A study by the Reserve Bank of India has found that interest rates adjust upward more completely during a tightening phase than they do downward during an easing phase.
According to the study, published in the RBI Bulletin, banks passed on 76.4 per cent of the repo rate increase through contract repricing, with lending rates rising by 191 basis points following a 250-basis-point hike between March 2022 and January 2025.
During the easing phase between February 2025 and March 2026, they passed on only 64 per cent of the rate reduction, with lending rates falling by 80 basis points following a 125-basis-point cut.
The study, titled "Transmission to Lending Rates of SCBs: Price versus Composition Effects in WALR", notes that changes in the weighted average lending rate reflect both changes in lending rates and shifts in loan-book composition. It is the work of the RBI researchers Yuvraj Kashyap, Kamal Gupta and Arvind Jha, and was reported by the Times of India.
That distinction does most of the work. Measured on the headline weighted average lending rate, which captures shifts in credit composition, the gap between the two phases does not merely narrow. It reverses: pass-through of 74.4 per cent in the easing phase against 72.4 per cent during tightening.
On the same measure, the 250-basis-point hike resulted in a 181-basis-point increase in the weighted average lending rate, and the 125-basis-point cut resulted in a 93-basis-point decline.
During easing cycles, the researchers note, banks protect net interest margins by raising spreads or shifting towards higher-yielding loans, weakening rate-cut transmission. In March 2026, fresh lending rates rose 12 basis points despite the easing stance.
Loans to micro, small and medium enterprises had the biggest impact on the fresh-loan average, with rates falling 122 basis points during the February 2025 to March 2026 easing cycle, against a 125-basis-point repo cut. Benchmark-linked repricing accounted for 118 basis points of that, reducing the system-wide average by 24 basis points, over 2.6 times any other sector's contribution.
Education loan rates fell the most, by 162 basis points, but contributed virtually nothing to the overall decline in the weighted average because of their small share of credit. Rupee export credit rates fell 161 basis points and contributed five basis points. Large-industry rates fell 67 basis points, contributing nine.
The two measures do not simply differ in magnitude. They disagree about the direction of the asymmetry. On contracted rates, increases travel better than cuts. On the headline average, cuts travel better than increases. The entire difference between them is the set of loans the banks chose to write.
This newspaper accepts both figures, which leads to a conclusion neither of them states. Transmission is not a property of interest rates at all. It is a property of the loan book, and the loan book is a decision.
A bank wishing to report a fully transmitted rate cut need not therefore cut a rate. It need only write more of the loans on which rates have already fallen. Education loans fell 162 basis points and moved the headline average by almost nothing, which means they are available in quantity at no cost to the published figure.
Taken to its proper conclusion, the weighted average lending rate can be driven to any value at all by lending exclusively at that value. A bank that makes a single loan has a weighted average lending rate equal to that loan, a pass-through of 100 per cent, and no composition effect left to explain.
The Department of Unnecessary Research has written to ask which of the two pass-throughs borrowers are repaying, and whether a borrower whose rate rose 12 basis points in March may claim the 93-basis-point decline as an average.
The answer is already in the title of the study. The average is not the rate anybody pays. It is the rate the lending adds up to.