Albert Hu and Ivan P'ng say yes link here figures here. My own reservation about this study: patent strength has a large positive impact on foreign direct investment; we might expect this to increase growth rates in patent intensive industries. The paper also has a good summary of the literature that examines the impact of patents on patenting (!!) and R&D. By-the-way Michele and I redid the Kanwar and Evenson regressions (using their data which they kindly provided us with) link here: when you account for the size of a country - obviously an important determinant of per capita R&D when domestic markets matter - the results are reversed.

when you account for the size of a country - obviously an important determinant of per capita R&D when domestic markets matter - the results are reversed.

Two questions:

1. If, as you say, the above is an important determinant, then why do you think the researchers did not take it into account in their data analysis?

2. Why do you believe that per capita R&D is an important determinant, and can you provide an example that illustrates how it is determined and demonstrates its importance?

David:

I am stunned. Still stunned. I am unable to communicate my level of stun. The paper you did using the Kanwar and Evenson data is outstanding. Incredible. Fantastic. I wish that you and Michele had treated your book "Against Intellectual Monopoly" the same way. Your approach was quite objective and balanced, and your results appear to be believable to me.

I wonder about a couple of things. Is it possible to gather more data to update closer to the present time? One of the things I noticed after reading your book was that you suggested the increase in copyright length should lead to an increase in output, but your data in the book did not show any increase in output. Yet, the number of books published in the United States has tripled or quadrupled in the last couple of years, and I was wondering whether that affected your analyses.

I think the ten years you arrived at for optimal length of patent sounds about right, with a couple of minor points. The reason I think the ten years sounds "about right" is that about 2/3's of all patents are effectively abandoned when the 11.5 year maintenance fee is due. About 1/3 of all patents are abandoned when the 3.5 year maintenance fee is due. Thus, even patent owners seem to think that about 10 years is roughly the effective life of a patent.

My primary minor point is that your regression is an average over many industries, and I wonder how much benefit there is to separate industries for shorter and longer lengths of protection. What is the relationship between optimal length of patents and the industries for which patents are issued? This paper is a great starting point for asking some of those questions.

I need to look over your paper with more detail. There was a lot of detailed math that takes more than a cursory examination to review. I note that you thought the optimum length for copyright was two years, which seems about optimum for movies, but I wondered whether that was true for books as well. I read your paper in insufficient detail to see whether you broke those out.

Again, congratulations on an outstanding paper. I love it. Do more like this one.

The data on books after 1988 is quite problematic. We used data on copyright registrations; registrations plunge after 1988, but that is not because fewer books were produced, but because the law was changed so that you no longer had to register the copyright. You refer to a large increase in the number of books published: I have two questions about that

1. Do you know where we might get data about this?

2. If the increase is that huge it probably has nothing to do with copyright, since nothing has really changed in copyright law for books the last couple of years. More likely it has to do with the fact that it is now possible to make money off of much smaller print runs (print-on-demand) and to appeal to more specialized audiences through the internet.

David:

You ask a difficult question. Here are the best numbers I can give you. Data up to 1998 is available from:

http://www.swivel.com/data_sets/show/1001304

After that, I had to dig from various sources. I am unable to find figures for the three years 1999, 2000 and 2001. The figures for 1996-1998 and after 2001 are as follows (most data is from http://www.bowker.com/index.php/press-releases):

1996 - 68,175 1997 - 64,711 1998 - 56,129

2002 - 150,000 2003 - 175,000 2004 - 295,523 2005 - 282,500 2006 - 274,416 2007 - 276,645 - 411,422 if you include on-demand and short run books (per Bowker press release).

It appears that the publishing industry went crazy shortly after 1998.

It's not clear to me where swivel got their data from. They say the US Census and have a link to a non-working page there. As far as I know the census doesn't collect data on the number of books published (different titles) just revenues. But I could be wrong; if the census does collect that data (even just every ten years) it would be very useful. Bowker is a good lead because they maintain the list of books in print. I'm sure it doesn't go back as far, but it should be possible to combine the two time series. I notice that they keep the "print-on-demand" market separate, as while the output of traditional titles has remained flat there are now about 50% as many on demand titles as traditional ones. The change in technology is a problem: most likely some of the flatness in traditional books is because the more marginal ones are now done as print-on-demand. There data would be useful for figuring out when print-on-demand became important.

David:

I must admit to failure for trying to gain more information with regard to print on demand. I did find some information from the census bureau regarding publishers (one link is below), but I do not see the number of titles printed anywhere. Sorry!

http://www.census.gov/econ/census02/data/industry/E511130.HTM