Assessing Education Risk

An actuarial perspective on student borrowing and outcomes

BY CINDY HU April 2026

---

In my experience, when people think about insurance, education rarely comes to mind. Insurance exists primarily to protect against life’s major risks: premature death, illness, disability or the destruction of valuable property. Yet for many individuals, as the Federal Reserve notes, the single largest investment they make before middle age is neither a home nor a business—it is their education.

At age 17 or 18, millions of students face one of the most consequential financial decisions of their lives: whether to attend university, where to enroll and how to finance the next four years. For those without full parental support or scholarships, the answer is typically debt. Student loans have become the dominant mechanism for financing higher education across much of the developed world. In the United States alone, outstanding student loan debt has ballooned to around $1.8 trillion, held by more than 42 million borrowers—making it one of the largest categories of consumer debt after mortgages and auto loans.¹

On average, borrowers carry roughly $37,000–$39,000 in federal loan debt, and about half of recent college graduates leave school with some form of debt. The prevalence of borrowing varies by institution type and region, but the scale is unmistakable: For many borrowers, student loans are not just a financing tool, they represent long-term financial obligations that can influence career choices, household formation and lifetime wealth accumulation.

From a cash-flow perspective, student loans appear well-structured. Payments are predictable, interest rates are known up front, and amortization schedules are clear. From a risk-allocation perspective, this structure places fixed obligations alongside variable and uncertain returns. Repayment is contractually fixed while the economic return on the cost of the education purchased remains uncertain and varies across individuals and employment circumstances.

Consider just a few examples. Outcomes can differ substantially even when students borrow the same amount. Earnings differ by field of study: Workers with STEM bachelor’s degrees have a median annual salary of about $98,000, compared to $69,000 for arts and humanities majors, highlighting how academic specialization can shape economic returns.²

Starting salaries also vary across fields: Engineering and computer science graduates often begin their careers near or above $76,000, while many humanities majors start closer to $48,000–$52,000. These wage disparities contribute to lifetime earnings differences by major—a pattern noted in research showing that lifetime earnings variation across fields can rival or even exceed the lifetime gap between college and high school graduates.^3,4

Institutional context matters too. Graduates from some selective universities report high median earnings above $140,000, experiencing very different economic outcomes than graduates from institutions with much lower median earnings.⁵

Timing also plays a role. Research on “recession scarring” shows that entering the labor market in a downturn can suppress earnings early in a career relative to stronger economic conditions, with effects that can influence long-term trajectory even for otherwise similar graduates.⁶

Under the prevailing financing model, these materially different outcome paths are funded with nearly identical debt contracts. A substantial dispersion of outcome risk is borne by the individual borrower. In effect, higher education is a risky investment financed predominantly with instruments that concentrate risk on the individual, in contrast to other high-stakes economic decisions, where some portion of uncertainty is shared, pooled or otherwise mitigated rather than borne entirely by a single party.

EDUCATION AS AN ASSET

Economists have long modeled education as an investment in human capital. Individuals incur costs today—tuition, living expenses and forgone earnings—in exchange for higher expected income in the future. On average, this investment does appear to pay off: Workers with a bachelor’s degree earn more than those with only a high school diploma. In 2023, workers with a bachelor’s degree earned, on average, about 61% more in annual wages than high school graduates. In another analysis, the typical U.S. college graduate earned about $1.19 million over a career, more than twice the lifetime earnings of a typical high-school graduate.^7,8

But averages conceal dispersion. Earnings vary not just across educational groups but within them. For example, among college graduates, the earnings distribution is broad: In 2021, one quarter of male bachelor’s degree holders earned less than about $56,000, and one quarter earned more than about $126,000, and similar variation exists for women. Moreover, some institutions produce cohorts in which graduates, on average, earn less than typical high school graduates.^9,10

The return on education depends on factors largely unknown when a student enrolls: academic success, labor market demand for particular skills, technological change, health, family obligations, regional economic conditions, and even macro shifts like recessions or pandemics. Although students can choose factors like major, field of study and institution beforehand, many of the most important determinants of long-term earnings are unpredictable at the time of enrollment. As a result, two students with identical tuition bills may experience materially different outcomes a decade later, with some college graduates earning less than the median high-school graduate and others earning well above the average.

From an actuarial lens, this is a classic problem of risk allocation, in my view. Outcome variance is high, but the financing mechanism does not reflect it. Unlike mortgages, student loans are not collateralized by an asset that can be repossessed. Unlike business loans, they are rarely contingent on the future performance of the underlying investment. And student loans are generally not dischargeable in bankruptcy, unlike some other forms of unsecured consumer debt. This means that borrowers must repay their loans regardless of whether their earnings are high or low.¹¹

From a financial perspective, borrowers commit to fixed repayment obligations that are not contingent on realized earnings. I would characterize this as exposing borrowers to a non-insured option on their future earnings. By taking on fixed, non-contingent debt to finance education, students commit to repayment obligations that do not adjust with realized outcomes. If earnings turn out high, the economic “payoff” materializes; if earnings are low, borrowers still owe the same payments, with no mechanism for shared downside risk.

RISK-SHARING MECHANISMS

Some attempts at risk sharing already exist within the current system of higher education finance, though they remain limited in scope and ambition. These mechanisms acknowledge—implicitly or explicitly—that education outcomes are uncertain and that fixed repayment obligations can be poorly matched to realized earnings. Yet in practice, they operate as incremental adjustments layered onto a fundamentally debt-based framework rather than as a rethinking of how education risks might be allocated.

The most prominent example is income-driven repayment (IDR). By tying monthly loan payments to a borrower’s income and offering forgiveness after extended repayment periods, IDR partially insulates borrowers from the worst financial consequences of low earnings. In effect, these plans introduce a degree of earnings contingency into an otherwise rigid contract, functioning as a crude form of income insurance. When income falls, payments fall as well, reducing default risk and preventing acute financial distress. However, the protection IDR provides is both incomplete and uneven.¹²

The system can be administratively complex, requiring active enrollment, annual income recertification and navigation of a shifting set of rules that might be poorly understood by borrowers. Many eligible individuals unintentionally fail to enroll or exit these plans, undermining their effectiveness as risk-sharing mechanisms. Moreover, forgiveness typically occurs only after decades, meaning that even under IDR, borrowers may spend much of their working lives constrained by debt obligations that influence career choice, savings behavior and household formation. Importantly, IDR does not pool risk ex ante (account for or price risk in advance) or price uncertainty into the financing contract itself; instead, it functions as a back-end safety net that absorbs some losses after the fact, often through taxpayer-funded forgiveness.

Income share agreements (ISAs) represent a more direct attempt to align financing with outcome uncertainty. Rather than borrowing a fixed principal, students agree to pay a percentage of future income for a defined period, explicitly tying repayment obligations to realized earnings. In theory, this structure transforms education financing from debt into a contingent claim on human capital, spreading risk across participants and eliminating the possibility of unaffordable fixed payments.¹³

High earners contribute more, low earners less, and those whose educational investment yields weak labor-market outcomes are protected from financial ruin. Despite this conceptual appeal, ISAs have struggled to scale. Regulatory uncertainty, concerns over consumer protection and questions about how to classify these contracts have limited their adoption. Adverse selection poses another challenge: students who anticipate lower earnings may be more likely to choose ISAs, raising costs and discouraging investor participation. As a result, most ISA programs remain small, institution-specific experiments rather than a systemic alternative to student debt.

Scholarships and grants also reduce exposure to education risk, but they do so only ex ante and for a subset of students. By lowering or eliminating up-front costs, they reduce the magnitude of borrowing required and therefore the severity of downside outcomes. However, they are not designed to hedge post-graduation uncertainty. Grants do not adjust based on realized earnings, nor do they provide protection for students whose labor-market returns fall short of expectations. Instead, they function as targeted subsidies, mitigating risk for some while leaving the broader population exposed to the same structural uncertainty.

INSURANCE INVOLVEMENT

Given the scale of investment and the degree of uncertainty involved, the limited role of insurance-based solutions in higher education finance is notable, in my opinion. Several explanations have been offered, centered on moral hazard, measurement difficulty and correlation risk. Each contains an element of truth, yet taken together, I believe they do not fully account for why education remains almost entirely uninsured. Here’s an examination of the three:

1. Moral hazard. If the downside risk of education were meaningfully reduced, students might alter their behavior in ways that increase expected losses. Educational outcomes are shaped in part by decisions that remain under the student’s control: choice of field of study, institution attended, academic effort, willingness to relocate and even the selection of employment offers accepted after graduation. Insurance against poor outcomes could, in theory, weaken incentives to pursue high-return majors, attend more rigorous programs or aggressively search for well-compensated employment. From an underwriting perspective, as I see it, this introduces the classic problem of insured parties influencing the probability and magnitude of loss after coverage is in place.
2. Measurement. The “loss” associated with education is typically difficult to define. Earnings are only one dimension of educational outcomes, and they evolve over decades rather than at a single point in time. Career paths are nonlinear, interruptions are common, and income trajectories may recover after early setbacks. Distinguishing between temporary underperformance and permanently low return is difficult, complicating both pricing and claims adjudication. Unlike property damage or mortality, educational outcomes lack a clear terminal event that triggers a payout, making standard insurance structures harder to apply.
3. Correlation. Educational returns are highly sensitive to macroeconomic conditions, and negative shocks tend to affect entire graduating cohorts simultaneously. Recessions, technological shifts or industry-wide contractions can depress earnings across large segments of the labor market at once. This undermines the basic insurance principle of diversification, where independent risks can be pooled and smoothed over time. From the insurer’s perspective, education risk resembles a systematic factor rather than an idiosyncratic one, raising concerns about capital adequacy during downturns.

These challenges are not unique to education, nor are they insurmountable. Insurers routinely underwrite long-duration risks with substantial behavioral components. Disability insurance, for example, must contend with moral hazard, imperfect observability, and outcomes that depend partly on individual effort and choices. Catastrophe risk, despite its extreme correlation, is managed through reinsurance, geographic diversification, and capital market instruments such as catastrophe bonds. Even life insurance involves underwriting future behavior—health choices, occupation and lifestyle—that affect outcomes over long horizons.

Seen in this light, the absence of education insurance appears to me less a consequence of fundamental infeasibility than of historical convention and the way existing financing structures became entrenched over time. Student loans became the default financing instrument long before modern data infrastructure, income tracking and risk markets could support more sophisticated designs. The result is a system in which one of the largest and riskiest investments individuals make remains financed almost entirely without formal risk pooling, even though comparable risk-sharing mechanisms exist in other domains.

EDUCATIONAL OUTCOMES

A more robust response to the risk inherent in higher education could begin by explicitly recognizing education as an insurable exposure rather than treating it as a near-riskless investment financed through fixed debt. This would require shifting the focus from repayment mechanics to outcome variability, and from individual balance sheets to pooled risk structures capable of absorbing uncertainty across people, time and institutions.

One possible approach is cohort-based earnings insurance. Under this model, students or graduates might pay a modest premium—either up front or over time—in exchange for protection against earnings falling materially below a defined benchmark. That benchmark could be tied to expected earnings for a given field of study, institution or graduating cohort, rather than to an absolute income level. Losses would be pooled across large populations, allowing adverse outcomes for some individuals to be offset by stronger-than-expected outcomes for others. Unlike income-driven repayment, which adjusts payments only after debt is incurred, earnings insurance could price uncertainty ex ante and explicitly insure against downside realizations rather than merely deferring them.

Institution-level risk sharing represents a complementary avenue. Universities could, in principle, retain partial exposure to their graduates’ post-completion outcomes, co-insuring the economic value of the education they provide. This could take the form of contingent payments, outcome-based rebates, or participation in pooled insurance structures tied to alumni earnings. Such arrangements could internalize some of the risk currently borne by students, potentially creating incentives for closer alignment among tuition levels, program offerings and labor market demand. Institutions might receive stronger signals about the economic performance of different programs, potentially encouraging investments in career placement, curriculum relevance and student support where returns are weakest.

Capital markets could also play a role in absorbing education risk. Just as longevity risk, catastrophe risk and mortgage credit risk have been securitized, diversified claims on human capital could be structured and distributed to long-term investors. Properly designed, such instruments could offer exposure to broad population-level earnings growth rather than idiosyncratic individual outcomes, making them suitable for pension funds, insurers and other investors with long horizons. By transferring risk away from individuals and into deep capital markets, these structures could increase the capacity to bear uncertainty while lowering the cost of protection for students.

Actuaries are particularly well-positioned to contribute to this reimagining of education finance. The challenge is fundamentally actuarial: modeling lifetime earnings distributions, quantifying variance and tail risk, accounting for cohort effects and macroeconomic shocks, and designing pooling mechanisms that remain solvent under stress. Translating uncertain, long-duration human capital returns into sustainable pricing structures is not unlike the problems actuaries have addressed in insurance, pensions, and social risk programs for decades. What is missing is not technical capability, but market and organizational structures for institutional adoption.

A LOOK AT RISK MANAGEMENT

Higher education is often framed as a pathway to opportunity, and for many individuals it is. But it is also a leveraged financial commitment made early in life, under conditions of profound uncertainty, with consequences that can shape economic outcomes for decades. Borrowers experience divergent results, and the current financing system does not account for the risks. There is a tension between how education financing is often structured and the variability in post-graduation outcomes.

Current student loan arrangements emphasize predictability of repayment while leaving outcomes largely uncertain. From a risk-management standpoint, this creates an imbalance between fixed obligations and variable returns, particularly for borrowers whose post-graduation earnings fall below expectations. Existing mechanisms, such as income-driven repayment, partially address this mismatch, but they do so primarily after outcomes are realized rather than by pricing uncertainty up front.

Examining education through an actuarial lens highlights familiar challenges: long time horizons, behavioral influences, cohort effects and correlated macroeconomic shocks. These features complicate risk pooling but do not distinguish education risk from other domains where actuarial methods have been applied successfully. Modeling earnings distributions, stress-testing outcomes across economic cycles and evaluating alternative allocation of risk are areas where actuarial analysis can add clarity, even when long-term directions remain unsettled.

As labor markets evolve and educational pathways diversify, understanding how uncertainty is distributed among individuals, institutions and broader systems becomes increasingly important. Bringing educational outcomes into the domain of risk management would not eliminate uncertainty, but it could transform how that uncertainty is shared.

Statements of fact and opinions expressed herein are those of the individual authors and
are not necessarily those of the Society of Actuaries or the respective authors’ employers.

Copyright © 2026 by the Society of Actuaries, Chicago, Illinois.