College cost, financial aid, and for-profit college: Not a market failure
College cost, financial aid, and for-profit college: Not a market failure
For a long time, college tuition increased at a faster pace than inflation, and while the causes are certainly multifactorial, one seems to stick out like a sore thumb: student aid. This holds true to a greater extent for private for-profit colleges (FPCs). It is however commonly believed, hastily though, that schooling should not be governed by market forces because for-profit colleges produce worse education and earnings outcomes compared to non-profit colleges. Studies suggest there are too many unknowns.
CONTENT
1. Student aid raises tuition.
2. For-profit colleges fare worse. Market failure?
3. In defense of for-profit colleges.
Studies after studies on for-profit colleges (FPCs) in the US indicate that free markets failed. Research design limitations leave so many unknowns, but the worst problem is that the economist researchers never attempt to explain why education is another exception, in which free markets don’t work and simply lead to overpricing. Instead of using sound theory to interpret data, they seem to use data analyses to build incoherent theories.
Free market critics focused on the unfairly high tuition of FPCs (known to be very dependent on student aid), and its positive relationship with student aid, leading to high student debts. What is not told is that sustained low costs (through subsidized loans) not only inflate credentials by creating too much attendance but also mask impractical majors since tuition rarely varies by field of study. Moreover, Caplan (2018) suggested that if education was expensive, the employers know they would miss hidden talents by not employing these no-college applicants:
Education still signals something, but lack of education is not the kiss of death. When asked, “Why didn’t you go to college?” “I couldn’t afford it” is a great excuse. Heavy subsidies take it off the table. Indeed, what excuses are left? “I’m a bad test taker”? “I didn’t feel like going to college”? “I figured I could learn better on the job”? Once the good excuses are gone, employers have little reason to stay open-minded.
The combination of grants and subsidized loans with credit expansion distorts resource allocations by incentivizing unsustainable (mal)investment in degrees in higher education, the kind of malinvestment predicted by the ABCT. This may partially explain why one third of bachelor’s degree holders are in a job not requiring the credential (McCluskey, 2017).
1. Student aid raises tuition.
McCluskey (2016) evaluates three competing theories of the increasing college costs. First, the cost disease, which postulates that labor-intensive industries will see their costs rise because their labor cannot be replaced with technology but their workers need to get paid more lest they go to higher-paid areas. Second, state and local funding, which postulates that cuts to direct state support have forced colleges to raise tuition. Third, Bennett’s hypothesis, which postulates that student aid (from the government) makes it possible for colleges to raise their prices. McCluskey’s review concluded that Bennett’s hypothesis is much more likely than the two others. This makes intuitive sense. If the purchasing power of the customers increases, the sellers will be able to increase the prices. But outcomes other than price increases are possible, such as decreases in aid from schools, or other effects which reduce the value of the aid to the student. While there is some evidence for Bennett’s hypothesis overall, findings are more robust among for-profit colleges. This is probably because these schools are heavily reliant upon federal aid.
Cellini & Goldin (2014) use administrative data from 5 US states, focusing on Title IV tuition premium. Title IV are institutions that meet criteria to be eligible to participate in federal student aid programs. Their regression uses Title IV institutions (vs non Title IV), school and program characteristics, as well as program, year and country fixed effects, to predict tuition. Programs in Title IV institutions charged substantially more (about ±50 log points) in tuition depending on sample restriction, and this gap increases as the passing percentage on the licensing exam declines. Within T4 institutions, programs that are ineligible for Title IV student aid have tuitions that are identical to those offered by NT4 institutions. These results lead them to conclude that T4 schools raise tuition above the cost of education to capture federal aid.
Lucca et al. (2019) studied the effects of increases in loan limits for subsidized and unsubsidized loans on tuition for undergraduate programs, using IPEDS, NPSAS, and Title IV Administrative Data from the Federal Student Aid Office. In their regression, the outcome variable is changes in sticker tuition (i.e., advertised tuition before any aid, grants, scholarships are applied), the predictor of interest is sensitivity of tuition changes to program cap changes, which is measured by the interaction between the changes in loans (either subsidized, unsubsidized or Pell Grants) and the changes in program maximum caps, along with additional controls, fixed effects of year and institution. A 1 dollar increase in the subsidized cap, unsubsidized cap, and Pell Grants result in a 64-cent, 20-cent, and 21-cent (non-significant) increase in sticker tuition. The effect is most pronounced for more expensive degrees and degrees offered by for-profit and 2-year institutions. These models are robust. Changes in loan supply does not impact enrollment and is not affected by the exclusion of the Great Recession time period. Models with subsidized and unsubsidized loans have slightly reduced coefficients when a couple covariates (either school types or non-tuition funding sources) are added. Subsidized loans are robust with respect to parallel trends but not unsubsidized loans.
Kargar & Mann (2023) studied the 2011 tightening of credit standards for Parent Loans to Undergraduate Students (PLUS) loans, using DOE and IPEDS datasets. A decline in tuition growth rates at colleges more affected by the tightening would support Bennett’s hypothesis. A first regression uses year effects after 2010 and the school’s fraction of students in 2010 from households <$48,000 (which captures the effect of credit tightening) and their interaction as predictors of school-level growth rates of PLUS loans, enrollment, tuition revenues, and expenses. A second specification replaces this fraction of students with dummies for four increasing bins to examine heterogeneity effects (although binning is never a good idea). The increase in fraction of these “low-income” students from 0 to 1 predicts a 1.85% decrease in tuition growth after 2010. This implies that the availability of PLUS loans increases the price of college. Based on their empirical analysis, they compute the fraction that students receive out of the total (consumer + producer) surplus created by the subsidy, using the equation I/(I+1) where I is the ratio of consumer to producer surplus. This ratio was 0.6, meaning that students receive only 60 cents of each dollar of subsidy. This leads them to conclude that Bennett’s hypothesis is best explained by market-power explanation, i.e., colleges charging very large markups over marginal costs.
Black et al. (2023, Tables 4-7) studied the impact of the Graduate PLUS Loan Program (Grad PLUS) which eliminated federal loan limits, using administrative data from Texas Education Research Center (TERC). Differential exposure to increased borrowing limits should provide causal effects of increased loan access on various outcomes (degree, earnings, prices). A first regression uses as predictors the interaction between student classified as constrained borrowers and entry cohorts which gained access to Grad PLUS loans, as well as student characteristics, entry cohort and program fixed effects, to predict outcome such as degree completion or annual earnings. Robustness check showed no violation of the assumption of parallel trends, no change in enrollment or student composition due to Grad PLUS. Gaining access to federal loans did not lead to educational gains, did not lead to increased long-run earnings, but led to a small increase in the probability of having any earnings during the academic year. A second regression uses as predictors the interaction between percentage of students borrowing at the pre-Grad PLUS federal loan limit and cohorts that entered after the Grad PLUS program was created, along with entry cohort and program fixed effects, to predict program price. A 1% increase in constrained students leads to a $79 increase in average annual Grad PLUS loan and a $60 increase in average cost of attendance after Grad PLUS.
2. For-profit colleges fare worse. Market failure?
Private for-profit colleges (FPC) have been criticized for their profit-seeking tendencies and need to satisfy the interests of shareholders, leading to increased prices for students. This problem is even more unpleasant since FPCs target minority, low-income, and non-traditional students (e.g., disabled, military, single parents). A large body of studies showed nil returns from attending for-profit 2-year college and negative returns from attending for-profit 4-year college (Liu & Belfield, 2020). The observation that FPC students do not have better outcomes than their public counterparts, but they accumulate much greater debt, led Cellini & Koedel (2017) to call for regulations to fix “asymmetric information”. In a response, Gilpin & Stoddard (2017b, Table 1) argued that if their data were disaggregated, several findings are noteworthy: 1) FPC graduates have much greater 3-year default rates than public graduates but not 2- and 4-year FPC students, 2) FPC students rely on federal loans more but loan balances among borrowers are more similar when comparing students at the same level of education, 3) FPC students are more likely to receive Pell Grants than their public counterparts. The Gainful Employment regulation championed by Cellini & Koedel, instead of improving employment or default rates, may well reduce the enrollment of disadvantaged and non-traditional students. Furthermore, as Looney (2024) observed, borrowing increased faster than net tuition. The reason for this discrepancy is so unclear that it cannot be attributed right away to the market power of FPCs. There is however some evidence that students choose to pay more for their education because they enrolled in more costly programs than did students in earlier years.
The negative outcomes associated with for-profit colleges (FPCs), often considered as true treatment effects, could well be a selection effect. It is well known that students select into colleges based on factors and characteristics that are difficult to observe. Why it matters is because FPCs tend to attract high-risk groups. Hoxby & Avery (2013) showed that disadvantaged (low-income) students tend to make poorly informed decisions about college. Cornaggia & Xia (2024) found that loan mismanagement is more prominent among male and non-white students and that financial literacy does not explain these group differences. Obviously, race differences in IQ and behavioral traits that are partially heritable and explain such differentials in poor decisions are typically ignored in economics research. There is no such a thing as equality of outcomes. All in all, these findings illustrate how the asymmetric information hypothesis used ad nauseam by economists as evidence of market failure is so disconnected from reality.
Methodological limitations, heterogeneity of effects, and alternative interpretations give no credence to the view that the failure of for-profit colleges to perform better must be attributed to market failures.
Experimental studies that submit fictitious resumes to real job openings illustrate this complexity in interpretation. Darolia et al. (2015) used logistic regressions, adjusting for city, occupation, time trend, race and gender, and found a 0.5% higher probability of positive employer response for applicants from community colleges relative to for-profit colleges, and this holds true within each education level. There are several explanations. Perhaps job applicants who attended for-profit or community colleges do not differ in skills valued by employers, or perhaps employers are unaware of differences in quality between these college sectors. Darolia et al. recognized that their analysis cannot capture skill differences that only become apparent to employers at the interview stage or later, or effects that arise because of differences in the ability of colleges to link students to employers. Deming et al. (2016) used regression and found these results: 1) a callback rate of -0.2% and +0.9% for for-profit associates (AAs) and bachelors (BAs) from online institutions in business jobs not requiring a BA, 2) a callback rate of -2% for for-profit BA that is mainly driven by BAs from online institutions rather than local institutions in business jobs requiring BA, 3) a callback rate of -5% and -0.5% for for-profit certificates in health jobs that don’t require a certificate and require a certificate, respectively. These results could not be adjusted for biases resulting from employers’ perceptions of human capital differences between college sectors or their perceptions of preexisting differences between students.
Liu & Belfield (2020, Tables 3-4) analyze the US National Student Clearinghouse, which tracks students as they transfer to other Title IV–eligible colleges. They utilize parallel data sets from two statewide CCSs: one state that ranks close to the national average of SES metrics (CCS-A) and one state that is more affluent than the national average (CCS-B). Using fixed-effects regression they found wage penalties to transfer to a for-profit college instead of to a public or private nonprofit college for CCS-A but found wage gains for CCS-B. The robustness of their main analysis is confirmed by an analysis using propensity score matching of students based on both personal characteristics and prior college performance. Interestingly, in the CCS-A, the fixed-effects regression showed that the students with no bachelor’s degree had slightly lower earnings while the students with bachelor’s degree had greater earnings when transferring to for-profit colleges.
Armona et al. (2022, Table 8) use the QCEW, IPEDS, NSLDS, CSD and US Census to study default rates at for-profit colleges. Their model handles selection effects by exploiting the fact that people are more likely to enroll in college when labor demand is low because the cost of foregone earnings is reduced. This is done by using the interaction of demand shocks for college with pre-existing postsecondary supply. The expectation is that there is a larger relative increase in for-profit college enrollments in areas with a higher baseline supply of for-profit colleges when this common decline in labor demand occurs. The regression model includes areas and state-by-year fixed effects as well as controls for demographic changes over time. They found that students who attend for-profit colleges, instead of public colleges, take on more debt, have higher default rates, and lower earnings (standard errors are very large).
Even more studies display heterogeneity of effects, which complicates interpretation.
Cellini & Turner (2019, Tables 6-7) analyze the US DoED data, focusing on students who received aid and participated in “gainful employment” programs under which every for-profit college program is held accountable for student outcomes. They employ a difference-in-differences model, using students who never attended college as a control group, and predictors such as post-education time period (when the individual completes or withdraws from the program) and its interaction with for-profit college, and some fixed effects. For-profit certificate students are 2.8% more likely to be employed, and earn $356 more than no-college students. Conditioning on employment status, for-profit students earn 4.5 less in log earnings compared to no-college students. Men see positive gains relative to no college attendance but not women for earnings, whereas both groups see slight negative losses for log earnings. The estimates of benefit-cost (null or slightly negative) may not be trusted due to limitations of the analysis.
Lang & Weinstein (2013) showed that data quality affects the results. In an updated study that uses the BPS transcript data, which provide more information about majors, they found that for-profit colleges do not perform worse than non-profit colleges although there is some heterogeneity that is due to selection effects. In an earlier study that uses the BPS survey data, they found that for-profit colleges performed worse than non-profit colleges.
Gilpin & Stoddard (2017a) argued that current studies have not analyzed the economic outcomes thoroughly, using the correct identification, or over a long enough period of time:
Studies using individual fixed effects (e.g., Cellini & Chaudhary, 2014) can control for differing permanent characteristics of students. However, it is still difficult to control for the potential possibility that FPCs educate students whose earnings’ trajectories would be lower at any institution. Furthermore, current waves of longitudinal studies have relatively young students (under 26, when the median for-profit four-year student is 31) and short windows after graduation. Furthermore, FPCs expanded rapidly through 2012, making it challenging to generalize from studies of the 1990s or early 2000s. Finally, the central identification problem is matching non-traditional FPC students with the relevant counterfactual. In many cases it may be non-enrollment, or enrollment in a different field of study, rather than enrollment in the same field at a traditional college. The mixed literature, combined with the empirical difficulties in rigorous identification, suggests a cautious policy approach until for-profits have experienced a longer period of study.
3. In defense of for-profit colleges.
Just like in other markets, and as predicted by theory, for-profit colleges are more responsive to the needs of the labor market (Deming et al., 2015; Gilpin et al., 2015) while public colleges are less able to expand capacity in a timely manner due to the many government requirements for approval on budgets, tuition, and new programming (Cellini, 2010; Deming et al., 2012; Gilpin & Stoddard, 2017a). FPCs have innovated in ways that have diffused through higher education. Despite much higher tuition levels, students still choose to enroll in these colleges. Plausible explanations could be that the public sector leaves the students with little other choice or that these FPCs offer programs more tightly coupled with local labor-market demand, which could explain why for-profit and community colleges are not exactly substitutes (Soliz, 2018).
It is therefore curious to observe that private for-profit schools produce worse educational outcomes. Classical theory considers market force processes under unrealistic assumptions such as perfect competition or pure markets. In practice, even when companies are privatized, some regulations are still running or added, and crowding out effects due to public companies are still active. Unlike non-profit colleges, for-profit colleges lack endowments, donations, or state funding that could subsidize tuition. This is why public colleges can lower their tuition. Fried (2011) showed that non-profit colleges do make huge profits, close to for-profit colleges. This is not known because non-profit colleges report their profits as expenses. When endowments/donations and state subsidies are taken into account, private and public colleges make 12,807 and 11,000 dollars per student, annually. Fried (2011) also noted that the barriers to entry are very high.
One highly debated regulation is the 90/10 rule, and running for several decades by now, which states that no more than 90% of a FPC revenue could come from federal funding sources if the institution received funding under Title IV of the Higher Education Act. Violation of this rule jeopardizes its eligibility to receive Title IV funds. The purpose was to prevent low-quality FPCs from existing solely on federal aid. Ward (2019, Table 3) investigated whether for-profit colleges raise tuition to avoid a violation of the 90/10 rule whenever their ratio is over 90%, using propensity score matching. There is at best a very weak evidence of increased tuition (coefficient only significant at 0.10). However, there is evidence of exits following a violation, which could be due to school closure, due to exiting the Title IV program, or due to purchase of merging. Overall, this means the regulation is either not detrimental or its negative impact is very hard to detect. The study is weakly informative since 3% of the schools violated the rule. The lack of significance despite evidence of increased tuition may suggest that schools just under the critical threshold are adjusting their tuition in small increments, making it harder to statistically identify such effects. More importantly, it is very likely that the regulation redirected resource allocation toward marketing and compliance activities by investing in tracking systems or staff to monitor the ratio of federal to non-federal revenue. Another possible undetected behavior is that FPCs could raise the price of specific services in order to get additional non-Title IV revenue and stay under the 90% threshold.
The above evidence illustrates well the poor theories advanced by these economist researchers. They seem to build their theory based on data, leading them to conclude that free market forces will cause students to be overcharged due to profit-seeking tendencies. They never ask themselves why markets work in other fields, but not in education. The theory that privatization leads to overpricing without corresponding quality gain has not been confirmed empirically.
References
Armona, L., Chakrabarti, R., & Lovenheim, M. F. (2022). Student debt and default: The role of for-profit colleges. Journal of Financial Economics, 144(1), 67–92.
Black, S. E., Turner, L. J., & Denning, J. T. (2023). PLUS or Minus? The Effect of Graduate School Loans on Access, Attainment, and Prices (No. w31291). National Bureau of Economic Research.
Caplan, B. (2018). The Case Against Education. Princeton: Princeton University Press.
Cellini, S. R. (2010). Financial aid and for-profit colleges: Does aid encourage entry? Journal of Policy Analysis and Management, 29(3), 526–552.
Cellini, S. R., & Goldin, C. (2014). Does federal student aid raise tuition? New evidence on for-profit colleges. American Economic Journal: Economic Policy, 6(4), 174–206.
Cellini, S. R., & Koedel, C. (2017). The case for limiting federal student aid to for-profit colleges. Journal of Policy Analysis and Management, 36(4), 934–942.
Cellini, S. R., & Turner, N. (2019). Gainfully employed?: Assessing the employment and earnings of for-profit college students using administrative data. Journal of Human Resources, 54(2), 342–370.
Cornaggia, K., & Xia, H. (2024). Who Mismanages Student Loans, and Why?. The Review of Financial Studies, 37(1), 161–200.
Darolia, R., Koedel, C., Martorell, P., Wilson, K., & Perez‐Arce, F. (2015). Do employers prefer workers who attend for‐profit colleges? Evidence from a field experiment. Journal of Policy Analysis and Management, 34(4), 881–903.
Deming, D. J., Goldin, C., & Katz, L. F. (2012). The for-profit postsecondary school sector: Nimble critters or agile predators? Journal of Economic Perspectives, 26(1), 139–164.
Deming, D. J., Goldin, C., Katz, L. F., & Yuchtman, N. (2015). Can online learning bend the higher education cost curve?. American Economic Review, 105(5), 496–501.
Deming, D. J., Yuchtman, N., Abulafi, A., Goldin, C., & Katz, L. F. (2016). The value of postsecondary credentials in the labor market: An experimental study. American Economic Review, 106(3), 778–806.
Fried, V. H. (2011). Federal higher education policy and the profitable nonprofits. Cato Institute Policy Analysis, No. 678.
Gilpin, G. A., Saunders, J., & Stoddard, C. (2015). Why has for-profit colleges’ share of higher education expanded so rapidly? Estimating the responsiveness to labor market changes. Economics of Education Review, 45, 53–63.
Gilpin, G., & Stoddard, C. (2017a). Does regulating for-profit colleges improve educational outcomes? What we know, what we don’t know, and what we need to find out. Journal of Policy Analysis and Management, 36(4), 942–950.
Gilpin, G., & Stoddard, C. (2017b). Does regulating for-profit colleges improve educational outcomes? Response to Cellini and Koedel. Journal of Policy Analysis and Management, 36(4), 953–956.
Hoxby, C., & Avery, C. (2013). The missing “one-offs”: The hidden supply of high-achieving, low-income students. Brookings papers on economic activity, 1, 1–65.
Kargar, M., & Mann, W. (2023). The incidence of student loan subsidies: evidence from the PLUS program. The Review of Financial Studies, 36(4), 1621–1666.
Lang, K., & Weinstein, R. (2013). The wage effects of not-for-profit and for-profit certifications: Better data, somewhat different results. Labour Economics, 24, 230–243.
Liu, V. Y. T., & Belfield, C. (2020). The labor market returns to for-profit higher education: Evidence for transfer students. Community College Review, 48(2), 133–155.
Looney, A. (2024). How Much Does College Cost and How Does it Relate to Student Borrowing? Tuition Growth and Borrowing Over the Past 30 Years. National Tax Journal, 77(3), 605–622.
Lucca, D. O., Nadauld, T., & Shen, K. (2019). Credit supply and the rise in college tuition: Evidence from the expansion in federal student aid programs. The Review of Financial Studies, 32(2), 423–466.
McCluskey, N. (2016). Most college cost fixes miss the root problem. Cato Institute, 13.
Soliz, A. (2018). The effects of the expansion of for-profit colleges on student enrollments and outcomes at community colleges. Educational Evaluation and Policy Analysis, 40(4), 631–652.
Ward, J. D. (2019). Intended and unintended consequences of for-profit college regulation: Examining the 90/10 rule. Journal of Student Financial Aid, 48(3), 4.