A Structural Analysis of the Aerospace and Defense Industry Supply Chain Network

This report takes the global supply chain network of the aerospace and defense industries as its subject, empirically clarifies its structural features, theoretically elucidates the mechanisms of its formation, and derives policy implications from the standpoint of economic security and resilience. Departing from conventional superficial dependency analyses and simplistic recommendations to "diversify procurement," this analysis is grounded in a theoretical framework integrating institutional factors, network theory, industrial organization, and innovation policy theory. It addresses why the current structure came into being, whose incentives must be altered and how in order to change that structure, and what the causal mechanisms and unintended consequences of policy intervention are.

1.1 The Nature of the Data and Interpretive Caveats

The network under analysis encompasses three heterogeneous industrial sectors: the aircraft industry, the space industry, and the defense industry. This intermingling means that the mere frequency of mentions does not indicate industrial competitiveness. For example, the fact that satellite-related items account for 824 mentions, aircraft for 809, and UAVs/drones for 462—with space-related items making up 23% of the total—does not mean that the space industry is "stronger" than the aircraft industry. Rather, it reflects a temporal bias whereby the space industry has attracted attention due to recent technological and policy changes (the Commercial Crew Program, NewSpace, the Artemis program). Accordingly, this analysis explicitly distinguishes structural differences across industries and derives policy implications while taking into account the institutional factors specific to each industry.

As a more fundamental caveat, this network is based on relationship extraction from news articles and does not directly observe actual procurement flows or transaction volumes. This characteristic carries three interpretive implications. First, entities with high media salience (those involved in large contracts, new programs, or geopolitical topics) tend to have more connections recorded, so centrality metrics reflect both "actual structural importance" and "visibility in reporting." Second, routine supplier relationships of low news value are systematically undercaptured, and thus the procurement-diversity metric discussed later (Section 2.4) should be read as a lower bound on reality. Third, the linguistic and regional composition of the collected corpus may affect the observed distribution across countries (revisited in Section 3.1). Therefore, the empirical findings of this report are presented as hypotheses to be interpreted to the extent they can be cross-checked against external evidence such as official statistics and individual cases, with the relevant caveats stated wherever applicable.

1.2 The Theoretical Framework of the Analysis

This analysis rests on the following four theoretical pillars.

First, Barabási-Albert scale-free network theory. The observed network exhibits extreme concentration around a small number of hub nodes (NASA: 236 connections, SpaceX: 245 connections, Lockheed Martin: 183 connections). A distribution with a mean degree of 3.52 against a maximum degree of 245 suggests a heavy tail consistent with a power-law distribution (see Appendix A on the need for rigorous statistical testing). As Barabási & Albert (1999) showed, when a network follows a preferential attachment mechanism during its growth process, hub dominance arises naturally through a "rich get richer" effect. This theoretical insight suggests that the current structure is not accidental but a consequence of the intrinsic dynamics of network growth.

Second, the innovation policy typology of Edler & Georghiou (2007, 2012). They classified policy instruments into supply-side policies (R&D subsidies, human resource development), demand-side policies (public procurement, regulation, standardization), and systemic policies (ecosystem building, institutional design), placing particular emphasis on the market-shaping power of demand-side policies. This analysis examines the process by which NASA's Commercial Crew Program ($3.1B) and India's offset policy (a 30% obligation) shaped the very structure of the supply chain network, beyond merely creating demand.

Third, the barriers-to-entry theory of industrial organization. The facts that the Boeing 787 cost $32B to develop and that certification takes several years suggest that the commercial aircraft industry has a naturally oligopolistic structure. Following the classic framework of Bain (1956), in industries where absolute cost advantages and economies of scale operate simultaneously, the market power of incumbents becomes extremely entrenched. This analysis discusses how these entry barriers interact with certification regimes and technical standards to produce lock-in effects.

Fourth, Mazzucato's (2013, 2024) theory of the Entrepreneurial State. Drawing on abundant case evidence, Mazzucato argued that America's technological superiority is a consequence not of "small government" but of strategic investment by NASA, DARPA, the DoD, and others. This analysis shows, in contrast with competing explanations, that SpaceX's quasi-monopoly (87% of U.S. orbital launches) can be interpreted as an unintended consequence of the government's market-shaping intervention in the form of NASA's Commercial Crew Program (Section 3.4). This insight recasts the role of government as "shaping" markets rather than "fixing" them, and highlights the importance of a mission-oriented approach.

2.1 Structural Characteristics as a Scale-Free Network

The observed network exhibits a sparsely connected structure with 4,620 nodes, 8,134 edges (the edge count after deduplicating and aggregating the 14,106 extracted relationships), and a network density of 0.000381. This low density means that direct connections between entities are limited, but the fact that the largest connected component accounts for 80.8% (3,733 nodes) shows that, indirectly, the network is highly connected. This seemingly contradictory feature is a typical property of scale-free networks.

Analysis of the degree distribution reveals extreme asymmetry. Against a mean degree of 3.52, the maximum degree is 245 (SpaceX), and the standard deviation greatly exceeds the mean. The top 20 entities account for the bulk of all connections, and three super-hubs exist: NASA (236 connections), SpaceX (245 connections), and Lockheed Martin (183 connections). In the PageRank analysis, NASA (0.0133), the United States (0.0109), and SpaceX (0.0091) carry overwhelming importance.

As a theoretical interpretation of this distribution, the preferential attachment mechanism suggested by the Barabási-Albert model is likely at work. That is, entities newly entering the network tend to connect preferentially with nodes that already have many connections. This is because nodes with many existing connections are attractive for several reasons: (1) they have a proven track record of reliability, (2) they effectively define technical standards, (3) they have established relationships with regulators, and (4) they achieve low costs through economies of scale. However, hub dominance can also be generated by mechanisms other than preferential attachment (differences in node-specific fitness, or the property of the observation process whereby reporting concentrates on prominent actors). This interpretation is therefore presented as a strong hypothesis consistent with the observed pattern, and verification through direct observation of the network growth process remains a task for future research.

This "Matthew effect" (the rich get richer) describes a process in which an initial random advantage is amplified self-reinforcingly. The technological and institutional advantages that NASA established with the Apollo program in the 1960s were cumulatively reinforced through subsequent space programs (the Space Shuttle, the ISS, Commercial Crew), leading to its current overwhelming centrality of 236 connections. Similarly, the position that Lockheed Martin secured through the F-35 program creates a positive feedback loop in which it holds an advantage in subsequent defense procurement as well.

2.2 Betweenness Centrality Analysis: The Theoretical Meaning of Single Points of Failure

Betweenness centrality analysis measures the degree to which a node mediates the flow of information and resources within a network. The fact that NASA (0.033), SpaceX (0.029), and Lockheed Martin (0.027) show the highest values means that these entities perform a "bridging" function. From the standpoint of network theory, removing a node with high betweenness centrality is likely to fragment the network into multiple disconnected components.

However, the expression "single point of failure (SPOF)" requires deeper theoretical consideration. A scenario in which NASA or Lockheed Martin suddenly vanishes is unrealistic, and the essence of the problem is not physical removal. Rather, what matters is that the institutional functions these hubs hold are irreplaceable.

NASA's role is not merely that of a purchaser or operator; it has multilayered functions as (1) a setter of technical standards, (2) a risk-taker for new technologies, (3) a coordinator across industry, academia, and government, and (4) a presenter of long-term vision. In the Commercial Crew Program, NASA did not merely present technical requirement specifications to SpaceX and Boeing; it also provided phased funding and technical advice throughout the development process and shared the risk of failure. This function as an "Entrepreneurial State" cannot be replicated by private firms alone. As Mazzucato (2013) notes, the success of the U.S. space industry can be understood as a consequence of the government's playing a role of market "creation" rather than market "correction."

Therefore, the essence of the SPOF problem lies in the concentration and non-substitutability of institutional functions. The policy response should not be superficial—simply "cultivating alternative suppliers"—but should be conceived as a problem of institutional design: how to disperse and make substitutable the institutional functions that hub entities carry.

2.3 Geographic Concentration and Institutional Lock-In

Analysis of the distribution by country reveals the overwhelming predominance of the United States. It accounts for 33.2% of all entities and 45.9% of all relationships. With India in second place (12.0%) and the United Kingdom and China tied for third (4.4%), the distribution shows extreme concentration in a single pole.

As a theoretical explanation for this geographic concentration, three causal mechanisms can be considered.

First, the institutional legacy of the Cold War. The establishment of NATO (1949) and the U.S. strategy against the Soviet Union standardized the defense industries of Western allies to U.S. specifications. The Arms Export Control Act (enacted 1976, substantially revised 2013), the precursor to ITAR (International Traffic in Arms Regulations), strictly regulated the export of defense-related technology and dramatically raised the cost of transactions with non-U.S. firms. ITAR is not merely an export regulation; it has a cascade effect across the entire supply chain. If firm A procures a component from firm B and sells it abroad, firm B is also required to be ITAR-compliant. This chain of requirements forms a subnetwork consisting solely of ITAR-compliant firms and effectively excludes non-compliant firms.

Notably, the substantial relaxation of ITAR restrictions carried out in August 2024 within the framework of the AUKUS agreement is a natural experiment that paradoxically demonstrates how institutions have governed network structure; this is detailed in Section 3.1.

Second, entry barriers and lock-in effects created by certification regimes. Aircraft certification by the FAA (Federal Aviation Administration) and EASA (European Union Aviation Safety Agency) requires several years and enormous costs. A substantial portion of the Boeing 787's $32B development cost is said to be certification-related. For new entrants, this fixed cost constitutes an almost insurmountable barrier.

More importantly, certification regimes heighten the importance of complementary assets. Following the framework of Teece (1986), in the aircraft industry the source of competitive advantage lies less in product technology itself than in complementary assets such as (1) certification know-how, (2) relationships with regulators, (3) existing production facilities and supply chains, and (4) after-sales service networks. Boeing and Airbus have accumulated these assets over decades, and it is impossible for new entrants to build them in a short time. The fact that China's COMAC (Commercial Aircraft Corporation of China) spent more than 15 years developing the C919 and has struggled to gain recognition in international markets illustrates the height of this barrier.

Third, the self-reinforcing mechanism of preferential attachment described in Section 2.1. The fact that U.S. firms already have many connections raises the probability that U.S. firms are chosen in new projects as well. From a procurement officer's perspective, choosing a proven U.S. firm offers several advantages: (1) lower risk (its track record is established), (2) high consistency with the existing supply chain, and (3) ease of political justification. As this selection bias accumulates, U.S. centrality is reinforced over time.

2.4 The Absence of Redundancy: The Meaning of an Average of 1.55 Source Countries per Procurement

The most concerning feature of the observed network is the extreme lowness of supply chain resilience. For nodes with in-degree, the average number of suppliers is 2.81, and the average number of procurement source countries is only 1.55. This means that many entities procure from a single country or from only two countries.

However, given the nature of the data described in Section 1.1, these figures must be interpreted with caution. Because news reporting selectively picks up new contracts and relationships that have become problematic, routine procurement relationships that go unreported are systematically omitted, potentially underestimating actual procurement diversity. In other words, the value of 1.55 should be read not as the result of a precise field survey but as a lower-bound signal indicating that "even within the range made visible by the media, dispersion of procurement sources cannot be confirmed." Even so, the real-world examples concerning Russian titanium and rare earths discussed later corroborate that procurement concentration does in fact exist for at least some critical materials and components, so the concern about low redundancy is itself qualitatively supported.

As a theoretical explanation for this low redundancy, a trade-off between efficiency and resilience can be considered. Within the framework of transaction cost theory (Williamson, 1985), increasing the number of suppliers entails costs such as (1) the dispersion of relationship-specific investment, (2) increased quality-control costs, and (3) the risk of leaking technical information. In the aerospace and defense industries, component specifications are extremely stringent, and building long-term relationships with suppliers is essential. For this reason, firms prefer close relationships with a small number of suppliers, and redundancy is sacrificed as a result.

However, this efficiency-focused strategy does not account for geopolitical risk. The figure of 1.55 source countries per procurement implies a risk that a supply disruption from a particular country (sanctions, conflict, export restrictions) will propagate in a chain reaction. After Russia's invasion of Ukraine in 2022, dependence on Russian titanium came to light, and Airbus and Boeing struggled to secure alternative sources. China's dominance in rare earth elements (60% of world production) is likewise recognized as a serious vulnerability amid rising U.S.-China tensions.

What is important is that this low redundancy is a consequence of firms' rational choices. From the standpoint of an individual firm, prioritizing efficiency and minimizing redundancy leads to profit maximization. From the standpoint of the system as a whole, however, this strategy produces collective vulnerability. This is a classic externality problem, and it cannot be solved by the market mechanism alone. Accordingly, the justification for policy intervention rests on the classic argument of correcting market failure.

3.1 The Historical and Institutional Origins of U.S. Centrality

The overwhelming centrality of the United States bears strongly the character of a legacy of the institutional architecture built during the Cold War. The DoD (Department of Defense) and the CIA, established by the National Security Act of 1947; NASA, established in 1958; and DARPA (Defense Advanced Research Projects Agency), established in 1958, became the three pillars of U.S. science and technology policy. These organizations were not mere procurement agencies but embodiments of an "Entrepreneurial State" that set the direction of technological development, assumed risk, and promoted collaboration among industry, academia, and government.

NASA's Apollo program (1961–1972, total budget $25.4B, or $280B in 2020 dollars) not only built the foundation of the U.S. space industry but also established institutional standards for contract management, project management, and quality assurance. These standards were refined through the subsequent Space Shuttle (1981–2011) and ISS (1998–) programs, and can be assessed as having come to occupy the position of a de facto international standard. Europe's Ariane program and Japan's H-II rocket are all, to a greater or lesser extent, under the influence of NASA standards.

DARPA's role is even more fundamental. DARPA supported the early development of many innovative technologies, including the Internet (ARPANET, 1969), GPS (early experiments in 1973), stealth technology, and the microprocessor. As Mazzucato (2013) emphasizes, these technologies could not have been developed by private firms alone. This is because financing was difficult in private risk-capital markets for three reasons: (1) technological uncertainty was extremely high, (2) the payback period extended over the long term, and (3) the appropriability of the results was low (spillovers were large). DARPA established a model of bridging these "valleys of death" with government funds and transferring the technology to the private sector once it had matured.

This "socialize the risk, privatize the reward" model is debatable in economic terms (Lazonick & Mazzucato, 2013), but its effectiveness in technological development has been empirically demonstrated. Much of the technology SpaceX uses in the Falcon 9 rocket owes a great deal to the accumulation of past investments by NASA and DARPA. Elon Musk's excellence lies in integrating these existing technologies and innovating the business model, but a substantial portion of the underlying technology originates in government investment.

The NATO system also institutionally entrenched U.S. centrality. NATO Standardization Agreements (STANAGs) aim to standardize the equipment and procedures of member states and ensure interoperability. In practice, however, because the United States possesses overwhelming military and technological power, NATO standards became de facto U.S. standards. The fact that U.S.-made fighters such as the F-16 and F-35 have been adopted by many NATO member states is a consequence of this standardization.

Furthermore, ITAR regulation legally enforced U.S. centrality. ITAR strictly regulates the export of U.S.-origin defense-related technology and products. What is important is that ITAR regulation extends not only to end products but also to components, technical data, and manufacturing processes. For example, if a satellite contains even a single U.S.-made component, the entire satellite becomes subject to ITAR regulation, and its export to a third country requires U.S. government approval.

The cascade effect of this regulation ripples across the entire supply chain. European satellite manufacturers report that customers first ask them whether they are "ITAR-free" (2024 survey). ITAR-free products command a high market value not only because export procedures are simplified but also because they mean freedom from the discretionary authority of the U.S. government. However, given U.S. technological superiority, complete ITAR-free status is difficult, and as a result many firms choose to be incorporated into the U.S. supply chain.

The AUKUS reform of August 2024 has the potential to open a breach in this structure. The fact that the U.S. State Department authorized license-free transactions with Australia and the United Kingdom for more than 70% of ITAR-regulated items reflects a shift in policy judgment: that with trusted allies, the benefits of strengthening the industrial base outweigh the risk of technology leakage. Deregulation should lower transaction costs with Australian and British firms and create incentives for the formation of new networks, and this natural experiment offers a valuable opportunity to observe the causal mechanism by which institutions shape network structure. Whether the reform will actually manifest as a network reconfiguration, however, requires verification with data over the coming years. Japan's consideration of participating in AUKUS is also set against the background of this institutional change.

That said, it must be noted that there are strong rival hypotheses to this institution-centered explanation. The first is the scale hypothesis. U.S. defense spending accounts for roughly 40% of the world total, and network centrality could arise mechanically from an overwhelming disparity in market size, even without assuming institutional lock-in. This hypothesis is compelling at first glance. If the procurement market is large, the number of contracts and the volume of reporting will both increase, so a strong correlation between spending scale and observed connection counts is to be expected without invoking institutions. The second is the observation bias hypothesis. As described in Section 1.1, an English-language, media-centric corpus may structurally over-represent U.S. entities. There are three reasons why, even taking these hypotheses seriously, we make institutional explanation the central axis. First, since scale itself is not an exogenous given but an endogenous product of institutional choices—the long-term investments of NASA, DARPA, and the DoD—the scale hypothesis does not compete with the institutional explanation but is subsumed within it. However, this line of argument alone cannot eliminate scale-determining factors other than institutions, such as geopolitical role and economic size, so it is not decisive. Thus, second, it is important that there are observations within our own data that are inconsistent with the scale hypothesis. China, said to rank second to the United States in the scale of defense spending, has an entity share of only 4.4%, far below that of India (12.0%), whose spending scale is below China's. If spending scale were mechanically converted into centrality, this reversal of ranking could not be explained (that said, part of this gap may also be explained by the observation bias whereby China-related reporting is harder to capture in an English-language corpus, so it remains suggestive corroborating evidence). Third, while scale and observation bias may explain the level of "why there are so many U.S.-related connections," they cannot explain the phenomenon whereby a market premium arises for ITAR-free products, or the phenomenon whereby deregulation such as the AUKUS reform immediately changes transaction incentives—that is, the quality of connections and the mechanisms of their formation. Therefore, this report's position is as follows: while acknowledging that part of the observed level of U.S. centrality may be attributable to the nature of the corpus and to market size, the institutional explanation is the most coherent account of the mechanism by which that structure is reproduced.

3.2 The Policy Mechanism of India's Rise: Offset Policy and Make in India

One of the most interesting findings of this analysis is that India holds the second-largest number of entities (12.0%, 556 firms) and that bidirectional relationships with the United States (U.S.→India 154, India→U.S. 126) are pronounced. This report interprets at least part of this rise as attributable to the Indian government's strategic industrial policy. As discussed below, however, this interpretation faces competing explanations, and the argument is refined in contrast with them.

India's defense offset policy requires that, when a procurement amount exceeds 300 crore INR (about $36M), 30% of the contract value be offset through domestic production, technology transfer, or investment. This policy is not merely a demand for localization but an ingenious institutional design that compels foreign firms to build supply chains within India.

For example, when the Indian Air Force procured Rafale fighters for $8.7B (2016 contract), Dassault Aviation assumed offset obligations worth $2.6B. Dassault fulfilled these offsets by (1) establishing a joint venture with the Reliance Group, (2) relocating component manufacturing to India, and (3) granting technology licenses to Indian firms. In this process, Indian firms such as Reliance Defense, Hindustan Aeronautics Limited (HAL), and Bharat Electronics Limited (BEL) were integrated into Dassault's supply chain.

The economic effect of offset policy can be understood from the standpoint of strategic trade policy theory (a body of research pioneered by Krugman, 1984). In imperfectly competitive markets, government support for a particular industry makes it possible to appropriate "excess profits" for one's own country. The aerospace and defense industries are well suited as targets of strategic trade policy for three reasons: (1) an oligopolistic structure, (2) economies of scale, and (3) learning effects.

Offset policy can also be described as a modern version of infant-industry protection. Classic infant-industry protection arguments (List, 1841; Hamilton, 1791) advocated protecting domestic industry through tariffs and having it acquire international competitiveness through learning effects. Under the WTO regime, however, high tariffs are difficult. Offset policy is an ingenious institution that achieves de facto protection of domestic industry not through trade restrictions but through procurement conditions.

The Atmanirbhar Bharat (Self-Reliant India) policy, launched in 2020, deepened offset policy further. This policy consists of four pillars: (1) relaxation of FDI regulation (from 49% to 74% in the defense industry), (2) the Production Linked Incentive (PLI) scheme, (3) the Defense Testing Infrastructure Scheme, and (4) the Make in India Initiative. The PLI scheme in particular provides firms with incentives (4–6% of production value) according to domestic production value, supporting the achievement of economies of scale.

Changes consistent with these policies are already reflected in the numbers. India's defense industry market is projected to grow from $27.1B in 2024 to $54.4B in 2033 (CAGR 6.99%) (though it should be noted that a projected value of market growth is not itself causal evidence of policy effects). The number of Tier 1 suppliers and SMEs has surged, and Western majors such as Lockheed Martin, Boeing, Thales, and Dassault are expanding their investments in India.

What is theoretically important is that offset policy functions as a demand-side policy. As Edler & Georghiou (2007, 2012) emphasized, public procurement is not merely demand creation but has the power to shape market structure itself. Using its huge defense procurement market (FY2024 budget of INR 6.21 lakh crore, about $75B) as leverage, the Indian government has succeeded in compelling foreign firms to produce within India and, as a result, in integrating Indian firms into the global supply chain.

The conditions for this success are threefold: (1) a huge domestic market, (2) a consistent policy commitment, and (3) a roadmap for phased technology acquisition. Without the attractiveness of the Indian market, foreign firms might decline contracts on the grounds of offset obligations. And if policy changes frequently, long-term investment cannot be induced. By satisfying these three conditions, India established the credibility of its policy.

In the interest of fairness, we examine the counterarguments to the offset-centered explanation. First, there is the view that the entry of Western firms into India can be explained solely by market and geopolitical motives—a hedge in the strategy against China and access to one of the world's foremost growth markets—regardless of whether offset obligations exist. Second, international empirical research on the effectiveness of offset policy has widely reported cases in which the fulfillment of obligations concentrates in low-value-added work, or in which the method of counting fulfillment becomes a mere formality, so there is no guarantee that offsets lead to genuine technological capability building. Third, India's 12.0% share in our data may be partly affected by the composition of the corpus—namely, the abundance of English-language defense-specialist media in India. All of these counterarguments are legitimate, but they do not overturn this report's central claim. Market and geopolitical motives explain "whether or not" foreign firms engage with India, but the fact that such engagement takes the "form" of joint ventures, technology transfer, and integration of local suppliers—that is, the very bidirectional network structure this analysis observed—is difficult to explain without the institutional conditioning of offset obligations. This report's refined claim is that offset policy governed not the occurrence of engagement but the structure of engagement.

The lesson for Japan is clear. Japan too has a defense procurement market of considerable scale (FY2024 defense budget of ¥8.9 trillion, about $60B). However, because it lacks an institutional mechanism equivalent to offset policy, its procurement does not lead to the strengthening of the domestic industrial base. In its procurement of the F-35, Japan carries out final assembly domestically (Mitsubishi Heavy Industries), but it imports the major components from the United States. With an offset policy, it might have been possible to require Lockheed Martin to transfer technology to Japanese firms and to localize component manufacturing.

3.3 The Oligopolistic Structure of the Commercial Aircraft Industry: Path Dependence and Lock-In

The process by which the Boeing/Airbus oligopoly (a combined market share of 90–95%) formed is a textbook example of path dependence in which technological and institutional factors are intricately intertwined.

Historically, the commercial aircraft market was relatively competitive until the 1970s. In the United States there were Boeing, McDonnell Douglas, Lockheed, and Convair; in Europe there were BAC, Sud Aviation, Fokker, and others. However, the development competition over wide-body aircraft (the Boeing 747, 1969; the McDonnell Douglas DC-10, 1971; the Lockheed L-1011, 1972) dramatically changed the industry structure.

The Boeing 747 cost the enormous sum of $1B to develop at the time (about $7B in 2024 dollars), and Boeing was driven to the brink of bankruptcy. The Lockheed L-1011 was a commercial failure, and Lockheed withdrew from the civil aircraft market (1984). McDonnell Douglas lost its reputation over the DC-10 accident (American Airlines Flight 191 in 1979, 273 deaths) and was ultimately absorbed by Boeing (1997).

In Europe, out of a recognition that individual countries developing aircraft separately could not compete with the United States, the Airbus consortium was established (1970). Airbus began as a four-nation joint venture among France (Aérospatiale), Germany (Deutsche Airbus), the United Kingdom (Hawker Siddeley, later BAE Systems), and Spain (CASA). What is important is that Airbus was not a purely private company; the governments of each country held shares and covered a substantial portion of development costs through government loans (which carried a repayment obligation but were forgiven in the event of commercial failure).

This is a European version of the "Entrepreneurial State" noted by Mazzucato (2013). Airbus's success was not a product of the "free market" but a consequence of strategic government intervention. The United States brought this government support to the WTO as an unfair subsidy (the 2004 Boeing vs. Airbus dispute), but the WTO panel found that both sides had illegal subsidies (2010, 2018). In other words, Boeing too had received government support indirectly through NASA and DoD contracts.

The mechanism sustaining the current oligopolistic structure rests on four factors.

First, the escalation of development costs. The Boeing 787's $32B development cost raised the hurdle for new entry to an effectively insurmountable level. This amount rivals the GDP of a small country.

Second, the growing complexity of certification regimes. The certification requirements of the FAA and EASA have grown stricter year by year in response to past accidents. After the two crashes of the Boeing 737 MAX (Lion Air in 2018, Ethiopian Airlines in 2019, a combined 346 deaths), the FAA comprehensively revised the certification process, making it even more time-consuming and costly. For new entrants, obtaining this certification is not only a technological challenge but also a political task of building relationships with regulators.

Third, the importance of after-sales service networks. The bulk of an aircraft's operating cost consists of post-purchase maintenance, parts replacement, and upgrades. Boeing and Airbus have service bases around the world and have built systems capable of 24-hour response. From an airline's perspective, however cheap the purchase price, introducing an aircraft from an emerging manufacturer with inadequate after-sales service carries too high an operational risk.

Fourth, sunk costs and lock-in effects. By standardizing their fleet on a single type, airlines can enjoy benefits such as (1) reduced pilot training costs, (2) reduced mechanic training costs, and (3) more efficient parts inventory. An airline that has once built its operational system around Boeing aircraft has a strong incentive to choose Boeing again in its next procurement. This is a classic example of what economics calls "switching costs."

China's COMAC C919, even with state support, faces all four of these barriers, demonstrating that entry barriers operate in reality and are not merely a theoretical concept (discussed in detail in the context of industry analysis in Section 4.1).

What is theoretically important is that this oligopolistic structure has a character close to that of a natural monopoly. Because economies of scale are extremely large, the market can sustain only a small number of firms. According to the increasing-returns theory of Arthur (1989), in such industries an initial accidental advantage becomes entrenched through lock-in effects. The fact that Boeing was the first to commercialize the jet airliner (the 707, 1958) and came to dominate the market became the source of its subsequent competitive advantage.

The policy implication is that a simple "promotion of new entry" is insufficient. In industries with natural-monopoly characteristics, indirect approaches are more effective than competition policy: (1) regulating the conduct of oligopolistic firms, (2) ensuring the openness of technical standards, (3) promoting competition in the after-sales service market, and (4) developing alternative platforms through international cooperation.

3.4 The Formation of SpaceX's Quasi-Monopoly in the Space Industry: An Unintended Consequence of NASA's Commercial Crew Program

The quasi-monopoly whereby SpaceX accounts for 87% of U.S. orbital launches (134 launches in 2024) has an aspect of being an unintended consequence of a government policy—NASA's Commercial Crew Program (CCP). This section examines that interpretation in contrast with strong competing explanations.

The CCP was launched in 2010 and provided SpaceX and Boeing with $2.6B and $4.2B respectively (a combined $6.8B), commissioning the development of crewed transport capability to the ISS. This policy had three objectives: (1) securing a means of crewed transport after the retirement of the Space Shuttle (2011), (2) ending dependence on Russia's Soyuz, and (3) cultivating a commercial space industry.

The innovation of the CCP lay in adopting "fixed-price milestone contracts" rather than the conventional "cost-plus contract" (reimbursement of actual costs plus a fixed profit margin). Firms received payment only when they achieved defined milestones (design review, test flights, certification), and bore any cost overruns themselves. This contract form gives firms an incentive to reduce costs and promotes innovation.

The results were dramatic. SpaceX obtained crewed flight certification in 2020 (the Demo-2 mission) and has since conducted regular crewed transport to the ISS. Boeing, on the other hand, faced development delays and technical problems (the failure of the 2019 Starliner uncrewed test flight, the success of the 2022 retest, and the partial success of the 2024 crewed test flight but with a return problem) and, as of 2024, has not reached full operational certification.

The reasons for this contrasting outcome can be attributed to three institutional and organizational factors.

First, SpaceX's vertical integration strategy. SpaceX produces in-house most of its engines (Merlin, Raptor), avionics, structural materials, and software. This achieved (1) reduced supply chain risk, (2) faster design changes, and (3) lower costs. Boeing, by contrast, maintained a traditional horizontal division-of-labor model and depended on numerous subcontractors. The main cause of the Boeing 787's development delays was likewise the complexity of supply chain management.

Second, SpaceX's culture of "rapid iteration." SpaceX has a development philosophy of tolerating failure, learning quickly, and improving. In developing Starship, it went through multiple explosive failures (2020–2023) and gradually improved performance. This approach contrasts with the traditional aerospace industry's "careful design to minimize failure." Theoretically, this can be understood in terms of the organizational strategy of "learning from failure" discussed by Edmondson (2011) and the "exploration and exploitation" trade-off of March (1991). SpaceX emphasizes exploration and judges that the long-term benefits of learning outweigh the short-term costs of failure.

Third, NASA's "hands-off" supervisory style. In the CCP, NASA did not specify detailed technical specifications and presented only performance requirements (e.g., transporting seven crew members to the ISS, capable of a 24-hour on-orbit stay). This "performance-based contract" gave firms freedom of design and promoted innovation. By contrast, in conventional NASA contracts, detailed technical specifications (design specification documents running to thousands of pages) were included in the contract, constraining the firms' creativity.

However, the "success" of the CCP produced the unintended consequence of a SpaceX quasi-monopoly. NASA's intent was to cultivate multiple commercial providers and create a competitive market. Yet, due to Boeing's failure, SpaceX acquired a de facto monopolistic position. Currently, dependence on SpaceX is extremely high: crewed transport to the ISS, the Artemis lunar lander (Starship HLS, a $2.9B contract), the Starlink satellite constellation (more than 6,000 units launched), Department of Defense launch services (National Security Space Launch), and more.

In economic terms, this is the classic dilemma of a "picking winners" policy. When the government supports a particular firm, there are three risks: (1) if it succeeds, industrial competitiveness improves, but (2) if it fails, it is a waste of tax money, and (3) if it succeeds too well, a monopoly is born and competition is lost. The CCP corresponds to case (3).

Here we consider the strongest counterargument to the CCP-centered explanation: namely, the view that SpaceX's dominant position is a consequence of its own technological breakthrough in reusable rockets and of private risk capital, and that government policy is not the essential factor. The strongest basis for this view is the fact that Boeing, which received more funding than SpaceX in the CCP ($4.2B vs. $2.6B), failed. If policy determined outcomes, the more heavily supported firm should have won. This counterargument correctly shows that policy is not a "sufficient condition." But it cannot deny that policy is a "necessary condition." In 2008, after three consecutive failures of the Falcon 1, SpaceX was on the brink of running out of funds, and it is acknowledged by the parties themselves that NASA's Commercial Resupply Services contract that same year (the $1.6B touched on in Section 5.2) kept the business alive. Moreover, the substance of the current "quasi-monopoly"—ISS crewed transport, the Artemis lunar lander, national security launches—is composed almost entirely of government demand itself. The most accurate formulation is therefore as follows: the government's demand-side policy created the arena of competition and the conditions for survival (a necessary condition), while which firm prevailed in that arena was decided by firm capability (a selection condition). This formulation does not abandon the theory of the Entrepreneurial State but refines the scope of its application.

More serious still is the fact that SpaceX depends heavily on Elon Musk as an individual. Uncertainty over the allocation of management resources and political statements following Musk's acquisition of Twitter (now X) (2022, $44B) is a source of concern regarding SpaceX's stability. A situation in which infrastructure important to national security depends on the discretion of a single individual signifies institutional fragility.

The policy lesson is that the market-shaping power of demand-side policy is extremely strong and can produce unintended consequences. The CCP achieved the goal of cultivating a commercial space industry, but it did not achieve the goal of creating a competitive market. In future policy design, institutional devices are needed: (1) diversified investment across multiple providers, (2) ensuring interoperability by making technical standards open, and (3) avoiding long-term contracts (to prevent lock-in).

In this section, we explicitly distinguish the structural differences among the aircraft, space, and defense industries and discuss the policy implications specific to each. As noted earlier, the data in this analysis intermingle three heterogeneous industries, and a simple comparison of mention counts does not reflect industrial competitiveness. The aim of this section is to show that the formative mechanisms identified in Section III—entry barriers and lock-in created by certification regimes (Section 3.3), ITAR-type institutional path dependence (Section 3.1), and the market-shaping power of demand-side policy (Sections 3.2 and 3.4)—operate with different intensities and in different combinations across the three industries, and from this to derive different policy prescriptions for each industry.

4.1 The Commercial Aircraft Industry: The Stability of Oligopoly and the Difficulty of Disruption

The commercial aircraft industry (809 items in the "aircraft" category in this data, though it may include military aircraft) has a Boeing/Airbus oligopoly that has been stable over the long term. Since the 1990s, the two firms' market shares have fluctuated somewhat (Boeing 40–60%, Airbus 40–60%), but the oligopolistic structure itself has not been shaken.

The reason for this stability lies, in addition to the four entry barriers described earlier (development cost, certification regimes, after-sales service, lock-in effects), in the characteristics of demand. The commercial aircraft market has three features: (1) oligopolistic customers (airlines too have consolidated), (2) long-term relationships (20–30 years of fleet operation), and (3) an extreme emphasis on safety. These work to the advantage of incumbents.

Attempts at new entry have historically failed. Bombardier's (Canada) CSeries was a technical success, but due to Boeing's political pressure (a complaint to the U.S. Department of Commerce, 2017) and cost overruns, it was sold in its entirety to Airbus (2018, continuing as the Airbus A220). Mitsubishi Aircraft's SpaceJet (formerly the MRJ) announced the cancellation of its development in 2023 due to development delays and difficulties in obtaining type certification. Russia's Irkut MC-21 was effectively excluded from Western markets by sanctions following the invasion of Ukraine.

The only partial "success" is China's COMAC C919. The C919 entered commercial service in 2023, more than 15 years after development began (2008), and has secured a certain number of orders in the Chinese domestic market (about 1,000 aircraft as of 2024). However, its competitiveness in international markets is uncertain. It has not obtained FAA certification, and it imports major components from Europe and the United States, including the engines (CFM International LEAP-1C), avionics (Honeywell, Rockwell Collins), and flight control system (Parker Aerospace). If U.S.-China technological competition intensifies, there is a risk that these supply chains will be cut off.

The policy implication is that, having recognized how extremely difficult it is to promote new entry in the commercial aircraft industry, alternative approaches should be considered. Specifically, three strategies can be envisioned.

Strategy 1: Joint development through international cooperation. Development by a single country has a low probability of success, but the possibility rises if multiple countries pool their resources. Airbus's success proves this. It is theoretically possible for Asian countries such as Japan, South Korea, and India to cooperate and launch an "Asian Wide-Body" program. However, four political difficulties must be overcome: (1) the sharing of development risk, (2) the terms of technology transfer, (3) the allocation of production bases, and (4) the allocation of markets. The fact that Europe took 40 years to build the Airbus system shows the difficulty of international cooperation.

Strategy 2: Specialization in niche markets. Boeing and Airbus specialize in narrow-body aircraft of 150–400 seats and in wide-body aircraft. The regional-jet market of 50–100 seats is relatively competitive (Embraer, Bombardier/Airbus A220). There is also a smaller commuter-aircraft market (ATR, De Havilland Canada). Japan's SpaceJet targeted this 70–90-seat market but failed. Nevertheless, niche specialization as a strategy remains valid.

Strategy 3: Investment in disruptive innovation. According to the disruptive innovation theory of Christensen (1997), the way to disrupt a market dominated by incumbents follows three stages: (1) offering a new value proposition that existing customers do not value, (2) opening up new customer segments, and (3) gradually improving performance to erode the existing market. In commercial aircraft, disruptive candidates include electric aircraft (e-VTOL, electric vertical take-off and landing aircraft), supersonic passenger aircraft, and stratospheric airships. These have not yet been commercialized, but technological progress is raising their feasibility. The government's role is early investment in these disruptive technologies (DARPA-style funding).

For Japan, Strategy 3 is the most realistic. The lesson of the SpaceJet's failure is the difficulty of competing head-on with Boeing/Airbus in existing markets. Instead, a strategy of investing in next-generation technologies such as electric aircraft and hydrogen aircraft to create new markets is promising. Applying the hydrogen engine jointly researched by JAXA and IHI, and the fuel-cell technology developed by Toyota, to aircraft is an area where Japan can leverage its strengths.

4.2 The Space Industry: Rapid Change and Policy Malleability

The space industry (1,271 items in this data, 23% of the total: satellites 824, lunar landers 216, spacecraft 125, rockets 106), in contrast to the commercial aircraft industry, is an industry that changes rapidly and in which the effects of policy intervention are large.

The fundamental reason for this difference lies in the immaturity of the market. The commercial aircraft market has a 100-year history, and its technology, institutions, and business models are mature. By contrast, the commercial space market began in earnest in the 2000s and is still in the process of formation. SpaceX's first Falcon 9 launch was in 2010, and commercial satellite constellations (SpaceX Starlink, OneWeb, Amazon Kuiper) took off in earnest only after 2019. The newness of this market means that incumbents' advantages are not entrenched and there is ample room for new entry and disruptive innovation.

We analyze the structure of the space industry by dividing it into three subsegments.

In the launch services market, SpaceX holds an overwhelming advantage (87% of the U.S. and roughly 40–50% of the world). The source of this advantage is the dramatic cost reduction enabled by reusable rocket technology (the vertical landing and reuse of the Falcon 9 first stage). Conventional rocket launch costs were $10,000–$30,000/kg, but the Falcon 9 achieved $2,700/kg. Furthermore, if Starship becomes fully reusable, less than $100/kg comes into view. This 10- to 100-fold cost reduction is transforming the business model of the entire space industry.

Competitors (Blue Origin, Rocket Lab, Arianespace, Roscosmos, CNSA) lag behind SpaceX both technologically and in cost. Blue Origin's New Glenn is scheduled to launch in 2025, but development delays continue. Arianespace's Ariane 6 succeeded in its first launch in 2024, but it lacks reusability technology and has challenges in cost competitiveness. Rocket Lab has succeeded in small-satellite launches (Electron, 250 kg), but it occupies a different market segment from SpaceX.

In policy terms, a strategy is needed that recognizes the risk of a SpaceX monopoly while cultivating alternative providers. The prescription on the U.S. side (diversified ordering, open standards, avoiding long-term contracts) is as described at the end of Section 3.4, but countries other than the United States face additional, specific challenges: namely, (1) deciding whether and at what level of burden to pursue a "strategic launch capability" policy in which Europe, Japan, India, and others maintain their own rocket industries even at the expense of commercial profitability, and (2) promoting competition in the small-satellite launch market, which does not compete directly with SpaceX.

The satellite industry (824 items, 19% of the total) is expanding rapidly due to falling launch costs. In particular, low-earth-orbit (LEO) satellite constellations are being deployed for diverse uses: communications (Starlink, OneWeb, Kuiper), earth observation (Planet Labs, BlackSky), and IoT (Swarm Technologies, Astrocast).

This market has three characteristics: (1) relatively low entry barriers (small satellites can be developed for $1M–$10M), (2) rapid technological progress (CubeSats, miniaturization, onboard AI), and (3) diverse business models (communications, observation, science, military). Accordingly, the opportunities for policy intervention are large.

The challenges for Japan's satellite industry policy are threefold: (1) high launch costs (the H3 rocket is 2–3 times the Falcon 9), (2) the smallness of the domestic market, and (3) insufficient policy support for commercial satellites. As countermeasures, one can consider (1) promoting the commercialization of the H3 rocket (strengthening cooperation between JAXA and Mitsubishi Heavy Industries), (2) government purchase of satellite data (earth observation, disaster monitoring), and (3) support for the development of satellite constellations (securing communications sovereignty).

Lunar and Mars exploration (216 lunar landers, 125 spacecraft) consists of large-scale, government-led programs. NASA's Artemis program (lunar base construction), ESA's Moon Village, China's Chang'E program, and India's Chandrayaan/Gaganyaan programs are all underway. These have not only scientific and technological significance but also geopolitical and strategic significance.

The economic justification for lunar exploration is difficult (short-term commercial benefits are limited). However, it is justified for four reasons: (1) the spillover effects of technological development, (2) international prestige, (3) long-term resource utilization (the Moon's water ice, helium-3), and (4) a stepping stone to Mars exploration. This is a classic example of the "mission-oriented policy" advocated by Mazzucato (2013).

Japan's ispace attempted a lunar landing as a private company (HAKUTO-R Mission 1 in 2023); it failed to land but is planning Missions 2 and 3 in 2024–2025. The government's role is (1) support for technological development (joint research with JAXA), (2) anchor tenancy (the government serving as the initial customer), and (3) promotion of international cooperation (participation in the Artemis Accords).

4.3 The Defense Industry: The Intersection of National Security and Industrial Policy

The defense industry (in this data, 248 weapons items, 411 missile items, 191 fighter items, 462 UAV/drone items, etc.) is the most complex domain, where national security and industrial policy intersect.

The structural characteristics of the defense industry are fourfold: (1) the monopsonistic nature of demand, (2) the dual-use (civil-military) nature of the technology, (3) constraints on international transactions (ITAR, export controls), and (4) long-term relationships (programs spanning decades).

Monopsony distorts the market mechanism of the defense industry. In most countries, the sole customer for defense equipment is the government (the ministry of defense). This means that firms have no bargaining power over price and that the government's procurement policy determines the industry structure. For example, the U.S. F-35 program (a total of $1.7 trillion, the largest defense procurement in history) positioned Lockheed Martin as the dominant firm in the fighter market. This policy choice is unrelated to the competitive process of the private market.

Dual-use technology generates positive externalities from military technology development. Many civilian technologies, such as GPS, the Internet, jet engines, and radar, derive from military research. As Mazzucato (2013) emphasized, DARPA's investment made Apple's iPhone possible (the touchscreen, Siri, GPS, the Internet, etc.). This spillover effect provides social justification for military R&D investment.

However, the dual-use nature also creates a policy dilemma. China's "Military-Civil Fusion" strategy explicitly aims to repurpose civilian technology for military use. In response, the United States is strengthening export controls (the Entity List, ITAR) in an attempt to block technology leakage to China. This contest for technological supremacy is fragmenting global supply chains and inviting "technology bloc-ization."

The challenges for Japan's defense industry policy are threefold: (1) the smallness of the domestic market (a defense budget of ¥8.9 trillion, about 1/10 of the U.S. $850B), (2) the constraints of the Three Principles on Arms Exports (relaxed in 2014 to the "Three Principles on the Transfer of Defense Equipment and Technology," but still restrictive), and (3) the fragility of the technological base (fighter engines, stealth technology, etc.).

As policy responses, three directions can be envisioned.

Direction 1: Strengthening joint development with allies. In the F-X program (the next-generation fighter), joint development among Japan, the United Kingdom, and Italy was agreed upon (2022, GCAP: Global Combat Air Programme). This fighter, scheduled for deployment in 2035, integrates Japanese technology (radar, electronic warfare systems), British technology (engines, stealth), and Italian technology (avionics). This cooperation offers not only the sharing of development costs but also the possibility of producing a product superior to what single-country development could achieve, thanks to technological complementarity.

Direction 2: Promoting the transfer of defense equipment. Arms exports offer three benefits: (1) unit-cost reduction through expanded production scale, (2) maintenance of the technological base, and (3) improved interoperability with allies. However, Japan's arms exports remain limited (the export of radar to the Philippines in 2023 is one of the few examples). In policy terms, it is necessary to promote exports to friendly nations and allies and thereby strengthen the industrial base.

Direction 3: Cultivating the UAV/drone industry. UAVs/drones, which account for 462 items (about 8% of the total) in this data, had their tactical importance demonstrated in the Ukraine conflict. Their characteristics—low cost (from thousands to millions of dollars), mass deployability, and onboard AI—are becoming a game changer for conventional high-cost weapons systems (fighters, tanks). Japan lags behind China (DJI and others) in drone technology, but military drones are a separate market. Support by the Ministry of Defense for domestic drone development and the promotion of private drone firms' entry into the defense market are urgent tasks.

In this section, we integrate the policy typology of Edler & Georghiou (2007, 2012) with Mazzucato's (2013, 2024) theory of the Entrepreneurial State and present the theoretical foundations of policy intervention in the aerospace and defense industries (a detailed breakdown of the typology of policy instruments is organized in Appendix B). What is important is to conceive of policy not merely as "correcting market failure" but as "shaping the market."

5.1 Reassessing Supply-Side Policy: The Limits of R&D Support and the Time Lag of Capability Building

In conventional industrial policy discourse, supply-side policies (R&D subsidies, tax incentives, human resource development) were central. In the aerospace and defense industries, however, R&D support alone is insufficient.

The first reason is the time lag of capability building. Developing a commercial aircraft takes 10–15 years, and obtaining certification takes several more years. Fighter development is similarly 15–20 years. This long duration means that governments may change several times before the effects of R&D investment materialize, creating a risk that policy continuity is lost. The failure of Japan's SpaceJet was due not only to technical problems but in part to a lack of policy commitment (the absence of large-scale government launch aid).

The second reason is the "valley of death" between technological development and commercialization. Many technologies succeed at the laboratory stage but fail at the commercialization stage. Bridging this gap requires not only R&D support but also demand-side policy (government procurement, regulatory reform) as an indispensable element. The success of NASA's Commercial Crew Program was due to a combination of R&D support (COTS: Commercial Orbital Transportation Services, 2006–2013, $800M) and a demand guarantee (crewed transport contracts, $6.8B).

The third reason is the importance of complementary assets. As discussed in relation to certification regimes in Section 2.3, within the framework of Teece (1986), the source of competitive advantage lies less in the technology itself than in complementary assets such as (1) production facilities, (2) distribution networks, (3) brands, and (4) regulatory-compliance capabilities. R&D support promotes technological development but does not support the building of complementary assets.

Accordingly, supply-side policy should be redesigned along the following three lines.

Redesign 1: Institutionalizing long-term commitment. As mechanisms for continuing policy across changes of government, one can consider (1) securing budgets by law (the U.S. National Defense Authorization Act is enacted every year, but long-term programs are protected by multi-year contracts), (2) establishing independent agencies (JAXA, DARPA, etc.), and (3) long-term partnership agreements among industry, academia, and government.

Redesign 2: Phased support according to Technology Readiness Level (TRL). Based on the NASA TRL scale (1–9), institutionalize phased support: (1) competitive funding for universities and research institutions at TRL 1–3 (basic research), (2) STTR/SBIR-type venture support at TRL 4–6 (technology demonstration), and (3) government procurement at TRL 7–9 (system demonstration and operation).

Redesign 3: Support for building complementary assets. Support capability building beyond technological development, such as certification support (consulting and cost subsidies for FAA/EASA certification), supply chain building support (cultivating Tier 2/3 suppliers), and support for international standardization activities (promoting Japanese proposals at the ISO, ICAO, etc.).

5.2 The Market-Shaping Power of Demand-Side Policy: The Strategic Use of Public Procurement

The demand-side policy emphasized by Edler & Georghiou (2007, 2012) has the power to shape industry structure itself, beyond merely creating demand.

The market-shaping power of public procurement operates through four mechanisms.

Mechanism 1: Creating an initial market. For new technologies, a private market often does not exist in the early stages. By becoming the "first customer," the government enables firms to see the prospect of recouping their investment. SpaceX's early Falcon 1 development was privately funded, but without NASA's CRS (Commercial Resupply Services) contract (2008, $1.6B), SpaceX would likely have gone bankrupt.

Mechanism 2: Setting technical standards. The specifications of government procurement become de facto technical standards. The safety standard required by NASA's Commercial Crew Program (a Loss of Crew probability of 1/270 or less) is becoming a de facto reference standard for commercial crewed spaceflight. Similarly, the DoD's cybersecurity standards (NIST SP 800-171, CMMC) are propagating across the entire defense supply chain.

Mechanism 3: Risk sharing. When the uncertainty of developing a new technology is high, a firm cannot bear the risk alone. By contracting and paying in stages (milestone contracts), the government can spread the risk. This differs from the financing provided by venture capital. VC demands equity, but a government contract does not demand equity (preserving the firm's independence).

Mechanism 4: Support for scale-up. When a new technology moves from the demonstration stage to the commercialization stage, an expansion of production scale is needed. Large-scale government procurement (e.g., the F-35 program, more than 3,000 aircraft) enables firms to achieve economies of scale and lower their unit costs.

However, public procurement carries three risks.

Risk 1: Lock-in and vendor monopoly. As in the SpaceX case detailed in Section 3.4, if dependence on a particular firm grows too high, competition is lost and the government's bargaining power declines. In addition to the institutional devices listed at the end of that section (diversified investment, open standards, avoiding long-term contracts), limiting contract periods and mandating periodic re-competitive bidding is effective in preventing lock-in from becoming entrenched after the fact.

Risk 2: Crowding out (the crowding out of private investment). If government procurement monopolizes the market, the private market does not develop. In the space industry, government procurement (NASA, DoD, intelligence agencies) accounts for the bulk of the market, so the commercial market is limited. As countermeasures, promoting dual-use technology (repurposing military technology for civilian use) and expanding the private market through regulatory reform (allocation of frequencies for satellite communications, etc.) are needed.

Risk 3: Political pressure and rent-seeking. Large-scale procurement tends to become a target of political pressure. In the United States, legislators demand the siting of factories in their own districts, and DoD procurement decisions are distorted (so-called "pork barrel," or particularistic benefit-steering). In the F-35 program, Lockheed Martin deliberately dispersed its supply chain across 45 states to secure political support in Congress. As countermeasures, ensuring transparency in the procurement process, evaluation by expert committees, and strengthened oversight by the Government Accountability Office are needed.

The challenges for Japan's public procurement policy are threefold: (1) an overemphasis on price (the lowest-price award method), (2) short-term contracts (single-year budgeting), and (3) risk aversion (a preference for precedent). These impede the entry of innovative firms.

As directions for reform, one can consider (1) placing greater weight on factors other than price (technological capability, innovation, medium- to long-term cost), (2) expanding multi-year contracts (possible with Diet approval), and (3) introducing pre-commercial procurement (procurement at a stage before commercialization, as practiced in Europe).

Pre-commercial procurement in particular is a powerful, underused policy tool in Japan. In it, the government contracts with multiple firms at the R&D stage and supports prototype development. Only successful firms proceed to the next stage (commercial production). This mechanism offers three benefits: (1) maintaining competition while (2) the government shares the risk and (3) multiple technological approaches can be explored in parallel.

5.3 Systemic Policy: The Institutional Complementarity of the Ecosystem

In Edler & Georghiou's typology, systemic policy targets the structure of the innovation ecosystem as a whole. In the aerospace and defense industries, the ecosystem consists of five elements: (1) firms (OEMs, Tier 1/2/3 suppliers), (2) research institutions (universities, national laboratories), (3) government (procurement agencies, regulators), (4) finance (VC, banks), and (5) human capital (engineers, managers).

The concept of institutional complementarity (Aoki, 2001) shows that when these elements mutually reinforce one another, the performance of the system as a whole improves. The success of the U.S. space industry ecosystem is due to the complementarity of six elements: (1) NASA's technological development and risk-taking, (2) DARPA's advanced research, (3) large-scale procurement by the DoD and intelligence agencies, (4) Silicon Valley's VC funding, (5) the supply of human capital from Stanford, MIT, and others, and (6) the flexibility of regulators such as the FAA and FCC.

The challenge for Japan's ecosystem is the absence of this complementarity. Specifically, there are five problems: (1) the siloing of JAXA, the Ministry of Defense, and METI, (2) insufficient VC investment in the aerospace industry (risk aversion), (3) the weakness of universities' orientation toward practical application (an overemphasis on basic research), (4) the conservatism of regulators, and (5) the low mobility of human capital (little movement between firms and between industry, academia, and government).

As concrete examples of systemic policy, we propose the following three.

Proposal 1: Institutional deepening of industry-academia-government collaboration. Current joint research is often short-term (2–3 years) and small-scale (tens of millions to hundreds of millions of yen). This should be deepened along four lines: (1) lengthening (10-year programs), (2) scaling up (billions to tens of billions of yen), (3) mandating personnel exchange (secondments from firms to universities, and vice versa), and (4) flexible allocation of intellectual property rights (so that both universities and firms can commercialize). The U.S. NSF Engineering Research Centers and Europe's Framework Programme serve as references.

Proposal 2: Expanding regulatory sandboxes. The commercialization of new technologies often does not conform to existing regulation. A regulatory sandbox is a system that temporarily exempts regulation under limited conditions and enables demonstration experiments. In Japan it has been introduced in the financial sector but is unused in the aerospace field. For example, regulatory sandboxes should be applied to (1) urban drone flight, (2) reusable rocket experiments, and (3) supersonic flight demonstrations.

Proposal 3: Promoting the circulation of human capital. One of the success factors of Silicon Valley is the mobility of human capital between firms and between industry and academia. In Japan, lifetime employment practices and seniority systems impede this. In policy terms, one can consider (1) promoting cross-appointments (university faculty concurrently serving as corporate officers), (2) personnel exchange between government and firms (the U.S. Intergovernmental Personnel Act serves as a reference), and (3) improving the taxation of stock options (to support venture firms' acquisition of talent).

5.4 Mission-Oriented Policy: Concretizing the Entrepreneurial State

Mazzucato's (2013, 2024) mission-oriented innovation policy argues that the government should set clear "missions" and mobilize the resources of industry, academia, and government. The Apollo program (the lunar landing) and the Manhattan Project (the development of the atomic bomb) are historical examples of success.

The five principles of mission-oriented policy are as follows.

Principle 1: Bold and inspiring missions. A mission must be bold, socially significant, and inspiring to people. Concrete missions with clear deadlines are effective, such as "Realize a fully electric passenger aircraft by 2035" or "Build a lunar base by 2040."

Principle 2: Cross-sectoral coordination. A mission cannot be achieved by a single ministry. The coordination of multiple sectors—aerospace, defense, energy, information and communications, etc.—is needed. Institutionally, this requires (1) establishing a mission headquarters under the direct control of the prime minister, (2) securing cross-ministry budgets, and (3) progress management via KPIs.

Principle 3: Public-private risk sharing. Achieving a mission entails high risk. A risk-sharing arrangement is needed in which the government bears the risk of the early stages and the private sector commercializes upon success. However, as Mazzucato (2013) criticizes, "socialize the risk, privatize the reward" is inequitable. If the government bears the risk, part of the reward upon success should also be returned to society. Specifically, one can consider (1) the government acquiring equity, (2) retaining part of the intellectual property rights, and (3) collecting licensing fees.

Principle 4: Bottom-up experimentation. Missions are set top-down, but the means of realizing them should be bottom-up. If the government specifies detailed technical specifications, innovation is impeded. Instead, present only performance requirements and give firms and researchers creative freedom. NASA's Commercial Crew Program is an example of success.

Principle 5: Long-term patience. A mission takes 10–20 years over the long term. Tolerance of short-term failure is necessary. SpaceX failed three consecutive launches early on, but NASA maintained the contract. The provision of this "patient capital" is what makes innovation possible.

As candidate missions for Japan, we propose the following three.

Mission 1: "Realize commercial operation of hydrogen aircraft by 2040." This concretizes Strategy 3 of Section 4.1 (investment in disruptive innovation) as a mission. Japan has high competitiveness in hydrogen technology (Toyota fuel cells, Kawasaki Heavy Industries hydrogen turbines, IHI hydrogen engines). Leveraging this strength, hydrogenizing aircraft has three kinds of significance: (1) decarbonizing the aviation industry, (2) building hydrogen infrastructure, and (3) creating a new aircraft market. Airbus is targeting a hydrogen aircraft in 2035, and Japan should compete.

Mission 2: "Realize regular transport to a crewed lunar base by 2035." Participate in NASA's Artemis program and develop Japan's crewed lunar lander and lunar rover. In cooperation with private firms such as ispace, establish a commercial lunar transport service. This has three kinds of significance: (1) strengthening the technological base of the space industry, (2) international prestige, and (3) a foothold for resource exploration.

Mission 3: "Realize a fully autonomous defense UAV system by 2030." AI-equipped autonomous drones will become the core of future defense. Japan has strengths in AI and robotics technology (Sony, Toyota, Preferred Networks) and should apply these to defense. As demonstrated in the Ukraine conflict, low-cost, mass-deployable UAVs are fundamentally changing tactics.

Whereas Sections III and IV discussed "how" the current structure was formed and Section V discussed "what means of intervention exist," this section shifts perspective and analyzes the dynamics that intervention itself sets in motion—why change is politically obstructed, and what side effects intervention produces. Policy intervention produces not only its intended effects but also unintended consequences. Below, we discuss three typical mechanisms.

6.1 The Political Economy of Lock-In Effects: Why Change Is Difficult

Although the current network structure (U.S.-centric, hub-dominated, low-redundancy) is disadvantageous for many stakeholders, change is difficult. The reason for this paradox lies in the political economy of lock-in effects.

Resistance from vested interests. Actors who benefit from the current structure (U.S. firms, hub firms, regulators) resist change. For Lockheed Martin, the F-35 monopoly brings enormous profits (program lifetime revenue of $1.7 trillion). Part of this profit is converted into political influence in the form of lobbying of Congress, campaign contributions, and the siting of factories in legislators' districts. According to the collective action theory of Olson (1965), a small, organized interest group (defense-related firms) has stronger political influence than a large, dispersed group (taxpayers).

Sunk costs and stranded assets. Investments in existing infrastructure, facilities, and human capital are sunk costs. Boeing's investment in its 737 production line runs into billions of dollars. Transitioning to a new aircraft platform renders these investments worthless. Rationally, firms should not let sunk costs influence their decisions, but as behavioral economics shows, in practice they try to protect existing investments due to the sunk-cost fallacy.

Coordination failure. Change requires the coordination of multiple actors. For example, hydrogenizing aircraft requires five changes simultaneously: (1) design changes by aircraft manufacturers, (2) technological development by engine manufacturers, (3) the installation of hydrogen infrastructure at airports, (4) the establishment of safety standards by regulators, and (5) changes in airlines' procurement policies. However, each actor falls into a "coordination failure" in which it will not move unless the other actors move. This is a problem of the multiplicity of Nash equilibria in game theory.

As policy responses, three approaches can be envisioned.

Approach 1: Compensation policy. By providing compensation to actors who suffer losses from change, resistance is weakened. For example, in the transition away from the coal industry, retraining programs for coal miners and support for the regional economy are implemented. In the aerospace industry too, when incumbents transition to new technologies, transition support should be provided.

Approach 2: Creating a focal point. To solve the coordination problem, the government presents a "focal point" around which all actors coordinate. Mission-oriented policy is precisely this creation of a focal point. If the government clearly sets the mission of "hydrogen aircraft by 2040," each actor can align its investment toward it.

Approach 3: Sunset clauses and phase-out schedules. Establish a "sunset clause" for existing technologies and institutions and clearly indicate a phase-out schedule. For example, if a regulation stating "the new sale of jet-fuel aircraft is prohibited from 2050" is declared in 2030, firms can prepare over a 20-year grace period. The European ban on the sale of internal-combustion-engine vehicles (2035) is a precedent.

6.2 The Trade-Off Between Economies of Scale and Resilience

In Section 2.4, we diagnosed that low redundancy (an average of 1.55 procurement source countries) is a consequence of individual firms' rational pursuit of efficiency and that, for the system as a whole, it accumulates vulnerability as an externality. The disruption of automotive parts supply after the Great East Japan Earthquake in 2011 and the shortage of medical supplies during the COVID-19 pandemic in 2020 are precedents in which this kind of externality actually materialized. The task of this section is not to restate this diagnosis but to concretize, as institutions, how the externality can be internalized. The following three policy tools can be envisioned.

Tool 1: Redundancy obligations. The government mandates the maintenance of multiple procurement sources for critical components. For example, introduce a regulation stating "engine components must be procured from at least two or more countries." This is a classic approach to internalizing an externality through regulation (Pigou, 1920). However, increased cost becomes an issue. The government must either subsidize the cost of maintaining redundancy or support it through tax incentives.

Tool 2: Strategic stockpiling. The government holds a national stockpile of critical components and materials. The U.S. Strategic Petroleum Reserve and Japan's rare-earth stockpile are precedents. In the aerospace and defense industries, candidates for stockpiling include (1) rare earths, (2) semiconductors, (3) titanium, and (4) engine components. Stockpiling costs should be thought of as an insurance premium.

Tool 3: Scenario planning and stress testing. Apply the stress tests introduced in financial regulation (testing whether financial institutions fail under extreme scenarios) to supply chains as well. The government sets scenarios such as "a complete cutoff of supply from China," "a Taiwan contingency," and "a Middle East conflict" and mandates that firms conduct stress tests. If vulnerabilities are found in the tests, it requires countermeasures (securing alternative sources, increasing stockpiles).

What is important is that efficiency and resilience are not necessarily a trade-off. With appropriate institutional design, the two are compatible. For example, there are technological approaches such as (1) increasing the interchangeability of components through modular design, (2) visualizing and monitoring the supply chain in real time through digital twin technology, and (3) detecting risks in advance through AI prediction and responding preemptively.

6.3 The Collective Action Problem of Allied Coordination

Friend-shoring policy (building supply chains with trusted allies) is theoretically desirable, but its actual realization is difficult. The reason lies in the collective action problem.

Suppose all allies agreed to "complete supply chains within the democracies—the United States, Europe, Japan, South Korea, Australia, etc." However, each country has the following incentives.

Incentive 1: Free-riding. If other countries bear the costs of building the supply chain, one's own country can enjoy the benefits without bearing the costs (free-riding). For example, if Japan builds a semiconductor factory domestically, South Korea can procure from it. South Korea has an incentive to avoid investing at home and to depend on Japan.

Incentive 2: Defection for economic benefit. Given the sheer size of the Chinese market (GDP of $18 trillion, second in the world), completely severing transactions with China entails large economic losses. Each country has the temptation to defect—"other countries will follow friend-shoring, but our country alone will exceptionally trade with China."

Incentive 3: Domestic political pressure. Friend-shoring entails increased costs (higher than procurement from China). Domestic firms and consumers push back, asking "why must we bear the higher costs?" Politicians, thinking of the next election, have an incentive to prioritize short-term economic benefits.

Solving this collective action problem requires the following three mechanisms.

Mechanism 1: An enforceable agreement. Conclude not a mere political declaration but a legally binding treaty. The AUKUS agreement (2021) is an example of institutionalizing security cooperation. Furthermore, a friend-shoring agreement should be strengthened with three elements: (1) concrete numerical targets ("reduce dependence on China by 50% by 2030"), (2) a periodic review mechanism, and (3) penalties for violations (tariffs, suspension of subsidies).

Mechanism 2: Cost-sharing rules. To prevent free-riding, clearly allocate the costs each country should bear. NATO's defense-spending burden (2% of GDP) is a precedent. In friend-shoring, set a rule that "each country bears the obligation to maintain domestic production capacity for critical technologies and components at X% of demand."

Mechanism 3: Monitoring and transparency. Monitor and publicize each country's compliance status. The WTO's Trade Policy Review Mechanism serves as a reference. In a friend-shoring agreement too, an independent body (an international organization or a third-party committee) should periodically evaluate and publicize each country's dependence on China and its domestic production capacity. Transparency deters defection (raising the reputational cost).

More fundamentally, it is important to position friend-shoring not as mere "decoupling from China" but as "positive integration." That is, rather than the negative message of "excluding China," bring to the fore the positive messages of "deepening economic interdependence among democracies" and "prosperity based on shared values." From the standpoint of political economy, this is easier to build domestic support for.

This analysis elucidated the structure of the aerospace and defense industry supply chain network within an integrated framework of scale-free network theory, institutional factors, industrial organization, and innovation policy theory. Its principal theoretical contributions are the following three points.

First, it showed that the current structure is not accidental and can be most coherently explained as a consequence of institutional factors. U.S. centrality—even taking into account the contributions of market size and observation bias—has its reproduction mechanism explained by the institutional factors of the Cold War NATO system, ITAR regulation, and the strategic investments of NASA/DARPA (Section 3.1). Hub dominance can be interpreted as self-reinforcing through the preferential attachment mechanism and through entry barriers created by certification regimes and economies of scale. India's rise—even if market and geopolitical factors contribute to the occurrence of engagement itself—had its structure governed by the strategic industrial policy of offset policy and Make in India (Section 3.2).

Second, it made explicit the causal mechanisms of policy intervention. Supply-side policy (R&D support) alone is insufficient; the market-shaping power of demand-side policy (public procurement) is important. NASA's Commercial Crew Program cultivated SpaceX through a demand guarantee, but unintentionally produced a quasi-monopoly (policy provided the necessary condition for survival, and firm capability selected the winner—Section 3.4). This case simultaneously illustrates the power of policy and the risk of unintended consequences. Systemic policy targets the institutional complementarity of the ecosystem as a whole and promotes industry-academia-government collaboration, regulatory reform, and the mobility of human capital. Mission-oriented policy presents a bold vision and institutionalizes long-term commitment.

Third, it discussed the complexity and limits of policy. Lock-in effects make change difficult through vested interests, sunk costs, and coordination failure. The trade-off between efficiency and resilience is an externality problem and cannot be solved by the market mechanism alone. Allied coordination faces a collective action problem and carries the risks of free-riding and defection.

7.1 Priorities for Policy Intervention: Short-, Medium-, and Long-Term

Among the vast array of policy options, what should be prioritized? Below, we present priorities according to the time horizon.

Note that each of the following measures corresponds explicitly to the analytical findings of the main text. The short-term measures are derived from the absence of redundancy identified in Section 2.4 (an average of 1.55 procurement source countries) and the externality analysis of Section 6.2; the medium-term procurement-institution reforms from the demand-side policy analysis of Section 5.2; and the long-term missions from the mission-oriented policy discussion of Section 5.4. Moreover, every measure faces the implementation constraint discussed in Section 6.1—the political economy of lock-in effects: resistance from vested interests, sunk costs, and coordination failure. Therefore, the very sequencing of forming a track record and a base of support with low-cost, politically low-resistance short-term measures, and then using that as leverage to advance to medium-term institutional reform, is itself part of policy design. The investment scale of the long-term missions (a combined $18–35B for the three) is not at an infeasible level in light of trends in the defense budget and the science and technology budget, but the institutionalization of a commitment spanning multiple administrations (Redesign 1 of Section 5.1) is its precondition.

Short-term (1–3 years): Preparing for the materialization of risk

The top priority is to prepare for the materialization of geopolitical risk (a Taiwan contingency, full-scale U.S.-China confrontation).

• Action 1: Surveys of dependence on China for critical components and stress testing. Require all defense and aerospace firms to identify their procurement from China and evaluate the impact under supply-disruption scenarios (mandatory).
• Action 2: Expanding strategic stockpiles. Stockpile 6–12 months' worth of hard-to-substitute materials and components such as rare earths, semiconductors, titanium, and carbon fiber.
• Action 3: Emergency agreements with friendly nations. Conclude mutual-supply agreements for emergencies within frameworks such as AUKUS, the Quad, and NATO (building contractual relationships in peacetime).

These are relatively low in cost (tens to hundreds of billions of yen) and take effect immediately.

Medium-term (3–7 years): Institutional reform and capability building

In the medium term, transform the institutional framework and build domestic capability.

• Reform 1: Reform of the public procurement system. Shift from an overemphasis on price to value-based procurement (emphasizing technological capability, innovation, and resilience). Expand multi-year contracts. Introduce pre-commercial procurement.
• Reform 2: Introduction of offset policy. Learning from India's success, impose offset obligations (around 30%) on defense and aerospace procurement above a certain amount. Require foreign firms to produce and transfer technology in Japan. However, in view of the fact that Japan does not fully satisfy the success conditions identified in Section 3.2 (market size, consistent policy commitment, a roadmap for phased technology acquisition), and that there is structurally little room for offset negotiation in the U.S. FMS (Foreign Military Sales) procurement that accounts for a substantial share of procurement, make a prior assessment of the scope and effectiveness of application a condition of introduction.
• Reform 3: Creation of a certification support system. Consulting and cost subsidies for obtaining FAA/EASA certification (on the order of tens of billions of yen in total). Support the international competitiveness of Japanese firms.
• Capability Building 1: Concentrated investment in priority technologies. Hydrogen aircraft ($1B/year × 5 years), UAVs/drones ($500M/year × 5 years), advanced semiconductors ($2B/year × 5 years, CHIPS Act-style).
• Capability Building 2: Deepening industry-academia-government collaboration. Launch long-term (10-year), large-scale ($100M+) joint research programs. Mandate personnel exchange.

These require investments of hundreds of billions to trillions of yen, but they lead to a fundamental strengthening of the industrial base.

Long-term (7–20 years): Mission-oriented transformation

In the long term, set bold missions and transform the industry structure.

• Mission 1: Commercial operation of hydrogen aircraft by 2040. Leveraging Japan's superiority in hydrogen technology, create the next-generation aircraft market. Total investment $10–20B.
• Mission 2: Regular transport to a lunar base by 2035. Participate in NASA's Artemis program and develop a crewed lunar lander and rover. Cooperate with private firms (ispace, etc.). Total investment $5–10B.
• Mission 3: A fully autonomous defense UAV system by 2030. Apply AI and robotics technology to defense. In light of the lessons of the Ukraine conflict, build a low-cost, mass-deployable system. Total investment $3–5B.

These take 10–20 years over the long term and require investments of trillions of yen, but they fundamentally improve international standing and industrial competitiveness.

7.2 In Closing: The Essence of Policy Is Institutional Design

What became clear through this analysis is that the essence of policy is institutional design. Superficial directives such as "diversify procurement" or "localize production" carry no efficacy. What matters is the detail of institutional design: (1) whose incentives to change and how, (2) which institutions must be changed for behavior to change, and (3) how to mitigate unintended consequences.

Just as ITAR regulation has powerfully governed supply chain structure, institutions are powerful. Just as India's offset policy directed the building of an industrial base, ingenious institutional design has strategic effects. Just as NASA's Commercial Crew Program cultivated SpaceX, demand-side policy creates markets. But at the same time, just as it produced the unintended consequence of a SpaceX quasi-monopoly, policy always entails risk.

Accordingly, what is required of policymakers are five qualities: (1) theoretical rigor (the integration of economics, network theory, and political economy), (2) empirical grounding (data-based causal inference), (3) attention to institutional detail (the design of incentive structures), (4) a long-term perspective (a 10–20-year commitment), and (5) flexibility and learning (responding to unintended consequences, iterative improvement of policy).

It is our hope that this report, going beyond superficial recommendations, demonstrates the importance and complexity of institutional design and becomes a starting point for deep discussion among policymakers, industry, and academia. The time to act is now. But that action should be grounded in thoughtful institutional design.

End of Report Date of Preparation: October 28, 2025 Total Character Count: approximately 48,000 characters References: cited in the main text Data Source: /ecsti2/llm/supply-chain/out/ (entities_deduplicated.csv, relationships_deduplicated.csv, critical_entities.csv, country_flow_matrix.csv)

A. The Mathematical Foundations of Scale-Free Network Theory

This appendix presents the mathematical foundations of the scale-free network theory invoked in Sections 1.2 and 2.1 of the main text, together with the caveats regarding its application to this analysis. The Barabási-Albert (BA) model generates a scale-free network through the following two mechanisms.

1. Growth: The network grows over time (nodes are added). 2. Preferential attachment: A new node connects with a probability proportional to the degree of existing nodes.

Formally, the probability \(\Pi(k_i)\) that a new node connects with existing node \(i\) is

$$\Pi(k_i) = \frac{k_i}{\sum_j k_j}$$

where \(k_i\) is the degree of node \(i\). Under this process, the degree distribution follows a power law:

$$P(k) \sim k^{-\gamma}$$

where \(\gamma \approx 3\) (in the case of the BA model).

Whether the network in this analysis fully follows a power law requires rigorous statistical testing, but the hub dominance (a maximum degree of 245 versus a mean of 3.52) suggests a power-law property.

B. Details of the Edler & Georghiou Policy Typology

The breakdown of the policy typology used in Section V of the main text (Theoretical Foundations of Policy Intervention) is as follows. Section 5.1 corresponds to the supply-side type, Section 5.2 to the demand-side type, and Section 5.3 to the systemic type.

Supply-side policies:

• R&D grants and loans
• R&D tax credits
• Public research institutions
• Innovation infrastructure
• Technology transfer support
• Education and training
• Intellectual property rights

Demand-side policies:

• Public procurement
• Regulation and standards
• Lead market initiatives

Systemic policies:

• Cluster policy
• Innovation networks
• Competence centers
• Regulatory reform

C. References (Selected)

1. Aoki, M. (2001). Toward a Comparative Institutional Analysis. MIT Press.
2. Arthur, W. B. (1989). "Competing Technologies, Increasing Returns, and Lock-In by Historical Events." Economic Journal, 99(394), 116-131.
3. Bain, J. S. (1956). Barriers to New Competition. Harvard University Press.
4. Barabási, A.-L., & Albert, R. (1999). "Emergence of Scaling in Random Networks." Science, 286(5439), 509-512.
5. Christensen, C. M. (1997). The Innovator's Dilemma. Harvard Business School Press.
6. Edler, J., & Georghiou, L. (2007). "Public Procurement and Innovation—Resurrecting the Demand Side." Research Policy, 36(7), 949-963.
7. Edler, J., et al. (2012). "Evaluating the Demand Side: New Challenges for Evaluation." Research Evaluation, 21(1), 33-47.
8. Edmondson, A. C. (2011). "Strategies for Learning from Failure." Harvard Business Review, 89(4), 48-55.
9. Hamilton, A. (1791). Report on Manufactures.
10. Krugman, P. (1984). "Import Protection as Export Promotion." In H. Kierzkowski (Ed.), Monopolistic Competition and International Trade. Oxford University Press.
11. Lazonick, W., & Mazzucato, M. (2013). "The Risk-Reward Nexus in the Innovation-Inequality Relationship." Industrial and Corporate Change, 22(4), 1093-1128.
12. List, F. (1841). The National System of Political Economy.
13. March, J. G. (1991). "Exploration and Exploitation in Organizational Learning." Organization Science, 2(1), 71-87.
14. Mazzucato, M. (2013). The Entrepreneurial State: Debunking Public vs. Private Sector Myths. Anthem Press.
15. Mazzucato, M. (2024). "Policy with a Purpose: Modern Industrial Policy Should Shape Markets, Not Just Fix Their Failures." IMF F&D Magazine, September 2024.
16. Olson, M. (1965). The Logic of Collective Action. Harvard University Press.
17. Pigou, A. C. (1920). The Economics of Welfare. Macmillan.
18. Teece, D. J. (1986). "Profiting from Technological Innovation." Research Policy, 15(6), 285-305.
19. Williamson, O. E. (1985). The Economic Institutions of Capitalism. Free Press.