A Bank for International Settlements (BIS) study found estimates of bitcoin on-chain transfer values can differ by as much as sixfold depending on measurement methods.

Cointelegraph, a blockchain outlet, reported on Sept. 15 that BIS researchers analysed 10 billion records from the bitcoin, ethereum and tron blockchains and said widely used crypto metrics may not accurately show real economic activity.

The key is bitcoin’s transaction structure. When users send coins, leftover amounts often return to the sender as change. Because this amount is recorded as a separate output on the blockchain, it can be double-counted as transfer value even though the funds did not move to another party.

Differences also emerged in how bitcoin market capitalisation is calculated. Conventional market capitalisation was sometimes 4 times higher than realised market capitalisation, which reflects the price at the time of the last movement at a given point. That is because widely used market capitalisation applies the current price uniformly to all circulating coins, while realised market capitalisation is calculated based on the price when each coin last moved.

These measurement issues were not limited to bitcoin. Ethereum’s structure, shaped by the widespread use of smart contracts, made it difficult to classify activity. Of about 67.5 million active smart contracts reviewed by the researchers, about 54 million could not be categorised under the study’s classification system. This showed that even with abundant on-chain data, it can be hard to interpret its meaning consistently.

The researchers also said stablecoins serve different purposes across blockchains, so simple aggregation can create distortions. USDT was more closely linked to decentralised finance (DeFi) activity on ethereum, while on the tron blockchain it had a stronger character as a means of payment and store of value.

The share held by smart contracts also differed widely. On ethereum, the share of USDT held by smart contracts exceeded 20 percent in 2022, but on tron it remained at about 1 percent.

This led to criticism that even for the same USDT, the nature of economic activity differs by network, and bundling and aggregating it as one can obscure how it is actually used. The researchers concluded that on-chain indicators should be treated as approximations, not direct measures of economic activity.

Markets are already seeing moves to distinguish between on-chain activity and adjusted metrics. Visa’s on-chain analytics dashboard, based on data from Allium Labs, shows both total stablecoin transaction volume and adjusted transaction volume. Visa explained that the aim of the adjustment method is to remove potential distortions such as high-frequency trading, bots, bridge routes and internal exchange operations.

Visa’s dashboard showed total stablecoin transaction volume over the past 30 days was $6.4 trillion, while adjusted volume was $313.1 billion. This suggests that even with the same on-chain data, the scale can vary widely depending on aggregation criteria, meaning that to identify real economic activity, it is necessary to examine not only transaction volume but also calculation methods and the nature of transactions.

The study suggests that how on-chain data are classified and interpreted matters more than the amount of such data. It said investors and policy authorities also need to consider network-specific characteristics and the limitations of aggregation methods when using on-chain metrics as a basis for judgment.