Gold and the Real Yield: Two Channels, Opposite Signs

Gold pays no coupon, so the question of what it earns has its answer somewhere else. It is in the bond market, in the yield on a Treasury that pays its holder after inflation. That yield is what a holder of gold gives up, and it is the standard account of why gold tends to weaken when real yields rise.

The more interesting case is the one where the yield rises and gold does not fall the way that story says it should. When that happens, a second channel may be acting on the price, with the opposite sign. This piece names both, says how quickly each can be seen, and sets out what each would look like in the data.

The First Channel: What Gold Gives Up

A Treasury inflation-indexed bond, a TIPS, pays a yield measured after inflation, owed by the US government. The Federal Reserve Board publishes the 10-year constant-maturity yield on those bonds every business day in its H.15 release, with the US Treasury as the source. Hold gold instead and the real yield is the income you forgo. When it rises, the cost of holding gold rises with it, all else equal. The sign is negative.

It is also the channel that can be seen fastest, because it is a market price. The 10-year inflation-indexed yield was 1.94% on 2 January 2026, reached its 2026 high of 2.95% on 5 October, and was 2.78% on 15 October (Federal Reserve H.15; history on FRED as DFII10). That snapshot has a date on it and will be stale by the time most people read this sentence; the series is public and its latest observation is a click away.

Whose behaviour does it describe? Holders who compare gold with bonds. The nearest primary evidence is the World Gold Council’s account of ETF flows. Its Q2 2026 report records gold ETF outflows of about 45 tonnes, “in response to weaker gold prices and, particularly in North America, upward adjustments to both inflation and interest rate expectations alongside a strengthening US dollar.” That is an account of ETF holders reacting to rates, inflation expectations and the dollar; the Council does not describe them as weighing gold against bonds. A real yield is a nominal rate less expected inflation, so upward revisions to rate expectations push it up and upward revisions to inflation expectations push it down. The Council says both moved. It does not say the real yield rose; the Federal Reserve series above does.

The Second Channel: A Buyer Making a Different Comparison

Central banks hold gold as a reserve asset. In the Council’s 2026 survey of central banks, which drew 76 responses, gold’s performance in crises, portfolio diversification and inflation hedging were among the key factors for holding it, and gold as a geopolitical hedge and as part of reserve diversification were key reasons for increasing allocations. None of the factors the Council’s summary names is that gold out-earns a bond.

That is not the same as saying official buyers ignore rates. In the same survey 92% of respondents said “interest rate levels” were relevant to their reserve management decisions. The survey does not say whether gold is bought despite the real yield or because of it. The reading offered here is narrower: the factors the summary names for holding gold are not about out-earning a bond, so a higher real yield does not remove them the way it can remove the case for an ETF holder. That is an interpretation of the survey, not a finding in it.

On this reading the sign is positive, and the size is not small. The Council puts central-bank purchases (its shorthand for central banks and other official institutions) at 289 tonnes for Q2 2026, “a record high for a second quarter” and a recovery “to the lofty levels that have been typical in the last four years” after “a notable Q1 slowdown following a downward revision to our data.” In the part of that section anyone can read, the Council does not tie that buying to the real yield. It says the geopolitical backdrop and softer gold prices are likely to have supported the Q2 pickup, and that high prices and country-specific liquidity needs influence the timing and scale of individual transactions. Over the first half as a whole, central-bank net demand was “the lowest for a first half since 2022.” In the survey, 89% of respondents expect global central-bank gold reserves to rise over the next twelve months, and a record 45% expect their own to rise.

The same quarter shows the pairing directly. In the Council’s Q2 2026 figures ETFs sold about 45 tonnes while central banks bought 289: two groups, moving in opposite directions.

Two Signs, Two Speeds

The channels differ in more than sign.

The real-yield channel is negative and immediate. It is priced every business day, so it can be watched as it happens. On this reading its signature is gold weakening while the real yield rises, though a stronger dollar would produce the same pattern.

The official-flow channel is positive and late. Purchases are reported after the fact, and the figures move: the Council’s own Q2 report says a sizable revision cut its Q1 estimate of central-bank demand from 244 tonnes to 57. A reader sees the first channel on the day and the second in a release. On this reading its signature is gold holding while the real yield rises, followed by official purchases that are consistent with it.

That asymmetry is the discipline. When someone says central banks are holding it up, the claim at that moment is an inference from timing. The data, when they arrive, can show that the buying happened and how large it was; they cannot show by themselves that the buying is what held the price.

What Each Would Look Like

Three conditions, each observable.

The real yield rises and gold falls with it. The first channel appears to be the larger one for now, and official buying is not visibly offsetting it.

The real yield rises and gold broadly holds. Something may be offsetting the first channel, and the second is the first place to look. The check is the next release: purchases at or above the recent pace are consistent with it, and purchases that fall away while the real yield stays high would put the support in doubt. The Council’s own outlook is that central banks “remain on course for another strong year, although likely lower than 2025,” so a smaller number is not by itself a surprise.

The real yield falls and gold falls anyway. Neither channel explains it. The dollar is a common next suspect, and The Two Faces of Gold sets out why a dollar shortage and a dollar debasement are different stresses that gold reads differently.

What This Does Not Say

It does not say where gold goes. Each condition above describes the market’s state, and none is a forecast. It does not say the second channel is permanent: the Council’s outlook is for another strong year, though likely a smaller one than 2025. And it does not rank the two channels against everything else acting on gold at the same time. The Asian Stress monitor reads gold in its cyclical tier, beside copper, the Nikkei, AUD/USD and US equities, for the reasons in Why Gold Joined the Framework.

What follows from a condition, whether it matters for a given currency pair at a given horizon against everything else, is a separate question, and Reading the Transmission Chain is where this site takes it up.