Gold clears $4,180 as the term premium trade finds a second expression

The textbook version of gold says it falls when real yields rise, because bullion pays no coupon and a higher real rate raises the cost of holding it. That relationship has been reliable enough to build models on for most of the past fifteen years. It has now been wrong for three weeks running. Ten-year real yields are up 29 basis points since August 21. Gold is up 9.7 per cent over the same stretch and closed Friday at $4,183 an ounce.

The model is not broken so much as incomplete. A real yield can rise for two reasons, and gold responds to them differently. If the real yield rises because expected growth has improved, gold falls, because the alternative asset has become more attractive on its merits. If the real yield rises because investors demand more compensation to hold duration, gold rises, because the thing being repriced is confidence in long-dated claims and bullion is the only major asset with no issuer standing behind it.

Every indicator available says the second case applies. The Bellweather Macro term premium estimate has moved from minus 12 basis points in June to plus 61 now. Five-year to ten-year forward inflation compensation has risen 27 basis points over the same period, faster than the five-year measure, which is the pattern associated with doubt about the destination rather than about the next few prints. Gold is not disagreeing with the bond market. It is agreeing with the part of the bond market that has been selling.

The equity row is the useful comparison. Stocks carry a correlation of minus 0.52 to the ten-year yield over the same window, meaning they fall as long yields rise, which is exactly what a portfolio holding both Treasuries and equities as a diversified pair does not want. Gold at plus 0.41 is doing the job the bond allocation was supposed to do, and it has been doing it since July.

We stopped treating duration as the hedge for the equity book in July, and the allocation went to cash and gold. Nothing since has argued for reversing it.

Theodora Vance, portfolio manager at Kestrel Asset Management

Two things would end this. The first is a thirty-year auction on Thursday that clears comfortably, which would cap the long end, compress term premium and remove the condition gold is responding to. The second is a services inflation print soft enough to restore the front end rally without the long end following it, which would look to bullion like a growth improvement rather than a compensation shock. Neither is the base case on most desks, which is why the positioning data shows net length still building.

The risk in that is ordinary and worth stating. A trade that has worked for three weeks attracts momentum money, and momentum money leaves faster than it arrives. Gold's realised volatility has risen from 11.4 to 16.8 annualised since August, the futures curve has moved into modest backwardation at the front, and both are signs of a crowded position rather than a structural one. The central bank bid is real. The trend followers sitting on top of it are not part of the thesis.