Paying above the market

Why would an employer pay more than the wage needed to fill the post?

Suppose you need a warehouse picker, and at the going rate the queue at your door is already long enough to fill the post tomorrow. Paying a quarter more would look like money thrown away for nothing you could not have had for less. Yet firms do it deliberately, and not only the ones flush with cash. What could the extra wage be buying that the market rate does not already buy?

Start by asking what the contract actually specifies. It buys hours — attendance, a body at a station. What the firm wants is effort, and effort is not written into it, because nobody can write down every judgement a picker makes across a shift. So the firm has bought one thing and needs another, and has to close the gap somehow.

The obvious instrument is dismissal. But look at what dismissal costs a worker paid exactly the market rate. He walks out and, by construction, an identical post at an identical wage is waiting. The threat is empty. A firm paying the going rate has, in effect, no penalty available to it at all.

Now raise the wage. The post carries something the next post does not — call it a rent, the surplus over what this worker could get elsewhere. Dismissal now costs him that stream, and each time he considers cutting a corner he is weighing the effort saved against the chance of being caught multiplied by what he would forfeit. The premium is not generosity. It is the manufacture of a penalty where none existed.

Notice the second thing the premium does without being asked for. If leaving is expensive, fewer people leave — and turnover is expensive to the firm too, in recruiting, in training, in the months a new hire takes to reach full speed. And a third: once word gets out that your posts pay above the rate, the queue at the door is not only longer but differently composed, because people with good outside options now bother to apply. Effort, retention, selection — one instrument, three effects.

Then push it out to the whole economy, which is where it turns strange. If the premium works, every firm wants one. But if every firm pays above the going rate, there is no going rate left to be above; the average simply rises. So what restores the penalty? Shapiro and Stiglitz's answer, in 1984, is that unemployment does. Wages settle above the level that would clear the market, more people want jobs at that wage than there are jobs, and the cost of dismissal becomes the expected wait before the next post. Involuntary unemployment stops being an embarrassment for the model and becomes the thing that makes it work.

Ford's five-dollar day in 1914 is the case everyone reaches for — roughly double the prevailing rate, announced by a firm whose assembly line had made the work punishing and whose turnover was enormous. Which of the three effects did it buy? Turnover collapsed, absenteeism fell, and the queue outside the plant became famous. That much is well documented. Whether the men on the line also worked harder per hour is far more difficult to establish, and the honest summary is that the retention channel is the best-supported part of the story.

"Efficiency wage" names a claim about cost per unit of effort, not about kindness: the wage per hour rises while the labour cost of getting a given job done falls. Where that is not true, the firm is simply overpaying, and no amount of theory rescues it.

The same claim tells you where the mechanism should not appear. Where output is countable and attributable to one person — fruit picked, garments sewn — the firm can pay for output directly, and piece rates do the job more cheaply than a standing premium. Efficiency wages belong to work whose quality is visible only in aggregate, and only later.

Two honest complications. The first is empirical. High-wage firms are usually also high-productivity, high-profit firms, and a firm sharing its rents with workers out of bargaining pressure or a fairness norm looks in the data almost exactly like a firm buying effort. Separating efficiency wages from rent-sharing is genuinely hard, and much of the evidence offered for one is compatible with the other. The second concerns the gift-exchange version of the story — that a wage read as generous is repaid with effort out of reciprocity. It is robust in laboratory experiments, but field experiments have generally found the effect small and short-lived, fading within hours or days. Reciprocity can be real without being durable enough to build a wage policy on.

• Why firms use piece rates in some jobs and standing premiums in others, and what that reveals about what is measurable.
• Whether minimum wages destroy jobs less than expected partly because they force firms onto an efficiency-wage margin they were already near.

Every term the collection defines is gathered in the glossary.