Goldman Sachs Invests in POP Bonds

by on August 2, 2012 at 8:51 am in Current Affairs, Economics, Law | Permalink

Goldman Sachs is investing in a New York City pay on performance bond (POP bond also called a social improvement bond). The Pop bond is based on recidivism rates for adolescents, as described by the NYTimes:

The Goldman money will finance a program called Adolescent Behavioral Learning Experience…which seeks to improve prospects for black and Latino adolescents. The jail program, which will offer counseling and education for an estimated 3,400 incarcerated adolescent men each year, will be run by two nonprofit organizations, Osborne Association and Friends of Island Academy, and overseen by MDRC.

…If the program reduces recidivism by 10 percent, Goldman would be repaid the full $9.6 million; if recidivism drops more, Goldman could make as much as $2.1 million in profit; if recidivism does not drop by at least 10 percent, Goldman would lose as much as $2.4 million.

…Currently, nearly 50 percent of young men released from Rikers reoffend within a year.

As I wrote earlier:

For Pop bonds to work it is critical that outcomes be measured and marked to an appropriate, randomized, control group. If not carefully monitored, the private sector will also excel at innovative and creative gaming at the public expense.

The involvement of Goldman Sachs makes me fear that my last sentence will prove prophetic.

1 Bill August 2, 2012 at 9:10 am

Good comment.

Sort of like some of the private medicare programs where you have to walk up three flights of stairs to register, or the programs which offer deductions from premiums if you use health clubs (ie, in effect, prices higher if you are unable to use health clubs because you are not ambulatory or in a nursing home). Funny we don’t randomnise more programs. Charter schools which send back kids to public schools or don’t offer ESL, etc.

2 Rahul August 2, 2012 at 9:36 am

Is Goldman treating this as an investment or a charitable donation?

3 IVV August 2, 2012 at 10:00 am

Probably depends on which side of the books you’re looking. An investment that acts as a charitable donation for tax purposes?

4 jpe August 2, 2012 at 11:16 am

However Community Reinvestment Act projects are designated. It’s not deductible, but it goes toward appeasing the regulators.

5 Yog Sothoth August 2, 2012 at 9:44 am

Agreed. Goldman is going to just Christ-rape the system here. Such a good idea in theory but once you factor in a little public choice into that theory you get a cash cow. BTW, is there a lot of public choice that models lobbying of the government? Seems like you could get support for antitrust laws, estate taxes, etc., which limit private concentrations of wealth out of that type of analysis.

6 Marc August 2, 2012 at 10:18 am

Could POP bonds be used to suppliment funding to schools in low-income districts? We already keep relatively meticulous records of school performance such as attendance, drop-out and graduation rates, as well as standardized testing (barring the debate about standardized test scores as an effective metric for performance). If we assume that lack of education is one of the driving forces behind perpetual poverty and crime (which have a high cost to local, state and federal governments), an increase in funding via the market may help reduce rates.

7 Bill August 2, 2012 at 10:24 am

Note on methodology

You don’t have to randomize if you pay on the performance of the entire pool–not just those participating in the program. In fact, if you pay for performance on the entire pool, that will incentivise Goldman to bring more persons from the pool (all incarcerated men) into the program. In that case, I don’t see why you need to randomize…Goldman would be paid just for the delta for the entire population. Also, if you randomize participation, what you are doing is NOT reaching those who the program could help, which seems strange.

8 Rich P August 2, 2012 at 10:48 am


That is definitely not true. Paying based on the performance of the entire pool would allow Goldman to bet on the effects of various outside trends on the population. For instance, job creation causes employment among parolees to climb. Recidivism would then drop among all former inmates, not just the ones in the intervention. This is precisely why randomized control trials are important.

If you pay based on the performance of the population rather than the intervention group, you are creating a new market where Goldman can make bets on societal trends, and allowing them to pick their bets (via POP) very carefully. Their shareholders will thank you, but I will not.

9 Bill August 2, 2012 at 1:05 pm

Rich P,

Re: “would allow Goldman to bet on the effects of various outside trends on the population”

This presumes there is no competition for the funding, doesn’t it. If the trend is visible to other bidders, it is discounted for.

You can think of it this way: bidding on the pool is the equivalent to community rating, where you have to take in anyone who presents themselves for coverage. If the benefit plan is uniform, ie, you can’t offer health club benefits, then you get the mean of the population, not a segment of it with lower costs.

Second, regarding the unemployment issue, this is also covered by bidding, but if you want, you could also write the contract to be adjusted on that parameter.

All of the above is simply applying both contract and insurance principals. Give me an example with numbers or sets and we’ll work through it.

10 Adrian Ratnapala August 2, 2012 at 10:51 am

I don’t get this. Presumably the amount of money put into this is only sufficient to give meaningful aid to some fraction of the population. And Goldman won’t put money into it if the payout depends on people from outside the that fraction.

11 Bill August 2, 2012 at 1:11 pm


This is an alternative to randomization, which also doesn’t reach the population that is excluded from the trial. I assume that you know the current rate of recidivism, that Goldman would have an incentive to reach first the easiest and then reach for the marginal ones that put it over the top for achieving an overall total group performance better than before. Given Rich’s comments, I would certainly 1) have competition and 2) identify the non-controllable variables that influence recidivism (unemployment rates, if Rich is correct, or say changes in the composition of the pool during both periods; those could be adjustments in the contract or you could pair it up with a different city as a control, etc.

12 Adrian Ratnapala August 2, 2012 at 10:55 am

One way to reduce gaming in this kind of thing is to post a substantial bounty (approx 50% of the payout) on anyone who can prove that the performance was up to scratch.

There will still be pressure to fudge the pre-arrange performance measures, but there will be pressure coming from opposite sides. The bankers want to create easy pickings, the bounty-hunters want to add gotcha clauses.

13 Bender Bending Rodriguez August 2, 2012 at 5:32 pm

??? That doesn’t make sense. I’m going to pay the POP holder their premium, plus another 50% to some schmo who proves that it worked? What if G-S proves that it works?

Or are you suggesting that there’s a 50% “clawback” clause in which someone can get a bounty for proving that it _doesn’t_ work?

That still doesn’t work: G-S games the system for the max payout, then turns around and proves that it doesn’t work, clawing back 50% of the bonus from itself.

14 ThomasH August 2, 2012 at 11:03 am

However this turns out, it should help develop the methodologies by which “outcomes [are] measured and marked to an appropriate, randomized, control group.”

15 dead serious August 2, 2012 at 11:31 am

True, because if Goldman excels at anything it’s identifying ways to game an existing system or enter a market before clearly-defined rules/regs are in place.

Not intended as a condemnation.

16 derek August 2, 2012 at 11:07 am

All great suggestions on how to avoid fraud and game rigging. Can I demand the same things before I pay my taxes?

17 Andrew' August 2, 2012 at 11:20 am

I see GS paying off victim witnesses. Hey, on the bright side it’s government backing into restitution.

18 R Richard Schweitzer August 2, 2012 at 11:27 am

OK. If the “program” is viable, where else will you get $9.6 million to give it a try? A tax on Big Sodas? Is the potential for gain to GS, a risk to the taxing authority. If so, what are they gaining (socially and in recidivism savings) for the exposure to GS gain?

19 Citi.zen August 2, 2012 at 12:26 pm

The payoff does not seem to justify the negative publicity and likely backlash. If we were talking billions, sure. But 2.4m for GS is very little.

20 Aaron August 2, 2012 at 12:40 pm

The maximum loss to GS here is $2.4mm. Do Alex (and commenters here) really think Goldman is going to go out of it’s way and risk further reputational damage to save $2.4mill on what is essentially a PR move?

21 Brock August 2, 2012 at 12:58 pm

It’s also possible that the “game” Goldman is planning on is actually fixing recidivism. After all, they might have figured out that NYC does such a poor job of getting these kids integrated back into society that they’ve already figured out how to do it for a profit (given the payback on the POP bond).

For instance, we know jobs and role models help. What if GS just paid these kids minimum wage to work in a mail room somewhere (so the mail room proprietor gets free labor) while getting a volunteer group to provide the mentorship they need? It might be that simple.

22 Bill August 2, 2012 at 1:13 pm

+1 “It’s also possible that the “game” Goldman is planning on is actually fixing recidivism.” Maybe that’s a game we want Goldman to play.

23 Mark August 2, 2012 at 2:51 pm

They don’t have to pay them to do any real job, just a lump sum conditional on not re-offending by whenever the cut-off is – there seems, roughly, to be enough cash in the game to make that potentially profitable. But if society did want to pay that sort of Danegeld it shouldn’t need Goldman-sized middleman fees to arrange it.

24 Leo August 2, 2012 at 1:18 pm

What if they just train the criminals to get caught less and give them better lawyers?

25 Bill August 2, 2012 at 2:26 pm

Are you assuming they aren’t getting that training in getting caught less in prison as a free good from other prisoners?

26 Erick August 2, 2012 at 2:45 pm

Even if they game the system, isn’t there some value to be gained? The corrections field is always in need of better data to help with identifying people more likely to reoffend.

27 Ray Lopez August 3, 2012 at 2:57 am

I think the concern for a random control group is the fact that crime rates have been dropping as the population ages, so who is to say that even without this program the rates would have dropped anyway? Hence Goldman Sachs may be just getting a free ride on a social trend, at taxpayer’s expense.

28 Joe August 2, 2012 at 4:30 pm
29 ziel August 2, 2012 at 8:38 pm

The program is being run by Bloomber’s own charity. From the article: The jail program, which will offer counseling and education for an estimated 3,400 incarcerated adolescent men each year, will be run by two nonprofit organizations, Osborne Association and Friends of Island Academy, and overseen by MDRC.

It doesn’t appear that Goldman has any actual control over the program, never mind designing or developing their own programs. They have only $2.4m at risk – a pittance. The current recidivism rate is 50%. A 10% reduction means they’re betting on a 45% rate after the program is completed.

I have no idea if “counseling and education” can bring about a 10% reduction in recidivism, and I’m guessing neither does Goldman. But we can be pretty sure that the program will provide some very nice young people who happen to be daughters and maybe even sons of some rich and powerful people in NYC (including, no doubt, executives at Goldman) the opportunity to be employed for a few years and to pad their resumes.

30 ziel August 2, 2012 at 8:56 pm
31 The Anti-Gnostic August 3, 2012 at 9:07 am

Youth Advocate/Cognitive Skills Facilitator at Friends of the Island Academy
Greater New York City Area | Education Management

Youth Advocate/Cognitive Skills Facilitator at Friends of the Island Academy, Co-Facilitor/Curriculum WriterII at Lyrics on Lockdown II, Facilita…
Co-Program Director of Fall 2011 viBestage at viBe Theater, Assistant Teacher at MCC Theater’s Youth Acting Lab, Co-Director of viBestages Fall P…
New York University, New York University (NYU)
I am a senior at NYU’s School of Individualized Study concentrating in Social Justice Education Through the Arts with a minor in Urban and Global…

I’m speechless.

32 jk August 2, 2012 at 11:42 pm

Recidivism backed securities

33 R Richard Schweitzer August 3, 2012 at 1:49 pm

A further thought: Retrain and relocate to a western or southwestern jurisdiction.

Comments on this entry are closed.

Previous post:

Next post: