中文Resources › History of Economic Thought › Topic 13
The Golden Age and the Capital Controversy
How postwar economics combined Keynes with neoclassical theory, how the two Cambridges built rival theories of growth and distribution, and how their debate over “capital” exposed a problem at the center of the neoclassical theory of distribution.
- The golden age, 1945–73
- The neoclassical synthesis and its critics
- Growth and distribution in two Cambridges (interactive)
- What is “capital”?
- Sraffa and the wage–profit frontier
- Reswitching (interactive)
- Who won, and does it matter?
- Kuznets: measuring growth and inequality
- Monopoly capital
- Development economics
1. The golden age, 1945–73
The quarter-century after the Second World War was the fastest period of growth that the capitalist world had ever seen. On Angus Maddison’s estimates, income per person grew by about 4 percent a year in Western Europe and about 8 percent a year in Japan between 1950 and 1973, against about 2.5 percent in the United States, which started from a much higher level. Unemployment in most rich countries stayed low for two decades, and inequality of income and wealth fell to the lowest levels of the century.
The institutions behind the boom
- Bretton Woods (1944). Exchange rates were fixed to the U.S. dollar, and the dollar to gold at $35 an ounce. The International Monetary Fund lent to countries with balance-of-payments problems, and the World Bank financed reconstruction and development. Crucially, the system allowed governments to control capital flows, which left them free to set interest rates for domestic goals such as full employment.
- Reconstruction and the Marshall Plan (1948–52). The United States gave about $13 billion in aid to Western Europe, which financed imports of equipment and helped stabilize governments facing strong communist parties.
- A capital–labor accord. Unions were strong and collective bargaining widespread. In the 1950 “Treaty of Detroit,” for example, General Motors and the United Auto Workers tied wages to the cost of living and to productivity growth in return for labor peace. Wages rose roughly in line with productivity, so the shares of wages and profits stayed fairly stable.
- The mixed economy. Governments took responsibility for full employment, built welfare states, nationalized some industries (especially in Britain and France), and in France and Japan practiced indicative planning and industrial policy.
The golden age was also the age of decolonization and the Cold War. Newly independent countries in Asia and Africa pursued planned industrialization, and the Soviet Union’s rapid growth in the 1950s turned economic performance into a contest between systems. The age ended between 1968 and 1973: profitability fell in most rich economies, inflation rose, the United States ended the dollar’s convertibility into gold in 1971, and the oil shock of 1973 brought stagflation (Topic 14).
Economics changed with the economy. It became more mathematical, more American, and more confident that it could manage the economy. Two very different research programs grew up in two Cambridges: at the Massachusetts Institute of Technology, in Cambridge, Massachusetts, around Paul Samuelson, Robert Solow, and Franco Modigliani; and at the University of Cambridge in England, around Joan Robinson, Nicholas Kaldor, Piero Sraffa, Richard Kahn, and later Luigi Pasinetti and Pierangelo Garegnani. Both claimed to be Keynes’s heirs.
2. The neoclassical synthesis and its critics
Building the synthesis
In 1937 John Hicks condensed Keynes’s book into the IS–LM diagram: an IS curve for the goods market, where investment equals saving, and an LM curve for the money market, where the demand for money equals its supply. The diagram made Keynes teachable, and it showed fiscal and monetary policy as two levers that shift the curves. Franco Modigliani (1944) showed that in this model unemployment persists only if money wages are rigid. Later, Arthur Pigou (1943) and Don Patinkin (Money, Interest, and Prices, 1956) argued that falling prices raise the real value of money holdings, which raises spending (the “real balance effect”), so that with fully flexible prices the economy would in theory return to full employment.
Paul Samuelson put these pieces together. His Foundations of Economic Analysis (1947) recast economics as constrained optimization; his textbook Economics (1948), the best-selling economics text for decades, taught that Keynesian policy could secure full employment, after which the familiar neoclassical results about markets and prices would hold. He called this the neoclassical synthesis: Keynes for the short run, Walras for the long run.
In 1958 A. W. Phillips found a stable negative relation between unemployment and the rate of wage inflation in British data since 1861. Samuelson and Solow (1960) presented the American version as a menu of choices for policy makers. The Kennedy administration’s Council of Economic Advisers, led by Walter Heller, used Keynesian arguments for the tax cut of 1964. By the late 1960s, Keynesian demand management was so widely accepted that even its critics spoke its language; in 1971 President Nixon reportedly declared himself “now a Keynesian in economics.”
The critics
The synthesis had critics on both sides. On the right, Milton Friedman argued that monetary policy mattered more than fiscal policy and that the Phillips curve trade-off would disappear once people expected inflation (Topic 14). On the left, the English Cambridge and other post-Keynesians argued that the synthesis had kept Keynes’s policy conclusions while throwing away his theory. Joan Robinson called it “bastard Keynesianism.” Axel Leijonhufvud’s On Keynesian Economics and the Economics of Keynes (1968) made a related argument from within the mainstream: that the “Keynesian economics” of the textbooks was not the economics of Keynes.
The difference mattered for policy. If unemployment comes from sticky wages, the long-run remedy is to make labor markets more flexible. If it comes from insufficient demand and unstable investment, flexible wages may make things worse (Topic 12), and the state must manage demand permanently.
3. Growth and distribution in two Cambridges
Harrod and Domar: the knife-edge
Roy Harrod (1939) and Evsey Domar (1946) extended Keynes from the short run to growth. Investment adds to demand, but it also adds to productive capacity. If the capital–output ratio v is fixed and households save a fraction s of income, the economy can grow steadily only at the warranted rate s/v, the rate at which new capacity is just matched by new demand. Full employment over time also requires growth at the natural rate n, the growth of the labor force plus labor productivity. Nothing makes the two equal, and any deviation from the warranted rate tends to grow: if firms invest more than expected, demand rises, capacity looks short, and they invest still more. The economy balances on a “knife-edge.” The two Cambridges gave opposite answers to the question of what brings the warranted and natural rates together.
The American Cambridge: Solow and Swan
Robert Solow (1956) and the Australian Trevor Swan (1956) let the capital–output ratio adjust. With an aggregate production function Y = F(K, L), firms can choose more or less capital-intensive methods, so capital per worker rises or falls until the economy settles on a steady state in which output grows at the natural rate. The saving rate affects the level of income but not its long-run growth rate, which comes only from technological progress; poorer countries should grow faster and converge toward richer ones. In this model distribution is determined by technology: the wage and the profit rate equal the marginal products of labor and capital (Topic 9).
Solow’s growth accounting (1957) then estimated that about seven-eighths of the growth of output per worker hour in the United States between 1909 and 1949 came not from more capital per worker but from a residual he labeled “technical change.” The Solow model became the foundation of mainstream growth economics, and later of endogenous growth theory (Topic 21).
The English Cambridge: Kaldor, Robinson, and Pasinetti
Nicholas Kaldor (1956) gave a different answer: distribution adjusts. Capitalists save a much larger share of their income than workers do. If investment rises, prices rise relative to wages, the share of profits goes up, and since profits are largely saved, total saving rises until it matches investment. Investment, driven by capitalists’ “animal spirits,” comes first; profits follow. Joan Robinson’s The Accumulation of Capital (1956) developed similar ideas and insisted that the economy moves through historical time, not between equilibria.
Luigi Pasinetti (1962) showed that even if workers save and own some capital, the long-run rate of profit depends only on the growth rate and on capitalists’ saving propensity. This is the Cambridge equation:
r = g / sc (rate of profit = growth rate ÷ capitalists’ saving propensity)
Workers’ saving drops out. Kalecki had made the same point in the slogan that capitalists earn what they spend, while workers spend what they earn. In this theory, the profit rate is not the marginal product of anything: it is set by accumulation and by the spending habits of the capitalist class.
Two theories of the rate of profit
Solow version: with Cobb–Douglas technology, the profit rate equals α × Y/K, and in the steady state K/Y = s/(g + δ), where s is the economy-wide saving rate and the depreciation rate δ is set at 4%. Here s is the average of sc and sw weighted by a 30% profit share.
4. What is “capital”?
The neoclassical theory of distribution says that, in competitive markets, the wage equals the marginal product of labor and the rate of interest (or profit) equals the marginal product of capital. Three “parables” follow, and they are what made the theory attractive: (1) the rate of profit is determined by the scarcity of capital, as the wage is by the scarcity of labor; (2) a lower rate of interest leads firms to choose more capital-intensive methods, so the demand curve for capital slopes down, just like any demand curve; and (3) the shares of wages and profits reflect the contributions of labor and capital to production. For these parables to hold, “capital” must be a quantity that can be measured before we know the rate of profit.
Robinson’s question
In “The Production Function and the Theory of Capital” (1953–54), Joan Robinson asked in what units capital is measured. Students, she complained, are taught to write output as a function of labor and capital and are then hurried on, in the hope that the student will forget to ask in what units C is measured. Real capital is a collection of different goods: trucks, machines, buildings, software. To add them up we need prices. But the price of a machine includes the profit earned on the capital used to make it, so the value of the capital stock changes when the profit rate changes, even if not a single machine changes.
Ten trucks and five machines
Each truck needs $10,000 of labor and $10,000 of capital inputs; each machine needs $10,000 of labor and $20,000 of capital inputs. Price = labor cost + capital cost × (1 + profit rate).
The same physical capital stock has a different “quantity” at each profit rate. If we need the profit rate to measure capital, we cannot then explain the profit rate by the marginal product of that capital without arguing in a circle. The value of capital also changes with the profit rate because of changes in relative prices, which later writers called price Wicksell effects, after Knut Wicksell, who had noticed the problem in the 1890s.
The American reply: the surrogate production function
The American Cambridge did not deny that real capital is heterogeneous. Its answer was that the one-good model is a useful simplification that more complex models would confirm. Solow treated the aggregate production function as an approximation to be judged by how well it fits the data. In “Parable and Realism in Capital Theory” (1962), Samuelson went further: he built a model with many different capital goods and showed that it behaved exactly as if there were a single capital good, with a “surrogate production function” from which wages and profits could be read off as marginal products. If that held generally, the parables would be safe.
Pierangelo Garegnani (1970) showed that Samuelson’s result depended on a special assumption: that in each technique, the consumer good and the capital good are produced with the same proportions of capital to labor. That is exactly the case in which prices are proportional to labor values (Topic 3), and in which the value of capital does not change with the profit rate. Drop it, and the surrogate production function disappears.
5. Sraffa and the wage–profit frontier
Piero Sraffa (1898–1983), an Italian economist whom Keynes brought to Cambridge in 1927, had already shaken the Marshallian theory of the firm in 1926, showing that Marshall’s supply curves were inconsistent with perfect competition; Joan Robinson’s Economics of Imperfect Competition (1933) grew out of that critique. He then spent decades editing Ricardo’s collected works and writing a short book, Production of Commodities by Means of Commodities (1960), subtitled Prelude to a Critique of Economic Theory.
Sraffa revived the classical approach of Ricardo. Commodities are produced by means of commodities and labor. Given the technique of production, prices and one distributive variable are determined once the other is known: if the profit rate is set, for instance, by the interest rate or by class conflict, the wage follows, and vice versa. Nothing in the system requires utility, demand curves, or marginal products. Sraffa also constructed a standard commodity, a composite of goods whose price does not change when distribution changes, and showed that in terms of it the relation between the wage w and the rate of profit r is a straight line:
w = 1 − r / R (R = the maximum rate of profit, when wages are zero)
For each technique there is a wage–profit frontier: the highest wage compatible with each profit rate. Higher profits mean lower wages, which puts conflict over distribution at the center of the theory, as it was for Ricardo and Marx. When everyday prices are used instead of the standard commodity, the frontier is a straight line only if every sector uses capital in the same proportion; otherwise it bends. When firms can choose among techniques, they pick whichever gives the highest wage at the going profit rate, or equivalently the lowest cost. The economy’s overall frontier is the outer envelope of the frontiers of all available techniques. Chapter 12 of Sraffa’s book pointed out, in a few pages, that the same technique could appear on that envelope at two separate ranges of the profit rate.
6. Reswitching
How the debate unfolded
In 1965 David Levhari, a student of Samuelson’s at MIT, published a proof in the Quarterly Journal of Economics that reswitching could not happen in an economy as a whole, as opposed to a single industry. If true, the neoclassical parables were safe. Luigi Pasinetti presented a counterexample at the World Congress of the Econometric Society in Rome that year. In November 1966 the Quarterly Journal published a symposium, “Paradoxes in Capital Theory,” with papers by Pasinetti, Garegnani, Michio Morishima, and others; Levhari and Samuelson’s own contribution was titled “The Nonswitching Theorem Is False.” Samuelson closed the symposium with “A Summing Up,” which included the example below.
Samuelson’s example
Two techniques produce one unit of the same good using only labor, applied at different dates:
- Technique A: 7 hours of labor, applied two years before the output is ready.
- Technique B: 2 hours applied three years before, plus 6 hours applied one year before.
Wages paid in advance must earn the going profit rate r until the output is sold, so in units of the wage the costs are
cost A = 7(1 + r)² cost B = 2(1 + r)³ + 6(1 + r)
Which technique is cheaper?
The line shows how much B costs relative to A at each profit rate. Shaded: the range where B is cheaper.
Setting the two costs equal gives 2(1 + r)² − 7(1 + r) + 6 = 0, with roots 1 + r = 1.5 and 1 + r = 2. So the techniques switch at r = 50% and switch back at r = 100%:
| Profit rate | Cheaper technique | What the neoclassical story would conclude |
|---|---|---|
| below 50% | A | A is chosen at low interest rates, so A must be the more capital-intensive technique. |
| 50% to 100% | B | B is chosen at higher interest rates, so B must be less capital-intensive. |
| above 100% | A again | But now A is chosen at an even higher interest rate, so A must be less capital-intensive than B. |
Both conclusions cannot be true. The ranking of techniques by “capital intensity” depends on the profit rate, so there is no general, profit-independent sense in which one technique uses more capital than another.
Reswitching and capital reversing
A related and in some ways more damaging result is capital reversing (or reverse capital deepening): moving to a technique that is chosen at a lower profit rate can reduce, rather than raise, the value of capital per worker. It can occur even when no technique ever comes back. Either way, a fall in the interest rate need not lead firms to use more capital per worker, and the downward-sloping demand curve for capital that the neoclassical parables need is not guaranteed. Without it, there is no assurance that a market for “capital” has a stable equilibrium, or that the profit rate measures capital’s scarcity.
Scholars are not born to live an easy existence.Paul Samuelson, “A Summing Up,” Quarterly Journal of Economics (1966), on the headaches reswitching causes for the old parables
7. Who won, and does it matter?
| Cambridge, Massachusetts | Cambridge, England | |
|---|---|---|
| Main figures | Paul Samuelson, Robert Solow, Franco Modigliani, David Levhari, C. E. Ferguson | Joan Robinson, Piero Sraffa, Nicholas Kaldor, Luigi Pasinetti, Pierangelo Garegnani |
| Starting point | Walras and Marshall: prices and distribution determined by preferences, technology, and endowments | Ricardo, Marx, and Keynes: distribution determined by social and historical forces; growth driven by investment |
| The profit rate | The marginal product of capital, reflecting its scarcity | Set by accumulation, capitalists’ saving, or class conflict; not the price of a scarce factor |
| Aggregate capital | A useful approximation; or unnecessary in a full general-equilibrium model | Logically incoherent outside a special case |
| Method | Equilibrium models, judged by fit with data | Historical time; skeptical of comparing equilibria as if the economy could move between them |
On logic, the English Cambridge won. Samuelson conceded that reswitching is possible and that aggregate capital cannot in general be measured independently of distribution. C. E. Ferguson, in a 1969 textbook on neoclassical production theory, concluded that relying on the theory was “a matter of faith.” Geoffrey Harcourt’s surveys (1969, 1972) recorded the outcome in detail.
Yet the American Cambridge kept the field. The aggregate production function stayed at the center of growth theory, macroeconomics, and empirical work. Its defenders made three replies:
- Approximation. Solow argued that the aggregate production function should be judged as an empirical tool, not as a description of reality, and that it works well enough for the questions it is used to answer. His Capital Theory and the Rate of Return (1963) also proposed analyzing the return on saving directly, without measuring a stock of capital.
- General equilibrium. Frank Hahn and others argued that the critique hit only the one-good parables. A full general-equilibrium model, with many capital goods each with its own price, does not need an aggregate K at all, and reswitching is no paradox in it.
- Rarity. Reswitching might be possible in theory but rare in practice.
The English Cambridge replied that general-equilibrium models give up the very parables (the scarcity theory of profit and the downward-sloping demand for capital) that made the neoclassical theory of distribution attractive, and that the approximation hides the question of how income is distributed between wages and profits. Joan Robinson, in her later work, turned the debate into a critique of equilibrium itself: an economy that has accumulated a given stock of machines cannot slide back and forth along a production function, because history cannot be undone.
Empirically, reswitching does seem to be rare. Studies of input–output data, such as Ochoa’s (1989) work on the U.S. economy, found measured wage–profit curves to be close to straight lines, which is also why labor values track market prices closely (see Topic 3). Han and Schefold (2006), comparing the techniques of OECD economies, found reswitching and capital reversing only in a small minority of cases. There is an irony here that Hunt emphasizes: the condition that makes reswitching unlikely is the same condition that makes the labor theory of value work well. Avi Cohen and Geoffrey Harcourt’s retrospective, “Whatever Happened to the Cambridge Capital Theory Controversies?” (2003), argues that the questions the debate raised, about what determines distribution and whether the profit rate measures anything real, were never answered, only set aside.
8. Kuznets: measuring growth and inequality
Much of the golden age’s economics rested on a new kind of data. Simon Kuznets (1901–1985), a Russian-born American economist, prepared the first official estimates of U.S. national income for the Senate in 1934, National Income, 1929–32, the ancestor of today’s GDP accounts. He was careful about what the numbers meant. The report warned that the welfare of a nation can scarcely be inferred from a measurement of national income, and Kuznets later argued against counting military spending and some government services as final output. He went on to compile long-run statistics of growth across countries, summarized in Modern Economic Growth (1966), and received the Nobel Prize in 1971.
In his 1955 presidential address to the American Economic Association, “Economic Growth and Income Inequality,” Kuznets looked at the few long income series then available, for the United States, Britain, and Germany. He suggested that inequality first rises as a country industrializes, because workers move from an equal but poor agricultural sector into a richer but more unequal urban one, and because savings concentrate among the rich; and that it later falls, as the urban population gains political power, wages rise, and the children of migrants move up. This is the inverted-U Kuznets curve. Kuznets warned that his evidence was thin, calling the paper perhaps “5 per cent empirical information and 95 per cent speculation.” Its optimistic message, that growth would reduce inequality by itself, nonetheless became a staple of Cold War economics and of development policy.
Later evidence was mixed. Cross-country data from the 1970s seemed to show the inverted U, but much of the pattern came from comparing Latin American countries, which are both middle-income and unequal, with others. And inequality rose again in most rich countries after 1980. Thomas Piketty (Topic 25) argued that the mid-century fall in inequality came from wars, depression, and policy, not from a natural stage of growth.
9. Monopoly capital
Marxian economics survived the golden age mainly outside the universities. Paul Sweezy (1910–2004), who left Harvard in the late 1940s, founded the magazine Monthly Review with Leo Huberman in 1949. His Theory of Capitalist Development (1942) had been the first systematic account of Marxian economics in English. Paul Baran (1909–1964), at Stanford, wrote The Political Economy of Growth (1957). Baran distinguished the actual surplus, what a society actually produces beyond its current consumption, from the potential surplus it could produce with its resources fully and rationally used; the gap measured the waste of capitalism. He argued that underdevelopment in the periphery was the product of its relation to the advanced capitalist countries, which drained its surplus.
The two drew on earlier work on monopoly and stagnation: Kalecki’s theory that prices are set as a markup over costs reflecting the “degree of monopoly” (1938), and Josef Steindl’s Maturity and Stagnation in American Capitalism (1952), which argued that the concentration of industry slows investment. Together they wrote Monopoly Capital (1966), published after Baran’s death. Their argument ran in steps:
- From competition to monopoly. The economy is dominated by a few hundred giant corporations that avoid price competition and compete instead through advertising, product design, and cost cutting. Prices rarely fall; costs do.
- The law of the rising surplus. Rising productivity with stable prices produces an economic surplus, the difference between what a society produces and the costs of producing it, that tends to rise over time.
- The absorption problem. Capitalists’ consumption and investment are not enough to absorb the surplus; investment is held back because new capacity would undercut the profits of existing plants. Left to itself, they wrote, “the normal state of the monopoly capitalist economy is stagnation.”
- Counteracting forces. Stagnation is held off by the sales effort (advertising, model changes, planned obsolescence), by government spending, especially on the military, and by imperialism. These absorb the surplus but are wasteful, and they shape the culture of the whole society.
The book had a large influence on the New Left of the 1960s. Harry Magdoff’s The Age of Imperialism (1969) documented the dependence of U.S. corporations on raw materials and markets abroad. John Kenneth Galbraith’s The New Industrial State (1967) described the planning power of large corporations from a liberal standpoint. Other Marxists criticized Baran and Sweezy for replacing Marx’s surplus value with a looser concept of “surplus” and for setting aside the tendency of the rate of profit to fall (Topic 7); in the 1970s, when profits were squeezed by wages and oil prices, their stagnation thesis looked less convincing. Hunt treats their work as a central part of the revival of critical political economy in the 1960s (ch. 19).
10. Development economics
The golden age also produced development economics, as newly independent countries asked how to industrialize. The early development economists shared a belief that markets alone would not transform poor agrarian economies, and that the state had to plan and invest.
- The big push. Paul Rosenstein-Rodan (1943) argued that a single factory may fail for lack of buyers, but many industries built together create markets for one another, so investment must be coordinated on a large scale. Ragnar Nurkse (1953) called for “balanced growth” across sectors.
- Unbalanced growth. Albert Hirschman’s The Strategy of Economic Development (1958) replied that poor countries lack the capacity to do everything at once. They should invest in industries with strong linkages, backward to suppliers and forward to users, and let the resulting shortages and pressures pull investment into other sectors.
- The dual economy. W. Arthur Lewis (1954) described an economy in which a modern capitalist sector grows by drawing labor from a traditional sector at a near-subsistence wage. Profits and investment rise while wages stay low, until the surplus labor is used up, a point later called the Lewis turning point. Lewis shared the Nobel Prize in 1979.
- Terms of trade and import substitution. Raúl Prebisch and Hans Singer (1950) argued that the terms of trade of primary-commodity exporters tend to decline against those of manufacturing exporters. Under Prebisch, the UN Economic Commission for Latin America (ECLA, founded 1948) advocated import-substituting industrialization behind tariffs. Gunnar Myrdal (1957) argued that trade and markets often widen inequalities between regions through “cumulative causation.”
- Stages of growth. Walt Rostow’s The Stages of Economic Growth: A Non-Communist Manifesto (1960) described five stages, from traditional society through a “take-off” to high mass consumption, and offered them as an alternative to the Soviet model.
By the late 1960s, import substitution was criticized for producing inefficient industries, chronic balance-of-payments problems, and inequality, while South Korea and Taiwan combined industrial policy with exporting. Dependency theorists argued that the problem was not a lack of capital but the structure of the world economy itself (Topic 11). From the 1980s, the Washington Consensus replaced planning with liberalization (Topic 14), but the questions of the early development economists returned in later debates about East Asia, China, the middle-income trap, and global value chains (Topic 19).
Readings
Core reading
Hunt, ch. 16 (Sraffa) and ch. 17–18 (contemporary orthodoxy, post-Keynesianism, Sraffian price theory).
Avi J. Cohen and G. C. Harcourt, “Whatever Happened to the Cambridge Capital Theory Controversies?” Journal of Economic Perspectives 17, no. 1 (2003).
Companion reading
Yueh, What Would the Great Economists Do?, ch. 9 (Joan Robinson) and ch. 12 (Robert Solow).
Joan Robinson, “The Production Function and the Theory of Capital,” Review of Economic Studies 21, no. 2 (1953–54).
Paul A. Samuelson, “A Summing Up,” Quarterly Journal of Economics 80, no. 4 (1966).
Baran and Sweezy, Monopoly Capital: An Essay on the American Economic and Social Order (1966), ch. 1–5.
Hunt, ch. 19 (the section on the performance of capitalism at the aggregate level).