The textbook story that wages equal marginal productivity is one of the most persistent simplifications in labor economics. It survived twentieth-century empirical challenges because the implications were politically convenient — if wages reflect productivity, then wage distributions reflect productivity distributions, and there is no distributional case for redistribution beyond what the market produces. The twenty-first-century empirical record has made the simplification much harder to defend.

What the Framework Says

In the perfectly competitive labor market of standard textbooks, firms hire workers up to the point where the worker's marginal product equals the worker's wage. Workers can move freely between firms; firms compete for workers. Any wage above marginal product prices out the firm; any wage below marginal product loses the worker to a competitor. The equilibrium wage is the marginal product, and the wage distribution is the marginal-product distribution.

This framework has the virtue of being a simple model that predicts certain things and is robustly false in others. Where it predicts well: rapid wage adjustment in high-skill specialized markets where workers really are mobile and firms really do compete (top software engineers, top finance professionals, top medical specialists). Where it fails: most of the labor market.

The Bargaining-Power Frame

The alternative framework, developed across the 20th and 21st centuries by labor economists from John Dunlop through David Card through Suresh Naidu, treats the wage as a bargained outcome between worker and firm, with the outcome depending on the parties' respective bargaining positions. The "competitive" wage is the particular case where worker outside options are strong enough that the firm cannot deviate below productivity. Most labor markets do not satisfy this condition.

The empirical work on this is substantial. Larry Katz and Lawrence Summers's 1989 "industry wage premiums" paper documented that similar workers earn very different wages depending on which industry they work in — the auto industry pays substantially more than retail for otherwise identical workers, even controlling for skill and credentials. This is not consistent with the competitive framework but is what the bargaining-power framework predicts.

The Empirical Decoupling

The most damaging empirical finding for the productivity story is the 1979-2024 decoupling of median wages from aggregate productivity. Productivity has roughly doubled; median compensation has risen ~10%. If wages equal productivity, this is impossible — productivity gains should flow to workers in the same proportion. The actual pattern is consistent with a sustained shift in bargaining power toward capital, which is what the bargaining-power framework predicts as the consequence of declining unionization, rising monopsony, and changes in industry structure.

The Specific Mechanisms

Several mechanisms explain how wages fall below productivity in practice:

  • Monopsony power in local labor markets: workers face limited outside options.
  • Non-compete agreements: legally restrict workers from moving.
  • Information asymmetries: workers don't know what comparable jobs pay.
  • Geographic friction: workers can't relocate cheaply.
  • Search costs: finding the next job takes time and effort that workers can't easily afford.
  • Switching costs: tenure-based benefits, retirement vesting, healthcare continuity all penalize movement.
  • Risk aversion: workers facing income volatility take the current job rather than searching for a better one.

Each of these mechanisms creates a wedge between productivity and wage. The cumulative wedge is what the bargaining-power framework calls "labor's share of the surplus" — the share that flows to workers out of the value they help produce.

The Skill-Biased Technical Change Argument

One mainstream defense of the productivity story is "skill-biased technical change". The argument is that technology has raised the productivity of high-skill workers more than low-skill workers, so the rising wage premium for skills reflects rising productivity differentials. This framework has substantial empirical support — the college wage premium has risen along with educational attainment — but it cannot fully explain the broader decoupling, because median wages have lagged productivity even within skill levels.

David Autor's later work has refined the framework into a "task-biased" model where automation substitutes for routine tasks while complementing non-routine tasks. This handles more of the empirical evidence but still does not explain the magnitude of the decoupling between median wages and median productivity.

The Rent-Sharing Evidence

One of the most direct empirical tests of the bargaining-power framework is "rent-sharing" — do workers in more-profitable firms earn higher wages than otherwise-identical workers in less-profitable firms? The competitive framework predicts no — workers should earn their productivity regardless of which firm employs them. The bargaining-power framework predicts yes — workers in firms with more surplus to share will capture some of it.

The empirical record clearly favors the bargaining-power framework. Rent-sharing is documented across industries, countries, and time periods. Workers in higher-rent firms earn higher wages, and the magnitude of the rent-sharing has declined over time, suggesting that workers' ability to capture firm rents has eroded as unionization declined.

The Honest Reading

The marginal-productivity theory of wages is a useful idealization that describes a narrow corner of actual labor markets. The bargaining-power framework describes most of the rest, and the empirical record over the last forty years has progressively favored it over the productivity story. The policy implications are substantial: if wages reflect bargaining power rather than productivity, then policy can move them — through minimum wages, through unionization support, through antitrust enforcement against monopsony, through non-compete restrictions. The productivity framework treats wage distribution as exogenous to policy; the bargaining-power framework treats it as a choice. Which framework you accept is partly empirical and partly political, but the empirical case has been steadily strengthening for the bargaining-power view.

What the Shift Implies for Policy

If wages reflect bargaining power rather than productivity, then the distributional outcomes of the labor market are policy-sensitive rather than fixed. Minimum-wage policy can move them. Union law can move them. Non-compete policy can move them. Antitrust enforcement against monopsony can move them. The productivity framework treats wage distribution as exogenous to policy; the bargaining-power framework treats it as a choice. Which framework you accept shapes the entire policy agenda for labor markets. The empirical case has been steadily strengthening for the bargaining-power view, but the policy implications have been only partially adopted in institutional practice.