The Right Answer Is a Conditional One
The subject of this series is the transmission from monetary-policy rate hikes to bank credit losses and profitability across economies and institutions. The final question is why one nominal increase can be absorbed in one setting but become a material loss in another, and what a decision-maker should do with that difference. The evidence supports a conditional judgment: the rate hike is common, but the state, contracts, starting balance sheets, and available buffers determine the size, speed, and location of the damage.
One shock, several receiving systems
The most useful synthesis is not a new formula. It is a disciplined way to keep four layers attached to one another.
The first layer is the common input: the policy rate rises. That move changes the price of financing, but it does not directly specify the payment on every loan, the value of every security, or the funding cost of every bank. The second layer is the condition in which the move arrives. A hike after a relatively loose monetary stance, with high inflation or a downturn, is not received by the same borrowers as a hike after a period of tighter conditions and stable incomes. The third layer is exposure: fixed- or floating-rate loans, refinancing dates, portfolio quality, market position, funding, securities, and liquidity. The fourth layer is the set of buffers that were built before or operate during tightening, including borrower-based measures, fiscal support, central-bank balance-sheet conditions, and the ability to avoid forced sales.
These layers interact rather than multiply mechanically. A high debt ratio is more informative when the debt resets quickly and borrowers have little income cushion. A floating-rate loan is more dangerous when the borrower was already weak, and it can be less damaging when the bank’s underwriting was conservative and the borrower has liquid assets. A deposit franchise can give a bank a temporary margin benefit, but that benefit may not survive if loan quality deteriorates or depositors move their funds. A liquid securities portfolio can prevent a fire sale, but it does not cure a weak loan book.
The BIS and IMF findings fit this layered account. At the BIS country level, a 1 percentage-point increase in the policy rate is associated with roughly a 0.1 percentage-point increase in the loan-loss rate over a three-to-five-year horizon; the paper’s rule-of-thumb summary gives about 0.12 percentage points at five years for a short-term-rate increase. At the BIS bank level, the average response is approximately 0.02 percentage points and becomes statistically visible only several years after the shock. Yet banks with poorer pre-existing portfolios experience almost a 0.1 percentage-point increase, with the response appearing in the first year. The difference between those numbers is exactly what a distribution of exposures should produce: a broad average can be modest while a vulnerable tail is material.
The IMF reaches the question through a different route. Its annual panel of 2,147 banks in 31 countries from 1995 to 2023 models loan-loss provisions, net interest margin, non-interest income, and return on assets as shares of average total assets. Higher rates increase provisions, especially for banks with flexible-rate loans and lower market share. They also increase the net interest margin on average, especially for deposit-funded banks. For banks with high NPL ratios and flexible-rate lending, however, the margin benefit is weaker or can turn negative because loans migrate into non-performing status and stop generating interest income. The average profitability result and the vulnerable-bank loss result are therefore compatible.
The studies are complementary because they answer at different levels and with different estimands. The BIS country outcome is an annual net loan-loss-provision ratio over net loans, estimated through local projections and followed over several years. Its bank-level exercise uses a different aggregation and sample.