Should Data Appear On Corporate Balance Sheets?

China has allowed companies to report data as assets on their balance sheets. The US might want to learn from that initiative.

Abstract gravity wave background – 3d rendered image. Hologram view, physical process. Futuristic AR topography view.

getty

We have known for quite a while now that corporate data can be very valuable. One of the big barriers to mass corporate adoption of AI is the absence of scrubbed and structured data in companies. Many firms continue to rely on legacy systems. I know of one multi-billion-dollar company that has a mind-boggling 16 identifiers for the same customer. Technical debt related to under-investment in data has become a major obstacle to digitization and AI adoption. Consultants, I know, pitch the promise of glitzy AI but get hired to do the mundane but very important job of cleansing the data to become AI ready.

One potential solution to this under-investment problem may be to allow firms to capitalize the cost of gathering, scrubbing and organizing the data on their balance sheets as assets. By no means is this a silver bullet as investment will continue to depend on the CEO’s awareness of the value of data, her other priorities and alternative uses of cash. But the move might help, on the margin. CEOs still care about earnings as a scorecard of their performance and capitalizing internal costs of acquiring data may increase investments in tracking and using data to learn even more about the firm’s operations and financing opportunities.

The Chinese reporting rule

In 2023, China took the lead in allowing firms to recognize data as assets – either as intangible assets or inventories, provided (1) data resources are identifiable, (2) the economic benefits of data resources are highly probable and attributable to the firm, and (3) the costs of generating or acquiring data resources can be measured reliably.

The new rule requires firms to segregate data for internal use and data for sale. Data for internal use refers to data used within the firm to support decision-making and enhance performance. Such data can be reported as intangible assets based on acquisition or development costs(e.g., costs incurred for collecting and processing data; not fair value). Like the treatment of D in R&D under IFRS, costs associated with developing data-related intangibles can be recognized as development expenditures (under intangible assets) during the development stage. These development expenditures will be transferred to intangible assets when they are ready for use.

In contrast, data for sale are data intended to be sold as products or services to external entities and are to be recorded as inventory in financial statements. Once data resources have been recognized as intangible assets or inventory, they are subject to impairment tests and amortization. In addition, firms are required to classify data based on their sources into purchased data, self-developed data, and other data and report the value of each category separately in the notes to financial statements.

The rule encourages extensive voluntary disclosure on the economic materiality of data including data sources, data processing and management, the application of data in firms’ operations, the impact of data resources on material operating, financing, and investing activities, the property rights over data resources, as well as the potential legal restrictions on data resources.

How has the experiment worked?

Hai Lu, chaired accounting professor at University of Toronto and Peking University, along with his co-authors, reports that the experiment has not worked all that well. Barely 2% of listed Chinese firms have capitalized data on their balance sheets. The primary concerns related to valuation based on cost, which may be over conservative relative to fair value, and property rights issues related to who owns the data (customer or company). On top of that, auditors will object to any rule that makes their verification task harder, regardless of how motivated the investor might be to access that information.

The way forward

My take is a bit more optimistic. Both concerns look fixable. How about asking firms to recognize costs but report fair value in footnotes (like SFAS 107 and 157 for financial assets)? Property rights issues are tricky but continued legislative activity will hopefully resolve questions about data ownership.

My worry is that the FASB has not even resolved R&D capitalization after decades of investor demand for more clarity. Data seems like a bridge too far for our glacially slow rule makers. With the increased focus on crypto and AI, perhaps the new SEC, under President Trump, might consider pushing for rules on when and how to capitalize data on balance sheets.

On the margin, such a move might accelerate firms’ investment in systems needed to capture the costs and benefits of data in their companies and hence in the eventual faster adoption of AI. If the rule makers don’t act, investors will continue taking wild shots at trying to guess the reliance on data as a new factor of production in companies. Regulators will compound their puzzling refusal to acknowledge that firms increasingly rely on an array of intangible assets such as data, as opposed to brick and mortar, to generate shareholder value.