By Eric Krebs, J.D. candidate, Harvard Law School
Criminal convictions in the United States have functioned as a kind of permanent civil mark—one that extends far beyond the formal sentence imposed by a court. A criminal record can bar people from jobs, housing, public benefits, and professional licenses, creating what scholars have called a form of “new civil death.” Somewhere between 19 and 24 million Americans have felony convictions, and far more have misdemeanor records. When arrests are included, roughly a third of U.S. adults have some form of criminal history. These records trigger a web of collateral consequences that persist long after any prison or probation term ends.
In response, numerous states across America have enacted laws allowing people to seal or expunge their criminal records, giving individuals the opportunity for a fresh start (on paper, at least). Understanding the effect of expungements, on both the individuals who receive them and the population at-large, is critical for scholars and advocates alike. How do these laws function—and do they actually work? How can states improve their administration and impacts?
In a landmark empirical study, J.J. Prescott, professor at the University of Michigan Law School, and Sonja B. Starr, professor at the University of Chicago Law School, sought to answer that question by analyzing expungement in Michigan. Their findings, published in the Harvard Law Review, offer the most rigorous evidence to date on whether expungement delivers on its promise. They sat down with host Jim Greiner on the most recent episode of Proof over Precedent to discuss their research process and findings.
Building the Data Infrastructure
Studying expungement is uniquely difficult because successful expungements are designed to disappear. By definition, expunged records are sealed or removed from public view, making them inaccessible to researchers. Prescott and Starr’s project took more than a decade largely because of this structural challenge.
To answer their research questions, they needed to link multiple administrative datasets: criminal history records from the state police, employment and wage data from the unemployment insurance system, and court records identifying expungement petitions. Each dataset was held by a different agency, governed by different privacy rules, and subject to legal constraints on disclosure. Ultimately, a separate state agency had to perform record linkage and de-identification before the researchers could analyze the data.
This process highlights a recurring challenge in access-to-justice research: the populations most affected by legal barriers are often the hardest to study. Data silos, privacy restrictions, and bureaucratic complexity can impede evidence-based policymaking—precisely in domains where evidence is most needed.
Public Safety and Recidivism
A common objection to expungement is that the public has a safety interest in knowing about past convictions, especially for employers or landlords. If records disappear, critics argue, dangerous individuals might reoffend without warning.
The data tell a different story. Among expungement recipients in Michigan, 7.1% were rearrested within five years, and just 2.6% were rearrested for violent offenses. Re-conviction rates were lower: 4.2% for any crime and only 0.6% for a violent crime. These rates compare favorably to the general population, suggesting that people who qualify for and obtain expungement are among the lowest-risk individuals in the criminal justice system. And to put the figures in starker relief, expungement recipients had recidivism rates an order of magnitude lower than that of the general post-prison population. (One recent study, for example, found that over 60% of people are rearrested within two years of being released from prison, and 77% are rearrested within five years.)
Of course, expungement does not alone cause lower crime rates, Prescott and Starr note. Michigan imposes a five-year waiting period before individuals become eligible for expungement, and a re-conviction during that time renders the person not eligible. Thus, the possibility of expungement can provide an incentive for avoiding crime in the years leading up to eligibility, and people who have stayed crime-free long enough to qualify for expungement are already likely desisting from criminal behavior. Nonetheless, the policy implication is powerful: by clearing the records of individuals who manage to “stay clean” for five years after jail, expungement does not appear to endanger public safety.
“Regardless of the cause, whether it’s selection and/or some kind of causal story,” says Prescott, “the main message is that you shouldn’t be worried—with an expungement system setup like Michigan has—that ‘dangerous people’ are having their records removed from the public view.”
The Uptake Gap
One of the most striking findings of the study concerns what Prescott and Starr call the “uptake gap.” “The bad news of our study was that under Michigan’s old regime in which you had to jump through a bunch of hoops to achieve an expungement, very few people who were legally eligible got them,” says Starr.
Among people legally eligible for expungement in Michigan, the authors estimated, only about 6.5% obtained expungement within five years of eligibility. This gap is not primarily due to judicial denial. Rather, most eligible individuals never applied. The authors identify several procedural barriers, including information deficits, costs and fees, distrust of legal institutions, lack of legal assistance, and administrative complexity.
For instance, Michigan required individuals applying for expungement to submit fingerprints, submit detailed criminal searches, pay for further background checks, visit police stations, and potentially submit to hearings. Some steps required individuals to resubmit official records of prior convictions which the courts and other agencies—in theory, at least—already had access to. “There are 10 to 12 steps that you have to go through,” says Prescott. “The complexity of the process [can be] overwhelming for many people who are in the category of individuals who are probably the most likely to benefit from this kind of relief.”
In sum, Prescott and Starr’s study illustrates how the mechanism of a law—how a benefit is administered—can be just as impactful as the law’s substance. Prescott and Starr’s study sheds light on how expungement programs can be reformed to improve substance, process, and outcomes.
If you’re interested in more on this topic, listen to our Proof Over Precedent podcast episode.