Sociology of Health & Illness Vol. 38 No. 8 2016 ISSN 0141-9889, pp.

1203–1216
doi: 10.1111/1467-9566.12450

Speculating on health: public health meets finance in

‘health impact bonds’
Rachel Rowe and Niamh Stephenson
School of Public Health and Community Medicine, The University of New South Wales,
Australia

Abstract Where modern public health developed techniques to calculate probability,

potentiality, risk and uncertainty, contemporary finance introduces instruments that
redeploy these. This article traces possibilities for interrogating the connection
between health and financialisation as it is arising in one particular example – the
health impact bond. It locates the development of this very recent financial
innovation in an account of public health’s role within governance strategies over
the 20th century to the present. We examine how social impact bonds for chronic
disease prevention programmes bring two previously distinct ways of thinking
about and addressing risk into the same domain. Exploring the derivative-type
properties of health impact bonds elucidates the financial processes of exchange,
hedging, bundling and leveraging. As tools for speculation, the functions of health
impact bonds can be delinked from any particular outcome for participants in
health interventions. How public health techniques for knowing and acting on risks
to population health will contest, rework or be subsumed within finance’s
speculative response to risk, is to be seen.

Keywords: finance, funding, budgeting, prevention, Chronic illness, long-term illness, risk,
biopolitics

Introduction

Early in 2013, a significant step towards the world’s first health impact bond was taken in the
municipality of Fresno, California. With demonstration project funds provided by a private
health foundation, over a two-year period venture capitalists and programme evaluators pro-
posed to model the link between reductions in demand for chronic asthma care with savings to
insurance companies and potential returns to investors (Cantor et al. 2013, Social Finance
2013). In December 2013, New Zealand’s Ministry of Health invited expressions of interest to
develop similar financial tools, and the Israeli Government presented a proposal to the fledg-
ling G8 Social Impact Investment Forum for the development of a social impact bond con-
nected to a 12-month intensive type 2 diabetes intervention (Samson 2013). These moves
extend ‘social impact bonds’ as have been launched in the UK (2010), ‘social benefit bonds’
in Australia (2011), ‘pay for success’ bonds in the USA (2012), and similar incarnations in
India, Belgium, the Netherlands, Ireland, Colombia, Canada, Germany (Azemati et al. 2013,
Social Finance 2012, 2014), and Malaysia (Tomkinson 2015) to the domain of disease

© 2016 Foundation for the Sociology of Health & Illness.

Published by John Wiley & Sons Ltd., 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA
1204 Rachel Rowe and Niamh Stephenson

prevention. The development of health impact bonds draws chronic disease and an emerging
financial market into the same frame.
In this article we examine potential alignments and points of disjuncture between public
health and financialisation, as they arise in an emerging mode of funding for disease preven-
tion initiatives. Our examination of social and health impact bonds foregrounds ongoing lines
of sociological inquiry as these bonds unfold. We want to open a space to consider how public
health techniques for knowing and acting on risks to population health might align with, con-
test, rework or be subsumed within the speculative response to risk in financial markets. Social
impact bonds in health help us to think about experimentation occurring across health and
finance. In the context of other types of private-public partnerships and privatisation, they are
not entirely exceptional. However, the development of social impact bonds calls for an exami-
nation of the financial concepts of risk at the forefront of this new form of investment in dis-
ease prevention. By tracing how health impact bonds function, this article contributes to the
sociological literature interested in the ways that techniques of measurement and valuation are
responsive to social pressures, as well as generative of particular epistemologies (Adkins and
Lury 2012).
Contemporary analyses of financial markets have emphasised their role within global capital
accumulation since the late 1970s (Bryan and Rafferty 2009, Strange 1988); within changing
governmental strategies, particularly in the management of public debt and reorganisation of
post-World War II social spending policies (Streeck 2013) and involving a range of new
actors, from non-financial corporations to households as both assets and debtors (Allon 2010,
Lapavitsas 2011, Martin 2007). The latter bodies of work, in particular, prompt consideration
of the various ways that financial tools seek and treat risk. They prompt us to consider the
multiple and uneven roles that finance plays within the porous and layered terrains of social
reproduction. We contend that health impact bonds are an example of financial innovation
seeking to expand opportunities for speculation on all-of-life or lifestyles through current
infrastructures of healthcare and welfare.
Beginning with a brief introduction to health impact bonds, this article then turns to the his-
toric development of techniques for knowing and managing risks to population health. Our
discussion pays particular attention to the specific context and challenges that the contempo-
rary phenomenon of chronic disease presents, as these areas of preventative intervention are
simultaneously cast as a growing challenge for public health and are flagged for potential
health impact bond funding. We then turn to financial accounts of risk. While measuring, mon-
itoring and anticipating risk may be nothing new to the calculus of public health, health impact
bonds open a range of new possibilities for capital. We suggest that the emergence of health
impact bonds demands an examination of how financialised conceptions of risk are taken up
in public health more broadly, and consider what the implications of this interplay may be.

What are impact bonds?

Impact bonds are instruments for social impact investment currently being trialled in several
countries. In a climate where non-government service providers must innovate to provide com-
petitive services, investors must demonstrate due diligence over programme governance and
objectives (KPMG 2012) and public health performance is measured against reductions in pub-
lic spending, it is hardly surprising that health impact bonds are being considered as a way to
finance preventative health programmes. (We note that preventative health is really a
misnomer, but use it here to reflect its common use.) Impact bonds extend traditional pay-
by-results mechanisms by enabling governments to pay investors retrospectively for
© 2016 Foundation for the Sociology of Health & Illness
 Public health meets finance in ‘health impact bonds’ 1205

programmes that have demonstrated particular outcomes. As financial commodities, health

impact bonds generate income streams that are linked to reduced future uptake of health and
welfare services, as well as anticipated increases in productivity (Mitropoulos and Bryan
2013). In this way, social and health impact bonds are packaged as enabling governments to
make immediate and future savings, by sharing the risk of future social problems with the pri-
vate market.1 The key features of health impact bonds are their stratification of risk exposure,
outcome measurement strategies and use of population data to establish counterfactuals.
Together, these act to commodify the risk profiles of programme participants, control groups
and potential programme participants in the future (Mitropoulos and Bryan 2013). We will
outline these key features with reference to the more established antecedents to health impact
bonds, that is, social impact bond pilots in recidivism and child removal.
In 2010, the world’s first social impact bond was designed for the One Service programme
based in Peterborough Prison, England. Its contract with investors was based on a reduction in
the frequency of re-conviction among programme participants, in comparison with a matched
national control group. In 2016, if the re-conviction rate of participants in the programme was
found to be at least 7.5% lower than the rate among controls, the UK Government and corpo-
rate partner Big Lottery Fund committed to pay investors a 13% return on their initial invest-
ment (United Kingdom Ministry of Justice 2014). The promised returns were based on
projected government savings in the cost of imprisoning a proportion of the 3,000 total partici-
pants who would participate in the intervention (Nicholls and Tomkinson 2013, Social Finance
2014). It was anticipated that in 2014 an interim return of 10% on principal would be paid to
investors. This was contingent on the re-conviction rate of the first 1,000 programme partici-
pants being 10% lower than among matched controls. In August of that year, the UK Govern-
ment announced that the re-conviction rate had fallen 1.6% short of the targeted cut-off for
early returns to investors (United Kingdom Ministry of Justice 2014). The Peterborough Social
Impact bond was cancelled later that year due to the introduction of a national programme for
people exiting prison, which would have confounded the One Service programme’s outcomes.2
Nonetheless, the Peterborough Bond was the first to demonstrate how the market value of
investments in a social impact bond could be related to changes in the overall risk profile of
participants in a prevention programme.
Meanwhile, in New South Wales, Australia, another social impact bond gave an example of
the first successful return to investors through this recent financial innovation. The market
value of investments in Uniting Care Newpin Social Benefit Bond, launched in 2013, are
linked to the restoration-to-family rates of children in the intervention compared with rates
among a control group (Wallace et al. 2014). Exposure to the risk of children staying in out-
of-home care forms the basis of the repayments mechanism in the Newpin bond. In 2014, a
7.5% return to investors was paid by the New South Wales Government, linked to the pro-
gramme’s attainment of a 60% restoration rate among participating families (Eyers 2014). The
Newpin Bond promises a maximum 15% interest rate per annum to investors, to be paid after
its full 7-year term (KPMG 2014). These examples demonstrate how social impact bonds link
investors’ interests to programme outcomes. However, rather than conclude that this form of
financing forges a collective interest in health and welfare, the technical aspects of how social
and health impact bonds are imagined and implemented need to be interrogated first.
In contrast to government bonds that entail a fixed rate of return, the repayment rates of
social impact bonds are set to fluctuate in correspondence with programme outcomes. While
some social impact bond pilots have involved the sale of tranches whose minimum repayments
are guaranteed, others have factored in greater risk for promised return. In Australia, in a con-
text where there has been limited private investment in public health care and welfare previ-
ously, the NSW Government has employed a cautious strategy to stimulate market confidence
© 2016 Foundation for the Sociology of Health & Illness
1206 Rachel Rowe and Niamh Stephenson

in these new financial mechanisms. In the Newpin bond, the NSW Government committed to
pay a minimum 5% interest per annum on investments for the first 3 years of the programme
regardless of its performance. In 2020, full investments will be returned if restoration rates
meet 55% over the total period, and a minimum return of 50% of the principal invested will
be repaid to investors if the restoration rate for children in the Newpin programme is found to
be less than 35% (KPMG 2014, Social Ventures Australia 2014). In the other Australian social
impact bond, which funds the Resilient Families Service operated by the Benevolent Society,
two tranches of risk were sold to investors. The low-risk tranche, which was purchased by 40
institutional or professional investors, guarantees full repayment of initial investments with an
additional maximum 10% per annum return if the programme’s target outcome is met after the
bond’s 5-year term (Benevolent Society 2013). The high-risk tranche, sold in this instance to a
group of 17 investors, only guarantees the repayment of principal invested if the minimum tar-
get programme outcome is met. High-risk investments in the Benevolent Society Bond are
linked to a maximum 30% per annum return, contingent on the degree to which the target pro-
gramme outcome has been met after the full term (Benevolent Society 2013). This cursory
view of the internal stratification of social impact bonds directs attention to their function as
instruments of risk.
Health impact bonds are being imagined similarly to these current examples of social impact
bonds in recidivism and child removal. Health impact bonds would make the probability of ill-
ness tangible through creating a commodity that speculates on the changing health profiles of
populations engaged in preventative health programmes. By design, governments would deter-
mine suitable health programmes for bond finance and engage intermediary actors to market
potential investment opportunities. Healthcare providers would administer the bond-financed
programme and external evaluators would monitor the attainment of particular outcomes
among the programme’s participants. Several governments, including New South Wales and
New Zealand, have already called for expressions of interest to develop impact bonds in the
areas of mental health, diabetes, asthma and other chronic diseases (KPMG 2012, New South
Wales Government 2015).
The calculus of outcomes is crucial to social and health impact bonds. In the first bond pilots,
binary outcome measures were perceived to be easier for potential investors to understand, and
therefore easier to market to investors (KPMG 2014). Mixed outcomes and corresponding mea-
surements have been necessary for some programmes. In the first health impact bond pilot, cur-
rently running in the Californian city of Fresno, returns to investors after a 2 year period that was
set to end in March 2015 depended on measuring a 30% reduction in emergency room visits and
a 50% reduction in asthma-related hospitalisations among intervention participants (Social
Finance 2013). Repayments were based on insurance claims data during the programme period,
in comparison with predicted claims based on participants’ previous insurance claims (Social
Finance 2013). A proposed bond-funded type 2 diabetes management programme in Israel would
be based on measuring future demand for diabetes treatments, take up of disability related allow-
ances and changing levels of workforce productivity (Samson 2013). In the Benevolent Society
Bond, repayments to investors are to be based on an ‘improvement percentage’ (Benevolent Soci-
ety 2013). This percentage will be based on the total uptake of certain interventions among the
programme participants, compared with a matched control group (KPMG 2014). These strategies
for defining outcomes share a dependence on longitudinal population data collection, the exis-
tence of sector-wide operations data for the determination of cost effectiveness, and a consoli-
dated approach (involving a significant mobilisation of resources) across all departments and
institutions that intersect with the population targeted by an intervention.
As we see in these examples, the existence of population health databases and national tar-
get screening rates (e.g. for type 2 diabetes, cardiovascular disease, chronic respiratory
© 2016 Foundation for the Sociology of Health & Illness
 Public health meets finance in ‘health impact bonds’ 1207

conditions and cancers) are part of the foundation required for impact bonds in health. The
NSW Treasury commissioned report on the initial phase of the social impact bond pilots notes
that the absence of a dedicated research and statistics institute on family organisation poses an
obstacle to the rigorous measurement of programme outcomes because insufficient population
data exists elsewhere (KPMG 2014). The tracing of outcomes among participants in an inter-
vention and suitable population controls is facilitated in some cases by the existence of dedi-
cated national research institutes. In the domain of public health, impact bonds will depend on
standardised data-collection and data linkage.
Impact bonds require sufficiently large populations to calculate probabilities with reasonable,
statistical precision (KPMG 2012, 2014). Harvard Kennedy School consultants on the develop-
ment of social impact bonds, known as ‘Pay for Success Bonds’ in the USA, suggest that pro-
grammes must have an intake of at least 200 people annually in order to rate the satisfaction
of targets and calculate returns to investors (Azemati et al. 2013). The Fresno health impact
bond involves 200 children, whose medical histories will form the counterfactual used to mea-
sure the degree to which the two key outcomes related to the intervention have been met. In
the Australian Newpin Bond, there will be approximately 700 families participating over the
course of the programme and a live control group is to include at least 300 children at each of
the three measurement periods during the 7 year programme period (Wallace et al. 2014). The
control group in this case is to be determined by applying the Newpin programme’s eligibility
assessment to children, and their parents, in areas that have comparable socio-demographic
characteristics to the areas where the Newpin programme currently operates (Wallace et al.
2014). Eligible families are to have been allocated to the control group at least 18 months
prior to the comparison of outcomes between the Newpin programme and standard interven-
tions (Wallace et al. 2014).
Across each of these examples, we can see that health impact bonds depend on a sophisti-
cated knowledge of population health, techniques for estimating how service uptake can be
expected to change over time, techniques for making risk profiles for different segments of the
population, and methods for forecasting impacts. Probabilistic risk calculation has enabled
health to be framed by sets of risk factors (Rockhill 2001). Whilst approaching populations in
terms of risk is not new for public health, health impact bonds draw on familiar instruments to
do something different with this risk. Before we examine that turn further, we first want to
consider aspects of the role that public health expertise has played in the governance of popu-
lations. This will contribute to understanding the kinds of expertise that financialisation
requires and re-works in order to expand into health impact investing.

Valuing futures: the population and modern disease prevention

Ideas about effective governance, since at least the late 18th century, have hinged on the cre-
ation of techniques that are able to identify and predict the implications of minor changes upon
the distribution of risk across populations. Classical liberal thought considered sovereign
decree or intervention as limiting society’s creative potential, and instead appreciated that pro-
ductive capacity could be fostered by creating particular, dynamic conditions and facilitating
exchange and contestation within a social body. The novelty of this perception was its
abstracted representation of the biological and social features of human life. It saw the popula-
tion as ‘a thick natural phenomenon . . . constantly accessible to agents and techniques of
transformation’ (Foucault 2007: 71). An objectified social body was constituted via a range of
techniques (such as the collection of morbidity and mortality data) that could observe general
biological processes, as well as measure, monitor and intervene in the statistical regularity of
© 2016 Foundation for the Sociology of Health & Illness
1208 Rachel Rowe and Niamh Stephenson

events (Foucault 2007). The techniques that enabled this conception of the population led to
the emergence of disease prevention as a possibility.
Examining shifts in the discipline of medicine, we can see that by the middle of the last
century attention had expanded from treating discrete lesions on the bodies of individuals to a
focus on disease aetiology within populations (Armstrong 1995). This included population
measurement and surveillance strategies that harnessed medicine for functions beyond the
treatment of disease, as well as engaging a broader range of non-medical actors in preventative
healthcare. Medicine’s central concern became the identification of modifiable variables associ-
ated with disease onset and progression. Armstrong (1995, 2012) has examined this shift in its
spatial, technological and temporal dimensions, suggesting that a ‘remapping of the spaces of
illness’ during the twentieth century correlated with the expansion of medical attention from
symptoms to risk practices. An interest in the constancy of rates of disease was pivotal to
identifying the targets of interventions to prevent disease (Foucault 2007). The clinic could not
afford the context in which to calculate probabilities or pre-empt risk in a population because
it offered only a glimpse of the already symptomatic patient. This propelled, or at least was
part of, the turn to community-based contexts within modern medicine.
Screening became a central technology by which to generate knowledge of the population.
Its development occurred during a time where the social contract expressed itself in terms of
collective interest; and techniques for appraising risk across the social body were integral to a
strategy to mitigate crises of capital. Health interventions in the Keynesian or New Deal period
embraced a promissory tone of ‘health for all’. While this grammar of universal inclusion fre-
quently obfuscated the needs and concerns of many marginalised social groups, it accompa-
nied post-World War II development projects, and within these, population screening
programmes. After World War II, the first population-wide screening programmes were mobi-
lised in the USA, Australia, Canada, New Zealand, and other countries with large immigration
programmes, to identify signs of tuberculosis prior to the development of symptoms
(Armstrong 2012, Australian Health Ministers’ Advisory Council 2008, Dara et al. 2013).
Over the decades that followed, it became widely understood that screening could show more
than the existence of disease. By identifying not only physiological but life-style risk factors
within the population, screening could also predict future incidence rates (Armstrong 2012).
The questionnaire, or population health survey, became a particularly important accompani-
ment to physical screening and helped to extend the parameters of medicine to the mental and
physical propensity for illness. Large-scale population health surveys, such as the Framingham
Heart Study (1948 to present) or the Nurses Health Survey (1976 to present) enabled the iden-
tification and prediction of pertinent lifestyle and social practices, and were able to link these
with disease progression. The Framingham Heart Study has identified disease patterns associ-
ated with nutrition, exercise and body mass from which the future likelihood of developing
cardiovascular disease can be estimated. In a particular individual body the existence of any or
all of these ‘risk factors’ (a term coined by Framingham researchers in 1961, Kannel et al.
1961), is not guaranteed to produce cardiovascular disease, but the presence of such factors
holds the statistical power to predict prevalence in a population. Such studies measure associa-
tions in order to enable public health and clinical responses that aim to stabilise risk.
Using knowledge of population risk factors has become integral to the design and develop-
ment of preventative interventions. Yet, the privatisation of risk displaces the original subject
of risk calculation from the population to the individual. We can already see this in current
contexts where, for example, unemployment benefits have become contingent upon particular
job-seeking activities, (Jessop 2007) financial security in old age has become dependent on
superannuation, and welfare based on personal assets such as housing wealth or inheritance
(Allon 2010, Cooper 2013, Langley 2008). It follows that much of the contemporary
© 2016 Foundation for the Sociology of Health & Illness
 Public health meets finance in ‘health impact bonds’ 1209

architecture of health care has been reconfigured to emphasise personal responsibility, choice
and insurance. There are two key points that we would make about this conjuncture, with rele-
vance to financial innovation in preventative programmes. First, the concepts of risk developed
within public health, illustrated by major studies such as Framingham Heart and Nurses
Health, are focused on prospective and potential occurrence. Second, that prevention has
increasingly come to extrapolate, or deduce individual health risks from the probabilities of
disease in a population. Otherwise put, the causes of incidence falsely reappear as the causes
of cases (Rockhill 2001). This matters, as we turn to the integral role played by chronic dis-
ease (i.e. the terrain in which health impact bonds are being proposed) in public health’s valu-
ing of potentiality via its expanding calculus of prevention, pre-emption and reduction of
disease.

Chronicity and prevention: the expanding temporality of disease

If the shift in public health interest from the symptoms of disease to its risk factors extended
the spatial and temporal dimensions of disease, in no field of health could this be more pro-
nounced than in chronic disease medicine (Armstrong 2014a). Modern prevention and the phe-
nomenon of chronic (non-communicable) disease have been shaped by each other.
Armstrong’s (2012) review of articles pertaining to chronic disease and its earlier incarnations
(e.g. degenerative disease), published in the Journal of the American Medical Association
(from 1883 to 2000), identifies that by the 21st century multiphastic screening for latent condi-
tions had largely been replaced by screening that targets lifestyle risk factors for chronic dis-
ease. Interventions aimed at preventing chronic disease followed suit, broadening their scope
from early diagnosis to health promotion programmes designed to tackle multiple, potentially
predictive risk factors enmeshed in social relationships, body shape, nutrition and the built
environment. This process meant that all-of-life and lifestyles came within the realm of medi-
cine. Moreover, as thinking about disease-risks slowly expanded to enrol the possible precur-
sors of disease within the preventative medical model, ideas about the temporality of disease
developed to include not only the early or latent stages of illness, but lifestyles that have the
potential to give rise to possible disease in the future.
The future potential of chronic disease unsettles notions of the present. Kreiner and Hunt’s
(2014) examination of the clinical treatment of ‘at-risk’ conditions points to the increasing con-
fluence of treatments for ‘risk’ and the treatment of immediate health threats. Participants in
their study who presented a borderline range of risk also frequently considered themselves to
already have the respective chronic disease. Within these perspectives, even terms like ‘partial
patient’ (Gillespe 2012, Greaves 2000) and ‘pre-diabetic’ appear to be surpassed. This is seen
in the ruptured binary of health and illness (Armstrong 1995), in treatment as prevention (Krei-
ner and Hunt 2014), as well as in contemporary accounts of the subjective experience of ill-
ness (Nettleton 2006, Reynolds Whyte 2012). The particular implications of chronicity are that
time in the present is stretched, enduring and future oriented (Cazdyn 2012). While there are
moments of acuteness in the disease trajectories of individuals (Reynolds Whyte 2012),
through the lens of the population, chronic diseases do not promise a sudden crisis or emer-
gency. Rather, chronic diseases offer the promise of predictability, imagined as a wave that
slowly increments to epidemic proportions. This sense of time figures in stabilising the con-
ceptual terrain for experimentation with health impact bonds.
Chronic disease has also extended the structure and aims of preventative or early interven-
tions beyond disease eradication to a patient’s ability to function. Armstrong (2014b) has
argued that the challenges presented by chronic illness in the 20th century have spurred
© 2016 Foundation for the Sociology of Health & Illness
1210 Rachel Rowe and Niamh Stephenson

changes in the organisation of health care. Indeed, the most rapid increase in medical interest
in chronic disease (Armstrong 2014b) occurred during the time frame as a significant debate
was occurring within the World Health Organisation and its associated networks. The WHO
Declaration of Alma Ata (1978) aimed to push medical interest across the globe past disease,
to health. In doing so, the Declaration argued for a preventative agenda focused on averting
the various risk factors associated with major chronic conditions. Writing decades later on the
cultural politics of the ‘chronic’, Cazdyn (2012) suggests that a major change in health care
has manifested in increasing emphasis on symptom management rather than cure. Within this
move, chronic disease medicine has invented new benchmarks for disease progression that are
illustrated in the shift in focus from degrees of disease severity to the capacity of target popu-
lations to undertake particular activities and live in so-called ‘productive’ ways. To some
extent, then, chronicity dispenses with the notion of terminality or acuteness, and re-positions
health and illness along a continuum of relative degrees of functionality (Cazdyn 2012).
Preventative interventions in this area have followed suit.
More recently, the community imagined as integral to the ‘health for all’ aims of Alma Ata
is being refigured as a market, and the expansion of market-based funding for preventative
health interventions is presented as a solution in the context of the enduring presence of dis-
ease. Clinical medicine is already a major market, about which much has already been written
(e.g. Clarke et al. 2003, Cooper 2008, Cooper and Walby 2014, Duster 2007). There have
been various attempts to economise prevention, from the Disability Adjusted Life Year (Ken-
ney 2015) through to actuarial arguments for mainstream health promotion campaigns over tar-
geted programmes interventions (Shiell et al. 2013). This means that before the introduction of
health impact bonds, much medical interest in chronic disease was already imbued with a con-
sequentialist reasoning, which functions to expand the temporality of disease and to render
health interventions in the present as market opportunities. This begs a question as to whether
health impact investing is best approached as an extension of marketisation processes already
familiar to public health, or whether it entails a significant shift; and we turn to this question
now.

From economising prevention to commodifying health risks

While context-specific mechanisms for hedging risks or arranging credit are as old as capital-
ism itself, the global market for financial commodities took off during a time of rapid change
in practices previously presided over by the nation-state. By the 1980s, most national econo-
mies had started to curb rates of social spending that coincided with politically guaranteed
full-employment (Streeck 2013), and had moved to enhance competition in the banking sector
and expand credit markets (Battellino 2007). Deregulation was key to generalising a new mode
of capital ownership from this period (Bryan and Rafferty 2006, 2009). This mode of owner-
ship has enabled trade in the contingencies related to underlying commodities without actually
trading the commodities themselves. A burgeoning market for financial securities has enabled
the development of a range of new tools focused on future potential. It is within these devel-
opments that we can locate health impact bonds as tools that enable the exposure of capital to
risk.
We are certainly not the first to point out how the preoccupation with risk has become an
organising force in contemporary society (Beck 1992, Giddens 1990, 1991, Lupton 1999) or
how neoliberalism harnesses insecurity as a dynamic force (Wacquant 2009). However, as
Bryan and Rafferty (2006) have also noted, sociological accounts of risk and security have
often failed to consider the specific ways that contemporary finance rationalises and works
© 2016 Foundation for the Sociology of Health & Illness
 Public health meets finance in ‘health impact bonds’ 1211

with probability, potentiality, uncertainty and risk. Securities sell the future performance of an
asset, and financial derivatives sell its exposure to risk alone (Bryan and Rafferty 2006, 2009).
An examination of financial interpretations of risk, return and exchange prompts us to consider
the novel form of speculation involved in health impact bonds. Unlike earlier funding arrange-
ments within the jurisdiction of public health departments, the current development of health
impact bonds enables investors to gain exposure to the contingencies associated with the
health of populations.
Examining them as financial instruments, health impact bonds translate health risks into uni-
versally exchangeable commodities. Already social impact bonds have been released on sec-
ondary markets where they can be traded between investors as an asset class in their own
right (Rose 2014, see also Benevolent Society 2013, Philipson 2011). This move essentially
de-links the experiences of health and sickness (or imprisonment, child removal and other
focuses of social benefit bonds) from what an investment in a health impact bond can do for
its owner.
The emerging market for social and health impact bonds draws the derivative-like properties
of these instruments into clearer view. Derivatives make risks commensurable. They can, like
money, be a way to store value and are able to be universally exchanged at rapid speeds and high
volumes (Bryan and Rafferty 2006). As Bryan and Rafferty (2006) have demonstrated, these
characteristics enable derivatives markets to bundle and blend a range of exposure positions to
different types of underlying assets. In this process, the character of the underlying component
(i.e. the occurrence or non-occurrence of child removal, or particular asthma or mental health
symptoms) upon which the derivative is based is irrelevant. Health impact bonds do not necessar-
ily depend on health or illness to produce returns when combined with a range of other invest-
ments they can be sorted into categories of risk, hedged against other probabilities (Mitropoulos
and Bryan 2013) or leveraged within corporate responsibility targets (KPMG 2012). What health
impact bonds trade in, regardless of the individual intentions of the actors that might purchase or
trade them, is the degree of, and diversification of, exposure to risk itself.

A new politics of prevention?

Just as increasingly constrained public health budgets have been the pretext for the exchanges
between military finance and emerging infectious disease interventions (Lakoff 2008), we must
also pay attention to the new modes of finance, health interventions and struggles for health-
care that might come to mutually constitute how the phenomenon of chronic disease is under-
stood. Here, one set of questions to ask concerns the kinds of interventions that would seem
to suit health impact bonds and in turn, what do health impact bond funding arrangements hin-
der, what types of programmes are unlikely to be supported and potentially lost, and what sub-
tle shifts might occur in public health planning as a result? First, since establishing
intervention and control groups of the required magnitude requires considerable coordination
and resources, and eligibility for the programme as well as for the control should be defined
and consistent, large non-government organisations are best positioned as primary contenders
to operate bond-financed programmes. Second, the focus on specific, clearly defined outcomes
has been celebrated as a way to encourage health service providers to innovate and change
methods in ways that they deem necessary to achieve the outcome target against which finan-
cial risk is being calculated (Hems 2009). One of the subsequent implications of health impact
bonds is their preoccupation with accurate measurement of outcomes, relative to their interest
in how outcomes are achieved. This consequentialist approach to the planning and evaluation
of preventative interventions is reinforced in health impact bonds. Early reviews of bond
© 2016 Foundation for the Sociology of Health & Illness
1212 Rachel Rowe and Niamh Stephenson

implementation have already noted the need to distinguish between data that supports health
and welfare workers to undertake their roles, and data that enables the assessment of cost
effectiveness (KPMG 2014). The pursuit of flexible programme delivery that focuses on binary
outcomes also has the potential to undermine attention to the processes employed and to unin-
tended consequences of programmes (Rockhill 2001). And third, the difficulty of locating suit-
able controls for small or specialist programmes makes them less likely to attract funding. In
such cases, it has already been acknowledged that social and health impact bonds may not be
a suitable mechanism by which to raise funds (KPMG 2014). It follows that if public health’s
role continues to expand beyond allocating taxation to preventative health care, to managing
private investment as funding for chronic health programmes, there will be implications in the
development and delivery of healthcare and for addressing present health inequalities. If a
function of public health becomes trading, or enabling trade in risk exposure, we can anticipate
social effects that reach beyond the immediate question of programme funding or programme
objectives.
During the modern Welfare State technical developments in health care and medicine
advanced techniques that measured probability, potentiality, risk and uncertainty within popu-
lations, and enabled preventative approaches within public health. The past few decades have
seen these techniques repurposed in ways that network economic, political and social gover-
nance. The identification of an emerging private market for preventative health finance should
be understood in the context of the broader tendency of capital to seek profit through offsetting
risk, expanding markets within reproductive spheres of labour, and enhancing labour’s produc-
tivity. When we speak of this tendency we must understand that capital is not a monolithic or
rational actor, but rather, it is a relationship constituted through manifold moments of
exchange, appropriation and circulation. Looking, as we have in this article, at the mechanisms
being used to roll out social impact bonds in health, we can see that these tools emphasise dif-
ferent concepts of risk. First, is the traditional public health concept of risk: where preventative
programmes seek to reduce the burden of disease in populations (and in doing so stabilise the
terrain for productivity and appropriation). Second, is a concept of risk central to exchange
and to circulation. Risk is necessary to accumulation, not only in the process of capturing prof-
its, but in managing and stabilising uncertainty as part of broader financial investment strate-
gies. It is the latter concept, or rationale for risk, that interests us here. Financial innovation
associated with preventative health activities occurs in the context of varied, locally specific
practices and responses to the social factors associated with disease incidence, progression
understandings of disease. It is also occurring in the context of insecurity, household unafford-
ability and stagnant wage levels for the huge majority of potential participants in the preventa-
tive health programmes for which health impact bonds are being considered. The current
development of health impact bonds, then, can be seen to extend modes of valuing population
health based on degrees of exposure to risk and deepen capital’s ability to speculate on life,
including moments of life that are not directly engaged in ‘productive’ activity within the tra-
ditional notion of labour contract (Lilley and Papadopoulos 2014, Papadopoulos, Stephenson
and Tsianos 2008).
To this end, public health has been able to provide the necessary methodologies to develop
health impact bonds. For the risks associated with a particular health outcome to become fun-
gible, health impact bonds require health risk profiles to be separable, dispersed and calculable
(Ewald 1991). As discussed earlier, public health techniques such as screening, surveillance
and modelling can constitute populations in this form. Moreover, public health’s growing
interest in chronic disease, with its emphasis on managing and intervening during the latent
and potential manifestations of disease, has played an important role in extending the poten-
tiality of the population. Chronic disease preventative interventions act to stabilise some
© 2016 Foundation for the Sociology of Health & Illness
 Public health meets finance in ‘health impact bonds’ 1213

conditions, and measure lifestyles, in ways that constitute them as apt objects for experimenta-
tion in novel modes of financial speculation.
The alignment of public health and financial markets via health impact bonds also leads us
to ask, as social volatility itself is harnessed for capital flows, what opportunities for alternative
ways of providing care might cease to appear and how might public health be transformed
through its uptake up financialised concepts of risk? How will the ways that various actors cre-
ate, re-interpret and/or resist the techniques of measurement required by these instruments, be
reflected in political pressures upon healthcare into the future? Where modern public health
has developed sophisticated techniques to calculate and monitor risk, finance appropriates this
risk to produce returns in the present. As commodities, personal and collective risks reappear
in fast and fluid circulation, as abstracted from the bodies where the risk of diabetes, asthma,
mental illnesses or other health issues becomes actualised. Social and environmental volatility
are required for health impact bonds to function as instruments of risk (Mitropoulos and Bryan
2014). As the risk associated with investments in the likelihood of future chronic disease can
be distributed, bundled or sold on; the experience of physical and psychological insecurity of
people who are the targets of health interventions can be reconstituted as experience that can
be hedged or swapped. That is, subjects who might previously have been cast as failed entre-
preneurs can be remade so that ‘everyone’ can be constituted as ‘an enterprise to be managed
and a capital to be made to bear fruit’ (Dardot and Laval, 2014: 302). It follows that finance
may make it possible for us to understand the ways in which we are connected across spatial
and temporal lines. However, the stakes may be that our collective health become even more
bound up in financial trading and capital accumulation. Challenging this predominant way of
valuing life will require a new range of analytical frameworks and close attention to how con-
temporary antagonisms (including struggles around the social determinants of health) shape,
contest and remake this shifting terrain.

Acknowledgements

The authors wish to thank Dr kylie valentine and two anonymous reviewers for their feedback
on an earlier version of this article.

Address for correspondence: Rachel Rowe, School of Public Health and Community Medicine,
UNSW Medicine, NSW, 2052, Australia. E-mail: rachel.s.rowe@gmail.com

Notes

1 Although most current social impact bonds are underwritten by State departments (Social Impact
Investment Taskforce 2014), the demonstration health impact bond being conducted in Fresno, Cali-
fornia will calculate savings to insurers, rather than to government.
2 We note here that social (and health) impact bonds operate and are affected by a broader political
context in which they exist. The Peterborough Bond was cancelled because a national reform, Trans-
forming Rehabilitation, was being launched that would provide an overlapping programme likely to
involve the same cohort. Furthermore, general incarceration rates are likely to fluctuate due to convic-
tion policy, policing strategy, poverty and levels of inequality. The number of people in English and
Welsh prisons has more than doubled over the past decade (United Kingdom Ministry of Justice
2013). The expansion of the prison industry in the USA is well documented elsewhere (e.g. Wacquant
2009).

© 2016 Foundation for the Sociology of Health & Illness

1214 Rachel Rowe and Niamh Stephenson

References

Adkins, L. and Lury, C. (2012) Introduction: special measures, The Sociological Review, 59, supplement
2, 5–23.
Allon, F. (2010) Speculating on everyday life: the cultural economy of the quotidian, Journal of Commu-
nication Inquiry, 34, 4, 366–81.
Armstrong, D. (1995) The rise of surveillance medicine, Sociology of Health & Illness, 17, 3, 393–404.
Armstrong, D. (2012) Screening: aping medicine’s temporal spaces, Sociology of Health & Illness, 34, 2,
177–93.
Armstrong, D. (2014a) Chronic illness: a revisionist account, Sociology of Health & Illness, 36, 1, 5–27.
Armstrong, D. (2014b) Revisionist or simply wrong? A rejoinder, Sociology of Health & Illness, 36, 7,
1116–7.
Australian Health Ministers’ Advisory Council (2008) Population Based Screening Framework. Canberra:
Commonwealth of Australia.
Azemati, H., Belinski, M., Gillette, R., Liebman, J., et al. (2013) Social impact bonds: lessons learned so
far, Community Development Investment Review. Available at http://www.frbsf.org/community-develop-
ment/files/social-impact-bonds-lessons-learned.pdf (Last accessed 9 June 2014).
Battellino, R. (2007) Australia’s experience with financial deregulation. Address to China Australia
Governance Programme, Melbourne, 16 July. Available at http://www.rba.gov.au/speeches/2007/sp-dg-
160707.html (Last accessed 1 November 2014).
Beck, U. (1992) Risk Society: Towards a New Modernity. London: Sage.
Benevolent Society (2013) Information memorandum: the Benevolent Society Social Benefit Trust No. 1.
Available at http://www.dpc.nsw.gov.au/__data/assets/pdf_file/0004/168340/TBS-Information-Memoran-
dum.pdf (Last accessed 20 April 2015).
Bryan, D. and Rafferty, M. (2006) Capitalism with Derivatives. New York: Palgrave Macmillan.
Bryan, D. and Rafferty, M. (2009) Financialisation and Marx: giving labor and capital and financial
makeover, Review of Radical Political Economy, 41, 4, 458–72.
Cantor, J., Mikkelsen, L., Simmons, B. and Waters, R. (2013) How we can pay for a healthy population? Innova-
tive ways to redirect funds to community prevention. Oakland: Prevention Institute. Available at http://
www.preventioninstitute.org/component/jlibrary/article/id-332/127.html (Last accessed 1 May 2015).
Cazdyn, E. (2012) The Already Dead: The New Time of Politics, Culture, and Illness. Durham: Duke
University Press.
Clarke, A.E., Shim, J.K., Mamo, L., Fosket, J.R., et al. (2003) Biomedicalization: technoscientific trans-
formations of health, illness, and US biomedicine, American Sociological Review, 68, 2, 161–94.
Cooper, M. (2008) Life as Surplus: Biotechnology and Capitalism in the Neoliberal Era. Seattle: Univer-
sity of Washington Press.
Cooper, M. and Walby, C. (2014) Clinical Labor: Tissue Donors and Research Subjects in the Global
Bioeconomy. Durham: Duke University Press.
Dara, M., Gushulak, B., Posey, D., Zellweger, J., et al. (2013) The history and evolution of immigration
medical screening for tuberculosis, Expert Review of Anti-infective Therapy, 11, 2, 137–46.
Dardot, P. and Laval, C. (2013). The New Way of the World: On Neoliberal Society (G. Elliot, trans).
London: Verso.
Duster, T. (2007) Medicalisation of race, The Lancet, 369, 9562, 702–3.
Ewald, F. (1991) Insurance and risk. In Burchell, G., Gordon, C. and Miller, P. (eds) The Foucault
Effect. Sydney: Harvester Wheatsheaf.
Eyers, J. (2014) NSW social impact bond returns 7.5 per cent. Sydney Morning Herald. 31 August.
Available at http://www.smh.com.au/business/nsw-social-impact-bond-returns-75-per-cent-20140831-10a
jsa.html (Last accessed 27 May 2014)
Foucault, M. (2007) Security, Territory, Population. Basingstoke: Palgrave Macmillan.
Giddens, A. (1990) The Consequences of Modernity. Cambridge: Polity Press.
Giddens, A. (1991) Modernity and Self-identity. Cambridge: Polity Press.
Gillespe, C. (2012) The experience of risk as ‘measured vulnerability’: health screening and lay uses of
numerical risk, Sociology of Health & Illness, 34, 2, 192–207.
© 2016 Foundation for the Sociology of Health & Illness
 Public health meets finance in ‘health impact bonds’ 1215

Greaves, D. (2000) The creation of partial patients, Cambridge Quarterly of Healthcare Ethics, 9, 1, 23–
37.
Hems, L. (2009) Social impact bonds: the potential for social enterprise. Available at http://
frenzy.fraynework.com.au/socialtraders2/opinion-pieces (Last accessed 1 June 2014).
Jessop, B. (2007) New Labour or the normalization of neo-liberalism, British Politics, 2, 3, 282–8.
Kannel, W.B., Dawber, T.R., Kagan, A., Revotskie, N., et al. (1961) Factors of risk in the development
of coronary heart disease – six year follow-up experience. The Framingham Study, Annals of Internal
Medicine, 55, 1, 33–50.
Kenney, K. (2015) The biopolitics of global health: life and death in neoliberal times, Journal of Sociol-
ogy, 51, 1, 9–27.
KPMG (2012) Investing For Social Success. Phase One: Feasibility Study and Concept Design. Decem-
ber. Wellington: KPMG.
KPMG (2014) Evaluation of the joint development phase of the NSW Social Benefit Bonds trial. January.
Sydney: KPMG. Available at http://www.benevolent.org.au/~/media/Benevolent/About/KPMG_Evalua-
tion_SBB_JDP_Final_Report_290114%20pdf.ashx (Last accessed 1 May 2015).
Kreiner, M. and Hunt, L. (2014) The pursuit of preventative care for chronic illness: turning health peo-
ple into chronic patients, Sociology of Health & Illness, 36, 6, 870–84.
Lakoff, A. (2008) The generic biothreat, or, how we became unprepared, Cultural Anthropology, 23, 3,
399–428.
Langley, P. (2008) The Everyday Life of Global Finance. Oxford: Oxford University Press.
Lapavitsas, C. (2011) Theorizing financialisation, Work, Employment, Society, 24, 4, 611–26.
Lilley, S. and Papadopoulos, D. (2014) Material returns: cultures of valuation, biofinancialisation and the
autonomy of politics, Sociology, 48, 5, 972–88.
Lupton, D. (ed) (1999) Risk and Sociocultural Theory: New Directions and Perspectives. Cambridge:
Cambridge University Press.
Martin, R. (2007) An Empire of Indifference. American War and the Financial Logic of Risk Manage-
ment. Durham: Duke University Press.
Mitropoulos, A. and Bryan, D. (2013) The futures trading of Social Benefit Bonds. Available at http://
s0metim3s.com/2012/11/25/sbbonds/ (Last accessed 1 February 2015).
Mitropoulos, A. and Bryan, D. (2015) ‘Social benefit bonds: Financial markets inside the state’ in Mar-
kets, rights and power in Australian social policy, eds. G. Meagher & S. Goodwin, Sydney: Sydney
University Press.
Nettleton, S. (2006) ‘I just want permission to be ill’: towards a sociology of medically unexplained
symptoms, Social Science and Medicine, 62, 5, 1167–78.
New South Wales Government 2015. Statement of opportunities: NSW Social Impact Investment Policy.
Available at http://www.dpc.nsw.gov.au/__data/assets/pdf_file/0003/168339/Statement_of_Opportuni-
ties_2015_WEB.pdf (Last accessed 20 April 2015).
Nicholls, A. and Tomkinson, E. (2013) Case study: the Peterborough Pilot Social Impact Bond. Oxford:
Oxford University. Available at https://emmatomkinson.files.wordpress.com/2013/06/case-study-the-
peterborough-pilot-social-impact-bond-oct-2013.pdf (Last accessed 1 May 2015)
Papadopoulos, D., Stephenson, N. and Tsianos, V. (2008) Escape Routes: Control and Subversion in the
21st Century. London: Pluto Press.
Patton, C. (2002) Globalizing AIDS. Minneapolis: Minnesota University Press.
Philipson, R. (2011) The potential for social impact bonds In New Zealand. Available at http://tinyurl.
com/m7oofn7 (Last accessed 1 November 2014).
Reynolds Whyte, S. (2012) Chronicity and control: framing ‘non-communicable’ diseases in Africa,
Anthropology and Medicine, 19, 1, 63–74.
Rockhill, B. (2001) The privatization of risk, American Journal of Public Health, 91, 3, 365–68.
Rose, S. (2014) Wait continues for social benefit bonds decision. Sydney Morning Herald. 17 September.
Available at http://www.smh.com.au/business/the-economy/wait-continues-for-social-benefit-bonds-deci-
sion-20140916-10hxw1.html (Last accessed 1 May 2015).

© 2016 Foundation for the Sociology of Health & Illness

1216 Rachel Rowe and Niamh Stephenson

Samson, O. (2013) Planning of a social impact bond to reduce development of type 2 diabetes in high-
risk pre-diabetics. Presentation to the G8 Social Impact Investment Taskforce. 5 December. Tel Aviv:
Social Finance. Available at http://tinyurl.com/kwjkwg3 (Last accessed 15 May 2015).
Shiell, A., Hawe, P. and Jan, S. (2013) The economic value of prevention, state of preventative health
2013. Available at http://anpha.gov.au/internet/anpha/publishing.nsf/Content/state-of-prev-health-2013#
footnote-backlink-e-55 (Last accessed 20 March 2015).
Social Finance (2012) A new tool for scaling impact: how social impact bonds can mobilize private capi-
tal for social good. Boston: Social Finance, Inc. Available at http://www.socialfinanceus.org/pubs/new-
tool-scaling-impact-how-social-impact-bonds-can-mobilize-private-capital-advance-social-good (Last ac-
cessed 1 February 2015).
Social Finance (2013) The California Endowment Awards grant to Social Finance and Collective Health. 25
March. Available at http://www.socialfinanceus.org/california-endowment-awards-grant-social-finance-and-
collective-health (Last accessed 1 May 2015).
Social Finance (2014) Peterborough Social Impact Bond reduces reoffending by 8.4%; investors on
course for payment In 2016. 7 August. Available at http://www.rockefellerfoundation.org/uploads/files/
a5de37d9-f46d-40b8-859c-2dcbdbc6098f-peterborough.pdf (Last accessed 1 November 2014).
Social Impact Investment Taskforce (2014) Impact investment: the invisible heart of markets. Available
at http://www.socialimpactinvestment.org/reports/Impact%20Investment%20Report%20FINAL%5B3%
5D.pdf (Last accessed 1 April 2015).
Social Ventures Australia (2014) About Newpin Social Benefit Bond. Available at http://www.commu-
nity.nsw.gov.au/docswr/_assets/main/documents/sva_newpin_factsheet.pdf (Last accessed 1 January 2014).
Strange, S. (1988) States and Markets. 2nd edn. New York: Bloomsbury.
Streeck, W. (2013) The Politics of Public Debt: Neoliberalism, Capitalist Development and the Restruc-
turing of the State, German Economic Review, 15, 1, 143–65.
Tomkinson, E. (2015) Malaysian innovation: building a social impact bond (SIB) pipeline, 3 February, Emma
Tomkinson: a stream of social consciousness (blog). Available at http://emmatomkinson.com/2015/02/
03/malaysian-innovation-building-a-social-impact-bond-sib-pipeline/ (Last accessed 1 March 2015).
United Kingdom Ministry of Justice (2013) Story of the prison population 1993–2002 England and Wales.
Available online at: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/21818
5/story-prison-population.pdf (Last accessed 1 November 2014).
United Kingdom Ministry of Justice (2014) Peterborough Social Impact Bond HMP Doncaster. Payment
by results pilot. Final reconviction statistics for cohort 1. 7 August. Available at https://www.
gov.uk/government/uploads/system/uploads/attachment_data/file/341682/pbr-pilots-cohort-1-results.pdf
(Last accessed 14 August 2014).
Wacquant, L. (2009) Punishing The Poor: The Neoliberal Government Of Social Insecurity. Durham:
Duke University Press.
Wallace, A., Hutley, N. and Harris Roxas, B. (2014) Evaluation of the Newpin SBB Program: Final
Evaluation Framework. Sydney: Urbis.

© 2016 Foundation for the Sociology of Health & Illness