When I was in law school, I was surprised (and fascinated) to learn how little scholars actually know about how patent laws affect innovation. My article Patent Experimentalism explains why this is such a hard empirical question, summarizes a lot of the empirical work that has been done, and analyzes the institutional design options (including policy randomization) to help make more empirical progress. Two of my favorites among the empirical pieces I discuss are by MIT economics professor Heidi Williams—one on IP-like contractual restrictions on human genes (summarized previously on this blog), and one on the skew in cancer drug R&D toward late-stage cancer patients (with Eric Budish and Ben Roin). In June, Williams posted a new paper that reflects on the challenges of measuring the relationship between patent strength and research investments, summarizes these two studies, and discusses directions for future research.
Although "a literal interpretation of the current set of available empirical evidence . . . would be that the patent system generates little social value," Williams explains that "drawing such a conclusion would be premature." The "dearth of empirical evidence" stems from two problems: measuring specific research investments, and "finding any variation (much less 'clean' or quasi-experimental variation) in patent protection." Her recent papers "identified and took advantage of new sources of variation in the effective intellectual property protection provided to different inventions." If you aren't familiar with those two papers, this piece contains a great summary.
Looking for similar kinds of variation seems like a promising avenue for future research, although Williams cautions against jumping quickly from her work to broad conclusions for patent policy. For example, while her work on contractual restrictions on human genes led to persistent decreases in follow-on research and commercial product development, her preliminary results from a follow-on project with Bhaven Sampat suggest that "on average gene patents have had no effect on follow-on innovation." As Williams notes, the U.S. Supreme Court has been concerned about the effects of patents on follow-on research in its recent forays into patentable subject matter, and perhaps further empirical work along these lines will help inform this muddled area of doctrine.