The Mankiw-Romer-Weil Model: Human Capital and Cross-Country Income

The Mankiw-Romer-Weil (MRW) model extends the Solow growth framework by treating human capital as a second reproducible factor of production, alongside physical capital, allowing it to explain much more of the variation in income across countries and reduce the “residual” that Solow’s original model left unexplained. Published in 1992, the model showed that once you account for schooling and skill development, poor and rich nations don’t need different production functions — the disparities follow naturally from different investment rates in physical and human capital.

The Solow model’s limitation

The original Solow growth model, published in 1956, treated capital (factories, machines, infrastructure) and labor as the two inputs to production. Given a savings rate, a population growth rate, and a depreciation rate, the model predicted steady-state income per capita.

But Solow ran into a famous puzzle: when researchers estimated how much of countries’ income growth should come from rising capital and labor, only a small fraction did. The bulk was left to “technological progress” — a residual that often accounted for 70% or more of growth. This was embarrassing. The model couldn’t explain most of what it set out to explain.

Mankiw, Romer, and Weil’s insight was simple: the missing piece wasn’t magic technology. It was human capital — the knowledge and skills workers accumulated through schooling and experience. By treating human capital as a third, accumulating factor (like physical capital), they could dramatically shrink that residual and explain income differences without invoking unexplained technology gaps.

Human capital as a factor of production

In the MRW framework, output depends on three factors: physical capital (K), human capital (H), and raw labor (L).

Output = A × K^a × H^b × L^(1-a-b)

Here, A is the technology level (held constant across countries), and a and b are the capital-share parameters. The key difference from Solow: H (human capital) is not fixed. It grows when people invest time in schooling and training.

Human capital accumulates much like physical capital: you invest by spending years in school instead of working, or by taking training courses. That investment yields a higher skill level, which persists and compounds. A person with a university degree earns more throughout their career than one with only high school — that income premium reflects the human capital invested.

Explaining the income gap

Solow explained that poor countries and rich countries converge to different steady states if they have different savings rates. But a savings rate of 15% cannot explain why one country earns $50,000 per capita and another earns $5,000.

MRW added a second dimension: countries differ not just in their rate of physical capital investment, but in their rate of human capital investment (measured roughly by school enrollment and years of education). Some countries send most young people to university; others send only a few. This difference in education investment is as important as capital investment in determining long-run income.

Once you measure education levels across countries and plug them into the MRW model, the predicted income gaps match reality far better. The model predicts that a country investing heavily in both physical and human capital will be rich; one investing in neither will be poor. The poor country isn’t cursed by inferior technology — it simply hasn’t invested in the skills needed to use capital productively.

The human capital accumulation equation

MRW models human capital growth as:

H = (h_0) × (years of school)

where h_0 is some baseline level and years of school is the average years of education in the population. Countries with higher school enrollment (as a fraction of the working-age population) accumulate human capital faster.

This is an oversimplification — quality of schools matters, not just years attended — but it makes the model tractable and focuses attention on education investment as central to growth, not peripheral.

Steady-state income and convergence

Like Solow, MRW predicts that countries converge toward a steady state in which income per capita is determined by: (1) the savings rate for physical capital, (2) the schooling rate (investment rate in human capital), (3) the population growth rate, and (4) the rate of technological progress (exogenous).

The key difference: two countries with the same physical-capital savings rate but different schooling rates will end up at very different steady-state incomes. A country that saves 20% of income for factories but sends only 5% of youth to secondary school will be poorer than one that saves 15% for factories but sends 40% of youth to school.

This aligns with empirical reality: some East Asian countries achieved high growth by prioritizing education even when physical capital was scarce (South Korea, Taiwan), while some oil-rich nations, despite capital wealth, remained lower-income because education investment lagged.

The residual: smaller but not zero

MRW’s empirical tests (using OECD data from the 1960s–1980s) showed that including human capital reduced the “Solow residual” substantially. Instead of technology explaining 70% of growth variation, it fell to 30–40%.

This is a major improvement. Yet a residual remains. Some of it reflects true differences in technology and productivity. Some reflects measurement error (schooling data is imperfect, quality varies). Some reflects institutions, governance, and other factors not in the model.

Applications and extensions

The MRW framework opened the door to “endogenous growth” theory, where long-run growth rates themselves depend on investment and policy choices, not just exogenous technology. Romer’s own work on knowledge accumulation and Lucas’s work on human capital became influential extensions.

Development economists use MRW-style reasoning to argue for education and skill investment in poor countries: it’s not altruism, it’s productive capital formation. A dollar spent on schooling may yield returns comparable to, or higher than, a dollar spent on roads or factories.

Limitations and critiques

The model oversimplifies. Education quality and type matter; years in school is a crude proxy. Returns to schooling vary by field and context. The model assumes human capital is interchangeable across countries, ignoring institutional and organizational differences.

Recent work emphasizes that how human capital is used depends on institutional quality, market openness, and corporate governance. Two countries with the same education level may see very different productivity if one has functioning capital markets and the other doesn’t.

Additionally, MRW treats technology as exogenous (falling equally on all countries), which sidesteps questions of why some countries innovate and others don’t. Modern growth theory tries to endogenize technology itself.

See also