Two days of price, then the give-back
The surprise arrived on Wednesday: the Treasury said it would spend double the expected amount on bond buybacks, and Scott Bessent said on Thursday the repurchases could be increased further. By Friday the decline in yields the announcement had triggered was largely unwound. Donald Trump said the same day that he had given no instruction to intervene, and that Bessent had acted on his own authority because he understood the issues.[1]
The operation itself is not large. The Treasury is doubling the size of its buybacks of longer-dated paper over the coming quarter and will spend at least 4 billion dollars per operation. Against a long end that climbed because the stock of government debt is growing and inflation is sitting above the Federal Reserve's 2 per cent target, that means a buyer at the margin. It does not change how much paper has to be placed.[1], [2]
Gold kept what the bond market gave back
Gold did not join that round trip. The spot price rose 2 per cent on Friday to 4,603 dollars, its highest since May 29, and closed a third consecutive higher week with a gain of 5 per cent. On the CME FedWatch tool, investors were putting a 67 per cent probability on rates staying unchanged next month and 33 per cent on a hike, and the dollar was heading for a weekly decline.[2]
Read together, the two prices point one way: the buyback moved the leg it was aimed at and left the other alone. A larger official buyer changes who is standing under long-dated paper on a given day. It does not change the deficit that issues the paper, nor the inflation reading that sets the compensation holders demand for carrying it. The competing account is that Friday's give-back was position-squaring in thin August trading rather than a verdict, and that the yield decline returns once the operations actually run.[2], [1]
Where the compensation question stands
Four sessions ago the question here was at what point a move in the compensation investors want for holding long-dated debt turns into a repricing of the policy path; that piece traced the 30-year yield to 5.324 per cent. This week answered part of it. The compensation leg moved on an official buyer and moved back, while the policy leg did not move at all. The observable test runs to the Federal Reserve's September meeting.[3], [2]
If the buyback is working only on the supply side, the probability of a hike should still be near 33 per cent when the committee meets and gold should still be holding its weekly gain. If that probability falls clearly and gold retreats with it, the reading that the buyback also touched inflation expectations gets stronger. The distinction is easier to follow by watching who is bidding under the paper than by watching the size of the operations.[2]