Few propositions in economics have been declared dead as often, or have refused burial as stubbornly, as the claim that a falling rate of profit is written into the way capitalist production develops. Textbooks settle the matter in a paragraph. Journals reopen it in special issues and long exchanges of articles. Between the paragraph and the special issue stands the reader who wants a plain answer to a plain question: does the law hold? The honest response begins with a refusal, because the question as usually posed bundles three different questions together, each with its own kind of evidence and its own state of resolution, and much of the celebrated quarrel consists of people answering one of them with material drawn from another.

The stakes explain the stubbornness. In the architecture of Marx’s critique, the tendency he announced is the point where the system’s success turns into its own difficulty: production grows more productive, capital accumulates, and the return on the accumulated stock is squeezed by the very mechanisation that made accumulation profitable. If the law holds in a strong form, crisis is not an accident that befalls capitalism from outside but a pressure generated inside its ordinary operation. If the law fails, that particular route from growth to breakdown closes, and the theory of crisis has to be rebuilt from other materials. Generations of economists, friendly and hostile, have understood what hangs on the answer, which is why a dispute that looks scholastic from a distance has absorbed so much serious labour at close range.

What follows separates the argument into the three tracks on which it actually runs. The first track is logic: what the law claims, in the texts Marx left, and whether the claim coheres on its own terms. The second track is mathematics: what can be proved, inside stated models, about mechanisation, wages and equilibrium returns. The third track is the record: what measured rates of return have done across long stretches of capitalist history. Each track gets its strongest arguments stated before any grading, each is graded separately, and the grading is allowed to differ from track to track, because it does. A reader who keeps the tracks apart will find that the debate, so confusing as a single shouting match, becomes a set of answerable questions, some of them answered.

The falling rate of profit and the argument that will not sit still

Marx announced the tendency twice over, in two different registers. In the notebooks of the late eighteen fifties, published long after his death as the Grundrisse, he treated the falling return on capital as the most consequential proposition in political economy, the law in which the real difficulties of the existing order announce themselves. In the manuscript that Engels later shaped into the third volume of Capital, the claim received its systematic dress: a chapter stating the law as such, a chapter parading the influences that work against it, and a chapter drawing out the internal contradictions that its operation produces. The double announcement matters for everything that follows in this article. The law was never a stray remark or a journalist’s flourish. It stood near the centre of the mature theory, and its author knew it carried weight.

Defenders in the next generations loaded it with even more. Henryk Grossman, writing in 1929, built an entire breakdown theory on the tendency, extending reproduction schemes across long runs of accumulation until the surplus available could no longer valorise the swollen capital stock. Ernest Mandel, a generation later, kept the law but refused the breakdown load, folding the tendency into a multi-causal account of long waves in which profitability, markets, class struggle and state policy interact. Between those poles the law has served as foundation stone, as one mechanism among several, and as expendable scaffolding, sometimes within the work of a single author. Any summary that begins by announcing what Marxists believe about the tendency has already gone wrong, because the tradition contains every position on it that a reasonable reader could occupy.

The challenges arrived in layers, and the layers correspond to the three tracks this article separates. Doubt about the law’s determinacy came first from inside the tradition, most sharply in Paul Sweezy’s 1942 presentation, which argued that Marx’s own premises leave the direction of the ratio unsettled. The formal challenge arrived in 1961, when Nobuo Okishio proved a theorem about technique choice and equilibrium returns that appeared to close the question against Marx altogether. The textual challenge matured later, as scholarly editions made the manuscripts fully available and editors such as Michael Heinrich asked whether Marx himself had retained confidence in the law during his final decade of work. The empirical challenge runs underneath all three, in the form of statistical reconstructions that disagree with one another about what the twentieth century actually did to profitability.

Each challenge is real, and each operates on its own track. Sweezy’s argument concerns what follows from premises. Okishio’s concerns what can be proved inside a model. Heinrich’s concerns what a set of manuscripts shows about their author’s mind. The statistical quarrels concern what happened in the world. None of these, stated carefully, even engages the others directly, which is precisely why the debate’s history is littered with exchanges in which both parties were right about the question they were answering and wrong about the question their opponent had asked. The remedy is not more passion. It is bookkeeping: a discipline of stating, for every argument in the dispute, which track it belongs to and what would count as settling it there.

The law stated in its own terms

A compact statement of the claim is necessary before its troubles can be appreciated, and the companion study of the tendency and its mechanism gives the derivation at full length. The rate at issue is the ratio between the surplus value produced in a period and the total capital advanced to produce it, the sum of constant capital, laid out on machinery, buildings and materials, and variable capital, laid out on wages. Accumulation, driven by competition, continually raises the mass of means of production that each worker sets in motion: one operative tends equipment that would once have required a roomful of hands. In Marx’s vocabulary the value composition of capital rises, and it rises because the technical composition underneath it rises. If the rate of surplus value, the ratio between the surplus extracted and the wage bill, were to stay fixed while the composition climbs, the arithmetic is unforgiving. The denominator of the profit ratio grows faster than the numerator, and the rate falls.

Marx complicated his own law in the act of stating it, and the complication is part of the claim rather than a retreat from it. The chapter on counteracting influences lists the forces that accumulation itself sets against the fall. Exploitation can intensify, raising the numerator. The same productivity growth that swells the stock of machinery also cheapens each machine, so the value of the constant capital grows more slowly than its physical mass. Wages can be pressed below the value of labour power. A growing reserve of unemployed labour holds wage growth down and opens low composition fields for fresh investment. Foreign trade cheapens both materials and wage goods. Joint stock organisation separates a layer of capital that accepts a dividend below the general rate. A law that arrives with its own brakes listed on the next page is not a prediction that the measured rate will decline year by year like a stone sinking in a pond. It is a claim about a pressure and about the forces arrayed against the pressure, all of them generated by the same process.

Two consequences follow for the debate. First, single episodes refute nothing. A decade of rising profitability is consistent with the law as stated, provided the counteracting influences were doing heavy work during that decade, just as a decade of decline is consistent with several rival explanations. Readers who demand that the law show itself in every statistical window are demanding something its author never promised. Second, the law’s friends cannot have the point both ways. A tendency claim that absorbs every contrary episode by pointing at its brakes risks becoming a claim that no evidence could embarrass, and the sharpest internal critics of the law, from Sweezy onward, have pressed exactly there: they ask what the premises actually fix and what they leave floating. That question, the question of determinacy, is the heart of the logic track, and it is a better question than the crude demand for a falling line on a graph.

Two compositions: what rises, and in what units

A distinction from Marx’s own text quietly governs the whole subsequent debate, and restoring it prevents a family of errors. He separated the technical composition of capital, the physical ratio between the mass of means of production and the number of workers who work them, from the value composition, the ratio between the money values of those means and of the wage bill, and he defined the organic composition as the value ratio insofar as it mirrors the technical one. The separation matters because the two ratios need not move together, and the wedge between them is the cheapening of the elements of constant capital operating at the level of definitions. A textile mill whose spindles per operative multiply tenfold across a century has undergone a revolution in technical composition. If each spindle, and the power that drives it, and the fibre it spins, have all cheapened dramatically across the same century, the value composition of the capital invested in that mill may have risen sedately, stagnated, or in particular decades even fallen. The law’s pressure operates on the value ratio, since profit is a ratio of values, while the mechanism’s engine, mechanisation, operates in the first instance on the physical ratio. Every version of the law therefore contains an implicit race between the physical multiplication of equipment per worker and the cheapening of each equipment unit, and the outcome of that race is an empirical and technological question, not a theorem.

This wedge explains several otherwise puzzling features of the debate’s history. It is why the same industrial landscape, visibly more mechanised in every decade, can host rising, falling or stationary measured rates without any of the observers hallucinating: the visible landscape reports the technical composition, the accounts report the value composition, and the cheapening wedge translates between them at a pace set by the machine producing industries’ own productivity. It is why Okishio’s construction, in which the cheapening is fully registered in the repricing of inputs, delivers its striking result, and why the temporal construction, in which the capital account remembers what was actually paid before the cheapening, can deliver the opposite path from the same physical facts. And it is why econometric tests of the law that proxy composition by physical capital per worker, without attending to the valuation of that capital, test a cousin of the law rather than the law itself. Readers who keep the two compositions distinct possess a solvent in which a surprising quantity of the literature’s apparent contradictions dissolve, leaving the genuine disagreements, determinacy, valuation and measurement, standing in clearer outline.

The working day: the first counterweight

The oldest counteracting influence in Marx’s system is not on the third volume’s list by accident of ordering; it is the one whose history he had written first and at greatest length. The struggle over the length of the working day, chronicled in the first volume’s chapter on that subject from factory inspectors’ reports and parliamentary inquiries, was the original arena in which the rate of surplus value was contested in practice. Every extension of the day beyond the hours needed to reproduce the wage raised the surplus extracted from each worker without any addition to the capital advanced, a pure lift to the numerator of the profit ratio, and the ferocity of the nineteenth century fight over hours, from the Ten Hours agitation to the legal normal working day, is unintelligible unless that arithmetic is kept in view. When legislation and organisation closed the extensive margin, capital migrated to the intensive and relative routes: speeding the machinery, filling the pores of the day, and above all cheapening wage goods so that the necessary portion of a now fixed day shrank. The law of the falling rate presupposes that whole history, because the rising composition it tracks is the shape accumulation took once the day could no longer be stretched, and its counterweights include the intensification that the same struggle forced into being.

The connection disciplines a lazy habit in the modern debate, which treats the rate of surplus value as a parameter to be dialled in models and rarely as the sedimented outcome of conflicts with dates, statutes and casualties attached. Sweezy’s indeterminacy, stated abstractly, says that exploitation can rise as composition rises. Stated historically, it says that whether exploitation rose depended on factory law, union density, the franchise, the supply of labour from countryside and empire, and the outcome of fights like those over the ten hours bill, none of which any theory of accumulation can deduce from the concept of capital alone. The logic track’s open verdict and the record track’s decompositions meet on this ground: the magnitudes the law leaves free are the magnitudes that politics and organisation decide, which is why the debate about a ratio keeps recruiting historians, and why its purely formal episodes, however brilliant, have never succeeded in dismissing them from the room.

An inheritance disputed: the classical economists on falling returns

Marx did not invent the idea that profitability declines as economies mature, and situating his version among its ancestors clarifies what is distinctive in it and what the later debate is really about. Adam Smith had explained falling returns by competition: as capitals multiply in a growing economy, they bid against one another, forcing returns down toward a floor set by the ordinary rate of interest in a fully stocked country. David Ricardo gave the tendency a harder mechanism. Population growth pushes cultivation onto worse land; the price of food rises; money wages must follow food prices upward if workers are to live; and profits, the residual after wages, are ground down between rising labour costs and rents that absorb the gains of the better soils. In Ricardo’s construction the fall is real, secular and rooted in nature’s diminishing generosity, and his remedy talk ran toward cheap imported corn, which is why the theory of profit became a weapon in the fight over the Corn Laws.

Marx studied both accounts closely in the notebooks published as Theories of Surplus Value, and his criticisms of them are the seedbed of his own law. Against Ricardo he objected that the squeeze story makes profit depend on an agricultural accident, as though the fate of industrial capital were hostage to soil quality, and that it treats wages through a crude population mechanism. Against Smith he objected that competition, which merely equalises returns among trades, cannot by itself depress the general level from which equalisation proceeds. His alternative relocated the cause inside production itself: not in nature’s stinginess or in the crowd of competitors, but in the rising mass of past labour, embodied in plant and materials, that each hour of living labour must set in motion and make fruitful. Where Ricardo’s rate falls because labour becomes more expensive to feed, Marx’s falls because labour becomes more heavily capitalised. The point of the genealogy for this article is that the family’s central quarrel has always been about which variable does the squeezing. The modern debate, with its composition effects ranged against wage effects in econometric decompositions, is Ricardo’s and Marx’s quarrel continued with better data, and Okishio’s theorem enters the same family argument from a third direction, by asking whether the production side mechanism can squeeze at all when wages in goods are held still.

What kind of law is a tendency?

The logic track harbours a philosophical question that is worth surfacing, because confusion about it fuels confusion everywhere else. What is a tendency claim, and what could count against one? In the sciences that economics imitates, tendency statements are familiar and respectable: a body tends to fall, gases tend to expand, although interference, containment and countervailing forces decide what any particular body or gas actually does. Such claims assert the operation of a mechanism, not the inevitability of an outcome, and they are tested in a distinctive way. The investigator asks whether the mechanism is present and active, whether its signature appears when interfering factors are controlled or accounted, and whether the countervailing forces that the theory itself names are strong enough, in the case at hand, to explain any gap between the tendency and the observed path. A tendency claim fails when its mechanism cannot be found at work, or when the theory must invent new brakes, unnamed in advance, to excuse every disappointment.

Measured by that standard, the law’s design is better than its reputation among philosophers of science suggests. Its mechanism is specified: accumulation raising the mass of means of production per worker. Its brakes are specified in advance, in print, in a dedicated chapter, rather than improvised after awkward data arrive. Its signature is identifiable in principle: periods and sectors where composition climbs fastest should show the strongest downward pressure, other things equal, and decomposition studies exist precisely to hunt for that signature. The genuine weakness lies elsewhere, in the comparative statics of forces that the theory does not equip with a common measure, the difficulty Sweezy isolated. A mechanism can be real, active and signed in the data while still losing, across any given half century, to the forces ranged against it. Whether the word law should be reserved for claims stronger than that is a question about honorifics. The substantive question is whether the mechanism operates as described, and that question is empirical in a way the tradition’s more metaphysical defences sometimes obscure. Readers who grasp this point hold the key to the whole debate, because they can see why the formal track’s results, which concern what mechanisms do inside models, and the record track’s results, which concern what mechanisms did inside history, are both needed and neither sufficient.

The track confusion, named

The namable claim of this article is the track confusion: most exchanges about the tendency fail because the participants are running on different tracks while believing they share one. One participant argues about a formal theorem, a second about what Marx wrote and meant, a third about a statistical series, and each produces genuinely strong material that cannot resolve the others’ questions. The confusion is not a sign of bad faith. It is the natural result of a dispute whose three subject matters share a vocabulary. The words rate, tendency, fall and law do duty on every track, so a sentence that is precise on one track sounds like an answer on the others.

Consider three scenes, each assembled from real habits of the literature rather than invented folly. In a seminar, a student announces that the law has been refuted, and produces, as the refutation, a chart of American profitability recovering after the early nineteen eighties. The chart may be accurate. What it cannot do is refute a theorem, because theorems are not refuted by charts; they are refuted by counterexamples inside their assumptions or by showing the assumptions empty. The student’s evidence belongs to the record, and it speaks, at most, to whether the tendency dominated its counterweights in one country across three decades. In a journal exchange, a defender of the law answers a statistical critic by quoting Marx’s chapter on counteracting influences, as though a passage from 1865 could arbitrate a dispute about capital stock measurement in 1970. The quotation is genuine. It settles nothing about depreciation conventions. In an examination script, a candidate demolishes the temporal interpretation of Marx’s value theory by observing that most economists do not use it, which confuses a sociological fact about the profession with a logical verdict on an argument.

The confusion also runs in the flattering direction. Defenders announce that the law has been vindicated when a long statistical decline is found, although a decline produced by rising wages would vindicate the profit squeeze account rather than Marx’s mechanism, and although the theorem on the mathematics track stands untouched by any series whatever. Critics announce that Marx has been buried when a formal result is proved, although the result governs a model whose assumptions, constant real wages and equilibrium comparison, are exactly the points a defender is entitled to contest. Victory on one track feels like victory in the whole war because each track’s result can be pronounced in the same sentence: the rate of profit tends to fall, or it does not. The sentence is shared. The propositions underneath it are not.

Naming the confusion gives the reader a working instrument. Before accepting any argument in this literature, ask three questions of it. Which track is the argument on? What would settle the question on that track? Is the evidence offered the kind that could settle it there? Arguments that survive these questions are the load-bearing ones, and the debate contains more of them than its reputation suggests. Arguments that fail are usually not dishonest; they are simply addressed to a question their author had not noticed was different. The tracks table that follows is offered as a map for applying the instrument at a glance, and the rest of the article walks each track in turn, slowly.

The three tracks table

The three tracks table compresses the whole dispute into a single grid. Its rows are questions rather than positions, which is the design decision that matters: partisans of every camp can sign the rows even when they fight over the cells. Its columns ask, for each track, what exactly is at issue, what the strongest argument looks like from each side, what sort of thing would settle the question, and whether settlement has arrived. Read down a column to see how different the three questions are. Read across a row to see a complete dispute in miniature, with its verdict attached and its verdict’s grounds exposed.

Track	The question at issue	Strongest case for the law	Strongest case against	What would settle it	Settled?
Logic	What does the claim say, in the texts Marx left, and does it hold together on its own terms?	Stated as a tendency with internal counterweights, the law survives single contrary episodes by design	With composition and exploitation both free to move, the premises alone do not fix the ratio’s direction	Agreement about which magnitudes the theory pins down and which it deliberately leaves open	Partly: the tendency reading is textually secure, its determinacy is not
Mathematics	Inside stated models of technique choice and wages, must mechanisation lower the equilibrium return?	Under temporal valuation, viable cost cutting can be shown to depress the rate, so the formal door stays open	Okishio’s proof shows that viable innovation at an unchanged real wage cannot lower the equilibrium rate	Agreement on the valuation convention and on whether equilibrium comparison is the right test	The proof is settled; its application is not
Record	Have measured returns on capital fallen across long stretches of capitalist history?	Several reconstructions find extended declines, and econometric work on American data finds the predicted pressure at work	Rival reconstructions, built on other choices about capital and depreciation, find recovery after the early nineteen eighties	Shared conventions for measuring the capital stock, depreciation and the boundary of the productive sector	No: findings move with the conventions chosen

Three cautions govern the table’s use. A settled cell means settled on that track, under that track’s standards, and exports nothing to neighbouring rows. An unsettled cell is not a euphemism for falsehood; several unsettled cells in this dispute contain propositions that may well be true and are certainly arguable. And the final column is a report on the state of argument, not a score in a contest between camps, because the same scholar can hold the winning position on one track and the losing position on another. Sweezy, to take the cleanest case, lost nothing of his standing by doubting the law’s logic while remaining one of its most lucid expositors.

Track one: the internal logic of the argument

Begin where the law begins, with its logical shape. Marx’s claim is built from three storeys. The ground floor is an accounting identity about where profit comes from: surplus value is produced by labour and by nothing else, so the mass of profit in a period is bounded by the surplus labour performed. The first floor is a dynamic claim about accumulation: competition forces each capital to raise the productivity of its workers by equipping them with more machinery and material, so the ratio between dead and living labour in production climbs over time. The top floor is the tendency itself: because the denominator of the profit ratio contains the whole accumulated stock while the numerator is fed only by living labour, a rising composition presses the rate downward unless the rate of exploitation rises fast enough to compensate. The claim’s logic is genuinely tight at each joint. Its vulnerability lives in the word unless.

What exactly does the law claim on its own terms?

Read strictly, the law claims a tendency, not a timetable. Accumulation raises the mass of means of production set in motion by each worker faster than it raises the surplus extracted from that worker, so the rate of profit is pressed downward, while named counteracting influences, themselves limited, work against the pressure.

The elaboration begins with the counteracting influences, because they are where the logic either deepens or dissolves. Each brake Marx names has a built in limit, and he says so. Exploitation can intensify, but the working day has a physical and social ceiling, and squeezing more surplus from each worker runs into resistance and exhaustion. Machinery cheapens, but cheapening applies to the wage goods as well, and a cheaper wage bundle raises the feasible rate of surplus value in a way that helps the numerator too; the brake and the accelerator are wired to the same engine. Wages can sit below the value of labour power, but not indefinitely and not without consequence for the workforce that capital needs. The reserve army disciplines wages, yet accumulation periodically drains the reserve and hands bargaining power back to labour. A defender of the law’s logic can therefore argue that the brakes are real but self limiting, while the pressure they resist is renewed by every round of accumulation. The tendency, on this reading, is not a hedge. It is the claim that the pressure outlasts its brakes across the long run of the system.

The critic of the logic answers at the same level, and the answer deserves its full force. To say that brakes are limited while pressure is persistent is to assert a ranking between forces that the theory nowhere measures. Nothing in the concepts of composition or exploitation supplies a common scale on which the cheapening of machinery can be weighed against the cheapening of wage goods, and without such a scale the assertion that one side wins in the long run is a conjecture wearing the dress of a deduction. Marx proved that a falling rate follows if exploitation is held still. He asserted, with reasons, that exploitation cannot rise without bound. Between those two results lies a gap that only an additional argument can fill, and the additional argument has to compare rates of change that the theory leaves free. This is Sweezy’s indeterminacy objection in its modern dress, and it has never been answered by quotation, because it is not a textual point. It is a point about what the premises contain.

Where does that leave the logic track? With a graded verdict rather than a trophy. The tendency reading is secure: Marx claimed a pressure, not a path, and critics who refute him with rising decades are answering a claim he did not make. The determinacy of the pressure is open: whether the pressure must dominate its brakes is a further proposition, arguable on both sides, which the tradition has sometimes smuggled into the law and sometimes honestly flagged as needing support from outside the law itself, from the theory of wages, of technical change, or of the state. Readers who keep those two judgements separate have understood the logic track. Readers who merge them, in either direction, have joined the track confusion.

The texts behind the law: manuscripts and an editor

Underneath the logic sits a prior question that belongs to the same track, because it asks what the claim is before asking whether the claim succeeds. The systematic statement of the law survives in a manuscript Marx drafted in the middle eighteen sixties and never prepared for the press. Engels published it in 1894 as Part Three of the third volume of Capital, and he was candid in his preface that his friend’s text was a first draft, in places barely more than notes toward an argument. The chapters on the law as such, on counteracting influences and on the exposition of internal contradictions are genuinely Marx’s. Their arrangement, headings and some of their transitions are the editor’s. How much doctrine can survive that admission is a question the series examines from the editorial side in its study of the Marx and Engels authorship question, and the short answer is: more than sceptics suggest, less than devotional readers assume.

How far did Engels shape the law’s printed form?

Engels assembled the third volume in 1894 from a manuscript Marx drafted three decades earlier, and he said openly that the text was unfinished. He ordered the material, supplied headings and smoothed transitions, but the chapters stating the law stand in the manuscript itself, so editing shaped presentation more than doctrine.

Presentation is not nothing, however, and the textual dispute sharpens once the whole archive is admitted as evidence. Marx returned to questions of profitability repeatedly in his final decade. He filled notebooks with mathematical fragments, followed the literature on interest and rent, and studied statistics with the appetite of a man who still expected to write. Two rival readings of that archive now stand in print. Michael Heinrich, an editor of the historical critical edition, argues that the fragments show a thinker growing less certain of his earlier constructions: Marx had come to doubt whether a general law of the rate could be demonstrated at all, since its two moving parts, composition and exploitation, cannot be ranked in advance, and the crisply systematic law of the printed volume is partly an artefact of Engels presenting a draft as a doctrine. Andrew Kliman answers that no retraction exists anywhere in the record, that Marx restated the tendency in writings from the same years, and that private hesitation over details is the normal condition of a working scientist rather than evidence of abandonment.

Both readings handle real documents, and neither can be dismissed by pointing at the other side’s motives, since both scholars have spent careers inside the archive they interpret. The honest grade for the textual strand is therefore narrower than either camp’s public summary. Settled: the law stands in Marx’s own manuscript, in his own ordering of topics, and was no invention of his editor. Settled also: Marx never published it himself, and the distance between a cherished draft and a finished doctrine is a distance he never closed. Contested: whether the notebooks of his last decade amount to quiet doubt, continued commitment, or the ordinary unfinished business of a research programme interrupted by death. Readers should notice what kind of evidence could move this strand. A dated letter retracting the law would move it. So would a late manuscript restating it with full confidence. Arguments about which interpretation feels more faithful move nothing, and they belong to biography rather than to the debate this article is mapping.

Inside the third volume: the law among its contradictions

Because the textual strand asks readers to care about particular chapters, those chapters deserve a reader’s summary of their own, and the summary sharpens several points the later debate blurred. The thirteenth chapter states the law almost arithmetically, with illustrative series in which a growing capital, divided between constant and variable portions in shifting proportions, yields a descending percentage return at an unchanged rate of exploitation. The fourteenth chapter then works through the counteracting influences one by one, in the order this article has already used, and its tone is not defensive but taxonomic, as though the author were inventorying the system’s own shock absorbers. The fifteenth chapter is the one posterity undervalued. There the falling rate is connected to the production of a relative surplus population, to the glut of commodities and of capital itself, and to the periodic devaluation through which accumulated wealth is written down so that accumulation can resume. Marx’s image is of capital becoming a barrier to its own expansion, a mass of value that cannot be valorised at the expected rate and so stands idle beside unemployed workers, an absurdity he presents as the system’s signature rather than as its accident.

Two features of that chapter matter for every later track. The crisis mechanism it sketches is not the smooth exhaustion of profitability across centuries but a violent reconciliation: the rate falls, paper and real capitals are devalued, the denominator of the ratio is slashed by destruction and bankruptcy, and the rate recovers on the ruins, which is why the law’s own text is compatible with long eras of healthy returns punctuated by collapses. Grossman’s breakdown reading had to flatten this rhythm into a one way ratchet to obtain its terminal result, and the flattening, visible to anyone who reads the fifteenth chapter whole, is the textual ground on which Mandel and other pluralists resisted him. The second feature is the chapter’s insistence that the same development which depresses the rate also swells the mass of profit, since a larger capital at a lower percentage can yield a greater absolute surplus than a smaller capital at a higher one. A falling rate is therefore not, in the text’s own presentation, a symptom of a shrinking pie. It is a ratio squeezed while the pie grows, and any empirical test that hunts for declining masses of profit has misread the claim it purports to test. More than one published refutation founders on exactly that rock, and spotting the error is a reliable early exercise in the track discipline.

Doubt from inside the tradition before the theorem

The formal track opens with a critic who cannot be filed as an enemy of the tradition, because he was one of its ornaments. Paul Sweezy’s The Theory of Capitalist Development, published in 1942, introduced English readers both to the law and to much of the continental literature around it, including the transformation critique associated with Ladislaus von Bortkiewicz. Having presented the law with scrupulous care, Sweezy then turned on its derivation. Marx, he observed, obtains the falling rate by allowing the organic composition to rise while the rate of surplus value stays fixed. But the two magnitudes are not independent dials. The productivity growth that equips each worker with more machinery also cheapens the goods workers consume, which means the same bundle of wage goods can be produced with less labour, which means the share of the working day needed to reproduce the worker shrinks and the rate of surplus value can climb. Drive composition upward and the mechanism that drives it simultaneously supplies the means to drive exploitation upward. Which force wins is not written into the premises.

What did Sweezy say could not be determined?

Sweezy held that the direction of the profit rate could not be settled from Marx’s premises alone. The productivity growth that raises the organic composition also cheapens wage goods, which permits a higher rate of surplus value, and with numerator and denominator both free to move, the ratio can travel either way.

Sweezy’s conclusion was measured, and its measured quality is what made it dangerous to the law’s standing. He did not declare the tendency false. He declared it underargued: the law, as derived, showed a possibility dressed as a necessity, and a theory of crisis that needed necessity would have to find its engine elsewhere. His own candidate was the difficulty of realising surplus value in consumption, the underconsumptionist line that he and Paul Baran later developed into the theory of monopoly capital, where the problem of a mature economy is absorbing a rising surplus rather than suffering a shrinking rate of return. Joan Robinson, writing her Essay on Marxian Economics in the same period from a position of critical sympathy, treated the falling rate with similar reserve, as one strand in Marx’s crisis thinking rather than its keystone. The significance of this episode for the whole debate is easily missed from outside. By the early nineteen fifties, the most sophisticated Marxist economics in English had already concluded that the law, in its strong form, could not bear the weight the tradition had placed on it, and had moved on. Okishio’s theorem, when it arrived, landed on ground that insiders had partly evacuated, which is one reason its reception outside the tradition, where it was trumpeted as a demolition, differed so sharply from its reception inside, where it often read as a rigorous version of doubts already in circulation.

Okishio’s theorem in plain language

Nobuo Okishio was a Japanese economist at Kobe University, formed in the mathematical tradition that took Marx’s reproduction schemes and the work of Piero Sraffa as starting points for exact reasoning. His paper Technical Changes and the Rate of Profit appeared in the Kobe University Economic Review in 1961, a venue obscure enough that the result travelled slowly, reaching wide attention only as it was taken up, extended and argued over during the following two decades. What he proved can be stated without algebra, and stating it without algebra is worthwhile, because the theorem’s force depends on assumptions that algebra tends to hide inside notation.

What is the Okishio theorem?

A result published in 1961 showing that if firms adopt only techniques that cut costs at prevailing prices and the real wage stays constant, the equilibrium rate of profit cannot fall. It is the principal formal challenge to Marx’s law, and remains contested.

Set the scene as the theorem sets it. The economy is described at a moment of settled prices, with a single uniform rate of return earned on capital in every line of production, the configuration classical and Marxian theory both call prices of production. Workers receive a real wage specified as a definite bundle of goods: so much food, housing, clothing and the rest per period of work, and that bundle does not change. A capitalist in some industry discovers or is offered a new technique, a different mix of machinery, materials and labour for making the same product. The capitalist’s test for adopting it is the only test a profit seeking owner can be assumed to apply: at the prices now ruling, does the new method produce a unit of output more cheaply than the old? If it does not cut costs, it is refused. If it does, it is adopted, and the adopter pockets an extra margin while prices still stand where they stood.

Now let the technique spread through its industry and let prices adjust until a uniform rate of return is restored everywhere, with the workers’ bundle exactly what it was before. Okishio proved that the new uniform rate cannot be lower than the old one. Where the innovation occurs in a basic industry, one whose product enters directly or indirectly into the production of everything, the new equilibrium rate is strictly higher. Where the innovation occurs in a luxury trade, whose product feeds no further production, the general rate is untouched. The intuition behind the algebra can be given honestly. The innovator’s cost saving is real output freed from producing the same goods. Since the wage bundle is fixed, none of that freed output is claimed by labour’s consumption. When competition generalises the technique and prices fall to match the new costs, the gain reappears as a higher ratio between the social surplus and the capital advanced across the economy. Productivity growth, inside this construction, enriches capital relative to a stationary standard of working class life. It cannot impoverish it.

A homely numerical sketch fixes the shape of the argument. Imagine a steel trade in which a unit of output sells at a settled price of one hundred, produced by a method whose materials and equipment wear absorb sixty and whose wage bill absorbs thirty two, leaving eight as return on the funds tied up. An engineer offers a method that uses more equipment and fewer hands: materials and wear rise to sixty six, the wage bill falls to twenty, total cost drops to eighty six at the prices standing in the market. The first adopter earns fourteen where rivals earn eight, and the method is plainly viable. Rivals follow or exit. As the method generalises, competition presses the price of steel down toward its new cost structure, and trades that buy steel find their own input bills lightened. Wages in goods are what they were. When the dust settles and one rate of return again rules across trades, the theorem says that rate stands at or above the level from which the whole motion began. The mechanisation that looked, from the doorway of the single firm, like a heavier burden of equipment per worker has raised the economy wide return on capital rather than depressing it.

The result was not left as a curiosity of one paper. Michio Morishima, in his formal reconstruction of Marx’s economics, and John Roemer, in his analytical work on exploitation and class, each rebuilt and generalised the theorem’s core, and both drew the moral that a falling equilibrium rate cannot be extracted from mechanisation alone under the stated conditions. Their authority matters because neither man was engaged in anti-Marxian polemic. Morishima’s project was to give Marx’s system the rigour of modern mathematical economics. Roemer’s was to rebuild its theory of exploitation on general equilibrium foundations. When scholars of that intent conclude that a proposition fails inside a framework, the failure is a finding, not a slogan, and the debate’s defenders have treated it accordingly, by attacking the framework’s assumptions rather than the proof’s correctness. That division of labour, proof granted and premises contested, is the mature shape of the formal track.

The viability criterion under the microscope

Everything in the theorem flows through one narrow gate: the rule that a technique is adopted only when it lowers cost at the prices standing in the market. The gate deserves microscopic attention, because each side’s best arguments are really arguments about what that rule describes. In Okishio’s own construction the rule is a rationality postulate. An owner who adopted a technique that cost more per unit than the method it replaced, at unchanged prices, would be choosing a smaller margin out of perversity, and a theory is entitled to abstract from perversity. The rule also encodes a claim about information: the adopter judges by prices that exist, not by prices that will exist after the adoption has spread, because those future prices are unknowable in detail and partly created by the adoption itself. Judged as a description of a single decision, the criterion is close to undeniable. Nobody invests to raise unit costs on purpose.

The aggregate is where the criterion’s innocence ends, and the gap between its two readings is the space in which several disputes live. For the theorem to go through, the social result of thousands of individually viable adoptions must be compared across two states of rest, and the comparison silently assumes that the only thing adoption changes is the technique in use. It does not change the scale of output that markets must absorb, the composition of demand, or the list of trades that exist. A heavier method in steel may be viable per ton while the industry’s profitable scale shrinks; a cheaper way of making a good may flood its market and destroy the price the viability calculation assumed. Defenders answer that equilibrium method exists precisely to hold such complications still while one mechanism is isolated, and that a critic who reintroduces every complication at once has not refuted the mechanism but declined the method. Critics answer that the complications are not decorations but the phenomenon, since capitalist technical change arrives as a disturbance, not as a re-sorting among resting points. This is the methodological form of the quarrel that the Shaikh section presents in substantive dress, and recognising the same argument in two costumes is further exercise for the track discipline.

Fixed capital adds a genuine technical thicket inside the gate, and honesty requires marking it. The homely sketch used earlier treated equipment wear as a smooth flow, but real plant lasts for years, and a technique’s cost advantage must be judged across the life of the machines it installs, whose later years will face prices nobody can quote at the moment of decision. Okishio himself extended his analysis toward durability, and the subsequent literature, including work by Morishima and by Roemer, reformulated the criterion for long lived equipment, in some versions requiring the adopter to weigh the whole stream of costs and receipts rather than a single period’s saving. The theorem’s friends regard these extensions as completions that preserve the moral: judge adoption by profitability over the relevant horizon, and viable innovation still cannot depress the settled rate at a fixed real wage. Its critics regard the extensions as concessions that the original gate was too narrow, and point out that across a machine’s lifetime the adopter is exposed to exactly the price slides that the temporal school places at the centre of its account. The thicket is real, the literature inside it is technical, and the safe summary for a general reader is that durability complicates the criterion without, so far, yielding a version of the dispute that either side regards as closed.

What the theorem settles and what it leaves alone

Precision about scope is the whole game here, so the theorem’s boundaries deserve a section of their own. Okishio’s result governs comparisons between equilibria. It says that if the economy were to settle at a uniform rate before an innovation and settle again afterward, with the wage bundle unchanged, the second settling point cannot show a lower rate. It says nothing, directly, about the path between the two points, about an economy that never settles, or about what happens to profitability while techniques are diffusing and prices are in motion. A tendency, its defenders observe, is a claim about historical motion, and a result about resting points may simply decline to address it. The theorem’s champions answer that equilibrium comparison is the standard instrument for isolating what a mechanism does, and that a mechanism which cannot lower the rate between rest points has a heavy burden when it claims to lower the rate along the way. Both remarks are fair. They mark the boundary of what has been proved.

The wage assumption is the second boundary, and it is load-bearing in the strictest sense. The theorem holds the workers’ bundle constant while productivity climbs, which means every gain from cheaper production accrues, in the eventual settlement, to the owners of capital. Let the bundle grow as productivity grows, which is what the long history of advanced economies broadly shows, and the arithmetic changes character. Okishio himself acknowledged that a sufficient rise in the real wage restores the possibility of a falling rate even inside his construction. The acknowledgement converts the theorem from a refutation into a fork. If one believes real wages are anchored by subsistence or by a customary standard that moves slowly, the theorem bites hard. If one believes wages are set by a contest whose outcome shifts with employment, bargaining institutions and the cycle, the theorem’s world is a special case, and the interesting question becomes which special case history inhabits.

Why does the assumed real wage carry so much weight?

Okishio holds the workers’ consumption bundle fixed while productivity rises, so every gain from a cheaper technique accrues to capital as it generalises. Let the real wage rise with productivity, as it has across long stretches of history, and part of the gain is absorbed by labour, which is exactly the opening through which the rate can fall.

Two further boundaries complete the survey. The theorem assumes that adopted techniques pass the viability test at the prices of the moment, judged one innovation at a time; it does not model a world in which firms must gamble on techniques whose payoff depends on prices that their own collective adoption will destroy, a world several critics argue is the actual one. And the theorem is silent about money, credit, unsold output and every other crisis mechanism, which is why even a complete surrender to Okishio on his own ground leaves Marxist crisis theory with other engines available, a point developed in the sections on the profit squeeze and on the stakes below. What the theorem settles, then, is one exact proposition: mechanisation by itself, at constant real wages, cannot depress the equilibrium rate of return. That proposition was worth settling, because a loose version of the law had traded on the opposite impression for a century. What the theorem leaves alone is nearly everything else the tradition argues about, and the refusal of both camps to respect that allocation of settled and unsettled ground is the formal track’s contribution to the track confusion.

The wage and profit frontier: the theorem in its modern dress

Readers who meet the theorem in contemporary dress will usually meet it as a picture, and the picture is worth describing because it converts the whole formal dispute into geometry that can be argued about precisely. Take the Sraffian wage and profit frontier: for a given system of production, the diagram traces every compatible pairing between the share of the net product going to wages and the uniform rate of return, the frontier sloping downward because a larger slice for labour leaves a smaller residual for capital at any given technique. A technique is a frontier. Technical change, on this representation, redraws the map. Okishio’s theorem, translated into the picture, says that when a new technique passes the viability test at the prices belonging to the old configuration, and the workers’ bundle is marked as a fixed point on the wage axis, the new frontier at that marked point lies at or outside the old one: the return available at the unchanged real wage is undiminished, and where basics are involved it is enlarged. The result’s elegance explains its hold on the profession’s imagination. A century of argument about tendency and countertendency is compressed into the question of how a frontier moves when its own points are used to judge the movement.

The picture also exposes, with unusual honesty, where the quarrels must live. Everything depends on the marked point. Move the wage point outward as productivity grows, in step with the historical pattern of advancing economies, and the new frontier can easily deliver a lower return at the new point than the old frontier delivered at the old one, which is the profit squeeze drawn as geometry. Freeze the point, and the theorem rules. The temporal school adds that the picture itself is a simultaneist artefact, since a frontier is a menu of alternative resting states and says nothing about the incomes and outlays, dated in calendar time, that firms actually experience while the economy travels between points on successive maps. Critics of that addition ask what a tendency claim could even mean if not a comparison among configurations. The frontier thus performs the same service for the formal track that the three tracks table performs for the whole debate: it does not settle the argument, but it shows every participant the exact location of the unsettled ground, drawn to scale.

The theorem’s journey through the journals

A result’s history shapes what a debate thinks it is about, and Okishio’s theorem had an unusually slow fuse. Published in a university review in Kobe in 1961, it circulated first among Japanese economists, for whom Marxian mathematical economics was a living research programme rather than a heritage subject. Its migration into English language argument ran through the nineteen seventies, when the revival of Sraffian economics gave it a natural audience and when the simultaneous equations framework it inhabits was becoming the common language of formal work on Marx. By the time survey writers took stock, in volumes such as Meghnad Desai’s Marxian Economic Theory and the Howard and King history of Marxian economics, the theorem had settled into its canonical role: the standing formal obstacle that any defence of the law must acknowledge before proceeding. Canonical roles distort. Generations of students met the theorem as a verdict pronounced by mathematics on Marx, stripped of the wage assumption and the equilibrium frame that this article has been at pains to keep attached, and a generation of defenders learned to respond by reaching for whichever reply lay closest, textual, empirical or methodological, without always noticing which track the reply belonged to.

The debate’s geography also deserves a word, because it explains some of its tone. The formal literature ran in journals of political economy and in the Cambridge inflected circles where Sraffa’s book was scripture; the textual literature ran through the German editorial project of the complete works and its associated scholarship; the statistical literature ran through American debates about that country’s postwar trajectory, conducted by researchers who often had limited contact with either of the other two conversations. Three scholarly worlds, three languages of rigour, one shared set of slogans. It would be surprising if such a structure produced anything other than the track confusion, and the surprise is rather that the cross references exist at all: that Kliman writes econometrics as well as textual defence, that Heinrich engages the formal indeterminacy argument on its own terms, that Duménil and Lévy preface their statistics with a theory of phases of capitalism. The integrators are the debate’s most useful citizens, and a reader assembling a shelf should prize them accordingly.

The real competition objection

One influential reply refuses the theorem’s scene setting rather than its logic or its wage assumption. Anwar Shaikh, whose work on crisis theory began in the nineteen seventies, argues that actual competition is not the stately equalisation of the models but a turbulent process in which capitals fight with the weapons at hand, and in that fight the criterion of survival is not the highest return on funds advanced. A firm survives by selling, and it sells by matching or beating the price its rivals can offer, which directs its attention to unit production cost above every other magnitude. The technique a firm reaches for is therefore the one that produces its commodity most cheaply per unit, and nothing in that reach guarantees that the winning technique also offers a handsome rate of return on the capital it requires. A highly mechanised method can shave unit costs while piling up an enormous charge of equipment per unit of output, so that when the method has spread, prices have fallen to its level, and the extraordinary margins of the pioneers have been competed away, the industry is left with a return on its swollen capital stock lower than the return it enjoyed before the innovation race began.

On Shaikh’s telling, the theorem’s viability test fails to describe this world. Okishio’s capitalist adopts only what raises profit at standing prices, computed with the pioneer’s whole position in view. The turbulent competitor adopts what cuts unit cost, with survival in view, and the two criteria part company exactly where mechanisation is heavy. Defenders of the theorem reply with equal precision: they argue that the alleged counterexamples characteristically smuggle in a technique that would not pass the original viability test once the adopter’s capital outlay is charged properly, or that they compare the new rate against a benchmark the theorem never promised to defend. The exchange has run for decades without either side conceding, and its persistence is itself informative. It shows that the formal track’s real contest is not over algebra but over which stylised description of capitalist decision making deserves to govern the model: the calm maximiser of the textbooks, or the embattled price fighter of Shaikh’s description. That is an argument about the world conducted through models, and it cannot be won by proof alone.

How firms actually choose: the evidence of practice

Shaikh’s objection to the adoption criterion invites an empirical question that neither pure theory can answer from its armchair: what do the people who choose techniques actually maximise? The evidence from management practice is more interesting than either camp’s stylised capitalist. Surveys of corporate financial officers, most famously the large American study by John Graham and Campbell Harvey published at the start of the two thousands, find that investment decisions in practice are governed by a mixture of net present value calculations, internal rate of return hurdles and, persistently, crude payback periods, the time a project takes to return its outlay, with the mixture varying by firm size and managerial education. Almost nobody in the surveyed population describes the decision as Okishio’s capitalist would, a comparison of unit costs at standing prices, and almost nobody describes it as a pure rate of return comparison either, since hurdle rates are notoriously sticky, set by convention and headquarters policy rather than recalculated against each project’s effect on the general return. The payback criterion deserves special notice from this debate’s participants, because a firm that ranks projects by speed of payback systematically favours techniques that recover their advance quickly, which can mean lighter, shorter lived equipment, the opposite bias from the heavy mechanisation the falling rate story requires, or can mean ruthless labour saving wherever wages dominate the outlay to be recovered, depending on the structure of costs.

None of this settles the formal dispute, and a guide committed to track discipline should say why. Okishio’s criterion is a modelling postulate about which adoptions survive competitive selection, not a claim about the contents of managers’ heads, and a selection argument can hold even when the agents’ stated reasons are a muddle, provided the market punishes the choices the postulate excludes. The practice evidence disciplines the postulate at a different joint: it reminds both sides that the economy’s actual selection machinery runs on conventions, accounting categories and organisational routines that neither the equilibrium model nor its temporal rival represents, and that the distance between the stylised adopter and the real investment committee is one more idealisation whose cost should be acknowledged when the conclusions are carried back to history. The formal track’s results survive the acknowledgment, as conditional results always do. What changes is the confidence with which they are pronounced over centuries of institutional variety, from the family workshops of Marx’s Manchester to the divisionalised corporations whose officers Graham and Harvey surveyed.

Simultaneous valuation and the world it assumes

The deepest reply to Okishio, and the one around which the formal track now organises itself, attacks neither the wage assumption nor the adoption criterion but the accounting underneath both. The tradition descending from Bortkiewicz through Sraffa to the modern textbooks values the inputs and outputs of a production period simultaneously. A ton of steel that enters a works in January and a ton embodied in the machines that leave it in December enter the equations at the same price, because the system of equations that determines prices of production has one price for each commodity, full stop. Simultaneity is not a casual habit. It is what makes the system determinate: with one price vector solving all the equations at once, the rate of return can be computed from physical data and the wage bundle, and Okishio’s result follows as a theorem about that computation.

Notice what kind of world simultaneity describes. It is a world photographed at an instant, in which the economy replaces itself exactly, inputs flowing in priced identically with outputs flowing out. Time, in such a photograph, does no work. Nothing is bought dear and sold cheap because the calendar advanced. Nothing is stranded by a change in relative prices between the day of purchase and the day of sale. Defenders of this accounting call it the only coherent way to state an equilibrium theory, and they are right that a theory of rest points needs it. Its critics call it a photograph mistaken for a film. In the film that actual capitals live through, inputs are bought at the prices of one season and outputs sold at the prices of another, and when technical progress is pressing prices downward, the gap between the two price sets is systematic rather than random. That gap is the crack into which the temporal interpretation inserts its lever.

It is worth pausing on a point of fairness that partisans on both sides sometimes skip. Nobody in this dispute accuses anybody of an arithmetical mistake. Okishio’s proof, granted simultaneity and his other premises, is correct, and the temporal school grants it. The temporal constructions, granted sequential valuation, are also correct, and careful simultaneists grant that. What divides the schools is which set of premises describes the object Marx’s law is about: a self replacing state contemplated by theory, or a sequence of monetary advances and returns unfolding in calendar time. That question is partly interpretive, partly methodological, and only at its edges empirical, which is why it has outlasted several attempts to declare it closed.

The temporal single-system interpretation

The temporal single-system interpretation, known by its initials TSSI, was developed from the nineteen eighties onward, principally by Alan Freeman and Andrew Kliman, with contributions from Guglielmo Carchedi and others gathered in the volume Marx and Non-Equilibrium Economics. Its name carries its two commitments. Single system: value and price are not two separate systems between which a transformation must be negotiated, but one system expressed in money, with the monetary expression of labour time serving as the bridge magnitude that lets aggregate value relations and aggregate price relations hold together. Temporal: magnitudes are determined in sequence. Inputs enter production bearing the prices actually paid for them; outputs emerge bearing new prices formed at the period’s end; the value transferred from means of production to product is the value those means carried when they were bought, not a value recomputed after the fact from the prices at which their replacements will trade.

Applied to Okishio’s scene, the temporal machinery changes the verdict without touching the theorem’s algebra. Let the innovator adopt a technique that passes the viability test at standing prices, exactly as the theorem requires. Let the technique generalise and output prices sag, as productivity growth presses them down. Now measure the rate of return the way an actual balance sheet must: funds were advanced at the old, higher prices for machinery and materials, while receipts arrive at the new, lower prices for the product. The capital laid out is a historical sum of money, and it does not shrink retroactively because replacements have become cheaper. On that measure, Kliman’s constructions show, a sequence of perfectly viable, cost cutting innovations can trace a falling path for the rate of return even while the physical surplus grows and even while the simultaneous computation, repricing the inputs at their new cheapness, reports the rate unchanged or improved. The full exposition of this reading, with its textual case from Marx’s own pages, is given in the companion treatment of the new interpretation and the temporal single system, and only its bearing on Okishio needs to be carried here: the theorem is true in its world, and its world is not the only one consistent with Marx’s text or with the calendar.

What changes when inputs carry the prices actually paid for them?

The temporal reading records what capitalists actually advanced at the start of the period, rather than repricing their inputs at end of period values. In a world of falling prices, the two records diverge, and a technique sequence that leaves the simultaneous rate unchanged or higher can register a falling rate when the money actually laid out is the measure.

The textual warrant the temporalists claim deserves statement in its strongest form, since the objection considered next is strong and deserves a worthy target. Marx, they note, computes with cost price, the sum actually expended, and remarks in the third volume that the value of the means of production consumed enters the product at the value those means possessed when they entered production. His own illustrative numbers proceed historically, capital by capital, without ever repricing an input at an end of period value his text had not yet established. On the single system side, they add that the famous aggregate equalities, total value equals total price and total surplus value equals total profit, can be sustained under their reading, which removes the motive for treating Marx’s value theory as a redundant detour. Whether these warrants persuade depends on the reader’s judgement about the manuscripts, and the textual strand of the logic track is the proper home for that judgement. The formal point stands independent of the textual one: sequential valuation is a consistent accounting under which Okishio’s conclusion does not follow, and consistency is all a reply to a theorem strictly needs in order to reopen the question.

A neighbouring answer: the New Interpretation

The linked companion article treats the temporal school together with a second monetary reading of Marx, and the two are so often shelved together that a short act of separation belongs in this guide as well. The New Interpretation, developed by Gérard Duménil and Duncan Foley in the nineteen eighties, shares the single system conviction that value and price form one monetary system rather than two, and it defines the bridge magnitude, the monetary expression of labour time, as the ratio between money value added and living labour performed. Where it parts from the temporal school is on sequence. The New Interpretation is largely content to work period by period with aggregates, preserving the equality between total profit and total surplus value by construction, and it does not rest its defence of Marx on sequential repricing of inputs. Its bearing on the falling rate is correspondingly indirect: its authors have generally been among the law’s measured critics, treating the tendency as one mechanism among several and directing their energy toward the statistics of the record track, where their reconstruction of American profitability set the terms that Kliman’s rival series then contested.

The separation matters for a practical reason. Polemics against the temporal reply frequently borrow arguments aimed at the New Interpretation, and defences of one monetary reading are quoted as though they secured the other. They do not. A reader can accept the monetary expression framework and reject sequential valuation, accept sequential valuation while doubting the aggregate equalities are the right hill to defend, or accept both, and each combination is occupied by serious scholars. This is the track discipline operating at fine grain: within a single family of interpretations, distinguishable claims travel under a shared label, and the label does argumentative work the claims have not earned. The habit of asking which claim, exactly, a given argument supports, is the whole method of this article compressed into a reflex.

The strongest objection to the temporal reply

The brief for this article instructs that one objection be stated at full strength, and it is the right instruction, because the objection is the most serious thing that can be said against the temporal rescue of the law. It runs as follows. A theorem was proved: under stated conditions, the equilibrium rate cannot fall. The reply does not dispute the proof. It changes the bookkeeping. Inputs are to be entered at one set of prices and outputs at another, and under the new bookkeeping the forbidden result becomes possible. But bookkeeping is chosen, and a convention chosen by the defenders of a proposition, after the proof against it is known, because under that convention the proposition survives, invites the suspicion that has attached to every rescue operation in the history of doctrine: the convention is doing the work that evidence and argument were supposed to do. Had sequential valuation delivered a rising rate, its inventors would not have brandished it as the authentic reading of Marx. The selection criterion is the conclusion.

The objection deepens when it turns from motive to substance. Simultaneous valuation, it argues, is not an arbitrary formalism but the bookkeeping of the market itself. Capital is not a museum of historical receipts. A machine bought dearly in one season, whose replacement is cheap in the next, has lost value in the only sense that matters to its owner, as the write downs on every balance sheet in every recession testify. To insist on carrying the machine at its purchase price when computing the rate of return is to protect the falling rate by protecting an accounting fiction, while the actual economy marks capital to market with indifference to anyone’s doctrine. And the wound is completed by an observation about what the temporal constructions actually exhibit: a falling rate constructed under falling prices is a rate measured against a shrinking measuring rod, fully compatible with a growing physical surplus, growing output and growing capacity to employ. The economics, says the objection, describes an economy getting richer. Only the accounting describes decline. A defence that changes the accounting rather than the economics has conceded the substantive point and kept the words.

That is the objection at full strength, and parts of it land. The answer available to the temporal school is nonetheless substantial. Simultaneity is also a convention, and a far more violent one when measured against the calendar: it asserts that the steel bought in January and the steel embodied in December’s product carry one price, although no market ever quoted that price for both transactions, and although the assumption’s only warrant is the determinacy it lends to a system of equations. Between two conventions, the one that respects the order in which money is actually advanced and actually recovered has at least a claim to realism that its rival lacks. Marx’s own computations, the temporalists add, proceed historically at the decisive points. And the measuring rod jibe can be returned: every rate of return in monetary economies is measured against money sums laid out in the past, which is why accountants distinguish historical cost from replacement cost rather than declaring one of them fictitious.

The honest concession, which this article makes on the debate’s behalf, is that the dispute has migrated. It began as an argument about whether capitalist development depresses profitability. It has become, in significant part, an argument about which idealisation misleads less when profitability is defined: the self replacing state in which time does no work, or the sequence in which stocks are carried at historical sums while prices slide. That is a real question, and methodological questions are not evasions. But it is a different question from the one the public argument imagines itself to be settling, and the migration explains why the formal track resists closure. To settle it, one side would have to show not that its bookkeeping is consistent, which both have shown, but that the other’s idealisation distorts some identifiable feature of capitalist motion that its own preserves. Attempts at exactly that demonstration fill the journals, and none has commanded general assent.

Where the formal dispute comes to rest

The secondary literature around the temporal reply illustrates the resting point precisely. Roberto Veneziani pressed the sharpest technical charges, arguing that the temporal constructions were underdetermined in some formulations and inconsistent in others, able to generate their results only by leaving key magnitudes insufficiently pinned down. Kliman and Freeman answered in kind, contending that the charges dissolve once the interpretation is stated with its own premises intact, and that several of the alleged inconsistencies arise from reimposing simultaneist requirements on a theory that rejects them, which begs the question in the charging document. David Laibman, defending the simultaneist tradition from within Marxian economics, has argued that equilibrium theorising of that kind is not a betrayal of Marx but the natural completion of his price theory, and that the temporal alternative purchases its results at the price of coherence with the rest of the system. Each exchange leaves both frameworks standing, internally articulated, with the disagreement relocated one level down, to premises about time, money and equilibrium that the disputants hold for reasons reaching beyond this single law.

Why can rival valuations each be internally consistent?

Each valuation builds its own coherent bookkeeping and proves its results inside it, much as rival geometries each prove their theorems. Consistency is a property of the system, not a verdict on the world, so the choice between the systems has to be argued from what each assumes about time, money and how prices actually form.

A reader is entitled to ask whether this is a scandal. It is not, and the comparison with other sciences calms the suspicion. Rival statistical frameworks in econometrics, rival equilibrium concepts in game theory and rival measures of national product all exhibit the same structure: consistency inside, contest at the border. The formal track’s mature finding can therefore be stated in one sentence that partisans of both schools have reason to accept, even when they dislike its even handedness. Okishio’s theorem permanently settles what mechanisation does to the equilibrium rate under simultaneous valuation at a fixed real wage; the temporal literature permanently settles that the same mechanisation, valued sequentially, can trace a falling path; and no further proof, on either side, can settle which of those settlements describes the tendency Marx asserted, because that question is about the fit between idealisations and a historical process, and fit is argued, not proved. The law’s formal status is thus neither refuted nor vindicated. It is conditional, and the conditions are now known with a clarity the nineteenth century could not have supplied.

Models as arguments: what a proof does to a debate

The formal track offers a clean specimen of a general phenomenon that students of economics should learn to recognise wherever it occurs: a proof does not end an argument, it rearranges one. Before 1961, the debate about the law ran on intuition and authority, with each side’s adherents convinced that the mechanism obviously did what they said it did. After 1961, nobody could responsibly assert the strong claim, that mechanisation as such depresses the equilibrium return at a fixed real wage, without answering the theorem, and the burden of proof acquired an address. That is a permanent change in the debate’s constitution, achieved not by new facts but by a new standard of explicitness. The temporal reply then performed the same service in reverse: by exhibiting a consistent neighbouring framework in which the forbidden path reappears, it showed that the theorem’s apparent finality had been borrowed from its framework, and the burden acquired a second address. Debates mature in exactly this way, by converting disagreements about conclusions into disagreements about premises that can be inspected one at a time.

Two corollaries deserve emphasis. A proof’s political afterlife is independent of its content: the theorem has been celebrated by writers who never examined its wage assumption and mourned by writers who never examined its valuation frame, because a result with a name and a date behaves in public argument like a trophy. And the correct response to a trophy result is the one this article has modelled throughout: neither worship nor resentment, but a request for the premises, followed by the separate question of whether the premises describe the territory the original claim was about. Readers who internalise that response own a piece of equipment that transfers to every formal dispute in economics, from growth theory’s aggregation quarrels to the modern literature on inequality, where theorems with trophies attached circulate in exactly the same way.

Limiting cases and a neighbouring exact result

Formal claims are best understood at their edges, and the theorem rewards a tour of its boundary cases. Innovation confined to a pure luxury trade, jewellery say, or the production of racehorses, leaves the general rate exactly where it stood, however dazzling the cost saving, because the cheapened product enters neither the workers’ bundle nor any production process, and the surplus available for distribution across capitals is unaltered in the magnitudes that matter. Innovation in the wage goods trades sits at the opposite pole of consequence: by cheapening the bundle’s contents it shrinks the labour time the bundle commands, enlarging the surplus share at any given money wage settlement and lifting the ceiling on the rate across the economy. Innovation in a basic industry that supplies every trade, energy or steel or transport, propagates its saving through all cost structures at once. The gradient among these cases is the analytical content behind the theorem’s quiet insistence that not all mechanisation is equal, and it supplies the record track with a testable refinement: if profitability movements are to be read through this lens, the sectoral location of technical change should matter to the aggregate outcome, a proposition decomposition studies can and do examine.

The neighbourhood of the theorem contains one more exact result that readers of this debate meet constantly and should not confuse with it. The Fundamental Marxian Theorem, established in the form associated with Morishima and generalised by Roemer, proves that positive profits in a price system exist exactly when surplus labour is performed in the corresponding value system: exploitation, in the precise sense of labour performed beyond what the wage bundle embodies, is the necessary and sufficient condition for profit to be positive. The result matters here as a demonstration of what formalisation can do for this tradition when it behaves. It neither saves nor damns the falling rate, since it concerns the existence of profit rather than its trend, but it rebuts in advance any lazy assurance that Marx’s framework dissolves under mathematical handling. Parts of the framework formalise beautifully. The falling rate is the part where formalisation delivered a conditional negative for the strong claim and, through the valuation dispute, a conditional reprieve, and the contrast between the two outcomes is the formal track’s honest portrait.

One innovation, two account books

Abstractions about valuation become vivid when a single episode is forced through both account books in turn, so consider the steel method from the earlier sketch once more, now followed through time. In the year of adoption, the pioneering firm advances money at the prices then standing: it orders the heavier equipment, pays the invoices its suppliers quote that season, and hires its smaller crew at the going wage. Suppose the equipment bill, reckoned at purchase, comes to a sum whose annual charge, spread across the kit’s working life, together with materials and the reduced wage bill, beats the old cost by a comfortable margin. The pioneer’s books, kept as businesses keep them, on historical sums, show a fat surplus over the year. Nothing in the temporal account disputes any of this; the dispute begins when the method has spread and the price of steel has slid to meet the new costs, for now the pioneer’s successors, and the pioneer itself in its later years, sell into a market whose prices were set by the cheapened technique.

Run the episode through the simultaneous book first. That book reprices everything at the settled values of the final configuration. The equipment in place is valued at what such equipment costs in the new configuration, where the machines that make machines have also been cheapened by parallel innovations; the wage bundle costs what it costs when wage goods are produced by improved methods; and the rate computed on those restated magnitudes comes out at or above the rate of the original configuration, as the theorem promises. The book is coherent, and its moral is that the economy’s capacity to generate a surplus over a fixed standard of workers’ consumption has grown. Now run the same episode through the temporal book. That book refuses the restatement. The capital advanced was a definite sum of money, laid out at definite past prices, and the return on that sum is computed against receipts at the prices the sliding market actually paid, year by year. On this book the later years of the installation can show a meagre or vanishing return on the historical advance, because the output price fell away beneath a capital sum that history had already fixed. The book is also coherent, and its moral is that owners who advanced money at the old prices were expropriated in part by the price slide their own industry’s progress produced.

Neither book is lying. They answer different questions, about capacity in one case and about the fate of sums actually advanced in the other, and the tragedy of the formal debate, if the word is permitted, is that both questions present themselves to intuition as the question, what happened to the rate of profit. Marx’s text, written by a man who thought in terms of money advanced and money recovered across the turnover of capital, leans naturally toward the second book in its illustrative passages, which the temporalists cite. The equilibrium constructions of his price theory, which require one price for each commodity within a period’s accounting, lean toward the first, which the simultaneists cite. The reader who has followed the episode through both books owns the dispute in a way no summary can confer: the disagreement is not about arithmetic, honesty or loyalty, but about which of two coherent questions the law should be understood as asking.

Remaining thickets: turnover, rent and the unfinished agenda

Several technical questions border the debate without belonging to its centre, and a guide owes its readers a map of the border posts. Turnover time matters because the rate that concerns owners is annual, while the rate in Marx’s formula is struck per turnover of capital; techniques that tie funds up for longer periods, in heavier plant or slower maturing processes, depress the annual figure through a channel distinct from composition, and formal treatments differ in how cleanly they fold turnover into the comparison of equilibria. Rent matters because a part of the surplus is siphoned to landowners before the industrial rate is struck, and the treatment of resource rents, swollen in eras of expensive energy and minerals, moves measured profitability in ways that have nothing to do with mechanisation. Joint production, where one process yields several products together, matters because it strains every valuation scheme yet devised and produced Steedman’s most damaging constructions. Finance matters because the share of profit accruing as interest, and the leverage with which returns on equity are geared to returns on total capital, can make the profitability that owners experience diverge from the profitability of production for decades at a time.

Each thicket has its own literature, its own specialists and its own unresolved core, and each is capable of absorbing a career. Their collective lesson for this article’s reader is one of scope. The three tracks organise the debate about the law’s core mechanism; they do not exhaust the economics of profitability, and a falling or rising measured rate in any concrete episode will typically have turnover, rent, jointness and financial structure braided through it alongside composition and wages. The honest analyst decomposes before pronouncing. That discipline, unglamorous as it sounds, is the working form of everything the track metaphor teaches at the level of theory.

Interest, credit and the rate that owners actually feel

A persistent source of public confusion in this debate is the slippage between the general rate of return on capital in production and the rates that financial pages report. Marx’s own architecture separates them with care. The surplus generated in production is divided, in the third volume’s later parts, between interest, claimed by the owners of money capital, and profit of enterprise, retained by the functioning capitalist who organises production. The rate of interest has its own determinants, in the supply and demand for loanable funds and the conventions of the money market, and across the nineteenth century it trended gently downward on the safest assets even as industrial fortunes multiplied, a pattern the classical economists read through their own theories of falling profit. Nothing in that pattern speaks directly to the tendency of the production side rate, and arguments that enlist bond yields as evidence for or against the law commit a small, common category error, mistaking the price of one slice of the surplus for the size of the whole.

Credit adds a second mediation that the debate’s cleaner models set aside. Leverage gears the return on owners’ equity to the return on total capital employed, magnifying it in booms and inverting it in slumps, while the expansion of credit can sustain accumulation through stretches in which the underlying production side return is already thinning, deferring the reckoning and concentrating it into a financial crisis when it arrives. Several interpreters of the late two thousands crisis, working inside the tradition this article surveys, argued exactly that sequence: a long erosion of production side profitability masked for years by financial expansion, until the mask itself became the crisis. Others, pointing to the extraordinary profits of the financial sector in the same years, read the episode as evidence that surplus was abundant and its problem was location rather than quantity. The article takes no side in that quarrel here; its relevance is disciplinary. Any attempt to bring the record to bear on the law must first decide how finance enters the accounts, as a claimant on surplus, as a producer of it, or as a veil over its trend, and the three decisions yield three different records from the same raw history. The thicket is not an excuse for agnosticism. It is the price of admission to the empirical conversation, payable in stated conventions.

The neighbouring challenge: Sraffa and redundancy

One neighbouring dispute must be sketched, because readers constantly import it into this one, although the full treatment belongs to the companion study of Sraffa and the Marxian system. Piero Sraffa’s slim Production of Commodities by Means of Commodities showed that, given the physical data of production and the real wage, prices and the uniform rate of return are determined without any reference to labour values. Ian Steedman’s Marx after Sraffa drew the conclusion without flinching: the value magnitudes of Marx’s theory are a redundant detour, since every result obtainable with them is obtainable, and some results are obtainable only, without them. Steedman also exhibited constructions, particularly under joint production, in which positive profits coexist with negative surplus value, which he took to sever the link between profit and exploitation at the theory’s root.

The distinction between this challenge and Okishio’s matters more than their frequent bundling suggests. Okishio works inside the shared framework: prices of production, a uniform rate, techniques chosen for profit. His question is what that framework implies about the rate’s movement, and his answer embarrassed the law from within its own house. Steedman’s question is whether the house needs its value theoretic foundations at all. A Marxist can concede Steedman’s redundancy point, rebuild exploitation theory on physical and monetary foundations as Roemer did, and still fight Okishio’s corner on wages and valuation; another can defend the value framework through the temporal reading and regard the Sraffian constructions as artefacts of simultaneity. The positions cross cut, and any map that lines the critics up on one side and the defenders on the other, with Okishio and Steedman as interchangeable exhibits, has merged two disputes that share personnel but not premises. Keeping them apart is the track discipline applied within the formal track itself.

The junction with the transformation problem

Every participant in this debate works in the shadow of a larger one, and a paragraph of orientation will save readers from conflating them. The falling rate is first derived in the vocabulary of values, from surplus value and the capitals it sets against, while the rate that competition equalises and owners experience belongs to the vocabulary of prices of production. The passage between the two vocabularies is the transformation problem, whose full treatment belongs to its companion article, and the falling rate debate inherits its fortunes without being reducible to them. Bortkiewicz’s original critique touched both territories at once, showing that Marx’s own sketch of the passage was incomplete, and much twentieth century argument ran the two disputes as a single front. The modern separation is healthier. The temporal school’s single system claim, if accepted, dissolves the junction by denying that two systems ever existed, which is one reason its adherents regard the interpretation as a master key. Simultaneist formalists, for their part, are content to state the falling rate question wholly in price terms from the outset, as Okishio did, letting values fall where they may. The reader’s discipline is to notice, in any given contribution, which vocabulary the rate belongs to and whether the author has earned the passage between them, since several famous refutations of the law are, on inspection, refutations of a value stated rate by evidence drawn from a price stated world, a track confusion conducted at the level of accounting languages rather than of questions.

Before there were statistics: the debate’s evidential childhood

It sharpens the record track to remember how young its evidence is. For the first seventy years of the law’s career, no participant could compute a national rate of return, because the statistical apparatus for doing so did not exist. National income accounting is a creation of the nineteen thirties and forties; capital stock estimates built by the perpetual inventory method are younger still. The nineteenth century argument therefore ran on the logic and textual tracks by necessity, fortified by impressions: the visible decline of interest rates on the safest securities across the century, the testimony of manufacturers about thinner margins, the brute fact of recurring commercial crises. Marx himself argued with the materials of a brilliant journalist and a British Museum reader, factory inspectors’ reports, parliamentary papers, the financial press, and his handling of them was formidable by the standards of his day. But no one should mistake the debate’s classical form, textual exegesis armed with anecdote, for a choice its participants made freely. It was the only form the available evidence allowed.

The record track’s arrival changed the debate’s centre of gravity in ways still working themselves out. Once series could be built, the question of what happened acquired answers of a new kind, and the older tracks were demoted, in the profession’s esteem, to preliminary skirmishes, although nothing about a series can answer the theorem or the manuscript, as this article has insisted. The demotion had one healthy effect and one unhealthy one. Healthily, it forced defenders of the law to specify which measured quantity their claim concerned, ending the era in which any inconvenient decade could be waved away. Unhealthily, it encouraged a generation of summaries in which the debate simply is the statistics, with the formal result demoted to a footnote and the textual questions to antiquarianism. The three tracks table is this article’s protest against that demotion, filed on behalf of proportion. Each track is a real tribunal. Each has limited jurisdiction. The mature debate is the one that keeps all three in session and forbids verdicts to wander between courtrooms.

Track three: the empirical record

Suppose both theoretical disputes were suspended and the question put to the data directly: have returns on capital in the advanced economies fallen across the long run? The record track’s first lesson is that this question, which sounds like the simplest of the three, has no answer independent of a chain of measurement decisions, each defensible and each consequential. The numerator requires a decision about what counts as profit, before or after tax, including or excluding the earnings of finance and of the unincorporated sector. The denominator requires a capital stock, and capital stocks are not found in nature; they are built, by cumulating investment and retiring assets under assumed service lives and depreciation profiles, and the builder must choose whether to value the result at historical cost, what was actually paid, or at current cost, what replacement would now require. The choice is not cosmetic. In an era of falling equipment prices, historical and current cost measures of the same physical stock diverge systematically, and a rate computed on one can trend down while its twin trends flat. Add the classification decisions, which industries count as productive of surplus, whether supervisory and commercial labour belongs in the wage bill of production or in the surplus’s deductions, and the honest researcher confronts a fan of possible series where the public argument imagines a single line.

The companion survey of the measured record in detail assembles the principal reconstructions; this article needs only their pattern of agreement and disagreement. Agreement first: nearly every serious reconstruction, whatever its conventions, finds American profitability lower at the start of the nineteen eighties than at its postwar height in the middle nineteen sixties, with a decline that spans roughly a decade and a half. On that stretch of the record the tendency’s friends and the profit squeeze school can both point to a fall, while differing over its cause. Disagreement begins with the decades that follow. Gérard Duménil and Dominique Lévy, in the most influential reconstruction from within the Marxist tradition, find a substantial recovery thereafter: on their measures, American profitability climbed back through the nineteen eighties and nineties to recover much, though not all, of the lost ground, a recovery they attribute to the political and managerial counteroffensive of the neoliberal period. Kliman, valuing the corporate capital stock at historical cost and treating depreciation by the tax based conventions he argues are conceptually required, finds no such recovery: on his reconstruction the American corporate rate continues its descent, and the crisis of the late two thousands arrives at the end of a long profitability decline rather than after a restored golden age.

Between those poles, the econometric literature has tried to test the mechanism rather than the line. Deepankar Basu and Panos Manolakos subjected American postwar data to formal testing and reported evidence consistent with the law’s specific machinery: a rising composition of capital exerting downward pressure on the rate, partially offset by movements in the rate of surplus value, with the net trend significantly negative across their sample. Work by Lefteris Tsoulfidis and collaborators on the Greek economy, and studies assembled for other national cases, report patterns in which profitability declines across long subperiods and composition effects are visible in decompositions. None of these studies commands the field, and each has drawn methodological rejoinders about variable construction and identification. What they collectively establish is narrower and more valuable than a verdict: the mechanism Marx specified can be found operating in the data when researchers look for it with the right decompositions, which is a different claim from the proposition that the mechanism always wins, and a stronger claim than the proposition that the whole law is a scholastic ghost.

Can a statistical series referee a theorem?

No, and the reason is structural rather than evasive. A theorem states what follows inside a defined model, and data drawn from an economy that matches none of the model’s assumptions exactly can neither confirm nor refute the deduction. Series can test the law’s empirical cousin, the claim that measured profitability trends downward, which is a different and answerable question.

The grading for this track follows from its structure. Settled: profitability in the leading economy fell across the postwar decades to an early nineteen eighties trough, on almost any convention a researcher chooses. Genuinely contested: the direction since, which turns on measurement philosophy as much as on the world, since the historical cost and current cost partisans are replaying, inside statistics, the same quarrel about how capital should be valued that divides the formal track. Open: whether the mechanism’s pressure, where econometrics finds it, is strong enough to dominate the long run, which is the empirical shadow of Sweezy’s logical question and inherits its difficulty. A track whose deepest dispute mirrors the other tracks’ deepest disputes is not thereby corrupt. It is thereby shown to be the same argument conducted in the medium of numbers, and the reader who has learned to separate the tracks will recognise the valuation quarrel wearing statistical dress and will not mistake a new costume for a new argument.

Decomposing the fall: what the American studies found

The record track’s most instructive genre is not the long series but the decomposition, the study that takes a measured decline and apportions it among rival mechanisms. Its founding exhibit is Thomas Weisskopf’s examination of the postwar American economy, published in 1979, which separated the movement of the profit rate into the contributions of a rising composition of capital, a rising share of wages pressing on profits, and shortfalls of demand that left capacity idle. His apportionment varied by subperiod in a way that has disciplined argument ever since: the wage share mechanism carried much of the explanation for the squeeze of the late nineteen sixties, while composition and realisation problems grew in weight in the decade that followed. A single national decline, decomposed, turned out to be a relay race in which different mechanisms carried the baton in different laps, a finding that embarrassed every monocausal reading in roughly equal measure.

Fred Moseley’s study of the same economy pressed a different variable into the foreground: the division of labour between production workers and the growing ranks of supervisory, commercial and administrative employees whose salaries, on his accounting, are paid out of surplus rather than adding to it. On Moseley’s reconstruction, a rising share of unproductive labour absorbed an increasing portion of the surplus across the postwar decades, so that the rate of profit relevant to accumulation fell even where narrower measures looked stable; stripping out the effect, in his counterfactual, removes much of the decline. Edward Wolff’s work approached from the productivity side, connecting the profitability downturn to the much studied slowdown in productivity growth, and finding the channels entangled, since slow productivity growth weakens the cheapening counterweight that otherwise offsets rising composition. The three research programmes do not agree, and were not designed to. What they share is the refusal to treat a falling line as self interpreting, and their disagreements are about measurable things: the classification of labour, the construction of the capital stock, the periodisation of the fall. That is what progress looks like on the record track, not a verdict descending from the clouds but an increasingly precise quarrel about quantities that better archives and better methods can, in principle, arbitrate. A reader choosing which reconstruction to trust should follow the same rule offered for the formal track: identify the premises doing the work, ask whether those premises are defended or merely assumed, and notice how much of the conclusion survives when a rival premise is substituted for one section of the calculation at a time.

Translating the national accounts: the measurement project

A distinct research programme deserves its own paragraph in any honest survey, because it tried to solve the record track’s problems at the root rather than arguing over their fruits. National statistical offices do not collect Marxian magnitudes. Their categories, built for Keynesian demand management and for tax administration, sort labour into industries rather than into productive and unproductive employments, treat government and finance by conventions alien to the theory, and value capital by rules designed for other purposes. Anwar Shaikh and E. Ahmet Tonak, in their 1994 study Measuring the Wealth of Nations, undertook the unglamorous labour of building a translation manual: a systematic reclassification of the American national accounts into the categories of Marx’s theory, sector by sector and flow by flow, with the decisions argued in the open. The resulting series, and the school of measurement work that followed their template in several countries, gave the record track something it had lacked, a set of magnitudes whose construction a critic could inspect line by line and redo differently where the judgement calls seemed wrong.

The translation project’s deepest lesson is that measurement in this field is itself theoretical work, not a preliminary chore that precedes it. Every decision in the manual, whether advertising labour produces value or merely circulates it, whether the state’s employees stand inside or outside the surplus’s production, embodies a reading of the theory, so that the resulting series test the law and the reading together. That entanglement does not make the enterprise futile; all mature sciences measure through theory laden instruments. It makes the enterprise honest in proportion as its decisions are exposed, and it explains why the record track’s arguments so often return, after a tour of the statistics, to questions of interpretation that a naive empiricism expected the numbers to banish. The tracks interpenetrate at the level of craft even while they remain distinct at the level of logic, and holding both truths at once is the advanced form of the discipline this article teaches.

Profit squeeze: the rival explanation

The most durable alternative to the law does not deny that profitability fell in the postwar decades. It assigns the fall to a different variable. Andrew Glyn and Bob Sutcliffe, studying Britain at the start of the nineteen seventies, argued that sustained full employment had strengthened labour’s bargaining position until real wages were rising faster than productivity, squeezing the share of profits in output and throttling investment. American versions of the argument, associated with Raford Boddy and James Crotty among others, read the late nineteen sixties through the same lens: tight labour markets, militant wage settlements, a profit share in retreat. The mechanism has a satisfying concreteness the grander theories sometimes lack. It names dates, institutions and struggles, and it explains why the profitability crisis arrived when it did rather than at the pace of a secular tendency grinding through centuries.

Its relation to the formal track contains the debate’s neatest irony. Okishio’s framework accommodates the squeeze without strain, because a rising real wage is precisely the condition under which his construction permits the equilibrium rate to fall. The theorem that was brandished as the law’s executioner turns out to contain, as a labelled special case, the rival theory of the postwar fall. Nothing about that irony rescues Marx’s mechanism, whose claim concerns composition, not wages; a fall caused by wage pressure vindicates the squeeze account and leaves the law exactly where Sweezy left it. But the irony disciplines both camps. Critics who cite the theorem against the whole tradition must explain why its own wage clause fails to describe the nineteen seventies. Defenders who cite falling postwar profitability as their evidence must show that composition, not distribution struggle, did the work. The squeeze thus functions in this debate as a standing lesson in causal attribution: a falling rate is a fact in search of a mechanism, and at least three mechanisms in the literature can produce one.

Kalecki’s alternative arithmetic

One major figure stands at an angle to the entire debate, close enough to matter and independent enough to embarrass its categories. Michał Kalecki, the Polish economist who arrived at much of Keynes’s system by a separate road and from Marxian starting points, built his account of profits on an identity of disarming simplicity: in a closed economy without a state, the aggregate profits of the capitalist class equal its investment plus its consumption, since workers, in his simplifying construction, spend what they earn. Capitalists, in the aphorism the theory generated, earn what they spend, while workers spend what they earn. The rate of return on the capital stock is therefore governed, in Kalecki’s world, by the decisions that drive investment, by the degree of monopoly that sets the mark up of prices over costs and hence the distribution between wages and profits, and by the saving habits of profit recipients, rather than by the composition of capital that drives Marx’s tendency. His business cycle theory made investment’s own volatility the engine of fluctuations, and his celebrated 1943 essay on the political aspects of full employment explained, from the profit side, why sustained full employment would meet resistance from the owners of capital even when it raised their earnings, since it undermined the discipline of the sack.

Kalecki matters to the falling rate debate in three ways. His distribution theory supplies the profit squeeze school with its analytical skeleton, since the mark up and the wage share are the levers his framework provides, and a rising wage share engineered by strong unions is a Kaleckian event before it is anyone else’s. His identity warns every participant that the aggregate mass of profit is demand determined in the short run, so that a falling rate can reflect faltering investment and realisation failure as readily as rising composition, and decompositions that ignore the demand side are incomplete by his standard. And his example demonstrates that the broad Marxian inheritance contains more than one complete, mutually irreducible theory of profitability, between which the evidence must arbitrate case by case. A debate that sometimes talks as though the tradition stood or fell with a single ratio looks, after an hour with Kalecki, like a family that has forgotten one of its own houses. Restoring him to the map is part of this article’s brief, since the three tracks organise the argument about the law, and Kalecki’s arithmetic is the standing reminder that profitability has other theorists, and the law’s rivals include cousins as well as opponents.

Brenner’s rival: overcapacity and the long downturn

One more named position belongs in the survey, because it contests the whole field’s framing of the postwar record rather than any single mechanism within it. Robert Brenner’s study of the long downturn, first presented in the late nineteen nineties, argues that the advanced economies’ profitability decline from the middle nineteen sixties was driven neither by rising composition nor primarily by wage pressure, but by chronic overcapacity in world manufacturing. On his account, lower cost producers in Germany and Japan entered markets whose prices had been set by earlier, higher cost incumbents, and the incumbents, locked into sunk plant and established positions, could not exit or adjust at the speed the new competition required. Prices were consequently held down across whole industries for years, margins were compressed from the demand side rather than the cost side, and the system’s problem was too much capacity chasing markets rather than too much machinery per worker or too strong a working class.

Brenner’s intervention matters to this article for two reasons. It demonstrates, from within the Marxist tradition and with dense empirical support, that even the agreed fact of the postwar decline sustains rival causal stories among scholars who share methods and sympathies, so that the record track’s grading must remain open at the level of mechanism even where the direction of the series is conceded. And it drew a celebrated counterattack, in a long exchange in the journal Historical Materialism and elsewhere, in which critics charged Brenner with neglecting the supply side dynamics his own evidence contained, while his defenders replied that the profitability schools had confused correlation with mechanism. The exchange is a model of the genre this article recommends: adversaries who share the data, dispute the decomposition, and know exactly which quantities would embarrass their own accounts. A student who reads one round of that exchange learns more about how the record track actually works than from a shelf of summaries, including, the author of this guide cheerfully concedes, this one.

Defenders beyond the temporal school

The temporal interpretation is the most discussed defence of the law in the formal literature, but the tradition’s defence has never rested on a single school, and the survey is incomplete without the alternatives. Ernest Mandel’s mature position, elaborated across Late Capitalism and his later writings on long waves, treats the tendency as a real pressure whose dominance alternates across long phases: in the downswing of a long wave the pressure governs, profitability sags and restructuring follows; in the upswing, a cluster of counteracting influences, cheapened inputs, new fields of investment opened by technological revolution, a defeated and reorganised working class, holds the pressure at bay for a generation. David Harvey’s reconstruction in The Limits to Capital keeps the law and multiplies its mediations, emphasising the spatial and temporal fixes by which surplus capital is displaced into new regions and long term projects, deferring the reckoning the tendency prepares. Anwar Shaikh’s defence, already examined, runs through the theory of real competition rather than through valuation. Howard and King, surveying this landscape in their history of Marxian economics, sorted the crisis theorists into families, fundamentalists who rest crisis on the falling rate, underconsumptionists, profit squeeze theorists and pluralists, and the taxonomy has stuck because it captures a real divergence of strategy among writers who share the canon.

What unites these defences, and distinguishes them from mere loyalty, is that each pays for the law’s survival with a definite concession. Mandel pays with mono-causality, demoting the tendency to first among several mechanisms. Harvey pays with immediacy, interposing the built environment and the credit system between the tendency and any observable crisis. Shaikh pays with the equilibrium framework itself, surrendering the clean comparisons in which the theorem lives. The temporalists pay with the concession recorded in the objection section, that the question has become partly one of rival idealisations. A reader can therefore grade the defences not by their conclusions, which agree, but by their prices, asking which concession costs the theory least. That is a far more interesting exercise than counting adherents, and it is available only to readers who have separated the tracks finely enough to see what each defender is actually buying.

The wage across the three tracks

No magnitude illustrates the track separation better than the real wage, because each track handles it in a different grammatical mood. On the mathematics track the wage bundle is an assumption, fixed by stipulation so that the effect of technique can be isolated, and the fiercest disputes there are really disputes about whether that stipulation drains the exercise of meaning. On the logic track the wage is a variable inside the theory, pressed downward by the reserve army and upward by accumulation’s demand for labour and by workers’ organisation, with Marx’s text supporting the picture of a contested magnitude whose level is set by class struggle within limits set by accumulation. On the record track the wage is an outcome to be measured, and its measured behaviour, rising across the long twentieth century in the advanced economies roughly alongside productivity, with plateaus and accelerations that vary by decade and country, is the single most consequential empirical regularity in the entire debate. A world in which real wages had stayed frozen at nineteenth century levels would have vindicated the theorem’s premise and made the law’s strong form plausible; the world that occurred, in which the bundle grew for generations, is the world in which Okishio’s own escape clause does heavy explanatory work and in which wage centred accounts of profitability movements acquired their standing.

Seeing the wage in three moods dissolves several stock polemics. When a critic announces that the theorem refutes Marx because wages are not in fact constant, the announcement confuses moods: the theorem never asserted constancy as a fact, and Marx’s theory, on the logic track, does not either. When a defender announces that rising wages refute the theorem, the same confusion runs in mirror image, since a conditional result is untouched by the failure of its antecedent in any particular era. What the record’s wage history actually does is relocate the argument. If profitability movements across the golden age and its aftermath were substantially wage driven, the interesting theoretical questions concern the determinants of the wage share, bargaining institutions, full employment policy and the reserve army’s modern forms, and the Marxist tradition contains rich resources for those questions that owe nothing to the falling rate’s fate. The law could lose on every track and leave the tradition’s economics of distribution standing. That observation, obvious once stated, is persistently forgotten by combatants who treat the tendency as the load bearing wall of the entire edifice, and the forgetfulness is itself a species of the track confusion, in which one proposition is made to carry the evidential weight of a whole research programme.

Counterweights at world scale

Marx’s list of counteracting influences reads, at first glance, like a miscellaneous shelf of exceptions. Read against the subsequent century and a half, two entries on the shelf swell into structural features of the modern world economy, and the debate’s treatment of them shows the logic track and the record track interacting at their best. Foreign trade, which Marx credited with cheapening both the elements of constant capital and the goods that enter workers’ consumption, has been generalised by container shipping, integrated supply chains and the incorporation of hundreds of millions of new workers into the world market, into a standing machinery for holding down the value of both components of the denominator and the wage bill alike. Defenders of the law who take the counterweights seriously must therefore explain why the pressure should be expected to prevail in an era unusually rich in counterweights, and the better defences do exactly that, pointing to the cheapening of wage goods as a force that raises the feasible rate of exploitation in step, to the exhaustion of fresh labour reserves as integration matures, and to the way trade itself raises composition in the exporting economies it modernises. Critics reply that a tendency whose dominance must wait upon the exhaustion of the world market’s margins has conceded the timescale that matters for any usable theory of crisis.

The cheapening of the elements of constant capital has undergone a similar promotion from footnote to headline. The information technology era delivered a collapse in the price of computing power and communications capacity so steep that whole categories of equipment cheapened faster than they accumulated, a development that sits squarely inside Marx’s named counterweight and that several reconstructions credit with supporting profitability in exactly the decades where the rival series diverge. Here the two great measurement philosophies of the record track meet the logic track’s own categories: valuing the stock at historical cost understates the cheapening’s effect on replacement, valuing at current cost registers it fully, and the choice between the books is the choice between asking what owners laid out and asking what their capacity would now cost to rebuild. Nothing in this section settles that choice. The point is that Marx’s own list, drawn in the eighteen sixties, contains the categories under which the twenty first century’s most distinctive economic facts file themselves, which is a large part of why a debate this old continues to recruit new combatants who find their own era prefigured in its terms.

Services, software and the shifting substrate

The economy whose composition the law describes has changed its material basis twice since the tendency was formulated, and each change tests the categories in instructive ways. The shift of employment toward services, which dominate the labour forces of the advanced economies, imports the dynamic William Baumol named the cost disease: activities in which productivity grows slowly, because the labour is itself the product, in care, teaching and performance, become relatively more expensive as wages set by the productive sectors rise around them. For the falling rate debate the shift cuts in two directions at once. Service capitals typically advance far less plant per worker than manufacturing capitals, so an economy tilting toward services exhibits a composition dynamic unlike the industrial paradigm the law was drawn from; yet the same shift swells exactly the categories of employment, commercial, supervisory and administrative, whose treatment divides the measurement schools, so that Moseley’s unproductive labour question grows more consequential as the substrate changes. Software and digital goods press the categories from the opposite side: products whose reproduction costs approach zero strain the labour theory’s ordinary bookkeeping and invite claims that the old ratios have lost their grip, claims the tradition’s defenders answer by relocating attention to the enormous fixed capitals, in data centres, networks and platforms, that the apparently weightless products rest upon. A guide need not arbitrate these frontiers. It need only record that the law’s mechanism was specified for an economy of mills and mines, that its translation to an economy of hospitals, logistics and code is genuine theoretical work rather than a reflex, and that the quality of that translation, sector by sector, is one of the places where the debate’s next generation will earn its keep.

The law in a world of unequal wages

The theorem assumes one economy with one wage bundle, and the world economy violates the picture in an instructive direction. Arghiri Emmanuel’s theory of unequal exchange, published at the end of the nineteen sixties, began from the observation that capital moves across borders far more freely than labour, so that rates of return tend to equalise internationally while wages, set by national histories and institutions, remain stubbornly unequal. In such a world the production of a commodity in a low wage economy, sold at prices formed on the world market, transfers value toward the high wage economies whose consumers and capitals capture the difference. The bearing on the falling rate runs through the counterweights. Trade with low wage regions cheapens the wage goods and materials of the high wage economies on a scale Marx’s foreign trade chapter could only sketch, propping up their profitability by exactly the mechanism he named, while the economies supplying the cheap goods industrialise under compositions and wage levels entirely their own. Asking whether the rate falls then requires specifying whose rate, in a system where the equalised world rate and the divergent national rates tell different stories, and the empirical literature’s drift toward global reconstructions in the debate’s statistical wing is a response to exactly that difficulty. Emmanuel’s own conclusions were contested, including by Marxists who disputed his wage centred mechanism, but the question he forced, about the unit of analysis appropriate to a law of capitalist development in an integrated and unequal world market, has never been unasked since.

Marxists who let the law go

A persistent misreading outside the tradition holds that the law functions as a loyalty test, so that abandoning it amounts to leaving Marxism itself. The personnel of the debate refute that reading by existing. Sweezy set the law aside as indeterminate and remained the dean of American Marxist economics for half a century, his school’s flagship journal treating the monopoly capital framework, not the falling rate, as the tradition’s working theory of mature capitalism. Steedman drew the Sraffian conclusion in full, dispensing not only with the law but with the value theory beneath it, and continued to write as a rigorous participant in the Marxian conversation, answered by its other participants as one of their own. Roemer’s analytical reconstruction kept exploitation and class while discarding the labour theory’s classical machinery and the tendency with it. Heinrich’s scepticism about the law comes from the most textually immersed position in the entire field, that of a scholar editing Marx’s manuscripts for the critical edition, a vantage from which no one has ever accused him of distance from the source.

The catalogue cuts in two directions, and both cuts are instructive. Against outsiders, it shows that the tradition’s internal range on this question is as wide as the profession’s, and that the debate is an argument among economists rather than a creed under siege. Against a certain kind of defender, it shows that rescuing the law is a choice made by some serious Marxists and declined by others of equal seriousness, so that the rescue cannot claim the tradition’s unanimous authority for its premises. What unites the rejectionists is not politics, a point this article has been instructed by its evidence to make firmly: Steedman’s formalism, Sweezy’s institutionalism, Heinrich’s philology and Roemer’s game theoretic instincts share no politics worth the name beyond an interest in whether the proposition is true. Acceptance and rejection of the law track methodological conviction, not allegiance, and any account of the debate that sorts its participants by loyalty has failed to read them.

The stakes: crises and the architecture of Capital

Why does one ratio carry such weight? Because of where it sits in the design of Marx’s masterwork. The first volume of Capital analyses production as if commodities exchanged at their values, and finds exploitation at the system’s heart. The third volume descends to the surface forms, prices of production, rent, interest, merchant’s profit, and there the rate of return is the magnitude through which capital experiences its own condition. Marx calls profit the motive power of production in that volume’s argument, the spur that drives accumulation and the gauge by which capitals judge their health. A secular pressure on that gauge is therefore not one malfunction among others. It is the system reporting, in its own preferred unit, that its success in raising productivity undermines the return on the wealth it has already accumulated. The law is the hinge between the theory of accumulation and the theory of crisis, and the series overview of the crisis theories built around that hinge traces what different generations hung upon it.

Who built the first breakdown theory on the law?

Henryk Grossman, whose 1929 study extended Otto Bauer’s reproduction schemes across many periods and argued that accumulation eventually leaves too little surplus value to valorise the swollen capital stock. His construction made the falling rate carry the full weight of a breakdown prediction, a load later Marxists largely declined to make it bear.

Grossman’s construction deserves its reputation for ingenuity and its fate as a warning. Starting from reproduction schemes that Otto Bauer had used to demonstrate harmonious growth, Grossman simply let the schemes run, period after period, with the composition of capital climbing at the stipulated pace, until the arithmetical moment arrived at which the surplus value produced no longer sufficed to fund both the accumulation the scheme required and the capitalists’ consumption at its customary level. Something, at that point, must give: accumulation falters, consumption is cut, or capitals are devalued in crisis. The breakdown was not prophesied for a date but exhibited as a tendency inscribed in the arithmetic of growth itself. Later Marxists mostly declined the load Grossman hung on the mechanism, for reasons this article’s grading supports. A breakdown theorem needs the pressure to dominate its counterweights with something like necessity, and necessity is exactly what the logic track cannot certify and what the formal track, after Okishio, cannot take for granted. Mandel’s alternative, in which the tendency operates as one force inside long waves whose turning points are settled by class struggle, technical revolutions and world market shifts, asks less of the law and has worn better for it.

The crisis connection cuts the other way as well, against careless use of the law by its friends. Several of history’s great crises arrived on the heels of prosperous, profitable years rather than at the end of long profitability declines; the crash of 1929 closed a decade remembered for exuberant returns, and the financial crisis of the late two thousands was preceded by years in which financial sector earnings reached extraordinary heights. Defenders answer, correctly, that sectoral booms can mask economy wide pressure and that a tendency is not a timetable, and the answer is available precisely because the law claims a pressure rather than a schedule. But the episodes enforce a discipline on everyone. If the law is to explain crises, it must do so through the mediations its own third volume supplies, devaluation, credit expansion, the frantic search for counteracting outlets, and not by pointing at a falling line and a subsequent crash as though adjacency were explanation. Pluralists in the tradition, who hold that no single mechanism generates all crises and that profitability, realisation difficulties, financial fragility and disproportionality take turns in the lead, occupy ground that the record track, with its mixed findings, does nothing to embarrass.

Reproduction schemes: the arithmetic of enough

Grossman’s construction repays a slower look, because it is the point where the law’s logic was first subjected to a fully explicit numerical discipline, decades before the theorem that made such discipline famous. Marx’s reproduction schemes, in the second volume of Capital, divide the economy into two great departments, one producing means of production and the other producing consumption goods, and trace the conditions under which each department’s output matches the other’s needs so that production can repeat and expand. Otto Bauer, the Austrian Marxist, extended the schemes in 1913 to answer Rosa Luxemburg’s contention that expanded reproduction was arithmetically impossible without outside markets, constructing a four year table in which accumulation proceeds smoothly with a rising composition of capital and a growing population. Bauer intended the table as a tranquilliser, proof that the system’s arithmetic contains no breakdown. Grossman took Bauer’s own numbers and simply refused to stop at year four. He let the table run for decades under its own stipulated rules, and he watched a point arrive at which the surplus value remaining after funding the stipulated accumulation could no longer sustain the capitalist class’s consumption at its customary scale, so that further accumulation at the given rate became arithmetically impossible without cutting one of the stipulated magnitudes.

The episode is a parable of the whole debate in miniature, and readers should extract its lesson carefully. Both men computed correctly. Bauer’s conclusion, harmony, and Grossman’s conclusion, breakdown, issue from the same table, differing only in the length of the run and in which stipulated magnitude each author was willing to let bend when the squeeze arrived. The schemes cannot arbitrate between them, because the schemes contain no theory of what bends: of whether capitalists confronted by the squeeze would accept lower consumption, slow the accumulation, devalue capitals through crisis, or revolutionise technique. Every later formal episode replays this structure. A model delivers a conditional; the conditionals on offer are individually impeccable; the argument migrates to the question of which stipulation history will violate first. Recognising the pattern inoculates the reader against a recurring disappointment, the expectation that the next model, being larger or cleverer, will finally relieve us of the burden of judgement about premises. It will not, and the debate’s maturity consists in having stopped expecting it to.

Grossman’s afterlife: breakdown and its critics

The breakdown construction had a contentious career of its own, and its fortunes track the logic dispute with unusual fidelity. Critics in the nineteen thirties, writing in the German movement’s journals and then in exile, seized on the arithmetic’s sensitivity: Grossman’s terminal moment arrived after a stipulated number of accumulation periods, and altering the assumed pace of accumulation or the growth of the workforce moved the breakdown earlier or later almost at will, which suggested to the critics that the result had been built into the parameters rather than derived from the system’s nature. The charge is the ancestor of every subsequent objection to tendency reasoning in this tradition, the suspicion that with enough free magnitudes, any terminal story can be manufactured. Defenders answered that the sensitivity was beside the point, since the construction claimed only that some such moment must arrive under sustained accumulation at a rising composition, whenever exactly it fell. Paul Mattick, the council communist writer who became Grossman’s most faithful later champion, restated the theory for postwar readers and wielded it against the managed capitalism of the postwar settlement, arguing that the state’s growing role in sustaining accumulation was itself evidence that the private mechanism had exhausted its room. Whether that restatement persuaded is another matter; its historical interest lies in showing how the law functioned, in the middle of the twentieth century, as the tradition’s principal theoretical bridge between economics and political expectation, a bridge whose load bearing capacity is precisely what Sweezy and Okishio, in their different idioms, were testing.

One decade as contested ground: the nineteen seventies

Every mechanism in this debate claims the nineteen seventies, and the decade therefore serves as the record track’s examination hall. The facts that all parties must accommodate are unusually rich: profitability in the advanced economies declined from its postwar heights; wage militancy and strike activity crested in many countries around the decade’s turn; a quadrupling of oil prices transferred income on a vast scale toward resource owners; manufacturing competition from rebuilt German and Japanese industry intensified; and the decade closed with a deliberate monetary tightening that raised real interest rates and broke the inflationary regime. The squeeze school reads the sequence from the labour market outward: full employment’s legacy empowered workers, settlements outran productivity, profits were compressed, and investment faltered for want of return. The composition school reads the same years as the delayed harvest of the postwar capital deepening, with the added weight of surplus absorbing employment in services and administration, on Moseley’s accounting, and points to decomposition results in which composition effects dominate the later laps of the decline. Brenner’s school reads the decade from the world market inward: overcapacity in traded manufacturing held prices down against all cost structures, squeezing margins from the revenue side while leaving unit costs, wages and composition as supporting actors. Rent based readings hand the decisive role to the oil transfers, a mechanism Marx’s framework accommodates under ground rent’s modern cousins while assigning it no place in the law itself.

The examiners’ lesson is that a decade, like a ratio, does not interpret itself. Each school accommodates the agreed facts by assigning the weights differently among mechanisms, and the assignments are not arbitrary, because each school’s weighting generates different expectations about the decades that followed: the squeeze school expects profitability to recover when labour is disciplined, as in its reading it did; the composition school expects recoveries to be partial and purchased by devaluation and restructuring; Brenner’s expects relief when excess capacity is finally scrapped or absorbed by new demand. Grading those expectations against the nineteen eighties and nineties is the record track’s live research programme, and it is a healthier spectacle than the headline quarrel suggests, because the schools are at last arguing about the same quantities with rival weights attached. A student could do worse than take the seventies as a first research topic in this field: the data are abundant, the positions are sharply drawn, and the episode teaches, in miniature, every skill the three tracks demand.

What a settlement would even look like

Debates that cannot describe their own resolution conditions tend to become permanent, so it is worth attempting the description track by track. On the mathematics track, a settlement is imaginable and its shape is known: not a new proof, since the conditional results on both sides are already proved, but a compelling argument, addressed to premises, that one idealisation of valuation and equilibrium is the appropriate test for a claim about historical tendency. Such an argument would have to come from the philosophy of modelling as much as from economics, showing either that self replacing states are the legitimate tribunal for tendency claims anywhere in science, or that sequential monetary accounting is the only tribunal that respects the object. Pieces of that argument exist; a version commanding assent across the schools does not, and until it arrives the track’s grade of conditional on both sides is final for practical purposes.

On the record track, settlement has a statistical shape and is therefore harder but more conceivable. It would require conventions for capital measurement defended on grounds independent of the conclusions they produce, series built under those conventions for several economies across a century or more, and decompositions robust enough that substituting a rival convention for one component at a time moves the findings in reported, bounded ways rather than reversing them. The field is closer to that standard than its polemics suggest, since the postwar decline commands broad assent and the live quarrel has narrowed to the decades after the nineteen eighties trough and to the apportionment among mechanisms. On the textual strand, settlement waits on the archive: the critical edition’s completion makes the manuscripts fully citable, and the Heinrich and Kliman readings will be testable against passages as they are published in their full contexts, although judgement about the weight of fragments will outlast the publication programme itself. Naming these conditions is not optimism. It is the recognition that each track’s unsettled state is unsettled for stated reasons, which is a very different condition from the fog in which the single sentence version of the debate conducts its business.

Grading the three tracks

The article can now collect its verdicts, track by track, in the open. On the logic track, the tendency reading of the law is settled doctrine about what Marx claimed, and any treatment that converts the claim into a forecast of monotonic decline is a misreading. The determinacy of the tendency is open. Sweezy’s question, which magnitudes the theory fixes and which it leaves to history, remains the live question, and the strongest defences of the law are those that answer it with auxiliary arguments about wages and technical change rather than with insistence. On the mathematics track, Okishio’s theorem is settled as a conditional result and is no longer honestly deniable by anyone who accepts its premises, among which simultaneity and the fixed wage bundle do the heavy work. The temporal counter constructions are likewise settled as conditional results. What remains contested is premise selection, and premise selection is a methodological and interpretive contest, not a theorem shaped hole waiting for a bigger proof. On the textual strand, the law’s presence in Marx’s manuscript is settled, its editorial shaping is documented, and the state of its author’s late confidence is a genuine scholarly standoff between archive based readings. On the record track, a long postwar decline into the early nineteen eighties is settled to a reasonable standard, the direction after that turning point is contested along measurement lines that replay the valuation dispute, and the mechanism’s detectable pressure in econometric work is a real finding that falls short of a coronation.

Set beside one another, those grades explain the debate’s strange double reputation. To a visitor, the argument looks interminable, a century and a half without a winner. To a participant, most of the territory is surveyed and fenced, and the fighting continues on a few well defined ridges where the remaining questions are hard in identifiable ways. Both impressions are accurate reports of different altitudes. The interminability belongs to the single sentence version of the question, does it hold, which cannot be answered because it is three questions. The progress belongs to the disaggregated version, where theorem, text and series have each yielded results that the next generation inherits rather than re-litigates. A reader who has absorbed the three tracks table owns the disaggregated version, and with it the ability to locate any new contribution to the debate within minutes: which ridge it fights on, what it would take to win there, and whether its author knows that the other ridges exist.

One further discipline follows from the grades, and it concerns the reader’s own position rather than the literature’s. It is entirely possible, and arguably the best supported stance available, to hold a different verdict on each track at once: to judge the tendency reading textually secure while its determinacy remains unproved, to grant Okishio his theorem and the temporalists their constructions in the same breath, to accept the postwar decline and suspend judgement on the decades that followed, and to regard Marx’s late confidence as a question the archive has not closed. Such a portfolio looks indecisive only to those still asking the single sentence question. To anyone who has separated the tracks it is simply what the evidence, sorted by jurisdiction, supports, and it has the practical virtue of being updatable in pieces: a new econometric result revises one holding, a newly published notebook another, without forcing the whole portfolio to be liquidated and rebuilt. Positions in this debate should be held the way the table is organised, row by row, and the reader who learns to hold them so has acquired the skill that the debate, at its best, exists to teach, quite apart from the fate of the law itself.

Why the argument has lasted so long

Longevity on this scale is never an accident, and the causes are worth naming because they predict where the debate goes next. The law sits at a junction where three great questions cross: what technical change does to distribution, how the value of a capital stock should be measured, and whether capitalist growth carries a self undermining dynamic. Each generation’s new instruments reopen the junction. Linear programming and Sraffian algebra reopened it in the nineteen sixties. Disequilibrium and temporal modelling reopened it in the nineteen eighties and nineties. Econometric identification strategies and the complete critical edition of the manuscripts are reopening it now, in the statistical and the textual registers at once. A proposition that can be re-asked in every new technical language the discipline invents does not die; it gets translated.

There is also an institutional cause, less noble and worth admitting in a guide that has promised candour. The law is the most exposed claim in the Marxian corpus, the one that most resembles a falsifiable prediction in a body of theory often accused of avoiding such exposure. That exposure attracts exactly the attention a claim of that shape deserves: critics gravitate to it because refuting it would be a result, defenders because holding it would be a vindication, and journals because the exchanges are guaranteed readers. The result is a selection effect in the literature. The law receives scrutiny out of all proportion to the space it occupies in Marx’s own crisis thinking, where breakdown was never his idiom and where the third volume’s contradictions chapter offers a far richer, less mechanical picture than the single tendency that posterity chose to fight over. Readers should hold both truths together: the debate is genuinely about the questions it states, and it is also larger than those questions strictly warrant, because it has become the arena in which the scientific standing of an entire tradition is symbolically contested. Symbolic arenas are not settled by evidence alone, which is the final reason the argument endures.

The debate’s next instruments

Projecting a dispute’s future is hazardous, but its instruments can be projected with some confidence, because in this debate the instruments have always driven the arguments. On the textual strand, the completion of the historical critical edition, whose volumes continue to appear, places the whole archive in citable form for the first time, including notebooks that earlier generations read in extracts or not at all; computational searching across that corpus will not settle the Heinrich and Kliman standoff by itself, since the standoff concerns the weight of fragments rather than their existence, but it will end the era in which either side could suspect the other of resting on a lucky selection. On the record track, the economic historians’ long run datasets, assembled from tax records, probate material and national accounts across centuries, are extending the measurable past backward toward the period in which the law’s author himself lived, which raises the prospect of testing the tendency against the nineteenth century economy that provoked it, an examination no previous generation could administer. On the formal track, agent based and disequilibrium modelling, in which heterogeneous firms adopt techniques under explicit rules and the aggregate rate is watched rather than solved for, offers a tribunal that neither the equilibrium school nor the temporal school fully owns, and early work in that idiom suggests the verdict will depend, instructively, on the adoption rules programmed in, which returns the argument to Shaikh’s question about what firms actually chase. A debate whose every instrument points back to its oldest questions is not going in circles. It is boring downward, through successive layers of technique, toward the small set of judgements about capitalist behaviour on which everything else depends.

Replication culture is the quietest of the new instruments and may prove the most corrosive to the debate’s old habits. The statistical wing’s classic series were built by hand, in an era when a rival who doubted a number had to rebuild a capital stock estimate from printed volumes to say so. The emerging norm, in economics generally, of published data and code alongside published results converts that heroic standard into a routine one: a reconstruction can be rerun under a rival depreciation rule in an afternoon, and sensitivity, which this article has repeatedly identified as the record track’s real question, becomes a table in the appendix rather than a suspicion in a footnote. Traditions of argument formed before that norm will feel the pressure unevenly, and the falling rate literature, with its deep dependence on constructed magnitudes, will feel it early. The likely consequence is not the debate’s end but its narrowing to the disagreements that survive convenience: the choice of idealisation on the formal track, the weighting of fragments on the textual, and the apportionment among mechanisms on a record whose arithmetic everyone can inspect. Those are the disagreements worth having, and they are quite enough to occupy another generation.

How to read the debate without drowning

The literature is forbidding in bulk and uneven in quality, but a path through it exists, and it follows the tracks. Begin with Marx’s three chapters themselves, the law as such, the counteracting influences and the internal contradictions, in the third volume of Capital; they are shorter than their reputation and stranger than any summary, and every later combatant is arguing about what those pages do and do not contain. Read Sweezy’s discussion next, for the cleanest statement of the internal doubt by a writer who wanted the theory to succeed. Then take Okishio’s paper, which is brief and, with the scene setting given in this article, accessible to a reader without formal training. The temporal case is best entered through Freeman’s chapters in the non-equilibrium volume and through Kliman’s book length defence of his reading of Marx, while Heinrich’s essay on crisis theory in the Monthly Review states the textual scepticism in its most compact and forceful form. For the record track, pair Duménil and Lévy’s reconstruction with Kliman’s rival series and read the methodological appendices rather than the headlines, since the appendices are where the real disagreement lives.

Working habits matter as much as the reading list. Annotate every contribution with the track it occupies before engaging its conclusions, and refuse, as a matter of personal discipline, to let a result migrate across tracks in your own notes. Keep the debate’s moving parts in a structured file, one entry per position, with its premises, its strongest result and its open flank recorded in your own words; a study store such as VaultBook suits exactly this kind of keeping, since a position summarised beside its premises can be rechecked in seconds when a new article claims to have settled what the file shows to be open. The reader who keeps such a file for six months will discover the debate’s worst kept secret: most new contributions restate positions the file already holds, in new notation, and the genuinely novel moves arrive rarely enough to be worth the wait.

What can a headline honestly say about the law?

Only that the claim is disputed at every level at which it can be tested, and that the dispute is genuine. A headline announcing that Marx has been refuted, or vindicated, is reporting one track’s local result as if it settled the others, which is the track confusion wearing a press badge.

Secular cousins: falling returns beyond the Marxian fence

The anxiety that mature capitalism outgrows its own profitability is not a Marxist monopoly, and a brief excursion across the fence repays the traveller. Alvin Hansen, addressing the American Economic Association in 1938, gave the anxiety its most influential mainstream form under the name secular stagnation: a mature economy, its frontier closed and its population growth slowing, would generate savings beyond the investment opportunities able to absorb them, settling into chronic underemployment unless the state supplied the missing demand. The mechanism differs from Marx’s at every joint, running through demography and the demand for capital goods rather than through composition and surplus, yet the family resemblance is unmistakable: a growth process that undermines its own continuation, diagnosed from the system’s internal arithmetic rather than from external shocks. Mainstream growth theory after the war domesticated a gentler version of the same idea in the concept of diminishing returns to accumulation, under which an economy’s return on capital declines as its capital deepens toward a steady state, a proposition taught to every undergraduate and structurally a cousin of the tendency, with the drama removed and the steady state presented as rest rather than as crisis.

The cousins matter to this article for a diagnostic reason. When economists outside the Marxian tradition discuss declining returns, they quarrel about measurement, about whether the relevant rate is a marginal product, a bond yield or a return on corporate equity, and about which countervailing forces, technical progress above all, should be expected to offset the decline, without anyone supposing that a single series or theorem could close the question. The maturity of that conversation, its comfort with conditional claims and rival measures, is the maturity this article has been recommending for the Marxian debate itself. Conversely, the Marxian debate’s obsessive clarity about premises, purchased through decades of formal combat, is a discipline the cousins could borrow with profit. The falling rate, seen from this altitude, is one province of a general problem that every theory of capitalist growth must face: what happens to the return on accumulated wealth as accumulation proceeds, and the province’s quarrels, however local their vocabulary, are contributions to the general problem rather than sectarian noise.

For students and exam candidates

Your examination risk on this topic is not ignorance; it is compression. Under time pressure, candidates collapse three questions into one and write that Okishio disproved Marx, or that the temporal school rescued him, and both sentences forfeit the marks that the question was designed to distribute. Structure your answer by the tracks. State the law as Marx stated it, a tendency with named counteracting influences, and note the determinacy question Sweezy raised. Present the theorem precisely, with its two governing assumptions, the viability of adopted techniques at standing prices and the constant real wage, and give its conclusion conditionally, because the conditionals are where the marks live. Then present the replies in ascending order of depth: the wage clause’s escape, Shaikh’s objection to the adoption criterion, and the temporal challenge to the valuation frame, closing with the objection that the temporal defence changes the bookkeeping and with the answer that simultaneity is also a convention. Grade each track in a sentence of its own, and let your conclusion be a set of conditional verdicts rather than an allegiance. An examiner who set this question is testing whether you can hold a structured disagreement in your head without flattening it, and the three tracks table, reproduced from memory and adapted, is the single most efficient scaffold for showing that you can.

For teachers and seminar leaders

Stage the material across two meetings and resist the urge to begin with the theorem’s critics. In the first meeting, set Marx’s three chapters beside Sweezy’s discussion and let the group discover the indeterminacy problem from inside the tradition, before any formal machinery arrives; students who have felt the logical gap in their own reading receive Okishio, in the second meeting, as the answer to a question they already own. In that second meeting, work the numerical sketch on the board, then change one assumption at a time: let the wage bundle grow, let the innovation land in a luxury trade, let the input prices be those of the previous season, and have the class report which change alters the verdict and which merely alters the arithmetic. Close by distributing the three tracks table with its verdict column blanked, and require each student to fill the column with a grade and a one line defence. The exercise teaches the article’s whole method in an hour: premises first, verdicts after, and no migration of results between tracks.

For researchers, historians, and journalists

Researchers entering this field should know where its load bearing citations sit and where its open seams run. On the formal side, the live seam is premise defence: the valuation dispute awaits an argument about idealisation that neither school can supply from inside its own equations, and contributions that engage the philosophy of modelling directly will matter more than another round of constructions. On the statistical side, the open seam is convention independence: reconstructions that report how their findings move under rival treatments of depreciation, sector boundaries and the classification of labour will displace the older literature faster than new series built on familiar choices. Historians of economics hold the textual seam: the critical edition’s volumes continue to appear, and the standoff over Marx’s late confidence will be arbitrated, if it is arbitrated, by scholars who can weigh fragments against the whole archive rather than by polemic. Journalists face the simplest discipline and the hardest temptation. The temptation is the verdict headline, refuted or vindicated, attached to whichever track’s result arrived last in the journals. The discipline is to ask any expert quoted on this subject a single question, which of the three questions their result answers, and to print the answer alongside the claim. A report that does that has told its readers something true and useful about a debate that has survived a century and a half of premature funerals: that its questions are separable, its partial results are real, and its persistence is a sign of structure rather than of failure.

Frequently Asked Questions

Q: Does the rate of profit actually tend to fall?

The careful answer separates the claim from its evidence. As a proposition inside Marx’s theory, the tendency names a pressure generated by accumulation itself: as the mass of plant and materials per worker climbs, the ratio between surplus and total advanced capital is squeezed unless exploitation intensifies enough to compensate. Whether that pressure dominates across history is the contested part. Statistical reconstructions find long declines in American profitability into the early nineteen eighties and disagree about the decades after; formal work shows the pressure cannot operate through one channel under stipulated conditions; and the counteracting influences Marx listed are documented forces, not debating inventions. The tendency is best reported as a real mechanism whose long run victory is unproved.

Q: What is the Okishio theorem?

The theorem is a 1961 result by the Kobe economist Nobuo Okishio about what competition does to innovation. Its setting is an economy resting at a uniform rate of return, with workers consuming a fixed bundle of goods. In that setting he proved that any technique a firm would actually adopt, meaning one that cheapens production at the prices of the day, leaves the economy’s settled rate of return as high as before once prices readjust, and raises it when the innovation strikes a basic industry. Since Marx’s law appears to predict the opposite, the result stands as the law’s sharpest formal obstacle.

Q: How do Marxists answer Okishio?

The replies form a ladder of increasing depth. The shallowest points to the theorem’s own wage clause: since the proof freezes workers’ consumption, a rising real wage restores falling profitability inside the model, and defenders argue that history supplies the rise. A second rung contests the adoption rule, following Shaikh’s claim that firms fighting for survival choose techniques by unit cost and can thereby mechanise the general rate downward. The deepest rung rejects the accounting frame, arguing with the temporal school that inputs must be valued at the sums actually advanced, under which viable innovation and a falling rate coexist in worked constructions. A fourth response abandons the field gracefully, accepting the theorem and rebuilding crisis theory on other mechanisms, as the monopoly capital school did.

Q: Did Marx himself doubt the law?

The documentary record supports a narrower statement than either camp prefers. No retraction exists: Marx never withdrew the law, and the systematic statement stands in his own eighteen sixties manuscript rather than in his editor’s invention. Against simple confidence stand the notebooks of his final decade, dense with mathematical fragments and statistical enquiries, which Michael Heinrich reads as evidence of growing doubt about whether the tendency could be demonstrated at all, given that its two driving magnitudes cannot be ranked in advance. Andrew Kliman reads the same archive as continued work on a retained theory, noting restatements and the absence of any renunciation. The defensible verdict is that doubt is documented as a scholarly possibility, not as a biographical fact.

Q: Is falling profitability the cause of crises?

Only in the strong breakdown version of the theory, associated with Grossman, does falling profitability carry the whole explanatory load, and that version has few defenders in its pure form. Marx’s own crisis chapter works differently: the tendency generates a mass of capital that cannot be valorised at accustomed rates, and crisis erupts through the resulting devaluations, credit strains and idle capacity, as one moment in a wider contradiction rather than as a mechanical trigger. Several great crises followed prosperous and profitable years, which any monocausal reading struggles to digest, while the profitability downturns of the nineteen seventies preceded a genuine crisis decade, which monocausal readings cite in their favour. The pluralist position, that profitability crises are one species among several, fits the record’s variety best.

Q: What did Sweezy argue against the law?

Writing in 1942 as a sympathetic expositor, Sweezy charged that the derivation quietly freezes one variable while freeing another. Marx lets the organic composition climb on the back of rising productivity while holding the rate of surplus value still, yet the same productivity that mechanises production also cheapens wage goods, enlarging the surplus share that workers’ consumption leaves behind. Once both magnitudes are allowed to move together, their ratio can travel in either direction, and the law’s claim shrinks from necessity to possibility. Sweezy therefore relocated the theory of crisis toward the difficulties of realising surplus value, the path he later developed with Paul Baran in the theory of monopoly capital.

Q: Is the temporal single system interpretation convincing?

It convinces in the way a successful possibility proof convinces: completely, about a limited point. The temporal constructions demonstrate that the falling path Okishio forbids is arithmetically available once inputs are carried at historical sums, and that demonstration permanently blocks any claim that the theorem settled the matter for all accounting frames. Whether the interpretation convinces as an account of Marx is a further question, on which textual scholars divide, and whether it convinces as economics depends on accepting that historical cost, rather than replacement cost, is the measure against which a tendency should be judged. Critics detect a convention chosen for its yield; defenders reply that simultaneity is equally a convention and a less realistic one. Many specialists therefore grant the interpretation its formal point while withholding assent from its stronger claims, a pattern of partial conviction that the debate’s grading should record honestly.

Q: Can the law be tested at all?

Different parts of it submit to different tests. The formal theorem cannot be tested by observation at all, since it states what follows inside stipulated assumptions; it can only be checked as mathematics and challenged at its premises. The textual question is testable against the archive, and each newly published notebook either strengthens or weakens the rival readings of Marx’s confidence. The tendency itself, understood as a mechanism, is testable in the way tendency claims generally are: by searching for the mechanism’s signature in decompositions of measured profitability, asking whether composition effects appear with the predicted sign and strength where counterweights are accounted. Basu and Manolakos’s econometric work on American data is the model of such a test, and its positive finding shows the law is not beyond evidence, merely beyond simple evidence.

Q: Which Marxists reject the falling rate of profit?

The roll crosses every methodological boundary in the tradition. Sweezy set the law aside as indeterminate in 1942 and built elsewhere. Ian Steedman discarded it along with the value framework beneath it, on Sraffian grounds. John Roemer’s analytical reconstruction proceeds without it. Michael Heinrich doubts it can be demonstrated and suspects its author came to doubt it too. The Monthly Review school, in practice, organised its economics around monopoly and stagnation rather than around a falling tendency, and Joan Robinson had treated the law with reserve a generation earlier still. Against the suspicion that rejection equals apostasy, note the company: editors of Marx’s manuscripts, founders of schools and the tradition’s most cited formalists all appear on the list.

Q: What would settle this debate?

Nothing single, because there is nothing single to settle. On the mathematics track, settlement would mean a victorious argument about which idealisation, equilibrium comparison under simultaneous valuation or sequential monetary accounting, is the legitimate tribunal for a tendency claim, and such arguments are won by persuasion across a profession, not by a further theorem. On the record track, it would mean measurement conventions defended independently of their results and a century of series, across several economies, whose decompositions survive having rival conventions substituted piecemeal. On the textual strand, it would mean the archive yielding a decisive document, a retraction or a confident late restatement, rather than further fragments for rival readings to contest. Expect partial settlements, track by track, and distrust any announcement of a total one.

Q: How does the three tracks table separate logic, mathematics and record?

By assigning every argument a jurisdiction. The logic row asks what the law claims and whether the claim coheres, and it is settled by reading and reasoning, not by data or by proof. The mathematics row asks what follows inside stated models of technique choice and valuation, and it is settled by checking derivations and contesting premises. The record row asks what measured profitability has done, and it is settled, where it is settled, by series built under declared conventions. Confusion begins when a verdict earned in one row is spent in another: a theorem brandished against a statistic, a quotation deployed against a derivation. The table’s columns reinforce the separation, since each row carries its own strongest arguments, its own settlement conditions and its own verdict, and a reader can watch a row being graded without the grade leaking sideways.

Q: What does Okishio’s theorem actually assume about real wages?

It assumes that the bundle of goods a worker’s wage commands stays exactly constant while techniques improve around it. The assumption is about real wages, goods per period of work, not about money wages, which can rise with prices freely inside the model. Its function is isolation: by freezing labour’s consumption, the proof ensures that every productivity gain accrues to the surplus, so that any change in the settled rate reflects technique alone. Okishio recognised the knife edge himself, noting that a sufficiently large rise in the real wage reopens the falling outcome his theorem excludes. The assumption is therefore best read as a diagnostic device, locating the precise condition on which the law’s simplest version fails, rather than as a description of any economy that has existed.

Q: Why do temporal and simultaneous valuation give different verdicts?

Because the two systems keep the capital account in different currencies of time. Simultaneous valuation prices every magnitude in a production period at one consistent set of values, so the steel a firm consumed and the steel embodied in its product carry the same figure, and profitability is measured against what the inputs would cost to replace at the period’s settled prices. Temporal valuation keeps the receipts and outlays in the order they occurred: inputs stand in the books at the sums genuinely advanced for them, outputs are sold at whatever prices rule when they reach the market. In a world of sliding prices the two ledgers part company systematically. A technique sequence can show a healthy return on replacement cost while showing a meagre return on the money actually committed years earlier, and each side’s verdict follows its ledger with complete internal consistency.

Q: How should a student present the debate without taking a false side?

Present every position in the form its own defenders would recognise, and label the jurisdiction of each piece of evidence as you introduce it. Begin with the law as tendency, not forecast. State the theorem with its assumptions attached, and give its conclusion the conditional grammar the proof earns: under these stipulations, the settled rate cannot fall. Introduce each reply at the level it targets, wage clause, adoption rule or valuation frame, and state the strongest objection to the deepest reply before answering it. Conclude with grades per track rather than a coronation, and where your own judgement leans, disclose it as judgement after the evidence, not as a premise smuggled into the exposition. That procedure is not neutrality as emptiness; it is fairness as method.

Q: What evidence would move the empirical track independently of the logic track?

Evidence that speaks to the tendency’s operation without borrowing authority from the law’s textual prestige or its formal fortunes. Long series for several economies, constructed under conventions announced in advance and varied systematically to show which findings survive the variation, would establish the direction of measured profitability as firmly as the subject allows. Decompositions that isolate the composition effect, asking whether periods and industries with faster capital deepening show stronger downward pressure on returns once wage movements and capacity utilisation are held apart, would test the mechanism rather than its shadow. Natural experiments would help most of all: episodes where wage growth was institutionally frozen while mechanisation raced, or where composition jumped for reasons independent of profitability itself. Such evidence could move the record track even if every logic and mathematics dispute stood exactly where it stands.