In his first Reith Lecture this year, Markets and Morals, Professor Michael Sandel discussed a couple of related examples in which economic incentives do not produce the intended system effect.
Firstly, an Israeli nursery. Parents would arrive late to pick up their children, causing cost and inconvenience to the staff. So the nursery introduced a system of fines, hoping to penalize these parents and encourage prompt pickup. As it turned out, the frequency of late pickup actually increased. It turned out that parents regarded this merely as a charge for extra time, and no longer felt guilty about the burden on the staff.
Secondly blood donation. In the UK, blood donation is voluntary; in the US, blood donors are paid, in order to encourage more regular donation. As it turns out, the US system produces more uncertainty and greater quality problems. Apparently blood donors no longer feel a sense of obligation in a market economy - blood is regarded as a commodity rather than as a gift.
As Professor Sandel puts it: Norms matter. This principle is not surprising to those familiar with Donella Meadows' idea of leverage points (PDF). Her paper identifies different degrees of effectiveness in intervening in complex systems. Economic incentives and punishments (rules of the system) rank 5th in her scheme, but the underlying value system (mindset or paradigm) ranks 2nd.