A Monetary History of the United States, 1867–1960

The book’s method is to decompose the money stock into a small number of components and then ask which one moved. The stock of money depends on the quantity of high-powered money — currency plus bank reserves — and on two ratios: the public’s ratio of bank deposits to currency, and the banks’ ratio of deposits to reserves. Almost everything interesting in the narrative is one of those two ratios moving.

This is a more useful frame than it sounds, because it locates a crisis somewhere specific. A panic is not a mood. It is the public converting deposits into currency, which drives the deposit-currency ratio down, which forces a multiple contraction in the money stock unless something offsets it. The behaviour is ordinary and individually rational; the aggregate effect is not.

The two ratios interact in a way that made the system progressively more fragile without anyone deciding it should be. Friedman and Schwartz work an example: a shift in the public’s deposit-currency ratio of the size seen in the 1893 panic would have forced a 6.3% reduction in the money stock at the deposit-reserve ratio then prevailing — and by 1907, the same shift would have forced 9.8%. The combined change in the two ratios had raised the money stock’s vulnerability to a run by more than half, in fifteen years, with no crisis in between to signal it.

The general lesson is the one worth carrying: leverage in a credit system accumulates in ratios that are nobody’s decision and appear on nobody’s statement. They are visible only if someone computes them, and they change what an identical shock does.

The authors are careful that the downturn beginning in 1929 was not, at first, unusual in monetary terms. The money stock declined slightly, which is more than happens in a mild contraction but not remarkable. What followed was.

In October 1930 the character of the contraction changed, and they date it that precisely. A crop of bank failures concentrated in Missouri, Indiana, Illinois, Iowa, Arkansas and North Carolina produced widespread attempts to convert deposits into currency. Deposits of suspended banks in November 1930 were more than double the highest figure recorded since the monthly series began in 1921.

The numbers immediately after are the ones that make the shape clear: 256 banks failed in November 1930 holding $180 million of deposits, and 352 failed in December holding over $370 million. The authors note the contagion started in agricultural areas that had absorbed the worst of the 1920s failures — and then observe that such contagion “knows no geographical limits”. Distress that had been correctly diagnosed as local for a decade stopped being local.

Across 1930 to 1933 the deposit-currency ratio fell to less than half its starting value. Friedman and Schwartz call it the most notable shift in that ratio in the ninety-three years their series covers. The banking system did not fail because the economy contracted; on their account the causation ran heavily the other way.

The Federal Reserve had been created in response to exactly this class of event, and specifically to make a run on deposits monetarily harmless — by supplying currency to a public that wanted currency, without forcing a multiple contraction of deposits. The authors’ verdict is flat: in practice it did not achieve that objective.

Their counterfactual is the part usually quoted and it is arithmetic rather than rhetoric. Suppose that in the first eight months of 1931 the System had raised its holdings of government securities by $1 billion instead of $80 million. Bank reserves would have risen enough to permit a multiple expansion of deposits in place of the multiple contraction that happened. They work through the offsets — banks would have borrowed less from the System, fewer failures would have meant less currency hoarding — and the offsets do not come close to cancelling it.

The gold drain and the discount-rate increases that followed Britain’s departure from gold in 1931 intensified the collapse, and here the authors are precise about why: those events would not have done that damage had they been accompanied by open market purchases offsetting the effect on high-powered money. The purchases were not made. The failure they identify is an omission, not an action.

They give the underlying reason as a priority ordering — external stability, meaning the gold parity, placed above internal stability, meaning the banking system — held by the Reserve System and by the wider community both. It is worth noting that this was a defensible position held by serious people, which is what makes it instructive rather than merely a blunder.

The most uncomfortable observation in the introduction is comparative. The money stock fluctuated more after 1914 than before it, even excluding the wartime increases. The “blind, undesigned, and quasi-automatic” working of the gold standard, they write, produced “a greater measure of predictability and regularity” than deliberate control exercised inside institutions built to promote monetary stability — their suggested reason being that its discipline was impersonal and inescapable. The authors do not use this to argue against central banking; they use it to argue that having the power to stabilise and having the judgement to use it are separate questions.

Their assessment of the reforms is selective and the selection is the point. Federal deposit insurance, enacted in 1934, “probably has succeeded, where the Federal Reserve Act failed” in making it impossible for a loss of confidence in some banks to become a general panic. Since the FDIC, they note, bank failures became a rarity.

What fixed the problem, on their reading, was not better forecasting or a wiser rate path. It was a structural change that broke the specific mechanism — the depositor’s incentive to run — and thereby removed the transmission from a local failure to a system-wide contraction. The lesson generalises: crises get solved by cutting the channel, not by anticipating the shock.

A quieter observation sits in the chapter on the 1920s and deserves more attention than it gets. Cagan’s explanation, which the authors report, for banks holding fewer excess reserves through that decade is the existence of the Federal Reserve as a “lender of last resort” — the backstop encouraged banks to trim reserves below what they would otherwise have held.

The development of the federal funds market both followed from that willingness and made it easier. An efficient market for repairing reserve deficiencies is unambiguously useful, and it also let every participant carry a thinner buffer.

So the institution created to make the system safer had, before it was tested, made every individual balance sheet thinner. That is not an argument against backstops. It is a reason to check whether a backstop has changed the behaviour of the people it protects before assuming it will hold at the moment it is needed.

The book’s central methodological claim is that a credit system fails through its plumbing, not through its headline price. The rate the public reads about is not where the failure lives; the failure lives in ratios that describe who will lend to whom, against what, and how thin their buffer is.

That is precisely the split The Credit Wall is built on. An aggregate spread near a forty-year low is the headline price. A bank hedging loans it just wrote, a syndication that will not clear, and paper changing hands below par are the plumbing. Friedman and Schwartz’s whole method says to weight the second.

The moral-hazard footnote reads directly onto vendor backstops. A chipmaker guaranteeing a customer’s loan, or agreeing to rent back hardware the customer cannot fill, is a backstop that changes what the protected party is willing to carry — and the question the book teaches you to ask is not whether the guarantee is good, but what everyone did differently because it existed.

It pairs with Big Debt Crises, which tracks the borrower’s balance sheet, and with The General Theory, which tracks the expected return that justified the borrowing. This one tracks the lenders, which is the side of the ledger the current record is moving on.