Finance formed from future income, once evidence can price it.
Fromstudent loan→Tohuman-capital credit
The concept imagines human-capital credits: capital that a person’s agent can spend across providers, repaid from future income into a fund that finances the next learner. Income-contingent finance already shows that training can be financed before a person has income to show for it. The evidence doesn’t yet show three things: that capability evidence lowers the cost of capital, that a revolving fund can stay solvent without subsidy, and that capital can avoid steering people toward whatever repays fastest. So finance comes after the evidence layer, as an expansion, and it starts with patient public or philanthropic capital.
The timing problem
The cost of becoming capable comes first, and the return comes later. Before a person has the capability that earns, they may need to pay for courses, simulations, assessments, equipment and time. The concept’s answer is human-capital credits [CN]. A state or another funder gives a person a pool of credits, and their agent spends them across whatever pathway it builds. Repayment starts only once the person earns. It can be partial, staged, or held in escrow until earnings pass a threshold. Repaid credits flow back to the fund.
- Capital
- Capability formation
- Paid work
- Income
- Repayment
- Recycled capital
- Next person’s formation
The financing question changes from should we fund this course of study? to should we fund this person’s path toward a capability they can demonstrate?
What already exists
Income-contingent finance is not new. In an income share agreement, a learner receives funding and pledges a share of future income above a threshold for a fixed term. Economically it sits closer to equity than to debt: the funder shares the downside, and the learner’s payments rise and fall with income [1]. Pricing each contract to the learner’s risk sends useful market signals. It can also steer people toward particular fields. Questions about regulation, enforcement, bankruptcy, taxation and securitisation remain open, as does the debate about whether such contracts come too close to indenture [1].
A 2023 study found the market split into two sectors: degree-granting institutions, and workforce training programs such as bootcamps. Contract terms interact in ways that make agreements hard to compare. About half of the institutions offering them didn’t provide the minimum information a learner needs to evaluate one. Historically underserved learners appeared more exposed to these agreements, though there was no evidence that they received systematically different terms. Public awareness was low, yet about 40% of U.S. adults said they would consider one [2].
Career impact bonds went further. They pair short-term training with catalytic capital, income thresholds and caps on payments. Early findings from one study show why this is hard. Some learners had difficulty understanding the terms, and more than half of those required to repay were not doing so after they graduated [3]. Programs that generate evidence like this give the finance layer the lessons it needs.
Evidence as underwriting information
The concept’s financial intuition is that evidence can make a person progressively more financeable.
- No evidence
- Learning evidence
- Assessment evidence
- Work evidence
- Income evidence
Each step could reduce uncertainty about whether investing in someone’s capability will produce future income. The review found no evidence yet that capability evidence reduces underwriting risk or the cost of capital. Evidence could even backfire. It may be noisy or stale, it may stand in for existing advantage, and verifying it adds cost.
Capability evidence may improve underwriting. Whether it lowers the cost of capital is an empirical question.
Go deeper
Whose choice does capital serve?
A private funder has good reason to prefer paths into high-paying work over paths into work that is socially valuable but lower-paid [1]. The review rates the claim that private-return finance will fund socially valuable, lower-paid capabilities as contradicted by the evidence. Finance could quietly bring back the steering this architecture set out to remove, with the pathway following the capital rather than the person’s intent.
The design has to keep financing separate from choice. That means standardised, comparable terms [2], underwriting that doesn’t penalise a chosen field, and public or cross-subsidised capital for pathways whose value is social rather than private.
Finance should widen the paths a person can choose, not narrow them to the ones that repay fastest.
Repayment meets real life
Income thresholds and caps protect people against low earnings. A fund still has to handle unemployment, capabilities that become obsolete, career changes, migration and hardship. For lower-income workers there is a further interaction. As pay rises, public benefits phase out. In one modelled healthcare career ladder, a worker could take about fifteen years to come out ahead of minimum-wage work once lost benefits were counted [4]. If repayment starts at the same point, it adds another deduction on top of that phase-out. Repayment thresholds should be set with benefit rules in view, and the person’s agent should show the combined picture (Shift 7).
Revolving is harder than repaying
A revolving fund must do more than collect some repayments. After losses, servicing, hardship, fraud and administration, it still has to finance the next cohort. Public capital can absorb social returns, but then the fund’s sustainability depends on fiscal support rather than repayment alone. That makes it useful to separate commercial capital, which needs a financial return, from human-capital infrastructure capital, which can accept longer horizons, social returns and first losses. In the early stages, public and philanthropic capital are likely to be needed for experimentation, equity and first-loss protection.
Why finance comes later
The first market doesn’t need finance to test whether evidence creates value (Shift 1). If the network can’t create value with conventional capital and public support, adding finance increases risk rather than solving the problem. And if the finance layer fails, the network doesn’t fail with it. It could still run as an evidence, apprenticeship or matching system funded by employers and public programs.
Not our invention
Income-contingent loans, income share agreements, career impact bonds, outcome-based funding, national training credits such as Singapore’s, and individual learning accounts in Europe all exist [2, 3, 5]. The addition here would be pricing finance on verified evidence that builds up across providers and employers, with repayment drawn from earnings wherever the person works.
What could go wrong
- Steering.Capital nudges people toward higher-paying fields.
- Opaque terms.People may not understand or be able to compare what they have signed [2, 3].
- Adverse selection.The people most likely to repay opt out.
- Weak collection.Repayments may not arrive, as early evidence warns [3].
- Stacked cliffs.Repayment and the phase-out of benefits combine into punishing effective marginal rates.
- Permanent subsidy.The fund needs so much subsidy that it becomes a grant program by another name.
How we’d test it
| The test | What we would measure | What would end the idea |
|---|---|---|
| Progressive financeability (Experiment 5) | Underwriting with no evidence, then with learning, assessment, work and income evidence: repayment, default, servicing cost, hardship, career choice, fairness across groups, capital per successful worker | Evidence adds no out-of-sample predictive power, or the fund needs subsidy for ordinary operations. The review’s planning threshold is at least 15% lower losses or subsidy, without a heavier burden on workers or exclusion |
| Steering | Pathway choices of financed and unfinanced people with similar intent | A systematic shift toward higher-paid paths that stated preferences don’t explain |
| Public economics (part of Experiment 7) | Lifetime public cost per sustained productive worker, against grants, vouchers and ordinary student finance | Higher cost per outcome once defaults and administration are counted |
How strong is the evidence?
- Established
- Income-contingent instruments can finance training before a person has income evidence [1, 2]
- Evidence-supported
- An early study of one career impact bond found some learners struggling with the terms and more than half of those required to repay not doing so [3]; risk-based private underwriting can steer the choice of field [1]
- Hypothesis
- Verified capability evidence improves underwriting enough to matter
- Unknown
- Whether a revolving fund stays solvent, whether the cost of capital falls, and whether socially valuable lower-paid capability can be financed without permanent subsidy
Still to confirm
Rules for income share agreements differ by jurisdiction. Confirm the regulatory position wherever a pilot runs.
Sources
- [1]Congressional Research Service, In Focus IF11269 on income share agreements (2019)Checked
- [2]RAND Corporation, study of the U.S. market for income share agreements (RR-A2649-1, 2023)Checked (summary page)
- [3]MDRC, Income Share Agreements to Finance Short-Term Career Training: Preliminary Findings from the Career Impact Bond Study (February 2024)Checked (preliminary findings)
- [4]National Governors Association, resource page on learning and employment records, including its account of the Federal Reserve Bank of Atlanta’s CLIFF toolChecked; the Atlanta Fed modelling it reports was not re-checked
- [5]Cedefop, Microcredentials for labour market education and training: the added value for end users (2023)Checked (executive summary and chapters 1–3)
- [CN]Open Commerce for Human Capital: concept document consolidated from the discussion (internal)Provided
Spotted a source we missed or a rule we misread? Tell us what we got wrong.