What inflation does the RBI actually fight?

In December 2025, the combined consumer price index inflation was 1.3%. That is headline CPI: the number that gets front-page space. It means the overall price level rose 1.3% in a year, on average, for the basket of goods and services an Indian household buys. But inside that single number there are two very different stories. Vegetables fell 18.5%. Pulses dropped 15.1%. Potato prices crashed 35%. Alongside that, doctor’s fees were up, education costs rose 3.4%, and health services climbed 3.4%. Strip out food and fuel from the headline, and what is left is core inflation. For December 2025, core inflation was 4.6%. That gap, between a headline near 1% and a core at 4.6%, is the fault line in Indian monetary policy.

Core inflation is not a separate index published by the Ministry of Statistics. It is a derived measure: take the all-items CPI and remove the food and beverages group, plus the fuel and light group. What remains is a basket of everything else, housing, health, education, transport, clothing, recreation. In the new 2024-base CPI, food and beverages has a weight of 36.75% (down from 45.86% in the 2012 series), while fuel and light components are tucked inside other groups. The resulting core gauge is watched by every bond trader and every member of the RBI’s Monetary Policy Committee.

Food and fuel prices are volatile and largely supply-driven. A late monsoon, a heatwave in Maharashtra, or a disruption to global crude oil can send these prices lurching from one month to the next. In December 2025, the Food and Beverages index inflation was -1.9%. That was driven by a record vegetable price plunge of -18.5%, itself pulled down by onion at -48.1% and potato at -35%. Those numbers are real and they matter enormously to households. But as a signal of where inflation is heading over the next two to three years, they are almost pure noise.

Core inflation strips that noise out. The idea, standard in central banking for decades, is that the prices of services like health, education and housing are “sticky”. Once a private hospital raises a consultation fee, it rarely cuts it back. A school that hikes tuition by 10% doesn’t reverse it next year. These price moves reflect demand-driven pressures in the economy, wage expectations, and firms’ pricing power. That is exactly the kind of inflation that monetary policy, through interest rates, can influence. As the RBI noted when it formally adopted inflation targeting in 2016, “food and fuel shocks are largely unresponsive to actions on the demand side”. So economists separate headline into two numbers: a noisy, weather-driven outer layer and a persistent, demand-driven inner layer.

Both are right, but they answer different questions. Headline CPI is, by definition, a measure of the cost of living. If your monthly food bill is dropping because onion is at ₹10 a kilo in Azadpur mandi, your wallet feels lighter. That is real to you. Headline captures that. Core CPI answers a different question: not what did things cost last month, but what underlying inflation trend is building? A 1.3% headline in December 2025 was essentially all due to collapsing food prices. Strip out food and beverages, and the rest of the basket was running at 4.3% inflation. Strip out food plus fuel, and core was 4.6%. The gap between 1.3% and 4.6% is the widest it has been in a while. It means the broad economy, outside the mandi, is still experiencing persistent price rises.

To see the divergence at a glance:

So if you are a policymaker trying to judge whether inflation is becoming entrenched in the system, you will trust the core reading more than the headline.

But core can also mislead. A prolonged spike in petroleum product prices eventually feeds into transport costs and into all goods that move by truck. That would show up in core with a lag. And if a food price shock persists for months, households may demand higher wages, turning a supply shock into a core inflation problem. So the two numbers need to be read together. As a rule of thumb, if headline is below core for a long stretch, as it is now, it suggests commodity prices or food are unusually soft. If headline races ahead of core, it is a supply-side spike that may pass.

By law, the RBI targets headline CPI. Since 2016, the government has set a target of 4%, with a tolerance band of 2% to 6%. If headline CPI stays outside that band for three consecutive quarters, the RBI must explain in writing to the government what went wrong and what it intends to do. The instrument is the repo rate, the rate at which the RBI lends to banks. When the Monetary Policy Committee raises the repo rate, borrowing becomes costlier for businesses and consumers, demand cools, and inflation ideally comes down. In late May 2026, the repo rate stood at 5.25%.

But the law targets headline, not core. And headline is dominated by the very food and fuel that the repo rate is weakest at controlling. This creates the central tension. RBI officials and MPC members watch core closely because it tells them whether domestic demand is overheating and embedding inflation into the economy. The RBI’s own analytical chapters in its monetary policy reports dissect core inflation trends, not just headline. In practice, the MPC uses core as a critical input to decide when to act on headline, even though the target is cast in headline terms. This dual-track approach is well understood in bond markets: a high headline number driven by a vegetable spike may not provoke a rate hike, but a steady creep of core above 5% almost certainly will.

As of early 2026, headline inflation on the new 2024-base CPI was 2.75% in January 2026, below the 4% target. Meanwhile, core inflation was still around 4.6% on the old series. The MPC faces a classic dilemma: cut rates to support growth when the headline looks anaemic, or hold steady because underlying price pressures refuse to die down. Those who argue for cutting point to a tepid private investment cycle and the risk of stalling an already moderate GDP expansion. They note that headline below 4% for an extended period will itself dampen inflation expectations. Those who argue for holding say that cutting rates when services inflation is above 6% and health and education costs are rising steadily is to risk letting the genie out of the bottle. Once consumers and firms begin to expect 5-6% price rises as normal, the RBI loses its most powerful weapon, credibility. The debate is live, and the minutes of the MPC are thick with dissents and qualifications. There is no clean answer, only a trade-off.

The repo rate is the lever, but it is a blunt one. Raise it, and banks are forced to pay more for funds from the RBI. They pass this on by raising interest rates on loans, for businesses, for home buyers, for a chai stall owner wanting a working-capital loan. Costlier credit means consumers postpone buying a fridge or a car, companies delay expansion plans, and aggregate demand in the economy slows. With less demand chasing the same goods and services, businesses find it harder to push through price increases. Inflation moderates. The process works with a lag of three to four quarters, often more. And it works much better on demand-driven inflation, think rising wages pushing up restaurant prices or housing rents, than on supply-driven food and fuel spikes. A 25-basis-point rate hike won’t make the monsoon better or bring down crude-oil prices. So when headline CPI surges because onion prices quadruple, the repo rate is an almost useless weapon. Raise it sharply to kill headline, and you risk choking off credit and employment for no gain on the vegetable front. That is why the split between headline and core is not an academic curiosity; it is the central dilemma of inflation management in a country where food still makes up over a third of the consumption basket.

The dilemma is distilled in the December 2025 numbers: headline at 1.3%, core at 4.6%. If the MPC were to judge that inflation is benign because headline looks dead, it could cut rates aggressively. But it would be cutting while services inflation runs at 6.2%, education costs climb, and health expenses keep rising. That could fuel a credit-driven consumption boom in the non-food part of the economy, eventually pushing core even higher. By the time that shows up in headline, say because food prices return to normal, inflation is already entrenched and the RBI is behind the curve. On the other hand, keeping rates high when headline is far below target, and when food prices are actually falling, can make the RBI look out of touch with the common person, who is seeing lower vegetable bills and wondering why EMIs aren't falling. This tension is not unique to India, but it is sharper here because food’s weight in the CPI, though falling from 45.86% in 2012 to 36.75% in the 2024 basket, remains among the highest for any major economy. The new 2024 series, which incorporates the Household Consumption Expenditure Survey of 2023-24, shows that Indians are spending less of their budgets on food and more on services. That shift, a textbook case of Engel’s law, may gradually reduce the dominance of food in the headline and make the RBI’s target and its instrument less mismatched. But for now, the conflict remains.

Headline CPI is your cost of living. It tells you what is happening to the prices of the things you actually buy every day, atta, sabzi, milk, petrol, LPG, medicine. It is the number that will shape your household budget and, indirectly, your vote. Core inflation is the trend the RBI steers by. It tells you whether the deep, slow-moving parts of the economy are heating up. Watching both tells you whether a price rise is a blip or a tide. When headline is low but core is sticky, as it is now, you know that the relief at the vegetable cart may not last and that pressure is building in the doctor’s clinic and the school fee counter. The RBI sees the same split you see. Whether it decides that the vegetable crash is the real story, or whether the sticky core demands a hard brake on the repo lever, is the most consequential judgment a six-member committee makes every two months.