US yields, and the two routes they take into Indian share prices

Why a US yield is anybody else's business

A US government bond yield is what the world's most reliable borrower pays to borrow. It is close to the return on money that is simply parked, so every other asset has to offer more than it. Raise that floor and everything above it has to reprice.

An Indian share sits a long way above that floor, but it is on the same ladder. It competes for the same global capital, and the capital has a choice. So a change in the yield on US government debt reaches an Indian share price by two routes that are worth keeping apart, because they act on different things and at different speeds.

The first is arithmetic. A share is a claim on money the company will produce far into the future, and converting future money into today's money is division by a discount rate. Move the rate, and the answer moves — before a single share has traded.

The second is behaviour. When safe dollar assets pay more, the return an Indian asset has to promise before a global allocator will hold it goes up too. Capital reallocates. That takes weeks, not milliseconds.

Neither route says anything about what the market will do. They say what pressure exists and where it enters. This article traces both, and then says where the chain is weaker than the standard telling admits. If you have not read what a yield is and how it differs from a coupon, start there; the mechanism below assumes it.

The first channel: what a distant rupee is worth today

A company is worth the money it will hand its owners, discounted back to today. Discounting is the arithmetic of that sentence: ₹100 arriving in ten years is worth less than ₹100 arriving tomorrow, and how much less depends entirely on the rate you divide by.

Here is the whole effect in one illustrative table. Take ₹100 due at four different horizons, and discount it first at 8% and then at 9% — a rise of exactly one percentage point. The figures are arithmetic you can redo on any calculator: present value = 100 ÷ (1 + r)ⁿ.

One rate change, four completely different answers. The distant rupee does almost all of the moving. Money due next year loses under 1% of its present value; money due in twenty years loses about a sixth of it. Nothing about the company changed in that table — the promised rupees are identical in every column.

Now the precision that most versions of this story skip. An Indian company's rupee cash flows are not discounted at a US yield. They are discounted at a rupee rate: an Indian government yield plus the extra return equity holders demand for taking business risk. The US yield reaches that rupee rate indirectly — through pressure on the currency, through Indian yields, and through the premium global capital demands for holding emerging-market equity at all. It is an anchor, not the divisor itself. Getting that wrong turns a real mechanism into a slogan.

Long-duration equity: the damage is not spread evenly

Read the table above again and a second conclusion falls out. If the size of the repricing depends on when the money arrives, then two companies facing the identical rate move take entirely different damage — purely because of the shape of their cash flows.

This is the same idea the library already owns for bonds. A long bond falls further than a short one on the same yield move because it is uncompetitive for longer, and the formal measure of that sensitivity is duration. Equities have the same property without the tidy number: a business whose value sits in cash flows twenty years out behaves like a long bond, and one that is already generating most of what it will ever generate behaves like a short one.

That is a mechanical claim about arithmetic, and it is worth separating from what any real share does. A real company also carries debt that reprices, may earn in dollars, and has an earnings outlook that is itself moving. Duration is one term in the answer, not the answer.

Here is the part that changes what you look at. An index is a weighted bag of durations, so an index-level move on a yield print is largely a statement about the index's composition. Two markets can face the same yield move and take different damage without either one disagreeing with the other about the yield. When a benchmark falls on a rate day, the useful question is not “did the market dislike the print” but “which duration bucket did the fall come from” — because if it came from a handful of long-duration names, the index number has told you about those names and very little about the rest of the market. Sector composition is doing more work in that headline than the headline admits.

The second channel: the hurdle every rupee of foreign capital has to clear

The arithmetic channel needs no investor to do anything. The second one is entirely about investors doing something.

A global allocator can hold safe dollar assets or take risk somewhere. When the safe option pays more, the return demanded from the risky option rises with it — not because India got worse, but because the alternative got better. Higher US yields raise the hurdle rate for holding an Indian asset. Whatever no longer clears that hurdle becomes a candidate to be sold, and the sale is denominated in rupees that then have to be converted back.

Which is why the currency and the flows are one channel and not two. Selling Indian equity means selling rupees. And for a dollar investor who has not hedged, the realised return is the Indian return plus the currency move — so a weakening rupee eats into the dollar outcome and raises the Indian return required to compensate for it. The two effects push the same way, which is what makes this channel self-reinforcing while it runs.

Two honest qualifications, both of which matter for how you read the data.

A flow number is a footprint, not a force. Foreign portfolio flow is a record of transactions that already happened. The mechanism is the hurdle rate; the flow figure is evidence about it, published after the fact and revised. Treating the flow series as the cause reverses the arrow.

The two channels do not arrive together. Discounting reprices on the yield print itself, in the same session. Reallocation of capital across countries takes weeks and runs through mandates, committees and hedging decisions. So an index can absorb a yield move on the day and still be facing the second channel a month later. Reading the day's close as the market's verdict on the yield conflates a fast mechanism with a slow one.

The same yield move, two different meanings

A yield tile cannot tell you why the yield moved, and the why decides whether an Indian company is worth less or not.

A nominal government yield is the sum of several things: what lenders expect real growth to pay, what they expect inflation to be, and the extra compensation they want for tying money up for a decade. The yield can rise because any one of those rose. The printed number is identical in each case. The implications for Indian earnings are not.

Call it the composition problem. Two moves of identical size in the same direction can carry opposite implications for what a company is worth, and the level on the tile distinguishes them not at all. Anyone reading a yield chart as a single variable is reading three variables that happen to be printed as one.

Where you look to separate them: the gap between a nominal yield and the inflation-protected yield of the same maturity is the conventional way to split the move into a real-rate part and an inflation-expectation part, with the residual usually labelled term premium. That is the conventional reading rather than a settled fact — the decomposition depends on the model used, and reasonable people produce different splits from the same two yields. It is still far more informative than the headline number on its own.

The denominator has two terms, and only one of them is on a tile

The rate an Indian company's cash flows get divided by has two halves: the safe anchor, and the extra return an investor demands for taking equity risk instead of lending safely — the equity risk premium. Two terms, one sum. Almost every telling of this story, including everything above, moves the first and quietly holds the second still.

The second does not hold still. It also gets no tile, on our page or anyone's, because it is not observable: it is backed out of prices, which means it is inferred from the very thing you were trying to explain.

The two terms do not move independently, and the sign of their relationship is not fixed. This is the leg most explanations skip, and it is where the earlier composition table pays off — because what the yield rose on also says something about which way the second term is leaning.

Those two rows are why an equity index and a bond yield can rise together for weeks without anything being broken: the anchor went up, the premium came in, and the sum the market actually discounts at barely moved. They are also why the second case bites harder than a table of discount factors implies — the divisor moved by more than the yield did.

And this, not window length alone, is the honest answer to why a yield-versus-equity correlation flips sign. The two series genuinely relate differently depending on which component of the yield is moving. A 30-day window sitting inside a growth-led stretch and a 90-day window spanning a term-premium-led one are measuring two different regimes and each reporting one number for it. A correlation that changes sign is a prompt to ask which component moved, not a fault in the data.

The trade-off in knowing this is real: the risk premium is not measurable, so none of it converts into a number you can monitor. What it buys is the habit of asking, on any yield move, whether the second term is leaning with the first or against it — and accepting that two careful people will answer that differently on the same day. That is a judgement, and it should be held as one.

Where the chain is weaker than it looks

An article that only builds the chain has not finished. Four places it is weaker than the standard telling admits.

Who the marginal buyer is decides how hard the flows channel bites. A market funded largely by foreign capital transmits it almost directly. One with a large, steady domestic bid — monthly systematic flows, insurance and pension money — can absorb foreign selling without the price clearing where it otherwise would. India's domestic bid has grown relative to a decade ago; that is an empirical claim about how the market is funded, it shifts over time, and no figure is attached to it here. None of it touches the arithmetic channel, which is division and does not care who is buying.

Indian policy is set for Indian conditions. The RBI's policy rate responds to domestic inflation and growth. It is not a follower of US policy, even though the currency creates a link that is real and that policymakers watch. The Indian discount rate can move less than, more than, or opposite to the US one over any given stretch.

Everything moves at once. Yields rarely move alone. The dollar, crude, volatility and flows all shift together, and a share price is absorbing all of them plus its own earnings news. Attributing a day's move to one macro variable is attribution, not measurement.

Correlation is not the mechanism. This is where most macro reading goes wrong, so it is worth being exact. A correlation says two series moved together over some window. It does not say which caused which, whether a third thing caused both, or whether the relationship holds outside the window you measured. A correlation is consistent with a channel; it is not evidence that the channel exists. The direction of the argument has to run from the mechanism to the data, never the other way — which is why every section above builds a channel first, and the one place a correlation appears, it appears as something the channel has to explain.

Reading this on the Macro page

Everything above is readable off any yield chart and a calculator. FNOTrader's Macro page assembles the pieces in one place, and three things about how it presents them are worth stating plainly, because two of them are counter-intuitive and one is a design choice people mistake for a fact.

Colour means impact on Indian equities, not the direction of the number. This trips up nearly everyone on first read. A tile turns green when the reading is supportive for Indian equities through the channel described above — so a falling dollar index shows green, because a softer dollar loosens global financial conditions. The colour is answering “what does this do to us”, not “did this go up”. Read it as an arrow on the number and you will misread every tile on the page.

The composite is a weighted average, and the weights are our judgement. The page scores the set as score = 100 × Σ(w_ic_i) ÷ Σ(w_i), with the dollar index at 0.20, the US 10-year yield at 0.15, Brent and foreign flows at 0.12 each, the yen cross and volatility at 0.10 each, and regime boundaries drawn at ±20. Those numbers are FNOTrader's modelling judgement about what matters most to Indian equities — a considered view, not a measured constant and not an estimated coefficient. A different reasonable weighting produces a different score from identical inputs. Anyone treating 0.20 for the dollar as a property of the world is making the same error as treating industry practice as regulation.

The heatmap is Pearson correlation on daily returns, over a 30, 60 or 90-day window. Short windows are noisy by construction, so a pair whose correlation flips sign between 30 days and 90 may be telling you only about the window — and, as the section above sets out, may instead be telling you that the two windows caught yield moves of different composition. Both readings are live, which is exactly why a single window number settles nothing. Check all three, then use the heatmap to ask whether the data is consistent with a channel you can already name — never to discover a channel.

The wider set of tiles, and how they are meant to be read together rather than one at a time, is the subject of the macro pillar. This article is the long version of one tile.

Common questions

Continue reading

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