What An Asset Is Worth With No Buyer

Ask four professionals what a single patent is worth and you will get four numbers that can differ by an order of magnitude. None of them will be lying. Each is answering a different question, and the question was never agreed. Start with the scale of what is being priced. Intangible assets make up about 92 per cent of the market capitalisation of the S&P 500 (Ocean Tomo, 2026). Most of the balance sheet of a modern company is not on the balance sheet. It sits in claims, marks, code and contracts, and almost none of it has a published price. The cost approach answers the narrowest question. Value equals the sum of what it would take to recreate the thing: direct engineering, materials, equipment time, the failed experiments, filing fees, prosecution, and renewals to the end of the expected life. Reproduction cost rebuilds every dead end. Replacement cost rebuilds only the path that worked. It is a floor and courts treat it as one. In the United States, the Georgia-Pacific factors allow cost as a reference point but prefer forward-looking methods. German courts accept cost as a minimum inside a licence analogy. Nobody's best answer. Then the case that breaks the method. Fraunhofer's mp3 patent family cost about EUR 50,000 to prosecute. The institute never manufactured a player. It licensed, and the family returned high double-digit millions a year (Beyer and Mackenstedt, 2008). A method that prices that family at EUR 50,000 is not measuring value. It is measuring effort.

The income approach is where professional practice lives, and it has six sub-methods that are really six different bets. Relief from Royalty is the workhorse for a single patent. You value the royalty the owner would have had to pay if they did not own it. That means you need a rate. Rates come from comparable transactions, and the published ranges are wide and industry-specific: pharmaceuticals land between 5 and 15 per cent of net sales, software between 5 and 12, electronics between 1 and 5 (ktMINE and the 2025 LES survey). Pick the bottom of one range and the top of another and the same patent moves by a factor of fifteen before you have projected a single sale. The 25 per cent rule is older and cruder. The licensee keeps 75 per cent of operating profit, the owner takes 25. It survives as a starting point and is widely criticised, because the ratio was never meant to be a formula. Multi-period excess earnings is the method for the primary asset in a purchase price allocation. You take the cash flows the asset generates and subtract a fair return on every other asset that helped: working capital, fixed assets, the workforce, other intangibles. What is left is attributed to the patent. Risk-adjusted net present value is the pharmaceutical method. Each phase carries its own probability of success, published by the industry, and only about 12 per cent of drugs that enter testing reach approval. The capitalised cost of a new drug runs at USD 2 to 4 billion (IQVIA, 2025). Real options is the last one, for platform technology where the value is not the cash flow but the option to keep going or stop. The probability that two valuers using the same patent and the same comparables land on the same number is not zero. It is close to zero.

Value does not sit in the middle of a patent portfolio. It sits in a thin tail, and the tail is the whole point. Gambardella, Harhoff and Verspagen (2008) put the median European patent at roughly EUR 300,000 and the mean at roughly EUR 3 million. That ten-fold gap is not a rounding error. It is the shape of the distribution. Scherer and Harhoff (2000) found that the top 10 per cent of patents account for 80 to 90 per cent of the value of a portfolio. Pakes (1986) worked from renewal data in France, the United Kingdom and Germany and concluded that the median patent is worth close to nothing. Around half of United States patents lapse before half their term. The owners simply stop paying. So the arithmetic everyone reaches for is wrong. Portfolio value is not the value of one patent multiplied by the count. It is a Pareto problem. You identify the outliers, you value those properly, and you score the rest as options that will probably never be exercised. I restructured a EUR 75 million industrial group once. The same logic ran through the asset register. Most of the book was carried at a number that no buyer would have paid, and the business survived on two or three assets nobody had bothered to value properly. The count told us nothing. The tail told us everything.

A patent is worth what it is worth only if it holds. That is the largest single destroyer of value in the entire field, and it is routinely left out of the model. Roughly 50 per cent of litigated patents are invalidated (Lemer, 2008; Hall and Harhoff, 2012). At the Patent Trial and Appeal Board in the United States, the institution rate for inter partes review runs at about 60 to 65 per cent (USPTO data). Those are not edge cases. They are the base rate. The practical effect is a premium. A patent that has survived a validity challenge is worth two to five times a comparable one that has never been tested, because the uncertainty that a buyer would otherwise carry has been retired in public. That is the single most important price inflection in the life of an asset, and it costs the least to explain. Half of the value is decided by a document most owners have never read.

Pick the bottom of one royalty range and the top of another and the same patent moves by a factor of fifteen before you have projected a single sale.

The same claim is worth different money in different countries, because the remedy changes with the border. The United States runs on damages. There is a jury, treble damages are available for wilful infringement, and there is no statutory cap. It is also slow and expensive: two to three years to trial, with both sides spending roughly USD 3 to 10 million. Germany and the Unified Patent Court run on the injunction. It is close to automatic, which is the point, and it arrives in about 12 to 18 months at a cost of EUR 200,000 to 1 million. China runs on market exclusion, with statutory damages capped at RMB 5 million and punitive damages up to five times for wilful conduct, at 12 to 18 months and EUR 50,000 to 300,000. The United Kingdom runs a discretionary injunction, 18 to 24 months, GBP 500,000 to 2 million, with the loser paying a share of the winner's costs. The same patent, the same claims, the same infringer, four different numbers. The premium in each jurisdiction has its own name. In the United States it is a damages premium. In Germany an injunction premium. In China a manufacturing premium. In Japan a predictability premium. In the United Kingdom a cost-shifting discount. A cross-border family is not one asset. It is a set of jurisdiction-weighted assets that happen to share a specification.

The mistakes in this field are not random. They cluster, and the biggest ones all push the number in the same direction. Revenue projections built with no market validation overvalue by 2 to 10 times. A discount rate borrowed from the entity's cost of capital and applied to a single asset's risk overvalues by 1.5 to 3 times. Ignoring the probability of invalidation overvalues by 2 to 5 times. Treating technical success as commercial success overvalues by 3 to 10 times. Every one of those is an error of optimism, and they compound. The cleanest evidence on that last one comes from citation studies. Hall, Jaffe and Trajtenberg (2005) found that forward citations do correlate with market value, but that the relationship explains at most 15 per cent of the variation. A well-cited patent is a technical achievement. It is not a price. Farre-Mensa, Hegde and Ljungqvist (2020, Journal of Finance) found something stranger. Winning the first patent raises a startup's probability of venture funding by 47 per cent, its probability of getting loan collateral by 76 per cent, and its probability of an IPO by 128 per cent. The effect is strongest for inexperienced founders and in capital-constrained regions. A second patent adds nothing statistically significant. Read that against the standard story. The first patent is worth most not because it excludes competitors, but because it opens a funding channel. The asset is the signal.

The professional answer is not a better formula. It is a protocol, and the protocol is boring on purpose. Run every applicable method independently, each with documented assumptions. Weight each one by the quality of its inputs. Take the weighted central estimate. Then report a range, the 25th to the 75th percentile across your sensitivity cases, and never a single point, because a single point claims a certainty nobody has. The Big 4 practice matches the protocol. Deloitte uses relief from royalty for patents and multi-period excess earnings for technology in an allocation. EY triangulates across all three approaches and runs Monte Carlo simulation for uncertainty. KPMG benchmarks from more than 20,000 public filings and 30 years of engagement data. Every one of them documents assumptions in a verification log that reads like a checklist: prior art searched, remaining life adjusted, invalidation probability applied, discount rate justified against asset risk rather than entity risk, royalty source cited, sensitivity run. The discipline is the deliverable. The number is a byproduct. I took a company through a Frankfurt listing. I signed a EUR 20 million personal guarantee and the company folded on Christmas Day 2008. What I learned then holds here. An asset has no price until two parties are willing to transact. Until that moment it has a method, a set of assumptions, and a range, and anyone who hands you a single figure has skipped the part that took the work. The probability that the next valuation you are quoted reports four methods, a range and its own assumptions is not high. Most people are pricing the middle and ignoring the tail that holds the value. What follows is public valuation method, summarised with its sources. It is not advice, it is not a valuation of any particular asset, and the authority on any given number is the report behind it.