Ronnie O'Sullivan yesterday gave us a nice example of the limitations of basic economics. He turned down the chance to go for a 147 break because the £10,000 prize for doing so was "too cheap."

This is an example of how incentives can backfire. Had there been no prize for doing so, I suspect O'Sullivan would have gone for the maximum, because he's a showman. Instead, he took umbrage at the meanness of the reward.

In this sense, he behaved just like the parents of children at the Israeli kindergartens described in a famous paper (pdf) by Aldo Rustichini and Uri Gneezy. They show that when the kindergartens introduced small fines for parents who collected their children late, the number of late-comers actually increased.

What happened in both cases is motivational crowding out: financial incentives can displace intrinsic ones. Small fines crowded out parents' desire to help kindergarten staff by being punctual, just as a small prize pot crowded out O'Sullivan's desire to play brilliantly. This is no mere curiosity. One reason for banks' serial criminality - PPI mis-selling. Libor rigging, forex fixing and so on - is that bonus culture has driven out any sense of professional ethics.

Daniel Pink, author of Drive, has described this crowding out as "one of the most robust findings in social science": see this paper (pdf) by Edward Deci and colleagues or this by Bowles and Reyes for more evidence.

You might object that incentives only backfire in these cases because they are too small: O'Sullivan would have gone for the 147 if the prize were bigger.

However, big incentives can also backfire. The Yerkes-Dodson law says that people can become over-motivated and so "choke" under pressure: other work by Uri Gneezy has established this experimentally. Who can forget John Terry's penalty miss in the 2008 Champions League final? (And who would want to?)

These are not the only ways in which incentives can fail. There are others:

- Performance-related pay can encourage people to hit targets at the expense of the organizations' wider goals. If teachers "teach to the test" they might fail to instil in children a love of learning, and if bosses are incentivized to hit quarterly earnings' targets they abandon longer-term strategy. For a formal model of this, see Benabou and Tirole (pdf).

- They can lead to figures being manipulated or fiddled.

- They can encourage a tunnel vision and so reduce (pdf) creativity. As Pink says:

Rewards, by their very nature, narrow our focus... Rewarded subjects often have a harder time seeing the periphery and crafting original solutions. (Drive p44-46)

- Big bonuses can signal that the task is very difficult, or even impossible. This might demotivate existing staff, or have adverse selection effects, insofar as only irrationally overconfident people will apply for the job.

- Bonuses can encourage people to free-ride on others' efforts. They can also encourage slacking, if workers give up after they have met their targets, or excessive risk-taking as workers become desperate to meet their targets.

Tim Worstall is right to say that the core concept in economics is that incentives matter. However, they can matter in unpredictable ways.

The point here is a simple one. Designing incentives - in companies, sport, public services or wherever - require careful thought. More thought, in fact, than is often given. I fear that, in the real world, "incentives" in fact serves an ideological function described by George Carlin:

Conservatives say if you don't give the rich more money, they will lose their incentive to invest. As for the poor, they tell us they've lost all incentive because we've given them too much money.