Inside Story

Will social impact bonds change the world?

The concept has spread like wildfire but the results, here and overseas, are mixed

Mike Steketee 4 October 2016 4471 words

Early supporter: NSW premier Mike Baird. CeBIT Australia


NSW premier Mike Baird says they have the “potential to change the world.” When he was still social services minister, Scott Morrison said he was “very keen” to explore them further. According to an OECD study, “few social policy tools have been disseminated so far and so fast.”

Social impact bonds, or social benefit bonds as they are called in New South Wales, are generating enough excitement to make Malcolm Turnbull proud. Not only that, they fit right into his innovation agenda, holding out the promise of a new way to tackle some of our most intractable social problems. They also tap into a growing market for investments aimed at doing good as well as earning a return.

The idea is to harness the resources of private investors to fund social programs at a time when governments feel increasingly constrained. The investors get a return only if the programs are successful, meaning some of the risk is transferred from taxpayers to private individuals and payments are based on outcomes rather than what the program happens to cost.

Pioneered in Britain six years ago, social impact bonds have spread rapidly, with sixty launched in fifteen countries. In Australia, Barry O’Farrell’s Coalition government in New South Wales was the first to announce a trial of the bonds in 2011, when Mike Baird, as treasurer, acted on a report commissioned by the former Labor government. Now Baird’s government is running two bonds aimed at reducing the rate of foster and institutional care for children. In July, it announced a third designed to reduce reoffending rates among ex-prisoners.

Baird wants to introduce two new social impact investments each year. He has outlined three priority areas: increasing access to early childhood education, encouraging more permanent foster care (including through adoptions), and using a new Aboriginal Centre for Excellence to help Indigenous children in western Sydney make the post-school transition. And as a sign of how far it is prepared to go down this path, the government is exploring four other areas – homelessness among veterans, waste management, the road toll and domestic violence. The new projects won’t all necessarily involve bonds: the government is also looking at potentially less complex alternatives such as government or private grants paid on the basis of outcomes.

Labor governments may not be quite as gung-ho about the idea, but Victoria, Queensland and South Australia have all announced pilot programs, or have plans to do so, in areas including drug and alcohol treatment programs, homelessness, prisoner reoffending and out-of-home care for Indigenous children.


As a concept, social impact bonds require a leap of thinking. Since governments took over providing mainstream social services from churches and charities during the first half of the twentieth century, it’s been assumed that the best way to ensure people are not left behind is to maintain nationwide programs funded by taxpayers.

Beyond routine social security payments, though, governments have a generally poor record in delivering services. Take, for example, shortcomings in child fostering and adoption, prisoner rehabilitation, and dealing with homelessness and long-term unemployment. There are arguments for trying something different.

At the same time, governments, particularly conservative ones, are looking to spend less in these and other welfare areas. As social services minister, Morrison was unequivocal about what he saw as the trend. Governments, he said, would get smaller in proportion to the size of the social challenges of the future:

But the non-government sector, the community, the private sector, will have to get bigger when it comes to addressing these challenges. This must extend to… private capital investment in addressing social needs – not charity… What I am basically saying is that welfare must become a good deal for investors – for private investors. We have to make it a good deal – for the returns to be there, to attract the level of capital that will be necessary in addition to the significant injection of capital and resources that is already provided by the states and the Commonwealth.

In the same speech, he said he was “very keen” to explore social impact bonds because “they have great potential for helping improve people’s lives while increasing public sector accountability.” When I submitted questions about the government’s plans to Morrison’s successor in the portfolio, Christian Porter, a spokesperson told me that the Coalition is exploring ways to develop a social impact investment market in Australia. “The government is open to looking at where social impact bonds might be a good fit in delivering outcomes for Australian communities,” she added.

Federal Labor is on board, too, at least in principle. “I am really interested in how to encourage and support the not-for-profit sector as they try to think of new ways of doing things,” shadow families and social services minister Jenny Macklin told me. “The good thing about social impact bonds is that people are trying out a new idea. At the moment I think there is still a lot of work to be done to see whether or not they will be effective. We need proper evaluation and a strong evidence base.”

In a speech last year, Porter also sounded a note of caution. Referring to various initiatives to increase housing affordability, including the South Australian trials of social impact bonds directed at homelessness, he observed that success had been “mixed” and “few projects have been able to scale up to the level of supply needed to make a real difference…”

Last year’s McClure report on welfare reform recommended expanding outcomes-based social investment models, including social impact bonds, or SIBs:

To date, evaluations of SIBs trialled overseas and in Australia have been promising. The vested interests of the parties in ensuring success has led to innovation and performance. A key benefit of SIBs is the opportunity they provide to “test and try” a multitude of different approaches. Where these solutions prove effective they could be scaled on a larger level.

This year’s federal budget included $96 million for a Try, Test and Learn Fund, which will invite bids from inside and outside government for innovative solutions to the problem of long-term welfare dependency among working-age people. Under its “investment” approach to social security, based on the New Zealand model, the government has commissioned an actuarial analysis, similar to that used by the insurance industry, to identify groups of people at risk of long-term dependency. As a result, young carers, young parents and young students will be the initial priorities for funding. The idea is to focus on policies or programs that, through prevention or early intervention, make a return on their investment – that is, that they save more money than they cost.


The first social benefit bond in Australia, launched in New South Wales in 2013, was Newpin (the new parent and infant network), which aims to help fostered children aged five years or younger return home and also to prevent the removal of those at risk of entering out-of-home care. Families go to Newpin centres, where staff help them work through problems such as the legacy of abuse they experienced as children, drug or alcohol addiction, and domestic violence. Caseworkers, typically trained in social work and early childhood development, and sometimes in psychology, focus on building respectful relationships with participants, in turn helping them to build better relationships with their children. Families are expected to attend two to four times a week for an eighteen-month period.

The program is run by Uniting, the services and advocacy arm of the Uniting Church in New South Wales and the Australian Capital Territory, and is partly funded by $7 million raised from fifty-nine investors, including wealthy individuals, family foundations, superannuation funds and Uniting itself. Payments to investors are based on the number of children successfully restored to the care of their families for at least a year.

According to the NSW government, “the bond targets a financial return of 10–12 per cent per annum for investors over its seven-year term.” In the first year, on the basis of a “restoration” rate of 60 per cent, the return was 7.5 per cent. In the second year, the return rose to 8.9 per cent and in the third year to 12.2 per cent, with a restoration rate of 65.2 per cent. This compares with a base restoration rate of 25 per cent, which the NSW Department of Family and Community Services considers to be the business-as-usual level. The maximum annual return is 15 per cent if a restoration rate of 70 per cent or better is achieved.

These are handsome returns that would be envied by most commercial investors in the current financial environment. Newpin investors do face risks: if the restoration rate falls below 55 per cent, they can lose up to half their funds. But they also have protections. Investors are assured of a minimum 5 per cent return, regardless of the restoration rate. And the proportion of restorations that fail because children are returned to out-of-home care within twelve months is assumed to be a maximum of 10 per cent, although the actual figure to date is 12 per cent.

The pilot: Britain’s then justice secretary Kenneth Clarke talks to prisoners at Peterborough Jail at the launch of the first social impact bonds scheme in September 2010. Chris Radburn/PA Wire

The families chosen for the programs are those considered most likely to benefit. “If the state child protection service thought there was definitely no chance of reunification, they would not refer to us,” says Liz Sanders, the head of Newpin. As a government official confirmed, “the aim is to identify people for whom the service can make the most impact and achieve the best outcomes.” This means that most of the 43,399 children in out-of-home care in Australia in mid 2015 – a 15 per cent increase over four years – are unlikely to be reunited with their families, even where this might be the best outcome. That figure includes 16,843 in New South Wales, which is disproportionally high in terms of share of population.

Compared to the scale of the problem, the numbers helped by Newpin have been small: after three years it had returned 148 children to their families, with eighteen going back to out-of-home care within twelve months, making a net restoration of 130 children. Another forty-seven were prevented from going into care. But sixty-six children left the program.

Nor does this kind of intensive social work come cheap. Contrary to the impression created by the Baird government, the $7 million provided by investors covers less than 20 per cent of the estimated $41 million the program costs over seven years. Another $9 million is allocated to the expected return to investors, the expected performance payment to Uniting, and a set payment to Social Ventures Australia – $50,000 in the first year, rising by 3 per cent in subsequent years – for managing the bond. To help meet Uniting’s commitments, the NSW government helps with an estimated cash flow of $50 million, which is supposed to be more than covered by the savings to the government ultimately generated by reduced out-of-home placements.

KPMG modelling for NSW Treasury before the launch of Newpin suggested that the program could save the state about $80 million by 2030. In 2014, the NSW government was reported to be planning to retain half these savings and use the other half to pay Uniting and bond investors.

But the government is not prepared to endorse this figure or commit to a current one. When I pressed the office of the NSW treasurer, Gladys Berejiklian, for current estimates of savings, the best government officials could offer was that further data was needed before a full analysis of gross and net savings was conducted.


Also aiming to reunite families, though structured differently from Newpin, is the second NSW program, Resilient Families, run by the Benevolent Society. Westpac and the Commonwealth Bank raised $10 million from investors for a five-year program to which the Department of Family and Community Services refers families with at least one child under six (including unborn children) assessed as being at risk of significant harm.

Under the program, caseworkers aim to visit families in their homes at least twice a week during an initial period of intensive support that also includes a twenty-four-hour call line. Families receive help dealing with substance abuse, domestic violence and other problems, and staff act as advocates on issues such as inadequate housing and welfare benefits.

“The funding has a lot of flexibility and we also can adapt very quickly,” says Claudia Lennon, the program’s manager. She cites as an example “Baby Ray,” an infant simulator doll used to help teach expectant parents how to care for a child. Imported from the United States, the doll is programmed to cry in certain circumstances – when its nappy hasn’t been changed, for instance – and to collect data on routines such as feeding, burping and nappy changes, as well as whether it has been shaken or had its head bumped. “It is a fabulous practical learning tool for parents expecting their first baby,” says Lennon. “It brings up things that parents would not have thought about, such as babies crying and not stopping.”

Two groups have invested in Benevolent Society bonds: a capital-protected class, which provided $7.5 million and receives a maximum return of 10 per cent a year; and a capital-exposed class, which put in the remaining $2.5 million and risks losing its money, but can achieve annual returns as high as 30 per cent. The minimum investment is $50,000. For each of the first two years, the return to the investor classes was 5 per cent and 8 per cent respectively in each year.

The NSW government provided $5.75 million up front to help establish the program and reduce the risk to investors. Assuming the scheme’s continued success, that amount will be deducted from the government’s payments at the end of the five-year bond. Westpac and the Commonwealth Bank did not charge for their services as financial intermediaries.

As with Newpin, payments are based on outcomes, but in this case there are three different measures rather than one. Two of them – the number of safety and risk assessments conducted on participating families, and the number of reports to helplines – showed a deterioration in each of the two years, which the Benevolent Society attributes at least in part to the “increased visibility” of family interactions under the program.

Despite this, Resilient Families was still able to report positive outcomes because two-thirds of the weighting is given to a reduction in children going into out-of-home care. The latest investor report puts this reduction at 27 per cent in 2014–15. What it doesn’t make clear is that the 27 per cent represents three children – a fall from eighteen to fifteen.


These figures drive home the point that social benefit bonds are not only in their early stages but are also operating on a very small scale. The NSW government is still developing principles to guide exactly how risk will be shared between government and investors and to determine actual, as opposed to assumed, benchmark costs – that is, what it currently costs the government to deliver services.

Innovation is one of the political selling points of social impact bonds. New South Wales is now leading the nation “in the innovative area of social impact investment,” Berejiklian declared in August last year.

But while the method of financing the program may be new, at least in the case of Newpin, the program’s approach is not. It originated in Britain as long ago as the early 1980s and was run by non-government organisations and funded by local government, which traditionally delivers social services in Britain. It was an expensive program, though, and fell victim to funding cuts.

Sanders, Newpin’s head in Australia, worked on the program in Britain before Uniting recruited her in 2005 to run it here. “Governments have been trying to find short-term solutions to these problems, so they love ten-week parenting programs,” she says. “That works for some families but it doesn’t work for the families we see because of entrenched intergenerational issues that need to be addressed.” These often include abuse suffered by parents when they were children, and recent or current drug and alcohol addiction. It was when Uniting was looking to expand the program from four to ten centres that it successfully tendered for one of the first two social benefit bonds in NSW.

Underlying that history is the potential conflict between innovation and the need to draw on a solid base of evidence. As government officers told KPMG during an evaluation of the trials, “if there is good data and a solid evidence base then it is likely that there are already effective programs in an area and there may be little in the way of service innovation.” Investors are also more likely to be attracted to a proven program.

Nevertheless, those involved in Newpin and Resilient Families praise their flexibility compared to sometimes lumbering bureaucracies with hard-and-fast rules. Parents who leave Resilient Families can rejoin the scheme, for instance, without needing to go through another government referral process. But there is no inherent reason why this should not be the case for any program delivered by non-government organisations, whether or not it’s funded by outside investors.

Supporters are keen to scale up the programs, confident that there is sufficient investor appetite to fund expansion. But they agree that the inhibiting factor is a shortage of personnel and other resources. This is not just a government funding issue: social workers and others in the field are mostly poorly paid, meaning that non-government organisations have trouble recruiting them.

It’s also the case that creating the bonds has been complex and extraordinarily time-consuming. KPMG calculated that the average number of staff hours spent on planning and developing each of the two NSW bonds was 11,712. It’s true that this was the first time the bonds had been introduced in Australia, and further experience is likely to reduce this figure. You’d certainly hope so.

Adding to the complexity is the involvement of a third party – namely, investors – and financial intermediaries. A 2011 report by the University of NSW Centre for Social Impact, headed at the time by former prime minister’s department head Peter Shergold, argued that social investors wanted a simple structure for the bonds and recommended against the use of a financial intermediary for a NSW pilot. But officials now say that this kind of financial expertise has been essential in establishing the structure.

As a result of these and other factors, some proposed schemes – here and overseas – have failed to make it beyond the conceptual stage. Back in 2012, the NSW government announced plans for three bonds: the third, designed to reduce prisoner reoffending rates and to be run by Mission Australia, didn’t proceed. Mission Australia is coy about the reasons. According to the government’s Office of Social Impact Investment, the decision not to go ahead was based on risks and challenges that included “performance targets, funding required, reputational risks to the parties involved, investor profile and legal,” as well as ongoing changes in prison policy and administration.

Negotiations for New Zealand’s first social impact bond, to help people with mental illness find employment, collapsed in July this year after a year of talks and $1.6 million in government spending on four proposals. The reported reason was that investors were unable to obtain guarantees about the security of their investment.

And sometimes the bond is issued, and fails. In the United States, a bond targeted at juvenile prisoners in New York was scrapped when it failed to reduce reoffending rates. The main investor, Goldman Sachs, lost a notional US$7.2 million as a result, but the blow was cushioned by a Bloomberg Philanthropies guarantee over most of the capital, resulting in its paying Goldman Sachs US$6 million.


While the savings to government from social impact bonds have the potential to significantly outweigh the costs of investor returns, administrative complexity and failed programs, the evidence to date is unclear. For each child who avoids going into out-of-home care, the NSW government saves between $30,000 and $45,000 a year, according to one official calculation. And that amount doesn’t include the indirect costs of potential future welfare benefits, health costs and crime. But the calculations are difficult to make: restoring a child to a family doesn’t guarantee he or she will have better life outcomes.

The jury is still out: the OECD’s Antonella Noya. Agence Française de Développement

Overseas experience suggests that savings are not always as large as anticipated. The UK National Audit Office reported in 2015 that a £3 billion Work Programme contract cost just 2 per cent less than would otherwise be expected. While that program isn’t financed through a social impact bond, it operates on the same principle: the government contracts with service providers to pay on the basis of the results achieved in finding work for long-term unemployed or those in danger of entering their ranks.

An OECD working paper warns of the risk of manipulating programs focused on outcomes. “It may be possible to game the results by selecting clients that are easiest to reach,” it says, “…while leaving those that would be most expensive without service.” Indeed, this was the experience in Australia with the Job Network, created when the Howard government privatised the Commonwealth Employment Service.

As a result, some in the welfare sector are wary of social impact bonds. As one policy adviser put it to me, an organisation committed to assisting the most disadvantaged, though not necessarily in the cheapest way, could be driven out by operators who promise quick, cheap results. “In the early days of the Job Network people made a lot of money, both for-profits and not-for-profits,” says this adviser. “They funded huge expansions in their charitable services by creaming funds from the Job Network. Then the government got wise and cut back. Who lost out? It was the unemployed.” In this race to the bottom, “it is very hard to see what the countervailing forces are, especially in a market environment where you set people up as competitors.”


Perhaps it will be different with social impact bonds. Many investors are motivated not just by the prospect of a financial return but also by a desire to improve people’s lives. But in its enthusiasm, the NSW government in particular has a political stake in the success of the bonds, creating a potential bias towards unnecessarily attractive investment returns.

In Britain, which has gone furthest down the social investment path, only philanthropists have put money into social impact bonds, according to one of their architects. That raises the question of whether the same goals can be achieved without the considerable cost and complexity involved in incorporating an investment return.

The OECD working paper, though not endorsed by the organisation as a whole, is cautiously supportive of social impact bonds but concludes that it is too early to make a definitive judgement. It argues, for example, that “few public authorities currently have the skills required to draw up complex results based contracts that are required for SIBs.” And on the other side, “few investors have detailed understanding of the types of social outcomes that are needed to address complex social challenges or of the barriers that need to be overcome to achieve better results. People who look at the world through a financial lens may think that everything can be fixed through incentive structures.”

In a second OECD working paper released this year, researchers Antonella Noya and Stellina Galitopoulou are also cautious:

SIBs have been costly instruments so far. They have entailed significant transaction costs that stakeholders should consider before embarking on them. Policy makers should evaluate carefully what is the value added for implementing a SIB for a policy intervention compared to a more traditional approach. However, transaction costs are expected to drop as more SIBs develop and there is a streamlined process for establishing them… Overall, while SIBs have achieved interesting results in some policy areas and triggered debates that can help reflect on how social services are being financed and delivered, additional knowledge and sound evidence need to be generated in order to reduce controversies around SIBs. The jury is still out.

The world’s first bond, designed to reduce reoffending rates among prisoners on short-term sentences at Peterborough Jail in Britain, operated for two years before the British government replaced it with a national rehabilitation scheme. While still based on payment by results, the new scheme will be implemented through a direct contract with the government.

This is despite the fact that the bond was judged by the government to have been a success, achieving an 11 per cent reduction in reoffending rates at a time when the national figure went up by 10 per cent. As with the Benevolent Society’s bond, the actual numbers don’t look as impressive: a reduction from 159 reconviction events for every 100 people released from prison to 141 – that is, from slightly more than one-and-a-half new offences per person to slightly less. Participation in the program was voluntary, presumably making it more likely to succeed, while the new program will be compulsory.

Victorian green bonds, issued by the state government in July at a coupon rate of 1.75 per cent, are an example of a less complex way of attracting private investment without a financial intermediary. Seventeen investors, including insurance and funds-management companies, contributed $300 million in a little over a day to finance LED traffic lights, mini-hydroelectric power stations, low-carbon buildings, a large-scale renewable energy power station and other projects.

All of this means that the question of whether social impact bonds can change the world, as Mike Baird believes, is still very much open. They may have a role in testing new approaches, but until there is much clearer evidence of net savings to taxpayers, it would be a courageous government that used them to run large programs.

This article was jointly funded by Inside Story and Australian Policy Online.