The previous four parts established the vocabulary and the structural logic. This last part grounds it in the capital arithmetic. The headline numbers (30-50% collateral funding reduction; the 56% intraday-repo cost reduction figure from the Kinexys Digital Assets stack) come out of one mechanism: collateral funded at end of day under a T+1 cycle becomes returnable intraday under real-time margin recalculation, and the reduction in average funded balance flows directly into balance-sheet cost.

The worked example

Take a clearing member running a futures book at a major derivatives clearinghouse. The member posts initial margin to the CCP against open positions, plus variation margin called daily to mark positions to market. Margin requirements are calculated end-of-day under the standard cycle, with intraday calls only in stress.

In the standard T+1 cycle, at the end of trade date T the CCP computes the day's variation margin and the next day's initial margin requirement, and issues margin calls. The member funds the calls overnight through its repo desk, posting eligible collateral (cash, treasuries, in some venues tokenised MMFs under the CFTC tokenised-collateral guidance) to the CCP's accounts. That collateral remains posted through the entire next trading day, regardless of whether the position reverses or unwinds during T+1. Even if the trader closes the position at 10am on T+1, the collateral does not return until the next end-of-day cycle.

The funding cost is the overnight repo rate (or the opportunity cost of cash that would otherwise be deployed) for the full holding period. For a position open for one day, the member funds collateral for two business days because the end-of-day cycle does not return collateral until after the position closes. For a position that turns over multiple times in a day, the member funds against each leg under the worst-case end-of-day snapshot, even if the net position is much smaller.

Now run the same book against a tokenised rail with real-time margin recalculation. The CCP recomputes margin continuously against the live position. As the position unwinds during T+1, the margin requirement falls; as it falls, excess collateral returns to the member intraday rather than waiting for end of day. The funding cost is now the integral of the actual margin requirement over time, not the worst-case snapshot.

For a typical mixed book, the average funded collateral balance falls 30-50% under real-time recalculation. The saving depends on how concentrated the day's positions are around peak risk: a mostly-intraday book with a small overnight residual sees the largest saving; a mostly-overnight book sees the smallest. The 30-50% range tracks the difference between funding the worst-case end-of-day snapshot and funding the time-weighted live exposure.

Where the 56% figure comes from

JPMorgan has published, in client case studies for Kinexys Digital Assets and the Tokenized Collateral Network, a 56% reduction in intraday repo financing cost relative to the equivalent overnight repo workflow. The number traces to the same mechanism. In conventional overnight repo, cash is borrowed end-of-day, held overnight, and repaid the next morning, with the funding rate applied to the full overnight period. In intraday repo over a tokenised rail, cash is borrowed at the moment it is needed, held only for the hours it is deployed, and repaid against atomic return of the collateral. The funding rate applies only to the hours the cash is in use.

The 56% saving is the blended midpoint across JPMorgan's client base. For clients whose intraday liquidity needs cluster in a few hours, the saving relative to overnight is closer to 80-90%. For clients whose use approaches a full business day, the saving compresses toward 20-30%. The headline figure is consistent with the broader 30-50% collateral funding reduction observation: same mechanism, different presentation.

The capital implications stack. Lower funded collateral balance means lower funding cost on the income statement. It also means lower balance-sheet utilisation, which under Basel III leverage ratio rules releases leverage capacity that can be deployed against other client business. For a balance-sheet-constrained clearing member, the leverage capacity release can be a larger economic effect than the direct funding saving, particularly at quarter-end when leverage ratios bind.

What this means for treasury PMs

The practical implication is that the operational unlock is structural, not incremental. The pre-tokenisation answer to clearing-collateral cost was to optimise the overnight funding rate, the haircut economics, and the collateral substitution policy. Each is a real lever, but each operates within the constraint of an end-of-day cycle that determines when collateral can move. A tokenised rail with atomic intraday settlement removes the constraint. The collateral can move in seconds, in either direction, against an atomic cash leg, on demand.

For a treasury function evaluating a switch, the relevant questions are: how much of the current funded collateral balance is structurally end-of-day driven versus position-driven; what fraction of the saving accrues to the institution versus the client; and what operational changes are required to integrate the tokenised rail into existing collateral, margin, and treasury systems. The answers are programme-specific. The mechanism is the same one this chapter has built up: atomic Model 1 DvP at intraday speed compresses the funding cost of collateral, the funding cost of intraday liquidity, and the cost of the post-trade workflow that wraps both.

The settlement primitives that make this possible are not novel inventions. They are the BIS Model 1 DvP framework from 1992 and the post-Herstatt PvP framework from 2002, implemented on infrastructure that compresses the historical funding penalty of atomic settlement to a level the original frameworks could not match. That is the chapter's load-bearing point: tokenisation does not invent new settlement primitives, it makes the regulator-recognised primitives operationally affordable. Operators who can articulate that to a counterparty are speaking the language the counterparty already uses. Operators who lead with "real-time" and "atomic" interchangeably are not.