Model the correspondent line as a decaying annuity, not fixed plumbing

Float income is what a bank earns for holding value in the gap between a debit and a credit. An FX spread is what it earns for standing in a currency position. Two of the four components Capgemini folds into its $230 billion figure are of that type, and only the other two, correspondent fees and transaction processing and settlement fees, are billed as services performed ^[3][9].

That split decides how the line behaves under pressure. A fee can be repriced or bundled. Income earned from elapsed time and parked inventory cannot be defended once the time and the inventory are gone.

The mechanics described to American Banker show how much of the revenue sits on parked inventory. Aaron McPherson of AFM Consulting said the banks stage things in advance and maintain liquidity in the reciprocal accounts ^[5]. Reconciliation is periodic, often monthly, and its stated purpose is to avoid tapping foreign exchange markets too often ^[6]. Monthly cadence is about twelve trips to the market a year ^[10], and the balances sit in place between those trips. Those balances are the asset the float income is earned on, and they exist because the settlement design assumes a gap.

The cost side does not move with the revenue side. The same reporting lists know-your-customer, anti-money-laundering and counter-terrorist-financing costs as a standing condition of participating, high enough to be prohibitive for smaller banks ^[7]. Nothing in the source suggests those obligations shrink when settlement rails change. A revenue line that thins while its compliance base holds is a margin problem, not an infrastructure problem, and it should be forecast accordingly rather than treated as the fixed cost of being connected.

Two cautions on the sizing. The $230 billion is a vendor estimate provided exclusively to the publication, and it is given as an aggregate with no per-component breakdown ^[3][11]. So the material supports a direction and not a magnitude: you can establish that more of the exposed revenue is time-and-position income than fee income, without knowing how much. American Banker also reports the system as slow and, at the same time, a large revenue opportunity for the institutions inside it ^[8]. Those two facts are the same fact seen from opposite ends of the trade.

Age is the weakest argument for durability here. Philliou traces the model to Italian merchant-banking houses such as the Medici, which used partner banks and agents in other cities to settle trade debts without physically shipping coins ^[1]. The design solved coin transport. Capgemini's Michael Levens calls the result the invisible plumbing of the global financial system ^[12], and plumbing is normally modelled as a cost centre you maintain. This one is a revenue centre several institutions have been drawing on, which is a different thing to lose.