Saturday, July 29, 2006

Patent value & court decisions

A recent bump in traffic has come from someone posting about my citation/value research in Silicon Investor.

One reply to the thread said this : "interesting blog... It is, and goes back to the Glimstedt study of citations, but I'd be happier if he were listing court decisions based on the citation method of valuation."

Actually the citation method is tied to court decisions, it was just not the focus of this particular research paper. I am only posting a brief summary of the full paper here, hence you didn't see the link back to the references which ties litigated patents (which I will loosely interchange with court decisions here) and citations. For your reference here's the portion of my research that ties them together:

It has been suggested that given the high cost of litigation, patents which are litigated are also typically considered valuable patents. Research with litigated patents have been correlated to high patent citations (Lanjouw and Schankerman 1997; Allison, Lemley et al. 2004) with Allison arguing strongly for the bi-directional relationship between litigated patents and value. Their research argument is based on the premise of the high relative cost of litigation as compared to the cost of merely obtaining the patent with almost 75x the cost to enforce through litigation compared to filing.

An interesting step in the research would be to dive more deeply into the litigated patents and pull out the citation counts, ultimatly seeing if there is a model that can be developed for monetary settlements and citations.

References above:
Lanjouw, J. O. and M. Schankerman (1997). "Stylized facts of patent litigation: value, scope, ownership." NBER Working Paper (No. 6297).
Allison, J. R., M. A. Lemley, et al. (2004). "Valuable Patents." Georgetown Law Journal 92(3): 435.

Saturday, July 8, 2006

Patent data: foreshadowing of value to come

As promised, here is part II – Patent Variables to measure internal patent value.

Why is this important for the practicing attorney? As one who prosecutes patents and gives advice to clients on which applications to pursue and which to let abandon, this will give scientific weight to your advice.

First I will give a short background on how I “measured value” objectively for the patents in my data set.

Dependent Variable, or “measured value”: Forward Citation Counts

The dependent variable, forward citations, is measured by the number of citations the granted patent has from other patents. The variable was operationalized by counting the frequency of citations that the particular patent receives from subsequent patents as filed in the USPTO. I also controlled for self-citing.

Is this valid? I propose it is – Starting as early as 1990 forward citations have been validated by other research, showing that a general conceptual patent value or quality definition can be linked to the number of citations and external references it receives. The reasoning for the use of citations as a measure of value is the same for scientific literature: the economic importance of a work should be correlated with the frequency to which it is cited as a benchmark for further developments. It has been suggested that given the high cost of litigation, patents which are litigated are also typically considered valuable patents, and research with litigated patents have been correlated to high patent citations.

Independent Variables: Strategic Patenting Decisions
I am suggesting that each of these variables has an influence into the final value of a patent.

Family Size. Family size is all related applications (continuations, continuation-in-part and divisionals) filed worldwide.

Breadth. Patent breadth or scope is a measure of the technological influence or boundaries that the patent encompasses.

Claim Count. Claims define the legal boundaries of the patent rights, thus the firm typically has incentive to claim as much and as widely as possible while the examiner may narrow or reduce the claims to ensure validity before granting.

Jurisdiction. Jurisdiction size is computed as the number of separate country jurisdictions in which patent protection was sought for the same application.

Provisional Basis. Provisional applications allow an inventor to file information relating to their invention claim priority to that filing date up to 1 year from the original provisional filing date

Priority Claim. Priority claim is the basis of whether the patent application was based on a previous application, such as a continuation or divisional.

If you wish to see how I operationalized any of these variables let me know and I will post some additional details.

Next up: Part III – Results How did each of these independent variables fair when put to the statistical tests? And, more to the point, how can we apply this from a patent filing strategy perspective?