An augmented market approach to patent portfolio valuation

Demand is increasing for quick, easy and relatively reliable methods of patent portfolio valuation in patent-intensive corporate transactions. An augmented market approach may be the solution

A little inaccuracy sometimes saves tons of explanation.

– Saki (Hector Hugh Munro)

In everything from mainstream media to expert publications, the price of a patent portfolio is frequently inferred by referencing a per-patent valuation implied in other comparable transactions – the so-called ‘market approach’.

While market approaches which use comparable transaction data to derive the value of a subject asset or business have been commonly adopted by financial professionals, applying them to patent portfolios is much more controversial and is in stark contrast to mainstream practice among IP professionals. For example, to value just one patent, IP professionals usually engage a technical crew to reverse engineer a potential infringing product and then work at a granular level, essentially word by word, on claim charting. For such professionals, it is inconceivable that a patent could be valued without claims, prior art and basic patent quality metrics (eg, backward and forward citations, number of claims and remaining life term) being considered.

Given this, many IP professionals have been highly critical of value indicators such as per-patent value. First, they believe that such indicators are meaningless because the value distribution of patents is highly skewed. Indeed, the values of individual patents vary substantially in a large portfolio, with a few key patents or core patents being worth significantly more than others and a sizeable portion worth nearly nothing. Second, even if per-patent value is meaningful for one patent portfolio, they insist that any per-patent value derived from one portfolio should not be applied to another, because no patents are identical, let alone patent portfolios.

There is no doubt that these criticisms and concerns are legitimate. For patent litigation and assertion, an IP practitioner would have to analyse each patent, make claim charts and conduct infringement analysis in as detailed a manner as possible. However, such a comprehensive analysis may not be required or justified based on cost and time constraints in other areas of IP practice. For example, only a handful of key patents will be claim charted and the overwhelming majority of these will simply be included as part of the portfolio in licensing and cross-licensing – especially if the subject technology areas or product categories are covered by a large portfolio.

More importantly, as patents and other IP assets play an increasingly important role in today’s economy, demand is increasing for cheaper, quicker and relatively reliable methods of patent valuation, especially in patent markets and corporate transaction markets.

As patent markets become increasingly developed, they offer a more efficient platform for patent owners and buyers to divest or invest in patent assets. In the early stages of negotiation for a patent transaction, neither buyer nor seller wants to devote significant amount of resources to claim chart patents or comprehensive due diligence; yet both parties are eager to know how much the subject patent portfolio would be worth in the current market. The market approach can offer a cost-effective way to conclude a preliminary valuation before either party commits additional resources.

In corporate transaction markets, technologies – including those covered by patents – are among the key motivations underlying deals such as mergers and acquisitions, private equity or venture capital financing and distressed assets divestiture or investment. Patent assets have been attracting more attention from corporate executives, investors, bankers and securities analysts, meaning that parties to negotiations encounter valuation issues in patents and technologies more frequently. While valuation reports with detailed analysis backed by claim charts and IP due diligence might be preferable, budget constraints and the highly limited time window render such analyses highly implausible. However, they do leave the door open for the market approach.

This article develops a practical procedure for patent valutation using the market approach. It first discusses the economic issues with this approach generally and those arising from its application to patent valuation specifically. It then reveals some insights into the pricing behaviour of patent portfolio transactions in an effort to lay an empirical foundation for the proposed method. Finally, it presents a three-step valuation method under the market approach for patent portfolio valuation.

Economics and methods of market approach

Valuation methods can generally be categorised as cost approaches, income approaches or market approaches. For patent valuation, the cost approach may be inappropriate – not only because identifying and allocating historical R&D costs to a specific patent portfolio is usually impracticable, but also because historical costs reveal little about the market value of patent assets today. The commonly used methods for asset valuation are either income approach or market approach. Income approach in the fields of finance and investment is also referred to as direct or absolute valuation, as it determines the valuation as the present value of the expected future cash flows or income through a discounting process to reflect the risk inherent in the assets.

Market approach in investment banking and security analysis is called indirect or relative valuation. It is based on the economic principle of no arbitrage, which dictates that similar assets which provide similar future cash flows and have a similar level of risk will sell for a similar price. Relative valuation determines value based on certain multiples, derived from a selected group of comparable transactions. For this reason, market approach or relative valuation is also known as the comparable company (or asset) method or multiple-based approach (or method of multiples).

While the procedures used by practitioners may vary, the market approach to valuation can generally be structured as a three-step process.

Identify a set of transactions comparable to subject assets

Transactions in the same industry sector as that of the subject assets are usually perceived to be comparable. For patent assets, this means that patent transactions in the same technology areas or field of use are preferred. However, valuation economics define ‘comparability’ as similarity in risk level and cash-flow pattern, which means that a market approach should not exclude transactions from other industry sectors or technology areas, as long as they demonstrate similar levels of risk and similar patterns of future cash flow.

Calculate price or value multiples by standardising valuations

The absolute amount of a valuation cannot be compared between comparable transactions or between comparable transactions and subject assets. Depending on how valuations are expressed, a practitioner can choose from various financial measures to standardise the valuations or compute the multiples. Financial measures include income statement items such as:

  • revenue, earnings before interest, taxes, depreciation and amortisation (EBITDA), earnings before interest and tax (EBIT) or net income;
  • a balance-sheet item such as a book value for equity or assets; or
  • certain cash-flow measures such as free cash flow to firm (FCFF) or free cash flow to equity.

The most common multiples in finance and investment include price to earnings ratio, price to book ratio, price to sales ratio, enterprise value (EV) to EBITDA ratio and EV/FCFF, among many others. A multiple can be interpreted as the price or value per unit of the financial measures. For example, an EV/EBITDA ratio of four means that each dollar of EBITDA is traded or sold at $4 in the market.

Non-financial measures are also used by investment bankers and securities analysts. For example, a multiple can be calculated as the ratio of acquisition price (in an M&A deal) to the number of subscribers of the target (price per subscriber) or as the ratio of market cap (or EV) to active users (price or EV per active user).

Naturally, for patent portfolio transactions, the multiple can be expressed as the ratio of transaction price to the number of patents in the portfolio or price per patent. Theoretically, and as shown in the empirical analysis section below, a per-patent or per-asset price can be meaningful and useful if it is derived from a large number of transaction samples. While no patents are identical and value distributions across different patent portfolios can vary dramatically, the calculated average price per patent based on a large size of samples in patent portfolio transactions shall converge on the theoretical or expected value per patent.

Apply representative multiples to financial or non-financial measures

In finance and investment, an analyst will adjust certain financial measures to account for extraordinary or other one-off events and may also adjust the valuation upwards or downwards based on the premium or discount associated with certain factors such as marketability, control status and size.

Similarly, when using a per-patent price calculated from comparable transactions, a practitioner should control and adjust for the valuation effects of various features and other factors specific to the subject patent portfolio.

Table 1Data of patent asset transactions – a general description

 

Number of samples

Patent sales/acquisitions

240

Of which with known counts of issued patents

215

Patent asset transactions with back-end payment

27

Patent application transactions

8

Total

275

Table 2Market transaction prices – per patent versus per asset

$

Sample size

Median

Weighted average

Patent asset portfolios

N=240

$203,000

$231,000

Patent portfolios

N=215

$254,000

$381,000

Single patent portfolio

N=48

$308,000

$924,000

Insights from deconstructing patent transaction prices

Before constructing a reliable method under the market approach for patent portfolio valuation, it is necessary to understand the pricing behaviour of the patent market. For example, what is the relationship between the number of patent assets and the prices? What are the major value components and which are contributors and which value destroyers? This section presents the most recent results from an ongoing research project which collects and analyses the market prices of patent portfolio transactions. The project was launched in 2012 as part of an effort to understand the pricing behaviour of the rapidly growing patent market.

All of the transactions collected were from publicly available sources; no confidential information or data was included in the study. Most of the samples were obtained through online searches of regulatory filings, news reports, analyst reports and other public sources. Other significant sources included major patent brokerage firms and data vendors – mainly ktMINE, IPOfferings and RoyaltySource. For a transaction to be included in the analysis, the payment and the count of patent assets in the portfolio must have been reported. Best efforts were then applied to collect other relevant information, including the time of the transaction, the type of seller and buyer, strategic intention, industry, technology type, the seller’s financial health, licensing-back clause, M&A deals and patent versus patent applications, among others.

Some of the data sources disclosed only an aggregate number of patents and patent applications or patent assets, and no exact count was reported for issued patents or patent applications. To accommodate this data issue, the project defines ‘patent assets’ or a ‘portfolio of patent assets’ to include both issued patents and patent applications. By contrast, ‘patents’ or a ‘portfolio of patents’ includes only issued patents.

Descriptive statistics

As of the end of 2015, the database created by the project had 275 samples of market transactions in patent asset portfolios, with the earliest samples dating back to the 1990s. As shown in Table 1, the vast majority of the records are straight patent asset sales or acquisitions. Also included are nearly two dozen patent-selling or privateering deals with significant back-end payments, nearly all of which involved patent transfer or exclusive licensing to non-practising entities (NPEs) for monetisation. The project has also started to collect data on transactions with patent applications only and several samples have been added to the data set.

While deals involving thousands of assets and millions or even billions of dollars tend to dominate the headlines, the patent market transaction data over the past two decades tells a different story. The median portfolio involves 11 assets, while the median price is about $2.43 million. Similarly, for deals with known counts of issued patents, the median number of patents is seven and the median price $1.83 million. In fact, 29% of patent asset portfolios have three assets or fewer and 38% of patent portfolios have three patents or fewer.

For the purpose of analysing patent portfolio prices, our analysis focuses mainly on transactions that involve at least one issued patent, with the back-end payment representing an insignificant portion of the deal value. Among the 240 such transactions in patent assets, 215 deals have known counts of issued patents. Table 2 summarises the median and weighted average price per patent and per asset.

The prices reported in this table (and in the rest of this section) are inflation adjusted, reflecting the dollar value as of December 2015. Based on the historical data, the median price per asset is $203,000 and the weighted average, $231,000. For transacted portfolios with known counts of issued patents, the median and weighted average price per patent is $254,000 and $381,000, respectively. Given that there were sizeable samples of deals which transacted only one issued patent, we were able to conduct meaningful analysis of such portfolios. There are 48 single-patent portfolios and the median price is $308,000 and average of $924,000, as reported in Table 2.

Finally, as shown in Figure 1, market prices of patent transactions fluctuate greatly over time, caused mainly by significant events in the IP industry and overall economy, as discussed below. IP valuation practitioners should bear this in mind when selecting comparable transactions. For example, to value a portfolio for sale or acquisition, the samples from the most recent transactions would be preferred. By contrast, when a patent damage expert tries to estimate the market price for a hypothetical negotiation years back, the market price back then should be used.

Figure 1. Patent market pricing since 2011 – median per patent versus per asset price

Econometric analysis

A hedonic model-like specification is designated to establish a relationship between the number of patent assets (or patents) and price, identify major value components and quantify the value impact of the components. Two econometric models are created for the analysis: one for patent assets transactions and the other for patent transactions. The dependent variable is transaction price, while independent variables include the counts of patent assets or patents, the organisation type of seller and buyer, strategic intention, industry, technology type, seller financial health, M&A deals, and patent versus patent applications, financial market index and major events (eg, the American Invents Act and Alice), among others. Except for counts of patents or patent assets and the NASDAQ index, all other variables enter the models as dummies.

With R2 of 83% ~ 84% for both models, the specifications seem to have explained the vast majority of the variances of the market prices of patents and patent assets; although for some independent variables, the significance levels of the coefficients vary across the two models. The major conclusions from the econometric analysis are summarised below.

Non-linear pricing: The non-linear relationship between the transaction price and the count of patents or patent assets is statistically significant. In other words, when the count of patent assets increases, the transaction price increases accordingly, but at a slower pace. This has a critical implication for patent portfolio valuation: the per-patent price derived from a very large portfolio should not be directly applied to valuing a much smaller one and vice versa.

Based on the econometric models, there are two ways to deal with this non-linear effect. First, use the econometric valuation model to value the portfolio and let the model take care of the non-linear effects and the effects of all other possible variables.

Alternatively, non-linear adjustment factors (NAFs) can be used to make a non-linear adjustment. In certain situations, we would have to apply the per-patent price derived from one portfolio to another with a dramatically different size. With the econometric valuation models, this can be accomplished in two steps. First, use the econometric models to generate a NAF based on the number of assets in the comparable (or benchmark) portfolio and the number of assets in the subject portfolio. Second, multiply the per-patent price derived from the benchmark portfolio by NAFs and then apply the adjusted price to the subject portfolio.

Technology premiums: Statistically significant price premiums are associated with several types of technology, including software, wireless and medical and pharmaceutical technologies – although the significance levels vary slightly across the patent model versus patent assets model.

Strategic premium: Patents or assets with strategic value to particular industries carry significant premiums. Examples of strategic patents include litigated patents, blocking patents, standard-essential patents (SEPs) or a large portfolio covering a wide range of technologies practised in an industry.

Licensing-back discount: Transactions that granted the seller a non-exclusive licence to the patent assets were priced with discounts. This makes sense from an economic point of view, because a seller’s use of the transacted technology would compete with revenue from the buyer.

Financial distress discount: The assets of sellers that were in financial distress or even in bankruptcy proceedings were usually sold at a discount. Most likely, the discount was a result of the seller’s desire to sell the assets quickly to meet liquidity demands or to satisfy requirements from creditors and new investors.

Another possible reason might be that the seller’s financial problem actually reflects the poor quality of the patent assets and bad patents are sold for less. However, literature in the relevant research fields, such as corporate financial distress and technology company failure, demonstrates that technology is only one of many factors – and most likely the least prominent – which can cause financial distress or tech company failure. When technology is among the important reasons, it is more likely that the technology is simply too far ahead of the curve. As a result, the argument that a patent owner’s poor financial performance implies weak patents should be taken with a pinch of salt.

NPE effect: NPEs are active players, as both buyers and sellers, but their pricing behaviours are usually unremarkable, except for one major player. This specific NPE was an extremely active buyer before 2012 and was able to acquire patents at prices that were significantly lower than those paid by other buyers. Overall, there is no evidence that NPEs are overcompensated. In fact, deals involving NPEs may have been underpriced – although this effect is not statistically significant.

America Invents Act effect: The America Invents Act was enacted in September 2011. To test whether it had any significant impact on the patent market, a dummy variable was introduced with a value of one for the months of October 2011 to June 2014, and zero for all other months. This analysis shows that the act significantly depressed the transaction prices of patent assets.

Alice effect: The US Supreme Court’s decision in Alice v CLS Bank on June 19 2014 sent shockwaves through all segments of the patent market. Similar to the method for testing for the effect of the America Invents Act, a dummy variable was designated, with a value of one for the months from July 2014 to December 2015. Both models conclude that Alice put significant downward pressure on the market price of patent portfolios, leading to much lower patent prices during the post-Alice period.

Augmented market approach to patent portfolio valuation

Guided by the economics of the market approach and the insights gained from this empirical analysis, we can formalise a method of patent portfolio valuation based on the prices from market transactions of patent portfolios. While this new method generally falls under the framework of a market approach, it is different from traditional market approach procedures – especially given that it is augmented or enhanced by econometric valuation models developed from market transactions. For this reason, the proposed method is referred to as an ‘augmented market approach’.

Figure 2 illustrates the three-step process of the augmented market approach.

Figure 2. Augmented market approach to patent portfolio valuation

Comparable transaction valuation

This step essentially reflects the traditional market approach, in the sense that this is the point at which valuations are concluded, based mainly on the per-patent prices calculated from comparable transactions. As discussed earlier, where possible, a practitioner should select the most recent samples of transactions to best reflect the current market sentiment. Depending on the data availability and sample size, it is recommended that three sets of comparable transactions be compiled for per-patent price calculation.

The first are samples that are controlled for size – due to the non-linearity revealed above, a practitioner should select a set of patent transaction samples that are of similar size to the subject portfolio. The samples should include transactions from all technology areas or industry segments, as justified previously.

The second are samples that are controlled for technology areas and/or industry segments – all transactions in similar technology areas and industry segments should be included in this set of samples, regardless of the size.

The third are samples that are controlled for both technology and size – the valuation determined from the per-patent price calculated from this set of samples is deemed to be the most indicative. However, due to the limited data availability, the sample size of this set can be relatively small.

If the transaction samples of similar size to the subject portfolio are extremely limited, it may be necessary to use samples of substantially different sizes and then adopt the NAF steps highlighted earlier.

Econometric model valuation

The second step of the augmented market approach calculates patent portfolio valuations using econometric valuation models, which are developed from the market transactions of patent portfolios. Depending on the features and characteristics of the subject patent portfolios, the following data items will be used as the inputs to run such econometric valuation models:

  • number of patents and patent applications;
  • patent ranking information which indicates the number of high-quality patents – other information may also be helpful, including the litigation status of the patents, number of blocking patents or SEPs;
  • technology areas or industry sectors;
  • seller’s financial health;
  • specific terms such as licensing-back and contingent payment; and
  • other information such as seller/buyer organisation type and involvement of brokerage firm.

The effects of industry-wide events and other possible economy-wide factors will be controlled by the econometric models.

Valuation determination

Based on the valuations obtained from the two steps above, a valuation or a range of the most likely valuation scenarios can be determined. Evidently, there will be other factors which are specific to the subject portfolio, and the effects of such factors are not reflected by the comparable transactions or controlled for in the econometric valuation models. As a result, further adjustments shall be made to take into consideration the valuation impact of these factors. These may well include any specific factors associated with the patent portfolio (eg, the remaining life term and prosecution history). It is also wise to take account of the seller’s requirements (eg, a seller may not want to deal with NPEs or may want to sell the patents to US companies only due to legal or regulatory issues) and the buyer’s status (eg, a buyer may have a business or strategic relationship with the seller, or the subject transaction may be negotiated as part of a large patent transaction or corporate transaction).

Action plan

The augmented market approach to patent portfolio valuation saves a great deal of time without sacrificing a significant amount of accuracy. It is a three-step process which consists of:

  • comparable transaction valuation based on per-patent prices calculated from three different sets of comparable transaction samples which are controlled for size, technology area or industry sector, and size and technology or industry at issue;
  • econometric model valuation based on econometric valuation models developed from actual transactions of patent portfolios, which usually require the following data items as major inputs:
    • number of patents and patent applications;
    • patent ranking, litigation status, blocking patents or SEPs;
    • tech fields and buyer or seller organisation type; and
    • seller’s financial status, licensing-back and brokerage involvement; and
  • valuation determination based on the valuations concluded from the comparable transactions and the econometric valuation models, further adjusting for factors specific to the seller, the buyer and the subject portfolio.

Jack Lu is a partner and chief economist of Intellectual Property Market Advisory Partners (IPMAP), LLC, Austin, Texas

Disclaimers and acknowledgments

The views expressed in this article are the author’s, not those of IPMAP, LLC or the data providers. The author would like to thank IPOfferings, ktMINE, RoyaltySource and several colleagues and friends for their help in data collection. Due to non-disclosure agreements, the data of each individual transaction will not be disclosed. Analysis of the aggregate data will be released periodically

Get unlimited access to all IAM content