Occupational licensing

Why does a customer accept a licence as proof of a skill they cannot judge themselves?

You hire an electrician. They rewire a socket, it works, you pay. Now ask the uncomfortable question: do you know whether the job was done well? The socket working proves almost nothing — an undersized cable, a loose terminal, a missing earth all give you a socket that works perfectly for years and then does not. You cannot judge the work before you buy it, and here is the strange part: you cannot reliably judge it after either. So on what basis did you choose this person?

Economists sort purchases by when you can tell what you got. Some qualities you can inspect beforehand — the size of an apple, the length of a plank. Some you learn on use: a meal, a haircut, a hotel room. And some you may never learn, because verifying them would require the very expertise you were hiring. A diagnosis you did not need, a repair that was not performed, a filling placed in a sound tooth: these are credence qualities, and the market for them has a hole in the middle of it.

Follow what that hole does. If buyers cannot tell good work from bad, they cannot pay more for good work — so the careful practitioner, who is slower and has more to cover, is undercut by the careless one, whose output looks identical at the moment of sale. Reputation patches some of this: repeat custom, neighbours, reviews. But reputation needs the failure to eventually surface and be traced back, and the difficulty with credence goods is that it often does not. A bad weld inside a wall surfaces, if at all, years later and to someone else.

So the buyer looks for something they can evaluate — not the work, which is beyond them, but a fact about the worker that someone competent has already checked. That is what a licence is: an assessment made by a third party, in advance, travelling with the person. It does not tell you this job was done well. It tells you this person once cleared a threshold set by someone with the relevant expertise. You are not judging the wiring; you are borrowing somebody else's judgement of the wirer.

Notice that the licence has to be hard to obtain to carry any information. If everyone can get one for a fee, it distinguishes nobody — which brings us straight to the other half of the story, because making a credential hard to obtain restricts who may work. That is not a side effect; it is the same fact seen from the supply side. Licensed practitioners earn a wage premium over comparable unlicensed workers — commonly estimated around ten to fifteen per cent, though how much is the licence and how much is the sort of person who obtains one remains argued — and prices to consumers are higher. Licences are usually tied to a jurisdiction, so a qualified person crossing a border may be unable to practise, which measurably reduces mobility.

And the quality side? This is where honesty is required. For a handful of occupations where errors are catastrophic and irreversible — anaesthesia, structural design, aircraft maintenance — almost nobody argues. But across the wider range of licensed trades, the evidence that licensing raises the quality of service delivered to consumers is weak and inconsistent; many studies find higher prices and restricted supply with no detectable improvement in outcomes. There is a perverse channel too: if licensing prices some customers out of the formal market, they do the work themselves or hire someone unlicensed, and average safety can fall even as the licensed tier improves. Scope has drifted, into occupations where the credence problem is thin and the harm mild — Louisiana required a licence to work as a florist for decades.

The two halves are not competing theories — they are the same mechanism described twice. A credential informs buyers precisely to the degree that it excludes people, so the informational benefit and the entry restriction cannot be dialled independently. Arguing about licensing is arguing about where on that trade-off a particular occupation belongs.

So the real test is not "is this skilled work?" but two questions asked together: how badly can the buyer judge the outcome, and how severe is the harm when it goes wrong? Where both are high, some verified threshold is hard to do without. Where the buyer can judge and the harm is mild, reputation does the job at lower cost.

There is also a middle option that is easy to forget, because the debate is usually posed as licence or nothing. Certification protects a title without restricting practice: anyone may do the work, but only the certified may claim the credential. That keeps most of the informational value and removes most of the entry restriction — which is why the "buyers cannot judge" argument, true as far as it goes, does not by itself get you to licensing.

• How the market for used cars unravels when quality is hidden, and what that predicts for services.
• Why licences rarely lapse or require re-testing, and what that implies about what they certify.

Every term the collection defines is gathered in the glossary.