WIPIP, University of Washington School of Law
Plenary Session 1: Innovation Policy
Stephanie Bair, Promoting the Useful Arts: Corporate Edition
87% of patents are assigned to organizations, not individuals. How to motivate individuals/employees? Assumption is usually that companies will offer optimal incentives, such as financial bonuses. But companies vary quite widely in incentive structures, both financial and more importantly nonfinancial. Amazon: works employees past endurance. Amazon thinks challenging employees is good for innovation. Google: free massages, food; policy allowing employees to take extended leave w/benefits to work for nonprofit, travel, etc. Google also thinks its approach is good for innovation.
Findings: workplaces that function like economic exchanges—bonus financially for innovation—don’t work very well. Social exchanges tend to motivate creativity more. Social exchanges are less formal than economic exchanges, like friend/neighbor relationship. Loose exchange of favors over time. Can be more efficient because of informality; reduces transaction costs. Built on trust, dignity, and respect. Other factors good for creativity: autonomy, competence, feeling of relatedness among employees, variety of tasks, work/life balance. Promoting sense of choice; avoiding undue control; giving credit for work; encouraging appropriate breaks/downtime. Conclusion: Amazon has it wrong. Controlling behaviors make employees feel less competent. [Does Amazon get fewer patents relative to its non-warehouse employee cohort size than Google? What other measures might we use?]
How to get companies to do what’s best? The case for private ordering. Evidence shows that creativity incentives support the bottom line, are related to employee satisfaction/productivity/loyalty; good press. Should want to do it of their own accord. (Unless employer class has taste for control/status inequality that overwhelms this, or is in a market with less need for ongoing large-scale innovation and just wants bodies. Cf. Edward Baptist’s excellent and terrifying The Half Has Never Been Told—physical torture can also elicit innovation and efficiency from human bodies.)
What causes these incentives to fail? Bounded rationality. Info processing limitations; status quo bias; conformity bias. Solution: company can use metrics. Info asymmetrics: potential employees may not have as much info as they do at Apple. Amazon announced new benefits after the bad press. Social norms: managers should face constraints for behaving poorly. Employment law: promote employee mobility.
Glynn Lunney, Copyright’s Excess
We know © exists to solve a problem of supply/demand and underproduction in the absence of legal protection; © tries to eliminate free riders to push paying demand = actual demand, making supply & demand intersect at optimal level. Copyright’s excess is: when we push © up we do it uniformly, not just for marginal works we want to incentivize—transfer wealth from consumers to producers of works we would have gotten anyway, with less ©. Traditional analysis: just redistribution, no welfare effects; maybe it’s in pockets of Congress, but that doesn’t matter. Plus, given uncertainty, might be incentive effects even for nonmarginal works.
But, traditional answer: higher prices mean higher deadweight losses and higher transaction costs to be balanced against marginal benefits. Deadweight loss will be half of what’s transferred to producers (in perfect model).
Real world application: demand curves are highly skewed. One song streamed 60 million times during 2005-before. 4 million songs on Spotify have never been streamed by anyone. 5 million times for Goo Goo Dolls song. To help one marginal song by $1 we have to give $12 to that song at the top. Very little goes to the marginal songs overall.
Music has seen a radical transfer of wealth from consumers to copyright owners from the 1970s to 2000s—followed by sharp decline in sales. What relationship between income and output in the music industry? Music revenue is down from peak in 2000. File sharing traffic is up—1000 petabytes a month, 1.25 billion albums a month if they were all albums; $75 billion in reduced deadweight loss. Albums released in US from 1996-2012: 30,000 to 100,000, then recession took us to 80,000, well above the peak revenue year of 1999. Billboard Hot 100: 5200 slots/year; number of unique songs: in 60s/70s it was about 750. Falls steadily until 2002 when it’s 300 songs, then up to 500, then recession and 400. Turnover isn’t exact proxy for high quality output—maybe we’re getting some super-high quality songs sticking around. Did quality peak in 1999? Rolling Stone thinks it peaks in 1970s (top albums ever), but Spotify might be less white-guy. 2014 play count: backside of Spotify distribution curve; median user is 28 and mostly younger; normalized, though, the play counts peak in 1983, when revenue was lowest, and 1997 is nadir in terms of what people still listen to today. So we didn’t see peaks in quality in 1999.
Why might we be getting fewer high-quality songs? New artists—peak revenue period in 1983-1999, there is a slight upward slope in the number of new artists. 25% of songs were by new artists, up to 40% at peak; filesharing kicks in and we go back to 15-20%. Why not as many new songs? Existing artists produce fewer hits when record sales are going up.
Hypothesis: backwards bending labor supply curve. Overpaying superstars reduces their creative output. Backstreet Boys & ‘Nsync were making $200 million/album. They produced fewer albums. Top 250 acts of all time, according to RIAA: early ones like Beatles produced average 14 albums in first 10 years of career (Beatles were 12 albums + EP); by the 1990s, that was down to 5. Lost: Sergeant Pepper’s, the White Album, Abbey Road, Let it Be. Adele’s delay between albums will predictably be greater too. Top artist hit production from 1962-present: number of hits v. record sales over first 10 years of career—the artists with lots of hits tend to be grouped on low income portion of scale.
More revenue didn’t yield more and better works of authorship. Higher revenue = more new artists, but fewer hits from existing artists; most of the new artists were one-hit wonders. Second effect (fewer hits) outweighed the first. Maybe we need to balance incentives for marginal artists v. incentives for most popular, instead of incentives v. access.
Laura Pedraza-Fariña, Scaffolding Innovation: The Role of Patents, Grants, and Informal Norms in Assembling Teams that Span Technological Domains
Lumpy structure of architecture of knowledge distribution. We’ve made vast improvements in the way we can cure cancer/prolong survival, particularly for childhood cancers, but the side effects include infertility; fewer advances in that area. Why this disconnect? Two communities involved in infertility/cancer research: oncologists, who are interested in understanding cell division; endocrinologists, who in part work on addressing infertility. Though when you ask cancer patients their main concerns, infertility is second only to fear of death, that research hasn’t happened. These two communities are not talking to each other though they have key complementary knowledge to address secondary infertility from cancer treatment.
Accounts of innovation incentives often assume free flow of information; don’t look at barriers to assembly of teams even once free-riding and market-demand problems are eliminated. Architecture of knowledge distribution/social barriers to flow of knowledge are often important. NIH sought proposals for interdisciplinary research—problems that can only be solved by cooperation among disciplines; oncofertility is one example.
Interviewed key informants: barriers and benefits to such research; what is the effect of current patent and grants policy/what should we do to encourage it? Requires someone in a position to unite two groups—an endocrinologist who fortuitiously ended up in charge of a cancer center. Benefits of collaboration: problem finding. Huge areas of unknowns—when the mouse ovary and monkey ovary people got together (hadn’t previously been talking), they realized that these ovaries behaved very differently and they needed to account for that instead of using mouse ovaries as human models. New field of research: oncofertility. New products: gel matrix to grow follicles into eggs in vitro. New social connections: ongoing collaborations between engineers, endocrinologists, oncologists, etc. People w/40 years of experience described it as the best program they’d been involved with in their lives. Intrinsically motivating to work in non-traditional teams.
Policy conclusion: create scaffolds. Temporary bridges may be enough to bring together communities previously separated by structural holes, b/c intrinsic motivation may take care of the rest. Patents are not the right tool for the job; the type of research creating nontraditional teams tends to be exploratory, low appropriability/high spillovers, long time to market—patents tend to distort incentives against these. Grants aren’t currently structured to do this b/c NIH institutes weren’t coordinating, but could be. Regulatory levers: FDA. Collaborative R&D.
W. Nicholson Price II, Timo Minssen, & Arti Rai, Patent Failures on Life Science Frontiers
Poster child for patents seems to be developing new drugs. Some literature says that’s not true across pharma/bioscience—antibiotics, orphan drugs, biosimilars, manufacturing, second uses for existing drugs, diagnostics—all these things are different.
Themes emerging from the literature: coordination—many of these policies cut across areas—FDA, PTO, Fed. Cir., Congress think of different things and don’t talk to each other. Policy academics in these areas also often don’t talk to each other. Maybe industry trade groups are better at figuring out how to play off regulations/regulators. Life sciences are different? Maybe health is special: human flourishing, market failure, FDA as giant regulator sitting on top of everything, gatekeeping market entry.