 Welfare economics is the branch of economics that studies market failure, that studies the failure of markets to deliver the efficient allocation of resources that we rely on markets we have come to expect from markets. Now the study of market failure through welfare economics offers the potential, the promise to be able to map the innovation process into the neoclassical framework by rationalizing the state interventions that we have observed historically when markets have failed. The state's role in the innovation economy complements the role of financial speculation. It funds upstream investments in science whose returns cannot be known in advance and enables commercialization of innovations and the speculation that attends it. But an historical fact, politically legitimate missions, not the observation of market failure, have driven state promotion of technological innovation at economically relevant scale. Two seminal papers each published some 60 years ago by two great and very different economists frame this discussion. The institutionalist Richard Nelson and the mathematical economist Kenneth Arrow will begin with Nelson who asks the right question, how much should we spend on basic research? Nelson specifies the neoclassical formula for the optimal allocation of resources, measuring net present value of the returns from investment versus the opportunity cost, the foregone activities that would have been funded. But he identifies the possibility of a positive externality. This positive externality justifies state action and principle if the key neoclassical conditions are met. Nelson is still active and rejects his early exercise in trying to understand the dynamics of innovation through the neoclassical production function. In fact, at an event sponsored by the Institute for New Economic Thinking in 2013, he said and I quote, market failure as defined in neoclassical welfare economics is a disastrous way to think about the role of the state. Now, as we saw in lecture one, Nelson invokes evolutionary economics as a much more productive approach. Now, Nelson goes on in this paper of 60 years ago to note that markets need not fail. He realistically asserts that when gains from invention are large and obvious, research funding is likely to be available. But how many times is it in fact true that gains from invention are large and obvious? That's a matter for empirical examination. But as we move upstream towards basic research, returns are progressively less obvious. Now uncertainty about future returns joins positive externalities as a source of market failure for investment in basic research. In 1972, the then Lord Rothschild conducted analysis of UK innovation policy for the then minister of education, Margaret Thatcher. He proposed a clear line between basic research to be funded by the state and applied research to be funded by private sector for-profit customers. But is this clear line legitimate? Donald Stokes, former dean of Princeton School of Public and International Affairs, contradicts Rothschild. He points out that the great chemist, biologist, Pasteur, discovered the principles of fermentation under a contract to increase the yields of the Belgian sugar beet growers. Pasteur's quadrant has entered the language of the economics of innovation. Finally, Nelson offers support for Schumpeter Mark II. The broad-based industrial giant simply has more potential applications for the output of R&D. Beyond Nelson's scope, this would at best relate to exploitative, not exploratory research and development and would still face the challenge of innovators dilemma defined by Clayton Christensen of Harvard. Customers of established companies want their suppliers just to do more of the same and the incumbent industrial giants don't want to cannibalize their existing businesses. Kenneth Arrow's analysis complements Nelson's, specifying three sources of market failure. Increasing returns to capital or labor violate the criterion for the efficient allocation of resources, including investment in R&D. Inappropriability, as Arrow spells it out, means that the funder cannot capture all the returns to their investment and equals Nelson's positive externalities. Uncertainty here means how can I optimize my utility or my production function over time when there are goods not yet invented which I may want to consume or use as intermediates in production? Arrow is especially famous for his collaboration with Gerard de Peux, which earned them the Nobel Prize in economics, in specifying the conditions necessary for quote, a general market equilibrium to be achieved. Crucially, that requires complete markets that enable agents to trade in all commodities ever available across the infinite array of all possible future states of the world. Arrow explicitly recognizes that such complete markets do not in fact exist. I have always read the Arrow-de-Bruh theorem as a disproof of the possibility of a general equilibrium ever existing. And Arrow's last sentence here is decisive. Our economic system does not possess markets for commodity options. Arrow goes on to address the peculiar economics of information. On the supply side, efficiency argues for free distribution of information because it's costless, which would in turn eliminate any incentive for producing such information. On the demand side, not only does information generate increasing returns, but customers can't assess its value until the information has already been purchased. Arrow, through these contradictions, explains why an efficient market in information cannot exist. Given the above, Arrow also falls back on Schumpeter Mark II. Not only does the giant corporation have more opportunities to appropriate the returns to R&D, but it can self-insure against the risk of failure. Ironically, correcting market failure depends on imperfect competition, but monopoly rents not only fund innovation, they reduce the incentive to innovate. Arrow's summary goes beyond Nelson. Not only will firms underinvest in R&D, if they can appropriate all the returns from the investment they do make, the result will be inefficient. The real-world interdependence of public and private investment in R&D demonstrates the relevance of Nelson's and Arrow's work. Patents that cite public funding have increased monotonically since World War II, especially patents awarded to corporations. Even as corporate willingness to invest in upstream science has in fact declined. Is this decline in investment in science, a function of maximizing value for short-term stockholders? Is it a function of shorter tenure of monopoly rents as the giants of the past, IBM, AT&T, Xerox, came under competitive assault? Is it in fact due to both of these forces? Moving from theory to empirics, is it possible to define the socially optimal quantum of R&D going back to Nelson's original question? If it were possible, how would the state provision and allocate it? This is where cost-benefit analysis is typically invoked. But such analysis has two sides, prior restraint versus post facto evaluation. Attempting to provide upfront analysis, upfront determination, that the benefits from investment will outweigh the costs, well, it's always easier to estimate the costs. Than to estimate the future benefits. If the rationale for state invention is to correct market failure, then state interventions will be evaluated by their achievement of an efficient allocation of resources. The inability to calculate which is what got us into this pickle in the first place. This is a black hole of self-referential attempts to escape from the reality of state and the state's role in funding innovation at the frontier. Now, as a practical matter, three scholars set out to quantify the magnitude of the gap between the social return to R&D and the value captured by the private companies funding it. The key is empirical identification of two different, two very distinct types of externality. Knowledge spillovers, the positive externality. My investment in R&D spurs additional investment in R&D by other firms. But then there's also the product market spillover, the negative externality. Successful R&D steals market share from competitors. This paper is driven by a methodological innovation. The key step is to position real firms in well-defined technology space and product market space and then measure the impact of their research and development on their own and their rivals metrics of performance. Patents, productivity, and R&D plus the change in market value are the instruments for quantifying technology and product market spillovers. Market value is used to measure the negative impact of product market spillovers. That is my successful R&D reduces your market value. This is an important point of vulnerability in the analysis. Since market value is subject to all sorts of influences, not directly related to the fundamental performance of the firm. In most cases, in fact, the empirical estimates track the theoretical predictions. R&D by neighbors close in technology space is associated with higher market value, more patenting, and increased total factor productivity. These are the positive spillovers. R&D by neighbors close in product market space, that is competitors, is associated with lower market value and with no effect on patents or total factor productivity. The authors translate their correlations into a structural model to generate quantitative estimates of returns to R&D. This is necessarily speculative, but it's also provocative. The marginal social return to research and development that they identify is almost three times the marginal private return. The team revisited the subject with more data five years later. Now the estimated wedge between social and private marginal return has increased from 37% to 44%. Thus, empirical academic work underwrites the role in the innovation economy of an active entrepreneurial state that addresses the market failures defined in theory by Nelson and Arrow. So now we turn from economic theory and empirics to political history and discovered that the effective motivation for state intervention was not market failure, it was national security. To begin, England and the First Industrial Revolution, the state's role on the demand side of technological change. Priya Satya documents the increased demand for guns by the British Army from the glorious revolution of 1688 to Waterloo in 1815 from 10,000 a year to millions per year. This demand pulled Britain's artisanal economy into radical innovations in industrial structure and made Birmingham, England the workshop of the First Industrial Revolution. Key factor, the state as the early collaborative customer whose demands were met by a radical increase in productivity due to Adam Smith's division of labor. On the supply side, it was the American Army that reduced the practice, the technological achievement that enabled mass production. The US Army's armories at Springfield, Massachusetts and Harper's Ferry, West Virginia perfected production with interchangeable parts. It started with guns, but it went on to sewing machines, typewriters, bicycles, automobiles. For our own digital revolution, the US Department of Defense was the decisive actor on both sides of the innovation economy. DoD operated through multiple channels. Its funding of research and development and the civilian spin-overs tend to be emphasized, perhaps overemphasized in the literature and the popular mind. But as with the British state in the 18th century, the role of the military as a creative and collaborative customer is fundamental. And note, in addition, the profound consequences of the Defense Department sponsoring the American Research University. The first analysis to my knowledge of the impact of Defense R&D on private R&D was recently published. Did the Defense Department crowd out private research and development? Do, for example, to an inelastic supply of research people? The answer is no. State research and development crowded in private research and development as a compliment and with the incentive for further private research. Fred Block has long been a leading scholar of the American public-private partnership that drove US leadership of the innovation economy in the second half of the 20th century. Leaving warfare and armaments out of the history of US industry is like the proverbial production of Hamlet without the prince. In addition to armories, the relevant American history includes Alexander Hamilton's report on manufacturers, Henry Clay's American system of investment in internal improvements, canals, and turnpikes. The Erie Canal itself, funded by bonds guaranteed by New York State, that made New York City the commercial and financial capital of the young republic. Federal and state land grants to railroad promoters, the Morrill Act that gave federal land to the states to fund the land grant universities, and of course, protective tariffs that allowed the American infant industries to mature into competitive standing versus the established British industries. Were any of these initiatives motivated by market failure? In fact, national development has overlapped with national security as a legitimizing mission, called out as such by Frederick List, a great neglected 19th century economist more than 150 years ago. By 1900, based on such initiatives specified by List, Germany had become the leading contender to supplant Britain as the leader at the scientific technological frontier. Germany invented the research university. Germany led the first science-based industries, chemicals and electrical engineering. Of course, from 1914, Germany then experienced the worst generation of any Western nation since the peace of Westphalia in the mid-17th century. While already the largest economy in the world by the end of the 19th century, the US lagged scientifically. During the Great Depression, state intervention was demanded as an emergency measure. But as the emergency retreated, so did the legitimacy of state intervention leading to the sharp Roosevelt recession of 1938. It was World War II that finally legitimized the role of the American state in the innovation economy. At the end of World War I, the elimination of military demand triggered a global depression. At the end of World War II, the conversion to peacetime conditions, at least in the US and to a lesser extent Britain, was cushioned by adoption of Keynes's one policy initiative that was effectively implemented, how to pay for the war through refundable taxes, rationing, and supply constraints. So the war generated a huge accumulation of liquidity that had not been consumed by inflation. And the post-war world that emerged appeared to be Schumpeter Mark II in action. In the context of this, Hart suggested a Nelson-Arrow convergence on policy that might have emerged. Would this peaceful consensus have proved stable and durable? In fact, it was the military mission that prevailed, not the social contract proposed by Vannevar Bush, who had led the Office of Scientific Research and Development throughout World War II in his report to FDR in 1945, Science, the Endless Frontier. A catalyst for the post-war continued investment of the American state in science and technology was the Korean War. Launched by North Korea in 1950, it was a relatively small war militarily, but a huge war economically and in terms of investment in R&D. In just 12 months, the defense budget went from $13 billion to $60 billion against GDP of just over $300 billion. If there had been no Korean War, it's very hard, if not impossible, to prove a counterfactual. But Hart comes close. In the evolution of the digital revolution, in the early days, the formative years, 1957 was an inflection point. The Soviet Union launched the Sputnik satellite, and DARPA, the Defense Advanced Research Project Agency, was a shocked response, along with massive increases in defense procurement and the first direct federal funding of higher education. The traitorous eight, who left shockly semiconductor to found Fairchild, were progenitors of the culture of Silicon Valley, although entrepreneurial betrayal had a much longer history than state funding of science. A generation before, for example, Walter Kreisler had left General Motors to start his own automobile company. Taking as a whole, this history suggests that we best think in terms of national systems of innovation versus market failure. Analytically, this offers a much broader synoptic frame in which to understand the economic role of the state in the innovation economy. And from a policy perspective, it integrates relevant state interventions with a complex dynamics of investment and innovation within the private sector. In the next lecture, we will turn to the complementary source of funding at the frontier, financial speculation.