social impact bond case

21
KS1003 Case Number 2026.0 This case was written by Pamela Varley, Senior Case Writer, for Dan Levy, Senior Lecturer in Public Policy, for use at the John F. Kennedy School of Government (HKS), Harvard University. HKS cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective manage- ment. (October 2013) Copyright © 2014 President and Fellows of Harvard College. No part of this publication may be reproduced, revised, translated, stored in a retrieval system, used in a spreadsheet, or transmitted in any form or by any means without the express written consent of the Case Program. For orders and copyright permission information, please visit our website at http://www.case.hks.harvard.edu/ or send a written request to Case Program, John F. Kennedy School of Government, Harvard University, 79 John F. Kennedy Street, Cambridge, MA 02138. Betting Private Capital on Fixing Public Ills: Instiglio Brings Social Impact Bonds to Colombia In July 2012, three freshly minted Harvard Kennedy School graduates—founding directors of a nonprofit startup called “Instiglio”—bought one-way tickets to Medellin, Colombia, gambling that in this city, they would make history with an ambitious new venture. The idea: to try out Social Impact Bonds, still experimental even in affluent nations, in Colombia and other low to middle-income countries. The SIB concept was, at the time, creating a buzz in policy circles. Audacious in introducing a new financing mechanism for experimental social programs—with private investors shouldering the financial risk of failure—SIB proponents hoped they had found a way to reverse a disappointing pattern, in which governments funded ineffec- tive programs year after year, but shied away from innovative experiments that might yield better results. For the public sector, SIBs offered a remarkable guarantee. If a SIB-financed social program failed to meet its performance targets, the government did not have to pay for it. Private investors fronted the cost and, when a program failed, absorbed the loss. If the program met or exceeded its performance targets, however, the govern- ment not only had to repay investors for the cost of the program, it also had to pay them a return on investment; the more successful the program, the higher the return. In this and other respects, SIB projects required govern- ment agencies to depart from their traditional procurement and budgeting norms. Even in the resource-rich envi- ronment of the United Kingdom and the United States, it had proved challenging to negotiate SIB contracts—and might prove more so in a less affluent country. But if it succeeded as its inventors hoped, the SIB model would achieve something extraordinary, bringing the power of market incentives to the pursuit of social aims. As the Instiglio founders headed to Medellin, they were excited by this potential—and by the thought of bringing it to Colombia, a rising star in the developing world. Social Impact Bonds: A Primer The world’s first SIB project, in Peterborough, England, was launched in 2010, a time of global recession. “Governments are stretched, charitable donations have fallen in the wake of the recession, returns on traditional investments are pitiful, and the capital markets are still confronting an image problem in the wake of the financial crisis,” summarized one journalist in October 2011. 1 With governments and investors casting around for new 1 “Social impact bonds target greater good; By combining societal goals with business acumen, pioneering class of investment asset aims to fill gaps charities, government can’t,” by Tara Perkins, The Globe & Mail, October 31, 2011. For the exclusive use of G. Gray, 2015. This document is authorized for use only by George Gray in 2015.

Upload: test123

Post on 04-Jan-2016

5 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: Social Impact Bond Case

KS1003

Case Number 2026.0

This case was written by Pamela Varley, Senior Case Writer, for Dan Levy, Senior Lecturer in Public Policy, for use at the John F. Kennedy School of Government (HKS), Harvard University. HKS cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective manage-ment. (October 2013)

Copyright © 2014 President and Fellows of Harvard College. No part of this publication may be reproduced, revised, translated,

stored in a retrieval system, used in a spreadsheet, or transmitted in any form or by any means without the express written

consent of the Case Program. For orders and copyright permission information, please visit our website at

http://www.case.hks.harvard.edu/ or send a written request to Case Program, John F. Kennedy School of Government, Harvard

University, 79 John F. Kennedy Street, Cambridge, MA 02138.

Betting Private Capital on Fixing Public Ills:

Instiglio Brings Social Impact Bonds to Colombia

In July 2012, three freshly minted Harvard Kennedy School graduates—founding directors of a nonprofit

startup called “Instiglio”—bought one-way tickets to Medellin, Colombia, gambling that in this city, they would

make history with an ambitious new venture. The idea: to try out Social Impact Bonds, still experimental even in

affluent nations, in Colombia and other low to middle-income countries.

The SIB concept was, at the time, creating a buzz in policy circles. Audacious in introducing a new financing

mechanism for experimental social programs—with private investors shouldering the financial risk of failure—SIB

proponents hoped they had found a way to reverse a disappointing pattern, in which governments funded ineffec-

tive programs year after year, but shied away from innovative experiments that might yield better results.

For the public sector, SIBs offered a remarkable guarantee. If a SIB-financed social program failed to meet its

performance targets, the government did not have to pay for it. Private investors fronted the cost and, when a

program failed, absorbed the loss. If the program met or exceeded its performance targets, however, the govern-

ment not only had to repay investors for the cost of the program, it also had to pay them a return on investment;

the more successful the program, the higher the return. In this and other respects, SIB projects required govern-

ment agencies to depart from their traditional procurement and budgeting norms. Even in the resource-rich envi-

ronment of the United Kingdom and the United States, it had proved challenging to negotiate SIB contracts—and

might prove more so in a less affluent country.

But if it succeeded as its inventors hoped, the SIB model would achieve something extraordinary, bringing the

power of market incentives to the pursuit of social aims. As the Instiglio founders headed to Medellin, they were

excited by this potential—and by the thought of bringing it to Colombia, a rising star in the developing world.

Social Impact Bonds: A Primer

The world’s first SIB project, in Peterborough, England, was launched in 2010, a time of global recession.

“Governments are stretched, charitable donations have fallen in the wake of the recession, returns on traditional

investments are pitiful, and the capital markets are still confronting an image problem in the wake of the financial

crisis,” summarized one journalist in October 2011.1

With governments and investors casting around for new

1

“Social impact bonds target greater good; By combining societal goals with business acumen, pioneering class of investment asset aims to fill gaps charities, government can’t,” by Tara Perkins, The Globe & Mail, October 31, 2011.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 2: Social Impact Bond Case

HKS Case Program 2 of 21 Case Number 2026.0

models, she noted, the SIB launch was well-timed. But the ideas at the heart of the SIB model grew from frustra-

tions that dated back many years. “The way our governments currently fund social service programs produces

inadequate performance and insufficient innovation,” said Jeffrey Liebman, an HKS economist and public policy

professor, with a special research interest in SIBs.2

These failures carried high and well-recognized social and fi-

nancial costs. Those who fell through society’s safety net often ended up in expensive, publicly-funded facilities

like homeless shelters, detox centers, or prisons.

In an effort to improve the efficacy of their social programs, some elected officials had tried in recent years to

shift to results-based contracting. The idea was to pay a service provider for achieving positive outcomes for its

clients rather than to pay for the delivery of services, per se. But in the social service realm, it was often difficult to

fairly assess a program’s performance. For one thing, different service providers tended to address different as-

pects of a social problem; thus it was hard to hold any one of them accountable for achieving the desired outcome.

In addition, many government-mandated performance evaluations measured success over the course of a single

year—often too short a timeframe to be meaningful. Also, most public sector contracts were proscriptive, allow-

ing little leeway for the provider to make programmatic adjustments. But there was an even more fundamental

problem. The idea behind the results-based model was, generally, to shift the risk of failure from the contracting

agency to the contractor. Ideally, the contractor fronted the cost of the work and was repaid if successful. But

most social service providers were nonprofit organizations that lacked the resources to do this. As a result, so-

called results-based contracts in this sector tended to provide relatively weak incentives—for instance, combining

a traditional cost-based contract with a modest bonus for a successful outcome.

Without clear, actionable information about the efficacy of their social programs, and with no confidence that

alternative approaches would be any better, governments tended to keep the programs they had, but not to add

anything new to the budget. The SIB model was designed to shake up this stalemate with an injection of private

money. Because investors fronted a SIB program’s cost and absorbed the loss if it failed, their risk was quite high.

Not only might they receive no return on investment, they might lose their investment capital altogether.3

To

make such a risk palatable, each SIB project would have to be carefully designed to give it every chance at success.

In theory, only the most promising programmatic approach would be selected for a SIB project, and only the most

proven and capable service providers would be invited to participate. The program contract would extend not just

one year, but several years. To address the many obstacles to success, the program would likely include a suite of

services, delivered by different specialized providers, but these would be overseen by a single social service con-

tractor, which would bear responsibility for coordinating the project and meeting the performance targets. These

performance targets would be set at a high level—but an attainable one, based on careful examination of existing

programs. Along the way, the program would be closely monitored for effectiveness, and adjusted as needed. A

2

“Social Impact Bonds/A Promising New Financing Model to Accelerate Social Innovation and Improve Government Perfor-mance,” by Jeffrey B. Liebman, Center for American Progress/Doing What Works, February 2011, http://www.americanprogress.org/wp-content/uploads/issues/2011/02/pdf/social_impact_bonds.pdf, retrieved September 20, 2013, p. 29. 3

For this reason, some observers have objected to the term “social impact bond” as a misnomer. In a traditional bond, inves-tors do not risk their capital.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 3: Social Impact Bond Case

HKS Case Program 3 of 21 Case Number 2026.0

financial intermediary—Liebman called it a SIBIO, or social impact bond-issuing organization—would broker the

deal: conducting due diligence, negotiating the contract, hiring the service provider, securing the financing, over-

seeing the funds, and, in general, managing the process. [See Exhibit 1 for process flow chart & Exhibit 2 for mon-

ey flow chart.]

This careful approach to program design would mitigate the financial risk but, SIB creators recognized, not

enough to entice most investors to the table unless they stood to earn a healthy return on investment for the pro-

grams that succeeded.4

This, in turn, made a successful SIB-financed project significantly more expensive than a

traditional government contract in absolute dollars, though—its creators argued—potentially less expensive per

unit of outcome.5

But SIB proponents readily acknowledged that in order to be worth the extra cost, the program

outcomes would have to be important to political leaders, both in their potential to deliver social benefits and in

their potential to avert higher government costs down the line. The earliest experiments in SIB contracts reflected

this principle. Pilot projects in the United Kingdom and in the United States involved programs to reduce the recid-

ivism rate of prison inmates, due for imminent release. If successful, the respective governments hoped they

would be able to reduce their long-term prison costs—especially if the inmate population shrank enough to allow

prison closures. [See Exhibit 3 for description/comparison of these two early SIB projects.]

The Instiglio Story

When five Kennedy School students6

came up with the idea for Instiglio in 2012, none of the early SIB experi-

ments had gone through an entire contract cycle, but the students were excited by the ideas at the core of the

model. They submitted their proposal—a SIBIO called Instiglio that would broker SIB projects in developing coun-

tries—to Harvard’s first annual “Public Sector Innovations” contest.7

In the spring of 2012, the Instiglio proposal

received the contest’s top prize—$10,000 and “incubation space” for further development.

Before receiving the award, all five students in the Instiglio group had been making other post-graduate plans.

Suddenly, they found themselves on the horns of a dilemma. To turn their proposal into reality meant gambling

that a smart but relatively young and inexperienced group could take an exciting but unproven idea to a challeng-

ing setting and make it work; that had “high risk” written all over it. But in the process of developing the proposal,

4

Though most of the early experiments revolved around investment in a single project, SIB creators recognized that only a very small number of investors would be willing to risk losing all their investment capital. To cushion the risk, they thought investors would prefer to invest in portfolios of SIB projects. Liebman noted that, even if two-thirds of SIB projects in a given investment portfolio succeeded (which he regarded as an optimistic scenario), these successful programs would have to yield a high return to compensate for the capital loss in the other third. To give the overall portfolio even a modest 5 percent annualized return, he argued, would require a 20 percent annualized return on the successful programs. A more solid 15 percent annualized re-turn would require a 30 percent annualized return on the successful programs. (“Social Impact Bonds,” pp. 18-19.) 5

Of course, if, at the end of the SIB contract, a program was deemed a success, the government would be free to enter into a more traditional government contract to continue it, going forward. 6

The original five were Michael Eddy, Michael Belinsky, Avnish Gungadurdoss, Madalina Pruna and David Bullon-Patton. 7

Harvard’s “i3” award for Public Sector Innovations was one of a suite of innovation contests/awards available to Harvard de-gree candidates in good standing. ( http://www.harvardi3.org/content/public-sector-innovation-award-presented-accenture, retrieved Sept. 16, 2013.)

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 4: Social Impact Bond Case

HKS Case Program 4 of 21 Case Number 2026.0

the students had grown increasingly attached to it. Winning the contest was a strong external endorsement. In

the end, three of the original five decided they could not pass up the chance to try it. The trio—Mike Belinsky

(MPP 2012), Avnish Gungadurdoss (MPA/ID 2012) and Michael Eddy (MPA/ID 2012)—became the founding direc-

tors of Instiglio and poured their energy into turning the brainchild into a reality, with much help along the way

from fellow students.8

Choosing Colombia

Their first step was to pick a country where they could try out their all-important first SIB project. Given the

complexity of the SIB model, the need to attract investors, and the “good government” spirit of SIBs, they looked

for a stable country with a credible government; well-placed, forward-looking political leaders; and proven service

providers. These criteria narrowed the range of possibilities by quite a bit. In the end, the Instiglio team chose Co-

lombia because, through HKS connections, they were able to secure an audience with Sergio Fajardo, the charis-

matic, progressive-minded governor of the state of Antioquia.9

Fajardo was a one-time mathematics professor

who had received his graduate training at the University of Wisconsin-Madison. Here, the Instiglio founders

thought, was as promising a political champion as they were likely to find. What’s more, he was prepared to meet

with them personally and listen to their proposal. Given the novelty of the SIB approach, that entrée was critically

important, observed Eddy, who would become the coordinator of Instiglio’s operation in Colombia: “In addition to

designing a new intervention, we were also selling rigorous impact evaluation, and then on top of that we were

selling the financing mechanism. So… there’s a lot to buy here.”10

In fact, Instiglio’s meeting with Fajardo in July 2012 went better than the founders had even dared hope. Fa-

jardo invited them to his home, where they met not only with him but—to their surprise—with his Secretary of

Planning, María Eugenia Ramos. Also in attendance were his chief of staff and three of his top personal advisors.

The Governor immediately understood and liked the SIB idea and asked Ramos to serve as the point-person for the

project. His own platform emphasized education. While he did not limit Instiglio to thinking about education, he

wanted the SIB project to align with the priorities of his administration, articulated in a development plan. The

Instiglio team readily agreed.

8

Within a few months, the Instiglio group would more than double in size, with the addition of Johannes Lohmann as Chief Financial Officer, Siegrid Holler as Country Coordinator, Julianna Arbeláez Jiménez as Public Relations Coordinator, and Lina Lorenzoni as Legal Consultant. Ilse Geyskens, who gave Instiglio its first significant donation in the fall of 2012, became chair of Instiglio’s Board of Directors in December 2012.

9 A few months later, in February 2013, the Wall Street Journal, Citi, and Urban Land Institute would name Antioquia’s capital

city of Medellin as “world capital of innovation” based on environment and land use; culture and livability; econom-ic/investment climate; progress and potential; places of power, education and human capital; technology and research; and mobility and infrastructure. (“Which Cities are the World’s Most Innovative?” by Trisha Riggs, Michael Mehaffy, Bendix Ander-son, and Leslie Braunstein, UrbanLand, The Magazine of the Urban Land Institute, March 1, 2013, http://urbanland.uli.org/economy-markets-trends/which-cities-are-worlds-most-innovative-winner/, retrieved May 19, 2014.)

10 Author phone interview with Eddy, April 26, 2013. Unless noted, subsequent quotations from and attributions to Eddy are

from this interview.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 5: Social Impact Bond Case

HKS Case Program 5 of 21 Case Number 2026.0

Given the newness of the SIB idea, the Instiglio group thought it risky to set up a SIB project that extended be-

yond the end of the Fajardo administration in December 2015. Thus, they decided, the project must be complet-

ed, evaluated, and, if successful, paid off before this date. This was a constraint, but the team did not believe it

insurmountable. Under their timeline, the organization would spend August selecting a project topic, September

and October designing it, November and December setting it up. Operations would begin in January 2013. Work-

ing closely with Ramos, the three immediately began to meet with key players across the administration to explain

the SIB idea, to brainstorm ideas for a good first project, and to talk through potential concerns.

A. Identifying a Promising Project for Antioquia

SIB creators had recognized from the start that, in any setting, it was important that a SIB project address an

issue of top political concern, in order to obtain the necessary buy-in from government. For one thing, govern-

ment had to be willing to pay a significant premium for a program success. For another, establishing a SIB contract

required the sustained involvement of several government agencies. In affluent countries, the cost-benefit calcu-

lation was likely to provide an additional motivation—in particular, the chance to reduce higher costs elsewhere in

government, as in the recidivism/prison example. In a developing country, this rationale might be present to some

degree, but with a less robust social safety net and less costly institutions, it was less likely that a SIB project would

deliver realizable cost savings in the short term. It should, though, deliver significant social benefits. For this rea-

son, the Instiglio team thought it crucial that Fajardo and his advisors select a project in which the social benefits

mattered deeply to them.

In developing a list of potential projects, the Instiglio team began by scouring the Governor’s 2012-2015 de-

velopment plans and soliciting project ideas from members of the Fajardo administration. They followed up by

researching each idea. From the start, they understood that data collection was likely to be less robust and user-

friendly in Antioquia than in the United States, but they nonetheless tried to find several key pieces of information

for each potential project, to see whether it met the criteria for a good SIB:

Scope of the problem/political priority?

Easy to identify target and control populations?11

An existing, promising program design to address this problem?

Local service providers who could successfully deliver this program?

Attainable outcome measures, aligned with program goals?12

11

It was important that the target and control groups be scrupulously comparable. Thus, for example, it would not be legiti-mate to compare the recidivism rate of prisoners who volunteered to participate in a social program with the general prison population. Both the Peterborough and New York City projects [see Exhibit 3] targeted an entire class of prisoners at a given prison, and each compared its outcomes with like prisoners elsewhere in the system. Thus, they could not be accused of “cher-ry-picking.” 12

Liebman stressed that the outcome measure must be “highly correlated with a program’s comprehensive social net bene-fits…. When measures are only weakly correlated with program success, or when only one component of a program’s impact

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 6: Social Impact Bond Case

HKS Case Program 6 of 21 Case Number 2026.0

To the extent possible, evidence of cost-effectiveness in combined social benefits and averted costs

elsewhere in the system?

This stage of the process proved more difficult than the group had anticipated. With Ramos’ active and en-

thusiastic help, the Instiglio team did find it easy to get on the calendar of social service experts and providers, and

with managers at all levels in the Fajardo Ministries of Health, Education, Gender Equity, Planning, and Finance.

These meetings were generally positive in tone, but the Instiglio group gradually came to realize that cordiality did

not necessarily imply full-hearted support, especially from the ranks of the public sector. Government managers

expressed two general concerns. First, Colombia had developed strict public procurement laws to guard against

corruption, and the SIB model—which required uncertain payments in the future rather than a defined payment at

the beginning—was a novelty. The Instiglio directors were aware of this, of course. Early on, they had met with a

well-placed Colombian lawyer who assured them that this mechanism would be possible under the existing legal

framework, but would require legislative approval. This reassured them that the problem would not be insur-

mountable. 13

But early on, they were not always able to persuade state officials that the legal issue would not,

ultimately, stop the project before it got off the ground.

Second, Colombia did not have a strong tradition of contracting out public services to nonprofit service organi-

zations. Most public services were provided by government agencies. Some Antioquia managers were therefore

concerned that the SIB/NGO projects would, effectively, become competitors to public sector programs and,

through their outcome evaluations, serve to embarrass the government. The Instiglio team tried to defuse these

anxieties by stressing that the programs were designed to complement—not compete—with existing government

programs, but they were unsure whether their arguments had been persuasive.

The bottom line was that for several reasons—lack of appropriate or readily accessible data, skepticism about

whether the SIB model was possible under the existing legal framework, concern that the SIB model posed a threat

to existing programs, or simple bureaucratic inertia—the Instiglio team found it difficult to get some of the basic

information they needed from government agencies in order to fully weigh and compare the program ideas sug-

gested to them. As they repeated their requests for data, it soon became evident that their ambitious idea of se-

curing agreement about a project by the end of August would not be possible. By October, however, the Instiglio

group decided that even if their information was incomplete, and even if none of the projects met all the criteria

for a good SIB project, they needed to move forward. They submitted a short list of four possible projects—all

high political priorities—to the Governor:

1. Preventing School Dropouts. Antioquia had a well-recognized problem with school dropouts. By the last two

years of high school, only 40 percent of eligible children were still attending school. Instiglio discovered that

can be measured, performance contracts based on the imperfect measure have the potential to distort performance toward that which can be measured.” (“Social Impact Bonds,” p. 21.) The classic example, in education, was standardized testing and the much-bemoaned problem of schools that ignored important parts of education in order to boost test scores—so-called “teaching to the test.” 13

This was also a problem in more affluent countries, and was handled in various ways. For instance, in Massachusetts, the money to pay for the program was placed in a special earmarked fund, so that a future administration would not have to make a new appropriation.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 7: Social Impact Bond Case

HKS Case Program 7 of 21 Case Number 2026.0

the United Nations Children’s Fund (UNICEF) operated a labor-intensive program in Antioquia called La Escuela

Busca al Niño, or the School Goes to the Students, which attempted to re-engage school dropouts who were,

by this point, involved in all kinds of other things—criminal gangs, narco-trafficking, illegal mining, child labor,

etc.—and persuade them to return to schooling in some form. It might be possible to expand this approach,

but finding a well-defined target/control population and tracking them would not be easy. Furthermore, be-

cause this program targeted students who had already left school and—in many cases—embarked on high-risk

activities, extensive outreach and rehabilitation work would be necessary to securely re-establish them in

school. Though Instiglio could find little conclusive data, such programs were “rumored to be” expensive, with

little clear evidence of success, according to Eddy. Alternatively, Dividendo por Colombia, a Colombian NGO

affiliated with the United Way, operated a program called Aceleración del Aprendizaje, aimed at primary

school students who were at least two years behind grade level—a group at high risk for dropping out. The

goal of this program was to make a focused effort to bring these students up to grade level. Though there was

limited reliable evidence to show that Aceleración was successful, the program had a generally positive repu-

tation by word-of-mouth.

2. Vocational Training for Youth. Antioquia’s unemployment rate for young people in the work force was 19.2

percent—dangerously high, as it pushed these vulnerable youths toward the underground economy and, in

particular, the country’s extensive criminal drug trade. The reason for the high rates, according to govern-

ment officials, was primarily a matter of poor job skills, probably combined with discrimination by employers.

A program implemented in 2002 attempted to get unemployed youth into the job market through a combina-

tion of classroom and on-the-job training conducted by accredited private training institutions. A rigorous

program evaluation, conducted by a respected organization, showed an employment increase of 5 to 7 per-

cent and significant long-term increases in women’s earnings. Despite these indications of success, the pro-

gram was later discontinued and the private providers were replaced by a large government-run training

agency, which had delivered far more mixed results, Eddy said. While it was tempting to consider using a SIB

to re-launch a program along the lines of the 2002 model, there was a catch. The 2002 program had been im-

plemented in Antioquia’s urban center, Medellin, which was relatively well-served by social programs. The

Governor was keen to try a job training program in the under-served countryside—and that would be far more

challenging. For any given unemployed rural worker, there would be far fewer employers and far fewer job

openings in commuting distance than for his or her urban counterpart. It would also be harder to match train-

ing to existing employment needs.

3. Prevention of Adolescent Pregnancy. Another top issue for the Fajardo administration was the issue of ado-

lescent pregnancy. Instiglio learned that roughly one out of every five girls aged 15 to 19 in Colombia was ei-

ther pregnant or had a child. In Medellin, each year more than 7,000 teenagers gave birth costing the gov-

ernment approximately at least $2.8 million. Both in Colombia and internationally, the track record of preg-

nancy-prevention programs was quite mixed overall. [See Exhibit 4 for a sample summary of 13 evaluations.]

In Colombia, however, a promising recent experiment, evaluated via a randomized control trial,14

showed a 13

percent reduction in unprotected sexual activity for participants of a semester-long school-based health edu-

14

In a randomized control trial, evaluators collected an outcome measure, thought to reflect the impact of the program, for both a target population that received program services and a control population that did not.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 8: Social Impact Bond Case

HKS Case Program 8 of 21 Case Number 2026.0

cation program. 15

Internationally, Instiglio found that the most successful programs tended to be group-based

and to target youngsters who were just becoming sexually active. In addition, they tended to combine health

education with information about how to obtain contraceptives, “life skills” discussions about decision-making

and goal-setting, and—for those at high risk—individual counseling. It therefore seemed possible to create a

SIB project that would add such services to the already successful health education model. The project might

run into political difficulties, however, as nearly a quarter of Medellin schools for seventh and eighth graders

had already invested in their own in-house sex education programs, run with their own personnel.

4. Prevention of Adolescent Delinquency. Perhaps the single biggest social problem in Antioquia was that of

adolescent delinquency, especially in rural areas. Many young people, especially in regions of the state desta-

bilized by territorial conflict, were actively recruited into criminal gangs. The underlying causes of the problem

were complicated. Many families had been displaced by the conflict, and were consequently struggling. In

addition, there was a well-entrenched illegal drug trade in these areas. If vulnerable adolescents could be re-

directed away from a life of crime, the country would likely save money in prison costs (over US $8000 per

prisoner per year), as well as the social benefit of greater social stability and productivity. But the youths were

dispersed and it was not immediately obvious how to identify a target/control population. Also, the evalua-

tions of previous or existing programs tended to be poor in quality, showed mixed results, or both, according

to Eddy. Nor was Instiglio able to find evidence of programs internationally that had proved particularly effec-

tive at decreasing delinquency in the midst of unrest. Still, for Antioquia government officials, finding a way to

address this problem was an absolute top political priority.

B. Proposing an Adolescent Pregnancy SIB Project in Medellin

In October of 2012—while continuing to work with Governor Fajardo and other Antioquia leaders to settle on

a good project to try at the state level—Instiglio also entered into a separate set of talks with the city government

of Medellin and its top elected official, Mayor Aníbal Gaviria. The city talks moved more quickly than those at the

state level, partly because, after brief deliberations, Mayor Gaviria expressed a strong preference for a SIB project

to reduce teen pregnancy. In December 2012, Instiglio enlisted the help of two service providers with extensive

knowledge of both local teen pregnancy and social service delivery in Medellin—Profamilia, a Colombian affiliate

of Planned Parenthood, and Bienestar Humano, a longstanding Medellin-based social service agency—to put to-

gether a proposal.

In developing this SIB proposal for Medellin, Instiglio had to make a number of critical decisions:

Deciding on the nature of the intervention.

Identifying the target group.

Identifying the control group.

Deciding on the length of the contract.

Selecting the best attainable outcome measure.

15

“Effectiveness and Spillovers of Online Sex Education: Evidence from a Randomized Evaluation in Colombian Public Schools,” by Alberto Chong, Marco Gonzalez-Navarro, Dean Karlan, Martin Valdivia, National Bureau of Economic Research, NBER Working Paper No. 18776, February 2013, http://www.nber.org/papers/w18776, retrieved October 29, 2013.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 9: Social Impact Bond Case

HKS Case Program 9 of 21 Case Number 2026.0

Choosing the service providers.

Defining success under the contract.

Deciding how much capital/profit to return to the investors.

Nature of the Intervention

Instiglio began by studying what Medellin had done, to date, in the area of reducing teen pregnancies,

and concluded that there was “a lot of silo-ing.” “We see the Secretary of Education doing a project over here,

the Secretary of Health doing a project over there,” said Eddy. At the same time, a review of hundreds of

pregnancy-prevention studies across the world revealed that the most successful programs tended to share

some common characteristics, according to Eddy. They targeted both boys and girls. They were group-based,

rather than one-on-one, and organized around the goal of risk-reduction—persuading teenagers either to de-

lay sexual activity, or to avoid unprotected sexual activity. Most important, they were comprehensive—that

is, they combined health education; information about how to access contraceptives; life skills (for example,

group discussions about self esteem, decision-making, and long-term planning); and, for kids most at-risk, in-

dividual counseling. [See Exhibit 4 for a summary chart of 13 evaluations.] Instiglio decided to propose a pro-

ject using this comprehensive approach.

Target Group. Studies also showed that prevention was most effective with children just on the verge of

becoming sexually active. In Colombia, that meant kids between the ages of 12 and 15. A program targeting

seventh and eighth graders held the best chance of reaching this group. The Instiglio team calculated that a

target group of 5,000 was optimal. In Medellin, there were 642 schools serving this population. At some of

these schools, the risk of teen pregnancy was relatively low and at others, a pregnancy-prevention program

was already in place (albeit a more limited model than the one Instiglio envisioned). At 100 schools, however,

the teen pregnancy risk was high and the school did not yet have any kind of pregnancy-prevention program.

Each year, about 13,000 students entered the seventh grade at these schools. The Instiglio team predicted

that most of these schools would probably be willing to participate in this SIB experiment—but it was always

possible that some would decline.

Control Group. The next challenge was to try to design an evaluation that could reliably isolate the impact

of the program from the multitude of other factors that might influence rates of unprotected sex during the

evaluation period. They decided on the use of a control group—that is, to compare students in the program

(treatment) schools to students in non-program (comparison) schools. Instiglio would have to propose a

method for selecting the treatment and comparison schools among the 100 potential schools in Medellin.

How Long? Instiglio thought that, with a concerted push, they could get a program up and running by

January 2014 (the beginning of the academic year in Colombia), but deciding on the duration of the program

was harder. The Instiglio team presented two options to the Medellin government. Under one option, the

program would follow students in both treatment and comparison schools for five years. From the point of

view of testing the SIB model, this was the better option, as it would give the program a good chance to make

mid-course corrections and prove itself. Under the second option, the program would follow a single cohort of

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 10: Social Impact Bond Case

HKS Case Program 10 of 21 Case Number 2026.0

students for 18 months. Though much shorter, this timeframe held a political advantage: it would finish within

the period of the Gaviria administration. The mayor and his advisors chose this latter option.

Outcome Measure. The outcome measure was a tricky matter. At first blush, it seemed self-evident that

the outcome measure should be the pregnancy rate. But the program was, by design, targeting young teens.

According to a 2006 study in Medellin, in any given year, 4.3 percent of girls between the ages of 10 and 19

bore a child.16

But just .38 percent bore a child between the ages of 10 and 14, while 8.1 percent bore a child

between the ages of 15 and 19. These figures were quite variable by neighborhood within Medellin. The low

incidence of pregnancy in the early teen years would make it difficult to detect the program’s impact on preg-

nancy, per se. To do so reliably would require a study period of at least five years’ length and a target popula-

tion of much larger size, by Instiglio’s calculation.

Given the time restrictions of the project, the Instiglio directors felt they had to measure something that

could be observed in a shorter time period, but that was still closely linked to adolescent pregnancies. What

made most sense, they concluded, was to measure the percent of students engaging in unprotected sex—the

group “at risk” for pregnancy. Based on the best studies they could find, it appeared likely that, absent a pro-

gram, somewhere between 11 and 28 percent of their target and control kids would fall in the “at risk” group.

The simplest idea was simply to interview students at the beginning and end of the program, in both

treatment and comparison schools, about whether they were engaging in unprotected sex. The challenge,

though, was how to get a reliable response from the students. In general, the Instiglio team knew, there was a

tendency for people to under-report behavior deemed undesirable or socially sensitive (the so-called “social

desirability bias”). Thus, they needed to come up with a method that addressed this well-known problem.

Service Providers. Instiglio proposed to engage two “leading operators” in Colombia to implement the

program—Profamilia and Bienestar Humano, the same organizations that had helped them with the program

design. With 34 offices and 225 clinics throughout Colombia, Profamilia was not only the primary family plan-

ning organization in the country, but was recognized throughout the region. In 2011, Profamilia had served

500,000 clients in Colombia. Under Instiglio’s plan, Profamilia would provide sex education, information about

how to access birth control, and individual counseling. Bienestar Humano, with 79 years of experience in

bringing about “positive behavioral change” in Medellin, had implemented a school-based pregnancy preven-

tion program for more than 1,600 students since 2009. Bienestar would run the life skills part of the program.

Defining Success. The next challenge was to figure out where to set the threshold for program success.

The general hope, of course, was that by the end of the evaluation period, a lower percentage of students in

the treatment population would report engaging in unprotected sex than in the control group—but how much

lower did the percent need to be in order for the program to declare success? The stakes for getting this num-

ber “right” were high. If the bar was not set high enough, then the Medellin government would pay a premi-

16

A nationwide survey of girls between the ages of 15 and 19, conducted in 2005, indicated that 21 percent had ever been pregnant. Nationwide, 9 percent of girls bore a child before the age of 20. Half of all Colombian women bore their first child before the age of 22. Figures drawn from “Evaluación del impacto del programa de Salud Sexual y Reproductiva en Adolescen-tes de Medellín, Fase cuantitativa,” Medellin Health Department, 2008.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 11: Social Impact Bond Case

HKS Case Program 11 of 21 Case Number 2026.0

um price for a program that was not necessarily any better than existing programs. But if the bar was set un-

realistically high, then the program would “fail,” the innovation would die on the vine, and the investors would

lose their capital—a bad outcome all around.

The Instiglio group thus set about gathering every performance evaluation they could find for programs

aimed at reducing teen pregnancy. The results, they found, were exceedingly mixed. [See Exhibit 4 for a sam-

ple summary of 13 evaluations.] A disappointingly large number of programs showed no impact, one way or

another. But Eddy thought the design of the evaluations was often so poor, it was hard to tell whether the

programs were actually ineffective or just badly evaluated. Thus, he said, “One of the most salient questions

[we faced] was what sort of impact could we reliably commit to, given the somewhat spotty nature of the evi-

dence?” But the team did have one important piece of data. One of Profamilia’s programs had undergone a

rigorous independent evaluation. It was a school-based program implemented through a computer-based

module to over 4,500 students across Colombia. The study had shown that the program was effective—

reducing unprotected sex by 13 percent. Moreover, the program was school-based, targeted at a similar popu-

lation and the study measured unprotected sexual activity in the same way that Instiglio proposed measuring

it.17

“Our program was going to be more intensive—and therefore more costly—than [this] intervention,” said

Eddy. Therefore, in internal deliberations, the Instiglio team decided that a 13 percent reduction in unpro-

tected sex was the minimum acceptable level for the new program. “Conceptually, we set that as a minimum

bar for us. In a sense, if we couldn't do better than the best alternative, then we'd consider that a ‘failure’ in

some regards.”

At the same time, they did not want the risk exposure for their investors to be too high. Thus, they decid-

ed that if the program did not achieve at least a 5 percent reduction in risky behavior, investors would lose

part or all of their investment capital. But only if the program achieved a 13 percent reduction or more would

investors actually earn a market-level return on investment. Between 5 and 13 percent, investors would re-

ceive all their investment capital and a small return—but not a market-rate return.

While it made sense to the Instiglio team to set the minimum performance level in this 5-to-13 percent

range, Eddy knew it might be difficult to persuade government officials that this performance level was high

enough. For many years, government leaders had been accustomed to setting far more ambitious perfor-

mance goals. In his state development plan, Fajardo, for example, had set as a goal a 20 percent reduction in

adolescent pregnancy statewide over the course of two-and-a-half years. These goals were aspirational and

not linked to specific interventions, but they were a common and accepted part of political discourse in Co-

lombia.

Finally, any impact evaluation included a level of statistical uncertainty—a product of the size of the target

and control groups, and the level of confidence in the results. Instiglio would have to decide how to handle

this uncertainty in deciding whether the program met, or did not meet, the success thresholds.

17

”Effectiveness and Spillovers of Online Sex Education: Evidence from a Randomized Evaluation in Colombian Public Schools,” by Alberto Chong, Marco Gonzalez-Navarro, Dean Karlan, Martin Valdivia, National Bureau of Economic Research, NBER Work-

ing Paper No. 18776, February 2013, http://www.nber.org/papers/w18776, retrieved October 29, 2013.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 12: Social Impact Bond Case

HKS Case Program 12 of 21 Case Number 2026.0

How Much Return to Investors? The final question—and a critical one—was how much return to offer

investors. In the Peterborough SIB project, investors might earn as much as 13 percent per year over eight

years, but only if the recidivism rate fell by 12.5 percent or more. For a more modest decrease of 7.5 percent,

investors would receive 10 percent per year. [See Exhibit 3.]

In the earliest SIB projects, a key selling point to government was the no-risk-for-a-failure guarantee. The

flip side of this guarantee, for investors, was the possibility of losing 100 percent of their investment capital.

In subsequent years, many SIB projects had opted for a more moderate version, in which government paid a

reduced rate if the program did not achieve the minimum performance standard. Investors might lose part,

but not all, of their investment. In Medellin, the Instiglio team decided to give Mayor Gaviria a choice be-

tween two payment schedules. One reflected full risk-transfer to investors—no payment unless the program

achieved a minimum success rate, with higher returns to investors for better program performance, above

that minimum success rate. The other included a minimum payment from government of 75 percent of pro-

gram costs, even if the program did not meet its minimum success rate. Since less risk was being transferred,

investors would receive a lower rate of return in this scenario.

To make sure the service provider had an incentive to succeed, Instiglio proposed that initial funding for

the program be divided 80 percent/20 percent between outside investors and the provider itself. At the end

of the project, the provider would receive a share of the reimbursement (and, if the program was successful,

return on investment) that was commensurate with that paid to the outside investors.

Instiglio, too, shared in the risks-and-reward equation. At low performance levels, Instiglio would suffer a

loss. At higher performance levels, it would receive a modest bonus. [See Exhibits 5 and 6 for payment

schedule by performance target, with and without a minimum payment.]

By the spring of 2013, Instiglio was nearing its one-year anniversary. The group had worked tirelessly—

first, to select a promising project and second, to design it so as to provide a good, solid test of the SIB model.

To get a green light on the project, however, Instiglio’s proposal would have to win over three critical players:

the government, the investors, and the service providers. With fingers crossed, they presented their pregnan-

cy-prevention proposal to Medellin Mayor Gaviria in May 2013, and waited for his verdict.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 13: Social Impact Bond Case

HKS Case Program 13 of 21 Case Number 2026.0

Exhibit 1. SIB Basics: Players & Flow Chart18

Though many variations were possible, there were five essential players in a SIB: the Government agency, the So-

cial Impact Bond-Issuing Organization (SIBIO), the Investors, the Service Provider(s), and the Independent Evalua-

tor. There were several basic steps in developing and running a SIB project:

18

One caveat. The final block of this chart reads, “If program unsuccessful: Government pays nothing; Investors lose investment capital.” In fact, this was a central tenet in the design of the earliest SIB projects. Over time, however, some SIB designers stepped back from this position, as many investors were unwilling to hold this much risk. Some of the later SIBs were designed so that if a program failed, the government would pay a portion—but not all—of the program costs and investors would lose a portion—but not all—of their investment. In such models, the potential return on investment was typically lower, as well.

GOVERNMENT decides to try a SIB project to address a chosen social issue

GOVERNMENT & SIBIO

negotiate a contract

SIBIO recruits INVESTORS

INVESTORS provide capital to SIBIO; SIBIO manages the money

SIBIO hires/pays SERVICE PROVIDER(S),

oversees project management

SIBIO (or GOVERNMENT)

hires INDEPENDENT EVALUATOR

EVALUATOR assesses Program

If program successful: GOVERNMENT makes performance-based payment to SIBIO; SIBIO repays INVESTORS their

capital + performance-based return on investment

If program unsuccessful:

GOVERNMENT pays nothing; INVESTORS

lose investment capital

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 14: Social Impact Bond Case

HKS Case Program 14 of 21 Case Number 2026.0

Exhibit 2. Flow of Funds in a SIB Contract19

19

“Social Impact Bonds/A Promising New Financing Model to Accelerate Social Innovation and Improve Government Perfor-mance,” by Jeffrey B. Liebman, Center for American Progress/Doing What Works, February 2011, http://www.americanprogress.org/wp-content/uploads/issues/2011/02/pdf/social_impact_bonds.pdf, retrieved September 20, 2013, based on chart, p. 11.

Investors

SIBIO

Service

Providers

Government

1. Working

Capital

2. Operating Funds

3. Performance-based Payment

4. Repayment + Return on

Investment

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 15: Social Impact Bond Case

HKS Case Program 15 of 21 Case Number 2026.0

Exhibit 3. Description/Comparison of Early SIB Projects in UK & US

The Peterborough SIB Project.20

The first social impact bond was developed in Peterborough, England in 2010 by

the United Kingdom Justice Ministry, working with a not-for-profit SIBIO called Social Finance, created in 2007.

The project goal was to reduce the recidivism rate for all male prisoners over 18, with a jail sentence of a year or

less, who were released from the Peterborough prison over the course of six years (approximately 3000 ex-

offenders, all told). If the project proved successful, the long-term hope was to expand it system-wide. Across the

United Kingdom, the recidivism rate for this population was 60 percent.

A group of 17 British and US investors collectively provided $8 million to fund the program upfront. Under the

contract, a primary service provider—One Service—was hired to oversee the work of several direct service provid-

ers (one working in the prison, one working with the families, one working in the community, etc.). If One Service

succeeded in reducing the recidivism rate by 7.5 percent, in comparison to a control group, the investors would

receive their money back plus a return on investment each year over eight years. For example, if the recidivism

rate dropped by 10 percent, investors would receive a return of 7.5 percent per year. At most, investors would

receive 13 percent per year for a recidivism reduction of 12.5 percent or more. Investors would begin to receive

this return at the four-year-mark, if an interim evaluation showed a recidivism drop of at least 10 percent. If One

Service did not reduce the recidivism rate by at least 7.5 percent over the life of the contract, the investors would

lose their $8 million and receive no return on investment.

The New York City SIB Project.21

Two years later, New York City developed the first US SIB project—also aimed at

reducing recidivism, but for a younger population: teenage boys, aged 16 to 18, jailed at Rikers Island between

September 2012 and August 2015 (about 3,000 per year). The recidivism rate for this population, within a year of

release, was just under 50 percent. New York City contracted with a SIBIO/program manager, MDRC, which in turn

contracted with two social service providers, Osborne Association and Friends of Island Academy, to deliver the

services. The financial agreement was slightly different from Peterborough’s. It relied on a single investor, Gold-

man Sachs, but used Mayor Michael Bloomberg’s access to his own foundation, Bloomberg Philanthropies, to

cushion Goldman’s risk. It also set a higher bar for “success” over a shorter time span.

Goldman Sachs provided project funding of $9.6 million, but Bloomberg Philanthropies guaranteed

the first $7.2 million of that sum. (In other words, even in the worst-case scenario, Goldman Sachs

would lose no more than $2.4 million of its investment capital.)

If the recidivism rate was reduced by 8.5 percent, Goldman Sachs would be repaid its $9.6 million

(half from the City of New York, half from Bloomberg Philanthropies).

20

“Peterborough Social Impact Bond,” Social Finance website, http://www.socialfinance.org.uk/sites/default/files/SF_Peterborough_SIB.pdf, retrieved October 8, 2013. 21

“Agenda, Scope, and Goals,” Social Impact Bond Project at Rikers Island, MDRC website, http://www.mdrc.org/project/social-impact-bond-project-rikers-island#agenda_scope_goals, retrieved September 13, 2013.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 16: Social Impact Bond Case

HKS Case Program 16 of 21 Case Number 2026.0

If the recidivism rate was reduced by 10 percent, the City of New York would repay the entire $9.6

million.

If the recidivism rate was reduced by more than 10 percent, the City of New York would also pay

Goldman Sachs a return on investment—for example, $600,000 for a 12.5 percent reduction, up to a

maximum of $2.1 million for a reduction of 20 percent or more.

If the program did not achieve at least an 8.5 percent reduction in recidivism, Goldman Sachs would

lose $2.4 million, and Bloomberg Philanthropies would lose $7.2 million.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 17: Social Impact Bond Case

HKS Case Program 17 of 21 Case Number 2026.0

Exhibit 3, cont’d. Key Contract Features in Peterborough & New York City SIB Experiments

Key Contract Questions Peterborough New York City

SIBIO Social Finance MDRC

Program manager One Service MDRC

Investors Group of 17 investors Goldman Sachs, with 75% of capital

guaranteed by Bloomberg Philan-

thropies

Contract duration 6 Years 3 Years

Target group Male prisoners of Peterborough

Prison, 18 & over, with sentences of

a year or less, due for release during

the contract period (approx. 3,000)

Male prisoners of Rikers Island, 16

to 18, due for release during the

contract period (approx. 9,000)

Control group Same prison population, in other UK

prisons

Baseline measure

Goal Reduce recidivism rate Reduce recidivism rate

Measure of outcome Change in recidivism rate, compared

to control group

Change in recidivism rate, compared

to baseline

Minimum definition of success Drop in recidivism rate of at least

7.5%

Drop in recidivism rate of between

8.5 & 10%

Maximum return on investment 13% each year for 8 years (for a

drop in recidivism rate of 12.5% or

more)

$2.1 million (for a drop in recidivism

rate of 20% or more)

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 18: Social Impact Bond Case

HKS Case Program 18 of 21 Case Number 2026.0

Exhibit 4. Effectiveness of 13 Programs Aimed at Reducing Teen Pregnancy

Instiglio looked at hundreds of studies about pregnancy-prevention programs before designing its program. Many

showed little evidence that the program had been effective. To give a flavor, this chart, derived from a 2006 study,22

shows summary data from 13 program evaluations. Of the 13, 8 programs were in the United States, 3 in other de-

veloped countries, and 2 in developing countries. In 9 studies, the results were based on self-report of the target

group. In the other 4, the results were based on laboratory tests. In all countries and both reporting categories, the

results were mixed. Three studies had significant positive effects, nine had insignificant effects, and one found

significant negative effects.

Pregnancy Reduction, Self-Report Pregnancy Reduction, Laboratory Report

Negative

Impact

No Significant

Impact

Positive

Impact

Negative

Impact

No Significant

Impact

Positive

Impact

Region

US

1 3 2 0 2 0

Developed

(non-US)

0 3 0 0 0 0

Developing

0 0 0 0 1 1

TOTAL 1 6 2 0 3 1

Setting

School 1 6 0 0 1 1

Community 0 0 1 0 1 0

22

Data drawn from an early draft of Kirby, Douglas B., B.A. Laris, and Lori A. Rolleri, “Sex and HIV education programs for youth; their impact on sexual behaviors of young people throughout the world,” Journal of Adolescent Health, 40:3 (2007), pp. 206-217.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 19: Social Impact Bond Case

HKS Case Program 19 of 21 Case Number 2026.0

Clinic 0 0 1 0 1 0

TOTAL 1 6 2 0 3 1

Strategy

Abstinence 1 0 0 0 0 0

Comprehensive 0 6 2 0 3 1

TOTAL 1 6 2 0 3 1

Sex

Males 0 2 0 0 0 0

Females 0 4 2 0 2 1

TOTAL 0 6 2 0 2 1

Average Age

9-13.9 1 3 0 0 0 0

14-17.9 0 3 2 0 3 1

TOTAL 1 6 2 0 3 1

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 20: Social Impact Bond Case

HKS Case Program 20 of 21 Case Number 2026.0

Exhibit 5. Payment Schedule by Performance Target, with Minimum Payment, in US Dollars23

Initial Program Funding………………… Government Payments at Contract’s End………………………….

Decrease in

size of at-

risk popula-

tion24

Share of

program

costs

fronted by

outside

investors

Share of

program

costs

fronted

by ser-

vice pro-

vider

Total pro-

gram

costs25

Government

payment to

outside in-

vestors26

Government

payment to

service pro-

vider27

Government

payment of

additional

SIB-related

costs28

Total gov-

ernment

payment

≥ 30% $1,325,567 $331,392 $1,656,959 $1,808,448 $452,112 $162,248 $2,422,808

25% $1,325,567 $331,392 $1,656,959 $1,765,360 $441,340 $148,513 $2,355,214

20% $1,325,567 $331,392 $1,656,959 $1,722,273 $430,568 $134,778 $2,287,619

15% $1,325,567 $331,392 $1,656,959 $1,670,567 $417,642 $118,296 $2,206,505

10% $1,325,567 $331,392 $1,656,959 $1,610,245 $402,561 $99,066 $2,111,872

5% $1,325,567 $331,392 $1,656,959 $1,463,746 $365,937 $52,366 $1,882,049

< 5% $1,325,567 $331,392 $1,656,959 $1,124,713 $281,178 $5,646 $1,411,537

23

Approximate data provided courtesy, Instiglio.

24

Reduction in the number of students engaging in unprotected sexual intercourse in the target group compared to the control group. 25

This refers to the cost of the 18-month program, per se, and does not include added costs associated with the SIB model, such as the cost of Instiglio’s time and expertise in setting up and managing the contract. Of these costs, 80 percent was to be fronted by outside investors ($1,325,567). The other 20 percent ($331,392) was to be fronted by the service provider, which would receive a commensurate share of the return on investment, if the program proved successful. 26

At the end of the contract, the outside investors would receive some or all of their money back, under this payment schedule, with a return on investment that depended on the success of the program. 27

At the end of the contract, the service provider would likewise receive some or all of its money back, under this payment schedule, with the same rate of return as that earned by the outside investors. 28

These proprietary costs include fees and taxes associated with setting up the investment fund and the cost of Instiglio’s time and expertise in setting up and managing the SIB contract. At the lower performance levels, the government payment would cover only a small portion of these costs. At the higher performance levels, the government payment would cover these costs and would provide Instiglio with a modest bonus.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.

Page 21: Social Impact Bond Case

HKS Case Program 21 of 21 Case Number 2026.0

Exhibit 6. Payment Schedule by Performance Target, without Minimum Payment, in US Dollars29

Initial Program Funding………………… Government Payments at Contract’s End………………………….

Decrease in

size of at-

risk popula-

tion30

Share of

program

costs

fronted by

outside

investors

Share of

program

costs

fronted

by ser-

vice pro-

vider

Total pro-

gram

costs31

Government

payment to

outside in-

vestors32

Government

payment to

service pro-

vider33

Government

payment of

additional

SIB-related

costs34

Total gov-

ernment

payment

≥ 30% $1,325,567 $331,392 $1,656,959 $1,903,241 $475,810 $192,466 $2,571,517

25% $1,325,567 $331,392 $1,656,959 $1,851,536 $462,884 $175,984 $2,490,403

20% $1,325,567 $331,392 $1,656,959 $1,799,830 $449,958 $159,501 $2,409,289

15% $1,325,567 $331,392 $1,656,959 $1,748,126 $437,031 $143,019 $2,328,176

10% $1,325,567 $331,392 $1,656,959 $1,636,098 $409,024 $107,307 $2,152,429

5% $1,325,567 $331,392 $1,656,959 $1,463,746 $365,937 $52,366 $1,882,049

< 5% $1,325,567 $331,392 $1,656,959 $0 $0 $0 $0

29

Approximate data provided courtesy, Instiglio.

30

Reduction in the number of students engaging in unprotected sexual intercourse in the target group compared to the control group. 31

This refers to the cost of the 18-month program, per se, and does not include added costs associated with the SIB model, such as the cost of Instiglio’s time and expertise in setting up and managing the contract. Of these costs, 80 percent was to be fronted by outside investors ($1,325,567). The other 20 percent ($331,392) was to be fronted by the service provider, which would receive a commensurate share of the return on investment, if the program proved successful. 32

At the end of the contract, the outside investors would receive their money back if (and only if) the reduction in risky behav-ior was at least 5 percent, under this payment schedule, with a return on investment that depended on the success of the pro-gram. 33

At the end of the contract, the service provider would likewise receive its money back if (and only if) the reduction in risky behavior was at least 5 percent, under this payment schedule, with the same rate of return as that earned by the outside inves-tors. 34

These proprietary costs include fees and taxes associated with setting up the investment fund and the cost of Instiglio’s time and expertise in setting up and managing the SIB contract. If the reduction in risky behavior was less than 5 percent, the gov-ernment would pay none of these costs. At the higher performance levels, the government payment would cover these costs and would provide Instiglio with a modest bonus.

For the exclusive use of G. Gray, 2015.

This document is authorized for use only by George Gray in 2015.