Loading field map
Innovation and research and development (R&D) form the subfield of industrial organization that studies how new products, processes, and business methods come into being, how firms decide to invest in creating them, and how market structure, competition, and public policy shape those decisions. The field sits at the intersection of the economics of knowledge, the theory of the firm, and competition policy. Its central puzzle is a paradox: innovation is widely understood to be the main engine of long-run economic growth and firm competitiveness, yet the conditions that maximize innovation do not align neatly with the conditions that maximize static efficiency in ordinary markets. Understanding that tension—and what, if anything, governments and firms should do about it—is the field's enduring core.
The subfield asks a cluster of related questions. Why do some industries innovate rapidly while others stagnate? What determines how much a firm spends on R&D, and how does that spending translate into new products or cost reductions? Does competition encourage or discourage innovation? How do firms capture returns from investments that produce knowledge, which is often non-rival (usable by many at once) and only partially excludable (hard to keep secret)? What is the right balance between rewarding innovators with temporary monopoly power and ensuring that new knowledge spreads quickly through the economy? And how should antitrust authorities treat collaborations, mergers, and dominant firms when innovation is at stake?
The stakes are unusually high. In standard economic theory, competitive markets push prices down to cost and allocate resources efficiently. But if firms cannot expect to earn supra-normal profits from successful innovation, they will underinvest in it. The patent system, trade secrecy, and other intellectual property protections exist precisely to create temporary monopolies as incentives. Yet monopoly power, once granted, can be used to suppress later innovation or to charge high prices. The field therefore grapples with a fundamental trade-off between dynamic efficiency (the rate of improvement over time) and static efficiency (how well resources are used at a given moment). This trade-off is not merely academic: it informs patent length and breadth, merger guidelines, antitrust enforcement against dominant firms, and government funding of basic research.
The modern subfield emerged in the mid-twentieth century, but its intellectual roots reach back to earlier discussions of invention and monopoly. Joseph Schumpeter, writing in the 1930s and 1940s, provided the founding vision. He argued that capitalism's defining feature was not price competition but "creative destruction"—the process by which entrepreneurs introduce new combinations that render existing products, firms, and even industries obsolete. Schumpeter famously suggested that large firms with market power were the most important engines of innovation, because they could afford the fixed costs of R&D, absorb the risks of failure, and expect to capture enough of the returns to justify the investment. This claim—that monopoly may be good for innovation—directly challenged the classical presumption that competition is always best.
For several decades, Schumpeter's hypothesis remained more of a provocation than a research program. The formal economics of innovation took shape in the 1950s and 1960s, when economists began modeling R&D as an investment decision under uncertainty. Kenneth Arrow's 1962 article on "Economic Welfare and the Allocation of Resources for Invention" was a watershed. Arrow formalized the argument that a competitive firm has weaker incentives to innovate than a monopolist would, because the competitive firm can only earn profits on the new product itself, while a monopolist can also avoid losing its existing profits. But Arrow also showed that even a monopolist underinvests relative to the social optimum, because it cannot capture the full consumer surplus generated by the innovation. This established the basic welfare framework: private returns to innovation are systematically below social returns, so some form of intervention—patents, subsidies, or public research—is justified.
The 1970s and 1980s brought the field into the mainstream of industrial organization through game theory. Economists began modeling R&D races: several firms invest in research, and the first to succeed wins a patent or a market advantage. These models, associated with economists such as Partha Dasgupta, Joseph Stiglitz, and Jennifer Reinganum, produced a striking and counterintuitive result: more competition in the R&D race can lead to excessive duplication of effort, as firms race to be first and dissipate the potential profits through redundant spending. At the same time, too little competition can lead to complacency. The relationship between market structure and innovation turned out to be non-monotonic—an inverted U, with moderate competition producing the most innovation. This finding, though derived from highly stylized models, became a central reference point.
A parallel development was the emergence of evolutionary economics, associated most strongly with Richard Nelson and Sidney Winter. Their 1982 book, An Evolutionary Theory of Economic Change, rejected the optimizing, equilibrium framework of mainstream economics. Instead, they modeled firms as following routines—habitual ways of doing things—and innovation as a process of variation, selection, and retention analogous to biological evolution. Firms search for better routines, but search is local and uncertain; market competition selects among firms with different routines; and successful routines are imitated or diffused. This approach emphasized path dependence, bounded rationality, and the difficulty of predicting innovation outcomes. It remains influential in empirical work on industry dynamics and in policy discussions about national innovation systems.
The field today is organized less around a single dominant paradigm than around several complementary approaches that answer different questions. These approaches coexist and often combine, but they rest on different assumptions and methods.
The oldest and most central approach asks how the intensity of competition affects R&D spending and innovative output. This literature grew directly out of Schumpeter's hypothesis and Arrow's formalization. Early empirical work in the 1960s and 1970s, using cross-sectional data on firms and industries, produced mixed results: some studies found that large firms and concentrated industries did more R&D, others found no relationship or even a negative one. The problem was that market structure and innovation are jointly determined—innovative firms grow and become dominant, while dominant firms may or may not continue to innovate—so simple correlations are hard to interpret.
Modern work in this tradition uses more careful econometric methods, including panel data and natural experiments, and it has largely moved away from asking whether monopoly or competition is "better" in the abstract. Instead, it asks how specific features of the competitive environment—entry threats, the number of close rivals, the ease of imitation, the length of patent protection—affect innovation incentives. A robust finding is that the threat of entry by a potential competitor can be a powerful spur to innovation by incumbents, even when actual entry never occurs. Another robust finding is that the relationship between competition and innovation depends on the distance to the technological frontier: firms near the frontier may innovate more when competition intensifies, while firms far behind may innovate less, because they have little hope of catching up and instead focus on survival.
A second major approach focuses on the design of intellectual property rights. The patent system is the primary policy instrument for addressing the underinvestment problem Arrow identified. The literature asks how long patents should last, how broad they should be, what should be patentable, and how patent rights interact with other aspects of competition policy.
The classic trade-off is between the incentive effect of patents (they raise the expected return to R&D) and the deadweight loss of monopoly pricing during the patent term, plus the possible chilling effect on follow-on innovation. A key insight from this literature is that patents are not simply a reward for invention; they are also a bargaining tool and a component of firm strategy. Firms may patent not to practice the invention but to block rivals, to build negotiating leverage in cross-licensing deals, or to signal quality to investors. The rise of "patent thickets"—dense webs of overlapping patents that any new entrant must navigate—has led to concerns that the patent system may sometimes impede rather than promote innovation, particularly in complex, cumulative technologies like software and semiconductors.
A related strand examines trade secrecy, which protects innovations that firms keep confidential. Trade secrecy is cheaper and faster than patenting but offers weaker protection, since independent discovery or reverse engineering is allowed. The choice between patenting and secrecy is a central firm decision, and it depends on the nature of the technology, the strength of patent enforcement, and the ease of reverse engineering.
A third approach, which has grown rapidly since the 1990s, is the empirical study of innovation using detailed firm-level and product-level data. This literature does not start from a grand theory of competition and innovation but instead measures innovation directly and asks what determines it. Innovation is measured in several ways: R&D spending, patent counts, patent citations, new product introductions, productivity growth, and, more recently, text-based measures from product descriptions and job postings.
This approach has produced several important findings. First, innovation is highly concentrated: a small number of firms account for a large share of R&D spending and patents, and within firms, a small number of researchers produce most of the output. Second, innovation is cumulative and complementary: most innovations build on earlier ones, and the ability to innovate depends on absorptive capacity—the firm's existing knowledge base and its ability to recognize and use external knowledge. Third, the returns to R&D are highly skewed and uncertain: most R&D projects fail, and the distribution of returns is extremely right-skewed, with a few blockbuster innovations generating most of the value. Fourth, R&D spillovers are large and pervasive: firms benefit substantially from the R&D of other firms, both within their industry and across industries, and these spillovers are often larger than the private returns to the innovating firm.
This empirical literature has also documented the rise of "open innovation"—the practice of sourcing ideas and technologies from outside the firm through licensing, alliances, acquisitions, and crowdsourcing—and the growing importance of "intangible capital" more broadly, including software, data, organizational capital, and brand equity. These developments have blurred the boundary between R&D and other forms of investment and have made the measurement of innovation more difficult.
The evolutionary approach, while less central to mainstream industrial organization, remains influential in several areas. It emphasizes that innovation is a process of search and learning under uncertainty, not a matter of choosing the optimal level of R&D spending. Firms differ in their capabilities, and these differences persist over time because capabilities are built through experience and are hard to transfer. Innovation is therefore not just about incentives; it is about knowledge, skills, and organizational routines.
This approach has been particularly influential in the study of "national innovation systems"—the network of institutions, including universities, government laboratories, financial systems, and firms, that shape a country's innovative performance. It has also informed the study of technological trajectories and paradigms: the idea that innovation within a given technology follows a predictable path until a radical new technology disrupts it. The evolutionary approach is less formal than the game-theoretic literature, and it has been criticized for being descriptive rather than predictive, but it provides a valuable corrective to the assumption that firms are rational calculators with perfect knowledge of their options.
These approaches are not rival paradigms in the sense of making incompatible claims about the world. They ask different questions and use different methods, and they often complement each other. The market structure literature provides the basic framework for thinking about incentives; the patent literature examines the specific policy instruments that shape those incentives; the empirical literature tests the predictions and measures the magnitudes; and the evolutionary approach supplies a richer account of the process by which innovation actually occurs.
There are, however, genuine disagreements. The most important concerns the role of market power. The Schumpeterian tradition, and much of the early game-theoretic literature, suggested that some degree of market power is necessary for innovation, because firms need to expect supra-normal profits to justify risky R&D investments. The empirical literature has complicated this picture by showing that competition from entrants can be a powerful spur to innovation, and that dominant firms often become complacent. A related disagreement concerns the patent system: some economists argue that patents are essential incentives, while others argue that they are often unnecessary, that they raise prices, and that they impede cumulative innovation. These disagreements are not settled, and they reflect different weightings of the static efficiency losses from monopoly against the dynamic gains from innovation.
Another important relationship is between the formal models and the empirical work. The game-theoretic models of R&D races are highly stylized and make strong assumptions about the nature of the innovation process—for example, that the first firm to succeed wins the entire market, or that the probability of success depends only on current R&D spending. These models are useful for generating intuitions, but they are not directly testable in their original form. The empirical literature has therefore moved toward more flexible models that can be estimated with data, often using structural econometric methods that allow for heterogeneity across firms and industries.
The field today is characterized by several ongoing developments. One is the increasing importance of data and digital technology. Much modern innovation is in software, data analytics, and artificial intelligence, which have different properties from traditional manufacturing R&D: they are often cheaper to produce, easier to copy, and more dependent on complementary investments in data and computing infrastructure. The economics of these technologies is not fully captured by the traditional patent-and-R&D framework, and the field is actively adapting.
A second development is the growing concern about declining dynamism in advanced economies. Empirical work has documented a slowdown in the rate of new firm formation, a decline in labor productivity growth, and a rise in market concentration in many industries. Some economists interpret these trends as evidence that competition has weakened and that dominant firms are using their power to suppress innovation. Others argue that the trends reflect the natural evolution of industries with high fixed costs and network effects, and that innovation remains vigorous. This debate has made the subfield highly relevant to current antitrust policy, particularly regarding large technology firms.
A third development is the globalization of innovation. R&D is increasingly conducted across national borders, through multinational firms, international collaborations, and global supply chains. This raises questions about how intellectual property is protected across jurisdictions, how innovation policies in one country affect innovation in others, and whether the gains from innovation are shared equitably across countries and workers.
Finally, the field has become more policy-oriented. Governments around the world use a variety of instruments to promote innovation—R&D tax credits, direct subsidies, public research funding, patent reform, and competition policy—and the subfield provides the analytical basis for evaluating these instruments. The central message of the field remains what it has been since Arrow: private markets will systematically underinvest in innovation because innovators cannot capture the full social value of their discoveries, but the policy remedies are not straightforward, because the same interventions that increase innovation incentives can also create monopoly power, reduce competition, and slow the diffusion of knowledge. Getting this balance right is the enduring challenge of the subfield.