Tokenised Money Market Funds: The Coupon is Proven, the Redemption Promise is Not

The front end is where these funds live, and the distinction matters more than it sounds. A tokenised money market fund holds Treasury bills and collateralised repo. Its income follows the policy rate, not the term premium. Thirty-year yields have dominated the summer headlines, and Japan's 30-year clearing 4% was the more dramatic story. Neither has much bearing on this segment.

What matters is the Fed. Rates have sat at 3.50% to 3.75% since July, when the committee held on a nine to three vote and all three dissents argued for an increase. August payrolls then printed 162,000, close to three times consensus, with unemployment steady at 4.1% and participation rising. Energy costs have stayed elevated, tied to slower supply chain normalisation around the Strait of Hormuz. Futures now price roughly 62% to 65% odds of a 25 basis point hike this month.

For this segment a hike is not a threat. It is revenue. The wrapper's commercial argument against a non-yield-bearing stablecoin is that the coupon rises with policy, and that argument gets stronger in September.

Why it matters for institutional players

Pass-through is already working, and it splits sharply by currency. On 6 September the dollar funds returned between roughly 3% and 3.6% on a 30-day basis, with the largest clustered at the top of that band. Franklin Templeton's BENJI sat at 3.57%, Ondo's USDY at 3.49%, WisdomTree's WTGXX at 3.46%, BlackRock's BUIDL at 3.44%, Spiko's USTBL at 3.39%. Those figures are as published; wrappers differ (registered MMF versus note or SPV), so the comparison is directional rather than like-for-like.

Euro product is a different business entirely. Spiko's EU T-Bills fund returned 2.18%. OpenTrade's EU bonds vault returned 1.98%. The same issuer, running a comparable structure, delivers roughly 120 basis points less in euro than in dollars. A Fed hike widens that spread unless the ECB moves with it. It also pulls corporate treasury demand towards dollar product at the moment Europe is trying to build a euro-denominated on-chain cash market.

The size gap is starker still. All non-US government debt tokenised together comes to $1.16bn, roughly one fourteenth of the dollar market. Spiko holds about three quarters of it, and its money market range passes $1bn across its dollar, euro and sterling T-bill funds, excluding its overnight swap product. Europe's on-chain cash market is close to being one firm.

Scale is concentrated at the product level. Tokenised US Treasury debt stands at $15.86bn, and only five products hold more than $1bn. BlackRock's BUIDL, tokenised and administered by Securitize, is the largest at $2.9bn, followed by Ondo at $2.7bn, Circle's USYC at $2.6bn, Franklin Templeton's BENJI at $2.5bn and WisdomTree's WTGXX at $1.2bn. Together they are roughly three quarters of the segment.

Holder concentration is the more revealing number, and it is worse. Our reading of on-chain balances for BUIDL, across all ten networks the fund trades on, finds 92 distinct addresses holding $2.89bn. Three of them hold 53%. Five hold 72%. Ten hold 89%. Fifty-four of the 92 hold less than $1m each and are, in balance sheet terms, noise. These are on-chain addresses, not beneficial owners; omnibus and venue wallets inflate the appearance of concentration. The Ethereum holder count has also fallen from 66 to 59 over 90 days while assets did not decline, so the visible base is narrowing rather than broadening.

The Big Whale's take

Most discussion of this segment treats round-the-clock redemption as a feature. It is better understood as an unpriced liability.

The ECB stated the mechanics plainly in April. Tokenised shares "may be redeemable instantly and 24/7", while the underlying fund "can only process subscriptions and redemptions at fixed cut-off times and with delays", a discrepancy increasing the risk of runs. Requests that accumulate overnight or across a weekend have to be met when the bill market reopens.

The BIS added the part with no precedent in traditional money funds. On-chain, redemptions are visible to every participant in real time. A queue is therefore public while it forms, which inverts the informational asymmetry that has historically slowed runs in regulated funds. An investor in Reserve Primary in 2008 learned about the outflows afterwards. Here they watch them.

The wrapper decides what else a holder can see. BUIDL holds a constant $1.00 net asset value and pays yield in additional tokens, so no price signal ever appears and the queue is the only evidence of stress. Accumulating products, where the token price itself rises, would show strain as a discount to net asset value. The largest fund in the segment has the least informative failure mode.

Combine that with the holder data and the scenario stops being abstract. A run here does not need a crowd. It needs three treasurers, and every other holder can watch the first one leave. The buffers behind the promise are thin by comparison. BlackRock's out-of-hours liquidity facility with Ethena was sized at $100m against a fund now holding $2.9bn, which is under 4% of assets. Fidelity's provision for FILQ runs from 21:00 to 01:00 UK time plus weekends, a defined window rather than a continuous order book. Those figures refer to disclosed lines; other bilateral capacity may exist.

Neither the ECB nor the BIS documents a single stress episode in the segment. That is the finding rather than the reassurance. A policy surprise on a Friday, a weekend of redemption requests from a handful of concentrated holders, a closed bill market and a publicly visible queue is a sequence that has never been tested.

None of this is an argument against the instrument. The yield is real and the pass-through is clean. It is an argument for two questions before sizing a position. What is the out-of-hours facility, in dollars, and who provides it. And how many other holders does the fund have, because in a fund with 92 addresses that question has an answer.