Who gets what — and what happened to the theory that tried to answer it?
When economics was born, the first question it asked was how the pie gets split. Ricardo called it “the principal problem in Political Economy.” Then the discipline decided the question was solved — factors earn what they add, full stop — and spent a century looking elsewhere. In 2014 a French economist with three centuries of data pulled it back to the center. This follows that one strand — how output gets divided — from Ricardo’s rent to Piketty’s $r > g$, and asks what survives of each answer along the way.
Voir comme graphe de débatHere is the spine the way the intellectual-lineage record holds it — one apparatus, the theory of distribution, traced from Ricardo’s 1817 tripartite split toward the modern empirical program. The era-organized textbooks teach these figures in separate chapters; this is the single argument that runs through all of them. Click any node to see where it sits.
The graph draws the classical-to-modern spine: Ricardo → Marx → the marginalist line through Marshall and Samuelson → Piketty. The middle rungs the thread leans on hardest — J.B. Clark’s marginal-productivity theory, the Cambridge controversialists Robinson and Sraffa, Kuznets, and the Atkinson-Saez data program — live in the prose below, not yet in the graph. Open the full network in the intellectual-lineage graph.
Distribution at the center
“To determine the laws which regulate this distribution is the principal problem in Political Economy.”
— David Ricardo, On the Principles of Political Economy and Taxation, Preface, 1817
Read that as a job description, not a flourish. The founders did not treat the split of output among landlords, capitalists, and workers as one topic among many. They built the entire apparatus to answer it. Distribution was the discipline’s reason for existing — and the rest of this thread is the story of how it stopped being that, and then became it again.
A note on scope before we start. This thread traces the theory of distribution — where the apparatus came from, what broke it, and what it now says output-splitting is. It does not adjudicate what to do about inequality; the policy question — wealth taxes, transfers, the efficiency-equity tradeoff — belongs to Is inequality a problem economics can solve?, which engages that debate directly. Here the policy face is cross-linked, never carried.
Ricardo’s machine has three claimants and one moving part. As population grows, cultivation spreads onto worse and worse land. The gap between what the best land yields and what the marginal plot yields accrues to landlords as rent. Wages are pinned near subsistence. Profit is what is left, squeezed between rising rents and a wage floor that will not fall. Run the logic forward and rent rises, profit falls, accumulation stalls, and the economy drifts toward a stationary state. That tripartite split — rent, wages, profit, each with its own law of motion — is the first general model of who gets what.
On land of quality $i$, output per unit of the same labor-and-capital input is $q_i$. Rent on that land is the surplus over the marginal plot actually in use, $q_m$: $\text{rent}_i = q_i - q_m$. The marginal land earns no rent by definition ($q_m - q_m = 0$). As population forces $q_m$ down, the rent wedge $q_i - q_m$ on every better plot widens — growth transfers the gains to landlords, which is exactly why Ricardo treated rent as a drain on the productive economy rather than a reward to it.
Imagine farmers forced to plant on ever-stonier ground as the population grows. The family that owns the good bottomland does nothing different, but the value of holding it climbs, because everyone else is scratching a living from rock. The landlord’s share rises not from working harder but from owning the scarce thing. Wages stay near the bone; profit gets pinched in the middle. The split is not a moral story yet — it is the arithmetic of scarcity working through who owns what.
There was no separate “distribution chapter” in classical economics, because distribution was the analysis. The full classical lineage — Smith’s engine, Ricardo’s rent and the diminishing returns under it, Malthus’s population floor, Mill’s stationary destination — is walked in History of Economic Thought Ch.3 (Classical political economy); the rent-and-distribution apparatus specifically is its second movement.
The tool underneath Ricardo’s rent — the declining marginal product of a factor applied to a fixed one — is the same one that prices labor and capital in modern micro. Its formal home is Economics Ch.5 §5.4 (Production functions), where “the worst land in use” becomes the marginal unit of any factor.
Take Ricardo’s apparatus at full strength first, because the modern reflex is to file it under “the corn model” and move on, and that reflex throws away the achievement. Ricardo made the split inspectable. Before him, “who gets what” was a matter of moral assertion; after him it was a system you could solve, where each share had a law and the laws moved together as the economy grew. He connected distribution to growth — shares shift in a predictable direction as land gets scarcer — and he handed the era’s hottest political fight, the Corn Laws, an analytical answer about who actually benefits from dear bread. He is also simply correct that scarcity generates rents; that piece never broke. The industrial setting he was writing into, the Britain whose population and factory towns were exploding around the Corn Law debate, is the spine of Economic History Ch.7 (The Industrial Revolution).
Then Marx took the same machine and turned it inside out. He kept Ricardo’s surplus — output over and above what it costs to keep the workforce alive and replaced — but he relocated where that surplus comes from. For Ricardo the wage-profit split is a fact of scarcity, impersonal as the weather. For Marx it is the rate of exploitation: the capitalist buys labor-power at its reproduction cost and pockets the gap between what the worker is paid and the value the worker’s labor produces. Distribution stops being natural arithmetic and becomes a relation of power.
Marx writes the working day as $v + s$: $v$, “variable capital,” is the part that reproduces the worker’s wage; $s$, “surplus value,” is the part appropriated by the owner of capital. The rate of exploitation is their ratio, $e = s / v$. A ten-hour day in which four hours cover the wage and six accrue to the owner gives $e = 6/4 = 1.5$ — a 150% rate. The number is doing one job: it makes the claim “the worker produces more than the wage reflects” into something with a magnitude, locating the conflict inside the split rather than in scarcity outside it.
Argue Marx’s distribution claim at strength and it holds more than the caricature allows. The nineteenth-century industrial worker did produce value the wage did not capture, and the bargaining asymmetry he named — one side can wait, the other has to eat — is real and was later re-formalized by perfectly mainstream tools (monopsony, principal-agent models). What is not being litigated here is whether the labor theory of value works as an account of how prices form; that is a different fight, and its formal coherence and the transformation problem belong to the value-lineage thread (forthcoming) — this stage engages Marx’s claim about the split, the rate of exploitation, not value theory’s internal mechanics. The surplus-value apparatus and the falling-rate-of-profit machinery sit in History of Economic Thought Ch.4 (Marx).
“Labor creates the value, capital just captures it” is half a truth wearing a whole one’s coat
The slogan is Marx’s distribution claim compressed for a placard, and it is live on the modern left every time a strike or a CEO-pay headline trends. It carries a real kernel — the bargaining asymmetry it points at is genuine — bolted to a value-theory claim that does not survive contact with how prices actually form. The trick is keeping the kernel without buying the coat.
Where this leaves us
Both classical answers put distribution at the center, and both left something durable. From Ricardo: scarcity generates rents, and the shares shift predictably as the economy grows — a result that never needed retracting. From Marx: the wage-profit split reflects bargaining power, not only natural scarcity — a point the mainstream later re-formalized once it had the tools. What broke is specific. The labor theory of value as the engine of the split did not survive, for reasons the value-lineage thread carries. And Marx’s immiseration prediction — that capitalism would grind real wages ever downward — was falsified hard: real wages in the advanced economies rose several-fold between his death and 1970, a record Economic History Ch.7 documents. The deepest thing the classicals got right is the thing the next era abandoned: that the division of output is a first-order question. Watching the discipline walk away from it is the next stage.
Three economists working independently in the 1870s, and one American writing in 1899, were about to dissolve the whole class-conflict framing with a single idea: maybe nobody takes anybody’s share. Maybe each factor simply earns exactly what it adds — and if that is true, then “exploitation” is not a crime but a category error.
The marginalist answer and its challenge
“It is the purpose of this work to show that the distribution of the income of society is controlled by a natural law, and that this law, if it worked without friction, would give to every agent of production the amount of wealth which that agent creates.”
— John Bates Clark, The Distribution of Wealth, 1899
Clark is not just saying factors are paid their marginal products. He is saying this is just — that the market hands each agent exactly what it contributes, so the distribution it produces is the one a natural law endorses. That is the strongest possible form of the “no exploitation” answer: not a denial that Marx had a point, but a claim that the point dissolves once you see the mechanism. If Clark is right, the class-conflict framing of Stage 1 is not refuted — it is unnecessary.
The marginal-productivity theory of distribution runs on one rule applied twice. A firm hires any factor up to the point where the last unit’s contribution to revenue equals its price: labor until the marginal revenue product of labor equals the wage, capital until the marginal revenue product of capital equals the rental rate. With competitive markets and constant returns to scale, those payments do something remarkable — they exactly exhaust the output, no surplus left over and no shortfall, every dollar of product traced to a factor that earned it. There is no residual for a class to fight over.
Each factor is paid its marginal revenue product: $w = MP_L \cdot P$ and $r = MP_K \cdot P$. With a constant-returns production function $Q = F(K, L)$, Euler’s theorem gives the product-exhaustion identity
$$F(K, L) = MP_K \cdot K + MP_L \cdot L,$$so factor payments add up to total output exactly — the Wicksteed adding-up result. This is the formal heart of Clark’s claim: under these conditions, “everyone is paid what they add” and “the payments sum to the whole product” are the same statement.
Picture the last worker hired. If they add more to the firm’s revenue than they cost, the firm hires another; if less, it sheds one. Hiring stops exactly where the last worker’s contribution equals their wage. Do the same for the last machine. When every factor is paid the value of its last unit’s contribution, and the recipe for output has constant returns, the payments happen to add up to precisely the output produced — nothing left in the pot. There is no surplus sitting on the table for anyone to have “taken.”
This is the spine of modern factor-price economics, and its formal home is the producer-theory and welfare apparatus in Economics Ch.5 §5.4 and the general-equilibrium treatment in Economics Ch.11 (Advanced microeconomics). The intellectual lineage — the 1871 marginalist break, its formalization through Walras and Marshall, and where Clark’s American variant sits in it — runs through History of Economic Thought Ch.5 (The marginalist revolution).
Can the theory measure the thing it claims pays capital its due?
“The phenomenon of switching back at a very low interest rate to a set of techniques that had seemed viable only at a very high interest rate involves more than esoteric difficulties. It shows that the simple tale told by Jevons, Böhm-Bawerk, Wicksell and other neoclassical writers cannot be universally valid.”
— Paul Samuelson, conceding the reswitching point, Quarterly Journal of Economics, 1966
Start with the theory at its strongest, because it earns it. At the level of a single firm in a single market, marginal-productivity pricing is simply how factor prices work: hire to the margin, pay the margin, and the math of competitive equilibrium does the rest. The product-exhaustion result is not ideology — it is a theorem, and it is what let the discipline drop the class-conflict framing for a reason rather than a preference. And the mainstream did not duck the hard question. Samuelson, the most formidable theorist of the age and the defender of the apparatus, took the Cambridge UK critique seriously enough to run the math himself — and conceded the technical points in 1966. That is the theory at its strongest: an apparatus whose own champion was willing to publish exactly where it failed.
“The production function has been a powerful instrument of miseducation. The student is told to assume all firms alike and to measure capital in ‘quantity of capital’ — and then is hurried on, in the hope that he will forget to ask in what units capital is measured.”
— Joan Robinson, “The Production Function and the Theory of Capital,” Review of Economic Studies, 1953
Now the critique, and it is a real formal win, not a heterodox tantrum. To say “capital earns its marginal product” for the economy as a whole, you need to add up the capital stock into a single number. But the value of that stock depends on the rate of profit — you discount the future earnings of machines to price them — and the rate of profit is precisely the thing marginal productivity is supposed to determine. The reasoning closes a loop: you need the answer to compute the input that gives you the answer. Sraffa’s 1960 demonstration made it concrete with reswitching — a production technique that is optimal at both a low and a high interest rate but not in between — which breaks the tidy story that cheaper capital always means more capital-intensive methods. Cambridge UK won the formal argument, and Samuelson said so in print.
“Marginal productivity is just ideology dressed as math” aims at the right target and overshoots it
The heterodox charge is that the whole apparatus exists to launder the existing distribution as “just.” Clark’s ethical framing gives the charge its hook, and the Cambridge result gives it teeth at the aggregate level. But the theory is not only ideology — at the level it was built for, it is a working description of how factor prices form, and treating it as pure propaganda throws away a true thing to score a rhetorical point.
Where this leaves us
Two findings have to be held at once, and the temptation is to drop one. The marginal-productivity theory survives as a partial framework: in competitive markets, factors do earn roughly their marginal products, and at the individual-firm and individual-market level the apparatus is undisturbed and indispensable. That is the concession the heterodox triumphalist version refuses to make. And yet Cambridge UK won the formal point: you cannot aggregate “capital” into one quantity independent of the profit rate it helps determine, so “capital in general earns its marginal product” rests on assumptions the controversy showed to be unsafe. That is the concession the mainstream-dismissive version refuses to make. Why did the mainstream move on anyway? Not by refuting Cambridge UK — it conceded — but because Sraffa’s positive program, distribution read off the physical input requirements of production, explained less than the marginalist alternative it was meant to replace. A formal win is not the same as a better theory. What the controversy leaves standing is a warning: “the market pays everyone what they are worth” is a defensible claim about competitive individual markets and an unsafe one about the whole economy’s class shares. That is the thread’s first formal correction, and it is a correction about method: aggregate-distribution claims rest on shakier ground than the elegant single-firm story advertises.
By the 1950s, economists were exhausted by the theory wars. So they did what people do when an argument will not settle: they stopped arguing and went to look at the data. And the data seemed to hand them a gift — a clean, hopeful law of how inequality behaves as a country grows rich.
The empirical turn
“The paper is perhaps 5 per cent empirical information and 95 per cent speculation, some of it possibly tainted by wishful thinking.”
— Simon Kuznets, “Economic Growth and Income Inequality,” presidential address to the American Economic Association, 1955
That is the man who gave distribution economics its organizing fact, warning his audience how thin the ground under it was. The fact was the Kuznets curve: inequality rises in early development, then falls — an inverted U you could read off the data without committing to any theory of who deserves what. For forty years it let economists stop arguing about distribution and treat it as a stage a country passes through. The caveat is what makes the inference defensible: Kuznets knew exactly how little his mid-century data could bear, and said so out loud. The field heard the curve and forgot the caveat.
The mechanism behind the curve is concrete. A poor agrarian country is fairly equal in its poverty. As it industrializes, workers move from low-inequality farming into higher-paying but more unequal urban industry, and for a while the gap between the two sectors widens the whole distribution — inequality climbs. Then, as the workforce finishes the move, mass education spreads, unions form, and the industrial wage structure matures, the gap compresses again and inequality falls. Plot inequality against income per head and you get an inverted U with a single hump. There is no formal model under it; the curve is the apparatus — an empirical generalization standing in for a theory.
Stylized, the claim is that inequality is a single-peaked function of income per capita, $I = f(y)$ with $f'(y) > 0$ for low $y$ and $f'(y) < 0$ for high $y$ — rising, cresting at some development threshold $y^*$, then falling. The content is entirely in the shape: one hump, traversed once, the same way by every industrializing country. Everything that follows in this stage is an attack on the claim that the shape is a law rather than a description of one passage of data.
Think of a country emptying its farms into its factories. At first only some people have made the jump to the better-paid city jobs, so the distance between the farm and the factory stretches the income ladder — inequality goes up. Once almost everyone has made the jump and schooling and bargaining catch up, the ladder’s rungs pull back together — inequality comes down. The promise hidden in the picture is the comforting one: keep growing and the inequality takes care of itself.
The arc from Kuznets’s curve to the program that overturned it — the long-run series, the top-income-shares method, the return of distribution as a measured object — is carried in History of Economic Thought Ch.17 (The new synthesis and modern pluralism), where the empirical turn in distribution sits beside the institutions-and-inequality literature.
A law of development, or an accident of when the data was collected?
“One must distinguish between inequalities in the income structure of a stable society and inequalities in the income structure of a society in the process of economic growth. The trend is from widening in the early phases to narrowing in the later phases.”
— Simon Kuznets, AEA presidential address, 1955
Take Kuznets’s inference at strength, because it was not a naive curve-fit. Given the data he actually had — a few decades of series for the United States, Britain, and Germany, all showing inequality falling through the mid-century — the inverted U was the responsible generalization, and the farm-to-factory mechanism under it is real and still operates in early-developing economies today. The curve organized four decades of development economics and gave the field a tractable empirical handle on a question theory had failed to settle. And Kuznets fenced his own claim: he told the profession it was speculation. The error was not his inference; it was the field’s decision to read a hedged hypothesis as a settled law.
“The reduction of inequality in France during the first half of the twentieth century was for the most part the chaotic consequence of war and of economic and political shocks. It was not the gradual, consensual, conflict-free process that Kuznets had in mind.”
— Thomas Piketty, on the long-run top-income series (Atkinson–Piketty top-incomes program)
Then the long-run data arrived and dissolved the law. When Atkinson, Piketty, and Saez reconstructed top-income shares back to around 1900 from tax records, the mid-century “downswing” Kuznets had extrapolated into a curve turned out to be an artifact of a specific catastrophe: the world wars destroyed capital, postwar marginal tax rates ran punitively high, and strong unions compressed pay. Once those conditions reversed after 1980, inequality climbed back toward early-twentieth-century levels. There was never a natural inverted U. There was a war-and-policy U, and Kuznets’s data happened to catch it on the downstroke. The compression he read as the next stage of development was the wreckage of two wars cooling off.
“Inequality naturally falls as countries get rich” is the Kuznets curve still drawing a paycheck
The development-optimist line — grow the economy and the inequality sorts itself out — is the Kuznets downswing repackaged as policy reassurance. It is half-alive: the farm-to-factory compression is real in genuinely early development. But as a forecast for rich economies it has been falsified by the very data that killed the curve, and the post-1980 record runs the other way.
Where this leaves us
The Kuznets curve is superseded as a law. The inverted U describes one episode — the mid-century rich world — under one set of conditions: capital destroyed by war, incomes compressed by high taxes and strong unions. It is not a feature of development. What survives is the narrower true thing: the structural-shift mechanism, farm to factory, really does compress inequality in early industrialization. What broke is the optimistic extrapolation that growth equalizes on its own once a country is rich. This is the thread’s second formal correction, and unlike the Cambridge one it is a correction about the facts rather than the method, and a settled one rather than a live mainstream split — the profession now treats the Kuznets curve as a description of one passage, not a law, and the long-run series settled it. The lesson generalizes past Kuznets, and it is the hinge of the whole thread: distribution does not follow a natural curve. It follows institutions, shocks, and policy. Which means the classical question — who gets the surplus, and why — was never answered when the discipline shelved it. It was only postponed.
If inequality is a policy outcome and not a natural law, then the question the classicals asked — who gets the surplus, and why — was only shelved, not solved. In 2014 a French economist with three centuries of reconstructed data pulled it back off the shelf, and it became the best-selling economics book of the decade.
The Piketty moment
“When the rate of return on capital exceeds the rate of growth of output and income, as it did in the nineteenth century and seems quite likely to do again in the twenty-first, capitalism automatically generates arbitrary and unsustainable inequalities. $r > g$.”
— Thomas Piketty, Capital in the Twenty-First Century, 2014
The data is the point. Piketty’s book put a nineteenth-century question back on the front page, and it did so not with a new theory but with a chart — two centuries of reconstructed top-income shares that no one before had assembled. The chart says what the Kuznets curve denied: the mid-century equality was the exception, not the destination. And $r > g$ names the engine — when wealth earns more than the economy grows, accumulated capital pulls away from labor income on its own, “the past devours the future.” This is the rung where distribution comes home to the center.
The modern program has two halves: a method and a mechanism. The method is the top-income-shares technique — reconstructing the income of the top slices of the distribution from tax records rather than household surveys, which lets you see the very top (where surveys miss the richest) and reach back a century (where surveys did not exist). Stitched across countries, this is the World Inequality Database. The mechanism is $r > g$: when the return on capital runs persistently above the growth rate, the stock of wealth grows faster than incomes, so the wealth-to-income ratio climbs and the owners of capital capture a rising share without doing anything new.
Piketty’s accounting rests on a long-run identity: the capital-to-income ratio settles at $\beta = s / g$, where $s$ is the net saving rate and $g$ the growth rate. A lower $g$ raises $\beta$ — slow-growing economies accumulate more wealth relative to income. Layer on the return $r$ on that wealth, and when
$$r > g,$$wealth compounds faster than the economy expands, so inherited and accumulated capital pulls ahead of labor income over time. Nothing in the identity is exotic; the force of the claim is empirical — that $r$ has typically exceeded $g$ over the long run, and the mid-century reversal that broke the pattern was war and policy, not the natural order reasserting itself.
Suppose a fortune earns 5% a year while the whole economy grows 2%. The fortune does not just stay large — it pulls steadily ahead of everyone whose income rises with the economy, generation after generation, and the gap widens by compounding alone. The heirs need do nothing; the arithmetic does the work. That is “the past devours the future”: yesterday’s accumulated wealth grows faster than today’s earned income, so dynasties form not from effort but from a return that beats growth.
The lineage home of the program — the Atkinson–Piketty–Saez turn, the post-2008 distributional reckoning, and where it sits in the field’s current pluralism — is History of Economic Thought Ch.17 (Modern pluralism). The policy apparatus it feeds — optimal taxation, the Ramsey rule — sits in Economics Ch.16 §16.7, though the policy fight itself this thread routes to the inequality walkthrough rather than carrying.
The historical record the program documents — the within-country widening that opened after 1980, and the post-2008 turn that put these charts on the front page — runs through Economic History Ch.18 (Globalization, financialization, and the great moderation) and Ch.19 (The 2008 crisis and after).
Here, for the first time in the thread, the predecessor frame is not a dead rung but a live opponent — and the honest engagement is the disagreement Piketty reopened, with both sides argued at strength. The Piketty-Saez-institutions reading: the post-1980 widening, the decline in labor’s share of national income, and the explosion at the top are not what a pure marginal-productivity story predicts if technology and factor supplies had simply stayed put. They track institutional change — union decline, the collapse of top marginal tax rates, financialization, the market power of superstar firms, and the $r > g$ accumulation dynamic. On this reading, distribution is shaped by power and policy well beyond what factor markets alone would deliver, and the marginal-productivity baseline explains the floor, not the deviations.
Now the other side at full strength, because it is not a strawman. The superstar-economics reading — Mankiw’s “Defending the One Percent” is its sharpest statement — says the top’s rise is best explained by skill-biased technical change and scale effects working through marginal products, not around them. Technology now lets one talented person serve a billion customers: the software a single team writes runs on every phone, the fund a single manager runs moves billions, the performer a single platform hosts reaches the planet. The marginal product of top talent genuinely exploded, so factors are still paid roughly what they add — the products just changed. On this reading, $r > g$ and the top-share charts are real, but the driver is technology rewarding scarce complementary skill, not institutions failing to restrain power.
Both channels are real, and the dispute that survives is about magnitude, not existence: how much of the post-1980 widening is technology-and-globalization rewarding scarce skill, and how much is institutions-and-power letting the top capture more than a frictionless market would grant. That is a genuine, live, mainstream-internal disagreement — and naming it as a magnitude question, not a winner-take-all clash, is the calibrated position, not a dodge.
“Wealth concentrates automatically” ($r > g$) is a real tendency, not an iron law — and the difference is the whole policy argument
Piketty’s $r > g$ entered general vocabulary as “the rich get richer no matter what.” The mechanism is real and the long-run data backs it. But “automatically” does heavy lifting — the mid-century reversal proves the tendency can be broken by policy, which is exactly why naming it a tendency rather than a law is what keeps the politics honest.
Where this leaves us
The arc is complete: center to margin and back. Piketty did not overturn marginal-productivity theory — he did something the thread had been building toward since Stage 2. He made distribution measurable over the long run, and the measurement showed that the questions classical economics asked — who gets the surplus, and why — are back as first-order questions, now with data Ricardo and Marx could only have dreamed of. The calibrated near-consensus is firm where it should be and honest where it cannot be. Firm: distribution is shaped by more than marginal products — Cambridge’s kernel and the empirical program both pointed here — and it is a first-order question again. Honest: the live disagreement is about magnitude, how much of the post-1980 widening is technology-and-globalization rewarding scarce skill (the superstar reading) versus institutions-and-power (the Piketty-Saez reading).
This thread commits, and says so on purpose: both channels operate, and the institutions-and-power channel is larger than the pre-Piketty mainstream allowed. Marginal products explain the baseline; institutions and power explain the deviations — and the deviations have been large. That is a position, not a hedge: a claim about where the weight sits, with the residual dispute named rather than buried. The thread closes on the irony it has been earning since Stage 1. Economics opened with distribution as “the principal problem,” spent a century treating it as a solved by-product of factor markets, and rediscovered — with two centuries of data the classicals never had — that the founders were right it was the central question all along.
Where this leaves us
Four engagement units, six conceptual rungs, each forced by a limit of the one before it:
- Ricardo and Marx — distribution at the center. Ricardo made the split inspectable and tied it to growth; Marx took the same surplus and relocated its source from natural scarcity to the rate of exploitation. Both put distribution at the heart of the discipline. The bargaining-power kernel survived; the labor theory of value and the immiseration prediction did not.
- Clark and Cambridge — the margin dissolves the conflict, then can’t measure itself. Marginal-productivity theory dissolved the class-conflict framing by paying each factor what it adds — and Cambridge UK won the formal point that aggregate “capital” cannot be measured independently of the profit rate it helps determine. The theory survives as a partial framework; the aggregate version rests on unsafe ground. (A correction about method.)
- Kuznets and the data — a law that was an accident. The inverted-U curve let the field treat inequality as a stage of development — until the long-run series showed the mid-century compression was a war-and-policy artifact, not a natural downswing. The Kuznets curve is superseded as a law; the structural-shift mechanism survives in early development. (A correction about the facts.)
- Piketty and $r > g$ — distribution comes home. The Atkinson-Piketty-Saez program made distribution measurable over two centuries and showed the classical questions are back. The calibrated verdict: distribution is shaped by more than marginal products and is first-order again; the live dispute is how much of the post-1980 widening is scarce-skill technology versus institutions and power.
Read end to end, this is the rarest shape a theory-history can take: a question that left the center and came back to it. Distribution began as the whole of economics — Ricardo’s “principal problem,” the thing the classical apparatus existed to answer. The marginalist turn made it look solved: if every factor earns what it adds, the split is just arithmetic, and the interesting questions moved to growth and stabilization. For most of a century that was the field’s settled posture, and it was not foolish — there were real reasons to treat distribution as a derived by-product during the growth-theory decades. The verdict is that those reasons were incomplete, not wrong.
What pulled distribution home was not a new theory but two things the earlier rungs lacked: a formal warning and a long dataset. The Cambridge controversy showed that “the market pays everyone what they’re worth” is a claim about competitive individual markets, not a theorem about the whole economy’s shares. The empirical-distribution program showed that the post-1980 widening, the labor-share decline, and the top-share explosion track institutions and power, not just marginal products — and that the mid-century equality everyone had taken as the destination was the exception. The honest place to stand is the one this thread has argued toward: marginal products explain the baseline, institutions and power explain the deviations, the deviations have been large, and the classicals were right that who-gets-what is a first-order question. The discipline spent a century learning that the hard way. The theory of distribution did not end; it came back with data.