There's an important economic principle behind the debate about bankers' bonuses. It's that individual self-interest sometimes conflicts with the collective interest.

What I mean is that it might well be rational for RBS to want to pay its bankers more to attract and retain good staff*. But the benefits to RBS of employing such people come at the expense of other banks. The star M&A banker wins business at the expense of rival banks; the good trader makes profits at other banks' expense and so on.

But what's true of RBS is true for any other bank. Each must offer big money to retain staff, the result of which is that aggregate banking profits are lower than they'd otherwise be. And over time, this means banks build up less capital than they would otherwise, with the result that the banking system as a whole is riskier than it'd otherwise be.

Banks are therefore engaged in a form of arms race. Just as if two rival nations compete to have the bigger military, the upshot will be that both are impoverished without any offsetting benefit, so banks are impoverished by high salaries.

There are several other examples of arms races, some of which are described by Robert Frank and Tom Slee, for example:

- Any individual might want to work longer than his colleagues to increase his chances of promotion. But if everyone does so, they all end up working longer than they'd like (pdf), without any improvement in their prospects for advancement.

- If people spend money to keep up with the Joneses - and they do - the aggregate result can be more debt and financial fragility without any rise in well-being.

- If education has a signalling benefit (it does), then each individual will feel the need to become at least as well credentialled as his peers. The result will be increased numbers getting degrees, but with a rising mountain of debt rather than an improvement in aggregate job prospects.

What we have in these cases is a form of market failure - because what's rational for any individual is collectively self-defeating. There is, therefore, a case in principle for government action to restrain pay. Hence the EU's bonus cap**.

In this context, the government has a dilemma; policies which might maximize RBS's value - paying to get the "best people" - might merely exacerbate collectively harmful behaviour.

I don't pretend to have any easy answer to these issues. I merely note that the issue of bankers bonuses falls into a set of behaviours which contradict the invisible hand theorem: the pursuit of self-interest does not always increase the collective good.

* It's also possible that bonuses have adverse incentives for the individual bank. For the purposes of my story, I'm ignoring this.

** Whether such caps work or not is another matter. Capping bonuses might lead to higher base salaries, and capping salaries might lead to more non-cash rewards such as company jets, prostitutes and yachts.