I've suggested that potentially the biggest cost of the Libor fiddle is that it might depress trust in the economy and hence long-run economic activity. Coincidentally, a new paper half-corroborates my fear. It says we can be "locked into cycles in which trust and trustworthiness emerge and collapse again and again."

Think of a trust game. In this, one person - the investor - decides whether to give money to another, the trustee. If he does give money, the experimenter triples the sum. The trustee then chooses whether to give any money back.

Experiments show that investors often do hand money over, and trustees hand some back. This contradicts the ideas that people are rational selfish maximizers, as this predicts that the trustee will keep all the money for himself.

But what if the game is repeated, with investors choosing trustees?

Imagine that some trustees then behave badly and don't return money. Investors who trust will then lose money. In evolutionary terms, they'll be selected against and will die out. Trust will then decline. And it might not be rebuilt merely by state regulation (pdf) of trustees. If it declines to zero, no investor will hand anything over to trustees, so trustees will end up with zero return as well.

But when trust is low, the trustee who can somehow signal his trustworthiness will attract more investors than his untrustworthy competitors. As an example, think of those 19th century food manufacturers such as Bird's, Colman's or Cadbury's who invested in brand reputation to signal that their products were less adulterated than their rivals'.

Trustworthiness will then be selected for, and so will spread. And as this happens, investors who trust will earn bigger returns, and so selection instead of favouring the untrusting will now favour the trusting. Both trust and trustworthiness will then rise. But when trust is very high and investors are handing over big money, trustees will have stronger incentives to run off with the money.

So, we get cycles of rising and falling trust and trustworthiness.

There are several implications here:

1. Homo economicus - the selfish rational maximizer - does not necessarily create a prosperous society. This is because homo economicus might have an incentive to run off with the money. A society of such people might therefore be a low trust one - and therefore poor.

2. Both trust and trustworthiness have positive externalities; they benefit others, not merely ourselves. This suggests they might be under-supplied by homo economicus.

3. There need not be stable equilibria of high or low trust. When trust is low, trustees have an incentive to be trustworthy. And when it is high, they have an incentive to defect. This gives us cycles.

4. Matthew's contention that "everything is leveraged" - small changes can have big and longlasting effects - is valid. A smallish loss of trustworthiness or trust can set us on the downward leg of a cycle. Which is why banks' misbehaviour is worrying.