Thursday, February 10, 2011

Outcome-based venture capital financing of social policy

Imagine this social policy experiment. Crimeland prison has among the highest prisoner recidivism rate (prisoners are convicted of another crime within one year of release) in Globonia. Then World Without Crime Foundation (WWCF) comes up with a proposal that commits to lower recidivism rate by atleast 50% (after adjusting for the national average decline) over three years. It would cost $50000 to implement the program over its three years.

So WWCF offers to finance the entire upfront investments in return for being given Prison Improvement Bonds. These Bonds would have 4 year maturity and would be redeemed with returns which are based on the percentage of reductions (over and above the promised 50% minimum) achieved with recidivism. However, if the experiment fails to yield the expected minimum returns, the investors get nothing and lose their principal.

David Leonhardt points to a real-world experiment with such bonds in Britain. The British Government has initiated a program at Her Majesty’s Prison Peterborough, where 60% of the prisoners are convicted of another crime within one year of release. A nonprofit group named Social Finance has raised about $8 million from investors and is implementing, in collaboration with the prison authorities, a program to help former prisoners find work, stay healthy and the like. Some 3000 prisoners are being covered under this, which started last year.

Investors will get their money back starting in 2014 — with interest — if the recidivism rate falls at least 7.5%, relative to a control group. If the rate falls 10%, the investors will receive the sort of return that the stock market historically delivers.

They form part of the emerging category of social policy financing - social impact bonds. It has also been called payment-by-performance by the British government officials. Non-profit groups like foundations pay the initial money for a new program and also oversee it, with government approval. The government will reimburse them several years later, possibly with a bonus — but only if agreed-upon benchmarks show that the program is working. If it falls short, taxpayers owe nothing. It is hoped that success with a few initial interventions could help build a mainstream social investment market that attracts financial institutions and retail investors.

The British government is also planning to raise about £5m to develop a further package of two or three more social impact bonds. These bonds could fund programmes reducing the number of children going into care, working with children in pupil referral units, diverting persistent women offenders from prison, and developing more effective drug rehabilitation projects. Schemes to tackle long-term health problems in the community, such as diabetes and asthma, could also produce big savings in acute hospital bills.

In the US, David Leonhardt also reports that the Obama administration is set to shortly propose seven pilot programs, costing up to $100 million, along these lines. The financing mechanism will be described as pay-for-success bonds. They are set to broadly focus on increase kindergarten readiness among low-income children; increase college completion rates; reduce criminal offenses and incarceration rates among minority youth; raise the future earnings of laid-off workers; reduce hospital readmissions among patients with chronic illness etc.

What are the advantages with such social policy venture capital funds? One, most importantly, it will bring in a culture of outcome evaluation into social policy spending. Two, governments can hedge against the downside risk of the intervention failing. It will ensure much greater bang for the buck with social policy spending. Three, the hedging against downside risks also makes it easier for governments to embrace innovative programs that would otherwise have not found the light of day for risk aversion and status quo bias.

Four, Governments strapped for cash would not need to cough-up resources upfront, especially for programs whose returns are likely to show-up only after a few years. They would need to make payments only on the successful implementation of the intervention. Five, non-government agencies, non-profit and for-profit, get the platform (with all the attendant logistical support) of government agencies to experiment on their initiatives. This would marry the professional expertise and commitment of the non-government agencies with the existing government systems. Six, it will enable more effective utilization of non-government funds. Today, much of these funds are frittered away on piecemeal interventions that have little policy value.

However, there are several formidable challenges that need to be surmounted before this approach can achieve its desired objective. Which interventions to select? Which outcomes to measure, with what parameter, and how do we benchmark them? What should be the baseline and expected outcome scores? What should be the appropriate control group?

There is the possibility of external agencies being entrusted perfectly doable projects and walking away with assured returns. Ensuring the selection of parameters that, with a reasonable degree of accuracy, measures outcomes is a difficult task and one that can be very easily subverted. Both the baseline calculation and the final outcome fixation should be done with adequate care and after rigorous due diligence. The final outcome should be adjusted for changes that would have taken place even without the intervention. In the absence of clear definition of the target population, the external agency will have an incentive to cherry-pick and present a distorted picture of its achevements. Finally, the control group should be selected with appropriate care so as to be representative with the treatment.

In fact, the details of such initiatives should be arrived at only through a rigorous professional exercise carried out by competent agencies, and devoid of political and anecdotal judgements. On a note of caution, atleast for the initial set of such financing interventions, it may be better to leave out economic cost-benefit analysis (and focus on the financial benefits by way of budgetary savings) from calculations of return on investment. It may be advisable to focus on interventions (or outcome measurement parameters) where the benefits are more easily quantifiable by comparison with a relevant control group. Further, decentralized interventions are more likely to succeed, at least in the initail stages, with such financing programs.

Update 1 (22/6/2012)

Nice Fixes column on social impact bonds. It points the work of an organization One Service, which has brought together four social service groups to provide comprehensive assistance to men released after serving short-term jail sentences at Peterborough Prison near London. It informs that the idea of social impact bonds originated from various thinkers, social service experts and captains of industry in Britain who formed a group called Social Finance to build a social investment market there.

One Service raised £5 million — about $8 million — from 17 investors in Britain and the United States, and the investors will get their money back only if One Service succeeds. The bondholders are mostly charitable groups who would normally give money away. Over the next six years, the recidivism rates of men released from Peterborough will be compared to the recidivism of a matched group of prisoners elsewhere. If Peterborough’s re-conviction rates are 7.5 percent less than the control group,the British government will repay the bondholders with interest. If that threshold isn’t met, investors lose their money, which means that technically it is not a bond. The better the recividism rates, the larger the payout for investors, which is capped at the equivalent of 13 percent per year over an eight-year period.

See also this McKinsey report on Social Impact Bonds. 

No comments: