Real yields, the Fed, and gold's case into Q4

The historical inverse relationship between gold and ten-year inflation-protected yields weakened materially from 2022 and has stubbornly declined to re-establish itself. Gold has appreciated through periods when the model implied it should have fallen.

The explanation is compositional. The correlation was built on a market dominated by Western investment flows. Official-sector buyers who are indifferent to carry now set the marginal price in many quarters.

It is worth being precise about what 'indifferent to carry' means in practice: a reserve manager diversifying away from dollar-denominated assets for strategic reasons is not comparing gold's zero yield against a Treasury yield in the way a hedge fund would. That buyer's decision function simply does not include the variable the old correlation depended on, which is why a model built on that variable has stopped working, not because the model was wrong before, but because the composition of the buyer base has changed.

If policy easing arrives on the timeline the market currently prices, real yields compress and the traditional relationship works in gold's favour again, with official demand as an additional support rather than the sole one.

The risk case is an inflation reacceleration that delays easing while investment flows stay absent. That combination produced the second-quarter consolidation and could repeat.

A further risk, less discussed, is that official-sector demand itself is not guaranteed to remain indifferent to price forever. If gold's price continues to rise faster than reserve managers' allocation targets require, some buyers may simply pause, as the Bank of Korea's own late entry into a domestic programme illustrates the opposite dynamic among newer buyers. A pause among the largest existing holders would remove the very support that has been offsetting the weak investment-flow leg.

Treat official-sector demand as the floor and the rate path as the accelerator. A portfolio sized on the floor and not on the accelerator survives a delayed easing cycle without being forced out of the position.

The cleanest test of whether the old correlation is truly dead, rather than merely dormant, would be a period of clearly rising real yields alongside continued, undiminished official-sector buying. If gold held up through that combination, it would confirm the compositional explanation. A renewed sharp fall under rising real yields would suggest the relationship is still latent and simply requires a large enough shock to reassert itself.