The diffusion deficit in scientific and technological power: re-assessing China’s rise

Disentangling diffusion

Central to assessments of national S&T capabilities is a basic distinction between innovation and diffusion.Footnote³ The former is defined as the first introduction of a new product or process, whereas the latter is defined as the spread of an innovation through a system or population (Schumpeter Citation1934, pp. 223–233). Thus, a state’s ability to introduce novel S&T advances (‘innovation capacity’) is different from its ability to spread new innovations throughout its domestic ecosystem (‘diffusion capacity’).

Of course, it is difficult to completely disentangle innovation and diffusion. To derive power from scientific and technological advances, states must both introduce and spread these advances. The two processes can overlap and interact, as evidenced by the fact that additional innovations often occur in the process of diffusion (Taylor, Citation2016, p. 231). Oftentimes, a state’s innovation capacity strongly correlates with its diffusion capacity.Footnote⁴ The state that introduces a new method can benefit from first-mover advantages, thereby also leading in the widespread adoption of that method. When the source of the technological advance is international, absorbing such innovations require tacit knowledge that is difficult to extract from its original context (Fadly & Fontes, Citation2019; Keller, Citation2004). China’s ‘indigenous innovation’ policies, for example, have boosted the capabilities of Chinese technology giants in key global production networks, which has aided the commercialization and adoption of emerging technologies by small and medium-sized firms (Malkin, Citation2020).

In theory and practice, however, a country’s innovation capacity and diffusion capacity can widely diverge. A country’s adoption rate of new technologies depends not just on its innovation capacity but also on many other factors, such as institutions for technology transfer, human capital, and openness to trade (Comin & Hobijn, Citation2010). Sometimes, the ‘advantages of backwardness’ allow latecomers to diffuse new technologies faster than the states that pioneer such advances (Gerschenkron, Citation1962). Cross-national studies on diffusion rates over multiple technologies have linked faster rates of intra-country diffusion with a later date of first adoption (Perkins & Neumayer, Citation2005; Pulkki-Brännström, Citation2009; Ray et al., Citation1969). To be clear, latecomer advantages do not determine a country’s diffusion capacity. Without sufficient supporting infrastructure and human capital, latecomer countries cannot absorb key technologies. Nevertheless, I highlight the ‘advantages of backwardness’ thesis because it captures the tension between innovation capacity and diffusion capacity.

Substantial differences between a country’s diffusion capacity and innovation capacity are not limited to developing countries. For example, while they both boast high living standards and highly developed economies, Sweden and Denmark have very different S&T systems, with the former investing heavily in frontier R&D and the latter concentrating on integrating new technologies into its production system (Edquist and Lundvall Citation1993). To compete in a world of globalized science and technology flows, the most advanced economies must also absorb and exploit innovations more effectively than the country in which it was first incubated. Per one estimate derived from data on the OECD countries over 135 years, 93 percent of total factor productivity increases in these high-income countries comes from knowledge that originated abroad (Madsen, Citation2007).

Additionally, leaders in the innovation of new technologies can also be laggards in the overall extent to which such advances penetrate a population of users. While the ‘advantages of backwardness’ argument primarily focuses on the speed of adoption, or the time from first use to saturation, innovation latecomers can also achieve a higher intensity of adoption (Comin & Mestieri, Citation2014, pp. 570–571). Take, for example, continuous casting, which is widely regarded as one of the most important process innovations in the history of the steel industry (Perkins & Neumayer, Citation2005, p. 795). The Soviet Union was one of the first countries to introduce continuous casting in the mid-1950s, but the diffusion of continuous casting throughout the country was very slow. In 1980 continuous casting produced only 10.7 percent of steel in the Soviet Union. This diffusion rate compared unfavorably with those of many countries that lagged in the initial innovation of continuous casting, such as Japan, which produced 59 percent of its steel with continuous casting in 1980, despite introducing the technology five years later than the Soviet Union (Poznanski, Citation1983, pp. 310–311).

Differential effects of diffusion and innovation capacity on growth outcomes

Careful work on the relationship between differential adoption of technology and growth trajectories further supports the need for independent analysis of diffusion capacity in assessments of national S&T capabilities. Using data on the diffusion of 15 technologies over 166 countries to study how technology adoption patterns affect cross-country income gaps, Comin and Mestieri distinguish between two different ‘margins of adoption’ (Comin & Mestieri, Citation2018). The initial adoption lag, closely linked to differences in innovation capacity, refers to the time between a pioneer’s introduction of a new technology and the first implementation of this advance in other countries. The intensive margin of adoption, more tied to diffusion capacity, captures cross-national differences in the technology’s intensity of use.

Measuring both margins of adoption is essential to comprehending how new technologies differentially advantage certain nations. In their analysis of the impact of these two margins on cross-country growth dynamics, Comin and Mestieri find that they contributed equally to the income differentials between Western and non-Western countries during the 19th century (Comin & Mestieri, Citation2018, pp. 168–169). Again, these two margins do not always converge. In some cases, leaders in the initial adoption of a technology can fall behind on the intensive margin of adoption. As demonstrated by the earlier example of continuous casting, compared to the Soviet Union, Japan faced an initial adoption lag but eventually led in the intensive margin of adoption because of its superior diffusion capacity (Figure 1).

Figure 1. Diffusion curves for continuous casting.

Display full size

Related work has shown that indicators of diffusion capacity can be more predictive of long-term growth outcomes than indicators of innovation capacity, which can be unreliable due to the long and uncertain lag between the initial introduction of a new technological advance and its impact on productivity growth. One analysis of technological shocks in the U.S. economy in the postwar period finds that two standard proxies of innovative activity, R&D intensity and patenting rates, have limited ability to predict fluctuations in total factor productivity. By contrast, activities related to broadly disseminating information about new products and processes track better with subsequent changes in productivity (Alexopoulos, Citation2011). Moreover, a study on the effect of human capital investments on long-term growth trajectories also favors measures of diffusion capacity. Leveraging U.S. county-level data, Maloney and Caicedo (Citation2017) trace enduring income differences back to two different types of human capital in 1880: one that captures more inventive activities, proxied by patenting density, and one capturing more adoptive activities, proxied by the density of engineers in a county. While both types had a positive effect on long-term growth trajectories, the effect of engineering density was stronger than the effect of patenting density.

The salience of diffusion capacity is only enhanced for rising powers close to the technological frontier, which boast leading firms that can quickly copy or license innovations. For advanced economies with a certain degree of absorptive capacity and access to these global networks, divergences in long-term economic growth are shaped more by imitation than innovation (Alnuaimi et al., Citation2012; Gries et al., Citation2017, pp. 320–321).

Indicators of innovation capacity and diffusion capacity

Incorporating both diffusion and innovation capacity into assessments of national S&T capabilities is only meaningful if reliable metrics for both dimensions exist. To further unpack differences between measures of innovation capacity and measures of diffusion capacity, I draw on the process-outcomes model (Geisler, Citation2000). By separating the flow of S&T from initial inputs to ultimate absorption in the economy into identifiable stages, each with distinct measurable activities, the model clarifies which metrics attest to innovation capacity and which attest to diffusion capacity (Figure 2). Under indicators of innovation capacity, I include measures of R&D inputs (e.g. funding, personnel), direct outputs of R&D (e.g. patents, publications), and the effectiveness by which R&D inputs translate into direct outputs (e.g. R&D efficiency). Under indicators of diffusion capacity, I include measures of intermediate outputs, such as the usage rates of new methods, and measures of the strength of the linkages between direct outputs and intermediate outputs, such as the robustness of industry-research institution collaborations (Geisler, Citation2000, pp. 243–266).

Figure 2. Innovation and diffusion capacity metrics, ordered by process outcomes model.

Display full size

Although innovation inputs and outputs represent only a portion of the S&T development process, they dominate assessments of national S&T capabilities in international relations scholarship. In fact, in one of the few articles dedicated to benchmarking S&T power resources, Robert Paarlberg states that S&T capabilities ‘can be measured in terms of either final scientific output or R&D input’ (Paarlberg, Citation2004, p. 126). Accordingly, Paarlberg’s measures of scientific and technological leadership include scientific papers published, patents registered, prizes won, innovation hubs, and R&D spending (Paarlberg, Citation2004, p. 126–133). Similarly, other work that benchmarks international S&T performance privileges innovation capacity indicators, such as patenting rates (Taylor, Citation2012, p. 119; Taylor, Citation2016, pp. 37–38) and Nobel Prizes (Gilpin, Citation1968, p. 29).

Among indicators of innovation capacity, R&D spending has become a magic number of sorts. Paarlberg writes that ‘perhaps the best way to measure the U.S. lead in science and technology’ is with R&D inputs (Paarlberg, Citation2004, p. 129). The U.S. National Science Board (NSB), one of the pioneers of national S&T indicators, also prioritizes data collection and analysis of R&D inputs and direct outputs. The NSB’s 1993 Science & Engineering Indicators claims, ‘The long term importance of R&D expenditures to technological preeminence, military security, and knowledge growth is axiomatic’ (National Science Board, Citation1993, p. 89).

This bias toward indicators of innovation capacity is reflected in power formulas used to calculate the power of states. Combing through an encyclopedia of all power formulas published from 1714–2010 (Höhn, Citation2011), I found 25 power formulas that incorporate scientific or technological capabilities as a separate element of national power.Footnote⁵ I could identify indicators of diffusion or innovation capacity for 14 of these formulas. This remaining sample strongly favored indicators of innovation capacity. Nine of the power formulas employ a mix of both innovation capacity indicators and diffusion capacity indicators, with six favoring the former and none favoring the latter.

The other five formulas exclusively relied on innovation capacity indicators. For instance, the Indian National Security Index, developed by India’s National Security Council Secretariat in 2002 to assess comprehensive national power, calculates S&T power using three innovation capacity indicators: patents granted to residents per mission, R&D expenditures as a percentage of GNP, and scientists and engineers as a proportion of R&D spending (Hwang, Citation2010). By contrast, none of the power formulas solely relied on diffusion capacity indicators.

S&T development is a complicated, messy process, so neatly sorting S&T metrics by associations with innovation capacity and diffusion capacity is a tough endeavor. Fortunately, the process-outcomes model functions as a helpful guide for identifying innovation capacity metrics and diffusion capacity metrics. These metrics reveal that the existing literature on power assessment is biased toward innovation capacity. A clear basis for differentiating between these two types of metrics will serve useful for the historical case studies of S&T assessments of rising powers, as well as the application to present-day assessments of China’s S&T capabilities.

The U.S. in the second industrial revolution (1860–1890)

The rise of the U.S. as an economic power took place amidst an era of remarkable technical breakthroughs, sometimes dubbed the ‘Second Industrial Revolution’ (Mokyr, Citation1998). If judging technological leadership in this period based on a country’s ability to generate novel S&T advances, few would have picked the U.S. as the leading candidate. The U.S. trailed Germany and other advanced economies on this front, leading many prominent American scientists to underrate the U.S.’s prospects for S&T leadership. Assessments based on diffusion capacity, however, tell a different story, as Americans excelled at the widespread adoption of innovations first introduced by other countries.

In the decades before World War I, the staying power of the U.S. S&T base confirmed the diffusion-centric view. Across various measures of industrial output and efficiency, the U.S. emerged as the preeminent economic power. Most notably, the U.S. surpassed Britain in per capita GDP and labor productivity around 1900 (Bolt & Luiten van Zanden, Citation2020; Broadberry, Citation2006, pp. 108–109). Did a failure to account for diffusion capacity contribute to flawed assessments of U.S. S&T capabilities? If theoretical expectations of a diffusion surplus hold, the case study evidence should reveal that innovation capacity indicators depicted the U.S. as a relatively weak S&T power. By contrast, the historical data on U.S. diffusion capacity should point toward the strength of U.S. S&T power in this period.

Indeed, the U.S. was far from an innovation leader during the second half of the 19th century. Scholarly and media reports criticized the U.S.’s failing in basic research during this period (Shryock, Citation1966, p. 73; The New York Times, Citation1860; Taylor, Citation2016, p. 9). American researchers lacked the financial support to compete with their European counterparts. By one estimate, in 1891 the entire U.S. research ecosystem supported only twenty-six ‘adequately endowed post-graduate fellowships in science’ (Cohen, Citation1976, p. 383).

European centers of excellence were advancing the technological frontier.Footnote⁷ Carl Snyder, in an influential article for the North American Review published in 1902, attributed European leadership in the great discoveries of the 19th century to their elite research institutions. The U.S. lacked institutions comparable to the German university system, the Royal Institution of London, or the College de France (Snyder, Citation1902, pp. 67–72; Dupree, Citation1964, p. 300). Nor were American cities competitive with London, Paris, and Berlin as innovative research clusters (Shryock, Citation1966, pp. 72–73). It is no surprise, then, that the best and brightest American researchers furthered their training at European institutions. This brain drain was an ‘almost wholly one-way direction of movement of graduate and postdoctoral students’ (Cohen, Citation1976, p. 359).

International comparisons of developments in chemicals, a key emerging technology in this period, provide more detailed indicators of the U.S.’s comparatively weak innovation capacity. From the inception of the Nobel Prize in 1901 to 1930, German and Britain researchers won almost three-fourths of the Nobel Prizes awards in chemistry, whereas just one American attained the top honor in that span (Thackray et al., Citation1985, p. 161). In 1899, German publications accounted for half of all citations in American chemical journals, essentially double the share attributed to American publications (Thackray et al., Citation1985, pp. 405–407). At the same time, American articles barely registered in the chemical journals of Europe, where the best research was published. According to one analysis of references in Annual Reports on the Progress of Chemistry, an authoritative British review journal, American publications accounted for only 7 percent of the citations in 1904 (Thackray et al., Citation1985, p. 157, 402; see also Macleod, Citation1971, p. 207).

In some cases, prominent critics of American science exaggerated U.S. deficiencies to press their case for additional R&D support (Kevles et al., Citation1980). The U.S. performed well on some indicators of innovation capacity, including patent counts and government R&D expenditures (The New York Times, Citation1860; Kevles et al., Citation1980, p. 32). Overall, though, the majority opinion was that the U.S. lacked the means to produce original breakthroughs. Across rankings of nations in terms of ‘productive scholarship’ during this period, America’s ‘most favorable classification’ was fourth place, after Germany, Great Britain, and France (Slichter, Citation1902, p. 12).

Judgements based on diffusion capacity, by contrast, paint the U.S.’s S&T capabilities in better light. Even though the U.S. lagged behind industrial rivals in scientists and researchers, the U.S. cultivated a broad-based system that trained more engineers and cultivated strong linkages that transferred knowledge between academic and industrial research institutions. Compared to arrangements at European peers, research and teaching in American universities were more closely tied to commercial opportunities, which ‘aided the diffusion and utilization of advanced scientific and engineering knowledge’ (Mowery & Rosenberg, Citation1993, p. 36). As a result, in many emerging technologies, the U.S. lagged in generating new breakthroughs but led in adapting them across many useful domains.Footnote⁸

Revisiting international comparisons in chemicals, despite lagging behind Germany in terms of chemical innovations and top chemists, the U.S. was the first to institutionalize the discipline of chemical engineering. This was necessary for the widespread diffusion of chemical processing, which came to transform a wide range of industries such as ceramics, food-processing, glass, metallurgy, petroleum refining, etc. A crucial step in this process was the emergence of unit operations, which broke down chemical processes into a sequence of basic operations (e.g. condensing, crystallizing, electrolyzing, etc.) that were common in chemical processing across a number of industries. Enabled by unit operations, the development of chemical engineering broke down the siloed divisions of industrial chemistry, which had been primarily oriented around the production of a very large variety of chemical products with little concern for the unifying principles between the manufacture of different products (Little, Citation1933, p. 7; Rosenberg, Citation1998, p. 176).

Though Germany dominated industrial chemistry, the U.S. led in cultivating a chemical engineering discipline that facilitated the gradual chemicalization of many industries. By 1913, the U.S. led the world in the production of chemicals (Murmann Citation2003), an indicator of the U.S. advantage in the intensive margin of adoption. American institutions of higher education, most notably MIT, were early adopters of the unit operations model and helped cultivate a common language and professional community of chemical engineering (Guédon, Citation1980, pp. 45–76; Rosenberg, Citation1998, p. 171). Rosenberg and Steinmueller conclude, ‘American leadership in introducing a new engineering discipline into the university curriculum, even at a time when the country was far from the frontier of scientific research, was nowhere more conspicuous than in the discipline of chemical engineering early in the 20th century’ (Rosenberg & Steinmueller, Citation2013, p. 1145).

Fittingly, one of the most colorful denunciations of America’s innovation capacity simultaneously underscored its strong diffusion capacity. In an 1883 address to the American Association for the Advancement of Science (AAAS), Henry A. Rowland, the vice president of the AAAS, denigrated the state of American science for its skew toward the commercialization of new S&T advances. Rowland expressed his disgust for media representations that upheld the ‘obscure American who steals the ideas of some great mind of the past, and enriches himself by the application of the same to domestic uses’, over ‘the great originator of the idea, who might have worked out hundreds of such applications, had his mind possessed the necessary element of vulgarity’ (Rowland, Citation1883, p. 242; Taylor, Citation2016, p. 9). Yet, it was America’s diffusion capacity – in all its obscurity and vulgarity – that sustained its growth to economic preeminence.

The Soviet Union in the early postwar period (1950–1970)

In the early postwar period, the Soviet Union shocked the world by launching the first satellite. Sputnik was just one of the many prominent Soviet scientific and technological achievements in the early decades of the Cold War. During this time, ‘the USSR appeared to all as a rising technological juggernaut (Taylor, Citation2016, p. 75)’.Footnote⁹ Ultimately, the Soviet Union’s total factor productivity growth slowed significantly in the 1970s; some studies, in fact, estimated a decline in factor productivity over the decade (Trachtenberg, Citation2018, pp. 86–87). The Soviet Union’s collapse in 1991 laid bare the inefficiencies of its productive engine.

Did neglect of the diffusion process contribute to misleading views of the Soviet Union’s S&T power? In cases of diffusion deficit, when a rising power’s innovation capacity substantially outpaces its diffusion capacity, innovation-centric assessments will overestimate its ability to sustain its rise. If theoretical expectations hold, evidence from this case should show that innovation-centric measures corresponded to a view of the Soviet Union as a S&T superpower. Additionally, diffusion-centric measures should expose the fundamental weaknesses of the Soviet Union’s S&T capabilities in this period.

An innovation-centered lens shaped views of the Soviet Union as a S&T superpower. The Soviet Union became a world leader in two key indicators of innovation capacity, R&D spending and the employment of scientists and engineers (Beckley, Citation2018, p. 35). A 1962 OECD report, one of the first efforts to systematically compare national S&T capabilities, found that the Soviet Union was near parity to the U.S. in these two indicators (Freeman & Young, Citation1965). By 1970 the Soviet Union led the world in R&D spending as a percent of gross national product (3.28 percent), exceeding the comparable U.S. figure of 2.57 percent (National Science Board, Citation1987, p. 236).

The notion of a ‘scientific manpower gap’ – specifically, that the Soviet Union was graduating two to three times as many scientists and engineers than the U.S. – took hold in U.S. discourse (Teitelbaum, Citation2014, pp. 32–36). Throughout the 1950s, this figure was ‘repeated ad infinitum’ by analysts and politicians, including the Central Intelligence Agency (CIA) director, the Atomic Energy Commission chair, the Assistant Secretary of Defense, and key members of Congress (Kaiser, Citation2006, pp. 1231–1234). In a 1955 speech, M.H. Trytten, a key director at the National Research Council highlighted the ‘startling fact’ that the Soviet Union was producing around twice as many science PhD graduates as the U.S. (Krige, Citation2000, p. 87).Footnote¹⁰

Detailed studies and compilations of national S&T indicators, though careful to note the differences between the two countries in counting science and engineering graduates, confirmed the Soviet’s strength in innovation capacity throughout the period. Footnote¹¹ Per 1985 figures, reported in the NSF’s Science and Engineering Indicators, the Soviet Union graduated the most ‘first university degrees’ in natural sciences and engineering, which was more than double the U.S. figure.Footnote¹²

Assessments of Soviet diffusion capacity presented a different picture. One particularly prescient CIA report, published in 1969, concluded that ‘the technological gap between the Soviet Union and the developed West is large and is probably widening’ (Central Intelligence Agency, Citation1969, p. 1). The crucial difference, to which the report devoted a full section, was the U.S.’s superior mechanisms to spread technologies, described as a ‘fast-acting, seemingly almost biological process’, compared to the Soviet Union’s mechanisms, which were ‘much more balky’ (Central Intelligence Agency, Citation1969, p. 24, 2). Limitations to the Soviet economy’s diffusion capacity included lack of competitive pressures, disincentives for firm-level technology planners to adopt new technologies out of fear that they would be forced to meet higher production targets as a result, and separation between the R&D system and the broader economy (Central Intelligence Agency, Citation1969; Cocka, Citation1980, pp. 220–240; Kontorovich, Citation1990, p. 255).

While the Soviet Union was at the forefront of introducing new technologies in many areas, it lagged in adopting technical improvements across a broad range of production processes. A 1977 assessment of Soviet progress in nine technology areas concluded that the Soviets were most successful in the initial research and pioneering of innovations and least successful in the diffusion of innovations across the economy (Amann et al., Citation1977).Footnote¹³ The Soviet Union was the innovation leader in extra-high-voltage transmission of electricity, for instance, but the U.S. and Great Britain were the diffusion leaders by 1970 (Amann et al., Citation1977). A similar story played out in machine tools and the oxygen converter process for steel, which only diffused to 12 percent of Soviet steel production (Berliner, Citation1973; Central Intelligence Agency, Citation1969, p. 47). As the CIA’s 1969 assessment concluded, ‘In no major branch of industry is the average level of Soviet technology in use on a par with that in the United States or Western Europe’ (Central Intelligence Agency, Citation1969, p. 5).

Similarly, the Soviet Union’s impressive success in a narrow set of mission-oriented technological feats – most dramatically captured by the launch of Sputnik – did not translate into broader technological trajectories. In domains relatively insulated from the economic system, such as rocket engines and advanced weapons systems, the Soviet Union performed exceptionally by investing high amounts of resources and talented personnel. Soviet scientists and engineers also made impressive contributions to the development of computers and semiconductors, and the Soviet Union successfully exploited military applications in these two domains. However, as these technologies spread to affect the entire economy, like with the advance of computerization in the 1960s and 1970s, the Soviet Union could not keep pace (Graham, Citation2013, p. 78).

Over time, the Soviet leadership recognized that diffusion capacity was the weak link. General Secretary Breznhnev commented in 1971, ‘If one examines all the links of the complex chain uniting science with production, it is not too difficult to see that the links connected with the practical realization of scientific achievements and their adoption in mass production are the weakest’ (Quoted in Cocka, Citation1980, p. 197)^. In the 1970s Soviet policymakers tried to set up institutions tasked with ‘the implementation and diffusion of new technology’ (Cocka, Citation1980, p. 219). Indeed, these reactions perhaps best capture the Soviet Union’s diffusion deficit.

Innovation-centrism in analysis of China’s S&T power

In debates over China’s S&T power, complex dynamics get reduced down to a magic word: innovation.Footnote¹⁴ Drawing on theories that identify innovation in leading sectors as the key mechanism behind historical power transitions, scholars frame the U.S.–China technological competition around which state dominates the creation of new technologies (Kennedy & Lim, Citation2018; Rapkin & Thompson, Citation2003).Footnote¹⁵ For instance, in their analysis of how and when a U.S.–China power transition could occur, Rapkin and Thompson (Citation2003, p. 333) concentrate on ‘China’s capacity to innovate’.

An evaluation of the international relations scholarship on China’s S&T capabilities provides some systematic evidence of a bias toward innovation capacity. I reviewed sixty articles across the following three journals: International Affairs, International Security, and Review of International Political Economy. From each journal, I selected the twenty articles that ranked highest in a Google Scholar search for ‘China’ and ‘technology’. Of the twelve articles that assessed China’s S&T capabilities in relation to a rising power’s ability to sustain its economic rise, over 80 percent of the articles’ assessments favored innovation capacity indicators.Footnote¹⁶

As was the case with overestimates of the Soviet Union’s S&T power in the 1970s, present-day assessments of China’s S&T power also heavily rely on indicators of innovation capacity, often citing China’s impressive performance in R&D expenditures, scientific publications, and patents.Footnote¹⁷ No indicator garners more attention than R&D expenditures. Accordingly, the expectation that China will soon surpass the U.S. in R&D spending has renewed fears of China’s growing S&T power. Citing China’s heavy investments in R&D, reports by influential commissions have warned that the U.S. faces a ‘tipping point in R&D’ (American Academy of Arts & Sciences, Citation2020, p. 11) and an ‘innovation deficit’ (MIT Committee to Evaluate the Innovation Deficit, Citation2015). Under the Global Power Index (GPI), a model developed by the Pardee Center at the University of Denver, a country’s share of global R&D expenditures accounts for half of its technological power.Footnote¹⁸ Frequently used by the National Intelligence Council, which leads the U.S. intelligence community’s strategic forecasting efforts, the GPI predicts that China’s national power will equal that of the U.S. in 2030 (National Intelligence Council, Citation2012, p. 16).

To be clear, some assessments of China’s innovation capacity do conclude that U.S. technological preeminence is durable. Beckley’s comparison of U.S. and Chinese innovative capabilities, published in 2011, finds that over the past twenty years the U.S. had increased its lead over China in many indicators, including a variety of patent and R&D metrics (Beckley, Citation2012, pp. 63–73). Likewise, Brooks and Wohlforth’s argument for sustained U.S. technological leadership primarily employs innovation capacity indicators, including triadic patent families and top-cited articles in science and engineering (Brooks & Wohlforth, Citation2016, p. 24).

Still, when the innovation capacity indicators from these previous studies are updated, nearly all the figures portray a narrowing U.S.–China gap in S&T capabilities, with China emerging as a near-peer competitor. I found the most recent figures for 20 S&T indicators employed in Beckley’s article.Footnote¹⁹ Apart from the number of universities ranked in the top 20 of the Academic Ranking of World Universities, China has substantially closed the gap with the U.S. across 14 of the 15 indicators of innovation capacity. Per one indicator, in 2009, foreign firms accounted for the majority of invention patents granted by China’s patent office. Now, according to updated 2019 figures, domestic Chinese firms account for 80 percent of invention patents granted. These preliminary results suggest that more detailed analysis of China’s diffusion capacity is necessary.

Decomposing diffusion and innovation capacity reveals china’s diffusion deficit

Does China’s diffusion capacity diverge significantly from its innovation capacity? In order to separately assess these two dimensions, I consult two influential and reliable indexes of national S&T capabilities: the Global Innovation Index (GII) and the Global Competitiveness Index (GCI). Among the dozen or so cross-national indexes that evaluate countries’ ability to develop and use new technologies, these two stand out for their objectivity and influence (Kennedy, Citation2017, p. 11). Experts on China’s science and technology ecosystems regularly cite the GII to judge China’s standing relative to the U.S. in global technology leadership (Chen et al., Citation2021, p. 4). Chinese scholars pay particular attention to the GCI. Since the index was established in 2004, they have published 253 articles that include the exact terms ‘comprehensive national power’ [zonghe guoli] and ‘Global Competitiveness Index’ [quanqiu jingzhengli zhishu] (author’s search of China National Knowledge Infrastructure database, conducted on May, 2, 2021; Pillsbury, Citation2000, p. 226, 373).Footnote²⁰

For each of the indexes, I sorted S&T indicators by their association with diffusion capacity and innovation capacity, according to the process-outcomes model introduced earlier. I focus on indicators for which it was relatively straightforward to determine whether they measured innovation capacity or diffusion capacity. For instance, the GII tracks the R&D expenditures of a country’s top three firms and the quality of a country’s top three universities – two obvious indicators of innovation capacity. The index also includes many indicators that clearly evaluate a country’s ability to disseminate new advances, such as the extent of research collaboration between universities and businesses. I excluded S&T indicators that could credibly measure both diffusion capacity and innovation capacity, such as school life expectancy and enrollment in tertiary education.Footnote²¹

This decomposition of the GII and GCI demonstrates that China’s diffusion capacity significantly lags behind its innovation capacity (Table 1). If judged solely by the 2020 GII’s indicators for the latter, China rates as a top-tier S&T power, boasting an average global ranking of 13.8. For reference, the 13th ranked country in the overall GII is Israel, widely recognized as a global leader in S&T. The gap between China and the U.S. on this innovation capacity subindex is very small: the U.S’s average ranking is 11.9 (Figure 3).

Figure 3. China’s diffusion deficit over time.

Display full size

However, if evaluated on GII indicators of diffusion capacity, China’s average ranking drops by 34 spots. The closest comparator based on the overall 2020 GII is the 47th ranked country, the Russian Federation, which few judge to be a prospective S&T superpower. When measured based on this diffusion capacity subindex, the gap between China and the U.S. is very wide: the U.S.’s average ranking is 26.9.Footnote²² The decomposition of the GCI produces similar results. China’s average global ranking in GCI indicators of innovation capacity is around 15; it falls to about 44 in terms of GCI indicators of diffusion capacity (Supplementary Appendix C).

This gap between China’s innovation capacity and diffusion capacity holds when raw scores are used instead of ranks. For some indicators, the shape of distributions could greatly deviate from the linear one imposed by ranks.Footnote²³ As an additional check, I compared China’s raw scores across diffusion and innovation capacity indicators with those of the U.S., France, and Israel – the latter two countries ranked just ahead of China in the overall 2020 GII. Relative to all three countries, China’s average score in the innovation capacity subindex was higher while its average score in the diffusion capacity subindex was lower.

China’s diffusion deficit extends beyond a single-year snapshot. Using the same procedures as the decomposition of the 2020 GII, I calculated the average rankings of the U.S. and China on the innovation and diffusion capacity subindexes for the GII in 2014 and 2017. In all three years, China’s average rank in diffusion-oriented indicators trailed its rank in innovation-oriented indicators by over 30 spots (Figure 3). While China has improved its average rank in the diffusion capacity subindex by about 5 places since 2014, the gap between China and the U.S. on this measure has only widened, as the U.S. improved its own diffusion capacity ranking by almost 10 places over the same period.

Notably, these results run counter to claims that China’s rising S&T prowess comes from its strategic advantage in deploying innovations at scale.Footnote²⁴ For instance, influential reports play up China’s capacity to adopt AI advances because it graduates more computer science students than competitors.Footnote²⁵ These analyses draw from a few examples of Chinese success at large-scale deployment in domains such as high-speed rail and mobile payments. Alongside the analysis that follows, the decomposition of these indicators cautions against overestimating China’s diffusion capacity.

China’s lethargic diffusion capacity is further confirmed by a detailed evaluation of its adoption of information and communications technologies (ICTs), generally considered the key drivers of future productivity growth.Footnote²⁶ While China has achieved a few noteworthy successes in ICT diffusion in consumer-facing applications – such as the spread of mobile payments and e-commerce – Chinese businesses have been slow to embrace digital transformation. China lags behind the U.S. in penetration rates of many digital technologies across industrial applications, including digital factories, industrial robots, smart sensors, key industrial software, and cloud computing (Alibaba Research Institute, Citation2019; Synced, Citation2020; Techxcope, Citation2020).

Furthermore, composite measures reveal a large gap between the U.S. and China in terms of the countries’ readiness to effectively spread and utilize ICT advances. China ranks 83rd in the world, 67 places behind the U.S., on the International Telecommunication Union’s ICT Development Index, which combines the level of networked infrastructure and access to ICTs and the level of use of ICTs in the society (International Telecommunications Union, Citation2017, p. 31).Footnote²⁷ China also significantly trails the U.S. in an influential index for adoption of cloud computing, which is essential to implementing AI applications and many other emerging technologies (Supplementary Appendix D). In 2018, U.S. firms averaged a cloud adoption rate of over 85 percent, more than double the comparable rate for Chinese firms (Wang and Chen Citation2020).

In sum, China faces a diffusion deficit. Therefore, most assessments of China’s S&T capabilities overestimate China’s capacity to convert technological breakthroughs into national productivity improvements because they privilege indicators of innovation capacity. A rebalanced evaluation of China’s potential for S&T leadership requires looking beyond multinational corporations like Huawei, first-tier cities like Beijing, and flashy R&D numbers to the humble undertaking of diffusion. This perspective brings a different cast of characters into the spotlight: smaller firms, fourth-tier cities that rarely feature in English-language coverage, and technology transfer mechanisms.