Gold and real rates: known limitation 1

Holding gold as an inflation hedge over short periods. Over decades gold has broadly kept pace with prices; over the horizons people actually hold it, five years or ten, the correlation with inflation is weak enough that anyone relying on it will spend most of that time confused about what they own.

Gold and real rates: known limitation 2

Reading the relationship with real yields as a rule rather than a tendency. The sign is more dependable than the size, and there have been multi-year stretches where it inverted entirely because something larger was setting the price. A framework that only works when nothing unusual is happening is not much of a framework.

Gold and real rates: known limitation 3

Forgetting that gold has a carrying cost and no coupon. It produces nothing, so the entire return has to come from the price, and against a positive real yield the holder is paying for the privilege. That is a defensible thing to pay for, and it should be paid knowingly.

Playbooks / Cross-asset

Gold and real rates

Gold pays no income, so its main competition is the real yield on a government bond, and the relationship between the two explains more of gold's behaviour than inflation does even though inflation is the reason most people say they hold it.

The rule

Ask why somebody holds gold and the answer is almost always inflation. Look at what actually moves the price over the horizons people hold it for, and inflation is not the best explanation available.

The better one is duller and more useful: gold pays nothing, so its competition is whatever a safe asset pays after inflation. That number has a name, a market price and a public series, and it explains more of gold's behaviour than the thing everybody cites.

The mechanism

Gold produces no income. No coupon, no rent, no dividend. Every return it delivers has to come from the price.

A government bond does pay, and after taking out expected inflation what is left is the real yield. That real yield is precisely what a gold holder gives up. It is the carrying cost of the position, and it is not a metaphor: it is a number you can look up.

So when real yields rise, holding gold gets more expensive, and when they fall it gets cheaper. The relationship between the two has generally run in opposite directions for that reason. The direction is the reliable part. The magnitude is not, and anyone quoting a precise sensitivity is claiming more than the data supports.

The number to watch is the ten-year inflation-protected Treasury yield, published as DFII10. It is worth pausing on why that series is the right one: it is a market price for the real return on a safe asset, not somebody's inflation forecast. The chartbook carries it, alongside breakeven inflation, which is the market's implied expectation and a different quantity entirely. Confusing those two is the commonest error in this area, and inflation and real returns separates them properly.

Why the inflation story disappoints

Over very long periods, a century or more, gold has broadly held its purchasing power. That is true and almost useless, because nobody holds anything for a century.

Over the horizons people actually hold it, five years, ten years, a working life, the link to realised inflation is loose. There have been inflationary periods when gold did badly and quiet periods when it did well. Somebody who bought it as an inflation hedge will spend most of the holding period unable to explain what they own, which is the state in which people sell at the wrong time.

Holding it because the real yield is low is a claim you can check. Holding it because inflation might rise is a claim that will not settle for years.

Where the relationship breaks

A framework that only works in normal conditions should say so.

There have been multi-year stretches when this relationship inverted or simply stopped mattering, because something larger was setting the price: a currency losing credibility, official buying at scale, or a general scramble for an asset that sits outside the banking system. In those periods gold is not being priced against a real yield at all. It is being priced as an escape.

That is not a flaw in the framework so much as its boundary, and knowing where a tool stops working is most of knowing how to use it. It is the same discipline as where a framework breaks: ask what would have to be true for this to be wrong, before relying on it.

The practical part

If gold is going in, it goes in sized, like anything else. A permanent quarter of the portfolio, which is what the permanent portfolio asks for, is a very large standing bet on the argument above being right. Position sizing applies, and gold has a particular tendency to be bought at the moment of loudest agreement, which is when it is most crowded and most expensive.

Then decide what you are actually buying. The metal, a fund that holds the metal, and a mining equity are three different assets. A miner is an operating business with costs, debt, jurisdictions and management, and it will not track the metal in the way a buyer expects. Read what any vehicle holds and what it is permitted to do with those holdings, which is the same scrutiny fund domicile applies to everything else on a fund factsheet.

And be clear about the role. Gold is not income, it is not a bond, and it does not behave like either. What it can do is behave differently from the rest of the portfolio in the specific conditions that hurt the rest of the portfolio, which is the only argument for it that survives contact with the evidence. That is a real argument, and it is a smaller one than the case usually made.

The arithmetic

Why the comparison is with real yields Gold pays nothing A bond pays a nominal coupon Real yield roughly the nominal yield less expected inflation Holding gold gives up the real yield. That giving-up is its carrying cost. The mechanism, stated plainly Real yields up the cost of holding gold rises Real yields down the cost falls So the two have tended to move in opposite directions, and the sign of that relationship is more reliable than its size. Where the real yield is published The 10 year US Treasury inflation protected yield, DFII10 at FRED. A market price, not a forecast. What the relationship does NOT do It does not predict a level. It does not hold every quarter. It has broken for years at a time when something larger was moving the price: a currency crisis, a central bank buying programme, a scramble for something outside the banking system.

Where it breaks

Holding gold as an inflation hedge over short periods. Over decades gold has broadly kept pace with prices; over the horizons people actually hold it, five years or ten, the correlation with inflation is weak enough that anyone relying on it will spend most of that time confused about what they own.
Reading the relationship with real yields as a rule rather than a tendency. The sign is more dependable than the size, and there have been multi-year stretches where it inverted entirely because something larger was setting the price. A framework that only works when nothing unusual is happening is not much of a framework.
Forgetting that gold has a carrying cost and no coupon. It produces nothing, so the entire return has to come from the price, and against a positive real yield the holder is paying for the privilege. That is a defensible thing to pay for, and it should be paid knowingly.
Sizing it by conviction. Gold is the asset most likely to be bought at the point of maximum agreement about why it must go up, which is exactly when it is most expensive and most crowded, and the sizing question is the same one every other holding faces.
Confusing the metal with the miners, or with a fund that lends its holdings. A mining equity is an operating business with costs, debt and management, and it does not track the metal. Read what any vehicle actually holds and what it does with it.

When to use it

Before adding gold to a portfolio, and whenever the case for holding it is being made loudly. The useful question is not whether inflation is coming but what the real yield is, because that is the number the holder is choosing to give up.

Sources

Last reviewed 28 August 2026. Educational research, not personal advice. Disclosure standards.