For these reasons, policymakers have been searching for ways to reduce the reentry barriers faced by people with criminal records, with states passing hundreds of criminal-records reforms in 2019 alone. One of the most popular approaches—thought by many to have the greatest potential upside—is enacting laws that allow criminal conviction records to be wholly expunged or, at least, sealed from public view. Expungement offers the possibility of attacking many reentry challenges at the same time; these laws typically enable individuals to apply for jobs, housing, schools, and benefits as though their convictions did not exist.

Today, a substantial majority of U.S. states provide some form of expungement procedure for otherwise valid adult convictions. Traditional versions of these laws require individuals to apply for record expungement, with burdensome procedural requirements. The scope of this relief is limited by stringent eligibility conditions: waiting periods (often of many years), limits based on the number of an individual’s past convictions, and constraints on the types of crimes that can be expunged.

Over the last few years, many states have adopted new or expanded expungement laws. In some cases, these new laws have loosened eligibility requirements. In others, they have focused on easing the procedural burdens. A few states now provide for the automatic expungement of certain adult criminal convictions. Notably, the first states to adopt such policies (Pennsylvania and Utah) did so with the near-unanimous support of Republican-dominated legislatures, a sign of the new bipartisan turn favoring expungement. Chambers of commerce have joined with civil rights and civil liberties groups to endorse many recent reform proposals. Many employers and landlords want to give opportunities to qualified applicants with records but cannot do so because of legal restrictions and fear of tort liability, which expungement laws in effect curtail.

Still, a chief difficulty for this growing reform effort has been the absence of hard evidence—about both the effects of expungement as well as the successes and failures of current approaches. In general, the data necessary to conduct a careful evaluation of expungement (most of all, the expunged records themselves) have been unavailable to researchers. In theory, expungement laws ought to make a big difference to people working to reclaim their lives after a criminal conviction, but policymakers have had lingering doubts. For example, some have suggested that in the age of the internet, records cannot be sealed effectively.

In the hope of improving matters, we recently conducted an empirical analysis of record clearing that we believe has made important inroads into the wasteland of what is known about expungement. Our formal article on this research, just out in the Harvard Law Review, examines how expungement works for those who are eligible and the relationship between expungement and important outcomes such as recidivism risk and employment success.

Through a data-sharing agreement with the state of Michigan, we were able to obtain complete, deidentified criminal records as well as employment information for a large sample of individuals with criminal convictions in Michigan, including all individuals receiving an expungement in the state as of March 2014. Michigan is a nice setting for such a study; it is a large, representative state with an expungement regime comparable to what is in place in many other states.

Our study—the first of its kind—draws three conclusions:

• Petition-based expungement policies result in very low uptake rates—that is, very few of those who are legally eligible actually apply for and receive an expungement. In Michigan, our evidence indicates that only 6.5% of eligible individuals receive expungements within the first five years of eligibility. This problem primarily stems from lack of knowledge of the law as well as the daunting bureaucratic and judicial processes applicants are required to navigate.
• People who receive expungements have very low rates of recidivism—even lower by some measures than the average crime rates for all Michigan adults. Just 4% of all expungement recipients in Michigan are reconvicted within five years, and most of these reconvictions are for nonviolent misdemeanors. These facts suggest that concerns over the public-safety risk of expungements are misplaced.
• Expungement recipients quickly exhibit much better employment outcomes—even after accounting for the recipient’s employment history and broader economic trends. Average quarterly wages rise by about 23% within a year of expungement, an increase driven mainly by unemployed or marginally employed people finding work. Although there are potential causal-inference challenges on this front, our data provide good reason to believe that at least most of this gain is caused by expungement.

In this article we summarize our findings.

Background: Criminal Records and Engagement

Criminal convictions have outsized consequences that last long after completing any formal sentence. First, people with criminal records have great difficulty finding a decent job. Field experiments, for instance, have found that employers are much less likely to call back otherwise-identical applicants if they have a record. Almost 90% of employers conduct criminal background checks for at least some employees and a large majority of employers do so for all employees. Convictions also trigger many “collateral” legal consequences, including licensing exclusions that put hundreds of professions off-limits, as well as limits on voting, access to firearms, educational loans, and many public benefits. Beyond these restrictions, employers report that they fear negligent-hiring liability if they knowingly hire someone with a record who later commits a crime on the job.

Because many Americans have criminal records, the aggregate effect of these consequences for society is huge. Moreover, this effect disproportionately falls on populations more likely to have a record, particularly black men, reinforcing racial disparities in socioeconomic outcomes generally.

Aiming to alleviate these difficulties, most states today allow for the expungement of adult criminal convictions under certain conditions. As a general matter, these laws lift statutory barriers to public employment, licensing, and benefits. They also typically give individuals the legal right to keep silent about an expunged conviction when interacting with a prospective employer or landlord. Employers and landlords cannot be held liable for failing to consider an expunged conviction. But information about an expunged conviction usually remains available for law enforcement purposes, like sentencing for a subsequent crime.

Eligibility requirements for expungement vary widely across states. Some states exclude certain crimes, like violent offenses, or restrict eligibility based on the extent of an individual’s criminal history. Almost all states impose a waiting period following conviction or completion of the sentence before a record can be expunged, during which the individual must stay crime free.

In recent years, there has been considerable focus on expanding the scope and improving the efficacy of these laws, including moving toward automatic expungement after some set number of years, as a few states have now done. While there have been other reform efforts targeting the challenges facing people with records, expungement offers potentially more significant relief by legally erasing the expunged conviction.

Research on effects / Despite the magnitude of reentry barriers, the scope of existing record-clearing laws, and the momentum of current reform efforts, empirical research on whether and how expungement laws improve matters has been disappointingly thin. The challenge has not been a lack of interest but rather a lack of data. Expunged convictions are generally unavailable and wage data are typically hard to obtain.

We know almost nothing about “uptake” behavior in the context of conviction expungement. The core question relates to the uptake rate: when people are legally eligible for expungement, how often do they apply for and receive it? Some recent work has found discouragingly low uptake in other post-conviction-relief contexts—for example, Colleen Chien’s study of three programs providing remedies such as sentence commutation or reduction of felonies to misdemeanors. But no study has explored similar questions in the context of expungement.

Existing research on the likely public-safety risk and employment consequences of expungement policies is also very limited. Although there are good theoretical reasons to believe that criminal-record expungement is likely to reduce recidivism, no empirical work scrutinizes post-expungement criminal behavior—and specifically whether expungement recipients are a significant threat to public safety.

As to employment consequences, ample empirical research indicates that criminal records undermine employment opportunities. However, it does not necessarily follow that expungement would undo this damage, either because the effects of even a temporary disability before expungement eligibility are difficult to shake or because it is difficult, once records are made public in databases and on the internet, to remove them from circulation. The most relevant research to date involves the tracking of a small sample of 235 clients of a law school clinic who sought expungement-like relief. The results point, tentatively and imprecisely, to employment gains.

A Comprehensive Study of Expungement

To effectively design and advocate for record-clearing reform, policymakers need evidence that derives from a rigorous study, one involving a large and representative sample of individuals with criminal records, an expungement law with typical eligibility and process features, and crucial measures of “success,” such as subsequent criminal behavior and employment outcomes.

Michigan presents such an opportunity. We were able to negotiate access to a high-quality dataset that provided, in deidentified form, the information we needed, including expunged criminal records. Michigan’s expungement law has been on the books for more than 50 years, although it has been amended (mostly expanded) in the last decade. It was, however, unchanged from 1983 to 2011, the period we study. There are tens of thousands of individuals who received relief before 2011, allowing for the measurement of long-term outcomes in a large sample.

To provide a few key details on this archetype expungement law, Michigan’s pre-2011 regime imposed a five-year waiting period, in the middle range compared to other state expungement laws. The five-year clock began running at sentencing or release from incarceration, whichever was later, and to be eligible at the end of the waiting period, the individual had to have no further convictions. The statute contemplates expungement for all types of crimes, including many violent felonies, but not traffic offenses, sex offenses, or felonies that carry potential life-imprisonment terms. The wide applicability of Michigan’s expungement law is again useful for research purposes because it allows comparisons of patterns across many types of offenders and offenses.

Michigan’s law did, however, stringently circumscribe eligibility based on record length—that is, the total number of criminal convictions a person has. Pre-2011, expungements were strictly limited to people with exactly one conviction on a single charge. Any other conviction at any time—even a second charge arising from the same incident—was disqualifying. These simple, bright-line eligibility requirements—although sometimes producing incoherent results (e.g., two minor misdemeanors [ineligible] versus a single violent felony [eligible])—nevertheless make it easier from a research perspective to code expungement eligibility and other variables related to the nature of the criminal record.

Michigan’s pre-2011 expungement law is also representative in terms of its procedures. As in most states today (including Michigan), someone who was eligible for expungement had to apply for relief. The application process was (and is) elaborate and under the law the court “may” grant the request in its discretion if doing so is “consistent with the public welfare.” Still, grant rates appear to have been fairly high. According to our conversations with the Michigan State Police, recent applications result in successful expungements about 75% of the time.

To study the operation and consequences of expungement, we analyze individual-level data containing full criminal histories for every individual receiving an expungement through June 2011, as well as for a large comparison group of similar people with records who did not receive expungements. These records were matched to state-agency wage and employment information for the same individuals, which allows us to study the employment outcomes of expungement.

The Uptake Problem

When assessing the effects of expungement laws, a basic first question is: do they actually help many people get their records expunged? Minimal uptake of record-clearing opportunities may be one of the most important, yet least recognized, hurdles facing reform efforts. Expungement laws cannot accomplish their policy purposes if very few people take advantage of them. And we find, indeed, that very few people do.

For our purposes, the “uptake rate” is the percentage of expungement-eligible individuals who receive an expungement over a given period of time. We began by identifying who these eligible people were, isolating a sample of individuals who had one conviction for an eligible crime on a single count and who were not reconvicted within five years of their sentencing or release. We used a cohort of almost 10,000 such individuals who were convicted between January 1999 and May 2001 and became expungement-eligible five years later. We limited the sample to those who were not sentenced to incarceration, which made it easier to calculate when the five-year waiting period expired. Relatively straightforward calculations using this sample show that just 6.5% of individuals eligible for expungement receive expungement during their first five years of eligibility. This rate is strikingly low given the many burdens of a criminal record.

How confident can we be about this number? We tested it by varying the assumptions we made to address certain data limitations, such as missing dates for a few observations and our lack of data on out-of-state convictions. These changes only minimally affected our uptake estimates, and we are quite confident that we are within a percentage point or two of the correct number. In addition, while we only tracked uptake for this relatively recent cohort for five years, our full dataset makes clear that most people who obtain expungements do so in those first five years. Using our data, we can extrapolate a lifetime uptake rate of less than 12% for those not sentenced to incarceration. And those who are sentenced to incarceration appear to have even lower uptake.

Although uptake was low in every subset of our sample, it was, relatively speaking, higher in some subsets than others. Some significant predictors of uptake included felony status (people with felonies were much more likely to receive expungements than those with misdemeanors, presumably because a felony record makes one more motivated to apply), gender (women had higher uptake rates), and county (there was considerable variation across Michigan). We also explored whether certain factors predicted when individuals received expungements and found most notably that they were especially likely to do so when they had recently experienced a job loss or decline in wages.

Why the low uptake? / What accounts for the expungement uptake gap? There are many significant benefits to a clean record (and no apparent costs), so the most obvious possibilities are that potential beneficiaries are unaware of the opportunity and that the opportunity, although attractive, is too difficult or costly to pursue to the finish line. We interviewed lawyers and advocates who work on expungement and reentry issues, and their experience confirms these hypotheses.

On the information side, many eligible individuals either do not know that the law allows record clearing or do not know that they are eligible. And counsel is rarely available to advise them. But though knowledge is necessary, it is certainly not sufficient. Pursuing an expungement takes time, effort, discipline, organization, and often courage. In Michigan, the application procedure includes 11 separate instructions, many of which contain more than one directive. The list is daunting and potentially confusing. Tasks include copying, mailing, obtaining fingerprints and records, going to a police station and a courthouse (multiple times), and much more. Many of these tasks bring applicants into direct contact with law enforcement (and perhaps victims). For individuals whose past experiences with the criminal justice system have been generally negative, this can produce profound fear and deep-seated pessimism about the likelihood of success. Petitioning for expungement is also costly, with the outlay totaling about $100, not including transportation expenses and time off from work.

Over the past 15 years, efforts to support people seeking expungement have grown in Michigan and elsewhere. State and city governments as well as nonprofits have sought to make potential applicants aware of expungement opportunities and have sponsored the development and dissemination of do-it-yourself legal materials. Expungement fairs have also become popular; these events offer information and hands-on assistance in completing application requirements, including having law enforcement and notaries onsite to help people with fingerprinting and required paperwork.

All of this suggests that uptake ought to be improving, and it is—at least in Michigan. In 2010, the number of expungements granted in the state exceeded 2,000, which is more than 60% higher than the count of expungements in 2004. In the last decade, these numbers have grown a bit further, although that is partly the product of slight expansions to eligibility. Still, efforts to help people navigate the process are far from a full solution: even doubling uptake rates would still leave more than 80% of eligible individuals without relief. And even with a lot of help, applicants remain, by all accounts, very frustrated and discouraged by the expungement process.

The only option that would be a true solution to the uptake gap is automating expungement, as several states have now done. States search their own data to identify eligible cases and designate records as sealed after the requisite time has passed. By eliminating bureaucracy, this approach makes expungement much less resource intensive for the individual and the state. Designing and implementing an automatic-expungement algorithm would require a modest initial investment that would vary by jurisdiction, although academic researchers and nonprofits are standing by to facilitate this step inexpensively. Once the process is in place, the cost per expungement should be close to zero, eliminating the burden of a petition-based regime for both the government and the recipient. And if expungement recipients become more likely to get jobs and pay taxes and less likely to collect government benefits, commit crimes, and require incarceration, the fiscal upside looks even more positive.

It is worth remembering that eligibility for expungement has been and remains quite strict in many states, including Michigan. Thus, the very low uptake rate is even more grim than it might at first seem—it’s a small fraction of a small fraction. Just 2,000 to 3,000 individuals in Michigan receive expungements a year, a trifle when compared to the roughly 300,000 new criminal convictions added to the rolls annually.

Automating expungement for eligible individuals is, then, not at all radical. It is efficient, fiscally sound, and pro-public safety. Notably, in the first two states that passed automatic-expungement reforms, Republicans—hardly a soft-on-crime crowd—controlled the legislatures, and the laws passed with very little dissent. It’s a commonsense reform that no one should oppose: if a state decides that expungement should be available under certain conditions, it seems wasteful and counterproductive to make it administratively overwhelming and costly to obtain.

Expungment and Criminal Recidivism

Strict eligibility rules make expungement rare, but perhaps limiting the scope of expungement is justified if record clearing entails significant public-safety risks. Opponents of expungement reform routinely allude to crime concerns. The suggestion is that members of the public need to know who has a criminal record so that they can best protect themselves and others from any future crimes committed by anyone with a record.

But our data from Michigan indicate that recidivism rates for expungement recipients are extremely low. In all likelihood, these low rates are primarily driven by expungement recipients being very low risk in the first place. But there are also many reasons to believe that expungement directly reduces rather than aggravates recidivism risk by improving economic and reintegration opportunities.

Calculating recidivism risk is fairly straightforward in our data. We track subsequent arrests and convictions over two- and five-year periods for a large sample of over 20,000 expungement recipients. Recidivism occurs if a recipient’s criminal record shows a post-expungement arrest or conviction in Michigan. (Federal and out-of-state arrests and convictions are not observed, suggesting that rates are slightly underestimated, but these are rare enough to make any effect negligible.)

Our analysis shows that 3.4% of recipients are rearrested within two years (1.8% reconvicted) and 7.1% are rearrested within five years (4.2% reconvicted). These overall numbers include all crimes, including minor and nonviolent ones that might matter less from a public-safety perspective. For violent crimes, only 2.6% of recipients are rearrested (0.6% reconvicted) and, for felonies, only 2.7% are rearrested (1% reconvicted).

Our calculations demonstrate that the recidivism risk of expungement recipients is very low in relative terms. In this group, recidivism rates are much lower than the rates found in most studies of criminal recidivism. Indeed, their risk of reoffense also appears to be significantly lower than the baseline crime risk posed by the general population, that is, all Michigan adults. In 2009 and 2010 combined, Michigan police made approximately 6.6 arrests per 100 adults in the population. A similarly timed expungement-recipient cohort only experienced 4.7 arrests per 100 individuals. This is especially remarkable given that expungement recipients (like criminal defendants generally) tend to come from disadvantaged backgrounds, which is normally associated with higher crime risk. Yet those with expunged records are less likely to offend than the average adult.

Our criminal-history data can be used in other ways to provide insight into important policy-relevant questions about expungement. For instance, policymakers might want to know whether recipient recidivism risk depends on the length of time since conviction. The data indicate that it does, but the differences between five elapsed years (the shortest period in our sample because of the waiting period) and 10 or more elapsed years are minimal. The implication is that if policymakers want to limit expungements to those with low crime risk, five years is a long enough waiting period to accomplish this purpose.

Policymakers might also care about whether recidivism risk varies by the type of crime being expunged. The data show that five-year rates are low for both recipients with expunged misdemeanor convictions—6.4% (arrest) and 3.8% (conviction)—and with expunged felony convictions—8.1% (arrest) and 4.6% (conviction)—although the rates are somewhat higher for those with an expunged felony. What about for those who have had violent felonies expunged? These rates are very close to the rates for felony convictions generally, and for violent-crime recidivism, the numbers are quite low: only 0.8% of individuals with an expunged violent crime are convicted of another violent crime in the first five years after receiving their expungement.

One important follow-up question, however, is whether we can draw any inferences about the likely consequences of expanding expungement eligibility. After all, expungement recipients are unusual. Each one had to wait at least five years, navigate a burdensome petition procedure, and survive a judge’s opportunity to refuse the requested relief. And in our study period, all were first-time offenders with just a single conviction. As reformers push for easier—even automatic—expungement for more types of crimes and with longer records and shorter waiting periods, can we expect these low rates of recidivism to hold?

The quick answer is probably not: if we expand expungement to a riskier pool, we will presumably see higher crime rates than we do in the narrow, low-risk pool we study. But any difference should not be dramatic, as long as waiting periods similar to Michigan’s five-year period remain in place. Criminological research tells us that the riskiest time for recidivism is during the first year or two after conviction or prison release. Individuals who remain crime free for several years beyond that pose little crime risk; their likelihood of committing a crime converges toward, and eventually equals, the level of the general population.

But asking whether the recidivism rates we measure will hold in more expansive or automatic expungement regimes is really the wrong question. Much more important is the causal effect of the law change: whether the recidivism risk of those who can now receive expungements is higher or lower than it would have been absent the expanded availability of expungement.

And there is excellent reason to believe that expungement, in fact, reduces crime. Research identifies many criminogenic factors associated with the drag of a criminal record. Unemployment, low wages and poverty, homelessness, poor education, stigma, and social exclusion are all associated with higher recidivism risk. If expungement alleviates even some of these conditions, it seems likely to reduce recidivism among recipients. It also follows that if reforms extend the effect of that reduction to a broader, higher-risk pool, the public-safety upside of record-clearing laws would only be larger.

Expungement opponents have never offered any empirical evidence that expungement somehow undermines public safety, and the logic underlying this theory crumbles under even a little bit of scrutiny. The idea seems to be that awareness of an individual’s criminal record allows members of the public to avoid victimization—that knowledge provides safety. But this is illogical as a strategy for protecting the public at large. People don’t conduct background checks on everyone they meet, much less on every stranger who could potentially harm them. Having a criminal record that appears in public databases inflicts many disadvantages, but it in no way disables an individual from committing future crimes. If anything, it might allow a particular employer or landlord to avoid the individual. But this only displaces risk to others in society—and if the individual ends up jobless or homeless, the risk will only be greater.

Expungement and Employment Outcomes

One important argument in favor of expungement policies is that they will improve the employment prospects of their beneficiaries—indeed, this is one of the arguments behind the claim that record clearing ought to reduce crime. Theory and evidence on the deleterious effects of criminal records support this argument, but empirical analysis of whether subsequently expunging criminal records can help is scarce.

Fortunately, wage and employment information captured in Unemployment Insurance Agency records for Michigan expungement recipients allows us to make important headway on this challenge. Our analysis effectively compares the experience of recipients before and after they were granted an expungement. It uncovers important gains in both employment rates and wages, and while there may be plausible alternative interpretations for some of our findings, on the whole the evidence suggests that record clearing has very positive employment effects.

Our analysis studies employment outcomes for all individuals of working age who received expungements between January 1998 and May 2011 on a known date, who appear (per driver’s license data) to be Michigan residents, and who have at least one quarter of employment and wage data at any time between 1997 and 2013. We run regressions that implicitly control for fixed demographic differences across individuals and explicitly control for individual-level pre-expungement employment-history trends as well as general economic conditions in Michigan.

Our results show that there is a substantial and statistically significant upward turn in the likelihood that an individual is employed after an expungement, where employment is defined loosely as having any positive wages through formal employment. The gains are especially sharp immediately after the expungement and level out after a year, but they hold over time. By the end of one year, expungement recipients’ prospects of being employed improve by 13%. With a more robust definition of employment (earning on average at least $100 a week from a covered employer), the improvement is even greater, about 23%. Wages likewise jump by about 23%.

The substance of our findings remains consistent if we use an array of alternative assumptions as well as alternative sample definitions, including dropping the years of the Great Recession. In addition, changes in employment prospects around the time of expungement appear to be large and positive for every subgroup, although the magnitude of the effects is much larger for women, which is broadly consistent with other research on job-access interventions (e.g., job training).

Of course, one critical question is whether the association we find between expungement and improved employment outcomes is one of correlation or causation. Did expungement cause the improvement? There are at least two alternative interpretations. First, people might apply for expungements when they are especially motivated to find a job. If so, this motivation (which presumably also leads to more intensive job-hunting) may produce an improvement in their prospects. A second and related concern is that the association might be driven by regression to the mean. If expungement-eligible people decide to apply for an expungement when their employment situation has deteriorated (as our data indeed suggest), the application for an expungement will likely be correlated with the “bounce back” from any unlucky downswing. Both of these dynamics may partly explain our results.

Yet, other features of our findings push in the opposite direction, suggesting that the gain can be causally attributed to expungement. For one, our data show that the timing of the upturn is associated with the receipt of an expungement rather than an application for one (which happens three to six months earlier on average). The motivation and mean-regression interpretations are both inconsistent with that pattern.

In addition, when we analyze employment changes for those who seek expungement very shortly after they become eligible (five to six years after their conviction or release), we find very similar improvements. This group is important because, for them, the timing of their expungement is far less likely to be attributable to a recent job loss or a sudden upsurge in motivation. Rather, it is simply driven by the expiration of a five-year waiting period. Even so, expungement appears to benefit this subsample nearly as dramatically as the full sample, strongly implying that the benefits are real.

Conclusion

Evidence on the operation and effects of expungement laws is essential as states debate the adoption and expansion of record-clearing policies. Although there is plenty of room for further research, we believe our recent work in Michigan—a representative state in criminal justice terms and one that boasts a fairly typical (and time-tested) approach to expungement—covers important new ground.

To be sure, our research does not address every policy question surrounding expungement. For example, some critics have worried that it restricts freedom of information too much. While this is essentially a philosophical objection that is beyond our paper’s scope, we note that very few people are absolutists about the idea that people’s background information ought to be public—much less that it should be maintained permanently by the state in databases that anyone can search. Laws protecting the privacy and confidentiality of an individual’s tax, financial, health, and educational data, for example, enjoy broad support, even though many of these records would also be of potential interest to employers, landlords, and others. Once individuals have served their sentence for their conviction (plus proven themselves with years of subsequent law-abiding behavior), we could similarly see their criminal record as a private fact that the state need no longer publicize.

Overall, even if there remain further issues to explore, our analysis of Michigan data supports expanding the availability of expungement, including making it automatic or at least procedurally easier to obtain. Expungement is associated with large gains in employment rates and wages, and a causal story seems highly plausible, at least for a significant share of the robust relationship apparent in the data. And recidivism rates among those with expungement are extremely low, so concerns about negative public-safety consequences ought to take a back seat in record-clearing conversations. Recidivism rates among expungement recipients may increase as eligibility expands, but an avalanche of research suggests that those rates would be even higher for the same individuals absent expungement. After all, a clear record allows people with past convictions to find employment and housing, and rebuild their lives. Those outcomes are good in and of themselves, but they also mean that broader expungement will make us safer.

Readings

• “Ban the Box, Criminal Records, and Racial Discrimination: A Field Experiment,” by Amanda Agan and Sonja Starr. Quarterly Journal of Economics 133(1): 191–235 (2018).
• “Expungement of Criminal Convictions: An Empirical Study,” by J.J. Prescott and Sonja B. Starr. Harvard Law Review 133(8): 2460–2555 (2020).
• “Redemption in the Presence of Widespread Criminal Background Checks,” by Alfred Blumstein and Kiminuri Nakamura. Criminology 47(2): 327–359 (2009).
• “Reducing Crime through Expungements,” by Murat C. Mungan. Journal of Economic Behavior & Organization 137(C): 398–409 (2016).
• “The Mark of a Criminal Record,” by Devah Pager. American Journal of Sociology 108(5): 937–975 (2003).
• “The Second Chance Gap,” by Colleen V. Chien. Michigan Law Review, forthcoming.
• “Unmarked? Criminal Record Clearing and Employment Outcomes,” by Jeffrey Selbin, Justin McCrary, and Joshua Epstein. Journal of Criminal Law and Criminology 108(1): 1–72 (2017).