Few questions in political economy are asked more often, or answered more carelessly, than whether the world economy has entered a crisis of capitalism. The phrase appears in headlines, in pamphlets, in seminar rooms and in the speeches of people who agree about nothing else. It is used to describe a bad quarter, a bank failure, a lost decade, a mood of public pessimism and the supposed final act of a system several centuries old, sometimes by the same writer in the same essay. A term stretched across so many different events stops discriminating between them, and a question built on such a term cannot be answered, because nobody has said what would count as an answer. The pages that follow refuse the usual shortcut. They offer no verdict on the present moment. They offer something more useful: a sorting instrument that separates three claims habitually fused into one, states what evidence bears on each, and records honestly how often the strongest of the three has been made before and how it fared.

The instrument is called the three-level diagnostic, and its discipline can be stated in a sentence. Before arguing about whether a crisis exists, specify the level at which the claim operates: a downturn inside a functioning order, the exhaustion of one institutional configuration of that order, or the failure of the order itself. Each level has its own evidence, its own historical precedents and its own test that could in principle settle it. Most public argument about capitalist crisis fails before it begins because the levels are traded against each other mid-sentence: evidence gathered at the first level is spent at the third, alarm appropriate to a regime change is attached to an ordinary recession, and calm appropriate to a recession is used to dismiss a genuine institutional exhaustion. Sorting the levels does not settle the argument. It makes the argument possible, which is the prior and harder task.

One claim organises everything ahead, and it deserves a name plain enough to carry. The claim is the level test: any assertion that capitalism is in crisis must specify which of the three levels it means, because the evidence appropriate to each differs in kind, and conflating them is what renders the long debate about capitalist breakdown unresolvable. A reader who absorbs nothing else can still use the test on the next headline. Ask which level the writer means. Ask what evidence is offered and whether it belongs to that level. Ask what observation would persuade the writer that the claim had failed. A writer who can answer all three questions is making an argument. A writer who cannot is performing a mood, and the distinction matters more than the conclusion reached, because moods cannot be corrected by evidence and arguments can.

Why the question resists a direct answer

The difficulty begins with grammar. The word crisis does double duty in ordinary language, naming both an event and a judgement about an event. A fever chart records a temperature; calling the temperature a crisis is a decision about what the temperature means. Economic language inherits this ambiguity in full. A fall in output is a measurement. A bankruptcy wave is a measurement. Whether those measurements amount to a crisis, and a crisis of what exactly, is an interpretive act performed on top of them, and the interpretation is where the disagreement lives. Two observers can accept every figure in a statistical release and still divide completely, because one reads a severe downturn inside a resilient order and the other reads the early symptoms of the order’s dissolution. The figures do not arbitrate between them. Only a prior agreement about levels can do that, which is why the diagnostic precedes the evidence throughout this article rather than following it.

Specification problems of this kind recur across the series of which this article forms a part. A question that looks empirical, answerable by gathering enough data, turns out on inspection to be a question about definitions wearing empirical dress. The data cannot be gathered sensibly until the definition is fixed, because the definition decides which data are relevant. Whether a downturn counts as a crisis in an order or a crisis of that order depends on what the order is taken to be: a set of markets, a set of institutions built around those markets, or a mode of production defined by wage labour and private accumulation. Each choice generates a different list of relevant facts. The Marxist tradition has always contained this insight in germ, since its founding texts distinguish between the crises that recur inside capitalist production and the hypothetical limit at which the relations of production themselves become fetters. What the tradition did not always do is hold the distinction steady under polemical pressure. The diagnostic is an attempt to hold it steady.

There is also an emotional economy to the question that honest treatment has to acknowledge. Declaring a terminal crisis has never been a neutral act. For critics of the existing order it promises that history is moving, that suffering has a direction and an end. For defenders it licenses a reassuring inversion: every survival becomes proof of permanence, every recovery a refutation of all criticism. Both camps therefore have reasons to prefer the loudest version of the claim, since the loud version is the one worth refuting and the one worth hoping for. The middle level, the exhaustion of a regime, is rhetorically unrewarding. It announces neither apocalypse nor vindication, only a long, difficult reorganisation whose outcome nobody controls. Yet the middle level is where most of the serious historical evidence sits, and a diagnostic that keeps returning attention there performs a service precisely because it frustrates both propagandas at once.

A diagnostic differs from a prediction in a way worth stating before any evidence is weighed. A prediction stakes itself on a future event and can be destroyed by the calendar. A diagnostic stakes itself on the present structure of an argument and can be destroyed only by showing that its distinctions fail to sort real cases. The physician who distinguishes a passing infection, a chronic condition requiring changed management, and organ failure is not evading the question of how ill the patient is. The distinction is the answer’s form, because treatment, prognosis and urgency all differ by category, and a single-word verdict would erase exactly the information a decision needs. Economic orders are not bodies, and the analogy must not be pressed into identity, but the logic of triage transfers: the level at which a problem sits determines what responses are possible, what timescales apply, and what would count as the problem deepening or lifting. Readers who finish this article still wanting a one-word answer to its title will at least know what that word would have to suppress.

The record of failed predictions, stated plainly

Credibility requires beginning where the tradition is weakest. Terminal crisis has been predicted, with confidence and with arguments that seemed strong at the time, in every generation since the early nineteenth century, and the system whose end was announced has outlived every announcement. This is not a peripheral embarrassment to be footnoted. It is the central fact any honest discussion of capitalist crisis must metabolise, because a theory of breakdown that has broken down empirically for two centuries carries a burden of proof that no amount of internal elegance can discharge. The diagnostic offered here is built partly as a response to that record: the level test exists because the undifferentiated claim kept failing, and kept failing in an instructive pattern, mistaking the death of configurations for the death of the order itself.

The pattern was visible early. The first generations of critics, writing amid the dislocations of early industrialisation, treated the new factory order as a brief convulsion that collapsing markets or its own misery would soon terminate. Sismondi, analysing the glut crises that followed the Napoleonic wars, saw in underconsumption a permanent structural flaw rather than a passing imbalance, and his reasoning was serious even where his horizon proved short. Marx himself was more careful than many of his followers. He expected crises to recur as a law of the system’s motion and spent the winter of a major commercial crisis, in the notebooks later published as the Grundrisse, working out the theory of a production order that generated its own interruptions. What he did not publish was a date, a timetable, or a mechanism that guaranteed termination. The famous confidence of the Communist Manifesto concerns the production of gravediggers as a social class, a political claim about agency, not a forecast that accumulation would cease of its own weight on schedule.

The generation of the Second International inherited the expectation and hardened it. When Henryk Grossmann published his breakdown law in 1929, arguing from Marx’s reproduction schemes that accumulation must mathematically exhaust the surplus available for continued expansion, the argument arrived months before a crash that seemed to confirm it. The nineteen thirties then staged the most plausible terminal scene the system has ever produced in its heartlands: bank failures in cascades, industrial output halved in the largest economy on standard historical estimates, unemployment reaching roughly a quarter of the labour force in the United States at the trough. Serious observers across the spectrum concluded that the order was finished. What followed instead was reconstruction: financial regulation, deposit insurance, welfare states, demand management, and a postwar expansion of a scale and duration nobody in that decade had forecast. The episode remains the single most important data point in this entire subject, and the article returns to it as a worked case below, because it shows both how close a level three scene can appear and how a level two reorganisation can answer it.

Later waves repeated the rhythm. The long postwar boom generated its own obituaries in advance: theorists of monopoly capital described an order condemned to stagnation, absorbing its surplus only through waste and military spending. The stagflation decade that closed the postwar settlement was widely read, from several directions, as the system’s general crisis, and a literature of final crisis flourished. The order responded with another reorganisation, harsher and more financial in character, and accumulation resumed on new terms. Even the great financial collapse treated elsewhere in this series, the most severe shock since the nineteen thirties, ended in reorganisation rather than termination, though the reorganisation’s costs and legitimacy problems remain debated. None of this proves that termination is impossible. It proves that termination claims require a standard of evidence far above the presence of severe distress, because severe distress has been the companion of every reorganisation the order has ever undergone. An instrument for judging crisis claims must encode that lesson structurally, not as a mood of caution but as a rule: distress is level one evidence, exhaustion of a configuration is level two evidence, and level three requires showing that reorganisation itself has lost its capacity, a showing nobody has yet completed.

The three levels of a crisis of capitalism

The diagnostic’s three levels can now be stated in their full form. A crisis in capitalism is a downturn that the order generates from its own normal operation and subsequently absorbs: output falls, firms fail, workers are discharged, credit contracts, and then, through mechanisms the cluster’s companion articles examine in detail, and the episode closes with the configuration that entered it still in place. A crisis of a regime of accumulation is a deeper event: one particular configuration of production, distribution, money and statecraft, the arrangement that for decades made growth self-reconciling, loses the fit among its parts, so that the ordinary repair mechanisms stop restoring what they used to restore and a reorganisation of the arrangement itself becomes unavoidable. A crisis of capitalism as such would be deeper still: a condition in which reorganisation itself fails, in which no rearrangement of those elements within the system’s defining relations can set accumulation going again. The three statements sound similar in a headline. In every other respect they are different claims, with different evidence and different histories, and the diagnostic exists to keep them apart.

What is the difference between a crisis in capitalism and a crisis of capitalism?

A crisis in capitalism is a downturn the system recovers from. A crisis of a regime of accumulation is the exhaustion of one institutional configuration, implying reorganisation. A crisis of capitalism as such would mean no reorganisation could restore accumulation, which is a far stronger claim, demanding evidence of an entirely different order before it can be believed.

Level one rewards a closer look before the levels are taken separately, because its mechanisms are the best documented part of the tradition’s crisis theory and because most real episodes begin in its register. Marx analysed the register from the simplest forms upward. Commodity circulation separates sale from purchase in time and place, which makes interruption possible in principle; the chains of payment and credit that develop on that basis make interruption actual, binding each producer to obligations that fall due whether or not the commodities behind them have found buyers. The mature analysis then follows the credit system to the point where a break anywhere propagates everywhere, which is why the tradition’s starting point remains the humble forms of production in the first volume. One celebrated passage of the third volume treats crisis as the violent reconciliation of contradictions that the system can manage in no gentler way: capitals that cannot valorise themselves are destroyed on paper and in fact, commodity stocks are written down, plant stands idle until its book value matches what the market will remunerate, and labour is thrown back onto the street until its price again permits profitable employment. Read without the surrounding volumes, the passage sounds apocalyptic. Read in context, it describes maintenance. The destruction is the mechanism by which proportionality is restored, and the tradition’s level one theory is therefore a theory of self-repair through damage.

Recovery has institutional as well as automatic components, and the mix differs across periods. In the nineteenth century, repair ran mostly through bankruptcy law, wage flexibility of a brutal kind, emigration, and the opening of new territories and industries to investment; the state intervened reluctantly and lenders of last resort were improvised late and partially. Across the twentieth century, the repair kit acquired public components: central banks willing to supply liquidity against panic, deposit guarantees that broke the logic of the bank run, unemployment insurance that sustained demand through the trough, and treasuries prepared to spend against the cycle. Each addition changed what a level one event looks like without changing its level. A downturn cushioned by automatic stabilisers is shallower than its nineteenth-century counterpart, and shallowness can disguise depth of cause, which is why the diagnostic warns against reading severity from headline pain alone. The level of an event is a property of what it does to the order’s arrangements, not of how frightening its worst month felt.

How regularly did commercial crises arrive in the nineteenth century?

Roughly once a decade in the leading economies, with a rhythm contemporaries recognised and named. The British sequence ran through the mid eighteen twenties, the late eighteen thirties, the late eighteen forties, the eighteen fifties, the mid eighteen sixties and the early eighteen seventies, each episode joining speculation, credit expansion and a sudden refusal of credit.

That regularity matters for calibration. An order that produces a serious interruption every ten years or so, across a century of uninterrupted long-run expansion, has made interruption part of its normal physiology. Observers inside that century who treated each new crisis as the last were not stupid; several of those episodes were severe, and severity invites terminal language. But the base rate was available to them as it is to us: the previous crisis, equally severe in feeling, had been followed by reorganisation at the level of firms and sectors, not by the end of the order. Base rates are the diagnostic’s memory. Any assessment that ignores them is not an assessment but a reaction, and reactions are precisely what the level test exists to slow down.

Level one also has a geography and a sociology that keep it from being an abstraction. Downturns strike regions and trades unevenly: a credit contraction that a diversified financial centre absorbs as a bad year can strip a single-industry town of its employer, its tax base and its young people within a decade. The diagnostic’s insistence that such an event remains level one for the system is not a denial of its weight for the people inside it. Levels classify the event’s relation to the order, not its moral size. Confusing those two judgements is another source of the debate’s heat, since a writer describing devastation truthfully at the local scale can sound, to a reader tracking the system’s aggregate, as if claiming a systemic collapse that the aggregates do not show. Both can be right at once, about different objects, and the level vocabulary is what allows that sentence to be said without contradiction.

Level two: when a whole configuration wears out

The second level is the diagnostic’s centre of gravity, because it is where the richest historical evidence and the most useful theory meet. The concept of a regime of accumulation comes from the French regulation school, a current formed in the nineteen seventies by economists such as Michel Aglietta, Robert Boyer and Alain Lipietz, who asked why the postwar decades had been so unusually stable and why that stability was visibly failing. Their answer was that stability is never a property of markets in the abstract. It is a property of a configuration: a way of organising production, a pattern linking wages to productivity, a monetary and financial arrangement, a set of state capacities, and a mode of insertion into the world economy, all fitting together well enough that the tensions of accumulation are absorbed rather than amplified. When the fit holds, the configuration earns the name of a regime. When the fit decays, the economy enters a structural crisis that no ordinary recovery ends, because what has failed is not demand or confidence but the machinery that used to reconcile the system’s moving parts.

The postwar configuration in the advanced economies, conventionally labelled Fordist after the production model at its core, shows the anatomy clearly. Mass production of standardised goods on moving assembly lines generated steady productivity growth. Collective bargaining and welfare institutions linked mass wages to that growth, so the workers who built the cars and appliances could buy them. Governments committed to maintaining demand and employment, and did so with tools ranging from public works to military budgets. An international monetary order anchored currencies and financed reconstruction, while cheap energy and expanding trade widened the field. None of these elements was designed as a system by a single hand; the regime was an emergent settlement among firms, unions, states and electorates. Its stability was nonetheless real, and its exhaustion was equally real: productivity growth faltered from the late nineteen sixties, the wage-productivity link became a distributional battleground, inflation and unemployment rose together in a combination the regime’s policy tools had not been built to treat, and the monetary anchor itself was abandoned. The crisis of that period was therefore not a downturn that happened to be long. It was the visible failure of a specific reconciliation machinery, and it ended only when a different configuration, more financialised, more global in its sourcing and less committed to full employment, had been assembled in its place.

What holds a regime of accumulation together?

Complementarity holds it together: each institution makes the others workable. Assembly line production pays for itself only when wages rise enough to absorb standardised output. Rising wages remain profitable only while productivity growth outruns them. Demand management succeeds only while a stable monetary framework keeps expectations anchored, and removing one strut shifts its load onto all the rest.

The general lesson the regulation school contributes to the diagnostic is that regimes are mortal but the order is not thereby shown to be. A regime is one solution to the reconciliation problem, evolved under particular technological and political conditions, and solutions wear out as conditions shift: the productivity frontier of a given production paradigm flattens, the compromises that distributed its gains harden into rigidities, the international setting that flattered it changes. Recognising a level two crisis is thus a claim about a configuration’s remaining life, not about the order’s. It predicts reorganisation: painful, contested, possibly lasting decades, and certain to be denounced in its early stages as the end of everything, because from inside a dying regime the death of its familiar world is genuinely what the experience feels like. The diagnostic treats that feeling as evidence about the regime, not about the system.

Level two claims are also, importantly, checkable in a way level three claims are not. A regime has specifiable components, and its exhaustion shows in specifiable places: the productivity performance of its core production model, the stability of its distributional settlement, the coherence of its policy instruments with the problems they face, the capacity of its financial arrangements to fund investment without recurrent panic. An analyst can therefore argue about a level two claim with evidence that both sides accept as relevant, and can be wrong in ways that later evidence reveals. Several historians of the postwar period who denied any structural break in the nineteen seventies, reading the stagflation as an ordinary downturn plus policy error, were answered by the subsequent decades, in which the old configuration was nowhere restored and its replacements differed from it in exactly the dimensions the regulation theorists had marked. The argument was real, and it could be lost. That possibility of losing is what makes level two the diagnostic’s model of an honest crisis claim.

Level three: the claim that no rearrangement can work

The third level asserts something categorically stronger: not that this downturn is deep, and not that this configuration is spent, but that the order’s repertoire of configurations is exhausted, that no rearrangement of production, distribution and regulation within the system’s defining relations can restore sustained accumulation. This is the breakdown claim in its strict form, and it is worth noticing how rarely it is argued in strict form even by those who use its language. Grossmann’s version rested on arithmetic drawn from Marx’s reproduction schemes: as the ratio of accumulated capital to living labour rises, the mass of surplus available relative to the accumulation requirement eventually fails, and the system chokes on its own growth. The argument was the most rigorous breakdown theory the tradition produced, and its reception is instructive. Critics showed that the result depended on holding constant exactly the variables that capitalist development varies: the rate of surplus value, the cheapening of the elements of constant capital, the opening of new spheres of investment. Marx’s own discussion of the falling tendency of the profit rate lists such counteracting influences in the same chapter, which is why several major Marxist economists, including figures of undoubted orthodoxy, have explicitly denied that any automatic breakdown theorem can be extracted from the theory. The dispute belongs to the cluster’s treatment of crisis mechanisms; here it matters as a demonstration that level three claims must survive the counteracting influences, not merely cite the tendency.

Rosa Luxemburg’s version approached the limit from the outside rather than from arithmetic. Accumulation, she argued, requires demand beyond what the system’s own workers and capitalists can supply, and therefore feeds on non-capitalist territories and strata, absorbing them until the planet offers no outside left. The argument had the great merit of tying crisis theory to imperialism and to real historical expansion. It also drew heavyweight criticism from within the tradition, most systematically from Bukharin, who denied that the reproduction schemes required an external consumer and treated the outside as an accelerant rather than a condition. Empirically, the closing of the geographical frontier did not coincide with the cessation of accumulation, since new outsides were repeatedly manufactured inside the order itself: subsistence sectors were commercialised, household production was marketised, public assets were privatised, and whole dimensions of life that stood outside the cash nexus in one generation stood inside it in the next. A level three argument built on a fixed inventory of outsides must reckon with the demonstrated capacity to produce new ones.

Why is the strongest claim the hardest to test?

Because it quantifies over futures. A level one claim is tested by waiting for recovery; a level two claim by watching whether a configuration’s components keep failing together. A level three claim asserts that every possible reorganisation will fail, which no finite run of failures can establish and no single success is permitted to refute.

The honest residue of level three is therefore narrower than its rhetoric. What can be shown, at any given moment, is a list: the reorganisations already attempted and their results, the constraints that defeated them, and the shrinking or otherwise of the plausible repertoire. What cannot be shown is the repertoire’s emptiness, since the repertoire is partly invented under pressure, and invention is the variable that has embarrassed every previous inventory. The diagnostic accordingly treats level three not as a forbidden thought but as a claim with an unusually heavy and partly unpayable burden, to be advanced, if at all, as a statement about the observed exhaustion of known strategies rather than as knowledge about all possible ones. Writers who perform that translation, from metaphysical certainty to an auditable list of failures, are doing level three work honestly. Writers who skip it are asking the reader to mistake the length of a crisis for the end of a system.

The level test in use

An instrument earns its keep in application, and the level test applies in four moves, each of which can be performed on any crisis claim in a few minutes. The first move names the level: in, of a regime, or of the system. The second move inventories the evidence offered and checks its passport, asking whether each item was issued at the level claimed or borrowed from a lower one. The third move asks for the falsifier, the observation the claimant would accept as defeat. The fourth move sets a timescale, since levels differ in their natural clocks: level one events resolve in quarters and years, level two transitions in decades, and level three claims, if they mean anything, in generations. A claim that will not state its clock is usually hoping to borrow urgency from the short cycles and significance from the long one simultaneously. The four moves do not require agreement about theory. They require only that a writer accept being read literally, which is why they meet resistance out of proportion to their modesty.

Worked briefly, the moves sort claims that public argument leaves in a heap. A newspaper argues that a stock market fall proves the system is finished: level three conclusion, level one evidence, no falsifier, no timescale, and the level test returns it for resubmission. An economist argues that the wage and productivity link in a particular economy has been broken for a generation and that no policy within the existing settlement has repaired it: a level two claim with level two evidence, falsifiable in principle by a durable restoration of the link, and the test passes it to substantive argument regardless of whether the argument finally convinces. A historian argues that a past decade’s turbulence marked the end of a monetary regime: again level two, now with the advantage that the subsequent reorganisation is itself evidence. The test is deliberately unimpressed by the seniority, passion or politics of the claimant. It grades the fit between claim and evidence, and it returns the same verdict forms to the order’s defenders: an editorial celebrating a strong quarter as proof of permanent systemic health fails exactly as its mirror image fails, level one evidence spent on a level three conclusion with the sign reversed.

Can one industry’s collapse remain a level one event?

Yes, and most industry collapses do. A sector can shrink to a fraction of its former size through technical change, import competition or exhausted demand while the wider order reallocates its labour and capital without structural strain. The event becomes more than level one only when the fall disables machinery the whole configuration depends on.

Why does the diagnostic ask about duration before depth?

Because depth misleads and duration discriminates. A very deep contraction that closes within a few years behaves, for the order, as a level one event with high amplitude, while a shallow stagnation persisting across a generation signals that the configuration’s repair mechanisms themselves are failing. Duration separates amplitude from structure before dramatic numbers argue.

Working notes kept alongside the texts repay the effort the test demands, and many readers of this series keep the three levels, the four moves and the worked examples together in the Marxism study notes maintained in VaultBook, where each new crisis claim encountered in the press can be filed against the level it asserts and the evidence it actually supplies. The habit sounds clerical and is in fact transformative, because a file of sorted claims, reviewed after some years, teaches the calibration lesson no single argument can: how often the level three file had to be emptied into the level two file by events, and how rarely the traffic ran the other way.

The three-level diagnostic

The table below is the article’s findable artifact, designed to be consulted before it is read around. Each row states a level; the columns give its definition, the evidence that would indicate it, the evidence that counts against it, its clearest historical precedents, and what would follow if the diagnosis were correct. Compression is the price of any table, and the surrounding sections exist so that the price is not paid in understanding: every cell is argued somewhere in the prose, and a reader who disputes a cell should be able to find the reasoning, and the evidence behind the reasoning, in the section devoted to that level. Used properly, the table is a checklist for interrogating claims rather than a vending machine for conclusions. It will not tell a reader which level describes any given moment. It will tell the reader exactly what would have to be true, and what would have to be shown, for each level’s description to fit.

Level	Definition	Evidence that would indicate it	Evidence against	Historical precedents	What would follow
Level one: a crisis in capitalism	A downturn generated by the order’s normal operation and absorbed by its ordinary repair mechanisms	Falling output, rising failures and unemployment, contracting credit, followed by liquidation, devaluation and recovery within the existing configuration	Recovery arriving on the historical schedule; the configuration’s institutions emerging strained but intact and unreplaced	The recurring commercial crises of the nineteenth century; the ordinary recessions of the postwar decades	Adjustment within the regime: losses allocated, weaker capitals eliminated, no redesign of the settlement required
Level two: a crisis of a regime of accumulation	The exhaustion of a particular configuration linking production, distribution, money and the state, so that its repair mechanisms stop restoring growth	Productivity failure in the core production model, a broken distributional settlement, policy instruments contradicting the problems they face, repeated level one events that stop resolving	A durable restoration of the configuration’s core linkages without redesign; productivity and distribution recovering together under the old rules	The ending of the postwar Fordist settlement in the nineteen seventies; the long price decline and reorganisation of the late nineteenth century	Reorganisation rather than replacement: a new configuration assembled over years or decades, with the order’s defining relations intact
Level three: a crisis of capitalism as such	A condition in which no reorganisation within the system’s defining relations can restore sustained accumulation	The demonstrated failure of successive, genuinely different reorganisations; repair mechanisms failing across institutional variants; the plausible repertoire visibly emptying	Any successful reorganisation, which refutes the claim for that period; the invention of new configurations, new sectors or new outsides that restart accumulation	None completed: the nineteen thirties are the nearest approach in the advanced economies and ended in reorganisation	Transformation of the order itself, by replacement, degeneration or reconstruction on different relations, none of which the diagnosis can schedule

Two cautions govern the table’s use. The first is that levels are assigned to episodes, not to eras in the lump: a single decade can contain a level one event inside a level two transition, and the interesting analytical work lies in saying which features belong to which stratum. The second is that assignment is retrospective by nature. During an episode, the evidence underdetermines the level, since a level two exhaustion announces itself initially as a level one downturn that refuses to end, and honesty requires carrying provisional assignments as provisional. Analysts who present a level assignment as a perception rather than an inference are claiming an authority the material cannot confer. The table disciplines the inference; it does not abolish its uncertainty, and the sections on the hardest historical cases below show the uncertainty being managed rather than concealed.

Sorting the evidence at the lower levels

Most evidence offered in crisis debates is genuine; the sorting problem is that it is filed at the wrong level. Financial episodes supply the clearest illustration. A banking panic is dramatic, photogenic and historically freighted, and it therefore attracts level three language almost automatically. Yet panics belong to the order’s normal pathology from its early centuries, and the interesting question a panic poses is not whether the system can have one but what the panic does to the configuration in which it occurs. Where the financial machinery is peripheral to the production settlement, a panic burns through speculation and subsides as a level one event. Where credit has become the connective tissue of the whole configuration, funding its investment, its consumption and its state, a panic can tear the settlement itself and open a level two transition. The most severe modern episode of this kind receives its own treatment in the series’ account of Marxism and the financial crash, which examines the mechanisms of that collapse in the detail they require; the diagnostic’s contribution is the prior question of what such an episode, however severe, is evidence about.

Contagion is the bridge concept between the levels, and it deserves precise handling. A shock travels from its sector of origin to the wider order through identifiable channels: debts owed by the failing sector to the financial system, supply chains that cannot substitute its output, employment concentrated where alternatives are scarce, and expectations that generalise from one balance sheet to all. Each channel is measurable in principle, and the strength of the channels, rather than the size of the initial failure, decides the event’s eventual level. This is why the diagnostic refuses to grade crises by their opening headlines. A modest failure sitting on dense channels can reorganise an economy; a spectacular failure on sparse channels can be absorbed as spectacle. Analysts who trace channels are doing level work. Analysts who quote the size of the first domino are doing theatre.

The historical panics of the nineteenth century repay a level reading because they include both kinds. The crisis of the mid eighteen forties in Britain, bound to railway speculation, ruined investors and contractors in large numbers and altered the financial landscape, yet the industrial order absorbed it and the railway network whose shares had collapsed went on to be built, becoming infrastructure rather than wreckage: a level one event of high amplitude inside an ascendant configuration. The long price decline that followed the panic of the early eighteen seventies behaved differently, as the worked case below argues, because it coincided with the exhaustion of the first industrial configuration’s productivity frontier and the contested birth of a corporate, steel and chemicals order. Same species of opening event, different level, and the difference lies in the configuration’s remaining vitality, which is precisely the variable the diagnostic exists to inspect.

Productivity, dynamism and debt: the slow evidence

Dramatic evidence arrives in crashes; slow evidence accumulates in trends, and the trends most often recruited for level three conclusions are three. Each deserves weighing at its proper strength. Productivity growth in the advanced economies has run slower over the decades since the postwar expansion than during that expansion itself, a pattern visible across most large economies though uneven among them and punctuated by a partial revival associated with computing in the years around the turn of the century. Slower productivity growth is a serious fact with serious consequences: it narrows the surplus from which wages, profits and public services are simultaneously paid, and it converts distributional conflict from a negotiation over increments into a contest over shares. At the level of the diagnostic, however, it is evidence about the prevailing configuration’s production model and the maturity of its technologies. It becomes level three evidence only on the further showing that no successor production model can restore the growth rate, a showing that the history of successive technological waves, each declared impossible from inside the exhaustion of the last, entitles the reader to demand in full.

Business dynamism tells a similar two-level story. Measures of enterprise formation, of the reallocation of jobs between firms, and of the share of activity carried by young companies have declined in several large economies across recent decades, again in the durable framing this subject requires: the comparison that matters is with the postwar pattern, not with any single year. A less dynamic business sector suggests sclerosis in the existing configuration: incumbents protected by regulation, finance or market position, entry discouraged, and the churn through which the order historically renewed its capital stock running slower. Sclerosis is a regime pathology with a long pedigree in the literature of institutional ageing. It is not, without much more, a property of the mode of production, since earlier configurations also sclerosed and were succeeded by more turbulent arrangements under the same defining relations.

Debt is the third slow trend and the most double-edged. Obligations relative to output have risen across households, firms and states over the long period since the postwar settlement, and each crisis of the modern configuration has been answered partly by moving losses onto public balance sheets, raising the stock that the next episode inherits. A level two reading is straightforward: the prevailing configuration increasingly finances its stability by borrowing against future income, which is a property of its monetary and fiscal settlement. The level three recruitment of the same fact argues that the borrowing is the system’s life support and that its continuation is arithmetically bounded. The diagnostic asks the level three claimant for the missing step: a demonstration that the bounding constraint binds under every monetary and institutional arrangement the order can adopt, including ones, such as the financial repression and rapid growth through which the postwar states reduced far heavier public debt burdens, that history has already exhibited. The fact of rising debt is not in dispute. Its level is, and the dispute is settled by arguments about arrangements, not by the size of the total.

Does slower productivity growth point beyond a regime problem?

Not by itself. Slowing productivity growth is the signature of a production model maturing, and it has appeared at the exhaustion of every successive technological configuration the order has inhabited. It supports a level two diagnosis beside a failing distributional settlement; a level three diagnosis only if no successor model can exist.

Secular stagnation and the stagnation tradition

Stagnation is the slow cousin of crisis, and the failure to distinguish them corrupts both vocabularies. A crisis is an event, a rupture in which the order’s contradictions are forcibly reconciled. Stagnation is a condition, a chronic shortfall of dynamism in which little ruptures because little is ventured. The mainstream of economics possesses its own stagnation theory, older than most of its critics realise. Alvin Hansen, addressing American economists in the late nineteen thirties, argued that the closing of the frontier, the slowing of population growth and the capital-saving character of new technology had permanently lowered the investment opportunities available to a mature economy, condemning it to chronic underemployment unless public spending filled the gap. The postwar boom refuted the forecast in its strong form, and the argument slept for decades before being revived in the twenty-first century by Lawrence Summers and others to explain persistently low interest rates, weak investment despite cheap money, and recoveries that left output below its earlier path. The revival matters here because it shows the stagnation question arising inside the citadel, from economists with no Marxist commitment, using the order’s own categories.

The Marxist tradition developed its parallel account through the theory of monopoly capital, associated above all with Paul Baran and Paul Sweezy, whose joint work argued that an economy dominated by giant firms generates a rising surplus it cannot absorb through consumption or productive investment, and therefore leans on waste, advertising, financial expansion and military expenditure to keep demand alive. The full statement of that tradition, its evidence and its critics, is given in the series’ treatment of monopoly capital stagnation theory, and the present article borrows only its diagnostic placement. Both stagnation theories, the mainstream and the Marxist, describe a condition of the prevailing configuration: its investment machinery, its demand structure, its demography. Both are therefore level two instruments whatever their authors intended, since both identify mechanisms that a reorganisation could in principle alter, through different technologies, different distributions of income, or different public investment regimes. Neither demonstrates that accumulation as such has lost its future, and their strongest versions do not attempt the demonstration.

What distinguishes stagnation from crisis?

Tempo and mechanism distinguish them. A crisis is a forcible reconciliation compressed into an event: balance sheets collapse, prices crash, and contradictions are settled violently in months. Stagnation is the absence of event, a long equilibrium of weak investment and slow productivity growth in which contradictions are not settled but indefinitely postponed.

The distinction carries a practical consequence for the level test. Stagnation evidence, however long the stagnation persists, cannot be converted into breakdown evidence by accumulation of years alone, because the quantity that matters for level three is not how long the order has underperformed but whether reorganisation remains possible. A stagnant configuration that a successor arrangement then replaces has refuted, by the replacement, the terminal reading of its own stagnation, and the postwar sequence supplies exactly this pattern at the level of the earlier Hansen forecast. Conversely, a crisis that appears to end a stagnation may merely inaugurate a new configuration whose dynamism later fades in its turn. Levels classify structures, and structures are identified by their mechanisms rather than by the calendar’s accumulation of gloom.

The profit-rate dispute, deliberately parked

No honest survey of crisis argument can omit the tendency of the rate of profit to fall, and no honest diagnostic can pretend the dispute around it is settled. Marx’s argument, developed in the third volume of Capital, holds that accumulation raises the ratio of dead to living labour in production, and that since living labour alone produces surplus value, the rate of profit on total capital tends downward unless counteracting influences intervene: cheapening of machinery and materials, intensified exploitation, foreign trade on favourable terms, and periodic devaluation of the existing capital stock. Whether measured profit rates across the advanced economies have in fact fallen over the long run, recovered, or oscillated within a band is a question fought over by specialists with rival datasets, rival deflators and rival treatments of finance and housing, and the fight is genuine: the same century can be made to display a falling, a flat or a rising series by defensible measurement choices. The full controversy, with the mechanisms of the rival crisis theories laid out for comparison, belongs to the series’ marxist crisis theory guide, and this article deliberately parks it there.

What the diagnostic contributes to the dispute is a placement rule. A falling profit rate, even if established beyond measurement quarrel, is a tendency operating inside accumulation, and Marx’s own presentation surrounds it with the mechanisms that counteract it. It therefore bears most directly on the frequency and depth of level one events and on the profitability climate in which regimes flourish or fail. To convert it into level three evidence, a claimant must argue that the counteracting influences are themselves exhaustible in principle, and that argument, as the Grossmann episode showed, is where breakdown theories go to be tested and where they have historically thinned. Parking the dispute is not dismissing it. It is refusing to let the most contested measurement in the field carry the largest conclusion in the field before its own house is in order.

The mechanisms beneath the diagnostic

Beneath the levels sits a family of mechanism theories, each explaining how capitalist production generates its characteristic interruptions, and the diagnostic is designed to be neutral among them while grading the conclusions built on them. Underconsumption theories locate the fault in distribution: wages, the system’s mass purchasing power, grow more slowly than the output workers produce, so realisation fails unless credit, exports or state demand bridge the gap. Disproportionality theories locate it in anarchy of production: investment in each branch proceeds without knowledge of the others, and the proportions among departments drift until some outputs cannot be sold and some inputs cannot be had. Profit squeeze theories locate it in success: long expansions tighten labour markets, wages rise faster than productivity, profits are compressed, and investment halts from the revenue side. The falling profitability argument locates it in the composition of capital itself. Each mechanism, where it operates, produces level one events with a distinctive signature, and each has implications for regimes, since a configuration can be understood partly as an institutional device for holding one or another mechanism in check: the Fordist settlement, on this reading, was an underconsumption remedy built into the wage relation, while the financialised successor managed demand through asset inflation and household credit instead.

The plurality of mechanisms is itself an argument for the diagnostic. Public crisis talk usually presupposes a single story, the story favoured by the speaker’s school, and treats the presence of distress as confirmation. But if several distinct mechanisms can each generate downturns, then distress alone cannot tell the observer which mechanism fired, let alone which level the episode occupies. Serious analysis works in the other direction: it identifies the mechanism from the episode’s internal pattern, asking whether the break came in realisation, in proportions, in profitability or in credit, and only then assigns a level based on the damage done to the configuration’s arrangements. The level test and the mechanism theories are thus complements rather than rivals. Mechanisms explain how trouble starts; levels grade how far it reaches; and the fashionable habit of skipping both steps, moving directly from a bad headline to a civilisational verdict, is the precise failure the three-level diagnostic was built to interrupt.

The counter-case at full strength

Part Six discipline in this series requires that the opposing case be built as strongly as its own advocates would build it, and the case against terminal crisis, and against careless crisis talk generally, is formidable. Its first pillar is the survival record already surveyed: an order that has absorbed industrialisation, world wars, the collapse of the international monetary system twice, the end of formal empires, the computerisation of production and the integration of billions of new workers into the world market, all while repeatedly being read its last rites, has earned a presumption that its adaptive repertoire is deeper than any observer’s inventory of it. The presumption is rebuttable, but the burden sits where the record puts it, on whoever claims that the repertoire is at last empty.

Its second pillar is the improvement in global living standards across the decades since the postwar expansion, which any honest treatment must present as a fact requiring accommodation rather than a slogan requiring rebuttal. Extreme poverty, measured on the international institutions’ standard lines, has fallen across that period from the condition of a large share of humanity to a much smaller share, with the largest absolute gains in the most populous economies of Asia. Life expectancy has risen on every continent. Literacy, electrification, access to sanitation and the survival of children have all moved in the same direction over the long run, at rates without precedent in earlier centuries. Regional reversals, stagnating pockets and the unevenness of the gains are real, and distributional questions about who captured the growth are legitimate and contested. But a picture of the order as a machine for generalised immiseration, whatever purchase it had on the early industrial decades, cannot be squared with the long global record, and theories of terminal crisis that quietly assume such a picture inherit its falsity. The tradition’s strongest internal critics have always known this: the claim was never that capitalism could not develop the productive forces, but that it developed them in forms that generated specific, analysable pathologies. Losing that precision to gain a slogan is a trade the evidence refuses to honour.

Its third pillar is the repeated opening of technological frontiers at the moments when exhaustion was being declared. The railway age was pronounced complete before electricity reorganised production and urban life. The electrical and chemical order was pronounced mature before computing remade information work, and computing itself was pronounced a productivity disappointment, in a famous economist’s quip about seeing computers everywhere except in the statistics, shortly before the statistics moved. The pattern does not prove that a frontier always waits; induction from past rescues is not a law of nature. It does establish that the frontier inventory has been systematically underestimated by contemporaries reasoning from the visible paradigm, which is exactly the reasoning a level three claim must perform. A claimant who cannot explain why the underestimation pattern should stop applying to the present inventory has not finished the argument.

What is the strongest evidence against a terminal reading?

The conjunction of the survival record with the living standards record. Every earlier configuration pronounced terminal was succeeded by reorganisation, and across the same span the material conditions of most of humanity measurably improved. Together they fix the burden of proof on any terminal claim: explain why reorganisation and improvement should both stop.

The objection: a diagnostic that refuses verdicts is an evasion

The strongest objection to everything proposed here must now be given its full force, because the diagnostic deserves to survive the best attack available rather than a convenient one. The attack runs as follows. Crisis language in the Marxist tradition, and in its popular descendants, is structured to be unfalsifiable. Growth is accumulation and therefore exploitation intensifying; stagnation is crisis by another name; expansion is unsustainable and merely defers the reckoning; contraction is the reckoning arriving. Every possible state of the system confirms the theory, which means no state of the system is evidence for it, and the word crisis has been emptied of content while retaining its thunder. A diagnostic that responds to this charge by sorting claims into levels, the objection continues, does not cure the evasion but professionalises it. Forever distinguishing, forever deferring, the analyst never has to say the thing the title asks. The question was whether the patient is dying, and the answer is a taxonomy. Taxonomies do not frighten anyone, mobilise anyone or advise anyone; they allow their author to sound rigorous while the historical moment, whatever it demands, passes unjudged. Worse, the levels are elastic enough to absorb any outcome retroactively: if recovery comes, the event was level one; if reorganisation comes, it was level two; if neither comes, patience, the timescale is generations. A framework that can narrate every sequel as confirmation has simply moved the unfalsifiability from the claim to the classification.

The objection gains strength from the genre’s actual record, and honesty requires conceding the ground it stands on. A large share of writing about capitalist crisis, including writing by authors of reputation, does proceed exactly as charged: the conclusion is fixed, the evidence is decorative, and every datum is pre-assigned to the confirmation column. Funeral orations for the system have been delivered on schedule through recoveries that their authors then described as the crisis deepening. Defenders of the order run the same racket in reverse, treating each survival as a refutation of all criticism and each boom as a certificate of eternal health. If the three-level diagnostic functioned merely as a more sophisticated vocabulary for the same performance, the objection would be fatal and the article a contribution to the problem.

The reply: every level carries its own falsifier

The reply begins by accepting the objection’s standard and then submitting to it. A framework is falsifiable level by level if, for each level, an observation can be named in advance that would defeat an assignment to that level, and the diagnostic names them. A level one assignment is defeated by non-recovery: if the downturn does not close on the historical schedule and its damage spreads to the configuration’s own institutions, the assignment fails and must be revised upward. A level two assignment is defeated by restoration: if the configuration’s broken linkages, its productivity engine, its distributional settlement, its policy coherence, are durably repaired without redesign, then the exhaustion claim was wrong and the episode is reclassified downward. A level three assignment is defeated by any successful reorganisation, which is the most fragile position of the three and deliberately so: the diagnostic loads the strongest claim with the weakest defences, so that holding it requires the continuous non-arrival of the very thing history keeps delivering. Elasticity is thereby removed where it matters. The classification can indeed narrate any sequel, but only by reclassifying, and reclassification is a public defeat for whoever assigned the original level, recorded in the analyst’s own file. What the objection describes as evasion is, operated honestly, a machine for recording defeats.

The charge of verdict refusal also dissolves on inspection of what the diagnostic actually issues. It issues verdicts constantly, at the levels where verdicts can be earned: that a given episode was a downturn within a functioning configuration, that a given settlement is exhausted and will not be restored, that a given terminal claim has not met its burden. What it refuses is one specific verdict form, the unconditional pronouncement on the order’s survival, because that verdict has a two-century record of being issued without the evidence its content requires. Refusing a verdict form that experience has discredited is not the absence of judgement. It is judgement applied to the history of judgements, which is what learning from a record means. The physician analogy returns with its proper limit: the triage officer who declines to pronounce a patient terminal on the evidence of a fever is not evading medicine, and the one who pronounces it on that evidence is not being brave. Both the order’s undertakers and its immortality salesmen are offered the same discipline, and the offer’s symmetry is the guarantee that the discipline is a method rather than a side.

One concession completes the reply, and it is made without reservation. A diagnostic is not a politics, a strategy or a consolation, and a reader who came to the title’s question wanting to know what is to be done will leave with the question intact. That is a real limitation, acknowledged rather than disguised. The article’s claim is narrower and, it maintains, prior: that the arguments people actually have about what is to be done are corrupted at the source when the diagnostic question is answered by slogan, and that clearing the corruption is work worth one long article in a series of three hundred. The wider question of what the tradition still offers the present is taken up in the series’ assessment of whether Marxism is still relevant, which this diagnostic supports but does not replace.

Worked case: the postwar settlement and its ending

The diagnostic is best learned on a case where the sequel is known, and the ending of the postwar settlement is the cleanest level two episode in the record. Consider what an analyst equipped only with the levels could have said at the start of the nineteen seventies, and what the subsequent decades did to each possible assignment. The configuration’s components were failing together in the pattern the regulation school later codified. The productivity growth of the core production model had slowed from its postwar pace. The wage-productivity bargain had become a battleground as unions defended real wages against the inflation that governments were using to dissolve distributional conflict. Policy instruments contradicted the problems they faced: demand expansion, the regime’s master tool, now fed price rises faster than employment, while monetary restraint deepened unemployment without quickly restoring profitability. The international monetary anchor collapsed at the decade’s opening, removing the framework within which national settlements had been negotiated. Each failure was documented at the time, and each was a failure of a named component rather than a general miasma of decline. A level two assignment was therefore available in real time, with named falsifiers: the assignment would have been defeated by a durable restoration of productivity growth under the old production model, a renewed wage bargain on the old terms, and policy instruments recovering their coherence. None arrived.

The terminal reading of the same episode was also available and widely taken. From one side, the stagflation decade was read as the general crisis of the system, the point at which its contradictions finally became unmanageable. From another, monetarist and market-liberal critics read it as proof that the postwar attempt to manage capitalism had itself been the disease, and their prescription was a reorganisation in the opposite direction. What actually unfolded was reorganisation without replacement: production restructured around flexible methods and global supply networks, finance liberalised and enlarged, the wage bargain replaced by labour market flexibility and household credit, the state’s role shifted from demand manager to guarantor of monetary stability and financial order. The defining relations of the order, wage labour and accumulation for profit, emerged intact and in some respects extended. The episode thus grades, in the diagnostic’s retrospective judgement, as a textbook level two crisis: severe, prolonged, configuration-ending, and system-preserving. Its lesson for the level test is the asymmetry of the obituaries. The level three readings of the nineteen seventies are remembered as rhetoric; the level two readings became the foundation of an entire school of economic analysis. Evidence, given time, does grade the graders.

A further lesson concerns pain. The transition between configurations was not a seminar exercise: it ran through industrial regions stripped of their industries, through unemployment rates in the old manufacturing heartlands that scarred a generation, through the deliberate use of recession as an instrument of restructuring. Analysts sometimes write as if assigning a crisis to level two, rather than level three, minimises it. The case forbids that reading. Regime transitions are among the most destructive peacetime processes modern economies undergo, destructive precisely because the old settlement’s protections are dismantled before the new one’s are built, and the people whose lives were organised around the old configuration pay the conversion costs. The diagnostic’s levels are grades of structural depth, not grades of human seriousness, and the confusion of those scales is one more way the public argument poisons itself.

Worked case: the long price decline of the late nineteenth century

The decades after the panic of the early eighteen seventies offer an earlier exercise in the same discipline, complicated by the fact that the episode’s very name is contested among historians. Prices fell, with interruptions, across much of the industrial world for over two decades. Agricultural districts suffered severely as cheap grain from new continents undercut old producers. Profit margins in the staple industries of the first industrial age were squeezed, and contemporaries spoke of a great depression in trade and industry, a phrase later historians have questioned because output and real incomes in the leading economies continued to grow through most of the period. The episode is a gift to the diagnostic precisely because its evidence points in two directions at once: the price and profit series describe distress, the quantity series describe expansion, and an analyst armed only with aggregate mood would be forced to choose a story by temperament.

The level reading dissolves the false choice. The configuration built on textiles, iron, steam and the family firm had matured; its productivity frontier was flattening in its core branches, and its price structure reflected both that maturity and the deflationary discipline of the gold standard. Simultaneously, a successor configuration was assembling inside the old one’s shell: steel replacing iron, chemicals and electrical engineering rising as leading sectors, the joint-stock corporation and the investment bank reorganising the ownership and finance of industry, and the great export of capital knitting the world economy together. Distress and expansion were both real because they belonged to different strata of the same transition: the declining branches generated the statistics of suffering, the rising branches the statistics of growth. An obituary for the order, of which the period produced several, mistook the first stratum for the whole. An insistence that nothing structural was happening, also heard at the time, mistook the second stratum for the whole. The diagnostic’s assignment, level two with strong level one episodes embedded, accommodates both bodies of evidence without amputating either, which is the test of a classification: not whether it flatters a prior, but whether it leaves less evidence orphaned than its rivals.

The case carries a second, quieter lesson about sources. Later generations inherit an episode largely through its loudest contemporary interpretations, and the late nineteenth century’s interpretations were produced by interested parties: protectionists campaigning against free trade, bimetallists campaigning against gold, agricultural lobbies, and the first theorists of imperialism reading the export of capital as the order’s final stage. Each had reasons to maximise or minimise the distress. The diagnostic’s insistence on components, on asking which linkage was failing and which was being born, is a defence against inheriting a lobby’s press release as history. Modern episodes arrive with louder lobbies and faster presses, and the defence is correspondingly more necessary.

Worked case: the depression decade as the hardest test

Every diagnostic should be confronted with the case most likely to break it, and for the three levels that case is the nineteen thirties. If any episode in the advanced economies deserved the level three assignment, this was it. The collapse began in finance and refused to stay there: banking systems failed in waves in the United States and Europe, industrial production fell by magnitudes outside living memory, world trade contracted catastrophically as states raised barriers and clung to gold, and the deflationary spiral ground on for years because the configuration’s own policy orthodoxy prescribed the medicine, balanced budgets and monetary defence, that deepened the disease. Unemployment on standard estimates reached roughly a quarter of the American labour force and comparable heights elsewhere, with all the political consequences the decade displayed. Theorists of breakdown did not lack for an audience. Nor did the decade’s other prophets: the period’s serious literature is full of considered judgements, from economists of several schools, that the order had entered its final phase and that recovery within its relations was impossible.

What defeated the terminal assignment was not a refutation in print but a reorganisation in fact, assembled from elements that the breakdown literature had not counted in the repertoire. Deposit insurance and bank regulation rebuilt the financial machinery on new terms. Governments learned, under the pressure of war mobilisation as much as of theory, to run economies at full employment by deliberate demand. Welfare states converted the unemployed from a revolutionary mass and a fiscal embarrassment into insured citizens with maintained purchasing power. Bretton Woods reconstructed the international monetary framework whose absence had turned national defences into collective disaster. The postwar expansion that followed was not a resumption of the nineteen twenties configuration but a new settlement, differing from its predecessor in its financial rules, its wage institutions, its state capacities and its international order, precisely the component list by which the diagnostic recognises a regime change. The hardest case in the record thus grades, with hindsight, at level two, and the grading carries the article’s central caution in its sharpest form: at the moment when level three seemed most nearly descriptive, the repertoire still contained a configuration nobody had forecast, assembled under pressures nobody had scheduled.

Intellectual honesty requires stating what this case does not show. It does not show that the repertoire is inexhaustible, only that it was not exhausted then, under the most unfavourable conditions yet recorded. It does not show that the reorganisation was inevitable or benign: it was purchased through a world war, and a different political sequence could have produced outcomes the diagnostic would grade differently. And it does not show that contemporaries who feared the worst were fools; given the evidence available inside the episode, before the new configuration’s components existed, pessimism was a defensible inference and optimism a faith. The diagnostic’s retrospective grades are instruments for calibrating future claims, not verdicts on the dead. Their message to the living analyst is procedural: the deeper the episode, the longer the honest analyst waits before assigning the highest level, because the highest level asserts something about a repertoire whose contents are revealed only under the pressure the analyst is watching.

The external limit: a different kind of boundary

One candidate for level three status differs in kind from everything discussed so far, and the diagnostic handles it by annex rather than by stretching a level. Climate change presents the order with a limit that does not arise from its internal ratios at all: the physical capacity of the atmosphere and the ecosystems to absorb the by-products of fossil-fuelled accumulation without transformations that damage the conditions of production itself. Marxist writing historically theorised limits as internal, arising from profitability, realisation or the composition of capital, and the tradition’s engagement with nature ran largely through the extraction of raw materials and the metabolic relation between town and country. A limit imposed by planetary chemistry, indifferent to the rate of profit and unresponsive to reorganisation of the wage bargain, does not fit the machinery of the three levels, which grade the order’s relations with itself. The series treats the ecological argument in its own right in the discussion of Marxism and ecological limits, and the present article confines itself to the diagnostic point.

Is a physical limit the same kind of problem as a market downturn?

No, and the diagnostic gains credibility by refusing to pretend otherwise. A market downturn is generated by the order’s own motions and can in principle be answered by rearranging them. A physical limit confronts every possible arrangement with the same chemistry, so the honest treatment is a standing annex: a boundary condition on all configurations.

The annex matters because physical limits interact with the levels without being reducible to them. A configuration heavily invested in carbon-intensive plant faces stranded assets and forced restructuring that can precipitate a level two transition on an accelerated schedule. A downturn that coincides with climate-driven harvest failures or infrastructure damage will be deeper than its internal mechanisms predict. Conversely, the investment programmes of an energy transition constitute a frontier of accumulation of the classic kind, absorbing capital and labour into new sectors, which is one reason the ecological question cuts in both directions in crisis argument. The diagnostic’s contribution is again separation: it prevents the physical limit from being used as a rhetorical trump that ends all level distinctions, and it prevents the level distinctions from being used to wave the physical limit away. Both evasions are common; both are refused here.

Late stage capitalism: a slogan examined

No phrase in this territory circulates more widely with less content than late stage capitalism, and the diagnostic’s treatment of it is a useful demonstration of the level test applied to vocabulary. In popular usage the phrase attaches to whatever feature of commercial life the speaker finds absurd or oppressive: a fee, a queue, an advertisement, a product of conspicuous uselessness, a corporate slogan of conspicuous emptiness. It functions as a sigh with footnotes, expressing the feeling that the order has become decadent, self-parodic and surely near its end. As analysis it offers nothing, because it specifies no mechanism, no level and no falsifier: no observation about commercial life could count against it, since cheerful efficiency and grotesque dysfunction are both recruited as symptoms of lateness. The phrase’s career illustrates the general law that the less a crisis term discriminates, the faster it travels.

The phrase has a scholarly ancestry that deserves rescue from its descendants. Late capitalism entered theoretical vocabulary through Werner Sombart’s periodisation of capitalist development into early, high and late phases, a scheme about organisational forms rather than a countdown. Ernest Mandel’s major work of the nineteen seventies revived the term as the name for the postwar configuration specifically, arguing that the order had entered a third technological age whose dynamics, credit superstructure and permanent arms economy he analysed in detail. Whatever the verdict on Mandel’s particular arguments, his usage was analytical: late capitalism was a stage with specifiable machinery, and late referred to a position in a sequence of configurations, not to a proximity to death he could measure. The popular descendant keeps the word late and discards the machinery, converting a periodisation into a prognosis. A reader meeting the phrase should ask the level test’s questions of it. Which level is being claimed? Usually none, or the third by implication. What evidence is offered? Usually an anecdote. What would count against? Nothing the speaker will name. The phrase is therefore best filed where the diagnostic files all such material: as evidence about the speaker’s experience of the prevailing configuration, which is real data about legitimacy and mood, mistaken only when presented as data about the system’s remaining lifespan.

Legitimacy and mood are not nothing, and the diagnostic should say so before leaving the slogan. Orders do not persist by accumulation arithmetic alone; they persist partly because enough people regard them as tolerable, improvable or inevitable, and widespread irony about an order’s products is a measure, however crude, of the thinning of its cultural credit. The regulation tradition itself treats shared expectations and norms as components of a configuration, which means a collapse of confidence in a settlement’s fairness can belong to a level two story. What the level test forbids is the silent conversion: from the true premise that many people experience the prevailing configuration as exhausted or absurd, to the conclusion that the order itself stands at its end. Between those propositions lies the whole history recounted in this article, of configurations that died culturally before they died economically, and of an order that buried them and continued in altered dress.

The uneven geography of crisis

Levels are assigned to episodes, and episodes have addresses. A recurring defect of crisis argument is its treatment of the world economy as a single patient whose temperature is taken at one financial centre. The record refuses the simplification. Configurations rise and exhaust at different times in different places; a regime crisis in one region has repeatedly coincided with exuberant accumulation in another, as production, capital and confidence migrate toward the ascending configuration. The long relative decline of particular industrial regions within wealthy economies, a level two story of configuration change written in local devastation, unfolded during decades when the same order was incorporating hundreds of millions of new industrial workers elsewhere and raising their incomes at historic speed. Both descriptions are accurate; a diagnostic that aggregates them into a single planetary mood discards the information that matters, which is the distribution.

The tradition’s own geography was alert to this from early on. Its theories of imperialism and uneven development treated the world order as a hierarchy of zones moving at different speeds, with crisis in the core and expansion at the periphery capable of coexisting for generations, each conditioning the other. World-systems analysis later formalised the picture into core, semi-periphery and periphery, with hegemonic cycles longer than any business cycle setting the frame. The diagnostic borrows the caution without adopting any full system: a level assignment should state its territory. A claim about the exhaustion of a configuration in the advanced economies is one claim; the same words applied to the world order are a different and much larger claim, requiring evidence about zones where accumulation is young, populations are still urbanising and the configuration being pronounced exhausted was never installed. Slippage between the two claims is among the most common and least noticed in the genre, and the level test’s territory question, asked alongside its level question, catches it.

Who declares a crisis, and what the declaration does

A crisis is partly an event and partly a speech act, and the diagnostic would be naive if it ignored the second half. To declare a crisis is to do things with words: to demand emergency measures, to suspend ordinary scrutiny, to reframe distributional choices as necessities, to claim the authority of the person who saw it coming. Governments declare fiscal crises to discipline opponents; firms declare existential crises to extract concessions from workforces; movements declare terminal crises to summon commitment; sellers of financial products declare crises to sell shelter from them. None of this makes declared crises unreal. It makes the declaration itself a datum to be examined with the same level questions, since the incentive structure of declaring rewards the highest level available: nobody mobilises a parliament, a workforce or a readership with the announcement that a configuration requires gradual reorganisation over two decades. The inflation of levels is therefore not an intellectual accident but a predictable output of the declaration market, and the reader who knows the market’s bias can discount accordingly, the way a reader discounts a headline’s adjectives.

The scholarly world is not exempt from the market. Careers are built on named crises, journals reward the dramatic framing, and a literature that announced the manageable difficulties of a maturing configuration would sell fewer copies than one announcing the end of an epoch. Against these incentives the diagnostic offers only its procedure, which has the virtue of being equally inconvenient to every camp. It asks the undertaker for a falsifier and the celebrant for one too; it asks the minister and the pamphleteer the same four questions; it files the answer, and it remembers the file. Traditions of argument improve when their defeats are kept, and the history surveyed in this article is largely a history of defeats unkept, each generation’s obituary written in ignorance of the last one’s fate. A modest filing discipline, levels marked and dates noted, is a genuine if unglamorous contribution to breaking the cycle.

Endings without successors

One question in the article’s territory is regularly mishandled by being fused with another: whether the order can end, and what would replace it if it did. The diagnostic separates them because their evidence differs entirely. The series’ companion treatment of the order’s present form, the analysis of Marxism and modern capitalism, examines the configuration question in its current dress; here the ending question is considered in the abstract form the level test requires. Historical modes of production have mostly ended without a successor standing ready in finished form. The slave economies of antiquity decayed across centuries into arrangements nobody had designed. Feudal relations dissolved unevenly, region by region, over a span longer than capitalism has yet existed, and the order that eventually succeeded them was assembled from elements, markets, wage labour, enclosures, whose combination no contemporary foresaw as a system. Endings in the record are typically degenerations, fragmentations and slow losses of reproductive capacity, punctuated by local collapses, rather than the clean succession of one total order by another on a date.

This matters for the level test because level three, stated carefully, asserts a failure of reproduction, not the arrival of a replacement. A claimant is therefore not required to produce the successor society in blueprint before the claim can be discussed, and defenders are not entitled to dismiss the claim by demanding one. What the claimant is required to produce is the level three evidence: the demonstrated exhaustion of reorganisation. Conversely, the absence of a visible successor is not itself evidence of exhaustion, since successors in the record have generally been invisible in advance, assembled retrospectively from components that contemporaries classified as marginal or degenerate. Both camps habitually commit the mirror fallacies here: one treats the missing blueprint as proof that no ending is possible, the other treats the missing blueprint as the ending’s mysterious confirmation. The diagnostic declines both, and keeps the question where it can be worked on: in the observed performance of reorganisations, attempted and completed, across the configurations the order has inhabited.

What would have to change for the assessment to move

A diagnostic that cannot say what would alter its reading is a creed, so the conditions belong in the open. The retrospective pattern of the record is that slow evidence accumulates at level two while level three remains undemonstrated; the observations that would disturb that pattern can be specified without pretending to forecast them. At the level of reorganisation, the significant observation would be successive failures of genuinely different configurations: not one settlement exhausting itself, which is the normal rhythm, but a sequence in which each attempted rearrangement, liberal and interventionist, financial and industrial, national and international in design, failed faster and more completely than its predecessor, with the intervals of stability shortening across the sequence. At the level of the production frontier, the significant observation would be a productivity stagnation that persisted across successive technological paradigms and institutional settings alike, surviving changes of regime that historically restarted it, since that pattern would separate the stagnation from any particular configuration and attach it to the order’s mature structure. At the level of finance, it would be debt dynamics that escaped each successive containment technology, so that the repertoire of monetary and fiscal expedients demonstrably shrank with use. None of these observations describes a bad year or a lost decade; each describes a pattern across attempts, and the diagnostic is explicit that nothing shorter can carry the weight, because everything shorter has already happened, repeatedly, inside the order’s continuing life.

The symmetric conditions deserve equal statement, since the diagnostic grades defenders as strictly as critics. The reading that all is structurally sound would be disturbed by a configuration’s failure that its managers could neither explain nor repair within its own terms, by distributional settlements collapsing faster than replacements could be negotiated, or by the slow evidence on productivity and dynamism deepening across a full generation without a successor model emerging from the order’s considerable research apparatus. A defender who cannot name such conditions is holding a faith symmetrical to the undertaker’s, and the level test was built to be unusable as a shield by either. What the instrument finally measures is not the system, about which it maintains a disciplined agnosticism at the highest level, but the quality of claims about the system, which it grades without mercy in all directions.

Misreadings this diagnostic exists to correct

Certain errors recur so reliably in this territory that correcting them is half the article’s purpose. The first holds that the recurrence of crises itself indicates an approaching end, as if repetition were escalation. The record shows recurrence as the order’s normal metabolism across two centuries of expansion, and nothing in a series of events licenses a conclusion about the series’ termination without an argument about changing structure. The second treats the phrase late stage capitalism as an analytical category with a definition and a literature of measurement behind it; as the section above argues, the popular phrase has neither, and its scholarly ancestor meant a periodisation, not a prognosis. The third confuses stagnation with crisis, merging a chronic condition with an acute event and thereby inflating the event count while draining the condition of its specific remedies. The fourth asserts that predicting collapse is a requirement of Marxist analysis, a duty from which doubters defect. The record of the tradition refutes it: major Marxist economists have denied any automatic breakdown, criticised breakdown laws from inside the school, and built reputations on the denial, and the founder’s own texts supply tendencies and countertendencies rather than a timetable. The fifth error runs in the opposite direction and belongs to the order’s celebrants: that surviving a crisis proves stability, as if each survival banked permanence. Survival proves that the repertoire was not empty on that occasion. It is evidence about a past capacity, and its conversion into a guarantee is the mirror image of the undertaker’s conversion of distress into doom.

Beneath all five sits the single confusion the level test was built to expose: the treatment of crisis as a property that events simply have, visible to anyone honest enough to look, rather than as a classification that arguments must earn by specifying level, mechanism and falsifier. The errors persist because each serves a rhetoric, and they will survive this article. What an article can do is give the reader who wants better arguments a portable procedure and a worked set of precedents, so that the next encounter with any of the five begins from recognition rather than persuasion. Recognition is not refutation, but it is the precondition of refutation, and in a debate this old and this loud it is no small gift.

The concept inside the tradition: a short history

The word crisis came to economics from medicine, where it named the turning point of an illness, the day on which the disease declared whether the patient would recover or die. The metaphor smuggled a prognosis into a description from the very beginning, and political economy has never fully expelled it. Marx used the term with the medical charge deliberately available, but his usage in the mature work is more differentiated than his popular image suggests. In his journalism of the eighteen fifties he tracked commercial crises as events with timetables, watching money market indicators the way a clinician watches a fever chart, and more than once expected an approaching crisis to detonate the political order of Europe. The expectations failed, and the failure is recorded in his correspondence with an irritation that later hagiography preferred to forget. The theorist of crisis was also, biographically, one of its most disappointed forecasters, and the mature theory can be read partly as the discipline he built from that disappointment: crises demoted from political detonators to structural events whose political consequences depended on organisation and circumstance.

Inside Capital the concept does systematic rather than prophetic work. Crisis appears as the moment in which the contradictions of the commodity form, sale and purchase split across time and space, money functioning as means of payment and therefore as a chain of obligations that can snap, are forcibly and temporarily resolved. The possibility of crisis is thus inscribed in the simplest forms of the system, while its actuality requires the developed machinery of credit and world market. This architecture explains a feature of the tradition that confuses outsiders: crisis for Marx is simultaneously a permanent possibility, a recurring actuality and a mechanism of restoration, and it is none of these exclusively. The breakdown reading seizes the first note, the complacent reading the third, and the level diagnostic is, in one of its aspects, simply the insistence that all three notes belong to the chord. Later systematisation flattened the chord in both directions. The Second International’s popularisers tended toward a mechanics of inevitable collapse; academic Marxism of a later period sometimes dissolved crisis into a functionalism in which every event confirmed the system’s adaptability. The original texts support neither simplification comfortably, which is why appeals to them settle so little in the contemporary argument and why the diagnostic grades claims rather than quoting authorities.

Depth, duration and the shape of an episode

Since levels cannot be read from a single statistic, the diagnostic teaches a habit of describing an episode’s shape before grading it, along three dimensions that can be assessed independently. Amplitude measures how far the principal magnitudes fall: output, employment, asset prices, trade. Breadth measures how far the damage spreads across sectors and regions, whether the contraction stays inside its sector of origin or generalises through the contagion channels described earlier. Persistence measures how long the magnitudes take to regain their previous path, and, more subtly, whether the recovery restores the previous trajectory or settles onto a lower one, leaving a permanent step down in the level of activity even after growth resumes. The three dimensions vary independently in the record. Episodes of terrifying amplitude have closed quickly and left the configuration intact; episodes of modest amplitude have persisted for decades and ended configurations. Grading by amplitude alone, the instinct of every headline, is therefore the least reliable of the three, and the diagnostic’s rule of asking duration before depth generalises: shape first, level second, verdict last.

Persistence deserves its modern name and its caution. Economists speak of hysteresis when a temporary shock leaves permanent scars, through workers whose skills decay in long unemployment, firms whose investment deferred becomes capacity never built, and regions whose decline compounds. Hysteresis complicates the level boundary in an interesting way, because a level one event with strong hysteresis can inflict level two quantities of damage without any failure of the configuration’s design: the machinery worked, but the episode was mishandled or merely unlucky, and the scars are the receipt. The diagnostic handles the complication by keeping the assignment tied to the configuration’s subsequent performance rather than to the episode’s cost. If the same arrangements, scars and all, go on to deliver a generation of functioning accumulation, the event remains level one in the structural ledger whatever its human price. If the scars become the mechanism by which the arrangements fail, through politics, demography or fiscal exhaustion, the episode is reclassified as the opening of a level two transition. Reclassification is not embarrassment; it is the system working, and the willingness to reclassify publicly is what separates an analyst from an advocate.

The repertoire: what reorganisation actually consists of

Claims about the exhaustion of reorganisation are usually made in ignorance of how various the observed repertoire has been, and an inventory disciplines them. Across the record, configurations have been rebuilt using moves drawn from a recognisable stock, combined in proportions that differed each time. Monetary frameworks have been redesigned wholesale: metallic standards adopted and abandoned, central banks founded, reformed and given new mandates, international monetary orders negotiated and later floated away. Financial architectures have been rebuilt after collapse: banking separated from speculation by law, deposit insurance invented, capital requirements imposed, resolution regimes created. The wage relation has been reconstructed repeatedly, from the suppression of combination through its legalisation to collective bargaining systems, statutory minima and, in the later configuration, deliberate flexibilisation. States have acquired and shed capacities in waves: poor relief, social insurance, public housing, demand management, industrial policy, privatisation. Corporate and legal forms have been reinvented, from the joint-stock company through the managerial corporation to the platform firm, each time altering what accumulation’s agent looks like. Trade and empire have supplied further moves: protected home markets, colonial preference, negotiated liberalisation, and the construction of regional blocs.

The inventory matters because level three asserts the stock’s exhaustion, and the assertion cannot be evaluated without knowing the stock’s historical creativity. Two features of the record stand out. First, the moves are combinatorial: no successful reorganisation used a single instrument, and the inventiveness lay in the combination, which means the space of unexplored combinations is vastly larger than the space of unexplored instruments, and claims to have surveyed it deserve scepticism proportional to that arithmetic. Second, several stock items were invented under duress by people who did not believe in them, in the middle of the episodes they answered. Deposit insurance was not waiting in a ministry drawer in the eighteen eighties; demand management at full employment was not the settled doctrine of any treasury before the nineteen forties. The repertoire is therefore not a cupboard that can be inventoried in calm times but a capacity that manifests under pressure, and the level three claimant is in the uncomfortable position of forecasting the failure of a capacity whose past performances were all surprises. The diagnostic does not conclude that the capacity is infinite. It concludes that its limits are not knowable in advance by the methods the genre uses, and that claims pretending otherwise should be graded as literature.

The state inside the picture

Older crisis writing often pictured the state as a spectator that crisis would sweep away or expose, and the record since has made that picture unusable. States are components of configurations, co-authors of their stability and participants in their exhaustion, and the growth of state capacity is itself one of the great reorganisation stories. The lender of last resort function was improvised in the nineteenth century out of repeated panics, codified by Walter Bagehot into rules that central banks then spent a century learning to extend to new kinds of institution and new kinds of asset. Each extension enlarged the set of level one events the system could absorb, and each enlargement created the moral hazard whose consequences stocked the next episode. The diagnostic reads this dialectic without embarrassment in either direction: public repair capacity is real, it changes level assignments by changing what configurations can survive, and it is not a deus ex machina standing outside accumulation, since the fiscal and monetary room it deploys is itself produced and limited by the accumulation it protects.

The consequence for crisis argument is a symmetry the genre often misses. Critics who treat every state rescue as proof that the system cannot stand on its own feet are assuming a definition of the system that excludes one of its permanent components; capitalism as historically practised has always included treasuries and central banks, and a configuration whose design incorporates public backstops is not cheating when the backstops operate. Defenders who treat each rescue as a costless vindication commit the opposite error, ignoring that each rescue draws on fiscal and political capital that is finite in any given configuration, and that a settlement can exhaust its repair credits before it exhausts its economic logic. The diagnostic’s level two lens handles both: the question is never whether the state intervened, but whether the intervention restored the configuration’s functioning or merely purchased an interval, and the answer is read from the subsequent performance of the linkages, not from the ideology of the intervention.

Financialisation graded at its proper level

The growing weight of finance within the advanced economies, its share of profits, its instruments, its political gravity, is among the most debated features of the prevailing configuration, and it illustrates the level test’s value on contested terrain. One literature reads financial expansion as the order’s senescence: capital, finding production unrewarding, migrates into speculation, and the resulting hypertrophy of claims on future income marks the autumn of the system. Giovanni Arrighi gave this reading its most elegant historical form, tracing earlier financial expansions in Genoese, Dutch and British cycles as the signal phase of a hegemonic order’s decline, after which leadership passed elsewhere. A second literature reads the same facts as one configuration’s solution to its predecessor’s problems: finance as the mechanism that globalised supply chains, disciplined national settlements, funded the technology sector and supplied the asset appreciation that substituted for wage growth in sustaining demand. On this reading financialisation is neither autumn nor spring but a regime’s operating system, with a regime’s characteristic virtues and pathologies, including a demonstrated tendency to generate the severe level one events that periodically threaten to open level two transitions.

The diagnostic does not choose between the literatures by taste; it specifies what each must show. The senescence reading owes an account of why financial expansion in earlier cycles accompanied the transfer of dynamism between centres rather than the death of the order, and why the same symptom should now be read at system level. The operating-system reading owes an account of the configuration’s repair capacity: whether a financialised settlement can survive its own crises without consuming, in each rescue, more of the fiscal and political capital that the next rescue will require. Both accounts, notably, are arguable with evidence, because both are at bottom level two arguments about the durability of a configuration and the direction of the world order’s geography, whatever civilisational language they borrow. Hyman Minsky’s financial instability hypothesis slots into the same placement with unusual neatness: his progression from hedge to speculative to Ponzi finance describes how stability within a configuration breeds the fragility that ends its calm phases, a mechanism theory for level one events whose severity rises with the configuration’s age. The diagnostic receives Minsky as it receives the others: a supplier of mechanisms, whose findings the level test then grades for reach.

Generations and the experience of unprecedentedness

Every generation meets its first great crisis as an unprecedented event, and the feeling is both sincere and statistically predictable. A worker whose adult life spans one configuration knows its downturns as weather; the first downturn that threatens the configuration itself arrives, by definition, without personal precedent, and the mind reaches for the largest category it owns. Memoirs of the nineteen thirties are full of intelligent people recording that the world was ending; so are the pamphlets of the eighteen seventies and the broadcasts of every severe modern recession. The diagnostic does not mock the feeling, which is information about the configuration’s cultural hold. It insists only that the feeling’s intensity is a constant across episodes of very different structural depth, and therefore cannot grade them. What varies, and can grade them, is the behaviour of the linkages afterwards.

There is a related trap in the opposite emotional direction, the survivor’s calibration error of those who lived through a severe episode and its reorganisation. Having watched the repertoire perform once, the survivor concludes it always will, and meets the next configuration’s exhaustion with the complacency of a person quoting a base rate at a regime change. The two errors alternate in the culture with the regularity of the cycle itself: the young pronounce endings, the old pronounce impossibilities, and the historical record, which contains both endings of configurations and survivals of the order, rebukes each in turn. An instrument like the level test cannot supply wisdom, but it can supply the questions that wisdom asks, in an order that slows both reflexes down: which level, what evidence, what falsifier, what timescale, what territory. Five questions are a modest inheritance to leave a reader. In a debate two centuries old and showing no sign of exhaustion itself, modest inheritances that survive contact with events are the only kind worth leaving.

How the levels interact

The levels are strata of one process, not sealed compartments, and their interactions are where analysis earns its keep. Level one events act as the selection environment in which configurations prove or lose their fitness: a settlement that absorbs a dozen downturns across a generation has demonstrated something real about its repair machinery, and each absorption both validates the design and, by encouraging leverage, complacency and the extension of every successful practice beyond its competence, sows the episode that will test it harder. Level two transitions are almost never announced by their participants as transitions; they begin as level one events that behave wrongly, recoveries that stall, instruments that stop working, and they are recognised as transitions only when the failures accumulate into a pattern that a new generation of analysts names. The naming itself is an event in the process, since a configuration widely believed exhausted loses the confidence on which its credit and cooperation rest, which is one reason incumbents resist level two language so fiercely and insurgents reach for level three language so eagerly.

Movement between levels also runs downward in significance over time. The terror of a level two transition fades as the successor configuration normalises; practices that arrived as emergency improvisations become the plumbing of ordinary life, and the previous settlement’s certainties become museum pieces that students must be taught to imagine. This normalisation is the deep reason crisis literature ages so badly: written at the moment of maximum uncertainty, it projects that uncertainty forward as a permanent condition, while the record keeps converting the unprecedented into the administrative. The diagnostic’s timescale question is the antidote it can offer. Ask of any crisis text how old its configuration’s troubles are, how many repairs have been attempted and with what half-lives of success, and the text will usually grade itself, its evidence sorting into the levels faster than its author sorted it. Analysis begins, very often, at the moment a reader stops asking whether a writer is optimistic or pessimistic and starts asking which stratum of the process the writer’s facts describe.

Reading an episode’s signature

Mechanisms leave signatures, and learning to read them is the practical skill that connects the diagnostic to the cluster’s crisis theories. An episode that begins with unsold inventories piling up in consumer goods while order books in capital goods still boom carries the signature of a realisation failure, the underconsumption pattern in which final demand falters first. An episode that begins with shortages, bottlenecks and price spikes in particular inputs while other branches glut carries the signature of disproportionality, the planless interdependence of branches asserting itself. An episode arriving at the end of a long boom with labour markets tight, wage settlements outrunning productivity and margins compressing across sectors carries the profit squeeze signature. An episode detonated in credit markets, where the failure of a class of assets freezes the payments system before the real economy has weakened, carries the financial signature, which Minsky’s progression describes from inside the borrowing units themselves. Signatures are rarely pure, and mature episodes mix them as contagion repaints the original pattern across the economy. But the discipline of asking which signature appeared first, before the generalisation, protects the analyst from the laziest inference in the genre: that because everything eventually falls together, everything fell for the same reason.

Signature reading also disciplines the level assignment, because mechanisms differ in their natural reach. A pure disproportionality break inside one supply network is the system’s routine self-correction and grades low unless the network is load-bearing for the configuration. A profitability failure generalised across the capital stock touches the engine of accumulation itself and bears watching at a higher level, though the record of profit squeezes absorbed by restructuring is long. A financial signature in a configuration whose demand model runs on household credit is a different event from the same signature in a configuration where credit is ornamental. Diagnosis, in this field as in the medicine from which its master word was borrowed, is the art of combining the mechanism’s identity with the patient’s constitution, and the constitution is the configuration. The levels name the possible depths of the illness; the signature names its kind; the constitution names the body in which kind and depth interact. An analysis missing any of the three is a horoscope with statistics.

Why configurations die quietly between crises

Exhaustion is mostly not an event. Between the dramatic episodes on which crisis literature concentrates, a configuration’s fitness is decided by slow processes that rarely make headlines: the productivity frontier of its core technologies flattening year by year, its distributional bargains hardening from settlements into entitlements that no party can renegotiate without electoral or industrial punishment, its most successful firms maturing from innovators into defenders of position, its policy doctrines ageing from instruments into incantations repeated after their conditions have passed. By the time a level one event arrives that the machinery cannot absorb, the outcome has usually been prepared by a decade of quiet deterioration, and the event itself is better understood as the audit than as the crime. This temporal structure explains a recurrent puzzle in the literature, the way level two transitions surprise contemporaries despite being, in retrospect, overdetermined. Attention is episodic and exhaustion is not; the diagnostic corrects the mismatch by directing the analyst to the inter-crisis indicators, the productivity trend, the settlement’s renegotiation costs, the doctrine’s fit to its problems, rather than to the crash that eventually prices them in.

The quietness has a second implication, for the timing of level assignments. If exhaustion accumulates between episodes, then the absence of crisis is not evidence of a configuration’s health, any more than the presence of crisis is proof of its death. Calm decades can be the sound of a frontier flattening under full employment. The celebrant’s favourite statistic, the length of the expansion, therefore grades as weakly as the undertaker’s favourite, the depth of the fall. Both are amplitude and duration measures of episodes; neither inspects the linkages. Analysts trained by the diagnostic develop a characteristic boredom with the question that dominates broadcasts, whether a crash is coming, and a characteristic interest in duller questions: whether the wage bargain can still be rewritten, whether the productivity engine has a successor in visible development, whether the state’s repair credits are replenished or drawn down. The history in this article suggests the dull questions have the better forecasting record, insofar as anything in this field forecasts at all.

Two centuries of configurations in outline

The level scheme gains solidity when the whole record is passed in review, because the sequence of configurations and their characteristic crises is the nearest thing the subject has to a laboratory. The order’s first industrial configuration, built on textiles, steam and the competitive family firm under a minimal state, generated the decennial commercial crises surveyed earlier and exhausted itself in the long price decline of the late nineteenth century, as its branches matured and its scale outgrew its organisational forms. Its successor, the corporate configuration of steel, chemicals, electricity, railways and finance capital under increasingly interventionist states, generated crises of a new scale, including the greatest crash in the record, and exhausted itself in the depression decade when its monetary orthodoxy and its absence of demand management converted a financial break into a civilisational emergency. The Fordist configuration that followed, built on mass production, the productivity wage bargain and the managed economy, generated the mildest downturns in the record and exhausted itself in stagflation when its frontier flattened and its bargains rigidified. The financialised configuration assembled from its ruins generated crises of increasing financial amplitude at shortening intervals in several economies, and its durability remains the live analytical question of the present literature, which the diagnostic treats with its usual refusal: the slow evidence against it is level two evidence, substantial and growing in the judgement of many specialists, and the level three conclusion remains, as ever, unearned by the evidence offered.

Read as a sequence, the record teaches two lessons simultaneously, and partisanship consists of learning only one. The first lesson is mortality: every configuration died, including the ones contemporaries experienced as the end of history, and the intervals between their deaths were measured in decades, not centuries. An analyst who treats the prevailing configuration as permanent is betting against the only perfect record in the subject. The second lesson is succession: every death so far was a succession, the order emerging from each interregnum with its defining relations intact and its geography, technology and class structure transformed. An analyst who treats the next succession as impossible is betting against the same record from the other side. The level test is simply these two lessons converted into procedure, and its famed refusal to answer the title’s question is the refusal to bet the procedure on either side of a record that rebukes both.

Designing versus discovering a configuration

A question the record raises, and the diagnostic needs for its level two prognoses, is whether configurations can be designed or only discovered. The postwar settlement, the most successful configuration in the record, was partly designed: its monetary order was negotiated at a conference table, its welfare components were legislated from manifestos, its full employment commitment was proclaimed in white papers. Yet its core, the productivity bargain between mass production and mass consumption, was nobody’s blueprint; it crystallised from practices firms, unions and states had improvised separately, and the designers of the monetary order did not foresee it. The financialised successor was even less designed, precipitating out of deregulation decisions taken for other reasons, technological changes in computing and communications, and the competitive pressure of restored capital mobility. The mixed verdict matters practically. If reorganisations were pure design, their exhaustion would invite a search for better designers, and level two prognosis would be a branch of policy advice. If they were pure emergence, prognosis would be meteorology. The record supports a middle position with real content: the components can be legislated, but the complementarity that makes components into a configuration cannot, since it depends on technologies, expectations and world conditions that no legislator commands. Reorganisation is therefore partly a lottery conducted by committees, which is a further reason level three claims, requiring the lottery to be unwinnable in principle, outrun what any observer can know.

The household scale, where levels are lived

Aggregate prose about configurations can drift far from the units in which economic life is actually experienced, and a section at household scale is a corrective the diagnostic owes its readers. For a household, the levels have translations. A level one event arrives as a lost job in a town where other jobs exist, a savings pot dented, a year or two of fear followed by reassembly on roughly the old terms. A level two transition arrives as the discovery that the old terms are not coming back: the industry that employed the district is not rehiring at any wage the household can accept, the house that anchored its wealth is valued in a market that no longer believes in the district, the skills certified in one configuration are discounted in the next, and the children’s rational strategy is departure. Nothing in the household’s experience distinguishes, in the living of it, between a configuration dying and a world ending, and the literature of every transition is full of testimony, accurate as testimony, that the world ended. The diagnostic’s insistence on levels is not a correction of that testimony but a refusal to let aggregate analysis borrow its authority. The world of a configuration does end; that is what a level two crisis is. The levels exist so that the ending of worlds can be discussed in its full seriousness without being confused, by either side, with the ending of the world.

The household scale also explains the political temperature of level two episodes better than aggregate statistics do. Distributional conflict inside a functioning configuration is a negotiation over growth’s increments, conducted in the expectation that tomorrow’s bargain starts from a larger total. The same conflict inside a dying configuration is a fight over a shrinking inheritance, in which each group’s defence of its position is rational and the sum of defences is paralysis. Policy doctrines built for the first setting fail in the second not because they are foolish but because the game has changed underneath them, and the failure of familiar doctrines is itself one of the most reliable surface signs that a level two transition is under way. Analysts who watch only the aggregate indices will miss the sign; analysts who watch the temperature of distributional politics without the level vocabulary will misread it as moral decline. The diagnostic exists in the space between those two mistakes.

The question behind the question

Nobody asks whether capitalism is in crisis out of idle curiosity, and the diagnostic works better when the motives stacked beneath the question are brought into the light. For some askers the question is biographical: a lost job, a foreclosed house, a degree that did not convert into the life it promised, and the wish to know whether the loss was personal failure or structural weather. For some it is political in the honourable sense: a judgement about whether to invest hope in reforming the prevailing configuration or in replacing the order, a choice on which lives and movements are genuinely staked. For some it is professional: the forecaster, the investor, the official, each needing a level assignment to price, to hedge or to govern. And for some it is eschatological in the strict sense, a secularised wish for an ending that would redeem the suffering the order undeniably produces, by giving it a terminus and a meaning. The diagnostic addresses all four askers differently. To the biographical asker it offers the household translations of the levels, which locate a private disaster in its structural weather without pretending the location consoles. To the political asker it offers the honest finding that the evidence disciplines strategy less than strategists wish, since the record supports neither the certainty that reform is futile nor the certainty that it suffices. To the professional it offers the procedure. To the eschatological asker it offers only the record, which is a cold gift: endings of configurations, yes, repeatedly, and no ending of the order, yet, in two centuries of asking. A question asked for four reasons cannot be answered with one word, and the feeling that it should be is the eschatological asker’s feeling wearing the others’ clothes.

Legitimacy: the cultural component of a configuration

The regulation tradition’s component list usually stops at production, wages, money and the state, but the worked cases keep suggesting a fifth component that the diagnostic should name: the stock of legitimacy a configuration commands, the widely shared sense that its arrangements are tolerable, its rewards roughly connected to contribution, and its hardships the price of a system that could not obviously be better. Legitimacy behaves like the other components. It accumulates during a configuration’s successful decades, it is drawn down by visible unfairness, by rescues that socialise losses while privatising gains, by the spectacle of rules rewritten for the well connected, and it exhibits the same complementarity: a settlement whose economic linkages still function can be destabilised by legitimacy failure, as the nineteen seventies showed when inflation, experienced as arbitrary theft, corroded consent for the wage bargain faster than the productivity figures alone would predict. The depression decade offers the darker demonstration. What transformed an economic collapse into a civilisational crisis in several countries was the exhaustion of belief in the old order’s competence across its own elites, a legitimacy bankruptcy that preceded and licensed the political catastrophes. Level assignments that ignore the cultural balance sheet will repeatedly be surprised by the speed of transitions whose economic components looked, on paper, repairable.

Measuring legitimacy is admittedly softer work than measuring productivity, and the diagnostic does not pretend otherwise. Its indicators are the kind historians use: the migration of serious opinion, the defection of a configuration’s own beneficiaries from its defence, the changing content of common sense about what is fair, possible and inevitable. Such indicators cannot be read off a release schedule, and they invite the analyst’s prejudices to masquerade as findings, which is a real danger the article has flagged in other contexts. But softness is not the same as irrelevance, and a component that decided the fate of the hardest case in the record cannot be omitted because it resists the spreadsheet. The level test handles the problem in its characteristic way, by converting the component into questions. Who is defecting from the settlement’s defence, and are they its former engineers or its permanent critics? Is the unfairness being alleged episodic, attached to a scandal, or structural, attached to the settlement’s normal operation? Are the alternatives being imagined rearrangements of the configuration or replacements of the order? The answers sort legitimacy evidence by level as the procedure sorts everything else, imperfectly and better than the alternatives on offer.

What remains genuinely in dispute

An article that has spent its length narrowing disputes should end the analytical work by marking the disputes that remain genuinely open, where the diagnostic organises the argument without settling it. Three stand out. The first concerns the productivity future: whether the slowdown across the decades since the postwar expansion reflects the exhaustion of a particular technological paradigm, in which case a successor paradigm restarts the engine, or a deeper feature of mature service economies, in which case configurations of every design inherit the constraint. Specialists of the first rank divide on the question, the measurement is honestly difficult, and the answer decides how much of the slow evidence belongs to level two and how much leans toward level three. The second open dispute concerns the state’s remaining repair capacity: whether the fiscal and monetary room consumed by successive rescues is replenished by growth and financial repression as after the postwar debts, or whether the prevailing configuration’s low growth and high obligations have converted the repairer into a source of fragility in its own right. The third concerns the direction of the world order’s geography: whether dynamism’s migration between regions represents the system’s healthy succession mechanism, one centre’s configuration replacing another’s, or a fragmentation in which no successor configuration achieves the coherence the role requires. Each dispute is arguable with evidence, each is being argued by serious people, and on each the diagnostic’s contribution is limited to keeping the levels, territories and falsifiers straight while the evidence accumulates. A map of open questions, honestly marked, is worth more than a verdict smuggled past them, and it is the correct note on which to approach the article’s close.

The limits of the diagnostic itself

The instrument must finally be turned on its own maker, because a classification that cannot state its limits will be used beyond them. The diagnostic grades claims; it does not generate knowledge of the economy, and a reader who applied its procedure flawlessly to bad evidence would produce well-formed error. Its assignments are retrospective in their reliable form, since only the sequel distinguishes a stalled recovery from a dying configuration with certainty, and its prospective use yields probabilities and watchlists rather than knowledge. It inherits the biases of its categories: the three levels were distilled from the European and North American record principally, and configurations elsewhere, assembled under different state traditions and inserted differently into the world economy, may sort imperfectly under them. It is silent, by design, about timing, and therefore useless for the purposes, speculation and electoral prophecy among them, for which most people reach for crisis language. And it carries a standing temptation of its own, the scholastic pleasure of sorting as a substitute for judging, which the objection section pressed hard and which the reply answered only on condition that falsifiers are named and defeats recorded. A diagnostic is a tool for keeping a debate honest. It is not a worldview, and the moment it is wielded as one it joins the genre it was built to correct.

A question that can finally be argued with

The title of this article asks whether a crisis of capitalism is under way, and the article has declined, on principle and at length, to answer in the form demanded. The decline is the answer’s substance rather than its absence. Sorted by the level test, the question dissolves into three questions that evidence can actually engage: whether a downturn is in progress that the prevailing configuration will absorb, as downturns have been absorbed across two centuries; whether that configuration is exhausting itself, a question on which the slow evidence about productivity, dynamism, debt and legitimacy bears directly and on which serious argument is possible; and whether the order’s capacity for reorganisation is itself exhausted, a claim whose burden no claimant has yet discharged and whose repeated premature assertion is the tradition’s most instructive failure. A reader who leaves with those three questions, the table that sorts them, and the habit of asking every new crisis claim for its level, its evidence, its falsifier and its clock, leaves with more than a verdict would have given. Verdicts expire with the next quarter’s figures. Procedures survive them, and in a debate entering its third century, the survival of a good procedure is the most useful legacy one article can leave.

The record beyond the core economies

The cases worked so far come from the order’s historic centres, because the breakdown debate was framed there, but a diagnostic claiming generality must test itself against the record of the economies that were inserted into the world order on terms they did not choose. The debt crisis that opened in Latin America in the early nineteen eighties is instructive. States that had borrowed heavily in the previous decade, in currencies they did not issue, under a development model that substituted imports behind protective walls, faced a creditor world whose monetary tightening converted their debts into a trap. The lost decade that followed, of stagnation, austerity and forced liberalisation, was experienced politically as national catastrophe and narrated internationally, in some quarters, as the failure of an entire development philosophy. The level reading is more precise. The import-substituting configuration, a specific settlement linking protected industry, state borrowing, urban wage earners and commodity exports, exhausted itself under the combined strain of its productivity limits and the changed world financial conditions: a level two exhaustion of a regional configuration, resolved eventually by reorganisation into a more open and financialised settlement whose own subsequent crises then require their own assignments. Nothing in the episode grades at the level of the world order, which was at that moment entering a long expansion of its financialised configuration, though the costs were borne with a concentration the core’s aggregates never register.

The East Asian crisis of the late nineteen nineties supplies the complementary test, an episode whose opening signature was financial and whose grading was contested in real time. Economies whose development model joined state-directed credit, export manufacturing and managed currencies suffered sudden capital flight, currency collapse and, in the worst case, a political transition. Terminal language flourished briefly, including predictions of the Asian model’s general ruin and of contagion to the world system. What the sequel delivered was reorganisation at different speeds: some economies restructured their financial and corporate arrangements and resumed rapid growth within the same broad model corrected, one large economy’s political order changed while its developmental configuration persisted, and the world order absorbed the episode without structural alteration. The level test, applied with its territory question, grades the episode as a set of national level two crises, one of them severe enough to change a regime in the political sense, inside a level one disturbance of the world configuration. The episode’s lasting analytical value is the demonstration that a crisis can be simultaneously a catastrophe for the societies inside it and a reorganisation signal for the system containing it, and that refusing either half of that sentence disqualifies an account from seriousness.

The aggregates that cannot settle it

Public argument about crisis runs on a handful of aggregate numbers, and part of the diagnostic’s service is explaining why none of them can assign a level. Output growth, the headline aggregate, measures the economy’s speed without inspecting its machinery: a configuration in its vigorous decades and a configuration living on credit expansion and asset appreciation can print identical growth figures for years, and the divergence appears only when the second’s financing fails. Unemployment measures the labour market’s slack without asking whether the slack is the cyclical kind that recovery reabsorbs or the structural kind left when a configuration’s industries depart. Stock market indices, the most quoted and least informative of the three, measure the expected profitability of listed firms under the prevailing configuration’s rules, and can therefore boom precisely as the configuration’s social foundations narrow, a pattern the record displays before several transitions. Inflation, the fourth aggregate, comes nearest to diagnostic value, since its postwar behaviour marked a configuration’s distributional bargain failing; yet even inflation grades ambiguously, reading as policy error at level one and as settlement failure at level two with identical figures in the release.

Level-sensitive evidence looks different in kind. It tracks the health of linkages rather than the speed of aggregates: whether productivity growth is broadening across sectors or narrowing into a few enclaves; whether wage growth and productivity growth are moving together or parting; whether investment is funding new capacity or the purchase of existing assets; whether the financial system’s lending is creating the income from which it will be repaid; whether the state’s repair operations are shrinking or growing relative to the economy they stabilise; whether the configuration’s successor technologies are visible in development or only in prospectuses. None of these series arrives in a single release, all require judgement in assembly, and several are contested in construction. That is the honest price of asking a structural question. Analysts who find the price too high are free to ask cyclical questions instead, and the diagnostic’s only demand is that the answers then be labelled with the level of the question that produced them.

The tempo of transitions

Level two assignments carry a temporal signature that helps distinguish them from the noise of ordinary politics: they are slow. The transition out of the first industrial configuration occupied roughly the two decades of the long price decline. The transition out of the depression-era corporate configuration, from the crash to the consolidated postwar settlement, occupied a decade and a half and a world war. The transition out of Fordism, from the first productivity faltering to the consolidated financialised successor, occupied a similar span depending on the dates a historian defends. A generation is therefore the natural unit of level two time, and the diagnostic builds the expectation into its procedure for two reasons. First, it protects level two claims from premature falsification: a settlement pronounced exhausted is not refuted by five subsequent years of functioning, since transitions include long plateaus in which the old machinery works after a fashion and its replacements exist only as experiments. Second, it protects the public from a subtler fraud, the conversion of a level two diagnosis into a promise of imminent transformation. A transition measured in decades offers no schedule for anyone’s hopes, and movements or governments that borrow the diagnosis to promise quick deliverance are misusing it in a way the diagnostic explicitly disowns. Patience is an unglamorous analytical virtue. In a field whose characteristic vice is the confusion of timescales, it is also the rarest.

Tempo also explains why the participants in a transition so often talk past one another without dishonesty. A finance minister lives in quarters and must answer for the next one; a union leader lives in the life of a bargaining round and a members’ mortgage horizon; a theorist lives in the configuration’s lifespan and finds both of the others myopic; a young worker lives in the whole of a career that the transition may consume before it completes. Each timescale generates its own rational crisis vocabulary, and the same event is, truthfully and simultaneously, an emergency at the shortest scale, a danger at the middle scale and a data point at the longest. The level test does not ask any participant to abandon a timescale. It asks each to declare it, so that the argument can proceed as a negotiation among declared horizons rather than as a shouting match among hidden ones. Declarations of timescale, like declarations of level, cost nothing but the pleasure of sounding absolute, and the article has argued throughout that the pleasure is overpriced.

Permanence claims: the mirror error at full strength

Symmetry requires grading the opposite genre with the same instruments, and the genre of permanence has a record as poor as the genre of doom. Its classic statements arrive at the peaks of configurations, when the prevailing settlement’s success is generalised into the order’s essence and the possibility of structural failure is retired from serious speech. The nineteen twenties generated a literature of the new era, in which the business cycle itself was pronounced conquered by scientific management and enlightened central banking; the pronouncement had half a decade to run. The postwar decades generated a quieter version, the end of ideology, in which the managed configuration’s stability was mistaken for the permanent settlement of the questions that had agitated earlier generations. The closing years of the twentieth century generated the loudest version, an end of history in which the liberal market order’s rivals had been exhausted not for a configuration’s lifetime but for good. Each statement was level three in form, a claim about the order’s permanent character, supported by level two evidence, the performance of a configuration in its strong decades, and each was falsified on schedule by the configuration’s mortality. The level test’s verdict on the genre is identical in structure to its verdict on breakdown claims and opposite in sign: permanence, like termination, is a claim about all future reorganisations, and the evidence of one configuration’s health cannot carry it.

The symmetry is not a rhetorical ornament; it is the diagnostic’s certificate of neutrality, and it has a practical consequence for how crisis claims should be heard in the aftermath of severe episodes. The moments when terminal claims flourish are the moments when permanence claims are being falsified, and the public argument, swinging between the two genres, systematically overshoots in both directions because it possesses only the two vocabularies. The middle register, in which configurations are born, flourish, sclerose and are replaced while the order persists in altered form, has no bestsellers and few slogans. It has the record. A reader who has internalised the three levels will notice the overshoot as it happens and discount both genres by the same factor, which is the closest thing to a forecasting advantage this subject offers: not knowledge of what comes next, but immunity to the two standard ways of being wrong about it.

A note on evidence: how the cases in this article were chosen

The worked cases were selected by a rule worth stating, since case selection is where arguments of this kind are usually rigged. The rule was adversarial: include the episode that most threatens the article’s own procedure, which is why the depression decade receives the longest case treatment and is graded with its full horror intact before its reorganisation is described. Include also the episodes whose evidence points in two directions, the long price decline above all, because a classification that only meets friendly data has not been tested. Exclude episodes whose documentation remains too contested for the grading to be more than a guess, which is why several famous modern episodes appear here only as pointers to their own articles in the series. The figures used are the standard historical estimates, stated as ranges or approximations where the scholarship ranges, and the article has described trends directionally wherever precision would be borrowed authority. Readers who wish to audit the cases will find the primary scholarly literatures organised around the regulation school for the postwar settlement, around the breakdown controversy for the nineteen thirties, and around the older debate on the great depression of prices for the nineteenth-century transition. The diagnostic does not fear that audit. It requests it, on the principle stated in the reply to the objection: a framework whose cases cannot be re-examined by its readers is asking for the faith it exists to refuse.

The working file: calibration over a decade

The level test’s final recommendation is a practice rather than a proposition, and it can be described concretely. Keep, in whatever study system a reader maintains, a file of crisis claims as they are made: the claim, its level as stated or implied, its evidence, its falsifier if one was offered, its territory and its date. Review the file at long intervals and grade the claims against the sequel, reclassifying where events require and recording the original assignment beside the revised one. A decade of such filing teaches lessons no treatise can, because the errors in the file are the reader’s own intellectual environment made visible. The distribution of errors is itself a finding. In most files kept through ordinary times, the dominant error is over-assignment by critics, terminal language attached to configuration events, followed at longer lags by the incumbents’ error, the refusal to reclassify a stalled recovery as a transition while the transition completes itself around them. A rarer and more interesting error runs through the work of the best analysts: correct level, wrong mechanism, the episode graded right for reasons the signature later contradicts. A file that teaches its keeper to recognise these three errors on sight has repaid the clerical labour many times over, since the errors are not historical curiosities but the standing temptations of every participant in the argument, including the author of this article and, the record suggests, every reader who has ever been certain about the level of the moment they were living through.

The file also corrects a distortion in how the debate’s history is remembered. Public memory retains the famous obituaries and forgets the thousands of sober level two analyses that aged well, because sobriety is not quoted. A personal file reverses the selection: it preserves what one’s own contemporaries actually argued, in the proportions in which they argued it, and it usually reveals that the diagnostic’s procedure was being practised, without the name, by the analysts whose work survived. Traditions improve by promoting their careful members’ methods into common property. That promotion, modest and procedural, is the whole ambition of the three-level diagnostic, and it completes the article’s argument where the argument began: with the claim that the question in the title cannot be answered as posed, and can be answered very well indeed once it is sorted into the questions it contains.

The fetters passage and the founder’s restraint

Level three has one canonical textual root, and intellectual honesty requires examining it rather than gesturing at it. In the preface to his 1859 contribution to the critique of political economy, Marx wrote that at a certain stage of development the material productive forces of society come into conflict with the existing relations of production, and that from forms of development of those forces, the relations turn into their fetters; an epoch of social revolution, the passage continues, then begins. Generations of readers have mined the sentences for a breakdown law, and the mining is understandable: fetters, conflict, revolution, arranged in a sequence, look like a mechanism with a guaranteed output. Read with the care the diagnostic recommends, the passage says both less and more interesting things. It specifies no timescale, and its author’s historical writings measure such epochs in centuries. It asserts a conflict between forces and relations, not the cessation of accumulation; fettered forces can grow slowly, unevenly and miserably for very long periods, as the late feudal centuries demonstrated in the example Marx had nearest to hand. And its operative concept, the fetter, is a level two concept in modern dress: a relation set that once developed the forces now obstructs them is precisely a configuration whose complementarities have inverted into rigidities. The passage describes the mortality of configurations and the possibility of epochal transition beyond them; the conversion of that description into a forecast about the present order’s remaining lifespan requires premises the passage does not supply, about the impossibility of new relation sets within the old mode, which is exactly the premise every level three claimant must argue rather than inherit.

The founder’s practice matched the restraint. Marx followed the crises of his lifetime with intense political hope and repeatedly, in letters, registered the hope’s disappointment and revised his expectations in light of the recovery that followed, a paper trail that the later obituary tradition would have done well to study. Engels, editing the legacy and surveying the scene from the eighteen nineties, wrote with unusual candour that history had shown their earlier expectations of imminent transformation to have been mistaken, and that the order’s productive development had outrun the timetable the movement had assumed. The point of recalling these admissions is not to recruit the founders against their followers, a game both camps play with quotations. It is to mark that the diagnostic’s central discipline, the grading of crisis claims against the record of failed terminal expectations, was practised first by the tradition’s founders on their own expectations. The level test is therefore less an external correction imposed on the tradition than a recovery of its founders’ own second thoughts, systematised. Critics of the tradition who cite its obituary habit as a refutation of everything in it, and followers who treat every hesitation as betrayal, are alike ignorant of the paper trail, and the article’s record sections exist partly so that neither ignorance can survive a careful reading.

War and the reorganisation shortcut

The worked cases keep meeting an uncomfortable accelerant, and candour requires a section on it. The two largest reorganisations in the record, the consolidation of the corporate order’s successor arrangements after the depression and the construction of the postwar settlement, were both mediated by world war: war destroyed capital on a scale that performed devaluation’s economic function without its bankruptcies, war mobilisation taught states the demand management they later applied in peace, war debts were inflated and grown away, and war’s solidarities legitimised the welfare constructions that peacetime politics had blocked for decades. Analysts have drawn opposite morals. One school concludes that the order’s deepest crises are resolved only by catastrophe, which is itself a species of level three pessimism about peaceful reorganisation. Another notes that the lesson drawn proves too much, since the transitions of the eighteen seventies and the nineteen eighties, comparable in structural depth, were accomplished without general war, and since war’s role in the famous cases was partly that of an interruption that froze and then concentrated changes already under way. The diagnostic grades the question as genuinely open and attaches its usual discipline: claims that reorganisation requires catastrophe must survive the peaceful transitions in the record, and claims that reorganisation is a normal administrative process must survive the depression case, where the peaceful repertoire was tried for a decade and failed until war enlarged it.

A neighbouring theory deserves its paragraph because it approaches the same territory from the order’s defence and converges, instructively, with parts of the diagnostic. Joseph Schumpeter, an admirer of Marx’s economics and an opponent of his politics, described the business cycle as the form of a process of creative destruction in which new combinations of production incessantly revolutionise the economic structure from within, destroying the old one as the price of the new. The phrase has been flattened by management literature into a celebration of disruption, but in its original setting it carried the diagnostic’s asymmetry in full: Schumpeter expected the process to continue transforming the order from inside, doubted that it contained any purely economic stopping point, and located the order’s eventual danger instead in its cultural and institutional by-products, the rationalisation of life and the hostility of the intellectual stratum it produced. A defender of capitalism who thought its crisis, if it came, would arrive through success and culture rather than through breakdown arithmetic is a useful embarrassment to both genres this article has graded. He reminds the undertakers that the most serious terminal arguments may not be economic in form, and the celebrants that applause for the engine is compatible with expecting the vehicle to be retired.

Worked case in slow motion: three decades of stagnation without collapse

One modern national record deserves separate treatment because it tests the diagnostic at its most sensitive joint, the boundary between stagnation, regime exhaustion and the terminal language both attract. Japan’s postwar configuration was among the most successful ever assembled: bank-centred finance joined to industrial groups, lifetime employment in the large-firm sector, an export engine disciplined by a developmental state, and growth rates across three decades that made its model a world exemplar. Its asset bubble’s collapse at the start of the nineteen nineties opened a banking crisis that the authorities, fearing the level one event in full, chose to manage by forbearance, and the economy settled into a long stagnation of weak growth, mild deflation and rising public debt that has furnished crisis rhetoric with a standing exhibit ever since. Terminal readings were plentiful in the early years and have recurred with each subsequent disappointment. The diagnostic’s grading, applied patiently, is more interesting than any of them.

The episode’s configuration components failed in the pattern level two describes, but in slow motion and with unusual social cushioning. The bank-industry nexus that had allocated capital during the catch-up decades became a machine for evergreening bad loans. The employment settlement persisted, protecting insiders while shifting a generation of new entrants into irregular work, a distributional deformation rather than a collapse. Deflation, mild in its annual figures, persisted long enough to entrench the expectations that perpetuate it, defeating successive monetary innovations that would have worked in the textbook’s economy. Public debt rose to levels that terminal commentary elsewhere treats as a countdown clock, yet the state’s financing costs stayed negligible for reasons rooted in the debt’s domestic currency, domestic holders and the central bank’s balance sheet, a combination the countdown genre, which reads debt ratios without their institutional wrapping, consistently mishandles. Meanwhile the society remained orderly, long-lived, technologically capable and rich by any historical standard, with its stagnation experienced as drift and disappointment rather than as breakdown. Graded by the diagnostic, the case is a level two exhaustion managed at extraordinary length, inside an order whose reproduction never came into question, and its lesson cuts against two camps at once. Against the undertakers: a rich capitalist economy can stagnate for a generation, with its model exhausted and its politics gridlocked, without anything resembling a terminal crisis arriving, which is among the strongest pieces of evidence in the modern record about the distance between level two and level three. Against the celebrants: the absence of collapse is compatible with a generation of forgone dynamism, deferred reform and quietly lowered horizons, so that survival, again, certifies far less than its salesmen claim. The case is the diagnostic’s patience made visible, and it is why the procedure asks about duration with such insistence: some episodes can only be graded by waiting longer than commentary prefers to wait.

Inside a transition: the texture of the argument

Because level two transitions last a generation, most analysts spend most of their lives inside one or inside the calm that follows one, and the texture of the public argument in each phase is distinctive enough to serve as a rough field guide. In a configuration’s confident decades, crisis talk is a fringe genre, indulged as entertainment and answered with performance statistics; the level test’s questions sound pedantic, because the configuration’s own functioning supplies a standing answer to them. As the frontier flattens and the bargains rigidify, the argument’s centre of gravity shifts: anomalies accumulate, the official doctrine’s explanations require more epicycles each year, and heterodox analysts gain audience share in proportion to the anomalies rather than to their own merits. This is the phase in which the diagnostic earns its keep, because it is also the phase in which both loud genres recruit most successfully, the undertakers mistaking the configuration’s accumulating failures for the system’s, the celebrants mistaking loyalty for analysis. The transition’s crisis years then scramble every calibration: amplitude spikes, the falsifiers seem to arrive weekly in both directions, and the temptation to convert a level assignment into a tribal badge becomes nearly irresistible. What distinguishes the analysts whose work survives such decades is a visible habit the diagnostic tries to codify: they separate the strata in public, they say which component has failed and which has not, they date their claims, and they publish their reclassifications with the same prominence as their predictions. The habit looks like modesty and functions as accuracy, since a transition’s evidence genuinely does arrive stratum by stratum, and only an analyst willing to be right in pieces can track it.

The aftermath has its own texture, less studied. Once a successor configuration consolidates, the previous transition is retrospectively compressed into a morality tale, its length forgotten, its outcome naturalised, and the analysts who graded it correctly are misremembered as having foreseen the successor in detail, which none did. The compression then sets the trap for the next cycle, because a public taught that transitions are swift and legible will demand swift legibility from the next one and reward the writers who fake it. Breaking that cycle is beyond one article’s power. But a reader who has watched the full sequence described once, in the record’s own cases, has a chance of recognising the phases as they recur, and of holding the level questions steady through the phase in which they are most unfashionable, which is the phase in which they matter most.

When the centre of the system moves

One further pattern in the record deserves a section because it generates a characteristic level error all its own: the migration of the world economy’s centre of dynamism from one leading region to another. The sequence is familiar in outline. A Dutch-centred commercial order gave way to a British-centred industrial one, which gave way in turn to an American-centred order of mass production and managed money. Each succession was accompanied, in the declining centre, by a literature of general capitalist crisis: the old centre’s exhausted configuration was experienced, by those whose world it had organised, as the exhaustion of the world itself. Each succession was also accompanied by a financial expansion in the declining centre, as its capital, facing narrowing opportunities in its historic industries, migrated into finance and foreign lending, the pattern Arrighi documented across the successions. From inside the declining centre, therefore, the evidence of senescence is genuine and abundant; it is simply evidence about a centre and its configuration, not about the order, whose dynamism has relocated rather than ceased. The level test’s territory question exists largely to catch this error, which is among the most consequential in the genre because it recruits real suffering, real industrial decay and real financial hypertrophy into a conclusion they cannot support.

The pattern’s forward use requires the same care as its backward use. Observers who note a present migration of dynamism between regions are observing, if the observation is sound, a succession process of the kind the order has undergone before, and the diagnostic grades claims about it at the level of configurations and centres unless the repertoire argument for the whole system is supplied. Succession is not smooth, and the intervals between a declining centre’s exhaustion and a successor’s consolidation have historically been the world order’s most dangerous decades, marked by rivalry, monetary disorder and war; the interregnum between British and American leadership is the standing example, and its dangers were systemic in effect without being terminal for the system. That combination, maximum danger with survival, is the hardest pattern for public argument to hold, addicted as it is to equating danger with doom and survival with safety. The three levels hold it without strain: a hegemonic transition is a level two process at the scale of the world configuration, capable of generating level one catastrophes of the largest amplitude, and compatible with the order’s continuation in a new geography. Analysts who can say that sentence, and resist improving it in either direction, have understood the article.

The succession pattern also explains a curiosity of the literature, the way each centre’s decline produces the richest crisis theory of its age. The theoretical schools this series surveys were disproportionately written in declining or threatened centres at moments of configuration stress, and their insights bear the mark of the location: acute about exhaustion, finance and empire, weaker about the successor configuration assembling elsewhere, whose analysts are usually too busy building it to theorise it. A reader using the diagnostic on the tradition itself will therefore grade even the classics with the territory question in hand, asking of each theory which centre’s experience it generalises. The question diminishes none of them. It restores to each its proper scale, which is the service the level test performs on every text it touches, beginning with the texts of the tradition that produced it.

For students and exam candidates

The three levels are, among other things, a ready-made essay architecture, and examiners reward the architecture because it converts a slogan question into an analytical one. Faced with a prompt asking whether capitalism is prone to terminal crisis, the strong answer opens by distinguishing the claim’s three forms, cyclical, configurational and systemic, and states the level test as its method in the first paragraph. The body then assigns the standard evidence to its levels: recurrent downturns to the first, the exhaustion of named settlements such as the postwar configuration to the second, and the breakdown arguments, Grossmann’s above all, to the third, with the counteracting influences and the founders’ own second thoughts entered against them. A paragraph on the counter-case, the survival record joined to the long improvement in global living standards, demonstrates the balance examiners look for, and a close that specifies what evidence would change the assessment earns the evaluation marks that verdicts alone never do. The commonest failure mode in weaker scripts is level smuggling: evidence introduced at one level and concluded from at another, usually a severe downturn doing duty for systemic failure. Naming the levels at the start, and auditing each paragraph against them before submission, is the single most reliable repair. Candidates should also handle the phrase late stage capitalism with visible care, noting its analytical ancestor in periodisation and its popular degeneration into mood, since the distinction is exactly the kind of precision such questions are set to test.

For teachers and seminar leaders

This material teaches best as an exercise in grading claims rather than as a doctrine to be transmitted, and the diagnostic table reproduces well as a seminar handout around which a term’s discussions can be organised. A productive opening session assigns each participant one famous crisis text, a pamphlet of the eighteen seventies, a breakdown argument from the interwar years, a stagflation-era essay, a financial crisis commentary, and requires a one-page grading: the level claimed, the level the evidence supports, the falsifier offered or missing, the territory assumed. The gradings, compared in discussion, teach the article’s content faster than any lecture, because students discover the level smuggling in their own first drafts before they are asked to find it in others. Later sessions can run the worked cases as problems: grade the depression decade using only evidence available by its midpoint, then again with the sequel known, and confront the class with the difference between the two gradings and what it implies about real-time analysis. Teachers should insist on the symmetry rule, that permanence claims from configuration peaks be graded with the same severity as breakdown claims from troughs, since students reliably find the rule uncomfortable in one direction and the discomfort is diagnostic of the seminar’s own moment. The aim throughout is calibration, a transferable skill: a student who can grade a crisis claim can grade a claim about technology, demography or any other field where levels of assertion are habitually confused. Close the unit by having the class draft the falsifier for its own consensus view, whatever it is; the exercise has never yet failed to improve the consensus.

For researchers, historians, and journalists

For working researchers the diagnostic’s value is as a filing system for a literature that badly needs one. Crisis scholarship accumulates without a shared scale, so that a monograph demonstrating a regime’s exhaustion and a monograph denying a system’s mortality can be shelved as contradictories when they are in fact answers to different questions, each possibly correct. Sorting the field by level before engaging it saves the wasted polemic that this series exists to reduce, and it clarifies what a new contribution must do: a level two argument owes component evidence, a level three argument owes the repertoire argument, and a referee can hold each to its proper debt. Journalists, who must assign levels under deadline with the whole declaration market shouting at them, are offered the five questions as a newsroom instrument: level, evidence, falsifier, timescale, territory. Put to any source, on the record, they convert a quote about doom or resilience into the beginning of a story about mechanisms, and they give readers the calibration information that straight quotation withholds. Historians will recognise the procedure as their own craft formalised, since the discipline’s best crisis writing has always separated the episode, the configuration and the order intuitively. The article’s request to all three professions is the same: keep the file, record the defeats, and let the record, rather than the moment’s loudest genre, set the burden of proof.

Frequently Asked Questions

Q: Is capitalism in crisis?

The question cannot be answered as posed, because the word crisis covers three different claims that the evidence bears on very differently. A crisis in capitalism, a downturn the order generates and absorbs, is a regular event that has occurred in most decades of the system’s history, so at that level the answer at almost any moment is that one has occurred within living memory or one is developing. A crisis of a regime of accumulation, the exhaustion of a whole configuration of production, wages, money and state, is a rarer and arguable condition that must be shown from the failure of named linkages. A crisis of capitalism as such, in which no reorganisation could restore accumulation, is the strongest claim and the one never yet demonstrated, despite two centuries of confident assertion. The honest answer is therefore the diagnostic itself: specify the level, and the question becomes answerable; refuse, and no evidence can settle it either way.

Q: What is the difference between a crisis in capitalism and a crisis of capitalism?

A crisis in capitalism is an event inside a functioning order: output falls, firms fail and credit contracts, and then the order’s repair mechanisms, liquidation, devaluation, cheapened inputs and public backstops, restore accumulation within the same institutional configuration. A crisis of capitalism in the intermediate sense is a crisis of one such configuration, the regime of accumulation whose production model, distributional settlement and policy framework have exhausted themselves together, so that repair no longer works and reorganisation is required. A crisis of capitalism in the full sense would be the failure of reorganisation itself, the demonstration that no rearrangement within the system’s defining relations, wage labour and accumulation for profit, could restart sustained growth. The differences are not verbal. Each level has its own evidence, its own precedents and its own test, and most confusion in the debate comes from gathering evidence at the first level and spending it at the third.

Q: What is secular stagnation?

Secular stagnation is the hypothesis that a mature economy can settle into a long condition of weak investment, slow growth and chronically low interest rates because the opportunities for profitable investment fall short of the savings seeking placement. The argument was first made by Alvin Hansen in the late nineteen thirties, who pointed to the closing of the frontier, slowing population growth and capital-saving technology, and it was revived in the twenty-first century by Lawrence Summers and others to explain recoveries that left output below its earlier path despite exceptionally cheap money. The Marxist tradition developed a parallel account in the theory of monopoly capital, in which giant firms generate a surplus they cannot absorb productively. In the three-level diagnostic, stagnation of either school is a condition of a configuration rather than an event within one, and it is therefore level two material: serious, long-lasting and open in principle to reorganisation, which distinguishes it from both a recession and a breakdown.

Q: Has capitalism survived predicted collapses before?

Yes, repeatedly, and the record is the starting point for any credible discussion of the question. Terminal crises were predicted in the early nineteenth century amid the dislocations of industrialisation, again by the generation that produced a formal breakdown law in 1929, again during the depression decade that followed, when the prediction seemed most nearly descriptive, again during the stagflation that ended the postwar settlement, and again after the most severe modern financial collapse. In each case the order responded not by continuing unchanged but by reorganising: new financial rules, new wage settlements, new state capacities, new technological bases. Survival therefore does not prove permanence, and the same record shows every configuration dying on schedule measured in decades. What the record does establish is a burden of proof. A claim that the repertoire of reorganisation is finally empty must explain why this exhaustion differs from all the previous ones that were announced with equal confidence and equal evidence of distress.

Q: What is a regime of accumulation?

A regime of accumulation is a configuration of institutions that, for a period measured in decades, makes capitalist growth self-sustaining by reconciling the system’s moving parts. The concept comes from the French regulation school, which analysed the postwar decades by identifying the configuration’s components: a production model, in that case mass production of standardised goods; a wage relation linking pay to productivity so that workers could buy what they produced; a monetary and financial framework; a set of state capacities for managing demand and welfare; and a mode of insertion into the world economy. The regime holds while the components complement one another, each making the others workable. It enters crisis when they cease to complement: the production model’s productivity frontier flattens, the wage bargain becomes a distributional battleground, and policy instruments contradict the problems they face. The concept matters because it names the middle level of crisis, exhaustion of a configuration, which implies reorganisation rather than either routine recovery or the end of the system.

Q: Does the productivity slowdown indicate a systemic crisis?

Not on its own, and the distance between what it shows and what it is often asked to show is a good illustration of the level test. Productivity growth across the advanced economies has run slower over the decades since the postwar expansion than during that expansion, with an uneven partial revival associated with computing. Slowing productivity growth is the classic signature of a production model maturing, and it has appeared at the exhaustion of each successive technological configuration, which makes it strong evidence at the level of a regime whose core engine is tiring. To raise it to the systemic level, a claimant must show the further proposition that no successor production model can restore the growth rate, a claim about the future of invention that no trend line can establish. The serious open dispute among specialists is whether the slowdown reflects one paradigm’s exhaustion or a deeper feature of mature service economies, and that dispute, not the slowdown itself, is where the level of the evidence will eventually be decided.

Q: Is climate change a limit to capital?

It is a limit of a different kind from the ones the crisis tradition historically theorised, and the diagnostic treats it in a standing annex rather than forcing it into the three levels. The levels grade the order’s relations with itself: profitability, realisation, the fit of institutions. Climate change confronts every possible configuration with the same physical facts, the atmosphere’s capacity to absorb the by-products of fossil-fuelled production without transformations that damage agriculture, coasts, infrastructure and health. No rearrangement of the wage bargain or the monetary framework alters that chemistry, although configurations differ greatly in how exposed their capital stock is and how rapidly they can substitute. The limit therefore interacts with each level, stranding assets, deepening downturns and supplying an investment frontier for energy transition, without being reducible to any of them. The tradition’s own engagement with nature ran mainly through raw materials and the metabolic relation between town and country, so the ecological argument is developed separately in this series rather than stretched over the crisis machinery.

Q: What would a terminal crisis of capitalism look like?

The diagnostic answers by specifying the evidence rather than the imagery, because the imagery, empty shelves, failed banks, mass unemployment, has all appeared in episodes that ended in reorganisation. A terminal crisis would be recognised by the failure of reorganisation itself, observed over time: successive, genuinely different configurations attempted, liberal and interventionist, financial and industrial, national and international in design, each restoring accumulation more briefly and feebly than the last; repair mechanisms, monetary, fiscal and institutional, losing effect across successive uses; productivity stagnation persisting across changes of regime that historically restarted it; and the plausible repertoire visibly emptying, so that informed observers could no longer name the next configuration’s components. Nothing in the record meets that description yet. The nearest approach, the depression decade, ended with the repertoire producing a settlement nobody had forecast. A terminal crisis is therefore a coherent concept with specified indicators, which is precisely why claims to have identified one can be graded, and why the grading so far has always returned them to a lower level.

Q: Why do Marxists keep predicting collapse?

Partly because the theory predicts crises and the step from crisis to collapse is rhetorically short, and partly because of the audiences the prediction serves. The founding texts supply tendencies toward breakdown alongside the counteracting influences, and readers under polemical pressure have repeatedly kept the tendency and dropped the countertendencies. A prediction of collapse also performs work: it promises adherents that history guarantees their side’s eventual vindication, it dramatises a movement’s literature, and it fits the eschatological shape that political hope often takes. Against the caricature, the record also contains major Marxist economists who explicitly denied any automatic breakdown, critics who dismantled breakdown laws from inside the tradition, and founders who recorded their own disappointed expectations and revised them. The predictive habit is therefore a feature of the tradition’s popular transmission more than of its strongest work. The three-level diagnostic is designed to break the habit’s mechanism, by requiring every collapse claim to specify the level at which reorganisation itself is alleged to have failed.

Q: Can capitalism end without being replaced?

Yes, and separating the two ideas, ending and replacement, removes a confusion that obstructs the whole debate. Historical modes of production have usually ended without a successor standing ready: the slave economies of antiquity decayed across centuries, feudal relations dissolved unevenly across longer spans still, and the orders that eventually followed were assembled retrospectively from elements contemporaries had classified as marginal. An ending, in the record, typically means a slow failure of reproduction, fragmentation and degeneration, not a ceremony of succession. Two consequences follow for the level test. A claimant arguing that the order’s reorganisation capacity is exhausted is not required to produce the successor society in blueprint before the claim can be discussed, since successors have never been visible in advance. Equally, the absence of a visible successor is not evidence that no ending is possible. The debatable question is confined to the evidence about reorganisation: attempted, failed or succeeding. Everything else is speculation about a future that history suggests will surprise all parties.

Q: How does the three-level diagnostic sort a crisis headline?

By putting four questions to the headline before its evidence is even examined. First, which level does the claim occupy: a downturn inside the order, the exhaustion of a configuration, or the failure of the order itself? Headlines rarely say, and the verb tenses usually smuggle the highest level into the phrasing. Second, what evidence is actually supplied, and at which level was it issued: a market fall, a bankruptcy or a bad quarter is level one evidence whatever the headline concludes from it. Third, what observation would the writer accept as showing the claim wrong: if none exists, the item is a mood rather than an argument. Fourth, over what timescale is the claim meant to be judged, since a level one claim is settled in quarters, a level two claim in decades and a level three claim across generations. A headline that survives the four questions with its evidence filed at its stated level has earned substantive engagement. The large majority do not survive the second.

Q: What separates a conjunctural shock from a systemic crisis in this framework?

The behaviour of the linkages after the shock. A conjunctural shock is an event, however violent, that the prevailing configuration absorbs through its ordinary repair mechanisms: losses are allocated, balance sheets are written down, policy instruments work as designed, and accumulation resumes within the same production model, wage relation and financial framework. A systemic crisis in the strict sense used here would be the condition in which those mechanisms themselves fail across attempted variants, so that no rearrangement within the order’s defining relations restores growth. Between the two sits the most common serious case, the shock that reveals a configuration’s exhaustion and opens a reorganisation. The separation is therefore not made by the shock’s size. A spectacular failure on sparse linkages can remain conjunctural, while a modest failure sitting on the channels a whole settlement depends on, credit, energy, a key wage bargain, can prove structural. The analyst grades the damage done to the arrangements, not the drama of the opening event.

Q: Why does the level test ask about duration and depth first?

Strictly, it asks about duration before depth, and the order matters because the two measurements carry opposite temptations. Depth, how far output, employment or asset prices fall, is the most available and the most misleading measurement, since the deepest contractions in the record have repeatedly closed within a few years and left their configurations intact, while shallow stagnations lasting a generation have ended them. Duration discriminates: distress that outlasts the historical recovery schedule, and downturns that stop resolving on time, indicate that the configuration’s repair mechanisms themselves are failing, which is level two information of the first importance. Asking duration first also protects the analyst from the declaration market, in which depth is quoted precisely because it inflames. The full shape of an episode has three dimensions, amplitude, breadth across sectors and regions, and persistence, and the level assignment should wait for all three. An assessment made on the first month’s amplitude is a reaction dressed as analysis.

Q: How should an analyst handle a crisis that stays at one level?

By resisting the professional temptation to promote it. A crisis that remains at one level, a downturn absorbed by the configuration, a sector destroyed without structural consequence, a stagnation that reorganises nothing, is a finding, and the analyst’s duty is to record it as one, with the same prominence given to more dramatic assignments. Three disciplines help. Keep the original assignment and its date in the file, so that later reclassification, if events force it, is visible as a revision rather than a quiet edit. Continue monitoring the linkage indicators rather than the headline aggregates, since a level that is stable in the aggregates can still be moving in the components. And report the negative result to whoever commissioned the analysis with undiminished confidence, because the alternative, inflating the level to justify the attention the episode received, is how the genre’s noise is manufactured. Most crises in the record stayed at the level at which they began. A literature that cannot say so plainly will misgrade the minority that did not.

Q: What would move a sectoral crisis to the system level in the diagnostic?

A chain of demonstrated transmissions, not a single dramatic failure. A sectoral crisis moves upward first to the level of the configuration when the sector’s fall disables machinery the whole settlement depends on: when its debts are held so widely through the financial system that its losses become the banks’ losses; when its output is an input for which other sectors cannot substitute, as with energy; when its employment is concentrated enough to break a regional or national wage settlement; or when rescuing it consumes the state’s repair capacity that other contingencies will need. Movement from the configuration level to the system level requires the further and much heavier showing described throughout this article: that the reorganisation the sectoral crisis forces itself fails, and fails in company with other attempted reorganisations, until the repertoire is demonstrably empty. Sectoral origin is therefore no bar to the highest grading, since several historic transitions began in one industry or one asset class, but each step upward must be earned by traced linkages and observed failures, never by the sector’s symbolic prominence.