Getting more "skin in the game" How latest evidence around social bonds suggests we can get more players involved in reducing re-offending, while reducing risk to the Crown

Steven Youngblood
Principal Policy Adviser, Department of Corrections

Author biography:
Steven Youngblood is a principal policy adviser at the Department of Corrections. He has also “had skin in the game” at the Ministry of Health, the New Zealand Defence Force and the Ministry of Education.

“When decision-makers have skin in the game – when they share in the costs and benefits of their decisions that might affect others – they are more likely to make prudent decisions than in cases where decision-makers can impose costs on others. [Skin in the game] is not just a useful policy concept but a moral imperative.” – Nassim Taleb (2013)


One of the challenges government faces when it is trying something new, particularly for its most challenging social problems, is that it carries all the risk – political and financial – if a project fails. This can be expensive, and can limit government’s willingness to take risks as well as its ability to innovate.

Social bonds, also known as “social impact bonds”, are an alternative financing and contracting tool which enable the non-government and philanthropic sector to become more involved in improving outcomes in the social sector. Social bonds can be used to tackle some of government’s most challenging problems (such as high rates of offending in particular areas of the country), and to shift some of the risk away from the public sector by introducing other forms of financing. Social bonds also support private sector social initiatives with powerful financial incentives. These incentives mean investors, not government, hold the financial risk – while ensuring that more players have “skin in the game”.

What are social bonds?

In essence, social bonds involve contracting private providers, such as NGOs, to deliver improved outcomes for a particular cohort. This allows governments to pay for social outcomes on a no-win / no-fee basis, while providing scope for innovation in service delivery.

This approach is different to the standard practice of contracting for specific activity or outputs, as a focus on outcomes allows providers to try new ways of achieving results – rather than being tied down to specific activities. With social bonds, the return to investors is contingent on the success of the programme in realising some level of saving to the government. *

Many countries are considering the use of social bonds to make headway into difficult social problems. Social bonds are currently underway in the United Kingdom, United States, South America, the Netherlands, South Africa, Belgium and Australia, and are being actively explored in Canada, Ireland and Israel. (New Zealand Ministry of Health, 2015). While there are around 100 social bond concepts being considered for development around the world, only a handful of social bond contracts have actually been implemented.

Australia is adopting social bonds at pace, with a range of initiatives currently being considered across the social sector. For example, Queensland is exploring how to use social impact bonds to address homelessness and re-offending. In addition, the Australian financial and NGO sectors are increasingly creating “impact investment” funds, which will be used to invest in social bonds within Australia.

Social bonds in New Zealand

In 2013, the New Zealand Government announced a series of social bonds trials led by the Ministry of Health (New Zealand Ministry of Health, 2013). The four areas being considered for social bonds include:

  1. reduction in youth offending in areas of need
  2. reduction in adult re-offending
  3. management of chronic illness
  4. support for people with mental illness to secure and sustain employment.

The Ministry is currently progressing the initiatives relating to mental illness and youth offending. These initiatives are in the feasibility phase of development, and are yet to be considered by Cabinet (New Zealand Ministry of Health, 2015).

How they work

The key characteristics of social bonds are (Chambers, 2015):

  1. The government contracts a broker to deliver a social outcome for a defined population / cohort, within a defined period of time. Payment rates (and conditions) are agreed, and funded from the savings to government of achieving these outcomes (such as savings from reduced re-imprisonment).**
  2. The broker issues bonds to investors to fund activity to achieve the contracted outcome. Investors could include any non-governmental or private sector organisation, such as banks, philanthropic organisations and / or charities.
  3. The broker contracts providers to deliver services to the target population.
  4. If the target outcomes are met, the government pays the broker their fee, providing a return to investors including interest. If the agreed outcomes are not achieved, government is not required to pay anything.

Shifting the risk away from government and towards investors presents a new level of risk for investors in their interactions with government, as investors now stand to lose their investment if outcomes are not met. This level of risk would therefore be reflected in the expected rate of return should the contract succeed.

Good data matters and is starting to become available

It is difficult to overstate the importance of good data to the success of social bonds. Thanks to initiatives such as the “investment approach to justice”, agencies now have the ability to identify particular individuals and their families who, due to their circumstances, are likely to have multiple and complex interactions with government. This approach to data is helping us to identify those individuals who are most likely to require targeted effort to reduce their chances of being a “frequent flyer” in the Corrections system.

For example, we know there are 3,834 young people in Northland between 15 and 24 years old who are at risk of poor long-term outcomes.***  We also know that Corrections is managing 525 of these young people, including 63 who are in prison. This level of detail allows us to identify those individuals under Corrections management whose risk factors increase their chances of re-offending, and create a cohort for a social bond.

Potential benefits of social bonds

A significant benefit of social bonds is the ability to involve more of the private sector in contributing to improved social outcomes. This provides an opportunity for government to ensure that more players have a stake in solving tricky social issues.

By paying for outcomes and not outputs, social bonds allow providers to trial innovative programmes and techniques to meet the goals of the contract. Private sector investors are likely to have a higher risk threshold than government, and as such may be prepared to try new approaches to solving old problems. Any innovations could then be adopted and, if appropriate, applied in a wider setting.

Perhaps the most significant benefit of social bonds is the “no win / no fee” aspect. This suits government and agencies well as it transfers the financial risk of project failure to investors. This means government can potentially make headway on difficult social issues without having to fund a whole programme, and can justifiably claim responsible use of tax-payer funds.

Potential risks of social bonds

Difficulty attracting investors

One possible risk is that the market for social investors is relatively untested in New Zealand. This, combined with the potential risk of losing any initial investment, could make it difficult to attract financing. The seemingly enthusiastic development of social impact funds in Australia, however, suggests that funds would be available for social investment opportunities in New Zealand. This could either be through the development of similar funds domestically, or through direct investment from Australian funds.

Attribution and gaming

Perhaps the most widely reported risks and concerns around social bonds relate to:

  1. attribution – confidence that outcomes are a direct result of provider efforts, and not some external influences
  2. gaming – concern that providers may be cherry-picking the “easiest to treat” individuals, making little difference to the hardest to reach within the cohort.

These are valid concerns – issues around attribution and gaming can undermine confidence in the efficacy of a programme. Such concerns are common amongst all outcome-based payment structures. However, if the target cohort is well defined – that is, it is made up of the most vulnerable / hardest to treat people (for example, those most likely to re-offend) then attribution and gaming become less problematic.

Evidence around social bonds

While there are over a hundred social bonds being explored or implemented around the world, there are two well-known examples that illustrate their success as a contracting / finance tool, both of which were put in place to reduce re-offending:

  1. Her Majesty’s Prison (HMP) Peterborough (UK)
  2. Riker’s Island (New York, USA).

Despite different outcomes for participants, both of these programmes are considered a success for their use of social bonds.

Peterborough (UK)

The Peterborough trial was the world’s first social bond trial, and took place between 2010 and 2015. The trial introduced an intervention referred to as “One Service” within the prison, and supported short-serving adult male offenders before and after release from prison. This involved contacting offenders before release to introduce case workers, have their needs assessed and plan resettlement activities. One Service then worked with offenders for 12 months following release, even if the offender returned to prison in that time.

Results after the first year showed sufficient reduction in reconviction rates, which put the Peterborough pilot on track for payment at the end of the trial. The results of the second cohort, and whether investors will receive their investment returned with interest, will be known later in 2016 (Disley, Giacomantonio, Kruithof and Sim, 2015).

Peterborough was originally intended to operate until 2017, but was ceased as a social bond in 2015 due to nationwide changes to the probation service in the UK, which incorporated many of the initiatives that One Service provided, including a payment by results component for providers. The sudden lack of a control group made it difficult to evaluate the success of the overall programme, despite strong interim results, and the programme was discontinued.

However, it is indicative of the success of the Peterborough trial that the first cohort experienced a reduction in re-offending. This has led to the unusual situation of regarding
Peterborough as a qualified success – it achieved outcomes but was prematurely concluded.

Rikers Island (New York)

The social impact bond for Rikers Island was the first of its kind in the USA, and was finalised in 2012 (Chen, 2012). This social bond focused on reducing re-offending for young offenders at Rikers Island, with somewhat challenging success conditions. The investors – the prominent investment bank Goldman Sachs – would only receive their full principal investment of $9.6 million USD back if recidivism dropped by 10 percent, and would be eligible for a return on their investment if recidivism dropped further.

The second factor that differentiated Rikers Island from other social bonds was that Goldman Sachs was only liable for a portion of their initial investment, which was $2.4 million USD, as the Bloomberg Foundation had guaranteed a $7.2 million loan to minimise the risk to the bank. This was intended to encourage investors to support social bonds, thus creating a market for future work. †

The Rikers Island trial did not achieve the desired reduction in re-offending, and investors did not receive their investment. While the trial was unsuccessful in its outcomes, it can be considered a success for social bonds as a contract and financing tool for government. Despite the failure to reduce re-offending among youth, the City of New York had not spent any money on a programme that did not deliver results.

Putting it all together – potential opportunities for social bonds with Corrections

Data and information are improving about where the biggest costs of offending lie. In particular, programmes such as the investment approach to justice are providing insights into which individuals are most likely to enter a cycle of offending / re-offending, as well as how much these individuals are likely to cost the government and society over their lifetimes. This data lets us identify cohorts of individuals that could be considered as part of a social bond.

Let’s consider a hypothetical scenario which could form the basis of a social bond. We know there are 525 young people in Northland aged between 15 and 24 who are currently managed by Corrections††,  at an annual cost of around $5.7 million.

Using data from the Integrated Data Infrastructure (IDI), we should be able to identify which of these individuals have risk factors that increase their likelihood of re-offending, as well as the estimated lifetime costs to government. Let’s say we identify a cohort of 300 individuals with a particularly high risk of re-offending†††, with an annual cost to Corrections of $3.9 million.

We might introduce a social bond to reduce  re-offending within that group by 10 percent, and offer $300,000 to investors if this is achieved. On the basis that the potential saving to government is estimated to be $390,000, which is 10 percent of $3.9 million, government is financially better off if the goal is met. If re-offending is not reduced by the 10 percent goal, then government is not out of pocket.

This is a simplistic example of how we might introduce a social bond into Corrections, with a range of other approaches that could be considered.Δ There is however, significant potential for social bonds to be another tool available in our efforts to reduce  re-offending.


The evidence suggests social bonds are increasingly seen as a viable social policy tool, and have the potential to contribute to difficult social policy problems such as reducing re-offending.

For social bonds to work, we would need to focus on identifying the right cohorts – this will ensure that effort is targeted to those who most need it, while reducing risk around attribution and gaming.

Social bonds are not a replacement for traditional government or agency work, and should rather be seen as another part of the toolkit for helping to resolve difficult social problems.

*Savings would come through reduced future expenditure, such as lower spending on prison sentences or health costs, and so on.

**In other words, future savings are “brought forward” to pay for outcomes now. For example, if we know the cost of keeping somebody on a jobseeker benefit is around $11,000 a year (at a total lifetime cost of $330,000), and the cost of keeping that individual off a benefit is around $8,000 a year ($240,000) – the net benefit is still $90,000 to the Government.

***Extracted from the Integrated Data Infrastructure (IDI) in 2015. Risk factors for youth include a child having a parent with a Corrections history, a CYF notification, a mother with no formal education, or they are not enrolled in education or training.

† It is worth noting that the then Mayor of New York City, Michael Bloomberg, was a proponent of introducing private financing into achieving social outcomes.

††This number includes 462 offenders managed in the community, and 63 in prison.

†††Based on 250 community and 50 in prison

ΔWe could, for example, consider graduated payouts for partial success, or interim payouts if particular milestones are met.

Chambers, C. (2015, August 10). Review of Social Impact Bonds. Retrieved November 26, 2015, from

Chen, D. W. (2012, August 2). Goldman to Invest in City Jail Program, Profiting if recidivism falls sharply. Retrieved November 26, 2015, from,-profiting-if-recidivism-falls-sharply-(re-mdrc’s-new-social-impact-bond-project).aspx

Disley, E., Giacomantonio, C., Kruithof, K., Sim, M. (2015). The payment by results Social Impact Bond pilot at HMP Peterborough: final process evaluation report. Ministry of Justice (UK) Retrieved November 22, 2016 from

New Zealand Ministry of Health (2015) Social Bonds - Progress to date. Retrieved February 3, 2016, from

New Zealand Ministry of Health (2013) Social Bonds: Proposal for a New Zealand Pilot (Rep.) Retrieved November 26, 2016, from

Taleb, N. (2013, September 09). Taleb on Skin in the Game | EconTalk | Library of Economics and Liberty. Retrieved March 2, 2016, from