Why
A '₩5 trillion saved' headline is the kind of number a PoC exists to take apart rather than repeat. In the National Assembly Budget Office's own model it is not a forecast — it is one cell: the top-right corner where the substitution rate is its highest (30%) and the fee spread its widest (the optimistic 1.5% merchant vs 0.1% stablecoin). The same table's opposite corner — 5% substitution, the conservative 0.6% spread — is ₩0.37T. Same formula, ~14× apart. The single defensible output of this model is the grid, and the single most-misleading way to report it is to quote the maximum. Building it makes that unavoidable to see.
The second thing the model teaches is what a fee is. The formula's last term is (merchant fee − stablecoin fee), and every won of that 'saving' was a won of someone's revenue — the card issuer's. So the number does not describe wealth created; it describes a transfer from issuers to merchants. The report is careful here in a way the headline is not: the issuer's loss is not one-to-one with net income, because a shrinking card business also sheds its own costs (points, mileage, rewards funded out of that fee). A faithful model carries both legs — the gross transfer and the partial cost-offset — instead of presenting a merchant gain as a system-wide gain.
It lands on the Jayverse economics thesis directly: a payment rail's fee is not friction to be wished away, it is an income statement. This pairs with the on-chain settlement-funding card (Visa × Credit Coop) as the other side of the same coin — that one is where stablecoin rails add a financing layer, this one is where they subtract an intermediary's fee — and both are only honest when you say whose revenue moves. The report's second half (volatility spillover) is a separate, data-science-shaped study: a Diebold-Yilmaz-style total connectedness index that sat at ~34% on average but rose to ~50% during the Feb–Mar stress window, which is its own PoC about when 'linkage is limited' stops being true.
How it works
The formula is the whole model
annual saving = card volume × substitution rate × (merchant fee − stablecoin fee)
Every output is that one line evaluated on a grid. The report fixes card volume at 2025 credit+debit ₩1,225.1T and varies the other two:
| Fee scenario | Merchant fee | Stablecoin fee | Spread |
|---|---|---|---|
| Conservative | 1.1% | 0.5% | 0.6% |
| Base | 1.3% | 0.3% | 1.0% |
| Optimistic | 1.5% | 0.1% | 1.4% |
The grid, not the headline
| Substitution → | 5% | 30% |
|---|---|---|
| Conservative (0.6%) | ₩0.37T (the floor) | ₩2.2T |
| Optimistic (1.4%) | ₩0.86T | ₩5.15T (the headline) |
The reported ₩5.15T and ₩0.37T are the two opposite corners of this surface — a ~14× spread driven entirely by which assumptions you pick. Quoting only the max is choosing the corner.
'Saving' names a transfer, and the transfer is not net
The last term is a fee, and a fee is revenue to whoever charged it. So the merchant's gain is the card issuer's loss — first-order, a transfer, not creation. But the issuer's net loss is smaller than the fee delta: a smaller card business also spends less on the points/mileage/rewards it funds out of that fee. A model that reports the merchant leg without the issuer's cost-offset overstates the system-wide effect. The channels most likely to substitute first — online checkout, cross-border remittance, platform commerce — are where that erosion would show up before general retail.
The other half: volatility spillover (a separate, data-science PoC)
Using Dec-2023→Jun-2026 data on USDT, USDC, BTC, KRW/USD, KOSPI, and the 3Y KTB rate, the report ran a dynamic connectedness (Diebold-Yilmaz-style) analysis. Findings worth modeling: stablecoin↔stablecoin linkage is high (USDT→USDC spillover ~43%), stablecoin↔FX is moderate (avg ~4%, max ~8–10%), and the total connectedness index averaged ~34% but spiked to ~49.86% during the Feb–Mar stress (BTC drop + Middle-East conflict). The takeaway is conditional: linkage looks limited in calm and rises in stress — exactly when a stability claim matters.
The four policy levers (context for any design)
Micro: (1) reserves ≥100% of issuance with cash/short-T-bill composition rules; (2) yield/rewards allowed but capped, to prevent over-issuance. Macro: (3) a 'significant stablecoin' designation (MiCA-style) carrying extra capital/liquidity, stress tests, disclosure, and issuance caps when usage is broad; (4) financial-stability measures aimed at the high-usage significant coins.
Where this lands in Jayverse
Build the sensitivity model as an economics PoC (the grid is the deliverable, not a number), pair it with the on-chain settlement-funding card as the 'fee subtracted vs financing added' two-sider, and route the spillover half to the data-science track (ds-var-cointegration, ds-time-series-arima) as a real connectedness-index exercise. Cross-ref: stablecoin-mcap-is-repackaged-money, what-needs-a-stablecoin, receipt-is-not-settlement.