AuthorHoffman, Elizabeth
  1. INTRODUCTION 192 II. BACKGROUND 197 A. Intellectual Property 197 1. Incentives for Inventing and Creating 197 2. Risk Preferences of Individuals and Firms with Respect to Creating 199 3. Prior Related Experiments on Intellectual Property 205 B. Framing Effects 207 1. Categorization Schemes 208 2. Light Computation 208 3. Emphasis and Priming 209 4. Imagine Yourself in a Context 211 III. DESCRIPTION OF EXPERIMENT 212 1. Experimental 2x2 Design 212 2. Demographic Variables and Baseline Risk Aversion 215 3. Subject Pool, Recruitment and Compensation 217 IV. RESULTS 221 V. DISCUSSION AND IMPLICATIONS 227 A. Limitations and Robustness 228 1. Invest in Invention Frame 228 2. Endowment Effects 230 3. No False Preferences 231 4. M-Turkers and Risk 232 B. Broader Implications 233 1. Implications for Intellectual Property 233 2. Implications for Contract Law, Corporate Law, and Other Areas 239 C. Social Welfare 241 VI. CONCLUSION 242 APPENDIX A: THEORETICAL FRAMEWORK AND IDENTIFICATION 244 A. Choice Theoretic Framework 244 B. Calibration to the Literature 247 C. Identification 250 APPENDIX B: ROBUSTNESS TEST: KEEP VS. INVEST IN INVENTION 253 APPENDIX C: ROBUSTNESS TEST: KEEP VS. INVEST 256 APPENDIX D: ROBUSTNESS TEST: ENDOWMENT ONLY 257 I. INTRODUCTION

    An emerging common wisdom holds that courts have made it "too hard" to obtain patent protection in critical industries. The origin of this criticism dates back at least as far as the United States Supreme Court's 2012 landmark opinion in Mayo Collaborative Services v. Prometheus Laboratories, Inc. (1) which (the argument goes) triggered a chain reaction of judicial opinions rendering patent rights progressively more difficult to secure. (2) Two years later, the Supreme Court decided Alice Corp. v. CLS Bank, (3) another opinion widely viewed as restricting patent rights. (4) And, barely three years after Mayo, the Federal Circuit cited it in invalidating a patent for a groundbreaking diagnostic test to detect fetal genetic conditions such as Down Syndrome early in pregnancy. (5) Before the test at issue was available, clinical diagnostic methods involved invasive techniques that materially endangered the health of the fetus. (6) But in 1996, doctors at Sequenom, Inc., a biotechnology company, discovered that maternal blood contains trace amounts of fetal DNA. (7) Having made this discovery, the same team developed a noninvasive blood test that could screen for fetal genetic conditions without endangering the fetus. (8) Sequenom's invention garnered it significant acclaim and prestigious awards for medical innovation. (9) The Federal Circuit was somewhat less impressed, and it invalidated the patent for failure to assert claims that were "significantly more" than a mere natural law. (10)

    Critics were quick to pounce. (11) Invention is already risky and costly enough, they argued, and this opinion made patent protection not only harder but also unpredictable, undermining the incentives to develop and finance critical new inventions. (12) As a result, they feared, risk-averse inventors and investors would stay away in droves, unjustly and inefficiently depriving society of many ground-breaking inventions such as Sequenom's. As Judge Kimberly Moore of the Court of Appeals for the Federal Circuit explained it in a recent dissent:

    The math is simple, you need not be an economist to get it: Without patent protection to recoup the enormous R&D cost, investment in diagnostic medicine will decline. To put it simply, this is bad. It is bad for the health of the American people and the health of the American economy. (13) The criticism recounted above seems intuitive, appealing, and powerful. But is it right? In this article, we interrogate it by deploying experimental methods to measure people's attitudes toward risk when investing in innovative activities. Although our inquiry produces a variety of insights, one in particular stands out: We uncover novel evidence that when confronted with an investment decision that is "innovation-related," people appear to become far more tolerant of risks than they are in other, economically equivalent settings. This result appears to be significant and robust, and it holds up regardless of whether one controls for subjects' age, gender, ethnicity, or several metrics of baseline risk aversion. Our results also persist when we vary the quantitative and qualitative risks involved, so long as the investment is tied to innovation. The effect appears to weaken substantially, however, when a risky option is framed simply as an investment opportunity, shorn of any invention-related dimension. Our interpretation of these findings is that the pursuit of invention--in concert with investing--introduces a critical interaction that operates to dampen people's manifest aversion to risk. In fact, we can even impute a quantitative size of this preference-dampening effect, by calibrating our results to a well-known set of risk tolerance measurement techniques in the economics literature. (14) Here, for the median subject in our study, we estimate that the innovation-related frame induces a reduction of manifest risk aversion of just under one-half of a standard deviation relative to our overall subject population. (15)

    To the extent that our results are generalizable, they have obvious implications for the "Goldilocksian" conundrum of patent protection--balancing the need to incentivize investors and inventors against the economic distortions from granting limited property rights to successful innovators. If inventors, entrepreneurs, and investors are comparatively more tolerant of risk in inventive settings, then patent policy may be able to incentivize value-enhancing innovation without throwing in a "premium" to compensate investors for their aversion to risk. Moreover, our results have broad implications outside of intellectual property, and in particular to the fast-developing areas of commercial and corporate law that must similarly wrestle with the question of how richly to incentivize financial investors in innovative industries.

    Several caveats to our analysis deserve specific mention before proceeding. First, as with all experimental findings, ours are subject to questions about the generalizability of our results in light of the subject pool. All of our experiments make use of either university students or workers on Amazon's Mechanical Turk platform ("M-Turk"). (16) Consequently, one might fairly question the representativeness of our subject pool relative to real-world inventors and investors, who actually participate in day-to-day innovation markets. The use of M-Turkers is sometimes singled out for criticism in this regard within the experimental literature, since it represents a population that is less capable of experimental control than conventional lab subjects. (17) We confront these concerns along multiple fronts. Foremost, we make sure to compensate our subjects with real monetary payoffs, so as to motivate and induce them to internalize the core financial tradeoffs we wish to study. Additionally, our dual-population study design allows us to draw comparisons between the university and M-Turk populations. Although we confirm the existence of differences (both demographic and behavioral) between these two populations, (18) the phenomenon of interest here (i.e., how innovation framing interacts with risk tolerances) remains remarkably consistent between the groups. (19) Although we cannot guarantee that these results would carry over to all real-world actors, their persistence across multiple distinct subject pools is at the very least encouraging.

    Second, although we believe our results deliver an important rejoinder to recent criticisms about courts' burgeoning stinginess towards patent holders, they do so in a particular and focused way: by showing that accommodations for risk preferences are perhaps unnecessary (or at least less necessary than one might think) in innovation-related contexts. A related but distinct criticism of the judicial opinions noted above is that they have simply made it costlier--even for a risk-neutral actor--to innovate or finance innovation because, for example, copying is insufficiently deterred. Our results have little to say about this dimension of the debate, other than to suggest that we may be able to confront the cost problem on its own terms, without also having to make significant additional allowances for risk aversion.

    The remainder of this Article consists of four parts. Part I discusses the motivation and background for our study, with particular emphasis on the oft-asserted argument that, within innovative industries, legal policy should accommodate risk aversion much like in other domains. Part II provides an overview of the experimental protocol, tying it to the relevant literature. Part III presents our core results, both for our baseline experiment and for a set of robustness experiments meant to stress test our core results to different environments. Part IV turns to implications, situating our findings within a variety of central legal puzzles regarding innovation. A series of appendices contain background technical derivations and provide additional statistical results.


    Before diving into our experimental enterprise, we first lay the foundation by providing a little background and context for our analysis. This Part describes the contours of some of the core behavioral theories that undergird much of intellectual property law and policy, focusing principally on patents. It further explores the assumptions that other scholars have made about the risks associated with intellectual property, including risks surrounding copying and risks surrounding creation. It then situates these theories against the literature on early-stage startup investing in technology firms, where--despite the asserted risks--there has long been significant appetite to invest. Finally, we...

To continue reading

Request your trial