Thank you.
Good afternoon, Mr. Chair, and members of Parliament. I appreciate the opportunity to testify before you today.
By way of a brief background, my name is Dr. David Juppe. I am the senior operating budget manager with the Department of Legislative Services in Maryland, which provides all support services for the Maryland legislature. For the past 26 years, I've been reviewing operating and capital budgets and providing fiscal policy advice for our budget committees. In addition, I teach undergraduate and graduate classes in budgeting. I spent a term as the president of the National Association of Legislative Fiscal Offices. I've provided budgetary training to the staff of the parliament in Mozambique, and in 2014 I provided some analysis and recommendations to the Government of Jordan in Amman on their budgetary processes in their review of the budget.
In 2012 my colleague, Kyle McKay, and I began examining the issue of social impact bonds, or pay for success, because of a proposal by our Department of Public Safety and Correctional Services that was designed to attempt to reduce recidivism rates below those in programs currently in place by that department. Based on our analysis, we found that the savings from the proposed program would be about $250,000 against program costs of about $4.1 million. We determined that the program would not save funding for the state, and we recommended against their adopting that program.
Turning to the concept of social impact bonds and pay for success generally, as I'm sure you're aware, a social impact bond is not in fact a bond. It's a unique form of financing program for government whereby a third party financier provides multi-year funding for non-profit or private sector service providers. They are given a multi-year source of funding, which reduces the risk they have in securing government funding each year. The concept of the social impact bonds is that this multi-year funding would help spur development of innovative ideas and ways of providing services for government. The other allure for government is that you only pay when you have successful outcomes, and no funding is required up front.
Social impact bonds have become more popular in the last few years. The first example was the Peterborough Prison in England. As the evaluations are coming in, we're starting to see that perhaps that program is not as successful as was initially hoped.
One of the reasons I think social impact bonds have become more popular recently is that since the recession of 2008 the government in the United States, and state governments especially, have not seen a robust economic recovery, as was hoped, and as we saw after the 2001 recession. Resources are limited for the providers of these services. This is one mechanism for providing additional funding for government services without government providing the funding up front.
There are a number of concerns and observations that I have drawn from our work looking at the topic, as well as looking at some other social impact bond proposals in the U.S.
One reason, and I think the most important, is that social impact bonds will result in higher costs to government. This is because you have additional costs for an intermediary to align the financing with the government and the provider. You also have to pay a rate of return to the financier once the program is completed and assuming the outcomes are successfully reached. Nothing prohibits government from directly contracting with these service providers and paying for the service directly. I think this has become an issue of government having limited resources and providers looking at other sources of funding to try to expand the pull so they can expand services.
From what I can see, the rate of return is not limited in any way. As we know, in the bond market, risk is measured by interest rates. The riskier it is that repayment may not materialize, the higher the interest rate a government is going to pay on a capital bond.
The social impact bonds or pay for success is a form of borrowing. If the program works then government will pay this rate of return, which happens to be whatever was negotiated, whether 10%, 15%, or 20%. There appear to be no limits on that amount.
In some instances the financing is guaranteed by a foundation, or at least a portion of the financing is guaranteed by a foundation to ensure that some rate of return is provided to the financiers. If that's the case, then it really obviates the entire point of transferring risk to the private sector, because you're guaranteeing some portion of the return.
Another major point I would like to make is that in our research we have found in many cases that the proposed savings are overstated. I have seen proposed reforms over many years of budgetary proposals, and in many instances, advocates have fallaciously used the fixed cost per case when evaluating the savings. For example, if it costs $30,000 per year to house an inmate, that includes both fixed costs for operating the facility and the variable cost. If you save one or two or ten inmates from returning to prison, you save a variable cost of food and supplies and medical costs and we've found that to be about $4,600 per inmate. You don't save the full $30,000 unless you close the entire facility.
In many cases, if advocates are proposing that the savings are going to be the full fixed and variable costs divided by the caseload, that's overstating the savings. In some cases advocates are suggesting that you have a cost avoidance, so you save cases from requiring more expensive care in the future. This is also very difficult to prove and doesn't result in immediate savings to the government's actual funding in the budget.
Funding logistics are also a concern that I observe in that one of the selling points for social impact bonds is that the government pays only if a program is successful. The problem I see with this is that if you have a multi-year program and the government has not been setting aside funding for this payment, then when the program is successful, you'll have to provide the entire amount of money plus the rate of return all at once, which would be difficult for governments to appropriate at one time.
Moreover if government sets aside funding in advance each year for the potential payment as we're seeing, for example, in the pay-for-success program in Massachusetts, it does not really realize those savings, because it still has to set aside those funds annually.
Because of this concept of a performance-based return on investment, I think rather than encouraging innovation, social impact bonds or pay for success will actually encourage a flight to quality. Investors are going to want to see programs that work and programs that are successful.
We've seen an example in Chicago, Illinois, where they just entered into a social impact bond in November 2014 for early education services using programs that have already proven to be effective over multiple years. Investors are not going to want to see these programs fail. So if you do have a situation in which multiple programs fail as they're trying to encourage innovation, then, getting back to our market-based model, you're going to see higher interest rates. Investors are going to demand a higher rate of return because there's higher risk.
Evaluation concerns that I have are first and foremost that, because you have this return-on-investment component, there is a greater pressure to produce results and you may have a situation where one study produces an outcome that's positive resulting in payment to the investors and to the service providers, but in many cases in public policy it can take years and sometimes multiple observations and multiple studies to determine if a program is really successful or not. In some cases I've seen it take 10 years before it was determined that a certain course of action really wasn't the best course of action, really wasn't working, and had to be ended. This places more of a premium on ensuring success early.
Also there's the question of methodologies and whether or not there's a treatment and control group and full randomization to ensure that fair and objective analysis and evaluation are completed. The U.S. Congress was considering social impact legislation last year in 2014, and I noticed in that legislation that they were considering allowing quasi-experimental designs, which may not require this sort of randomization.
In conclusion, social impact bonds are a new concept that has recently become more popular. A number of states and countries are looking at this concept, but I believe it is a more expensive way of providing services. It is essentially government by credit card. You're potentially borrowing funds, and it's not necessarily good fiscal policy. If you can pay for these services, pay for them directly. Why would you pay a third party intermediary a rate of return that is in no way limited or restricted?
Secondly, I would point out that instead of encouraging innovation, there would be a flight to quality, and so I don't see the concept of innovation there.
Finally, I would just say there are logistical issues as to whether or not government actually saves money, because you either have to set aside funds or you'll have a very large payment to make upon the completion of the proposal.
Thank you very much.