The Buyer Strike That Wasn’t: Two Auctions, Record Demand, and What It Means for Your MPF Bond Fund

Core proposition: A month ago, the market whispered “buyer strike” — nobody wanted US government debt, and yields were supposedly spiralling out of control. Last week (6–8 October), the US Treasury ran three consecutive auctions, and buyers answered with record-sized bids. For MPF members, the conclusion is practical: the enemy of your bond fund is not “no buyers”, it is your own blind-priced switch during auction week.

1. The Macro Matrix: Four Numbers from Auction Week

The 7 October 10-year auction deserves a closer look. It cleared 1.7 basis points through the when-issued yield — buyers accepted a lower return than the market price to secure the bonds. The 2.77 bid-to-cover ratio was the highest of the last 50 auctions. Primary dealers were left with just 2.5%, the lowest share at a 10-year auction in records going back to 2003. The next day’s 30-year auction was solid too: 5.618%, up 31bp from the prior month, with over 93% absorbed by end investors.

Yields fell after the auctions: the 10-year from an intraday 5.368% to 5.227%, closing 13 October at 5.22%; the 30-year from 5.732% to 5.56%. Both remain 24-year highs, but auction week’s message is clear: at these prices, buyers showed up.

Policy lines up with that reading. CME data puts the odds of a 28 October Fed hike at only about 19%; consensus is for a hold. Tonight’s September CPI (8:30pm HKT, consensus 3.6% year-on-year) is the next card to fall.

2. Backtest and Win Rate: How the “Strike” Thesis Got Falsified

The “buyer strike” story began with September’s 7-year auction: USD 44bn, 5.085% (the highest since April 1993), bid-to-cover just 2.21. On that single data point, the conclusion looked plausible. Three weeks later, two harder-to-sell long-bond auctions gave the opposite answer.

The key is separating sellers from buyers. This sell-off’s sellers are forced liquidators — hedge funds deleveraging, mortgage hedgers, AI-company bond-hedging desks — not a buyer walkout. Three pieces of evidence:

1. Supply has not risen. The Treasury has not increased coupon auction sizes since April 2024, and in August committed to holding them unchanged “for at least the next several quarters”. Prices fell, but not because more paper arrived.
2. Buyers voted with real money. Asset managers’ net long position in 10-year futures reached 2.83 million contracts (about USD 283bn), the most in CFTC data back to 2006. At the 12 August 10-year auction, non-dealer buyers took USD 38.2bn of USD 42bn on offer — also the most since 2003.
3. The historical win rate favours buyers. BMO notes that in the eight instances since 2020 when a 10-year auction was nearly as strong as 7 October’s, the following 30-year auction was weak three-quarters of the time — yet this time both were strong. Even the most pessimistic historical pattern broke.

The Treasury also bought back at least USD 4bn of 20–30-year bonds on 8 October (roughly double the usual size), soaking up more long-end supply. Put together, the win-rate map is clear: betting on “auction failure” has been a losing trade for two straight weeks.

3. The Switching Blueprint: How to Position the Bond Sleeve

Auction week’s lesson is not “bond funds are safe”. It is “do not switch on headlines”. The blueprint:

1. Benchmark first, gut feel never. The DIS Core Accumulation Fund’s statutory reference portfolio holds 37% bonds. If your bond sleeve sits far below that, the problem is not “auctions” — your allocation itself has drifted from benchmark. The Age 65 Plus Fund is roughly 80% bonds: members near retirement should expect auction-week NAV noise by design.

2. Do not confuse the 30-year headline with your fund. A 30-year bond has a duration of roughly 15 years: a 100bp yield rise costs about 15% in price. But MPF bond funds are far shorter (Fidelity’s Hong Kong Bond Fund, for example, runs about 3.8 years). The same 100bp costs your fund about 3.8%, not 15%. Selling a 3.8-year-duration fund on a 30-year scare number is a units error.

3. Carry is now your cushion. A 10-year yield of 5.22% means bond-fund coupon income of roughly 5% and up. At 3.8 years’ duration, about 139bp of yield rise over the next 12 months (5.3 divided by 3.8) merely eats one year’s carry. That is the big difference from 2022, when carry was only about 2% and the cushion was far thinner.

4. The next supply test is already scheduled: November refunding week. The Treasury announces its quarterly refunding on 4 November, with auctions on 9 November (3-year), 10 November (10-year) and 12 November (30-year). Watch two numbers then: whether the dealer share returns toward the 13% average of 2024–25, and whether auction sizes grow. If neither happens, the “buyer strike” thesis can be formally filed away.

Illustrative maths: On the average balance of HKD 338,950, with a 30% bond sleeve (about HKD 101,685), a 100bp yield rise costs a 3.8-year-duration fund roughly HKD 3,864. The auction-week swing — the 10-year falling from 5.368% to 5.227% (14bp) — maps to NAV noise of only about HKD 539 on that sleeve. Auction-week noise is worth knowing about. It is not worth switching over.

4. Forward-Price Protection: Execution Discipline for Auction Weeks

MPF’s forward-pricing mechanism (T+1/T+2 NAV) makes auction weeks especially dangerous: auction results land around 1am HKT, but your switch instruction prices off a later NAV. The discipline:

• No instructions on auction days. Around the 7–8 October auctions, yields moved over 10bp in a single session. An instruction placed on auction day buys at the post-auction price, not the price you saw.
• CPI day is a no-go zone too. Tonight’s 8:30pm CPI release typically reprices the bond market within hours. Wait for the number to land and the market to digest it; do not gamble blind ahead of the data.
• Split large switches, and avoid FOMC week. The 27–28 October FOMC meeting is the next volatility window. If a switch is not urgent, wait until after the November refunding week — and still avoid auction days.
• Do not chase the post-auction bounce. The 14bp yield fall after the auctions was not a buy signal — just the normal unwind of pre-auction panic. Allocation decisions follow benchmarks and duration, not single-day moves.

Bottom line: Last week’s auctions proved that buyers queue up for 10-year paper at 5.3% and 30-year paper at 5.6%. The “buyer strike” was a good story, but the data vetoed it. What MPF members really need to guard against is not a bond market with no buyers — it is paying, at an unknown price, for somebody else’s panic during auction week.