate effect.

Intergenerational income transmission varies across commuting zones (CZs). I investigate whether children’s educational outcomes help to explain this variation. Differences among CZs in the relationship between parental income and children’s human capital explain only one-ninth of the variation in income transmission. A similar share is explained by differences in the return to human capital. One-third reflects earnings differences not mediated by human capital, and 40% reflects differences in marriage patterns. Intergenerational mobility appears to reflect job networks and the structure of local labor and marriage markets more than it does the education system.

The synthetic control method (SCM) is a popular approach for estimating the impact of a treatment on a single unit in panel data settings. The “synthetic control” is a weighted average of control units that balances the treated unit’s pre-treatment outcomes as closely as possible. The curse of dimensionality, however, means that SCM does not generally achieve exact balance, which can bias the SCM estimate. We propose an extension, Augmented SCM, which uses an outcome model to estimate the bias due to covariate imbalance and then de-biases the original SCM estimate, analogous to bias correction for inexact matching. We motivate this approach by showing that SCM is a (regularized) inverse propensity score weighting estimator, with pre-treatment outcomes as covariates and a ridge penalty on the propensity score coefficients. We give theoretical guarantees for specific cases and propose a new inference procedure. We demonstrate gains from Augmented SCM with extensive simulation studies and apply this framework to canonical SCM examples. We implement the proposed method in the new augsynth R package.

We discuss the potential role of Universal Basic Incomes (UBIs) in advanced countries. A feature of advanced economies that distinguishes them from developing countries is the existence of well developed, if often incomplete, safety nets. We develop a framework for describing transfer programs, flexible enough to encompass most existing programs as well as UBIs, and use this framework to compare various UBIs to the existing constellation of programs in the United States. A UBI would direct much larger shares of transfers to childless, non-elderly, non-disabled households than existing programs, and much more to middle-income rather than poor households. A UBI large enough to increase transfers to low-income families would be enormously expensive. We review the labor supply literature for evidence on the likely impacts of a UBI. We argue that the ongoing UBI pilot studies will do little to resolve the major outstanding questions.

We study the impact of post-1990 school finance reforms, during the so-called "adequacy" era, on absolute and relative spending and achievement in low-income school districts. Using an event study research design that exploits the apparent randomness of reform timing, we show that reforms lead to sharp, immediate, and sustained increases in spending in low-income school districts. Using representative samples from the National Assessment of Educational Progress, we find that reforms cause increases in the achievement of students in these districts, phasing in gradually over the years following the reform. The implied effect of school resources on educational achievement is large.

The Earned Income Tax Credit (EITC) is a refundable tax credit that promotes work. Research has shown that it also reduces poverty and improves health and education outcomes. The maximum credit for families with two or fewer children has remained flat in inflation-adjusted terms since 1996. Over the same period, earnings prospects have stagnated or diminished for many Americans, and prime-age employment rates have fallen. This paper proposes to build on the successes of the EITC with a ten percent acrossthe-board increase in the federal credit. This expansion would provide a meaningful offset to stagnating real wages, encourage more people to enter employment, lift approximately 600,000 individuals out of poverty, and improve health and education outcomes for millions of children.

Many Unemployment Insurance (UI) recipients do not find new jobs before exhausting their benefits, even when benefits are extended during recessions. Using Survey of Income and Program Participation (SIPP) panel data covering the 2001 and 2007 to 2009 recessions and their aftermaths, we identify individuals whose jobless spells outlasted their UI benefits (exhaustees) and examine household income, program participation, and health‐related outcomes during the six months following UI exhaustion. For the average exhaustee, the loss of UI benefits is only slightly offset by increased participation in other safety net programs (e.g., food stamps), and family poverty rates rise substantially. Self‐reported disability also rises following UI exhaustion. These patterns do not vary dramatically across household demographic groups, broad income level prior to job loss, or the two business cycles. The results highlight the unique, important role of UI in the U.S. social safety net.

Chetty, Friedman, and Rockoff (2014a, b) study value-added (VA) measures of teacher effectiveness. CFR (2014a) exploits teacher switching as a quasi-experiment, concluding that student sorting creates negligible bias in VA scores. CFR (2014b) finds VA scores are useful proxies for teachers' effects on students' long-run outcomes. I successfully reproduce each in North Carolina data. But I find that the quasi-experiment is invalid, as teacher switching is correlated with changes in student preparedness. Adjusting for this, I find moderate bias in VA scores, perhaps 10-35 percent as large, in variance terms, as teachers' causal effects. Long-run results are sensitive to controls and cannot support strong conclusions.

The years since the 2009 end of the Great Recession have been disastrous for many workers, particularly those with low human capital or other disadvantages. One explanation attributes this to deficient aggregate labor demand, to which marginal workers are more sensitive. A second attributes it to structural changes. Cyclical explanations imply that if aggregate labor demand is increased then many of the post-2009 patterns will revert to their pre-recession trends. Structural explanations suggest recent experience is the “new normal.” This paper reviews data since 2007 for evidence. I examine wage trends to measure the relative importance of supply and demand. I find little wage pressure before 2015, pointing to demand as the binding constraint. The most recent data show some signs of tightness, but still substantial slack.

In this paper, we review the most prominent provision of the federal income tax code that targets low-income tax filers, the Earned Income Tax Credit (EITC), as well as the structurally similar Child Tax Credit (CTC). We frame the paper around what we see as the programs’ goals: distributional, promoting work, and limiting administrative and compliance costs. We review what is known about program impacts and distributional consequences under current law, drawing on simulations from the Tax Policy Center. We conclude that the EITC is quite successful in meeting its three goals. In contrast, most of the benefits of the CTC go to higher income households. In addition to analyzing current law, we assess possible reforms that would reach groups – for the EITC, those without children; for the CTC, those with very low earnings – who are largely missed under current policy.

Large-scale social experiments were pioneered in labor economics, and are the basis for much of what we know about topics ranging from the effect of job training to incentives for job search to labor supply responses to taxation. Random assignment has provided a powerful solution to selection problems that bedevil non-experimental research. Nevertheless, many important questions about these topics require going beyond random assignment. This applies to questions pertaining to both internal and external validity, and includes effects on endogenously observed outcomes, such as wages and hours; spillover effects; site effects; heterogeneity in treatment effects; multiple and hidden treatments; and the mechanisms producing treatment effects. In this Chapter, we review the value and limitations of randomized social experiments in the labor market, with an emphasis on these design issues and approaches to addressing them. These approaches expand the range of questions that can be answered using experiments by combining experimental variation with econometric or theoretical assumptions. We also discuss efforts to build the means of answering these types of questions into the ex ante design of experiments. Our discussion yields an overview of the expanding toolkit available to experimental researchers.

Economists often use test scores to measure a student’s performance or an adult’s human capital. These scores reflect nontrivial decisions about how to measure and scale student achievement, with important implications for secondary analyses. For example, the scores computed in several major testing regimes, including the National Assessment of Educational Progress (NAEP), depend not only on the examinees’ responses to test items, but also on their background characteristics, including race and gender. As a consequence, if a black and white student respond identically to questions on the NAEP assessment, the reported ability for the black student will be lower than for the white student—reflecting the lower average performance of black students. This can bias many secondary analyses. Other assessments use different measurement models. This paper aims to familiarize applied economists with the construction and properties of common cognitive score measures and the implications for research using these measures.

Social Security Disability Insurance (SSDI) awards rise during recessions. If marginal applicants are able to work but unable to find jobs, countercyclical Unemployment Insurance (UI) benefit extensions may reduce SSDI uptake. Exploiting UI extensions in the Great Recession as a source of variation, we find no indication that expiration of UI benefits causes SSDI applications and can rule out effects of meaningful magnitude. A supplementary analysis finds little overlap between the two programs’ recipient populations: only 28% of SSDI awardees had any labor force attachment in the prior calendar year, and of those, only 4% received UI.

Unemployment Insurance benefit durations were extended during the Great Recession, reaching 99 weeks for most recipients. The extensions were rolled back and eventually terminated by the end of 2013. Using matched CPS data from 2008-2014, we estimate the effect of extended benefits on unemployment exits separately during the earlier period of benefit expansion and the later period of rollback. In both periods, we find little or no effect on job-finding but a reduction in labor force exits due to benefit availability. We estimate that the rollbacks reduced the labor force participation rate by about 0.1 percentage point in early 2014.

Analysts often examine the black-white test score gap conditional on current family income. We describe a method for identifying the gap conditional on the family’s permanent income. Current income explains only about half as much of the black-white test gap as does permanent income, and the gap among families with the same permanent income is only 0.2 to 0.3 standard deviations in two commonly used samples. When we add permanent income to the controls used by Fryer and Levitt (2006), the unexplained gap in third grade shrinks below 0.15 SDs, less than half of what is found with their controls.

Four years after the beginning of the Great Recession, the labor market remains historically weak. Many observers have concluded that "structural" impediments to recovery bear some of the blame. The author reviews such structural explanations, but after analyzing U.S. data on unemployment and productivity, he finds there is little evidence supporting these hypotheses. He finds that the bulk of the evidence is more consistent with the hypothesis that continued poor performance is primarily attributable to shortfalls in the aggregate demand for labor.

More than 2 years after the official end of the Great Recession, the labor market remains historically weak. One candidate explanation is supply-side effects driven by dramatic expansions of unemployment insurance (UI) benefit durations, to as long as 99 weeks. This paper investigates the effect of these extensions on job search and reemployment. I use the longitudinal structure of the Current Population Survey to construct unemployment exit hazards that vary across states, over time, and between individuals with differing unemployment durations. I then use these hazards to explore a variety of comparisons intended to distinguish the effects of UI extensions from other determinants of employment outcomes. The various specifications yield quite similar results. UI extensions had significant but small negative effects on the probability that the eligible unemployed would exit unemployment. These effects are concentrated among the long-term unemployed. The estimates imply that UI extensions raised the unemployment rate in early 2011 by only about 0.1 to 0.5 percentage point, much less than implied by previous analyses, with at least half of this effect attributable to reduced labor force exit among the unemployed rather than to the changes in reemployment rates that are of greater policy concern.

In the early 2000s, a highly selective university introduced a “no-loans” policy under which the loan component of financial aid awards was replaced with grants. We use this natural experiment to identify the causal effect of student debt on employment outcomes. In the standard life-cycle model, young people make optimal educational investment decisions if they are able to finance these investments by borrowing against future earnings; the presence of debt has only income effects on investment decisions. We find that debt causes graduates to choose substantially higher-salary jobs and reduces the probability that students choose low-paid “public interest” jobs. We also find some evidence that debt affects students' academic decisions during college. Our estimates suggest that recent college graduates are not life-cycle agents. Two potential explanations are that young workers are credit constrained or that they are averse to holding debt. We find suggestive evidence that debt reduces students' donations to the institution in the years after they graduate and increases the likelihood that a graduate will default on a pledge made during her senior year; we argue this result is more likely consistent with credit constraints than with debt aversion.

The EITC is intended to encourage work. But EITC-induced increases in labor supply may drive wages down. I simulate the economic incidence of the EITC. In each scenario that I consider, a large portion of low-income single mothers’ EITC payments is captured by employers through reduced wages. Workers who are EITC ineligible also see wage declines. By contrast, a traditional Negative Income Tax (NIT) discourages work, and so induces large transfers from employers to their workers. With my preferred parameters, $1 in EITC spending increases after-tax incomes by $0.73, while $1 spent on the NIT yields $1.39.

Despite extensive public infrastructure spending, surprisingly little is known
about its economic return. In this paper, we estimate the value of school facility
investments using housing markets: standard models of local public goods imply
that school districts should spend up to the point where marginal increases would
have zero effect on local housing prices. Our research design isolates exogenous
variation in investments by comparing school districts where referenda on bond
issues targeted to fund capital expenditures passed and failed by narrow margins. We extend this traditional regression discontinuity approach to identify the dynamic treatment effects of bond authorization on local housing prices, student
achievement, and district composition. Our results indicate that California school
districts underinvest in school facilities: passing a referendum causes immediate, sizable increases in home prices, implying a willingness to pay on the part of marginal homebuyers of $1.50 or more for each $1 of capital spending. These
effects do not appear to be driven by changes in the income or racial composition of
homeowners, and the impact on test scores appears to explain only a small portion
of the total housing price effect.

Growing concerns over the inadequate achievement of U.S. students have
led to proposals to reward good teachers and penalize (or fire) bad ones. The
leading method for assessing teacher quality is “value added” modeling (VAM),
which decomposes students’ test scores into components attributed to student
heterogeneity and to teacher quality. Implicit in the VAM approach are strong
assumptions about the nature of the educational production function and the
assignment of students to classrooms. In this paper, I develop falsification tests
for three widely used VAM specifications, based on the idea that future teachers
cannot influence students’ past achievement. In data from North Carolina, each
of the VAMs’ exclusion restrictions is dramatically violated. In particular, these
models indicate large “effects” of fifth grade teachers on fourth grade test score
gains. I also find that conventional measures of individual teachers’ value added
fade out very quickly and are at best weakly related to long-run effects. I discuss
implications for the use of VAMs as personnel tools.

Nonrandom assignment of students to teachers can bias value-added estimates of teachers’ causal effects. Rothstein (2008, 2010) shows that typical value-added models indicate large counterfactual effects of fifthgrade teachers on students’ fourth-grade learning, indicating that classroom assignments are far from random.This article quantifies the resulting biases in estimates of fifth-grade teachers’ causal effects from several valueadded models, under varying assumptions about the assignment process. If assignments are assumed to depend only on observables, the most commonly used specifications are subject to important bias, but other feasible specifications are nearly free of bias. I also consider the case in which assignments depend on unobserved variables. I use the across-classroom variance of observables to calibrate several models of the sorting process. Results indicate that even the best feasible value-added models may be substantially biased, with the magnitude of the bias depending on the amount of information available for use in classroom assignments.

Data from college admissions tests can provide a valuable measure of student achievement, but the non-representativeness of test-takers is an important concern. We examine selectivity bias in both state-level and school-level SAT and ACT averages. The degree of selectivity may differ importantly across and within schools, and across and within states. To identify within-state selectivity, we use a control function approach that conditions on scores from a representative test. Estimates indicate strong selectivity of test-takers in “ACT states,” where most college-bound students take the ACT, and much less selectivity in SAT states. To identify within- and between-school selectivity, we take advantage of a policy reform in Illinois that made taking the ACT a graduation requirement. Estimates based on this policy change indicate substantial positive selection into test participation both across and within schools. Despite this, school-level averages of observed scores are extremely highly correlated with average latent scores, as across-school variation in sample selectivity is small relative to the underlying signal. As a result, in most contexts the use of observed school mean test scores in place of latent means understates the degree of between-school variation in achievement but is otherwise unlikely to lead to misleading conclusions.

The Supreme Court has held repeatedly that race-based preferences in public university admissions are constitutional. But debates over the wisdom of affirmative action continue. Opponents of these policies argue that preferences are detrimental to minority students -- that by placing these students in environments that are too competitive, affirmative action hurts their academic and career outcomes.

This article examines the so-called "mismatch" hypothesis in the context of law school admissions. We discuss the existing scholarship on mismatch, identifying methodological limitations of earlier attempts to measure the effects of affirmative action. Using a simpler, more robust analytical strategy, we find that the data are inconsistent with large mismatch effects, particularly with respect to employment outcomes. While moderate mismatch effects are possible, they are concentrated among the students with the weakest entering academic credentials.

To put our estimates in context, we simulate admissions under race-blind rules. Eliminating affirmative action would dramatically reduce the number of black law students, particularly at the most selective schools. Many potentially successful black law students would be excluded, far more than the number who would be induced to pass the bar exam by the elimination of mismatch effects. Accordingly, we find that eliminating affirmative action would dramatically reduce the production of black lawyers.

Schelling (“Dynamic Models of Segregation,” Journal of Mathematical Sociology 1 (1971), 143–186) showed that extreme segregation can arise from social interactions in white preferences: once the minority share in a neighborhood exceeds a “tipping point,” all the whites leave. We use regression discontinuity methods and Census tract data from 1970 through 2000 to test for discontinuities in the dynamics of neighborhood racial composition. We find strong evidence that white population flows exhibit tipping-like behavior in most cities, with a distribution of tipping points ranging from 5% to 20% minority share. Tipping is prevalent both in the suburbs and near existing minority enclaves. In contrast to white population flows, there is little evidence of nonlinearities in rents or housing prices around the tipping point. Tipping points are higher in cities where whites have more tolerant racial attitudes.

Racial segregation is often blamed for some of the achievement gap between blacks and whites. We study the effects of school and neighborhood segregation on the relative SAT scores of black students across different metropolitan areas, using large microdata samples for the 1998–2001 test cohorts. Our models include detailed controls for the family background of individual test-takers, school-level controls for selective participation in the test, and city-level controls for racial composition, income, and region. We find robust evidence that the black–white test score gap is higher in more segregated cities. Holding constant family background and other factors, a shift from a highly segregated city to a nearly integrated city closes about one-quarter of the raw black–white gap in SAT scores. Specifications that distinguish between school and neighborhood segregation suggest that neighborhood segregation has a consistently negative impact while school segregation has no independent effect, though we cannot reject equality of the two effects. Additional tests indicate that much of the effect of neighborhood segregation operates through neighbors' incomes, not through race per se. Data on enrollment in honors courses suggest that within-school segregation increases when schools are more highly integrated, potentially offsetting the benefits of school desegregation and accounting for our findings.

School choice policies may, by aligning administrators’ incentives with parental demand, yield improved efficiency in educational production (Milton Friedman, 1962; John E. Chubb and Terry M. Moe, 1990). But Eric A. Hanushek (1981) cautions: “If the efficiency of our school systems is due to poor incentives for teachers and administrators coupled with poor decision-making by consumers, it would be unwise to expect much from programs that seek to strengthen ‘market forces’ in the selection of schools” (p. 35, emphasis added). Poor decision-making is not required; parents may rationally choose schools with “pleasant surroundings, athletic facilities, cultural advantages” (ibid., p. 34) over those that most efficiently pursue academic performance; they may prefer poorly run schools with good peer groups over those that are more effective but enroll worse students (J. Douglas Willms and Frank H. Echols, 1992, 1993); or they may simply be unable to identify effective schools (Thomas J. Kane and Douglas O. Staiger, 2002). Any factor that leads parents to choose any but the most effective available schools will tend to dilute the incentives for efficient management that choice might otherwise create.

This study examines the distribution of student outcomes across schools within metropolitan housing markets for evidence on parental demand. Economists have long noted that parents’ choices among residential locations are potentially informative about how more complete choice systems may operate (Charles M. Tiebout, 1956; Melvin V. Borland and Roy M. Howsen, 1992; Caroline M. Hoxby, 2000; Rothstein, 2005). I ask whether school effectiveness plays a sufficiently important role in these decisions to create meaningful incentives for more productive school management.

In Grutter v. Bollinger, Justice O’Connor conjectured that in 25 years affirmative action in college admissions will be unnecessary. We project the test score distribution of black and white college applicants 25 years from now, focusing on the role of black–white family income gaps. Economic progress alone is unlikely to narrow the achievement gap enough in 25 years to produce today’s racial diversity levels with race-blind admissions. A return to the rapid black–white test score convergence of the 1980s could plausibly cause black representation to approach current levels at moderately selective schools, but not at the most selective schools.

In her opinion in Grutter v. Bollinger, Justice Sandra Day O’Connor concluded that affirmative action in college admissions is justifiable, but not in perpetuity: “We expect that 25 years from now, the use of racial preferences will no longer be necessary to further the interest [in student body diversity] approved today.”

The rate at which racial gaps in precollegiate academic achievement can plausibly be expected to erode is a matter of considerable uncertainty. In this essay, we attempt to evaluate the plausibility of Justice O’Connor’s conjecture by projecting the racial composition of the 2025 elite college applicant pool. Our projections extrapolate past trends on two important margins: Gaps between the economic resources of black and white students’ families, and narrowing of test score gaps between black and white students with similar family incomes. Just as the last decades have seen considerable narrowing of gaps on each margin, further progress can be expected over the next quarter century.
Our central question is whether this progress will plausibly be fast enough to validate Justice O’Connor’s prediction. We are well aware of the hazards inherent in our exercise: No such distant projections can be definitive. Nevertheless, by relying on reasonable historical assumptions that are arguably optimistic, we develop a baseline case for assessing the likelihood of O’Connor’s forecast.

We conclude that under reasonable assumptions, African American students will continue to be substantially underrepresented among the most qualified college applicants for the foreseeable future. The magnitude of the underrepresentation is likely to shrink—in our most optimistic simulation, somewhat over half of the gap that would be opened by the elimination of race preferences will be closed by the projected improvement in black achievement.

Still, it seems unlikely that today’s level of racial diversity will be achievable without some form of continuing affirmative action. If the Supreme Court follows through with O’Connor’s stated intention to ban affirmative action in 25 years, and if colleges do not adjust in other ways (such as reducing the importance of numerical qualifications to admissions), we project substantial declines in the representation of African Americans among admitted students at selective institutions.

Our analysis proceeds from the assumption that the most likely future course will resemble past trends. Substantial changes in educational policy, in school effectiveness, and in income inequality would all have important effects on black test score distributions and on the admissions landscape.

The methods used in most SAT validity studies cannot be justified by any sample selection assumptions and are uninformative about the source of the SAT's predictive power. A new omitted variables estimator is proposed; plausibly consistent estimates of the SAT's contribution to predictions of University of California freshman grade point averages are about 20% smaller than the usual methods imply. Moreover, much of the SAT's predictive power is found to derive from its correlation with high school demographic characteristics: The orthogonal portion of SAT scores is notably less predictive of future performance than is the unadjusted score.