The Share

The discipline that now spends its days on allocation — how scarce means are matched to competing ends — was founded on a different question, which its founders thought the deeper of the two. David Ricardo stated it at the opening of his great work and repeated it in his letters with the emphasis of a man naming his life's subject: to determine the laws that regulate the distribution of the produce of the earth among the three classes of the community is, he wrote, the principal problem in political economy. Not how the produce is made, nor how it is exchanged, but how it is divided — among the landlord who takes rent, the capitalist who takes profit, and the laborer who takes wages. For Ricardo the question of distribution was not a corollary of the theory of value; it was the point of having a theory at all.

The modern subject keeps the question but simplifies its terms. Land, which loomed so large for Ricardo's agrarian Britain, is folded into capital, and the three-class division becomes a two-way split: labor's share of the national product against the share of capital — wages and salaries on one side, profit, interest, and rent on the other. The figure shows the division as a single bar, the whole of what a nation makes carved into the portions its makers and owners carry off. The labor share is the fraction of that bar painted as wages, and the whole of this volume is an inquiry into what determines where the dividing lines fall — and, more pointedly, into whether they fall where they do because of what each factor contributes, or because of the power each can bring to bear.

It is worth pausing on why distribution, and not allocation, sat at the center for the classics, because the reason is the volume's reason too. Allocation is a question about efficiency — about getting the most from given means — and on efficiency, as The Surplus argued at length, a competitive economy does remarkably well. Distribution is a question about justice and about power, and on those a competitive economy is, as that volume also argued, silent: it can be perfectly efficient and grotesquely unequal in the same breath. The classics kept distribution central because they understood that the size of the product and its division are different questions, and that the second is the one over which societies actually fight. This volume is the table's fullest reckoning with that second question, asked of the whole economy at once — and it receives that question, already sharpened, from the two volumes that name it as their destination.

For the first half of the twentieth century, the answer to Ricardo's question appeared to be that there was no question — that the division of the product was simply, mysteriously, fixed. Arthur Bowley, assembling the statistics of British national income, found labor's share so steady across the decades that the regularity took his name: Bowley's Law, the constancy of the share of wages in the national product. The finding was striking precisely because nothing in theory required it; the share might have drifted anywhere, and instead it sat, decade upon decade, near the same two-thirds.

Nicholas Kaldor raised the observation to a principle. Among the half-dozen "stylized facts" of economic growth that he catalogued in the late 1950s — the regularities any theory of growth ought to reproduce — were the rough constancy of the rate of profit, of the ratio of capital to output, and, crowning them, of the shares of labor and capital in national income. The facts were stylized, idealizations meant to be roughly rather than exactly true, but they hardened into foundations. The workhorse models of modern growth were built to honor them; and the constancy of the share in particular did a great deal of quiet ideological work, for it made distribution look like a parameter of nature rather than an outcome of struggle. If labor's share is fixed, then labor and capital rise together, each taking its appointed cut of an expanding pie, and there is nothing much to fight about — and nothing much for economics to say.

John Maynard Keynes, contemplating the stability of labor's share in 1939, called it a bit of a miracle — and the phrase is more apt than he intended, for miracles, examined closely, tend to dissolve into ordinary events misdescribed. The constancy would prove vulnerable on three fronts, each developed in a later plate. It may have been, in part, an artifact of the mathematics economists used to measure it, a constancy assumed by the production function and then "discovered" in the data it had been fitted to — the burden of the next plate. It was, even taken at face value, the constancy of a particular era and a particular set of institutions — the mid-century decades of strong unions, regulated finance, and managed trade — rather than a timeless law. And it has, in any case, ended: since around 1980 the labor share has fallen across the advanced economies, which settles the matter, for whatever held the share roughly steady for a season was plainly contingent, and not a law at all. The miracle expired; and its expiry is the event this volume exists to explain.

The constancy of the previous plate has a mathematical twin, and the resemblance is so close as to be nearly an identity. In 1928 Charles Cobb and Paul Douglas proposed the production function that bears their names — output as a product of capital and labor each raised to a fixed power — and fitted it to the data of American manufacturing. Under competitive factor pricing, that function has a property that should give the reader of the last plate pause: the share of each factor is exactly the exponent attached to it, and so is constant, no matter how much capital is accumulated relative to labor. Pile capital ever higher; the labor share does not budge. The Cobb–Douglas function does not merely permit a constant share — it guarantees one.

This is the first crack in the miracle. If the function economists chose to describe the economy has constant shares built into its very form, then finding constant shares in data fitted with that function is perilously close to finding what one assumed. The suspicion was sharpened by Henry Phelps Brown and, more recently and more sharply, by Jesus Felipe and John McCombie: because the national accounts satisfy an identity — value added is, by definition, the sum of the wage bill and profits — a Cobb–Douglas regression can appear to fit the aggregate data, and to "confirm" both marginal-productivity pricing and constant shares, even when no aggregate production function of that kind exists at all. The good fit may register the accounting identity rather than any law of technology. The constancy of the share, in short, was not so much a discovery about the world as a property of the curve drawn through it.

To see what is at stake one must let the function generalize. The Cobb–Douglas is the special case, in a wider family, where the elasticity of substitution between capital and labor — the ease with which one can be swapped for the other as their relative price changes — is exactly one. The figure shows why that single number governs everything. If the elasticity is one, capital deepening leaves the share untouched, and the economy keeps its constant share by construction. If the elasticity exceeds one, capital and labor are easy substitutes, and accumulating capital — even as its marginal product falls — raises capital's share and lowers labor's. If the elasticity is below one, they are complements, and accumulation does the reverse, raising labor's share. Whether the relentless accumulation of capital shifts the division toward capital or toward labor is thus not a matter of principle but of a single empirical parameter — and the answer decides how to read the modern decline.

This is precisely the reading that Loukas Karabarbounis and Brent Neiman offered, in 2014, for the global decline of the labor share. The relative price of investment goods has fallen steeply for decades — machines, and above all computers, have grown cheap — and if capital and labor are gross substitutes, with an elasticity above one, then cheap capital is substituted in for labor, and capital's share rises as labor's falls. The decline, on this account, is not a failure of competition or a shift of power but the orderly response of a competitive economy to a change in relative prices, working through an elasticity that happens to exceed one. It is the orthodoxy's own explanation of the fall — and, like the orthodoxy, it stands or falls on a parameter the profession has not agreed upon.

The orthodox theory of the functional distribution is the second plate of The Wage raised to the scale of the whole economy. John Bates Clark extended the marginal principle from the wage of labor to the return of every factor: each is paid the value of its marginal product, so that capital's share is the marginal product of capital times the quantity of capital, divided by output, and labor's share is the corresponding expression for labor. The division of the product is, on this account, a matter of pure technology — the partial derivatives of the production function, evaluated at the quantities employed — and Clark, never shy of the moral implication, drew it out: each factor, and so each class, receives precisely what it contributes to the product. Distribution is not robbery but desert; the share is earned.

An obvious worry attends the theory, and its resolution is one of the prettier facts in economics. If every factor is paid the value of its last unit's contribution, why should those payments happen to use up exactly the whole product — no more, leaving the firm insolvent, and no less, leaving an unexplained surplus? Philip Wicksteed posed this "adding-up problem" in 1894, and the answer is a theorem of pure mathematics, due to Euler. If the production function exhibits constant returns to scale — if doubling every input exactly doubles output, the natural assumption for an economy that can replicate its activities — then the sum of each input multiplied by its marginal product equals total output, identically. Pay each factor its marginal product under constant returns, and the product is exhausted to the penny: the shares are the output elasticities of the factors, and they sum to one, with nothing left over and nothing wanting. The theory is, in this precise sense, internally consistent.

The achievement is genuine, and it should be granted before it is qualified: here is a complete and consistent account in which the entire product is explained by the contributions of the factors that made it, the arithmetic closing exactly. But the authority the theory has carried rests on two foundations, and both will give way. The first is competition — that factors are paid their marginal products requires the price-taking, frictionless markets that The Wage's monopsony plate and this volume's plate on market power both deny; where employers or firms have power, factors are paid less than their products, and the adding-up leaves a residue of pure profit the theory cannot name. The second foundation is deeper and stranger: to speak of "the marginal product of capital" at all, one must be able to measure the quantity of capital whose product is in question — and that, the next plate shows, cannot in general be done. Clark's mathematics is sound. His leap from paid its marginal product to paid what it deserves is a moral claim wearing a theorem's clothes; and even the theorem, at the level of the aggregate, rests on a measurement that the most rigorous critics of the theory were about to prove impossible.

The orthodox theory of the share rests, at the aggregate, on a quantity it cannot define, and the demonstration of this — conducted between Cambridge in England and Cambridge in Massachusetts across the 1950s and 1960s — is the most consequential and least known episode in the modern theory of distribution. The difficulty is deceptively simple to state. To say that capital earns its marginal product, one must speak of the quantity of capital whose marginal product is in question. But capital is not a substance; it is a heterogeneous heap of produced things — looms, furnaces, buildings, code — that can be added into a single magnitude only by valuing them, in money. And the value of a capital good is the present worth of the future earnings it yields, which depends on the rate of profit used to discount them. The quantity of capital thus depends on the rate of profit — and the rate of profit is precisely what the marginal product of capital was supposed to determine. The theory must know the rate of profit before it can measure the capital whose marginal product is meant to explain the rate of profit. It reasons in a circle.

Joan Robinson opened the attack in the 1950s; Piero Sraffa supplied, in 1960, the rigorous apparatus; and the younger economists of the English Cambridge — Pasinetti, Garegnani, and others — pressed it home. The defenders, at the Massachusetts Institute of Technology, were the most formidable in the profession: Paul Samuelson and Robert Solow. Samuelson tried to rescue the aggregate by constructing a "surrogate production function," an as-if aggregate that would behave as the simple theory required. And then the English Cambridge produced the results that broke it — results so counterintuitive that they had to be demonstrated by explicit example before they were believed.

The first result is reswitching. A technique of production — a particular recipe of inputs — may be the most profitable choice at a low rate of profit, give way to a different technique as the rate of profit rises, and then return as the most profitable choice when the rate of profit rises higher still. The figure shows it: technique β is chosen at both ends of the range, α in the middle, the economy switching from one to the other and back. The second result, capital-reversing, is its companion: a lower rate of profit, which the simple theory says should coax the economy toward more "capital-intensive" methods, can instead be associated with less capital per worker. Together these break the spine of the marginal-productivity theory. That theory needs the demand for capital to slope downward — a lower price of capital (a lower rate of profit) calling forth a greater quantity — exactly as the demand for any factor is supposed to. Reswitching and capital-reversing show that the relationship need not be monotonic at all: one cannot line techniques up from less to more capital-intensive as the rate of profit falls, because the same technique can sit at both ends of the line. There is no well-behaved demand curve for aggregate capital, and so no reading of capital's share off its scarcity.

What is remarkable is that the defenders conceded. Samuelson, in a 1966 paper with the wry title of a summing-up, granted that reswitching is logically possible, that the phenomena his critics described were real, and that the simple parable told by the founders of the marginal theory — capital as a single substance whose return measures its scarcity — could not be universally valid. It was as clean a surrender as the discipline has witnessed, and its implication for this volume is exact and large. If there is no coherent aggregate quantity of capital, then there is no aggregate marginal product of capital, and the claim that capital's share of the national product is fixed by technology — that the division between wages and profits is read off the production function — has no general foundation. The orthodox account of the share, so consoling and so elegant, was shown by its own most rigorous practitioners to rest on a quantity that cannot in general be made to exist.

The doors this opens are the subject of the plates that follow. If technology does not fix the division between wages and profits, then something else must — and the candidates are the ones this table has been assembling all along: the classical surplus, with its open degree of freedom; the balance of power between organized labor and organized capital; the degree of monopoly in the product market. The capital controversy did not prove that distribution is governed by power. It proved that the theory which said distribution is governed by technology cannot, in general, be true — and that is enough to make the question of power live again. The Wage and The Surplus both flagged this controversy at their edges; here it stands at the center, the hinge on which the whole inquiry into the share turns from technology toward power.

If technology cannot pin the division between wages and profits — and the capital controversy has just shown that, in general, it cannot — then the question becomes what does, and the classical tradition, made rigorous by Piero Sraffa, gives the cleanest answer the subject possesses. In his treatise of 1960 on the production of commodities by means of commodities, the same work The Surplus drew upon, Sraffa showed that once the technical methods of production are given — the recipe of inputs for each output — the system determines all relative prices and one of the two distributive variables, but only once the other distributive variable is supplied from outside. Fix the real wage, and the rate of profit and all prices follow; fix the rate of profit, and the real wage and all prices follow. The technology determines everything but the division itself. There is exactly one degree of freedom, and the economy's own mechanics do not close it.

The figure gives the degree of freedom its shape. For a given technique there is a frontier — a downward-sloping tradeoff between the real wage and the rate of profit, the one rising only as the other falls. At one extreme, a rate of profit of zero, the whole net product accrues to labor as wages; at the other, a wage of zero, the whole surplus accrues to capital as profit; and between these limits the division slides continuously. The frontier is a zero-sum boundary: given the technique, every gain to labor is a loss to capital, and conversely. Technology, having drawn this frontier, falls silent. It does not say where on the frontier the economy will sit — and since, by the previous plate, no aggregate marginal product can be appealed to, the point must be selected by something the technology does not contain.

What selects the point is, in this tradition, social and institutional. One may take the real wage as given — set by the historical and customary condition of the working class, by what struggle and convention have made the accepted standard of life, the socially determined subsistence of The Wage's third plate — and let the rate of profit fall out as the residual. Or one may take the rate of profit as given — anchored, some Sraffians argue, by the rate of interest that monetary authorities set — and let the real wage fall out instead. Either way, a distributive variable is fixed from outside the economy's technology, by forces of power, custom, or policy, and the whole division follows from it. This is the formal vindication of everything Smith and Ricardo intuited: the frontier is technology's contribution, the point upon it is power's, and the share is what power, operating within the limits technology sets, finally chooses. It is worth noting, too, that Sraffa built his analysis around a constructed "standard commodity," a composite designed to serve as an invariable measure of value against which the wage–profit tradeoff could be read without distortion — which is to say he was hunting, like this whole series, for an honest numéraire.

Grant the previous plate — that distribution has a degree of freedom technology leaves open — and the most natural candidate to fill it is the one Smith named at the very start: the balance of power between those who hire and those who work, now writ across a whole economy. Marx gave the mechanism its enduring image. Unemployment, he argued, is not merely a personal misfortune but a systemic instrument: a "reserve army of labor," a standing pool of the unemployed, disciplines those in work, tempers their demands, and holds the wage — and the wage share — down. Some slack in the labor market is not a malfunction from capital's point of view; it is a source of capital's bargaining power. It follows that the share should move with the tightness of the labor market: when unemployment is low and the reserve army thin, labor's hand strengthens and its share rises; when unemployment is high, the reverse.

Michał Kalecki drew the political conclusion with a clarity that still startles. In a famous essay of 1943 on the politics of full employment, he reasoned that if governments learned to maintain genuine full employment, the discipline of the sack would vanish, workers' confidence and bargaining power would swell, and the wage share would climb — and that, precisely for this reason, the owning and rentier interests would come to oppose full-employment policy, not on economic grounds but on grounds of class power, generating a political cycle in which booms are cut short before labor grows too strong. The level of unemployment a society chooses to tolerate is, on this view, not a technical datum handed down by the natural rate but a distributional decision in disguise — a verdict on the share, enforced through the macroeconomy. And Kalecki gave the share a second determinant that the rest of this volume will magnify: the "degree of monopoly," the markup of prices over costs that firms' market power allows. The greater that markup, the smaller the wage share — distribution as a function of how much power firms wield in their markets.

This is the bargaining theory of The Wage and the division of The Surplus lifted to the aggregate. The point on the frontier — labor's share of the whole product — is settled, on this account, by the institutions and the balance of organized power: by the density of unions and the reach of collective bargaining, by minimum wages and employment protections, by the level of unemployment the central bank and the government permit, and by capital's own outside option, the ever-present threat to move production somewhere cheaper. When those forces favor labor, the share rises; when they favor capital, it falls. And the matter does not stop at fairness, for distribution feeds back into growth. The post-Keynesian economists Amit Bhaduri and Stephen Marglin distinguished "wage-led" from "profit-led" economies: in the first, a higher wage share, by financing mass consumption, raises demand and growth; in the second, a higher profit share, by financing investment, does. Whether redistributing toward labor expands or contracts the pie is thus itself contingent — a question this volume hands on to The Growth.

The theoretical plates have established that the share could move — that no law of technology pins it, and that power may fill the degree of freedom that technology leaves open. The empirical plates now establish that it has moved, and decisively. Since around 1980, across the United States and most of the advanced economies, labor's share of the national product has fallen — by several points of the whole, the precise figure depending on choices the eleventh plate will scrutinize. Loukas Karabarbounis and Brent Neiman documented the decline as a global phenomenon, present in the majority of countries and the majority of industries; Michael Elsby and his colleagues traced its American course, with much of the drop concentrated after 2000. The miracle Keynes named has plainly expired.

The decline is broad — spread across countries and across industries — and that breadth is itself a clue, for it argues that common, global forces are at work rather than the peculiarities of any one nation's laws. The candidate explanations are, almost exactly, the plates of this volume and its companions, each a different answer to the question of what filled the open degree of freedom and pushed the point down the frontier. There is the cheapening of capital working through an elasticity of substitution above one, the orderly-competitive story of the third plate. There is globalization — the entry of hundreds of millions of low-wage workers into the world economy, the offshoring of labor-intensive production, the "China shock" — which weakened the outside option of labor in the rich countries and shifted the bargain against it. There is automation, the machine of The Wage and The Surplus, displacing labor from the tasks capital learned to do. There is the decline of unions and the broad erosion of worker bargaining power, the conflict story of the seventh plate, which some economists now argue is the largest single factor. There is the rise of market power, the subject of the next plate. And there is measurement, the subject of the eleventh — the possibility that some of the decline is an artifact of how the share is counted.

These explanations are not rivals to be adjudicated so much as forces to be weighed, for they operate at once and reinforce one another, and the share is, as the seventh plate's edge warned, over-determined. Cheaper capital and weaker unions and fiercer import competition and greater corporate power all push the same way, and disentangling their contributions — assigning the fall its percentages of cause — is the hardest and most contested work in the field. What is not contested is the direction and the rough magnitude: labor's portion of the product has shrunk, and with it the comfortable mid-century picture of labor and capital rising in lockstep. The question Ricardo placed at the center of the subject, and that the constancy doctrine had lulled the profession into ignoring, is once again the live question — and the explanations on offer divide, almost without remainder, into those that credit productivity and those that credit power.

The instinctive reading of a falling labor share is that capital's share must have risen to meet it — that the loss to labor is the gain to the owners of machines and buildings. The most important recent work on distribution suggests this reading is largely wrong, and that what rose was something the two-factor framing of the first plate cannot name: not the return to capital, but pure profit, the rent of market power. Jan De Loecker, Jan Eeckhout, and Gabriel Unger estimated that the average markup of price over marginal cost in the American economy rose from about a fifth above cost in 1980 to something over half above cost by the 2010s, the increase concentrated in the most powerful firms. A rising markup is, by definition, a rising share of revenue that goes neither to labor nor to the replacement and return of capital, but to profit in excess of both — economic rent, the surplus that The Surplus taught us to watch.

The mechanism by which this lowers the labor share was anatomized by David Autor and his colleagues under a vivid name: the rise of the superstar firm. Industry after industry has grown more concentrated, "winner-take-most," dominated by a handful of extraordinarily productive giants that operate at unusually low labor shares — vast sales spread over relatively few workers, with much of the value flowing to profit, to intellectual property, to the platform itself. As economic activity reallocates toward these superstars, the labor share of the whole economy falls, even if the labor share within the typical firm holds steady; it is the shift of weight toward the low-labor-share giants, the between-firm reallocation, that drags the aggregate down. The fall of labor's share and the rise of corporate concentration are, on this account, one phenomenon seen from two sides.

The reframing this forces is the deepest point of the plate, and it dissolves the very dichotomy the volume began with. When the return to capital is measured properly — using the genuine, and falling, cost of capital rather than treating all non-labor income as capital's — it emerges, in Simcha Barkai's work, that both the labor share and the true capital share declined over recent decades, while the share of pure profit rose substantially. The portion that labor lost did not, for the most part, become capital's; it became rent — the profit of market power, accruing to the owners of brands, patents, networks, and platforms. The contest, properly seen, is therefore not labor against capital at all but labor, together with productive capital, against the rentier: against those who collect the surplus that power, not contribution, commands. This carries the inquiry straight to The Power, the table's tenth element, and to The Rent, its second; and it joins the markup on prices to the markdown on wages of The Wage's eighth plate, for a firm with power in its product market often wields power in its labor market too, and the two are a single scissors, cutting the labor share from both blades at once.

Everything so far has concerned the functional distribution — the division of the product between labor and capital as factors, abstract categories of production. But people do not live as factors; they live as persons, in households, and what they care about is the personal distribution: how income and wealth are spread across the population, how far the top stands above the bottom. For much of the twentieth century these were treated as two separate inquiries, the functional belonging to the theory of production and the personal to the sociology of inequality. The signal achievement of recent work is to have shown that they are one inquiry, because the bridge between them is a single fact: capital is owned far more unequally than labor is supplied.

That fact is the hinge. Labor income, however unequal, is spread across the great mass of people who work; capital income flows overwhelmingly to the top, for the ownership of wealth is concentrated at a height that the distribution of wages never approaches. It follows immediately that a shift in the functional share — from labor toward capital, or toward the pure profit of the ninth plate — is a shift of income toward the already-rich, and so a rise in personal inequality, by sheer arithmetic of who owns what. The falling labor share is not merely a fact about factors; it is rising personal inequality wearing a macroeconomic disguise. To explain the explosion of top income shares since 1980, one cannot avoid the falling labor share; to explain the falling labor share's bite, one cannot avoid the concentration of capital ownership. The two distributions have re-fused.

It is Thomas Piketty who, in 2014, gathered this into a synthesis on the grand scale, with two centuries of data behind it. The U-curve of the figure is its emblem: top income and wealth shares stood high before the First World War, were compressed to a mid-century low by the shocks of depression, war, and inflation and by the policy response of progressive taxation, strong unions, and the welfare state, and have climbed again since about 1980 toward their old summits. Around this history Piketty wrapped a theory. Capital's share, he wrote, is the product of the rate of return on capital and the capital–income ratio — the stock of wealth measured in years of national income — and that ratio tends, in the long run, toward the saving rate divided by the growth rate, so that when growth is slow, wealth piles up to a high multiple of income. And his famous inequality, that the rate of return on capital has historically exceeded the rate of growth, implies that accumulated and inherited wealth grows faster than the economy, concentrating ownership and returning society toward the patrimonial capitalism of the nineteenth century, in which inheritance, not earnings, commands the heights. The mid-century compression, on this telling, was the anomaly — the product of catastrophe and reform — and the rising inequality of our own age the resumption of capitalism's deeper tendency.

The figure above tracks one slice of the distribution — the top hundredth — across a century. This one shows the whole of it at a single moment, and lets you set how unequal it is. The straight line is perfect equality: the poorest tenth of people would hold a tenth of the income, the poorest half exactly half. The bowed curve is a real distribution, in which the poorest hold less than their number and the richest far more; the further it sags from the straight line, the more unequal the society, and the shaded crescent between the two — its size as a fraction of the whole lower triangle — is precisely the Gini coefficient. Drag the inequality up and watch the curve bow, the crescent swell, and the share left to the bottom collapse.

Inequality (Gini)

Richest tenth holds—

Poorest half holds—

Poorest fifth holds—

Richest hundredth—

Two corrections keep the plate honest, and both lead back into the volume. The first is that Piketty's engine quietly depends on the same contested parameter as the third plate's: for capital's share to rise as wealth accumulates, the return on capital must not fall too quickly as capital grows abundant — which requires an elasticity of substitution above one, exactly the disputed assumption of Karabarbounis and Neiman. If capital and labor are complements rather than easy substitutes, accumulation would erode the return faster than the stock rose, and capital's share would fall, not climb. The second correction is Matthew Rognlie's, and it is sharper still: the measured rise in capital's share over recent decades is concentrated almost entirely in housing — in the imputed rents of owner-occupied homes and the soaring value of residential land — so that, net of housing, the long-run rise in the share of productive capital is small. If so, the story is less Piketty's tale of machines out-earning workers than a Ricardian tale of land and scarcity, which hands the question to The Rent. And there is a third complication the next plate will press: a large part of top incomes, especially in America, is recorded not as capital income at all but as labor income — the soaring pay of executives, financiers, and the "working rich," the superstar dynamics Sherwin Rosen described, in which scale lets the best-placed performers command the whole market. The top of the distribution scrambles the categories: its income is booked as wages but functions, often, as the rent of position and power.

Ricardo placed the division of the product at the center of the subject, and this volume has circled back to him by the long road of the modern theory. The road's lesson can now be stated. The share of the national product that goes to labor was thought, for the better part of a century, to be roughly constant — a benevolent miracle, a sign that labor and capital rose together and that the division was a settled fact of nature. The constancy was in part an artifact of the very function economists fitted to the data, which had constant shares built into its form; and it has, in any case, ended, the share falling across the advanced world since around 1980. The orthodox theory that would have explained the constancy — each factor paid its marginal product, the division read off the production function, the whole arrangement technologically given and, the orthodoxy added, deserved — was shown by its own most rigorous practitioners, in the capital controversy, to rest on a notion of aggregate capital that cannot in general be made coherent. There is no well-defined marginal product of capital in the aggregate, and so no technological law that fixes capital's share.

What that demolition leaves is a frontier and a freedom. Technology, in the classical-Sraffian reckoning, draws the wage–profit frontier — the menu of divisions a given state of the arts permits — but it does not choose the point upon it. That single degree of freedom is filled from outside, by the balance of power: by the size of the reserve army of the unemployed and the level of slack a society tolerates, by the strength of unions and the reach of collective bargaining, by the threat of capital to move and the worker's freedom to refuse, by the degree of monopoly firms wield in their markets. The long decline of the labor share is the macroeconomic record of that balance shifting against labor — and its leading modern explanation, the rise of market power and the superstar firm, recasts the fall not as capital's gain but as the rising share of pure profit, the rent of power, so that the contest is revealed to be not labor against capital but labor, together with productive capital, against the rentier. And because the ownership of capital and the collection of rent are concentrated at the very top, the falling labor share is, at the same time and by the same arithmetic, the return of an inequality among persons not seen since before the First World War.

So the question the volume opened with — is the split between labor and capital set by productivity, or by power? — can be answered, and the answer is the one this whole table keeps arriving at: by both, but not equally, and not in the way the orthodoxy supposed. Productivity, through technology, draws the frontier of the possible; power chooses the point upon it. Where the orthodox conditions hold — true competition, a coherent production function — the share tracks the marginal products and wears the face of nature. Where they fail, as the capital controversy proved they must in general and as monopsony and market power show they do in fact, the share is filled by the strength of the parties, and its movements record not changing contributions but changing power. To read the falling labor share as a law of technology — as the orderly response of a competitive economy to cheaper machines — is to mistake one possible reading for the only one, and to perform, perhaps without intending it, the oldest ideological service economics can render: the naturalization of a distribution of power as a fact of nature. The constancy doctrine performed that service by making the division seem fixed; the marginal-productivity gloss performed it by making the division seem deserved. Undoing both is the purpose of this volume.

And here, at the scale of a whole economy, the two volumes that pointed to this one converge. The Wage asked what sets the price of labor, and found that the wage sits within a range that bargaining power decides, the equality of wage and marginal product a property of a frictionless world that no real labor market reaches. The Surplus asked who keeps the value an exchange creates, and found that economics can size the surplus with precision but falls silent on its keeping, which power decides. The Share is the ratio of the one to the other, aggregated and made national: the wage is what labor receives; the product, surplus and all, is what labor helps to create; and the share is the distance between them, summed across the whole economy. That distance — between what labor makes and what labor gets — is the single contest this entire table was built to read, and the labor share is its largest and most legible measure. It is, in the strict sense the series intends, the greatest of all the prices: the price of an entire class's labor, set against the whole of what that labor helps to produce, and quoted, like every price, as a verdict on a hidden contest rather than a fact of nature.

The verdict, being the largest, hands on the most. It hands to The Power, the tenth element, the structures — monopoly, monopsony, concentration, and the law that permits or restrains them — that confer the strength which fills the open degree of freedom. It hands to The Rent, the second, the rising portion of the product that is neither wage nor return to productive capital but pure rent, and the land and housing in which Rognlie found so much of it hiding. It hands to The Tax, the seventeenth, the truth that the distribution struck in the market is not the final word, for the state redraws it through taxation and transfer, and the mid-century compression was in no small part its work. It hands to The Growth, the twentieth, the question whether a pie divided more toward labor grows faster or slower — whether the size and the division are, as the old separation assumed, independent, or whether, as the wage-led and profit-led theories insist, they are bound together. And it hands to The Norm, the twenty-sixth, the question economics cannot answer from inside itself: not what the share is, but what it ought to be — what portion a society believes those who labor are owed.

This volume has answered the positive question and declined the normative one, as its discipline must. What portion of the product goes to labor is set, within the limits technology draws, by power — by the bargain struck and re-struck in every market the other volumes describe, summed into a single national figure. What portion should go to labor is a question of justice, and the table can only clear the ground for it, by refusing to let an outcome of power masquerade as a dictate of nature. The share is not nature's; it is ours. It is not given; it is chosen — if not by any of us deliberately, then by all of us together, through the institutions we build and tolerate. And it is not closed but open, as in honesty it must remain — for technology underdetermines it, power fills the gap, the very number is a construct laden with the theory it is meant to test, and the question of what we owe one another is not the kind of question a production function was ever going to settle. That is the whole of what this table has to teach, said once, at the scale of a nation: that the price of labor, like every price, is a verdict on a contest — and that the contest, and so the verdict, remains open.