Association for Public Policy Analysis & Management

There are compelling arguments for investing in programs and policies that improve the rates of college enrollment and completion (Holzer 2008).  However, the U.S. has been losing ground to other countries in the percentage of its workforce that is college educated (Miller et al. 2009).  If we are to “once again have the highest proportion of college graduates in the world” (Obama 2009), we need to invest our scarce education dollars on strategies that maximize the chances students will complete high school with the skills required to be admitted to college and that college-ready students enroll.  Meeting this challenge requires that we not only understand what strategies work to improve youths’ preparation for and enrollment in college, but that we also understand the relative cost-effectiveness of those strategies.

This study extends the work of a recently completed systematic review of evidence on the effectiveness of college access programs (Maynard et al. 2014) to include an examination of the relative cost-effectiveness of those programs.  The study focuses on 18 different college access programs that were evaluated between 1990 and 2014 using strong experimental or quasi-experimental designs.  The paper applies the “ingredients approach” to cost estimation (Levin and Belfield 2014) and combines the cost and impact estimates to generate two cost metrics for each intervention model—(1) estimates of the cost per participant of delivering the services and (2) estimates of cost per unit gain in college access (i.e., the cost invested to “generate” each additional college enrollee).  Where feasible, the cost analysis relies on estimates provided by the study authors. However, for studies that do not include a cost analysis, we have generated cost estimates using available descriptive information on the programs, supplemented with interviews with program staff (where that is still feasible) about the quantities of various ingredients used in the programs and published estimates of the costs of those ingredients. Costs are discounted to reflect the average years between the program expenditure and the observed gains in college enrollment and they are denominated in 2014 dollars.

Both the impacts of the program and their costs vary considerably.  For example, impacts on college enrollment rates range from zero to over 25 percentage points, and costs range from less than $100 per participant to thousands of dollars per participant. Moreover, there is not a strong correlation between program costs and impacts on college enrollment rates.