New Technology, Same Anxieties: Economic Freedom, Labor Markets and the AI Transition
Almost from the moment ChatGPT was launched in 2022, fears that artificial intelligence (AI) will produce a “job apocalypse” have abounded. Popular concerns regarding technological unemployment have been stoked by predictions from technology company executives that AI could eliminate half of all entry-level jobs in the next one to five years or that work will become optional, among other headline-grabbing forecasts. Lawmakers across the country have reacted to these concerns with a variety of policy proposals, ranging from state-mandated investments intended to offset potential labor market disruptions to (once again) implementing a universal basic income to even the government taking ownership stakes in AI companies.
However, as a growing body of research has found, widespread AI-driven disruption in labor markets has yet to be seen. Rather, AI’s labor market impact has so far varied depending on the particular application. Where it can augment (i.e., complement) human effort, AI has been shown to actually yield employment and wage gains. Where AI can effectively replicate human task completion, however, evidence suggests AI has yielded some displacement, with one study finding that this has led to some declining employment opportunities among entry-level workers. Overall, the aggregate economic effects of AI on labor markets remain modest, though the picture continues to evolve rapidly.
As models improve and become more widespread, it seems reasonable to expect that AI will meaningfully alter the composition of work and the demand for certain kinds of skills. For some occupations, this could entail substantial automation, potentially exposing workers to displacement. The key policy question is, therefore, how to help workers transition into new opportunities with minimal disruption. That is, how to minimize the costs of transitioning into an “AI-enabled” economy, while also enjoying the productivity benefits that AI allows us. While most policy proposals aimed at addressing this possibility have mainly called for greater government intervention into markets, an alternative approach would be to remove regulatory barriers that impede the ability of workers to adjust to the new equilibrium created by AI.
Governments have gotten quite creative at finding new ways to impede labor market competition and fluidity, but the oldest and perhaps most salient barrier to entry is occupational licensing. Reforming, or ideally eliminating, these barriers would go a long way towards creating the kind of institutional flexibility that can minimize the time spent unemployed or searching for opportunities, and permit the fuller realization of the potential benefits of AI.
Shock Absorbing Institutions
It is often argued in policy debates that economic shocks or the introduction of novel technologies create “scarring” that is concentrated in certain regions or local economies in the United States whose industries were particularly exposed to these disruptions. Job losses resulting from increased foreign competition, recessions, or automation compound into regional economic decline and social malaise. This argument maps cleanly into support for job retraining programs, enhanced social safety nets, and broader arguments for targeted industrial policies aimed at correcting the effects of these shocks.
Yet this view ignores the fact that economic shocks and structural transformations do not occur in a vacuum, but are mediated by the preexisting institutional environment, the formal laws, rules, and regulations within which individuals and firms operate. The extent to which a particular region suffers long-term economic dislocation or successfully adapts is not simply the product of the shock itself, but of how that shock interacts with the regulatory environment in which adaptation takes place. The quality of institutions, therefore, plays a significant role in shaping economic resilience and adaptation.
Research suggests that economically freer institutions help to lighten the blow of recessions and other economic disruptions. Broadly speaking, such institutions are characterized by lower barriers to entrepreneurship, more flexible labor markets, lighter regulatory burdens, and stronger protections for private property. For instance, recent research has found that U.S. metropolitan areas that undertook economic liberalizations before the Great Recession experienced comparatively faster recoveries in employment rates and incomes than less free areas. Looking internationally, another study found that economically freer countries experience less severe crises and have shorter recovery times than less free countries. By imposing fewer and less burdensome barriers on entrepreneurs and workers to reallocate resources following a business downturn, economically free institutions help to make the areas governed by them more resilient and adaptive.
Occupational Licensing, Disruptive Technology, and AI
Given the benefits of economic freedom in mitigating the effects of crises, it seems reasonable to expect that it may help regions adapt to disruptive technological change. Considering AI’s significant overlap with labor markets, the rules and regulations governing those will exert a significant effect on how its impacts are felt. Labor markets, when relatively free of regulation, provide workers with the incentive to seek out the highest valued use of their time, net of search costs. Employers, meanwhile, face an incentive to seek out and retain quality workers, bidding them away from competitors with more generous compensation. Over time, this process results in worker earnings roughly equaling what economists refer to as their “marginal revenue product.”
This market process, however, does not operate seamlessly. There are costs to seeking out and applying for jobs, in addition to the costs to employers of retaining employees and measuring their quality. These transaction costs can affect how long it takes workers to find work and be hired, potentially increasing the time between jobs.
Such costs are heightened by regulatory barriers that limit entry into various occupations. Since 1960, the proportion of U.S. workers required to have a license to practice has grown significantly, rising from 5 percent of the total labor force to 25 percent by 2020. Licensing requirements, ostensibly created to ensure the quality and safety of the goods and services covered, frequently function as barriers to entry that protect current providers and impede labor-market adaptation. As AI changes the composition of work and demand for particular skills, occupational licensing may make it more difficult for displaced workers to transition into growing occupations, slowing labor-market adjustment.
Recent research from labor economists finds evidence of licensing having these effects. A study examining the impact of licensing on occupational mobility found that for both employed and unemployed workers, licensing requirements function as a barrier to entry to both switching occupations and entering new ones. Other research has found that licensing reduces geographic mobility since credentials tend to be state-specific, and also contributes to earnings inequality by limiting entry into highly paid occupations.
Examining the interaction between licensing requirements and automation, another set of economists found that licensing is consistently related to lower income mobility. More importantly, they also found that reducing occupational licensing requirements can offset more than 90% of the negative effects that automation exposure has on workers’ ability to move up the income ladder. While their work deals with robotic automation, the mechanism is the same. Labor market rigidities extended the adjustment period for displaced workers, whereas jurisdictions with fewer regulations allowed workers to adjust much more quickly.
The existing evidence suggests that occupational licensing could potentially limit the ability of workers to adapt to changes brought about by the diffusion of advanced AI models, placing added strains on social safety nets and possibly increasing demand for government interventions to halt the spread of such technology. These effects alone could significantly erode the potential economic benefits of AI, both by lengthening the adjustment period for displaced workers and by limiting entry into occupations that will grow as a result of AI-driven productivity increases.
Occupational licensing also has a negative effect on entrepreneurship. This is significant since it is primarily through competitive entry and entrepreneurial experimentation that innovations and technologies are introduced. A growing body of research has found that restrictive licensing regimes reduce entrepreneurship, both by reducing sales among self-employed firms and by limiting new entry into markets. Given that AI has been shown to boost small business formation through the reduction of startup costs, licensing could limit the adoption of AI tools by constraining one of the primary channels by which innovations are introduced.
Technological Problems Require Institutional Solutions
AI stands poised to be the next general-purpose technology, akin to electricity, microprocessors, or the automobile. As with these previous technologies, the creation of new opportunities, products, and knowledge that is brought about is likely to outweigh the potential destruction of old forms of work and production. Current fears about AI producing a mass of structurally unemployed workers are simply not borne out by existing labor studies.
Technological innovations, like other economic changes and shocks, are filtered through the existing institutional structure in which economic activity takes place. To address the possibility of labor market scarring, policymakers, rather than attempting to devise job retraining programs or income redistribution schemes, should focus on getting the “rules of the game” right. Occupational licensing deregulation is one policy lever officials should give strong consideration to. As noted above, occupational licensing deregulation can reduce the downside risks created by automation by lowering the costs faced by displaced workers of switching jobs or starting a new job after a period of unemployment. The legitimate functions of licensing, assuring consumers of the quality of the goods and services they are purchasing, can be more effectively provided by private certification providers operating in competitive markets.
To be sure, such deregulation will likely need to go hand-in-hand with reforms to state and local social safety nets to ensure incentives to find new work remain, as well as adjustments to collective bargaining arrangements where these impinge on AI adoption. Nevertheless, ensuring that labor markets are free and flexible is a better solution to the potential disruption resulting from to AI’s diffusion than the alternatives.