Posted on Tuesday, June 11, 2019

Mohini Singh, ACA, Director of Financial Reporting, CFA Institute and Campbell Pryde, President and CEO, XBRL US 

In June of 2018, the U.S. Securities and Exchange Commission (SEC) finalized a rule replacing the requirement that operating companies and mutual funds file in conventional XBRL with a requirement that they file using Inline XBRL.

The XBRL standard was originally published in 2003, 16 years ago. Inline XBRL is an additional XBRL standard published in 2013 that aligns XBRL with human readable report formats used on web-sites. But is Inline XBRL the right approach? We sometimes hear arguments that XBRL is “old technology”. And that it’s time to replace the standard with technologies like blockchain or artificial intelligence (AI).

Both blockchain and AI have been suggested as technologies that should be considered instead of the XBRL data standard. But what are they, really? Can a technology replace a standard? Or is XBRL complementary to AI and blockchain? Let’s look at the facts:


It’s usually described as a distributed, decentralized, public ledger. The “blocks” in blockchain contain records of information: 1) transactions (for example, the date, time, and amount of a purchase), 2) the digital signature of the buyer and seller of the transaction, and 3) a unique identifier called a “hash” that allows us to tell it apart from every other block. The “chain” in blockchain is the links between all the blocks. Each time a new transaction occurs, it is added as a permanent block to the chain.

Most blockchains are public and therefore all transactions on the chain can be viewed, but because the buyers and sellers are represented through digital signatures, their identity is masked.

The purpose of the blockchain is to establish trust (that a transaction has occurred, and the amount has been paid) among untrusted parties (when you don’t know the identity of the parties to a transaction). In the absence of blockchain, trusted third parties like banks, brokers, or big retail distributors, like Amazon, facilitate transactions between two parties who don’t know each other. The middleman serves an important role because it can verify and check identities, confirm that the transaction actually occurred, and that it was conducted for the amount that both parties agreed upon. The blockchain eliminates the need for such centralized authorities because it contains all the data about the transaction and is viewable by all parties, but it masks the identity of the parties. And the blockchain provides an audit trail that never goes away.

So will blockchain replace XBRL??

Unlikely. Blockchain is not a data standard. And XBRL is not a distributed ledger system. Replacing one with the other would be like replacing the English language with an iPhone. Both are used to communicate, but one is the standard, and one uses the standard. In fact, just as you need to be able to speak a language to communicate, blockchain needs a data standard to record information.

One example of how blockchain needs data standards is a smart contract. These are digitized contractual obligations that reside on the blockchain. A smart contract between two parties may specify that if the debt coverage ratio of one party falls below a certain level, an action is triggered by the digital contract. As the concept of the smart contract was developed, it became clear that reliable, consistent, machine-readable data is necessary for smart contracts to be fulfilled. The only way to enable access to consistent, machine-readable data is through universally accepted data standards. Smart contracts that rely on data prepared using a financial data standard can automatically trigger an action without the need for human intervention.

In point of fact, the development of blockchain makes the need for standards essential. The excitement and interest in blockchain based technologies has raised awareness about the lack of financial standards, and focused technology enthusiasts on identifying existing standards that can be leveraged, and developing new ones where needed.

Artificial intelligence.

AI is about creating technology that allows computers and machines to function in an intelligent manner, as opposed to natural intelligence, which is displayed by humans. Machines with AI can mimic cognitive functions, like learning and problem solving. Examples of AI are self-driving cars, and drones. AI can be used to detect fraud in the banking industry, or to assist in identifying appropriate healthcare treatments, by analyzing large amounts of data.

The XBRL data standard renders information into machine-readable format, eliminating the need for manual data entry. AI is sometimes referred to as a tool that could render text or numeric facts into machine-readable information. In theory then, couldn’t AI be used to read a financial statement and understand the reported values without the need for human intervention? If so, wouldn’t that eliminate the need for humans to XBRL-ize their financial data?

If we have AI, do we need XBRL?

Most definitely. Much of AI is based on machine-learning – deriving information and making decisions by identifying patterns in data. An AI system needs to review of lots of examples, drawn from lots of data, and supported by an open data ecosystem with unambiguous standards for structuring and labeling that data. The quantity and quality of the data supporting an AI system is a key factor in the usefulness of the AI system.

The U.S. Government is taking a close look at how technologies like AI can be used and at understanding the potential risks in AI. In 2018, the Department of Homeland Security published a paper on AI, noting “As AI technologies become more widespread, efforts to ensure that they work as intended become more critical. Our interviews and literature indicated that the application of a standard or combination of standards—such as analytic, research, legal, regulatory, moral, ethical, technical, industry, data, and information security— can help to reduce the risk of adversary exploitation.“

Standardized data, like XBRL for financial data, is crucial to building a data ecosystem that can support AI. For an AI system to effectively extract, understand, analyze, and learn from vast quantities of data, requires access to data that is clearly and unambiguously defined. When it comes to financial data, that can only be XBRL. As with blockchain, XBRL data standards cannot be replaced by Artificial Intelligence. AI needs XBRL to provide unambiguous, consistent data to drive machine learning.

The last word

XBRL has been around for a long time, but a data standard is not a technology. And unlike technologies, standards evolve to adapt to new technologies. They actually improve over time, and with greater use.

As the AI and blockchain communities have evolved, and as they investigate new use cases for these technologies, they recognize that mature, widely used data standards are critical to the smooth functioning of AI and blockchain applications. They need data standards that support numerous information collection and analysis systems, and that produce good quality, consistent, clean, digitized data that is interoperable with other data.

The XBRL standard, because of its ability to create machine-readable, consistent, unambiguous data, is a perfect fit to drive technologies like AI and blockchain.