Incentives Matter: Incentive Design Matters More

<>

Listener: 22 March, 2008.

<>

<>Keywords: Social Policy; Statistics;

<>

<>“Incentives matter” is a routine part of economists’ thinking. Tax a commodity, its price goes up and there is an incentive to reduce consumption. Reduce the tax, the price goes down and the incentive goes the other way.

<>

<>But many real-life circumstances are more complex, and the right incentives are more difficult to work out. Consider long-term welfare recipients. Some are going to remain there till death, but others would be better off if they had jobs, were earning their own livings and were better integrated into society. (Taxpayers would be better off, too.)

<>

<>There have been various attempts to design incentives that help beneficiaries to rejoin the workforce. Typically, someone has a good idea, it is implemented and then we have no follow-up as to whether it works or not. The Canadians have been more systematic. They randomly divided a group of welfare recipients into two. One group was left on the current regime, the other was put on a self-sufficiency project (SSP) that gave its participants generous incentives to leave welfare and get full-time work.

<>

<>Getting a job is not easy. It needs time and a rearrangement of one’s life. The SSP group was given a year to get a job, and the scheme then gave the members a full-time earnings subsidy for the next three years.

<>

<>Given the 5600 participants, each with their own special circumstances, this experiment was difficult to evaluate. It was investigated by Canadian David Card, based at the University of Berkeley, and New Zealander Dean Hyslop who works in our Treasury. Their paper, published in the prestigious journal Econometrica, has been given the even more prestigious Ragnar Frisch prize, which is awarded every two years for the best applied econometrics article published in the past five years’ issues. (Frisch was one of the two first Nobel Laureates in economics.)

<>

<>The study’s findings about the long-term effects of the SSP are salutary. In summary, they are zilch. The scheme gave some beneficiaries the incentive to get a job earlier than if there had been none. But after it came to an end, their participation in the labour force was much the same as those who had been ineligible. More of both groups were employed than when the scheme started (not everyone, of course), but those without the incentive merely took longer to get a job. The substantial public spending on the incentives to get people off the scheme did so in the short run, but gave no long-term return.

<>

<>Even more disappointing, the programme seems to have done nothing for the wages paid to the subsidised. I have had the view that on-the-job workers acquire experiences and skills that should lead them to progress up through the workforce. Getting a beneficiary into a job should trigger this improvement. Because they got their jobs earlier, this theory predicts, the subsidised should have been better paid than the unsubsidised in the long run.

<>

<>Card and Hyslop were unable to find any such effect. Instead, they found the additional employment by the SSP group was largely at or near the minimum wage, and there was no significant wage growth over time.

<>

<>Formally, the study concludes the short-term incentives had an effect, but only in the short term. There were no long-term gains. The public spending was a subsidy, but it was not an investment in getting beneficiaries into good-quality jobs in the long run.

<>

<>This does not mean we should give up trying to get suitable long-term beneficiaries back in the workforce earlier. But it seems likely that incentive schemes such as the SSP give a short-term boost but are not particularly effective in giving a long-term one. My hunch is that we probably need to look more at upgrading the beneficiaries’ skills and, in the case of parents, providing appropriate childcare facilities.

<>

<>So, “yes”, incentives matter. That’s the easy part. Designing incentives matters more. That’s the hard part.

<>

<>D Card & DR Hyslop “Estimating the Effects of a Time-limited Earnings Subsidy for Welfare-leavers”, Econometrica, Vol 73, No 6 (November 2005) 1723-1770.