SCIENCE
Innovation Economics
Abstract. Innovation Economics investigates the generation, implementation, appropriation, diffusion, financing, distribution and social evaluation of economically valuable novelty. Its fundamental explanatory challenge arises because ideas can be reused, scientific discoveries have uncertain and dispersed applications, technological capabilities are cumulative, and institutions distribute both incentives and the returns to innovation. This monograph reconstructs the field from the early economics of entrepreneurial new combinations to research-based growth, evolutionary economics, national innovation systems, patent economics and contemporary causal evidence on public research, tax incentives and emerging technologies. It distinguishes invention, research, innovation and diffusion; specifies how official innovation measurement is constructed; develops elementary welfare and growth relationships as interpretive tools; and examines innovation across firms, sectors, regions and countries. The comparative synthesis identifies five recurring questions: who initiates inventive search, who finances the uncertainty, who appropriates benefits, which organizations and infrastructures permit use, and how private decisions combine into long-run social outcomes. Throughout, the treatment differentiates established models, empirical estimates, institutional descriptions and the author's integrative interpretations. It offers a comprehensive disciplinary synthesis without claiming a newly established model or historical priority.
Keywords: Innovation economics; research and development; invention; technological change; knowledge spillovers; endogenous growth; creative destruction; innovation systems; absorptive capacity; intellectual property; patents; diffusion; R&D tax incentives; venture finance; competition; productivity; industrial policy; general-purpose technologies; clean innovation; digital economy; innovation measurement
1. The Subject of Innovation Economics
Innovation Economics is the study of the economic processes through which novel knowledge, products, production methods and organizational practices become useful, viable and socially consequential. The discipline asks how scarce resources are allocated to uncertain searches, how information moves between organizations, how new combinations reshape competition, and which institutions determine whether discoveries produce widely shared economic value. Its unit of analysis can be a research project, laboratory, firm, collaborative network, university, sector, region or national economy. Its outcome variables range from product quality and cost reduction to employment, environmental performance, firm entry, productivity and welfare.
The field draws a decisive distinction between novelty and realized economic use. The Oslo Manual 2018 characterizes an innovation through a new or improved product made available to potential users or a business process brought into use by an enterprise; novelty to the firm suffices for some statistical purposes. An invention may be technically ingenious without reaching implementation. Research generates knowledge, but innovation can involve organizational recombination, practical learning and adoption from others without formal laboratory expenditure. Diffusion extends the effects of an original innovation across users and places.
The economics of innovation must therefore explain both a supply of possibilities and a demand for workable improvements. The invention of a technique does not guarantee local adoption, while adoption may deliver substantial welfare gains even in the absence of a local discovery. The research boundary with Technological Change is maintained by giving priority here to incentives, organizational choices, knowledge institutions and innovation governance; the consequences of technology remain present where needed to evaluate those choices.
2. Fundamental Definitions and Classifications
A product innovation modifies the good or service delivered; a business-process innovation modifies the way activities are performed. Changes may be technological, organizational, logistical or commercial. The fourth edition of the Oslo measurement framework simplifies earlier classifications into these two principal categories and instructs statisticians to distinguish activities from outcomes, novelty from economic significance and firms with innovation activity from firms with implemented innovations. Survey reports are thus observations generated by a measurement convention, rather than a universal inventory of all valuable novelty.
Research and experimental development, or R&D, has a narrower definition. The Frascati Manual 2015 requires activities to be novel, creative, uncertain, systematic and transferable or reproducible. Basic research seeks new understanding without a particular immediate application; applied research addresses practical objectives; experimental development systematically uses knowledge to create or improve products and processes. A marketing redesign, routine software maintenance or installation of a known machine can constitute business innovation yet fail the Frascati R&D criterion.
These definitions establish a measurement hierarchy: R&D expenditure is an input; scientific publications and patents are partial intermediate outputs; implemented innovations are events; productivity, resilience and welfare are possible outcomes. The hierarchy has no fixed one-to-one correspondence. A large research budget can fail, a breakthrough can be difficult to patent, and a useful process innovation can be learned from a supplier. Empirical investigation has to specify which level it identifies before interpreting the observed association.
3. Historical Origins: Division of Labor, Entrepreneurship and Development
Innovation appeared in economic reasoning long before a separate innovation-economics field formed. Classical discussions of specialization and mechanization connected changes in knowledge and tools to the organization of production. The Industrial Revolution supplied a historical setting in which machines, standardized components, transport infrastructure and industrial training transformed feasible production and exchange. Subsequent economic history revealed that invention, the making of equipment, financing, managerial experimentation and the construction of markets rarely proceeded in the same sequence or at the same speed.
Joseph Schumpeter gave the modern field a distinctive interpretation. In The Theory of Economic Development, first published in German in 1911 and translated into English in 1934, he treated development as the carrying out of new combinations and assigned entrepreneurs a coordinating function. Innovation could introduce goods, methods, markets, supplies or industrial organizations; credit could mobilize purchasing power for activities that departed from inherited routines. Entrepreneurial profit was linked to temporary advantages generated by such change.
Schumpeter's historical contribution joins three analytic levels: an economic system governed by established routines, the actor who departs from them, and changes in competitive structure induced by the departure. This legacy informs later creative-destruction research, although the subsequent literature does not share one universal account of entrepreneurs, monopoly or optimal innovation. The endurance of the historical insight lies in identifying economic change as an organized process that must overcome the inertia of existing production arrangements.
4. Basic Research, Information and Market Failure
The economic properties of knowledge complicate the conventional market exchange of goods. An idea can often be used repeatedly at low marginal replication cost, while its discovery initially requires uncertain labor and capital. Knowledge can spill beyond the investor who funded its discovery; a scientific result can benefit unknown industries or future generations. Yet knowledge can also be protected by secrecy, data access, complementary know-how or legal rights. Its appropriability varies enormously by field.
Richard Nelson's economics of basic scientific research describes why benefits from general scientific understanding may accrue too diffusely for private firms to finance at the socially desirable level. Kenneth Arrow's allocation of resources for invention analyzes uncertainty and the information paradox confronting transactions in inventions: a purchaser may need to learn what knowledge is useful, even though disclosure changes the seller's bargaining position.
For a prospective project, the private expected value is the probability-weighted stream of appropriable returns less its costs. Social value includes benefits to consumers, imitators, workers and future research users, as well as external costs. A gap between private and social returns may justify collective institutions. Yet the gap can run in either direction for particular innovations: patent races, duplicative research, congestion, rent destruction and negative externalities can produce excessive private search. The direction of corrective policy must follow the identified mechanism rather than a generic assumption that all additional R&D is beneficial.
5. The Formal Economics of Knowledge and Research
A simple nonrivalry representation writes the feasible use of an idea A by multiple firms as A available in each production process without proportionally depleting its stock. Unlike a physical machine, additional uses need not reduce the knowledge available to an existing user. The restriction is economic: exclusion rights, complementary equipment, skilled labor and local infrastructure can prevent access. Romer's endogenous technological change formalizes nonrival, partially excludable ideas and the importance of research decisions under imperfect competition.
Consider an illustrative stock-of-knowledge identity: A(t+1) = (1 − δ)A(t) + I(t), where I denotes newly usable knowledge and δ the obsolescence or loss of effective usability. This is an accounting representation chosen for exposition, not an estimate or a newly claimed theory. In some settings knowledge does not physically depreciate; instead practical relevance and access can decay. A stylized project-decision condition is E[appropriable net payoff] ≥ opportunity cost. Social evaluation replaces appropriable payoff with full incremental social surplus, including spillovers, learning, displacement and external costs.
The distance between these inequalities explains innovation policy and firm strategy more clearly than the proposition that science mechanically generates growth. The empirical problem is to recover the relevant counterfactuals, probabilities and incidence of benefits. Observable expenditures identify neither expected marginal social value nor the quality-adjusted stock of usable knowledge. Even estimates of average return cannot automatically determine the optimal subsidy for an additional project at a different frontier.
6. Endogenous Growth and Creative Destruction
The neoclassical growth tradition treats technological progress as a major determinant of long-run output per worker while typically imposing its aggregate evolution from outside the model. Endogenous-growth approaches move knowledge generation inside the decision system. In Romer's research-based economy, intentional investment develops new designs, nonrival ideas support cumulative productivity and human capital influences research capacity. Market equilibrium reflects the partially excludable character of innovations and the fixed costs of discovery.
Aghion and Howitt's creative-destruction growth model introduces quality-improving innovations that render earlier technologies less profitable. Incumbents can rationally fear that present discoveries will be displaced by tomorrow's research. Market incentives incorporate appropriability, technological spillovers and business-stealing effects. The model therefore allows actual research intensity to deviate from the welfare-maximizing level in more than one direction. Competitive innovation can be vigorous while some projects convey little net social gain; basic research can be neglected despite high future spillovers.
The principal theoretical distinction concerns the origin of growth. An accumulation model emphasizes the production and reuse of ideas; a quality-ladder model emphasizes replacement and rent destruction. Both require assumptions about the research technology, population, market power and diffusion. They produce conditional comparative statics rather than invariant quantitative laws. Empirical policy analysis should ask whether the industry features expanding design varieties, quality improvements, network effects, learning from foreign technologies or a combination of mechanisms.
7. Evolutionary Economics and Technological Trajectories
Innovation frequently proceeds through experiments whose outcomes, future markets and engineering properties cannot be fully known in advance. Evolutionary economic approaches study routines, selection, learning and cumulative paths of capability development. Giovanni Dosi's technological paradigms and trajectories describes sustained improvement within a prevailing problem-solving direction and discontinuities when a new technological paradigm appears. Search is guided by organizational knowledge and technical constraints, so the entire space of possible discoveries is rarely surveyed.
The classic overview by Kline and Rosenberg further challenges the image of research moving in one uninterrupted line through development to market. Engineers discover problems in production, users reveal needs, markets change during testing, and feedback travels among design, operations and scientific inquiry. The research laboratory may be indispensable at one stage and less central at another. Commercial adoption involves complementary services, training, standards and maintenance.
Path dependence is an important implication of these historical and organizational accounts. A successful standard creates installed infrastructure and expertise that may favor incremental improvement while raising the cost of transition. A new paradigm can initially look inferior on familiar performance metrics yet open different possibilities. Evaluation must follow the relevant time horizon and competing feasible alternatives. Economic selection is neither a guarantee of technically best solutions nor an assurance that every locked-in convention is inefficient; switching costs and interdependence are real components of welfare.
8. National and Regional Innovation Systems
Innovation takes place inside connected institutions: enterprises, universities, public laboratories, schools, investors, regulators, purchasers, professional communities and standards bodies. The interactive-learning perspective associated with Lundvall emphasizes user-producer relationships, cumulative competence and the organization of learning. Nelson's comparative national innovation systems documents substantial differences among the institutions supporting technological innovation in 17 countries. Similar R&D spending levels can coexist with different educational systems, organizational forms, finance regimes and innovation outcomes.
A system is an analytical description of linked institutions rather than a guarantee of coordinated performance. Knowledge flows can be blocked by poor incentives, incompatible regulations, geographic inequality, weak purchasing capacity or disconnected research agendas. Firms also import ideas, equipment and expertise, so national boundaries can conceal transnational networks. Effective comparison distinguishes mechanisms that travel across jurisdictions from institutional packages whose outcomes depend on local history.
Regional innovation systems focus on the geography of specialized labor, proximity to universities, supply chains and dense professional networks. Agglomeration can increase the circulation of tacit information and ease matching between firms and skills. Concentration can simultaneously raise rents, housing costs and entry barriers. Comparing places requires attention to selection: innovative people and firms may choose productive locations, complicating causal attribution to policy-designed clusters. The policy question is which missing link is actually constrained, and whether public intervention changes the counterfactual location or merely subsidizes an activity that would have occurred elsewhere.
9. Absorptive Capacity and Organizational Learning
A stock of published information becomes valuable only when organizations can understand and implement it. Cohen and Levinthal's absorptive-capacity analysis identifies the capability to recognize the value of external information, assimilate it and apply it commercially. Prior experience, skilled employees and internal R&D contribute to that capability. A company can rationally invest in internal scientific competence even when it intends to license or learn much of its technology from outsiders.
Absorption has several costs: translation of information into product requirements, adjustment of processes, training, integration with legacy systems and changing the incentives of employees. The appropriability of external knowledge cannot be assessed independently of these complementary expenses. An adoption subsidy that lowers the price of equipment may have weak effects if firms lack engineers, reliable power, cybersecurity capacity or managerial incentives to reorganize.
The measurement problem is subtle. Researchers often use R&D intensity, skill composition, patents or collaboration indicators as proxies for absorptive capacity, but the proxies are influenced by firm quality, product markets and management. Bloom and Van Reenen's management-practices evidence shows strong associations between management organization and firm performance across surveyed countries; it does not establish that every management intervention causally generates equivalent innovation gains. Organizational learning belongs inside innovation analysis precisely because the same external idea can yield different outcomes across firms.
10. Users, Suppliers and Sources of Novel Products
Innovation search begins from many locations. Producers observe engineering constraints and margins; users experience unsatisfied needs; suppliers contribute materials and components; public organizations specify mission requirements; academics supply fundamental theories and measurement. Von Hippel's lead-user research argues that users with needs that precede the general market may reveal valuable information about emerging products. This source of novelty is especially relevant where established market surveys sample buyers who have never encountered the new use case.
User innovation does not imply that large firms or formal R&D become irrelevant. Producers may be best placed to scale a design, obtain regulatory approval, guarantee quality, manage manufacturing defects and create distribution. Design benefits can therefore be distributed between originators of an idea and owners of complementary assets. In industrial procurement, early customers may finance testing but can also specify a narrow architecture that inhibits later diffusion. The organization of a supplier-customer relationship changes the technology that is worth searching for.
A disciplined account of innovation sources asks four practical questions: who holds the relevant problem information, who can perform experiments, who controls complementary resources and who can bear implementation risk. These questions help explain the joint occurrence of closed research laboratories, supplier partnerships, public consortia, open-source ecosystems and lead-user communities. Each arrangement solves part of the discovery and coordination problem under different terms of ownership and exchange.
11. The Economics of Appropriability and Complementary Assets
Innovators seldom capture all the value generated by their work. Patents, secrecy, lead time, trademarks, scale and learning curves can protect different margins of advantage. Teece's analysis of profiting from technological innovation shows that innovators may lose profits to imitators or to the owners of complementary manufacturing, distribution and service capabilities. The firm with the original discovery need not be the organization best positioned to deliver a reliable product at scale.
Appropriability should be measured relative to an innovation's stage. Secrecy may protect a manufacturing recipe while offering little defense for an easily reverse-engineered interface. A patent may secure a technical claim yet leave the holder without capital for clinical trials or market access. Open publication can accelerate ecosystem development while exposing the originator to competitors. Licensing can spread a useful invention and reward the owner, but negotiation and anti-commons problems can arise with overlapping rights.
This framework links industrial organization to innovation economics. A firm's boundaries depend partly on which complementary assets can be contracted and which require integration. The welfare evaluation must then separate profits captured from net new value produced. A legal monopoly can redistribute rents without changing total technological capability; conversely, a temporary proprietary return can induce risky work that society would otherwise forgo. The optimal arrangement depends on imitation costs, sequential improvements, bargaining frictions and the technologies of use.
12. Patents, Scientific Disclosure and Cumulative Invention
Patents can trade a period of exclusive rights for disclosure of an invention; actual scope, validity, exclusions and duration vary across jurisdictions. A patent system seeks to maintain incentives for expensive experimentation while permitting subsequent learning. Scotchmer's cumulative-research problem exposes the difficulty of rewarding both foundational inventors and developers of later improvements. Strong early rights may encourage pioneering research yet raise the costs of follow-on use; narrow rights can expand access while reducing pioneer incentives.
A patent is an observed legal event, not a direct measure of social value. Applications and grants respond to filing costs, law, industry composition, strategic portfolios and enforcement expectations. WIPO's World Intellectual Property Indicators 2025 summarizes administrative patent, trademark and design activity, including observations from 2024. Such series permit comparisons in filing activity, but they require adjustments for origin, office procedures, families, field and time before they can support claims about inventive productivity.
The literature on software patents illustrates how cumulative digital development complicates appropriation and patent quality. Innovation policy must also consider trade secrecy, prizes, public procurement, copyright, standards and open licensing, because different knowledge structures require different institutions. The scientific question is how each institutional arrangement changes marginal invention, disclosure, entry, diffusion and litigation costs.
13. Patents as Economic Indicators
Patent counts record particular steps in a system of legal protection. They exclude unpatented process know-how, service innovations, organizational change and many discoveries intentionally placed in the public domain. Differences in the propensity to patent can dominate differences in invention rates. Patent citations may indicate technological links and sometimes market significance, but examiner practices, technological classes and aging generate measurement distortions.
In Market Value and Patent Citations, Hall, Jaffe and Trajtenberg relate patent and citation metrics to the market value of firms in their historical sample. Their results support carefully adjusted use of citations as indicators; they do not convert every citation into an invariant monetary value. A patent family spans legal filings concerning related inventions, and the time needed to observe subsequent citations creates truncation for recent cohorts.
The same principles apply to scientific publications and venture funding. Publication counts reflect incentives, language, database coverage and research traditions; funding reflects financing conditions as well as project prospects. A country can increase measured innovative activity by encouraging filings or by changing tax definitions, while measured productivity remains unchanged. A robust innovation dashboard triangulates R&D, implementation surveys, IP data, productivity, skills, trade and diffusion, stating both denominators and observation windows.
14. Research Productivity and R&D Capital
The empirical productivity approach treats accumulated research spending as an imperfect stock of knowledge capital, often with a lag and an assumed depreciation rate. Griliches's R&D productivity assessment explains why researchers must grapple with the definition of the R&D stock, output price indexes, correlated regressors, spillovers and simultaneity. Profitable firms may spend more on research because they are successful; firms may also become successful because research improves their capabilities. This two-way connection is central to identification.
One illustrative stock accounting method is K(t) = (1 − δ)K(t−1) + R(t), where R denotes inflation-adjusted research expenditure and δ is an assumed depreciation rate. If K(t−1) = 100, R(t) = 20 and δ = 0.10, the arithmetic yields K(t) = 110. This numerical example verifies the mechanics of the accounting convention; it does not establish any economy's actual R&D stock or depreciation rate. Alternative initialization rules or price deflators change the estimate.
Research productivity is similarly sensitive to what constitutes output. Bloom, Jones, Van Reenen and Webb's ideas-getting-harder evidence studies rising researcher effort and falling measured research productivity in several domains. The results challenge simple assumptions of a constant aggregate ideas-production efficiency, while leaving room for differences across fields, spillover horizons and measurement conventions. It is therefore unwarranted to infer a single globally declining rate of useful creativity from one research-output indicator.
15. Knowledge Spillovers and Social Returns
A spillover occurs when one firm's research changes another actor's opportunities without a full market payment for the effect. The effect can be positive through shared knowledge, skills and suppliers, or negative through product-market displacement. Jaffe's technological-opportunity study used firms' positions in technological space to explore both kinds of interaction. Later empirical work developed separate measures of technological relatedness and market rivalry to avoid confusing competitive losses with knowledge benefits.
Bloom, Schankerman and Van Reenen's technology-spillover estimates find that positive spillovers dominate in their U.S. firm setting, with estimated gross social returns to R&D at least twice private returns. The result is specific to the sample, instruments, variables and model; it is evidence for a large underinternalized benefit in that setting, not a universal multiplier to apply to every grant application. Jones and Williams's social-return calculation provides a complementary growth-theoretic argument suggesting substantial research underinvestment in their calibration.
The direction of a marginal public intervention is a separate question. Projects can displace private spending, compete for scarce scientists, duplicate work, intensify harmful technology or create international benefits unavailable to domestic taxpayers. Spillovers establish the need to examine social incidence rather than assume that the investor's private financial return is sufficient. They also make open science, mobility and collaborative standards relevant instruments of innovation policy alongside direct R&D subsidies.
16. Finance, Uncertainty and the Research Funding Gap
Research expenditures are especially difficult to finance because experimental results may fail, knowledge is often intangible collateral, investors lack the developer's technical information and a project may have a long interval before revenue. The capital structure of an innovative firm therefore matters. Internal cash flow, specialized equity, strategic partnerships, customer prepayments, bank loans, venture capital and grants allocate different claims and risks. Hall and Lerner's financing of R&D survey finds strong evidence of elevated financing costs for many small and young firms, while evidence for large established firms is mixed.
Equity investors can diversify across unsuccessful projects and participate in large gains from successful commercialization. However, venture investors concentrate on ventures with scalable markets and plausible exits. Public science, infrastructure, clinical research, rural technologies and slow-adopting environmental applications may have valuable external benefits without fitting that financial template. Debt is better suited to projects with pledgeable assets and predictable cash flows; it can be burdensome when knowledge has little liquidation value.
The financing gap is conceptually distinct from the knowledge-spillover gap. Even a fully appropriable project can be underfunded because outsiders cannot ascertain its quality. Conversely, an easily financed innovation can have low social returns after environmental or displacement effects. Practical interventions must diagnose which friction binds: credit constraints, missing technical capability, spillovers, demand uncertainty or weak property rights. Financing instruments also change who owns the eventual results, which can affect disclosure, distribution and the direction of future search.
17. Venture Capital, Grants and Entrepreneurial Selection
The innovation value of a financing instrument depends on its selection process and effect on subsequent choices. Venture capital can bring governance, technical networks and access to follow-on funding as well as capital. Yet fund returns and venture success are highly skewed, and observed venture-backed growth combines selection with treatment. A successful portfolio is evidence of project choice and incentives together; it cannot isolate the causal contribution of the financier without a credible counterfactual.
Sabrina Howell's evidence from R&D grants uses ranked applicants to a U.S. Department of Energy small-business program. Near the funding threshold, an early-stage award increased subsequent venture investment and business innovation outcomes. The study highlights the value of financing prototypes under severe uncertainty. Its estimate belongs to the applicant population and public program investigated; it does not establish that every grant program doubles financing probabilities.
Policy reviews such as Lerner's study of public venture initiatives emphasize governance, local capacity and the danger of attempting to manufacture investment ecosystems through isolated subsidies. Programs can select visible applicants that would already attract funding, distort project choice or distribute benefits to connected incumbents. A robust design specifies additionality, timing, follow-on commercialization and regional spillovers. Long time horizons and transparent rules can matter as much as the nominal monetary budget.
18. R&D Tax Incentives and Direct Subsidies
Tax credits reduce the private price of eligible research, while grants select particular projects or applicants. Their economic effects vary because firms differ in taxable income, constraints, administrative access and opportunities to relabel ordinary spending. Bloom, Griffith and Van Reenen's international R&D tax panel finds that lower user costs were associated with higher R&D intensity across nine OECD countries from 1979 to 1997, with stronger long-run effects than short-run responses.
Dechezleprêtre and colleagues' research-tax threshold design exploits a change in U.K. eligibility cutoffs to identify increased R&D and patenting among affected firms. This design is informative near the eligibility threshold, and its estimated response should not be extrapolated without adjustment to large multinationals, very different tax systems or substantially different research sectors. Eligibility thresholds may also induce strategic firm behavior, an issue the study explicitly considers.
A tax incentive offers broad decentralized selection but may support projects with low external value; a grant can target public priorities but is vulnerable to information and governance errors. Procurement creates demand for outcomes rather than paying only for inputs. Policy comparison should include administrative costs, additional private spending, spillovers, the public cost per genuinely additional project, and long-run diffusion. The scientifically relevant performance ratio is not merely subsidized R&D divided by foregone tax revenue; it is incremental social benefit compared with a credible no-policy counterfactual.
19. Public Science, Universities and Mission-Oriented Research
Public research institutions sustain knowledge with long and uncertain horizons. Universities educate skilled workers, publish cumulative methods, create instrumentation and train scientists who move through labor markets. Government laboratories and mission agencies finance expensive facilities, data infrastructure and high-risk exploratory work. Such activities can generate public goods whose uses are not known to the original funders. Nelson's basic-research argument makes this a central case for collective investment.
The relationship between public science and business invention can be investigated rather than presumed. Azoulay and collaborators' NIH funding-rule study analyzes how publicly supported biomedical research influences later private patenting. Its identification design aims to separate project selection from funding effects. Evidence of downstream private activity indicates a channel through which scientific knowledge enters commercial development; it does not imply that patents exhaust the public health or scholarly value of the research.
Public procurement, research missions and challenge prizes add other mechanisms. A mission can coordinate engineering standards, early demand and complementary infrastructure; it may also impose narrow specifications or become vulnerable to administrative lock-in. Evaluation should state whether a mission seeks broad knowledge, a measurable technological capability, a public service outcome or domestic industrial capacity. These aims are related yet distinct, and each requires different evidence of success.
20. Innovation Diffusion and Adoption Dynamics
The first discovery of a technology and its widespread economic influence are separated by adoption decisions. Users compare expected gains with equipment costs, required training, switching costs, uncertain reliability and compatibility. Network effects and experience can accelerate use after an initial threshold. The fitted adoption curve often resembles an S-shape, but that shape alone does not reveal the causal mechanism: heterogeneous profitability, informational contagion and delayed supply can generate superficially similar trajectories.
Griliches's hybrid-corn diffusion research showed how market density, profitability and region-specific supply influenced the spread of agricultural technology. Mansfield's industrial imitation study extended the analysis to rates at which firms copied innovations in several industries. Both studies displaced the idea that availability alone produces immediate universal use. A technology may be invented globally yet require localized complementary innovations before it can produce value.
Comin and Hobijn's cross-country diffusion study examines 15 technologies in 166 countries over two centuries. Their reported average adoption lag of 45 years conceals substantial differences across countries, eras and technologies. Earlier and later waves should therefore not be compared without respecting changes in communication, capital markets and infrastructure. Diffusion analysis asks when a technology was available, what qualifies as adoption, how intensively it was used and whether benefits reached marginalized populations.
21. Learning Curves, Complementarities and Implementation Costs
Technologies alter feasible production only after actors learn how to operate them. Technical knowledge may reside in manuals, but operational knowledge often emerges from maintenance, repeated tasks, failure analysis and local adjustment. Such learning can make later adopters more productive than early adopters, or give early adopters advantages that are hard to reverse. Innovation is consequently embedded in work practices and supply chains as well as equipment and code.
The organizational-capital evidence of Brynjolfsson, Hitt and Yang helps explain why measured returns to digital equipment depend on complementary intangible investments. A costly software installation may initially disorganize a firm while staff are trained and processes redesigned. Static cost-benefit calculations can therefore miss transition losses, organizational capital and delayed improvements.
The complementarity problem affects aggregate statistics. If a technology becomes cheaper but its effective use requires scarce professionals, expenditure on hardware can rise before observable output does. If quality improvements are inadequately captured by price indexes, recorded output can understate benefits; if labor displacement is ignored, aggregate productivity can conceal distributional costs. A credible diffusion study should pair adoption dates with intensity, complementary investments, exposure to disruptions and counterfactual productivity. It should not infer success from installation counts alone.
22. Competition, Concentration and Innovation Incentives
Competition can encourage firms to escape rivals through better products and lower costs, while anticipated appropriable rents can finance expensive exploration. The relative importance of these channels depends on imitation speed, sunk costs, entry barriers, product substitution and how far firms stand from the technological frontier. Incumbents may defend existing product lines, purchase emerging rivals or invest heavily in research. Entrants can disrupt markets yet also depend on incumbent platforms, distribution and manufacturing.
The creative-destruction framework formalizes the replacement of innovation rents and makes clear that incentives are intertemporal. A rival's future success can reduce today's investment value, while prospects of a breakthrough can stimulate entry. Akcigit and Ates's business-dynamism analysis interprets rising U.S. concentration and weaker dynamism through a model in which declining knowledge diffusion is important. The study offers a calibrated mechanism and supporting evidence rather than a one-factor explanation of every sector.
Policy instruments include merger review, contestable access to essential inputs, interoperability, patent examination, market entry rules and R&D incentives. Their effects on discovery should be evaluated jointly with quality, prices, labor conditions and consumer choices. Concentration by itself is not a complete index of innovative performance. A concentrated market can support expensive frontier research, but exclusionary practices can prevent complementors from using and improving a platform. The central empirical task is to distinguish scale efficiencies from mechanisms that suppress rival experimentation.
23. Industrial Organization and Innovation Strategy
The innovation decisions of firms are strategic. A company may choose incremental improvement, a radical technical approach, acquisition, licensing, standards participation or secrecy based on expected rival behavior. Its value proposition is created within supply chains, after-sales support and regulation. Strategic complementarity means that two investments can be valuable only when undertaken together; incompatibility can cause technologically capable projects to fail commercially.
Teece's complementary-assets account helps explain when inventors benefit from vertical integration and when they can license discoveries to established producers. In another direction, the lead-user evidence suggests that source information may originate outside a company's formal research department. Firms choose governance arrangements in response to where information resides and how contractible it is.
Industrial ecosystems also shape the direction of innovation. A dominant technical interface can create a large developer market, while privately imposed changes to an interface can disrupt investments by complementors. Open standards can expand interoperability but face collective-action problems in financing maintenance. Public policy must evaluate arrangements by their effects on entry, experimentation, cumulative improvements and the final cost to users. The desirable configuration depends on sector-specific transaction costs rather than a general presumption for either openness or integration.
24. International Diffusion, Trade and Catch-up
Countries can raise productivity by inventing at the world frontier or by adapting and deploying technologies pioneered elsewhere. The second channel can be economically dominant when local production methods are far from best practice. However, importing equipment is only the first stage of transfer. Learning requires engineering skills, spare parts, standards, reliable infrastructure, organizational competence and access to markets. Technology can be formally available and practically inaccessible.
The comparative national-system studies show how schooling, industrial institutions and research networks condition technological learning. The long-run diffusion evidence documents persistent international variation in the adoption of specific technologies. These findings support attention to implementation capacity and to the incentives faced by domestic firms and public agencies. International licensing, foreign direct investment, migration, research collaborations and trade in intermediate goods can each transfer knowledge, subject to their own restrictions and distributional consequences.
Catch-up policy raises difficult questions of sequencing. A government may prioritize broad digital connectivity and technical education before expensive frontier laboratories; in other sectors targeted research centers are necessary to absorb specialized knowledge. Infrastructure and institutions are complementary, so expenditures cannot be evaluated as isolated inputs. Cross-country comparisons must also control for differences in income, population, resource endowments and the composition of industry. A high patent-filing country and a high adoption-productivity country may exemplify different forms of innovative capability.
25. Innovation, Structural Transformation and Economic Development
Innovation changes the relative productivity and viability of sectors. Improvements in agriculture can release labor for manufacturing and services; logistics and information technology can create new service markets; medical innovations can extend productive life and change household spending. The transitions depend on wages, capital, demand, training and international trade. Modern technology can also increase productivity within traditional sectors without replacing them.
The central development question is whether a new capability generates broad employment, domestic supplier learning and public revenue, or remains an isolated enclave. A capital-intensive laboratory complex may create valuable science while offering few immediate jobs; a low-cost organizational improvement in food distribution may raise welfare with minimal patent activity. Both are innovation, but their social effects require different measures. Development policy thus combines technical capacity with institutions that diffuse benefits throughout the productive structure.
A persistent difficulty is identifying innovation's causal effect independently of the institutions that foster it. Fast-growing regions attract researchers and venture capital; stronger education systems may generate both patents and more productive firms. Simple cross-country correlations between innovation indexes and income levels combine reverse causation, selection and joint determinants. The correct conclusion is comparative: productive transformation requires the discovery and adaptation of workable methods together with the capacity to implement them, while the precise intervention depends on local constraints.
26. Innovation and the Distribution of Income
Innovation can raise aggregate income while redistributing work opportunities, bargaining power, profits and consumer benefits. Process improvements may lower prices and expand demand, yet displace tasks from particular occupations or locations. Product innovations can create new industries and employment but demand different qualifications. Intellectual property and platform positions can concentrate returns, while open diffusion can spread productivity gains. Aggregate gains therefore do not settle the question of compensation or inclusion.
Acemoglu and Restrepo's robot-exposure study finds negative employment and wage effects in more exposed U.S. commuting zones over its historical sample. These are local labor-market effects of industrial robots under a particular empirical strategy, not proof that every innovation reduces total employment or that future automation will reproduce the same magnitude. General equilibrium responses may include new tasks, lower prices, increased demand and movement between industries.
The spatial distribution of adjustment costs matters. Workers may face retraining expenses, limited housing mobility and a mismatch between the location of job losses and new opportunities. Policy can support adaptive skills, relocation, portable benefits and competition in labor and product markets. The analytical objective is to evaluate the gains and losses attributable to each feasible innovation policy, including effects on people whose welfare is poorly measured by patent and venture-financing statistics.
27. Environmental Innovation and Directed Research
Climate and environmental innovation involves two interacting market failures. Polluting activity can impose damages on third parties, while researchers may capture only part of the social gains from cleaner technologies. The two-market-failures analysis explains why environmental regulation and innovation support may be complementary. A research subsidy alone can fail to create a market for low-emission products if pollution remains underpriced; an emission price can induce adoption but may leave beneficial upstream knowledge underfunded.
Acemoglu and colleagues' directed-technical-change model studies how research can be redirected toward cleaner inputs under environmental constraints. Its results depend on substitution opportunities, research dynamics and policy timing. A well-designed transition compares expected emission benefits, supply-chain effects, economic costs and the contribution to subsequent global knowledge. Public procurement and technical standards can create early markets, while international diffusion can multiply benefits beyond the initial subsidizing country.
Clean innovation also produces distributional and geopolitical questions. Energy systems use physical infrastructure with long asset lifetimes; technological choice can affect regional employment, material imports, land use and household energy bills. Measuring patent growth in a green class is therefore only one indicator. A full welfare analysis traces actual deployment, lifecycle performance, rebound behavior and the distribution of costs across time and geography. The direction of innovation is an economic choice shaped by institutions and prices.
28. Health and Biomedical Innovation
Biomedical innovation combines uncertain research with unusually demanding validation, safety and public-health responsibilities. A therapeutic target, laboratory prototype, clinical trial, manufacturing process and distribution network are economically and scientifically distinct achievements. Strong incentives for commercial discovery can coexist with underinvestment in prevention, neglected diseases and research whose benefits are difficult to appropriate. Patent rights can finance trials and still create affordability concerns at the point of care.
The public biomedical research evidence shows a channel from publicly funded university research to later private patenting. This connection helps explain why basic science and applied commercialization should be treated as complementary components of a wider system. It does not measure all clinical benefits, nor does it prove that any specific patent improves health outcomes. Additional clinical evidence, safety regulation, real-world effectiveness and access conditions are necessary for evaluation.
Public policy can finance upstream science, clinical infrastructure, product procurement or manufacturing readiness. Each intervention changes different risks: scientific feasibility, regulatory evidence, scale, demand, or affordability. In fields with substantial public funding and concentrated intellectual property, the terms of licensing and procurement can influence who ultimately benefits. Innovation economics supplies the incentive and resource-allocation analysis; health economics and biomedical sciences supply essential outcome measurement and clinical standards.
29. Digital Innovation and General-Purpose Technologies
Digital technologies often lower the cost of storing, processing, copying and transmitting information, thereby changing both innovative production and the spread of its results. Goldfarb and Tucker's digital-economics synthesis organizes economic effects around changing search, replication, transportation, tracking and verification costs. These cost reductions can generate new products, business processes and business models. At the same time, data control, platform access, cybersecurity and interoperability reshape the distribution of returns.
General-purpose technologies have broad applications and require continuing complementary innovation. Electricity, digital computation and contemporary machine learning are useful examples of broad technological families, but their realized impacts depend on organizational adoption, complementary skills and institutional constraints. The intangible-asset account shows why investments in computers cannot be separated analytically from organizational capabilities. Apparent delays between technical breakthroughs and recorded productivity can arise from these adjustment costs and from measurement problems.
The new frontier of generative systems creates questions about automated search, research complementarity, reproducibility, intellectual property, data access, energy costs and the replacement or augmentation of professional tasks. At the November 2026 research cutoff, the long-run quantitative effects of particular foundation models remain an open empirical question requiring realized adoption and causal outcome data. The field should analyze each application through production tasks, incentives, diffusion and welfare, avoiding the assumption that technical benchmarks directly translate into aggregate economic gains.
30. Innovation Measurement and Composite Rankings
Measurement begins by specifying which construct is observed. The Oslo Manual's firm survey definitions capture implementation and activity, while the Frascati R&D definitions classify systematic creative work. WIPO's Global Innovation Index 2026 compares 139 economies through a composite set of indicators; its 2026 ranking places Switzerland, Sweden and the United States first, second and third. Ranking is a property of the stated index, indicator coverage and weights, not a physical measurement of a country's total innovative value.
A transparent dashboard reports multiple dimensions rather than only one score. Expenditure intensity should be associated with a clear denominator, such as domestic R&D as a share of GDP; patent statistics require families, field adjustments and filing dates; implementation surveys require representative samples and harmonized question wording; productivity requires price and quality measurement. The denominator for an innovation rate matters: the share of firms reporting a product innovation can rise because the number and composition of firms change.
Composite indexes support benchmarking but can obscure trade-offs between research inputs, institutions and realized outcomes. Country ranks may change when coverage changes even if underlying performance is constant. The more useful analytical question is whether a jurisdiction's research capacity, diffusion infrastructure and market conditions improve on consistent measures over time. Where the available data are descriptive, the manuscript should preserve that description and refrain from causal claims about institutional superiority.
31. Empirical Identification and Evaluation Design
Estimating the causal impact of innovation faces simultaneity, selection, measurement error, delayed outcomes and spillovers to untreated actors. Firms with strong prospects may spend more on R&D; government grants may select better researchers; patent productivity may be driven by regulatory changes. Panel fixed effects absorb stable differences but cannot alone remove changing unobserved confounders. Instrumental variables require both relevance and an exclusion restriction. Regression discontinuity designs identify local effects near a threshold under continuity assumptions.
The empirical contrast is visible across the literature. Firm-spillover estimation constructs technological and product-market distances, while R&D tax threshold research exploits an administrative eligibility rule. Grant-ranking comparisons use near-cutoff differences in finance, and robot-exposure analysis relies on variation across commuting zones. Their treatment populations, time horizons and outcomes cannot be pooled into a single universal innovation elasticity.
Useful evaluation distinguishes additionality in R&D input, technological output, diffusion, private returns and social welfare. For a subsidy, one should measure both extra spending and substitution away from unsubsidized work. For procurement, actual service performance and supplier spillovers matter. For a science institution, long citation lags, worker mobility and downstream industries complicate attribution. Pre-analysis plans, robustness checks, independent replication and publication of null results improve confidence. Bibliographic existence is evidence that research was conducted; it is not equivalent to verification of its underlying dataset.
32. Comparative Analytical Synthesis
The established literatures reveal a recurring economic sequence. A knowledge opportunity must become recognizable; a capable organization must finance uncertain exploration; research results must be assembled with complementary assets; users must adopt the resulting improvement; and a larger economy must absorb its benefits and adjustment costs. The sequence can repeat, overlap and reverse, as interactive innovation research has shown. Each stage involves different information, institutions and measures of success.
The first interpretive conclusion concerns bottlenecks. An economy with many patents may be limited by diffusion; a system with efficient adopters may be limited by frontier research; a technically strong laboratory may be limited by financing for clinical or industrial scale; a highly funded enterprise may be limited by missing demand. Therefore a policy instrument chosen solely because its label contains innovation is analytically incomplete. The binding constraint must be identified at the margin and situated within the chain of use.
The second conclusion concerns incidence. Private profits, social spillovers and distributional gains do not coincide automatically. A policy can efficiently stimulate research but concentrate its rewards; a widely shared productivity improvement can be created by a small private firm yet benefit consumers primarily; a technically impressive innovation can impose environmental or labor-adjustment costs. The correct welfare account therefore includes who bears risk, who receives returns and how benefits move through networks and places.
The third conclusion concerns time. New ideas, research projects, patents, products, infrastructure and productivity belong to different observation clocks. The long-run adoption evidence and the research-productivity findings together caution against reading current research expenditure directly as near-term output growth. They do not establish one universal innovation delay. Evaluations must specify discovery, implementation, diffusion and spillover horizons before drawing policy conclusions.
The fourth conclusion concerns institutional variety. The division of knowledge among users, firms, universities and states makes learning relational. Patents, grants, scientific publication, procurement, standardization and competition rules are components of a policy portfolio whose effectiveness depends on the sector and institutional capacity. The authorial contribution of this monograph is this explicit comparative synthesis of existing evidence across incentives, capabilities, diffusion and welfare. It is an analytical interpretation of established research, without naming or claiming a new theory, doctrine or conceptual system.
33. Open Scientific Problems and Research Agenda
Several major questions remain difficult because innovation crosses organizational boundaries and long periods. Measuring the value of nonpatented organizational innovation remains a challenge. Distinguishing genuinely new knowledge from the rearrangement of existing knowledge is complicated by digital reproducibility and cross-firm learning. Estimating long-horizon public R&D returns requires tracking spillovers through researchers, industries and national borders. The welfare effect of intellectual property depends on follow-on invention and enforcement costs that are often incompletely observed.
Digital and machine-learning applications introduce further empirical questions: how automated scientific assistance changes research productivity, whether access to data and compute concentrates discovery, how reliably productivity gains diffuse to small firms, and how labor-market transitions evolve. Environmental innovation presents a related challenge in coordinating the direction of research with deployment incentives and verifying the actual environmental performance of technologies. Cross-border knowledge flows and security restrictions create new questions about fragmented research networks.
The discipline advances through improved linked microdata, comparable longitudinal innovation surveys, causal evaluations, historical case research and transparent replication. The next research dossiers for this Science could independently examine innovation measurement, appropriation and patent design, the social return to R&D, diffusion and development, public innovation finance and directed clean innovation. Each dossier requires a separate passport and original verified corpus, while remaining connected to this single canonical Science. Institutional and quantitative uncertainty should be documented as part of the result rather than hidden behind aggregate rankings.
Conclusion
Innovation Economics studies how economically valuable novelty is generated, organized, financed, implemented, appropriated, distributed and reproduced through learning. Its modern identity connects Schumpeter's entrepreneurial new combinations, Arrow's and Nelson's information problems, theories of endogenous growth and creative destruction, evolutionary trajectories, institutional systems, and econometric evidence on research incentives and diffusion. The field explains why the existence of knowledge does not guarantee its use, why private research returns can diverge from social benefits, why patent activity is a partial measure of innovation, and why policy must consider both inventive search and the capacity to absorb results.
Three conclusions govern the scientific interpretation. First, invention, R&D, innovation and diffusion are empirically distinct processes, each requiring appropriate definitions and outcome measures. Second, research incentives operate inside organizations and institutions whose complementary capabilities affect the direction, scale and social incidence of technological change. Third, the effectiveness of public and private innovation strategies depends on identifying binding constraints, causal counterfactuals and the beneficiaries of change across time and geography. These conclusions summarize the established evidence and the monograph's comparative analysis. They provide a foundation for later, separately authorized research dossiers without asserting a new economic doctrine.
References
- OECD and Eurostat. 2018. Oslo Manual 2018: Guidelines for Collecting, Reporting and Using Data on Innovation. 4th ed. Paris: OECD Publishing; Luxembourg: Eurostat. https://doi.org/10.1787/9789264304604-en.
- OECD. 2015. Frascati Manual 2015: Guidelines for Collecting and Reporting Data on Research and Experimental Development. Paris: OECD Publishing. https://doi.org/10.1787/9789264239012-en.
- World Intellectual Property Organization. 2026. Global Innovation Index 2026: Powering Entrepreneurs at the Frontier of Science. Geneva: WIPO. Accessed November 16, 2026.
- World Intellectual Property Organization. 2025. World Intellectual Property Indicators 2025. Geneva: WIPO.
- Schumpeter, Joseph A. 1934. The Theory of Economic Development: An Inquiry into Profits, Capital, Credit, Interest, and the Business Cycle. Translated by Redvers Opie. Cambridge, MA: Harvard University Press. Original German edition 1911.
- Arrow, Kenneth J. 1962. “Economic Welfare and the Allocation of Resources for Invention.” In The Rate and Direction of Inventive Activity: Economic and Social Factors, 609–626. Princeton, NJ: Princeton University Press for NBER.
- Nelson, Richard R. 1959. “The Simple Economics of Basic Scientific Research.” Journal of Political Economy 67 (3): 297–306. https://doi.org/10.1086/258177.
- Griliches, Zvi. 1957. “Hybrid Corn: An Exploration in the Economics of Technological Change.” Econometrica 25 (4): 501–522. https://doi.org/10.2307/1905380.
- Mansfield, Edwin. 1961. “Technical Change and the Rate of Imitation.” Econometrica 29 (4): 741–766. https://doi.org/10.2307/1911817.
- Dosi, Giovanni. 1982. “Technological Paradigms and Technological Trajectories: A Suggested Interpretation of the Determinants and Directions of Technical Change.” Research Policy 11 (3): 147–162. https://doi.org/10.1016/0048-7333(82)90016-6.
- Kline, Stephen J., and Nathan Rosenberg. 1986. “An Overview of Innovation.” In The Positive Sum Strategy: Harnessing Technology for Economic Growth, edited by Ralph Landau and Nathan Rosenberg, 275–305. Washington, DC: National Academy Press.
- Lundvall, Bengt-Åke, ed. 1992. National Systems of Innovation: Towards a Theory of Innovation and Interactive Learning. London: Pinter Publishers.
- Nelson, Richard R., ed. 1993. National Innovation Systems: A Comparative Analysis. New York: Oxford University Press. https://doi.org/10.1093/oso/9780195076165.001.0001.
- Romer, Paul M. 1990. “Endogenous Technological Change.” Journal of Political Economy 98 (5, Part 2): S71–S102. https://doi.org/10.1086/261725.
- Aghion, Philippe, and Peter Howitt. 1992. “A Model of Growth Through Creative Destruction.” Econometrica 60 (2): 323–351. Working-paper and bibliographic record: https://www.nber.org/papers/w3223.
- Cohen, Wesley M., and Daniel A. Levinthal. 1990. “Absorptive Capacity: A New Perspective on Learning and Innovation.” Administrative Science Quarterly 35 (1): 128–152.
- Teece, David J. 1986. “Profiting from Technological Innovation: Implications for Integration, Collaboration, Licensing and Public Policy.” Research Policy 15 (6): 285–305. https://doi.org/10.1016/0048-7333(86)90027-2.
- von Hippel, Eric. 1986. “Lead Users: A Source of Novel Product Concepts.” Management Science 32 (7): 791–805. https://doi.org/10.1287/mnsc.32.7.791.
- Griliches, Zvi. 1979. “Issues in Assessing the Contribution of Research and Development to Productivity Growth.” Bell Journal of Economics 10 (1): 92–116. Bibliographic record hosted by NASA NTRS.
- Jaffe, Adam B. 1986. “Technological Opportunity and Spillovers of R&D: Evidence from Firms’ Patents, Profits and Market Value.” American Economic Review 76 (5): 984–999. NBER Working Paper 1815.
- Scotchmer, Suzanne. 1991. “Standing on the Shoulders of Giants: Cumulative Research and the Patent Law.” Journal of Economic Perspectives 5 (1): 29–41. https://doi.org/10.1257/jep.5.1.29.
- Jones, Charles I., and John C. Williams. 1998. “Measuring the Social Return to R&D.” Quarterly Journal of Economics 113 (4): 1119–1135. https://doi.org/10.1162/003355398555856.
- Hall, Bronwyn H., and Josh Lerner. 2010. “The Financing of R&D and Innovation.” In Handbook of the Economics of Innovation, Vol. 1, 609–639. Amsterdam: Elsevier. https://doi.org/10.1016/S0169-7218(10)01014-2.
- Bloom, Nicholas, Rachel Griffith, and John Van Reenen. 2002. “Do R&D Tax Credits Work? Evidence from a Panel of Countries 1979–1997.” Journal of Public Economics 85 (1): 1–31. https://doi.org/10.1016/S0047-2727(01)00086-X.
- Bloom, Nicholas, Mark Schankerman, and John Van Reenen. 2013. “Identifying Technology Spillovers and Product Market Rivalry.” Econometrica 81 (4): 1347–1393. https://doi.org/10.3982/ECTA9466.
- Comin, Diego, and Bart Hobijn. 2010. “An Exploration of Technology Diffusion.” American Economic Review 100 (5): 2031–2059. https://doi.org/10.1257/aer.100.5.2031.
- Bloom, Nicholas, Charles I. Jones, John Van Reenen, and Michael Webb. 2020. “Are Ideas Getting Harder to Find?” American Economic Review 110 (4): 1104–1144. https://doi.org/10.1257/aer.20180338.
- Hall, Bronwyn H., Adam Jaffe, and Manuel Trajtenberg. 2005. “Market Value and Patent Citations.” RAND Journal of Economics 36 (1): 16–38.
- Howell, Sabrina T. 2017. “Financing Innovation: Evidence from R&D Grants.” American Economic Review 107 (4): 1136–1164. https://doi.org/10.1257/aer.20150808.
- Azoulay, Pierre, Joshua S. Graff Zivin, Danielle Li, and Bhaven N. Sampat. 2019. “Public R&D Investments and Private-sector Patenting: Evidence from NIH Funding Rules.” Review of Economic Studies 86 (1): 117–152. https://doi.org/10.1093/restud/rdy034.
- Dechezleprêtre, Antoine, Elias Einiö, Ralf Martin, Kieu-Trang Nguyen, and John Van Reenen. 2016. “Do Tax Incentives for Research Increase Firm Innovation? An RD Design for R&D.” NBER Working Paper 22405. https://doi.org/10.3386/w22405.
- Acemoglu, Daron, Philippe Aghion, Leonardo Bursztyn, and David Hemous. 2012. “The Environment and Directed Technical Change.” American Economic Review 102 (1): 131–166. https://doi.org/10.1257/aer.102.1.131.
- Jaffe, Adam B., Richard G. Newell, and Robert N. Stavins. 2005. “A Tale of Two Market Failures: Technology and Environmental Policy.” Ecological Economics 54 (2–3): 164–174. https://doi.org/10.1016/j.ecolecon.2004.12.027.
- Acemoglu, Daron, and Pascual Restrepo. 2020. “Robots and Jobs: Evidence from US Labor Markets.” Journal of Political Economy 128 (6): 2188–2244. https://doi.org/10.1086/705716.
- Goldfarb, Avi, and Catherine Tucker. 2019. “Digital Economics.” Journal of Economic Literature 57 (1): 3–43. https://doi.org/10.1257/jel.20171452.
- Akcigit, Ufuk, and Sina T. Ates. 2023. “What Happened to US Business Dynamism?” Journal of Political Economy 131 (8): 2059–2124. https://doi.org/10.1086/724289.
- Bloom, Nicholas, and John Van Reenen. 2007. “Measuring and Explaining Management Practices Across Firms and Countries.” Quarterly Journal of Economics 122 (4): 1351–1408. https://doi.org/10.1162/qjec.2007.122.4.1351.
- Bessen, James, and Robert M. Hunt. 2007. “An Empirical Look at Software Patents.” Journal of Economics & Management Strategy 16 (1): 157–189. https://doi.org/10.1111/j.1530-9134.2007.00136.x.
- Lerner, Josh. 2009. Boulevard of Broken Dreams: Why Public Efforts to Boost Entrepreneurship and Venture Capital Have Failed—and What to Do about It. Princeton, NJ: Princeton University Press.
- Brynjolfsson, Erik, Lorin M. Hitt, and Shinkyu Yang. 2002. “Intangible Assets: Computers and Organizational Capital.” Brookings Papers on Economic Activity: Microeconomics 2002 (1): 137–198.