Without the right incentives, social benefit bonds can leave us in a bind
The Conversation, by Paul Oslington
The NSW Treasury’s current trial of social benefit bonds is attracting a lot of attention, including a session devoted to bonds and other social enterprise innovations at the Economic Society of Australia’s recent Australian Economic Forum in Sydney.
As Dr Caralee McLiesh from Treasury pointed out, they are not really bonds. The government is not a paying fixed return, nor is it guaranteeing the principal. Instead, the government agrees to pay investors conditional on agreed outcomes being achieved by the social service organisation with the funds provided by the investors.
Essentially, they are outcome-based contracts – like those through which an increasing proportion of Australian social services are delivered – plus a financing mechanism that attempts to overcome the restrictions social service organisations face in raising capital through the market.
We certainly need new models for the partnership between governments and social service organisations. The most significant experiment so far in this area – the Commonwealth’s Job Network (or JobSearch) – is increasingly looking like the old and unlamented Commonwealth Employment Service, with various church and other logos on the door rather than governments logos. The rhetoric of outcome-based payments, which free up innovative providers to do what is needed, has been hampered by the risk aversion and micromanagement of government. The current system only keeps working because of the goodwill of the providers, especially the church-related providers, but serious doubts remain about its sustainability.
Are social benefit bonds a mechanism for this new partnership that is needed? Perhaps we must not get carried away by the hype, as they are only applicable to a relatively small subset of social services and it is early days yet. Social bonds share many of the advantages and problems of existing outcome-based contracts.
There a number of ways social bonds could turn out badly for government, and those that government serves:
Social bonds will not go anywhere unless there are robust mechanisms for evaluating outcomes for which the government pays. They are likely to be a disaster if an overenthusiastic government proceeds without working this out properly.
We know from the previous experience of outcome-based contracts that measuring social outcomes is nontrivial. Even defining outcomes is difficult, with a tension between fundamental goods that might be (or might not be) widely agreed upon by all involved in the area, and the need to specify outcomes in a way that is measureable.
Another difficulty is specifying the appropriate counterfactual. If the government pays a provider for all outcomes, then it will be overpaying the provider for those who would have got a job anyway, and creating perverse incentives for the provider. An example in employment is ignoring (or “parking”) the most disadvantaged workers who had little chance of getting a job and making money out of those who can be placed with little or no effort.
Determining the probability of an unemployed individual obtaining a job is very hard, and economists’ estimates usually rely on averaging and results from randomised trials, which can prove problematic. Another difficulty is obtaining reliable data on the chosen measures in an environment where the financial and political stakes are high.
The thinness of the social evaluation market is more of a problem for implementing social bonds than the thinness of the finance market, which most of the discussion has so far focused on. There are not enough people with high-level quantitative economic skills and knowledge of social services that would be needed if outcome-based contracting or social bonds are to become widespread.
As well as the shortage of people with these skills, they and the consulting firms they are likely to be working for have exactly the same twisted incentives as the ratings agencies whose actions contributed to the mispricing of complex financial assets during the GFC. If the investor or social service provider pays for the evaluation, the incentive is for the consultant to overstate outcomes. If the government pays, outcomes are likely to be understated.
Social benefit bonds as described above result in the investor taking on all the risk. But to sell the bonds, the government would have to take some of the risk, in the form of guaranteed payments to the investor. However as the government takes on more risk to sell the bonds, this changes the incentive structure: it weakens the incentives for good performance, and increases the likelihood that the government will end up paying for dud programs.
If this ends up happening, the government will end up paying both the substantial transaction costs associated with an issue of social bonds and paying for ineffective social programs.
Bonanza for lawyers and consultants
These groups will do very well out of social bonds regardless of whether they work well or the social programs are effective. It would be ironic if a scheme designed to reward performance instead of activity in social services ends up most benefiting lawyers and consultants who make their money out of activity.
Cost-shifting within government
As social bonds become more widespread, I can imagine a scenario where the Commonwealth is backing social bonds for a provider to transfer costs from the Commonwealth budget to the state budget, while another provider holds a bond that transfers costs the other way. Both bond issues are profitable for the governments concerned, the providers and investors, but no net social benefit is achieved the same thing could happen with bonds backed by different budget units at the same level of government. If bonds become widespread it will be a difficult problem to avoid, especially as there are incentives for all involved to overlook the problem.
Turning social benefits into private benefits
A potentially more damaging transfer is involved where providers achieve a specified outcome in a way that imposes costs on others in society. For instance, in employment services, the incentives to place workers can lead providers to pay wage subsidies which wipe out apprenticeship and training opportunities to young workers, and hence destroy social value in the long term. This is exactly the privatisation of gains and socialisation of losses that was rightly criticised during the GFC. Government employees, in my view, are at particularly risk of losing sight of social value, designing bond programs that yield financial benefits for their department, which under current public service reward systems are likely to yield private career benefits, but that destroy harder to measure social value.
A common thread is the lack of expertise in designing incentives, and the lack of appreciation of differences between private and social value. Both are taught in good, advanced economics programs, and more people who are involved in social services need this kind of training.
We certainly need new models for engagement of government with social service organisations. Current arrangements are unsustainable and the risk of doing nothing is great. Social bonds may be part of the answer, but we need to get the incentives right to avoid these bad endings.