Most debates about helping low-wage workers start and end with the minimum wage. A more useful question is often different: what stops this worker from moving into a better-paying job they could already do? For millions of Americans, part of the answer is a government license. Before you can cut hair, do nails, install drywall or work as a security guard, many states require you to pay fees, finish months of training and pass exams. That is a barrier written into law, and it falls hardest on people with the least money and time to spare.

Friedman's other chapter

Milton Friedman's argument against wage floors is well known; we cover it in What Milton Friedman actually argued about the minimum wage. Less often read is Chapter IX of Capitalism and Freedom (1962), "Occupational Licensure."

In it, Friedman separates three forms of regulation that are often lumped together. Registration means a practitioner simply has to list their name with the government. Certification means a government agency vouches that someone has met a standard, but people without the certificate are still free to do the work. Licensure means nobody may do the work without the government's permission. He argues that the case for each step is weaker than for the one before. Certification gives consumers the information they need about quality, while licensure takes away their choice of whether to rely on it.

His political point is the one that has aged best. The pressure for licensing, he observes, rarely comes from consumers who were harmed. It comes from people already in the occupation, who have the most to gain from limiting how many newcomers can compete with them. He applies this most fully to medicine, the hardest case, and concludes that even there licensure is not justified and that certification would do. That conclusion built on his earlier research with Simon Kuznets, Income from Independent Professional Practice (1945), which argued that restrictions on entry helped explain why physicians earned more than comparable professionals.

Few people, including many free-market economists, would go as far as abolishing medical licenses. But the argument is much stronger for the low-income occupations that licensing has since spread into.

How many workers need a license?

The answer depends on how you ask, and the honest range is roughly one-fifth to three-tenths of the workforce.

  • The Bureau of Labor Statistics asks about licenses in the Current Population Survey. In its 2025 annual averages, 21.6 percent of employed people held a license, and 24.0 percent held a certification or license. (Because the October 2025 survey was not collected, these are 11-month averages.)
  • Morris Kleiner and Alan Krueger's specially designed survey, later published in the Journal of Labor Economics (2013), found 29 percent of workers were licensed and 35 percent were licensed or government-certified.
  • The Obama White House's 2015 report put the share in occupations licensed at the state level at about 25 percent, up from less than 5 percent in the early 1950s.

Whichever figure you prefer, licensing now reaches far more workers than labor unions or the federal minimum wage. The BLS put the union membership rate at 10.0 percent in 2025, and the share of hourly paid workers earning at or below $7.25 at 1.0 percent in 2024.

Licenses for low-income jobs

The 2015 report, Occupational Licensing: A Framework for Policymakers, was prepared by the Treasury Department, the Council of Economic Advisers and the Labor Department. It is worth reading precisely because it did not come from licensing's usual critics. Among its findings:

  • Over 1,100 occupations were licensed in at least one state, but fewer than 60 in all 50. Funeral attendants were licensed in nine states and florists in one.
  • Michigan required three years of education and training for a licensed security guard, while most other states required 11 days or less.
  • In half the states, an applicant could be denied a license over any criminal conviction, whether or not it was relevant to the job or recent.

The Institute for Justice, a public-interest law firm that litigates against licensing rules, has tracked the same problem for 102 lower-income occupations across three editions of its License to Work study (2012, 2017 and 2022). The third edition covers all 50 states, the District of Columbia and, for the first time, Puerto Rico. In the institute's discussion of the release, a co-author summarized the findings:

  • More than 2,700 licenses across the 50 states and D.C. for those 102 occupations.
  • On average, a license required nearly a year of education and experience, at least one exam and $295 in fees. Those figures exclude tuition and lost earnings; IJ's separate research put average cosmetology school tuition at $16,000.
  • 88 percent of the occupations studied were unlicensed in at least one state.
  • An entry-level emergency medical technician needed about a month of training on average, while a cosmetologist needed about a year.

IJ is an advocacy group, but these are counts of state requirements that anyone can check. The pattern they describe is hard to defend on safety grounds alone: when most states let people do a job without a license, or ask for a month of training to save lives but a year to cut hair, something other than public protection is setting the rules.

For a low-wage worker, a year of unpaid training is not a paperwork hurdle. It is a year of income that a parent working two jobs probably cannot give up. A licensing wall decides whether they can move from a $12 job to a better-paying trade at all.

What the research finds

Three findings recur in the economics literature.

Licensed workers earn more. Kleiner and Krueger found that licensing is associated with about 18 percent higher wages, while government certification has a much smaller association with pay. The 2015 report summarized estimates of a 10 to 15 percent wage gap between licensed and unlicensed workers with similar education and experience, and noted that methods that better control for worker differences tend to find smaller effects, under 10 percent, and sometimes none. These are wage premiums for people who got in, not gains for workers in general.

Consumers pay more, and employment falls. The White House report cited price effects of 3 to 16 percent and noted that in several studies licensing did not measurably improve quality. The most comprehensive recent study is Kleiner and Evan Soltas's "A Welfare Analysis of Occupational Licensing in U.S. States," published in the Review of Economic Studies in 2023. Comparing occupations that are licensed in some states and not others, they find that licensing raises wages and hours but reduces employment.

The net effect is a loss, mostly borne by workers. Kleiner and Soltas estimate an average welfare loss of 12 percent of the occupation's total surplus. Workers bear about 70 percent of that cost and consumers about 30 percent. Higher wages compensated workers for only about 60 percent of the cost of the training the licenses required. The workers who never entered the occupation because of the barrier count too, and they do not show up in a wage premium.

Licensing, mobility and employer power

Licenses are issued state by state, so moving across a state line can mean starting over. Janna Johnson and Kleiner's study of 22 licensed occupations, "Is Occupational Licensing a Barrier to Interstate Migration?" (American Economic Journal: Economic Policy, 2020), found that workers in occupations with state-specific licensing exams moved between states at a rate 36 percent lower than members of other occupations. Occupations that use national exams showed no such effect, and reciprocity agreements increased moves among lawyers. The 2015 report found that about 35 percent of military spouses in the labor force worked in jobs requiring a state license or certification, and that they were ten times more likely than civilians to have moved across state lines in the past year.

This matters for the strongest argument for a minimum wage. In Monopsony: the strongest case for a minimum wage, and its limits, we explain why employers can hold wages below what workers produce when workers have few alternatives. A wage floor tries to patch over that power after the fact, and it only helps in a narrow range. Licensing reform attacks the cause. Every license that stops a cosmetologist from taking a better offer in the next state leaves the worker with fewer options and the current employer with more power. Removing those barriers raises what workers can command in the market without fixing a single price.

Universal recognition and other reforms

The most important reform of the last decade started in Arizona. House Bill 2569, approved on April 10, 2019, requires Arizona licensing boards to license new residents who have held a license in good standing in another state for at least a year, have passed any required exam there and have no unresolved discipline. Boards may require supervised practice for applicants licensed fewer than five years. The Goldwater Institute, which designed the model, reports that Arizona was the first state to pass universal recognition and that more than 10,000 people have obtained a license through it.

The idea spread quickly. The Archbridge Institute's 2026 State Occupational Licensing Index counts 28 states with some form of universal licensing recognition. The laws vary in strength. Some include a "substantially similar" clause that lets boards reject applicants whose home-state training differs from local rules, and some require residency. Those versions help fewer workers.

Recognition only helps people who already hold a license. For workers trying to get their first one, the reforms that matter are the ones the 2015 report recommended: limit licenses to real health and safety risks, weigh costs against benefits before creating new ones, and allow certification or registration where a license isn't needed. Add to those reforms of blanket criminal-record bans, which keep people with records out of jobs that would help them stay out of trouble. None of it puts a price floor on anyone's labor.

The case for licensing

Licensing has real defenders, and some of their points hold up.

  • Some licenses protect people. The 2015 report itself says carefully designed licensing can improve quality and safety and give workers a clear path to build skills. Few people would want an unlicensed surgeon or pilot. Friedman's alternative, certification plus liability for fraud and negligence, works better when consumers can judge quality after the fact than when one mistake is catastrophic.
  • Consumers seem to value some of it. Kleiner and Soltas find that consumers' higher willingness to pay offsets about 80 percent of the price increases licensing causes. That suggests licensing does deliver some quality or assurance people care about. They also find more favorable welfare effects in occupations where licensing is more common.
  • Licensed workers are workers too. Many licensed workers earn modest incomes, and the wage premium is real money to them. Reform that lowers the premium, by letting more people in, has losers among people already in the occupation.
  • The evidence is uneven. Wage effects shrink with better methods, quality studies are mixed rather than uniformly negative, and many results come from a small group of researchers using a limited set of occupations.

Our answer is that these points argue for keeping licenses where the risk is serious and the requirements match it, not for the current system. The best evidence shows workers carrying most of the cost. The same jobs are done unlicensed in other states without obvious harm, and certification can give consumers quality signals without banning anyone from work. More of the relevant research is collected in our research tracker.

The bottom line

The minimum wage debate asks how much the law should force employers to pay for a job. Licensing reform asks how many jobs the law should forbid people to take. For a low-wage worker, the second question can matter more. A $295 fee and a year of unpaid training can be the difference between a lifetime of entry-level pay and a trade that pays well, and the first job ladder only works if the higher rungs are open.

This is the site's thesis in its most practical form. Instead of mandating higher wages and hoping the jobs survive, remove the barriers that keep people out of better-paying work. Licensing reform has unusual allies across parties, as the Obama administration's report and Arizona's law show, and it doesn't ask employers to pay more than a job produces. If you want the broader argument, start with The case against the minimum wage: 9 arguments, with the evidence.

Sources and further reading