I. Introduction
My colleague Grover Rutter (see his article in this edition) has presented an excellent review of how to treat patents from a financial and tax reporting standpoint. However, how do you determine the real value of a patent? That begs the question: What factors are relevant to the real value of a patent? You have to know what factors into a valuation before you can address the valuation issue. Let’s start with some definitions, and then address this issue. After that, lets look at the PatentValuePredictor model and some actual valuation data and trends provided by the model.
II. General Valuation Theories
Valuation is an accounting term which means a lump sum of money payable to receive the future benefits of an asset at a particular time. There are three generally accepted accounting theories for valuing assets: market, cost, and income. Market theory values an asset as the present value ascribed to similar assets in an active public market. Cost theory values an asset by the cost of replacing the asset. Income theory values an asset by the present worth of the net anticipated economic benefit of the asset. Can we apply any of these theories to value patents?
III. Valuation Theories as Applied to Patents
Market theory valuation of patents has little or no utility because no two patents are similar enough for the sales price of one to define the value of another. Of course, you can say that patent licensing and sale applies market theory to reach a market price. The problem is that most patents are not bought or sold in an arms length negotiation, and therefore do not have an objective sale price. Even when they are licensed or sold, the transaction is usually clouded by other factors including tech transfer or line of business transfer.
Cost theory is generally inapplicable since a patent cannot be replaced. That is, once the invention is generally known, it is no longer patentable.
Income theorym is applicable in certain circumstances. Patents have known finite terms. If you can determine the income resulting from ownership of a patent over that term, you can assign a value to the patent just like you can assign a value to a long term bond.
Conventional methods using income theory to value a patent analyze micro economic data to determine the anticipated economic benefit of owning the patent. This micro economic data includes market data indicating the gross sales and net income derived from the sale of products attributable to the patent, and any revenue derived from licensing the patent. Applying income theory to micro economic data to value a patent is labor intensive, costly, and complex. This method should include an analysis to determine the meaning of the claims of the patent, a comparison of products to the claims of the patent to determine what products are actually covered by the patent, a determination of the size of the market covered by the patent, and a determination of the cost advantage of the patented technology compared to alternative technologies for that market. A micro economic analysis can be used to prove damages in patent infringement litigation. However, a micro economic analysis of a patent is often cost prohibitive for purposes of business valuation, capital allocation, taxes, and licensing. Moreover, the data necessary for members of the public to perform micro economic analysis of patents is simply not available. This is because that data includes relationships between patents, product lines, product line specific costs and earnings information, and licensing royalty rates and terms. Companies rarely release that type of information to the public. Thus, micro economic analysis of patents is often not feasible.
IV. Problems With Generally Applying Income Theory
I hasten to point out that even income theory valuation based upon micro economic analysis has limited utility in most commercial settings, as opposed to its application in patent infringement litigation. Why? Because patents and products do not have a one to one relationship. They have a many-to-many relationship. As a result, you cannot simply evaluate the value of a patent once you know the financials relating to certain products that the patent covers. To illustrate this point, consider the following two hypothetical situations.
FIRST HYPOTHETICAL SITUATION – UNUSED PATENTS:
A company owns ten patents. The first patent covers a first product, and the company manufactures that product for a hefty profit. The second through tenth patents do not cover the first product, but each one covers some alternative potential product that, if produced, could effectively compete with the first product. No one produces anything covered by the second through tenth patent. Do you allocate all value to the first patent? Surely the other nine patents have actual value to the company! How do you allocate income attributable to sales of the first product to the ten patents in order to assign value to each one of the ten patents?
SECOND HYPOTHETICAL SITUATION – THE MANY-TO-MANY CONUNDRUM:
There are three competing products in a particular niche market and five relevant patents. Patents 1, 2 and 3 each cover the first product. Patents 1, 2, and 4 cover a second product. Patents 2 and 5 cover the third product. You also know of the existence of prior art that indicates a likelihood that some claims in patents 1 and 5 are invalid. As an additional complication, what if it was unclear whether certain of the products were in fact covered by certain ones of the five patents. That is, what if the issue of infringement was not cut and dried? Confusing? You bet! Even if you knew the sales and profit margins for the various products in the hypothetical situations just noted, there would be no simple or logical way to assign values to the various patents. This is all too often the reality when comparing patents and products: there exist many-to-many patent-to-product relationships of uncertain bounds.
What have I told you so far? I have told you that classical approaches to valuation are inadequate. It is time for a new approach.
V. The PatentValuePredictor Theory for Valuing Patents
Now let me tell you about the PatentValuePredictor model for valuing patents. First, you should know that this model is implemented as a web service, and it provides valuations for all U.S. patents and (a provisional valuation of) published U.S. patent applications in real time.
How does the PatentValuePredictor model work? The PatentValuePredictor model simplifies the valuation determination problem by reformulating the problem. It does not attempt to address the many-to-many relationship noted above, and it does not attempt to find and use micro economic data relevant to any particular technology niche. Instead, it substitutes for the foregoing many-to-many quandary and the (generally unavailable) microeconomic data an estimate of an annual sales covered by the patent. The model generates a nominal annual sales covered by the patent based solely upon measurable properties of the patent document and the value of the Gross Domestic Product (GDP). I won’t bore you with the details of the model in this article. You can those details of the model in my earlier article entitled “A Macro-Economic Model Providing Patent Based Company Financial Indicators and Automated Patent Valuations ” posted in the publications sections of both www.PatentValuePredictor.com and www.Neifeld.com. Suffice it to say here that there is a heuristic relationship between measurable properties of patent documents and patent value. For example, generally speaking, the broader the claim protection, the more valuable the patent.
I will tell you that there are good points and bad points about the PatentValuePredictor patent valuations. First, the valuations are clearly statistical in nature and therefore imperfect. However, there is no such thing as perfection in valuation. Moreover, there is as far as I know, no other completely objective and generally applicable method of valuing patents. Furthermore, the Web implementation of the PatentValuePredictor model provides immediate results, and it is far less expensive (currently $100 per patent valuation) than any other method of which I am aware. Finally, as the size of an evaluated patent portfolio grows, the PatentValuePredictor model’s portfolio valuation becomes statistically more accurate. See for example the corporate patent portfolio value charts in my earlier article “A Macro-Economic Model Providing Patent Valuation and Patent Based Company Financial Indicators” posted in the publications sections of both www.PatentValuePredictor.com and www.Neifeld.com.
Finally, there are some other interesting statistics I would like to share with you that relate to valuation of patents. These statistics are derived from the PatentValuePredictor model.
First, there are currently 1,726,307 enforceable patents. To determine actual dollar values, the PatentValuePredictor model currently assumes that the entire GDP is covered by patents. The current GDP is $11.252 trillion. Therefore, the PatentValuePredictor model indicates that each enforceable U.S. patent covers, on average, annual sales of about $6.5 million (that is, the GDP divided by the number of currently enforceable patents). However, profit is, generally speaking, only a small fraction of gross sales, and old patents near the end of their term have reduced value. That explains why the PatentValuePredictor model determines an average value of enforceable patents is only about $2.8 million. To get this result, we calculated the current valuation of each one of the 1,726,307 enforceable patents, and then calculated the average value.
The PatentValuePredictor model indicates that the bulk of the most valuable patents are and have for many years been in the Pharmaceutical or Biotechnology (Pharma/Bio) technology areas. The chart below shows the currently ten most valuable patents and their technology area.
TEN CURRENTLY MOST VALUABLE PATENTS (AS OF 3/11/2004)
Patent
Issued
Current Value ($)
Assignee
Technology
6,517,866 2/11/2003 1,797,722,689 Pfizer Inc. Pharma/Bio
6,500,987 12/31/2002 1,570,968,527 Teva Pharmaceutical Industries Ltd. Pharma/Bio
6,566,344 5/20/2003 1,481,848,538 Idenix Pharmaceuticals, Inc. Pharma/Bio
6,465,496 10/15/2002 1,408,931,126 Teva Pharmaceutical Industries, Ltd. Pharma/Bio
6,452,054 9/17/2002 1,220,308,695 Teva Pharmaceutical Industries, Ltd. Pharma/Bio
6,221,640 4/24/2001 1,194,927,644 Cubist Pharmaceuticals, Inc. Pharma/Bio
6,071,970 6/6/2000 1,107,999,343 NPS Pharmaceuticals, Inc. Pharma/Bio
6,319,919 11/20/2001 1,081,784,355 Davis; Bonnie (Syosset, NY) Pharma/Bio
5,610,034 3/11/1997 1,071,288,767 Alko Group Ltd. Pharma/Bio
6,022,716 2/8/2000 1,069,310,287 Genset SA Pharma/Bio
While the Pharma/Bio tech area has held the lead for most valuable patents, the relative value of the most valuable patents has been increasing for decades. The charts below show the ten most valuable patents issued respectively in 1983, 1993, and 2003, and a relative measure of their value. Note in the sequence of three charts below the trend of the relative value to increase over the decades.
TEN MOST VALUABLE PATENTS ISSUED IN 1983
Patent Issued Relative Value When Issued Assignee Technology
4,399,282 8/16/1983 1,343 Kabushiki Kaisha Yakult Honsha Pharma/Bio
4,375,514 3/1/1983 1,256 Schering, Aktiengesellschaft Pharma/Bio
4,372,948 2/8/1983 974 Kureha Kagaku Kogyo Kabushiki Kaisha Pharma/Bio
4,374,829 2/22/1983 661 Merck & Co., Inc. Pharma/Bio
4,396,617 8/2/1983 660 Duphar International B.V. Pharma/Bio
4,399,276 8/16/1983 605 Kabushiki Kaisha Yakult Honsha Pharma/Bio
4,369,189 1/18/1983 551 Union Carbide Corporation Pharma/Bio
4,410,537 10/18/1983 507 Burroughts Wellcome Co. Pharma/Bio
4,399,148 8/16/1983 499 Union Carbide Corporation Pharma/Bio
4,372,953 2/8/1983 490 Otsuka Pharmaceutical Company, Limited Pharma/Bio
TEN MOST VALUABLE PATENTS ISSUED IN 1993
Patent Issued Relative Value When Issued Assignee Technology
5,252,474 10/12/1993 1,696 Merck & Co., Inc. Pharma/Bio
5,256,558 10/26/1993 969 The Trustees of Rockefeller University Pharma/Bio
5,258,502 11/2/1993 868 Massachusetts Institute of Technology Pharma/Bio
5,268,273 12/7/1993 824 Phillips Petroleum Company Pharma/Bio
5,182,263 1/26/1993 823 Hoffmann-La Roche Inc. Pharma/Bio
5,187,241 2/16/1993 763 International Business Machines Corporation Pharma/Bio
5,262,568 11/16/1993 756 State of Oregon Pharma/Bio
5,198,563 3/30/1993 695 Phillips Petroleum Company Chem/Polymer
5,227,405 7/13/1993 690 Duke University Pharma/Bio
5,196,524 3/23/1993 679 Eli Lilly and Company Pharma/Bio
MOST VALUABLE PATENTS ISSUED IN 2003
Patent Issued Relative Value When Issued Assignee Technology
6,517,866 2/11/2003 3,374 Pfizer Inc. Pharma/Bio
6,566,344 5/20/2003 2,646 Idenix Pharmaceuticals, Inc. Pharma/Bio
6,602,861 8/5/2003 1,252 Research Corporation Technologies, Inc. Pharma/Bio
6,531,282 3/11/2003 1,225 Oligotrail, LLC Pharma/Bio
6,605,606 8/12/2003 1,109 Miravant Pharmaceuticals, Inc. Pharma/Bio
6,665,641 12/16/2003 884 ScanSoft, Inc. Software
6,602,503 8/5/2003 861 Biogen, Inc. Pharma/Bio
6,596,332 7/22/2003 841 Nestec S.A. Foods products
6,602,499 8/5/2003 824 The General Hospital Corporation Pharma/Bio
RE038073 4/8/2003 804 Research Corporations Technologies, Inc. Pharma/Bio
VI. Conclusion
Conventional valuation models are not generally applicable to patents. In addition, where they are applicable, they are labor intensive, and often require a series of assumptions. The PatentValuePredictor model is unconventional in the sense that it applies a modified version of income theory in which a nominal cash flow is determined based upon a macro economic model that applies heuristic rules. It has the benefits of being completely objective, applicable to all patents, inexpensive, and quick. Application of that model to the universe of U.S. patents indicates that the most valuable patents have generally been in the pharmaceutical and biotechnology technology area for the last three decades.