Gold and real rates

What it costs to hold something that pays nothing

Gold produces no income. So the cost of owning it is not a fee — it is the yield you gave up by not holding a safe interest-bearing asset instead. And because gold is a claim on purchasing power rather than on rupees, the yield that matters is the safe yield after inflation: the real rate. When the real rate falls, holding gold costs less. When it rises, it costs more.

That is the whole mechanism, and it is worth stating plainly before anything else, because almost every popular explanation of gold reaches for fear instead. Fear is real and it does move the price. But fear is an event, and the real rate is a standing condition — it is there on quiet days too, quietly changing what the asset costs to keep.

Nothing in this article says where gold goes next, or where real rates go next. Neither question is answerable and neither is the subject. What is describable is the route: how a change in one reaches the other, and where that route runs out. This is one input taken apart in depth; reading the macro signals together covers how gold sits alongside the dollar, yields, crude and volatility, and why no tile is meant to be read alone.

The real rate is the rent you decline to collect

Start with the comparison an owner is actually making, whether or not they put it in these words. You can hold a bar of gold, or you can hold a short government bond. The bond pays you something each year. The gold pays you nothing. So far the bond wins by the size of its coupon, and that gap is what people mean when they say gold has a carrying cost.

But the coupon is paid in rupees, and rupees lose purchasing power at whatever rate prices are rising. Gold does not promise you rupees; it is a lump of metal whose whole return is whatever someone will pay for it later. Comparing a rupee income against a thing that holds no rupees means stripping the inflation out of the income first. Subtract expected inflation from the nominal yield and what is left is the return in purchasing power — the real rate. That is the number gold is actually competing against.

Now the mechanism does its work. A safe asset offering a healthy real return makes gold expensive to hold, because every year of holding metal is a year of that real return declined. A safe asset offering nothing after inflation makes gold cheap to hold, because the alternative has stopped paying anything worth having. And a negative real rate — a safe yield below the rate at which prices are rising — means the supposedly prudent choice loses purchasing power slowly and reliably. Against that, an asset that merely earns nothing is no longer the imprudent one.

One refinement, and it is the kind of detail that separates a mechanism from a slogan. Gold is not costless to keep: there is storage, insurance, a locker, or the expense ratio of a fund that holds it for you. That makes gold's own carry slightly negative rather than zero. So the point at which gold stops being the dearer thing to own is not a real rate of zero — it is a real rate a little below zero, by roughly the cost of keeping the metal safe. If you keep gold in a bank locker, the rent on it is part of this arithmetic, not a separate household expense.

The specific mistake this corrects is comparing gold's price change with a nominal deposit rate. “Gold did better than my fixed deposit” compares a price move against an interest yield, over a window chosen after the fact, with inflation left out of both sides. It is not a wrong calculation so much as a comparison of two different kinds of number. The decision-relevant quantity is the real rate, because that is what the holding is costing you while you wait. Inflation and interest rates are the two halves of it.

Two roads to the same real rate — and the one that catches people out

The real rate has two moving parts, and either can move it. That is why two apparently opposite stories — “gold rose because the central bank turned dovish” and “gold rose because inflation is running hot” — are the same mechanism described from different ends. Nominal yields falling and expected inflation rising both shrink the same difference.

The last row is the one worth sitting with, because it breaks the rule most people carry around. Falling nominal yields are not automatically good for gold. If yields are falling because the market has stopped believing in inflation altogether, and expectations are falling faster than the yields are, then the real rate is going up and the opportunity cost of holding metal is going up with it. Cash and short bonds gain purchasing power in that state. Gold's carrying cost is at its worst there, not its best.

This is also the answer to a question that trips up readers of the same argument applied to shares. Equity valuation runs off the discount rate, and a falling discount rate lifts the present value of distant earnings — the mechanics are in US yields and Indian valuations, and in duration for the bond version of the same arithmetic. Gold has no earnings to discount, so the rate reaches it by a different route entirely: not through a valuation, but through what the alternative pays. Same variable, two unrelated channels. That is why gold and equities can respond to the same rate move in the same direction on one occasion and opposite directions on another without either relationship having broken.

One honest caveat on the inputs. Expected inflation is not observable — nobody can read it off a screen. In the United States it is inferred from the gap between ordinary Treasury yields and inflation-linked ones, which is a market price and therefore an estimate with its own moods. So the real rate is itself a constructed number, not a measured one, and two people using different inflation measures will disagree about its level while usually agreeing about its direction.

What the mechanism does not explain

A mechanism that explained everything would be a suspicious mechanism. Real rates set the opportunity cost of holding gold; they do not set the price on their own, and there are identifiable buyers whose demand carries no yield term at all.

• Official reserves. Central banks hold gold as a reserve asset. That decision is about what a country's reserves should consist of and who could freeze them, not about a yield comparison. A reserve manager acting on those grounds is indifferent to the carry argument this article is built on.
• Jewellery and household demand. A wedding purchase is not a yield trade. Demand of that kind answers to income, to the wedding and festival calendar, and to custom. None of those carries a rate term either.
• Acute stress. In a genuine scramble for safety, gold trades on liquidity and fear for a while, and the carry argument goes quiet until the scramble ends. Occasionally it inverts: in a forced-selling episode holders sell what they can sell, and gold falls with everything else precisely because it is easy to sell.
• The dollar leg. Gold is quoted in dollars, so a change in the dollar changes the quoted price with no change in gold. That channel is described in the dollar index and emerging markets, and it matters twice over for an Indian buyer — see the next section.

So the mechanism does not license reading the price off the real rate. A period in which real rates rose and gold rose alongside them would refute nothing here — it would show one term in a sum being outvoted by the others. The right claim is that real rates move the carrying cost, always and mechanically, not that they move the price, sometimes. Carrying cost is one input into what buyers will pay. It is a large one, and it is not the only one.

That distinction matters for how you read a correlation. Our Macro page shows a grid of the standard measure of how tightly two series moved together over a window you pick — 30, 60 or 90 days, computed on daily returns. The measure is Pearson correlation, and gold is the input where people most often mistake the grid for the argument. A gold-to-Nifty coefficient tells you whether two series leaned the same way over the days in that window. It cannot tell you which moved first, whether either caused the other, or whether a third thing moved both — and with gold a third thing very often did, since the dollar sits inside both legs.

A correlation that flips sign between the 30-day and the 90-day window is telling you about the window, not about a change in the world. The wiring of the global economy does change, but rarely inside six weeks; a 30-day sample changes constantly. The usable discipline is to name the channel first and only then check whether the data over some window is consistent with it. Run that in reverse and you have measured a coincidence to two decimal places. Gold's reputation as a hedge against equity drawdowns rests, in most retellings, on exactly this inversion: an observation over selected windows, promoted to a property of the asset.

An Indian buyer holds two positions, not one

Here is the layer that global explanations of gold leave out entirely. The world price is quoted in dollars. An Indian buyer pays rupees. So the rupee price of gold is the dollar price multiplied by the number of rupees a dollar costs — two legs, multiplied, and only the first of them is about gold.

The arithmetic is short enough to redo. These are illustrative round numbers chosen so the multiplication is easy, not market levels: no live price appears anywhere in this article. Rupee return = dollar return × (new rate ÷ old rate).

Row three is the cushion people describe without naming. An 8% fall in the dollar price arrived as a 1.1% fall in rupees, and the difference was not gold doing anything — it was the rupee. Row five is the same effect with the gold leg removed: a domestic gain of 5% with the metal exactly unchanged.

The cushion is structural, not luck, and the reason is worth spelling out. A strong dollar pushes on both legs at once. It lowers the dollar price of gold, because gold is priced in the currency that just went up — and, as a reading rather than a mechanism, because the conditions that lift the dollar often include a higher real return available in dollars, which is the carry argument arriving by the other door. It also weakens the rupee, through the channels set out in the dollar index and emerging markets. One cause, two legs, opposite signs in the multiplication — so they partially offset by construction rather than by coincidence.

Be careful about what that does and does not establish. That an offset exists is a matter of wiring, and you can reason it out before looking at any data. How much of any particular fall it absorbed, over what stretch, is a separate and empirical question — it needs a computed series, a stated window and an as-of date, and no such figure appears in this article. “The rupee has always cushioned it” is a claim about a sample somebody has usually not specified.

And the trade-off, because a dampener works in both directions. The same wiring trims the gains: a gold rally powered by a weak dollar is the same dollar move that firms the rupee, and a firmer rupee eats part of the rupee return. Row two. An Indian gold holder owns an unhedged short position in the rupee that they did not choose and cannot separately size, and it flattens the outcome at both ends. Whether that is a feature depends entirely on what the holding is for — which is a question a purchase made out of custom or reassurance never has to answer, and the one the common money mistakes article puts at the centre of it.

Two further Indian-specific wedges sit between the world price and what you pay, and neither has anything to do with real rates. The domestic price is a landed price: the world price converted at the exchange rate, plus import duty and local levies, with tax on purchase on top. A change in the duty rate moves the Indian price in a step while the world price stands still. And the form matters — jewellery carries making charges that are not recoverable on resale, which is a cost of the object rather than a cost of the metal. That last cost is part of why borrowing against household gold is a lending category in its own right here, with its own RBI loan-to-value ceilings: it raises money from the metal without paying the making charges away a second time.

Which real rate is yours

Now a distinction that almost nothing written on this topic makes, and that changes which number an Indian reader is actually looking at.

The real rate that sets the world price of gold is an American one. The real rate that measures your opportunity cost is an Indian one. They are different numbers, they are set by different central banks, and they do not have to move together.

The reason is that these are two different questions wearing the same words. The dollar price of gold is set in a global market where the dominant alternative to holding metal is a dollar-denominated safe asset, so it is the dollar real yield that shapes what global buyers will pay. But the alternative you are declining is not an American bond. It is an Indian deposit or an Indian government bond, earning an Indian nominal rate, eroded by Indian inflation. That is the rent you are actually not collecting.

The consequence is uncomfortable and precise. A position in gold can be getting cheaper to hold in the world's terms — a falling US real rate, supporting the dollar price — while getting dearer to hold in yours, if the Indian real rate is rising at the same time. Both statements are true simultaneously and they are about different things: one is about the price, the other is about your cost. Reading a single global commentary about “real rates and gold” will give you the first and silently omit the second.

There is a measurement problem attached, and it should be stated rather than papered over. The US real rate can be read approximately off traded inflation-linked bonds. India has no comparably deep market in them, so an Indian real rate has to be constructed: take a nominal yield and subtract an inflation number you have chosen — retail inflation, wholesale inflation, a core measure, last year's print or a forecast. Different choices give materially different answers about the level. Treat the Indian real rate as an estimate with a visible construction, not a quantity you can look up, and be suspicious of any argument that turns on its second decimal place.

One structural note, kept general because the specifics change. An instrument that pays a coupon on top of the gold price — a gold-linked bond rather than metal or a fund holding metal — alters this arithmetic directly, because part of the forgone yield is handed back. Whether such an instrument is on sale at any given time is a separate question, and not one this article answers. The point is the structure: the carrying cost is a property of the wrapper as much as of the metal, and two ways of owning the same gold can have different carries.

Why a hedge carries a minus sign in our composite

Because the score is not rating gold. Gold carries a weight of 0.03 and a negative sign, so rising gold pushes the composite down — and every tile on the page answers exactly one question, which is what this does, mechanically, to Indian equities. On that question our reading has two parts. A gold rally, especially one running alongside a rising dollar, is most readable as hedging demand: buyers paying up for protection, which is the same information a volatility index carries and a caution sign for the residual claim, which is equity. And second, a domestic channel that a global model would not have: gold is an enormous household asset in India, and strength in it can pull household savings toward metal and away from shares. Neither part is a claim that gold is a bad thing to own.

That is the sign. The weight is the other half of the entry, and it sits inside this arithmetic. FNOTrader's Macro page reduces a long list of inputs to one number. Each component is scored for its effect on Indian equities, multiplied by a weight, summed, and divided by the total weight so the result sits on a fixed scale: score = 100 × Σ(wₓ · cₓ) ÷ Σ(wₓ). Twenty weighted market inputs go into that sum and their weights total 1.33; foreign-flow is folded in at scoring time for a further 0.12, so the divisor reaches 1.45 when everything reports and falls below it when a feed fails. Which is why the divisor is the sum of the weights rather than one. Each component's contribution is capped, so a single input having a wild day cannot dominate the total. A score above +20 or below −20 is labelled a regime rather than a wobble.

This is the same convention that governs every tile on the page, and it is the one visitors find hardest. Colour shows the scored effect on Indian equities, not the direction the number moved. A falling dollar shows green. A rising USD/JPY also shows green, because a weak yen is the condition under which the yen-funded carry trade stays intact. There is no “up is red” rule to learn; each tile carries a sign we chose, and the sign is the entire content of the colour.

The weight, the sign and the cut-offs are our judgement, not measured constants. Nobody has established that gold is worth 0.03 of anything, or that a regime begins at 20 rather than 18. They encode a considered view, applied consistently so that the same view is expressed the same way every day. A different considered view would set them differently and would not be wrong. Treat the composite as a stated opinion with a formula attached, not as a reading off an instrument. The full scoring mechanics belong to the article on reading the page as a whole.

And now the honest part, which is the reason gold's weight is small rather than large. The single sign cannot distinguish the two reasons gold rises. A falling real rate makes gold cheaper to hold — and a falling real rate is the same condition that lifts the present value of distant corporate earnings, which is supportive of equities rather than a warning about them. So a rally that reflects a shrinking carrying cost and a rally that reflects fear arrive at the tile as one identical observation, take the identical minus sign, and mean close to opposite things for shares. That limitation is not fixable inside a one-number-per-input design. It is the argument for reading gold beside the yield tiles and the volatility tiles rather than on its own: gold up with volatility up is one configuration, gold up with real yields falling and volatility calm is another, and the tile alone cannot separate them.

The price of the hedge, in arithmetic

Nothing in a portfolio is free, and the cost of gold is unusually easy to state, because it is the mechanism itself running for as long as you hold.

Take an illustrative safe real yield of 2% a year — a made-up round number, not a market level. Decline it for ten years and the safe alternative has grown by a factor of 1.02 to the tenth, which is 1.22. You have given up about 22% of purchasing power against that alternative before gold has done anything at all. Gold has to make that up in price just to draw level. At an illustrative 4% real yield the same ten years cost about 48%, since 1.04 to the tenth is 1.48. The carrying cost compounds exactly the way returns do, which is the part that gets left out of the sentence “gold is insurance”.

That arithmetic is the trade-off in its clearest form. What you buy for it is a holding whose value does not depend on any borrower paying you back or any company earning anything — protection against a specific and real set of outcomes. What you pay is a certain, compounding, forgone real yield, incurred every year whether or not those outcomes ever arrive. Insurance you never claim on is not a failure; it is the ordinary result. But it is a cost, and a cost stated as a percentage of net worth is the only version of it a household can actually reason about. Where a holding sits in the plan is the subject of asset allocation; the arithmetic above is what that decision is trading away.

Note what the mechanism does not license. Everything here describes the cost of a position and the channels that change it. None of it says what any real rate will do next, which way gold will move, or how much of anything anyone should hold. Those are different questions, and the first two of them are not answerable.

Where to look at this

The reading is the part that has to be yours. The mechanical part — keeping the series together, on one scale, over comparable windows — is what the Macro page in FNOTrader's Options Analytics app does.

It shows gold alongside the dollar index, US yields across the curve, credit spreads and both volatility indices, each as a tile coloured by its scored effect on Indian equities, with the weights on display rather than hidden, and the composite and its regime label computed from them. The correlation grid runs Pearson on daily returns over a selectable 30, 60 or 90-day window, so a relationship can be checked against more than one sample before it is believed.

The weights and signs are ours and are published so they can be argued with. The correlations are arithmetic on price series and carry the limits described above whoever computes them.

Common questions

Continue reading

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