Are firms efficient institutions for responding to uncertainty, as Coase thought? Or are they, as Marxists believe, means whereby capitalists exploit workers? A new paper by Ernst Fehr and colleagues provides experimental evidence.

They split subjects into principals and agents, and asked them to try to form either employment contracts or sales contracts. Under sales contracts, the agent performed a set task. Under employment contracts, the principal could assign the agent to one of three possible tasks; one of which paid well in one state of the work, one of which paid well in another, and a third which was inefficient in aggregate but very profitable for the principal.

This set-up highlights the difference between Marxian and Coasian theories of the firm. For Coasians, employment contracts are better because they allow the principal to respond to uncertainty by assigning workers to the more efficient task when that task cannot be identified in advance: as Coase said (pdf), "it seems improbable that a firm would emerge without rthe existence of uncertainty." But for Marxists, the danger is that they allow workers to be exploited - to be assigned to the third task.

Fehr and colleagues found that, in one-shot encounters where employment contracts were struck, 51% of principals exploited agents. "The Marxian idea that power can be used for exploitation is real" they conclude.

Why only 51%? It's because there's a norm of fairness which stops some principals exploiting workers. This norm is a two-edged sword. On the one hand, it promotes efficiency, as it encourages agents to enter into the more flexible employment contract rather than sales contracts in the belief they'll be treated fairly. On the other hand, though, this belief might prove mistaken - and so the fairness norm actually facilitates exploitation.

However, in repeated encounters, the prevalence of exploitation dropped to 21%. This is because employers wanted to build a reputation for fairness which they could use to encourage workers to stick to employment contracts.

Simple as it is, this gives us a framework to pose the question: under what conditions are we likely to have Coasian rather than Marxian firms?

One is where there's a strong norm of fairness. You can read the campaign for a living wage as an effort to build such a norm.

Another is where firms have a desire for a reputation as a "good" employer. This is more likely to be the case under conditions of near-full employment, where they have to compete for for workers.

A third is the existence of strong unions. Fehr and colleagues say:

To the extent to which reputational forces alone are insufficient for solving the employers' moral hazard problem, labor unions and labor legislation can play an efficiency enhancing role by constraining the employers' ability to assign the workers inefficient tasks.

This corroborates my suspicion that strong unions can be good for an economy.

There is, however, a fourth possibility - for workers to have an outside option such as welfare benefits that allow them to reject exploitative contracts.

The fact that many of capitalism's supporters reject this fourth course makes me suspect that what they are interested in is not so much efficient Coasian firms as the power of capital to exploit workers.