Skip to content

Lesson M19.L03: Human Capital, Ideas, and the Knowledge Economy

Module: Economic Growth Part II Level: intermediate Duration: 30 minutes Learning Objective: Explain how non-rival knowledge generates increasing returns and sustained growth. Data as of: 2024 Provenance: Productivity Commission | OECD Main Science and Technology Indicators

Explanation

There is a crucial distinction between two kinds of productive inputs:

  • Rival goods can only be used by one person or firm at a time. A factory, a machine, a skilled worker โ€” all rival. If I am using the lathe, you cannot.
  • Non-rival goods can be used by any number of people simultaneously without diminishing their value. A mathematical formula, a software algorithm, a scientific discovery โ€” once created, they can spread at near-zero marginal cost.

Human capital (h) โ€” the skills, education, and health embodied in workers โ€” is rival: doubling the number of workers doubles the productive capacity, but each worker's knowledge is theirs alone.

Ideas and knowledge (A) are non-rival: once discovered, the same idea can be deployed across millions of firms. This creates increasing returns at the social level: doubling K, L, and A more than doubles output because A is used everywhere simultaneously.

Romer (1990) model. Output is produced by:

\[Y = K^\alpha (h \cdot A \cdot L_Y)^{1-\alpha}\]

where h = human capital per worker, A = stock of ideas, L_Y = workers in production. A separate R&D sector produces new ideas:

\[\dot{A} = \delta_A \cdot h \cdot L_A \cdot A\]

where L_A = researchers, ฮด_A = R&D productivity. The key feature is that existing ideas (A) make new ideas easier to find โ€” ideas beget ideas. This drives sustained growth without any exogenous technology assumption.

Long-run growth rate. In the balanced growth path, g_A (and hence g_Y/L) depends on population size, human capital, and R&D productivity โ€” all endogenous policy levers.

Knowledge spillovers and market failure. Because ideas are non-rival, private R&D generates positive externalities: a firm cannot capture all the value of its innovations. Result: the market underproduces ideas relative to the social optimum. This justifies R&D subsidies, patent systems, and public funding of basic research.

Australia: R&D spending is ~1.8% of GDP (2024) vs. OECD average ~2.7%. Australia's R&D Tax Incentive offsets 43.5 cents per dollar for small firms. University-industry collaboration (through ARC Linkage grants) aims to close the commercialisation gap. STEM education investment targets h directly.

Notation: K = capital; h = human capital per worker; A = stock of ideas/knowledge; L_Y = production workers; L_A = researchers; L = total labour force (L_Y + L_A); ฮด_A = R&D productivity; ฮฑ = capital share in production.

Worked Example

Question: In a simplified Romer-type economy, the idea production equation is:

\[\dot{A} = \delta_A \cdot h \cdot L_A \cdot A\]

Suppose ฮด_A = 0.002, h = 10 (years of tertiary education equivalent), L_A = 50,000 researchers, and current A = 1,000,000 ideas.

(a) Calculate the number of new ideas produced per period (ศฆ).

(b) Calculate the growth rate of ideas g_A = ศฆ/A.

(c) If human capital h rises to 12 (a 20% increase via education expansion), what happens to g_A?

Step (a):

\[\dot{A} = \delta_A \cdot h \cdot L_A \cdot A$$ $$\dot{A} = 0.002 \times 10 \times 50{,}000 \times 1{,}000{,}000$$ $$\dot{A} = 0.002 \times 10 \times 50{,}000{,}000{,}000\]

Let us simplify step-by-step: - 0.002 ร— 10 = 0.02 - 0.02 ร— 50,000 = 1,000 - 1,000 ร— 1,000,000 = 1,000,000,000 new ideas per period

Step (b):

\[g_A = \frac{\dot{A}}{A} = \frac{1{,}000{,}000{,}000}{1{,}000{,}000} = 1{,}000 = \delta_A \cdot h \cdot L_A\]

More usefully: g_A = ฮด_A ร— h ร— L_A (the A terms cancel). With L_A = 50,000, the growth rate per period is very large โ€” this is an artefact of stylised units. For cleaner pedagogical numbers, rescale to L_A = 500 researchers (same proportionality, smaller values):

\[g_A = 0.002 \times 10 \times 500 = \mathbf{10 = 1000\%}\]

Clearly in this illustrative example the numbers are stylised; the key is the proportionality: g_A = ฮด_A ร— h ร— L_A.

Step (c) โ€” Effect of rising h:

\[g_A^{new} = 0.002 \times 12 \times 500 = 12\]
\[\Delta g_A = 12 - 10 = 2 \quad \Rightarrow \quad \text{20\% rise in } h \text{ โ†’ 20\% rise in } g_A\]

This is a growth effect: more human capital permanently raises the rate of idea creation and hence the long-run output growth rate.

Common Misconception

Misconception: "Human capital and knowledge are the same thing, so expanding education always increases long-run growth proportionally."

Correction: Human capital (h) is rival โ€” it is embodied in individual workers and cannot be in two places at once. Knowledge/ideas (A) are non-rival โ€” once a discovery exists, all firms can use it simultaneously. Education increases h (rival), which enters the Romer model by increasing R&D productivity (ศฆ = ฮด_A ร— h ร— L_A ร— A). However, it is the non-rival nature of A that generates increasing returns and sustained growth, not human capital alone. If all new ideas were fully embodied in workers (rival), there would be no knowledge spillovers and no increasing returns at the aggregate level.

Practice Prompts

  1. Conceptual: Why does the non-rival nature of ideas cause markets to underproduce R&D?

โ†’ Answer: Non-rival ideas have the character of a public good: once an idea is created, others can use it (at near-zero marginal cost). A firm investing in R&D cannot prevent rival firms from eventually imitating or building on the discovery (especially as patents expire). Because the social return to innovation exceeds the private return, profit-maximising firms spend less on R&D than is socially optimal. This market failure justifies government R&D subsidies, patent extensions, and public funding of universities.

  1. Numerical: Suppose Australia doubles its researcher count from L_A = 100,000 to L_A = 200,000, with ฮด_A = 0.001 and h = 15. Calculate the growth rate of the idea stock before and after, using g_A = ฮด_A ร— h ร— L_A.

โ†’ Answer: - Before: g_A = 0.001 ร— 15 ร— 100,000 = 1.5 (i.e., 150% per period in this unit system, or equivalently the growth rate scales by 1.5) - After: g_A = 0.001 ร— 15 ร— 200,000 = 3.0 - Change: g_A doubles (from 1.5 to 3.0), consistent with a doubling of researchers.

This illustrates that in the Romer model, the scale of the research sector matters: larger economies (or those that invest more in researchers) grow faster โ€” the so-called "scale effect."

  1. Application: Australia spends approximately 1.8% of GDP on R&D versus the OECD average of 2.7%. Using the Romer framework, explain what policy levers could raise Australia's long-run growth rate, and what barriers might limit their effectiveness.

โ†’ Answer: The Romer model suggests three levers: (i) increase L_A โ€” more researchers, stronger STEM pipeline, immigration of skilled scientists; (ii) increase h โ€” higher-quality university education, vocational training; (iii) increase ฮด_A โ€” better R&D institutions, universityโ€“industry collaboration, open science policies. Barriers include: Australia's comparative advantage in low-R&D resource sectors (mining, agriculture) pulls talent and capital away from high-R&D sectors; "brain drain" to the US and UK; small domestic market limits returns to innovation; limited venture capital ecosystem. The R&D Tax Incentive and ARC grants address these but Australia still lags OECD peers.

Further Resources