Was the Great Depression preventable?
In 2002, a sitting Federal Reserve governor stood up at Milton Friedman’s ninetieth birthday and formally apologized for the worst economic disaster in American history. “You’re right, we did it.” Six years later that same governor was running the Fed during another crash and got to test the apology. The question of whether 1929 could have been stopped is not a historical curiosity. It is the operating manual every central bank now reads from.
Voir comme graphe de débat“We did it. We’re very sorry.”
“Let me end my talk by abusing slightly my status as an official representative of the Federal Reserve. I would like to say to Milton and Anna: Regarding the Great Depression. You’re right, we did it. We’re very sorry. But thanks to you, we won’t do it again.”
— Ben Bernanke, then Governor of the Federal Reserve, at a conference honoring Milton Friedman’s 90th birthday, University of Chicago, November 8, 2002
A central banker, on the record, in front of the author whose book had spent forty years arguing that the Fed caused the Great Depression, accepting the charge in three sentences. This is not how central banks talk. By 2002, it was how this one could.
What does “preventable” mean for an event of this size? US industrial production fell roughly 47 percent. Unemployment hit 25 percent in 1933. The money stock contracted by a third. A counterfactual that erases all of that is not what anyone is claiming. The serious claim is narrower: under a different policy response — one specific lever pulled differently — the 1929 recession would have been a 1929 recession, painful but recoverable, not a four-year cascade. The question is which lever.
By 2002 the mainstream answer had a name and an author. The Friedman-Schwartz thesis: the Fed could have flooded the banking system with liquidity, prevented the cascading bank failures of 1930–33, and kept the money stock from collapsing. That this was the mainstream answer is itself the historically interesting fact. In the 1950s and 1960s the standard teaching was almost the opposite — a demand collapse no monetary tool could have reached. Something changed. History of Economic Thought Ch.10 (The Counter-Revolution) covers how monetarism displaced primitive Keynesianism on the macro questions; this walkthrough is downstream of that shift.
Bernanke could apologize on behalf of the Fed in 2002 because the discipline had reached something close to consensus on the proximate cause. Friedman and Schwartz’s A Monetary History of the United States (1963) had spent four decades being absorbed — first contested, then accepted, then taught as the default by the 1990s. The standard graduate macro class in 2000 taught Friedman-Schwartz on the Depression the way it taught Solow on growth. Dissenting positions were still in the room, but as positions to be aware of, not as the answer. A sitting Fed governor — an institution famously allergic to self-criticism — could deliver the apology because the apology was no longer controversial.
So the answer to “was it preventable?” circa 2002 was — in the room where the apology was given, with the author it was given to — basically yes. A different Fed response in 1930–33 would have produced a different outcome. That answer was the pre-2008 consensus. The interesting question is where it came from and what it still leaves out, because the consensus is older than it looks and shakier than its delivery suggests.
To see how the Fed-could-have-stopped-it answer became the default, you have to read the book Bernanke was apologizing to. It is one of the most quietly successful counter-revolutions in twentieth-century economics, and it runs through a single chart.
The Great Contraction
“From the cyclical peak in August 1929 to a cyclical trough in March 1933, the stock of money fell by over a third… The contraction is in fact a tragic testimonial to the importance of monetary forces.”
— Milton Friedman and Anna Schwartz, A Monetary History of the United States, 1867–1960 (Princeton, 1963), Chapter 7, “The Great Contraction”
A third of the US money stock vanished between the 1929 peak and the 1933 trough. Not because Congress voted it away, not because households chose to hold less of it, but because the banking system collapsed and the Federal Reserve did not stop the collapse. The chart is the argument: this didn’t have to happen.
The mechanics are not subtle. A fractional-reserve banking system has a money stock several times larger than the central bank’s monetary base, the multiplier coming from banks lending a fraction of every deposit. When depositors panic, banks that cannot meet withdrawals fail; each failure destroys deposits, which were money. The central bank’s job here is mechanical: lend freely against good collateral, flood reserves so the multiplier doesn’t collapse. This is the lender-of-last-resort function, formalized by Bagehot in 1873 and known to every central banker since. Economics Ch.16 (Monetary and Fiscal Theory) works through the money-multiplier algebra and the demand for base money; the lender-of-last-resort doctrine itself was written into central banking by the Bank of England a century before the Fed inherited it.
Friedman and Schwartz’s argument: between 1930 and 1933, through four distinct banking panics — Caldwell in November 1930, the Bank of United States that December, the wave following Britain’s exit from gold in autumn 1931, the cascade into the March 1933 bank holiday — the Federal Reserve did not perform its core function. Economic History Ch.12 (Interwar Monetary Collapse) walks through each panic, and sets the monetarist reading beside the three rival explanations the historical record has to adjudicate between.
The mechanism is clean. Let $M$ be the money stock, $B$ the monetary base, and $m$ the multiplier: $M = m B$, with $m = (1+c)/(c+r)$ where $c$ is the public’s currency-deposit ratio and $r$ is banks’ reserve-deposit ratio. In a panic, both $c$ and $r$ rise — the public hoards cash, banks hoard reserves — so $m$ falls sharply. To hold $M$ constant, the central bank must raise $B$ enough to offset the fall in $m$. Between 1929 and 1933 the Fed in fact let $B$ rise only modestly while $m$ collapsed; the product $M$ fell roughly 33 percent. Friedman-Schwartz’s claim is that a Fed expansion of $B$ sufficient to stabilize $M$ was within the institutional means available; the failure was one of policy, not capacity.
Imagine the money supply as water in a bathtub. The Fed controls the faucet. Banks have a hole in the bottom of the tub: when depositors panic, the hole gets bigger. The Fed’s job in a panic is to turn the faucet up faster than the hole drains. Between 1930 and 1933 the hole opened wider four times. The Fed nudged the faucet. A third of the water drained out.
The Friedman-Schwartz case at its strongest. The Federal Reserve was created in 1913 specifically to prevent the kind of liquidity panic the United States had suffered repeatedly in the late nineteenth century — 1873, 1893, 1907. The Aldrich Commission wrote the lender-of-last-resort function directly into the Fed’s charter. The tools, the mandate, the precedent from Bagehot — all available. During the four panics of 1930–33 the Fed sterilized gold inflows, kept the discount rate elevated for months, and allowed thousands of country banks to fail without intervention. Friedman and Schwartz’s reconstruction: the death of Benjamin Strong, the New York Fed governor who had effectively run policy through the 1920s, left an institutional vacuum. The remaining Board had no dominant figure and a pre-Keynesian instinct that monetary tightness was a moral correction. The 1930 New York Fed wanted to act and was overruled. The 1932 open-market operations that finally expanded the base were too small, too late. The Fed’s own minutes read like a manual of what not to do.
Friedman-Schwartz are right on the narrow question: the Fed had the lender-of-last-resort tool, didn’t use it, and the money stock fell by a third because the Fed allowed it to. A larger Bagehot intervention during the 1930 and 1931 panics would have prevented some material fraction of the bank failures. That much survives every subsequent revision. The harder question is whether the Fed could have done what Friedman-Schwartz say it should have done, given the international monetary order it actually faced.
Because the Fed in 1930 was not a free agent. It was a participant in a global system whose rules had been rebuilt after the First World War, on terms that constrained every central bank in the system. To understand what the Fed could and couldn’t do, you have to look outside the United States.
Golden fetters
“The gold standard, far from being synonymous with stability, was the principal threat to financial stability and economic prosperity between the wars. Insofar as the gold standard provided a mechanism for transmitting destabilizing impulses from one country to the next, it provided that transmission mechanism for the Great Depression itself.”
— Barry Eichengreen, Golden Fetters: The Gold Standard and the Great Depression, 1919–1939 (Oxford University Press, 1992), p. 4
Friedman and Schwartz wrote about the Depression as an American story. Eichengreen, three decades later, wrote it as an international one. The pivot changes what the Fed could plausibly have done.
The interwar gold standard, reconstructed at Genoa in 1922 and joined by Britain in 1925 and France in 1928, locked participating central banks into a hard constraint. A country on gold had to convert its currency into gold at a fixed parity on demand, tethering the monetary base to gold reserves. Run a balance-of-payments deficit, lose gold, and the central bank was obliged to raise rates to defend the parity. The system had no symmetric adjustment mechanism: surplus countries (the US and France through the late 1920s) could sterilize gold inflows; deficit countries had to deflate.
This is the open-economy trilemma in its starkest form — any two of fixed exchange rates, free capital movement, and independent monetary policy, but not all three. Economics Ch.17 (Open Economy Macroeconomics) covers the modern derivation. The interwar gold standard chose the first two; monetary policy — the lever Friedman-Schwartz wanted the Fed to pull — was the lever not available. A Fed expansion large enough to backstop the US banking system in 1931 would have triggered a gold outflow that forced the Fed to reverse course or break parity.
Formally: under the gold standard, the central bank’s asset side is constrained by $B \le \phi G$, where $B$ is the monetary base, $G$ is gold reserves, and $\phi$ is the statutory backing ratio. A discretionary expansion of $B$ to offset a collapse in $m$ runs into the constraint when $G$ falls. Eichengreen-Sachs (1985) test the implication directly: regressing recovery dates on gold-standard exit dates across ten European countries, the relationship is sharp. The earliest exiters (Britain, Sweden, Denmark in 1931) recovered fastest; the Gold Bloc (France, Belgium, Netherlands, Switzerland) staying in until 1935–36 sustained the deepest contraction. The cross-country variation does what within-country counterfactuals cannot: it identifies the constraint causally.
Picture every central bank in 1930 connected by a length of chain, with one end of the chain anchored to a vault of gold in each capital. Each chain has the same length. If any country’s central bank tries to print money to fight a depression, the chain pulls taut and gold starts pouring out of the vault. The country either gives up and stops printing — or it cuts the chain. Britain cut the chain in September 1931 and recovered. France held on until 1936 and didn’t.
Eichengreen’s case at strength: the Friedman-Schwartz indictment misallocates the responsibility. The Fed of 1931 could not have done what Friedman-Schwartz wanted without first breaking gold — an act the entire international financial architecture, central bank cooperation agreements, and the cultural commitment of every senior central banker was constructed to prevent. A large open-market expansion in 1931 was, by the doctrine of the time, monetary surrender. The constraint was real and binding, not a personnel issue.
The cross-country evidence is hardest to argue with. Take ten countries in 1930–36, order them by gold-exit date, and the order almost monotonically tells you when they recovered. Britain leaves in September 1931, bottoms out in 1932; Sweden leaves the same month, recovers fastest in Europe; the United States effectively leaves in March–April 1933 and starts climbing immediately. The Gold Bloc — France, Belgium, Netherlands, Switzerland, Poland — rides the depression down into 1935 and 1936. Economic History Ch.11 (Gold Standard Era) sets up the prewar regime and the lender-of-last-resort doctrine the Bank of England wrote; Ch.12 (Interwar Monetary Collapse) carries the exits.
By the 1990s the mainstream had absorbed this. Bernanke himself: his Princeton-era research on the international transmission of the Depression is explicitly in Eichengreen’s tradition, not Friedman’s. The 2002 apology was delivered by someone who already accepted the constraint, not by a pure monetarist.
Eichengreen is right that the gold standard was the binding constraint, and the modern consensus has absorbed him. The post-Eichengreen verdict: the Fed could have prevented the worst of the bank failures only by breaking gold earlier, and staying on gold while the banking system collapsed was the policy error that compounded the original one. It was preventable, but prevention required institutional courage of a kind the international monetary order of 1929 was specifically built to discourage.
All of which still treats monetary policy as the relevant lever. A different tradition argues that monetary policy — on or off gold, with or without LOLR — was simply not the level at which 1929 had to be fought. To see that argument at strength, you have to read Keynes writing in real time, and then you have to read someone who thought the whole framing was wrong from the beginning.
Magneto trouble, or no trouble at all?
“We have magneto trouble… We have involved ourselves in a colossal muddle, having blundered in the control of a delicate machine, the working of which we do not understand. The result is that our possibilities of wealth may run to waste for a time — perhaps for a long time.”
— John Maynard Keynes, “The Great Slump of 1930,” The Nation and Athenaeum, December 20–27, 1930
December 1930. Six years before The General Theory, Keynes is already saying the existing macroeconomic vocabulary cannot describe what is happening. The mechanics’ usual monetary tools, the ones Friedman-Schwartz would later focus on, will not reach the broken part.
Keynes’s frame, made explicit in 1936 and visible in this 1930 essay, is demand-side. Investment and consumption collapse together; the real interest rate the economy needs to clear at is deeply negative; nominal rates cannot go below zero. Monetary policy cannot produce the real-rate stance the economy requires. What is needed is direct demand support — government spending, sustained until private demand recovers. The zero-lower-bound machinery this argument depends on — why a liquidity trap breaks the usual monetary transmission — is worked formally in Economics Ch.15 (New Keynesian Economics); History of Economic Thought Ch.8 (The Keynesian Revolution) covers how The General Theory built the frame around exactly this case.
The Austrian frame, sharpened in Rothbard’s America’s Great Depression (1963, same year as Friedman-Schwartz, opposite verdict), reads the disaster as the boom, not the contraction. Fed easy money throughout the 1920s — on this account driven by Strong’s desire to help Britain sustain the pound at an overvalued parity — generated systematic malinvestment, capital flowing into long-duration sectors not justified by underlying saving. History of Economic Thought Ch.6 (The Austrian Tradition) covers the Mises-Hayek-Rothbard lineage and the capital-structure theory the boom-causation indictment rests on.
Take Keynes at strength. By 1931 the short rate was effectively zero. Banks sat on excess reserves they did not lend; households hoarded cash; firms looking at collapsed demand had no incentive to borrow even at zero. The Friedman-Schwartz prescription — flood the system with reserves — produces banks with even more excess reserves; the transmission channel from monetary base to spending is broken. Krugman’s 2009 framing was that 2008 supplied the natural experiment: the Great Depression historian running the Fed, doing exactly what Friedman-Schwartz prescribed, still required massive fiscal expansion to lift the economy off the floor. Christina Romer’s work on 1933–37 adds historical evidence: monetary devaluation did most of the early lift, but the 1937–38 recession — following premature fiscal contraction — shows demand was doing real work.
Now the Austrian case at strength, louder in public discourse than its academic standing suggests. The argument is not “let banks fail.” The argument is that “could the Fed have prevented it?” is the wrong question — the Fed had already caused it. Throughout the 1920s the Fed under Strong kept US short rates artificially low, partly to help Britain maintain the pound, partly to sustain the New York equity boom, driving credit expansion not warranted by underlying saving. When recognition came in 1929, the malinvestments had to liquidate; every subsequent intervention — Hoover’s wage-floor exhortations, the 1932 tax increase, Smoot-Hawley, NIRA price-fixing, the RFC — prevented the relative-price adjustment needed to reallocate capital out of the bubble sectors. On the modern Austrian reading (Ron Paul’s End the Fed, the Bitcoin-standard literature), the 2008 balance-sheet expansion and the 2021–23 inflation are the postponed consequence of refusing to let 1929 burn itself out.
“Federal Reserve inflationary policy in the 1920s and Herbert Hoover’s interventionist response to the crash, rather than laissez-faire capitalism or private speculation, created and prolonged the depression.”
— Murray Rothbard, America’s Great Depression, 1963
The boom was the disaster
The Austrian charge isn’t that the Fed did too little in 1931. It’s that the Fed did too much in 1925. Modern credit-cycle research (Borio, the BIS) has absorbed the boom-pattern empirics without absorbing the prescribed remedy. The boom-causation insight survives. The let-it-burn cure does not.
Both contain a piece of the truth. Keynes is right: 2008 showed monetary expansion at the floor has weaker traction than the pure-monetary reading implied, and fiscal did real work in 1933–37. The claim that monetary policy alone could have prevented the Depression overstates what monetary policy can do at the zero bound.
The Austrians are partially right on a narrower point than their public advocates claim. The boom-causation claim — that 1920s Fed policy generated credit-driven asset-price excess bound to correct — is now reflected in serious work outside the tradition: Claudio Borio and the BIS have built two decades showing financial-cycle credit expansions reliably precede severe recessions across countries and monetary regimes. The Austrian prescription does not survive. Countries that intervened earliest (Sweden, Britain, the United States) recovered first; the Gold Bloc that let deflation run bled longer. The human cost of the four-year contraction — 25 percent unemployment, the political wreckage that fed the rise of fascism — is not a price any defensible framework asks a society to pay for a theory of capital-structure reallocation. Booms cause busts is absorbed; busts must be unimpeded is not.
All four positions — Friedman-Schwartz on monetary contraction, Eichengreen on the gold-standard constraint, Keynes on demand and the zero bound, the Austrians on credit-driven boom dynamics — turn out to be present in the modern synthesis. The synthesis is not a compromise; it is a verdict, tested by a natural experiment in 2008.
“We know what not to do this time”
“I came to the Federal Reserve having spent a good part of my professional career studying the Great Depression… The Federal Reserve’s response to the recent financial crisis was based on lessons learned about what went wrong in the 1930s.”
— Ben Bernanke, reflecting on the 2008–09 response, paraphrased across The Courage to Act (2015) and Brookings interviews
Six years after the apology, the man who delivered it was chairing the Federal Reserve through a banking crisis that on most indicators looked like 1929 starting over. The 2008 response is the test of whether the 1929 verdict was right.
The modern synthesis integrates all four mechanisms. Friedman-Schwartz: Fed failure to perform LOLR allowed the money stock to collapse a third. Eichengreen: the gold standard bound monetary policy, and central banks took the constraint as binding even when the cost was banking collapse. Keynes: at the zero lower bound, monetary policy alone would have been insufficient — fiscal support was needed. Austrian credit cycle: a decade of overaccommodative 1920s policy under the gold standard’s skewed adjustment built up the credit excess 1929 was the recognition of.
In 2008 the Bernanke Fed applied this synthesis. LOLR aggressively. Not constrained by gold (severed in 1971). Coordinated with a Treasury and Congress that delivered, however imperfectly, fiscal expansion. Subsequent macroprudential reform — Dodd-Frank, stress tests, elevated capital and liquidity requirements — addressed the credit-cycle accumulation phase the Austrians had been right about in pattern. Economic History Ch.19 (The 2008 Crisis and After) walks through 2008–10.
A constituency still reads this whole synthesis as the wrong lesson. The gold-standard revivalist tradition — Ron Paul’s congressional advocacy through the 2008 and 2012 primaries, the Bitcoin-as-digital-gold movement traceable through Saifedean Ammous — argues that the post-1971 fiat dollar enabled exactly the credit-driven boom dynamic that produced 1929 and 2008, and that the 2008 response stored up 2021–23 inflation as a delayed consequence.
Steelman the case: a strict gold standard or hard-money successor would impose discipline on central banks that discretionary fiat regimes cannot. Strong’s 1920s easy money, the 1990s Greenspan put, the 2000s housing accommodation, the 2010s low rates, the pandemic expansion — a single pattern of overaccommodation, asset-price excess, and bailout at the expense of holders of money. A hard-money rule would break the cycle. The argument has internal consistency and a historical record (lower prewar long-run inflation, fewer asset-price crises under binding rules) to point at.
Now read it against the chapter just walked through. The interwar gold standard — the actual instance of a hard-money rule binding central banks during a major shock — is the architecture that turned 1929 into the four-year cascade. The discipline gold imposed was the discipline of refusing to act as lender of last resort during banking panics. The 2008 response the revivalists object to is what prevented 2008 from becoming 1932. Hard money would discipline the central bank; what the case does not address is that the discipline is the disaster, when the disaster is a banking panic.
So: was the Great Depression preventable? The walkthrough’s position, stated plainly.
It was preventable in the sense that any one of three levers, pulled differently, would have made it materially shallower. A Fed willing to perform Bagehot during the 1930–33 panics, accepting that this required breaking gold. A coordinated international abandonment of the interwar parities in 1930 or early 1931, before the cascades hardened. A fiscal expansion of meaningful size while political conditions still permitted one. Any of these — not all three — would have produced a different outcome. The 1929–33 cascade was the joint product of failures on each lever, not the unavoidable consequence of a single broken machine.
It was not preventable in the sense that some downturn from 1929 was near-inevitable given 1928’s credit-cycle conditions. A decade of overaccommodative policy channeled into an asset-price boom no central bank was prepared to lean against produced the starting condition. The credit-cycle insight does not give a counterfactual with no recession; it gives one where the recession is recognized, absorbed, and exited within the normal range of business-cycle duration.
2008 was the test. A Fed that had absorbed Friedman-Schwartz, Eichengreen, Keynes, and the credit-cycle empirics took a starting condition that on most indicators looked worse than 1929 and produced a recession in which unemployment peaked at 10 percent rather than 25 percent. The 2022 Nobel to Bernanke is the institutional seal. Whether the response stored up later inflation is a real argument, but no serious macroeconomist accepts a framing in which sustained 10-percent unemployment is preferable to 2021–23 inflation. The apology of 2002 was paid back in 2008.
The synthesis the 2008 response operated under is itself contested — the size of the fiscal multiplier, the framework for thinking about ordinary recessions, whether the 2008 response itself contributed to the fragilities of the following decade. These are live questions, taken up in the walkthroughs on what causes recessions and whether economists caused 2008. The Great Depression is the reference case both are arguing about.