Shan Alexander writes:
> I am looking for information on evaluating a patent portfolio as a financial
> asset where the worth of the patents themselves is evaluated independently
> of the worth of the technology they cover. Each patent in each country will
> require individual assessment but are there any underlying principles that
> can be used? If anyone has any experience in this field or can guide me to
> any papers, etc. on the subject, I should be very pleased to hear from them.
I have recently written a paper on exactly this subject, but it may not
be what you are looking for. That will depend on whether you want a
"how-to" article that will tell how to to make optimal decisions (e.g.,
whether or not to file or renew), or an analysis of (presumably) optimal
decisions made by others. My paper is of the latter kind. If you want a
paper of the former kind, you might have a look at a book by Russell
Paar, published by John Wiley, called something like "Valuing
Intellectual Property." It uses standard financial analysis to value the
cash flows arising from patents. You would have to make some special
assumptions to discriminate between the value attributable to the patent
as distinct from the invention.
My paper:
Somewhat briefly, the idea is this: Suppose you have a portfolio of
patent families (family = patents on the same invention in different
countries). Each patentable invention is considered a random draw from a
distribution of inventions (and here I am talking, as you said, about the
value of patent(s) on an invention as distinct from the technology). You
can think of this as the "quality" of the (patents arising from the)
invention. In addition, in each country you also make a random draw that
reflects the value of your patent in each particular country. The common
quality of the invention, plus the country-specific value, are multiplied
by the size of the market in each country (and, optionally, shifted by
such things as exports and the R&D/sales ratio), to generate an initial
annual return to patent protection, valued in dollar terms. This initial
return is then assumed to depreciate annually at a fixed rate, until the
return the patent generates is less than the cost of renewal, at which
point the patent lapses. The present value of the annual returns until
lapse, net of application and renewal fees, is the value of the patent as
a financial asset at the time of filing. The sum of present values across
all countries in which the applicant finds it profitable to file is the
returns to patent protection on the invention, i.e., its value as a
patentable asset.
What this model generates is a set of parameters that is specific to each
portfolio; it doesn't tell you anything about the value of the individual
members of the portfolio. However, you can then use a technique called
Monte Carlo simulation to generate simulated patents that behave like the
members of the portfolio, so you can compute the portfolio's mean,
variance, quantiles, etc. You can also compute confidence intervals for
the value of the portfolio, compare mean patent values in different
countries, etc.
I have successfully tested this model on a random portfolio drawn from the
1974 patent application cohort. I estimated that, conditional on the
applicant having filed in at least one country outside the priority
country (so that we know the invention was not abandoned as of the Paris
Convention deadline), the average value of patent protection was about
$245,000 per "international" family, in 1974 dollars. However, over half
the total value generated worldwide was attributable to the top 5% (~3000)
of all patent families (which were worth between about $1 million and $40
million apiece); the bottom 50% of all international families were worth
less than $45,000 each.
If you want to read the paper, it is available in PostScript format via
anonymous ftp to econ.yale.edu in the /pub/putnam directory; it is called
intpat.ps. I welcome comments.
I stress that this model is still experimental, that it was developed for
academic purposes, and I am continually working on it. In particular, I
am in the process of trying to determine whether this method is
sufficiently precise to be applied to firm-size portfolios---at this point
I honestly don't know.
I hope this helps.
Jon Putnam
P.S. For a different method of valuing patents that applies when you
have renewal data (which I don't) but don't have patent family data
(which I do), see, Pakes and Simpson, "Patent Renewal Data," _Brookings
Papers on Economic Activity: Microeconomics_, vol. ??, no. 3, 1989.