The Value of a University Is Not Determined by Entrance Congestion—Defending Academic Support for "Low-Selectivity Universities" Through Evidence and Norms
Abstract
This paper addresses the question of whether public academic support for universities with low entrance selectivity, commonly referred to as "F-rank" universities (hereinafter "low-selectivity universities"), can be justified. The proposition this paper defends is as follows.
Entrance selectivity is not a direct metric for measuring the value-added of education or net social benefits. Therefore, it cannot be justified to uniformly exclude the students, faculty, researchers, and the intellectual functions maintained at these institutions from public support solely on the basis of low selectivity. The presence and method of support should be determined by comparing the value-added of education, research and regional functions, costs, alternative pathways, and the impact on students.
There is a distinction that must be made in advance. There are at least three layers of "support for universities." First, support for the opportunity for students to learn. Second, support for education, research, and regional functions. Third, support for the survival of a specific school corporation. What this paper directly defends is the eligibility for participation in the first and second, not an unconditional guarantee of the third. Furthermore, this paper does not prove the stronger proposition that "increasing institutional subsidies for low-selectivity universities more generally than at present is more efficient than other policies." That judgment requires separate verification of value-added using Japanese data and a cost-benefit comparison with alternative policies.
The following sections examine, in order: (1) the nature of selectivity metrics, (2) U.S. quasi-experimental studies on the causal effects of university education on marginal students, (3) comparisons with alternative pathways, (4) evidence from Japan and the UK, (5) normative grounds regarding the fairness of opportunity, and (6) responses to major counterarguments concerning signaling, costs, declining birthrates, and fiscal policy. Finally, the paper presents design principles for policy and the conditions under which a reduction in support would be justified.
1 Introduction—Deconstructing "What Are We Supporting?"
The phrase "supporting universities" conflates different targets. It includes supporting the opportunity for students to learn, supporting the conditions for faculty to provide education, supporting the foundation for researchers to continue their inquiries, and supporting the survival of the organization known as a school corporation. These are logically independent and can be separated in terms of policy. Even if a university corporation fails in its management and requires integration or withdrawal, there is no necessity for the learning opportunities, earned credits, research materials, and educational functions of the faculty to disappear along with it. Support for entities other than the corporation can continue in the form of student grants, guarantees for transfer to other universities, succession of curricula, and the preservation of research data and library resources. The first principle of this paper is "support people and functions, not corporations."
Beyond that, whether support for a certain university (or more precisely, a faculty or program) is socially worthwhile must be judged by comprehensively considering the value-added by education, external effects on research, the region, and public human resources, the necessary costs, the realistic alternative pathways for students, and who bears the benefits and burdens. Entrance selectivity can be one of the explanatory variables for predicting any of these, but it is not a metric that measures them directly. I will first confirm this point from the definition of the metric.
2 What Does the "F-Rank" Metric Measure?
The "BF (Border-Free)" classification from Kawaijuku, which is the origin of the colloquial term "F-rank," is a category assigned when it is impossible to set a deviation score range (borderline) where the probability of passing is 50%, due to reasons such as having few unsuccessful applicants in the previous year's entrance exam. This classification is assigned by faculty, department, and entrance exam method, not to the university as a whole, and Kawaijuku itself explains that entrance difficulty rankings do not indicate educational content or social standing [1].
In other words, what BF indicates is whether or not there was enough selection occurring in that entrance exam to set a 50% pass line; in other words, the selection pressure at the time of the entrance exam. Selection pressure is determined by the number of applicants, enrollment capacity, location, tuition, name recognition, the local 18-year-old population, entrance exam methods, and the application behavior of students. It is a quantity distinct from how much students grew through education (value-added), what roles they played after graduation, or what kind of research and knowledge preservation took place there. Selectivity is not uninformative because it can be related to outcomes through student composition, peer effects, and resources, but at the very least, selectivity cannot be equated with the value-added of education. The "entrance congestion" mentioned in the title of this paper is used as a metaphor for this selection pressure.
Demographics highlight this point. According to a 2025 survey by the Promotion and Mutual Aid Corporation for Private Schools of Japan, 316 out of 594 private universities (53.2%) did not meet their enrollment capacity. The background to the improvement from the previous year (59.2%) is pointed out to be the first reduction in enrollment capacity in 22 years and a temporary increase in the 18-year-old population [2]. All other things being equal, a decline in the local 18-year-old population can push down the number of applicants and selection pressure. According to estimates by a Ministry of Education, Culture, Sports, Science and Technology (MEXT) study group, the number of university entrants will decrease from approximately 627,000 in 2021 to approximately 460,000 in 2040, a decline of about 27% [3]. In the future, it is inevitable that the number of universities and faculties that become "low-selectivity" will increase structurally, regardless of the quality of education.
Furthermore, using selectivity as a criterion for subsidies has problems in terms of incentive design. Under a system that reduces subsidies due to low selectivity, universities can improve their evaluation by tightening enrollment capacity and rejecting applicants rather than improving education. An allocation that disadvantages universities that have expanded educational opportunities and favors universities that have strengthened selection is inconsistent with the goals of educational policy.
3 Post-Graduation Income and Value-Added Are Different Things
The most frequent error in university evaluation is to regard the average income of graduates as the educational power of the university. Post-graduation income conflates the differences between students that existed before enrollment and the differences added by the university during their enrollment. If universities are compared without removing the former, the differences due to the selection of entrants are counted as educational effects. Students at high-selectivity universities are often blessed with academic ability, household finances, information, and human networks before enrollment, so even if their post-graduation income is high, it is not necessarily the case that the university created all of it. Conversely, even if the income level of graduates from low-selectivity universities is relatively low, it may have improved compared to the counterfactual of not having gone to university.
Therefore, the policy question is not "Do graduates of low-selectivity universities earn more than graduates of elite universities?" but rather "What changes for the same person depending on whether they were able to enter that university or not?" Research approaching this question has been accumulated since the 2010s using methods such as Regression Discontinuity (RD) design, which utilizes the boundaries of admission criteria.
4 U.S. Quasi-Experimental Studies—Positive Effects of University Education on Marginal Students
4.1 Florida International University: Zimmerman (2014)
Zimmerman used the GPA admission criteria of Florida International University (the largest public university with the lowest selectivity in the state university system) to compare students with observable attributes near the threshold. Academically marginal students who just barely exceeded the threshold and gained admission had approximately 22% higher income 8 to 14 years after high school graduation, and this gain exceeded the cost of university; the gain was larger for male students and those eligible for free school lunches (low-income group) [4]. The rate of return per year of four-year university education was estimated to be approximately 8.7%, which was almost the same level as the average rate of return for all Florida high school students. At least in this setting, the common belief that "university education has little effect on academically marginal students" does not hold.
4.2 35 Public Universities in Texas: Mountjoy (2026)
Mountjoy's study, published in the Quarterly Journal of Economics, addresses a broader institutional scope [5]. By utilizing admission criteria across all 35 public four-year universities in Texas and comparing students who just barely met the criteria with those who just missed them, the study found that students who were marginally admitted received approximately one more year of four-year university education, their bachelor's degree attainment rate increased by about 12 percentage points, and their long-term earnings rose by approximately 8%. The estimated internal rate of return was 26% for the students themselves, 16% for society as a whole, and 7% for government finances.
There are two policy-relevant findings. First, there was almost no systematic relationship between university selectivity and the causal income gains obtained by marginal students. Conventional indicators that view average graduate income as a measure of a university's value-added overestimated causal value-added by about twofold and the correlation between selectivity and value-added by about threefold. Second, analysis by margin suggested that the effects were greater for students who would have lost access to four-year university education entirely if rejected, rather than for those on the threshold of being able to attend a 'more selective university.' Much of the marginal benefit of university education may stem from avoiding the loss of access to higher education, rather than from moving up one rung on the prestige ladder.
These are local average treatment effects (LATE) at the admission thresholds of specific public university systems in the United States, and the figures cannot be directly extrapolated to Japanese private universities. This is because tuition levels, labor markets, and entrance examination systems differ. However, these results are incompatible with the premise that university education for marginal students is uniformly ineffective. For Japan, the same type of verification using Japanese data is necessary—this is the accurate conclusion.
5 Comparison with Alternative Paths: University Is Not Always the Best Option
University attendance is not always the best option. Mountjoy (2022) decomposed the effects of expanding access to two-year community colleges in the U.S. into two causal margins: 'democratization'—the inclusion of groups who would not have received higher education otherwise—and 'diversion'—the redirection of groups who would have otherwise enrolled directly in four-year universities. While a significant positive value-added was confirmed for the former, negative effects occurred for students diverted from four-year to two-year institutions [6]. The distinction between democratization and diversion itself has been a classic point of discussion since Rouse (1995) [7].
The principle derived from this is that the value of an educational institution is not determined in isolation, but by comparison with the next-best alternative for the individual. The effect of the same university differs depending on whether the person would have gone to a different university, attended a vocational school, secured stable employment, or become unemployed if they had not enrolled in a low-selectivity university. This result indicates the need to support appropriate matching and ease of transition between universities, short-cycle higher education, professional education, vocational training, and employment, rather than making enrollment rates themselves the goal. The defense in this paper is conditional on this.
6 Evidence from Japan and the U.K.: Positive Averages, but Large Variance
6.1 Japan
The study by Nakamuro and Inui, which controlled for shared genetic and family factors by comparing identical twins, estimated the wage return per year of education at approximately 10% [8]. While twin comparisons do not control for individual-specific ability differences and are not an analysis by university rank, they show that the explanation that 'high-income earners only have high incomes because of their innate ability' is insufficient.
Shima (2018) is a study that directly addresses low-deviation-score private universities. This study estimated an average return on educational investment of approximately 5% for a group of low-deviation-score private universities, and also estimated expected wages for one private university and one social science department with a Benesse deviation score of less than 45 located in a regional area, primarily based on the industries and company sizes of male graduates' employment. The results showed an average expected return of approximately 4% and a median of approximately 5%; while about two-thirds were positive, about one-third did not show a positive return based on the assumptions adopted and the estimated expected wages [9]. As this is not a study that tracked actual lifetime income and is a case study of one university and one department, it is not possible to estimate the effect of all low-selectivity universities in Japan or the probability that an individual will actually suffer a loss from this. What can be said from this study is that the financial returns of low-selectivity private universities cannot be assumed to be uniformly zero.
6.2 The United Kingdom
In the U.K., lifetime earnings of degrees, controlling for prior academic ability and family background, have been estimated primarily for students in England using data linking tax records and education records (LEO). The 2020 estimates by the IFS (Institute for Fiscal Studies) showed a net gain of approximately £130,000 on average for men and £100,000 for women on a discounted present value basis; while enrollment was financially positive for about 80% of students, it was estimated that about 20% would have been better off financially if they had not enrolled [10]. It should be noted that this is a non-experimental estimate and cannot fully control for unobserved ability and preferences. The 2026 updated estimate, which reflected an additional seven years of actual income data, revised the average individual net return downward by approximately 30% from the previous estimate [11]. Furthermore, a separate 2025 study (LSE), which added non-monetary benefits such as health and happiness based on the 2020 estimates, estimated the total lifetime return for the average entrant at approximately £150,000, and approximately £80,000 even for students near the admission margin (using students at about 40 low-selectivity universities as a proxy), but this does not reflect the 2026 downward revision, and the figures may change [12].
There are two implications that can be drawn from the evidence in Japan and the U.K. First, it is highly possible that students who attend universities that accept many marginal students will have a positive return when measured by total lifetime returns on a monetary basis. Second, because the variance is large, even if the average for all students is positive, there may be a segment of about 20-30% for whom the return is negative. It is precisely because of this variance that a system based on the idea that 'you should borrow money at your own risk because you can expect a profit' cannot work. Looking at the individual level, educational investment is uncertain in terms of future income, and even if it fails, one cannot sell human capital to recover the principal. However, at the societal level, the returns on educational investment are large and span a wide range. If the costs of education are borne solely by individuals, it will lead to underinvestment below the socially optimal level. What is required is social risk-sharing, such as grant-based support, income-contingent repayments, and completion support.
7 Fairness of Opportunity: The Distributive Consequences of Limiting Support to High-Selectivity Tiers
According to the 2025 OECD report, in Japan, 72% of 25- to 34-year-olds whose parents have received higher education obtain higher education qualifications, whereas this figure is only 43% for those whose parents' highest level of education is upper secondary education or equivalent. This 29-percentage-point gap is larger than the OECD average (approximately 25 percentage points) [13]. Access to higher education remains strongly skewed across generations.
If public resources are concentrated only on high-selectivity universities in this situation, policy may reinforce a cycle where family educational resources produce high entrance exam scores, high entrance exam scores attract generous public resources, and those resources become the next generation's educational resources. Students at high-selectivity universities should naturally be supported as well, but a distribution where those who have already received many educational resources receive even more in the future only puts public support on the side of reproducing existing inequalities.
The normative basis of this paper is not an exhaustive list of theories, but is narrowed down to the following two points. First, entrance exam scores are influenced not only by an individual's effort and potential, but also by the education they have received up to that point, family income, parental education, regional educational environment, illness or disability, care responsibilities, and information access. Therefore, making entrance exam scores alone a sufficient condition or an exclusion condition for support eligibility can be contrary to fairness of opportunity. Second, the appropriateness of support should be determined by a comparison between social net benefits and realistic alternatives, and selectivity is not a proxy for this (Sections 2-4). Note that from a position that emphasizes fairness of opportunity, investment in primary and secondary education may be prioritized over additional investment at the university level. The argument of this paper is that 'there is no basis for uniformly excluding the low-selectivity tier from support at the university level,' and it does not determine the optimal allocation between educational stages.
In fact, there is descriptive evidence suggesting that the social value of a university depends on 'to whom it opens its doors.' In a study by Chetty et al. using U.S. tax and education linkage data, when social mobility is measured by 'acceptance rate of low-income students × upward mobility rate after graduation,' many examples are observed where mid-tier public university groups rank higher than the most selective private universities [14]. While this is not an estimation of causal effects, it shows what evaluations that look only at the success per graduate might miss.
In Japan, a new higher education support system (grant-type scholarships and tuition reduction/exemption) was created in 2020, and from fiscal year 2025, it has been expanded for multi-child households to include tuition and admission fee reductions up to the government-defined upper limit without income restrictions (subject to asset requirements and institutional requirements) [15]. The design, in which support is tied to the individual student rather than the institution and can be used at a wide range of universities and vocational schools that meet the requirements, is consistent with the first principle of this paper. At the same time, it should be noted that academic requirements regarding attendance rates, credits earned, and grades, as well as institutional requirements regarding educational systems, finances, and information disclosure, are imposed, and it is not an unconditional grant.
8 Benefits Beyond Income
Adults in Japan who have received higher education score an average of 33 points higher on the OECD Survey of Adult Skills (PIAAC) literacy assessment than those whose highest level of education is upper secondary or equivalent [13]. While this is not a direct causal effect, it indicates that higher education history is associated not only with income but also with information-processing skills.
The benefits of higher education can also include increasing the number of people who can understand and utilize specialized knowledge in the workplace and in civic life—for example, nurses reading research findings, teachers avoiding misinterpretation of statistics, municipal employees conducting policy evaluations, and citizens scrutinizing medical information or political claims. These benefits are a separate pathway from cutting-edge knowledge production, and the two are complementary. However, the magnitude of this type of externality cannot be directly estimated from the studies cited in this paper. Here, I limit myself to the observation that the objective function of higher education policy can include not only the strengthening of the top end of the distribution but also the raising of the average level of comprehension.
9 Research Funding Allocation—Concentration on Prestige Lacks Proof of Efficiency
There are fields in research funding that require large-scale equipment or long-term teams, making a certain degree of concentration essential. However, it does not follow that all additional research funding should be directed to the currently most prestigious universities. There is uncertainty in the value of research, and because prestige and resources operate in a cycle where past prestige attracts research funding and talent, which in turn generates results and further prestige, one cannot judge the efficiency of allocation based solely on the observation that famous universities produce many results.
Regarding empirical evidence at the researcher level, Fortin & Currie (2013), who analyzed natural science grants in Canada, report that the relationship between research funding and outcomes such as papers and citations is positive but weak, and that holders of large grants have lower output per dollar [16]. Ohniwa et al. (2023), who analyzed KAKENHI grants in the life sciences and medical fields in Japan, also report that small-scale grants of less than 5 million yen per researcher per year, and categories distributed widely beyond top performers, have higher efficiency in creating emerging research themes per investment amount [17]. Both are analyses at the researcher/research-funding unit level and do not directly compare base funding allocation between university ranks. Therefore, what can be said here is that, at the very least, there is doubt about justifying additional allocation based solely on prestige without verifying the marginal benefit of research funding concentration. An allocation strategy that combines concentrated investment in large-scale facilities with broad-based research funding, small-scale exploratory grants, and shared use of equipment data is worth considering.
10 Regional Functions—Neither Endorsing the Existence Effect Theory nor the Worthlessness Theory
The summary of the MEXT review committee states that private universities educate approximately 80% of undergraduate students and are responsible for fostering essential workers and industrial talent, research, and regional revitalization, thus advocating for support including base funding. At the same time, it also calls for prioritization based on functions and outcomes, reorganization and integration, and the smooth withdrawal of universities that cannot guarantee quality [3]. While this document indicates the recognition of policy authorities and does not independently verify the social effects of private universities, it can be referenced as a framework that does not treat support and reorganization as mutually exclusive.
On the empirical side, Yanagiura & Tateishi (2024), who examined the regional economic effects of small, non-research-oriented private universities in Japan using a prefectural panel from 1955 to 2015, report that the positive relationship between the number of universities and per capita prefectural GDP was strong mainly during the period of high economic growth, weakened thereafter, and that no sustained effect on regional human capital or innovation was confirmed, with effects remaining limited to temporary capital investment stimulus [18]. Although this study does not directly define and compare low-selectivity universities, the results do not support the assumption that the mere existence of a university generates sustained regional effects.
What follows from this is the conclusion to question the functions of a university, not its mere existence. If regional effects are to be used as a basis for support, it is necessary to specifically design and verify functions such as human resource development that meets regional demand (nursing, childcare, education, welfare, etc.), connections to local employment, joint projects with local governments, hospitals, and small and medium-sized enterprises, continuing education for working adults, and the sharing of libraries and research facilities.
11 Responding to Counterarguments
11.1 "Isn't it just signaling in the end?"
If the high income of degree holders functions only as proof to employers of their original ability rather than as a result of skill formation, then a significant portion of private benefits is not social benefit. If everyone competes to obtain higher degrees, losses such as credential inflation also occur. The studies in Sections 4 and 6 also do not fully decompose wage increases into human capital formation and signaling, and if wages are viewed as productivity, there is a possibility that social returns are overestimated.
This counterargument is valid. However, what follows from it is not the cessation of support for low-selectivity universities, but a design requirement to link support to actual learning and capability formation rather than the issuance of degrees. This would include measuring learning outcomes before and after enrollment, making graduation requirements substantive, connecting with qualification exams and internships, auditing grade evaluations, tracking learning achievement, and enabling mutual transition with non-university qualification and vocational education. In addition, the more the signaling hypothesis holds true, the weaker the basis for linking subsidies to entrance rankings (selectivity) becomes, and the stronger the basis for linking them to exit substance (learning outcomes) becomes. A combination of broad access, rigorous achievement standards, and sufficient learning support is a response to this requirement.
11.2 "Education for students with low academic ability is too costly"
If learning support, small-group instruction, basic education, and life support are necessary, the cost per student can be high. However, high cost and negative cost-benefit are different things. According to Mountjoy's (2026) estimates, public university education for marginal students had an internal rate of return of 7% even for government finances [5]. Furthermore, if part of the cause of learning difficulties lies in the education system or family environment prior to the student's arrival, it is not fair to treat the cost of remediation solely as the inefficiency of the university that finally accepted the student.
In terms of evaluation design, ranking universities by the raw average of their graduates creates an incentive to select only students who are easy to educate. This is the same structure as evaluating hospitals solely by the health of discharged patients, which would result in hospitals that refuse severe patients receiving high evaluations. What is needed is value-added evaluation that adjusts for student characteristics at the time of enrollment. At the same time, because the mechanical application of a single indicator invites indicator manipulation, a combination of multiple indicators—such as learning outcomes, completion rates, student debt, employment, qualifications, student composition, and regional functions—along with qualitative review and auditing is required.
11.3 "Since the population is declining, not all universities can remain"
That is correct, and this paper does not argue for the survival of all corporations. Under the prospect of a significant decline in the number of students entering higher education [3], integration, reorganization, and withdrawal are inevitable. However, the withdrawal of a university is different from the closure of a regular store. Students have entrusted their earned credits, qualification programs, academic records, scholarships, housing, job hunting activities, and graduation qualifications to the university over several years. Withdrawal must be accompanied by guarantees of education until graduation for current students, credit transfer and transfer guarantees, preservation of academic and research data, support for commuting and moving expenses, succession of curricula, and alternatives for regional access to higher education. There are multiple paths to preserving functions without preserving the corporation, such as integration, joint operation, online course sharing, inter-university consortia, conversion to short-term higher education, and becoming a hub for adult education. The distinction between academic support and organizational rescue is a distinction that includes this withdrawal design.
11.4 "Japanese universities are already too protected by public funds"
According to 2022 data published in the OECD's Education at a Glance 2025, government expenditure on higher education in Japan was $8,184 per student in purchasing power parity, below the OECD average of $15,102, and the share of public funding in higher education costs was 37.5% in Japan compared to the OECD average of 67.4% (as higher education expenditure includes R&D costs, this is not a simple comparison of tuition support amounts) [13]. While optimal expenditure cannot be determined by international comparisons alone, this at least does not align with the image that "Japanese university education is excessively protected by public funds."
Under a system centered on private institutions with high household burdens, there are students who face a dispersion of expected returns (Section 6) and completion risks. If public funds are uniformly reduced under these conditions, there is no guarantee that exit will occur in order of lowest educational quality. This is because university exit is influenced not only by quality but also by location, demographics, household constraints, and reputation. Families with information and assets can move to other options, but students who cannot afford relocation costs or tuition may be forced to exit higher education entirely.
12 Policy Design Principles
Justifying support for low-selectivity universities is different from justifying unconditional institutional subsidies. The design principles derived from Sections 1–11 are as follows.
First, tie support to individual students and make it portable. This includes grant-type scholarships, tuition waivers, living expense support, income-contingent repayments, and emergency grants. Given the dispersion of returns, the depth of completion support and repayment protection is necessary.
Second, support not just access but completion. This includes foundational subjects, individual tutoring, academic advising, mental health, disability and childcare support, and flexible scheduling. A system that admits students, collects tuition, and makes it the individual's responsibility if they fail to graduate cannot be called support.
Third, evaluate by department or program rather than by university. Educational effectiveness, qualification attainment, research capacity, and regional demand differ by department.
Fourth, look at value-added adjusted for entry background, employment stability, and public roles, rather than raw employment rates or average wages. Do not undervalue socially necessary fields like childcare, education, welfare, and nursing based solely on average wages, even if they are low-paying.
Fifth, combine the concentration of research funds into large-scale facilities with broad foundational expenses, small-scale exploratory grants, and equipment/data sharing.
Sixth, force the exit of organizations that do not meet quality standards. If there is false advertising, significantly low educational reality, fraudulent grading, continuous financial crisis, or a lack of student protection, public support should be suspended, and integration or withdrawal should be pursued. However, even in such cases, students must be protected through the withdrawal design described in Section 11.3.
13 Conditions Under Which Support Reduction Is Justified
I will clarify what evidence would overturn the defense in this paper. The criterion is the following:
If reliable comparisons show that the social net benefit of support for a particular educational program, including student transition costs, consistently falls below the social net benefit of realistic and accessible alternatives (such as transfer support to other universities, professional education, vocational training, or employment support), then the reduction, redesign, or abolition of support for that program is justified.
The judgment criteria for this comparison include changes in learning, completion, employment, income, health, and social participation compared to the counterfactual of similar students; external effects such as research, regional impact, public human resources, and knowledge preservation; necessary costs and opportunity costs; and the potential for improvement through additional support. Even if there are some benefits, if a more beneficial alternative is realistically available at the same cost, redesign is justified. We should not impose a burden of proof that is favorable only to the defense.
Furthermore, it must be confirmed that this judgment cannot be substituted by "because it is a bottom-tier (BF) university." While the possibility that BF status is statistically correlated with the above judgment criteria cannot be denied, none of these are measured directly [1]. Selectivity may be a trigger for scrutiny, but it is not a criterion for determining support eligibility.
14 Conclusion
I will summarize the arguments of this paper in line with the scope of the evidence. First, entrance selectivity is not a direct indicator of educational value-added and fluctuates with demographics. Second, in specific US public university systems and at the margin of admission, positive causal effects of university education on academically marginal students have been confirmed, and the relationship between selectivity and causal value-added was weak [4][5]. The premise that education for marginal students is uniformly ineffective cannot be maintained. Third, the effect of advancing to higher education is determined by comparison with the next-best option, and if the comparison is wrong, negative effects can occur [6]. Fourth, evidence from Japan and the UK suggests that while the average return on higher education, including for the low-selectivity tier, can be positive, the segment for whom returns are negative cannot be ignored [9][10]. This is precisely why support should take the form of student grants, completion support, risk sharing, value-added evaluation, and withdrawal design, rather than unconditional subsidies. Fifth, allocating support exclusively to the high-selectivity tier means that public funds are validating intergenerational educational inequality [13].
At the same time, not all universities and departments have the same value, not all students benefit financially, not all corporations need to be preserved, and university education is not necessarily the best path for everyone. Quality assurance and exit systems are necessary. However, these reservations support the conclusion that the conditions and design of support must be precise, not that uniform exclusion based on low selectivity is justified.
Being low-selectivity is not a sufficient reason to refuse public support. The question to ask is what value that educational program adds to whom, and whether it is worth the cost compared to realistic alternatives.
References
[1] Kawaijuku, 'Entrance Exam Difficulty Ranking Table' and its guide, Kei-Net. https://www.keinet.ne.jp/exam/ranking/
[2] The Promotion and Mutual Aid Corporation for Private Schools of Japan (2025), 'FY2025 Trends in Private University and Junior College Enrollment Applications'. https://www.shigaku.go.jp/files/shigan_doukouR7.pdf
[3] Ministry of Education, Culture, Sports, Science and Technology (2026), 'Summary of Deliberations by the Council on the Future of Private Universities Walking with Society Toward 2040' related materials. https://www.mext.go.jp/content/20260611-mxt_koutou01-000050360_11.pdf
[4] Zimmerman, S. D. (2014). The Returns to College Admission for Academically Marginal Students. Journal of Labor Economics, 32(4), 711–754. https://doi.org/10.1086/676661
[5] Mountjoy, J. (2026). Marginal Returns to Public Universities. Quarterly Journal of Economics, 141(1), 429–497. https://academic.oup.com/qje/article/141/1/429/8376650
[6] Mountjoy, J. (2022). Community Colleges and Upward Mobility. American Economic Review, 112(8), 2580–2630. https://doi.org/10.1257/aer.20181756
[7] Rouse, C. E. (1995). Democratization or Diversion? The Effect of Community Colleges on Educational Attainment. Journal of Business & Economic Statistics, 13(2), 217–224.
[8] Makiko Nakamuro & Tomohiko Inui (2012), 'Estimating the Returns to Education Using Twin Data', RIETI Discussion Paper 12-E-076 (Non-technical summary). https://www.rieti.go.jp/jp/publications/nts/12e076.html
[9] Kazunori Shima (2018), 'The Utility of University Education: Mean and Variance—Focusing on Low-Deviation-Score Private Universities', Quarterly Personal Finance, Autumn 2018 Issue, Yu-Cho Foundation. https://www.yu-cho-f.jp/wp-content/uploads/2018autumn_articles03.pdf
[10] Britton, J., Dearden, L., van der Erve, L., & Waltmann, B. (2020). The Impact of Undergraduate Degrees on Lifetime Earnings. IFS Report R167. https://ifs.org.uk/publications/impact-undergraduate-degrees-lifetime-earnings
[11] Institute for Fiscal Studies (2026). New Estimates of the Impact of Undergraduate Degrees on Lifetime Earnings. https://ifs.org.uk/publications/new-estimates-impact-undergraduate-degrees-lifetime-earnings
[12] Frayman, D. (2025). The Full Returns to Higher Education in the UK. CEP Occasional Paper No. 71, London School of Economics. https://cep.lse.ac.uk/pubs/download/occasional/op071.pdf
[13] OECD (2025). Education at a Glance 2025: Japan Country Note. https://www.oecd.org/en/publications/education-at-a-glance-2025_1a3543e2-en/japan_8f0a8541-en.html
[14] Chetty, R., Friedman, J. N., Saez, E., Turner, N., & Yagan, D. (2017). Mobility Report Cards: The Role of Colleges in Intergenerational Mobility. Opportunity Insights. https://opportunityinsights.org/wp-content/uploads/2018/03/coll_mrc_summary.pdf
[15] Ministry of Education, Culture, Sports, Science and Technology, 'New System for Higher Education Support' (Established in FY2020. From FY2025, tuition and other fees are reduced or exempted up to the national limit for multi-child households without income restrictions. Subject to asset, academic, and institutional requirements). https://www.mext.go.jp/a_menu/koutou/hutankeigen/index.htm
[16] Fortin, J.-M., & Currie, D. J. (2013). Big Science vs. Little Science: How Scientific Impact Scales with Funding. PLOS ONE, 8(6), e65263. https://doi.org/10.1371/journal.pone.0065263
[17] Ohniwa, R. L., Takeyasu, K., & Hibino, A. (2023). The Effectiveness of Japanese Public Funding to Generate Emerging Topics in Life Science and Medicine. PLOS ONE, 18(8), e0290077. https://doi.org/10.1371/journal.pone.0290077
[18] Yanagiura, T., & Tateishi, S. (2024). Local Economic Impact of Small, Non-research Private Universities: Evidence from Japan. Economics of Education Review, 102, 102576. https://ideas.repec.org/a/eee/ecoedu/v102y2024ics0272775724000700.html
Note: Many of the quasi-experimental studies cited in this paper estimate local causal effects within specific countries and institutional contexts. Applying these findings to low-selectivity private universities in Japan requires careful consideration of differences in tuition, labor markets, and entrance examination systems. Furthermore, the definition of 'treatment'—whether it refers to 'admission eligibility,' 'years of enrollment,' or 'degree attainment'—varies by study. Numerical data and institutional information are based on public information available as of the time of writing (August 2026).