There is No Patent Bubble, Nor NPE Mana

Shortly after the Nortel transaction and Google’s acquisition of Motorola Mobility in the summer of 2011, some industry observers quickly warned us that patent market was a bubble (here, here and here). The debate over the patent bubble has been going on since then (examples can be seen here, here and here). Some were saying that the patent bubble has already burst (here and here), some saying it’s about to, while still others keeping hailing the booming patent market.

To be sure, all of the concerns over the patent bubble are legitimate, and as always, rational debate is beneficial to the healthy development of patent market. There is no doubt that most of the opinions expressed were based on the observers’ experience and the information available to them at the time. Unfortunately, unlike in the well-established financial markets where transaction information and price data are mostly available for research and analysis, the prices and deal terms in patent transactions are usually kept secret by the parties. Except for meeting certain regulatory requirements (such as SEC filing in the US) for publicly-traded companies, there is usually not much additional motivation for the parties to release the prices and deal terms in patent transactions.

The lack of disclosure leads to the scarcity of data, and what comes with the scarcity are the incompleteness and obscurity, all of which lead to misinterpretation of the data and information. More importantly, misinterpretation, in turn, can lead to mispricing and market inefficiency when the misinterpreted data is applied to value patents for transaction. For example, after the Nortel transaction and Google’s acquisition of Motorola Mobility, some observers noticed that both deals were concluded on a $750K per patent basis. Therefore, as the story goes, market price per patent was about $750K per patent.

Obviously, the basket of assets that Google acquired for $12.5 billion, which included both IP and other tangible/intangible assets, is quite different from the 6000 patent and patent applications Nortel sold. Also, as discussed in some commentaries, pricing of both deals largely reflected the dynamics and strategic concerns leading to the transactions, which were mostly specific to the parties in the deals (see here and here). This raises many interesting questions, not only regarding how to interpret Nortel and Google transactions specifically, but more generally, about how to interpret and apply market prices for patent portfolio valuation. For example:

• Has patent market pricing changed significantly since Nortel and Google-Motorola Mobility deals?
• Is per patent price meaningful across different transactions?
• Can a simple average price per patent be applied to other transactions?
• How to adjust strategic value specific to certain deals to derive a more “reasonable and fair” price that is meaningful and useful for other transactions?
• How to adjust other factors such as industry differences; seller/buyer organization type; patent vs. patent applications; and a wide variety of other payments and considerations such as licensing back, options to purchase, covenant not to sue/not to compete, product purchase payment scheduling and financing etc.?

In an effort to address some of the issues above, I started a project to collect and analyze the patent sales data and information. It is an on-going project with the following long term goals:

• Analyze and interpret the price information in patent market transactions;
• Decompose price data to identify value components and to quantify component premiums and discounts;
• Derive fair market price based on adjustments made for various premiums and discounts;
• Use the model and insights derived from the analysis to value patent portfolios.

More samples will be added to the data pool, and analysis and results will be released periodically. The complete analysis based on the data collected as of mid-September 2012 will be presented at Workshop 3-K of 2012 LES (USA & Canada) Annual Meeting at Toronto, 10/14-10/17, 2012. This article will focus on two of the major topics discussed during this most recent round of analysis, the patent bubble concern and the role of nonpracticing entity (NPE) in patent sales market.

Detect Bubble in Patent Sales Market

Prior to analysis, the data has to be processed appropriately, and various adjustments have to be made to reflect the economics underlying the transactions. First of all, the payments are adjusted by inflation using the CPI indexes as of the transaction dates and those in June 2012. Second, a net payment for the patent portfolio needs to be estimated. This involves different adjustments based on accounting and financial data released. The step is especially important for the patent portfolios transacted as part of mergers and acquisitions or other assets-package sales.

One of such examples is Google’s acquisition of Motorola Mobility mentioned earlier in this article. After the announcement, some of the observers simply took the total payment of $12.5 billion and divided it by 17,000, the number of patents, thereby reaching a per patent price of $746 thousand. However, Google acquired the company’s operating assets and the patents are only part of the basket of the assets, although a significant part. One of the analysts estimated the fair market value of the patents as about $4.5 billion. According to Google’s SEC filing, however, the basket of “patents and developed technology”, including patents, patent applications and other forms of technologies, was worth $5.5 billion in fair market value. Therefore, adjustments have to be made accordingly for the Google-Motorola Mobility deal to be included in the analysis.

After the adjustments, a per patent price is calculated for each transaction with the data available in payment and number of patents. The weighted average price is computed as the sum of the payments divided by the sum of the numbers of patents across all of the portfolios analyzed. To test the hypothesis that the Nortel transaction had fundamentally changed the pricing in patent sales market and might have caused patent assets bubble, the samples of patent transactions are divided into two groups, a pre-Nortel group and a post-Nortel group. Table 1 below summarizes the basic descriptive statistics of the two groups.

As shown in the table, the median and average per patent prices of the pre-Nortel deals were significantly higher than those of post-Nortel ones.  However, the weighted average price of the post-Nortel transactions was higher than that of pre-Nortel ones, indicating that the data in the post-Nortel period might have been skewed by a few much larger and more expensive portfolios transacted. However, based only on the data in the table, the hypothesis ofa patent bubble in post-Nortel period can’t be rejected nor validated.

NPE in Patent Sales Market

Further efforts to test the hypothesis shifted the research focus to another interesting phenomenon in the debate over patent bubble, that is, the complete absence of NPE in the discussions. As one clicks through the links at the beginning of this article, it is easy to see that the discussions unanimously traced the same origin of the patent bubble, that is, the patent race among large practicing companies. It is a little surprising, especially in light of the frequently-seen and mostly negative coverage about NPE’s role in other major areas of IP business such as licensing and litigation.

There is no doubt that NPEs have played an important role in patent sales market. As summarized in an earlier study by Santa Clara University Law School Professor Colleen Chien, overwhelming majority of the patents in the market before 2010 was sold to NPEs. Also, the Knowledge@Wharton article cited above actually compared the roles of NPEs and practicing companies played in the market before and after Nortel transaction. The article concludes that the bull patent market was fueled, not by NPEs (or patent trolls as called in the article), but mainly by the “mutually assured destruction between combatants in competitive industries”.

Now, the question is, is the inconclusive hypothesis test in patent bubble caused by the differences in pricing behaviors between NPEs and non-NPEs? Conceptually, it is certainly possible. To further explore this possibility, the samples of patent transaction are divided into two categories, non-NPE and NPE; and then within the NPE category, two sub-groups of NPE buyer and NPE seller. The basic descriptive statistics are summarized in Table 2.

The statistics in the table indicates that the average prices of the deals with non-NPE parties are two to three times of the prices of those with at least one party being NPE. Especially, NPE buyers seem to pay average prices that are closer to what non-NPEs are paying, while NPE sellers are likely to receive the lowest prices among all market players. Reading the data in Table 2, it is tentative to conclude that the inconclusive hypothesis test in patent bubble might have been caused by the fact that the higher prices realized in the patent race among non-NPEs were offset by the lower prices paid or received by NPEs.

However, it worth to noticing that the basic descriptive statistics in the tables above may not be able to reveal the important effects many other factors might have on the prices of the patent portfolios transacted. In other words, the basic statistics can explore the data only in one dimension, such as pre- and post-Nortel or NPE vs. Non-NPE. While this one-dimensional analysis is helpful, it fails to address the impacts of other factors as listed earlier in this article, including strategic value, industry difference, patents vs. patent applications, and other payments and considerations. Obviously, a new approach is needed.

Econometric Analysis and Conclusions

The most important step of this research project is the econometric analysis on the patent sales data. A hedonic-model-like specification is designated to decompose and adjust the effects of various factors. In addition to the number of patents, strategic factors, industries, and other relevant variables mentioned above, the model also includes two dummy variables to specifically address the issues discussed in this article: a time dummy variable to separate those deals done before and after the Nortel transaction, and an organization type dummy variable to indicate the status of NPE or non-NPE. Finally, to obtain insights about any possible differences between NPE seller and NPE buyers, two additional dummy variables are introduced in the model.

The discussions about the modeling and the conclusions will be presented at Workshop 3-K of 2012 LES (USA & Canada) Annual Meeting at Toronto, 10/14-10/17, 2012. The major conclusions related to this article are as follows:

1. After adjusting various other factors, the coefficient of the time dummy variable is not statistically significant, indicating that Nortel deal did not fundamentally change the market pricing of patent portfolios. In other words, patent market has not been a bubble.
2. Although the basic statistics in Table 2 point to the possibility that the inconclusive hypothesis test in patent bubble might have been caused by the offsetting effects in pricings between NPEs and non-NPEs, the econometric analysis does not support this possibility. In other words, after adjusting the effects of all other factors, there is no difference in pricings between the transactions with at least one part being NPE and those with both parties being non-NPEs.

Finally, the analysis in this study offers further support to the conclusions I reached in my recent NPE researches (here and here) and in one of my recent presentations, that is, NPE is simply a business model, and there is no systematic evidence to prove that NPEs behave differently than other players in the licensing market and patent sales market. This said, regarding other studies based on the litigation-related data, such as the ones claiming that NPEs have caused substantially high direct and indirect costs to non-NPEs (here and here), I have no empirical evidence to make additional comments, besides those made in my NPE papers cited earlier in this paragraph. However, two of the recent studies by Professors David Schwartz and Jay Kesan (here and here) do offer very compelling analysis about the data and methodological issues in the NPE cost studies based mainly on the data in NPE-related litigations.

Disclaimers and Acknowledgments

The views expressed in this article are the author’s, not those of Applied Economics Consulting Group, Inc. or the data providers. The time spent on the research leading to this article was not charged to Applied Economics Consulting Group, Inc. or its clients.

I would like to thank RoyaltySource and several colleagues and friends for the help in data collection.

Bound by NDAs, data of each individual transaction will not be disclosed. Analysis of the aggregate data will be released periodically.