Working Papers 


The Heterogeneous Long Run Effects of Universal Preschool


Working Paper, March 2023 (awaiting disclosure) Joint with Alex Smith and Jonathan Eggleston


Using universe tax data and a difference-in-differences approach that leverages the introduction of universal preschool in Georgia and Oklahoma, we show that universal preschool generated no average increases in earnings for individuals for whom the program was available in early childhood. These average null results obscure significant heterogeneity in the effects of preschool. A simple theory of treatment contrasts explains heterogeneity in effects across natural subgroups and those constructed intentionally to maximize heterogeneity. Data driven approaches to identify treatment heterogeneity validate the importance of treatment contrast in explaining the observed effects, while also suggesting alternative forms of treatment heterogeneity that were not ex ante apparent.  


Which Workers Left the Labor Force Post-COVID Recession and Why?: Filing, Tax Credit Take-Up, and Labor Force Engagement by Occupation


Working Paper, March 2023 (awaiting disclosure review) Joint with Laura Kawano, Bruce Sacerdote, and Michael Stevens


While aggregate unemployment has returned to pre-COVID levels, labor force engagement remains suppressed. We explore how both stimulus policy and labor market exits during the pandemic impact tax filing behavior and utilization of tax-based safety net programs. We use a panel of tax data and newly assembled information on worker occupation to document who exited the labor force and why. While exit is widespread, those most likely to exit in the wake of the pandemic were those with the weakest pre- pandemic rates of labor force engagement, including individuals in their twenties and sixties, those with low earnings, and those working in the restaurant or leisure and hospitality sectors. Increased exit is not explained by eligibility for unemployment insurance or stimulus payments, school closures, physical proximity of work, or mass layoffs. Nor is the Great Resignation associated with additional upgrading of occupations or firm quality. Labor force disengagement is acute in counties that had larger shocks to out of home mobility, despite fully recovered job postings and rising wages in these areas. Individuals who leave the labor force do not (on average) make up the missing income nor are they more likely to live with their parents. However, gig economy work and non employee compensation replace about 20% of the losses of W-2 income. The facts suggest that the willingness to work for low pay decreased and preferences for leisure and job flexibility increased among a set of workers who have found ways to get by with less income.


The Effect of Income During Infancy: Evidence from a Discontinuity in Tax Benefits (forthcoming, Quarterly Journal of Economics) Online Appendix


Working Paper, February 2022. Joint with Jonathan Eggleston and Alexander A. Smith


We provide evidence of the long-run effect of income provided during the first year of life. We take advantage of the January 1 birthdate cutoff for child-related tax benefits, which results in families of otherwise similar children receiving substantially different amounts of income. Using the universe of administrative tax data in selected years, we show that an additional $1,000 of income during the first year of life increases young adult earnings by at least 1-2%, with larger

effects for males. These effects show up earlier in terms of improved math and reading test scores and a higher likelihood of high school graduation. Estimates of parental behavior in the years after birth suggest an important role for liquidity during the year after birth.


Breaking the Cycle:  The Intergenerational Effects of an Anti-Poverty Program in Early Childhood (forthcoming, Journal of Political Economy)


Working Paper, January 2022. Joint with Chloe Gibbs


Despite substantial evidence that resources and outcomes are transmitted across generations, there has been limited inquiry into the extent to which anti-poverty programs actually disrupt the cycle of bad outcomes.  We leverage the rollout of the United States’ largest early childhood program, Head Start, to estimate the effect of early childhood exposure among mothers on their children’s long-term outcomes.  We find evidence of intergenerational transmission of effects in the form of increased educational attainment, reduced teen pregnancy, and reduced criminal engagement in the second generation.  These effects correspond to an increase in discounted second-generation wages of 6 to 11 percent depending on specification.  Exploration of earlier outcomes suggests an important role for changes in parenting behavior and potential non-cognitive channels.



Fighting Crime in the Cradle: The Effects of Early Childhood Access to Nutritional Assistance (forthcoming, Journal of Human Resources)

(previously circulated as “Fighting Crime in the Cradle: The Effects of Early Childhood Food Stamp Access”) #JHR Threads Explainer


Working Paper, June 2019. Joint with Alex Smith


Despite the extraordinary social costs of crime, relatively little is known about the early life determinants of later criminal behavior. We explore the effect of access to nutritional assistance in early childhood. Using variation in the rollout of the Food Stamp Program (FSP) in the 1960s and 70s, combined with criminal conviction data from North Carolina, we find that FSP availability in early childhood leads to large reductions in later criminal behavior. Each additional year of FSP availability in early childhood reduces the likelihood of a criminal conviction in young adulthood by 2.5 percent. FSP availability has particularly strong effects on the most costly crime types for society: violent and felony convictions. These effects are substantially larger for non-whites, consistent with the higher levels of FSP participation in this population. Analogous estimates derived from the FBIs Uniform Crime Report data suggest similar reductions in arrests for violent crime. These results reveal an important additional benefit of the FSP and suggest a potential link between childhood nutrition and later criminal behavior. Even under conservative assumptions, the discounted social benefits from the FSP's later crime reduction exceed the costs of the program over this time period.


The Bottom Line on College Advising: Large Increases in Degree Attainment


Working Paper, October 2021. Joint with Ben Castleman


We combine a large multi-site randomized control trial with administrative and survey data to demonstrate that intensive advising during high school and college leads to large increases in bachelor's degree attainment. Novel causal forest methods suggest that these increases are driven primarily by improvements in the quality of initial enrollment. Program effects are consistent across sites, cohorts, advisors, and student characteristics, suggesting the model is scalable. While current and proposed investments in postsecondary education focus on cutting costs, our results suggest that investment in advising is likely to be a more efficient route to promote bachelor's degree attainment.


The Effect of Early Childhood Education on Adult Criminality: Evidence from the 1960s through 1990s (forthcoming, American Economic Journal: Economic Policy)


Working Paper, September 2020. Joint with John Anders and Alex Smith


We investigate the impact of early childhood education on adult criminal behavior, leveraging changes in policies that occurred over a number of decades. Using variation across birth cohorts generated by the rollout of Head Start (for those born in the 1960s and 1970s) and Smart Start (for those born in the 1980s and 1990s), along with administrative crime data that include the birth county of all individuals convicted of a crime in North Carolina, we find that improvements to early childhood education led to large (20 percent) reductions in the likelihood of a serious criminal conviction in adulthood. These reductions were concentrated in high poverty counties. While the benefits generated by each program in the form of crime reduction account for a large portion of the costs of the education provided, we find substantial relative gains from the targeting of funds to high poverty areas and to areas without existing access to subsidized care.


You Can’t Handle the Truth: The Effects of the Post-9/11 GI Bill on Higher Education and Earnings


NBER Working Paper #29024, July 2021. Joint with Laura Kawano, Bruce Sacerdote, Bill Skimmyhorn, and Michael Stevens


The Post 9/11 GI Bill (PGIB) is among the largest and most generous college subsidies enacted thus far in the U.S. We examine the impact of the PGIB on veterans’ college-going, degree completion, federal education tax benefit utilization, and long run earnings. Among veterans potentially induced to enroll, the introduction of the PGIB raised college enrollment by 0.17 years and B.A. completion by 1.2 percentage points (on a base of 9 percent). But, the PGIB reduced average annual earnings nine years after separation from the Army by $900 (on a base of $32,000). Years enrolled effects are larger and earnings effects more negative for veterans with lower AFQT scores and those who were less occupationally skilled. Under a variety of conservative assumptions, veterans are unlikely to recoup these reduced earnings during their working careers. All veterans who were already enrolled in college at the time of bill passage increase their months of schooling, but only for those in public institutions did this translate into increases in bachelor’s degree attainment and longer-run earnings. For specific groups of students, large subsidies can modestly help degree completion but harm long run earnings due to lost labor market experience.


Can Information and Advising Affect Postsecondary Participation and Attainment for Non-Traditional Students? Evidence from a Large-Scale Experiment with the U.S. Army


NBER Working Paper 30665, November 2022. Joint with Kelli Bird, Ben Castleman, and Bill Skimmyhorn


Lack of information and advising prior to college matriculation may contribute to poor post-secondary outcomes among non-traditional students. We conducted a large-scale, multi-arm field experiment with the U.S. Army to investigate whether a package of research-based personalized information and access to advising affects postsecondary choices and attainment among a large non-traditional adult population. We find no impact of the intervention on whether veterans enroll in college, on the quality of their college enrollment, or on their persistence in college. Our results suggest that influencing non-traditional populations’ educational decisions and outcomes will require substantially more intensive programs.





The Effect of Reduced Student Loan Borrowing on Academic Performance and Default: Evidence from a Loan Counseling Experiment             


Journal of Public Economics, 202, 2021.


Student loan borrowing for higher education has emerged as a top policy concern. Policy makers at the institutional, state, and federal levels have pursued a variety of strategies to inform students about loan origination processes and how much a student has cumulatively borrowed, and to provide students with greater access to loan counseling. We conducted an experiment to evaluate the impact of an outreach campaign that prompted loan applicants at a large community college to make informed and active borrowing decisions and that offered them access to remote, one-on-one assistance from a loan counselor.  The intervention led students to reduce their unsubsidized loan borrowing by 7 percent, resulted in worse academic performance, and increased the likelihood of loan default during the three years after the intervention occurred. Our results suggest policy makers and higher education leaders should carefully examine the potential unintended consequences of efforts to reduce student borrowing, particularly in light of growing evidence regarding the counter-intuitive positive relationship between reduced borrowing levels and worse student academic and financial outcomes. 


Exploring College Counselor Effectiveness


AEA, Papers and Proceedings, 109, 2019.


Several recent studies have investigated the impact of college advising on whether students enter and succeed in postsecondary education. These papers generally find positive impacts from advising, yet few explore the extent of heterogeneity in impacts between program counselors. We take advantage of quasi-random counselor assignment to explore variation and correlates of college counselor effectiveness using rich administrative data on adviser-advisee interactions from a multi-site randomized control trial of a college advising program.  Large overall effects on college enrollment from the advising intervention are consistent across counselors; there is little relationship between counselor characteristics or behaviors and student outcomes.  


Fighting for Education: Veterans and Financial Aid


Journal of Labor Economics, 37(2), 2019. [Pre-publication version]

Winner of Econ of Education Distinguished CESifo Affiliate Award (best paper presented by young economist)


The Post-9/11 GI Bill brought about the largest expansion in veteran education benefits since the end of World War II, increasing annual benefit expenditures from $3 billion to more than $13 billion dollars annually; in 2014, this comprised nearly one-third of college grant and scholarship aid provided by the federal government and exceeded the amount of college grant and scholarship aid provided by all states combined. Leveraging the variation provided by the benefit expansion, I explore the effect of financial aid on the degree attainment of older nontraditional students. To determine these effects, I exploit variation over time and across types of recipients, as well as geographic variation in the size of the benefit expansion. I find that this financial aid program increased degree attainment by five to six percentage points, a 25 percent increase. This increase was driven by new enrollment among those who might not otherwise have attended, as well as by increased persistence among those who might otherwise have dropped out before degree completion.  


Aid and Encouragement: Does a Letter Increase Enrollment Among UI Recipients?


American Economic Journal: Economic Policy, 10(3), 2018. Joint with Sarah Turner

[NBER Working Paper No. 23374, April 2017.]

Survey and Letter Appendix


For individuals who experience job loss, enrollment in post-secondary programs may provide an opportunity to improve future employment outcomes; yet, decisions to enroll may be hampered by insufficient information about the benefits and costs of enrollment.  One explanation for this is that Unemployment Insurance recipients are not fully aware of the federal financial aid and enrollment opportunities available to them. Indeed, incomplete knowledge of financial aid availability and eligibility during the recent recession may have limited the extent to which workers receiving Unemployment Insurance (UI) took advantage of collegiate opportunities. In May of 2009, an executive initiative and guidance from the Department of Labor and the Department of Education encouraged states to send letters to UI recipients. These letters: (1) suggested training as an avenue to better job security and higher wages, (2) informed UI recipients about the Pell grant program and educational opportunities, and (3) suggested that UI recipients may be given special consideration for Pell grant receipt. We use the variation in the sending of Pell letters within and across states to identify the effect of this information on the college enrollment decisions of millions of individuals who began drawing UI between May 2009 and November 2010. Using the 2008 panel of the Survey of Income and Program Participation, we find that individuals who are likely to have received a letter are four to five percentage points (40%) more likely to enroll in college within the following six months. A back of the envelope calculation suggests that the intervention resulted in an increase in enrollment of at least half a million UI recipients.  The effects are larger for women than men. The most pronounced effects are for black UI recipients and UI recipients with relatively low earnings potential at baseline, suggesting that differences in the accessibility of information about program benefits and how to access them may contribute to inequality.  These findings have implications for a large set of government programs where a well-coordinated information intervention may be substantially more effective than raising the generosity of existing government programs in increasing participation.


From the Battlefield to the Schoolyard: The Short-term Impact of the Post-9/11 GI Bill


Journal of Human Resources, 50(2), 2015. [Pre-publication version]


The Post-9/11 GI Bill brought about an enormous expansion in veteran education benefits, roughly doubling the average maximum benefit level and generating large variation in the magnitude of benefit expansion across states. Using a difference-in-differences strategy, I explore how enrollment of older, non-traditional students responds to educational subsidies.  I use a model of investment in education quality to illustrate how a modification in the structure of benefit provision changes veterans’ incentives to invest in human capital.  The transition from a conditional cash transfer to a more in-kind benefit affects the relative prices of different types of education, pushing veterans to invest in more expensive (plausibly higher quality) schooling.  Using a difference-in-differences strategy, combined with state level variation in the degree of benefit expansion, I find that the higher level of benefits increased college enrollment of separated veterans by between fifteen and twenty percent, while also shifting the composition of enrollment towards four-year schools.


Out of Work and Into School: Labor Market Policies and College Enrollment During the Great Recession


Journal of Public Economics, 124, 2015. Joint with Sarah Turner [Pre-publication version


The Great Recession brought large increases in unemployment and college enrollment; we examine how the combination of changing state labor market conditions and state-specific variation in Unemployment Insurance (UI) interact to affect enrollment outcomes. We identify a substantial role of the UI program in affecting post-secondary enrollment choices. We provide some of the first evidence that the duration of UI affects a displaced individual’s propensity to enroll, and suggestive evidence that these effects are larger in states with more inclusive approved training laws. These findings identify a substantial overlap between UI policy and post-secondary enrollment decisions, indicating the potential importance of UI in not only providing income but also facilitating investments in skills.


Enroll or Enlist: Credit Constraints, College Aid, and the Military Enlistment Margin


Economics of Education Review, 51, 2016. [Pre-publication version]


Money for education consistently ranks among the top answers recruits give when asked why they joined the military. A potential explanation for the prevalence of this enlistment motivation is that the actual value of veteran education benefits exceeds the expected net present value for individuals who want to attend college but face borrowing constraints. Theory predicts that only constrained individuals will respond differently to a change in costs depending on the resources available at the time of the change.  Costs that accrue during a period of lower resources place a greater strain on credit-constrained individuals who are unable to borrow from future earnings to finance them. A policy that reduces the cost of immediate (lower resources) and post-enlistment (higher resources) college enrollment should result in a reduction in the enlistment of college-prepared individuals only if these constraints are present. I explore this question by examining the enlistment response of individuals to financial aid shocks. I find that the introduction of a merit-aid program decreases the probability that a male enlists in the military by .06 percentage points (a six percent reduction), and that these effects are concentrated among applicants that are more likely to qualify for merit scholarships.  These effects are located mainly in low-income areas, supporting the argument that the effects on enlistment are a result of easing financial constraints.  A back of the envelope calculation suggests that between 15 and 25 percent of merit-aid eligible likely recruits face constraints.


Property Taxes and Politicians: Evidence from School Budget Elections

(previously circulated as “Do Median Voters Matter: Evidence from School-Budget Elections”)


National Tax Journal, 69:3, 2016. Joint with Tom Dee.


Recent studies provide mixed evidence on whether electoral pressure influences policy choices. This study examines this question in a unique setting: local school districts where the policy outcome, property taxes, has unusually high visibility and salience. We exploit the sharp discontinuity created by the annual majority-rule school-budget elections in New Jersey’s school districts. Using panel data from over 3,600 district-by-year elections, we find that local politicians are responsive to a modest change in support for a budget (around the 50 percent threshold) resulting in a reduction in contemporaneous property taxes (by $180 per pupil or 1.7 percent) as well as the proposed tax bill for the following year. However, these tax reductions do not generally persist because budget rejections also trigger subsequent increases in both voter turnout and support for school spending, suggesting the stability of Tiebout equilibria.