EXPLANATORY AND POLICY SIGNIFICANCE
What implications can we draw for innovation theory and policy from the continuum of excludability and from the existence of highly nonexcludable information goods of the sort just canvassed? Perhaps the most fundamental conclusion is that a patent system will predictably and systematically distort private investment decisions regarding innovation, overstating the value of highly excludable information goods and understating the value of highly nonexcludable ones. As a result of these distortionary effects, patents will fail to provide sufficient private returns to enable investment in certain information goods that clearly offer a net social benefit. Indeed, the valuable innovations neglected by patents will in some cases be comparatively more valuable than the ones patents do incent. Finally, increases in patent protection will tend to exacerbate these distortions by channeling ever more resources toward comparatively less valuable (but more excludable) innovations and away from an increasingly larger domain of highly valuable, less excludable ones.
Below we elaborate on these effects and draw out their implications for the conventional theory of patents, for patent and innovation policy, and for debates about information policy.
The Potential Distortions of Patents
An optimally efficient system of innovation incentives would provide signals to private parties regarding the expected returns from innovative activity that directly tracked the underlying social value of the activity. Patents, however, link the expected private returns not to social value simpliciter, but rather to the portion of social value that can be effectively (or cheaply) extracted through the exercise of exclusionary rights. But there is no reason to think that variations in the ease or costs of exclusion are correlated with the underlying social value of different information goods. Reasoning in ideal terms, patents will drive innovative effort and investments away from an optimally efficient allocation providing the greatest net social value and instead toward information goods that may provide lower net social value but higher private value owing to lower costs or barriers to effective excludability
There are two distinct kinds of ideal-type distortion at issue here. The first type reflects the fact that there are some highly nonexcludable goods whose development a patent system will fail to incentivize because the private returns appropriable using patents remain lower than the private costs of creation or validation of the good. (141) To be sure, it is to some extent a familiar point that patents will only incentivize a subset of the universe of net-beneficial innovations. As others have observed, transaction costs and other barriers to perfect price discrimination and tailored licensing mean that a patent system will fail to produce some net-beneficial innovations because some of the social surplus from innovations will go uncaptured by the private innovator (with the innovator's share further reduced by the limited duration of patent protection, etc.). (142) This claim is an extension of the point recognized above--that patents do not yield perfect appropriability because of limits on their scope, duration, and so forth. And indeed an "optimizing" response has developed in the literature; that theory urges the creation of ever more expansive, fine-tuned property rights, so as to capture all net-beneficial innovations in pursuit of a global optimum. (143) Our point, however, is different in two respects.
A first difference lies in the divergent prescriptions that these two arguments recommend. For highly nonexcludable goods, the standard "optimizing" response to the transaction cost problem--namely, to increase the strength of patent protection or the ability of patentees to extract a greater share of the surplus from transactions--will be ineffectual in remedying the underlying skew between social value and private appropriability. In fact, this intervention will have the opposite effect: strengthening patent rights will further distort the signal that exclusion rights transmit to make relatively excludable goods still more appealing targets of investment in comparison to relatively less excludable goods (as discussed below). Importantly, the features that make patents ineffectual at inducing the creation of highly nonexcludable goods do not apply to other institutional approaches to innovation. That is, if the government funds the creation of an information good such as a checklist (as it in fact did in our example), the innovation can be distributed without exclusion and thus without the need to confront the normative, technological, and institutional barriers to the enforcement of exclusion rights.
A second, somewhat subtler, difference between the transaction-cost problem and our argument lies in the force of our claim not only for optimizing but also non-optimizing views. For non-optimizers, the fact that patents will fail to incentivize all net-beneficial innovations is less concerning so long as we can assume that the innovations we obtain from a patent system will tend, as a whole, to be more beneficial than the ones we forego. And that standard assumption is appropriate if patents operate symmetrically on all kinds of information goods (that is, if symmetric transaction costs are the main problem with recouping value); if that is so, then at any given level of patent protection, the innovations that are incentivized will be those with a higher ratio of (privately appropriable) social value to private costs. (144) But where the constraints on private appropriation are not symmetrical across categories or types of innovation--as is the case for highly nonexcludable information goods--then some innovations that go under-incented may hold out greater ratios of social value to cost. It is easy to imagine, for instance, that there may be unincentivized lifestyle interventions that are not only net beneficial, but more beneficial than an incentivized statin drug (because the intervention is cheaper or generates more social value, or both). For such cases, alternative or supplemental innovation approaches will not just increase the overall amount of valuable innovations that we are able to obtain as long as we are willing to devote more social resources to this sector; they also hold out the promise of improving the efficiency of expenditures even if we keep them at the existing level. Shifting some resources from the patent system to alternatives will provide a greater welfare "bang for our buck."
If we wish to realize the social benefit from these highly nonexcludable innovations that remain unprofitable even under a patent system, then we must pursue alternative innovation policies, such as prizes, public funding, or commons-based approaches. Critically, the problem of nonexcludability points to a domain of innovation that patents, whatever their scope, cannot adequately address. And this holds even if our focus is not on more upstream or basic research, but rather solely on downstream or directly implementable interventions. Even for the latter, we cannot conclude that the most efficient system of innovation could rely solely on exclusion rights. This necessity of supplementing patents with some alternative policies has a pointed implication for innovation policy analysis that bears emphasizing. Any policy, such as prizes or public funding, that would generate more valuable, highly nonexcludable innovations than patents would not merely supplement the patent system, but would, at least in this respect, outperform it. This particular superiority should then be added to our understanding of the virtues of nonexclusionary approaches to innovation. Of course, nonexclusionary approaches have many possible disadvantages, too. Nothing we say here, for example, contradicts the concern that governments may make wasteful investments or may be susceptible to inappropriate influences. When deciding on the proper mix of institutional approaches, these possible costs must be weighed against the possible benefits-benefits that now must be understood to include the ability to generate investment in highly nonexcludable goods.
Precisely this sort of comparative-institutional approach was what Demsetz advocated in his original articulation of the allocative case for patents, when he cautioned against the "nirvana fallacy" of evaluating the actual operation of one system against an ideal version of another. (145) Yet, the continuum of excludability reveals that the allocative case for patents is itself premised on a flawed idealization: that of a one-to-one relationship between property/exclusion rights and private appropriability of market value. Correcting for this flaw boosts the comparative case for alternatives to patents--or, more precisely, the case for a broad ecology of innovation policies that includes a significant, expanded role for other institutional approaches.
A second type of potential distortion is presented by the fact that patents may not only fail to incentivize some net-beneficial goods, but also affirmatively jeopardize the creation of such goods by diverting resources away from them. (146) Consider again a lifestyle intervention that is more net beneficial, but less excludable, than a statin drug. At some low level of patent protection, it may be the case that the lifestyle intervention holds out greater private returns than the less valuable drug, and the returns are sufficient to recoup the capitalized costs of developing and validating the intervention. However, as patent protection increases, the private appropriability from the drug may increase to a point that it becomes the more profitable project. And, assuming increasing costs of capital (i.e., an upward-sloping supply curve for investment dollars), it may crowd out the lifestyle intervention entirely...