BIS Paper: Bitcoin Transfer Figures Can Swing Sixfold on Method Alone

A Bank for International Settlements working paper published Sept. 15 lands on an uncomfortable truth for anyone who quotes on-chain data with confidence. Bitcoin transfer-value estimates varied by as much as a factor of six across the measurement approaches the authors tested. That isn't a claim of a universal sixfold error. It's messier than that, because it means the same public ledger can support wildly different headline totals depending on choices the analyst makes.

The paper, working paper 1377, identifies transaction aggregation as one of three structural sources of divergence, alongside smart-contract programmability and cross-chain comparisons. Where you draw the line around a single transfer changes what counts as economic activity. Grouping matters, and the ledger doesn't tell you how to group. The authors frame their conclusions as their own, not as official BIS positions, which is worth keeping in mind before anyone cites this as a regulatory ruling.

The contract data makes the problem concrete. The authors classified 13 million active contracts, including roughly 1.4 million tokens, and found trading activity heavily concentrated around stablecoins. Heavy token issuance and fast contract proliferation make real economic activity harder to separate from noise. So a dashboard can apply its calculation consistently and still publish a total built on assumptions nobody reading it can see.

Stablecoins sharpen the point. The same asset means different things on different chains. On Ethereum, stablecoin activity tracks closely with smart-contract interactions. On Tron, stablecoins sit outside smart contracts far more often, a pattern the authors read as transactional and store-of-value behavior.

So cross-chain rankings that treat every recorded unit as economically equivalent are comparing apples to stablecoins, basically. Those rankings get published anyway, and they get cited in board decks.

Here's where it bites on my beat. Regulators weighing licensing regimes and disclosure rules lean on these same indicators, and MiCA compliance already forces firms to justify numbers they report. If the underlying measure can move sixfold on methodology alone, then the methodology is the disclosure. The authors want granular, data-bounded estimates that state their assumptions up front, and that's the correct ask.

Capital follows clarity. Right now the clarity is thin.