Endogenous Growth Theory

Endogenous Growth Theory is a macroeconomic framework within Business & Economics Theory that explains sustained long-run growth in per-capita Gross Domestic Product as an internal outcome of economic incentives, purposeful investment in research and development (R&D), human capital formation, and technological knowledge spillovers. Developed in the 1980s and early 1990s by Paul M. Romer (1986, 1990), Robert E. Lucas Jr. (1988), Sergio Rebelo (1991), and Philippe Aghion and Peter Howitt (1992), the theory overcame a central limitation of the neoclassical Solow–Swan Model, which attributed long-run per-capita growth to an unexplained exogenous rate of technological progress. By modeling how economic policies and market structures govern the creation of non-rival ideas, endogenous growth theory demonstrates why saving rates, R&D subsidies, intellectual property regimes, and education permanently alter an economy's long-run growth rate.

The microeconomic foundation of endogenous growth rests on the fundamental distinction between rival physical goods and non-rival ideas, articulated by Paul Romer (1990). Physical inputs such as machinery, silicon wafers, and skilled labor are purely rival: if a lathe or an engineer is employed on one assembly line, that exact physical unit cannot simultaneously be used in another factory. Consequently, replication of physical inputs (K and L) exhibits constant returns to scale:

Knowledge and technological blueprints (A)—such as the polymerase chain reaction, a sorting algorithm, or the design of a microprocessor—are non-rival: once discovered, a blueprint can be replicated across any number of factories simultaneously at near-zero marginal cost without depleting the blueprint itself. When knowledge A is included as a produced input alongside rival inputs K and L, the aggregate production function exhibits increasing returns to scale across all three inputs jointly:

By Euler's theorem on homogeneous functions, a production function with increasing returns to scale cannot be sustained under perfect competition where all inputs—including ideas—are paid their marginal products, because total factor payments would exceed total output. Endogenous growth models resolve this tension through two mechanisms: (1) external knowledge spillovers across competitive firms (Romer, 1986; Lucas, 1988), or (2) monopolistic competition in which innovators earn temporary patent rents that cover the fixed upfront sunk cost of R&D (Romer, 1990; Aghion & Howitt, 1992).

The simplest analytical class of endogenous growth models eliminates diminishing marginal returns to reproducible capital altogether, yielding the AK Model formalized by Frankel (1962) and Sergio Rebelo (1991). When aggregate capital K_t is interpreted broadly as a composite of physical capital and human capital that accumulate together, the aggregate production function is linear in K_t:

where A > 0 is a constant technology parameter measuring the marginal and average product of broad capital. With a constant saving rate s ∈ (0, 1), population growth n, and depreciation rate δ, the law of motion for per-capita capital k_t = K_t ÷ L_t is k̇_t = sAk_t − (n + δ)k_t. Dividing by k_t yields the steady-state growth rate of per-capita output (g_y):

Unlike the Solow–Swan model—where f′(k) → 0 as k → ∞—the marginal product of broad capital in the AK model remains permanently equal to A. Provided sA > n + δ, the economy sustains permanent positive per-capita growth without exogenous technological progress, and a permanent increase in the saving rate s (via Fiscal Policy or capital tax reform) permanently raises the long-run growth rate g_y.

In the Uzawa–Lucas Human Capital Model (Lucas, 1988), workers divide their non-leisure time between producing final goods (u ∈ (0, 1)) and accumulating human capital through education (1 − u). Human capital h_t accumulates according to ḣ_t = φ(1 − u)h_t, where φ > 0 measures educational productivity. Because existing human capital makes future learning more productive, human capital accumulation is linear in h_t, generating endogenous steady-state growth g = φ(1 − u^*) reinforced by external productivity spillovers across skilled coworkers.

Second-generation endogenous growth models explicitly model a dedicated research sector that allocates scientific labor (L_A) to invent new technologies.

Romer's (1990) Expanding-Variety Model and the Scale Effect

In Romer (1990), final output is produced using labor (L_Y) and a continuum of A_t specialized intermediate capital goods: Y_t = L_Y^1−α ∫₀^A_t x(i)^α di. In the research sector, researchers (L_A) draw upon the entire accumulated stock of public knowledge (A_t) to invent new blueprints at rate Ȧ_t = δL_AA_t, so the growth rate of technology is g_A = Ȧ_t ÷ A_t = δL_A. Because innovators do not capture the positive "standing-on-shoulders" spillover that their blueprint provides to future researchers, the decentralized market equilibrium allocates too few workers to R&D relative to the Pareto Efficiency social optimum.

Charles I. Jones (1995) noted that the linear specification g_A = δL_A implies a strong scale effect: since the number of scientists and engineers in advanced economies rose more than fivefold during the second half of the twentieth century, per-capita GDP growth rates should have accelerated fivefold, whereas empirical growth remained stable near 2% per year. Jones resolved this puzzle with the semi-endogenous knowledge production function:

where φ < 1 captures the "fishing-out" effect (easy discoveries are found first, making subsequent breakthroughs progressively harder). Setting g_A = Ȧ_t ÷ A_t constant and differentiating with respect to time yields the steady-state growth rate:

In the Jones model, public R&D subsidies and higher research shares permanently raise the level of income along the balanced growth path, while sustained growth requires an expanding global pool of researchers (n > 0) to offset the rising complexity of frontier discovery.

Schumpeterian Creative Destruction

Philippe Aghion and Peter Howitt (1992)—awarded the Nobel Memorial Prize in Economic Sciences—formalized Joseph Schumpeter's concept of creative destruction. In their quality-ladder framework, entrepreneurs invest in R&D to achieve vertical quality improvements of size γ > 1 over incumbent intermediate goods. Successful innovators displace the incumbent monopolist and capture patent rents until they are in turn displaced by the next innovator. Consequently, R&D faces two opposing welfare distortions: an intertemporal spillover effect (innovators underinvest because future generations build on their quality step forever) and a business-stealing effect (innovators overinvest from a social perspective because they do not internalize the destruction of the incumbent's rent).

Why does a competitive market generally underinvest in basic research and development?

Because scientific ideas are non-rival and only partially excludable, a firm that funds fundamental research cannot prevent competitors and future researchers from using the resulting knowledge to design subsequent products. Because the private return captured by the innovator is smaller than the social return across the entire economy (positive knowledge spillovers), decentralized market equilibrium results in R&D investment below the socially optimal level, justifying public science grants, R&D tax credits, and patent protection.

What is the "standing on shoulders" versus "stepping on toes" effect in R&D?

formulated by Charles Jones (1995), the "standing on shoulders" effect (φ > 0) refers to the positive intertemporal spillover whereby past discoveries (such as calculus, electricity, or the internet) make current researchers more productive. The "stepping on toes" effect (0 < λ < 1) refers to duplication and congestion when many researchers work simultaneously on the same scientific problem, causing a doubling of research labor L_A at a point in time to increase new discoveries by less than double (L_A^λ).

How does competition affect innovation in the Aghion–Howitt Schumpeterian framework?

Empirical and theoretical work by Aghion, Bloom, Blundell, Griffith, and Howitt (2005) demonstrates an inverted-U relationship between product-market competition and innovation. In neck-and-neck industries where firms operate at similar technological levels, stronger competition spurs R&D as firms innovate to "escape competition" and restore profit margins. In laggard sectors far behind the technological frontier, intense competition reduces post-innovation rents and discourages R&D investment (the Schumpeterian effect).

1. Romer, P. M. (1986). Increasing returns and long-run growth. Journal of Political Economy, 94(5), 1002–1037. https://doi.org/10.1086/261420
2. Lucas, R. E., Jr. (1988). On the mechanics of economic development. Journal of Monetary Economics, 22(1), 3–42. https://doi.org/10.1016/0304-3932(88)90168-790168-7)
3. Romer, P. M. (1990). Endogenous technological change. Journal of Political Economy, 98(5, Part 2), S71–S102. https://doi.org/10.1086/261725
4. Aghion, P., & Howitt, P. (1992). A model of growth through creative destruction. Econometrica, 60(2), 323–351. https://doi.org/10.2307/2951599
5. Jones, C. I. (1995). R&D-based models of economic growth. Journal of Political Economy, 103(4), 759–784. https://doi.org/10.1086/262002