Bank for International Settlements researchers conduct data inspection and testing of XRP ledgers

Bank for International Settlements researchers test XRP ledgers as verification layer for official statistics

According to a working paper released on September 2, researchers affiliated with the Bank for International Settlements (BIS) tested XRP Ledger (XRPL) as a verification layer for the integrity and source of official statistics. The prototype system records fingerprints of encrypted datasets rather than publishing underlying statistics on the chain for public verification. In controlled tests conducted by the researchers, the measured median release time delay was three to five seconds. By comparing the document with XRP Ledger records, the verification process takes about one to two seconds.

The authors point out that the experimental system was not intended for production use and remains unmaintained prototype software. The prototype connects XRP Ledger to the Statistical Data and Metadata Exchange (SDMX) standard, an international standard used by institutions including BIS to exchange statistical information.

Prototype technical principles: Data integrity verification

The system does not link a complete statistical data set, but calculates an encrypted fingerprint representing the data and records it on XRP Ledger's Devnet, allowing recipients to check whether downloaded files are consistent with the originally released version.

Official statistics are usually distributed through websites, databases and automated data feeds. The recipient must trust that the downloaded file is authentic and that the content has not changed since the publisher published it. BIS tested a way to prove that official numbers have not been tampered with, by "stamping" documents on the open source public XRP Ledger. This technical demonstration shows how to solve this problem by independently verifying records.

The researchers designed the prototype to provide an independent verification record. Prior to release, the software converts SDMX files to a standardized format and calculates the SHA3-512 cryptographic hash value. The hash function is similar to a unique digital fingerprint, and even if tiny parts of the source file change, it will produce different results. Therefore, the recipient can double the calculation and compare the result to the fingerprints recorded on the chain.

This prototype supports checking entire files or selected statistical series within them. When multiple series are included, the system combines their hashes into a single Merkle Root. This allows multiple data components to share a single on-chain record while maintaining independent verifiability. The root node is anchored through an XRP Ledger transaction, and a reference to the transaction is then embedded into the SDMX file, allowing the recipient to obtain the information needed to locate the records on the chain and complete the comparison.

This design means that XRP Ledger does not establish the accuracy of the statistical data itself, but only provides evidence that the verified data matches the version connected to the original record.

Performance Indicators and Network Environment Description

The paper reported that under controlled test conditions, the median release delay was approximately three to five seconds. The verification time is approximately 1 to 2 seconds. Release delays cover the process of creating and confirming blockchain records, while verification involves recalculating fingerprints and checking them against information stored in relevant XRP Ledger transactions.

The researchers emphasized that these results should not be regarded as guaranteed performance under production conditions. Measurements were made through an experimental system running on XRPL Devnet, rather than a public main network used for real transactions. Devnet provides XRP for testing through faucets, allowing developers to experiment without using market-value assets. Its activities, reliability requirements and operating environment are different from the XRP Ledger main network.

Researchers chose XRPL in part because of its relatively fast transaction confirmation speed and low transaction costs. However, the experiment did not directly compare XRPL's performance to Ethereum, Solana, licensing databases or traditional digital timestamp services.

Authentication and Expansion Potential

Confirming that the data has not changed solves only part of the authentication problem. Malicious actors can create a valid hash for fraudulent files and publish it from unrelated blockchain addresses. To this end, the prototype introduced a publisher identity system, utilizing a W3C verifiable credential signed by an identity key associated with the publisher's XRP Ledger address. The recipient can use the credential to check whether the entity issuing the document controls the recognized address. Therefore, the verification process evaluates both data integrity and publisher identity.

Researchers say the architecture may eventually support Zero-Knowledge Proofs. Such technologies allow organizations to prove specific facts about restricted data without fully disclosing the data set. In addition, the potential uses of artificial intelligence agents are also pointed out. As automated systems increasingly retrieve statistical data without having to manually identify the source one by one, machine-readable verification records can allow software to reject files that have been tampered with or misattributed.

These are still proposed extensions. The published codebase provides experimental code rather than production services for central banks, national statistical agencies or artificial intelligence companies.

The use of XRP Ledger does not necessarily require XRP tokens

The prototype uses the XRP Ledger infrastructure because each record on the chain requires transactions, but it does not use XRP for payments, liquidity management, settlement, or cross-border transfers. Anchoring each record only on the main network requires a small transaction fee, and the specific amount depends on network conditions and the publication method chosen by the organization.

This distinction is critical because using XRP Ledger does not necessarily create substantial demand for its native assets. Data publishers can anchor many statistical series through a single Merkle root, reducing the number of transactions required.

Still, the experiment expands the types of applications tested on XRPL beyond payments and tokenized financial assets. The network is also developing institutional functions covering licensing transactions, lending and asset issuance. Recent data shows that XRP Ledger's orderbook trading volume has increased by 79%, while the number of daily traders has declined, indicating an increasing concentration of online activity. But these developments do not establish that statistical prototypes will enter production; they only provide a broader context for testing different financial and non-financial applications in the XRPL ecosystem.

Production deployment still requires further testing

Additional security, governance and operations work will be needed before official statistics providers can rely on the prototype. Institutions need to develop policies on publisher keys, credential leaks, failed transactions, and corrections to previously released datasets. The permanence of blockchain records presents another challenge: Official statistics are often revised after institutions receive new information or correct errors. Production systems must distinguish between legal revisions and unauthorized changes without implying that the earliest version is still valid.

Publishers also need to decide whether to rely on public blockchain, run additional infrastructure, or combine blockchain records with existing digital signature systems. Privacy regulations may restrict what metadata can appear in public transactions.

The authors say their approach is not limited to SDMX and can also adapt to other structured formats, including XBRL, which companies and regulators use for financial reporting. No implementation deadline is currently provided. The codebase states that the software is not maintained, and BIS has not announced plans to deploy the program in its official statistical publications.

Therefore, the scope of the verification results is narrower than the claim that BIS uses XRP Ledger. BIS affiliated researchers demonstrated that under controlled conditions, experimental systems can use XRPL Devnet to authenticate statistical files in seconds.