Patents And Copyrights : Cumulative Innovation : Causal Evidence from the Courts

Cumulative Innovation : Causal Evidence from the Courts

Alan Dershowitz, Harvard Law Professor On the OJ Simpson Case | Murder | Facial Recognition
Patents And Copyrights : Cumulative Innovation : Causal Evidence from the Courts

This paper estimates the causal effect of patent rights on cumulative innovation. Using the patent invalidation decision of the US Federal Circuit Court of Appeal as its variable of interest. And the citation after the invalidation decision as the outcome of interest. The authors find that on average, over all the fields. Invalidation leads to a 50% increase in subsequent citations to the focal patent. Moreover, the authors find that the impact of patent invalidation is highly heterogeneous. With large variations across patents and technology fields in ways that are consistent with the blocking effect of patents arising from bargaining failures. Lastly, the authors also find that the blocking effect is concentrated in patents owned by large firms that appear to block small innovators. 

Governments use patent rights as tools to increase R&D incentives and to promote follow on innovation. But recently, among academics and policy makers there is a growing consensus that patents do the opposite and impede innovation. Patent rights tend to increase transaction costs. (Because you must pay to use the innovation) and there tends to be a lot of patent litigations as well.

The paper’s findings support and disagree in part with the academics and policy makers. The authors findings provide a good reason to believe that a wholesale scaling back of patent rights may not be the appropriate policy. This is because patent rights block cumulative innovation in very specific environments. And the authors suggest that government policies should target and address these environments. Moreover, the authors recommend designing policies and institutions that facilitate more efficient licensing. Which is the key to removing the blocking effect of patents and promoting cumulative innovation.

The authors based their empirical work on two data sets:
The decisions of the Court of Appeals for the Federal Circuit. And the US Patent and Trademark Office (USPTO) patent data set. For the OLS regression the authors regressed a dummy variable Invalidation  on citation after the invalidation decision controlling for docket number. In addition, data of the decision, patent identification number, name of the three judges involved, and a few other variables. With the OLS regression used by the authors there is a problem of self-selection of patents, causing an endogeneity problem. For example a positive shock to the value of the underlying technology may increase the citations to the pattern. (Positive correlation between outcome and control). And the patentee may heavily invest in the case to avoid invalidation (negative correlation between treatment and control). Therefore, by not controlling for the positive shock we are underestimating our causal impact of patent invalidation on cumulative innovation. 
To overcome the endogeneity problem. The authors make use of an IV that arises from the random allocation of the three judges. For the panel that decides the invalidation decision. This random allocation of judges with a high propensity to invalidate cases are not assigned to cases. Because of unobservable characteristics that may affect their decision to invalidate the patent. Moreover, the substantial variation across judges in the propensity to vote for patent invalidity implies a strong first stage. Any additional effect that case-specific unobservability may have on the decision to invalidate the patent (e.g. info revealed during the litigation process) are removed by dropping the patent p from the JIP IV. 

AI Financial Planning with the World’s Leading Tech – Rebellion Research

Patents And Copyrights : Cumulative Innovation : Causal Evidence from the Courts Patents And Copyrights : Cumulative Innovation : Causal Evidence from the Courts

In addition to strong first stage and randomness. We can argue for a strong exclusion restriction because the judges were assigned randomly. If, for example, a judge from the Federal Circuit had a celebrity status. Then that could increase the media attention of the patent and the innovation the patent covers. Then the citation for that patent may increase, hence violating the exclusion restriction. However, the judges on the Federal Circuit are not that well known. (Only judges on the Supreme Court could be argued to have celebrity status). Therefore, one could state that the violation in the exclusion restriction is somewhat irrelevant. On top of this, when 12 of the cases that did go to the Supreme Court were dropped by the authors. The causal effect of patent protection on citations did not change significantly.

However, the monotonicity assumption for the IV could be violated. For example, a judge may have a tendency to invalidate a patent in the technology field. But not in the pharmaceutical field. If that judge was predominantly picked for the cases with technology patent then their propensity to invalidate would be high. Implying that the probability of invalidation is high. Than if that same judge was picked for the a case involving a pharmaceutical patent. Then the probability of invalidation is low even if the judge had a high propensity to invalidate. This would imply that the invalidation decision is not consistent with the monotonicity assumption. However, the judges being randomly assigned mitigates this discrepancy, and one can work with this IV. 

The IV specification instructs the Invalidation dummy with the JIP. And returns an estimate that reveals a statistically significant positive effect between citations and invalidation by the Federal Circuit. The difference between the IV and the OLS specification shows the importance of controlling for endogeneity of invalidation. 

To further explore the heterogeneous effect of patent rights the authors make use of the MTE. To show that there is a substantial heterogeneity in the effect of patent protection on cumulative innovation. The finding of an increased MTE also helps identify mechanisms. That drive the increase in citations that we observe after Federal Circuit invalidation. The MTE increases with the probability of invalidation based on observable characteristics.

As a result, those with a high chance of invalidation also contribute the most to cumulative innovation in their field. Those patents with a high chance of invalidation are also characterised by unobservable factors that make invalidation less likely. An example of unobservable factors is legal enforceability which is observable to the patentee but unobservable to the licensees. This asymmetric information can lead to bargaining failure in licensing negotiations. Hence, when the Federal Circuit invalidates such a patent. It facilitates access to the technology that was blocked by the bargaining failure, which then results in a high MTE.

The heterogeneity of the effect of the patent rights exist for two reasons:

  1. Concentration of patent ownership in the technology field. (Empirically, if fragmented then they have to negotiate with a higher number of patentees. Which raises cost, but theoretically the additional patent in a fragmented environment. Depends critically on the degree to which patents are complements). Table VI confirms the empirical viewpoint. Since the effect of patent invalidation is small and statistically insignificant among patents in concentrated fields. Whereas it is large and significant in fragmented technology fields.
  2. Complexity of the technology field. In complex fields, new products incorporate numerous patentable elements. When products typically incorporate many patentable inputs, and these are held by different owners, multiple negotiations are needed. This increases the chance of bargaining failure. Table VI also verifies this: the effect of invalidation is more than twice. As large in complex technology areas as compared to non complex technology areas.
Patents And Copyrights : Cumulative Innovation : Causal Evidence from the Courts Patents And Copyrights : Cumulative Innovation : Causal Evidence from the Courts

Overall, the findings indicate that the fragmentation of patent ownership. In addition, the complexity of technology fields are key empirical determinants of the relationship between patent rights and cumulative innovation. The results found by the authors confirm that there is a blocking effect of specific innovations in biotechnology and medical instruments (since these fields are complex technology fields) . Furthermore, the authors found a heterogeneous causal effect of patent rights on innovation, therefore any policy implication will have to target specific technology areas to preserve innovation incentives.

The authors investigate whether the blocking effect of patent rights works through reducing the number of later innovators or on the intensity of their downstream innovation. The results of the IV estimates of the patent invalidation effect on citations by different size groups vary. The estimate of the blocking effect on the total number of external citation reveals that the blocking effect. Is concentrated exclusively on citations that patents of large firms receive from small innovators:
invalidation of a large firm patent increases small firm citation by about 520%. Which is consistent with the earlier estimate of 50% for the average blocking effect in the overall sample.

The authors’ findings provide good reason to believe that a wholesale scaling back of patent rights. May not be the appropriate policy. This could be because patent rights block cumulative innovation in very specific environments. And this suggests that government policies to address this problem should be targeted. Therefore, the authors’ results seem to lend more to designing policies. In addition, institutions that facilitate more efficient licensing. Which is the key to removing the blocking effect of patents and promoting cumulative innovation.

In general, the paper is well written and makes use of the IV without any extreme violation of the assumptions. Even when there is a violation of assumptions (such as the celebrity status of the judges). The causal effect does not change significantly, implying the paper’s findings are robust. On top of that, the violation of the monotonicity I used is weak. (And the best example I could come up with). Because the judges are selected at random. Thus there is no tendency for a single judge to work more with technology patents over pharmaceutical patents. Hence, the propensity for judges to invalidate should not favour technology patents over pharmaceutical patents. Therefore, I find the causal effects and the policy implications the authors suggest are robust.

Patents And Copyrights : Cumulative Innovation : Causal Evidence from the Courts Written by Alberto Galasso and Mark Schankerman

Patents And Copyrights : Cumulative Innovation : Causal Evidence from the Courts