The bosses' pay con-trick is getting worse. The pay of average FTSE 100 directors rose 49% last year, with the average CEO getting ''3.86m.

I say this is a con for several reasons.

1. The link between pay and firm performance is asymmetric (pdf). In good times, bosses' pay rises a lot, but in bad times, it doesn't fall so far. In fact, the biggest influence on bosses' pay is not so much company performance as simply the size of the firm.

2. Big bonuses and performance-related pay are not technically necessary to elicit performance. There's plenty of evidence that bonuses are sometimes ineffective at improving effort, sometimes actually counter-productive, and sometimes inferior to fines .

3. The claim that bosses must be paid a lot because their pay is set in a global market is silly. CEOs are paid more in the US than UK, and yet few Brits have become CEOs of US firms; one of the very few exceptions was Martin Sullivan who was CEO of, ahem, AIG. Holding down UK bosses' pay is unlikely to lead to a mass emigration.

4. It's not even clear that the average boss has much effect upon corporate performance. Take three facts:

- We know from football that changing coaches has no effect upon team performance. If bosses of organizations as small as 11 men can't turn around performance, what hope do they have for larger organizations? As Warren Buffett said, when a boss with a good reputation takes over a business with a bad one, it is the business that keeps its reputation.

- Jonathan Haskel and colleagues have found that 80-90% of total factor productivity growth (pdf) comes from plants exiting and entering the industry, rather than from internal productivity growth. This suggests that bosses do less than widely thought to improve organizations' efficiency.

- The death rate of companies is not only high, but statistically distributed (pdf) in a similar way to that of the extinction of species. This suggests that bosses can no more foresee (pdf) or prevent the demise of their firms than species can foresee or prevent their own extinction. Which suggests that bosses know less than we suppose.

In saying all this, I'm not taking a Marxist view. It was, remember, Hayek who told us that central management was flawed because of the impossibility of aggregating dispersed data.

5. The idea that bosses matter, and must be paid accordingly, plays upon our cognitive biases, such as:

- Fundamental attribution error. We assume that individual must be responsible for outcomes, and downplay the role of environmental or situational factors. So if a firm does well, we attribute its success to management rather than a lack of competition or macroeeconomic forces or just luck.

- Outcome bias. We believe that if something happened, it was inevitable. When a company succeeds, therefore, we look for causes - and the fundamental attribution error directs us to management. We forget that success might be due to luck, or to factors outside management's control.

- Salience effects. I'll concede that a handful of managers - Steve Jobs, Jack Welch, Arsene Wenger - can achieve great things. But these are exceptions. Looking at them and inferring that the average manager has great powers is like looking at Kate Moss and inferring that obesity doesn't exist.

Given all this, you might wonder what the real reason is for bosses' high pay. Simple. Power. Bosses , generally, might not have the power to create super-efficient high-performing firms, but they do have the power to extract rents from shareholders and workers. Like some City traders, they must, in effect, be bribed not to plunder the firm's assets. From the point of view of shareholders, the small theft that is a multi-million pay-packet is better than the large theft of wilful mismanagement.