The brokered patent market in 2014

Technology distribution

When we receive a package from a broker, we use the package materials to categorise the package according to our taxonomy of technical areas. It is a two-tiered classification, with 16 general technical categories and 98 sub-categories. As seen in Figure 2, the distribution of the general technologies is heavily skewed towards software and cloud computing. With the recent cases since CLS Bank v Alice, it will be interesting to see whether there is a drop-off in software portfolios on the market.

The most common three categories – application software, cloud computing and system infrastructure – represent almost 50% of packages. The popularity of these categories demonstrates the market’s current focus on internet computing. Additionally, we have observed a significant drop in the proportion of packages in the semiconductor category. Only 10% of packages from the 2014 market were semiconductor packages, down from 20% in the 2013 market. While the numbers declined, it is possible that this does not reflect a change in the overall semiconductor portfolio percentages, as our clients were more focused on semiconductor buying in 2013; this may have skewed our 2013 market dataset towards semiconductor packages. Also, we believe that because of the difficulties in detecting infringement, semiconductor package sales continue to be a challenge.

The good news is that with more packages on the market and increasing diversity, there are assets available from every technology category.

Figure 2. Package distribution by tech category

Shift to smaller packages

In the 2014 market the average package size decreased significantly. The average package contained 12.6 total assets and 7.8 US-issued patents, down from 18 and 9.3, respectively, in the 2013 market. The percentage of packages with 10 or fewer assets was similar from 2013 to 2014 (it increased from 68% to 71%), but there was significant growth in the number of packages containing between 11 and 25 assets. Considering the drop in total patent assets, it does not appear that large packages were simply being broken up into smaller ones. It is possible that the assets that might have been in larger packages are no longer being brought to the brokered patent market.

Figure 3. Frequency of package sizes by total number of assets

Pricing

In the 2014 market we saw that 83% of packages contained pricing guidance. This is a slight fall from the 2013 market, where 87% of packages had pricing guidance. It is likely that this drop is due to our methodology change. Additional packages that we now capture and are outside of our clients’ buying areas are discussed less with brokers; therefore, there are fewer opportunities for pricing guidance to be added after initial receipt of the package. However, the proportion of packages with pricing guidance remains significantly higher than the 56% in the 2012 market. Additionally, we are getting this pricing information more easily than in previous years. Now, brokers often give pricing information to us in the initial email or in the package documentation. Those that do not provide pricing in that format are more likely to provide it in response to email requests, in contrast to previous years when many brokers would provide pricing only via phone calls.

Figure 4. Distribution of asking prices, top and bottom 5% removed

Distribution of asking prices

The distribution of package prices has shifted significantly to lower-priced packages. Sixty-two percent of packages had an asking price of less than $1 million, as opposed to 53% in the 2013 market. Also, this year, we saw no packages with an asking price of $20 million or greater.

Figure 5. Per asset asking price in relation to broker pricing guidance

From the 2013 market to the 2014 market, the average asking price per asset has dropped by 12%, from $305,000 to $269,000. The average asking price per US patent has dropped by almost twice as much – 23% – from $467,000 to $360,000. This – along with the drop in the percentage of packages that sell – shows that the market is softening. We have also heard anecdotally from multiple brokers that prices are falling; therefore, we believe that the drop in asking price is evidence of sales prices falling proportionally. We believe that the actual sales price of sold packages is still 35% below the broker’s asking prices as calculated in last year’s article.

Despite the drop in average asking price per asset, the 2014 market still has a comparable maximum asking price per asset. However, the standard deviation of asking prices has fallen significantly, which implies that prices are beginning to normalise. This normalisation is likely due to the dominant brokers, as discussed above, pricing their packages consistently; but the process is still in progress. The outliers are similar to those in the 2013 market and the standard deviation is still high at approximately 75% of the average asking price (compared to approximately 85% in 2013 market).

Surprisingly, broker provision of EOU is not boosting asking prices as seen last year. In the 2013 market we saw a 27% mark-up for packages where EOUs were provided; in the 2014 market this mark-up is less than 5%. This comports with anecdotal evidence from brokers that EOU is becoming a more standard expectation. In the 2013 market 33% of packages had broker-provided EOUs, while in the 2014 market 47% had them. We are also seeing some brokers ask sellers for upfront fees to cover some of the costs of preparing the offering package, including EOUs.

Figure 6. Asking prices per US-issued patent by tech category

Pricing per asset/per issued US patent

Pricing for public deals is also dropping significantly, but it remains much higher than the pricing guidance on the open market. The ROL Group has compiled a list of public patent sales from 2010 to present. The list comprises 35 transactions involving public companies such as Yahoo!, Kodak, Panasonic and General Electric. These deals include more than just patents and often include licences and non-patent assets in the mix. The average price per asset for all of the deals is $827,000 (2010 to 2014). However, when we review the more recent deals (only deals from 2012 to 2014), we have $673,000 per asset, showing a downward pricing trend in the price of patents.

Despite this downward trend, both the public deal sets are priced significantly higher than the asking prices on the 2014 market. The public deal values are for completed deals only, whereas the asking prices are not sale prices. Historically, we estimate sale prices to be around 65% of asking prices, which means that the separation between open market deal pricing and public deal pricing is even greater than represented in the graph.

Asking prices by tech category

The average asking prices per US-issued patent per tech category were calculated for every tech area with at least five packages on the market after the top and bottom 5% of data points were removed. It is evident that tech categories are strong drivers for asking price, as the category with the highest asking price per asset: communication services demanded 207% of the asking price per US-issued patent for the semiconductor category (see sidebar explaining wireless category pricing). These results do not necessarily correlate with the distribution of the number of packages on the 2014 market by technology category discussed above. Internet computing categories, application software, cloud computing and system infrastructure software are the most common categories on the market and demand a high asking price. However, the fourth most common category, semiconductors, demands the second-lowest asking price. This implies that the high supply of internet computing categories is matched or exceeded by demand; whereas demand for semiconductor packages, which are also in high supply, is significantly lower.

Key diligence data

To run an efficient patent-buying programme, buyers need to set effective buying criteria. These help to eliminate packages that fail to provide meaningful value to the buyer. Typical buying criteria include technology fields, maximum price and remaining life of the key assets.

Last year’s article focused on clients’ pass decisions in the common diligence paths for a package. Through an analysis of packages that were assigned a diligence workflow for a client, the following section examines the changes to that path and the importance of particular criteria.

At 70%, technology fit remains the primary reason for passing (ie, not recommending a bid) on a package. This high percentage is unsurprising, given that a wide net is cast to find packages to present and the technology area is the first and quickest criterion to be evaluated.

Figure 7. Cumulative sales by years from package listing date

“Actual market adoption is too small” (12%, up from 6%) and “evidence of use fails to map properly” (down 5% from 13%) have switched places as the second and third most common reasons for passing on a package. Additionally, the likelihoods of these reasons for passing occurring have changed drastically; “actual market adoption is too small” occurred twice as often as last year and “evidence of use fails to map properly” occurred less than half as frequently. Our clients are more focused on immediate market impact and are therefore quickly skipping packages in smaller markets to avoid spending time to look at the EOUs where the business impact is not sufficiently meaningful.

The other significant change is a decrease in passing due to “unresolved prior art” – 4%, down from 7%. Buyers are being more selective in other categories earlier in the diligence process. This means that fewer packages get to the stage where prior art is analysed and therefore fewer packages are eliminated from consideration for prior art reasons.

Overall, being selective early in diligence benefits buyers because eliminating packages earlier in the process means that they spend less time and money analysing a package on which they are going to pass. EOU analysis and prior art searches in particular are often more labour intensive and therefore more costly than other diligence activities.

Long-term sales view

Switching to a new sample set of all of the unique public packages we have received (961), we analysed package sales by year listed. This sample set includes packages that were analysed in our previous papers on the 2012 market and the 2013 market and 53 packages previously analysed for long-term sales from 2009. Our methodology uses the US Patent and Trademark Office (USPTO) assignments database to identify sales (if at least one patent in a package is found to have a sale assignment, that package is treated as sold). However, this approach does exhibit a lag due to delays in recording assignments. Also, this data is presented on a calendar-year basis rather than a market-year basis.

In our article on the 2013 market, we observed an extended sales process and, in aggregate, the results remain similar to those from last year. We then looked at the data by listing year to see whether the sales cycle from first entry on the market has changed. By comparing the time to sale of sold packages listed in 2009, 2011 and 2013 (timeframe based on receipt date to recorded assignment date of each package), a stark contrast emerges.

We observed that the overall chance of selling a package has fallen from 51% if listed in 2009 to 33% if listed in 2011. A drop from 51% to 33% reflects some of the difficulties that we hear brokers are experiencing in selling packages. Although the market may be robust, it has become tougher. Bear in mind that this sales data necessarily lags behind the actual market (by up to 18 months); we believe that the current market is actually worse.

An additional result of our analysis is that the period in which to culminate a sale has dropped. Considering packages listed in 2011, no additional sales have been recorded in the past year. Further, for a package listed in 2011, only 6% of listed packages sold in the second year on the market, as compared with 15% of 2009 packages selling in their second year. The third year on the market was even grimmer for 2011 packages: only 3% sold, as compared to 13% of 2009 packages. This suggests that buyers are establishing formal evaluation processes and making decisions more rapidly.

Our estimate for packages listed in 2013 was calculated in the following manner. Packages listed in 2013 that had been on the market for more than 12 months – listed January 2013 to May 2013 – determined the sales rate for 2013 packages in the first year on the market. Then, the sales rates from packages listed in 2011 and 2012 were used to determine how quickly sales would fall off in subsequent years.

Sales

For our sales analysis, we used the sold dataset of 164 sold packages (961 listed packages). However, we now look at the same data from the perspective of the date of sale (per the recorded assignment). This allows us to assess what was actually sold during a calendar year (note that data showing a 2014 purchase/sale year reflects only sales of packages received by May 31 2014 and with assignments recorded by July 7 2014).

Figure 8. Percentage of sales with EOUs by sale year

Sales by EOU provided

As discussed in the pricing section above, a broker providing EOU no longer shows an increase in asking price. However, EOU does increase the chance of sale. Packages with EOUs provided accounted for more than half of the sales in the first five months of 2014. This is a distinct advantage considering that most of these sales came from the 2013 and 2014 markets (as defined in the previous sections), where the likelihood of EOU being provided was only 33% and 41%, respectively. Putting these numbers in context, a broker has about a 25% better chance of a sale if there is EOU. In the tougher market, including EOU seems prudent.

Sales by bid due date/receipt date

We also analysed how much time a buyer really has to bid on a package. While there is no pressure to bid quickly on undesirable packages, we wanted to focus on packages that did sell in order to calculate when buyers need to make buying decisions. Table 5 shows that the middle of the distribution tends to range from between four and six months after the bid due date for sales occurring in recent years. However, because this is the date on which the sale is recorded at the USPTO, the bids are in fact placed significantly earlier. Many buyers do not immediately record their sales with the USPTO after the deal closes – the average delay is three months, based on spot-checks of a random sub-sample of the sold packages. If you add an additional two-week timeframe between reaching an agreement and closing the sale, the centre of the curve shifts from five months after the bid due date to closer to one-and-a-half months.

For sold packages, there is a 25% chance that the agreement between the buyer and seller was made on or around the bid due date and a 50% chance that an agreement was made by one-and-a-half months after the bid due date. This is a change from previous years, as buyers must now get their bids in on time or request additional time to respond from the broker if they do not want to miss opportunities.

On the surface, this observation seems to contradict the fact that we see many bid due dates pushed back by the broker, but the rapid timetable for sold packages is masked by the 67% of packages that will never sell and the bid due date getting pushed back on unsold packages. Due to the operationalisation of buying programmes, the packages that do sell are selling fast.

Sellers

In order to determine what types of entity were successfully selling their assets, we categorised all selling entities into types. We found that as expected the sales occurring from January 1 2013 to July 7 2014 were mostly by operating companies, followed by individual inventors. However, there was also a slight increase in NPE sold packages: during 2013 and 2014, NPE sales accounted for 3% more of the market than they did across sales from all years (37.5% above the overall average). We will continue to look for more NPE sales in the next year.

Despite the slight increase in NPEs selling their assets, no NPE sold more than one package. Seven operating companies, on the other hand, were part of the nine entities that sold more than one package from January 1 2013 to July 7 2014. These sales accounted for 21% of sold packages, 32% of sold assets and 35% of sold US-issued patents. More impressively, the average sales rate (sold packages/listed packages) of these nine companies is 69% – more than twice the expected rate of sale on the open market.

Figure 9. Distribution of seller type by sale year

Buyers

Just as we did with the sellers, we classified all buyers into types. We found that the sales occurring from January 1 2013 to July 7 2014 were purchased by a much higher percentage of operating companies (46%) than observed in our previous paper. Last year, we found that 35% of sold packages listed before 2013 were purchased by operating companies. Additionally, purchases by NPEs have fallen from 56% to 38%, while purchases by defensive aggregators have increased from 8% to 15%. The decrease in NPE purchasing activity likely relates to a decrease in buying by Intellectual Ventures while waiting to fund its Invention Investment Fund 3. We have started to see Intellectual Ventures resume buying, so the NPE number may rise in the coming year.

Figure 10. Distribution of buyer type by sale year

Repeat buyers

During this period 59 buyers purchased 112 packages and 14 buyers purchased multiple packages. Repeat buyers account for 58% of the purchased packages, with the top four buyers accounting for 35% of the purchases.

Full market size

We extrapolated our data on pricing, sales and packages to determine the size of the 2014 market. We size the market for deals with fewer than 200 assets in the package sold by brokers to the open market. We estimate $260 million ($283 million last year) in annual sales, with broker fees of about $52 million ($57 million last year). The biggest differences from last year are as follows:

• There are 22% more packages in the estimated market;
• The average assets per package had decreased;
• The sales window has remained at 36 months, although only 1% of total sales are expected to occur after to 24 months; and
• Our estimated sales rate has decreased.

To derive the market size, we first estimated the number of new packages brought to the market in a year (764, as discussed above). We then used the 36-month sales rate (22% sell), recognising that the number could be as high as 32% (predicted above). As noted above, sales rates calculations are delayed by the sales window and assignment data (up to 18 months), so we reduced the expected sales rate based on our current estimates of 22%. We used this same practice in last year’s paper, reducing the sales rate from 50% to 30%, and our estimate was confirmed by this year’s number being 32%. Then, we discounted the sales price from the asking price (65% of the asking price).

We believe that the actual sales price is one of the more difficult areas to estimate because so few transactions are reported. We adjusted the number of US-issued patents per package to 10.45 to take into account the large number of small packages and the pricing differences of big packages versus small packages. While the average number of US-issued assets per package is 12.6, the median is only four and a greater number of smaller packages sell. We believe that the price and number of assets could be 50% below or above this number. Using the adjusted number of issued US patents per package brings us to a total market of $260 million per year.

Using an average commission rate of 20%, the revenue from this market for brokers is $52 million per year.

We back-tested the market size by estimating the average loaded labour rate per broker ($300,000 a year). This gives 173 brokers. Assuming that four brokers work in each brokerage shop, this results in 58 brokerages (our data shows 58). Each brokerage would bring about 13 packages to the market per year. However, our data shows that a few brokers bring many packages to the market, with the majority bringing a few packages.

This 2014 market size estimate seems to reflect a comparative calculation base on the estimated 2013 market size in last year’s paper. Last year we estimated the 2013 market to be 624 packages per year; therefore, we are seeing a 22% increase, to 764 packages, in package flow as discussed above. On the surface, this appears to be a drastic increase in market size, but the number of packages is only one element in calculating market size. Other factors include the smaller package size and lower prices. After accounting for the 16% decrease in average US-issued patents per package and the 12% decrease in asking price per US-issued patent, the result is that the total value of the market decreased by 10% from 2013 to 2014. This would produce a 2014 market size of $254 million (compared to $283 million in 2013).

Opportunities

With more than 60 new packages coming to market each month, the opportunity to find something to fit your needs exists. Eighty percent of the packages have a target asking price below $2 million, allowing companies to purchase assets to address specific problems at a relatively low cost.

We are seeing an even lower percentage of packages selling than last year, indicating a continued downward trend. Corporate buyers continue to be seen as significant buyers and have overtaken NPEs as the greatest buying segment.

We continue to evaluate, bid on and buy patents for our clients and will continue to track the data and refine our database. This being our third year of reporting on our data, and having collected a larger segment of the market data, we have had the opportunity to look into sales trends on a deeper level. As soon as the data started answering these questions, new detailed and in-depth questions presented themselves. We look forward to continuing to answer these.