October 24, 1929. Around 1:30 PM. Richard Whitney, vice president of the New York Stock Exchange, walked from the Morgan offices across Broad Street to the US Steel trading post and bid 205 for ten thousand shares — roughly ten points above the last print. The gesture was theatrical. Thomas Lamont and a cluster of Morgan-led bankers had pooled capital that morning to stop the panic. The panic paused. By the following Monday it was back. Here is the question this desk has been chewing on: what does anyone actually mean when they say the 1929 crash *broke the gold standard*? Because it did not break it. Not that day, not that year, not for three more years. To understand what actually broke, and when, and by how much, you have to walk a decision tree. We are going to ask you three questions. Each answer routes you to a different diagnosis of the mechanics, and at the bottom we reconcile them.
Question 1: Do You Believe the October 1929 Crash Itself Severed the Dollar's Link to Gold?
This is a chronology question, and chronology in this episode is unforgiving. The reason it matters is that most retellings compress three and a half years of slow institutional failure into a single catastrophic week. The compression hides the mechanics.
If Yes
You are missing somewhere between 39 and 51 months, depending on which link you mean. The stock market crashed in October 1929. The US did not suspend citizen convertibility until Executive Order 6102, signed April 5, 1933. The external convertibility — foreign central banks redeeming dollars for gold — was halted during the bank holiday of March 1933 and formally revalued by the Gold Reserve Act of January 30, 1934. Whatever broke in October 1929, it was not the dollar's gold backing.
And here is the detail that nobody who opens with the 1929 crash ever mentions. Between the end of 1929 and the end of 1932, the Federal Reserve's monetary gold stock *rose*, from approximately $3.997 billion to approximately $4.226 billion, per the *Banking and Monetary Statistics of the United States, 1914-1941* series. More gold in the vault, not less. Something else was breaking, and it was not the vault.
If No
Good. Hold this in mind: the dollar remained legally convertible to gold at $20.67 per ounce through all of 1930, 1931, and 1932. During those three years, M1 money supply contracted from roughly $26.4 billion to $19.4 billion — a fall of about 26.5%, per Friedman and Schwartz's *A Monetary History of the United States* — while the Treasury's gold stock was rising. That paradox is the whole article.
Question 2: Was the Gold Standard a Mechanically Binding Constraint on the Federal Reserve?
This is the central historiographical fork. Barry Eichengreen's *Golden Fetters* (1992) argues yes. Friedman and Schwartz argue no. The arithmetic matters here more than the adjectives.
If Yes
You are in the Eichengreen camp. The evidence is cross-sectional: the UK left gold in September 1931 and bottomed in 1932. The US held until 1933 and bottomed later. France held until 1936 and was the last major economy to recover. The correlation is clean in any dataset from the period.
But *binding* needs numbers.
OK so here is where it gets really interesting, and I am going to show the entire gold cover calculation because nobody ever shows it and the whole argument rides on it.
Math teardown — the Federal Reserve's gold cover ratio, 1929 vs 1932.
The Federal Reserve Act of 1913 required the Fed to hold a 40% gold reserve against Federal Reserve notes in circulation, plus a 35% reserve against member bank deposits at the Fed. That is the legal ceiling on how much the Fed could expand without breaching its own statute.
End of 1929. Federal Reserve gold stock: approximately $3.997 billion. Federal Reserve notes outstanding: approximately $1.910 billion. Member bank deposits at the Fed: approximately $2.355 billion. Required gold backing: (0.40 × 1.910) + (0.35 × 2.355) = 0.764 + 0.824 = $1.588 billion. Free gold — the gold the Fed could have deployed without breaching the cover — was therefore approximately 3.997 − 1.588 = $2.409 billion.
End of 1932. Gold stock: approximately $4.226 billion. Federal Reserve notes: approximately $2.778 billion (currency hoarding had pushed this up). Deposits: approximately $2.509 billion. Required gold: (0.40 × 2.778) + (0.35 × 2.509) = 1.111 + 0.878 = $1.989 billion. Free gold: 4.226 − 1.989 = $2.237 billion.
Now the payoff line. The Fed had roughly $2.2 billion in free gold at the end of 1932. At the 40% note cover ratio, that free gold could have legally supported an additional $5.5 billion in Federal Reserve note expansion — more than the entire 1929-1932 M1 contraction being unwound in one stroke.
Put another way: the legal gold ceiling was not the binding constraint. The vault was not empty. It was half full of currency the Fed was refusing to release.
If No
You are in the Friedman-Schwartz camp and the math above is yours. Your extended argument is that the Fed misread its own statute, panicked about free gold, let runs on banks cascade into suspensions, and compounded the contraction by declining to act as lender of last resort. More than 9,000 commercial banks suspended operations between 1930 and 1933, per FDIC historical series. The Fed had gold. It did not have nerve.
*Federal Reserve Bulletin, November 1929. The New York Fed discount rate was cut from 6% to 5% on November 1 — six days after Black Tuesday.*
*The Hoover administration publicly reaffirmed the dollar's gold convertibility in at least a dozen statements between October 1929 and March 1933. The gold pledge was a political commitment long before it became a mechanical constraint.*
Either camp is consistent with the arithmetic. The camps disagree about causation and institutional psychology. The numbers themselves are not in dispute.
Question 3: Was the January 1934 Devaluation the Mechanical Turning Point?
This question matters because the 1933-34 policy sequence had multiple moving parts — the bank holiday, the gold embargo, the executive order on citizen holdings, and finally the formal revaluation — and practitioners disagree about which one actually turned the ship.
If Yes
You are reading the Gold Reserve Act of January 30, 1934 as a price-level intervention, and the arithmetic is blunt.
The official gold price moved from $20.67 per troy ounce to $35.00. The rise:
35.00 / 20.67 = 1.6933 — a 69.33% increase in the dollar price of gold.
The reciprocal — the dollar's value *in* gold — fell by:
1 − (20.67 / 35.00) = 1 − 0.5906 = 0.4094 — a 40.94% devaluation of the dollar.
The Treasury's existing gold stock, valued at $4.033 billion at the old parity, was revalued at $4.033 × 1.6933 ≈ $6.828 billion at the new parity. The paper profit — approximately $2.795 billion — was transferred to the newly created Exchange Stabilization Fund, per Section 10 of the Act itself.
The price-level effect was immediate, and honestly this is my favorite detail in the whole episode. Wholesale prices had been falling at roughly 10% per year from 1929 through early 1933. Within weeks of the April 1933 gold embargo — nine months *before* the formal Gold Reserve Act fixed the new parity — wholesale prices were rising at an annualized rate of around 14%. The market priced in the coming devaluation the moment it stopped believing the old parity. The Act of January 1934 ratified what the market had already done.
If No
You are pointing at the March 1933 bank holiday, and you have a case. Roosevelt closed every commercial bank in the country on March 6, 1933 — four days into his presidency — under the Emergency Banking Act. When sound banks reopened on March 13, deposits began flowing back immediately. Currency in circulation, which had spiked from hoarding, fell approximately $1.4 billion in the two weeks after reopening, per Federal Reserve Bulletin data for March-April 1933.
The turning point for bank runs was March 1933. The turning point for the price level was the embargo of April 1933 and the revaluation of January 1934. These are different turning points for different variables. Both are correct depending on what you are measuring.
If You Answered Everything
Work through the combinations and you get a map of the historiography.
No-Yes-Yes — the crash did not break anything, the gold standard was binding, the 1934 devaluation fixed the price level. This is the modern synthesis, Eichengreen-adjacent, mainstream in the academic literature since *Golden Fetters*.
No-No-Yes — the Fed had gold to spare but failed institutionally, and the 1934 devaluation was nonetheless the formal turning point. This is Friedman's position updated by Bernanke's 1983 paper on the non-monetary effects of the banking crisis.
No-Yes-No — the gold standard was binding and the bank holiday was the turning point. This is the position most historians of banking take, because they are measuring bank balance sheets and not price levels.
What every combination has to agree on is the arithmetic. Gold stock rose from $3.997 billion to $4.226 billion between end-1929 and end-1932. M1 fell by approximately 26.5% over that window. The dollar was devalued 40.94% against gold in 1934. More than 9,000 banks suspended operations in the intervening three years. These numbers are not in dispute in any serious primary or secondary source we have read. What *is* in dispute is what caused what. The decision tree does not collapse the dispute. It tells you where inside it you are standing.
We would revise this framing if a new primary source — an unpublished Federal Reserve Board meeting transcript from the Hamlin, Miller, or Harrison papers, say — established that the free gold doctrine was an operationally binding constraint inside the New York Fed in 1931-32, rather than a post-hoc rationalization written into the record later. Until such a source surfaces, the $2.2 billion of 1932 free gold remains the single number that determines whose camp the mechanics actually belong to. The number sits in the archive. It is waiting for whoever wants to argue the other side.