Search the archive.

Despite the perception of the US as the epitome of private sector-led wealth creation, in reality it is the State that has been engaged on a massive scale in entrepreneurial risk taking to spur innovation. In this chapter four key successful examples are given: the roles of DARPA (the US government's Defense Advanced Research Project Agency), SBIR (Small Business Innovation Research), the Orphan Drug Act (the EU passed its own Orphan Drug Act in 2001, imitating the US act passed in 1983) and the National Nanotechnology Initiative. What they share is a proactive approach by the State to shape a market in order to drive innovation. The insight gained is that other than being an entrepreneurial society, a place where it is culturally natural to start and grow a business, the US is also a place where the State plays an entrepreneurial role, by making investments in radical new areas. The State has provided early-stage finance where venture capital ran away, while also commissioning high-level innovative private sector activity that would not have happened without public policy goals backing a strategy and vision.

So far I have argued that while the level of technological innovation is integral to the rate of economic growth, there is no linear relationship among R&D spending, the size of companies, the number of patents and the level of innovation in an economy. What does seem to be clear, however, is that a necessary precursor for innovation to occur is to have a highly networked economy, with continuous feedback loops established between different individuals and organizations to enable knowledge to be shared and its boundaries to be pushed back. This chapter attempts to illustrate that at the frontiers of knowledge, simply having a national system of innovation is not enough. Over time, more impressive results can be achieved when the State is a major player operating within this system. This role does not necessarily have to take place at a national level (although it can) and should not only involve long-term subsidies to certain companies ('picking winners'). Rather the State, through its various agencies and laboratories, has the potential to disseminate new ideas rapidly. It can also be nimble, using its procurement, commissioning and regulatory functions to shape markets and drive technological advance. In this way it acts as a catalyst for change, the spark that lights the fire.

In reality, the whole idea of governments crowding out the private sector is usually false. Government investment often has the opposite effect. When structured strategically it can crowd in private investment, stimulating funding that might not have happened otherwise and expanding national output, which benefits public and private investors alike. Public investment in Apollo inspired rapid advances in computing and digital technology through contracts with the private sector. Relatively small government stakes in Airbus have helped build the world’s biggest aircraft company, with operations and suppliers across Europe. The history of technological breakthroughs shows that public investment, particularly when made early in the innovation process, absorbs major uncertainties and long-term risks that private investors can be reluctant to take on. Indeed, public investment in the early, high-risk stage of areas like nanotechnology, biotechnology and green technology were critical to the later proliferation of small start-ups, some of which later scaled up. An alternative theory must thus be based on market shaping and market co-creation, a concept we return to in Chapter 6.

Our findings show that mission-oriented innovation policies generate a larger and positive effect both on the level of GDP and on private R&D investment, as compared to generic public expenditures. The high value assumed by the cumulative multiplier on impact and after four quarters might be motivated by the fact that mission-oriented innovation policies are based on contracts, partnerships and cooperative agreements between public-private organizations, which all reflect the focus these projects have on well-defined objectives (Mowery, 2010, 2012). The launch of this kind of public program requires a co-investment between the private and public sectors to finance the large fixed initial costs, such as new physical structures (e.g. laboratories and equipment), the payment of highly skilled labor and the related research activities. A higher multiplier effect could be also motivated by the fact that a great part of R&D expenditure is salary payments for R&D workers (Goolsbee, 1998), who have a higher marginal propensity to consume compared to profits’ earners. Moreover, the massive peak effect of the cumulative multiplier on impact and the dynamics of the cumulative effect in the subsequent periods (see Table 4) are in line with Perotti's (2004) findings, showing that the cumulative multiplier associated with total public investment in the US generates a peak effect on impact and declines in the subsequent time periods. Finally, the positive effect on private R&D produced by mission-oriented policies is significant on impact, whereas the effect of generic public expenditures is lower and becomes significant only after seven quarters. This confirms the idea that targeted innovation policies produce a strong and direct effect on the R&D investment decisions of firms, leading to new directions for technical change.

To sum up, our findings confirm the theoretical intuitions argued in Section 2 as well as in Deleidi and Mazzucato (2019). Specifically, mission-oriented innovation policies are able to produce a larger fiscal multiplier and to determine a stronger direct crowd-in effect than generic public expenditures. The crowd-in effect estimated on private R&D is in line with the findings put forward by Moretti et al. (2019). Conversely, our findings are in contrast with the idea that military R&D expenditures generate a crowding-out effect on private R&D investment (see, among others, Goolsbee, 1998; Wallsten, 2000; Bronzini and Iachini, 2014). Finally, consistent with the recent hysteresis literature (Yellen, 2016; Fatás and Summers, 2018), our findings show that changes in government spending produce positive and permanent effects on the output level.

Second, grantees are in fact accountable for grant-funded research. Each grant operates within a context of ongoing funding streams, reporting obligations, and repeat players. Even though any individual grant may lack its own strong accountability mechanisms, the practical need to get the next grant creates accountability for grant recipients. 19

Downside risk invites the most straightforward rebuttal. The government does not bear downside risk alone. Researchers also face downside risks from project failures. While the NIH does not require success from its funded projects - science is risky, and innovative science more so - nevertheless it is easier to generate data, and especially to publish in prestigious peer-reviewed journals, if research achieves its stated goals. This bias in favor of positive results has its own powerful negative consequences for science, 190 but it does keep some of the risk of failure squarely on the researcher. Failing to receive or renew grant funding results in a range of consequences that can hit a researcher hard, including shame among peers, the inability to hire (or the need to fire) subordinates, denial of tenure or promotion, and the end of a lab and a career. 191

The most important reporting of grant progress comes not in response to the current grant but in applying for the next grant. Grants terms are measured in years; researcher careers are measured in decades (or, at least, most researchers hope so). Failure to get subsequent grants can result in the downsizing of a lab or the end of a career, making researchers repeat players. 180 As Gallini and Scotchmer noted, the "moral hazard" of non-contingent ex ante funding for a single grant "is overcome because future grants are contingent on previous success." 181 They argue that in practice, grants "operate much like prizes, with the wrinkle that a researcher must convince the sponsor in advance that his output might be worthy of a prize. For this purpose, his reputation might suffice, and in some cases, much of the research has already been completed." 182

One of the greatest problems, which we return to in Chapter 9, has been the way in which such reductions in spending on R&D have coincided with an increasing 'financialization' of the private sector. While causality may be hard to prove, it cannot be denied that at the same time that private pharma companies have been reducing the R of R&D, they have been increasing the amount of funds used to repurchase their own shares—a strategy used to boost their stock price, which affects the price of stock options and executive pay linked to such options. For example, in 2011, along with $6.2 billion paid in dividends, Pfizer repurchased $9 billion in stock, equivalent to 90 per cent of its net income and 99 per cent of its R&D expenditures. Indeed, from 2003 to 2012, Pfizer spent amounts equivalent to 71 per cent and 75 per cent of its profits, respectively, on share buybacks and dividends (Lazonick 2014, 55). Amgen, the largest dedicated biopharma company, has re- purchased stock in every year since 1992, for a total of $42.2 billion through 2011, including $8.3 billion in 2011. Since 2002, the cost of Amgen's stock repurchases has surpassed the company's R&D expenditures in every year except 2004, and for the period 1992-2011 was equal to fully 115 per cent of R&D outlays and 113 per cent of net income (Lazonick and Tulum 201 1). The fact that top pharma companies are spending a decreasing amount of funds on R&D at the same time that the State is spending more—all while increasing the amount they spend on share buybacks—makes this particular innovation ecosystem much more parasitic than symbiotic. This is not the 'crowding out' effect: this is free-riding. Share buyback schemes boost stock prices, benefitting senior executives, managers and investors that hold the majority of company stock. Boosting share prices does not create value (the point of innovation), but facilitates its extraction. Shareholders and executives are thus 'rewarded' for riding the innovation wave the State created. In Chapter 9, 1 look more closely at the problem of value extraction and ask whether and how some of the 'returns' from innovation should be returned to the employees and State that are also key contributors and stakeholders in the innovation process.

Unfortunately the same problem seems to be appearing in the emerging clean technology sector. In 2010, the US American Energy Innovation Council (AEIC), an industry association, asked the US government to increase its spending on clean technology by three times to $10 billion annually, with an additional $ 1 billion given to the Advanced Research Projects Agency-Energy (Lazonick 201 lc). On the other hand, companies in the council have together spent $237 billion on stock repurchases between 2001 and 2010. The major directors of the AEIC come from companies with collective 2011 net incomes of $37 billion and R&D expenditures of approximately $10 billion. That they believe their own companies' enormous resources are inadequate to foster greater clean technology innovation is indicative of the State's role as the first driver of innovation or of their own aversion to taking on risks—or both.

Third, some doubt whether peer review is accurate: is it good at sorting out good ideas and grant applications from bad ones? The answer to this seems to be a cautious and qualified "yes." Figuring out whether peer review accurately identifies projects most likely to succeed is challenging; basic research, in particular, is typically likely to fail, and paradigm-changing research is perhaps the most likely to fail, almost by definition. 166 Evidence suggests that peer review can probably discriminate sound applications from seriously flawed applications. 167 However, beyond that distinction, scholars debate whether better-scored grants are actually more productive. 168

[*32] Overall, deep peer involvement, whether in the process of seeking applications and therefore identifying areas of innovation (broad or narrow) or in the process of choosing projects and individuals to fund, casts a substantially different light on the grant-funding process. Peer involvement in picking projects has its flaws; it might involve bias, it might suggest safe science, and it is certainly imperfect at identifying the best projects for funding. Similarly, the process of identifying areas of potential innovation, which relies both on peer involvement and on targeting by agency or other government actors with an eye toward social welfare priorities (or patronage, or pork), has its own flaws and idiosyncrasies. But of course, so does the principal alternative - a market-based system that relies on the incentives of private actors to decide what innovation is best to pursue, not based on evaluations of scientific merit or social welfare value, but on a calculus of what profits are appropriable through an imperfect intellectual property system or otherwise. 169 Grants aren't perfect; they're just different, and more interestingly different than is often assumed.

//DH

Making this limited case for choosing within cleantech is eased somewhat by the practical reality of implementing government grants and prizes. In the case of research grants, the government must pick winners with each project, thereby necessarily choosing one specific technology over another. Of course, the overall portfolio of grants may seek to be technology-neutral, but in practice the decision-making process typically requires choosing particular technologies. For example, according to the recently-passed 21st Century Cures Act, which includes prizes for innovation in healthcare, experts will pick particular projects.163 One might try to imagine the most technology-neutral prize possible—e.g., a prize to the first person who produces energy with no more than x units of carbon at y cost per unit of energy. In practice though, prizes are likely to be technology specific: e.g., competitions for battery, wind, or solar design. Indeed, setting broad targets for climate prizes may perversely increase the subjectivity in judging the winner, something that many supporters of prizes seek to avoid.164 And even scholars who have been critical of the government’s role in picking winners have proposed climate prizes that target certain technologies and certain sectors.165 Notably, in all of these

Policymakers may have imperfect knowledge about which innovations are likely to contribute the most to social welfare, but,

we hope, they are less likely than the utility-blind patent system to favor innovations that are affirmatively harmful to society. We do not intend to seem overly sanguine about the state’s ability to direct innovation. Certainly, research grants have supported the development of drugs and chemicals that turned out to cause grave social harms.399 Our claim is more modest. Once we recognize the existence and scale of antisocial innovation, our confidence in the patent system should decrease.

And yet the idea that government cannot pick winners is historically inaccurate.48 In fact, during periods of technological shifts, government can play a critical role in co-ordinating industrial efforts and setting standards that create markets. But they must make the decision to pick a strategy – and tilt the playing field in that direction. We saw this in the 1990s, when the government of South Korea recognized the enormous potential of high-definition (HD) technology. At the time, the electronics industry was undergoing a shift from analogue to digital products, and South Korea was still a mass-producer of analogue TVs. To build the necessary capabilities to shift to HD products, the South Korean government set up a committee dedicated to co-developing HD TV – comprised of three ministries and more than a dozen private firms, universities and government research institutes – which resulted in the creation of a ‘grand research consortium’. The consortium, which was led by the Video Industrial R&D Association of Korea, included the Korea Electronics Institute, the Korea Institute of Industrial Technology, Samsung, LG, Hyundai, Daewoo Electronics and other private-sector firms. With $100 million in combined funding from the government and the private sector, it focused on the technology transfer and absorption from the USA and Japan. The government co-ordinated the work of leading companies to develop standards for digital TV emerging from the USA while also encouraging competition between the companies.49 Meanwhile, Korean companies established research teams and centres in the USA that were close to universities and other research institutes.50 In October 1993, after the consortium presented the first prototype for digital TV broadcasting and receiving, the Korean government moved to support the second phase of the project – the industrialization and commercialization of a new prototype. Two years later, the consortium started the development and miniaturization of ASIC chips, and many companies competed to scoop the contract for the final commercial product – which was ultimately launched by Samsung in 1998.51

Not all strategies of this type succeed, of course; that is part and parcel of trial and error. But sometimes picking winners is confused with state support for troubled industries, for example when the UK government attempted to help the domestic car industry by forming British Leyland in 1968 and shipbuilding by creating Upper Clyde Shipbuilders that same year. Neither company was successful. But picking winners in the sense of backing what are deemed to be the innovations, sectors and businesses of the future is quite different from trying to keep ailing industries and companies alive. And in the narrow sense of selecting a winning business, governments, just like venture capitalists, will win some and lose some. The same year that the US government made the $535 million guaranteed loan to Solyndra, it also made a similar loan of $465 million to Tesla – now a global leader in the electric revolution that is transforming the automotive industry. China is the world’s biggest producer of pencils because it set out to develop a competitive industry, not because it started with comparative advantages in technology or the supply of graphite. Chinese state-owned firms invested in technology and labour, and the government provided cheap finance, tariff protection of domestic producers, lax forest management policies to keep wood cheap, and generous export subsidies.52

The path-dependent direction that the economy follows under 'free-market' conditions is problematic, particularly when the world is confronted with great societal challenges such as climate change, youth unemployment, obesity, aging and inequality. In addressing these challenges, the State must lead—not by simply fixing market failures but by actively creating and shaping (new) markets, while regulating existing ones. It must direct the economy towards new 'techno-economic paradigms', in the words of the technology and innovation scholar Carlota Perez. Usually, these directions are not generated spontaneously from market forces; they are largely the result of strategic public-sector decision making.

Indeed, nearly all the technological revolutions in the past—from the Internet to today's green tech revolution—required a massive push by the State. Silicon Valley's techno-libertarians might be surprised to find out that Uncle Sam funded many of the innovations behind the information technology revolution. The iPhone is often heralded as the quintessential example of what happens when a hands-off government allows genius entrepreneurs to flourish, and yet the development of the features that make the iPhone a smartphone rather than a stupid phone was publicly funded. The iPhone depends on the Internet; the progenitor of the Internet was ARPANET, a program funded in the 1960s by the Defense Advanced Research Projects Agency (DARPA), which is part of the Defense Department. The Global Positioning System (GPS) began as a 1970s US military program called NAVSTAR. The iPhone's touchscreen technology was created by the company FingerWorks, which was founded by a professor at the publicly funded University of Delaware and one of his doctoral candidates, who received grants from the National Science Foundation and the CIA. Even SIRI, the iPhone's cheery, voice-recognizing personal assistant, can trace its lineage to the US government: it is a spinoff of a DARPA artificial-intelligence project.

Overall, then, the system differs markedly from the simplified version presented in the critique of grants. Are government bureaucrats making uninformed decisions about what scientific projects get funded? Not really. It is true that staff and leaders at the NIH are involved in the process: the Center for Scientific Review processes initial applications and assigns them to review groups, Scientific Review Officers run the Scientific Review Groups in the initial peer review, NIH staff collates scores and prepares funding reports for the Councils, and NIH Directors or their delegates make the final decisions. But the key determinant of funding is initial peer review. Several scientists with expertise in the field read applications; determine how they fare on significance, investigator qualifications, innovation, approach, scientific environment, and overall impact on the field; and write up scores, critiques, and reasoning. Then those scientists meet, discuss the most promising grants, and decide their final scores - which projects are most worthy. That's mostly it. Grants are ranked, and the grants judged most worthy are funded until the funding runs out (with a bit of wiggle room).

While the process does not involve the market aggregating private information held by firms, it does involve the aggregation of relevant information. The grant applicants themselves disclose what they know of the innovation's potential value and their own capacities in the grant application. Peer reviewers see that information, have their own information about the field, and often can directly compare projects proposed by different researchers in the same field. And agency personnel can provide broader perspectives about government information. This process is a far cry from the notion of an uninformed bureaucrat simply sitting in a room and "picking [*30] winners."

Grants can fund innovation whose social welfare value exceeds the market value of the product. Rachel Sachs notes,

Where the general population's willingness and ability to pay for a particular drug track the social value it contributes, patents are thought to provide a relatively efficient way of incentivizing the development of socially valuable drugs. But each of these factors - willingness to pay and ability to pay - presents a well-known bias, through which innovation incentives will be directed away from certain types of treatments or diseases with high social salience. 216

Part IV describes the rich tools the grant system supplies to policymakers, focusing on grants' two key comparative advantages. First, grants can support innovations whose social value exceeds their appropriable market value. This describes basic research; because later applications of basic research are variable and unpredictable, it has substantial spillovers (positive knowledge externalities), and is undersupplied by private firms relative to its social benefit. 25 Private firms also generate inadequate information about which basic research is worth funding. But a panel of experienced peer reviewers, [*9] combined with disclosures from grant-seeking researchers, may be able to make precisely that determination. At a broader level, the market systematically undervalues some forms of innovation because market demand does not reflect social welfare value. A powerful example is innovation targeting diseases of the poor; because the poor often cannot pay for drugs, market signals do not reflect the social welfare benefits of developing those drugs. The grant system's reliance on non-price signals brings risks of inefficiency or cronyism, but its incorporation of non-market information also allows different, useful allocation of funds beyond what markets would pick.

Second, grants can directly support innovation enablers - the people, institutions, processes, and infrastructural resources involved in innovation - in a way largely unavailable to other forms of directed innovation incentives, especially patents and prizes. Basic research serves this role when it provides the grounding for later research, but it is only one example. Grants can develop human capital by providing training or otherwise enabling the research of young scientists who will have longer careers ahead of them. Grants can also target the processes or institutions of innovation by providing resources specifically for interdisciplinary research (to build collaborations and boundary-crossing networks) or for institutions (to provide physical or other resources for collections of individuals). Finally, they can support infrastructure, including datasets that enable future innovation, such as the Precision Medicine Initiative's All of Us dataset or the Human Genome Project (both NIH-funded).

The reality of the current grant system belies these three critiques. Part III describes the grant system as it functions today, with substantial emphasis on grants awarded by the NIH - perhaps the world's most prominent grant funder - to researchers at other institutions, and rebuts each critique. 18 First, the mechanics of grant application, review, and funding refute the narrative that grants are allocated by information-poor bureaucrats. The grant system uses a rigorous process of peer review to determine which proposals will be funded. Part of this process involves detailed applications, which requires potential grant recipients to share their own private information about the likely costs and potential value of the proposed research. The evaluation itself leverages the expertise of scientists with relevant experience and knowledge. And the entire process is coordinated by agency representatives who combine their own scientific background with knowledge about the innovation priorities of the NIH and the government more generally.

3. Competition Harms. Some innovations are designed to harm the competitive process itself. Many innovations, of course, are created to give the inventor a competitive advantage.227 When superior new products or processes harm competitors but benefit consumers, that is typically a sign of healthy competition.228 The Google search engine, for example, resulted in the quick demise of lesser search companies like AltaVista and Ask Jeeves.229 But internet users were better off. The economist Joseph Schumpeter referred to this process as “creative destruction.”230 In some cases, though, innovations are employed to harm competitors in ways that do not make consumers better off.

Predatory pricing algorithms are an example of anticompetitive antisocial innovation. Lina Khan has argued that Amazon has used its sophisticated pricing algorithm to selectively undercut the prices of rivals selling on the Amazon marketplace in an effort to eliminate or co-opt competitors.231 But the threat of algorithmic predatory pricing is broader than Amazon. Indeed, Christopher Leslie has shown how pricing algorithms facilitate predatory pricing more generally.232 In particular, artificial intelligence can make price wars easier to win by allowing predatory firms to target their rivals’ customers for exclusionary below-cost pricing while continuing to profit off their own loyal customers.233 This type of algorithmic predatory pricing will harm competition and ultimately make consumers worse off. Firms can also use pricing algorithms to collude and raise consumer prices. For example, tenants’ groups have alleged that corporate landlords use a pricing algorithm offered by software company RealPage to inflate rental prices across the country.234 The U.S. Department of Justice weighed in on the tenants’ side, observing that “[t]oday, software algorithms can be employed to fix prices—and this modern machinery may be easier and more effective than past methods of price fixing.”235

Policymakers may have imperfect knowledge about which innovations are likely to contribute the most to social welfare, but, we hope, they are less likely than the utility-blind patent system to favor innovations that are affirmatively harmful to society. We do not intend to seem overly sanguine about the state’s ability to direct innovation. Certainly, research grants have supported the development of drugs and chemicals that turned out to cause grave social harms.399 Our claim is more modest. Once we recognize the existence and scale of antisocial innovation, our confidence in the patent system should decrease.

//DH

But, governments only need to disproportionately subsidize cleantech innovation until the productivity gap between cleantech and dirtytech knowledge stocks is closed. Once the system reaches this technology tipping point, cleantech innovation will be more productive and more profitable than dirtytech innovation, meaning that innovation blocks will begin flowing to cleantech without additional government support. Thus, in contrast to a tax on carbon or an innovation subsidy for spillovers, which should be maintained in perpetuity, innovation snowballing can be addressed by a large, but temporary, government investment in cleantech R&D.

//DH

To some, the implications of innovation snowballing may seem unintuitive and to others, obvious. Whatever one’s reaction, the structure of the analysis is profoundly different from conventional economics. In the conventional story, a new technology can reduce the cost of producing cleantech. But the key piece that the conventional story misses is the dynamics: how that new technology not only lowers cleantech production costs today but draws in future cleantech innovators, progressively shifting production toward cleantech, creating a snowball effect. Those new innovators are drawn in because, as cleantech becomes cheaper to produce, future innovations are closer to commercialization and profitability. Furthermore, as cleantech becomes cheaper and closer to becoming the cutting-edge technology, more people will use the technology, increasing the scale over which cleantech innovators can deploy their innovation and make a profit. Once cleantech surpasses the tipping point of being the cheapest energy technology, cleantech innovation will snowball at a greater rate than dirtytech—thereby ensuring a virtuous cycle. The conventional story entirely misses these dynamics. Rather, it treats technological progress as if—once there is a breakthrough technology in solar that makes it cheaper to produce energy from the sun than from fossil fuels—technological progress will continue as if no such breakthrough had occurred and entrepreneurs would continue innovating in dirtytech at the same rate, even once it has been overtaken by solar. If policy is driven by this conventional economic view, the government loses the opportunity to push toward the tipping point and thereby efficiently and cheaply deploy the private sector to achieve greater gains for environmental quality at lower cost.