Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Climate Science. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 24 January 2021

Energy Innovation and Policyfree

  • John A. AlicJohn A. AlicIndependent Consultant


Stabilizing atmospheric greenhouse gases will require very large reductions in energy-related carbon dioxide emissions. This can be achieved only through continuous innovation, aggressive and ongoing. Fast-paced innovation, in turn, depends on rapid and widespread diffusion, adoption, adaptation—in short, on technological learning. These processes are integrally linked, as virtuous circles, through feedback loops embedded in economic markets. The overall dynamics are fundamentally incremental.

Pundits and policymakers, nonetheless, sometimes seem to hope that “breakthroughs” will emerge to sweep existing energy technologies aside. Such hopes are misplaced, for two reasons. If breakthroughs are construed as something “new under the sun,” they are rare and unpredictable, and policymakers have few tools to foster them. Energy technologies, after all, have been intensively explored over the past two centuries: the physical constraints are well understood and there are few reasons to expect research to lead to anything fundamentally new. Infant technologies, second, tend to perform poorly, and to be quite costly. Improvements come over time though technological learning. Inputs to this sort of learning range from field service experience to “just-in-time” research. Economic competition provides much of the driving force.

The dynamics just sketched are broadly representative of the evolutionary paths traced by past energy technologies—wind and steam power, gas turbines, nuclear power, and solar photovoltaic (PV) cells and systems. Similar paths will be followed if prospective innovations such as carbon capture and storage, small nuclear reactors, or schemes for tapping the energy of the world’s oceans begin to mature and diffuse. Over the next several decades, the world should expect to work with existing technologies in various stages of maturation that can and will—because this is inherent in the process of innovation—advance on technical measures of performance (e.g., energy conversion efficiency) and come down in costs (in most cases) through continuous improvement.

This sort of innovation is first and foremost the work of profit-seeking businesses, enterprises that conceive, develop, introduce, and market new technologies. These firms exploit publically funded R&D; just as important historically, government procurements have created initial markets, including the first PV cells and also the gas turbines that many utilities now buy for electric power generation, the early versions of which were based on designs for military aircraft. A major task for energy-climate policy is to create similarly viable market segments in which new and emerging technologies can gain a foothold, as a number of governments have done for battery-electric vehicles. Direct and indirect subsidies—financial preferences as provided in some countries for battery-electric vehicles, and market set-asides, as for biofuels in Europe, Brazil, and the United States—insulate firms from potential competition, creating opportunities to push forward technologically, overcoming early handicaps, such as high costs and poor performance, associated with emerging technologies. The implication: Effective innovation policies must provide powerful incentives for profit-seeking businesses. This is true worldwide, although mechanisms will differ from country to country.


Naive depictions of innovation often highlight paradigm-busting developments such as GPS-enabled smartphones or jet engines and their close relatives, gas turbines, with transformative impacts—in the case of the jet engine and gas turbine on both aviation and electrical power generation—the effects of which roll like waves through a formerly pacific sea. To shift the metaphor: “A global race is underway to find the great white whale of 21st century energy: the elusive super battery. It’s a race that will shape energy geopolitics for decades to come, experts say” (Vasile, 2015). This is hyperbole, of a sort that misleads. Any “super battery” will be based on well-known principles of electrochemistry and almost certainly represent developments in one or another battery system (meaning, basically, anode, cathode, and electrolyte) that scientists and engineers have been exploring for decades. Such accounts seem to mistake the origins of innovative waves for the sources of their momentum. In fact, it is ongoing technical advance—incremental innovation, sometimes plodding, sometimes inspired—that provides the driving force for transformation. Without these repeated pushes the initial impetus would die out.

Many years ago, Simon Kuznets (1972, p. 437), a pioneer in the study of innovation and 1971 Nobel Laureate in economics, observed that “a major technological innovation requires a long period of sustained improvement, and many significant complementary innovations (some of them also major but derivative) before its ramified and significant effects … are realized.” The complementary innovations may be dispersed widely across economic sectors. Many will have little visibility beyond those directly involved, as engineers, scientists, and technicians, or entrepreneurs and managers. Yet these complementary innovations—of the sort Mokyr (1990) calls microinventions—open new markets and applications, fostering new demand and stimulating still more innovation. Except that sometimes the circle turns vicious, as with fossil fuels, with what had once been viewed as incidental impacts—release of carbon dioxide (CO2) and other greenhouse gases (GHGs) to Earth’s atmosphere—creating another set of dynamics, these unsustainable.

Already, the complex of innovations responding in part to demand for low-cost energy with zero or near-zero net carbon emissions (e.g., with emissions balanced by take-up in non-atmospheric sinks) has yielded technologies capable in principle of satisfying total demand worldwide (IPCC, 2012, p. 10). So it is diffusion that will largely pace decarbonization (however interpreted). Of course innovation almost always and almost everywhere proceeds continuously: the attributes of known, existing technologies are not static, with ongoing technical gains leading to increased performance (however measured) and, in many cases, reduced costs—fostering further applications and speeding diffusion. The task for energy innovation policy is to accelerate the process.

Practitioners understand all this, even if they might articulate it differently. Yet for others, including some policymakers and much of the public, the lessons have sometimes been obscured by headlines and sound bites. Thus when “Tech billionaire” Bill Gates predicts a “‘clean energy breakthrough’ within 15 years” one must read down and do a bit of head-scratching to grasp his full meaning. Gates is arguing that the “energy miracle” will arise through something akin to Edisonian trial-and-error: “Each dead end will teach us something useful and keep us moving forward” (all quotations from Murray, 2016). This is a description of technological learning, a process in which knowledge accumulates through the repetitive diagnosis of experience, extraction of lessons learned, and corrective action, after which the cycle repeats. This in fact is true of all experiential learning. It is the (implicit) strategy followed by infants in learning to walk and talk and also the back-and-forth between theory and experiment at the frontiers of science. It is the way in which search processes for more efficient solar photovoltaic (PV) cells proceed (Shah, Torres, Tscharner, Wyrsch, & Keppner, 1999) and, with rather fewer inputs from science in the early years and more from cut-and-try engineering, the way jet engines and gas turbines developed (Scranton, 2011). The Internet too, although a very different sort of innovation, emerged from the confluence of multiple streams of mostly incremental advances in software, microelectronics, and fiber-optic telecommunications (Alic, Mowery, & Rubin, 2003).

Unsurprisingly, given the way innovation works, new technologies may have murky origins. Simultaneous invention has been common (Allen, 1983). As in the search for new knowledge through science, competitive emulation and extension is a hallmark of technological advance. Scientific discoveries sometimes lead quickly to technological innovations, as with the PV cell. In other cases, unexpected new findings heralded as preludes to innovation lead to little—the case so far with high-temperature superconductivity. Predictions of loss-free electrical power transmission, super-efficient motors and generators, and energy storage in superconducting magnetic rings followed soon after the 1986 discovery by IBM scientists in Zurich (Office of Technology Assessment [OTA], 1988). None have yet been realized.

Other innovations reach the market only to vanish. Battery-electric vehicles, common in the early years of the auto industry, lost out to internal combustion engines, then reappeared after more than half a century. The gas turbine, first demonstrated in France early in the 20th century, could not compete; efficiencies were less than 5% (the ratio of the energy theoretically available in the fuel to the useful output of the turbine), well under other available technologies (Schlaifer & Heron, 1950, p. 325). Decades later, after a succession of technical advances mostly in component parts (such as compressors), the first jet-propelled planes flew in Germany and Britain. These too were inefficient; they guzzled fuel and also needed constant maintenance. Militaries continued to invest (mostly through R&D and procurement contracts with private firms) because jet propulsion promised overwhelming advantages in aerial combat, as Table 1 shows. As costs came down and technical performance improved, utilities began, in the 1970s, to buy gas turbines for peaking power (Williams & Larson, 1988). At first, these were based on aircraft power plants. Specialized designs followed, and in the form of combined cycle installations—another old idea, using the turbine’s hot exhaust, otherwise wasted energy, as a source of heat for another power unit—gas turbines in some cases became the low-cost alternative for base-load generation (Horlock, 1995). Ramifications continue. Nearly all stationary gas turbines burn natural gas, and as utilities employed them in growing numbers the share of world electricity output produced with natural gas as fuel rose from around 15% in 1990 to nearly 22% in 2013 (IEA, 2015a, p. 586). Because gas turbines can run on almost any sort of fuel, should sustainable biofuels or hydrogen (e.g., produced with solar energy) become available, turbine power plants could produce low carbon or carbon-free electricity without the storage requirements of large-scale solar and wind energy installations.

Table 1. Jet Engine Performance, 1945–2015.

ca. 1945

ca. 2015

Maximum thrust (pounds)


> 100,000

Fuel consumption (pounds of fuel per pound of thrust per hour)


0.3 +

Thrust-to-weight ratio

~ 1.5

> 8

Time between overhauls (hours)


> 30,000

Source: Based on Alic (2007, p. 112).

Early PV cells too converted only 4–5% of incident sunlight into electricity. Decades later, PV systems were three times more efficient (Green, Emery, Hishikawa, Warta, & Dunlop, 2016), far cheaper, and had become cost-effective sources of terrestrial power generation in a rising number of locations. Public-sector procurements, by defense agencies as well as space agencies, got PV technologies started, just as for jet engines. Mandated quotas for biofuels—blending at some specified percentage with petroleum (gasoline or diesel)—function differently but aim at similar outcomes (Alic, 2015). The market set-aside insulates biofuels firms from normal economic forces, in this case price competition with petroleum, giving the more innovative firms a chance to survive long enough to bring their production costs down to levels comparable to those for gasoline or diesel fuel refined from fossil sources.

The gas turbine and PV examples illustrate the ways in which innovation normally proceeds. These sorts of gains should be prime targets of energy-climate policy. The discussion below in several places contrasts energy-climate technologies with digital computers and integrated circuits (ICs). Differences that matter for policy will be highlighted. These differences can be stark, especially in pace of innovation (slow for energy compared to digital electronics) and in physical limitations (far more constraining for energy).


Joseph Schumpeter (1934) taught us that innovation takes place through creative destruction: new technologies displace old; some firms survive, others disappear. Schumpeter also pointed to commercialization as a signpost marking the end of some often-lengthy period of conceptual exploration, idea generation, experimental development, and testing that precedes market introduction or other application. Schumpeterian perspectives, basically evolutionary, continue to underlie much of the analytical work on innovation processes, in the literatures of economics and management especially. In the usual analogies with biological evolution, idea generation replaces mutation and market processes rather than environmental pressures select innovations for survival (for one example, see Mokyr, 1990). Preoccupation with transformative innovations and disruptive impacts also reflects Schumpeter’s continuing influence, as pointed out many years ago by another pioneer in innovation studies, Nathan Rosenberg (1963, p. 424), who wrote that Schumpeter’s “towering intellectual leadership … has led to an excessive concern with … the circumstances surrounding the initial ‘breakthrough,’ and to a neglect of … the cumulative impact of relatively small innovations …”

Schumpeter worked in Europe and the United States during the first half of the 20th century and drew his insights from these two parts of the world. Attention turned to countries elsewhere after World War II, as evidence of the role of technological change in driving economic growth and development accumulated. Many studies of technology transfer and diffusion, of productivity trends, and convergence, or not, between wealthy industrialized countries and the rest have shown that national institutions and national circumstances continue to shape the origins, development, and diffusion of technology, globalization notwithstanding. Testifying to the significance of local conditions, in India, with something like one-fifth of the population lacking electricity (IEA, 2015b, p. 28), small local enterprises, many of them single proprietorships, have sprung up to market stand-alone solar installations (Singh, 2016). Many operate on a cash basis as part of India’s large informal economy.

Regardless of national system, innovation takes place chiefly in the private sector. Profit-seeking businesses conceive, develop, and introduce nearly all new technologies of consequence, albeit sometimes with substantial funding from governments, through military or other contracts.1 On occasion, innovations do come from state enterprises. From the 1950s, France, with its dirigiste approach to industrial policy, looked to Électricité de France (EDF, partially privatized in 2005) for development and sales of nuclear power equipment in competition with firms, mostly privately owned but often subsidized, based in Britain, Canada, and the United States (Cowan, 1990), and also, since 2001, Areva, which has sought to develop small, modular, gas-cooled reactors. China too has continued to pursue state-led development, especially in “strategic” industries—a group that includes coal, oil, and electrical power. The country has had mixed results in drawing in foreign know-how that will help Chinese-owned firms in moving upmarket, and also in spurring indigenous innovation (Fu & Gong, 2011). China’s rulers have tried to loosen the grip of officialdom on state enterprise; hobbled by perverse incentives, they “have historically been unable to take on the role of green innovators” (World Bank & Development Research Center of the State Council, the People’s Republic of China, 2013, p. 246).

Brazil, in some contrast, used its leverage over foreign direct investment to combine know-how resident in multinational corporations with local technological capabilities in creating the world’s only reasonably viable biofuels industry (Meyer et al., 2013; Alic, 2016). Policies put in place in the mid-1970s, when Brazil’s trade deficit ballooned as oil prices rose, took two forms: incentives for growing sugarcane—superior in terms of life cycle GHG emissions as a biocrop to maize (corn), a common feedstock for bioethanol elsewhere—and requirements that automakers build vehicles able to run on almost any percentage of ethanol mixed with gasoline. The country’s military government imposed these policies by fiat, and private firms carried out the technical work.

Wealthy economies built out their energy infrastructures long ago, during the era of industrialization that so fascinated Schumpeter. Quick fortunes could once be made in electrification, oil and petroleum, and automobile manufacture. In rapidly growing economies still building their energy systems, on the other hand, financial opportunities abound. Yet that does not guarantee innovation: poor countries tend to be short of technological and managerial know-how, and in some cases cronyism and corruption mark the preferred path to profits (just as in wealthy countries). Even so, what policymakers will have to do, in wealthy and poor countries alike, is to link energy-climate innovation with the future profit streams of private firms.

Analyzing Innovation

Schumpeterian generalizations aside, no unified view of technological (or social) innovation can be said to exist. As two of the pioneers in economic analysis of technical change noted many years ago, “Our knowledge is balkanized.”2 This remains broadly true, despite a good number of informative syntheses (see, e.g., for energy-related work: Jacobsson & Bergek [2004]; Hekkert, Suurs, Negro, Kuhlmann, & Smits [2007]; Klagge, Liu, & Campos Silva [2012]).


Innovation studies are balkanized across industries as well as disciplines, and studies of energy industries and energy innovation can seem a bit insular in light of approaches taken in exploring other sectors—perhaps because energy drew relatively little attention in academic circles until the oil crises of the 1970s and developed thereafter in isolation from the mainstreams of both innovation analysis and industry analysis. Thus readers may encounter terminology such as RD&D, for research, development, and demonstration, seldom employed elsewhere, as if demonstration were unique to energy-related industries rather than a common element in all technology development, with some energy analysts going so far as to add deployment (making RDD&D), as if technologies could be marched out like armies.

Interactions between industries that supply energy (fuels, electricity) and those that produce goods and services that consume energy (autos, construction, energy-intensive manufacturing such as primary metals and cement) ramify in ways that matter for energy-climate policy. Those who design and construct homes and office buildings seldom pay the heating and cooling bills, slowing adoption of energy-conserving design principles. Lithium-based batteries, to take a very different example, reached consumer markets in the 1970s in handheld electronic devices and, after many technical improvements, by Japanese firms especially, migrated to automobiles around 2010. While new-car purchasers, in the United States especially, have not paid much attention to fuel costs, leaving regulatory policies the primary spur to fuel- and CO2-saving innovations, economic forces bear much more directly on trucking companies and airlines. Depending on the routes flown as well as the ups and downs of prices, jet fuel accounts for 30% or more of airline operating costs, with direct effects on profitability. Airlines therefore press manufacturers of airframes for energy-saving design features, and these manufacturers in turn draw in technologies from suppliers in other industries, as do automakers dealing with regulations on fuel economy and emissions, now including CO2. To combine light weight and low cost in components such as doors, bumpers, and windshield frames, automakers evaluate composite materials, light metals such as aluminum and magnesium, and various grades of steel. Whereas in 2000 they could select from around 100 grades of steel, 15 years later steelmakers were offering some 200 grades (Gehm, 2016). Digital power-plant control systems designed around ICs and solid-state sensors cut energy consumption in both aircraft and road vehicles. Locational information promises further savings through more nearly optimal routings. For cars and trucks, vehicle-to-vehicle communications and cloud-connected driver-assistance technologies should help reduce energy-wasting traffic congestion.

As these examples suggest, “smart customers” contribute to innovation through expressed demand and feedback. Effective contributions depend on technological expertise. Here, electrical utilities compare poorly to other industries. Lumpy investment patterns have led to dependence on outside consultants and contractors rather than internal sources of expertise. Some utilities have had trouble in sorting good advice from bad and in monitoring construction work, by the evidence, most notably, of the troubled courses of more than a few nuclear power projects. Internal R&D provides an indicator of the technical capacity of an organization (Cohen & Levinthal, 1990), and since the 1990s many power companies, after privatizing, have cut back already low levels of R&D spending (Sterlacchini, 2012), suggesting further degradation in the ability to make good decisions on technologies offered by suppliers. One implication: should governments begin to mandate carbon capture and sequestration (CCS) installations on existing or new power plants, fueled by coal, natural gas, or biomass, at least some utilities might stumble. CCS systems strip CO2 from flue gases so it can be stored or otherwise kept from entering the atmosphere. Equipment is available but immature, and quite costly both to install and operate. Poor choices would push costs still higher. Electricity bills would rise, and consumer resistance with it, undercutting prospects for CCS—a technology that has, after earlier optimism, drawn increasing skepticism as a means for decarbonizing power generation (Maddali, Anand Tularam, & Glynn, 2015).


Much as policymakers, analysts, and entrepreneurs stand to benefit from useful measures of innovation, they are lacking. Input measures start and mostly stop with spending on research and development, an accounting category; like all such, it can only be defined in a somewhat arbitrary fashion. The most common output indicator, patents, has limited relevance for most industries. Inadequate measures of innovation hamper both policymaking and business management.

In the 1960s, the Organisation for Economic Co-operation and Development (OECD) codified definitions of R&D in its Frascati Manual (OECD, 2015). Originally devised by science ministries to capture the sort of work they themselves funded, the definitions—of basic research, applied research, and development—slight non-research aspects of technical advance (Godin, 2002; Freeman & Soete, 2009). Private spending must in any case be estimated based on surveys. Largely for historical reasons, service-producing firms and small enterprises have been under-represented (despite the reputation small firms enjoy as innovation seedbeds), and estimates also suffer from spotty responses and accounting and budgeting systems within firms inconsistent with the OECD definitions (especially in the service-producing sector). Some companies, finally, choose to minimize or exaggerate R&D spending to influence the perceptions of financial markets (Lev, 2005).

Over the years, survey coverage and quality have improved. Yet the larger issue remains: categories set up in the 1950s to track spending on science do not align well with the innovation-related activities of private firms. Studies have shown repeatedly that many technical advances stem from ordinary engineering, of a sort rarely considered R&D (e.g., Pavitt, 1984). Sales and marketing employees, still further removed from the usual notions of R&D, likewise spur a good deal of innovation, for instance by channeling technical information between firms (Schrader, 1991); these employees may be highly skilled, and engaged unambiguously in technical work that contributes to innovation, yet it will not be called R&D. Much the same is true of the many well-documented contributions users make to innovation (von Hippel, 1988).

On the output side, the indicators are still worse. Most patents are worthless. Leaving aside special cases such as chemicals and pharmaceuticals in which product composition can be protected, empirical studies find linkages between patents and innovation to be weak (Moser, 2013; Hall, Helmers, Rogers, & Sena, 2014). Analysts work with patents because they have nothing better at hand (e.g., Haščič & Migotto, 2015).

Shortcomings notwithstanding, innovation indicators matter for policy. Comparisons of R&D across countries and over time underpin appeals for increased spending (see Policy: Beyond R&D). When oil prices rose in the 1970s, many governments boosted public support of energy-related R&D, only to cut funding when prices dropped. Ever since, energy analysts have argued that these portfolios should be rebuilt.


Some forces in innovation ecosystems push new technologies into being, others pull. Push comes from science and invention: basic research on the genetics of algae, for example, could at some point lead to synthetic fuels more sustainable than those made from terrestrial biomass. Pull comes from market forces such as expressed demand for functional improvements and governments (regulations, defense requirements). CCS mandates, for example, would force increased private-sector spending on R&D, testing, and demonstration.

The longstanding consensus among students of innovation is that both sets of forces, technology push and demand pull are important (Mowery & Rosenberg, 1979), depending on circumstances. New products such as the microprocessor may reflect anticipated demand, even though users have no clear sense of their value before commercialization. These sorts of innovations sometimes follow from ideas “in the air,” as one of the developers of the microprocessor said of the “idea of a ‘CPU [central processing unit] on a chip’” (Faggin, 1992, p. 146). In this case, two firms designed microprocessors over the same time period, but only one, Intel, took its device to market, and then without any real sense of how big an innovation this would be (Aspray, 1997). In another way, too, this example undercuts the rhetoric accompanying many appeals for “radical” breakthroughs: while transformative, the microprocessor resulted from a straightforward exercise in chip design, without anything that might be called research.

Radical vs. Incremental Change

Hurd (1981, p. 164) provided as good a definition as any of what makes for radical innovation in calling the digital computer “a new class of object the world had never seen before.” Early computers offered big gains in speed over the adding machines and electro-mechanical punched card equipment they supplanted, yet their performance soon came to seem primitive: since 1950 computing speed has increased by perhaps 100 trillion times, some 12 orders of magnitude. These gains resulted from a very large number of mostly small advances in architecture, software, and componentry—disk storage, transistors, then IC chips. Moore’s law, depicting the number of transistors on an IC, has become widely familiar. From a handful of transistors per chip in the early 1960s, the numbers had passed into the millions by the late 1980s. Seen from inside the microelectronics industry, this was a consequence of a great many innovations. Some had disruptive consequences. The shifts from bipolar chips in the 1970s to MOS (metal oxide semiconductor), and then to a particular type of MOS, complementary metal oxide semiconductor, or CMOS, left some companies behind—creative destruction at work. Yet only the most detailed Moore’s law plots reveal their impacts (as bumps or wiggles in the trend over time). What seemed disruptive, arguably radical, from the inside, from the outside appeared incremental. Much of the time, radical innovation is in the eye of the beholder.

Beyond this, radical innovations display three general characteristics. First, as illustrated by the microprocessor, reduction to practice often follows from pursuit of objectives “in the air.” Unsurprisingly then, simultaneous or near-simultaneous developments have been common, as with the microprocessor—and the jet engine, developed independently from the mid-1930s in Britain and Germany. Third, even though early jet engines worked well enough to power fighter planes, they exhibited numerous and quite serious limitations, as related earlier.

We can also distinguish between two categories of “something new.” Jet engines/gas turbines and digital computers represent combinations or recombinations of system components, some of them preexisting. More efficient compressors, based on industrial practice, helped make jet engines practical; so did heat-resistant alloys developed decades earlier for metal cutting, from which high-temperature components were fabricated. In early digital computers, vacuum tubes originally developed for radios served as electronic switches. System-level developments contrast with indivisible, unitary innovations such as PV cells and the transistors that replaced vacuum tubes in computers. At the same time, if photovoltaics represent discontinuity, something new “under the sun,” the basic physical phenomenon, conversion of light into electricity, had been known for more than a century (Shah, Torres, Tscharner, Wyrsch, & Keppner, 1999); it was just that no one had figured out how to exploit the effect.

The radical/incremental distinction, in sum, should be viewed in terms of impacts rather than origins. Box 1 includes three further energy examples, illustrating in various ways the common pattern of high costs and poor performance in infancy.

Box 1. Ongoing Innovation: Three Examples

Fuel cells convert the chemical energy in a carrier such as hydrogen into electricity with no intervening steps. First demonstrated in the 19th century, their attractions include high efficiency and better power density than common batteries (i.e., rapid release of energy). High costs and short lifetimes, associated with expensive and short-lived catalysts such as platinum, restricted early applications to spacecraft (Perry & Fuller, 2002). Since the 1970s, auto companies and suppliers have worked to develop practical fuel cells as alternatives to batteries for electric vehicles. Commercialization began in 2015–2016, with small numbers of fuel cell–powered autos trickling into selected markets (Carney, 2015).

Flow batteries, unlike fuel cells, can be recharged electrically; unlike conventional batteries, they store chemical energy in liquid rather than solid electrolytes. This makes them suitable for large-scale storage in conjunction with intermittent renewable power sources such as wind and solar. Technical feasibility was demonstrated in the 1970s, but investors balked at the front-end capital costs and R&D slowed (Boicea, 2014). Since about 2005, R&D has picked up, responding to new demand for capacity leveling and backup as renewables increase their share of electric power output.

In a third story yet to be fully resolved, numerous companies have pursued processing of the cell walls of plant matter to make cellulosic biofuels. If sourced from agricultural residues, cellulosic feedstocks will reduce price competition between food and fuel and lessen land-use impacts (land clearing to grow either food or bioenergy crops releases substantial amounts of CO2). U.S. legislation has required blending of cellulosic biofuels with gasoline since 2009, a mandate unmet for lack of production capacity, the technology having proved unexpectedly recalcitrant. Optimistic firms plunged into the guaranteed market, and within a few years a number had exhausted their financing and entered bankruptcy—destruction without, as yet, much creation (Alic, 2015). Nor have the failures been limited to small, entrepreneurial start-ups: Abengoa, a diversified multinational headquartered in Spain with biofuels among its lines of business, sought protection from its creditors in 2015 (Neumann & Román, 2015). Even so, a number of commercial-scale plants opened, beginning in 2014. Longer-term viability remains to be determined; like any synthetic, cellulosic fuels will have to compete on price with petroleum, absent direct or indirect subsidies (price supports, market set-asides).

Technological Learning

Nuclear power, with solar photovoltaics one of only two fundamentally new energy conversion systems to emerge in the second half of the 20th century, illustrates a further aspect of performance improvement over time: called by Nathan Rosenberg (1982) learning by using. In the United States in the mid-1970s, capacity factors for nuclear power—the ratio of delivered electrical output to that theoretically available (i.e., in round-the-clock operation)—averaged only 50–60%. By the early 2000s, the average had climbed to 90% (EIA, 2015, p. 129). Most of the plants in question had been designed in the 1960s: their “technology” was largely fixed even as capacity factors rose. What happened? Technicians, engineers, and managers identified and implemented operating practices and maintenance procedures that reduced the frequency of unexpected outages and extended the intervals between planned shutdowns. Only then did nuclear power become a reliable source of base-load generating capacity.

Performance gains have many sources. A wind turbine manufacturer may redesign a bearing mount to increase rigidity and reduce vibration in response to reports from the field. This is a minor change that might have major effects on in-service performance. Flutter in jet engine/gas turbine fan and compressor blades—destructive vibrations—remains a poorly understood phenomenon (Vahdati, Smith, & Zhao, 2015). Yet solutions can usually be found, through instrumentation, computer simulation, and design modifications, if the phenomenon turns up during testing or early service. In some cases, companies turn to what might be called “just-in-time” research; for examples, see Consortium for Science, Policy & Outcomes (CSPO) (2012, p. 18). In such ways innovation proceeds.

Analysts often represent gains in performance over time as learning curves or experience curves based on figures-of-merit such as cost, productivity, or, for energy systems, efficiency (Nagy, Farmer, Bui, & Transcik, 2013). Moore’s law is one example. Figure 1 shows gains over several decades in jet engine propulsive efficiency (not the same as energy conversion efficiency).

Figure 1. Efficiency Gains in U.S. Military Jet Engines.

Notes: Thrust-specific fuel consumption (vertical axis) at rated military power (without afterburning). Turbofans, or fan jets, bypass some air around the core of the engine, increasing the efficiency with which thrust is generated, hence the thrust-specific fuel consumption (the quantity plotted); thermal efficiency (the parameter of interest for applications other than jet propulsion) does not change.

Dates on the horizontal axis correspond to the end of development and testing and beginning of low-rate production. They are approximate to the nearest year. Over the first two decades especially, engine manufacture commonly began well before the completion of development, with design modifications subsequently incorporated in efforts to resolve stubborn technical difficulties.

Source: CSPO (2012, p. 16).

Whether or not a break in trend lines, as in Figure 1, signals relatively fundamental innovation depends on the criteria. Turbofan (or fan jets) yielded significant gains in aircraft fuel consumption because only a fraction of the intake air passes through the compressor; the rest bypasses the interior part of the engine, exiting through an external fan at a much lower velocity than the exhaust gases leaving the turbine at the rear. This, according to physical laws, converts energy into thrust more efficiently. It was by no means a new idea when implemented: one of the co-developers in Britain of the jet engine, A. A. Griffith, “drew an engine scheme in 1941 with a bypass ratio of 8, a suggestion not far from what is now regarded as optimum for subsonic propulsion” (Cumpsty, 2003, p. 157). What made turbofans practical two decades later? Better components, such as more efficient compressors, resulting in part from more accurate mathematical analysis of blades and flow passages, aided as time passed by digital computation. As an innovation, the turbofan configuration has no relevance for units designed to produce shaft power, as in electric power generation; in these applications, all the gas must pass through and drive the turbine stages; better components, of course, do help. Turbofans are also quieter. They count as major innovations, in aircraft propulsion only, and as another idea “in the air” long before reduction to practice.

By contrast with gains in computing performance, the rates of improvement for jet engines/gas turbines must appear painfully slow, less than a single order of magnitude. When Pratt & Whitney introduced a geared turbofan engine in 2015 (another old idea newly commercialized), airlines could expect fuel burns to decline by around 15% (Economist, 2016)—certainly significant, but no sort of break from the longer-term trend.

PV performance too shows steady performance gains, within constraints set by physical laws. For the original and still most common PV cell, single-crystal silicon, efficiencies have roughly tripled over six decades or so, to a little over than 15% (Green, Emery, Hishikawa, Warta, & Dunlop, 2016). Considerably greater efficiencies can be achieved, at higher cost, with other materials and cell configurations, and much work goes into alternatives such as quantum dots and intermetallics called perovskites. To those directly involved some of this will seem quite fundamental—to almost anyone quantum dots must seem exotic (the name refers to nanoscale regions that confine charge carriers so that incident sunlight can strip them off in greater numbers). Yet for energy-climate policy, if not for research policy, what matters is overall PV system cost, and the share of these costs traceable to the PV cells themselves has already fallen below 30% (Lewis, 2016).

Publicly Funded R&D

Each year the International Energy Agency (IEA) reports spending by member countries on “public energy technology research, development and demonstration,” or RD&D. During the 1970s, spurred by rising oil prices in the wake of Arab embargos and the Iranian revolution, totals rose to a little over $20 billion.3 As the second oil shock faded away, RD&D budgets declined to the $10 billion range. Around the turn of the 21st century governments began to rebuild their portfolios, with the 2014 total coming to about $17 billion. The United States topped the 2014 RD&D list at $6.3 billion, followed by Japan at $3.3 billion and France and Germany as the only other IEA members spending more than $1 billion.

The IEA adds demonstration to the standard R&D categories.4 For most countries in most years, this adds relatively little to reported totals. The presumption has been that government support will often be necessary for commercialization of energy technologies. Private firms generally undertake testing and demonstration as part of normal engineering activity, but for big, costly, and technically complex undertakings they may be unwilling to bear the full risks. Historically, nuclear power was the principal case, taking nearly three-quarters of all IEA-reported funding in the 1970s (IEA, 2015c). Government demonstration funds have also flowed to synthetic fuels plants, so-called clean coal technologies, and CCS.

Comparisons with spending by firms that make energy-consuming products and systems point to the continuing inadequacy of public spending. R&D for Pratt & Whitney’s geared turbofan engine cost the company a reported $10 billion–plus (over more than two decades, a slow pace because oil prices were low over much of the period) (Economist, 2016). For Boeing’s 787, designed and manufactured with lightweight composite materials to reduce fuel consumption per passenger-mile, the R&D bill came in at something over $28 billion (the 787’s troubled launch contributed to cost overruns) (Economist, 2015). By contrast with such totals, public spending on energy-climate RD&D can only look modest. The “Mission Innovation” plan unveiled in at the Paris Conference on Climate Change in late 2015 calls for a doubling of “governmental and/or state-directed clean energy research and development investment over five years” (Mission Innovation, 2015). That will make a difference. How much of a difference will depend on how effectively governments spend the added funds.

Big government programs such as those that have targeted nuclear power, synthetic fuels, fusion energy, and CCS attract attention and evaluation. Governments also support small-scale energy-related research projects, and in very large numbers—work conducted in universities, in public laboratories, and sometimes in industry. These projects and problems are too many and too specialized for routine outside assessments—who but experts in the same or closely related fields can reach reasoned judgements concerning quantum dot research?—and internal reviews can be self-serving. Managerial oversight has also suffered because energy ministries, which historically served multiple constituencies, often fighting for favors (coal vs. oil, fossil fuels vs. nuclear vs. renewables), had no compelling focus on which to base priorities. Comparisons with defense ministries are revealing (CSPO, 2012). Despite widespread and sometimes justified complaints of “waste, fraud, and abuse,” defense agencies, with their existential missions, have mostly done a reasonably effective job of supporting technology development.

Policy: Beyond R&D

Policies that leave the technical work to private firms have worked reasonably well in fostering energy-related innovation. Restrictions on sulfur dioxide emitted by coal-burning power plants led to rapid technical improvements by suppliers competing to sell scrubbers to utilities (Alic, Mowery, & Rubin, 2003, pp. 31–32). Automobile fuel economy has increased quite dramatically worldwide since the 1970s (BP, 2016, p. 24), for reasons including mileage standards. Regulations capitalize on the ability of profit-seeking firms to manage R&D effectively, skills that firms in any industry must build if their competitive success depends on management of technology. This is one of the advantages of carbon pricing as a route to mitigation of climate change. Low energy prices dampen incentives for innovation. Raising the costs of energy, through a carbon tax or through restrictions on CO2 emissions, would induce private spending on R&D and give firms specific targets to aim for, which is conducive to effective management. The effects would differ for nuclear power compared to plants burning fossil fuels (and biomass), as illustrated in Box 2.

Box 2. Carbon Pricing as a Focus for R&D and Power Plant Design

Energy conversion efficiency in fuel-burning power generation leveled off decades ago in the range of 30–40%, for reasons of costs more than technology (Hirsh, 1989, p. 4). As any thermodynamics text will explain, peak operating temperatures determine the efficiency of so-called heat engines, a class that includes both steam and gas turbine plants. For gas turbines (and jet engines), hot-section temperatures have limited efficiency from the beginning. Over the years, better materials—“superalloys” tailored to withstand ever higher temperatures—have been found, along with protective surface coatings, and engineers have learned to design component parts with integral cooling passages. Operating temperatures have climbed steadily for decades (although they have yet to approach the maximum, set by the flame temperature of fuels burning in air) and so has efficiency (Perepezko, 2009). Both the materials—complex alloys made with expensive starting materials (nickel, cobalt, molybdenum, niobium) refined to high levels of purity—and their processing (for the most critical applications, oriented columnar crystals)—are costly. Yet this makes little difference in overall cost structures: the most expensive parts are needed only in the hottest locations in the engine, and they make up a minor share of total manufacturing costs.

Constraints for steam plants differ. These plants are massive compared to gas turbines, built with pipes, fittings, valves, and components such as boilers, superheaters, reheaters, and the turbine itself that are expected to operate reliably for decades. (Gas turbines are easier to maintain, repair, and overhaul.) As a result, materials costs have been a limiting constraint. The materials of choice for high-temperature, high-pressure service have been alloy steels less exotic than the superalloys found in the hot sections of gas turbines. They are produced in large volumes compared to superalloys, at lower costs (but still high compared to more common grades of steel), in compositions that permit cutting, shaping, and welding. In essence, gas turbines use the best materials available, almost regardless of cost, to achieve the best possible performance, while steam plants are designed on the basis of trade-offs between materials cost and performance. In principle, it is easy enough to raise steam plant efficiency: the means (e.g., so-called supercritical and utrasupercritical cycles—see Breeze, 2012, for a brief account) have been textbook staples for generations. In the traditional approach to steam plant design, front-end capital costs will be estimated and balanced against the expected lifetime savings in fuel purchases that might be achieved through higher operating temperatures.

A price on carbon, if it did not rule out new fossil fuel plants entirely, would shift the design point toward higher temperatures. Materials suppliers would ramp up R&D on alloys suited to supercritical and ultrasupercritical plants. (China has already begun to insist on such plants [International Energy Agency, 2015d, p. 30), and India could follow suit [International Energy Agency, 2015a, pp. 88 and following].) For gas turbine plants, carbon pricing would have little effect, because it is the availability of the best materials, more than their costs, that has been the binding constraint. R&D would continue much as in the past (ceramics have long been a target). The viability of nuclear plants should increase in principle (in practice public attitudes would perhaps continue to govern), and a sufficiently high price on carbon might well erase their capital cost disadvantages. Carbon pricing would also encourage R&D on new reactor configurations, standardized, passively safe, and modular (to control manufacturing and construction costs). In such ways does policy choice influence technological trajectories.

Energy-Climate Technology Policies

Any policy classification must seek some sort of balance between over-simplification and over-complication. The main categories of energy-climate innovation policies include energy prices themselves, carbon pricing, whether direct (e.g., a carbon tax) or indirect (e.g., quantitative restrictions such as cap-and-trade), other environmental regulations (e.g., restrictions on other GHGs), and measures that can that target particular technologies (e.g., R&D on high-temperature materials, subsidies for battery-electric vehicles). Only this last group will get further attention. And while “nontechnology” measures such as competition policy (antitrust) can have considerable impact on innovation (Alic, Mowery, & Rubin, 2003), discussion of these too will be omitted.

Table 2 then groups policies into in three broad categories:


Direct financing of knowledge creation through R&D


Measures that induce private spending on innovation or foster commercialization through knowledge utilization


Diffusion-related policies intended to speed applications through information and learning

Table 3 subdivides each of the three. Some of the 16 categories in Table 3 could be further segmented. Financial subsidies, for instance, come in many varieties. Still, the threefold division highlights the extent to which governments can, if they choose, call on non-R&D policies to spur innovation, including adoption and diffusion, of particular technologies and technology families. Category III measures that foster individual, organizational, and social learning have been especially undervalued.

Table 2. Major Policy Categories for Support of Innovation.a



Energy-Climate Examples

I. Funding for Knowledge Generation through R&D

Discovery of new knowledge and refinement or extension of existing knowledge in disciplinary fields, including social sciences, and across disciplines. May or may not be linked with particular societal objectives. Time horizons may be short or long.

Research on perovskite solar photovoltaic cells.

Engineering development and refinement of processes for capturing CO2 from fossil fuel–burning power plants.

Consumer attitudes concerning energy-efficient appliances.

II. Direct or Indirect Support for Commercialization

Foster marketplace introduction of new technology through some form of subsidy for either product conception and design (goods or services) on the supply side or purchase and application by users on the demand side.

Low-interest loans for investment in production capacity for electric-vehicle batteries; customer rebates for purchase of qualifying vehicles.

III. Support for Knowledge Diffusion

Articulation, elucidation, and evaluation of technical knowledge.

Amplification and accelerated circulation and application of knowledge.

Financial support for specialized training of engineers and technicians.

Development and codification of technical specifications for drop-in biofuels.

Educational websites for informing consumers.

Notes: aExcludes regulatory measures, which sometimes act as powerful inducements to innovation.

Source: Based on Alic, Mowery, and Rubin (2003).

Table 3. Sixteen Policies for Supporting Innovation.



I. Funding for Knowledge Generation through R&D

1. R&D contracts with, or grants to, nonprofits

Many governments emphasize basic research conducted in universities.

2. Intramural R&D by government laboratories

Often mission-oriented (e.g., defense, agriculture).

3. R&D contracts with private firms (fully funded or cost shared)

Normally mission-oriented; some governments (and, e.g., the European Union) support R&D with the intent of improving industrial competitiveness.

4. R&D contracts with consortia or collaborations

Proprietary interests of participating organizations may limit R&D to generic, pre-competitive work.

5. R&D tax credits

Unlikely to alter firms’ risk/reward assessments. Can be difficult to target.

II. Direct or Indirect Support for Commercialization

6. Patents

The stronger the protection, the weaker the incentives for diffusion through imitation or circumvention.

7. Tax credits or production subsidies for firms bringing innovations to market

Aim to push technologies into the marketplace from supply side.

8. Tax credits or rebates for purchasers of new technologies

Aim to create demand pull, in contrast to technology push (above).

9. Procurement

Powerful stimulus when government is a major customer, as in defense.

10. Demonstration

Intended to validate technologies viewed by private firms as too risky for investment of own funds in design, development, and testing (as illustrated by, e.g., first-generation nuclear power plants).

11. Monetary prizes

Administratively simple, once rules have been set, but effectiveness largely unresolved (historically poor).

III. Support for Knowledge Diffusion

12. Education and training

Many established channels act slowly (e.g., university degree programs).

13. Codification of technical knowledge (e.g., screening, interpretation, and validation of R&D results, support for databases)

Often linked with regulation, as for aviation safety.

14. Technical standards

Compromises among competing interests may lock in inferior technologies; governments not infrequently set standards that advantage local firms relative to foreign-owned companies.

15. Technology/industry extension

Time consuming, costly to reach large numbers of firms or individuals.

16. Publicity, persuasion, consumer information

Competing interests may attenuate, perhaps distort, the message.

Source: Based on Alic, Mowery, and Rubin (2003).

Governments everywhere acknowledge the necessity of basic research (Category I). Not all countries can expect to contribute much to new science, yet participation builds human and organization skills, hence absorptive capacity (Cohen & Levinthal, 1990), which is vital for adoption and diffusion. More broadly, new knowledge feeds incremental advance and on occasion may yield discoveries leading to more radical innovations. The chief questions for Category I policies concern priority-setting in specialized fields that only experts can understand in depth, and adjusting these priorities over time. In the eyes of most observers, peer review works, if imperfectly, for extramural research. Oversight of work conducted in a government’s own laboratories has sometimes been problematic. When public agencies fund questionable extramural R&D, and unsatisfactory outcomes result, interested parties, including research groups that competed for funds and lost, will complain; if money is misspent internally, those on the outside may not even know.

The non-R&D policies in Categories II and III pull innovations into applications and economic markets. These are essential for commodity products that must compete on price, absent subsidies, with locked-in energy products and energy services such as fossil fuels, some of which have been subsidized for decades. Pitfalls for financial incentives in Category II begin with rewards channeled to politically influential supplicants. Other Category II policies, such as innovation prizes, have questionable records, notwithstanding their popularity in some circles (Khan, 2015).

Category III policies generally draw less attention than either R&D spending or financial subsidies. This is unfortunate. Technical information circulates continuously among economic actors, moving across industry boundaries and national borders through formal channels such as inter-firm licensing and strategic alliances (e.g., Kogut & Zander, 1993), as well as less formal ties among engineers and scientists (Schrader, 1991). Information exchange amplifies the dynamics within virtuous circles linking R&D with customers and end users. Governments should give Category III policies greater attention (by, for instance, minimizing the blockage to information exchange created by intellectual property protections).

Most of all, policymakers need to recognize that it is demand that fuels the sort of just-in-time research needed to pull new technologies into the global energy sector. This has happened with military technologies, as illustrated by rapid advances in aviation technology at the time of World War I and the jet engine at the time of World War II (Alic, 2007). If, for example, means are found of protecting ocean energy (and offshore wind energy) equipment from the long-term corrosive assaults of seawater and the brute force of ocean waves and currents, it will be because the forces of demand helped guide and prioritize R&D. This is one of the chief reasons why governments should continue to emphasize Type II measures, despite frequent criticism by economists and others of subsidies.

Even if carbon pricing becomes widespread, governments will continue to work with policy portfolios from measures such as those in Table 3. In principle, portfolio construction begins with identification of technologies with the greatest potential for achieving the most important objectives—needless to say, a politically fraught task. This is part of the reason the Intergovernmental Panel on Climate Change (IPCC) takes a highly granular approach to technical analysis, assembling massive amounts of information, without making much of any effort to convey a sense of priorities. (The International Energy Agency’s broad-brush approach, at something of an opposite pole, sticks mostly to generalities on both technologies and policies, as for example: “A systems approach that implements integrated policies in whole buildings or multiple buildings can facilitate synergies across different technologies and actors …” [IEA, 2016, p. 67].) Yet policy choice depends in the end on lawmakers, who cannot always call on staffs with the ability (and the time) to sort through and untangle outside assessments, or even those of reports issued by agencies of their own government. The consequences are familiar. Vacuums open, making it easier for the politically powerful to get their way, as illustrated by corn ethanol interests in the United States and palm oil biodiesel interests in East Asia.

Roadmaps with expert rankings of technical prospects—for example, for the many possibilities in solar PV (e.g., new materials such as perovskites, layered cell structures, concentrators), with concrete R&D objectives, timelines, and recommended funding levels—could help decision-makers, the more so if revised at frequent intervals to reflect the latest research findings. (The IEA produces what it calls roadmaps, but these are better viewed as scenarios.) While politics will always intervene, policymakers can only benefit from expert advice on, say, how best to support adoption of flow batteries for grid storage and backup: What, for instance, might be the relative merits of investment grants and subsidies versus aggressive standards requiring utilities to invest in or purchase renewable power to create demand pull for storage? These are the sorts of questions that governments seeking better energy-climate innovation policies will have to address.


Technological innovation is fundamentally incremental; breakthroughs exist mostly in the eye of the beholder, and the usual precondition for major impacts will be continuous improvement through incremental innovation. Continuous improvement itself has many sources other than R&D, and policymakers have more tools for encouraging incremental innovation than for seeking to foster radical technological change. Because most innovation (as opposed to research) takes place in private, profit-seeking firms, government policies should balance support for “technology push” through R&D spending with support for “technology pull” from markets and end users. This implies that, while governments should budget greater sums for energy-climate R&D, it is even more important that they manage publicly funded R&D effectively. What will be needed in the years ahead are practical strategies and structured priorities that capitalize upon and provide incentives for private firms, guided by realistic roadmaps rather than the wish lists that too often emerge from government agencies with their own bureaucratic agendas.


  • Alic, J. A. (2007). Trillions for military technology: How the Pentagon innovates and why it costs so much. New York: Palgrave Macmillan.
  • Alic, J. A. (2015). Biofuel battles: Politics, policy, and the Pentagon. Energy Research and Social Science, 10, 10–18.
  • Alic, J. A. (2016). Decarbonizing transport: What role for biofuels? In T. Van de Graaf, B. K. Sovacool, A. Ghosh, F. Kern, & M. T. Klare (Eds.), Palgrave handbook of the international political economy of energy (pp. 397–416). London: Palgrave Macmillan.
  • Alic, J. A., Mowery, D. C., & Rubin, E. S. (2003). U.S. technology and innovation policies: Lessons for climate change. Arlington, VA: Pew Center on Global Climate Change.
  • Allen, R. C. (1983). Collective invention. Journal of Economic Behavior and Organization, 4, 1–24.
  • Aspray, W. (1997). The Intel 4004 Microprocessor: What constituted invention? IEEE Annals of the History of Computing, 19(3), 4–15.
  • Boicea, V. A. (2014). Energy storage technologies: The past and the present. Proceedings of the IEEE, 102(11), 1777–1794.
  • BP. (2016). BP world energy outlook 2016 edition. BP p.l.c.
  • Breeze, P. (2012, April 1). Pushing the steam cycle boundaries. Power Engineering International.
  • Carney, D. (2015, November 2). Fuel cell futures: No longer a dream. Automotive Engineering, 14–21.
  • Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35(1), 128–152.
  • Consortium for Science, Policy & Outcomes (CSPO). (2009, September). Innovation policy for climate change. Washington, DC: CSPO.
  • Consortium for Science, Policy & Outcomes (CSPO). (2012, March). Energy innovation at the Department of Defense: Assessing the opportunities. Washington, DC: CSPO.
  • Cowan, R. (1990). Nuclear power reactors: A study in technological lock-in. Journal of Economic History, 50(3), 541–567.
  • Cumpsty, N. (2003). Jet propulsion (2d ed.). Cambridge, U.K.: Cambridge University Press.
  • Economist. (2015, November 7). New aircraft: Take-off delayed.
  • Economist. (2016, January 2). Flying’s new gear.
  • Energy Information Agency (EIA). (2015, November 24). Monthly energy review. DOE/EIA-0035(2015/10). Washington, DC: EIA.
  • Fagerberg, J., & Mowery, D. C. (2006). Oxford handbook of innovation. New York: Oxford University Press.
  • Faggin, F. (1992, March). The birth of the microprocessor. BYTE, 17(3), 145–150.
  • Freeman, C. (1995). The “National System of Innovation” in historical perspective. Cambridge Journal of Economics, 19(1), 5–24.
  • Freeman, C., & Soete, L. (2009). Developing science, technology and innovation indicators: What we can learn from the past. Research Policy, 38, 583–589.
  • Fu, X., & Gong, Y. (2011). Indigenous and foreign innovation efforts and drivers of technological upgrading: Evidence from China. World Development, 39(7), 1213–1225.
  • Gehm, R. (2016, March). Multi-material structures move MPG upward. Automotive Engineering, 18–21.
  • Godin, B. (2002). The numbers makers: Fifty years of science and technology official statistics. Minerva, 40(4), 375–397.
  • Green, M. A., Emery, K., Hishikawa, Y., Warta, W., & Dunlop, E. D. (2016). Solar cell efficiency tables (version 47). Progress in Photovoltaics: Research and Applications, 24, 3–11.
  • Hall, B., Helmers, C., Rogers, M., & Sena, V. (2014). The choice between formal and informal intellectual property: A review. Journal of Economic Literature, 52(2), 375–423.
  • Hall, B. H., & Rosenberg, N. (2010). Handbook of the economics of innovation. Vols. 1 and 2. Amsterdam: Elsevier.
  • Haščič, I., & Migotto, M. (2015, June 22). Measuring environmental innovation using patent data. OECD Environment Working Papers No. 89. Paris: OECD.
  • Hekkert, M. P., Suurs, R. A. A., Negro, S. O., Kuhlmann, S., & Smits, R. E. H. M. (2007). Functions of innovation systems: A new approach for analysing technological change. Technological Forecasting and Social Change, 74, 413–432.
  • Hirsh, R. F. (1989). Technology and transformation in the American electric utility industry. Cambridge, U.K.: Cambridge University Press.
  • Horlock, J. H. (1995). Combined power plants—Past, present, and future. Transactions of the ASME: Journal of Engineering for Gas Turbines and Power, 117, 608–616.
  • Hurd, C. C. (1981). Early IBM computers: Edited testimony. Annals of the History of Computing, 3(2), 163–182.
  • Intergovernmental Panel on Climate Change. (2012). Special report on renewable energy sources and climate change mitigation—Summary for policymakers: A report of Working Group III of the IPCC and technical summary. Geneva, Switzerland: IPCC.
  • International Energy Agency (IEA). (2015a). World energy outlook 2015. Paris: IEA.
  • International Energy Agency (IEA). (2015b). India energy outlook: World energy outlook special report. Paris: OECD.
  • International Energy Agency (IEA). (2015c). Key trends in IEA public energy technology research, development and demonstration (RD&D) budgets. Paris: IEA.
  • International Energy Agency (IEA). (2015d). Tracking clean energy progress 2015. Paris: OECD.
  • International Energy Agency (IEA). (2016). Energy technology perspectives 2016. Paris: OECD.
  • Jacobsson, S., & Bergek, A. (2004). Transforming the energy sector: The evolution of technological systems in renewable energy technology. Industrial and Corporate Change, 13(5), 815–849.
  • Kelly, P., & Kranzberg, M. (Eds.). (1978). Technological innovation: A critical review of current knowledge, Volume I—The ecology of innovation. San Francisco: San Francisco Press.
  • Khan, B. Z. (2015). Inventing prizes: A historical perspective on innovation awards and technology policy. Business History Review, 89, 631–660.
  • Klagge, B., Liu, Z., & Campos Silva, P. (2012). Constructing China’s wind energy innovation system. Energy Policy, 50, 370–382.
  • Kogut, B., & Zander, U. (1993). Knowledge of the firm and the evolutionary theory of the multinational corporation. Journal of International Business Studies, 24(4), 625–645.
  • Kuznets, S. (1972). Innovations and adjustments in economic growth. Swedish Journal of Economics, 74(4), 431–451.
  • Lev, B. (2005). Remarks. In C. Corrado, J. Haltiwanger, & D. Sichel (Eds.), Measuring capital in the new economy (pp. 562–566). Chicago: University of Chicago Press.
  • Levin, R. C., Klevorick, A. K., Nelson, R. R., & Winter, S. G. (1987). Appropriating the returns from industrial research and development. Brookings Papers on Economic Activity, 3, 783–831.
  • Lewis, N. S. (2016). Research opportunities to advance solar energy utilization. Science, 351(6271), aad1920.
  • Liu, H., & Liang, D. (2013). A review of clean energy innovation and technology transfer in China. Renewable and Sustainable Energy Reviews, 18, 486–498.
  • Maddali, V., Anand Tularam, G., & Glynn, P. (2015). Economic and time-sensitive issues surrounding CCS: A policy analysis. Environmental Science & Technology, 49, 8959–8968.
  • Meyer, D., Mytelka, L., Press, R., Dall’Oglio, E. L., Teixeira de Sousa, P., Jr., & Grubler, A. (2013). Brazilian ethanol: Unpacking a success story of energy technology innovation. In A. Grubler & C. Wilson (Eds.), Energy technology innovation: Learning from historical successes and failures (pp. 275–291). Cambridge, U.K.: Cambridge University Press.
  • Mission Innovation. (2015, November 30). Mission innovation: Accelerating the clean energy revolution: Joint launch statement. Retrieved from
  • Mokyr, J. (1990). The lever of riches: Technological creativity and economic progress. New York: Oxford University Press.
  • Moser, P. (2013). Patents and innovation: Evidence from economic history. Journal of Economic Perspectives, 27(1), 23–44.
  • Mowery, D., & Rosenberg, N. (1979). The influence of market demand upon innovation: A critical review of some recent empirical studies. Research Policy, 8, 102–153.
  • Murray, J. (2016, February 24). Bill Gates: World will deliver “clean energy breakthrough” within 15 years. Guardian.
  • Nagy, B., Farmer, J. D., Bui, Q. M., & Transcik, J. E. (2013). Statistical basis for predicting technological progress. PLoS ONE, 8(8), e52669.
  • Nelson, R. R. (Ed.). (1993). National innovation systems: A comparative analysis. New York: Oxford University Press.
  • Nelson, R. R., & Winter, S. G. (1977). In search of a useful theory of innovation. Research Policy, 6, 36–76.
  • Neumann, J., & Román, D. (2015, November 25). Spain’s Abengoa files for creditor protection. Wall Street Journal.
  • Office of Technology Assessment. (1988, June). Commercializing high-temperature superconductivity. Washington, DC: OTA.
  • Organisation for Economic Co-operation and Development (OECD). (2002). Dynamising national innovation systems. Paris: OECD.
  • Organisation for Economic Co-operation and Development (OECD). (2015). Frascati manual 2015: Guidelines for collecting and reporting data on research and experimental development, the measurement of scientific, technological and innovation activities. Paris: OECD.
  • Pavitt, K. (1984). Sectoral patterns of technical change: Towards a taxonomy and a theory. Research Policy, 13, 343–373.
  • Perepezko, J. H. (2009). The hotter the engine, the better. Science, 326, 1068–1069.
  • Perry, M. L., & Fuller, T. F. (2002). A historical perspective of fuel cell technology in the 20th century. Journal of the Electrochemical Society, 149(7), S59–S67.
  • Rogers, E. M. (2003). Diffusion of innovations (5th ed.). New York: Simon & Schuster.
  • Rosenberg, N. (1963). Technological change in the machine tool industry, 1840–1910. Journal of Economic History, 23(4), 414–443.
  • Rosenberg, N. (1982). Learning by using. In N. Rosenberg (Ed.), Inside the black box: Technology in economics (pp. 120–140). New York: Cambridge University Press.
  • Schlaifer, R., & Heron, S. D. (1950). Development of aircraft engines [and] Development of aviation fuels: Two studies of relations between government and business. Boston: Harvard University Graduate School of Business Administration.
  • Schrader, S. (1991). Informal technology transfer between firms: Cooperation through information trading. Research Policy, 20, 153–170.
  • Schumpeter, J. A. (1934). The theory of economic development. Translation of 1911 German edition. Cambridge, MA: Harvard University Press.
  • Scranton, P. (2011). Mastering failure: Technological and organisational challenges in British and American military jet propulsion, 1943–57. Business History, 53(4), 479–504.
  • Shah, A., Torres, P., Tscharner, R., Wyrsch, N., & Keppner, H. (1999). Photovoltaic technology: The case for thin-film solar cells. Science, 285, 692–698.
  • Singh, K. (2016). Business innovation and diffusion of off-grid solar technologies in India. Energy for Sustainable Development, 30, 1–16.
  • Sterlacchini, A. (2012). Energy R&D in private and state-owned utilities: An analysis of the major world electric companies. Energy Policy, 41, 494–506.
  • Vahdati, M., Smith, N., & Zhao, F. (2015). Influence of intake on fan blade flutter. Journal of Turbomachinery, 137(8), 081002-1–081002-10.
  • Vasile, Z. (2015, April 7). Why future superpowers will run on super batteries. Christian Science Monitor.
  • von Hippel, E. (1988). The sources of innovation. New York: Oxford University Press.
  • Williams, R. H., & Larson, E. D. (1988). Aeroderivative turbines for stationary power. Annual Review of Energy, 13, 429–489.
  • World Bank and Development Research Center of the State Council, the People’s Republic of China. (2013). China 2030: Building a modern, harmonious, and creative society. Washington, DC: World Bank.


  • 1. National systems of innovation have usually been taken to comprise the complex of institutions and institutional arrangements that exert some sort of influence, as for example through provision of education and training, intellectual property protection, and research. See, for example, Nelson (1993); Freeman (1995); OECD, (2002); and Hekkert, Suurs, Negro, Kuhlmann, and Smits (2007). For defense R&D and procurement, see, for example, Alic (2007).

  • 2. Nelson and Winter (1977, p. 47). Kelly and Kranzberg (1978), make much the same point, p. iii: “state-of-the-art knowledge [of the innovation process] is fragmented along disciplinary lines. As a direct result of this fragmentation, research results have become scattered throughout a very large number of professional journals.” For later reviews, see, for example, Rogers (2003); Fagerberg and Mowery (2006); and Hall and Rosenberg (2010).

  • 3. Figures here and following in 2014 US dollars from the IEA database. For a summary, see International Energy Agency, 2015c. The IEA membership includes most but not all OECD countries, and the surveys, while they report much of the world’s publicly financed energy RD&D, do not cover spending by China, India, Russia, and Brazil, among others.

  • 4. For an explanation of differences from the standard OECD (Frascati) categories, see OECD (2015, pp. 336–337).