Bitcoin records every transaction on a public ledger, but that data does not produce one fixed measure of economic activity.
A September 2026 BIS working paper found Bitcoin transfer-value estimates can differ by as much as six times. The result changes depending on how analysts count transaction outputs and remove funds returned as change.
The finding directly affects widely cited Bitcoin on-chain volume figures. Different methodologies can produce sharply different totals from the same public transaction records.
The BIS study examined Bitcoin transaction data covering 2009 through 2026. Its dataset included about 1.3 billion transactions and 3.6 billion transaction outputs.
Researchers calculated an upper estimate, an adjusted estimate, and a conservative lower estimate for Bitcoin transfer volume.
The upper estimate counted all transaction outputs without making adjustments. That method effectively treats every output as transfer activity.
Researchers then removed self-transfers, where funds return to a sending address. These outputs most likely represent change rather than separate economic transfers.
The adjusted method reduced measured on-chain flows by up to six times. The paper notes that markets widely report the unadjusted upper figure as transfer value.
Bitcoin uses an unspent transaction output, or UTXO, model instead of an account-based system.
Each Bitcoin transaction consumes existing outputs and creates new ones. Users must spend an entire UTXO even when transferring only part of its value.
The study gives an example involving a four-Bitcoin UTXO. A user can spend it to transfer 1.5 BTC.
The transaction sends 1.5 BTC to the recipient and returns 2.5 BTC to the sender as change.
However, transaction data cannot always identify which output represents change. Bitcoin can send change to a new address controlled by the original sender.
Raw volume calculations can therefore include returned funds alongside actual transfers. Different assumptions for identifying those outputs produce different Bitcoin on-chain volume figures.
The study also found that excluded self-transfer values increased sharply after March 2016. Researchers linked the increase to greater address reuse.
(adsbygoogle = window.adsbygoogle || []).push({});The paper states that public blockchain records remain transparent, but methodology determines how researchers translate those records into economic measures.
The authors describe on-chain indicators as “noisy approximations,” not direct measures of economic activity. They say methodological choices and underlying assumptions can alter widely used crypto metrics.
For traders and analysts, checking methodology means identifying whether a volume figure includes self-transfers and change outputs.
The BIS researchers use bounded estimates instead of treating one calculation as the definitive transfer value. They also call for explicit assumptions when measuring blockchain activity.
The study states that a range of estimates provides a better measure of economic activity and trends than one figure.
