Information markets and Social Policy Bonds

Robert W Hahn is co-founder and executive director of the American Enterprise Institute-Brookings Joint Center for Regulatory Studies. His paper, Using Information Markets to Improve Policy, briefly refers to my work on Social Policy Bonds. I respond to his comments.

Horesh's proposal differs from ours in two other ways. First, it does not pay for incremental progress toward a specific goal. That is, the bond would only pay the fixed amount if a specific objective were achieved.

I have three main answers:

First, let's say our targeted objective is to reduce unemployment from 10 per cent to 3 per cent. We issue Social Policy Bonds that become redeemable for a fixed sum when unemployment is sustained at 3 per cent or below for, say, two years. This is a remote objective, but would that put people off buying the bonds even if they did not particularly want to make a long-term investment that might take many years until redemption? Not necessarily, because if they undertake unemployment-reduction projects now, the market value of their bonds will rise, even if the effect of their activities is only to reduce unemployment to, say, 8 per cent. The market will most likely interpret such a reduction as a reduction in the cost of getting the percentage down to the required 3 per cent target. The market value of the bonds will therefore rise, and the original purchasers can sell and realise a profit, once they have carried out their unemployment reduction projects.

Governments routinely issue bonds redeemable 10, 20 or more years into the future, but very few people who buy them at issue intend to hold them all the time until redemption. They can profit from shorter-term variations in bond prices.

Second, the redemption terms of the bonds could be formulated to explicitly reward incremental progress. In the interests of simplicity I have not gone into detail about this, but time-based provisos could easily be added to the redemption terms of any Social Policy Bond issue. For example, bonds could be issued that targeted a reduction of unemployment of 10 to 3 per cent. They might be redeemable for $1000 each, but if, say, the 3 per cent target were achieved within five years (and sustained for a further two years) the issuers could make them redeemable for $1500. If the target is not achieved for 30 years, the redemption value could collapse to zero. Of course, bond issuers could post more sophisticated time-based redemption conditions.

Third, and more trivially, bond issues could target intermediate targets themselves. So bonds could be issued that targeted a sustained drop in unemployment to 6 per cent rather than 3 per cent. Or two or more unemployment rates could be targeted by two or more bond issues.

Second, in the absence of costless bargaining and coordination among firms holding Social Policy Bonds, individual firms will not have adequate incentives to help meet the performance objective under his proposal because of problems with free riding.

I think this would be more accurate if instead of 'will not have adequate incentives…' it read 'would have reduced incentives…'. Even if bargaining and coordination were costly firms would, under certain conditions, help meet targeted objectives. Certainly, the direct benefits of bondholding would not always be apportioned strictly corresponding to bond holdings. But:
Firms would be looking at how worthwhile it is to themselves to undertake activities that would raise the value of their bonds. They might not be deterred from carrying out these activities just because other firms might benefit disproportionately more. Company executives and workers respond to all sorts of incentives, and are not deterred from maximum performance just because large shareholders will receive much greater financial rewards.

Firms could leverage their actual bondholding by buying call options or futures for the bonds, thereby maximising the returns from their target-achieving activities.

However, perhaps more important is that bargaining and coordination in the age of the internet are, in fact, very close to costless.

[Horesh] does not address how to obtain information on the costs and benefits of a policy prior to implementation.

This is true for new policies. But issuers of Social Policy Bonds will easily limit the maximum cost of their policy because they determine the redemption value of each bond, and the total number of bonds in circulation. And once a bond has been issued that, say, targets the reduction of unemployment from 10 to 8 per cent, then the market value of the bonds will generate and reveal masses of information about the costs of this policy, costs and benefits of particular unemployment-reduction projects and, importantly, costs of reducing unemployment still further. (I go into this in detail in my books.)

Technorati tag: economics