Commodities and Indian inflation: what actually reaches the price you pay

A world price is the start of the journey, not the end of it

A commodity reaches an Indian consumer price through four filters, in order: the exchange rate turns a dollar price into a rupee one, the tax and subsidy structure decides how much of the move is allowed through, the item's share of the consumption basket decides how much the surviving move matters, and time decides when any of it shows up. Each filter can shrink the move, and what reaches a shelf is rarely what the chart showed.

That is why the two charts people compare — a commodity price and an inflation print — are not two views of the same thing. One is a price a trader can transact at right now. The other is an average of what households actually paid, over a month, for a basket in which that commodity is one line among hundreds.

The filters multiply, they do not add. A commodity that doubles in dollars but carries a small basket weight, sits behind a subsidy and is bought mostly by industry rather than households can move the consumer price index by almost nothing while dominating every market screen. The reverse also happens: a modest move in something India imports directly, eats immediately and taxes as a percentage arrives nearly in full.

The first filter is the least interesting and the most often forgotten. India pays in dollars for what it imports, so the relevant number is never the world price alone — it is the world price multiplied by the exchange rate. A commodity that is flat in dollars while the rupee weakens has become more expensive here, and no commodity chart shows that. The currency leg is set out in what the rupee does to an Indian portfolio; here it is simply the first multiplication in the chain.

The multiplication, with illustrative numbers

Put numbers on it, chosen to make the arithmetic legible rather than taken from any actual series.

Say a commodity rises 20% in dollars over a quarter, and over the same quarter the rupee weakens 2% against the dollar. The landed rupee price is up about 22%, because the two effects compound rather than add. Now say the item accounts for 3% of the consumption basket. If every paisa of that 22% reached the shelf, the arithmetic contribution to the overall price level would be 0.22 × 0.03, or about 0.7 percentage points.

Then apply the wedge. Suppose the item carries a fixed per-unit duty that does not move with the world price, and suppose that block is large enough to dilute the rise by half by the time it reaches retail. The arriving move is 11%, and the contribution is around 0.3 percentage points. Same commodity, same world move, half the effect — and nothing about the commodity changed. What changed was the structure sitting between it and the buyer.

Two things are worth taking from that arithmetic rather than from the specific figures. The first is that basket weight is a hard ceiling on the direct effect: an item's own line in the index cannot contribute more than its weight allows, however dramatic the price move, so a small-weight item is arithmetically incapable of driving a headline print through its own line. Anything beyond that ceiling has to arrive through some other item's weight — diesel raising the price of a vegetable is crude arriving under the food line, not under the fuel one — which is the second round, and it is slower and smaller. The second is that the wedge is where the discretion lives, and the next section is about who exercises it.

Pass-through is a decision, and it is taken by someone

Here is the point most coverage of this subject skips. Pass-through is not a property of a commodity. It is a policy choice, taken commodity by commodity, and it can be reversed.

The reasoning is an accounting identity rather than an opinion. A rise in the world price of something a country imports is a real cost, and real costs do not evaporate. Somebody pays. There are exactly three accounts available:

1. The household, if the retail price is allowed to rise. The shock is recorded as inflation.
2. The government, if duty is cut or subsidy raised to hold the retail price down. The shock is recorded as forgone revenue or higher expenditure — a wider fiscal deficit.
3. The company, if the retail price is held and the difference comes out of a producer's or distributor's margin. The shock is recorded as a corporate earnings problem.

There is a fourth adjustment, and it is worth naming because it is invisible in all three accounts: buyers can simply buy less. That shows up as a quantity change, not a price change, so it never appears in an inflation print at all — which is one reason a subdued print is not by itself evidence that a shock was absorbed painlessly.

The fuel version of this — excise, state tax and oil marketing margins — is worked through in full in crude oil and the Indian economy, and it is the case everyone knows. It is also the least representative one, because fuel is only a single instrument out of several. The wider set is what makes Indian commodity inflation hard to read from a chart:

• Ad valorem import duty — a percentage of value, so it rises in rupees when the world price rises and amplifies the move. It is also the quickest thing to change: a duty cut takes effect on the date the notification names, with no legislative cycle in between.
• Specific duty — a fixed amount per unit, unchanged whether the commodity is cheap or dear. It dilutes a percentage move automatically, without anyone deciding anything.
• Subsidy — the state pays the difference between what a producer needs and what a buyer is charged. Here pass-through to prices can be close to zero while pass-through to the budget is close to total.
• Administered and regulated prices — where a regulator sets the tariff, the input cost reaches the bill only when a tariff order allows it, in the size the order allows, on the cycle the order runs on.
• Quantity instruments — export duties, export restrictions, stock limits and releases from a buffer stock. These do not adjust a price; they sever the link between the domestic price and the price the same good fetches abroad.

Fertiliser is the cleanest example, because it is the case where the identity is most visible. A large part of India's fertiliser input is imported and priced globally, while the price a farmer pays is controlled or heavily subsidised. When the world price of a fertiliser input rises, the farm-gate price largely does not, and the increase shows up in the subsidy bill. The shock does not enter the inflation statistics as a fertiliser price at all — it enters the budget. It can still reach food prices later and indirectly, through what farmers decide to grow, but by then it is unrecognisable as the thing that moved. Anyone tracking that commodity through the consumer price index is watching the wrong instrument.

The trade-off, which is real and rarely stated alongside the relief: none of these instruments makes the cost disappear. A duty cut narrows the tax base at the moment revenue is wanted; a larger subsidy widens the deficit, and the extra borrowing is placed in the same bond market that prices every other rupee asset, so the cost reappears as a question about interest rates rather than as a line on a shopping bill. An export restriction holds a consumer price down by holding a producer's realisation down, which is a transfer between two domestic groups, and it changes the price signal the next sowing decision is made against. Suppressed pass-through is displaced pass-through.

Which produces the observation the whole article exists for. The same commodity move can look completely different in the data across two years, with no change in the commodity and no change in the economy — only a change in which account was chosen. Comparing this year's inflation response with a previous episode's, without checking whether the duty and subsidy settings were the same, compares two different policy regimes and calls the difference a fact about the world.

Not one channel — six, with different mediators

Grouping commodities by what stands between them and the shelf is more useful than grouping them by what they are. The mediator decides both how much arrives and how quickly.

Read the third column, not the first. Two commodities can move identically in dollars and reach the price level through completely different doors, at completely different speeds, in completely different sizes. A commodity chart tells you about the first column and nothing about the other two.

Gold is the odd entry and worth a sentence, because it breaks the pattern the other five follow. Every other commodity here is an input to something: crude becomes freight and packaging, metal becomes a machine, fertiliser becomes a crop. Gold becomes an ornament and then sits in a locker. So it has a direct effect on the measured price level — households buy it, and the price they pay is collected — and essentially no second round, because nothing downstream is manufactured out of it. It produces the strange result that an asset price rally is recorded as consumer inflation. Why the metal's price moves at all is a different question, answered by the relationship between gold and real rates.

Four arrival times, and the mistake that comes from ignoring them

The specific mistake worth naming: watching a commodity chart and expecting the next month's inflation print to look like it. They are measuring different things over different windows, and even where the channel is real the timing rarely lines up.

Four speeds, in order:

1. Same month. Items households buy in their raw form — fuel at the pump, cooking gas, edible oil, gold. The price collected is close to the price that moved.
2. Weeks to a quarter. The freight round. Diesel reprices the cost of moving everything, and that reaches shelf prices as contracts renew rather than on the day the barrel moved.
3. Quarters, and diluted. The manufacturing round. A metal or a polymer is a modest share of a finished good's cost, sitting under a producer's margin that can absorb a move before passing it on, and under contracts written months earlier.
4. Possibly never. Anything the wedge fully absorbs. A subsidised input, a regulated tariff that is not revised, a duty cut sized to offset the move.

There is also a purely mechanical timing effect that gets read as news. An inflation print is a comparison with the same month a year earlier. So a commodity that rose sharply twelve months ago and has been flat since still produces a high reading until that old rise drops out of the comparison window — and on the month it drops out, the reading falls without any price falling. That is arithmetic in the measure, not an event in the economy. The general mechanics of a price index are set out in inflation explained.

The consequence for a reader is a change in what to look at rather than a change in what to expect. If you want to know whether a commodity move is reaching households, the fuel and food sub-indices of the consumer price release answer it directly, and they answer it with a lag that is knowable rather than guessable. The commodity chart cannot answer it at all.

The largest part of the basket is not global at all

Everything above assumes the commodity is imported. For the biggest block of the Indian consumption basket, that assumption mostly fails.

Food is the largest single group in the Indian consumer price basket, and a large part of it — cereals, vegetables, milk, most fruit — is grown domestically and priced domestically. Its swings are driven by rainfall, sowing area, storage, disease and the state of a road, none of which appears on any global commodity screen. A vegetable price spike is a supply event in a district, not a signal from a world market.

Where the world price does reach food, it does so by two specific routes. The first is direct import: an item India does not grow enough of, edible oil being the clearest case, in which the world price and the duty rate largely set the domestic one. The second is export parity: for something India both grows and exports, a domestic seller can always choose the export market, so the world price sets a floor under the price at home even though nothing is being imported. That second channel is exactly what a quantity instrument is designed to cut, which is why export restrictions on food are a price-control tool as much as a trade one.

This is also why the distinction between the headline number and the underlying one keeps coming up. The measure with food and fuel stripped out — conventionally called core inflation — exists because those two groups swing on supply events that a policy rate cannot address: a central bank cannot make it rain, and raising rates does not produce vegetables. But the target the Reserve Bank of India works to is defined on the headline index, food and fuel included, which is the tension underneath most commentary about what policy ought to do. The reaction function itself belongs to the article on RBI policy; what matters here is that the commodity part of an inflation print is the part policy is least able to act on and most obliged to count.

The wholesale read-across — a named way to get this wrong

India publishes a wholesale price index alongside the consumer one, and the two routinely tell different stories about the same commodity move. The wholesale number spikes; the consumer number barely registers; someone declares one of them wrong.

Neither is wrong. The gap between them is the wedge, measured.

The wholesale index covers goods at the point of bulk transaction — no retail margin, no retail tax structure, and no services at all. The consumer index covers what households actually spend on, which includes rent, education, health, transport fares, telecom and domestic help. None of those is manufactured out of a commodity, and together they occupy a large part of the basket that no commodity move can reach.

So a commodity shock hits an index made almost entirely of goods with full force, and hits an index containing a large services and housing block with much less. Reading a wholesale spike as an advance warning of the consumer print is the error — call it the wholesale read-across. The two indices have different baskets, different coverage and different collection points, and the difference between them is structural rather than a lag waiting to close.

What the wholesale number is genuinely good for is the corporate question rather than the household one. It is close to the cost side of a manufacturer's income statement, so a widening gap between wholesale input prices and consumer output prices describes a margin being squeezed somewhere in the chain — which is a claim about company earnings, not about the cost of living.

Reading the commodity tiles on the Macro page

Everything above is mechanism. A dashboard is an observation, and the two have to be kept apart — particularly here, where the mismatch between them is larger than on any other part of the page.

FNOTrader's Macro page carries six commodity tiles: Brent, WTI, gold, silver, copper and natural gas. Only three of the six carry any weight in the composite score. Brent carries 0.12 with a negative sign, gold 0.03 negative, and copper 0.04 positive. WTI, silver and natural gas are displayed and charted but not scored at all — they are there for context.

The score itself is score = 100 × Σ(wᵢ·cᵢ) / Σ(wᵢ), where each input is scored for its effect on Indian equities and then weighted. Twenty weighted tiles sum to 1.33, and foreign institutional flow is folded in separately at a weight of 0.12 — it is not one of the tiles — for twenty-one contributions and a divisor of 1.45 when every feed reports. A feed that fails drops out of the numerator and the denominator together, so 1.45 is the all-present maximum rather than a constant. That gives the three commodity tiles these ceilings, which are derived from the weights rather than measured:

Read that last row twice. The entire commodity complex on the page can move the composite by less than the dollar index can on its own, and the regime boundaries sit at ±20 with a stress override at −35. That is not a claim that commodities matter less than the dollar to Indian inflation; it is a statement about what this particular score was built to summarise, which is the risk read for equities on the day.

Those weights, signs and cut-offs are FNOTrader's design judgement, not measured constants. No regression produced them and no dataset selected them. A reasonable analyst would set them differently — heavier on crude if they weighted the fiscal channel, lighter if they thought the rupee tile already carries the oil effect — and the composite is best read as one view expressed numerically rather than as a measurement. None of those numbers is printed on a tile either — a tile shows the label, the level, the day's change, the 5-day change and an expandable note. The weights ride in the payload behind it, so a reader looking at the page cannot see which tiles are doing the work.

Three features of the commodity tiles are read wrongly more often than not.

1. Colour shows the effect on Indian equities, not the direction of the number. Falling Brent shows green because cheaper crude relieves the import bill; rising copper also shows green, because copper is read as a global growth proxy rather than as a cost. Rising gold shows red, because the model treats a gold rally as hedging demand. Two commodities moving up on the same day can therefore show opposite colours, and both are correct under the convention.
2. The unweighted tiles do not follow that convention, which is a known wrinkle rather than a design. A tile with no weight has no impact score, so its colour falls back to raw direction: rising is green. On a day when the whole energy complex rises, Brent goes red and WTI goes green, on the same barrel. Read the unweighted tiles as context, not as signals.
3. Each input is clamped before it is weighted. A tile's contribution saturates at a per-tile daily move — a 3% day for Brent, 2% for gold — beyond which a bigger move adds nothing. So a 3% crude day and a 30% crude day contribute the same amount to the score. The composite measures agreement across channels, not severity within one, which is precisely why the stress override has to bypass the average: one input in genuine distress would otherwise be diluted by twenty calm ones. That is why the stress state also fires on a single reading — a one-day VIX spike, or a fast move in the yen — whatever the average says.

And the boundary that matters most for this topic. The Macro page cannot see pass-through, and is not trying to. A tile reads a one-day change in a globally traded price. Pass-through is measured monthly, arrives over weeks to quarters, and is mediated by duty and subsidy decisions that no market price contains. The tile is a same-day risk read; the inflation channel described in this article runs on a different clock entirely, and expecting the first to anticipate the second is a category error rather than a data problem.

The same caution applies to the correlation heatmap, only more so. It computes the ordinary linear measure — Pearson correlation — on daily returns of the Nifty against a fixed list of ten drivers, over the last 30, 60 or 90 sessions, and three of those ten are commodities. A correlation there tells you how an equity index and a commodity moved together day by day. It says nothing about whether the commodity reached a consumer price, because a price level is not in the calculation at all. Short windows are noisy, and a correlation that flips sign between the 30-session and the 90-session window is describing the window. The full reading of the page is in the pillar on reading macro signals.

Common questions

Continue reading

More in Macro & Markets · App: Options Analytics · Definitions: glossary · Free tools: calculators · All: every article