Recent Arguments against the Gold Standard
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
No. 728 June 20, 2013
Recent Arguments against the Gold Standard
by Lawrence H. White
Executive Summary
The presidential primary contests of 2011– well-trained academic economist can describe
12 brought renewed attention to the idea of on the whiteboard an ideal monetary system that
reinstituting a gold standard. The 2012 Re- produces greater stability in the purchasing pow-
publican Party platform ultimately included a er of money than a gold standard does—or scores
plank calling for the creation of a commission higher on whatever one criterion the economist
to study the issue. favors—while sparing us a gold standard’s re-
The favorable attention given to the idea of source costs by employing fiat money. But other
reinstituting a gold standard has attracted criti- well-trained economists have proposed different
cism of the idea from a variety of sources. Con- criteria, and even a flawless central bank cannot
sidered here are the most important arguments pursue all criteria with one policy.
against the gold standard that have been made More important, fiat standards in practice
by economists and economic journalists in re- have been far from perfect monetary systems.
cent years. We need to examine historical evidence if we
A few recent arguments are novel to some ex- want to come to an informed judgment about
tent, but not all add weight to the case against whether actual gold-based systems or actual fiat-
a gold standard. Several authors identify genu- based systems display the smaller set of flaws.
ine historical problems that they blame on the I find that the most automatic and least man-
gold standard when they should instead blame aged kind of gold-based system—a gold stan-
central banks for having contravened the gold dard with free banking—can be expected to out-
standard. perform a gold standard with central banking
Gold standards, being real-world human in- and to outperform the kind of fiat monetary
stitutions, fall short of perfection. No doubt a systems that currently prevail.
Lawrence H. White is a professor of economics at George Mason University and an adjunct scholar with the
Cato Institute.We need Introduction in 2008 reappear in the recent literature.
to examine Other arguments are novel to some extent,
The presidential primary contests of but not all add weight to the case against a
historical 2011–12 brought renewed attention to the gold standard. Several authors identify gen-
evidence if we idea of reinstituting a gold standard. At least uine historical problems that they blame on
four candidates spoke favorably about the the gold standard, when they should instead
want to come gold standard. One suggested a “commission blame central banks for having contravened
to an informed on gold to look at the whole concept of how the gold standard.
judgment about do we get back to hard money.” The 2012 Re- Bernanke told the students at George
publican Party platform ultimately included Washington University, “Unfortunately gold
whether actual a plank calling for the creation of just such a standards are far from perfect monetary sys-
gold-based commission, explicitly viewing it as a sequel tems.”7 We can all agree that gold standards,
systems or to the U.S. Gold Commission of 1981: “Now, being real-world human institutions, fall
three decades later . . . , we propose a similar short of perfection. There is no doubt that a
actual fiat-based commission to investigate possible ways to well-trained academic economist can describe
systems display set a fixed value for the dollar.”1 on the whiteboard an ideal monetary system
The favorable attention given to the idea that, through the flawlessly timed and flaw-
the smaller set of of reinstituting a gold standard has attracted lessly calibrated policy actions of a central
flaws. criticism of the idea from a variety of sources. bank, produces greater stability in the pur-
In the popular press, Atlantic writer Matthew chasing power of money than a gold standard
O’Brien has expounded on “Why the Gold does—or scores higher on whatever one crite-
Standard Is the World’s Worst Economic rion the economist favors—while sparing us a
Idea,”2 while Washington Post columnist Ezra gold standard’s resource costs by employing
Klein has declared that “The problems with fiat (noncommodity) money.8 But other well-
the gold standard are legion.”3 On the more trained economists have proposed different
scholarly side, Federal Reserve Chairman and criteria, and even a flawless central bank can-
former Princeton economics professor Ben not pursue all criteria with one policy.
Bernanke, guest lecturing at George Wash- More important, fiat standards in prac-
ington University on the history of monetary tice have been far from perfect monetary sys-
policy in the United States, in the words of tems. We need to examine historical evidence
the New York Times’ account, “framed much if we want to come to an informed judgment
of this history as a critique of the gold stan- about whether actual gold-based systems or
dard, which was dropped in the early 1930s actual fiat-based systems display the smaller
in a decision that mainstream economists set of flaws. We need to recognize the variety
regard as obviously correct, hugely beneficial of institutional arrangements that the world
and essentially irreversible.”4 Well-known has seen under gold standards and likewise
University of California–Berkeley economist under fiat standards. In particular, we need
Barry Eichengreen has offered “A Critique of to distinguish an “automatic” gold-stan-
Pure Gold.”5 dard system—like the classical gold standard
In a Briefing Paper published by the Cato in countries without central banks—from
Institute, I addressed a number of then- the interwar gold-exchange system that was
common theoretical and historical objec- managed or mismanaged by the discretion
tions to a gold standard, sorting those that of central bankers. I find that the most auto-
have some substance from those that are ill- matic and least managed kind of gold-based
founded.6 Here I consider the most impor- system—a gold standard with free bank-
tant arguments against the gold standard ing—can be expected to outperform a gold
that have been made by economists and eco- standard with central banking, and to out-
nomic journalists since then. Some of the perform the kind of fiat monetary systems
less-substantial arguments that I criticized that currently prevail.
2What follows are critical analyses of the a recently realized round number), those
leading recent arguments against a gold holdings are worth $444.6 billion. Current
standard. I spell out each argument as crit- required bank reserves (as of October 2012)
ics have made it, and evaluate its logical are less than one fourth as large, $107.3 bil-
and historical merits. I begin with the least lion. Looked at another way, $444.6 billion
substantial arguments, and proceed to the is 18.4 percent of the current money supply
weightier. measure “M1” ($2,417.2 billion as of Oc-
tober 22), which is the sum of currency in
Claim 1: There Isn’t Enough Gold to circulation and checking-account balances.
Operate a Gold Standard Today That is a more than healthy reserve ratio by
Personal finance columnist John Wag- historical standards.11
goner recently claimed in USA Today that Waggoner labors under several miscon-
“there’s not enough gold in the world to re- ceptions. First, gold standards have histori-
turn to a gold standard.”9 He explained: cally required only fractional reserves—that
is, the holding of enough gold to back only
In the gold standard, the amount of a small portion of the money supply. So long
currency issued is tied to the govern- as banks or the government can satisfy the
ment’s gold holdings. The price of actual demand of conversion of money to
Gold standards
gold would have to soar to accommo- gold, fractional reserves do not make a gold have historically
date U.S. trade in goods and services. standard into a “kind of semi-gold stan- required only
. . . Total gold owned by the [United dard.” Second, it is not generally true that
States] government—including the “the amount of currency issued is tied to the fractional
Federal Reserve and the U.S. Mint— government’s gold holdings.” It is true only reserves—that is,
is 248 million ounces. That’s about if the government monopolizes the issue of
$405 billion dollars at today’s prices, gold-redeemable currency and the holding
the holding of
hardly enough to support a $15 tril- of gold reserves, but history offers 60-plus enough gold to
lion economy. examples of competitive private-note issue back only a small
under historical gold and silver standards.12
The government could use a kind Third, the vulnerability of the average reserve portion of the
of semi-gold standard, limiting the ratio to government manipulation is not in- money supply.
amount of money printed to a per- evitable. It can be avoided by leaving com-
centage of its gold reserves. For exam- mercial banks to determine their own reserve
ple, it could say that at least 40% of ratios, as in historical free banking systems.
all currency outstanding be backed
by gold. This would limit the money Claim 2: The Gold Standard Is an
supply, but be vulnerable to govern- Example of Price-fixing by Government
ment manipulation—revising the lim- Barry Eichengreen writes that countries
it downward to 5%, for example. using gold as money “fix its price in domes-
tic-currency terms (in the U.S. case, in dol-
Waggoner’s figures of 248 million ounces lars).” He finds this perplexing:
and $405 billion are approximately correct,
but his claim that the price of gold would But the idea that government should
have to soar to make that an adequate stock legislate the price of a particular
of gold reserves is not. The August 31st Sta- commodity, be it gold, milk or gaso-
tus Report of U.S. Treasury-Owned Gold puts the line, sits uneasily with conservative
U.S. government’s total holdings at 261.5 Republicanism’s commitment to let-
million ounces.10 (The source of Waggoner’s ting market forces work, much less
lower figure is unclear.) At a market price with Tea Party–esque libertarianism.
of $1,700 per fine troy ounces (to choose Surely a believer in the free market
3would argue that if there is an increase Claim 3: The Volatility of the Price of
in the demand for gold, whatever Gold Since 1971 Shows that Gold Would
the reason, then the price should be Be an Unstable Monetary Standard
allowed to rise, giving the gold-min- Eichengreen argues that “gold’s inherent
ing industry an incentive to produce price volatility” makes it unsuitable to “pro-
more, eventually bringing that price vide a basis for international commercial
back down. Thus, the notion that the and financial transactions on a twenty-first-
U.S. government should peg the price, century scale.”15
as in gold standards past, is curious at Klein declares, “The problems with the
the least.13 gold standard are legion, but the most obvi-
ous is that our currency fluctuates with the
To describe a gold standard as fixing gold’s global price of gold as opposed to the needs
price in terms of a distinct good, domestic of our economy.”16 It is not entirely clear
currency, is to begin with a confusion. A what “our currency fluctuates with the glob-
gold standard means that a standard mass al price of gold” means in this declaration. If
of gold (so many troy ounces of 24-karat it means that, for a country that is part of an
gold) defines the domestic currency unit. The international gold standard, the purchasing
currency unit (dollar) is nothing other than power of domestic currency moves with the
a unit of gold, not a separate good with a world purchasing power of gold, then it is
potentially fluctuating market price against true, but it fails to identify a problem. The
gold. That $1, defined as so many ounces world purchasing power of gold was better-
of gold, continues to be worth the specified behaved under the classical international
amount of gold—or, in other words, that x gold standard than the purchasing power of
units of gold continue to be worth x units fiat money has been since 1971. If it means
of gold—does not involve the pegging of to invoke the volatility of the real or dollar
any relative price. Domestic currency notes price of gold since gold was fully demone-
(and checking-account balances) are denomi- tized in 1971, it identifies a problem, but it
nated in and redeemable for gold, not priced is a problem experienced under a fiat stan-
in gold. They don’t have a price in gold any dard and not under a gold standard. Today,
more than checking account balances in our demonetized gold rises and falls in price as
current system, denominated in fiat dol- savers and investors rush into and out of
lars, have a price in fiat dollars. Presumably gold as a hedge against fiat-money inflation.
Eichengreen does not find it curious or ob- The respected University of California–
The world jectionable that his bank maintains a fixed San Diego economist and blogger James D.
purchasing dollar-for-dollar redemption rate, cash for Hamilton makes an argument that is less
checking balances, when he withdraws cash ambiguous, but puzzling nonetheless. Ham-
power of gold at its automatic teller machine. ilton charts how much the average dollar
was better- As to what a believer in the free market wage would have varied if it was initially fixed
would argue, surely Eichengreen under- in ounces of gold but instead was paid in the
behaved under stands that if there is an increase in the de- dollar equivalent as the price of gold varied
the classical mand for gold under a gold standard, what- between January 2000 and July 2012.17 He
international ever the reason, then the relative price of gold observes that “if the real value of gold had
(the purchasing power per unit of gold over changed as much as it has since then, the
gold standard other goods and services) will in fact rise, dollar wage that an average worker received
than the that this rise will in fact give the gold-min- would need to have fallen from $13.75/hour
purchasing power ing industry an incentive to produce more, in 2000 to $3.45/hour in 2012.” That sounds
and that the increase in gold output will in alarming, but in fact it is of very little signifi-
of fiat money has fact eventually bring the relative price back cance. It is relevant only if the behavior of the
been since 1971. down.14 “real value” (purchasing power) of gold is in-
4dependent of the monetary regime so that the fact: “It is true that the biggest concern I It makes little
purchasing power of gold-backed currency have about going back on a gold standard sense to attribute
would fluctuate on the world market. Such a today—that it would tie the monetary unit
calculation would be relevant if a small open of account to an object whose real value can volatility in the
economy (say, the Bahamas) should unilat- be quite volatile—was not the core problem real price of gold
erally adopt the gold standard today. That associated with the system of the 19th cen-
would indeed be a bad idea.18 But thoughtful tury.” He then continues: “But the fact that
to the growth in
advocates of the gold standard propose that this wasn’t the core problem with the gold demand from
it should again be an international standard. standard in the nineteenth century does not steadily rising
Hamilton’s calculation is completely irrel- mean that it wouldn’t be a big problem if we
evant to that proposal. A Lucas critique ap- tried to go back to the system in the twenty- incomes.
plies: observations drawn from a world of fiat first century.”20
regimes are not informative about the behav- But it’s unlikely that purchasing-power in-
ior of the purchasing power of money under stability would be any more of a problem for
an international gold standard. a present-day international gold standard.
Hamilton anticipates such an objection Hamilton attributes “recent movements in
and has a reply ready: the real value of gold” to “the surge in income
from the emerging economies rather than
[G]old advocates respond with the U.S. monetary policy,” citing data showing
claim that if the U.S. had been on a global gold jewelry sales up strongly in 2010
gold standard since 2000, then the over 2009, led by large increases in sales to In-
huge change in the real value of gold dia, Hong Kong, and mainland China.21 It is
that we observed over the last decade reasonable to suppose that demand for gold
never would have happened in the first jewelry rises with income. But real income
place. The first strange thing about this in India and China is rising fairly steadily. It
claim is its supposition that events and makes little sense to attribute volatility in the
policies within the U.S. are the most real price of gold to the growth in demand
important determinants of the real from steadily rising incomes.
value of gold. According to the World Hamilton’s drawing of a trend from two
Gold Council, North America accounts data points, moreover, is not a careful read-
for only 8% of global demand.19 ing of the data source he cites. Even if we
focus exclusively on 2010 over 2009, only a
This, too, is irrelevant to the evaluation of small fraction of the extraordinary increase
proposals for an international gold stan- of 69 percent in gold jewelry sales to India
dard. By the way, Hamilton’s 8 percent fig- can possibly be attributed to India’s real in-
ure is North America’s share of global pur- come growth, which was 10 percent that year
chases of new gold jewelry, a nonmonetary according to the International Monetary
and flow measure, rather than its share Fund. The income-elasticity of demand for
of the stock transactions demand to hold gold jewelry is nothing like 6.9 if we observe
monetary gold, which under an interna- longer-run trends. The text of the article con-
tional gold standard would presumably be taining the data provides a clue to the lion’s
closer to North America’s 30 percent share share of that one year’s increase: “Histori-
of world output. cally savvy gold buyers, India’s influx of buy-
The purchasing power of money was ing implies an expectation that gold prices
more stable under the classical interna- still have much higher to go. The [World
tional gold standard (1879–1914) than it Gold Council] says that ‘Indian consumers
has been under fiat money standards since appeared almost universally to expect that
1971. In a blog entry a few days after the the local gold price was likely to continue ris-
one just quoted, Hamilton recognizes this ing.’”22 That is, Indians did not buy so much
5gold jewelry in 2010 just for ornamentation, silver. Under a reliably anchored monetary
but also as an investment or inflation hedge. system this source of commodity price vola-
Likewise, the article notes, “many in China’s tility would disappear.
middle class are looking to gold as a means The answer to Cowen’s first question—
for long-term savings and a possible hedge why put your economy at the mercy of “es-
against inflation.” sentially random” supply and demand
If we look at additional years of the data, we shocks for gold?—is that, to judge by the
see that global gold jewelry sales in 2010 were historical evidence, doing so engenders less
down from the levels of 2007 or 2008, which volatility than the alternative of putting your
is hardly consistent with the hypothesis that economy at the mercy of a central bank’s
gold demand is rising mainly due to rising monetary policy committee. Monetary sup-
emerging-economy income. If we look at the ply and demand shocks under fiat money
article’s entire 2004–10 range of sales data for systems have been much larger. Under
gold in all forms, we see as much or more vola- the classical gold standard, changes in the
tility in investment sales of gold (bars, coins, growth rate of the base money stock were
medallions, exchange-traded funds) as in jew- relatively small—perhaps surprisingly small
elry sales. Absent fiat inflation hedging, there to those who haven’t looked at the numbers.
Inflation- is little cause for concern about the volatility The largest supply shock, the California Gold
hedging demand of demand for gold or gold’s real price. Rush, caused a cumulative world price level
is volatile Like Hamilton, the respected George Ma- rise of 26 percent (as measured by the United
son University economist and blogger Tyler Kingdom’s Retail Price Index) stretched over
because inflation Cowen23 also expresses concern about vola- 18 years (1849–67), which works out to an
expectations are tility in the real price of gold: inflation rate of only 1.3 percent per annum.
As Cowen recognizes, gold discoveries the
volatile under Why put your economy at the mercy size of California’s are hardly likely today.25
unanchored of these essentially random forces? I Barry Eichengreen also worries that vola-
monetary believe the 19th century was a rela- tility in the demand for gold would persist
tively good time to have had a gold even in an international gold standard:
systems. standard, but the last twenty years,
with their rising commodity prices, There could be violent fluctuations
would have been an especially bad in the price of gold were it to again
time. When it comes to the next twen- become the principal means of pay-
ty years, who knows? ment and store of value, since the
demand for it might change dramati-
In a later blog entry, Cowen adds, “I think a cally, whether owing to shifts in the
gold standard today would be much worse state of confidence or general econom-
than the 19th century gold standard, in part ic conditions. Alternatively, if the price
because commodity prices are currently of gold were fixed by law, as under
more volatile and may be for some time.”24 gold standards past, its purchasing
Cowen does not directly address the pos- power (that is, the general price level)
sibility that the current volatility of several would fluctuate violently.26
commodity price series, most importantly
that of gold, is principally caused by the The concern that Eichengreen expresses in
inflation-hedging prompted by our current his first sentence seems baseless. It would
fiat monetary systems. Inflation-hedging require a separation of monetary functions
demand is volatile because inflation expec- such that gold serves as the commonly ac-
tations are volatile under unanchored mon- cepted medium of exchange, but a unit of
etary systems. Inflation-hedging involves something else (what?) serves as the unit
other commodities in addition to gold and of account. Only under such a peculiar ar-
6Figure 1
Composite Price Index 1750 to 2003, January 1974 = 100 (logarithmic scale)
Logarithmic Scale
Source: Jim O’Donoghue, Louise Goulding, and Grahame Allen, “Consumer Price Inflation since 1750,” Office
for National Statistics [UK] Economic Trends 604 (March 2004): 38–46.
rangement could one ounce of monetary stabilizing the demand to hold money rela-
gold have a fluctuating price. In every his- tive to income (or stated inversely, it better
torically known system where gold or gold- stabilized velocity) than the fiat money sys-
redeemable claims were the principal means tem that followed it.27 He explained:
of payment, a specified amount of gold also
defined the pricing unit. Since the move in 1971 toward flex-
The concern Eichengreen expresses in ible exchange rates and the complete
his second sentence, that under a gold stan- divorce of United States monetary
dard dramatic shifts in the demand for gold management from the objective of
would result in “violently” fluctuating price a pegged gold price, it is clear that
levels, seems also to lack merit. The histori- the nominal anchor for the mone-
cal evidence shows that price levels during tary system—weak as it was earlier
the classical gold standard of 1821–1914 did [under Bretton Woods]—is now entire-
not fluctuate any more violently than the ly absent. Future monetary growth The classical
fiat money era post-1971. Figure 1 shows and long-run inflation appear now gold standard
price index movements in the United King- to depend entirely on the year-to-year
dom over 253 years under gold and paper “discretion” of the monetary author- constrained
sterling standards. ity, that is, the Federal Reserve. Not inflation in a
There is a good reason why the demand surprisingly, inflationary expectations
for monetary gold did not change dramati- and their reflection in nominal interest
more credible
cally under the classical gold standard. As rates and hence in short-run inflation way, better
Robert Barro noted 30 years ago, the clas- rates have all become more volatile. pinning down
sical gold standard constrained inflation in
a more credible way, thereby better pinning Volatility of inflation and expectations of inflationary
down inflationary expectations and better volatility of inflation did diminish during expectations.
7When the “Great Moderation” after the 1980s, leading economic historian—inconsistent
productivity but since 2006 they have returned. In the with the historical record of the gold stan-
14 years between August 1991 and August dard. First, as Eichengreen surely under-
growth allows 2005, the annual U.S. Consumer Price In- stands, the condition for the price level not
particular goods dex inflation rate (year-over-year, observed falling isn’t an unlikely or “magical” exact
monthly) stayed between 1 and 4 percent, equality (=) between the rate of growth in
to be produced at a band of just 3 percentage points. But be- the stock of monetary gold and the rate of
lower cost, those tween July 2008 and July 2009, the year-over- growth in the output of other goods and
goods become year inflation rate went from a high of 5.5 services (which proxies for demand to hold
percent to a low of minus 2.0 percent, a swing monetary gold for transactions), but rather
cheaper in both of 7.5 percentage points in a single year. It that the rate of growth in the stock of mone-
real and nominal has since risen as high as 3.9 percent. As tary gold is as at least as great (≥) as that of the
terms. long as the Fed retains discretion, inflation rate of growth of output. How rare was that?
expectations will remain variable. Not very. During the period of the classical
gold standard, given that the long-run aver-
Claim 4: A Gold Standard Would Be a age inflation rate was close to zero, this con-
Source of Harmful Secular Deflation dition was met about half of the time. The
“The most fundamental argument index numbers compiled by O’Donoghue,
against a gold standard,” writes Cowen, “is Goulding, and Allen in fact show a few more
that when the relative price of gold is go[ing] years of a rising, rather than a falling, price
up, that creates deflationary pressures on index during the 93 years from the United
the general price level, thereby harming out- Kingdom’s resumption of the gold standard
put and employment.”28 Eichengreen offers in 1821 to its departure in 1914.30 Over the
a similar criticism: period as a whole, the compound inflation
rate was one-tenth of 1 percent per annum.
As the economy grows, the price level It is true that if the output of goods and
will have to fall. The same amount of services grows too fast for the stock of mon-
gold-backed currency has to support etary gold to keep up, the price level falls.
a growing volume of transactions, In such an environment, when productivity
something it can do only if the prices growth allows particular goods to be pro-
are lower, unless the supply of new duced at lower cost, those goods become
gold by the mining industry magically cheaper in both real and nominal terms. 31
rises at the same rate as the output Such deflation, which results from rapid
of other goods and services. If not, growth in real output, can hardly be a cause
prices go down, and real interest rates for regret.
become higher. Investment becomes Eichengreen’s case for fearing deflation
more expensive, rendering job cre- under a gold standard overlooks the im-
ation more difficult all over again.29 portant historical findings of Atkeson and
Kehoe.32 Examining inflation rates and real
Eichengreen concludes: “The robust invest- output growth rates for 17 countries over
ment and job creation prized by the gold more than 100 years, they found that there
standard’s champions and the deflation is no link between deflation (falling prices)
they foresee are not easily reconciled, in oth- and depression (falling real output) outside
er words.” In a nutshell, he maintains that of one extraordinary episode, the Great De-
vigorous economic growth is at war with it- pression period of 1929–34. Their evidence
self under a gold standard because the mon- suggests to them that the Great Depression
ey stock won’t keep up. should be considered “a special experience
Eichengreen’s argument here is theo- with little to offer policymakers consider-
retically incorrect and—surprisingly from a ing a deflationary policy today.” Outside of
8the Great Depression, in their database “65 creating unsold inventories of goods, lead-
of 73 deflation episodes had no depression” ing to recessionary cutbacks in production
(and most of these deflations without de- and employment until prices and wages
pression “occurred under a gold standard”), decline sufficiently to clear the markets for
while 21 of 29 depressions occurred without goods, labor, and money balances (a classic
deflation. We consider the Great Depression discussion is provided by Yeager 1956.)36
in more detail below, but the Atkeson-Kehoe A good deflation involves no such un-
evidence makes it clear that the combination planned inventory accumulation, so it does
of rapid deflation and rapid output shrink- not depress output. In terms of the standard
age of 1930–33, which occurred under the equation of exchange, MV = Py, a good defla-
interwar system managed (or mismanaged) tion has the price level P falling contempora-
by central banks, was unlike experience un- neously with real income y rising. A bad de-
der the much milder deflations of the classi- flation has P falling with a lag (and y falling
cal gold standard. in the interim) behind a shrinking money
We need to recognize the basic distinction, stock M or shrinking velocity of money V.
which applies under any monetary standard, Bad deflation was a major problem in the
between a good deflation and a bad deflation. early 1930s, as a series of banking panics
Selgin,33 Atkeson and Kehoe,34 and Bordo, led to the hoarding of currency by the pub-
We need to
Landon-Lane, and Redish35 have made this lic and the stockpiling of reserves by banks recognize the
distinction conspicuously clear, but Eichen- (events that can be described either as a fall basic distinction,
green neglects it, as does Bernanke routinely. in the velocity of base money or a fall in the
In brief, a good deflation is a situation where quantity of broader money). It was briefly a which applies
the price level falls because output grows problem during the pre-Fed banking panics under any
more rapidly than the money stock. It is a in the United States. But banking panics are
situation of ongoing approximate monetary not caused by being on a gold standard (see
monetary
equilibrium, involving no significant excess Claim 6 below). standard,
demand for money and therefore no sig- The nonconflict between deflation and between a good
nificant excess supply of goods at any date’s robust growth is evident during the most
price level. Prices fall one by one as the selling extended deflationary period under the clas- deflation and a
prices of particular goods follow their costs sical gold standard in the United States, bad deflation.
of production downward. Real living stan- the 15 years from 1882 to 1897. The Gross
dards rise as goods become cheaper. A defla- Domestic Product deflator (as constructed
tion driven by real growth does not make real by Romer 1989), which is a measure of the
growth more difficult to sustain. price level, fell from 8.267 to 6.383, a com-
A bad deflation, in a world with some de- pound inflation rate of approximately –1.7
gree of downward price and wage stickiness, percent per annum.37 Over the same period,
is a situation where prices fall as a lagged real GDP grew at the healthy rate of ap-
response to an unexpected shrinkage in the proximately 3.0 percent per annum. Robust
money stock or a spike in money demand. investment and real income growth were
(The degree of price and wage stickiness is easily reconciled with deflation. The similar
lower in a system where the expected infla- experience in Britain during the same period
tion rate is lower, but stickiness was not zero has sometimes been called a “great depres-
even under the classical gold standard when sion,” but use of that label confuses defla-
the long-run expected inflation rate was tion, which did happen, with falling output,
near zero.) Such shocks create a monetary which did not.38
disequilibrium, an unsatisfied demand to The same confusion is evident when po-
hold money at the existing price level. Con- litical commentator Bruce Bartlett writes
sumers and businesses cut their spending that “while a gold standard provided sta-
for the sake of adding to money balances, ble purchasing power over long periods of
9time, that was only because inflations were percent to keep the anticipated real interest
subsequently offset with debilitating de- rate constant. Therefore an anticipated defla-
flations.”39 In fact, as the 1882–97 period tion has no effect on the cost of investment. A de-
shows, and as Atkeson and Kehoe show cline in the price level greater than anticipat-
more generally, deflations under the clas- ed over the period of a loan does raise the ex
sical gold standard were not debilitating.40 post real interest rate paid on the loan. But
That is, they were not associated with falling such an unanticipated decline, occurring af-
output. Bartlett is mistaken in thinking that, ter an investment loan was taken out, does
as a consequence of deflation, “there were not raise the interest rate at the time of the
greater economic instabilities, higher unem- loan contract, and thus cannot make invest-
ployment and longer recessions during the ment more expensive.
gold-standard era.” Despite a weak banking To be fair, Eichengreen may have had in
system, the record of the gold-standard era mind (and simply neglected to specify) the
before 1914 in the United States does not in one atypical set of conditions where his ar-
fact show greater economic instabilities or gument would apply. Namely, if the nomi-
longer recessions than the post–World War nal interest rate is already near or at the zero
II era.41 lower bound, then the nominal rate cannot fall
Atkeson and Kehoe also address specifi- enough in response to a large downward shift
cally the case of slow-growing Japan in re- in the anticipated inflation rate to keep the
cent decades, which has often been cited as ex ante real interest constant. The ex ante real
evidence of the depressing effect of falling or interest rate then does rise. This was a prob-
negative inflation.42 They show that Japan’s lem during the extreme deflation of 1930–
growth rate began falling around 1960, 32; three-month Treasury rates fell close to
while its inflation rate began falling around zero at the end of 1932. Below I argue that
1970, suggesting that the former is a secular this deflation—under the Federal Reserve’s
trend independent of the latter. They aptly watch—was not due to the gold standard, but
comment: “Attributing this 40-year slow- due to its contravention. The zero low bound
down to monetary forces is a stretch.”43 may be a problem today under the Federal
Returning to the quotation from Eichen- Reserve’s deliberate policy of ultralow short-
green, let us consider his claim that when term interest rates. During the period of the
prices go down “real interest rates become classical gold standard, there were no cases of
higher” with the result that “[i]nvestment an anticipated deflation so great as to bring
becomes more expensive, rendering job the nominal interest rate close to zero or cre-
creation more difficult.”44 The statement ate a lower-bound problem.
unfortunately fails to keep straight the stan-
dard distinction between two kinds of real Claim 5: A Gold Standard too Rigidly
interest rates, ex ante (anticipated) and ex Ties the Government’s Hands
post (retrospective). The identity that de- One of the slides for Ben Bernanke’s lec-
fines a real interest rate is: (1 + real interest ture at GWU reads as follows:45
rate) = (1 + nominal interest rate) ÷ (1 + infla-
Deflations under tion rate). The inflation rate in question can The strength of a gold standard is its
the classical gold either be an anticipated rate or a rate mea- greatest weakness too: Because the
sured retrospectively. Correspondingly, the money supply is determined by the
standard were not derived real interest rate can either be antici- supply of gold, it cannot be adjusted
debilitating. That pated or retrospective. The standard theory in response to changing economic
is, they were not of the Fisher Effect tells us that when (say) conditions.
a drop to minus 1 percent from 0 percent
associated with annual inflation is anticipated, the nominal Note the passive wording: be adjusted. Adjust-
falling output. interest rate also drops by approximately 1 ed by whom or by what? On a previous slide
10Bernanke indicated that he was assuming an Federal Reserve carrying its own weight, suc- The historical
automatic gold standard, without a central cessfully adjusting the money supply to con- record does not
bank able to do any significant adjusting of ditions.47 That is, the Fed has not reduced
the money supply. But under a gold stan- cyclical volatility in the economy. show the Federal
dard, a change in the money supply can also Bernanke apparently thinks that mar- Reserve carrying
be brought about by market forces. Under a ket determination of the money supply is a
gold standard, market forces in gold mining, weakness because it eliminates the option to
its own weight,
minting, and banking do adjust the money use monetary policy to reduce the unemploy- successfully
supply in response to changing economic ment rate (or in economists’ jargon, rules out adjusting the
conditions, that is, in response to changes exploiting the short-run Phillips Curve). Ac-
in the demand to hold monetary gold or cording to the New York Times account of his money supply to
to hold bank-issued money. The supply of GWU lecture, Bernanke told the class that conditions.
bank-issued money is not determined by the being on the gold standard “means swearing
supply of gold alone. If such a market-driven that no matter how bad unemployment gets
change counts as the supply being adjust- you are not going to do anything about it.”
ed—and why shouldn’t it?—then Bernanke’s True, an automatic gold standard does elimi-
statement is false. The money supply does nate the option to respond to the unemploy-
adjust in response to changing economic ment rate. But that is a feature, not a bug.
conditions.46 Any economist who takes to heart the case
But perhaps the Bernanke slide’s phrase that Kydland and Prescott have made for the
“cannot be adjusted” only intends to say benefit of rules over discretion in monetary
that under a fully decentralized and auto- policy will recognize that such a restraint is a
matic gold standard there is no central mon- strength rather than a weakness.48
etary policy committee or other small group When job seekers recognize the central
of people who can deliberately adjust the ag- bank’s intention to use monetary expansion
gregate money supply. Under that reading to reduce unemployment, they will raise
the statement is true. But read that way the their inflation-rate expectations and thus
statement does not deny that market forces their reservation wage demands. Monetary
will adjust the money supply appropriately. expansion will then only ratify their ex-
Bernanke neglects to provide a compara- pectations, not surprise them, and thereby
tive analysis here. One might, with equal or will achieve only higher inflation and no
greater justice, invert his statement and say, reduction in the unemployment rate. Just
“The strength of a fiat standard is its great- as Ulysses strengthened his ability to sail
est weakness too: because the money supply home, past the island of the Sirens, by tying
is not automatically determined by market himself to the mast and plugging his helms-
forces but by the discretion of a committee, man’s ears with wax, so too a monetary sys-
it can change in ways that are inappropri- tem strengthens its ability to achieve the
ate to changing economic conditions.” The good outcome it can achieve by foreswearing
comparative historical question remains: un- other goals. Kydland and Prescott identify
der which system—automatic adjustment by the goal as zero inflation, but more gener-
market forces under a gold standard or de- ally the goal is to facilitate trade—including
liberate adjustment by central bankers on a intertemporal trade—most efficiently.
fiat standard—is the money supply better ad-
justed to economic conditions? Those who Claim 6: A Gold Standard Amplifies
understand why central economic planning Business Cycles (or Fails to Dampen
generally fails should presume that market them as a Well-managed Fiat Money
guidance works better, absent a persuasive System Does)
rebuttal showing that money is an excep- In response to my 2008 piece, Tyler Cow-
tion. The historical record does not show the en wrote:49
11My main worry with the gold stan- bilities (Canada allowed nationwide branch-
dard is simply the pro-cyclicality of ing), and the rules (originally imposed to
the money supply. . . . For instance help finance federal expenditures in the Civil
would you really want a contract- War) requiring note-issuing banks to hold
ing money supply in today’s envi- federal bonds as collateral (no such rules op-
ronment? And yes credit crunches of erated in Canada). The banknote restriction
this kind happen in market settings prevented banks from issuing more notes
too so you can’t blame it all on Alan during seasons of peak currency demand,
Greenspan. which in turn led to reserve drains every
autumn (not seen in Canada). Because pan-
Cowen’s worry here does not appear to be ics are not inherent to a gold standard, but
about the pro-cyclicality of the gold sup- rather to a banking system weakened by legal
ply. Gold mining is actually countercyclical restrictions, the pre-1933 panics do not in-
with respect to the price level: that is, a fall- dict the gold standard, but rather indict legal
ing price level denominated in gold units restrictions that weaken banks. While Ber-
raises the purchasing power of gold and nanke was correct to say in his lecture that
so increases global mining output. For any “The gold standard did not prevent frequent
The U.S. banking single economic region, the price-specie-flow financial panics,” neither did it cause them.51
panics, both mechanism is likewise countercyclical with Financial Times columnist Martin Wolf
under the pre-Fed respect to the price level, meaning a falling expresses a worry similar to Cowen’s, that
local price level attracts gold from the rest of a gold standard with fractional-reserve
system and in the the world. Cowen instead appears to worry banking is inherently pro-cyclical: “In good
1930s, came from about the supposed pro-cyclicality of bank- times, credit, deposit money and the ratio
issued money (deposits and banknotes) as a of deposit money to the monetary base ex-
legal restrictions result of bank runs and credit crunches. He pands. In bad times, this pyramid collapses.
that weakened worries that the banking system is prone to The result is financial crises, as happened re-
the banking contract its liabilities in a downturn, and peatedly in the 19th century.”52 In fact, free
thereby to amplify the economy’s contrac- banks did not exhibit exuberant swings in
system, not from tion. their reserve ratios. 53 Less-regulated bank-
the United States The inside money supply does fall in ing systems were more robust than Wolf
being on the gold a banking panic if there are runs for base suspects, as seen not only in Canada but
money, whether that base money is metallic also in Scotland, Sweden, Switzerland, and
standard. or fiat.50 But it is not true that a gold stan- other systems without central banks under
dard or free banking makes the banking sys- the gold standard. Repeated financial crises
tem prone to bank runs and credit crunches. were a feature of the 19th-century banking
The U.S. banking panics, both under the systems in the United States and England,
pre-Fed system and in the 1930s, came from weakened as they were by legal restrictions,
legal restrictions that weakened the banking but not of the less restricted systems else-
system, not from the United States being on where.54
the gold standard. Comparing the United
States to Canada illustrates this strikingly. Claim 7: The Gold Standard Was
Canada was equally on the gold standard, Responsible for the Deflation that
and had a similar agricultural economy, but Ushered in the Great Depression in the
experienced no panics. Its banking system United States
was far less restricted and consequently far The most prominent set of criticisms of
stronger. The most important legal restric- the gold standard among academic econo-
tions on U.S. banks were the prohibition of mists in recent years blames the gold stan-
interstate branching, which would have al- dard for creating the Great Depression in
lowed better diversification of assets and lia- the United States and for then spreading
12it internationally. Douglas Irwin summa- units much higher than before the war, and
rizes the case and identifies its most cited much higher than postwar price levels mea-
source:55 sured in gold units. As Robert Mundell
noted in his Nobel lecture, large volumes of
Modern scholarship regards the De- European gold flowed to the United States,
pression as an international phenom- which continuously remained on gold (al-
enon, rather than as something that though the federal government embargoed
affected different countries in isola- gold exports in 1917–19).59 The gold inflow
tion. The thread that bound countries substantially raised the U.S. dollar price level
together in the economic collapse was during the war. Despite a major correction in
the gold standard. Barry Eichengreen’s 1920–21, “the dollar (and gold) price level”
1992 book Golden Fetters is most com- remained 40 percent above “the prewar equi-
monly associated with the view that librium, a level at which the Federal Reserve
the gold standard was the key factor kept it until 1929.”60 For the United States,
in the origins and transmission of the this meant that the price level would eventu-
Great Depression around the world.56 ally have to fall.
Meanwhile in Europe, wartime money
The piece of evidence most often cited for printing had pushed the price levels in the
this view is “[t]he fact that countries not United Kingdom, France, and other coun-
on the gold standard managed to avoid the tries much higher than 40 percent above
Great Depression, while countries on the their prewar levels. For the United Kingdom
gold standard did not begin to recover until and France to return to the gold standard
they left it.”57 (that is, to reinstitute convertibility at a de-
This section addresses the “factor in the fined parity between the domestic monetary
origins” charge. The next section addresses unit and gold), even without further U.S. de-
the “transmission” charge. flation, would require some combination of
James D. Hamilton argues that “between devaluation and deflation. Mundell points
1929 and 1933, the U.S. and much of the rest out that some notable staunch defenders of
of the world were on a gold standard. That the gold standard, such as Charles Rist and
did not prevent (indeed, I have argued it was Ludwig von Mises, saw devaluation as a more
an important cause of) a big increase in the prudent option than a painfully large defla-
real value of gold over that period. Because tion. Mises is reported to have criticized the
the price of gold was fixed at a dollar price recommendation that a deflation should be
of $20/ounce, the increase in the real value undertaken to reverse the effects of wartime
of gold required a huge drop in U.S. nomi- inflation by remarking that, once you have
nal wages over those years.”58 Because wages run a man over with a truck, you do him
were sticky downward, the drop in nominal no favor by putting the truck in reverse and
demand for labor created a massive loss of driving over him in the other direction.
employment. France chose to adjust the franc’s gold
To understand the deflation of 1930–32, content downward (to devalue) fully in pro-
we need to review the deflation of the inter- portion to its lost purchasing power, which To understand
war period as a whole. And to understand enabled them to keep the postwar franc the deflation of
the interwar deflation as a whole, we need price level. The United Kingdom and most
to review the monetary events of World War other countries chose to restore the prewar
1930–32, we need
I. During the war, the major combatant na- gold content to the monetary unit, which to review the
tions suspended the gold standard in order forced a major downward adjustment in the deflation of the
to print copious amounts of money to fi- price level to reverse most of the wartime in-
nance war expenditures. At war’s end they flation. As Mundell put it, “The deflation of interwar period
were left with price levels in local currency the 1930s was the mirror image of the war- as a whole.
13The global time rise in the price level that had not been occurred sooner had the Fed not increased
deflation of the reversed in the 1920–21 recession.”61 Ma- its expansionary efforts from June 1927 to
zumder and Wood detail the economic logic December 1928. The Fed finally tightened
interwar period of this reversal in an important recent pa- credit in early 1929 to moderate the rapid
was not due per, and show how the movement of prices rise in stock market share prices.
parallels the pattern seen in resumptions of In the view famously spelled out by Mil-
to the world’s the gold standard at the old parity following ton Friedman and Anna J. Schwartz in their
being on the gold previous wartime inflations.62 A Monetary History of the United States,66 what
standard. The global deflation of the interwar “might have been a garden-variety recession,
period, in other words, was not due to the though perhaps a fairly severe one,” became
world’s being on the gold standard. It was the Great Depression when bank runs were
due to many countries leaving the gold stan- allowed to shrink the broader money supply
dard, inflating massively while off the gold dramatically.67 The Fed stood idly by, not
standard, and then resuming the gold stan- trying to counter the shrinkage, while “the
dard at the old parity (not devaluing to accom- stock of money fell by over a third” between
modate the inflated price level). August 1929 and March 1933.68 The result-
Attempts to reduce the demand for mon- ing inflation rates in 1930, 1931, and 1932
etary gold through international coordina- were deeply negative: –6.4, –9.3, and –10.3
tion among central banks came to naught. percent, respectively.
The Federal Reserve System, and especially In Golden Fetters, Eichengreen charges
the Bank of France, absorbed large amounts that “the gold standard was responsible
of gold by sterilizing inflows to block the rise for the failure of monetary and fiscal au-
in prices that otherwise makes a region’s in- thorities to take offsetting action once the
flow self-limiting.63 They were not acting in Depression was underway.”69 More specifi-
accordance with the gold standard. Rather, as cally, he claims that the gold standard “was
Ben Bernanke puts it, “in defiance of the so- the binding constraint preventing policy-
called rules of the game of the international makers from averting the failures of banks
gold standard, neither country allowed the and containing the spread of financial
higher gold reserves to feed through to their panic.”70 Friedman and Schwartz, however,
domestic money supplies and price levels.”64 had already provided some evidence to the
The U.S. recession that became the Great contrary. They showed that the Fed during
Depression, according to the National Bu- this period was not obeying the dictates of
reau of Economic Research business-cycle the gold standard, but was in fact violating
chronology, began once the previous busi- them by sterilizing gold inflows.71 The U.S.
ness expansion ended in August 1929. Pric- gold stock rose in 1931 and again in 1932,
es began to fall three months later. Monthly but the Fed prevented bank reserves and the
data show the consumer price index rising money supply from expanding and thereby
up until November 1929, with December prevented a moderation of the downward
the first month of decline. The arrival of de- pressure on prices and output. If not the gold
flation cannot then have been the initiating standard, what stopped the Fed from ex-
cause for the expansion turning into reces- panding? Most plausibly, to judge by its own
sion. Better explanations for why the boom pronouncements at the time, we can blame
did not continue are beyond our subject the Federal Reserve Board’s adherence to a
matter here, but some contemporary observ- now-discarded credit policy doctrine known
ers, such as F. A. Hayek, argued that the Fed as the Real Bills Doctrine, which held that
had amplified the boom to an unsustain- the issuance of short-term, self-liquidating
able degree by deliberately expanding credit loans would ensure that the created money
to keep wholesale prices from falling.65 In would go to real goods, and thus the lending
Hayek’s view, a milder downturn would have would be non-inflationary.72
14Eichengreen acknowledges that the Fed pression spread across the world via the fixed
had “extensive gold reserves,” but none- exchange rate gold standard.”76 In Eichen-
theless maintains that it “had very limited green’s earlier words, the international gold
room to maneuver.”73 A more recent study standard “transmitted the destabilizing im-
coauthored by Anna J. Schwartz, Michael D. pulse from the United States to the rest of
Bordo, and Ehsan U. Choudhri provides ad- the world.”77 This description of events has
ditional evidence that, in fact, the Fed had some truth to it, but is misleadingly incom-
more than enough spare gold reserves (in plete. The destabilizing impulse, as empha-
excess of its legally mandated gold cover re- sized in the previous section, came from the
quirements) to offset the contraction of the Federal Reserve and Bank of France steriliz-
broad money supply and thereby offset the ing gold inflows and thereby absorbing ever-
downward pressure on real output.74 They greater amounts of gold. “These policies,” as
summarize their findings as follows:75 Bernanke has noted, and not the gold stan-
dard as such, “created deflationary pressures
[T]he United States, . . . holding mas- in deficit countries that were losing gold.”78
sive gold reserves . . . , was not con- Even more important, as discussed above,
strained from using expansionary counties such as the United Kingdom were
policy to offset banking panics, defla- already headed for deflation once they decid- The interwar
tion, and declining economic activ- ed to return to the gold standard at their pre- period shows
ity. Simulations, based on a model of war parities while their price levels were well us a case where
a large open economy, indicate that above their prewar (and equilibrium) levels.
expansionary open market operations The interwar period shows us a case where central banks—
by the Federal Reserve at two critical central banks—not the gold standard—ran not the gold
junctures (October 1930 to February the show. To put it mildly, they failed to run
1931; September 1931 through January it as well as the classical gold standard. As
standard—ran
1932) would have been successful Richard H. Timberlake has emphasized, it the show.
in averting the banking panics that is illogical to blame the international gold
occurred, without endangering convert- standard for the interwar disaster.79 The
ibility [through losses of gold reserves]. international gold standard worked well in
Indeed had expansionary open market the prewar period, when central banks were
purchases been conducted in 1930, the less active in trying to manage gold flows
contraction would not have led to the (and in many countries, such as the United
international crises that followed. States and Canada, did not yet exist). Blame
for the unfortunate results of the interwar
Specifically they find that, under a simulated system rests instead on decisions to resume
program of large open-market purchases to the gold standard at the old parity and on
offset the contraction of the broader money the discretionary policies of central bankers.
supply, “U.S. gold reserves would have de- The illogic is compounded when the failure
clined significantly but not sufficiently to of the discretionary interwar central bank-
reduce the gold ratio below the statutory ing system is taken to provide evidence in
minimum requirement.” support of giving central banks more discre-
tion than they have under an automatic in-
Claim 8: The Gold Standard Was ternational gold standard.
Responsible for Spreading the Great The interwar experience does carry a les-
Depression from the United States to the son for advocates of reinstating an interna-
Rest of the World tional gold standard. It indicates that the in-
The second part of the “Golden Fetters” ternational gold standard works best when
indictment, to quote a recent statement of it works most automatically. A valid point
it by Michael Bordo, is that “The Great De- is therefore made by Bernanke’s lecture
15You can also read
