Lewis Dual-Sector Model

A 1954 paper that won a Nobel 25 years later

Sir W. Arthur Lewis, a Saint Lucia-born economist working at the University of Manchester, published "Economic Development with Unlimited Supplies of Labour" in The Manchester School in May 1954. The paper opened a question that mainstream postwar economics had been studiously ignoring: how does a country with masses of rural underemployment industrialise? Standard one-sector neoclassical labor market models offered no insight — they assumed full employment and continuous wage adjustment, both clearly wrong for the colonial and post-colonial economies Lewis had spent his career studying. His answer reframed the problem in two-sector terms, and the framework was simple enough to teach, formal enough to extend, and predictive enough to match observed dynamics. It became the founding document of modern development economics. Lewis shared the 1979 Nobel Memorial Prize in Economic Sciences with Theodore Schultz — the first Black recipient of any Nobel in a category other than Peace — explicitly cited for "pioneering research into economic development research with particular consideration of the problems of developing countries."

The two sectors

Lewis's first move is to refuse the one-sector picture. A poor country, he insists, has two distinct economies coexisting inside the same political border. The traditional sector is the village. Most of the labor force lives there, working family land or in petty craft production. It is organised by custom and kinship, not by markets; people are paid in food and shelter rather than wages; and output is split among all members of the household irrespective of who contributed how much. The modern sector is the colonial-era plantation, the port, the railway, the early factory. It is small, urban or enclave, and runs on capitalist principles: wage employment, profit motive, and capital accumulation. Crucially the two sectors do not share a labor market. Wages are set independently in each, and labor moves between them only by migration, not by daily price adjustment.

The two sectors differ on every economic dimension a textbook tracks:

Dimension	Traditional / rural	Modern / capitalist
Organising principle	Kinship, custom	Profit, wage contract
Labor payment	Average product (output shared)	Marginal-product-priced wage above subsistence
Marginal product	≈ 0 (surplus labor)	Strictly positive; above the wage
Capital intensity	Very low	Rises over time via reinvestment
Output use	Almost all consumed	Profits reinvested as capital
Technology	Stagnant, traditional	Adopting modern industrial methods
Labor supply curve	Inelastic at the village level	Perfectly elastic at the Lewis wage

Surplus labor and the zero marginal product

The signature assumption of the model is that the marginal product of labor in the traditional sector is zero — or near zero — over a long range. Lewis is careful: he does not claim every Indian or Egyptian peasant is literally non-productive. He claims that over some range of the labor force, withdrawing workers does not reduce total agricultural output. The mechanism is institutional: the family farm is a fixed plot of land worked by everyone who lives there, and output is divided by family members rather than by individual contribution. Under those conditions the last worker may contribute very little — perhaps nothing — to total output, but still receives an equal share of it. Average product (what each person eats) is strictly positive; marginal product (what the last worker adds) is not. The wedge between the two is what Lewis calls the surplus.

This was empirically defensible in 1954. China, India, sub-Saharan Africa, Latin America: all had labor-to-land ratios in agriculture far higher than late-19th-century Europe had had during its own industrialisation. The idea that an additional pair of hands on a rice paddy already worked by six adults adds nothing to the harvest survives modern econometric scrutiny in many settings, though the size of the surplus varies. The number can be inferred by comparing the productivity of workers who migrate to cities to the productivity of those they left behind: if rural output does not fall when migrants leave, surplus labor existed.

The Lewis wage

The modern sector capitalist faces a labor supply curve that is, for practical purposes, horizontal. Any number of workers can be hired by setting a wage just above the rural subsistence — what Lewis calls the institutional wage — typically estimated as the average rural product plus a premium of around 30 percent. The premium compensates the migrant for several things at once:

• Higher urban cost of living — rent, food bought rather than grown, transport.
• The psychic cost of leaving family, village, and known ways of life.
• The loss of village insurance: when illness or crop failure hits, the village shares; the city does not.
• Some search cost and the risk of urban unemployment during the move.

The key fact is that this wage does not rise as the modern sector grows. With a vast rural reservoir, capitalists never have to bid wages up to find workers. The supply schedule is flat at the Lewis wage, and the marginal cost of additional labor is constant. This is what makes the dynamics so different from a neoclassical labor market in which wages and employment rise together.

w_L = w_rural + premium
    ≈ (average product on the farm) × 1.30  (Lewis's stylised number)

Profits, reinvestment, and capital deepening

Inside the modern sector, the production function is standard: output depends on capital and labor, and the marginal product of labor falls as more workers are added to a given capital stock. The capitalist hires labor up to the point where marginal product equals the wage. Everything above that — the area between the marginal-product curve and the horizontal Lewis wage — is profit.

The crucial behavioural assumption is that capitalists reinvest. They save a high fraction of profits and accumulate capital. Each round of reinvestment shifts the marginal-product-of-labor curve up: more capital makes each worker more productive. At the same wage, employment expands. More employment generates more profit, profit is reinvested, the curve shifts again, employment expands further. This is the engine of the model — a self-reinforcing cycle in which growth in the modern sector is funded entirely by the surplus extracted from its own cheap labor.

One way to see why this is powerful: in a single-sector neoclassical model, growth in employment requires wage rises, and wage rises eat into profits, slowing accumulation. In the Lewis world, employment growth is free of wage pressure for a long time. The savings rate of the economy effectively rises as the share of national income earned by capitalists rises, and the rural-to-urban transfer of labor pays for the entire industrialisation programme. Lewis showed this could explain why savings rates rise during industrial take-off — a fact otherwise hard to reconcile with neoclassical theory.

The Lewis turning point

This cannot go on forever. The rural surplus is finite. Each migrant pulled into the modern sector reduces the rural labor force by one. At some point — call it the turning point — the rural marginal product is no longer zero. Pulling one more worker out now reduces farm output. Equivalently, the rural average product is now higher (the same harvest is divided among fewer people), so the wage needed to lure a migrant rises. The horizontal Lewis supply curve bends upward into a normal upward-sloping schedule.

The transition is sharp in the model and reasonably sharp in the data. Three observable signatures mark a country crossing it:

1. Real urban wages of low-skill labor begin rising rapidly after decades of stagnation, as employers bid against each other and against a tightening rural reserve.
2. The agricultural terms of trade improve — farm output prices rise relative to manufactures, because fewer workers per acre raises the rural marginal product and thus the supply price of food.
3. Labor shortages emerge in coastal/export manufacturing regions before they emerge in the interior, because that is where migration flows reach their margin first.

Before the turning point, growth feels like cheap labor abundance; after it, growth feels like a wage push. Productivity must rise to justify the higher wages, and the economy starts looking less like Lewis and more like a neoclassical full-employment model.

China — the textbook case

China has been a more textbook example of the Lewis model than any country Lewis himself studied. Between 1980 and the mid-2020s, the country urbanised from roughly 19 percent to over 65 percent of population. The estimated cumulative rural-to-urban migration is over 250 million people — the largest peacetime movement of labor in human history. Real manufacturing wages stayed almost flat in dollar terms through the 1980s and 1990s and only began rising slowly in the early 2000s, despite GDP growth of 8 to 10 percent annually. Capital accumulation in coastal export manufacturing absorbed migrants at constant nominal wages for two decades. Profit margins in light industry — textiles, electronics assembly, toys, footwear — were high because the labor input was so cheap. Profits were reinvested at rates over 40 percent of GDP.

Then, around the late 2000s, the system started running short of workers. Coastal cities reported migrant-worker shortages from 2004 onward. By 2010, real wages of unskilled migrants — measured by surveys of factory pay in Guangdong and Zhejiang — were rising at 10 to 15 percent per year in real terms. Between 2009 and 2019, nominal wages of unskilled rural-to-urban migrants roughly tripled. Cai Fang and other Chinese labor economists explicitly date China's Lewis turning point to roughly 2010, identifying it as the structural reason why Chinese export competitiveness in low-wage manufacturing was eroding and why production was relocating to Vietnam, Bangladesh, and Cambodia. The post-2010 Chinese economy is no longer a Lewis economy; it is a normal labor-constrained economy in which wages and productivity must rise together.

India — still pre-turning point

India presents the opposite case. As of 2024 roughly 42 percent of the Indian workforce remains in agriculture — a share that China had already fallen below by the 1990s. Real wages of casual rural workers have risen but slowly, with periods of decline. Per-capita income at which the rural labor share fell decisively below 30 percent — the rough threshold that historically precedes a Lewis turning point — has not yet been reached in India. The signature of an India-style stalled transition is twofold: many rural workers do migrate to cities, but they end up in informal-sector occupations — street vending, day labor, rickshaw driving, domestic service — rather than in productive modern manufacturing. The capitalist accumulation engine is therefore weakened, because the modern formal manufacturing sector is too small to absorb labor at scale. Restrictive labor laws in formal industry, the dominance of small-scale firms below the regulatory threshold, slow infrastructure, and the relative weakness of export-oriented manufacturing are all cited as binding constraints. The Lewis lens predicts that until India develops a labor-absorbing modern sector, wages will stay compressed and structural transformation slow.

Empirical estimates and rough numbers

Country	Crossed turning point?	Approximate date	Ag share of employment	Post-crossing wage signal
Japan	Yes	~ 1960	32 % (1960) → 6 % (2020)	Wages tripled in the 1960s
South Korea	Yes	~ 1975	50 % (1970) → 5 % (2020)	Real wages quadrupled 1975-1990
Taiwan	Yes	~ 1970	37 % (1970) → 5 % (2020)	Sharp wage rise late 1970s
China	Yes (consensus)	~ 2010	69 % (1980) → 22 % (2023)	Real manufacturing wages ~3× post-2010
Vietnam	Approaching	2025-2030 (projected)	40 % (2020)	Wages rising 6-8 %/yr in real terms
India	No	—	42 % (2024)	Casual rural wages flat to slowly rising
Sub-Saharan Africa (avg.)	No	—	~ 50 % (2024)	Limited modern-sector absorption

Extensions and successor models

• Fei-Ranis (1961, 1964). John Fei and Gustav Ranis formalised the Lewis model with explicit production functions for both sectors, traced the turning point precisely, and emphasised the role of agricultural productivity growth. In their version, the turning point can be brought forward by green-revolution-style improvements in farm yield — a hopeful note Lewis himself was sceptical of.
• Harris-Todaro (1970). John Harris and Michael Todaro pointed out that Lewis assumed every migrant landed a modern-sector job, contradicting massive observed urban unemployment. They replaced the certain modern wage with an expected wage equal to the urban wage times the probability of employment. Migration equilibrium leaves urban unemployment as a persistent feature, even when wage gaps remain. The Harris-Todaro model is the canonical migration model in development economics and a direct child of Lewis.
• Two-sector with productivity in agriculture. Modern variants relax Lewis's assumption of stagnant agriculture and add productivity growth in both sectors. This allows for the possibility that agricultural surplus can rise even without labor leaving, and that the turning point can be brought forward by farm modernisation.
• Informal urban sector (de Soto, ILO). A persistent finding in poor countries is that the modern capitalist sector is too small to absorb migrants, who flow instead into an informal urban sector of street vendors, day laborers, and unregistered workshops. Modelling this as a third sector — not capitalist, not rural — captures dynamics Lewis missed.
• Dual-sector with structural transformation. Recent work by Diego Restuccia, Berthold Herrendorf and Akos Valentinyi recasts dual-sector dynamics in a neoclassical general-equilibrium framework. They quantitatively replicate the wage path and the timing of the turning point for countries like South Korea using productivity gap estimates rather than zero marginal product as the primitive.

Critiques

• Agriculture is not stagnant. The green revolution showed in the 1960s and 1970s that rural productivity can rise dramatically, breaking the assumption of a fixed marginal product. T.W. Schultz, Lewis's Nobel co-laureate, made this his central argument.
• The informal sector eats the model. Most rural-to-urban migrants in modern developing economies do not enter modern factories; they enter slums and informal service work. The crucial Lewis link — migration into capitalist accumulation — is broken.
• Profits do not always reinvest. Capital flight, luxury consumption by elites, and remittance-financed real estate investment can divert profits from productive accumulation. The Lewis engine sputters if reinvestment is weak.
• Labor-displacing technology. Modern manufacturing is far more capital-intensive than the 1950s factories Lewis observed. An expanding modern sector may absorb fewer workers per unit of output than Lewis's arithmetic assumes — a problem amplified by automation.
• The wage is institutionally fragile. The Lewis wage assumes a stable subsistence anchor in the village. Rising rural living standards, rural-to-urban remittance flows, and political mobilisation can push the institutional wage up even before the turning point is reached, compressing profit margins and slowing accumulation.

Policy implications

The Lewis framework yields a set of policy heuristics that have shaped development practice for seventy years:

• Do not subsidise agriculture out of the surplus. Keeping the rural wage low and the urban-rural gap wide accelerates the labor transfer. Stalin's brutally extractive collectivisation in the 1930s and the Soviet "scissors crisis" of the 1920s were policy implementations of this logic — for better or, mostly, much worse.
• Invest in modern manufacturing — especially labor-absorbing export industry. Cheap-labor industrialisation is the textbook Lewis growth path. Every successful Asian Tiger followed it.
• Worry about the timing of the turning point. A country approaching it must reorient — productivity must replace cheap labor as the source of competitiveness. Failure to make the transition produces middle-income stagnation. China's post-2010 reform debate is exactly this transition problem.
• Watch rural-to-urban migration as a leading indicator. Slowdowns in migration are an early signal of the turning point approaching.

Common pitfalls in applying the model

• Mistaking the average product for the marginal. Rural workers may eat well — average product is positive — while still being marginally redundant. The two are different numbers and only the marginal matters for the Lewis surplus.
• Equating "modern sector" with "urban". Many urban workers in poor countries are informal and not part of the capitalist accumulation engine. Confusing urbanisation with industrialisation muddles the Lewis story.
• Assuming the turning point is a single date. It is a transition window, often a decade long, identified after the fact from wage dynamics. Real economies grade between Lewis and neoclassical labor markets.
• Ignoring labor demand. Lewis emphasises supply (the rural reserve). But modern-sector growth requires capital and demand for its output. A country can have abundant surplus labor and still fail to industrialise if export demand or capital accumulation falters.
• Treating it as a global growth model. Lewis described a particular structural transformation. It does not describe a rich-country labor market and was never meant to.