Heckscher-Ohlin Model

The question Ricardo could not answer

By 1900, Ricardo's theory of comparative advantage was a hundred years old and still the only systematic answer economists had to the question "why do countries trade?" Ricardo's answer was technology: each country was endowed with different production functions, and even if one country was absolutely worse at making everything, it would specialise in whatever it was relatively least bad at. The argument was logically airtight and empirically suggestive but it left a basic puzzle untouched: why are technologies different across countries? Surely a Lancashire spinning machine is a Lancashire spinning machine wherever it is set down. If knowledge can cross borders, technological differences should be eroded by imitation. So what then is the durable, geographically rooted thing that makes trade happen?

The Swedish economist Eli Heckscher proposed an answer in a 1919 article ("The Effect of Foreign Trade on the Distribution of Income") and his student Bertil Ohlin developed it into a full theory in his 1933 book Interregional and International Trade. Their move was to set all technological differences aside — assume every country uses the same production function for every good — and instead let countries differ in what they have to work with: their endowments of factors of production. Labour, capital, land, and now we would add human capital and natural resources. The Heckscher-Ohlin theorem, in one sentence, is:

A country exports the good whose production intensively uses the factor with which the country is relatively abundantly endowed.

This is endowment-based comparative advantage. China is labour-abundant relative to capital, so it exports labour-intensive textiles and apparel and imports capital-intensive machinery from the United States. The United States is capital-abundant, so it exports machinery, aircraft, and pharmaceuticals and imports labour-intensive consumer goods. Australia is land-abundant — and the land happens to be good for cattle and wheat — so it exports beef and grain. The pattern of trade is read off the map of endowments.

The canonical 2×2×2 setup

The model in its textbook form has the most parsimonious structure imaginable. There are two countries (call them Home and Foreign), two goods (call them cloth and steel), and two factors of production (labour L and capital K). The assumptions are:

• Identical technology. Both countries know how to produce both goods using the same production functions Q_cloth = F(L, K) and Q_steel = G(L, K).
• Different endowments. Home has more labour per unit of capital than Foreign: L_H / K_H > L_F / K_F. So Home is labour-abundant, Foreign is capital-abundant.
• Different factor intensities. Cloth uses labour intensively (high L/K ratio at any given factor prices); steel uses capital intensively. We assume no "factor-intensity reversal" — the ranking does not flip as factor prices change.
• Factors mobile within a country, immobile across countries. Workers can move from cloth to steel inside Home but cannot emigrate to Foreign.
• Perfect competition, constant returns to scale, full employment of factors, no transport costs, no tariffs, and identical homothetic preferences across countries.

From this list the model derives four major results: the Heckscher-Ohlin theorem (pattern of trade), the Stolper-Samuelson theorem (1941, distributional effects of trade), the Rybczynski theorem (1955, effect of endowment changes on output), and the factor-price equalization theorem (1948, the most striking and most controversial).

Autarky: why prices differ before trade

Before trade opens, each country produces whatever it consumes — autarky. Because Home is labour-abundant, labour is comparatively cheap there: the wage-rental ratio w/r is low. Cloth, the labour-intensive good, can therefore be produced cheaply in Home. Foreign has the opposite configuration: capital is abundant, the wage-rental ratio is high, and the labour-intensive good is expensive while the capital-intensive good is cheap.

Notation: let p ≡ p_cloth / p_steel be the relative price of cloth. In autarky:

p_H^A < p_F^A
(cloth cheap in labour-abundant Home;
 cloth expensive in capital-abundant Foreign)

This is the price gap that opening trade will arbitrage. Trade flows from cheap to expensive: cloth moves from Home to Foreign, steel moves from Foreign to Home, and the gap closes from both ends.

Opening trade: specialization and convergence

Once trade opens and goods can flow freely between Home and Foreign, the law of one price (in the absence of transport costs) forces the relative price p to be the same in both countries. Call this world price p^W; it lies between the two autarky prices: p_H^A < p^W < p_F^A.

In Home, the relative price of cloth has risen from p_H^A to p^W. Cloth becomes more profitable, so Home reallocates resources from steel to cloth — but it does not specialise completely. Both goods are still produced (incomplete specialisation), just in different proportions. Home now produces more cloth than it consumes and exports the surplus; it produces less steel than it consumes and imports the shortfall. Foreign does the mirror image: it specialises (partially) in steel, exports steel, imports cloth.

The remarkable fact about Heckscher-Ohlin equilibrium is that the volume of trade depends only on the difference in endowments. The more dissimilar two countries are in their factor proportions, the more they trade. Two identical economies would have no reason to trade at all — a prediction that turns out to be sharply violated in reality and is one of the model's most studied failures.

Stolper-Samuelson: who wins, who loses

The Stolper-Samuelson theorem (Wolfgang Stolper and Paul Samuelson, 1941) takes Heckscher-Ohlin's general-equilibrium structure and reads off its distributional implications. The theorem says:

A rise in the relative price of a good causes a more-than-proportional rise in the real return to the factor used intensively in producing that good, and a fall in the real return to the other factor.

Apply this to the opening of trade in our two countries:

Country	Abundant factor	Exported good (intensive use)	Price ↑	Winners	Losers
Home (e.g. China)	Labour	Cloth (labour-intensive)	p_cloth ↑	Workers	Capital owners
Foreign (e.g. US)	Capital	Steel / machinery (capital-intensive)	p_steel ↑	Capital owners	Workers

This is the formal grounding for what political economists had observed for decades: in a capital-abundant rich country, opening to trade with a labour-abundant poor country raises returns to capital (and to highly-skilled labour) and lowers real wages for less-skilled workers. The theorem does not say everyone is worse off — Heckscher-Ohlin says aggregate gains from trade are positive. It says distributional consequences are sharp: a sub-group of factor owners is unambiguously worse off, in real terms, even though the country as a whole gains. That asymmetry is the engine of trade politics.

The empirical signature is visible in US data from the 1980s onward: the college-wage premium rose, blue-collar manufacturing wages stagnated in real terms, and the share of national income going to capital rose. Trade with China is not the only cause — skill-biased technological change is a major rival explanation — but Stolper-Samuelson is the standard channel through which trade contributes to rising within-country inequality.

Factor-price equalization: the magic claim

The factor-price equalization theorem (Samuelson 1948, 1949) is the most striking result of the Heckscher-Ohlin framework, and the one that strains credulity the most. It says:

Under the model's assumptions, free trade in goods alone — with no movement of labour or capital across borders — is sufficient to drive the wage rate and the rental rate of capital to the same level in every country.

The mechanism is indirect. When Home exports cloth, it is effectively exporting the labour embodied in that cloth. Foreign, by buying the cloth, is effectively importing labour services. Trade in goods is implicit trade in factors. As trade volume rises, the implicit factor flows are large enough that, under the right conditions, w_H = w_F and r_H = r_F exactly.

The "right conditions" are restrictive: identical technologies, no transport costs, no factor-intensity reversal, both goods produced in both countries (no complete specialisation), and balanced trade. In practice, wages between rich and poor countries differ by factors of 10 or more even after decades of trade liberalisation. Factor-price equalization is best read as a limit theorem — a statement about which way prices move (toward equality) rather than how far they actually go.

The empirical evidence is mixed: trade liberalisation episodes (NAFTA, EU enlargement, China's WTO accession) do produce wage convergence between trading partners, but the convergence is partial and slow, attenuated by everything the textbook model assumes away.

The Leontief paradox

The first serious empirical test of Heckscher-Ohlin came from Wassily Leontief in 1953. Leontief had just developed input-output analysis (for which he would win the 1973 Nobel) and had detailed 1947 tables of how much labour and capital US industries used per dollar of output. He computed two numbers: the capital-to-labour ratio embodied in a representative bundle of US exports, and the same ratio for a bundle of US import-substitutes (goods the US imported but could have produced domestically).

The 1947 United States was unambiguously the most capital-abundant economy on Earth — the war had destroyed much of Europe and Japan's capital stock while leaving America's intact and expanded. Heckscher-Ohlin therefore predicted, sharply and unambiguously, that US exports should be more capital-intensive than US imports. Leontief found the opposite:

    K/L (exports)  <  K/L (import-substitutes)
US exports were ~30% MORE labour-intensive than import-substitutes

The result was a shock. It dominated trade-theory journals for two decades and was replicated by Bhagwati (1964), Baldwin (1971), and many others on various data sets, with varying degrees of confirmation. Several resolutions were offered:

• Human capital matters. "Labour" is not one factor. American workers in 1947 were much better educated and trained than workers in trading partners; the labour embodied in US exports was disproportionately skilled labour. Treat human capital as a separate factor and the US becomes human-capital-abundant; the paradox softens or disappears.
• Natural resources matter. US import-substitutes included resource-intensive goods (minerals, certain raw materials). Treat natural resources as a third factor, deduct them, and the comparison cleans up.
• Demand reversal. If US preferences were unusually biased toward capital-intensive goods, the US could end up exporting labour-intensive goods despite being capital-abundant. The model assumes identical preferences; relax that and the prediction weakens.
• Factor-intensity reversal. A good might be labour-intensive in one country and capital-intensive in another. If so, the simple HO sign predictions break.
• Productivity-adjusted factors. Trefler (1993, 1995) showed that allowing technology to differ across countries — measured as productivity-adjusted endowments — substantially repairs the model's empirical performance.

None of these resolutions is fully satisfying. The Davis-Weinstein (2001) "factor content of trade" exercise concluded that with a richer factor list and country-specific productivity, the sign of Heckscher-Ohlin predictions is recovered for most factor categories. But the magnitudes still fail — countries trade about a quarter of what the model predicts. The puzzle has shifted from "why does the sign reverse?" to "why is the volume so low?"

Variants and extensions

• Heckscher-Ohlin-Vanek (1968). A multi-good, multi-country, multi-factor generalisation. The HOV theorem predicts the factor content of trade in terms of factor endowments, which is what Trefler and Davis-Weinstein actually test.
• Rybczynski theorem (1955). A complement to Stolper-Samuelson. At constant prices, an increase in a country's endowment of one factor causes a more-than-proportional rise in the output of the good using that factor intensively, and an absolute fall in the output of the other good. Used to think about the effects of immigration, capital accumulation, and labour-force growth on industry composition.
• Specific-factors (Ricardo-Viner) model. Allows one factor to be tied to a specific industry (e.g. land specific to agriculture, capital specific to manufacturing) while another is mobile across sectors. Short-run alternative to HO with cleaner distributional intuition.
• New trade theory (Krugman 1979, 1980, Helpman 1981). Adds increasing returns and product differentiation. Explains intra-industry trade — Germany and France trading cars — which HO cannot. Does not replace HO; runs alongside it for differentiated goods.
• Eaton-Kortum (2002), Melitz (2003). Modern quantitative trade models. EK gives the gravity equation and country-pair trade flows; Melitz adds firm heterogeneity. They subsume HO as a special case rather than rejecting it.
• Trade-and-wages literature. Empirical studies (Wood, Lawrence-Slaughter, Feenstra-Hanson, Autor-Dorn-Hanson "China Shock") apply Stolper-Samuelson logic to measure the actual wage impact of trade with developing economies.

Where the model fits the data, and where it does not

Factor	Empirical fit	Comment
Land	Strong	Argentina, Australia, Canada export grain and beef. Cleanest HO confirmations.
Mineral / energy	Strong	Saudi oil, Chilean copper, Australian iron ore. Endowment is literally in the ground.
Skilled labour (human capital)	Moderate	Skill-abundant countries export skill-intensive goods. Works for OECD pairs.
Physical capital	Weak	The original Leontief paradox lives here. Improved but not solved.
Unskilled labour	Moderate	China and Bangladesh export labour-intensive goods — but so does technologically-leading Vietnam.
Aggregate trade volume	Weak	The "missing trade" problem (Trefler 1995): actual factor-content of trade is ~25% of HO prediction.
Intra-industry trade	Fails	~50–80% of OECD trade is intra-industry. HO predicts inter-industry trade only.

Where Heckscher-Ohlin shows up in policy and history

• China's accession to the WTO (2001). Textbook HO event: a massive injection of labour-abundant labour into the world trading system, predicted to lower wages of less-skilled workers in capital-abundant rich countries. Autor, Dorn, and Hanson's "China Shock" papers measure this in US manufacturing regions and find effects consistent with Stolper-Samuelson.
• NAFTA (1994). Bringing labour-abundant Mexico into a free-trade area with capital-abundant US and Canada produced predicted patterns: US exports of capital-intensive intermediates rose; Mexico's exports of labour-intensive assembled goods (apparel, autos) rose; US manufacturing wages in exposed industries fell.
• Brexit and trade with the EU. The factor-content arguments are messier because the UK and EU have similar factor endowments — Stolper-Samuelson predicts small distributional effects, which roughly matches observed labour-market evidence.
• Resource curse and Dutch disease. Rybczynski applied to natural resources: a windfall increase in a country's endowment of an extractable resource shifts production toward that resource and squeezes out manufacturing, often appreciating the real exchange rate.
• Immigration debates. HO and the specific-factors model give competing predictions for how immigration affects native wages. HO with multiple goods predicts no wage effects from immigration if the country can adjust its output mix (Leamer-Levinsohn); specific-factors predicts wage effects on substitutable native workers (Borjas, Card).

Worked example: comparing US-China trade flows

Use rough numbers for 2020. The United States has roughly 5 trillion dollars of physical capital per worker (averaged over its 165 million-strong workforce). China has roughly 200,000 dollars of physical capital per worker (over its 800 million workforce). The capital-to-labour ratio is therefore about 25 times higher in the US.

K/L  (US)    ≈ 5 × 10⁶ USD / worker
K/L  (China) ≈ 2 × 10⁵ USD / worker
Ratio        ≈ 25 ×

Heckscher-Ohlin predicts: the US exports goods that intensively use capital, China exports goods that intensively use labour. Actual 2020 bilateral trade flows (top categories by dollar value):

US exports to China	China exports to US
Aircraft and parts	Smartphones and tablets
Semiconductors	Toys, sporting goods, apparel
Soybeans and corn	Furniture, plastics
Pharmaceuticals	Footwear, textiles
Automobiles (high-end)	Automobiles (low-end EVs, parts)

The picture is broadly consistent with HO: US exports concentrate in capital-intensive and human-capital-intensive goods (aircraft, chips, pharma) plus land-intensive agricultural products (soybeans — the US is also land-abundant). China's exports concentrate in goods that even today use comparatively more labour per unit of value. But notice the overlap in automobiles — both countries export them, of different sub-types. That intra-industry overlap is what HO cannot explain and what new trade theory was built for.

Common pitfalls

• Confusing "abundant" with "absolutely large". Heckscher-Ohlin is about ratios of factor endowments, not absolute amounts. The US has more workers than Sweden in absolute terms, but Sweden has a higher capital-to-labour ratio, so Sweden is capital-abundant relative to the US.
• Forgetting that HO predicts aggregate gains. The gains-from-trade result and the distributional result are separate. Stolper-Samuelson identifies losers within each country; HO says total income is higher with trade. Both claims are simultaneously true.
• Treating factor-price equalization as a literal prediction. The theorem holds under restrictive assumptions and even then only when both goods are produced in both countries. In practice it indicates direction of movement, not endpoint.
• Using HO to explain intra-industry trade. Most modern trade between similar economies is intra-industry. HO is silent on this; reach for Krugman-Helpman or Melitz instead.
• Reading the Leontief paradox as a refutation. Leontief showed that the simplest HO with two homogeneous factors fails the US 1947 test. Extensions with human capital and natural resources mostly rescue the sign of the prediction; the volume failure remains.
• Ignoring factor-intensity reversal. The clean two-factor theorems assume the factor ranking of goods does not flip as factor prices change. Real production functions can violate this; when they do, the sign predictions of HO are not robust.