Skip to content
News
CommentaryEuropean Union

BIS Working Paper Finds On-Chain DeFi Metrics Unreliable Without Methodological Care


Key points

  • Bitcoin transaction values can differ by up to a factor of six depending on the measurement approach applied to raw blockchain data.
  • The authors classify approximately 13 million active smart contracts, of which around 1.4 million are tokens, highlighting the scale of proliferation that complicates identifying economically meaningful activity.
  • Stablecoin behaviour diverges by chain: Ethereum usage is more associated with smart-contract interactions, while Tron usage more commonly occurs outside smart contracts, suggesting transactional or store-of-value motives.
  • The BIS paper draws on data from Mercurius, a platform covering Bitcoin, Ethereum, and Tron, and proposes a measurement toolkit using granular estimates, explicit assumptions, and technical disaggregation.
  • The central finding is that on-chain indicators should be treated as noisy approximations rather than direct measures of economic activity, a conclusion with direct implications for any model or regulatory framework that treats them as ground truth.

A Bank for International Settlements (BIS) working paper argues that widely used measures of cryptoasset and decentralised finance (DeFi) activity are far more sensitive to methodological choices than practitioners typically acknowledge, and should be treated as noisy approximations rather than direct readings of economic activity. Drawing on granular transaction data from Mercurius, a platform covering Bitcoin, Ethereum, and Tron, the authors identify three structural sources of divergence: how Bitcoin transaction values are aggregated, the proliferation of spurious smart contracts, and the difficulty of comparing activity across chains where the same instrument can serve different economic functions.

The numbers are striking in their range. Bitcoin transaction values alone can vary by a factor of six depending on how blockchain data are processed, and the authors classify roughly 13 million active contracts of which around 1.4 million are tokens, many with no clear economic significance. Trading activity across the chains studied is found to be highly concentrated and centred on stablecoins, but even stablecoin behaviour is not uniform: Ethereum stablecoin usage skews toward smart-contract interaction, while Tron usage more often sits outside smart contracts, consistent with transactional or store-of-value purposes rather than DeFi participation.

For operators building data pipelines, risk models, or regulatory submissions that rely on on-chain metrics, the paper’s practical implication is that headline aggregates from public blockchains embed technical artefacts alongside genuine economic signals, and the two are difficult to separate without explicit methodological choices. The authors propose a toolkit centred on granular, bounded estimates with declared assumptions, technical classification, and disaggregation. Policymakers and researchers reading on-chain data without that discipline risk drawing conclusions from measurement noise rather than economic substance.

Original source

BIS HQ working papers

bis.org