Comment from Tokenization Systems

This is a supplemental comment to the comment Tokenization Systems filed on this joint FinCEN-OFAC docket on April 18, 2026, which is on the public record as comment FINCEN-2026-0100-0005. It adds one focused point and does not restate the prior comment. The point concerns Recommendation 6 of our comment of record, which urged the agencies to require that Permitted Payment Stablecoin Issuer risk assessments address reserve custody relationships, including concentration risk. As filed, that recommendation rested on a single third-party descriptive source, the Federal Reserve FEDS Note by Du, Sonawane, and Watsky (DOI 10.17016/2380-7172.3958), which documents the custodial concentration channel during the March 2023 Silicon Valley Bank episode. That source shows that the channel exists; it does not establish the causal magnitude or timing of the resulting peg disruption. Since the April filing, first-party causal evidence on exactly that question has matured. A working paper produced within our research program, "Did Reserve Concentration Cause Stablecoin Fragility? Evidence from the SVB Shock," entered our record on June 8, 2026. Because it post-dates the April filing, it could not have been cited in the original comment. A within-unit event study of USDC's hourly price path documents a depeg trough of 9.77 percentage points at the tenth hour of the event window opened by the Silicon Valley Bank failure and Circle's disclosure of its exposure, measured against flat pre-event trends. The recovery breaks at the joint Federal Reserve, Treasury, and Federal Deposit Insurance Corporation backstop announcement, and that break is detected by a formal joint structural-break (Chow) test significant at the announcement hour (F = 7.41, p = 0.0012). Two placebo tests are consistent with an event-specific channel rather than generalized market stress: a non-event weekend (p = 0.322) and the Silvergate Bank collapse (p = 0.300), the latter a crypto-focused bank that held no stablecoin reserves and therefore opened no reserve-loss channel. The exposure that produced this trough was approximately 8 percent of the issuer's total reserves, a small share in percentage terms. With the reserve bank in receivership and weekend banking closed, the issuer's primary-market redemption throughput fell over the weekend to roughly seven percent of its seizure-day level, impairing the arbitrage that normally holds the token at par. Read together, these measurements carry an arithmetic the agencies can verify directly: even a complete loss of the Silicon Valley Bank exposure would have left roughly 92 cents on the dollar of reserve backing, yet at its minute-level intraday low the token traded near 89 cents, below that worst-case floor. The market was pricing not only the potential loss of the concentrated reserves but the impairment of the redemption mechanism itself; a reserve-custody concentration transmits through both channels at once. The study's cross-sectional difference-in-differences estimate is reported as a descriptive magnitude only; the causal weight rests on the within-unit event study and the structural-break test. This evidence indicates that a single-bank reserve-custody concentration shock produced a large and abrupt deviation from peg whose recovery trajectory broke at the policy backstop announcement, a path consistent with policy-driven recovery rather than market self-correction. The shock operated through the issuer-to-reserve-bank relationship, not through the issuer's direct customer relationships. That is precisely the relationship Recommendation 6 asked the agencies to bring within the special-standards-of-diligence framework. The attached PDF develops this single point in full.