The previous part covered what tokenisation moves toward. This one covers what TradFi is moving away from. The question is why a regulated institution would carry the cost of building, supervising, and capitalising a new ledger when the existing rails work, more or less, for the existing book. The honest answer is that the existing rails do not work as well as the institutions running them would like. Four pain points are doing most of the work in the current investment case.
Working capital trapped in settlement floats
The standard US equity trade settled on T+3, then T+2, then on 28 May 2024 moved to T+1 across NYSE- and Nasdaq-listed names (SEC press release). Each step compresses the window in which the buyer's cash and the seller's securities are both committed but neither has changed hands at the legal level. That window is funded somewhere, usually with a combination of clearing-house margin and a clearing member's balance sheet. The DTCC's six-month look-back put the post-T+1 reduction in National Securities Clearing Corporation margin at around 41% (DTCC report).
The same arithmetic runs across the T+2 cash leg of bond, derivative, and FX trades. The funding cost of the float on the existing rail is some number of basis points; the equivalent on a tokenised rail with atomic delivery-versus-payment is, in the limit, zero. That delta is why a clearing house, a CSD (central securities depository), or a major dealer will commit ops, tech, and risk staff to a tokenisation programme even before any new revenue is on the table. Chapter IV on settlement finality explains why the legal layer has to keep up before the operational saving is real.
Manual reconciliation across siloed ledgers
Banks reconcile. They reconcile their general ledger against their core banking system, against the CSD, against custodians, against clearing houses, against correspondent banks, against the prime broker. Each reconciliation pays a team of people to spot, investigate, and resolve breaks. The cost shows up in the operations line, the operational-risk capital charge under Basel, and the latency it adds to every customer-facing process.
A tokenised instrument on a shared ledger does not, by itself, eliminate reconciliation. The two-ledger model described in Chapter I, Part 2 keeps the bank's general ledger and the chain in parallel, and the reconciliation surface between them is where most of the operational risk now lives. What it does change is the cardinality. Instead of a fan of bilateral reconciliations to multiple counterparties, an institution reconciles to one shared record visible across the consortium. A ledger like Partior or Project Ensemble is, operationally, a structural reduction in reconciliation overhead even before any cash leg is tokenised.
Cross-border payments still routed through correspondent banking
Cross-border payments remain one of the stubbornest parts of the existing rail. SWIFT messages settle through chains of correspondent banks, each adding latency, fees, and a reconciliation break. SWIFT's own gpi (global payments innovation) tracking shows median end-to-end times well under an hour for high-quality corridors, but a long tail still settles in days for non-G10 currency pairs (SWIFT gpi observatory). The G20 cross-border roadmap, coordinated through the Financial Stability Board, targets 75% of cross-border payments settled under one hour by 2027 (FSB roadmap). Doable for well-served corridors, less so for the rest.
The tokenised alternative, illustrated by Partior, mBridge, and Project Agorá, routes the cash leg directly between bank ledgers (or central-bank ledgers) without the correspondent chain. Partior clears multi-bank, multi-currency tokenised settlements in roughly 120 seconds. mBridge graduated from BIS Innovation Hub coordination in 2024 and is now operated by participating central banks. Each is a credible rebuttal to the "this is all marketing" critique that the SWIFT gpi numbers, on their own, license.
Fragmented post-trade infrastructure adds operational risk
The post-trade stack at a major bank is a museum. Different asset classes route through different clearing houses and CSDs, each with its own messaging standard, settlement cycle, and bilateral connectivity. Custody, transfer-agent, and corporate-actions data flow through separate systems that batch overnight. The CPMI's 2024 tokenisation taxonomy frames the institutional case for shared ledgers as, in large part, a case for collapsing this fragmentation.
Group 1a tokenised assets under Basel SCO60 carry a 2.5% capital add-on for infrastructure risk, which sounds expensive until it is set against what the legacy stack already costs implicitly. The arithmetic is not always favourable to tokenisation, but it is now actually possible to run, which it was not before SCO60.
The pain points above are why a CFO will sign the cheque. Part 3 takes the architectural fork that follows once the cheque is signed: permissioned chains, permissionless chains, and the hybrid topology that most institutional programmes end up running.