I. The Dare
REPORTED On Tuesday, September 8, speaking at Southern Methodist University, Treasury Secretary Scott Bessent said: "I am the house now, so when we intervene with the Japanese yen, I have pretty good insight into what the Bank of Japan is going to do, what Japanese policymakers are going to do." Bloomberg and Fortune reported it as a challenge to traders to bet against him.
DOCUMENTED The bond market that day: the 10-year Treasury closed at 4.80 percent and the 30-year at 5.25 percent. Twenty days later, on September 28, the 10-year closed at 5.24 percent and the 30-year at 5.56 percent. The 10-year first closed at or above 5.00 percent on September 15.
The Old Goat has covered this Treasury Secretary before, in The Bessent Pattern. This dispatch takes a narrower question: what did Treasury actually do, and what did the market do in response?
II. What Treasury Did
DOCUMENTED The yen. Japan's Ministry of Finance said in an August 3 statement that it bought yen on July 31 (U.S. Eastern Time) in coordination with the U.S. Department of the Treasury, under a September 2025 joint statement, and that it "will not hesitate to conduct further joint intervention." The statement gives no amounts. The size of the U.S. side has not been disclosed. On September 15, Bloomberg reported Bessent calling the U.S. share "nominal." The Old Goat treats any dollar figure for the American operation as unconfirmed.
DOCUMENTED FRED's dollar-yen noon buying rate was 159.16 on July 31, 156.96 on August 3, 158.90 on August 10, 154.24 on September 8, and 157.18 on September 25. The yen strengthened after the joint action and then gave much of it back.
DOCUMENTED The buybacks. On August 19 Treasury announced that its buyback operations in the 10-to-20-year and 20-to-30-year sectors would double from a maximum of $2 billion to at least $4 billion each, effective September 9 through November 4, to provide "liquidity support." Treasury's own data shows what happened next:
| Date | Sector | Max size | Accepted | Offered | 30-yr yield change |
|---|---|---|---|---|---|
| Aug 11 | 10-20yr | $2.0B | $2.00B | $7.40B | -1bp |
| Aug 18 | 20-30yr | $2.0B | $2.00B | $19.87B | -3bp |
| Aug 19 (announcement) | -- | -- | -- | -- | -9bp |
| Sept 10 | 10-20yr | $6.0B | $5.19B | $10.49B | +9bp |
| Sept 24 | 20-30yr | $6.0B | $4.08B | $10.47B | +7bp |
The ceiling reached $6 billion, triple the original $2 billion, and Treasury accepted less than the maximum on both of the largest operations even though offers exceeded it. Treasury's data does not say why. Buybacks are paid for with cash raised from other borrowing, which Krugman's earlier column likened to paying down part of a mortgage by running up a credit card balance. REPORTED
III. What The Market Did
DOCUMENTED The daily change in the 30-year yield since March 1 has a standard deviation of about 4 basis points. The August 19 announcement day, down 9 basis points, was a move of a bit over two standard deviations in the direction Treasury wanted. By August 21 the 30-year had given back 8 of those 9 basis points, closing at 5.27 percent against 5.28 percent the day before the announcement. On September 10, the day of the $6 billion 10-to-20-year operation, the 10-year rose 12 basis points and the 30-year 9. On September 24, the day of the $6 billion 20-to-30-year operation, both rose 7.
INFERENCE That is not proof that the buybacks failed. Yields move for many reasons on any given day, three operations are a small sample, and no one can say where yields would have gone without them. What the record does show is narrower and harder to dispute: the announcement's one-day gain was gone within two sessions, and the two largest operations were followed by yields rising, not falling. The buyback program has not, so far, been visible in the yield data.
IV. What Actually Moved
Krugman's primer lays out four candidate explanations: the Iran War's energy-driven inflation pushing central banks to raise short rates; an AI-driven investment boom; a glut of Treasury supply; and what he calls the "debasement trade," where investors demand a premium against eventual default or inflating away the debt. He says his own guess is mainly the AI boom, reinforced by the war, and promises a fuller argument next week. REPORTED
Here is what the public data says about each, and it is the part the Old Goat can add.
- Inflation expectations did not do it. DOCUMENTED From February 27, the last close before the war, to September 28, the nominal 10-year yield rose 127 basis points, from 3.97 to 5.24 percent. The inflation-protected (real) 10-year yield rose 118 basis points, from 1.72 to 2.90 percent. The market's 10-year breakeven inflation rate rose only 9 basis points, from 2.25 to 2.34 percent. Roughly 93 percent of the rise is in the real yield. Investors are demanding more return above inflation, not pricing in more inflation.
- The Fed moved. DOCUMENTED The Federal Open Market Committee voted 12-0 on September 16 to raise its target range a quarter point, to 3.75 to 4.00 percent, effective September 17, and said inflation "remains elevated." The effective fed funds rate had sat near 3.63 percent from December until then. The 2-year Treasury yield rose from 3.38 to 4.92 percent between February 27 and September 28. That is the market pricing a higher path for policy well beyond one hike.
- No credit stress. DOCUMENTED The investment-grade corporate bond spread over Treasuries was 0.85 percentage points on February 27 and 0.83 on September 28. Krugman treats a narrow spread as consistent with an oversupply-of-Treasuries story. It also means corporate lenders see no new default risk.
- It is not only American. DOCUMENTED OECD monthly averages on FRED show the U.K. 10-year rising from 4.43 percent in February to 4.99 percent in August and Germany's from 2.75 to 3.18 percent. Krugman argues cross-border capital flows can carry a U.S. investment boom into foreign bond markets. REPORTED
INFERENCE A move that is mostly real yield, alongside a Fed hike, a repricing of the 2-year, and stable breakevens, fits a story about the price of capital rather than a story about inflation fear. It fits Krugman's AI-capex explanation and it fits a heavy-supply explanation. It does not by itself separate them. The Old Goat has not verified the AI-spending projection that carries his lean, which comes from a paper by Stijn van Nieuwerburgh cited in the primer, and treats that part as his argument, not settled fact.
One figure to handle carefully: the primer says federal debt has passed 100 percent of GDP for the first time since World War II. FRED's series for debt held by the public shows 98.7 percent as of the first quarter of 2026, its latest reading. The claim may reflect newer data or a different measure, and the Old Goat has not confirmed it.
V. Two Corrections To The Record
DOCUMENTED An earlier Krugman column said rates had not been this high since "the fading days of the dotcom bubble." FRED says otherwise. The 10-year at 5.24 percent is the highest close since June 12, 2007 (5.26 percent), and the 30-year at 5.56 percent is the highest since June 14, 2004 (5.58 percent). The primer's later phrase, "the peak of the 2000s housing bubble," is the accurate one for the 10-year.
The Old Goat has also removed from this site earlier, unsourced claims about Secretary Bessent's personal bond trading, for which no primary source was ever found. This dispatch relies only on published Treasury, Federal Reserve, Japanese Ministry of Finance and FRED data.
VI. What To Watch
- Krugman's Part II. It should either firm up or complicate the AI-boom lean. This dispatch will be revisited then.
- Treasury's refunding statement and the November 4 end of the enlarged buyback window. Whether Treasury extends the program is the plainest signal of how it judges the results.
- The size of the U.S. yen operation. The Exchange Stabilization Fund's reports will eventually show it.
- The real 10-year yield. It is the number that moved. If it keeps climbing while breakevens sit still, inflation is not the story.