Advancing Evidence-Based Policymaking to Solve Social Problems

Advancing Evidence-Based Policymaking to Solve Social Problems

JEFFREY B. LIEBMAN

Advancing Evidence-Based Policymaking to Solve Social Problems

Much is known about how to address social problems, but much remains unknown. Here is a plan to help disseminate tested interventions and identify new approaches that work.

Despite spending billions and billions of dollars each year, the United States is simply not making rapid enough progress in addressing a range of social problems. Many children are poorly prepared to start or advance in school. Many prisoners end up back in jail after being released. Many new and displaced workers lack skills to succeed in the workplace. Many people suffer from chronic illnesses such as diabetes and asthma. In the face of these and other social problems, the nation has either failed to develop effective solutions, failed to prove that the solutions work, or failed to scale up the solutions that do work.

These failures have adverse consequences for the affected individuals and their communities; they also damage the nation’s economic health. At the most basic level, the nation’s prosperity depends on the productivity of its workforce and the share of residents who are employed. But when there are 7 million youth who are neither in school nor working, when half of low-income fourth-graders are not reading at even a basic level, when 1 in 15 African American men is incarcerated, and when the nation ranks no better than average in college graduation rates among developed countries, the United States is clearly not maximizing the potential of its workforce. The results are felt in a lower standard living for individuals and in reduced economic growth for the nation.

Solving such social problems, difficult under any circumstances, is complicated by the nation’s continuing fiscal woes. At every level of government, policymakers are facing the same challenge: How do we continue to innovate and make additional progress in addressing the nation’s problems when budget cuts are making it difficult, if not impossible, to hold onto the gains we have already made? The only way to keep making progress in this fiscal environment is to produce more value with each dollar that government spends. Doing so will require better use of evidence in policymaking.

The good news is that over the past decade, new evidence- based practices have emerged, at the federal, state, and local levels that simultaneously offer the potential to speed up progress in addressing social problems and to make better use of taxpayer dollars. Further, there are a number of clear steps that the federal government can take to promote the use of these practices throughout all levels of government.

If these evidence-based approaches to policymaking spread widely, we will achieve better outcomes with government expenditures by replacing less-effective government programs with programs that work better, and we will develop new, more-effective approaches. But if our goal is to make significant progress in addressing our most serious social problems, simply expanding the use of these strategies is unlikely to be enough to produce the results we require. We need to supplement the wide diffusion of these practices with a more-focused approach that aims to supply solutions for specific high-priority populations. To this end, I propose that the government launch two initiatives—that will span a decade and target entire populations of at-risk individuals in specific communities.

The various evidence-based practices that have emerged in recent years fall into five general categories, based on the challenges they are intended to address. These challenges are:

Subsidizing learning and experimentation to develop new solutions. After 40 years of applying rigorous evaluation methods to social policy, researchers and the government have learned a lot about what works and what does not. But for many of the most important social problems, there are no proven, cost-effective, scalable strategies. The challenge is how to finance needed program development and experimentation. If the government or a philanthropy funds 10 promising early childhood interventions and only one succeeds, and that one can be scaled nationwide, then the social benefits of the overall initiative will be immense. Therefore, there is a need for funding mechanisms that pay for the learning—and the failure—that are necessary to ultimately come up with successful solutions.

The simplest way to increase social innovation is to fund it directly through government or philanthropic grant competitions and innovation funds. Alternatively, governments can guarantee to “buy” the results of social innovations. In producing private consumer goods, innovators know that if they produce a terrific product, there is a market waiting for it. But “customers” do not necessarily exist who will scale up successful social innovations, and this may inhibit philanthropies and social entrepreneurs from investing in developing them. Alternatively, governments can offer large prizes for the development of solutions to important public problems, an approach that has worked in a number of more technical areas.

Increasing the amount of evidence on what works. Faster and less expensive ways are needed to find out which innovative solutions work. Today, the outcomes of most programs are never measured, much less evaluated by comparison to a rigorous counterfactual scenario. Occasionally, a program will be subjected to a large program evaluation that, by the time the results are available, may indicate whether or not the program worked in the past but not whether it is still working, given that conditions may have changed.

One approach to finding answers is prioritizing evaluation funding. Spending a few hundred million dollars more a year on evaluations could save tens of billions of dollars by identifying which programs work and generating lessons to improve programs that do not work. Federal agencies should be given the authority to set aside a portion of their program funding for evaluation, and individual grants can carry a requirement to include evaluations.

Also, granting agencies can structure grant competitions to simultaneously reward interventions that have developed rigorous evidence of effectiveness and to generate new evidence on promising practices that have not yet been evaluated. For example, the i3 program run by the Department of Education (DOE) provides competitive grants to local education agencies to expand innovative practices to improve student achievement, increase high-school graduation rates, or increase college enrollment and completion. The program uses three tiers of grants: scale-up grants to fund practices for which there is already strong evidence, validation grants to fund promising strategies for which there is currently only moderate evidence, and development grants to fund “high- potential and relatively untested” practices. Most of the scale- up grants have provided approximately $3 million each, most of the validation grants approximately $15 million each, and most of the development grants approximately $50 million each. If more agencies used this tiered approach, then groups that depend on grants to support the development of social interventions would probably work to obtain the added evidence that would bring larger grants.

In the past, a major obstacle to building up evaluation evidence has been the cost associated with data collection (an in-person survey can easily cost $1,000 or more per sample member). The information technology revolution offers the means to improve evaluations considerably. Government administrative data systems can be used to provide important outcome measures on a monthly or quarterly basis and to compare the performance of different providers. Evaluations have also been limited by the delay (often 5 to 10 years) between when an evaluation is proposed and when evidence becomes available. Although some delay is necessary to measure the long-run impacts, with information technology and the availability of large administrative data sets measuring key outcomes such as earnings histories, college enrollment, and criminal trajectories, there is the opportunity to do evaluations more rapidly to provide useful assessments of current program effectiveness. It is time to bring the private-sector practices of rapid-cycle innovation and use of data for continuous improvement to the public sector.

Making greater use of evidence in budget and management decisions. There are several challenges to making budget decisions based on evidence of program effectiveness. For many programs conclusive evidence does not exist. Even when evidence of program effects does exist, it can be analytically challenging to extrapolate from that evidence to broader policy effects or to systematically compare the relative effectiveness of different policy options. When funding is distributed to lower levels of government or is used to finance provision by numerous private-sector service providers, effectiveness can vary widely in different locations. In some cases, underfunding is what causes programs to be ineffective. Thus, the best policy response to poor performance can be to increase funding. On the management side, the main challenge is to produce evidence on a high enough frequency to allow a feedback loop between management practices and outcomes.

In addition to technical issues, there are often political obstacles. Legislatures sometimes earmark funding for particular providers, thwarting efforts to award grants through merit-based competitions. An evidence-based initiative that threatens to reduce the market share of an incumbent provider who has friends in the legislature may not endure, whereas an ineffective program that is politically connected may be immortal.

Several strategies have been developed to tackle such issues. The tiered evidence approach used by the DOE’s i3 grant competitions to direct funds to programs with the best track records has been adopted by several other federal agencies, including the Department of Labor (DOL) and the Department of Health and Human Services (HHS). Even without tiered standards, grant competitions can make evidence of success a selection criterion. Grants made through noncompetitive means, such as formula grants, can also make the adoption of best practices a requirement for some or all of their funding.

Agencies also are doing more—and could do yet more— to compare the performance of different providers. For example, the federal Head Start program has historically renewed existing providers automatically. But under new regulations designed to respond to concerns about uneven quality at Head Start sites, providers are awarded five years of funding and then evaluated, with the bottom 10% of performers required to compete with other potential providers for funding. This approach of re-competing the worst performers is a good model for types of social services where it is not easy to scale the size of operations up or down.

In a related step, the government can publicize the performance of all providers and then give the individuals who are eligible to be served a voucher to choose among them. This approach is used in the Australian employment program, Jobs Services Australia, where the government compiles results on reemployment rates and other quality metrics for each service provider and publishes a rating (one to four stars) for each.

Another approach addresses the fact that in policy areas where dozens of different interventions have been evaluated in separate research studies, often using different outcome measures and different evaluation methodologies, it can be difficult for policymakers to make sense of all of the evidence, especially when the cost of the interventions varies widely. A review of the evidence by a policy-focused team of experts can help determine the relative cost-effectiveness of different approaches. The Washington State legislature created the Washington State Institute for Public Policy to produce studies of this sort. The institute uses a benefit/cost model to review the impact of various criminal justice policies, and state policymakers nationwide are using the findings to determine which policies are the most cost-effective for reducing recidivism.

Targeting efforts around spreading proven practices has also proved useful. New York City’s Center for Economic Opportunity provides an example. The center, established in 2006 and supported by a $100 million annual fund, focuses on developing and evaluating innovative approaches to reducing poverty. Initiatives that are successful then become eligible for permanent funding through the city budget. The center has produced more than 50 initiatives, many of which have been evaluated with randomized controlled trials. The federal Charter Schools Program, which funds successful, high-quality charter schools to expand enrollment or set up additional schools, is another example. Since 2010, the DOE has awarded competitive grants to support the expansion of 34 existing high-performing charter schools and the creation of 251 new high-quality charter schools. In both situations, the same teams that set up the original models are involved in the replications. Although narrow-purpose programs can sometimes be excessively rigid and stifle innovation, when they support proven innovations they can be another way to spread successful practices.

Rather than managing programs based on the quantity of services provided, government agencies need to track outcomes for specific target populations and manage their programs to achieve outcome goals.

Simply cataloging successful practices can help them to spread. The DOE’s What Works Clearinghouse, the Department of Justice’s CrimeSolutions.gov, the Substance Abuse and Mental Health Services Administration’s National Registry of Evidenced-Based Programs and Practices, and the Department of Labor’s Clearinghouse of Labor Evaluation and Research are helpful steps toward making evidence on what works more available. Private-sector efforts such as the Coalition for Evidence-Based Policy’s Social Programs that Work list are similarly useful.

One of the most direct ways to use evidence in allocating resources is to stop funding things that are found not to work. An example of this occurred early in the Obama administration when federal funding for the Even Start program, which provides literacy training to parents and children in homes where English is not the first language, was eliminated after a randomized control trial found that outcomes were no better for those served by the program than for those in the control group. This decision was controversial. Supporters of the program argued that it had improved after the period covered by the evaluation, but the cancelation remained in force.

A leadership strategy called PerformanceStat also has proven valuable. In this approach, the chief executive or a top deputy of an agency holds regular meetings with key staff to review up-to-date data on progress toward achieving performance goals. The PerformanceStat movement began at the local level with New York City’s Compstat and NYCStat, Baltimore’s CitiStat, and Maryland’s StateStat. The method has since spread to other locations, and uses of the techniques have corresponded with falling crime in New York City, the demolition of blighted properties in New Orleans, improved Medicaid claims processing in Los Angeles County, and reduced homelessness in Washington, DC. In the past two years, this approach has blossomed within the federal government as well, with nearly two dozen agencies using the strategy.

Making purposeful efforts to target improved outcomes for particular populations. Most government social service funding is dedicated to purchasing slots in programs. Programs are managed to deliver a defined set of services to a fixed number of people rather than to achieve any particular outcome. Multiple government programs often provide “stove-piped” assistance or services to a given individual, with none of them accountable for getting the individual to achieve success. Rather than managing programs based on the quantity of services provided, government agencies need to track outcomes for specific target populations and manage their programs to achieve outcome goals. Successful population-focused efforts will generally require extensive collaboration with many nongovernmental community partners, including businesses, nonprofit service providers, and philanthropies.

A number of approaches can help address these issues. The Greater Cincinnati Strive Partnership is perhaps the best example of an effort to define a target population and coordinate services in a strategic way to make sure everyone in the population receives the services they need to succeed. Strive is an initiative in Cincinnati, Ohio, and two neighboring cities in Kentucky—Covington and Newport—that aims to improve student achievement “from cradle to career.” The partnership has brought together a range of community members to focus in a data-driven way on achieving eight outcomes: kindergarten readiness; fourth-grade reading proficiency; eighth-grade math proficiency; high school graduation rates and ACT scores; and postsecondary enrollment, retention, and completion.

The partnership takes an “every child” approach and regularly tracks the number of students who are not achieving target outcomes to form strategies to deliver the services necessary to raise the number of successful outcomes. Different community organizations are taking the lead on various components of the initiative. Kindergarten readiness has risen in all three communities, and the Cincinnati Public Schools became the first urban school district in Ohio to be rated “effective.”

An important component of population-focused efforts is matching the right services to the right individuals. Some services work better for particular population subgroups than for others, and some are cost-effective only for certain population subsets. New York State, as part of its Work for Success employment initiative for the formerly incarcerated, is developing a client matching system to connect prisoners with the appropriate set of services upon release from prison.

Another approach is setting outcome-focused goals and managing them. The Department of the Interior’s initiative to reduce violent crimes in Indian communities illustrates the power of goal setting, frequent and population-focused performance measurement, replication demonstrations, and scaling. Over three years, violent crime has fallen by 55% at four reservations. The agency is now working to expand the initiative to two additional reservations.

Spurring innovation and aligning incentives through cross-sector and community-based collaboration. Outside of education, most social-service provision is done by private providers, with most of the financing coming from government. This means that government’s task is to create an environment with its grant competitions and procurements so that providers and their philanthropic partners can be successful in innovating, producing evidence on what works, and scaling up effective services.

The federal government has recently taken a number of steps toward that goal. The Social Innovation Fund, administered by the Corporation for National and Community Service, makes grants to grant-making intermediaries that match the federal awards dollar for dollar with funds raised from other sources. Each intermediary then runs a competitive process to make grants to expand community-based nonprofits with evidence of strong results, and the grantees are required to match the grants they receive. To date, the fund has awarded a total of $138 million to 20 intermediaries, which have selected 197 nonprofit subgrantees, resulting in $350 million in nonfederal cash match commitments. Each grant is expected to be rigorously evaluated, though given the relatively small size of the grants and the early stage of development of some of the organizations receiving grants, it is unclear how many impact evaluations will actually emerge from the initiative.

The federal government has also used waivers and performance partnerships to encourage community-level innovation. Disadvantaged communities and individuals are often the recipients of services from multiple federal programs spanning several federal agencies. In many cases, this funding is not well coordinated, and no one is responsible for whether or not the combination of funding produces successful outcomes. In his 2013 budget, President Obama proposed to allocate $200 million in existing funding to Performance Partnerships in which states and localities would be given the flexibility to propose better ways to combine federal resources in exchange for greater accountability for results. The initial proposals were targeted at the areas of disconnected youth and neighborhood revitalization. Several states gave examples of the projects they would like the flexibility to undertake. For example, Iowa was interested in developing a coordinated approach to providing services to high-risk youth—those involved in the child welfare, juvenile justice, mental health, and vocational rehabilitation systems.

More broadly, federal waiver authority is an important tool for developing innovative approaches. By letting state and local governments test and evaluate new strategies, they can demonstrate solutions that can then be spread through legislative changes. Nine states are currently testing new strategies for serving children and families involved in the child-welfare system. To be successful, a waiver program must not only give states the flexibility to try new solutions, but it must also incorporate rigorous evaluation so that informed decisions can be made about whether to spread the new strategies to additional jurisdictions.

An important category of initiatives in which state and local governments are being given the flexibility to combine federal funding streams in creative ways is that of place-based initiatives. These include the Department of Housing and Urban Development’s Choice Neighborhood Program, which invests in improvements in housing, schools, transportation, access to employment, and the Sustainable Communities Initiative, a joint effort of several federal agencies that supports regional planning and development efforts.

The availability of a federal grant competition can serve as an action-forcing event that enables government, philanthropic, and private partners at local levels to come together and produce an innovative plan to improve outcomes. Perhaps the most successful example of this has been the DOE’s Race to the Top Fund, which offered grants to states for educational reform efforts and required all of the necessary stakeholders to be partners in the state proposals. Notably, many of the states that did not receive federal funding nonetheless decided to implement their proposed reform plans. As with waiver authority, the key to producing learning from initiatives such as this is not simply to promote innovation but also to rigorously assess the results of the different strategies, so that these state initiatives can truly function as “laboratories of democracy.”

Federal-state partnership funds represent another approach. Many federally funded social programs are administered by state and local governments, and efforts to improve the administration of these programs require cooperative efforts across levels of government. In 2010, Congress authorized the Partnership Fund for Program Integrity Innovation, managed by the Office of Management and Budget (OMB), which received $32.5 million to be spent over a multiyear period. The OMB set up a Collaborative Forum with representatives from state and local governments, nonprofits, federal agencies, and participation by the public to come up with ideas for pilot projects that could demonstrate best practices and inform policy decisions. The OMB approves pilot concepts and then transfers funds for the pilot to a lead federal agency, which implements the pilots in collaboration with state and local partners. As one example, the Department of Justice is working with three state and local juvenile justice agencies to develop a cost-effectiveness scorecard. By showing the cost-effectiveness of different evidence-based juvenile justice interventions, the score- card will help program leaders make better service contracting decisions and also help frontline service providers make better decisions about particular interventions for youth.

Pay for Success initiatives using social impact bonds (SIBs) represent a particularly novel approach to financing social programs. Under the most common model, the government contracts with a private-sector intermediary to obtain social services and then pays the intermediary entirely or almost entirely based on the achievement of performance targets. If the intermediary fails to achieve the minimum target, the government does not pay. Payments typically rise for performance that exceeds the minimum target, up to an agreed-upon maximum. The intermediary obtains operating funds by raising capital from commercial or philanthropic investors who provide upfront capital in exchange for a share of the government payments. The intermediary uses these operating funds to contract with one or more service providers to deliver the interventions necessary to meet the performance targets.

New York City launched the nation’s first SIB initiative. The city is using this approach to finance services for 16- to 18-year-olds who are jailed at Rikers Island, with the aim of reducing recidivism and related budgetary and social costs. Services are being delivered to approximately 3,000 adolescent males per year, from September 2013 to August 2015. MDRC, a nonprofit, nonpartisan education and social policy research organization, is serving as the intermediary, overseeing day-to-day implementation of the project and managing the two nonprofit service providers that are delivering the intervention. Goldman Sachs is funding the project’s operations through a $9.6 million loan to MDRC. The city will make payments that range from $4.8 million if recidivism is reduced by 8.5% to $11.7 million if recidivism is reduced by 20%. Bloomberg Philanthropies is guaranteeing the first $7.2 million of loan repayment.

Although SIBs are still a highly experimental approach to financing social programs, they are also highly promising because they directly address all five of the challenges involved in disseminating evidence-based practices. Because SIBs shift the risk of failure from taxpayers to the private sector, they enable governments to try innovative solutions. Because rigorous real-time impact assessment is an essential component, SIBs generate additional evidence on program impacts. Because an intervention that results in a successful SIB will likely be scaled up whereas an unsuccessful one will not receive further funding, the evidence from a project is highly likely to be influential in budget decisions. Because SIBs assign specific populations to service providers and hold them accountable for outcomes, they avoid the traditional fragmented and slot-based approach to service provision. Because SIBs are multiparty contracts that combine government, service providers, and private investors in a multiyear effort to achieve performance goals, they promote cross-sector collaboration.

Spreading the new practices

Despite the encouraging progress, it remains the case that most government spending is not allocated based on evidence or with a focus on innovation or performance. Even in federal agencies that have been the most creative in developing new ways to allocate grant funds, such as the DOE and HHS, the bulk of spending is distributed through conventional grants and other traditional mechanisms. Therefore, it is an important priority to spread these new strategies widely: to additional programs within the federal agencies that are already using them, to additional federal agencies, and to more state and local jurisdictions.

Dissemination will occur faster and be more likely to survive changes in administrations if Congress and the president take action to support the adoption of effective approaches. There are a variety of legislative and administrative actions that would help spread these practices. These include:

Providing funding authority for evaluations. All agencies that administer social programs should receive authority, similar to that currently provided to the DOL, to reserve a portion of program spending to fund program evaluations. The DOL’s authority is better crafted than similar authority that has been provided to other agencies because it gives the agency the flexibility to use the evaluation funds to evaluate the highest-priority initiatives rather than tying the funds to evaluation of specific programs. The agency authority should be provided only to agencies that establish a chief evaluation officer position or have a similar office that is dedicated to producing independent and rigorous evaluation evidence about the agency’s programs.

In addition, a small evaluation fund should be provided to OMB for cross-agency evaluation initiatives. The administrative structure could be similar to that of the Partnership Fund described above and would enable coordinated, cross-agency evaluation planning for related programs. OMB would be encouraged to consult with Appropriations Committee leadership in deciding which programs to evaluate and to demonstrate the potential benefits of planned studies.

Expanding the use of tiered evidence standards in grant competitions. In high-priority policy areas where sufficient evaluation evidence is available, agencies should be encouraged to use tiered evidence standards in grant competitions. Structuring a grant competition in this way ensures that the largest share of federal funding goes to practices that have been shown to be effective, while also investing smaller amounts of funds in developing evidence about promising but not yet proven approaches. Grant competitions that use tiered evidence standards are expensive to administer because they generally require agencies to recruit outside evaluation consultants to review the evidence base of each proposal received. Thus, they are worth using only for large, high-priority grant competitions.

Reserving a portion of formula funding for proven practices. The majority of federal social spending is administered through formula funding to state and local governments, rather than through grant competitions. As there currently is a limited evidence base for most activities funded via formula funding, it would not make sense to restrict the use of these funds only to evidence-based practices. Instead, a portion of these funds should be used to build the evidence base. Then, over time, as the evidence base becomes strong enough, formula funding provided to state and local governments should begin to require that a portion of such funds be spent on programs that have been proven effective. Depending on the policy area and the amount of evidence that exists, this requirement could start by stipulating that 1% of such funds be allocated to proven practices, then have the requirement rise over time to 5%.

Directing the OMB to submit an annual report on evidence-based techniques and practices. The 2010 Government Performance and Results Modernization Act was important because it codified the new agency performance reporting framework in a way that makes the framework more likely to endure. Similarly, Congress could take steps to show that it supports the recent executive branch evidenced-based initiatives. At a minimum, a congressional committee could have an annual hearing at which it invites the OMB director and some agency representatives to describe their progress in using evidence-based techniques. Congress could also require the OMB to produce an annual report listing evidence-based practices by agency and reporting the percentage of each agency’s grants that were made using evidence-based techniques. Many such reporting requirements quickly devolve into time-wasting compliance exercises. Therefore, if such a report is requested, the requirement for producing it should be sunsetted after five years and apply only if Congress also rescinds several existing OMB reporting requirements.

Directing a specific government agency to take charge of producing cost-effectiveness reports. The congressional budget and appropriation committees should jointly identify five priority policy areas each year and ask either the Congressional Budget Office or the Government Accountability Office to produce a report in each area identifying what evidence currently exists about the outcomes being produced by federal spending in that area; what is known about the relative cost-effectiveness of different strategies in the area; what promising strategies exist that have not been rigorously evaluated; and where promising strategies are lacking and what funding strategies and incentives could be used to spur the development of new solutions. Potential areas for such reports include job training, early childhood education, and school dropout prevention, among many others. This added requirement would probably require an increase in the agency budget of at least $3 million per year, but this modest investment would enable Congress to target resources more effectively. Once the agency has gone through the effort of producing an initial report in each area, it should be updated every two or three years.

Compiling federal program evaluations into a comprehensive Web site. In 2010, the OMB issued guidance requiring agencies to make information readily available online about all federal program-impact evaluations that are planned or already under way. This effort was intended to be analogous to the HHS clinical trial and results data bank (ClinicalTrials.gov) that aims to prevent drug companies from hiding negative trial results. Although some agencies complied by posting their evaluations on their agency Web sites, the OMB never moved forward to compile all of the agency information into a comprehensive federal evaluation Web site. In addition to preventing the suppression of results, a comprehensive federal evaluation Web site would help Congress and the public understand which programs have been evaluated, which ones haven’t, and what the results showed. A challenge with government transparency efforts is that they often provide a large quantity of information without the interpretive context necessary for consumers of the information to use it. Thus, any federal effort of this sort will likely need to be accompanied by a parallel effort by organizations like the Coalition for Evidence-Based Policy to interpret the information.

Making administrative data more accessible for measuring outcomes. Evaluation studies that wish to use government administrative data to measure program outcomes now typically need to go through extensive one-off negotiations to arrange access. Establishing a standardized way for evaluators to access common federal data sources would increase the number of evaluations that can occur.

The 10-year challenge

If the various evidence-based approaches to policymaking spread widely, the government will achieve better outcomes, will stop spending money on programs that do not work, and, most important, will spur the development of new, more-effective intervention strategies. But if the goal is to make significant progress in addressing the nation’s most serious social problems, simply expanding the use of these approaches is unlikely to be enough. It will be necessary to supplement the wide diffusion of these practices with a more-focused approach that aims to find solutions for specific populations.

Consider the challenges of reducing recidivism among ex-offenders, raising fourth-grade reading and math skills among children living in high-poverty neighborhoods, preventing youth from dropping out of high school, helping chronically unemployed individuals obtain and keep jobs, raising community college completion rates, eliminating chronic homelessness and homelessness among families, and helping developmentally disabled youth make successful transitions into the adult workforce, among many others. Although dozens of programs spend billions of dollars serving these populations, services are delivered in a highly fragmented manner. One program provides mental health services, another provides housing services, and still another provides job-readiness training. But in most cases, no one is responsible for ensuring that a specific cohort of individuals in a particular community achieves successful outcomes. Moreover, it is not just federal government programs that determine whether success is achieved. Results depend on the joint actions of multiple levels of government, diverse not-for-profit and for-profit service providers, private businesses, and philanthropic and other community partners.

With a goal of overcoming this fragmentation and producing the community-level collaboration and innovation necessary to make real progress on the nation’s most persistent social problems, I propose the 10-Year Challenge. Congress and the president should work together to identify 10 social problems where it is a national priority to find solutions. All of the problems would be ones where the specific individuals in the population to be served can be identified and baseline outcomes can be established; these two factors will provide an observable baseline against which improvement can be measured.

Through a grant competition, 10 communities would be selected for each problem—100 communities overall—in an effort to transform outcomes for the specific population within 5 to 10 years. A single agency would be the granting agency for each initiative, though many of them will require cross-agency collaboration. The granting agency would first issue planning grants of $250,000 each to several dozen communities that demonstrate a commitment to cross-sector partnerships and organizational capacity to develop competitive proposals. Final awards would be made to those communities that successfully propose a data-driven, collaborative approach to transform delivery of services to achieve measurable improvements in outcomes for specific cohorts of individuals.

In selecting winning proposals, the agency should look at such criteria as the project’s design, the level of commitment of community partners, and the likelihood based on existing evidence, that the project will make significant progress in addressing the target social problem;; the potential for the project to produce rigorous evidence that would add to what is currently known about the effectiveness of particular strategies; the extent to which the proposed strategies represent significant advances over current practices; and the potential for scaling the project within the given state and to other states or other similar populations.

In the projects, a target population could be statewide (for example, all youth in a state who are aging out of the juvenile justice or foster-care system in a given year) or much more geographically targeted (for example, all preschool-age youth in a particular neighborhood). In most cases, interventions will involve 1,000 to 2,000 individuals per year. With fewer individuals, it will be hard to have enough statistical precision to know what the results are; whereas budgetary resources are unlikely to permit significantly larger groups to be served, except in cases in which the cost per person served of the intervention is very low.

The projects would spend, on average, $10 million a year on services (with flexibility depending on the nature of the intervention and the size of the community and the target population). The grants would cover one-third of the cost of service provision, with state and local governments providing one-third and private community partners covering one- third. The federal government would also provide up to $1 million per year per project for evaluation expenses. In total, the federal share of the initiative would cost approximately $400 million per year. Funding for the initiative could come from repurposing existing grant and formula funding, or from a congressional appropriation. In addition, the federal government would waive program rules as necessary so that communities could take existing funding streams and use them in more flexible ways to fundamentally redesign systems.

Pay for Success

In the second initiative that I propose, the government would become a strategic partner with state and local governments in expanding current Pay for Success/SIB projects into areas where state and local activity has the potential to achieve important federal policy objectives or produce significant federal budget savings. The highest priority would be to create an initiative around early childhood interventions including home visiting, early learning, and preschool, or initiatives spanning birth to second grade. Evidence of success in early childhood programs is strong, but state and local governments are having trouble establishing Pay for Success projects in this area because much of the financial benefit of these interventions accrues to the federal government in the form of lower Medicaid and transfer program spending and higher federal tax revenue. Federal involvement could enable the economics of these projects to work.

As another example, if a state government can set up a Pay for Success project that helps disabled individuals return to work rather than apply for federal disability insurance benefits, the federal government should offer to reimburse the state government for any success payments it makes in the project.

The federal government should also consider a broader role as a strategic partner with state and local governments in establishing SIB projects. The rationale for a broader federal role is that if one state or local government discovers a solution to a social problem, it will have tremendous value to the nation as it gets scaled nationwide. As mentioned above, a process through which the federal government subsidizes learning at the state level can help overcome the underinvestment in learning that is likely to occur when individual jurisdictions cannot capture the full value of their discoveries. The federal subsidy could come in the form of an extra prize-like payment to investors in successful SIB projects. Alternatively, the federal government could “backstop” a portion of the losses in unsuccessful projects.

To support this effort, Congress should give the DOE and HHS the authority to repurpose a total of up to $125 million of existing funds to support five state or local Pay for Success initiatives. The funds would be used to match, on a dollar-for-dollar basis, state or local success-based payments. A typical-sized project would provide $10 million of services per year for four years and serve approximately 1,000 families per year. With a state/local commitment of $25 million and a federal commitment of $25 million, success-based payments of up to $50 million would be possible. Funds repurposed in this way would remain available until expended, rather than expiring at the end of the fiscal year.

During the past decade, governments have shown tremendous creativity in coming up with new approaches to foster experimentation and learning and to allocate spending to social programs with the best evidence bases. The results of the programs funded through the new mechanisms will greatly add to the knowledge base regarding which policies and programs work in addressing serious social problems and which do not. If the new evidence-based practices continue to spread, the amount of learning will accelerate and the country will reap fiscal and economic benefits.

But without a strategic effort to develop approaches that target entire populations of at-risk individuals in specific communities, it is unlikely that the nation will move the dial on its most pressing social problems. What is needed is a decade in which we make enough serious attempts at developing scalable solutions that, even if the majority of them fail, we still emerge with a set of proven solutions that work.


Jeffrey B. Liebman () is the Malcolm Wiener Professor of Public Policy at Harvard University’s John F. Kennedy School of Government. This article is adapted from Building on Recent Advances in Evidence- Based Policymaking, a white paper he prepared for the Brookings Institution’s Hamilton Project.