SF Fed study finds stablecoin Treasury demand only partly offsets China's retreat

A San Francisco Fed study concludes that stablecoin issuers' growing Treasury holdings only partly offset China's retreat from the market, according to The Defiant's account of the paper, and the limit is tenor: issuers concentrate in short-term debt while the decline in Chinese holdings has landed mostly on longer-dated securities, so the two flows meet at different points on the curve.

A net figure obscures that if it is read as substitution, because demand added at the front end does not replace demand withdrawn years out, and stablecoin issuers' buying does not compete for the paper the retreat has left behind. Whether the front end needed the support is a separate question from whether the long end lost it, and the paper's arithmetic does not answer the second.

The study's second condition reaches further: additional Treasury demand from stablecoins depends on who buys the stablecoins, which turns the question from total market value to the mix of holders. A demand estimate pegged to stablecoin market capitalization assumes the next holder behaves like the existing average, but the authors' caveat makes the next holder the variable, so the Treasury demand that arrives with a growing float is not a fixed multiple of it.

Which way that mix moves is the part of the market this publication has been tracking. Licensed stablecoins are moving into bank rails, checkout and agent billing, where float grows as a byproduct of payment volume, and a reserve book that tracks transaction flow accumulates differently from one that tracks trading inventory, in which a trader's reserve is inventory that can unwind as quickly as it built. The paper does not resolve which buyer dominates; it says the answer determines whether the demand persists.

Stablecoin reserves and tokenized Treasury funds draw on the same short end of the curve, putting them in competition for collateral that the $1.7 trillion collateral-mobility forecast around tokenization assumes will keep moving, while Dune's 0.006% monthly turnover for tokenized assets is the counterweight because issuance has run ahead of circulation. The Fed study lands on the same seam: issuer demand is real and concentrated at the front end, and how much it grows is a question about buyers.

The offset is real where it lands and conditional beyond it, so a straight-line demand projection would be doing arithmetic the paper's own condition does not license. The thing to watch is not the size of the float but the identity of the marginal holder, which the stablecoin market is currently changing faster than it is growing.