Babies, bathwater and desperately seeking savings
Instead of throwing the baby out with the bathwater, we should take greater care to put our precious infants in the right tubs, says Steve Goldberg, arguing that we should be more selective about what to fund through social impact bonds.
In a provocative blog post, Adrian Brown argues that “the flawed logic of cashable savings risks undermining Social Impact Bonds". Adrian, who is a principal at the Boston Consulting Group and author of Lighting the Touch Paper, The First Billion and other seminal papers on social finance, highlights several legitimate concerns about the predominant economic model for SIBs. But he overshoots the mark when he leaves us with two equally unattractive options: (1) relying on far-fetched savings that create perverse incentives, and (2) fighting over budget scraps to fund untested innovations with elusive economic value.
Fortunately, things aren’t quite that bleak. There are a number of savings-based SIBs that make sense and many more than don’t. The lesson isn’t that savings are a flawed basis for SIBs, but that we should be more selective about what interventions are funded that way.
Raising capital from private investors whose repayment is based on generating offsetting savings can be a sound way of powering performance improvements that traditional government contracting cannot. But, as Adrian rightly points out, in most cases “the cashable savings simply aren't large enough or... they fall across multiple parts of government making them difficult to aggregate.” He’s also correct that “the complexities of SIBs and risk-transfer involved make them an inherently expensive procurement option".
These are sound reasons for being more discerning about how we design SIB transactions. As Professor Jeffrey Liebman and his colleagues at the Harvard Kennedy School’s SIB Technical Assistance Lab recently wrote, “It is an open question how often governments will be interested in signing on to projects that, for example, produce budget savings equal to 70 per cent of their costs along with significant nonmonetizable social benefits (e.g., reduced crime, higher earnings, better health)".
Adrian’s observations suggest three remedies. First, if commissioners are considering SIBs where “the ‘price’ paid per outcome... doesn't reflect the full, risk-weighted cost to the provider of achieving that outcome,” they need to contend with Adrian’s concerns about perpetual subsidies and value-for-money.
Second, we must be far more creative about solving the accounting problem of capturing savings that are diffused among siloed departmental budgets. Two promising U.S. approaches are the Results First initiative of the Pew Center on the States and the Actionable Intelligence for Social Policy project at the University of Pennsylvania. We also need to adopt pragmatic ways of measuring the true costs and benefits of prevention programs. Excellent models have been developed by the Social Research Unit at Dartington in the U.K., and the Washington State Institute for Public Policy and the Urban Institute in the U.S.
Third, we shouldn’t do SIBs that are too small in relation to the fixed costs. As Professor Liebman’s team recently acknowledged, “The overhead costs of the SIB financing mechanism, including fees for legal counsel, intermediary costs, evaluation expenses, and costs associated with investor due diligence, are primarily fixed costs and will constitute a smaller proportion of the total project as the size of the intervention grows. In most cases, these costs are only worth incurring for a SIB contract worth at least $20m” (about £13m, a threshold we have yet to reach).
In order to reach his sweeping conclusion that the cashable savings model undermines SIBs, Adrian makes two points with which I respectfully disagree. First, he maintains that focusing on savings distorts government policymaking. “When government chooses to spend money on services such as nursery education or social care it does so because it believes these services are socially valuable and worth paying for,” not “to help governments save money". He worries about creating the impression that budgetary savings are “the primary value of SIBs".
When it comes to designing SIBs, government savings are one means to the end of improving the lives of grievously disadvantaged people. But “savings” without improved social outcomes are either nonsensical or just disguised budget cuts.
Second, Adrian asserts that “the main benefit of paying for outcomes without specifying activities is that this helps to drive innovation". From this premise he argues that payment-by-results contracts should be “paid for by reallocating the existing budget rather than tapping future savings".
The genius of the Peterborough social impact bond is not the development of new offender-support programs, all of which predated the pilot project. Rather, it is the financial innovation comprised of:
(1) raising long-term funding up-front from third parties,
(2) in amounts sufficient to provide integrated service packages to address complex problems,
(3) delivered by a dedicated intermediary exercising vigilant performance management informed by robust data collection, and
(4) the flexibility to make appropriate course corrections.
Peterborough poses no trade-offs between social impacts and financial results. Government savings and investor returns depend entirely on improved results for ex-offenders. The performance benchmarks are based on all offenders leaving the prison, whether or not they elect to participate in the re-entry program. So investors don’t get paid unless the One*Service attracts enough participants over three long years and equips them with the skills and resources they need to break the vicious cycle of recidivism. This is not a case where “cashable savings needlessly imposes the economic logic of the investor onto the government commissioner".
In the U.S., we’ve long debated (and continue to debate ad infinitum) whether the primary objective of social finance is to spur new innovation or expand innovations that have already been proven to be effective. Initiatives like the White House’s Social Innovation Fund and Grantmakers for Effective Organizations have come down on the side of “scaling what works", and federal grants and state procurements for SIBs are strongly trending in the direction of “evidence-based practices", disfavoring outcomes-based finance for social R&D. For example, New York State (with the pro bono support of the Kennedy School, funded by the Rockefeller Foundation) just released an RFP for up to $30 million in SIBs that says “services must have strong, credible evidence that supports their link to the proposed outcomes". When it comes to funding embryonic ideas, many (but not all) Americans believe that’s the province of philanthropy.
I can readily agree with Adrian that “commissioners should regularly look across the portfolio of services they buy and ask themselves which are the least effective". But I can’t share his optimism that “decommissioning these services frees up current spending that can be used to fund SIBs in policy areas where innovation is most urgently needed".
Under the dire fiscal circumstances in which both the U.K. and the U.S. find themselves, the idea of taking money away from existing programs, even if their effectiveness is doubtful, and transferring those funds to “innovative” initiatives, will arouse determined opposition. Moreover, government is notoriously maladroit at choosing effective programs to fund.
For example, Adrian is certainly correct in saying that “reducing homelessness is an outcome that society should value regardless of whether it pays for itself". But the difference between the wish and its fulfillment is vast. In 2010, President Obama launched a plan to end chronic homelessness by 2015. Although this is something we actually know how to do (and which can pay for itself), it’s clear that we are going to fall well short of the goal. Similarly, independent evaluations confirm that home-visiting programs funded by the Maternal, Infant, and Early Childhood Home Visiting Program (MIECHV) achieve exceptional results and save gobs of money, but Congress probably won’t reauthorize MIECHV when it expires in 2014. “Politicians... don’t tend to spend their efforts on expensive programs that show benefits many years down the road.”
Two former U.S. federal budget experts recently asked “Can Government Play Moneyball?”, referring to the American book and film showing how the payroll-deficient Oakland Athletics thrived by choosing players based on their statistical performance rather than following the hunches of baseball’s wise men:
“Based on our rough calculations, less than $1 out of every $100 of government spending is backed by even the most basic evidence that the money is being spent wisely. As former officials in the administrations of Barack Obama (Peter Orszag) and George W. Bush (John Bridgeland), we were flabbergasted by how blindly the federal government spends. In other types of American enterprise, spending decisions are usually quite sophisticated, and are rapidly becoming more so: baseball’s transformation into ‘moneyball’ is one example. But the federal government—where spending decisions are largely based on good intentions, inertia, hunches, partisan politics, and personal relationships—has missed this wave.”
As MDRC, the intermediary for the first U.S. SIB, recently observed, we have a long way to go on SIBs that don’t rely on savings:
“SIBs have been proposed for programs that are intended to realize government savings in a relatively short time period. These kinds of projects are probably the right place to start in building support for SIBs. However, the goal of most social programs is not primarily to save money but to improve the lives of low-income and at-risk individuals and families. SIBs could be structured to encompass other socially desirable goals that do not lead to government budget savings but do lead to societal improvements, so long as government can decide what it is willing to pay to achieve specific goals. SIBs could be designed to finance a range of different outcomes from increasing high school graduation rates and persistence in college, to improved cognitive and behavioral skills for young children, or better mental health outcomes for adolescents. All of these areas have promising, and perhaps even some proven, interventions with the potential to be scaled up. And additional funding for these kinds of programs is in at least as short supply as funding for programs that may generate short-term savings. But thus far, we have not seriously asked ourselves what we are willing to pay for this kind of success. Whether that amount would be sufficient to cover program costs and pay an acceptable return to investors is an open question worth exploring.”
Instead of throwing the baby out with the bathwater, we should take greater care to put our precious infants in the right tubs. In cases where savings have to make the investment math work, we should focus on evidence-based interventions where there is compelling cost-benefit analysis confirming that expected savings will cover the true costs of services (including the fixed costs of SIB infrastructure) plus reasonable returns to investors.
In the many cases where sufficient savings aren’t reliably available, diverting current spending to unproven innovations is likely to prove harrowingly difficult. Instead, we’ll need to figure out how desired outcomes like improved employment, health and educational achievement can be plausibly “monetized” to attract new sources of sustainable funding from private capital markets.