March/April 2008

As Good as Gold?

For the first time in many years, the monetary arrangements of the United States have become an issue in the 2008 presidential race. The subprime crisis and the decline in the foreign exchange value of the dollar have raised questions about the performance of the Federal Reserve Board. One candidate has proposed ending the post-1971 experiment with an unanchored fiat dollar issued by the Federal Reserve and returning to a gold standard with private money issue. Critics have raised a number of theoretical and historical objections to the gold standard. Some have called the gold standard a "crazy" idea. The gold standard is not a flawless monetary system. Neither is the fiat money alternative. In light of historical evidence about the comparative magnitude of these flaws, the gold standard is not a crazy idea.

The "gold standard" generically means a monetary system in which a certain mass of gold defines the monetary unit (e.g., the "dollar") and serves as the ultimate medium of redemption. For example, during the "classical" gold standard period (1879–1914), the U.S. dollar was defined as 0.048 troy oz. of pure gold. Inverting the defined ratio, 1 ounce of pure gold was equivalent to US$20.67. Gold coins need not, and historically did not, form the predominant medium of exchange in a financially sophisticated economy. Issuers of paper currency and checkable deposits, normally private commercial banks but also a government central bank if one exists, make their notes redeemable for gold and hold gold coins and bullion as reserves for meeting redemption demands. Because of the banks' contractual obligation to redeem in gold, the volume of paper currency and deposits — the everyday means of payment — is geared to the volume of gold. So what are the key objections to the gold standard?

"A gold standard leaves the quantity of money to be determined by accidental forces."

There is a germ of truth to this concern. A gold standard does leave the quantity and purchasing power of money to be determined by the forces of supply and demand in the market for gold. There can be "accidental" shifts in the supply and demand curves to which the quantity and purchasing power of money will respond. Our current fiat standard, by contrast, leaves the supply of money to the decisions of a committee (namely, the Federal Open Market Committee of the Federal Reserve System). The practical question is: under which system are the quantity and purchasing power of money better behaved?

As is well known, the stock of gold did not grow at a perfectly steady rate during the era of the historical gold standard. Some increases in gold output — such as the Yukon discoveries and the development of the cyanide process — were responses to previous increases in demand and the purchasing power of gold and thus helped to stabilize the purchasing power of gold over the long run. Other increases resulted from accidental discoveries. The largest such "supply shock" in the 19th century was the 1848 discovery of gold in California. The outpouring of gold from California reduced the purchasing power of gold around the world, or in other words, generated an inflation of the price level. But how large an inflation? The magnitude was surprisingly small. Even over the most inflationary interval, the general price index for the United States rose from 5.71 in 1849 to 6.42 in 1857 (year 2000=100), an increase of 12.4 percent spread over eight years. The compound annual price inflation rate over those eight years was slightly less than 1.5 percent. Twenty-two years later, when the gold standard was finally restored following its suspension during the Civil War, the purchasing power of gold had actually risen slightly (the price level was slightly lower).

The economic historian Hugh Rockoff, in an examination of the output of gold, concluded that "it is fair to describe the fluctuations in the supply of gold under the classical standard as small and welltimed." He found that supply of fiat money in the postwar United States (1949–79), by contrast to the behavior of gold under the classical gold standard, had both higher annual rates of growth and a higher standard deviation of annual growth rates around decade averages.

In a study covering many decades in a large sample of countries, the Federal Reserve Bank of Minneapolis economists Arthur Rolnick and Warren Weber similarly found that "money growth and inflation are higher" under fiat standards than under gold and silver standards. Specifically, they reported, "The average inflation rate for the fiat standard observations is 9.17 percent per year; the average inflation rate for the commodity standard observations is 1.75 percent per year."

This result was not driven by a few extreme cases; in fact, in computing the average rates of inflation Rolnick and Weber deliberately omitted cases of hyperinflation (which occurred only under fiat money). Still, "every country in our sample experienced a higher rate of inflation in the period during which it was operating under a fiat standard than in the period during which it was operating under a commodity standard." Peter Bernholz of the University of Basel adds that "a study of about 30 currencies shows that there has not been a single case of a currency freely manipulated by its government or central bank since 1700 which enjoyed price stability for at least 30 years running."

The evidence thus indicates that growth in the stock of gold has been slower and steadier in practice than growth in the stock of fiat money. Of course, U.S. inflation is thankfully not as high as 9 percent today, but at 4.3 percent (CPI, year-overyear) it is currently more than twice as high as Rolnick and Weber's figure for commodity standards. Under a gold standard, the price level can be trusted not to wander far over the next 30 years because it is constrained by impersonal market forces. Under a fiat standard, the future price level depends on the personalities of yet-to-beappointed monetary authorities and is thus anybody's guess.

The blogger Megan McArdle gets things almost exactly backward when she writes, "The gold standard cannot do what a wellrun fiat currency can do, which is tailor the money supply to the economy's demand for money." Under the gold standard, market forces do in fact automatically tailor the money supply to the economy's demand for money. The economics of gold mining operates to match world supply with world demand at a stable price level (though admittedly large demand shocks can take years to be accommodated), and the "price-specie-flow mechanism" quickly brings gold from the rest of the world into any single country where demand for money has grown. We can only imagine a well-run fiat-currency-issuing central bank trying to match these properties. We cannot observe any central bank that has actually managed it.

"A gold standard would be a source of harmful deflation."

The inflation rate under the gold standard averaged close to zero over generations, being sometimes slightly positive and sometimes slightly negative over individual decades. Rolnick and Weber, as quoted above, found an average inflation rate of 1.75 percent over the sample of gold and silver episodes reported in the published version of their paper; an earlier version using a different sample arrived at an average rate of -0.5 percent. In 1879 the United States resumed gold redemption for the U.S. dollar, which had been suspended since the Civil War. Between 1880 and 1900 the United States experienced one of the most prolonged periods of deflation on record. The price level trended more or less steadily downward, beginning at 6.10 and ending at 5.49 (GDP deflator, base year 2000=100). That works out to a total decline of 10 percent stretched over 20 years. The deflationary period was no disaster for the real economy. Real output per capita began the period at $3,379 and ended it at $4,943 (both in 2000 dollars). Total real per capita growth was thus a more-than-healthy 46 percent. (Real GDP itself more than doubled.)

Monetary economists distinguish a benign deflation (due to the output of goods growing rapidly while the stock of money grows slowly, as in the 1880–1900 period) from a harmful deflation (due to unanticipated shrinkage in the money stock). The gold standard was a source of mild benign deflation in periods when the output of goods grew faster than the stock of gold. Prices particularly fell for those goods whose production enjoyed great technological improvement (for example oil and steel after 1880). Strong growth of real output, for particular goods or in general, cannot be considered harmful.

It would be possible for the central bank under a fiat money standard to offset productivity- driven declines in some prices by expanding the quantity of money in order to drive others prices upward, thus eliminating deflation "on average." But there is no social benefit in doing so. Falling costs of production in steel (i.e., productivity gains) do not discourage investment in steel. A gradual anticipated deflation does not discourage investment, especially not when productivity gains are driving growth in the first place.

Nor does a deflation penalize debtors once it comes to be anticipated, because nominal interest rates adjust downward to reflect anticipated repayment in dollars of higher purchasing power.

"The gold standard was responsible for the U.S. banking panics of the late 19th century and the monetary contraction of 1929–33 and thereby for the Great Depression."

The U.S. monetary contraction of 1929–33 is the prime example of a harmful deflation. It should be noted that it happened on the Federal Reserve's watch. The episode should be blamed not on the gold standard, but on the combination of a weak banking system and a befuddled central bank. The U.S. banking system was prone to runs and panics in the late 19th century, and continued to be so through the 1929–33 episode in which the Fed stood by and did not supply replacement reserves to keep the money stock from contracting. Other countries on the gold standard — for example Canada — had no banking panic in 1929–33 (nor did Canada have panics in the late 19th century), so the gold standard couldn't have been responsible for the panics. Rather the panics were due to completely avoidable legal restrictions (namely the ban on branch banking, and compulsory bond collateral requirements making the supply of banknotes "inelastic") that weakened the U.S. banking system.

"The benefit of a gold standard (restraining inflation) is attainable at less cost by properly controlling the supply of a fiat money."

Although growth in the stock of fiat money could in principle be as slow as (or slower than) growth in the stock of gold under a gold standard, it has not been so in practice, as already noted. Alan Greenspan actually used to recommend controlling the fiat money supply to mimic the pricelevel behavior of a gold standard. In response to questioning at a 2001 congressional hearing, Greenspan said: "Mr. Chairman, so long as you have fiat currency, which is a statutory issue, a central bank properly functioning will endeavor to, in many cases, replicate what a gold standard would itself generate."

Fiat money regimes have not, however, accomplished price stability as fully as the gold standard did. Although inflation is less severe today than it was 30 years ago, experienced inflation rates, and the expectations of future inflation rates embodied in long-term interest rates, have remained higher than corresponding rates under the classical gold standard.

"A gold standard is no restraint at all, because government can devalue or suspend gold redemption whenever it wants."

A similar claim could be made about any other restraint in the Constitution. And yet constitutional rules are useful. By authorizing only a limited set of government activities, ruling others simply out of bounds, they save the public the trouble of trying to weigh every potential activity on a cost-benefit basis.

An important problem in fiat money regimes, as famously identified by Finn Kydland and Edward C. Prescott, is the lack of an enforceable commitment not to use surprise monetary expansion and resulting inflation as a temporary stimulus to the economy. When the public knows that the central bank would be tempted to use surprise inflation, the public rationally expects higher-than-optimal inflation. The central bank has to deliver higher-than-optimal inflation to avoid a negative surprise. An unfortunate standoff is reached at a higherthan- optimal inflation rate (which, being fully anticipated, provides no economic stimulus). A gold standard avoids this trap. Like tying Ulysses to the mast, it achieves better results by removing the option (to use surprise inflation) that leads to ruin. Of course, a gold standard is not the only possible rule for constraining the creation of money. Alternatives include a Friedmantype money-growth rule or an inflation-targeting rule. But the gold standard has a longer history and is the only historically tested rule that does not presuppose a central bank.

Leaving money issue in the hands of private banks rather than a government institution, as the United States did before 1913, removes the option to use surprise monetary expansion one step further. It remains true that government can suspend the gold standard in an emergency, as both sides did during the U. S. Civil War, but the spirit of the gold standard calls for returning to the parity afterward, as the United States did. Judging by long-term interest rates and the thick market for long-term bonds under the post-bellum classical gold standard, the risk of permanent devaluation or suspension was considered small.

"Fiat money is necessary so that a lender of last resort can meet liquidity needs of the banking system."

History shows that a lender of last resort would hardly be needed with a stable monetary regime and a sound banking system (again it is instructive to contrast the United States with Canada in the 19th century). In the rare cases such a lender might be needed, bank clearinghouses can play the role.

"The gold standard is an example of price fixing by government."

The gold standard doesn't fix a price between dollars and gold any more than the traditional British measurement system fixes a price between pints and quarts. The fixed relationship is a matter of definition. A gold standard defines the dollar (or whatever the name of the monetary unit) as a specified mass of gold. Dollars are not a separate good from gold.

"The United States can't recreate the classical international gold standard by itself."

I have saved for last what I think is the strongest objection to unilateral return to the gold standard. The United States would not enjoy the benefits of being on an international gold standard if it were the first and only country whose currency was linked to gold. At least two major benefits would be missing: (1) the United States would not enjoy fixed exchange rates with the rest of the world (of course, we're already living with that disadvantage today), and (2) the purchasing power of gold would not be as stable. The purchasing power (or relative price) of today's demonetized gold has been much less stable than that of gold under the 19th century's global gold standard, because the demand to hold gold today is largely a speculative rather than a transactions demand. With only one economy on gold — albeit a large economy — monetary use of gold would likely remain the tail rather than the dog. Thus even in the unlikely event that the United States were to elect a president committed to a pro-gold policy, that president would be prudent to try to cultivate similar commitments from the governments of the other leading economies of the world before taking the United States down the yellow brick road alone.


Under the gold standard the issue of common money by banks is restrained by the cost of acquiring gold, which is determined by impersonal supply-and-demand forces in the gold mining market. Because of the issuers' contractual obligations to redeem in gold and the corresponding prudential need to hold gold reserves, the dollar volume of paper currency and deposits — the stock of money — is geared to the volume of gold. Growth in the stock of money is governed by market forces rather than by government fiat. A gold standard does not guarantee perfect steadiness in the growth of the money supply, but historical comparison shows that it has provided more moderate and steadier money growth in practice than the present-day alternative, politically empowering a central banking committee to determine growth in the stock of fiat money. From the perspective of limiting money growth appropriately, the gold standard is far from a crazy idea.

Historical problems of U.S. banking instability, sometimes blamed on the gold standard, turn out on closer inspection to have had been rooted in banking regulations that inadvertently weakened U.S. banks. Gold standard countries like Canada that avoided the peculiar banking restrictions of the United States also avoided the instability. As we discovered in the greatest banking panic, that of 1929–33, having a Federal Reserve System capable of overriding the gold standard did not eliminate the problem of weakness in the U.S. banking system.

Other supposed historical problems, like price deflation due to goods production outgrowing gold production, turn out not to have been actual problems.

A gold standard does entail resource costs of mining the gold that is lodged in bank vaults. But so too does a fiat standard entail resource costs, primarily in the form of the deadweight costs of inflation. All in all, because the costs of a gold standard are reasonably small in relation to its benefits, the gold standard is not a crazy idea.

Lawrence H. White is the F. A. Hayek Professor of Economic History at the University of Missouri–St. Louis and an adjunct scholar of the Cato Institute. He is the author of Competition and Currency, Free Banking in Britain, and The Theory of Monetary Institutions. A longer, footnoted version of this article can be found here.