Accelerating Community College Graduation Rates: A Benefit–Cost Analysis

ABSTRACT

This article reports a benefit–cost evaluation of the Accelerated Study in Associate Programs (ASAP) of the City University of New York (CUNY). ASAP was designed to accelerate associate degree completion within 3 years of degree enrollment at CUNY’s community colleges. The program evaluation revealed that the completion rate for the examined cohort increased from 24.1% to 54.9%, and cost per graduate declined considerably (Levin & Garcia, 2012; Linderman & Kolenovic, 2012). The returns on investment to the taxpayer include the benefits from higher tax revenues and lower costs of spending on public health, criminal justice, and public assistance. For each dollar of investment in ASAP by taxpayers, the return was $3 to $4. For each additional graduate, the taxpayer gained an amount equal to a certificate of deposit with a value of $146,000 (net of the costs of the investment). Based on these estimated returns, a cohort of 1,000 students enrolled in ASAP would generate net fiscal benefits for the taxpayer of more than $46 million relative to enrolling in the conventional degree program. ASAP results demonstrate that an effective educational policy can generate returns to the taxpayer that vastly exceed the public investment required.

KEYWORDS:

• Community college
• cost–benefit analysis
• degree completion
• evaluation framework
• higher education interventions
• policymaking
• public investment

Acknowledgments

The authors wish to extend thanks to Donna Linderman, Colin Chellman, Zineta Kolenovic, Simon McDonnell, Brooks Bowden, and two anonymous reviewers who provided valuable comments.

Funding

Special thanks to the New York City Center for Economic Opportunity for support of this study.

Notes

1. City University of New York Office of Institutional Research and Assessment’s analysis of data from the Integrated Postsecondary Education Data System revealed that nationally, only 16% of urban community college students graduate within 3 years (CUNY Office of Institutional Research and Assessment, 2012).

2. When ASAP began in 2007, the 3-year CUNY community college graduation rate was 24% for skill-proficient students, based on a CUNY analysis of student-level data from the CUNY Institutional Research Database, and 13% for all first-time full-time community college students who entered in fall 2004 regardless of skills proficiency (CUNY Office of Institutional Research and Assessment, 2012).

3. Borough of Manhattan, Bronx, Hostos, Kingsborough, La Guardia, and Queensborough Community Colleges.

4. The second ASAP cohort admitted in 2009 also realized a 3-year graduation rate of 55% versus 23% for a comparison group of similar students (Linderman & Kolenovic, 2013).

5. The evaluation was conducted by Metis Associates. The results of the optimal matching for CUNY ASAP conducted by Jing Zhu are available in Linderman and Kolenovic (2012, Appendix C). Other covariates used in the propensity score-matching analyses are applying for financial aid (Pell and Tuition Assistance Program [TAP]), cumulative grade point average, and cumulative credit.

6. Updated analyse for recent years and cohorts, some using random assignment (which ensures unbiased impact estimates with external validity limitations), have been provided by Scrivener, Weiss, and Sommo (2012), Scrivener and Weiss (2013), Kolenovic, Linderman, and Karp (2013), and Scrivener et al. (2015). The latter evaluation estimated that for students with development education needs in 3 colleges, 40% of the ASAP program group received a degree within 3 years compared with 22% of the control group, using random assignment to ASAP and a control group.

7. For arguments explaining the importance of cost–benefit and cost-effectiveness analyses for policy and decision making, see Levin and McEwan (2001), Evans and Ghosh (2008), and García (2015).

8. In other words, the counterfactual used in the evaluations of the program’s impacts by CUNY and MDRC is the increased number of graduates among ASAP participants versus the traditional program or comparison group. The benefits and costs of an additional degree under ASAP and following the traditional program are estimated by comparing the net benefits, after costs are accounted for, of graduating versus not graduating (i.e., receiving a high school diploma) in both cases.

9. Following the cost–benefit analyses described herein, the benefits of a higher educational attainment (associate degree [AA] completion) are measured relative to the benefits of high school completion. Because CUNY was producing graduates from both ASAP and in the traditional program, we calculated the incremental benefits and costs (relative to those of high school completion) for associate degree completion of the two alternatives. The benefits of completing an associate degree were similar for the two alternatives—as graduation requirements are the same and students are comparable at the baseline on observable characteristics, as determined by propensity score matching. However, the cost per graduate to the taxpayer differs depending on whether the graduate came from the treatment or the comparison group. In our analysis, we compared the benefits and costs of completing an associate degree relative to high school completion under both ASAP and the traditional program. It should be noted that experimental estimates for later cohorts in randomized controlled trials have yielded comparable results of the program’s impact to the ones used in this study, with graduation rates in the treatment group doubling those in the control group, even under different student eligibility requirements (Scrivener et al., 2015).

10. All comparisons of benefits and costs were based on differences between high school graduates and associate degree graduates. We attempted to include analyses for partial completers but were not able to obtain the necessary data. The omission of benefits to partial completers imparts a downward bias to benefits.

11. The average age of a 3-year graduate was 23 years for ASAP, based on a mean starting age of 20 years (see Linderman & Kolenovic, 2012, p. 17).

12. The debate on discount rates is generally resolved in this range for social investments, but it is still under debate in the literature (Benzoni & Chyruk, 2015; Moore, Boardman, & Vining, 2013).

13. Both independent studies and government studies by the Department of Justice or the Department of Health and Human Services have provided systematic information for the nation and states that link education to crime and to health, respectively. But systematic studies at local levels are rare to nonexistent. The relation between education and income and its yield of government tax revenues are available from the U.S. Department of Commerce, and, particularly, the Bureau of the Census, as well as state and local agencies. A summary of sources of information utilized throughout the analysis, by category of benefits, is available upon request.

14. The estimates of the net benefits to the student are also conservative, as they are limited by data availability to the higher personal income generated by the investment and omitting personal benefits from improvements in their health (see McMahon, 2009, for an estimate of individual health benefits), lower probability of criminal involvement, and less dependence on public assistance.

15. Blomquist et al. (2014) is a recent example of an evaluation estimating the social value of higher education.

16. FTE students is a commonly used measure for enrollments that adjusts the number of students by the number of credits of study they have undertaken.

17. These institutional costs did include some student costs in terms of tuition and fees for non-ASAP students. However, in the analysis, we assumed that all the institutional costs were paid by the taxpayer, which further contributes to providing a conservative estimate for the net benefits (minus costs) to the taxpayer. These portions of costs could be partially double-counted in this study, but their nominal magnitude has no major consequences for the results and conclusions. See a detailed table and explanations in Levin and Garcia (2013).

18. http://www.cuny.edu/admissions/tuition-fees.html

19. Forgone earnings were calculated as the difference between the average earnings of a full-time worker with a high school degree and the average earnings of an enrollee in a public institution (aged 20–22 years), both expressed as PV at age 23 years and in 2010 dollars.

20. http://www.cuny.edu/admissions/tuition-fees.html

21. Educational attainment in the ACS and Census data was defined as (a) high school graduate or GED, regular high school diploma, GED or alternative credential, and (b) associate degree, type not specified.

22. We refer to earnings and income interchangeably. They are not identical. Earnings are derived from wages and salaries; income includes not only wages and salary, but also rents, dividends, and interest from property, etc. However, for high school and associate degree populations, the proportion of income from earnings is usually 95% or greater.

23. See Levin and Garcia (2013) for information on associations between education and categories of benefits examined in the article, as well as for breakdowns by gender or race in some cases.

24. Detailed descriptive information of the ACS data by year is available upon request. Linderman and Kolenovic (2012) provided the descriptive information for the students in the first ASAP cohort and in the comparison group (see Table 1). In terms of race breakdown, the ACS demographics are in between the demographics for the treatment group and the control group. In terms of gender, about 54% of students in ASAP (and 53% in the comparison group) were women, relative to 50% in ACS. Mean household income in ACS was higher than mean household income for CUNY students. As explained in Levin and Garcia (2013), estimates of the PV of earnings by gender and race showed that the difference between earnings (and hence taxes) for an associate degree over high school completion for each gender and race group was relatively similar (it varied from about $310,000 to about $362,000). Therefore, any difference in the demographic composition of CUNY students relative to the Census data would not be expected to seriously distort the results, on average.

^25. Tax contributions were calculated based on earnings, as a proxy for an individual’s total income. For the groups studied in the report, earnings constituted more than 90% of income.

^26. Available at http://users.nber.org/~taxsim/taxsim-calc9.

^27. The NBER’s tax simulator allows for specifying only the state but not the exact city where the individual lives. In contrast, the calculations corresponding to other taxes calculated in the section “Additional Tax Revenues from City Income, Property, and Sales Taxes” (city income, property and sales taxes) were estimated using information pertaining to individuals who reside in New York City.

^28. Rouse (2007) and Levin et al. (2007) also estimated tax contributions for single individuals. See tax contributions from earnings for married people in Levin and Garcia (2013), Appendix B. Total taxes for married people were, on average, 15% lower than for single people.

^29. MEPS reports years of education of the individuals at the time the individual joined the study rather than degree received (i.e., if an individual earned any educational credential afterward, his educational attainment would be underreported). We assumed 12 years of education is equivalent to a high school degree and 14 years of education is equivalent to an associate degree. We limited our analysis to data for individuals residing in the Northeast region, within a metropolitan statistical area.

^30. According to the definition of variables in Integrated Public Use Microdata Series census microdata for social and economic research), this variable reports how much pretax income (if any) the respondent received during the previous year from various public assistance programs commonly referred to as “welfare.” Assistance from private charities was not included, but the following were included: federal/state Supplemental Security Income payments to elderly (age 65 years and older), blind, or disabled persons with low incomes; Aid to Families with Dependent Children; and General Assistance.

^31. For selection of discount rate chosen, please see “Methods” section.

^32. Another excluded benefit is the higher likelihood of students in ASAP enrolling in 4-year institutions and pursuing more education. Based on Linderman and Kolenovic (2012), CUNY-ASAP estimates that transfer rates to a 4-year institution of ASAP students who received the associate degree are higher than those of associate degree recipients in the comparison group. Specifically, 72.4% of ASAP associate degree recipients versus 62.2% of associate degree recipients in the comparison group transferred to a baccalaureate institution. Strumbos and Kolenovic (2016) estimated the effects of ASAP on enrollment in a baccalaureate program, associate degree attainment, bachelor’s degree attainment, any degree attainment, and time to degree using information from the first three cohorts (6-year outcomes). They estimated effects of 9, 21, 7, and 18 percentage points on the four first outcomes, respectively, and a −1.51 semester reduction in the time-to-degree outcome (propensity score-matching results). Eight-year outcomes for ASAP’s first cohort of students (fall 2007 cohort, the only cohort so far with 8 years of data) showed that 33.5% had earned a bachelor’s degree (a gain of 8 percentage points relative to the comparison group). With a longer timeframe that makes data available, future work by CUNY is expected to provide information on differences in labor market outcomes for the two groups.

^33. The survey examined the cohorts graduating every 3 years, a sample composed of nearly 13,000 graduates. See Levin and Garcia (2013), Figure 4.

Additional information

Funding

Special thanks to the New York City Center for Economic Opportunity for support of this study.