A Gold Standard Does Not Require Interest-Rate Targeting

Stephen Moore and Herman Cain, the two recent nominees to the Federal Reserve Board of Governors, have in the past suggested returning to a gold standard (although Moore now says he favors merely consulting a broad range of commodity prices as leading indicators). In response, a number of recent op-eds criticized the idea of reinstating a gold standard. The critics unfortunately show little theoretical understanding of the mechanisms by which a gold standard works, and consult no evidence about how the classical gold standard worked in practice.

I don’t seek to defend the nominees, who I think are poor choices on other grounds that have been enumerated by Will Luther. And I don’t seek here to answer many common criticisms of the gold standard, since I have tried to do that here and here. I want to focus on one novel criticism. It stems from imagining that a gold standard regime works like our present regime in the sense that the central bank uses a short-term interest-rate target to steer the economy toward its long-run goal. The only difference is that the central bank pursues a constant dollar price of gold rather than another nominal goal like a gradually rising price-level or nominal-income path.

Thus a Washington Post reporter, Matt O’Brien, declares that a gold standard is “a disaster” and “might be the worst guide to setting policy.” That he sees the gold standard as a “guide to setting policy” already signals a misconception. O’Brien comes to the “disaster” conclusion by starting from the false premise that the wild price volatility of today’s demonetized gold tells us how volatile the price of gold would be under an international gold standard absent domestic central bank action. To offset that potential price volatility, he supposes, the central bank would have to undertake wild and often inappropriate swings in its interest rate policy. If you look at the track record of the classical gold standard, however, you don’t find such wild central bank policies. In the United States, you don’t even find a central bank.

In a classic article on “Econometric Policy Analysis: A Critique,” the Nobel-laureate monetary economist Robert E. Lucas enunciated what has come to be known as “the Lucas Critique.” The Critique warns us against assuming that a statistical relationship observed under one policy regime would persist under a different policy regime. Under the regime of the classical gold standard, a newly minted US $10 dollar coin contained .48375 troy ounces of gold. Alternatively put, the gold definition of the dollar equated 1 troy oz. of gold to $20.67. The dollar price of gold in the US did not vary from that par value (except within the narrow band around parity set by the cost of shipping gold in or out, estimated at less than ±0.33% of par value) despite the absence of offsetting central bank policy. Matt O’Brien’s view is inconsistent with the historical record of the classical gold standard.

Where did O’Brien get his initial false premise and its incorrect implication about monetary policy under a gold standard? He refers to a piece by the economist Menzie Chinn on the blog Econbrowser that asks the question: “What Would It Take to Implement Cain’s Gold Standard, Interest-Rate-Wise?” As its title suggests, Chinn’s piece takes it for granted that a gold standard is implemented by having a central bank adjust interest rates as necessary to maintain a constant nominal price of gold. This is an odd conception of a gold standard because, as already noted, the United States didn’t have a central bank while it was on a gold standard before 1914. The mechanism for maintaining the dollar-gold parity was something quite different: The redeemability of dollar deposits and banknotes for gold coin or bullion, and international transfers of gold, enabled the quantity of money to adjust endogenously to bring about monetary equilibrium at the given parity. The redemption was performed mostly by private commercial banks before 1914.[1]

With the founding of the Federal Reserve (which opened in 1914), and after the 1933 mandate that all commercial banks and individuals turn in their monetary gold to the federal government, the right to redeem dollar-denominated claims to gold could now be exercised only by foreign central banks, and only against the US Treasury. The decentralized automaticity of the classical international gold standard was gone, and central banks ruled the roost. In the mid-1960s President Johnson began restricting foreign redemption of dollars, and in 1971 President Nixon ended it. Since then the dollar price of gold has been free to fluctuate as the public hedges against fiat-money inflation and speculates about a variety of other risks.

Chinn’s piece does not look at the classical gold standard period in the US. He plots data on the dollar price of gold only from 1968 to the present, then regresses the dollar price of gold during that non-gold-standard period on a short-term Treasury bond yield. In disregard of the Lucas Critique, he then quite remarkably draws conclusions about the working of a gold standard. He writes:

Stabilizing the price of gold in US dollars requires adjusting the interest rate (akin to how the exchange rate is managed). … [A] return to the gold standard would imply that the Fed funds rate would have to be about 15 percentage points higher than it was in January 2000 in order to keep the dollar’s value stable at January 2000 levels — a rate 18 percentage points higher than actually recorded in March 2019.

Chinn’s estimate of the required Fed funds rate is based on the coefficients produced by his regression of recent gold prices on a Treasury yield rate. He reports the following point estimate of the relation between “the log price of gold and the real 3 month Treasury yield, estimated over the 1968M03-2019M02 period”:

pgold = 6.423 – 10.210 fedfunds

He infers from the negative sign that the Fed can raise interest rates to lower the price of gold. To maintain a constant dollar-gold parity, he then supposes, the Fed must raise interest rates (its only policy tool considered) to offset what would otherwise be a rise in the dollar price of gold. But that is plainly not how the parity is or was maintained under a gold standard as conventionally defined or as historically experienced. Rather, the parity is maintained by redeemability of dollars into coined gold at the defined par rate. The dollar money stock endogenously adjusts—either via the price-specie-flow mechanism a la David Hume or via goods arbitrage a la McCloskey and Zecher—until it is consistent with the quantity of money demanded at the defined parity.

Chinn’s idiosyncratic conception of how a gold standard works is inconsistent with at least two historical facts. First, as a recent working paper by Christopher Hanes shows, the Bank of England, which did vary its discount rate to manage gold flows, never had to raise its rate above 7 percent during the classical period 1880-1913. There is nothing remotely like a 20 percent rate to be seen. Second, as already noted, the United States maintained a fixed dollar-gold parity over the same span without any central bank. A central bank varying an interest-rate policy target obviously cannot be the key to explaining how the US maintained a fixed dollar-gold parity before 1914. Having a central bank adjust the interest rate can hardly be a requirement for keeping the dollar at its defined par value with gold. Although there is no central bank policy rate to track, the observed range of rates on prime commercial paper in the United States remained between 3 percent and 6.5 percent during 1888-1914, as shown in a published paper by Gene Smiley.

The Lucas Critique may not always be pertinent criticism, as found by a paper that Professor Chinn has cited. But the Critique provides a highly relevant warning against extrapolating from the behavior of the dollar price of gold under our current fiat regime to the behavior of the dollar price of gold under a gold standard regime. As already noted, the dollar price of one troy ounce of gold does not vary under a gold standard (except within the very narrow band between the gold import and export points). Rather, the dollar sticks to its definition in terms of gold due to redemption and equilibrating money flows. If one were to repeat Chinn’s exercise with data from the classical gold standard, regressing the log dollar price of gold on a constant plus the real 3-month Treasury rate, I venture to predict that he would find very different coefficients. Namely, the constant would be the log of $20.67, and the coefficient on the Treasury rate would be zero.

Whatever the demerits of Stephen Moore and Herman Cain as potential Federal Reserve Governors, the working of a gold standard should not be misunderstood or misrepresented as part of the argument against them. As Tyler Cowen has noted, one of the nominees’ chief shortcomings is their loyalty to a President who gives partisan monetary policy advice, whereas a great merit of an automatic gold standard system is that it provides a barrier against partisan manipulation of money. Both supporters and critics of the nominees should make a real effort to study the self-adjusting mechanisms and track record of the classical gold standard before they absurdly proclaim it a disaster.

[1] George Selgin and I have explained why redemption by private commercial banks, constrained by competition and the rule of law, is more reliable than redemption by monopolistic bodies with sovereign immunity.

[Cross-posted from Alt-M.org]