Category Archives: Innovation

“Dynamic Commercialization Strategies for Disruptive Technologies: Evidence from the Speech Recognition Industry,” M. Marx, J. Gans & D. Hsu (2014)

Disruption. You can’t read a book about the tech industry without Clayton Christensen’s Innovator’s Dilemma coming up. Jobs loved it. Bezos loved it. Economists – well, they were a bit more confused. Here’s the story at its most elemental: in many industries, radical technologies are introduced. They perform very poorly initially, and so are ignored by the incumbent. These technologies rapidly improve, however, and the previously ignored entrants go on to dominate the industry. The lesson many tech industry folks take from this is that you ought to “disrupt yourself”. If there is a technology that can harm your most profitable business, then you should be the one to develop it; take Amazon’s “Lab126″ Kindle skunkworks as an example.

There are a couple problems with this strategy, however (well, many problems actually, but I’ll save the rest for Jill Lepore’s harsh but lucid takedown of the disruption concept which recently made waves in the New Yorker). First, it simply isn’t true that all innovative industries are swept by “gales of creative destruction” – consider automobiles or pharma or oil, where the major players are essentially all quite old. Gans, Hsu and Scott Stern pointed out in a RAND article many years ago that if the market for ideas worked well, you would expect entrants with good ideas to just sell to incumbents, since the total surplus would be higher (less duplication of sales assets and the like) and since rents captured by the incumbent would be higher (less product market competition). That is, there’s no particular reason that highly innovative industries require constant churn of industry leaders.

The second problem concerns disrupting oneself or waiting to see which technologies will last. Imagine it is costly to investigate potentially disruptive technologies for the incumbent. For instance, selling mp3s in 2002 would have cannibalized existing CD sales at a retailer with a large existing CD business. Early on, the potentially disruptive technology isn’t “that good”, hence it is not in and of itself that profitable. Eventually, some of these potentially disruptive technologies will reveal themselves to actually be great improvements on the status quo. If that is the case, then, why not just let the entrant make these improvements/drive down costs/learn about market demand, and then buy them once they reveal that the potentially disruptive product is actually great? Presumably the incumbent even by this time still retains its initial advantage in logistics, sales, brand, etc. By waiting and buying instead of disrupting yourself, you can still earn those high profits on the CD business in 2002 even if mp3s had turned out to be a flash in the pan.

This is roughly the intuition in a new paper by Matt Marx – you may know his work on non-compete agreements – Gans and Hsu. Matt has also collected a great dataset from industry journals on every firm that ever operated in automated speech recognition. Using this data, the authors show that a policy by entrants of initial competition followed by licensing or acquisition is particularly common when the entrants come in with a “disruptive technology”. You should see these strategies, where the entrant proves the value of their technology and the incumbent waits to acquire, in industries where ideas are not terribly appropriable (why buy if you can steal?) and entry is not terribly expensive (in an area like biotech, clinical trials and the like are too expensive for very small firms). I would add that you also need complementary assets to be relatively hard to replicate; if they aren’t, the incumbent may well wind up being acquired rather than the entrant should the new technology prove successful!

Final July 2014 working paper (RePEc IDEAS). The paper is forthcoming in Management Science.

“How do Patents Affect Follow-On Innovation: Evidence from the Human Genome,” B. Sampat & H. Williams (2014)

This paper, by Heidi Williams (who surely you know already) and Bhaven Sampat (who is perhaps best known for his almost-sociological work on the Bayh-Dole Act with Mowery), made quite a stir at the NBER last week. Heidi’s job market paper a few years ago, on the effect of openness in the Human Genome Project as compared to Celera, is often cited as an “anti-patent” paper. Essentially, she found that portions of the human genome sequenced by the HGP, which placed their sequences in the public domain, were much more likely to be studied by scientists and used in tests than portions sequenced by Celera, who initially required fairly burdensome contractual steps to be followed. This result was very much in line with research done by Fiona Murray, Jeff Furman, Scott Stern and others which also found that minor differences in openness or accessibility can have substantial impacts on follow-on use (I have a paper with Yasin Ozcan showing a similar result). Since the cumulative nature of research is thought to be critical, and since patents are a common method of “restricting openness”, you might imagine that Heidi and the rest of these economists were arguing that patents were harmful for innovation.

That may in fact be the case, but note something strange: essentially none of the earlier papers on open science are specifically about patents; rather, they are about openness. Indeed, on the theory side, Suzanne Scotchmer has a pair of very well-known papers arguing that patents effectively incentivize cumulative innovation if there are no transaction costs to licensing, no spillovers from sequential research, and no incentive for early researchers to limit licenses in order to protect their existing business (consider the case of Farnsworth and the FM radio), and if potential follow-on innovators can be identified before they sink costs. That is a lot of conditions, but it’s not hard to imagine industries where inventions are clearly demarcated, where holders of basic patents are better off licensing than sitting on the patent (perhaps because potential licensors are not also competitors), and where patentholders are better off not bothering academics who technically infringe on their patent.

What industry might have such characteristics? Sampat and Williams look at gene patents. Incredibly, about 30 percent of human genes have sequences that are claimed under a patent in the United States. Are “patented genes” still used by scientists and developers of medical diagnostics after the patent grant, or is the patent enough of a burden to openness to restrict such use? What is interesting about this case is that the patentholder generally wants people to build on their patent. If academics find some interesting genotype-phenotype links based on their sequence, or if another firm develops a disease test based on the sequence, there are more rents for the patentholder to garner. In surveys, it seems that most academics simply ignore patents of this type, and most gene patentholders don’t interfere in research. Anecdotally, licenses between the sequence patentholder and follow-on innovators are frequent.

In general, it is really hard to know whether patents have any effect on anything, however; there is very little variation over time and space in patent strength. Sampat and Williams take advantage of two quasi-experiments, however. First, they compare applied-for-but-rejected gene patents to applied-for-but-granted patents. At least for gene patents, there is very little difference in terms of measurables before the patent office decision across the two classes. Clearly this is not true for patents as a whole – rejected patents are almost surely of worse quality – but gene patents tend to come from scientifically competent firms rather than backyard hobbyists, and tend to have fairly straightforward claims. Why are any rejected, then? The authors’ second trick is to look directly at patent examiner “leniency”. It turns out that some examiners have rejection rates much higher than others, despite roughly random assignment of patents within a technology class. Much of the difference in rejection probability is driven by the random assignment of examiners, which justifies the first rejected-vs-granted technique, and also suggested an instrumental variable to further investigate the data.

With either technique, patent status essentially generates no difference in the use of genes by scientific researchers and diagnostic test developers. Don’t interpret this result as turning over Heidi’s earlier genome paper, though! There is now a ton of evidence that minor impediments to openness are harmful to cumulative innovation. What Sampat and Williams tell us is that we need to be careful in how we think about “openness”. Patents can be open if the patentholder has no incentive to restrict further use, if downstream innovators are easy to locate, and if there is no uncertainty about the validity or scope of a patent. Indeed, in these cases the patentholder will want to make it as easy as possible for follow-on innovators to build on their patent. On the other hand, patentholders are legally allowed to put all sorts of anti-openness burdens on the use of their patented invention by anyone, including purely academic researchers. In many industries, such restrictions are in the interest of the patentholder, and hence patents serve to limit openness; this is especially true where private sector product development generates spillovers. Theory as in Scotchmer-Green has proven quite correct in this regard.

One final comment: all of these types of quasi-experimental methods are always a bit weak when it comes to the extensive margin. It may very well be that individual patents do not restrict follow-on work on that patent when licenses can be granted, but at the same time the IP system as a whole can limit work in an entire technological area. Think of something like sampling in music. Because all music labels have large teams of lawyers who want every sample to be “cleared”, hip-hop musicians stopped using sampled beats to the extent they did in the 1980s. If you investigated whether a particular sample was less likely to be used conditional on its copyright status, you very well might find no effect, as the legal burden of chatting with the lawyers and figuring out who owns what may be enough of a limit to openness that musicians give up samples altogether. Likewise, in the complete absence of gene patents, you might imagine that firms would change their behavior toward research based on sequenced genes since the entire area is more open; this is true even if the particular gene sequence they want to investigate was unpatented in the first place, since having to spend time investigating the legal status of a sequence is a burden in and of itself.

July 2014 Working Paper (No IDEAS version). Joshua Gans has also posted a very interesting interpretation of this paper in terms of Coasean contractability.

“Agricultural Productivity and Structural Change: Evidence from Brazil,” P. Bustos et al (2014)

It’s been a while – a month of exploration in the hinterlands of the former Soviet Union, a move up to Canada, and a visit down to the NBER Summer Institute really put a cramp on my posting schedule. That said, I have a ridiculously long backlog of posts to get up, so they will be coming rapidly over the next few weeks. I saw today’s paper presented a couple days ago at the Summer Institute. (An aside: it’s a bit strange that there isn’t really any media at SI – the paper selection process results in a much better set of presentations than at the AEA or the Econometric Society, which simply have too long of a lag from the application date to the conference, and too many half-baked papers.)

Bustos and her coauthors ask, when can improvements in agricultural productivity help industrialization? An old literature assumed that any such improvement would help: the newly rich agricultural workers would demand more manufactured goods, and since manufactured and agricultural products are complements, rising agricultural productivity would shift workers into the factories. Kiminori Matsuyama wrote a model (JET 1992) showing the problem here: roughly, if in a small open economy productivity goes up in a good you have a Ricardian comparative advantage in, then you want to produce even more of that good. A green revolution which doubles agricultural productivity in, say, Mali, while keeping manufacturing productivity the same, will allow Mali to earn twice as much selling the agriculture overseas. Workers will then pour into the agricultural sector until the marginal product of labor is re-equated in both sectors.

Now, if you think that industrialization has a bunch of positive macrodevelopment spillovers (via endogenous growth, population control or whatever), then this is worrying. Indeed, it vaguely suggests that making villages more productive, an outright goal of a lot of RCT-style microdevelopment studies, may actually be counterproductive for the country as a whole! That said, there seems to be something strange going on empirically, because we do appear to see industrialization in countries after a Green Revolution. What could be going on? Let’s look back at the theory.

Implicitly, the increase in agricultural productivity in Matsuyama was “Hicks-neutral” – it increased the total productivity of the sector without affecting the relative marginal factor productivities. A lot of technological change, however, is factor-biased; to take two examples from Brazil, modern techniques that allow for double harvesting of corn each year increase the marginal productivity of land, whereas “Roundup Ready” GE soy that requires less tilling and weeding increases the marginal productivity of farmers. We saw above that Hicks-neutral technological change in agriculture increases labor in the farm sector: workers choosing where to work means that the world price of agriculture times the marginal product of labor in that sector must be equal to world price of manufacturing times the marginal product of labor in manufacturing. A Hicks-neutral improvement in agricultural productivity raises MPL in that sector no matter how much land or labor is currently being used, hence wage equality across sectors requires workers to leave the factor for the farm.

What of biased technological change? As before, the only thing we need to know is whether the technological change increases the marginal product of labor. Land-augmenting technical change, like double harvesting of corn, means a country can produce the same amount of output with the old amount of farm labor and less land. If one more worker shifts from the factory to the farm, she will be farming less marginal land than before the technological change, hence her marginal productivity of labor is higher than before the change, hence she will leave the factory. Land-augmenting technological change always increases the amount of agricultural labor. What about farm-labor-augmenting technological change like GM soy? If land and labor are not very complementary (imagine, in the limit, that they are perfect substitutes in production), then trivially the marginal product of labor increases following the technological change, and hence the number of farm workers goes up. The situation is quite different if land and farm labor are strong complements. Where previously we had 1 effective worker per unit of land, following the labor-augmenting technology change it is as if we have, say, 2 effective workers per unit of land. Strong complementarity implies that, at that point, adding even more labor to the farms is pointless: the marginal productivity of labor is decreasing in the technological level of farm labor. Therefore, labor-augmenting technology with a strongly complementary agriculture production function shifts labor off the farm and into manufacturing.

That’s just a small bit of theory, but it really clears things up. And even better, the authors find empirical support for this idea: following the introduction to Brazil of labor-augmenting GM soy and land-augmenting double harvesting of maize, agricultural productivity rose everywhere, the agricultural employment share rose in areas that were particularly suitable for modern maize production, and the manufacturing employment share rose in areas that were particularly suitable for modern soy production.

August 2013 working paper. I think of this paper as a nice complement to the theory and empirics in Acemoglu’s Directed Technical Change and Walker Hanlon’s Civil War cotton paper. Those papers ask how changes in factor prices endogenously affect the development of different types of technology, whereas Bustos and coauthors ask how the exogenous development of different types of technology affect the use of various factors. I read the former as most applicable to structural change questions in countries at the technological frontier, and the latter as appropriate for similar questions in developing countries.

Tunzelmann and the Nature of Social Savings from Steam

Research Policy, the premier journal for innovation economists, recently produced a symposium on the work of Nick von Tunzelmann. Tunzelmann is best known for exploring the social value of the invention of steam power. Many historians had previously granted great importance to the steam engine as a driver of the Industrial Revolution. However, as with Fogel’s argument that the railroad was less important to the American economy than previously believed (though see Donaldson and Hornbeck’s amendment claiming that market access changes due to rail were very important), the role of steam in the Industrial Revolution may have been overstated.

This is surprising. To my mind, the four most important facts for economics to explain is why the world economy (in per capita terms) stagnated until the early 1800s, why cumulative per-capita growth began then in a corner of Northwest Europe, why growth at the frontier has continued to the present, and why growth at the frontier has been so consistent over this period. The consistency is really surprising, given that individual non-frontier country growth rates, and World GDP growth, has vacillated pretty wildly on a decade-by-decade basis.

Malthus’ explanation still explains the first puzzle best. But there remain many competing explanations for how exactly the Malthusian trap was broken. The idea that a thrifty culture or expropriation of colonies was critical sees little support from economic historians; as McCloskey writes, “Thrifty self-discipline and violent expropriation have been too common in human history to explain a revolution utterly unprecedented in scale and unique to Europe around 1800.” The problem, more generally, of explaining a large economic X on the basis of some invention/program/institution Y is that basically everything in the economic world is a complement. Human capital absent good institutions has little value, modern management techniques absent large markets is ineffective, etc. The problem is tougher when it comes to inventions. Most “inventions” that you know of have very little immediate commercial importance, and a fair ex-post reckoning of the critical parts of the eventual commercial product often leaves little role for the famous inventor.

What Tunzelmann and later writers in his tradition point out is that even though Watt’s improvement to the steam engine was patented in 1769, steam produces less horsepower than water in the UK as late as 1830, and in the US as late as the Civil War. Indeed, even today, hydropower based on the age-old idea of the turbine is still an enormous factor in the siting of electricity-hungry industries. It wasn’t until the invention of high-pressure steam engines like the Lancanshire boiler in the 1840s that textile mills really saw steam power as an economically viable source of energy. Most of the important inventions in the textile industry were designed originally for non-steam power sources.

The economic historian Nicholas Crafts supports Tunzelmann’s original argument against the importance of steam using a modern growth accounting framework. Although the cost of steam power fell rapidly following Watt, and especially after the Corliss engine in the mid 19th century, steam was still a relatively small part of economy until the mid-late 19th century. Therefore, even though productivity growth within steam was quick, only a tiny portion of overall TFP growth in the early Industrial Revolution can be explained by steam. Growth accounting exercises have a nice benefit over partial equilibrium social savings calculations because the problem that “everything is a complement” is taken care of so long as you believe the Cobb-Douglas formulation.

The December 2013 issue of Research Policy (all gated) is the symposium on Tunzelmann. For some reason, Tunzelmann’s “Steam Power and British Industrialization Until 1860″ is quite expensive used, but any decent library should have a copy.

“Identifying Technology Spillovers and Product Market Rivalry,” N. Bloom, M. Schankerman & J. Van Reenen (2013)

R&D decisions are not made in a vacuum: my firm both benefits from information about new technologies discovered by others, and is harmed when other firms create new products that steal from my firm’s existing product lines. Almost every workhorse model in innovation is concerned with these effects, but measuring them empirically, and understanding how they interact, is difficult. Bloom, Schankerman and van Reenen have a new paper with a simple but clever idea for understanding these two effects (and it will be no surprise to readers given how often I discuss their work that I think these three are doing some of the world’s best applied micro work these days).

First, note that firms may be in the same technology area but not in the same product area; Intel and Motorola work on similar technologies, but compete on very few products. In a simple model, firms first choose R&D, knowledge is produced, and then firms compete on the product market. The qualitative results of this model are as you might expect: firms in a technology space with many other firms will be more productive due to spillovers, and may or may not actually perform more R&D depending on the nature of diminishing returns in the knowledge production function. Product market rivalry is always bad for profits, does not affect productivity, and increases R&D only if research across firms is a strategic complement; this strategic complementarity could be something like a patent race model, where if firms I compete with are working hard trying to invent the Next Big Thing, then I am incentivized to do even more R&D so I can invent first.

On the empirical side, we need a measure of “product market similarity” and “technological similarity”. Let there be M product classes and N patent classes, and construct vectors for each firm of their share of sales across product classes and share of R&D across patent classes. There are many measures of the similarity of a vector, of course, including a well-known measure in innovation from Jaffe. Bloom et al, after my heart, note that we really ought use measures that have proper axiomatic microfoundations; though they do show the properties of a variety of measures of similarity, they don’t actually show the existence (or impossibility) of their optimal measure of similarity. This sounds like a quick job for a good microtheorist.

With similarity measures, all that’s left to do is run regressions of technological and product market similarity, as well as all sorts of fixed effects, on outcomes like R&D performed, productivity (measured using patents or out of a Cobb-Douglas equation) and market value (via the Griliches-style Tobin’s Q). These guys know their econometrics, so I’m omitting many details here, but I should mention that they do use the idea from Wilson’s 2009 ReSTAT of basically random changes in state R&D tax laws as an IV for the cost of R&D; this is a great technique, and very well implemented by Wilson, but getting these state-level R&D costs is really challenging and I can easily imagine a future where the idea is abused by naive implementation.

The results are actually pretty interesting. Qualitatively, the empirical results look quite like the theory, and in particular, the impact of technological similarity looks really important; having lots of firms working on similar technologies but working in different industries is really good for your firm’s productivity and profits. Looking at a handful of high-tech sectors, Bloom et al estimate that the marginal social return on R&D is on the order of 40 percentage points higher than the marginal private return of R&D, implying (with some huge caveats) that R&D in the United States might be something like 3 times smaller than it ought to be. This estimate is actually quite similar to what researchers using other methods have estimated. Interestingly, since bigger firms tend to work in more dense parts of the technology space, they tend to generate more spillovers, hence the common policy prescription of giving smaller firms higher R&D tax credits may be a mistake.

Two caveats. As far as I can tell, the model does not allow a role for absorptive capacity, where firm’s ability to integrate outside knowledge is endogenous to their existing R&D stock. Second, the estimated marginal private rate of return on R&D is something like 20 percent for the average firm; many other papers have estimated very high private benefits from research, but I have a hard time interpreting these estimates. If there really are 20% rates of return lying around, why aren’t firms cranking up their research? At least anecdotally, you hear complaints from industries like pharma about low returns from R&D. Third, there are some suggestive comments near the end about how government subsidies might be used to increase R&D given these huge social returns. I would be really cautious here, since there is quite a bit of evidence that government-sponsored R&D generates a much lower private and social rate of return that the other forms of R&D.

Final July 2013 Econometrica version (IDEAS version). Thumbs up to Nick Bloom for making the final version freely available on his website. The paper has an exhaustive appendix with technical details, as well as all of the data freely available for you to play with.

“Is Knowledge Trapped Inside the Ivory Tower?,” M. Bikard (2013)

Simultaneous discovery, as famously discussed by Merton, is really a fascinating idea. On the one hand, we have famous examples like Bell and Gray sending in patents for a telephone on exactly the same day. On the other hand, when you investigate supposed examples of simultaneous discovery more closely, it is rarely the case that the discoveries are that similar. The legendary Jacob Schmookler described – in a less-than-politically-correct way! – historians who see patterns of simultaneous discovery everywhere as similar to tourists who think “all Chinamen look alike.” There is sufficient sociological evidence today that Schmookler largely seems correct: simultaneous discovery, like “lucky” inventions, are much less common than the man on the street believes (see, e.g., Simon Schaeffer’s article on the famous story of the dragon dream and the invention of benzene for a typical reconstruction of how “lucky” inventions actually happen).

Michaƫl Bikard thinks we are giving simultaneous discovery too little credit as a tool for investigating important topics in the economics of innovation. Even if simultaneous discovery is uncommon, it still exists. If there were an automated process to generate a large set of simultaneous inventions (on relatively more minor topics than the telephone), there are tons of interesting questions we can answer, since we would have compelling evidence of the same piece of knowledge existing in different places at the same time. For instance, how important are agglomeration economies? Does a biotech invention get developed further if it is invented on Route 128 in Massachusetts instead of in Lithuania?

Bikard has developed an automated process to do this (and that linked paper also provides a nice literature review concerning simultaneous discovery). Just scrape huge number of articles and their citations, look for pairs of papers which were published at almost the same time and cited frequently in the future, and then limit further to articles which have a “Jaccard index” which implies that they are frequently cited together if they are cited at all. Applying this technique to the life sciences, he finds 578 examples of simultaneous discovery; chatting with a randomly selected sample of the researchers, most mentioned the simultaneous discovery without being asked, though at least one claimed his idea had been stolen! 578 is a ton: this is more than double the number that the historical analysis in Merton discovered, and as noted, many of the Merton multiples are not really examples of simultaneous discovery at all.

He then applies this dataset in a second paper, asking whether inventions in academia are used more often (because of the culture of openness) or whether private sector inventions are used more often in follow-up inventions (because the control rights can help even follow-up inventors extract rents). It turns out that private-sector inventors of the identical invention are three times more likely to patent, but even excluding the inventors themselves, the private sector inventions are cited 10-20% more frequently in future patents. The sample size of simultaneous academic-private discovery is small, so this evidence is only suggestive. You might imagine that the private sector inventors are more likely to be colocated near other private sector firms in the same area; we think that noncodified aspects of knowledge flow locally, so it wouldn’t be surprising that the private sector multiple was cited more often in future patents.

Heavy caveats are also needed on the sample. This result certainly doesn’t suggest that, overall, private sector workers are doing more “useful” work than Ivory Tower researchers, since restricting the sample to multiple discoveries limits the potential observations to areas where academia and the private sector are working on the same type of discovery. Certainly, academics and the private sector often work on different types of research, and openness is probably more important in more basic discoveries (where transaction or bargaining costs on follow-up uses are more distortionary). In any case, the method for identifying simultaneous discoveries is quite interesting indeed; if you are empirically minded, there are tons of interesting questions you could investigate with such a dataset.

September 2012 working paper (No IDEAS version). Forthcoming in Management Science.

“Back to Basics: Basic Research Spillovers, Innovation Policy and Growth,” U. Akcigit, D. Hanley & N. Serrano-Velarde (2013)

Basic and applied research, you might imagine, differ in a particular manner: basic research has unexpected uses in a variety of future applied products (though it sometimes has immediate applications), while applied research is immediately exploitable but has fewer spillovers. An interesting empirical fact is that a substantial portion of firms report that they do basic research, though subject to a caveat I will mention at the end of this post. Further, you might imagine that basic and applied research are complements: success in basic research in a given area expands the size of the applied ideas pond which can be fished by firms looking for new applied inventions.

Akcigit, Hanley and Serrano-Velarde take these basic facts and, using some nice data from French firms, estimate a structural endogenous growth model with both basic and applied research. Firms hire scientists then put them to work on basic or applied research, where the basic research “increases the size of the pond” and occasionally is immediately useful in a product line. The government does “Ivory Tower” basic research which increases the size of the pond but which is never immediately applied. The authors give differential equations for this model along a balanced growth path, have the government perform research equal to .5% of GDP as in existing French data, and estimate the remaining structural parameters like innovation spillover rates, the mean “jump” in productivity from an innovation, etc.

The pretty obvious benefit of structural models as compared to estimating simple treatment effects is counterfactual analysis, particularly welfare calculations. (And if I may make an aside, the argument that structural models are too assumption-heavy and hence non-credible is nonsense. If the mapping from existing data to the actual questions of interest is straightforward, then surely we can write a straightforward model generating that external validity. If the mapping from existing data to the actual question of interest is difficult, then it is even more important to formally state what mapping you have in mind before giving policy advice. Just estimating a treatment effect off some particular dataset and essentially ignoring the question of external validity because you don’t want to take a stand on how it might operate makes me wonder why I, the policymaker, should take your treatment effect seriously in the first place. It seems to me that many in the profession already take this stance – Deaton, Heckman, Whinston and Nevo, and many others have published papers on exactly this methodological point – and therefore a decade from now, you will find it equally as tough to publish a paper that doesn’t take external validity seriously as it is to publish a paper with weak internal identification today.)

Back to the estimates: the parameters here suggest that the main distortion is not that firms perform too little R&D, but that they misallocate between basic and applied R&D; the basic R&D spills over to other firms by increasing the “size of the pond” for everybody, hence it is underperformed. This spillover, estimated from data, is of substantial quantitative importance. The problem, then, is that uniform subsidies like R&D tax credits will just increase total R&D without alleviating this misallocation. I think this is a really important result (and not only because I have a theory paper myself, coming at the question of innovation direction from the patent race literature rather than the endogenous growth literature, which generates essentially the same conclusion). What you really want to do to increase welfare is increase the amount of basic research performed. How to do this? Well, you could give heterogeneous subsidies to basic and applied research, but this would involve firms reporting correctly, which is a very difficult moral hazard problem. Alternatively, you could just do more research in academia, but if this is never immediately exploited, it is less useful than the basic research performed in industry which at least sometimes is used in products immediately (by assumption); shades of Aghion, Dewatripont and Stein (2008 RAND) here. Neither policy performs particularly well.

I have two small quibbles. First, basic research in the sense reported by national statistics following the Frascati manual is very different from basic research in the sense of “research that has spillovers”; there is a large literature on this problem, and it is particularly severe when it comes to service sector work and process innovation. Second, the authors suggest at one point that Bayh-Dole style university licensing of research is a beneficial policy: when academic basic research can now sometimes be immediately applied, we can easily target the optimal amount of basic research by increasing academic funding and allowing academics to license. But this prescription ignores the main complaint about Bayh-Dole, which is that academics begin, whether for personal or institutional reasons, to shift their work from high-spillover basic projects to low-spillover applied projects. That is, it is not obvious the moral hazard problem concerning targeting of subsidies is any easier at the academic level than at the private firm level. In any case, this paper is very interesting, and well worth a look.

September 2013 Working Paper (RePEc IDEAS version).

“Patents and Cumulative Innovation: Causal Evidence from the Courts,” A. Galasso & M. Schankerman (2013)

Patents may increase or hinder cumulative invention. On the one hand, a patentholder can use his patent to ensure that downstream innovators face limited competition and thus have enough rents to make it worthwhile developing their product. On the other hand, holdup and other licensing difficulties have been shown in many theoretical models to make patents counterproductive. Galasso and Schankerman use patent invalidation trials to try and separate out the effect, and the broad strokes of the theory appear to hold up: on average, patents do limit follow-up invention, but this limitation appears to solely result from patents held by large firms, used by small firms, in technologically complex areas without concentrated power.

The authors use a clever IV to generate this result. The patent trials they look at involve three judges, selected at random. Looking at other cases the individual judges have tried, we can estimate the proclivity to strike down a patent for a given judge, and thus predict the probability a certain panel in the future will strike down a certain patent. That is, the proclivity of the judges to strike down the patent is a nice IV for whether the patent is actually struck down. In the second stage of the IV, investigate how this predicted probability of being invalidated, along with covariates and the pre-trial citation path, impact post-trial citations. And the impact is large: on average, citations increase 50% following an invalidation (and indeed, the Poisson IV estimate mentioned in a footnote, which seems more justified econometrically to me, is even larger).

There is, however, substantial heterogeneity. Estimating a marginal treatment effect (using a trick of Heckman and Vycatil’s) suggests the biggest impact of invalidation on patents whose unobservables make them less likely to be overturned. To investigate this heterogeneity further, the authors run their regressions again including measures of technology class concentration (what % of patents in a given subclass come from the top few patentees) and industry complexity (using the Levin survey). They also denote how many patents the patentee involved in the trial received in the years around the trial, as well as the number of patents received by those citing the patentee. The harmful effect of patents on future citations appears limited to technology classes with relatively low concentration, complex classes, large firms with the invalidated patent, and small firms doing the citing. These characteristics all match well with the type of technologies theory imagines to be linked to patent thickets, holdup potential or high licensing costs.

In the usual internal validity/external validity way, I don’t know how broadly these results generalize: even using the judges as an IV, we are still deriving treatment effects conditional on the patent being challenged in court and actually reaching a panel decision concerning invalidation; it seems reasonable to believe that the mere fact a patent is being challenged is evidence that licensing is problematic, and the mere fact that a settlement was not reached before trial even more so. The social welfare impact is also not clear to me: theory suggests that even when patents are socially optimal for cumulative invention, the primary patentholder will limit licensing to a small number of firms in order to protect their rents, hence using forward citations as a measure of cumulative invention allows no way to separate socially optimal from socially harmful limits. But this is at least some evidence that patents certainly don’t democratize invention, and that result fits squarely in with a growing literature on the dangers of even small restrictions on open science.

August 2013 working paper (No IDEAS version).

“Technology and Learning by Factory Workers: The Stretch-Out at Lowell, 1842,” J. Bessen (2003)

This is a wonderful piece of theory-driven economic history. Everyone knows that machinery in the Industrial Revolution was “de-skilling”, replacing craft workers with rote machine work. Bessen suggests, using data from mid-19th century mills in New England, that this may not be the case; capital is expensive and sloppy work can cause it to be out of service, so you may want to train your workers even more heavily as you deepen capital. It turns out that it is true that literate Yankee girls were largely replaced by illiterate, generally Irish workers (my ancestors included!) at Lowell and Waltham, while simultaneously the amount of time spend training (off of piece-wages) increased as did the number of looms run by each worker. How can we account for this?

Two traditional stories – that history is driven by the great inventor, or that the mill-owners were driven by philanthropy – are quickly demolished. The shift to more looms per worker was not the result of some new technology. Indeed, adoption of the more rigorous process spread slowly to Britain and southern New England. As for philanthropy, an economic model of human capital acquisition shows that the firms appear to have shifted toward unskilled workers for profit-based reasons.

Here’s the basic idea. If I hire literate workers like the Yankee farm girls, I can better select high-quality workers, but these workers will generally return home to marry after a short tenure. If I hire illiterate workers, their initial productivity is lower but, having their family in the mill town, they are less likely to leave the town. Mill owners had a number of local methods to collude and earn rents, hence they have some margin to pay for training. Which type should I prefer? If there exist many trained illiterate workers in town already, I just hire them. If not, the higher the ratio of wage to cloth price, the more I am willing to invest in training; training takes time during which no cloth is made, but increases future productivity at any given wage.

Looking at the Massachusetts mill data, a structural regression suggests that almost all of the increase in labor productivity between 1834 and 1855 was the result of increasing effective worker experience, a measure of industry-specific human capital (and note that a result of this kind is impossible without some sort of structural model). Why didn’t firms move to illiterate workers with more training earlier? Initially, there was no workforce that was both skilled and stable. With cloth prices relatively high compared to wages, it was initially (as can be seen in Bessen’s pro forma calculation) much more profitable to use a labor system that tries to select high quality workers even though they leave quickly. Depressed demand in the late 1830s led cloth prices to fall, which narrowed the relative profitability of well-trained but stable illiterate workers as compared to the skilled but unstable farm girls. A few firms began hiring illiterate workers and training them (presumably selecting high quality illiterate workers based on modern-day unobservables). This slowly increased the supply of trained illiterate workers, making it more profitable to switch a given factory floor over to three or four looms per worker, rather than two. By the 1850s, there was a sufficiently large base of trained illiterate workers to make them more profitable than the farm girls. Some light counterfactual calculations suggest that pure profit incentive is enough to drive the entire shift.

What is interesting is that the shift to what was ex-post a far more productive system appears to hinge critically on social factors – changes in the nature of the local labor supply, changes in demand for downstream products, etc. – rather than on technological change embodied in new inventions or managerial techniques. An important lesson to keep in mind, as nothing in the above story had any Whiggish bias toward increasing productivity!

Final working paper (IDEAS version). Final paper published in the Journal of Economic History, 2003. I’m a big fan of Bessen’s work, so I’m sure I’ve mentioned before on this site the most fascinating part of his CV: he has no graduate degree of any kind, yet has a faculty position at a great law school and an incredible publication record in economics, notably his 2009 paper on socially inefficient patents with Eric Maskin. Pretty amazing!

“Does Knowledge Accumulation Increase the Returns to Collaboration?,” A. Agrawal, A. Goldfarb & F. Teodoridis (2012)

The size of academic research “teams” has been increasing, inexorably, in essentially every field over the past few decades. This may be because of bad incentives for researchers (as Stan Liebowitz has argued), or because more expensive capital is required for research as in particle physics, or because communication technology has decreased the cost of collaboration. A much more worrying explanation is, simply, that reaching the research frontier is getting harder. This argument is most closely associated with my adviser Ben Jones, who has noticed that while team size has increased, the average age star researchers do their best work has increased, co-inventors on inventions has increased, and the number of researchers doing work across fields has decreased. If the knowledge frontier is becoming more expensive to reach, theory suggests a role for greater subsidization of early-career researchers and of potential development traps due to the complementary nature of specialized fields.

Agrawal et al use a clever device to investigate whether the frontier is indeed becoming more burdensome. Note that the fact that science advances does not mean, ipso facto, that reaching the frontier is harder: new capital like computers or Google Scholar may make it easier to investigate questions or get up to date in related fields, and certain developments completely subsume previous developments (think of, say, how a user of dynamic programming essentially does not need to bother learning the calculus of variations; the easier but more powerful technique makes the harder but less powerful technique unnecessary). Agrawal et al’s trick is to look at publication trends in mathematics. During the Soviet era, mathematics within the Soviet Union was highly advanced, particularly in certain areas of functional analysis, but Soviet researchers had little ability to interact with non-Soviets and they generally published only in Russian. After the fall of the Soviet Union, there was a “shock” to the knowledge frontier in mathematics as these top Soviet researchers began interacting with other mathematicians. A paper by Borjas and Doran in the QJE last year showed that Soviet mathematics were great in some areas and pretty limited in others. This allows for a diff-in-diff strategy: look at the change in team size following 1990 in fields where Soviets were particularly strong versus fields where the Soviets were weak.

Dropping papers with a Russian-named coauthor, classifying papers by fields using data from the AMS, the authors find that papers in Soviet-heavy fields had the number of coauthors increase from 1.34 to 1.78, whereas Soviet-weak fields teams grew only from 1.26 to 1.55. This difference appears quite robust, and is derived from hundreds of thousands of publications. To check that Soviet-rich fields actually had influence, they note that papers in Soviet-rich subfields cited Soviet-era publications at a greater rate after 1990 than Soviet-poor subfields, and that the increase in coauthoring tended to be driven by papers with a young coauthor. The story here is, roughly, that Soviet emigres would have tooled up young researchers in Soviet-rich fields, and then those young coauthors would have a lot of complementary skills which might drive collaboration with other researchers.

So it appears that the increasing burden of the knowledge frontier does drive some of the increase in team size. The relative importance of this factor, however, is something tough to tease out without some sort of structural model. Getting around the burden of knowledge by making it easier to reach the frontier is also worthy of investigation – a coauthor and I have a pretty cool new paper (still too early to make public) on exactly this topic, showing an intervention that has a social payoff an order of magnitude higher than funding new research.

Oct 2012 working paper (no IDEAS version). As a sidenote, the completely bizarre “copyright notice” on the first page is about the most ridiculous thing I have seen on a working paper recently: besides the fact that authors hold the copyright automatically without such a notice, the paper itself is literally about the social benefits of free knowledge flows! I can only hope that the copyright notice is the result of some misguided university policy.


Get every new post delivered to your Inbox.

Join 188 other followers

%d bloggers like this: