Understanding Endogenous Growth Theory: Key Concepts and Critiques

Search Fintrev

Search our guides, reviews and glossary.

Endogenous growth theory holds that an economy’s long-run growth rate is determined inside the economic system, by decisions about saving, education, research and policy, rather than by a pace of technological progress that arrives from outside. “Endogenous” means generated within the model. The theory explains why technology and know-how improve, and it treats that improvement as something people, firms and governments can speed up or slow down.

Exogenous vs. endogenous growth

The theory was built as a response to the neoclassical model developed by Robert Solow in the 1950s. In the Solow model, capital per worker is subject to diminishing returns: each additional machine adds less output than the one before. Saving more raises the level of income, but the economy eventually settles on a steady state. The only source of sustained growth in output per person is technological progress, which the model assumes grows at a fixed rate it does not explain.

That left a puzzle. If capital earns diminishing returns, poor countries with little capital should grow faster than rich ones and catch up. Many did not. Endogenous growth theorists in the 1980s argued that the missing piece was the economics of knowledge itself.

How the theory works

Three mechanisms appear across the main models.

Nonrival ideas. A design, formula or piece of software can be used by any number of people at once. Romer’s 1990 paper, “Endogenous Technological Change,” modeled ideas as nonrival but partly excludable: patents and secrecy let inventors earn a return, which gives firms a reason to pay for research.

Spillovers. Knowledge created by one firm leaks to others through hiring, imitation and published research. Romer’s 1986 paper, “Increasing Returns and Long-Run Growth,” used such spillovers to show that the economy as a whole can enjoy increasing returns even if individual firms do not.

Human capital. Robert Lucas’s 1988 paper, “On the Mechanics of Economic Development,” made accumulated skills a growth engine. Educated workers are more productive and also make the people around them more productive.

The simplest formal version is the AK model:

Y = A × K

Here Y is output, A is a constant productivity term, and K is a broad measure of capital that includes physical, human and knowledge capital. Because there are no diminishing returns to K, the growth rate of output is:

g = s × A − δ

where s is the saving (investment) rate and δ is the depreciation rate.

Worked example (hypothetical)

Assume A = 0.30, s = 0.20 and δ = 0.05.

• Growth = 0.20 × 0.30 − 0.05 = 0.01, or 1% per year.
• Raise the saving rate to 0.25: growth = 0.25 × 0.30 − 0.05 = 0.025, or 2.5% per year.

In the AK setting, the higher saving rate raises growth permanently. In the Solow model, the same change would lift income to a higher steady-state level, after which growth would return to the exogenous rate of technological progress.

Romer’s 1990 model adds an idea-production function. In a simplified form, the growth of the knowledge stock depends on the number of researchers: more people working on R&D means faster growth in ideas, and therefore in output per person.

How it is used

The theory gives economists a framework for asking which policies can change long-run growth in GDP per person. Its central tenets include:

• Competition encourages product and process innovation.
• Investment in infrastructure, education, health and telecommunications may yield increasing returns.
• Private R&D, rewarded by property rights and patents, is a key source of technological progress.
• Human capital is a primary input, not a side effect of growth.
• Entrepreneurship creates new firms, jobs and further innovation.

Knowledge-intensive sectors such as software and telecommunications fit the model well, since a single design can be copied at almost no cost. The theory also reframes the classic factors of production: knowledge becomes an input that can be produced and accumulated rather than taken as given.

Limitations and critiques

Hard to measure. Human capital, knowledge stocks and spillovers lack clean statistical counterparts. Critics argue that the theory rests on assumptions that cannot be measured accurately, which makes empirical validation difficult.

Scale effects. Early Romer-style models imply that a larger research workforce should produce permanently faster growth. Charles Jones pointed out in 1995 that the number of researchers in advanced economies rose substantially without a matching rise in growth rates. His response, the semi-endogenous growth model, keeps R&D as the engine but lets ideas become harder to find, so long-run growth ends up tied to population growth rather than policy alone.

Knife-edge assumptions. The AK result depends on returns to capital being exactly constant. Slightly diminishing returns bring back a Solow-style steady state; slightly increasing returns produce explosive growth.

Policy ambiguity. Patents encourage invention but restrict diffusion, and subsidies can be misdirected. The theory says policy can raise growth but leaves the right mix and scale of intervention open.

Distinctions readers confuse

• Endogenous variable (econometrics). In statistics, including CFA Level 2 material, an endogenous variable is one correlated with the error term in a regression or determined jointly within a system of equations. That technical meaning is unrelated to growth theory.
• Immiserizing growth. Jagdish Bhagwati’s concept describes export-led growth that worsens a country’s terms of trade enough to leave it worse off. It is a trade theory, not a growth-engine model.
• Exogenous growth. The Solow tradition, in which technology grows at an assumed, unexplained rate.

FAQs

The Bottom Line

Endogenous growth theory matters to economists, students and policymakers who want to know which choices can change long-run growth in output per person. It ties growth to research, human capital, competition and investment, which gives policy a real role. Measurement problems, scale effects and knife-edge assumptions mean the theory leaves the right mix and size of intervention open.