Friday, 16 November 2012

Patent Thickets

Patent thickets
I have been very kindly invited to comment on Research into the use of Patents, based on papers produced and sponsored by the IPO , a conference today at the Big Innovation Centre.
Here are some notes on " A Study of Patent Thickets By Bronwyn Hall  , Christian Helmers , Georg von Graevenitz , Chiara Rosazza‐Bondibene. It's great stuff and this is my take.
  1. Their exam question is this:
    "This study investigates a question posed by the UK Intellectual Property Office concerning patent policy and SMEs: Are patent thickets a barrier to entry and how do they affect small and medium sized enterprises (SMEs)?"


Page 45 amplifies what they are looking at very clearly
"one of the functions of the patent system is to allow inventors to exclude others from practicing their invention. The implication of this fact is that in technology areas where there are large numbers of patents, it might be more difficult for new firms to enter because the technology space is effectively covered by patents held by existing firms. By itself, this is not necessarily a phenomenon requiring some kind of policy intervention, as it is to be expected if the patent system is doing its job. However, in sectors where firms must draw on technologies for which their competitors hold patents in order to produce, it is possible that the presence of many overlapping patents held by incumbent firms could discourage the entry of new firms with novel ideas, because such entry requires negotiating access to a prohibitively large number of other technologies in order to incorporate their invention(s) in a product."

So what do they study? Again, page 46 is helpful here
In order to capture the idea that some sectors may be characterized by collections of patents held by different firms, but at least some of which are jointly required for production, we use the previously described measure of patent thickets developed by Graevenitz et al. (2012, 2011), henceforth vGWH. The idea of this measure, which is based on patent applications to the EPO, is that it can proxy for the extent to which a sector contains many patents with possibly overlapping claims. Because it identifies situations where groups of firms are applying for similar patents that potentially block each other, it identifies technology areas where there is active patenting by existing firms that have strategic relationships with one another. As argued earlier, such technology areas are usually those where products are also complex and draw on technologies held by multiple firms. The inquiry we undertake here is whether UK firms are discouraged from entering such technology areas. Therefore we examine the influence of this measure on the probability that a UK firm enters a technology sector, where entry is defined as the priority year of the first patent in the relevant technology sector that is applied for at the European Patent Office (EPO) or the UK Intellectual Property Office (UKIPO). The sample we use for estimation includes all the firms with at least one patent application at the IPO in the UK or the EPO during the 2001‐2009 period.



  1. As they note, entry by young firms is key for growth: Miranda, Halitwanger and Jarmin. So this is critical.


  1. There's been a huge rise in patent filings


  2. In some important areas, technology is getting more cumulative. That means that patents can hold up future innovation. Increased use of standards is one example.


  1. Patent assertion entities.

Here the evidence, as I read it, is mixed

Here the evidence, as I read it, is mixed
Recently, there has been an enormous increase in patent infringement cases filed by patent assertion entities (PAEs) in the US.14 The increase as well as number of high‐profile cases, such as NTP vs. RIM or Eolas vs. Microsoft, triggered a heated debate on the role of PAEs in facilitating the so‐called market for technology.
Recent empirical evidence by Tucker (2011) suggests that PAE litigation has a negative effect on innovation carried out by alleged infringers.
Helmers and McDonagh (2012) look at patent cases at the Patents Court for England and Wales that involve PAEs . In contrast to the US, they cannot find any significant increase in the number of cases involving PAEs in the UK over the 9‐year period 2000‐2008. They find that only in one minor aspect of the cases a PAE was successful in asserting infringement.
In fact, most of the cases before the Patents Court that involve PAEs are cases in which manufacturers successfully seek the invalidation of patents owned by PAEs.
If low patent quality is associated with patent thickets, this would imply a link between thickets and PAE litigation. That is, if low quality patents provide incentives for PAEs to acquire and assert such patents, we would expect to see PAEs to assert relatively more patents in areas in which patent thickets exist.
Other findings are however seems to be that these entities acuatlly take out high value patents. So this is not clear.


Nonetheless an important conclusion
PAE litigation  occurs  mostly  in  technological  fields  that  are  also  affected  most  by  patent thickets, such as the information and communication technologies.  


  1. More astonishing, to me at least, is the evidence eon patent offices#
    Quillen et al. (2009) indicate that a comparison of the USPTO application backlog at the end of 2008 with the Net Disposal rate in 2008 a 60 month examination backlog emerges


    The most recent figure comparable to those provided by Harhoff and Wagner (2009) suggests that average duration of examination at EPO has increased to 5.24 years for 2011.


  1. And here is an amazing stat.
USPTO patent examiners spend on average only 18 hours working on each patent over a period of three years.


  1. They then need a measure of thickets:
This is very hard. I think idea is this. Firm A submits a patent, patent 1. The EPO does a search on anything that might be "prior art" that is, that might be a source (another patent say, calling the patent into question). These sources are called Type X or Type Y documents. For patent 1, this search is a published list of e.g. other patents or other documents.
A "triple" is a case where there are three firms, A, B and C. Each of those firms owns some part of the "prior art", that is, each of the other firms appears on the list. This then implicitly limits the claim on the other two recent applications of each of the two other firms.



  1. What do they find? Table 2 finds triples are very high in semi-conductors, telecoms and digital comms.

  1. There has been big growth in triples




  2. What is the effect on entry?
Here's where it gets tricky. They take company accounts data. But the vast majority of firms don't patent, so they have to get rid of many of those firms. So they end up with 29,000 firms who can potentially enter into 34 technology sectors, 2002-2009. They don't look at whether a new firm enters industry X. That's hard to do, since the firms are often classified by SIC not product (e.g. all transport, not domestic airline). , Rather they look at whether a small or large firm patents for the first time in one of the 34 technology sectors. A visual summary of what they find is in their Figure 13


Note that the red line is in the negative part of the y axis. They summarise the picture as follows:
Figure 13, overlays the coefficients of aggregate patenting and triples density as a function of firm size on the actual size distribution of our firms. From the graph one can see that (my bullet points on their text)
  1. "The impact of aggregate patenting in a sector is higher and more variable than the impact of the triples density.
  2. Firms in the lower range of the size distribution (assets less than 10,000 pounds) are much more likely to enter a sector with high aggregate patenting if they enter at all, but
  3. their hazard of entry falls 15 per cent if the triples density doubles in that sector.
  4. On the other hand, for the few firms in the upper range of the size distribution (assets greater than 100 billion pounds), the hazard of entry falls only 7 per cent if the triples density doubles."


Finally, they also document that the effect on entry of a 1 sd change in the number of triples is very concentrated in the high triples classes e.g. telecoms:




  1. They are then very nuanced in their conclusion.


Our results suggest a substantial and statistically significant negative association between the density of thickets and the propensity to patent for the first time in a given technology area.
As we find thickets to affect entry negatively, there is a strong indication that thickets represent some kind of barrier to entry in those technology areas in which they are present. However, we must emphasize that the simple finding of a barrier to entry created by patent thickets is not proof positive that reducing that barrier and increasing entry would lead to welfare improvements in the innovation/competition space. Rather it is the existence of evidence that the presence of thickets reduces entry combined with the large literature we have reviewed that shows that currently patent systems do not work as well as they should. This literature documents quality issues with patents in technology areas affected by patent thickets, a large decline in the relationship between R&D spending and patenting in some sectors and a substantial increase in resources devoted to patent litigation leading to the partial or complete revocation of patents in areas identified as prone to thickets. All of this may lead one to the conclusion that the operation of the patent system could use some improvement.


I admire the thoughtfulness and scholarship in this paper very much. I wonder though if the effects they find are actually an understatement of the problem. The entry they study is, I would guess is dominated by already incumbent firms taking out patents for the first time. but the important effects outlined by Haltiwanger et al are for newly established firms entering into product markets de novo. For them, the barriers tentatively identified here will be even higher.

I do predict however that all this will have a positive effect on the entry of…lawyers.

This is just one area of IP which could use some reform, for more examples, see this post on "Why Didn't Google Start in the UK?"