BIS Tests XRP Ledger Prototype to Verify Official Statistics in Seconds

Points of Focus

• BIS researchers tested an XRP Ledger proof of concept for verifying the origin and integrity of official statistical datasets.
• The prototype recorded median publication times of three to five seconds and verification times of one to two seconds.
• A single XRPL entry can anchor fingerprints for thousands of data sets, with possible extensions to XBRL reporting, zero-knowledge proofs and AI agents.

Researchers behind a Bank for International Settlements working paper tested XRP Ledger (XRPL) as part of a system for independently verifying official statistics.

A BIS working paper published Sept. 2 describes a proof of concept that generates cryptographic fingerprints for statistical data sets and anchors a summary value on XRPL. Recipients can use the ledger record to confirm a data set’s publisher and detect changes made after publication.

XRP Ledger anchors statistical data fingerprints

The research centers on SDMx, or Statistical Data and Metadata eXchange, a standard widely used to distribute official statistics. According to the paper, existing SDMx workflows lack a built-in cryptographic mechanism for confirming data provenance and integrity.

The prototype fingerprints each data set, groups the hashes in a Merkle tree, and records a single root on XRP Ledger. The statistical files remain offchain, while signed credentials and an onchain attestation registry allow recipients to verify the publisher and file integrity with one ledger lookup.

The experiment adds a data-verification use case to the growing range of institutional activity around XRP Ledger. Recent developments have included higher RLUSD collateral limits for institutional borrowers on Bybit and work around the respective roles of XRP (XRP) and RLUSD on the same ledger.

Prototype verifies data sets in 1 to 2 seconds

Researchers ran the implementation on XRPL DevNet using synthetic SDMx data sets. Median publication latency for canonicalization, hashing, aggregation, and offchain bookkeeping was 3-5 seconds, with anchoring confirmation following XRPL finality. Verification took 1-2 seconds under the test conditions.

The tests were conducted under controlled proof-of-concept conditions. Because one XRPL entry can cover thousands of data sets, ledger fees become negligible at moderate batch sizes, leaving processing and storage as the main costs.

BIS also released an open-source reference implementation under an Apache 2.0 license. The method is data-format-agnostic and can extend beyond SDMx to structured statistical and regulatory formats, including XBRL.

The paper identifies zero-knowledge proofs as a possible next step for verifying selected properties of a data set without exposing the underlying information. Researchers also point to automated verification by AI agents, which could check provenance and integrity before statistical data enters downstream analysis.