THE fraction of enterprise value of large US companies represented by tangible assets — things like real estate and inventory — has fallen from 50% to 20% over the past 15 years. Moreover, only about 25% of enterprise value is in identifiable intangible assets such as patents, copyrights, trademarks, customer lists, and data files. The remaining 55% is in assets too intangible to even identify clearly — things like technical knowledge, corporate culture, going-concern value, and customer relationships.

A recent article in Business Law Today calls attention to something less tangible still — the economics of negative information assets. How do you assess and safeguard corporate know-how about products that didn’t work, dead-end research or failed experiments at a time of heightened staff mobility?

For companies with little in the way of tangible assets or recurring revenue — ones valued primarily on expected future products and services — gauging the worth of what a company knows is key, and much of that knowledge is about approaches that don’t work. Moreover, that knowledge only has value if the company can protect it.

Just as normal matter exists in a sea of dark matter and dark energy that make up 95% of the universe, positive business information (such as product specifications) exists in a much larger sea of negative information (such as ideas that don’t work). Advertisers work to create brand names — positive information assets — but also to warn against an innumerable list of dangers from head (dandruff) to toe (athlete’s foot), or make “where’s the beef” criticism of competitors’ products. Such negative information assets increase the value of a product by making the alternative seem worse. Employee handbooks list a few positive admonitions, and much longer lists of what not to do.

Not only are intangible assets — especially negative information assets — growing rapidly in economic value, they are also getting harder to protect. Employee mobility has grown since 2008, especially in technology fields; and the government is working to increase it further with Federal Trade Commission efforts to ban non-compete clauses and the proposed Workforce Mobility Act of 2023. Courts are getting more reluctant to grant patents and other intellectual protection, especially on software. Work-from-home means information spills out to uncontrolled locations, and of course the internet makes transmitting information far easier than in the past.

From an investors’ perspective, this means that only about 20% of what you pay for an average stock covers assets that can be easily insured and protected against theft or damage; and that creditors can seize and sell in a bankruptcy with some hope of recovery. Stock investors are hoping for future cash flows to be generated mostly by intangible assets, in many cases assets that cannot even be identified clearly and may be negative. Creditors can look to only a small portion of enterprise value to secure loans.

At least with positive information there is some hope of tracking down theft and suing for recovery. If an employee goes to a competitor who soon launches a product similar to yours, you will notice. You can demonstrate the similarities in court, and perhaps refer to patent or other protection.

But what if an employee goes to a competitor and designs a different product from yours, and the research effort is sped up because your former employee learned on your dime what dead ends to avoid? What will you point to in court to prove the theft? What kind of patents protect failed ideas?

Most of my professional experience is in finance. A hedge fund employee who copies her former employer’s strategy can be identified. But behind every implemented strategy are hundreds or thousands of alternatives that were tested and rejected. An employee who knows those results can find a good new strategy much faster than a researcher starting from scratch.

One prominent example of this is the high-profile lawsuit by Alphabet, Inc.’s Waymo alleging that its former engineer Anthony Levandowski took to Uber Technologies, Inc. information about dead-end designs and approaches unsuited to market that saved Uber substantial time and money, and were critical to its efforts to develop self-driving cars.

Moving from an investor’s to a social perspective, intellectual protection and trade secret law attempts to balance the social good of encouraging research and development, with the other social good of competition. Patents force applicants to reveal their secrets in return for 20 years of protection. Companies are allowed to take some measures to protect trade secrets, but not to prevent former employees from making a living in their fields.

Negative information does not fit into either model. You can’t force former employees to remake mistakes they identified while working for you. You can’t get a patent on a failed idea, one that forces other people to pay you money to avoid the failure. From a social standpoint these don’t balance any considerations, they just impose perverse costs.

There is a related issue in academic research — it’s rare to publish negative results. A vast amount of negative information knowledge is systematically excluded from the literature in many fields.

If the trend toward increasing negative information asset value and decreasing ability to protect it continues, it should lead to fragmentation in research. If every company pursues one idea and liquidates if its idea fails, negative information becomes public, and no entity has to protect it. Negative information assets are a problem only for large companies that survive the failure of many research paths, and thereby accumulate large negative information assets that cannot be insured or protected.

One solution for a large company is to put researchers in siloes, so individual employees have knowledge of at most one dead-end. But this eliminates valuable cross-fertilization of research and gets in the way of rational employee development. Many hedge funds do this, but it is a problematic solution in fields like information and medical technology where the scientific culture is more open, and there is more communication between academic and for-profit activities and more diverse specialists are needed for projects.

I don’t know how this will play out, but I think it represents one of the biggest challenges to big tech company valuations. With increasing employee mobility and weakening non-compete protections, big tech is losing the ability to protect its fastest-growing type of asset. I can’t see any good solution — either in research strategy or law — but it’s something investors should consider seriously.

BLOOMBERG OPINION