The gold standard, as an international institution, existed for less than 40 years, emerging first, and by accident, in England, and more than a century and a half later, spreading by a rapid series of independent, but interrelated, decisions to the United States, Germany, and most of Europe and much of the rest of the world. After its collapse with the outbreak of World War I, reconstruction of the gold standard was thought by many to be a precondition for recreating the stable and prosperous international order that had been brutally demolished by the Great War. But that attempt ended catastrophically when the restoration of the gold standard was subverted by the insane gold-accumulation policy of the Bank of France and the failure of the Federal Reserve and other national monetary authorities to heed the explicit warnings of two of the leading monetary theorists of immediate postwar era, R. G. Hawtrey and Gustav Cassel, that unless the monetary demand for gold was kept from increasing as a result of the resumption of convertibility, a renewed gold standard could trigger a disastrous deflation.
But despite its short, checkered, and not altogether happy, history as an international institution, the gold standard, in its idealized and largely imagined form has retained a kind of nostalgic aura of stability, excellence and grandeur, becoming an idiom for anything that’s the best of its kind. So no, Virginia, there is no Santa Claus, and the gold standard is not the gold standard of monetary systems.
My own impression is that most, though not all, supporters of the gold standard are smitten by a kind of romantic, unthinking, and irrational attachment to the idea that the gold standard is a magic formula for recovering a lost golden age. But having said that, I would also add that I actually think that the gold standard, in its brief first run as an international monetary system, did not perform all that badly, and I can even sympathize with the ultimately unsuccessful attempt to restore the gold standard after World War I. I just think that the risks of scrapping our current monetary arrangements and trying to replace them with a gold standard recreated from scratch, over a century after it ceased to function effectively in practice, are far too great to consider it seriously as a practical option.
Not long ago I was chided by Larry White for being unduly harsh in my criticism of the gold standard in my talk at the Mercatus Center conference on Monetary Rules for a Post-Crisis World. I responded to Larry in this post. But I now find myself somewhat exasperated by a post by Stephen Cecchetti and Kermit Schoenholtz on the blog they maintain for their money and banking textbook. I don’t know much about Schoenholtz, but Cecchetti is an economist of the first rank. They clearly share my opposition to gold standard, but I’m afraid that some of their arguments against the gold standard are misguided or misconceived. I don’t write this post just to be critical; it’s only because their arguments reflect common and long-standing misconceptions and misunderstandings that have become part of the received doctrine about the gold standard that those arguments are worth taking the time and effort to criticize.
So let’s start from the beginning of their post, which is actually quite a good beginning. They quote Barry Eichengreen, one of our most eminent economic historians.
“Far from being synonymous with stability, the gold standard itself was the principal threat to financial stability and economic prosperity between the wars.” Barry Eichengreen, Golden Fetters.
That’s certainly true, but notice that the quotation from Eichengreen explicitly refers to the interwar period; it’s not a blanket indictment.
After quoting Eichengreen, Cecchetti and Schoenholtz refer to a lecture delivered by Ben Bernanke when he was still Chairman of the Federal Reserve Board. Quoting this lecture is not a good sign, because back in 2012 after Bernanke gave the lecture, I wrote a post in which I explained why Bernanke had failed to provide a coherent criticism of the gold standard.
In his 2012 lecture Origins and Mission of the Federal Reserve, then-Federal Reserve Board Chair Ben Bernanke identifies four fundamental problems with the gold standard:
- When the central bank fixes the dollar price of gold, rather than the price of goods we consume, fluctuations in the dollar price of goods replace fluctuations in the market price of gold.
- Since prices are tied to the amount of money in the economy, which is linked to the supply of gold, inflation depends on the rate that gold is mined.
- When the gold standard is used at home and abroad, it is an exchange rate policy in which international transactions must be settled in gold.
- Digging gold out of one hole in the ground (a mine) to put it into another hole in the ground (a vault) wastes resources.
Bernanke’s first statement is certainly correct, but his second statement ignores the fact that the amount of new gold extracted from the earth in a year is only a small fraction of the existing stock of gold. Thus, fluctuations in the value of gold are more likely to be caused by fluctuations in the demand for gold than by fluctuations in supply. The third statement makes as much sense as saying that since the US economy operates on a dollar standard transactions must be settled in hard currency. In fact, the vast majority of legal transactions are settled not by the exchange of currency but by the exchange of abstract claims to currency. There is no reason why, under a gold standard, international transactions could not be settled by abstract claims to gold rather than in physical gold. I agree with the fourth statement.
Consistent with Bernanke’s critique, the evidence shows that both inflation and economic growth were quite volatile under the gold standard. The following chart plots annual U.S. consumer price inflation from 1880, the beginning of the post-Civil War gold standard, to 2015. The vertical blue line marks 1933, the end of the gold standard in the United States. The standard deviation of inflation during the 53 years of the gold standard is nearly twice what it has been since the collapse of the Bretton Woods system in 1973 (denoted in the chart by the vertical red line). That is, even if we include the Great Inflation of the 1970s, inflation over the past 43 years has been more stable than it was under the gold standard. Focusing on the most recent quarter century, the interval when central banks have focused most intently on price stability, then the standard deviation of inflation is less than one-fifth of what it was during the gold standard epoch.
Annual Consumer Price Inflation, 1880 to 2016
Source: Federal Reserve Bank of Minneapolis.
I am sorry to say this, but comparing the average rate and the variability of inflation under the gold standard (1880-1933) and under a pure dollar standard (1973-2016) is tendentious and misleading, because the 20-year period from 1914 to 1933, a period marked by rapid wartime and post-war inflation and two post-war deflations, was a period when the gold standard either was not functioning at all (1914 to about 1922) or was being unsuccessfully reconstructed. Now it would be one thing to conclude from the failed attempt at reconstruction that a second attempt at reconstruction would be futile and potentially as disastrous as the first attempt, but that doesn’t seem to be what Ceccheti and Schoenholz are arguing. Instead, they include data from 20 years when the gold standard was either not operating at all or was operating dysfunctionally to make an unqualified comparison between the performance of the US economy under the gold standard and under a pure dollar standard. That simplistic comparison conveys almost no useful information.
A fairer comparison would be between the gold standard as it operated between 1880 and 1914 and either the dollar standard of the post-Bretton Woods era (1973 – 2016) or the period from 1991 to 2016 when, according to Cecchetti and Schoenholz, the Fed adopted an explicit or implicit inflation target as its primary policy objective. Changing the gold-standard period in this way reduces the average inflation rate from 0.86% to 0.23% a year and the standard deviation from 5.06% to 2.13%. That comparison is not obviously disadvantageous to the gold standard.
What about economic growth? Again, the gold standard was associated with greater volatility, not less. The following chart plots annual growth as measured by gross national product (gross domestic product only came into common use in the 1991.) The pattern looks quite a bit like that of inflation: the standard deviation of economic growth during the gold-standard era was more than twice that of the period since 1973. And, despite the Great Recession, the past quarter century has been even more stable. To use another, simpler, measure, in the period from 1880 to 1933 there were 15 business cycles identified by the National Bureau of Economic Research. That is, on average there was a recession once every 3½ years. By contrast, since 1972, there have been 7 recessions; one every 6 years.
Annual GNP Growth, 1880-2016
Source: FRED and Romer (1986)
Again, this comparison, like the inflation comparison is distorted by the exogenous disruption associated with World War I and its aftermath. Excluding the 1914 to 1933 period from the comparison would make for a far less one-sided comparison than the one presented by Cecchetti and Schoenholtz.
Finally, consider a crude measure of financial stability: the frequency of banking crises. From 1880 to 1933, there were at least 5 full-fledged banking panics: 1893, 1907, 1930, 1931, and 1933. Including the savings and loan crisis of the 1980s, in the past half century, there have been two.
But if we exclude the Great Depression period from the comparison, there were two banking panics under the gold standard and two under the dollar standard. So the performance of the gold standard when it was operating normally was not clearly inferior to the operation of the dollar standard.
So, on every score, the gold standard period was less stable. Prices were less stable; growth was less stable; and the financial system was less stable. Why?
We see six major reasons. First, the gold standard is procyclical. When the economy booms, inflation typically rises. In the absence of a central bank to force the nominal interest rate up, the real interest rate falls, providing a further impetus to activity. In contrast, countercyclical monetary policy—whether based on a Taylor rule or not—would lean against the boom.
I don’t know what the basis is for the factual assertion that high growth under a gold standard is associated with inflation. There were periods of high growth with very low or negative inflation under the gold standard. Periods of mild inflation, owing to a falling real value of gold, perhaps following significant discoveries of previously unknown gold deposits, may have had a marginal stimulative effect under the gold standard, but such episodes were not necessarily periods of economic instability.
At any rate it is not even clear why, if a countericyclical monetary policy were desirable, such a policy could not be conducted under the gold standard by a central bank constrained by an obligation to convert its liabilities into gold on demand. Some gold standard proponents, like Larry White for example, insist that a gold standard could and would function better without a central bank than with a central bank. I am skeptical about that position, but even if it is correct, there is nothing inherent in the idea of a gold standard that is inconsistent with the existence of a central bank that conducts a countercyclical policy, so I don’t understand why Cecchetti and Schoenholtz assert, without argument, that a central bank could not conduct a countercyclical policy under a gold standard.
Second, the gold standard has exchange rate implications. While we do not know for sure, we suspect that current U.S. advocates of a shift to gold are thinking of the case where the United States acts alone (rather than waiting to coordinate a global return to the gold standard). If so, the change would impose unnecessary risks on exporters and importers, their employees and their creditors. To see why, consider the consequences of a move in the global price of gold measured in some other currency, say British pounds. If the pound price of gold changed, but the dollar price of gold did not, the result would be a move in the real dollar-pound exchange rate. That is, unless the dollar prices of U.S. goods and the dollar wages of U.S. workers adjust instantly to offset gold price fluctuations, the real dollar exchange rate changes. In either case, the result would almost surely induce volatility of production, employment, and the debt burden.
If the US monetary policy were governed by a commitment to convert the dollar into gold at a fixed conversion rate, the dollar price of gold would remain constant and exchange rates of the dollar in terms of the pound and other non-gold currencies would fluctuate. We have fluctuating exchange rates now against the pound and other currencies. It is not clear to me why exchange rates would be more volatile under this system than they are currently.
More broadly, a gold standard suffers from some of the same problems as any fixed-exchange rate system. Not only can’t the exchange rate adjust to buffer external shocks, but the commitment invites speculative attacks because it lacks time consistency. Under a gold standard, the scale of the central bank’s liabilities—currency plus reserves—is determined by the gold it has in its vault. Imagine that, as a consequence of an extended downturn, people come to fear a currency devaluation. That is, they worry that the central bank will raise the dollar price of gold. In such a circumstance, it will be natural for investors to take their dollars to the central bank and exchange them for gold. The doubts that motivate such a run can be self-fulfilling: once the central bank starts to lose gold reserves, it can quickly be compelled to raise its dollar price, or to suspend redemption entirely. This is what happened in 1931 to the Bank of England, when it was driven off the gold standard. It happened again in 1992 (albeit with foreign currency reserves rather than gold) when Britain was compelled to abandon its fixed exchange rate.
Cecchetti and Schoenholtz articulate a valid concern, and it is a risk inherent in any gold standard or any monetary system based on trust in a redemption commitment. My only quibble is that Cecchetti and Schoenholtz overrate the importance of gold reserves. Foreign-exchange reserves would do just as well, and perhaps better, than gold reserves, because, unlike gold, foreign-exchange reserves yield interest.
Third, as historians have emphasized, the gold standard helped spread the Great Depression from the United States to the rest of the world. The gold standard was a global arrangement that formed the basis for a virtually universal fixed-exchange rate regime in which international transactions were settled in gold.
Here Cecchetti and Schoenholtz stumble into several interrelated confusions. First, while the gold standard was certainly an international transmission mechanism, the international linkage between national price levels being an essential characteristic of the gold standard, there is no reason to identify the United States as the source of a disturbance that was propagated to the rest of the world. Because the value of gold must be equalized in all countries operating under the gold standard, changes in the value of gold are an international, not a national, phenomenon. Thus, an increase in the demand for gold causing an increase in its value would have essentially the same effects on the world economy irrespective of the geographic location of the increase in the demand for gold. In the 1920s, the goal of reestablishing the gold standard meant restoring convertibility of the leading currencies into gold, so that price levels in all gold standard countries, all reflecting the internationally determined value of gold, were closely aligned.
This meant that a country with an external deficit — one whose imports exceed its exports — was required to pay the difference by transferring gold to countries with external surpluses. The loss of gold forced the deficit country’s central bank to shrink its balance sheet, reducing the quantity of money and credit in the economy, and driving domestic prices down. Put differently, under a gold standard, countries running external deficits face deflationary pressure. A surplus country’s central bank faced no such pressure, as it could choose whether to convert higher gold stocks into money or not. Put another way, a central bank can have too little gold, but it can never have too much.
Cecchetti and Schonholtz are confusing two distinct questions: a) what determines the common international value of gold? and b) what accounts for second-order deviations between national price levels? The important point is that movements in national price levels under the gold standard were highly correlated because, owing to international arbitrage of tradable goods, the value of gold could not differ substantially between countries on the gold standard. Significant short-run changes in the value of gold had to reflect changes in the total demand for gold because short-run changes in the supply of gold are only a small fraction of the existing stock. To be sure, short-run differences in inflation across countries might reflect special circumstances causing over- or under-valuation of particular currencies relative to one another, but those differences were of a second-order magnitude relative to the sharp worldwide deflation that characterized the Great Depression in which the price levels of all gold-standard countries fell simultaneously. It is always the case that some countries will be running trade deficits and some will be running trade surpluses. At most, that circmstance might explain small differences in relative rates of inflation or deflation across countries; it can’t explain why deflation was rampant across all countries at the same time.
The shock that produced the Great Depression was a shock to the real value of gold which was caused mainly by the gold accumulation policy of the Bank of France. However, the United States, holding about 40 percent of the world’s monetary gold reserves after World War I, could have offset or mitigated the French policy by allowing an outflow of some of its massive gold reserves. Instead, the Fed, in late 1928, raised interest rates yet again to counter what it viewed as unhealthy stock-market speculation, thereby intensifying, rather than mitigating, the deflationary effect of the gold-accumulation policy of the Bank of France. Implicitly, Cecchetti and Schoenholtz assume that the observed gold flows were the result of non-monetary causes, which is to say that gold flows were the result of trade imbalances reflecting purely structural factors, such as national differences in rates of productivity growth, or propensities to save, or demand and supply patterns, over which central banks have little or no influence. But in fact, central banks and monetary authorities based their policies on explicit or implicit goals for their holdings of gold reserves. And the value of gold ultimately reflected the combined effect of the policy decisions taken by all central banks thereby causing a substantial increase in the demand of national monetary authorities to hold gold.
This policy asymmetry helped transmit financial shocks in the United States abroad. By the late 1920s, the major economies had restored the pre-World War I gold standard. At the time, both the United States and France were running external surpluses, absorbing the world’s gold into their central bank vaults.
Cecchetti and Schoneholtz say explicitly “the United States and France were running external surpluses, absorbing the world’s gold into their central bank vaults” as if those surpluses just happened and were unrelated to the monetary policies deliberately adopted by the Bank of France and the Federal Reserve.
But, instead of allowing the gold inflows to expand the quantity of money in their financial systems, authorities in both countries tightened monetary policy to resist booming asset prices and other signs of overheating. The result was catastrophic, compelling deficit countries with gold outflows to tighten their monetary policies even more. As the quantity of money available worldwide shrank, so did the price level, adding to the real burden of debt, and prompting defaults and bank failures virtually around the world.
Cecchetti and Schoenholtz have the causation backwards; it was the tightness of monetary policy that caused gold inflows into France and later into the United States. The gold flows did not precede, but were the result of, already tight monetary policies. Cecchetti and Schoenholtz are implicitly adopting the sterilization model based on the price-specie-flow mechanism in which it is gold flows that cause, or ought to cause, changes in the quantity of money. I debunked the simplistic sterilization idea in this post. But, in fairness, I should acknowledge that Cecchetti and Schoenholz do eventually acknowledge that the demand by central banks to hold gold reserves is what determines actual monetary policy under the gold standard. But despite that acknowledgment, they can’t free themselves from the misconception that it was a reduction in the quantity of money, rather than an increase in the demand for gold, that caused the value of gold to rise in the Great Depression.
Fourth, economists blame the gold standard for sustaining and deepening the Great Depression. What makes this view most compelling is the fact that the sooner a country left the gold standard and regained discretionary control of its monetary policy, the faster it recovered. The contrast between Sweden and France is striking. Sweden left gold in 1931, and by 1936 its industrial production was 14 percent higher than its 1929 level. France waited until 1936 to leave, at which point its industrial production was fully 26 percent below the level just 7 years earlier (see here and here.) Similarly, when the U.S. suspended gold convertibility in March 1933—allowing the dollar to depreciate substantially—the financial and economic impact was immediate: deflation turned to inflation, lowering the real interest rate, boosting asset prices, and triggering one of the most powerful U.S. cyclical upturns (see, for example, Romer).
This is certainly right. My only quibble is that Cecchetti and Schoenholtz do not acknowledge that the Depression was triggered by a rapid increase in the international demand for gold by the world’s central banks in 1928-29, in particular the Bank of France and the Federal Reserve.
Turning to financial stability, the gold standard limits one of the most powerful tools for halting bank panics: the central bank’s authority to act as lender of last resort. It was the absence of this function during the Panic of 1907 that was the primary impetus for the creation of the Federal Reserve System. Yet, under a gold standard, the availability of gold limits the scope for expanding central bank liabilities. Thus, had the Fed been on a strict gold standard in the fall of 2008—when Lehman failed—the constraint on its ability to lend could again have led to a collapse of the financial system and a second Great Depression.
At best, the charge that the gold standard limits the capacity of a central bank to act as a lender of last resort is a serious oversimplification. The ability of a central bank to expand its liabilities is not limited by the gold standard in any way. What limits the ability of the central bank to expand its liabilities are gold-cover requirements such as those enacted by the Bank Charter Act of 1844 which imposed a 100% marginal reserve requirement on the issue of banknotes by the Bank of England beyond a fixed fiduciary issue requiring no gold cover. Subsequent financial crises in 1847, 1857, and 1866 were quelled as soon as the government suspended the relevant provisions of the Bank Charter Act, allowing the Bank of England to increase its note issue and satisfy the exceptional demands for liquidity that led to the crisis in the first place.
Finally, because the supply of gold is finite, the quantity available to the central bank likely will grow more slowly than the real economy. As a result, over long periods—say, a decade or more—we would expect deflation. While (in theory) labor, debt and other contracts can be arranged so that the economy will adjust smoothly to steady, long-term deflation, recent experience (including that with negative nominal interest rates) makes us skeptical.
Whether gold appreciates over the long-term depends on the rate at which the quantity of gold expands over time and the rate of growth of demand for gold over time. It is plausible to expect secular deflation under the gold standard, but it is hardly inevitable. I don’t think that we yet fully understand the conditions under which secular deflation is compatible with full employment. Certainly if we were confident that secular deflation is compatible with full employment, the case for secular deflation would be very compelling.
This brings us back to where we started. Under a gold standard, inflation, growth and the financial system are all less stable. There are more recessions, larger swings in cons umer prices and more banking crises. When things go wrong in one part of the world, the distress will be transmitted more quickly and completely to others. In short, re-creating a gold standard would be a colossal mistake.
These conclusions are based on very limited historical experience, and it is not clear how relevant that experience is for contemporary circumstances. The argument against a gold standard not so much that a gold standard could not in principle operate smoothly and efficiently. It is that, a real gold standard having been abandoned for 80 years, recreating a gold standard would be radical and risky undertaking completely lacking in a plausible roadmap for its execution. The other argument against the gold standard is that insofar as gold would be actually used as a medium of exchange in a recreated gold standard with a modern banking system, the banking system would be subject to the potentially catastrophic risk of a flight to quality in periods of banking instability, leading to a disastrous deflationary increase in the value of gold.
HT: J. P. Koning