Plate v in print

The Decline

Karabarbounis and Neiman; the labor share has fallen across the world since about 1980 — by cheapening machines, or by the rising power of capital.

The share that held for a century has, since about 1980, fallen across nation after nation and trade after trade; and one reading lays the fall at the feet of cheapening machines, which firms now buy in place of workers — making the decline the impersonal work of technology, not the hand of power.

— the argument of this plate, after Loukas Karabarbounis and Brent Neiman, 2014

The constancy of labor's share — Douglas's celebrated finding, the rock on which the productivity theory of distribution was built — turns out to have been the feature of an era, not a law for all time. Since around 1980, labor's share has fallen, and not in one country alone but across most of the world. Loukas Karabarbounis and Brent Neiman documented the global scope of that decline, and offered an explanation that, strikingly, swings the argument back toward the productivity side: that the fall is the work not of power but of cheapening machines. The plate confronts the great fact the rest of the volume must reckon with — that the share, long thought fixed, has moved — and the first and most influential account of why, an account that locates the cause in technology rather than in any shift of power between labor and capital.

For most of the twentieth century, labor's share of national income hovered near its Bowley's-Law value, varying with the business cycle but always returning to it, and this stability was taken to be deep — a fact about the nature of production. Then, beginning in the early 1980s, it broke. Across the United States, across Europe and Japan, across many emerging economies, and within the great majority of individual industries taken one by one, labor's share of income declined — by several percentage points globally, a large movement in a number once thought fixed. Karabarbounis and Neiman's signal contribution was to show how general the decline was: not an American peculiarity, not confined to a handful of sectors, not an artifact of shifting from labor-heavy to capital-heavy industries, but a broad, global, within-industry phenomenon that any adequate theory of distribution would now have to explain. The decline is, all by itself, a refutation of the simplest productivity story, for if the share were truly pinned by the production function at its three-quarters value, it could not have moved at all.1

1. For most of the twentieth century labor's share hovered near its Bowley's-Law value, varying with the cycle but returning — a stability taken as deep. From the early 1980s it broke: across the US, Europe, Japan, emerging economies, and within most individual industries, labor's share fell several points globally. Karabarbounis and Neiman showed the decline was general — not American, not a few sectors, not a shift between industries, but broad, global, and within-industry. By itself it refutes the simplest productivity story: a share pinned by the production function could not move.

Karabarbounis and Neiman traced the decline to a specific and technological cause: the falling price of capital goods. Over recent decades, the relative price of investment goods — machinery, and above all the equipment of the information age, computers and their kin — has fallen dramatically, as we have grown ever more skilled at producing them. Cheaper capital invites firms to substitute it for labor, to buy the machine rather than hire the worker. And here a single parameter becomes decisive: the elasticity of substitution between capital and labor, which measures how readily one can be swapped for the other. If capital and labor are gross substitutes — if a fall in the price of capital draws in more than proportionally more of it — then the cheaper, more abundant machines capture a rising share of income, and labor's share correspondingly falls. This is the mechanism they propose, start to finish: the price of capital fell; firms, finding substitution easy, shifted toward it; and because the substitution was easy, capital's share rose while labor's declined. The great fall is explained without any firm gaining market power, any union being broken, or any politics shifting at all.2

2. Their cause is technological: the falling price of capital goods. The relative price of investment goods — machines, above all computers — has fallen sharply as we produce them better. Cheaper capital invites firms to substitute it for labor. The decisive parameter is the elasticity of substitution: if capital and labor are gross substitutes (a price fall draws in more than proportionally more capital), cheaper machines capture a rising share and labor's falls. Price fell, firms substituted, capital's share rose — no power shift required.
⁂

What makes this account important for the volume is its character. It is, fundamentally, a productivity-and-technology explanation of the falling share — a story in which the decline is the impersonal working of technical change, and not the hand of power at all. No firm needed to gain market dominance, no union needed to be broken, no law needed to be rewritten in capital's favor; cheaper machines and the easy substitution of capital for labor were sufficient by themselves. In this reading the decline is the marginal-productivity theory still faithfully operating, only with a moving production function: as capital cheapens and its quantity swells, the marginal-product logic itself reassigns the shares, handing more to the now-abundant machines and less to labor. It is the productivity answer updated for an age of falling machine prices, and it stands as the principal rival to the power explanations the volume has built — Kalecki's markup, Goodwin's struggle — locating the cause of the great decline squarely in technology rather than in the balance of class power. The same fact that the power theories claim as their vindication, the technological theory claims as its own.3

3. The account's character matters: it is a productivity-and-technology explanation of the fall, not power. No firm gained dominance, no union broke, no law was rewritten — cheaper machines and easy substitution sufficed. The decline is the marginal-productivity theory still operating, with a moving production function: as capital cheapens and swells, the marginal-product logic reassigns shares toward machines. The productivity answer, updated — the chief rival to Kalecki's markup and Goodwin's struggle, claiming the same fact for technology.

But the technological account is contested at nearly every joint, and the disputes return power to the picture. The crucial parameter — an elasticity of substitution greater than one — is far from settled; many careful estimates find it less than one, which would predict that cheaper capital raises labor's share, the exact opposite of what occurred. And the capital whose share rose, examined closely, proves a suspicious category. Much of the measured increase turns out to be not productive machinery at all but housing — the soaring value of land and dwellings — or intellectual property, or, most tellingly, pure profit, the rents earned by firms with market power. If the rising capital share is in substantial part a rising profit share, then the decline of labor's share is not the work of cheap machines but of the very market power that Kalecki described: concentration, fattening markups, the long weakening of labor. The technological story and the power story thus contend over a single fact, each able to point to evidence, and the measured numbers do not, by themselves, declare a winner. Even the cause of the decline is a battleground between productivity and power.4

4. The technological account is contested. The crucial elasticity above one is unsettled — many estimates find it below one, predicting the opposite. And the "capital" whose share rose is suspicious: much is housing, or intellectual property, or pure profit — the rents of market power. If the rising capital share is largely a rising profit share, the decline is Kalecki's power story — concentration, markups, weakened labor — not cheap machines. Technology and power contend over one fact; the numbers do not decide.

So the volume arrives at its central tension, now fully formed. Labor's share has fallen — that much is beyond serious dispute. But why is the very question the whole volume has been circling: is the fall the impersonal work of productivity and technology, as Karabarbounis and Neiman's cheap machines would have it, or the work of power — rising concentration, vanishing unions, the political tilt toward capital — as Kalecki's markup and Goodwin's struggle would suggest? The honest answer is that both forces are at work, and that disentangling their contributions is among the hardest problems in all of economics, for the same data are consistent with either emphasis. What the flagship can do is step back from the measurement and ask the deeper question the decline forces upon us: whether the division of the product is governed, in the end, by laws of nature that we merely discover and must accept, or by arrangements of power and politics that we choose and could remake — and what the long sweep of capital's share, traced across centuries rather than decades, reveals about which. The next plate is the divergence.5

5. The central tension, fully formed. Labor's share has fallen — beyond dispute. But why? The impersonal work of technology (cheap machines), or of power (concentration, vanishing unions, the tilt toward capital)? Both operate, and disentangling them is among economics' hardest problems — the same data fit either emphasis. The flagship steps back to the deeper question: is the division a law of nature we must accept, or an arrangement of power and politics we could remake — and what does capital's share across centuries reveal? Next: the divergence.

The constancy of labor's share was the feature of an era, and the era has ended. Since around 1980 the share has fallen across most of the world and most industries, refuting the simplest productivity story, for a share pinned by the production function could not move. Karabarbounis and Neiman traced the fall to cheapening capital goods — machinery, and above all computers — substituted for labor wherever substitution was easy, a technological account that lays the decline at the door of technical change rather than power. But the account is contested at every joint: the crucial elasticity may run the other way, and the capital whose share rose may be not machinery but housing, or intellectual property, or profit — the rent of market power. Technology and power contend over the same fact, and the numbers alone do not decide between them. The next plate is the divergence.