It's fitting that Mervyn King should have been a director at Aston Villa - because that job, like his previous one, has given him a close-up view of the failure of mainstream economics.

I'm referring to the idea that wages equal marginal product. Aston Villa's wage bill this season was higher than Leicester's. This difference is not reflected in their marginal product. In fact, Villa owner Randy Lerner is sitting on a loss of over £200m.

This highlights an under-appreciated distinction: people are often not paid their actual marginal product, but their expected marginal product. Villa paid Joleon Lescott and a ragbag of Carlos Kickaballs £50,000 a week because they expected them to save the club from relegation. Such expectations proved over-optimistic.

This is not an idiosyncratic example. Across whole professions, wages can be inflated because expected marginal product exceeds actual marginal product. Fund managers, for example, are paid millions for mediocre performance because gullible investors over-estimate their ability to beat the market. I suspect the same is true of corporate CEOs: managerialist ideology causes remuneration committees to over-estimate the extent to which an heroic leader can improve the company, and so pay too much. If bosses were paid their actual marginal product, Fred Goodwin and Dick Fuld would have had salaries of minus billions of pounds. They didn't.

I'd add that there are at least three other reasons why pay can deviate from marginal product, especially at the top end.

First, value is often determined not merely by individual's production, but by team production, which makes it difficult to identify an individual's marginal product. As Lars Syll says (pdf):

It is impossible to separate out what is the marginal contribution of any factor of production. The hypothetical ceteris paribus addition of only one factor in a production process is often heard of in textbooks, but never seen in reality.

For example, the same manager, for example, might or might not add value depending (pdf) upon whether his skills are a good match or not with the organizations.

This can cause pay to exceed marginal product because the fundamental attribution error leads hirers to over-estimate individuals' contributions and under-estimate organizational factors - as happened, for example, when ITV hired Adrian Chiles and Christine Bleakley on big money in the mistaken hope that they would attract viewers.

Secondly. wages can exceed marginal product in efficiency wage situations, where workers must be bribed not to steal the firms' assets. This explains why bankers are paid so much. I suspect the same applies to CEOs.

Thirdly, power matters. Dani Rodrik has shown that even controlling for productivity workers are better paid in democracies (pdf) because these are associated with stronger workers' bargaining power. In the same spirit, economists at the IMF show that CEOs pay is lower when trades unions are stronger.

All of this is to vindicate a point made by Joe Stiglitz - that evidence for the validity of marginal product theory, especially at the top end, "remains thin".

Now, you might object that some of the examples I've given are of disequilibrium: Fred Goodwin's job didn't last long, and Villa players' prospects aren't great. But all theories are true if you ignore the exceptions. Life is lived in disequilibrium, and a few years of egregious deviations from marginal product can generate big inequalities.

It could be that marginal product theory - just like simple-minded talk of incentives - is as much ideology as science.