Archive for February 2012

Pay for Success in Deval Patrick's budget

I just got around to posting this, but here is the recent development from Massachusetts regarding a Social Innovation Financing Trust Fund - a facility to pay for social impact bonds.

     Section 35XX. (a) There shall be established and set up upon the books of the commonwealth a trust to be known as the Social Innovation Financing Trust Fund, in this section called the trust, for the purpose of funding contracts to improve outcomes and lower costs for contracted government services, in this section called pay for success contracts, subject to the requirements of subsection (b).
     (b) Notwithstanding any general or special law to the contrary, the secretary of administration and finance, in this section called the secretary, may enter into pay for success contracts. Each contract shall include: (1) a requirement that a substantial portion of the payment be conditioned on the achievement of specific outcomes based on defined performance targets; (2) an objective process by which an independent evaluator will determine whether the performance targets have been achieved; (3) a calculation of the amount and timing of payments that would be earned by the service provider during each year of the agreement if performance targets are achieved as determined by the independent evaluator; (4) a sinking fund requirement under which the secretary shall request an appropriation for each fiscal year that the contract is in effect, in an amount equal to the expected payments that the commonwealth would ultimately be obligated to pay in the future based upon service provided during that fiscal year, if performance targets were achieved; and (5) a determination by the secretary that the contract will result in significant performance improvements and budgetary savings across all impacted agencies if the performance targets are achieved.
     (c) The secretary, in his discretion, may provide that payments in future years under such contracts shall constitute a general obligation of the commonwealth for which the full faith and credit of the commonwealth shall be pledged for the benefit of the provider or providers of the contracted government services, but the total amount of payments under such contracts secured by such a pledge of the full faith and credit of the commonwealth shall not in the aggregate exceed $50,000,000.
     (d) The secretary shall be the trustee of the trust, shall administer the trust and shall ensure that all funds appropriated as described above are deposited in the trust and shall make payments from the trust in accordance with the terms and conditions of the contracts, without further appropriation. The secretary shall provide a status report on all contracts no later than February 1 of each year to the house and senate committees on ways and means.

Source: http://www.mass.gov/governor/legislationeexecorder/legislation/an-act-making-appropriations-for-the-fy2012.html

Pay for Success in Obama's 2013 Budget

Straight out of the White House 2013 Budget:


Pay for Success in Domestic Programs. Many traditional Government social programs fit one of two molds: prescriptive programs that stifle innovation by specifying eligible providers and activities, or flexible block grants that fail to focus on results. To ensure taxpayers get the best possible return on their investment, the Administration is testing a new program model—Pay for Success—in which the Government provides flexibility for how services are delivered and pays for results after they are achieved. The working capital for a Pay for Success project generally comes from private investors that bear the risk of failure, but receive a financial return if the project succeeds. Projects use and build evidence-based practices to improve the lives of vulnerable target populations, reducing their need for future Government services and cash assistance. Over the course of 2012, the Administration is launching a small number of Pay for Success pilots in criminal justice and workforce development. The President’s 2013 Budget reserves a total of up to $109 million to test this new financing mechanism in a broader range of areas including education and homelessness. If successful, Pay for Success projects offer a cost-effective way to replicate effective practices and support continuing innovation as Federal resources become more constrained.

White House, The Budget for Fiscal Year 2013, "Cutting waste, reducing the deficit, and asking all to pay their fare share," page 45, http://www.slideshare.net/whitehouse/the-presidents-budget-for-fiscal-year-2013.

From the information age to the analysis age

I love this new NYT article on the rise of data.  The next innovations will come not from the ability of people to gather more data, but from new ways of analyzing data - combining it, slicing it up, extrapolating it for predictions, and so on.

UPDATE:

Just discovered Kaggle,  an amazing website that capitalizes on the rise of "data scientists" by throwing them into a tournament against each other.

More here:
http://www.businessweek.com/magazine/kaggles-contests-crunching-numbers-for-fame-and-glory-01042012.html

Essex in UK to use social impact bonds to address foster care

I wrote earlier about Essex County in England exploring social impact bonds. Now Essex County has announced that they will use a social impact bond (SIB) to reduce their residential care population. There are 1,535 residential care children in Essex. (I use residential care to describe all children under government care in Essex.) The SIB will take 11% of that population, 170 kids, and try to move about half of them (90 kids) out of residential care using Multi Systemic Treatment, one of several textbook treatments available for addressing this population. (Some others are Functional Family Therapy, Multidimensional Treatment Foster Care, and Family Integrated Transitions.)

Description of treatment. Multisystemic therapy is an intensive, home-based intervention for chronic, violent, or substance abusing juvenile offenders between the ages of 12 and 17. Trained therapists work with the youth and his or her family in a home, school, or community setting, with an emphasis on addressing the causes of delinquency. Service duration averages 60 hours of contact over four months, with each MST therapist working in a team of four therapists and carrying a caseload of four to six families

Effectiveness of treatment. One of the best academic studies of this treatment was in the population of youth who committed sexual crimes. The 89% of participants who completed the study had 83% fewer arrests for sexual crimes and 70% fewer arrests for other crimes than did the control group. (Source: http://www.ncbi.nlm.nih.gov/pubmed/19170451.)

Cost of treatment. According to the famous WSIPP cost-benefit analysis of various treatments, the marginal cost of MST is $4246. (Marginal cost is the cost of serving an additional person, so it is usually lower than the average cost, which is the total cost of treatment of the population divided by the number of persons treated.) Therefore the total cost of serving this population is at least $721,820.

Return on investment. The Essex press release mentions that MST is 4x as pricey as residential care. Using marginal cost, that means if the SIB moves 90 kids out of the residential care system, it will save Essex $1.52 million. The maximum savings (if it avoids 170) is $2.88 million. If we use 2x marginal cost for the estimate, the target savings is $3m and the maximum is $5.76m.

Other estimates: According to NSPCC, residential care costs 1,500 pounds/kid/week ($2300/week, or $118,000/year), while cost of treatment is 8,000 pounds/family. This puts target savings at $10.6 million, and maximum savings of placing all 170 kids of $20 million, assuming only 1 year of foster care avoided. Note, however, that these numbers are different from the 4:1 cost-to-treatment ratio mentioned in the Essex County release.

Direct vs. strategic budgetary costs. NSPCC mentions savings accruing to Essex in several ways: "In theory, over time, as the number of preventable family breakdowns are reduced, fewer children would enter the care system, leading to "cashable" savings for the Council, such as through the closure of a residential care home." This is a stretch goal for most interventions. The first savings accruing from a SIB is a direct budgetary reduction in marginal cost: if you place a kid out of residential care, you no longer have to house and feed that kid, reducing government cost. The second savings comes as you successfully place enough kids to close down an entire home -- or to avoid building a new one to meet expected rise in residential care population.

Control group. According to the Essex press release, the residential care population dropped by 5% from 1630 to 1535 last year. If we expect the same population decrease this year, that means we should expect 89 kids to leave residential care without any additional intervention. The Essex social impact bond takes 170 of the 1535 for treatment and will try to move 90 out of residential care. The cynic might say that the bond's work is easy because if it picks well, it might capture the 89 who would have otherwise left residential care anyway, as predicted by the trend. The optimist might say that the bond's work is hard, because it tries to get a 50% reduction in a population that has historically seen a 5% reduction. In either case, the control group should account for the trend in residential care numbers in the county.

Partners. Essex worked with Social Finance UK to develop this social impact bond.

Timeline. An earlier document had the following timeline for Essex:
  • Sept 2011 - start implementing the SIB 
  • Sept 2012 - interim outcomes review 
  • March 2013 - final outcomes review 
Some remaining questions.
  • Services:  What services and service providers will be chosen for this SIB?   What is the full range of services that Essex delivers to its population of 1,535, and what effect do those services have on the population?
  • Funders: Will the funders for this intervention differ from those that funded Peterborough?
  • Savings:  How much of the savings will Essex pass on to the service provider? 
  • Evaluation:  How will the 170 kids be selected from the population of 1,535?  How will Essex attribute efforts to the treatment? Will the remainder of the population be used as a control group, or will the control be constructed in a more scientific way?
Sources.
  • Essex press release
  • WSIPP
  • Borduin CM, Schaeffer CM, Heiblum N., A randomized clinical trial of multisystemic therapy with juvenile sexual offenders: effects on youth social ecology and criminal activity, J Consult Clin Psychol. 2009 Feb;77(1):26-37.