What Actually Sets a Wage

In a perfectly competitive labour market, a wage is set by what the last worker hired adds to the firm's revenue.

The argument is tidy. A firm keeps hiring while each additional person adds more than they cost, and stops when the two are equal. Many employers competing for the same workers bid the wage up to that level, because anyone paying less loses their staff to someone paying more. Pay equals marginal product, everywhere, for everyone.

For this to work you need many employers competing for each worker, full information about what is available, and costless switching. Take a moment with that list and consider how many labour markets you have encountered that satisfy it.

The alternative is monopsony, which is the buying side's version of monopoly, and the crucial thing is that it does not require one employer.

A monopsonist faces an upward-sloping supply of labour: to hire one more person, it must offer more, and, in practice, raise everyone else's pay too, because paying a newcomer more than incumbents is unsustainable. So the cost of one extra worker exceeds that worker's wage. The profit-maximising response is to hire fewer people at a lower wage than a competitive market would produce.

What generates monopsony power is anything that makes switching hard: distance in a rural area, specialised skills with few buyers, non-compete clauses, visa sponsorship tied to an employer, the cost and time of searching while working full-time. Recent empirical work suggests it is widespread rather than exotic, and that measured labour market concentration in many local markets would trigger antitrust scrutiny if it were a product market.

There is a second departure from the competitive model, and it runs the other way.

In 1914 Henry Ford roughly doubled wages to five dollars a day. The competitive model says this is throwing away money. Ford's turnover had been running at over 300% a year (the plant was hiring several people for every position annually) and after the increase it collapsed, along with absenteeism, while productivity rose enough that the policy appears to have paid for itself.

This is an efficiency wage: paying above market to raise productivity, reduce turnover, attract better applicants and make dismissal a real threat. It explains something the competitive model struggles with, which is why observably identical firms pay different amounts for identical work, indefinitely, without the low payers being competed out.

And that persistence is not a curiosity. Matched employer-employee datasets, linking individuals to the firms they work for, consistently find large firm effects on pay. The same worker moving between firms sees their wage change substantially; observably similar people doing similar jobs earn very different amounts depending on where they are.

Which firm you are in turns out to be a major determinant of what you are paid. In a competitive model this cannot happen, and it happens everywhere.

Then there is the fact any account of modern wages has to explain. In the United States, productivity and median compensation tracked closely until roughly the early 1970s and have diverged substantially since, with output per hour rising much faster than the typical worker's pay. Other rich countries show the same divergence, smaller.

Part of the gap is measurement: output is deflated by one price index and consumption by another, and non-wage benefits like health insurance are frequently excluded from the wage series while counting as compensation. Serious analysts disagree about how much of the gap survives careful measurement. What is not seriously disputed is that labour's share of national income fell, and that the median worker's share of that share fell too.

Finally, a phrase worth interrogating every time it appears.

A hospital says it cannot recruit nurses because there is a shortage. In a competitive market, a persistent shortage is impossible: the price rises until the market clears. A shortage that lasts for years is therefore not a fact about the supply of people but a fact about the wage being offered, or about the administered pay scale that prevents it from moving.

This is not an argument that any particular employer is being unreasonable. Pay scales in health and education are set nationally and constrained by budgets, and a hospital genuinely cannot unilaterally raise them. It is an argument about what the word "shortage" is doing in that sentence, which is concealing a price that is not allowed to adjust behind a description of the world.