Vacancies alongside unemployment

Why do unfilled jobs and jobless workers persist side by side instead of clearing each other away?

Open a newspaper in almost any month of any year and you will find both complaints side by side: employers who cannot fill posts, and people who cannot find work. The obvious reading is that something has gone wrong — the wage is stuck, or the schools taught the wrong things, or somebody is not trying.

But the coexistence is not occasional. It is permanent, in every economy that measures both, in booms as well as slumps. A phenomenon that never goes away is rarely a malfunction. So what would have to be true for empty jobs and idle workers to be the normal state of a labour market that is working?

Begin by noticing that a job and a worker are not the two halves of a commodity trade. A tonne of wheat is interchangeable with any other tonne, and a price can clear the market in an afternoon. A job is a particular set of tasks in a particular place at a particular time, and a worker is a particular bundle of skills, constraints and preferences. Each side must find, evaluate and agree with a specific other. That takes time, and time is the whole story.

Watch what time alone does, with no failure of any kind. Suppose that in each month two per cent of employed people separate from their jobs — quits, dismissals, firms closing — and that an unemployed person has a thirty per cent chance of finding work that month. In a steady state the two flows balance: the inflow into unemployment, two per cent of those working, must equal the outflow, thirty per cent of those unemployed. Solve it and unemployment settles at 0.02 divided by 0.32, about 6.3 per cent. Nobody in that economy is unemployable and no wage is stuck. The stock exists because the average search lasts about 1/0.30, a little over three months, and three months of anybody's time is a stock.

Do the same on the other side. Hires each month are thirty per cent of that 6.3, about 1.9 per cent of the workforce. If a firm takes on average a month to fill a post, then at any instant roughly 1.9 per cent of the workforce's worth of posts are open. A stock is a flow multiplied by a duration, and neither duration is zero, so neither stock can be.

That is the core answer, and it is almost embarrassingly simple: both stocks are the queue that any matching process with a positive search time must carry.

Push further, though, because delay is not the only thing going on. Ask why anyone waits at all. An unemployed worker who accepts the first offer ends their unemployment immediately; many do not. They hold out because a match is durable, and a bad one is expensive to leave — so it can be rational to reject a poor offer and keep looking, exactly as it can be rational for a firm to leave a post open rather than hire someone who will do the job badly for five years. Both sides are choosing search over a bad match. Waiting is a decision, not a symptom.

Which changes who bears what. The firm with an unfilled post loses some output; the worker without a job loses their entire income. That asymmetry is why the ability to wait matters so much: a worker with savings, a partner's earnings, or unemployment insurance can decline a poor offer and take a better-fitting job, which is measured as more unemployment and generally produces better matches. Who can afford to keep searching shapes what the market produces, not just how long it takes.

Now the sharp question. If both stocks are just the queue, how would we ever detect a market that is genuinely broken? Here the test is clean. Plot vacancies against unemployment over many years and ordinary business cycles trace a downward-sloping curve: when demand is strong, vacancies are high and unemployment low, and in a slump the reverse. Movement along that curve is the cycle. What the frictional account predicts is that the curve stays put. If instead it shifts outward — more vacancies than before at the same unemployment rate — then the same raw materials are producing fewer matches, and something real has degraded: skills that no longer fit the posts, workers in the wrong regions, or a screening process that has become slower.

So the refuting observation is specific and observable. A stable curve says the coexistence is friction. A curve that has moved outward and stayed there says mismatch. Both have been seen, and confusing them is the usual error in the newspaper version of this argument.

Three refinements.

This is a different claim from wage stickiness. That argument is about the price refusing to move downward and what a cut costs a firm. The point here is stranger: even with wages perfectly flexible, both stocks would remain positive, because the obstacle is not the price but the time it takes for a specific person and a specific job to find and assess each other.

Second, "frictional" does not mean small or benign. Three months of average search is a serious cost, and the queue lengthens sharply when demand falls. An equilibrium need not be comfortable.

Third, a measurement caveat: a posted vacancy is not always a real job. Postings are cheap, and firms advertise to build candidate pools, to satisfy internal rules, or out of inertia. Some part of the vacancy stock was never available to anyone, which inflates apparent mismatch without any worker having failed at anything.

• Why the same shock produces a brief queue in one country and a decade-long one in another.

Every term the collection defines is gathered in the glossary.