Their statistical metric: The group measures patent quality by noting the number of citations the patent receives by other patents. Thus the report is

Their statistical metric: The group measures patent quality by noting the number of citations the patent receives by other patents. Thus the report is saying that the number of new patents that wind up being cited by other new patents has declined, which means in essence, they are only generally useful to the original patent holders. And from that measure they determine there has been a 20% drop in "citations". What they don't state is whether they normalized their data for the increase in size of the body of patents from which to cite. In 1990, there were less than five million US patents, while in 2011 there are over eight million US patents (roughly a 60% increase in volume). Without normalizing the citation numbers against the volume increase (i.e., considering the effects of inflation), the cited 20% lower figure is meaningless. It becomes an apples to oranges comparison.