Will It Work?: Feasibility Assessment in Pay for Success
Private investors often use feasibility assessments to help guide investment decisions. In the simplest sense, a feasibility assessment is an analysis of how successful a project can be. As private investors emerge in the world of social finance, through social innovation funding, they expect a similar type of analysis. For social programs, feasibility assessments are developed from a comprehensive understanding of the program and its impact.
A critical component of any feasibility assessment is the population analysis. In the private sector, these techniques are used to understand the market. Population analytics profile a customer base, and from this description, investors can learn a lot about who is using a product or service and the potential to expand into new markets. As you can imagine, these estimates are also very important in social finance transactions. Investors and providers alike can learn a lot by analyzing who a program intends to serve, who they actually serve, and who could potentially be served if the program were expanded.
As part of NCCD’s Pay for Success project, I am on a team that is developing feasibility assessments for three different program providers in the United States. Given that some of the analytic techniques are borrowed from the private sector, we found it is useful to differentiate the study groups that feed into the population analysis. The first group we examine are the clients who have actually received the service/intervention. We call this group—you guessed it—the service population. Often, the best way to start compiling information on the service population is by collecting data from the providers themselves. Analyzing client characteristics (such as demographics, service history, assessments, program time/completion, and the impact/effect of the service) provides valuable insights for the population analysis.
Next, we examine the target population. This is the group of clients the program was intended to serve and can be very similar to or different from the service population. Most programs clearly define the target population in the program description’s eligibility requirements. Unlike service populations, target populations are almost always developed from administrative data (where this information is found depends on the program’s clientele). Target populations are developed by profiling clients who match the program eligibility requirements. Then, similar to the service population, all available client characteristics are gathered to describe the target population with as much specificity as possible.
Examining service and target populations are not new concepts in program evaluation. The intersection of these two populations often help evaluators understand fidelity (meaning the extent to which the program is serving the people for whom it was intended). In a feasibility assessment, these analyses provide an additional function. As part of the feasibility assessment, investors need to know the potential impact of program expansion. This is the scaling estimate. The service and target population analyses provide critical insight into this question.
To develop the scaling estimate, insights from the service and target population are applied to the population at large. For example, if the analyses suggest that a program works well for children with a distinct set of characteristics in one location, the population estimate would calculate how many kids in the entire community have those exact same characteristics. Like the target population analysis, this estimate often requires administrative data (sometimes from several different systems). The scaling estimate is calculated by determining the maximum number of potential program participants and the expected impact if the program were given unlimited resources.
The key to developing the best population analytics is availability and access to quality information that describes both the program and the population at large. Programs that have worked with system administrators to develop protocols and safely exchange information have a distinct advantage when it comes to developing population analytics. Given that these estimates are critical to understanding programmatic impact and attracting investments from the private sector, it is always a good idea for programs to start building data infrastructures so that the effect of their services can be properly estimated.