This is a brief of the study of the article Re-evaluating the labor market effects of occupational licensing: Longitudinal evidence across states. Humanit Soc Sci Commun 12, 240 (2025). https://doi.org/10.1057/s41599-025-04497-5
We present new data on regulatory restrictions across states and occupations between 2017 and 2022 to study the labor market effects of occupational licensing. First, we document three stylized facts: (a) regulatory restrictions across states and occupations have grown by nearly a factor of three since 2019, (b) increases in regulatory restrictions are concentrated in occupations with lower median hourly wages and higher within-occupation inequality, and (c) states that expanded regulatory restrictions tend to have lower GOP vote shares. Second, exploiting variation across occupations within the same state-year, we find that a 10% rise in regulatory restrictions leads to a 2% rise in hourly wages, but a 4% decline in employment. Both the employment and wage effects are concentrated in lower wage jobs, as well as among respondents with professional licenses, even after controlling for differences across industry.
Introduction
The share of the workforce subject to occupational licensing requirements has surged from roughly 5 to 25 percent between 1950 and 2013 (Kleiner and Krueger 2013). (Footnote1) While these licensing requirements are associated with higher pay and benefits among those with the license (Gittleman and Kleiner 2016; Gittleman et al. 2018), they are also linked with lower interstate mobility (Carpenter et al. 2017; Johnson and Kleiner, 2020; Knepper et al. 2022), labor supply and entry (Blair and Chung 2019), and competition (McLaughlin et al. 2017). A major empirical challenge for estimating causal effects in these studies, however, is that state licensing reforms are correlated with other political and economic factors. Nevertheless, understanding the presence and effects of licensing restrictions on workers comes at an important time, especially given the expansion of artificial intelligence, automation, and regulatory reform; these discoveries will require individuals become even more adaptive and responsive to change as technological advances rapidly take form.
This paper introduces a new methodological approach to understand the cross-sectional and time series properties of occupational licensing. First, unlike traditional approaches that have focused on binary indicators for occupational licensing requirements, we introduce a strategy for obtaining more comprehensive and continuous measures based on all available state regulatory data. The algorithm proceeds in two steps. We first collect over 600 state regulatory and federal regulatory documents, and manually label them as either containing or not containing occupational licensing text. (Footnote2) We additionally label the documents that contain occupational text based on the comprehensiveness of the text and the occupation described. We subsequently use those documents to train and cross-validate a machine learning classifier (e.g., logit or random forests) that reliably predicts a document’s relationship to licensing out of sample for all regulatory documents across states. (Footnote3) Our machine learning approach classifies regulatory documents based on their language, allowing us to identify words and groups of words that are commonly associated with manually identified cases of occupational licensing. (Footnote4) These data-driven features of our approach allow us to create a fully comprehensive and continuous measure of licensing requirements, which are especially useful since policymakers are most interested in how varying degrees—not just the presence—of licensing stringency and amount will influence economic outcomes.
Second, using the developed measure of licensing across states and occupations, we benchmark our new continuous measure with existing data on licensing reforms and document three novel facts: (a) regulatory restrictions across states and occupations have grown by nearly a factor of three since 2019, (b) increases in regulatory restrictions are concentrated in occupations with lower median hourly wages and higher within-occupation inequality, and (c) states that expanded regulatory restrictions tend to have lower GOP vote shares. Next, we use our measure to revisit the literature on the labor market effects of occupational regulatory restrictions. Exploiting variation across occupations within the same state-year, we find that a 10% rise in regulatory restrictions leads to a 2% rise in hourly wages, but a 4% decline in employment. We also compare employment and earnings between licensed and unlicensed individuals (i.e., professional certification), revealing that state-level increases in licensing restrictions primarily impact those with licenses.
Our paper is related to a growing empirical literature on the effects of occupational licensing on labor markets. Ever since Kleiner (2006), there has been a general recognition that these restrictions can raise quality in a market by establishing certain standards and that licenses cover nearly a third of the labor force (Kleiner and Krueger 2010, 2013). (Footnote5) While licensing restrictions can raise wages similar to that of unionization (Kleiner and Krueger 2010), the bulk of the literature also finds that licensing restrictions adversely affect employment and career outcomes. For example, Johnson and Kleiner (2020) show that individuals exposed to state-specific licensing exams have 36% lower migration rates than their peers in other occupations. Furthermore, Kleiner and Xu (2023) show that licensing restrictions also reduce labor market fluidity (i.e., movement in and out of occupations). Taking into account possible effects on quality, Kleiner and Soltas (2023) find that the net effects are negative: a 12% decline in surplus where workers bear 70% of the losses and consumers 30%. (Footnote6)
Our paper adds a novel dimension of longitudinal and cross-sectional variation. Unlike prior studies that have largely relied on cross-sectional variation in licensing restrictions by state and occupation, and have therefore been subject to concerns about endogeneity, we exploit within-state variation across different occupations. For example, Gittleman et al. (2018) find that licensed workers have not only higher wages, but also higher employment rates. We replicate these results, but show that increases in licensing restrictions reduce employment rates. This could reflect broader general equilibrium effects, i.e. by stifling job mobility and the returns to investing in workers, employment is adversely affected. Although our estimates rely on cross-sectional variation across occupations, they overcome classic concerns about the endogeneity of state policy confounders and nonetheless are a step forward to understanding the causal effects of licensing restrictions on labor market outcomes. Our paper also closely complements recent evidence from Bae and Timmons (2023) who show the positive effects of universal licensing reforms on employment ratios.
Conclusion
There is a large and growing literature on the effects of regulation on the labor market and the allocation of talent. In particular, the literature on occupational licensing has pointed towards a sharp increase in individuals covered by licensing restrictions and a negative effect on employment, even though the incumbents may benefit in the form of higher wages. There is also growing concern that these licensing restrictions stifle career mobility and labor market dynamism.
We introduce new data by collecting state regulatory documents from all available states since 2017 and apply a machine learning classifier to count the number of restrictions. Our measure exhibits a strong correlation with existing measures of occupational licensing, but reflects more of the intensive margin of regulation. We document an increase in licensing restrictions with the growth concentrated in lower wage occupations and more Democratic states. We subsequently exploit within-state variation and find that increases in regulatory restrictions are associated with declines in employment, but increases in hourly wages.
Our results are consistent with the arguments of policymakers who would further reduce the prevalence of occupational licensing. Occupational licensing reform has many proponents at the highest levels of politics. For example, the Obama White House issued a 77-page report outlining many of the problems that occupational licensing creates, as well as a framework for reforming such laws (White House 2015). Similarly, and more recently, the Trump White House recently issued an executive order titled “Unleashing Prosperity Through Deregulation” mandating that, for every new regulation proposed by federal agencies, at least ten existing regulations must be identified for elimination. The challenge is that the White House can do little about state policies, and even reform-minded coalitions at the state level often find it difficult to overcome the political inertia created by laws that both protect a specific group of individuals’ jobs from competition and artificially elevate their wages. Nevertheless, federal leadership can encourage both implicitly and explicitly state-level action. Our results highlight results effects along an intensive margin, not only an extensive margin, which suggests that modifications to occupational licensing laws and regulations that simplify or reduce their net burden, but fall short of removing them altogether, could still substantially reduce their harm.
Our findings also raise several questions for future research. First, our methodological approach could complement existing and more manual based approaches: human labelers could help provide context for AI-based methods to better understand the context of state legislation. Second, our results underscore the importance of longitudinal variation in recovering possible causal effects. While we do not have much variation within occupations over time, future research could trace mobility decisions among individuals who are in the same occupation, but move states. Third, we know little about how occupational licensing interacts with other regulatory distortions, like land use restrictions that curtail the housing supply. We leave these questions to future work.
Authors’ (Makridis, C.) comment:
The share of the U.S. workforce subject to occupational licensing requirements has surged over the years. While licensing has been associated with higher pay and benefits for those who obtain credentials, research suggests it also restricts labor mobility, limits workforce entry, and reduces competition (e.g., see Morris Kleiner, Edward Timmons, among others).
Our recently-published work introduces a novel methodological approach to measuring occupational licensing restrictions. Trained on over 600 state and federal regulatory documents, we create a continuous measure of licensing restrictions and use it to document three empirical patterns and estimate some effects on wages and employment.
Empirical Results:
1) Licensing restrictions have grown nearly threefold since 2019 with the most significant increases concentrated in lower-wage occupations and those with higher within-occupation inequality.
2) States that expanded regulatory restrictions tended to have lower GOP vote shares, suggesting a political dimension to licensing expansions.
3) A 10% increase in licensing restrictions leads to a 2% increase in hourly wages, but a 4% decline in employment.
Comparisons between licensed and unlicensed individuals from the CPS data before/after show that state-level increases in licensing primarily impact those with licenses.
The results support the growing bipartisan support for licensing reform, including the recent efforts from the administration to reduce the regulatory burden through the 1-to-10 rule (among others). The biggest burden of these often falls on the lower wage workers!
The full article was originally published in Nature.
