The Banks That Feel It First

This chapter focuses on why banks in the same economy can experience very different consequences from one policy-rate increase. The question is no longer only how the macro shock reaches borrowers; it is which loan book, funding model, securities portfolio, and liquidity position receives that shock. A bank-level view makes the rate-hike/credit-loss relationship concrete: it shows where losses concentrate, where a temporary margin gain comes from, and which buffers can change the outcome.

A loan book carries its past into the next rate cycle

A bank’s exposure begins with the contracts it already holds. If a large share of its loans carries a floating rate, the policy move can be passed to borrowers relatively quickly. The interest bill rises while income may not, and the probability of missed payments increases. If the bank’s loans are fixed for a period, the direct payment effect is delayed. That delay can be valuable, but it is not the same as eliminating risk: borrowers may still refinance later, lose employment, face weaker collateral, or see their businesses weaken as demand slows.

The BIS country analysis finds that economies with a high prevalence of floating-rate loans experience larger and faster increases in credit losses. The comparison uses a threshold of more than half of household loans at floating rates and covers only 46 countries for which the relevant information is available. The additional loss response is about 0.11 percentage points in the first year after a 1 percentage-point policy-rate increase. The number should be read as a comparison between estimated states, not as a tariff applied to every floating-rate loan.

The IMF reaches the same directional conclusion from bank-level data, but it has to construct an indirect measure. The paper combines information about mortgage-rate flexibility in the country where a bank operates with the bank’s mortgage share and its business description. Banks that mainly lend to non-financial corporations or consumers can be classified as more flexible-rate exposed because those loans are often shorter-term or linked to market rates. Direct, complete information on each bank’s fixed- and variable-rate stock is not available in the source data, so this is a proxy rather than a precise contract inventory.

That limitation matters. A country known for fixed-rate mortgages can still contain banks with large corporate or consumer books that reprice quickly. Conversely, a loan labelled “floating” may have a cap, a reset interval, or a borrower able to hedge the payment. The papers establish a robust direction—faster repricing tends to increase the sensitivity of provisions—but not a universal conversion from loan type to default probability.

The second part of the loan-book diagnosis is quality before the hike. A bank with a high share of impaired loans has less room for another deterioration. Some borrowers may already be close to default; collateral recoveries may be uncertain; and management may have less capacity to absorb an additional provision charge without reducing capital or lending. When the rate rises, a weak portfolio can turn a modest increase in debt service into a concentrated wave of new impairments.

The BIS bank-level evidence makes this starting condition visible. Banks with an above-country-average share of impaired loans before tightening experience an increase in loan-loss rates of almost 0.1 percentage points after a 1 percentage-point policy-rate increase. The response is significant in the first year.