The Rate Cap Is a Constraint, Not a Guarantee
For the state consumer-loan caps at the center of this discussion, the important question is not simply how high the legal rate may be. It is how the rule changes the set of borrowers a lender can serve profitably. Understanding that institutional setting and the sequence of lender responses helps distinguish a loan that becomes cheaper from a loan that disappears, changes form, or moves to someone else.
What the three state caps actually covered
The New York Fed study examines three states that enacted consumer usury limits during the period from 2016 to 2022: South Dakota, Illinois, and North Dakota. Each cap was set at 36 percent, inclusive of fees. The affected products included payday loans, installment loans, and auto-title loans—forms of consumer credit that can carry high annualized prices, particularly when a relatively small balance is combined with meaningful underwriting, servicing, or collection costs.
The timing was not identical. South Dakota’s Initiative Measure 21 became effective on November 16, 2016. Illinois’s SB 1792, the Predatory Loan Prevention Act, took effect on March 23, 2021. North Dakota’s SB 2103 became effective on August 1, 2021. The staggered dates matter because the study compares changes around each state’s intervention rather than treating all three laws as one event.
The laws also did not cover every institution that might lend to a household. Banks and credit unions were expressly exempt. That exemption creates a possible route for substitution: if a payday or installment lender reduces supply, a mainstream institution might fill part of the gap. But “could fill the gap” is a market possibility, not a result guaranteed by the statute. A bank or credit union may have different underwriting systems, minimum loan sizes, customer relationships, risk tolerances, or cost structures. The fact that one lender is legally allowed to charge a different rate does not mean that institution will offer the same borrower a comparable product.
The report’s institutional details also help interpret evidence about lender presence. State regulatory information summarized in the paper shows that the number of affected licenses fell by 19 percent in South Dakota and by more than 50 percent in Illinois in the year after the caps; the paper reports no comparable figure for North Dakota. Those numbers suggest contraction among affected licensees, but they are not a complete account of credit flows. A license can disappear without showing exactly how many borrowers lost access, and a lender can remain present while changing eligibility, product design, or target customers.
This is why a rate cap should be understood as a constraint on a market rather than as a universal discount. A price ceiling binds only when the price that would otherwise be needed is above the permitted maximum. For a relatively safe borrower, the lender may already be able to make a viable loan below 36 percent. For a riskier borrower, expected losses may make the same permitted rate inadequate. The rule is identical, but its force depends on the borrower’s risk and the economics of the product.
Risk is not the only consideration. Lenders must also decide whether their available capital, staff, data, and operating systems can support a particular customer segment under the new rule. The New York Fed authors emphasize funding and technological constraints as possible reasons that lenders may redirect capacity. This matters because a lender with limited capacity does not face only two choices—continue exactly as before or close. The lender may stop making loans to one segment and use its resources in another.
The distinction between a covered lender and an exempt lender therefore matters twice. First, it determines which loans are directly constrained by the cap. Second, it determines whether a replacement market is even plausible. If an exempt institution serves a different clientele or requires a different underwriting process, the legal exemption may exist without producing a practical substitute for the borrower who was previously using high-cost credit.
The law sets the maximum price, but the lender determines which response is feasible for a particular risk group.
The studied laws are therefore narrower than a general statement such as “the government capped interest rates.” They targeted selected products, in selected states, with important institutional exemptions. That boundary becomes even more important when comparing these results with a federal proposal aimed at credit cards: the mechanism may be related, but the product and the institutions covered would not be identical.
The inclusion of fees is an especially consequential detail. If a lender can freely replace an interest charge with a fee, a nominal rate ceiling may leave the effective price largely untouched. The state limits in this study were described as inclusive of all fees, so that particular route around the stated percentage was constrained. Even then, inclusion does not determine the outcome by itself. A lender can still adjust eligibility, loan size, term, renewal policy, or the type of customer it targets. The rule closes one pricing path; it does not prescribe the entire lending model.
The reported license changes illustrate the difference between a market presence and a borrower outcome. A decline in licenses is compatible with lenders exiting, consolidating, or changing their legal or operating structure. It does not reveal whether the remaining lenders expanded, whether another institution supplied replacement credit, or whether the affected borrowers stopped borrowing. For that, the study needs household-level measures of balances and accounts, and even those measures cannot identify every product or informal source of funds. Legal coverage, lender presence, and household access are related layers of the market, not interchangeable statistics.
The same distinction helps organize the possible responses. Broad exit should be visible in fewer affected lenders, though a license measure is imperfect. Rationing should appear as less credit for the borrowers whose risk makes the ceiling binding. Substitution should appear as a compensating expansion through another lender or product for those borrowers. Restructuring could preserve an account while changing its terms or eligibility. Reallocation would combine a contraction for the most exposed borrowers with expansion for a safer group. These signatures can overlap, but they ask different empirical questions. A credible interpretation should not treat any one of them as the automatic meaning of a falling rate.
What a binding ceiling lets a lender do
Consider two applications for loans of similar size. One comes from a borrower whose repayment history makes losses relatively unlikely. The other comes from a borrower whose history signals a much greater chance of nonpayment. Before a cap, the second loan may be priced higher to compensate for that risk. If the law prevents the lender from charging the necessary rate, the lender faces a reduced permitted return without a corresponding reduction in the risk being taken.
At that point, rationing is the most direct response. The lender can decline the application, offer less credit, or stop renewing an existing account. In the credit market, rationing does not necessarily mean a formal sign saying “no.” It can appear as a narrower eligibility rule, a smaller principal, fewer accounts, or the absence of a new offer when an old one expires. The common feature is that the riskiest borrower receives less credit at the permitted price.
Substitution is a different response. It occurs when another lender provides a comparable loan after the original lender withdraws or reduces supply. The distinction is important because a fall in one lender’s accounts does not by itself prove that total household access fell. To establish substitution, one would want evidence that another institution or product expanded for the same affected borrowers. The study’s reported high-risk outcomes do not show a compensating replacement through mainstream lenders, but the credit-bureau measures do not trace every possible source of funds.
Restructuring is different again. A lender may alter the loan’s term, underwriting process, fee design, or product structure in an effort to operate within the cap. Some changes may be legal adaptations; others may test the boundary of the rule. The report identifies product restructuring and partnerships with exempt lenders as possible ways a lender might respond. It also discusses evasion, including attempts to change the structure of a product or route lending through an institution not covered by the limit. These possibilities should not be collapsed into ordinary rationing: they preserve some form of lending, but potentially change who bears the cost and how the loan works.
Market power creates another distinct possibility. If an affected lender had been charging more than was necessary to cover risk and cost, a cap might reduce the price while leaving the quantity of credit broadly intact. In that case, the lender absorbs a smaller margin rather than withdrawing from the borrower. This is one reason standard price theory is a prediction, not a complete description of every market. The outcome depends on how close the old price was to the lender’s constraint and on how much discretion the lender had before regulation.
The reallocation path begins when the lender’s capacity is limited and the cap makes some existing loans unattractive. A loan to the riskiest borrower may no longer be viable at 36 percent, while a loan to a somewhat safer borrower remains viable at that rate. Staff time and underwriting capacity released from the first group can then be directed to the second. The safer borrower does not receive credit because the law commands the lender to serve them. They receive it because the new constraint changes the relative attractiveness of different applications.
This is a sorting process, not necessarily a literal transfer of one loan or one dollar. The lender may close an account held by a high-risk borrower and later originate a different account for a safer borrower. The timing, amount, and terms can all differ. What connects the outcomes is the allocation of limited lending capacity, not an assumption that the same funds can be traced from one household to another. That distinction also explains why a lender can report continued activity after a cap while the highest-risk borrowers experience a meaningful contraction.
The cap changes the ranking of applications at the margin. Before the rule, a risky loan might have generated enough revenue to justify the expected loss and the work required to administer it. After the rule, a safer loan may offer the lender a better risk-adjusted use of the same staff or capital even if both loans carry the same legal maximum. A lender that wants to preserve volume can therefore become more selective. Selectivity is not the same as a neutral improvement in underwriting: it can move the threshold at which a borrower is served and leave those just below it with fewer options.
The process can also be uneven within a nominal risk group. Credit scores summarize repayment risk, but they do not turn borrowers into identical units. Some people near the boundary may remain eligible, while others with similar scores may face different product economics, local competition, or lender policies. The study’s use of risk deciles is valuable because it reveals broad patterns without implying that every household in a decile had the same experience. It is a way to detect distributional movement, not a claim that risk score alone explains each lending decision.
That process can produce a counterintuitive distributional result. The borrower most exposed to high-cost lending may lose an account, while a borrower one or two steps higher in the credit-score distribution gains an account. From the lender’s perspective, both outcomes can be part of one adjustment: withdraw where the ceiling binds and expand where it does not. From the borrower’s perspective, however, the policy is experienced differently. One household sees an unaffordable product removed; another sees a new opportunity; neither experience can be inferred from the legal ceiling alone.
The New York Fed report does not directly measure every link in this chain. It observes household balances, open accounts, and delinquency, then uses comparisons across states, time, and risk groups to estimate how those outcomes change. It does not identify each lender’s capital budget, show every rejected application, or follow a dollar from one borrower to another. The reallocation interpretation is supported by the pattern of pooled results, but the specific operational decision—whether a lender hired staff, bought alternative data, changed its product, or partnered with an exempt institution—remains an explanation rather than a directly observed fact in the supplied evidence.
That limitation does not make the mechanism irrelevant. It tells us how to use it. The mechanism is a map of possible transmission paths, while the empirical estimates tell us which outcomes actually moved in the observed sample. If the later data show a decline for the riskiest borrowers and an increase for moderately safer ones, the capacity-constrained model helps explain how both can happen. It should not be used to claim more detail about firm behavior than the evidence contains.
The most defensible conclusion from the institutional setting is therefore conditional. A 36 percent cap can bind for some high-risk consumer loans while leaving safer loans feasible. Once it binds, lenders can ration, find substitutes, restructure, evade, use exempt partners, or redirect capacity; these responses have different implications for access and borrower welfare. The next analytical task is to see how the study distinguishes those distributional outcomes from ordinary differences across households and states.