When a Longer Loan Becomes a Prudential Question
This series examines Colombia’s January 2023 reform that required higher provisions for selected long-maturity consumer loans. The central question is not simply whether the rule made borrowing more expensive. It is whether a prudential measure can make lenders better prepared for future losses without tightening credit uniformly—and, if the average remains calm, where the adjustment appears instead. The answer matters because a stable system-wide number can coexist with a meaningful change in who receives long-term credit and on what terms.
The problem was not simply long loans
Colombia’s reform arrived after a rapid post-pandemic credit expansion. During the recovery of 2021 and 2022, consumer credit grew quickly, especially through personal loans and credit cards. New loans were increasingly issued with maturities longer than five years, even though the existing stock of consumer credit was still dominated by shorter loans. A longer repayment horizon can make a loan more useful to a household by lowering the scheduled payment. It also leaves more time for inflation, unemployment, interest-rate changes, or a borrower’s income shock to disrupt repayment.
That trade-off is easy to miss when attention stays on the monthly installment. A lower scheduled payment can make a loan appear easier to service today, but the lender must carry the exposure across more future conditions. Income may change, collateral may lose value, and the economy may pass through a downturn before the final payment. The relevant prudential question is not whether a long loan is automatically irresponsible. It is whether the institution has recognized the additional uncertainty while the loan is still being originated, rather than waiting for missed payments to reveal it.
A maturity rule also affects the allocation of credit, not only its total volume. If a lender becomes less willing to fund a 10-year loan, a household might receive a smaller principal, offer more collateral, accept a shorter term, or fail to obtain that product from that lender. Another institution might continue to lend on the original terms. The eventual system average could therefore look stable even while the available choices change for a particular borrower or group of lenders.
That vulnerability became more visible in the second half of 2022. Short-term delinquency indicators were rising while inflation remained high, monetary conditions were tightening, economic activity was expected to slow, and unemployment pressures were building. The concern was therefore not that every long loan was unsound. It was that a rapidly expanding segment was extending repayment risk further into an uncertain future, just as households’ capacity to service debt was coming under pressure.
This distinction is important. A long maturity is not itself evidence of a bad borrower or a bad loan. It is a reason to recognize that the lender may be exposed to more states of the world before the debt is repaid. Prudential policy tries to make that exposure visible in the institution’s accounts before losses actually arrive.
The wider Colombian adjustment reinforces why the reform should not be treated as a single-cause explanation for credit conditions. The IMF’s 2024 assessment describes credit growth slowing sharply, consumer-loan nonperforming loans rising, and lending standards tightening while monetary policy remained restrictive. It attributes the broader change to several forces together, including tighter macroeconomic policy, higher consumer-loan provisioning requirements, and lender decisions. The maturity-specific rule is one targeted intervention inside that larger adjustment.
That setting creates a measurement problem as well as a policy problem. When a rule is introduced during a period of high interest rates and slowing demand, a fall in lending after the intervention cannot automatically be assigned to the rule. Nor can a stable average prove that no lender changed course. To understand the reform, one must separate the buffer it directly affects from the many forces that were changing the amount and composition of credit at the same time.
This series keeps those questions distinct. It asks what the maturity rule did to institutions’ preparation for loss, what it did to the terms of newly originated loans, and whether the answer differed by lender. That distinction is more useful than a simple verdict that the policy either “worked” or “failed.”
What Colombia changed
In November 2022, Colombia’s Financial Superintendence issued External Circular 026, effective January 1, 2023. For covered new consumer loans, the rule added a factor for the risk associated with longer maturity and higher leverage. Credit cards, revolving credit, and payroll loans to pensioners were excluded. The additional factor was 1.1 for loans with maturities above 72 months and up to 108 months, and 1.4 for loans beyond 108 months. In practical terms, the reform raised the expected-loss calculation by 10 percent in the first band and 40 percent in the longest band relative to the existing calculation.
The thresholds are not a new interest-rate schedule for borrowers. They are points at which the institution’s expected-loss calculation changes. That difference matters: the rule first changes the lender’s internal cost of carrying the loan, and only then might the lender decide whether to alter the contract. A bank that absorbs the cost passes no immediate charge through to the customer; a smaller or more exposed institution may protect its position by changing the amount or collateral it offers. The policy is therefore aimed at a risk calculation, while its distributional consequences emerge through lender choices.
A provision is not a fee that the borrower pays to the regulator. It is an accounting expense that sets aside capacity for a possible future loss; the paper notes that provisions do not involve a cash outflow, even though they affect reported expenses and profitability. The rule could still influence borrowers indirectly. A lender facing a higher cost for carrying a long loan might raise its interest rate, reduce the principal it is willing to disburse, require more collateral, prefer a shorter maturity, or stop serving some applications. Alternatively, a lender with sufficient capital, margins, or scale might absorb the cost and leave the contract largely unchanged.
That gives the reform two distinct policy objectives. The first is resilience: if long-term borrowers default, the institution should have a more adequate buffer against the loss. The second is moderation: by making riskier long-term lending more expensive to carry, the rule might discourage an excessive expansion of that segment. These objectives can diverge. A lender may build a larger buffer while continuing to offer the same terms, especially if the affected loans are only a narrow part of its portfolio. Conversely, a lender may protect its balance sheet by tightening contracts without materially improving the system’s overall loss coverage.
The design was deliberately narrow. It did not reprice every existing consumer loan or impose the same increase across the entire portfolio. It targeted newly originated loans whose repayment horizon crossed one of two maturity thresholds. That narrowness makes the rule easier to study, but it also limits what a result can mean: evidence about these new loans is not automatically evidence about credit cards, revolving balances, existing borrowers, or the whole household-credit market.
The narrow scope also clarifies what a successful intervention would look like. If the purpose is to prepare institutions for losses on a particularly exposed maturity segment, a higher coverage ratio is the relevant early signal. If the purpose is to cool credit growth, one would instead look for fewer or smaller long-term originations, higher prices, more collateral, or a shift toward shorter maturities. Those signals need not move together. A rule can succeed on the first test and have little effect on the second.
It is equally important not to confuse the absence of a direct charge with the absence of a burden. Provisions affect an institution’s reported expenses and may put pressure on profitability. The lender then decides how much of that pressure to bear, how much to pass through, and which loans to prioritize. The reform’s economic meaning lies in that chain—from risk recognition, to balance-sheet capacity, to credit allocation—not in the regulatory percentage alone.
The question hidden by the average
The natural first question is whether the reform tightened credit overall. The study’s answer is more subtle. Using granular supervisory data and comparisons designed to account for the fact that long loans already differ from short loans, it finds that the rule increased provision coverage. Under the strictest specification, coverage rose by about 0.166 percentage points at the 72-month threshold and about 0.86 percentage points at the 108-month threshold. That is evidence that the rule achieved its direct buffer-building purpose.
This first result is about preparedness, not realized performance. A higher coverage ratio means that the institution has recognized more potential loss in its provisioning position. It does not by itself show that defaults later fell, that households became safer, or that the lender became more profitable. Those outcomes would require a longer view of loan performance and institutional accounts than this opening result provides.
But the same study finds no statistically distinguishable average change in interest rates, loan amounts, or loan-to-value ratios at either threshold. Higher provisions were not broadly passed through as higher prices, smaller loans, or more collateral in the aggregate estimates. That does not mean that every lender left every contract untouched. It means that the average of the observed new-loan market does not reveal a general tightening along those measured margins.
“Not statistically distinguishable from zero” is a narrower conclusion than “nothing changed.” The estimates may miss small effects, changes in approval decisions, shifts into exempt products, or responses that differ across institutions. The result says that the study cannot identify a robust average movement in the selected contract measures under its comparison strategy and observation window. That careful wording is essential because the policy question is about allocation, while the most visible statistics are averages.
The difference between those statements is the key to the case. If larger institutions can absorb the added provisioning cost while smaller institutions retrench, their opposing responses can cancel in a system-wide average. Borrowers connected to different lenders may then experience different versions of the same rule. A neutral average can describe the market as a whole while concealing a sharper constraint at the edge of the market—particularly for loans beyond nine years, where the additional provisioning factor is largest.
There is another reason to read the average carefully: the study observes loans that were originated, not every application that was considered. A smaller lender could respond by approving fewer long-term loans, but that decision would not appear as a lower loan amount among the loans it still made. A household could also move to another lender or choose a different product. Those possibilities do not invalidate the observed results; they define their boundary. The evidence is strongest about the terms of originated loans and weaker about the full set of opportunities that borrowers faced.
The average result also should not be read as a promise that competition neutralized risk for everyone. If larger lenders can keep offering long loans while smaller lenders pull back, the system may remain supplied in aggregate while becoming more dependent on institutions with greater scale. That can preserve access for some borrowers and narrow it for others. The distribution of the adjustment is therefore part of financial stability’s real-world meaning, even when it is not visible in the headline average.
The rest of the series follows that distinction. First, it explains how expected loss turns maturity into a balance-sheet cost. It then shows how the study separates the reform’s effect from the pre-existing differences between long and short loans. Finally, it asks why stronger buffers, stable average terms, and tighter conditions at smaller lenders can all be true at once. The practical judgment is consequently conditional: Colombia’s rule appears well suited to strengthening preparedness for long-maturity losses, but whether it moderated credit—and who bore the adjustment—requires looking beyond the average.