Criminal Records and Licensure in Five Allied Health Professions: Is There Evidence of a Disparate Impact on Historically Marginalized Groups?

Abstract

Context: In health care, licensing is pervasive. Restrictions on applicants with criminal records may have a disparate impact on historically marginalized groups. There is bipartisan interest in evaluating whether occupational licensing requirements are too strict.

Methods: The authors analyze how 12 representative states (California, Colorado, Connecticut, Delaware, Florida, Illinois, Missouri, New York, Ohio, Pennsylvania, South Dakota, and Texas) respond when people with criminal records apply for a license for five entry-level allied health professions (dental hygienist, occupational therapy assistant, physical therapy assistant, radiologic technologist, and respiratory therapist).

Findings: With one exception for one allied health profession, all states require their licensing boards to consider past serious criminal convictions. A majority of states require the conviction to be substantially related to the scope of professional duties for it to provide a basis for disqualification. Most states make it difficult for applicants with criminal records to determine whether they may obtain a license.

Conclusions: State licensing boards have considerable discretion in handling applicants with a criminal record. The trend is toward fewer restrictions, but more could be done to increase the transparency of state licensing board guidelines, practices, and procedures—particularly in the states that still rely on a “good moral character” test.

In the modern economy, licensing is pervasive—particularly within health care. Fully 20%–25% of jobs in the United States now require a state-issued license, but the percentage is closer to 75% in the one-fifth of the US economy that is devoted to health care (NCSL 2020; Kleiner and Vorotnikov 2017; White House 2015; Kleiner and Kruger 2013). In recent years, there has been bipartisan interest in evaluating whether licensing practices are unduly restrictive. The Biden administration's 2021 executive order on promoting competition in the American economy notes that “overly restrictive occupational licensing requirements can impede workers' ability to find jobs and to move between States” (White House 2021). The Trump administration's 2020 executive order on increasing economic and geographic mobility similarly observes that “overly burdensome occupational licensing requirements can impede job creation and slow economic growth, which undermines our Nation's prosperity and the economic well-being of the American people” (White House 2020). Both executive orders included provisions intended to reduce the burdens of occupational licensing. The Obama administration issued an extensive report recommending various steps to restrict the scope of substantive requirements imposed by occupational licensing regimes (White House 2015). The topic has also been a policy perennial for the US Federal Trade Commission, although the state action doctrine limits the FTC's ability to bring cases against overly burdensome licensing regimes (FTC n.d.).

At the state level, occupational licensing has steadily expanded to cover a substantial share of the economy (NCSL 2020; Kleiner and Vorotnikov 2017). The COVID-19 pandemic has prompted multiple states to ease up on telemedicine and reciprocity for certain licensed professions, although it remains to be seen whether those waivers will remain in effect as the pandemic recedes (Hentze and Herman 2021; CSG 2020).

Applicants with criminal records face significant challenges (Sibilla 2020). The Obama administration's 2015 report on occupational licensing describes the basic dynamics:

Others have noted similar concerns about the ability of applicants with criminal records to obtain a license and about the impacts of licensing on military spouses and immigrants (Ewald 2019; Umez and Pirius 2018; White House 2015; Vallas and Dietrich 2014). More broadly, it is important to ensure that the costs imposed by current occupational licensing regimes are justified by documented benefits (Kleiner 2015).

Since a license is required for many occupations in health care, which is a large, robust, and growing area of the economy, restrictions on applicants with criminal records can foreclose substantial employment opportunities for those seeking to make an honest living. Furthermore, to the extent that occupational licensing restrictions affecting allied health professions (AHPs) prevent workers from marginalized groups from securing better jobs and economic opportunities, they have the potential to entrench economic and racial inequality. These restrictions may also be another example of inefficient regulatory barriers in the economy (Lindsey and Teles 2017).

These problems are compounded when licensing restrictions are framed in ambiguous and expansive terms, such as limiting entry to applicants of “good moral character.” Open-ended exclusionary standards make it difficult for applicants with a criminal record to determine whether they will be able to obtain a license, and uncertainty seems likely to deter many potential applicants.

To quantify the regulatory structure of these exclusionary provisions, we examine how 12 representative states (California, Colorado, Connecticut, Delaware, Florida, Illinois, Missouri, New York, Ohio, Pennsylvania, South Dakota, and Texas) handle applicants with criminal records. We focus on five entry-level AHPs (dental hygienist, occupational therapy assistant, physical therapy assistant, respiratory therapist, and radiologic technologist). Our data set covers 24% of the US states and 50% of the US population.

All 12 states license four of the five AHPs we study (dental hygienist, occupational therapy assistant, physical therapy assistant, and respiratory therapist), and a majority of those states (8 of 12) license the fifth AHP (radiologic technologist). Every state requires consideration of whether an applicant for a license has a criminal record, and each state provides a list of categorical and discretionary exclusion criteria (i.e., criminal offenses that fall into each category). About half the states limit the scope of discretionary exclusions to convictions that are substantially related to the scope of services in the AHP, but the other half do not have such restrictions. Only 3 of the 12 states prohibit their licensing boards from evaluating “moral character” when evaluating licensing applicants.

Most states make it difficult to determine the factors that state licensing boards will consider in evaluating an applicant with a criminal record. Unless a licensing board concludes that arrests go to moral character (in the states that use that standard), only actual convictions should matter in evaluating licensing applicants. Some states have a mechanism for potential applicants to obtain an advance determination of whether their criminal record disqualifies them from obtaining a license before undergoing the necessary training, but most do not.

In combination, these dynamics seem likely to deter individuals with a criminal record from even considering entering an AHP. And because the likelihood of having a criminal record is not uniform across the population, excluding applicants with a criminal record seems likely to have a disparate impact across various subpopulations (principally race and gender) and has the potential to affect health equity.

Occupational licensing of AHPs shows no signs of going away entirely, but a significant number of states have recently moved away from open-ended moral character tests and have adopted an array of categorical and discretionary exclusions. It remains to be seen how state licensing boards will implement these provisions, which give them more discretion in certain domains (i.e., greater reliance on permissive exclusions) and less discretion in other domains (i.e., requiring past criminal offenses to be substantially related to the scope of services provided by the AHP). Regardless, much more can be done to inform potential applicants of how state licensing boards will handle their particular case, which will reduce barriers to entry and has the potential to reduce any disparate impact.

Background and Methods

Licensure

Licensure is a state-imposed restriction on market entry, backed up by civil and criminal penalties. For example, in California, the unauthorized practice of medicine (i.e., without a license) can be punished by up to three years in jail and fines of up to $10,000. Unlicensed practice of an AHP generally carries less strict (but still material) penalties.

Licensure was initially limited to a small number of high-status professions, but it has now spread throughout large swathes of the economy (NCSL 2020; Kleiner and Krueger 2013). There is considerable variation in which professions are regulated through licensure, and states vary significantly in the share of the workforce that is subject to occupational licensing, with five states licensing less than 15% of their workforce, and three states licensing more than 30% (Carpenter et al. 2018; White House 2015). At least one state (Louisiana) licenses floral arranging, even though there is no evidence of a heightened risk of death or serious injury associated with purchasing flowers from unlicensed florists (Weissmann and Dieterle 2018).

Licensing has both benefits and costs. Licensing can increase quality and protect the public against those without the necessary skill and moral character to deliver certain types of services (Kleiner and Krueger 2013). The public interest theory of licensure is that licensure is necessary to protect the public from the adverse consequences of an unregulated market, and state licensing boards will carry out their obligations in a public-regarding manner. Licensed providers will behave in ways that justify the trust that is reposed in them (Smith 1776).

A public choice perspective offers a more skeptical take, with licensure used to increase the relative earnings of the profession, with relatively little impact on the quality received by consumers (Stigler 1971; Friedman 1962). Less regulated markets can function quite well, particularly when information intermediaries can evaluate the quality of services and consumers can make the cost–quality trade-offs that they prefer (Barton 2001, 2003; Hyman and Silver 1997–1998). In practice, licensing boards reflect the interests of the regulated profession in restricting competition from new entrants, rather than the public interest (Ohlhausen and Luib 2016). This problem is not limited to licensing boards; regulatory agencies often create barriers to entry and adopt policies that favor the interests of incumbents (McConnell 1966). Licensure is also not a guarantee of good behavior; a significant number of licensed providers will betray the trust that is put in them (Liu and Hyman 2021; Silver and Hyman 2018).

A historical perspective offers a different and more positive take on the benefits of licensure. Until late in the 19th century, entry to the health care market was entirely open (Mohr 2013; Starr 1982). The result was a proliferation of “irregular” providers, including purveyors of patent medicine and snake oil, “hydropathy,” phrenology, and magnetic medicine (Swenson 2021; Janik 2014; Mohr 2013; Starr 1982). Although the state of “regular” medicine at the time was unimpressive, “regular” providers were convinced that open entry was responsible for considerable pain, suffering, and death—and they lobbied state legislatures to enact licensing regimes (Swenson 2021; Mohr 2013; Starr 1982). After the US Supreme Court upheld West Virginia's licensing regime in Dent v. West Virginia (1889), other states quickly followed. Licensure resulted in improved academic training of physicians, higher professional status and income, and better quality of care (Swenson 2021; Kleiner and Krueger 2013; Starr 1982). Midwifery provides a similar story; licensure resulted in lower maternal mortality and nonwhite infant mortality (Anderson et al. 2021).

Licensure clearly improved the quality of health care in the early 20th century, but it has proven difficult to establish an ongoing positive impact of licensure on health outcomes (Kleiner 2006). Strict licensing requirements can actually reduce the overall distribution of high-quality services, as consumers who are unable to afford the increased costs of obtaining services from a licensed professional do without, engage in self-help, or turn to less safe (but cheaper) alternatives (Kleiner 2015; Kleiner 2006; Carrol and Gaston 1981).

Our analysis extends the information in the National Occupational Licensing Database (NOLD), a database constructed by the National Conference of State Legislatures, and builds on a 50-state study of occupational licensing barriers for ex-offenders (Sibilla 2020). The NOLD database covers 48 licensed occupations (including the five AHPs that we study) and all 50 states (NCSL 2020). The data was collected in 2017, and it reflects the following data points for each licensed occupation: whether there was a “good moral character” requirement; whether there was a blanket ban on ex-offenders; whether there were rehabilitation requirements; whether there had to be a relationship between the offense and the licensed occupation for it to provide a valid basis for exclusion; and whether there were any limitations on the scope of the licensing board's inquiry into an applicant's criminal record. The Institute for Justice (IJ) Barred from Working study analyzed whether states were employing “best practices for ensuring economic opportunity for ex-offenders without jeopardizing public safety” and scored each state from 0 to 100 (Sibilla 2020: 5).

For the five AHPs we study, table 1 summarizes the licensing framework in our 12-state sample and nationwide, using data from the NOLD. Each row in table 1 reflects the share of states in each column that had a particular occupational licensing requirement or restriction. The last row in table 1 presents the average score assigned in the Barred from Working study (Sibilla 2020).

Table 1 demonstrates that our sample of 12 states is reasonably representative of the entire United States, although our sample has a higher percentage of states with a categorical prohibition (16% vs. 9%) and a lower percentage of states that require a nexus between the past offense and the licensed occupation for it to provide a basis for exclusion (5% vs. 12%). In fairness, the NOLD data dates from 2017 and our data is from 2020, and our findings should be interpreted with these differences in mind.

For our five AHPs, we analyze in more granular detail the specific categorical and discretionary (i.e., “permissive”) criminal offenses that provide a basis for denying a state license. We also examine the factors that state licensing boards use when implementing permissive exclusions, and the extent to which such information is accessible to applicants.

Criminal Records, Race, and Health Equity/Health Justice

As noted previously, it has been estimated that between 70 and 100 million Americans have a criminal record. The 2015 report on occupational licensing by the Obama administration points to some of the complexities with accepting such figures at face value:

Regardless of how we elect to define a “criminal record,” certain demographic groups are still overrepresented relative to their share of the population, even after taking account of differences in the base rates of offending (Beck 2021; NRC 2014).

The literature on health care disparities and health equity/health justice is vast, and we do not attempt to summarize it here. Suffice it to say there are a substantial number of health (and health care) disparities, and education, income, and race all seem to matter. Whether cast as “health equity” or “health justice,” there has been considerable academic interest in these dynamics (Matthew 2022; Wiley et al. 2021; Benfer et al. 2020).

Methods

We focus on the specific state-level statutory and/or regulatory licensing requirements relating to applicants with a criminal record for five entry-level AHPs: dental hygienist, occupational therapy assistant, physical therapy assistant, respiratory therapist, and radiologic technologist. All five AHPs require an associate's degree or certificate to obtain a license. To ensure adequate geographic and demographic coverage, we used a stratified semirandom sample to identify 12 states to study (California, Colorado, Connecticut, Delaware, Florida, Illinois, Missouri, New York, Ohio, Pennsylvania, South Dakota, and Texas), and we collected data during late 2020. Our data set includes all six of the largest states (i.e., states with populations >12 million); four states selected randomly from the 37 midsized states (i.e., states with populations between 1 million and 12 million); and two states selected randomly from the remaining eight small states (i.e., states with population <1 million). Our data set covers 24% of the states and 50% of the US population. Stated differently, our sample covers 100% of the largest states, 11% of the midsized states, and 25% of the smallest states.

We collected information from federal government sources on all five AHPs, including median annual wages, expected growth (percentage increase from 2020 to 2030, compared to an overall expected growth of 8% for all professions), and the share of workers in each AHP who are female, African American, and Hispanic. Table 2 summarizes the results.

As table 2 makes clear, all five AHPs have impressive median annual wages, ranging from $50,000 to $77,000. By comparison, the median household income in the United States was $68,000 in 2020 (Shrider et. al. 2021).

All five AHPs have a disproportionate share of female providers, particularly dental hygienists (94%) and occupational therapy assistants (86%). Three of the five AHPs (dental hygienist, physical therapy assistant, and radiologic technologist) have fewer African Americans than their share of the working population (11.9%). All five of the AHPs have fewer Hispanics than their share of the working population (4.7%).

For each state and AHP, we used online resources (i.e., Lexis, Westlaw, and the websites for each state government and occupational licensing board) to collect the relevant state-level statutes and/or regulations involving licensure. For each, we coded whether licensure regulates entry; how a past criminal record was considered in deciding whether to grant a license (specifically, whether exclusion was categorical or discretionary and which offenses were listed as categorical vs. discretionary; whether past offenses were material to the licensing decision if they were not relevant to the practice of the AHP; and the temporal cutoff [if any] for considering past convictions); types of evidence for considering whether an applicant has been rehabilitated; relevance of moral character determinations to the licensing decision; and the degree of transparency of this information to those interested in pursuing each AHP. Finally, we analyzed the association between state-level attributes and IJ state scores on statutory protections for licensing applicants with criminal records (Sibilla 2020).

Findings

Table 3 presents a top-level summary of our findings. As the table indicates, seven states license all five of the AHPs in our sample, and four states license four of the five AHPs (in all four of these states, the nonlicensed AHP is radiologic technologist). Table 3 also shows that states rely on a mix of categorical and permissive exclusions, but permissive exclusions predominate. Indeed, 6 of the 12 states do not have any categorical exclusions. With one exception (radiologic technologists in Ohio), all AHPs are subject to either categorical or permissive exclusions in all states where they are licensed.

Six states have an explicit “moral character” requirement for granting a license, covering a total of 19 licensed AHPs (35% of the total AHP-state observations in our data set). Three states (California, Missouri, and Pennsylvania) explicitly prohibit the state licensing board from evaluating moral character, covering 13 licensed AHPs (23% of all AHP-state observations in our data set). We note that Pennsylvania changed its licensing regime effective December 28, 2020; had we coded the statute that was previously in effect, fully 30% of the total AHP-state observations in our data set would be subject to a moral character test.

Finally, five states have language on their website indicating that they engage in a case-by-case process of adjudication to determine whether an applicant qualifies to obtain a license. Of course, the implementation of permissive exclusions necessarily requires a case-by-case process of adjudication, but this “case-by-case” language might be interpreted by unsophisticated applicants as indicating that “anything goes” when state licensing boards are evaluating the importance of an applicant's criminal record, irrespective of what state licensing boards are actually doing.

Table 3 does not indicate the specific criminal offenses that trigger categorical and discretionary exclusion. Table 4 provides a breakdown by offense and AHP of the nine most frequently listed criminal offenses that trigger categorical exclusion. We provide detail on less frequently listed categorical exclusions in the appendix, table A-1.

As Table 4 indicates, the crimes that individual states have identified as meriting categorical exclusion are very serious criminal offenses. Table 5 provides a similar breakdown of the 16 most frequently listed permissive exclusions. We provide detail on less frequently listed permissive exclusions in the appendix, table A-2.

Almost all the permissive exclusions in table 5 involve serious criminal offenses, although not as uniformly serious as the categorical exclusions in table 4. Critically, the categorical and permissive exclusions listed in tables 4 and 5 require an actual conviction; simply being arrested for any of the specified offenses does not provide a valid basis for exclusion. The same cannot be said of states that empower their licensing boards to consider moral character when evaluating applicants. For example, the Texas board that licenses dental hygienists has determined that “criminal behavior is highly relevant to an individual's fitness to practice dentistry, dental hygiene, and dental assisting. Therefore, all arrests, criminal convictions or deferred orders, prosecution, or adjudication . . . must be reported” (TSBDE 2020).

Some states provide additional guidance on how permissive exclusions should be evaluated and implemented. Table 6 provides detail on the specific factors that must be weighed when implementing a permissive exclusion as well as the transparency of information regarding state board licensing practices and procedures.

As table 6 makes clear, a bare majority of states require a past criminal conviction to be substantially related to the scope of services provided by the AHP for it to provide a valid basis for exclusion, while a slightly larger majority make the relationship between the crime and the nature/duties of the profession a factor that the licensing board must evaluate (56% vs. 64%). Other factors include how much time has passed since the criminal conviction, the nature and gravity of the offense, and whether the applicant has been rehabilitated. As we detail in appendix table A-3, licensing boards are similarly directed to consult an extensive list of factors in deciding whether an applicant has been rehabilitated. Of course, none of these factors apply if an applicant has committed an offense that provides a categorical basis for exclusion. Stated differently, the factors listed in table 6 only apply if the applicant has committed an offense that provides a discretionary basis for exclusion or if the state employs a good moral character test. The listed factors channel and constrain the discretion of licensing boards, although boards still have considerable flexibility in handling applicants with criminal records.

What about transparency? Almost all state licensing boards have websites that provide links to the actual laws and regulations they are enforcing, but much less effort has been invested in providing context that makes the information useful and usable to applicants. As table 6 shows, less than 25% of licensing boards list or summarize the factors they apply on their websites, forcing applicants to parse complex statutory and regulatory language. In addition, most state licensing board websites provide only limited information about their operations, decision-making processes, and past precedents, making it difficult or impossible for applicants with less-than-pristine records to make accurate predictions about their prospects. Potential applicants might reasonably fear the worst, deterring them from even applying.

In fairness, licensing boards in a growing number of states have a process for potential applicants to find out whether their criminal record is disqualifying before they apply (Sibilla 2020). However, the process can be time-consuming and shame-inducing, and it may require the submission of costly certified court records.

Those considering whether to obtain the necessary training might look to sources other than the licensing board for information on whether a criminal record will disqualify them from obtaining the necessary license. For example, student counselors at vocational training schools or at community colleges might be a more accessible and useful source of information than the website of a state licensing board. We are not aware of a data source we could use to evaluate the frequency with which potential applicants rely on such sources.

Finally, in many states there has been a trend toward a more open approach to considering applicants with criminal records for occupational licenses (IJ 2022; NCSL 2020). Eight of the twelve states in our data set (66%) revised their licensing regulations in the three years before our study in ways that made it easier for applicants with criminal records to obtain a license. If we use a slightly longer time frame and take a national perspective, a clear majority of states (39 states plus the District of Columbia; 78%) have done the same (IJ 2022; Sibilla 2020). However, some of these states explicitly excluded health care–related occupations from these reforms, so one must examine each licensed occupation and state to determine the applicable requirements (IJ 2022; Sibilla 2020).

To what extent do state-specific factors explain the observed variation in occupational licensing practices for applicants with a criminal record? We used regression analysis (both ordinary least squares and spatial autoregressive models) to examine this issue, using the IJ Barred from Working (Sibilla 2020) scores as our dependent variable. As detailed previously, higher scores indicate states that have adopted more permissive practices for handling licensing applicants with criminal records. Our spatial model allows the dependent variable (score), independent variable (gross state product), and error term of the model to be spatially dependent. We define our spatial weighting matrix based on the distance between each state.

Table 7 evaluates whether state-level social, economic, political, and demographic factors might explain the large observed variation in these scores. We used two different measures to examine partisan intensity: whether the state government (i.e., control of the legislature and the governorship) was all Republican or all Democrat (with the other combinations constituting the omitted category), and the share of votes in each state for the Democrat and Republican candidates in the 2020 presidential election.

As table 7 reflects, depending on the regression, we find some evidence that larger states have higher IJ Barred from Working scores (models 5 and 6), and more diverse states have lower IJ Barred from Working scores (models 1, 5, and 6). In two of the models (models 5 and 6), states with a higher percentage of African Americans had lower IJ Barred from Working scores. In one model (model 6), states with a higher percentage of Asian Americans and higher income inequality had lower IJ Barred from Working scores. The share of Hispanics, state economic activity (gross state product, or GSP), and partisanship were not associated with IJ Barred from Working scores in any of our models. Our (non-)findings involving partisanship are consistent with earlier research, which indicates that criminal justice reform is the result of cross-party coalitions (TPPF 2021; CSGJC 2021; Keller 2016). Stated differently, imposing barriers to occupational licensing of applicants with criminal records is a game that both political parties have played—and neither appears to “own” reform of that issue.

Our spatial autoregressive models indicate that geographic proximity matters. The spatial lags for the IJ Barred from Working score (dependent variable) and GSP (independent variable) are both significant, while the lagged error term is not significant. These results indicate that the IJ Barred from Working scores of geographically proximate states are associated with one another, and the GSP of a state is also associated with the geographically proximate states' IJ Barred from Working scores.

In table 7, our independent and dependent variables are all from 2020. As an alternative, we also fit the same set of independent variables to the updated (November, 2021) IJ Barred from Working scores (Sibilla 2021). Our overall findings are qualitatively similar, but the magnitude generally increases. Violent crime gains significance in most models. The results using the 2021 IJ Barred from Working scores suggest the possibility of a lag in the impact of social, economic, and political factors on state licensing practices involving ex-offenders. We report a full set of regression results using the 2021 IJ Barred from Working scores in appendix table A-5.

Discussion

When regulating entry, the basic problem is error costs. The licensing board wants to admit applicants who meet the requisite requirements and will behave themselves once they are let loose on an unsuspecting public. Conversely, the licensing board also wants to reject applicants who fail to meet the requisite requirements, or who meet the entry requirements but will misbehave and exploit the trust reposed in them at some point in the future. Expressed in terms of error costs, the licensing board wants to maximize the number of correct decisions (i.e., the sum of true positives and true negatives) while minimizing the number of incorrect decisions (i.e., the sum of false positives and false negatives).

The difficulty of that task is substantially increased because licensing boards are in the business of making what is, for most providers, effectively “once and forever” up-or-down decisions based on incomplete information about a provider's likely future behavior. For understandable reasons, legislators and state licensing boards have long thought that an applicant's criminal record is helpful in determining whether an applicant belongs in the true positive or true negative box. The list of categorical exclusion offenses in table 4 is heavily weighted toward severe crimes, and the list of permissive exclusion offenses in table 5 includes multiple offenses that might well be relevant to an assessment of whether an applicant has the “right stuff” to be a licensed professional. In fairness, the “conviction” part of the equation is probably doing a lot of the work for the specific offenses in question.

To be sure, attitudes about criminal offenses can change over time; possession and/or consumption of marijuana will be evaluated very differently today from how it was evaluated in 1970, when the Controlled Substances Act classified it as a Schedule I drug (with no acceptable medical uses and a high potential for abuse). The same permissive approach is less likely for the possession and/or consumption of heroin and Oxycontin, indicating the complexity of developing and deploying hard-and-fast rules for handling such matters, as opposed to open-textured standards that are implemented in a discretionary fashion.

Licensing boards also know they can make mistakes, and a false positive (i.e., granting entry to someone who should not have received a license) is more likely to harm the general public (and more likely to result in unfavorable publicity for the licensing board) than a false negative (i.e., refusing to grant entry to someone who should have gotten a license). The result is that state licensing boards follow a “better safe than sorry” approach in evaluating applicants with criminal records, particularly for jobs in health care (Ewald 2019). The same dynamic affects other regulatory agencies that control market entry, such as the FDA.

Traditional politics matters as well. The categorical and permissive exclusions we describe are typically enshrined in statutes enacted by state legislatures, and the dynamics we have described above (i.e., better safe than sorry, particularly when dealing with criminals) make it easy to understand how these provisions get adopted in the first place. Stated differently, these restrictions do not appear to be an entrenchment scheme enacted at the behest of the AHPs, whose political power is likely quite limited. Reform will require legislative action; courts will not overturn even deeply silly exclusionary provisions (e.g., Gurrola v. Duncan 2022). That said, for those who believe that entrenchment is a one-way ratchet, the recent trend toward more permissive licensing of applicants with criminal records suggests that matters are more complicated.

These restrictions on entry may protect some patients from exploitation, but they do so by excluding some applicants entirely and making the task of obtaining a license more difficult and uncertain for many applicants. What are the likely consequences of this approach? Because criminal records are not distributed randomly across the population, these restrictions will predictably affect some subgroups more than others. Women will be less affected than men. African Americans and Latinos will be more affected than European Americans, who will be more affected than Asian American/Pacific Islanders. These predictions are broadly consistent with our findings in table 2, although more detailed longitudinal data on AHP demographics at the state level will be necessary to fully examine this issue.

By limiting the ability of applicants with criminal records to obtain a license, these restrictions will also lower the future earning potential of denied and deterred applicants. Since income and health are linked, limitations on entry that disproportionately affect the ability of members of certain demographic groups to secure well-paying and secure jobs in health care are likely to worsen both their health status and the health status of family members who depend on them for financial support, compounding the health equity problems outlined previously. In addition, previous research has found that some patients are more willing to follow prescribed treatments if their physician is from the same racial or ethnic group (Alsan, Garrick, and Graziani 2019; Silver et al. 2019). If the same dynamic applies to AHPs, these restrictions on entry might have analogous follow-on consequences for population health. Stated differently, to the extent that there are biases in the criminal justice system, the restrictions we describe have the potential to import these same biases into the health care system.

We also note an irony of organizational design. Discretion is often an engine of discrimination (Cole 1999). States are simultaneously giving licensing boards more discretion in certain domains (i.e., greater reliance on permissive exclusions) and less discretion in other domains (i.e., prohibiting the use of moral character tests and categorical exclusions, and requiring past criminal offenses to be substantially related to the scope of services provided by the AHP). It remains to be seen how state licensing boards will actually implement these commands.

Of course, multiple factors influence the demographics of the AHPs we study. Indeed, as table 2 indicates, the share of occupational therapy assistants and respiratory therapists who are African American is greater than their share of the working population, and far greater than their share of dental hygienists and physical therapy assistants. By comparison, Hispanics make up a very small share of dental hygienists and radiologic technologists, and their share of all five AHPs is smaller than their share of the working population.

Another complication is that all five of the AHPs we study are dominated by women. Men may be reluctant to enter a female-dominated AHP, even if they do not have a criminal record (Miller 2017). These patterns suggest that additional research and better data will be necessary to disentangle and quantify the various factors that influence the decision to obtain the necessary training and licensure to practice an AHP.

Additional complexity is introduced by the fact that the African American males who are able to obtain a license disproportionately benefit when those with criminal records are excluded, and that “licensed minorities and women experience smaller racial and gender wage gaps than their unlicensed peers” (Blair and Chung 2020: 1). A theoretical model by the same authors similarly found that “the licensing premium is higher for workers who are members of demographic groups that face a higher cost of licensing. The predictions of the model can explain, for example, the empirical finding in the literature that occupational licenses that preclude felons close the racial wage gap among men by conferring a higher premium to black men than white men” (Blair and Chung 2021: 201). A study of licensed dental assistants found that a secondary license (for conducting X-rays) reduced the wage gap between non-Hispanic white and minority dental assistants by 8% (Xia 2021). Stated differently, focusing exclusively on those who are excluded elides the distributional and financial consequences for applicants able to navigate the exclusionary restrictions.

Licensing board websites almost always contain hyperlinks to the statutes they are enforcing, but statutory language is often impenetrable. Almost nothing has been done to make such information more useful and usable to applicants, let alone to those considering undergoing the necessary vocational training. It is one thing to know that some crimes constitute a permissive basis for exclusion, and entirely another to know that the state board ultimately awards a license to 5% (or 95%) of applicants with those crimes on their record.

If state licensing boards were more transparent about such matters, it would be relatively simple to create an online algorithm that would ask potential applicants a few questions (e.g., what state and AHP are you considering? What specific criminal offenses were you convicted of? When were you convicted? Does the criminal offense in question have anything to do with the services provided by the AHP?) and then spit out the actual probability of obtaining the desired license in the specified state. The failure to make public the “common law” of how state licensing boards handle applicants with criminal records is an underappreciated factor deterring such individuals (who are disproportionately likely to be from historically marginalized groups) from even considering obtaining the training to become a licensed worker in an AHP.

Finally, federalism (i.e., state-imposed restrictions, rather than a national standard) is both a strength and a weakness. On the one hand, federalism allows states to experiment with various strategies for dealing with licensing applicants with criminal records—with some states retaining the traditional global “moral character” touchstone—while other states have embraced more concrete categorical and discretionary exclusion factors and spelled out a framework that state licensing boards must use in evaluating individual applicants. On the other hand, state-specific licensing limits geographic mobility and worsens supply shortages, thereby increasing costs and reducing access. You pays your money and you takes your chances.

Conclusion

We find some evidence that restrictions on whether ex-offenders can obtain occupational licenses have had a disparate impact on some (but not all) historically marginalized groups in some (but not all) of the five AHPs we study. Multiple states have recently loosened these restrictions, but the impact of these changes remains to be seen. State licensing boards retain significant discretion in handling applicants with criminal records—even in the states that have banned the requirement that applicants demonstrate “good moral character.”

It is ultimately an empirical question whether the benefits from any given set of restrictions on entry exceed their costs, including the potential for disparate impact on historically marginalized groups. That said, there are no easy answers here—only trade-offs among competing values and tragic choices (Calabresi and Bobbit 1978).

Acknowledgments

Courtney Stone Mirski (Georgetown University Law Center) provided excellent research assistance. We appreciate the helpful comments from Jim Morone and three anonymous peer reviewers.

References