Those Who Came to Set the Price

This edition follows the sequence of events. What happened over four days is its structure.

Start with what moved in opposite directions on the same day: time zones created the asymmetry. Because of the Earth's rotation, Asia on 8/19 traded the shock (19-year-high yields and a lost diplomatic route), while a few hours later, the US traded the remedy (Treasury intervention). Asia's 8/20 V-shaped reversal priced in that remedy.

#13 concluded: this is the world's most important price, yet nobody officially offers to set it. On August 19, the first arrival stepped into that place. The US Treasury doubled long-bond buybacks per operation from USD 2 billion to at least USD 4 billion. The sectors are 10–20 years and 20–30 years; the period is September 9 to November 4. The stated purpose is supporting long-bond liquidity.

Buybacks Start with what that term means. The Treasury repurchases its own bonds circulating in the market. It looks like central-bank quantitative easing (QE), but is something quite different. ⑴The buyer is the Treasury, not the Fed. ⑵The money is raised by issuing new short-term Treasuries rather than printing it. ⑶Money and total Treasury debt therefore stay unchanged. Only the maturity mix changes: retire long bonds and issue short ones. This policy does not expand money; it replumbs the Treasury reservoir.

The distinction matters because an intervention that leaves totals unchanged leaves a different fingerprint.#13 split yields into real yields (money's true rental cost after inflation) and breakeven inflation (BEI, market inflation expectations inferred from inflation-linked bond prices). An intervention using supply and demand to suppress the long-bond premium (term premium: extra yield for longer maturity) should lower real yields alone, leaving BEI still. An intervention that expands money moves BEI.

We observed the former fingerprint.

On the announcement day, 8/19, the entire decline was in real yields (−9bp). BEI stayed at 2.30. Inflation expectations did not fall; supply and demand suppressed the term premium. The intervention left a textbook fingerprint.

Two things make this intervention unusual. First, it has a time limit. The window closes on November 4. That date is not a coincidence: the day after the US midterms (11/3) and the next quarterly refunding announcement. It is a logical date in the quarterly schedule, but also concentrates firepower just before the election and decides on extension the following day. The assessment date is therefore not the start, but the end date. Where long-term yields go after the window closes answers #13's question (gauge ③ tracks two D-counts).

Second, the Fed is on the other side. Two interpretations surround Chair Warsh, who took office in May (Senate confirmation 54 to 45, the most divided ever). Reports suggest the Fed could tighten more aggressively to offset Treasury easing. Another reading is that Warsh dislikes Fed intervention itself, so he will leave the strongest tools unused, with the Treasury fighting alone. These interpretations point in different directions, so we place them side by side. Both share one fact: the Fed has much stronger tools—Operation Twist, selling short bonds to buy long ones, and yield curve control (YCC), capping long-term yields—but is holding them unused.

Back to §01: why did Korea alone fall −5.80%? A side-by-side view of Asia on 8/19 sharpens the question.

The fear index, VIX, started at 15 and fell further. This was not global risk aversion. Korea also split: KOSPI −5.80% versus KOSDAQ −1.17%. The epicenter was large exporters (Samsung Electronics −7.8%; SK hynix −9.8%). Something selectively pressed Korea's large exporters.

The suspect is the exchange rate. Dollars earned by the memory boom produced a current account surplus (the country's earnings from trade and income), totaling over 12 months USD 266.2 billion, or 13–14% of Korean GDP. June alone contributed USD 49.7 billion, and it was accelerating. Converted into won, those dollars lifted the currency. The chart measures how far, against more than just the dollar:

If this were dollar weakness, the won would rise only against the dollar. Instead, it has risen against the euro, yen and yuan too, by around 10% against all of them (4-pair average +10.7%, maximum–minimum spread 1.7pp). The timing overlaps: the won's weakest day was 6/5 (KRW 1,559.5 per dollar), while KOSPI's record high was 6/22 (9,115). The stock market turned down just after the won turned.

Why does this hurt large exporters? Samsung and SK hynix earn dollars but report in won. Samsung's Q2 operating profit of KRW 89.4 trillion was earned at exchange rates around KRW 1,500 per dollar. The same dollar earnings are 7–8% smaller in won around KRW 1,390. The boom earns dollars; dollars lift the won; the stronger won reprices the stocks that created the boom: a boom weakening itself. It is the mechanism of Dutch disease, in which resource exports strengthen a currency and hurt manufacturing. Korea's resource happens to be memory. The concentration of damage in large-exporter KOSPI rather than domestic-focused KOSDAQ, and the won's distinct rise, fit this picture.

The next announcement came from Seoul. After the 8/19 close, on the evening KOSPI fell −5.80%, SK hynix filed: KRW 40.00434 trillion, 24.07 million shares (3.3% of shares outstanding), to be bought over 3 months from 8/20 and canceled in full. The largest ever among Korean listed companies, with a pledge to return at least 50% of cumulative 2025–2027 free cash flow (FCF: cash earned minus investment, which the company can actually dispose of). The next day, reports said Samsung Electronics would convene its board in August to approve more than KRW 100 trillion in shareholder returns, including a special cash dividend.★Samsung's case is still a report, not a filing. Until a DART resolution appears, this edition treats it as unconfirmed (gauge ① watches).

KRW 40 trillion and KRW 100 trillion. It is easy to admire the size. The next question is how much these amounts can actually reach share prices. Use three rulers: market capitalization, cash generation and actual market purchases.

Start with cancellation. Trading flows are modest: spread over 3 months, purchases are KRW 630 billion daily, 7% of SK hynix's daily trading value. That cannot mechanically produce a +12.73% daily rise. The per-share ceiling is clear too. Canceling 3.3% of shares raises remaining holders' share by 1/(1−0.033), or+3.4%, the full arithmetic effect. Yet the shares moved+12.73% the day after the announcement, four times the mechanical effect. That gap is the announcement's real content. The market bought not KRW 40 trillion of order flow, but the pledge to return at least 50% of cumulative 2025–2027 FCF. If annual FCF stays at KRW 26 trillion, that exceeds KRW 39 trillion over 3 years. The KRW 40 trillion is its first execution. The repricing concerned the shareholder-return regime.

The dividend ruler is different. A dividend buys no shares. After the record date, the price loses the payout (ex-dividend). KRW 100 trillion merely transfers 6.3% of market capitalization to shareholders' accounts; it does not by itself raise per-share value. Still, the figure matters for two reasons. First, the funding size: 2.6 years of annual FCF cannot be a recurring payment. It must be one-off, hence "special." Second, where the cash goes. Cancellation gives the controlling holder no cash, only +0.68pp of ownership. Dividends send cash up the ownership ladder. The two companies chose forms that fit a publicly visible need in its governance structure: a property settlement's cash demand on one side, succession on the other. This is observation, not a causal claim that those needs determined the choice.

Three assets outside center stage sent signals in the same week. All are observations from this week.

① Oil — Calm prices, shrinking buffers. The Hormuz blockade is in its 6th month. Daily transits have fallen from 110 vessels to 7 (−94%) yet Brent trades in the USD 92s, well below March's USD 126 peak. Bypass pipelines, including the Suez pipeline, carry 2.4 million barrels daily. Calm prices do not mean calm conditions. Divide the global commercial-inventory buffer above its operating floor by the depletion rate to get the runway, which in the central scenario is 67 days (2.2 months), right on the attrition scenario's 2-month defense line (optimistic 135 days, pessimistic 31: scenarios, not forecasts). Bypass infrastructure leads toward the Mediterranean, so Asia-bound bypasses run at only 19% of nominal capacity. Average prices look calm, but Asia's volumes are thin. This directly affects Korea and Japan.

② Crypto — Higher prices, not new money. In the 8/19 US rally, bitcoin gained +7.9% and ether +19.3%. Total crypto market capitalization rose by USD +113 billion in a day, while stablecoin supply, the inflow fuel, changed by USD +360 million: a 300x gap. This suggests new money did not drive the rise. Of USD 1.91 billion liquidated over 24 hours that day (forced closure of leveraged positions), 91% were short bets. Rising prices forced short sellers to buy back, amplifying the rise: a short squeeze. This is a one-off mechanism. Continued gains need a turn in stablecoin inflows. Context: crypto fell −53% from last October's peak and bottomed on 7/1. It is now a +18% rebound from the low, not a recovery (−22% year to date).

③ Japanese households — Quietly moving from deposits to government bonds. People opened every sluice gate in this edition: the Treasury and corporate boards. Japanese households are opening theirs without a resolution. In monthly retail government-bond issuance, fixed 5-year bonds that lock in rates have led floating 10-year bonds for 17 consecutive months (since April 2025). This fiscal year's first 5 months already issued JPY 4.1 trillion, an annualized JPY 9.8 trillion, versus last year's JPY 6.2 trillion:+59%. As 30-year JGB yields passed 4% and deposits lagged, household savings began moving on their own. Tens of millions of small sluice gates need no board: the interest rate is the handle.

This week, issuer intervention moved US Treasuries; exchange rates and return announcements moved Korean stocks; bypass routes and inventories moved oil; liquidation moved crypto; and interest rates themselves moved Japanese households. In all five, prices followed where the water actually flowed rather than forecasts.

The picture comes together by retracing the Treasury secretary's moves before buybacks. ⑴He invoked Iran to press down Oil prices and lower rates (verbal intervention). ⑵He hinted at buying the yen (exchange rates). When those did not work, ⑶he turned to Buybacks as a physical measure. Oil, exchange rates and interest rates—the three prices examined separately in §02, §03 and §05—are already treated as one plumbing system at the policy level.

Four observations remain:

⑴ Does the fingerprint persist? Only real yields have moved so far. If BEI starts moving, intervention has reached inflation expectations and changed character. That is precisely gauge ④'s trigger condition.

⑵ The day after the window closes. November 4, the day after the election and the next issuance-plan announcement. A time-limited intervention is assessed at its end. If long yields return after the window closes, intervention has delayed the price rather than set it.

⑶ A paradox — intervention may hasten yen-carry unwinding. The US–Japan 10-year yield gap has narrowed to 176bp (trigger line 150bp). Treasury suppression of US long-term yields narrows it further. The narrower the gap, the less remains from borrowing cheap yen to buy US assets. Conditions ripen for Japanese money to come home: repatriation. A US remedy may hasten a reversal from Japan.

⑷ A new competitor for the reservoir. More than USD 40 trillion in federal debt is not the only force pushing long yields higher. AI hyperscalers' long-term corporate bonds have begun competing with Treasuries for the same long-duration demand. Fiscal funding and AI capital expenditure draw from the same reservoir. This leads directly to the next edition (§08).

This table is the gauge panel above the title. Like #13, it is a live widget that updates with new observations after publication. Publication values remain in this table. Event needles (①②③) cannot be classified automatically and carry human-verification dates. Only ③'s D-count is recalculated daily.

Two earlier needles were retired. The mandatory treasury-share cancellation D-count was absorbed into ② (SK hynix KRW 40 trillion); the 2027-09-06 deadline itself remains valid. The Insurance Business Act/Samsung Life needle had low information value and was removed. Oil runway and ERP cracks are out of this edition: oil remains in §05, and compute/ERP belongs to the next edition.

This four-day record closes. When #13 ended with nobody offering to set the price, we awaited bond vigilantes' footsteps. Instead came the issuers themselves: the Treasury offering to buy its own bonds, two boards their own shares. Offstage, exchange rates, bypass routes, positioning and interest rates each pushed prices.

The boldest among those coming to set prices has not appeared yet. It is in Chicago: a futures market trying to price compute. Across the Pacific is a country trying to make that price 0. Next: The Value of Compute.