Joff Wild

A few days ago I wrote a blog around a note written to me by James Bessen and Michael Meurer about comments I had made concerning their book Patent Failure. Part of their note featured the claim that the frequently recounted tales of IBM making $1 million plus from patent licensing were “an urban legend”. I expressed surprise at this and asked Jim and Michael to come back with further information. Jim has now done so:

Does IBM get over $1 billion a year in profits from licensing its patents? This claim has been widely circulated, but we challenge it as an "urban legend" in our book, Patent Failure. Although this is a very minor point in the book (we use IBM's licensing revenues as one measure to check economists' estimates of patent value), since Joff sees this as an issue of credibility, it is worth explaining how we arrived at that conclusion.

In fact, almost all of the information we used comes from IBM's Annual Reports. Like all good urban legends, this one begins with an element of truth. Beginning in 2000, IBM began including a statement about its "IP and licensing royalties" which have been running over $1 billion. Now this sounds like a billion of patent licensing revenues and perhaps IBM wanted it to sound like that (this was the height of the dotcom hype).

But this figure is not found in IBM's actual Income Statement. Furthermore, the breakdown in the report shows that it consists of a lot else other than patent licensing revenues. For example, in the 2003 report, "IP and licensing royalties" consist of:

$560m "Sales and other transfers of IP"

$270m "Custom software development"

$340m "Licensing/royalty based fees"

The first category consists of the valuation of IP (divined how?) largely in divisions that are sold or spun off. Not patent licensing. The second category is not patent licensing either.

The third category includes patent licensing but it also includes technology licensing. Like many companies, IBM distinguishes patent licensing from technology licensing; the latter frequently involves the transfer of trade-secrets, technical know, it may include training, loans of personnel, laboratory access, etc.

People in IBM's patent department have told me that about 40% of the "licensing/royalty based fees" is actual patent licensing.

40% x $340m = $135m, or $111m in 1992 dollars (for comparison to our figures).

Now that is a very successful licensing program, but it is not a billion dollar licensing program.

Again, I am very grateful to Jim for taking the time to explain his and Michael’s thinking. But, again, I find it very difficult to accept their conclusions.

Let’s go back to where this started. It was with an assertion made by Jim and Michael that in 1999 non-pharma and chemical US public companies made just over $3 billion in profits as a result of the patents they own. In their book (which the publishers have now sent me) they state that such a figure represents: “The gross positive incentives provided to innovators by worldwide patents above and beyond the profits that could have been earned without patents.”

Yet, having said this, in the analysis above, they only factor in the revenues made by IBM from straight patent licensing. What about the $560 million from sales and other transfers of IP? Did the existence of patents in those sales not affect transaction costs? Would IBM have got the price it did in a sale if certain patents had not been involved? What about custom software development? As I understand it, this involves IBM providing customised software solutions to clients based on IBM proprietary rights, some of which – at least – will be patents and the know-how around how to work those patents to best effect. Does this not count?

So, while I am prepared to concede that IBM has never obtained $1 billion a year from straight patent licensing, I remain to be convinced that IBM’s patent portfolio has not enabled it to obtain premium prices in both sales and other transfers of IP, as well as in custom software development. Maybe it has not, but Jim and Michael do not explore this.

In the end, we are left with the fact that direct IP licensing, sales and development has seen IBM generate $1 billion or so plus a year since the mid-1990s. Jim and Michael have not shown that the figure would have been pretty much the same if patents did not exist because they have not fully considered all the ways in which patents have played a role in generating this revenue. Of course, patent revenue does not equate to patent profit, but my guess is that in IBM’s case it does equate to much more than the $135 million Jim and Michael talk about above. All this is before you factor in the myriad other ways that patents have helped to enhance the IBM offering – from the ability to exclude competitors, through the chance to conduct more focused R&D, to the entry they give to collaborative projects with other organisations. And if Jim and Michael are wrong about IBM, what does this tell us about all the other assumptions they have made? After all, IBM is just one of thousands of publicly quoted US companies, albeit one that understands patents and their potential better than most.

Anyway, enough of this already: the IAM blog is now officially closed for business on the Patent Failure front. I have the book and will be reading it through over the coming days. Then I hope to arrange an interview with Jim and Michael. This will run in the next issue of IAM, publishing at the end of May. I already have a lot of questions to ask them.