What causes recessions?
Demand shocks? Supply shocks? Financial panics? The schools of thought still disagree on the basics.
The business cycle fact book
"In October 2009, the unemployment rate hit 10%. 15 million Americans were out of work. What just happened — and could we have stopped it?"
The 2008 crisis destroyed \$10 trillion in household wealth. Home prices fell 33% peak to trough. Lehman Brothers — a 158-year-old firm — vanished in a weekend. And the most powerful economic institutions on the planet were caught flat-footed.
Before you can argue about what causes recessions, you need to know what a recession actually looks like — not the textbook definition (two quarters of falling GDP), but the empirical signature that every theory must explain.
The business cycle. Economic output doesn't grow smoothly. It expands, peaks, contracts, hits a trough, and expands again. The NBER officially dates U.S. recessions going back to 1854. There have been 34 of them. No two are alike, but they share a family resemblance.
The stylized facts. Burns and Mitchell catalogued the regularities in 1946, and they've held up across seventy years of subsequent data:
- Investment is the most volatile component — 3–4 times more volatile than GDP. Firms slash capital spending first because investment is forward-looking and easily deferred.
- Consumption is smoother than GDP — households draw down savings or borrow to maintain living standards. This is consumption smoothing in action.
- Unemployment rises fast and falls slowly — it took 6 months for unemployment to jump from 5% to 10% in 2008. The recovery took six years.
- Cycles are irregular — expansions have lasted from 12 months to 128 months. There is no clock.
In a recession, the economy shrinks — but not evenly. Business investment collapses (firms stop building factories), while consumer spending dips only modestly (people still buy groceries). Unemployment spikes. These patterns repeat in every recession, from 1929 to 2020. The facts are agreed upon. What causes them is where economists start fighting.
These facts are the scorecard. Volatile investment, smooth consumption, countercyclical unemployment, irregular timing. Every theory in the next four stages will be judged against this pattern.
But the interpretation divides the field before anyone fires a shot. If recessions are deviations from a smooth trend, then something has gone wrong and needs fixing. This is the Keynesian view. But if the trend itself moves — if "potential output" shifts with productivity — then recessions might be movements of the trend, not departures from it. Same data, radically different implications for policy. That disagreement drives the next four stages.
"The expansion will end, as all expansions do. The question isn't whether — it's when, and whether policymakers will be ready."
— Larry Summers, The Washington Post, 2018
Is a recession always around the corner?
Expansions don't die of old age — they're murdered by policy mistakes, financial imbalances, or external shocks. But the longer an expansion runs, the more fragilities accumulate. The stylized facts tell you what recessions look like. They don't tell you when the next one hits.
Are recessions deviations or adjustments?
"Recessions are not simply smaller booms. The dynamics of contraction are qualitatively different from the dynamics of expansion. Unemployment rises like a rocket and falls like a feather."
— Neftci (1984), confirmed by McKay & Reis (2008)
This asymmetry is one of the most robust facts in macroeconomics. Recessions are sharp; recoveries are slow. Any theory that treats booms and busts as symmetric mirror images is missing something fundamental about how economies contract. The standard linear models (which we'll build in Stages 2–5) struggle with this asymmetry — it takes nonlinear dynamics, financial frictions, or search-and-matching models to capture it.
"GDP measures market production, not welfare. A recession that shifts activity into the informal sector, home production, or leisure may be less severe than the headline number suggests."
— The measurement critique (Becker, Nordhaus, others)
Fair point — but the 15 million people who lost their jobs in 2008 weren't enjoying extra leisure. The measurement debate matters at the margins, but during deep recessions, GDP captures something very real: millions of people want to work and can't find work. That's the phenomenon we need to explain.
Where this leaves us
The facts are agreed upon. The interpretation is where the war happens. Investment plunges. Consumption holds. Unemployment spikes. Every theory must match these patterns. The question is why — and that's where the schools split. In 1936, Keynes offered the first answer that stuck: people simply stop spending.
A quarter of the workforce is idle. Factories sit empty. Classical economists say: wait. Wages will fall, and the market will clear. One man looked at the breadlines and said they were wrong. His answer launched a revolution — and ninety years of argument.
The Keynesian answer
"The difficulty lies not so much in developing new ideas as in escaping from old ones."
— John Maynard Keynes, The General Theory of Employment, Interest, and Money, 1936
Keynes wrote this in 1936. The Great Depression was still raging. Classical economics said it couldn't happen — markets clear, wages adjust, full employment is the natural state. A quarter of American workers were unemployed. The theory said they didn't exist.
Keynes's answer was radical: recessions happen because people stop spending, and the economy can get stuck below full employment because prices and wages don't fall fast enough to clear markets. This was heresy. Say's Law — the classical bedrock — held that supply creates its own demand. Every act of production generates income that gets spent. A general shortfall of demand is impossible.
Keynes said: look outside your window.
The Keynesian cross. Start with a radical assumption: firms produce whatever is demanded. If demand drops from \$20 trillion to \$18 trillion, firms cut production to match — laying off workers, shuttering factories. Output is determined by demand, not supply capacity.
The multiplier amplifies the initial shock. A \$100 billion fall in investment doesn't reduce GDP by \$100 billion — it reduces it by \$100 billion times the multiplier. Laid-off workers cut spending, causing further layoffs, causing further spending cuts. The economy spirals downward.
With $MPC = 0.8$, the multiplier is 5. But this is the naive version — crowding out, taxes, and imports all leak spending out of the chain and shrink the effective multiplier to something closer to 1.0–1.5 in practice.
Imagine a factory town where the factory closes. Workers lose income and spend less at local shops. Shop owners earn less and cut their own spending. The initial hit cascades through the economy, making the total damage larger than the original shock. That's the multiplier in reverse — and it's why recessions can snowball from a single trigger into economy-wide downturns.
The AD-AS framework. The aggregate demand–aggregate supply model is the workhorse. AD slopes downward; short-run aggregate supply slopes upward because of sticky nominal wages. A recession is a leftward shift of AD — because investment collapses, confidence evaporates, or credit freezes. The key mechanism is price stickiness: if all prices adjusted instantly, the economy would snap back. It's because they don't that demand shortfalls translate into real output losses.
$$AD: \quad Y = C(Y - T) + I(r) + G + NX$$The prescription follows directly: if the problem is insufficient demand, boost demand. Government spending, tax cuts, monetary expansion — push AD right and restore output. This was the intellectual foundation for the New Deal, the 2009 stimulus, and the COVID fiscal response.
"The stimulus should have been at least \$1.2 trillion. The plan was nowhere near big enough to fill a gap of around \$2 trillion a year."
— Paul Krugman, January 2009
"Was the 2009 stimulus too small?"
Christina Romer told Obama it needed to be \$1.2 trillion. Congress passed \$787 billion. The sluggish recovery became the central exhibit in the biggest fiscal policy debate of the century — and the Keynesian demand story is the framework you need to evaluate it.
Is insufficient demand a real cause of recessions?
"In the long run, we are all dead. Economists set themselves too easy, too useless a task if in tempestuous seasons they can only tell us that when the storm is past, the ocean is flat again."
— John Maynard Keynes, A Tract on Monetary Reform, 1923
This is the Keynesian case in one sentence. Classical economics said: wages will fall, markets will clear, full employment will return. Keynes said: how long? Five years? Ten? The Great Depression lasted a decade under classical policies. "Eventually" is not a policy. The human cost of waiting for self-correction — a generation's lost earnings, skills, and hope — is the cost Keynes refused to accept.
"If demand shortfalls cause recessions, why would demand fall? Rational agents don't suddenly stop spending without reason. The demand decline is a symptom, not a cause."
— The classical and RBC counter-argument (Say, Lucas, Prescott)
This is the deepest objection to the Keynesian story: it describes recessions but doesn't explain what triggers them. "Animal spirits" — Keynes's term for waves of optimism and pessimism — was an honest admission that he couldn't fully explain the initial shock. The demand story tells you what happens after spending falls. It's weaker on why it falls in the first place. That gap will matter.
Where this leaves us
The Keynesian revolution gave us the first workable theory of recessions: demand shortfalls, amplified by the multiplier, sustained by sticky prices. The evidence from the Great Depression, 2008, and COVID is compelling — output fell far below any plausible measure of supply capacity, and millions were involuntarily unemployed. But the framework has gaps: it doesn't fully explain what triggers demand collapses, and it asserts price stickiness without explaining why prices are sticky. Those gaps opened the door for a devastating counterattack.
For three decades, Keynesians dominated economic policy. Then Milton Friedman looked at the same Great Depression and reached the opposite conclusion: the problem wasn't too little government spending. The problem was too little money.
The RBC revolution
"Inflation is always and everywhere a monetary phenomenon in the sense that it is and can be produced only by a more rapid increase in the quantity of money than in output."
— Milton Friedman, The Counter-Revolution in Monetary Theory, 1963
The monetarist counterpoint. Friedman blamed the Fed for the Great Depression — not lack of spending, but lack of money. The Fed allowed the money supply to contract by a third between 1929 and 1933. In Friedman's telling, the central bank didn't fail to act — it actively caused the worst economic catastrophe of the twentieth century.
Friedman and Schwartz's A Monetary History of the United States (1963) was the intellectual bomb that cracked the Keynesian consensus. Their argument was forensic: trace the money supply through every recession and expansion, and a pattern emerges. Monetary contractions precede downturns. Monetary expansions precede recoveries. The Great Depression wasn't a failure of demand — it was a failure of the Fed.
The expectations-augmented Phillips curve. Friedman's theoretical contribution was equally devastating. The original Phillips curve (1958) suggested policymakers could buy lower unemployment by accepting higher inflation — a permanent tradeoff. Friedman and Phelps demolished this:
Inflation $\pi_t$ deviates from expected inflation $\pi_t^e$ only when output $Y_t$ deviates from potential $Y^*$. Crucially: once expectations adjust, the tradeoff vanishes. You can't permanently lower unemployment by running inflation — people catch on, demand higher wages, and unemployment returns to its "natural rate." The only lasting effect is higher inflation.
Friedman's argument: if the government tries to keep unemployment low by running inflation, workers eventually figure it out. They demand higher wages to compensate. Employers raise prices. The economy ends up with the same unemployment but higher inflation. You can fool people once, but you can't fool them forever. The 1970s proved him right.
The RBC thesis. In 1982, Kydland and Prescott went further than Friedman. Their Real Business Cycle model proposed that recessions are the optimal response of rational agents to changes in productivity. No sticky prices. No coordination failures. No role for monetary or fiscal policy. Just technology shocks rippling through a perfectly competitive economy where every agent is doing exactly the right thing.
subject to the production function $Y_t = z_t K_t^\alpha N_t^{1-\alpha}$, where $z_t$ is a technology shock. When $z_t$ falls, the return to labor and capital both decline. Hours worked fall (intertemporal substitution of leisure), consumption dips mildly (consumption smoothing), and investment plunges (the forward-looking firm sees lower future returns). The model generates the Stage 1 stylized facts — all without a single market failure.
The RBC story: imagine you're a farmer and it rains for a month straight. You can't plant, the return on working your land is near zero. So you take time off, fix the barn, read a book. Is this a "failure"? Or is this the smart thing to do when conditions are bad? RBC says the whole economy works like this — when productivity dips, people rationally scale back until conditions improve. No intervention needed. No intervention wanted.
Why most economists reject the RBC substance. What are "negative technology shocks"? The economy literally gets worse at producing things? Technology occasionally stagnates, but it rarely goes backward. The model requires enormous intertemporal substitution of leisure, but microeconomic estimates of labor supply elasticity are far too small. And workers in recessions say they're involuntarily unemployed. Most damning: the Volcker disinflation of 1981–82 — a deliberate, announced tightening of monetary policy — caused unemployment to hit 10.8%. A purely real model can't explain why printing less money causes real people to lose real jobs.
"What if recessions aren't failures of the economy — what if they're the economy working correctly? What if people rationally choose to work less when productivity falls?"
— The Real Business Cycle thesis (Kydland & Prescott, 1982)
Are recessions actually efficient?
If RBC is right, government stimulus during recessions doesn't fix a problem — it creates one. It pushes people to work more than is optimal given the current state of technology. The policy implication is radical: do nothing. Let the recession run its course.
Are recessions optimal or pathological?
"The Great Depression was not a failure of capitalism. It was a failure of government — specifically, of the Federal Reserve System, which allowed the money supply to shrink by one-third."
— Milton Friedman & Anna Schwartz, A Monetary History of the United States, 1963
This reframing was revolutionary. Keynesians said the Depression proved markets fail. Friedman said it proved government fails — the very institution Keynesians trusted to fix the problem. His evidence was powerful: the Fed tightened monetary policy in 1928–29 to curb stock speculation, then allowed thousands of bank failures that shrank the money supply. Ben Bernanke, at Friedman's 90th birthday in 2002, said: "You're right, we did it. We're very sorry. We won't do it again." Six years later, as Fed chair, he made sure they didn't.
"Movements in employment reflect the willingness of workers to substitute leisure intertemporally in response to changes in the real wage. What we call 'recessions' are periods when rational agents choose to work less because the return to labor is temporarily low."
— Edward Prescott, Nobel Prize lecture, 2004
Prescott is making the strongest possible claim: unemployment in recessions is voluntary. Workers aren't constrained; they're optimizing. This is the logical consequence of the model, and Prescott had the intellectual honesty to say it out loud. Most economists find it absurd as a description of the Great Depression or 2008. But the model's quantitative performance — matching the volatility of key aggregates — forced the profession to take it seriously as a methodology, even while rejecting the conclusion.
Where this leaves us
Friedman won two arguments decisively. First: the Fed caused the Great Depression by allowing the money supply to collapse. This is now mainstream. Second: the long-run Phillips curve is vertical — you can't permanently buy lower unemployment with inflation. RBC's lasting contribution is twofold: the insistence on microfoundations, and the DSGE computational toolkit. Its substantive claim — that recessions are efficient — is largely rejected. But Kydland and Prescott won the methodology war so completely that even their fiercest critics now use their framework. The next stage shows the result: a synthesis that keeps the RBC architecture but puts Keynesian content back inside.
By the early 2000s, the wars seemed over. New Keynesians had taken the RBC framework, added sticky prices and monopolistic competition, and built a model that central banks could actually use. The "New Keynesian synthesis" was the consensus. A leading macroeconomist declared the field's problems essentially solved — one month before the worst financial crisis in eighty years.
The New Keynesian synthesis
"The state of macro is good."
— Olivier Blanchard, NBER Working Paper 14259, August 2008
Published August 2008. Lehman Brothers collapsed one month later. The worst financial crisis since the Great Depression was about to reveal everything the "good" state of macro had missed.
The New Keynesian DSGE model — the synthesis of Stages 2–3 — was elegant. Take the RBC skeleton (microfounded, dynamic, stochastic). Add two ingredients: monopolistic competition (firms have pricing power) and sticky prices (they don't adjust every period). Now demand shocks matter, monetary policy has real effects, and the economy can be stabilized by a well-designed central bank.
Calvo pricing. Each period, a random fraction $1 - \theta$ of firms get to reset their price. The rest keep last period's price. With $\theta = 0.75$, the average firm resets annually. This micro-level stickiness creates macro-level consequences: when demand falls, firms that can't cut prices reduce output instead.
The three-equation model. The canonical NK model fits on one page:
- Dynamic IS (demand): $\tilde{y}_t = \mathbb{E}_t[\tilde{y}_{t+1}] - \sigma^{-1}(i_t - \mathbb{E}_t[\pi_{t+1}] - r_t^n)$
- NK Phillips Curve (supply): $\pi_t = \beta \mathbb{E}_t[\pi_{t+1}] + \kappa \tilde{y}_t$
- Taylor Rule (policy): $i_t = r_t^n + \phi_\pi \pi_t + \phi_y \tilde{y}_t$
Together these determine output, inflation, and the interest rate. Set $\theta = 0$ (perfectly flexible prices) and it collapses to RBC. As $\theta$ rises, demand shocks become more powerful. The degree of price stickiness is what separates the two paradigms — not philosophy, but an empirical parameter.
The NK model says: the economy works like RBC most of the time, but prices don't adjust smoothly. When demand falls, firms that haven't updated their prices yet are stuck selling at the old price and cut production instead. The central bank can help by cutting interest rates, which stimulates spending and closes the gap. Think of the central bank as a thermostat for the economy — it works beautifully as long as the temperature doesn't drop below what the heating system can handle.
What 2008 exposed. The baseline NK model had no banking sector. No leverage. No contagion. No credit freezes. When Lehman collapsed and credit markets froze, the model that was supposed to explain recessions couldn't explain the recession that was actually happening. The thermostat hit zero — the zero lower bound — and the heating system broke.
Post-crisis extensions. The profession scrambled. Financial frictions entered through the financial accelerator (Bernanke, Gertler, Gilchrist). Heterogeneous agents replaced the representative consumer (HANK models showed that distributional effects change aggregate dynamics qualitatively, not just quantitatively). The zero lower bound became a first-class feature rather than a footnote.
"When the zero bound binds, the government spending multiplier can be 1.5 or above. The key mechanism is the elimination of monetary offset."
— Christiano, Eichenbaum & Rebelo, 2011
"Was the 2009 stimulus too small?" (revisited)
Stage 2 introduced this question through the Keynesian lens. Now you have the full NK framework. At the zero lower bound, crowding out disappears and monetary offset is impossible — the multiplier roughly doubles. The 2009 stimulus question looks different with this machinery.
Is the NK synthesis the final word?
"The 2008 crisis was a devastating blow to the profession's self-confidence, and rightly so. We had convinced ourselves that the big problems were solved. We were wrong."
— Olivier Blanchard, "Rethinking Macroeconomic Policy", IMF, 2010
To his credit, Blanchard wrote the mea culpa as well as the premature victory lap. His 2010 paper catalogs what the profession missed: financial frictions, nonlinear dynamics, the limits of conventional monetary policy. It launched a decade of model-building that incorporated the lessons of the crisis — but many of those lessons should have been incorporated before 2008, not after.
"When the zero bound binds, the government spending multiplier can be 1.5 or above, and the welfare multiplier can be even larger. The key mechanism is the elimination of monetary offset."
— Lawrence Christiano, Martin Eichenbaum & Charles Rebelo, Journal of Political Economy, 2011
This is the canonical ZLB result. The NK model shows that when rates hit zero, two forces that normally suppress the fiscal multiplier are simultaneously disabled: government borrowing doesn't push up rates (no crowding out) and the central bank won't fight the expansion (no monetary offset). This rescued fiscal policy from the theoretical irrelevance that Ricardian equivalence and monetary offset had imposed on it. But it's a conditional rescue — the multiplier is large only when the ZLB binds.
Where this leaves us
The NK synthesis is the best available framework for understanding recessions. Both demand and supply shocks matter. Monetary policy has real effects. Price stickiness creates output gaps that represent genuine welfare losses. But the 2008 crisis exposed real blind spots: financial instability, distributional effects, and the institutional fragility that makes some economies resilient and others brittle. The framework wasn't wrong — it was dangerously incomplete.
The NK model could explain demand shocks, supply shocks, and policy errors. But 2008 showed the most devastating recessions arise from something none of these models captured: cascading financial failure. Credit booms, leverage cycles, bank runs, and contagion — the financial friction frontier is where the field is now.
The financial friction frontier
"Credit booms systematically precede financial busts. The pattern is so reliable it should be at the center of any theory of recessions. Yet neither RBC nor mainstream NK models explain why financial excess builds up or why the resulting crash is so severe."
— Summary of the post-crisis critique (Romer, Stiglitz, Minsky tradition)
The deepest critique of everything built in Stages 1–4. Standard models can incorporate financial shocks after the fact, but they don't explain why credit booms happen, why banks make bad loans, or why the system builds up fragility until it shatters. The Minsky hypothesis — stability breeds instability — describes the pattern. Formalizing it remains an open challenge.
The financial accelerator. Bernanke, Gertler, and Gilchrist (1999) showed how small shocks get amplified through the financial system. When asset prices fall, firms' collateral loses value, their borrowing capacity shrinks, investment drops, which reduces output and asset prices further. A modest downturn becomes a severe one because financial markets amplify rather than absorb shocks.
HANK models. Kaplan, Molin, and Violante (2018) replaced the representative consumer with heterogeneous agents. Some households are wealthy, some are hand-to-mouth, some are in between. This matters enormously: in a representative-agent model, a recession hurts "the economy." In a HANK model, it devastates the bottom quartile while barely touching the top. Distributional effects change aggregate dynamics qualitatively — a $1 transfer to a credit-constrained household generates more spending than $1 to a wealthy one.
The Minsky cycle. Stability breeds instability. During long expansions, risk premia compress. Banks make riskier loans. Leverage rises. Regulators relax. Everyone believes the good times will continue — until they don't. The transition from "hedge finance" (income covers debt payments) to "speculative finance" (income covers interest but not principal) to "Ponzi finance" (income covers neither; relies on asset appreciation) is the Minsky cycle. The 2008 crisis followed this script almost exactly.
What this means for recession theory. The post-crisis research program hasn't overthrown the NK synthesis — it's extended it. Financial frictions, heterogeneous agents, and occasionally binding constraints (like the ZLB) are being bolted onto the DSGE chassis. The question is no longer "demand vs. supply" but "which combination of shocks and amplification mechanisms dominates this particular episode?"
"The crisis was not a failure of the New Keynesian model. It was a failure of the economists who used it to assume that financial markets could be safely ignored."
— Michael Woodford, "Financial Intermediation and Macroeconomic Analysis", Journal of Economic Perspectives, 2010
Did the NK consensus fail in 2008?
The baseline NK model has no banking sector, no leverage, no financial panics. The worst financial crisis in 80 years hit the economy through channels the workhorse model didn't contain. The model survived — but only by bolting on the financial frictions it had previously assumed away.
Is the NK synthesis the final word?
"The basic three-equation model captures the essential mechanisms — demand determination of output in the short run, a meaningful role for monetary policy, and eventual price adjustment. Extensions to handle financial frictions, heterogeneity, and occasionally binding constraints are natural developments within the framework, not refutations of it."
— Michael Woodford
Woodford is right that no alternative paradigm offers comparable rigor and empirical performance. Post-Keynesian and Austrian approaches have important insights but lack the formal discipline that makes models testable and policy-relevant. The NK synthesis is imperfect but improvable — and the post-crisis extensions demonstrate its adaptability.
"Stability is destabilizing. Long periods of tranquility encourage increasingly risky financial structures — hedge, speculative, and Ponzi finance. The system generates its own crises from within."
— Hyman Minsky (posthumously vindicated by 2008)
Minsky's insight — that the seeds of crisis are sown during the boom — is absent from the NK framework, where financial instability must be imposed as an exogenous shock. The 2008 crisis wasn't caused by an asteroid hitting the economy; it was caused by the economy's own internal dynamics. A theory of recessions that can't explain why booms produce busts is missing something fundamental. Agent-based models and network approaches may eventually capture these endogenous dynamics, but they're still at the frontier, not the mainstream.
The verdict
What causes recessions? The honest answer is: it depends on the recession. Demand shocks cause most mild downturns — confidence falls, spending drops, and sticky prices translate that into lost output and jobs. Supply shocks (oil crises, pandemics) hit the economy's productive capacity directly. Financial crises amplify both types of shocks through leverage, panic, and credit contraction. The NK synthesis is the best available framework because it accommodates all three channels, but it remains weakest on the financial channel — the one that produced the most devastating recession in living memory. No single model explains all recessions, and the profession should be more honest about this. The question "what causes recessions?" has a plural answer, and pretending otherwise is how we get caught off guard.
Where this leaves us
We started with 15 million Americans out of work and a question: what just happened? Five stages later, here's what you know:
- The facts constrain every theory (Stage 1). Investment plunges. Consumption holds. Unemployment spikes fast and recovers slow. Cycles are irregular. These patterns are non-negotiable — any theory that can't match them is wrong.
- Keynes: demand is the problem (Stage 2). When people stop spending, the multiplier amplifies the shock, sticky prices prevent self-correction, and the economy gets stuck below potential. The prescription: boost demand. It works — but it doesn't explain what triggers the collapse.
- Friedman and the RBC revolution (Stage 3). The Fed caused the Great Depression by allowing the money supply to collapse. You can't permanently buy lower unemployment with inflation. And Kydland & Prescott's radical claim — recessions are efficient — was rejected on substance but transformative in methodology. The DSGE framework they built became the standard tool, even for their critics.
- The NK synthesis: all of the above, and still not enough (Stage 4). Sticky prices inside the RBC framework. Demand and supply both matter. Central banks can stabilize — until the ZLB binds. Financial crises were the blind spot that 2008 exposed.
- The financial friction frontier (Stage 5). Credit booms precede busts. The financial accelerator amplifies small shocks into large crises. HANK models show recessions devastate the bottom quartile. The Minsky hypothesis — stability breeds instability — is the insight the mainstream still hasn't fully absorbed.
The profession's honest admission is that the relative importance of different causes varies by episode. The Great Depression was a demand catastrophe amplified by monetary policy failure. The 1970s stagflation was a supply shock that exposed the limits of demand management. The 2008 crisis was a financial implosion that standard models barely captured. COVID was sui generis. There is no single answer to "what causes recessions?" because recessions are not a single phenomenon — they are a family of pathologies with overlapping symptoms and distinct etiologies.
The next time someone confidently tells you "recessions happen because..." and offers a single cause, you have five frameworks to evaluate the claim. The answer that matters is: which mechanisms dominate this recession, right now? That's not a cop-out. That's what it means to actually understand the question.