This week, The New Zealand Initiative released a report assessing social impact bonds in a New Zealand context. The report identifies both opportunities and challenges. It makes a number of recommendations for realising their potential.

Social impact bonds aim to improve social outcomes by tapping into private sector commercial, philanthropic and social service delivery expertise to a greater degree than in the past.

First launched in the UK in 2010, they have taken off internationally, with over 100 launched to date, including two in New South Wales. New Zealand's Ministry of Health is working on a pilot bond for New Zealand.

In their starkest form, government pays only for success; the costs of failure fall entirely on private parties. Cash-strapped social service providers have to look to socially-inclined investors, foundations and/or philanthropists to cover their social service delivery costs in the meantime. All capital and hoped-for investor returns are lost if the better outcomes are not achieved.

The transfer of risks from taxpayers to private parties obviously sharpens the incentives for providers. Less obviously, it should also encourage delivery innovation. Those whose capital and return is at risk have a strong incentive to ensure that providers are not over-looking promising innovative possibilities.

Internationally targeted outcomes to date encompass public safety and recidivism, rough sleeping, chronic homelessness, and better outcomes for at-risk children and their families. One bond in California targets improved outcomes for children with asthma.

The interim results from the UK's first social impact bond are encouraging. An 8.4% reduction has been achieved in rates of reoffending amongst a cohort of recently-related short-term prison inmates, compared with a control group.

Despite their promise (and considerable hype), their success is yet to be demonstrated. The UK's 8.4% reduction was below the 10% target, so no interim payment for success was triggered. But a payment for success in 2016 is plausible. To the best of our knowledge, to date no social impact bond has triggered a success payment.

The focus on improved outcomes represents a major change from long-standing public sector funding for the delivery of services. What outcomes should be targeted? How can they best be measured? How much should be paid for success? How would this be distributed across affected votes?  Whom should be contracted? Such questions are challenging.

One recommendation is that a special unit headed by the Treasury be set up to coordinate government agencies and provide a one-shop stop for potential providers and intermediaries.