Moretti & Wilson: Do State Incentives for Innovation Work?

Do state incentives for innovation work? As I recently discussed in a short presentation at IP Scholars on August 8, state and local incentives for innovation, from R&D tax credits to competitive awards for research or commercialization, have become increasingly common as states attempt to build regional "clusters" of innovation like Silicon Valley. Due to the importance of proximity and localized knowledge spillovers in generating innovation, in theory, state and local innovation incentives that result in self-sustaining "clusters" of innovation could have significant effects on patenting activity and other measures of innovation.

In this important new study, State Incentives for Innovation, Star Scientists and Jobs: Evidence from Biotech, Daniel Wilson, Senior Economist at the Federal Reserve Bank of San Francisco, and Enrico Moretti, Professor of Economics at U.C. Berkeley, perform the first state-wide analysis of state innovation incentives specifically targeted at the biotech industry. Their study investigates the effects of both biotech-specific incentives (offered in 11 states as of 2010) and general R&D tax credits (offered in 34 states as of 2010) on biotech activity in the states over the period from 1990-2010. While I cannot remark on the quality of Moretti and Wilson's data set or analysis, their results are fascinating and provide some insight on several crucial questions.

1. Do state biotech and R&D tax incentives enhance local biotech activity?

The study suggests yes—at least with respect to a few performance benchmarks. Adoption of R&D tax credits and biotech-specific incentives (e.g. tax credits, exemptions, start-up loans, or grants directed primarily at the biotech sector) raised the number of "star scientists" (defined as scientists whose patenting average over the past 10 years is within the top 5% nationally) in the state, and led to higher total employment in the biotech industry. They also increased the number of pharmaceutical manufacturing and R&D establishments in the state. Finally—of high relevance to the states, if not necessarily to innovation—the incentives had an indirect effect on employment in local services, such as retail and construction.

2. Are the incentives cost-effective for local policymakers?

The study gives no insights here. We don't know whether the effects of state innovation incentives are short-lived or how localized the benefits are, and there is no realistic way to judge whether policymakers are overpaying for any benefits they get. Just like with patents, we cannot go back and look at the counter-factual: what would have happened if the states had not used these incentives? But again—in theory—due to agglomeration benefits and knowledge spillovers, states adopting the incentives could experience long-term gains, not just a short-term push in local employment numbers.

3. What are the effects on national levels of innovation?

Here the study is inconclusive, perhaps slightly negative. In prior work, Wilson performed a state-by-state analysis of general state R&D tax credits from 1981-2004. He showed the average effective credit for state R&D credits increased four-fold over this period and in many states is now more than the federal R&D credit. But he found that although state R&D tax credits increased investment in R&D within the state, the result was mainly due to drawing R&D away from other states. In other words, "the net national effect of R&D tax incentives on R&D spending [was] near zero."

In the Moretti and Wilson study, a few findings suggest the same thing may be true for state biotech incentives. First, most of the gains in star scientists were due to relocation from other states. The incentives had limited effect on patenting activity of scientists already in the state. They had zero effect on academic researchers at non-profit universities not receiving the subsidies (suggesting those researchers weren't necessarily learning more from those researchers that did benefit directly.) And they had limited effects on general numbers of patent filings.

However, I would not take too much from these conclusions about patent activity. First, as the authors point out, research in biotech can take a long time to come to fruition and generate patentable inventions. Second, as we know, patents do not equal innovation, and plenty of innovation happens without patenting. It's possible the corporations or individuals benefitting from the awards focused their efforts on developing nonpatentable innovations or commercial applications for existing inventions. It's also possible they chose not to patent and relied on trade secrecy or market lead time.

4. Federalism implications?

This study also has important implications for scholars on economic theories of federalism, such as Robert Cooter (Berkeley Law). An important question raised by Cooter's and others' scholarship on the "ideal" number of governments, is the extent to which local knowledge and expertise may facilitate states' efforts at growing innovation clusters. For example, is there any comparison here between state or city provisioning of "local public goods" (e.g. parks, fire prevention) and local support for innovation? Because information is a "pure" public good, states do not internalize all the benefits of their innovation policies. Thus, one could argue that we should simply nationalize the whole system, as we do for IP or national security. On the other hand, if these incentives really do result in clusters of innovative activity that would not otherwise occur, maybe they are net winners.

Another issue, which should be important for scholars of "horizontal federalism"—who study the interactions between the states in passing laws and policies—is whether state innovation incentives produce negative effects for other states. If they are (as Wilson's study on state R&D tax credits suggests) a zero-sum game in which states or cities are compelled to pay to attract "star scientists," maybe they should be prohibited under the dormant Commerce Clause, as the Sixth Circuit suggested in Cuno v. DaimlerChrysler Corp. (striking down city and Ohio state tax benefits to encourage a jeep manufacturer to expand in Toledo, since vacated on standing grounds). Then again, maybe R&D tax credits are an exception—a simple locational incentive disguised as an innovation incentive. As I've discussed, states can use a variety of other incentives to promote innovation besides R&D tax credits that may not be as vulnerable to a race to the bottom.

I am glad that economists like Moretti and Wilson are interested in these questions and look forward to reading more of their invaluable empirical work.