Endogenous growth theory is a family of models in macroeconomics that explains sustained economic growth through mechanisms operating within the economy, rather than through technological progress assumed to arise independently of economic decisions. Its central mechanisms include investment in human capital, the creation of knowledge, and profit-motivated innovation. “Endogenous” means that these mechanisms are explained by the model itself. In many versions, incentives and policies can influence the long-run growth rate, not merely the level of output. (nobelprize.org)
Origins and the neoclassical benchmark
The principal benchmark is the Solow–Swan model, in which diminishing returns to physical capital prevent capital accumulation alone from sustaining growth in output per worker indefinitely. Continued long-run growth requires technological progress, whose rate is normally specified outside the model. A higher saving rate raises the long-run output level and produces transitional growth, but does not permanently increase the growth rate per worker when technological progress is unchanged. (nobelprize.org)
Modern endogenous growth research developed prominently during the 1980s and early 1990s. Paul Romer’s 1986 paper modeled growth associated with knowledge accumulation and increasing returns. Robert Lucas’s 1988 analysis emphasized schooling and learning through production. Romer’s 1990 model explicitly explained technological change through investment decisions by profit-maximizing agents. These contributions established distinct but related approaches rather than a single universal model. (journals.uchicago.edu)
Knowledge, spillovers, and incentives
A central distinction concerns nonrival goods. A machine cannot normally be used simultaneously by unlimited producers, whereas an idea or production design can be used repeatedly without being depleted. Knowledge can nevertheless be partly excludable: its use may be restricted through secrecy or intellectual property. Nonrivalry therefore does not imply that every idea is a freely accessible public good. (nber.org)
A knowledge spillover occurs when knowledge generated by one producer benefits others without full compensation. Such an externality separates the private return to creating knowledge from its social return. Spillovers can counteract diminishing returns at the aggregate level even when individual firms face conventional production constraints. Other models explain innovation through monopolistic competition, allowing innovators to earn profits that help finance the initial cost of developing new designs. (nobelprize.org)
Major model families
AK models. The simplest representation is the AK model, with production expressed as
[ Y=AK, ]
where (Y) is output, (K) is an accumulable capital stock, and (A) is a constant productivity parameter. With a fixed saving share (s), depreciation rate (\delta), and no population growth,
[ \dot K=sY-\delta K, \qquad g_Y=g_K=sA-\delta. ]
These equations illustrate how accumulation can sustain growth without an externally imposed increase in (A). The crucial assumption is the absence of diminishing returns to the relevant accumulable stock. Increasing returns or market imperfections are therefore not necessary features of every endogenous growth model. (nber.org)
Human-capital models. These explain growth through the accumulation of skills and productive capabilities. Individuals allocate time between production and education, while learning by doing builds specialized capabilities through experience. Lucas’s framework distinguishes an individual’s own productivity gains from external benefits associated with the surrounding stock of human capital. (doi.org)
Research-driven models. Firms devote resources to research and development because successful discoveries can generate profits. In Romer’s 1990 model, research produces new designs, expanding the variety of productive inputs. The allocation of skilled workers between research and goods production helps determine growth. Knowledge creation is thus connected to economic incentives rather than treated as an unexplained residual. (nber.org)
Schumpeterian models. Drawing on Joseph Schumpeter, these emphasize creative destruction: improved technologies replace older ones and erode incumbent producers’ profits. Philippe Aghion and Peter Howitt’s model, published in 1992, makes research-generated improvements the source of growth. Expected future innovation can discourage current research by shortening the period during which an invention earns returns. (nber.org)
Policy implications
Endogenous growth models provide frameworks for examining research subsidies, education funding, taxation, and patent protection. In Romer’s model, the market equilibrium allocates too little human capital to research. In creative-destruction models, however, innovation also imposes losses on existing producers; this “business-stealing” effect can offset beneficial spillovers. Consequently, decentralized innovation need not always be uniformly below the socially desirable level. (nber.org)
The implications also extend to trade. Access to larger markets and knowledge pools can alter innovation incentives; Romer’s model predicts growth benefits from integration into world markets. Effects of taxation likewise depend on how taxes change incentives to accumulate physical and human capital. These are model-dependent results, not unconditional predictions that every intervention permanently increases growth. (nber.org)
Empirical challenges and extensions
A major challenge concerns “scale effects”: some early research-driven models predict that a larger research workforce produces a permanently higher growth rate. Charles Jones’s 1995 studies found these predictions difficult to reconcile with industrialized economies’ time-series evidence, including comparatively stable growth despite expanding research effort. (www-leland.stanford.edu)
Semi-endogenous growth theory modifies the knowledge-production mechanism. Research still generates technological progress internally, but declining returns to discovery can make the long-run growth rate depend on exogenous parameters such as population growth. Policies may then change income levels and transitional growth without permanently changing the growth rate. This distinction separates explaining the production of technology from explaining every determinant of sustained growth endogenously. (www-leland.stanford.edu)