Inflation, and how it is measured

Reading for India · about 10 min

The answer

Inflation is the rate at which the general level of prices rises over time. A government measures it by pricing a fixed list of goods and services every month, weighting each item by how much an average household spends on it, and comparing this month's total to the same month a year ago.

Why this costs you money

You have money in a savings account or a fixed deposit. The bank tells you the rate. You think that rate is what you earn.

It is not. What you earn is the rate minus inflation. If the deposit pays 6.5% and prices rise 5%, you earned 1.5%. That is the only number that buys anything.

Here is the more expensive version of the same mistake, and almost nobody catches it.

The official inflation number is a weighted average of what an average household in the whole country spends. You are not that household. In India's current CPI basket, food and beverages carry a weight of about 36.75% . If you are a salaried person in a city, food is probably 12% to 18% of your spending. Rent, school fees, transport and health are far more.

So when the headline says inflation is 3%, and food is what is cheap this year, your own cost of living may have risen 6%. You then plan your savings, your salary expectation and your retirement number using a figure built from somebody else's shopping.

The loss is slow and it is large. Planning a 30-year retirement on a 4% assumption when your real number is 7% is not a small error. It is the difference between a plan that works and a plan that runs out.

How it works

Every inflation number in the world is built in the same 4 steps.

1. Find out what people buy. A statistics agency runs a large household spending survey. India's current CPI series uses the Household Consumption Expenditure Survey of 2023-24. This produces the list of items and the share of the budget each one takes.

2. Fix the weights. Each item gets a weight equal to its share of household spending. The weights are the whole game. An item with a 26% weight moves the index 26 times as much as an item with a 1% weight, for the same price change.

3. Collect prices every month. Staff record the price of the same item, in the same shop, in the same quantity.

4. Compare to a year ago. The weighted total is the index. The inflation rate is the percentage change in that index from the same month last year.

The words you will see

Headline inflation is the whole basket.

Core inflation is the basket with food and energy removed. Those 2 groups move for reasons interest rates cannot touch — a bad monsoon, a war, a pipeline. Core is the slower signal underneath.

CPI is the Consumer Price Index. It prices what households buy, at retail.

WPI is the Wholesale Price Index. It prices goods before they reach you. India's WPI contains no services at all, which is why it can say something completely different from CPI.

PCE is the Personal Consumption Expenditures price index, used in the United States. Similar idea, different weights.

The weights, side by side

GroupIndia CPI (2024 base)US CPI-U (Dec 2025)
Food and beverages36.75%13.70%
Housing and utilities17.67%44.47%
— of which shelter alone35.63%
Transport8.80%16.32%
Health6.10%8.42%
Clothing and footwear6.38%2.37%
Energy / fuel and light5.49%6.38%

Those 2 columns describe 2 different lives. That is the point of this article.

What it tells you, and what it does not

It does tell you the average rate at which prices are rising for the country, and therefore what the central bank is reacting to.

It does not tell you your own cost of living. It is an average, and averages hide the spread.

It does not include asset prices. House prices, share prices and gold as an investment are not in CPI. The US index does include owners' equivalent rent, an estimate of what a homeowner would pay to rent their own house, at a weight of about 26.2%. That is a rental estimate, not a house price — and because US rents are surveyed on a rotating schedule, part of what you read is history.

A base change breaks the comparison. India moved from a 2012 base to a 2024 base and the food weight fell from about 45.86% to about 36.75%. Nobody's grocery bill changed that day. The measuring instrument changed.

The decision rule

Use the headline number for one purpose only: guessing what the central bank will do. Use your own number for everything about your own money.

If your spending is concentrated in a group carrying a small official weight, the headline will understate your inflation whenever that group rises faster than average — and overstate it whenever that group falls.

The second half matters. Your personal rate is not always higher. A household that spends heavily on food, in a year when food prices fall, is doing better than the headline suggests. The rule is not "the government understates inflation". The rule is "the weights are not yours, so the number is not yours".

Try this now

Five minutes. You will finish with your own inflation rate, from your own spending.

  1. Open your bank app or UPI app and pull up the last 3 full months of transactions. Most apps show a monthly spending summary or let you download a statement.
  2. Add the 3 months together and sort the total into 7 buckets: food and groceries (including delivery and restaurants), housing (rent, or home loan interest, plus electricity, water, gas), transport (fuel, cabs, train, servicing), health (premiums, doctors, medicines), education (fees, books, classes), clothing, and everything else.
  3. Turn each bucket into a percentage of your 3-month total. Those 7 numbers are your personal weights. Write them beside the official weights above.
  4. Open the latest CPI release. In India that is the monthly press release on mospi.gov.in, which prints an inflation rate for each group. In the United States it is the CPI news release on bls.gov, which does the same.
  5. For each bucket, multiply your weight by that group's published inflation rate. Add the 7 answers. That is your personal inflation rate.

What you should see. Two things.

First, your weights will not look like the official weights. A salaried person in an Indian city usually finds a food weight around half of 36.75%, and housing and transport weights well above the official figures. An American who owns a home outright finds a shelter weight nothing like 35.63%.

Second, your personal rate will usually differ from the headline by 1 to 3 percentage points, in one direction or the other. That gap is the number you have been ignoring in every savings calculation you have ever done.

Keep the 7 percentages. The next article uses them again.

Three real cases

1. India's CPI base revision, first published January 2026the ruler changed, not the room India replaced its 2012-base CPI with a 2024-base series built on the 2023-24 consumption survey. The food and beverages weight fell from about 45.86% to about 36.75%. Housing was extended to rural households. CDs and DVDs came out; headphones went in. The first print on the new series was 2.75% for January

  1. Anybody

comparing that to a 2019 figure is comparing 2 different baskets.

2. United States, June 2022headline and core told opposite stories Headline CPI rose 9.1% over 12 months, the largest increase in 40 years. Core CPI rose 5.9%. The gap was energy: the energy index was up 41.6% and gasoline alone up 59.9%. Food was up 10.4%. Shelter, the largest single component, was up only 5.6%. A person who walked everywhere and held a fixed mortgage experienced very little of that 9.1%. A person driving 100 kilometres a day experienced far more than it.

3. India's switch of anchor from WPI to CPI, 20142 official numbers, 2 answers Until 2014 the number in Indian public discussion was the Wholesale Price Index. A committee chaired by Urjit Patel recommended in January 2014 that the RBI anchor policy to consumer price inflation instead, and the RBI adopted it. The reason is instructive: WPI has no services and prices goods before they reach a household, so in some years it moved opposite to the prices people were actually paying. Both numbers were correctly calculated. They measured different things.

The question that resolves it

A novice reads an inflation number and asks: is it high or low?

An expert reads the same number and asks: what is in the basket, and what weight does it carry?

A 3% print driven by falling vegetable prices, with rents and school fees rising 8%, is not a benign number for a city household. It is an uncomfortable one wearing a calm face.

What would make this wrong

If the official basket were close enough to every household's basket, personal inflation rates would cluster tightly around the headline and this article would be a technicality. Household-level studies find the opposite: the spread across households in the same country and month is wide, and widest when food or energy prices move sharply.

Two honest limits. Your personal rate is an estimate, because you are applying national group rates to your own weights, and prices in your town are not the national average. And your weights change — a year containing a wedding or a hospital admission gives weights you should not use for a 30-year plan.

None of this means the official number is dishonest. It is a correctly calculated average. Averages are simply not descriptions of individuals.

In India

CPI is published by the Ministry of Statistics and Programme Implementation, around the 12th of the following month, for rural, urban and combined populations. The current series has a 2024 base. A separate Consumer Food Price Index is published alongside it.

WPI is published by the Office of the Economic Adviser, Ministry of Commerce and Industry, around the 14th. Goods only. It matters for contract escalation clauses and for reading input costs, not for your cost of living.

The target. Under the Reserve Bank of India Act, the central government sets an inflation target for the RBI. It is 4% CPI inflation with a tolerance band of 2% to 6%. In March 2026 the government retained the same target and the same band for a further 5 years. If average inflation stays outside the band for 3 consecutive quarters, the RBI must report to the government in writing and say what it will do.

That target is on headline CPI, not core.

In the United States

CPI is published by the Bureau of Labor Statistics in the second week of the following month. It is the number in every headline. Weights are updated annually. It also adjusts Social Security payments and many tax brackets, so it has direct legal consequences.

PCE, published by the Bureau of Economic Analysis at the end of the month, is the index the Federal Reserve actually targets. Its weights come from business surveys rather than household surveys, and it counts spending made on a household's behalf — most importantly employer-paid and government-paid medical care. That gives medical care a much larger weight in PCE than in CPI, and usually makes PCE inflation run slightly below CPI inflation.

The target is 2% on the PCE price index over the longer run. The Federal Reserve adopted an explicit numerical goal in January 2012. It set that goal itself; Congress did not. That is the opposite of the Indian arrangement. Core PCE is the series quoted most often in Fed communication.

Where they differ, and what that tells you

1. Food weight. Food is about 36.75% of India's CPI and about 13.70% of the US CPI. Indian headline inflation is substantially a food number. A good monsoon can pull it down by more than a percentage point with nothing else changing. The same food move barely registers in the US headline.

2. Headline against core, as the official target. India's target is on headline CPI. The Fed's target is on core PCE in practice. So the RBI is formally accountable for a number a monsoon can move, and the Fed is not. The RBI therefore has to explain, repeatedly, why it is not reacting to a food spike that interest rates cannot influence.

3. Housing. Shelter is 35.63% of the US CPI, and about 26.2 percentage points of that is an estimate rather than a transaction. Housing and utilities is 17.67% in India. So US inflation is dominated by an estimated, slow-moving rental number, and Indian inflation by food prices that can move 5% in a month.

What that tells you. Stop comparing the 2 headline numbers; they are not comparable. More usefully: in India, to tell a policy problem from a weather problem, look at core. In the United States, to tell a live problem from an old rental number still working through the index, look at core services excluding shelter. In both countries the step is the same. Open the components. The headline should be the last thing you read, not the first.

Carry this

  • Inflation is a weighted basket. The weights are the whole number.
  • The official basket is not your basket. Compute your own weights once and keep them.
  • What you earn on savings is the rate minus inflation. Your inflation, not the headline.

Knowledge check

Q. Two countries. In each, headline consumer inflation prints at 6% this month, and food inflation is 11%.

  • In Country A, food carries a 37% weight in the basket.
  • In Country B, food carries a 14% weight in the basket.

A salaried city household in each country spends about 15% of its budget on food. For which household is the headline number more misleading, and why?

Explanation. The distortion is the gap between your weight and the basket's weight, multiplied by how fast that group is moving.

In Country A the household spends 15% on food while the basket assumes 37%. That is a 22 percentage point gap, applied to a group rising at 11%. The headline carries a large amount of food inflation this household is not experiencing, so its true rate is well below 6%. In Country B the gap is 1 percentage point, so the headline is close to right.

The third option is the tempting one, and for a good reason: the food inflation rate really is identical, so it feels like the food effect should be too. It is not, because an index is weighted. The same price change enters the 2 indices at 37% strength and at 14% strength. The rate is half the calculation. The weight is the half almost nobody looks up.