When the north last headed south: revisiting the 1930s.
Reinhart, Carmen M. ; Reinhart, Vincent R.
ABSTRACT The U.S. recession of 2007-09 is unique in the post-World
War II experience in the broad company it kept. Activity contracted
around the world, with the advanced economies of the North experiencing
declines in spending more typical of the developing economies of the
South for the first time since the 1930s. This paper examines the role
of policy in fostering recovery in that earlier decade. With nominal
short-term interest rates already near zero, monetary policy in most
countries took the unconventional step of delinking currencies from the
gold standard. However, analysis of a sample that includes developing
countries shows that this was not as universally effective as often
claimed, perhaps because the exit from gold was uncoordinated in time,
scale, and scope and, in many countries, failed to bring about a
substantial depreciation against the dollar. Fiscal policy was also
active--most countries sharply increased government spending--but was
prone to reversals that may have undermined confidence. Countries that
more consistently kept spending high tended to recover more quickly.
**********
The financial and economic dislocations of the past two years have
been sharp and widespread. Yet there is ample precedent for such
crises--and for the economic adjustment that follows to be wrenching.
Among the advanced economies, those earlier crises occurred either
before World War II or in open economies that were out of sync with the
global cycle. (1) Crashes and severe contractions have been more common
in emerging market economies. In the current episode, however, activity
collapsed in unison in developed and developing countries around the
world. Indeed, the rarity of current circumstances is why we rely on an
event three-quarters of a century old, the Great Depression, as the main
comparator. (2)
Given the importance of that precedent in understanding the current
contraction, it is useful to cast a sharp focus on the role that policy
actions played in shaping recovery in the 1930s. Unconventional monetary
policy action has been called (Svensson 2003) a "foolproof
way" of preventing deflation, especially in an open economy that
can generate additional demand through depreciation of its currency. But
when the global pie is shrinking, such action may be less effective. In
the 1930s, moving off the gold standard bought fiscal authorities in
many countries more space for stimulus because their central banks had
room on their balance sheets to purchase more government securities and
to generate additional income. It also allowed each country to devalue
relative to gold. (3) Those actions, however, were mostly uncoordinated
in time, scale, and scope. As a consequence, the record of success among
countries abandoning the gold standard, both in avoiding a severe
contraction and in speeding the recovery, is quite mixed. The 1930s also
saw massive increases in government spending in many countries, but
fiscal authorities were prone to reverse themselves. As a result, some
of the direct benefits of that spending were offset by harmful effects
stemming from its volatility.
I. The Appropriate Precedent
The Business Cycle Dating Committee of the National Bureau of
Economic Research has put the peak of the current U.S. cycle at the end
of 2007. There is no equivalent formalism at the world level, but
indicators for most other countries started turning down about six
months later, consistent with the view that the United States led the
way down. Robert Barro and Jose Ursua (2008) have demonstrated that
occasional large, adverse shocks hit national economies without the
reason for those shocks always being clear. The current episode is
particularly unusual because so many economies around the world
contracted simultaneously.
Table 1 provides a historical perspective on the rarity of events
like those of recent years, by documenting changes in real exports
during past systemic crises from 1890 to today, for samples ranging from
35 to 111 countries. The episodes included in the table are those that
saw spikes in the number of banking crises worldwide, as reported by
Carmen Reinhart and Kenneth Rogoff (2009). As is evident from the table,
it is not unprecedented for a majority of countries to experience
declines in real exports coincident with systemic financial crises. Many
of the median changes listed in the last column are negative, and the
largest declines (which the preceding column reports for each country)
are quite large indeed. The scale of the most recent experience,
however, has only one precedent, namely, the early 1930s: more than
four-fifths of countries in both periods saw contractions in exports of
greater than 15 percent. The scope of the problem also distinguishes the
Great Depression and the current, second "Great Contraction":
only in those two episodes did virtually all of the nations of the world
witness shrinking trade flows. No other crisis period in the past
century matches that experience.
The commonality of the experience in these two episodes makes an
examination of the setting of policy in the 1930s relevant for
consideration today. We consider the actions of the monetary and those
of the fiscal authorities in turn.
II. Monetary and Exchange Rate Policy during the Great Depression
The painful adjustment in activity around the world during the
early 1930s strained the confidence of many public officials in the
speed with and extent to which the market system would correct itself.
As a consequence, the range of policy response was wide. The major form
that monetary policy experimentation took was to expand central bank
balance sheets by lowering the gold content of the home currency. As
will be discussed below, countries devalued relative to gold at
different points over the decade and by different amounts. The mechanism
through which this proved expansionary can best be understood by
considering a single country' s experience.
In the United States, the key decision in the early 1930s that
shifted the stance of monetary policy decisively toward ease was not
made by the nation's principal monetary authority, the Federal
Reserve. Rather, it was the devaluation of the dollar in terms of gold
by newly inaugurated President Franklin Roosevelt, (4) followed by a
sharp increase in gold inflows as a result of political instability in
Europe, that produced a marked relaxation of monetary conditions,
through a large increase in high-powered money. (5) As shown in figure
1, high-powered money in the United States (essentially, currency in
circulation, vault cash, and bank deposits with the Federal Reserve)
increased by 60 percent from March 1933 to May 1937 (the trough and
peak, respectively, of the business cycle); the M1 measure of the money
supply expanded by about the same amount from 1933 to 1937. Short-term
nominal interest rates, proxied in the bottom panel by the three-month
Treasury bill rate, were already close to zero. Thus, in the decade from
1932 onward, policy impetus cannot be measured by reference only to the
level of the short-term interest rate.
[FIGURE 1 OMITTED]
Then as now, the size and composition of the monetary
authority's balance sheet had the potential to influence financial
markets and the economy. An enlarged balance sheet also provided fiscal
authorities more space to be aggressive, if they felt so inclined. (6)
All this meets the definition of "unconventional" monetary
policy and quantitative easing (as in Bernanke and Reinhart 2004). In
the standard rendering, there were three acts to this episode of
quantitative easing.
In the first act, in 1932, U.S. policymakers extended their mistake
of the prior three years of not addressing a crisis of confidence. After
the stock market crash of 1929, the public sought to build up a cushion
of safe assets. For households, this meant holding more currency; for
banks, the demand for reserves rose. Declines in asset values and
increased demand for liquidity strained the financial system, leading to
a daisy chain of bank failures, which further heightened demand for safe
assets. (7) High-powered money did expand, but by too little to offset
increases in desired currency and reserve holdings, as detailed by
Milton Friedman and Anna Schwartz (1963) and by Philip Cagan (1965). (8)
In the second act, President Roosevelt's decision to devalue
relative to gold in 1933 triggered an expansion in the monetary
authority's balance sheet and sent a clear signal of the intent to
reflate. (9) In the final act, policymakers repeated their initial
mistake and contracted policy by sterilizing gold inflows in 1936 and
increasing reserve requirements in 1937, stalling the expansion of
high-powered money. (10) This third act highlights the danger of a
premature exit from policy accommodation, as Christina Romer has
recently pointed out. (11) It is the middle act, the move off the gold
standard, that has been most widely praised and that offers the best
evidence that unconventional policy action can spur recovery. But
although the U.S. experience can be interpreted that way, the wider
international record is more mixed.
Devaluing a nation's currency in terms of gold has three
distinct effects. (12) First, the home-currency value of the monetary
authority's resources expands. If, as in the U.S. case in 1933,
short-term interest rates are near the zero bound, this amounts to
unconventional monetary policy. Second, if other countries remain at an
unchanged gold parity (or devalue by less than the home country), the
exports of the home country become priced more competitively on world
markets. Third, devaluation might be interpreted as a regime switch,
signaling higher inflation in the future and therefore working to lower
real interest rates immediately.
Table 2 gives a year-by-year chronology of countries' exits
from the gold standard during the 1930s, along with some information
about the course of economic contraction and recovery in each country.
The first column reports the year that output peaked--usually 1928 or
1929. The second column reports the peak-to-trough decline in real
output. This was, indeed, a wrenching contraction, with the 29 percent
decline in the United States among the worst. Small open economies that
were reliant on commodity production, such as Chile, Nicaragua, and
Uruguay, were hit especially hard. Closed economies, such as Italy and
Portugal, in contrast, fared better.
The last column in the table provides a metric for recovery: the
number of years it took for output to return to the previous peak. This
seems an intuitive way to measure a downturn, but it is also quite
conservative. Ongoing expansion in potential output implies that a
return to prerecession output is not synonymous with an elimination of
economic slack. What is striking in this column is how varied was the
experience and how long was the typical path to recovery.
In the event, abandoning the gold standard was not a foolproof
solution for economic recovery. Figure 2 plots for each country in table
2 the peak-to-trough decline in real GDP against the number of years it
took after 1929 for the country to devalue or leave the gold standard.
There is no obvious association between the timing of the devaluation
and the severity of the downturn. Early leavers (those in 1929 and 1930)
experienced output contractions ranging from 13 to 36 percent. Late
exiters (from 1933 onward) suffered output declines from 6 to 32
percent.
In his work with different coauthors on the interwar gold standard,
Barry Eichengreen has argued that devaluation against gold was an engine
of reflation. (13) The simple scatterplot in figure 2 suggests that the
benefits were not always evident. But the figure implicitly differs from
the prior literature in three ways: the choice of the measure of
activity, the window of observation, and the country coverage; these
differences can be addressed systematically.
[FIGURE 2 OMITTED]
Table 3 reports regressions that seek to explain various measures
of economic recovery in the 1930s for different sets of countries. The
first column reproduces what might be called "exhibit A" for
those arguing that the change in the exchange rate regime was crucial in
fostering economic recovery. That column, following the literature,
relies on information collected in real time by the League of Nations.
The change in industrial production from 1929 to 1937 for the 19
countries for which data are available from that source is regressed
against the number of years after 1929 that the country exited the gold
standard. For these 19 countries, leaving the gold standard had a
statistically significant effect on industrial production over that
common time period. Indeed, the coefficient on the timing variable is
quite large. Leaving the gold standard at the start rather than at the
end of the period prevented a decline in industrial output of more than
one-half.
It might be argued that the more pronounced effect on industrial
production in part reflects the higher cyclical amplitude of this
narrower slice of economic activity. It could also be due to the greater
dependence of manufacturing on international trade. The second column
therefore repeats the exercise for the same countries but uses the
change in real GDP per capita in place of the change in industrial
production. The results, although smaller, remain statistically
significant and quantitatively important for this broader measure of
activity. According to this estimate, delaying the exit from the
beginning to the end of this fixed window is associated with about a 20
percent loss in real GDP per capita.
The previous literature's use of the League of Nations sample
puts particular weight on the experience of large countries and of
countries in Europe. The third column therefore broadens the sample to
include 39 countries, including many in Latin America. Although the
coefficient on the timing variable remains negative, it is no longer
statistically significant. Thus, some of the purported benefits of the
1930s regime switch are apparently sensitive to the country set.
In addition, because countries left the gold standard at different
times, the literature's use of a single time period to measure
recovery across that experience might be inappropriate. An alternative
is to determine the width of the observation window country by country.
We ran regressions using the date of exit from the gold standard to
explain, first, the peak-to-trough decline in real GDP per capita (fifth
and sixth columns in table 3), and second, the number of years it took
for real GDP per capita to return to its precrisis level (final two
columns). Because of the varied country set, the table reports both the
simple regression using the date of exit and an augmented one that also
includes dummy variables for whether the country was an Axis power, in
Latin America, or a member of the British Commonwealth. In no case does
the date of exit from the gold standard help to explain the depth or
duration of the downturn, confirming the message from the earlier
scatterplot.
All told, the evidence that countries exiting the gold standard
early fared better is apparently fragile. Once one expands the sample to
a broader set of countries and considers other measures of the business
cycle, leaving the gold standard early was not always a reliable route
to a shorter or less severe recession.
Why did exiting the gold standard generate so little benefit in the
larger sample? The answer in part was already evident in table 2.
Countries left the gold standard at different times. Moreover, when they
did leave, the range of variation in bilateral exchange rates vis-a-vis
the U.S. dollar was wide, indicating that policymakers did not follow a
common roadmap. The greatest number of countries left in 1931, but those
that did so in 1932 adjusted by more. A few that moved to some form of
floating arrangement saw their currencies unhelpfully appreciate against
the dollar.
The important point for the United States is that almost all of
these devaluations relative to gold produced an appreciation of the
dollar, which added to the force of contraction domestically. Not until
1933 was some of that force pushed back, and even then the dollar still
appreciated against the currencies of many economies. For
contemporaneous observers, these swings in bilateral exchange rates
smacked of "beggar thy neighbor" policy. From that experience
was born a mistrust of floating exchange rates and a desire for a more
managed system, famously expressed by Ragnar Nurkse (1944), among
others. The net effect of these currency changes was to worsen the
external drag on the U.S. economy, exactly when the appropriate policy
was to reflate.
Some sense of the net external drag can be gotten from figure 3,
which plots effective exchange rate indices between the United States
and five country groups. The base is set to 100 in 1929, and the shaded
area represents the range from plus to minus 15 percent of that parity.
There are three main messages. First, the range of variation of nominal
exchange notes was fairly narrow, except in Latin American countries,
suggesting that external relative price adjustment was not the crucial
means of rebalancing. Second, Canada's vaunted embrace of floating
exchange rates produced a result suggestive of considerable management
in that market outcome. (14) Third, most of the lines follow a track
above 100 (that is, an appreciation of the U.S. dollar), implying that
exchange rates worked to offset some of the domestic U.S. monetary
policy stimulus.
[FIGURE 3 OMITTED]
This brings to mind Robert Mundell's (1968) insights about the
N + 1 currency problem. In a system of N + 1 floating exchange rates,
depreciation of the N currencies must come from an appreciation of the N
+ 1 currency. This creates a need for the economy using that anchor
currency to overcompensate with domestic stimulus for that force of
external restraint. The advantage of the gold standard was that all N
could cheapen their currencies without putting a special burden on any
one nation, given that the N + 1 price was the value of gold. In the
event, however, the adjustment was not so smooth.
III. Fiscal Policy during the Great Depression
Sustained fiscal impetus in the major countries was similarly
needed in the 1930s. And it was tried in many countries, but seldom
consistently. Indeed, in many cases fiscal policy contracted as the
national economy shrank, worsening the downturn. This record of
policymaking is summarized in table 4, which reports for a group of 30
countries the year in which economic activity hit its cyclical low, as
well as real government spending in that year, indexed so that the 1929
level equals 100. (15) We rely on government spending to measure the
fiscal impetus, rather than the more commonly used budget balance, for
two reasons. First, revenue typically falls off in economic
contractions, irrespective of policy intent, thus worsening the fiscal
balance without necessarily providing much impetus. Second, we are
somewhat more confident about the reliability of spending data over time
and across countries than about that of revenue (see Kaminsky, Reinhart,
and Vegh 2005).
The countries in table 4 are listed according to the co-movement of
government spending and the economic cycle, from the most procyclical to
the most countercyclical. Quite clearly, fiscal policy sometimes
imparted considerable restraint rather than stimulus. As the third
column shows, real government spending contracted in at least one year
in 24 of the 30 countries, sometimes by a large amount.
The United States was not among the countries where the trend of
real government spending amplified the business cycle: by the trough in
1933, real government spending was almost twice its level of 1929.
Figure 4 plots real government spending in the United States and Canada,
again indexed to 100 in 1929. There were three distinct episodes of
large increases in spending, first at the end of President Herbert
Hoover's administration in 1932 and then under Roosevelt in 1934
and 1936. Contrary to the popular perception, Hoover did significantly
enlarge the role of government. (16) In contrast, the governments of the
United Kingdom and the Nordic countries provided less impetus, and
fiscal policy in the Latin American countries was decidedly procyclical
until 1934. The last group was no doubt hampered by a lack of access to
funding, as well as institutional problems evident in Argentina and
Brazil, among other countries.
[FIGURE 4 OMITTED]
Although fiscal impetus was forceful in some countries, in almost
all it was also erratic. Figure 4 further reveals that each of the three
large increases in spending in the United States and Canada was followed
by some retrenchment. The impetus from government spending in the United
States in 1932, 1934, and 1936 appeared on track to provide considerable
lift to the economy, but after each of those years real spending dropped
off, imparting an arithmetic drag on expansion. The fact that fiscal
expansion has been aggressive in many countries in 2009 works to help
contain the contraction in the global economy. That it will continue to
do so is far from assured, if history is any guide.
Table 5 examines the ebbs and flows of real government spending
across countries from 1929 to 1939. The first column reports the most
conventional measure of spending volatility, the standard deviation of
annual percentage changes in spending. Fiscal policy was indeed volatile
in this period, with six countries posting a standard deviation of
spending of more than 30 percent. A second indicator of the
inconsistency of fiscal policy is the frequency with which government
spending sharply reverses course. We calculate the "amplitude"
of such a reversal as the sum of the percentage changes in spending in
two consecutive years in which the first year sees a rise in spending
and the second a decline. The second column of table 5 reports the
amplitude of the largest such reversal in real spending growth for each
country, the third lists the year of that reversal, and the fourth
reports the number of times such reversals exceeded 10 percentage points
in amplitude. Again and again, fiscal policy lacked follow-through in
providing consistent impetus. Every one of the countries in the table
experienced a reversal of real spending in at least one year of the
decade, and all but one country suffered at least one reversal with an
amplitude of more than 10 percentage points.
This volatility of fiscal spending could, in principle, have
blunted some of the force of the fiscal impetus if it rendered economic
planning more difficult. Table 6 examines the extent to which such a
mechanism was at work: using data from a sample of 30 countries from
1929 to 1939, the table reports regressions that attempt to explain the
variation in the depth and duration of the business cycle with a measure
of the growth of real government spending standardized by its
volatility; to be precise, it is the average annual change in real
government spending from 1929 to 1939, divided by its standard
deviation. These regressions were performed with and without dummy
variables for regions and for whether the country was an Axis power; the
regressions including the dummy variables also include the logarithms of
real GDP per capita and population in 1928, to capture any effects of
country size.
As is evident from the first pair of regressions, the depth of the
cycle appears unrelated to the volatility of government spending.
However, the remaining regressions at least produce coefficients that
match the intuition. Higher standardized spending hastened the return of
output to precrisis levels and added to real GDP growth. However, the
standard deviation of spending is in the denominator of that explanatory
variable, implying that greater volatility of real government spending
tended to delay economic recovery and to reduce the net change in real
GDP per capita. Thus, there may have been real costs associated with
policy wavering.
IV. Some Lessons
Unconventional monetary policy and aggressive fiscal policy were
used extensively in the 1930s, in a considerable number of countries.
They were not, however, employed consistently. Monetary policy was
hampered by beggar-thy-neighbor problems as countries devalued relative
to gold at different times and by different amounts. As a consequence,
countries derived less benefit from exiting the gold standard than they
could have, if indeed they saw any benefit at all. The United States was
in the vanguard of aggressive use of fiscal policy at the central
government level, but there and in many other countries this fiscal
impetus was partly reversed soon after. The net effect was to raise
volatility--and therefore uncertainty--and potentially to lessen the
stimulus provided.
A message from the 1930s is that national authorities must
recognize that the openness of the global economy sometimes works to
blunt the effectiveness of policy in one country. In the 2000s the N + 1
currency has been the U.S. dollar, whose special reserve-currency status
meant that the United States received flight-to-safety flows even as it
was the epicenter of the financial crisis. (17) Like the appreciating
U.S. dollar in the first part of the 1930s, this flight to the safe
haven by capital holders outside the United States, by bolstering the
dollar, augments the forces of restraint at home.
Such a force may strengthen the case for concerted fiscal stimulus,
but here an unpleasant reality intrudes: financial markets do not view
all countries alike. Some have a history of uncertain repayment of their
debt. Indeed, as shown by Reinhart, Rogoff, and Miguel Savastano (2003),
some countries are "debt intolerant" and tend to default at
debt-to-income ratios that elsewhere would be an entry ticket to
European Monetary Union under the Maastricht Treaty. Progress in
institution building has been significant in many of these emerging
market economies. But national authorities take that lingering lack of
acceptance very seriously and are unlikely to act in a fashion that
threatens a reminder of earlier excesses. This implies that the advanced
economies may be the only agents with significant scope for fiscal
stimulus during a global crisis. (18)
ACKNOWLEDGMENTS We have benefited from the comments of our
discussants, the editors, other participants at the Brookings Panel
conference, and Kenneth Rogoff. We also thank Meagan Berry, Adam Paul,
and Gregory Howard for their assistance.
References
Akerlof, George A., and Robert J. Shiller. 2009. Animal Spirits:
How Human Psychology Drives the Economy, and Why It Matters for Global
Capitalism. Princeton University Press.
Barro, Robert J., and Jose F. Ursua. 2008. "Macroeconomic
Crises since 1870." BPEA, no. 1: 255-350.
Bernanke, Ben S. 2004. Essays on the Great Depression. Princeton
University Press.
Bernanke, Ben S., and Vincent R. Reinhart. 2004. "Conducting
Monetary Policy at Very Low Short-Term Interest Rates." American
Economic Review 94, no. 2: 85-90.
Board of Governors of the Federal Reserve System. 1943. Banking and
Monetary Statistics, 1890-1941. Washington.
Cagan, Phillip. 1965. Determinants and Effects of Changes in the
Stock of Money, 1875-1960. Columbia University Press.
Calvo, Guillermo A., and Carmen M. Reinhart. 2002. "Fear of
Floating." Quarterly Journal of Economics 117, no. 2: 379-408.
Choudhri, Ehsan U., and Levis A. Kochin. 1980. "The Exchange
Rate and the International Transmission of Business Cycle Disturbances:
Some Evidence from the Great Depression." Journal of Money, Credit
and Banking 12, no. 4, part 1: 565-74.
Eichengreen, Barry. 1992. Golden Fetters: The Gold Standard and the
Great Depression 1919-1939. Oxford University Press.
Eichengreen, Barry, and Kevin H. O'Rourke. 2009. "A Tale
of Two Depressions." VoxEU (June).
www.voxeu.org/index.php?q=node/3421.
Eichengreen, Barry, and Jeffrey Sachs. 1985. "Exchange Rates
and Economic Recovery in the 1930s." Journal of Economic History.
45, no. 4: 925-46.
--. 1986. "Competitive Devaluation and the Great Depression: A
Theoretical Reassessment." Economics Letters 22, no. 1: 67-71.
Eichengreen, Barry, and Peter Temin. 2000. "The Gold Standard
and the Great Depression." Contemporary European History 9, no.
2:183-207.
Friedman, Milton, and Anna Jacobson Schwartz. 1963. A Monetary
History of the United States, 1867-1960. Princeton University Press.
Hamilton, James D. 1988. "The Role of the International Gold
Standard in Propagating the Great Depression." Contemporary Policy
Issues 6, no. 2: 67-89.
Hsieh, Chang-Tai, and Christina D. Romer. 2006. "Was the
Federal Reserve Constrained by the Gold Standard During the Great
Depression? Evidence from the 1932 Open Market Purchase Program."
Journal of Economic History 66, no. 1: 140-76.
Ilzetzki, Ethan, Enrique Mendoza, and Carlos Vegh. 2009. "How
Big (Small) Are Fiscal Multipliers?" University of Maryland.
James, Harold. 2009. The Creation and Destruction of Value. Harvard
University Press.
Kaminsky, Graciela L., Carmen M. Reinhart, and Carlos A. Vegh.
2005. "When It Rains, It Pours: Procyclical Capital Flows and
Policies." In NBER Macroeconomics Annual 2004, edited by Mark
Gertler and Kenneth Rogoff. MIT Press.
Maddison, Angus. 2004. Historical Statistics for the World Economy:
1-2003 AD. Paris: Organisation for Economic Co-operation and
Development. www.ggdc.net/maddison/.
Meltzer, Allan H. 2003. A History of the Federal Reserve, Volume 1:
1913-1951. University of Chicago Press.
Mitchell, Brian R. 2003a. International Historical Statistics:
Africa, Asia, and Oceania, 1750-2000. London: Palgrave Macmillan.
--. 2003b. International Historical Statistics, Europe, 1750-2000.
London: Palgrave Macmillan.
--. 2003c. International Historical Statistics: The Americas,
1750-2000. London: Palgrave Macmillan.
Mundell, Robert A. 1968. International Economics. New York:
Macmillan.
Nurkse, Ragnar. 1944. International Currency Experience: Lessons of
the Interwar Period. Geneva: League of Nations.
Officer, Lawrence H. 2001. "Gold Standard." In EH.Net
Encyclopedia, edited by Robert Whaples (October 1).
eh.net/encyclopedia/article/officer.gold.standard.
Orphanides, Athanasios. 2004. "Monetary Policy in Deflation:
The Liquidity Trap in History and Practice." North American Journal
of Economics and Finance 15, no. 1: 101-24.
Reinhart, Carmen M., and Vincent R. Reinhart. 2008. "Is the US
Too Big to Fail?" VoxEU (November).
www.voxeu.org/index.php?q=node/2568.
Reinhart, Carmen M., and Kenneth S. Rogoff. 2009. This Time Is
Different: Eight Centuries of Financial Folly. Princeton University
Press.
Reinhart, Carmen M., Kenneth S. Rogoff, and Miguel A. Savastano.
2003. "Debt Intolerance." BPEA, no. 1 : 1-62.
Romer, Christina D. 1992. "What Ended the Great
Depression?" Journal of Economic History 52, no. 4: 757-84.
Svensson, Lars E. O. 2003. "'Escaping from a Liquidity
Trap and Deflation: The Foolproof Way and Others." Journal of
Economic Perspectives 17, no. 4: 145-66.
Comment and Discussion
COMMENT BY
HSIEH, CHANG-TAI Why did monetary policymakers fail to stem the
Great Depression of the 1930s? A conventional view, largely due to Barry
Eichengreen and Jeffrey Sachs (1986) and Ben Bernanke (2004), holds that
the gold standard was key. Adherence to the gold standard forced
economies experiencing capital outflows to contract and was the key
mechanism by which deflation was spread throughout the world. The link
to gold also prevented central banks from acting as the lender of last
resort when faced with a financial panic, and it placed constraints on
fiscal authorities who might otherwise have engaged in expansionary
spending or tax policies. A key stylized fact that supports this
interpretation is that among the industrialized countries, the depth and
length of the depression were correlated with how long a country stayed
on the gold standard. The downturn was more muted in countries, such as
the United Kingdom, that were among the first to exit the gold standard,
and longer in countries, such as France, that were among the last.
This paper by Carmen Reinhart and Vincent Reinhart challenges this
interpretation of the role of the gold standard by marshalling new
facts. Figure 2 of their paper shows that the statistical relationship
between the date of exit from the gold standard and the magnitude of the
depression disappears when the sample is broadened beyond the
industrialized countries. For example, several countries that were
mainly primary commodity producers were among the first to leave the
gold standard (most of them in 1929, two full years before the United
Kingdom's departure in 1931), yet suffered some of the worst
downturns. If one interprets this finding as evidence that adherence to
the gold standard did not affect the depth and severity of the Great
Depression, it potentially changes the standard interpretation of its
causes. The question is whether this reinterpretation is warranted.
In fact, the authors provide two interpretations of their new
finding. First, they argue that it shows that departure from the gold
standard is less effective "when the global pie is shrinking."
Although this interpretation might be correct, the paper presents no
evidence to support it. If the global pie was shrinking, by definition
it was shrinking for industrialized and nonindustrialized countries
alike. Why, then, might such a decline weaken the effectiveness of
exiting the gold standard more for the latter than for the former? For
example, is a larger share of manufacturing output in the
nonindustrialized countries exported, making these countries more
susceptible to downturns in world export markets? More generally, does
the response to exiting the gold standard differ depending on whether a
country is more or less dependent on world trade? Or is the argument
that the nonindustrialized countries specialize largely in commodities,
whose price elasticity of demand is less than that of other goods? In
that case, what might be driving the results is that the output
elasticity to changes in the terms of trade for the nonindustrialized
countries is not the same as that for the industrialized countries.
The second interpretation offered by the paper harkens back to the
argument by Ragnar Nurkse (1944). In brief, the argument is that one
country's departure from the gold standard might not necessarily
translate into a decline in the terms of trade if other countries are
depreciating against that country's currency at the same time. (As
an aside, it would be useful if the paper couched the discussion in
terms of the real exchange rate, that is, net of changes in domestic
prices or wages on both sides.) Again, here it would be useful to know
how exactly this interpretation fits with the observation that departure
from the gold standard was associated with economic recovery in
industrialized but not in nonindustrialized countries. The paper
emphasizes the fact that exchange rate adjustment in most countries came
largely at the expense of the United States. This may well be true for
1931 and 1932, but the United States' departure from the gold
standard in 1933 was quickly followed by a recovery. The paper needs to
show that departure from the gold standard was associated with
depreciation for the industrialized countries and not for the
nonindustrialized countries. A casual reading of figure 3 suggests that
the evidence on this point is not clear. Over all, the European
countries (other than the United Kingdom and the Nordic countries) did
see their currencies depreciate relative to the U.S. dollar. The
Canada-U.S. exchange rate, however, was basically unchanged. The pound
sterling actually appreciated against the U.S. dollar, yet this was the
country where the downturn was the smallest. On the other hand, the
Latin American currencies saw the largest depreciation against the U.S.
dollar, yet these were the countries where the downturn was most severe.
Let me propose a third explanation. To interpret the correlation
between the timing of exit from the gold standard and subsequent
economic outcomes as causal, one needs to be sure that the cross-country
variation in the timing of the exit from gold is not driven by forces
that might also drive the economic outcomes one is measuring. In his
book Golden Fetters (1992), Barry Eichengreen provides a wealth of
narrative evidence that the degree of commitment to the gold standard
among the industrialized countries was largely driven by ideology and
domestic political considerations. Although these political forces might
also have an independent effect on economic policy (that is, other than
through their effect on exchange rate policy), this is not the same as a
story where, for example, countries that exited the gold standard, or
exited sooner, were the ones that were hit the hardest by adverse
economic shocks. And the argument that variation in the degree of
commitment to the gold standard is exogenous to economic forces is less
plausible for the nonindustrialized than for the industrialized
countries. For example, isn't the fact that Argentina, Brazil,
Australia, New Zealand, Uruguay, and Venezuela (table 2 in the paper)
were the first countries to exit the gold standard driven by the severe
decline in world prices for their exports? If so, then how does one
disentangle the effect of the decline in export prices from the effect
of exit from the gold standard? It might well be the case that the exit
from the gold standard stimulated output, but this effect is overwhelmed
by the economic shocks that prompted the country to exit the gold
standard in the first place.
In sum, the new facts presented by Reinhart and Reinhart have the
potential to overturn what we thought we knew about the causes of the
Great Depression. But much more needs to be done to show that their
interpretation of their new facts--that the timing of departure from the
gold standard did not contribute to recovery from the Great
Depression-is the right one.
REFERENCES FOR THE HSIEH COMMENT
Bernanke, Ben S. 2004. Essays on the Great Depression. Princeton
University Press.
Eichengreen, Barry. 1992. Golden Fetters: The Gold Standard and the
Great Depression 1919-1939. Oxford University Press.
Eichengreen, Barry, and Jeffrey Sachs. 1985. "Exchange Rates
and Economic Recovery in the 1930s." Journal of Economic History
45, no. 4: 925-46.
Nurkse, Ragnar. 1944. International Currency Experience: Lessons of
the Interwar Period. Geneva: League of Nations.
GENERAL DISCUSSION Linda Goldberg thought it worth recalling the
specific circumstances in which exchange rate changes can make a
difference toward recovery from a recession. The classic mechanism is
expenditure switching: changes in exchange rates change the relative
prices of goods in different countries. But the amount by which such
changes help a country in recession depends in part on the degree to
which production is vertically integrated. The proportion of U.S.
imports that consists of components and raw commodities rather than
final goods for consumption is higher today than in the past, and this
limits the effects that one can expect through the exchange rate
channel. Goldberg also suggested introducing fnancial globalization
variables into the analysis. For example, there is evidence that the
more globalized banks are less sensitive to U.S. monetary policy than
other banks, because the globalized banks are able to transfer liquidity
among their different subsidiaries. This does not make monetary policy
completely ineffective, but it does shift the incidence of monetary
policy to those countries that are host to the global counterparties in
these intrafirm capital transactions.
Alan Auerbach pointed out that state and local spending was a much
larger share of U.S. government spending in the 1930s than it is today.
In the current recession, state and local government responses have
tended to be procyclical, and this effect needs to be taken into
account. Auerbach also observed that the paper dealt only with
government spending, and he suggested looking at the tax policy response
to the recession in different countries as well. Finally, he wondered to
what extent recent fiscal policy actions in different countries have
been expressly designed to avoid international leakages, perhaps in
response to the greater openness of economies in general.
CARMEN M. REINHART
University of Maryland
VINCENT R. REINHART
American Enterprise Institute
(1.) Reinhart and Rogoff (2009) provide many comparisons and a full
explanation of the data.
(2.) Eichengreen and O'Rourke (2009) provide useful
comparisons to that episode as well.
(3.) This strategy is discussed in Eichengreen (1992) and Romer
(1992).
(4.) There were two steps in this process. Executive Order 6102 in
April 1933 lowered the gold content of the dollar and prohibited the
public from holding gold. The value of the dollar in terms of gold was
lowered again with the Gold Reserve Act of 1934.
(5.) Eichengreen (1992), Romer (1992), and Bernanke (2004) explain
the mechanics. Important earlier contributions include Choudhri and
Kochin (1980) and Hamilton (1988).
(6.) Open market purchases of Treasury securities can lower yields
on government debt if assets are imperfect substitutes for each other.
Even if they are perfect substitutes, the swap of interest-bearing
government debt for non-interest-bearing reserves works to lower debt
service. Also, a decline in real interest rates improves measures of
debt sustainability. Two issues arise, however. First, the macroeconomic
effects will depend on whether the public capitalizes the income stream
of central bank profits. Second, paying interest on reserves lessens the
reduction in the debt burden associated with open market purchases.
(7.) James (2009) argues that these two episodes are distinct. The
year 1929 marked a major asset revaluation, and 1931 was a year of
banking collapse.
(8.) This failure can be explained as the Federal Reserve being
either hamstrung by the gold standard (as argued in Eichengreen 1992) or
focused too much on reserve supply rather than reserve demand (an in
Meltzer 2003). Either case amounts to a lack of willingness to use the
appropriate policy tools, not a lack of ability. Hsieh and Romer (2006)
show that a short-lived monetary accommodation in 1932 did not trigger
concerns in markets or among policymakers about a destabilizing exit
from the gold standard.
(9.) Romer (1992) stresses the multiplier effects of the former:
Eichengreen and Temin (2000) emphasize the change in the zeitgeist as
rekindling inflation expectations.
(10.) Meltzer (2003) and Orphanides (2004) review this experience.
(11.) Christina D. Romer, "The Lessons of 1937," The
Economist, June 18, 2009.
(12.) Eichengreen and Sachs (1986) work through the effects in a
simple model.
(13.) See Eichengreen (1992), Eichengreen and Sachs (1985), and
Eichengreen and Temin (2000).
(14.) Indeed, "fear of floating" in the Calvo and
Reinhart (2002) sense seems evident.
(15.) Because of data limitations inherent in a large historical
sample, the table uses statistics on central government spending only,
which is problematic for countries with a federal system that allows
discretion at the state or province level, such as the United States and
Argentina, among others. Local budgetary pressures may [lave
necessitated spending retrenchment that offset federal impetus. An
additional issue is that real government spending is constructed using
nominal spending from Mitchell (2003a, 2003b, 2003c), deflated by
available price indexes. Again this is done for comparability across
countries, but these measures do not always align well with readings
from the national income and product accounts, where available.
(16.) Akerlof and Shiller (2009) applaud the aggressiveness of both
Hoover and Roosevelt but lament the unevenness of their policies.
(17.) We raised this point in Reinhart and Reinhart (2008) when we
asked whether the United States was "'too big to fail."
Note the parallel with the discussion of the Federal Reserve's
failure in the early 1930s. Policymakers need to recognize that
safe-haven flows increase demand, necessitating even greater increases
in supply.
(18.) Another reason the advanced economies may have to shoulder
more of the burden is systematic differences in fiscal multipliers
across the North and the South, as discussed in Ilzetzki, Mendoza, and
Vegh (2009).
Table 1. Declines in Real Exports during Crisis Episodes
Countries
experiencing real
export decline
(percent of total)
Greater
No. of than
Episode Year countries Any 15 percent
Barings crisis 1890 35 34.3 5.2
1891 35 57.1 8.6
Panic of 1907 1907 73 31.5 4.1
1908 75 66.7 20.0
Commodity crash 1920 70 48.6 31.4
1921 73 76.7 54.8
Great Depression 1929 94 43.6 13.8
1930 94 88.3 48.9
1931 95 100.0 88.4
1932 95 80.0 62.1
Sterling crisis 1967 104 48.5 23.3
End of Bretton 1973 111 11.7 4.5
Woods regime
First oil shock 1975 110 41.8 29.1
Latin American 1981 106 60.4 31.1
debt crisis 1982 108 62.3 29.2
Nordic crises 1991 93 57.0 26.9
Exchange Rate 1992 95 36.8 14.7
Mechanism crisis
Tequila crisis 1995 105 23.8 11.4
Asia, Russia, 1997 109 40.4 13.8
LTCM crises (b) 1998 107 51.4 22.4
September-1911 2001 108 74.1 28.7
"Great Contraction" 2008 87 86.2 52.9
2009 (a) 42 100.0 92.9
Averages 1890-1939 70 41.1 18.0
1957-2008 99 33.0 14.6
Countries
experiencing real
export decline
(percent of total)
Largest Median
decline change
Episode Year (percent) (percent)
Barings crisis 1890 -18.0 2.2
1891 -47.5 -1.0
Panic of 1907 1907 -27.4 6.9
1908 -33.6 -5.4
Commodity crash 1920 -60.9 1.4
1921 -73.7 -19.8
Great Depression 1929 -36.2 1.3
1930 -51.2 -13.9
1931 -73.5 -33.3
1932 -53.8 -17.1
Sterling crisis 1967 -92.0 0.4
End of Bretton 1973 -79.4 39.1
Woods regime
First oil shock 1975 -78.3 1.8
Latin American 1981 -70.7 -3.6
debt crisis 1982 -77.2 -4.0
Nordic crises 1991 -75.8 -1.4
Exchange Rate 1992 -65.6 3.9
Mechanism crisis
Tequila crisis 1995 -79.3 9.4
Asia, Russia, 1997 -83.8 2.8
LTCM crises (b) 1998 -62.4 -1.4
September-1911 2001 -39.8 -9.6
"Great Contraction" 2008 -74.1 -16.6
2009 (a) -62.2 -36.6
Averages 1890-1939 -82.2 4.5
1957-2008 -92.0 9.2
Sources: Reinhart and Rogoff (2009, appendix A); League of Nations,
Statistical Yearbook, various issues; national sources; Maddison
(2004); Mitchell (2003a, 2003b, 2003c).
(a.) Through April.
(b.) LTCM, Long Term Capital Management, the large hedge fund that
failed in 1998.
Table 2. Depth and Duration of the Great Depression by Year of Exit
from the Gold Standard
Peak-to-trough Years until
Year of decline in real return to
business GDP per capita (a) precrisis
Country cycle peak (percent) real GDP
December 1929 and 1930 exits from gold standard
Australia 1926 17.3 10
New Zealand 1929 17.8 7
Argentina 1929 19.4 15
Brazil 1928 13.3 8
Uruguay 1929 36.1 17
Venezuela 1929 24.1 6
1931 exits from gold standard
United Kingdom 1929 6.6 5
Austria 1929 23.4 10
Canada 1928 29.0 12
Finland 1929 6.1 5
Germany 1928 17.8 7
Japan 1929 9.3 4
Norway 1929 1.9 3
Sweden 1930 4.8 4
Chile 1929 46.6 16
El Salvador 1928 11.3 9
Hungary 1929 11.4 7
India 1929 8.2 31
Korea 1928 12.7 5
Malaya (b) 1929 17 35
Mexico 1929 31.1 16
Portugal 1929 2.4 2
1932 exits from gold standard
Colombia 1929 3.8 3
Costa Rica 1928 15.7 9
Greece 1930 6.4 4
Nicaragua 1929 43.0 24
Peru 1929 25.4 6
Romania 1931 8.0 7
1933 exits from gold standard
United States 1929 28.9 10
Guatemala 1930 23.6 6
Honduras 1931 32.0 36
Philippines 1929 13.1 8
1934 exits from gold standard
Italy 1929 6.4 6
1935 exits from gold standard
Belgium 1928 10.4 11
1936 exits from gold standard
France 1929 15.9 10
Netherlands 1929 16.0 21
Switzerland 1929 9.8 9
Netherlands East 1929 14.3 9
Indies (c)
Poland 1929 24.9 9
Sources: Reinhart and Rogoff (2009); Eichengreen (1992); League
of Nations, Statistical Yearbook, various issues; Officer
(2001); Maddison (2004); Mitchell (2003x, 2003b, and 2003c).
(a.) GDP is measured in 1990 international Geary-Khamis
dollars.
(b.) Present-day Malaysia and Singapore.
(c.) Present-day Indonesia.
Table 3. Regressions Explaining the Depth and Duration of the
Great Depression by Date of Exit from the Gold Standard,
Dependent variable: Change in industrial
production or real GDP per capita,
1929-37 (percent)
Industrial Real GDP Real GDP
Independent variable production (b) per capita (b) per capita
Constant 45.19 13.75 9.29
(9.13) (5.74) (4.81)
Exit from gold standard -7.33 -2.74 -1.99
(years after 1929) (2.35) (1.48) (1.35)
Dummy for Axis power
Dummy for Latin America
Dummy for British
Commonwealth
Adjusted [R.sup.2] 0.36 0.17 0.06
Standard error of 19.75 12.41 15.80
the regression
Dependent Dependent
variable: variable:
Peak-to-trough Years until
decline in return to
GDP per real precrisis
Independent variable (percent) capita level (percent)
Constant 18.10 7.87 10.18 5.85
(3.37) (4.52) (2.57) (3.92)
Exit from gold standard -0.35 1.02 0.21 0.86
(years after 1929) (0.95) (0.95) (0.72) (0.83)
Dummy for Axis power 3.56 -1.47
(5.49) (4.76)
Dummy for Latin America 14.81 5.31
(3.92) (3.40)
Dummy for British 6.28 5.77
Commonwealth (5.43) (4.71)
Adjusted [R.sup.2] 0.00 0.31 0.00 0.10
Standard error of 11.07 9.62 8.44 8.35
the regression
Sources: Authors' regressions using data from League of Nations,
Statistical Yearbook, various issues; Eichengreen (1992); Officer
(2001); Maddison (2004); Mitchell (2003,,2003',2003c).
(a.) Sample consists of 39 countries except where noted otherwise.
Numbers in parentheses are standard errors.
(b.) Sample consists of 19 countries.
Table 4. Real Government Spending in Selected Countries, 1929-36 (a)
Real Largest
Year of government annual
trough in spending decline in
real GDP at trough real spending
Country per capita (1929= 100) (percent)
Chile 1932 53.0 34.4
Peru 1932 55.7 25.7
Venezuela 1932 73.8 33.2
Finland 1932 79.4 28.7
Austria 1933 79.9 21.8
Germany 1932 94.1 9.8
Netherlands East Indies 1934 94.9 4.0
Brazil 1931 96.4 15.4
Mexico 1932 96.9 7.2
Japan 1931 99.7 8.9
Colombia 1931 102.0 32.7
Norway 1931 105.5 None
New Zealand 1932 106.1 3.7
Argentina 1932 110.2 3.9
Uruguay 1933 110.7 13.3
Hungary 1932 111.8 10.2
India 1938 112.6 9.9
Poland 1933 114.0 None
Australia 1931 115.7 3.1
Belgium 1932 116.2 0.2
Greece 1932 117.5 38.8
United Kingdom 1931 118.3 None
Korea 1932 120.2 None
France 1932 138.9 None
Canada 1933 149.0 11.9
Portugal 1936 151.9 3.9
Sweden 1933 152.3 None
Netherlands 1934 154.2 3.6
Italy 1934 178.5 31.1
United States 1933 191.6 2.1
Median 111.3
Standard deviation 31.9
Year of
largest
decline in
Country real spending
Chile 1932
Peru 1932
Venezuela 1931
Finland 1932
Austria 1933
Germany 1931
Netherlands East Indies 1931
Brazil 1931
Mexico 1931
Japan 1931
Colombia 1929
Norway None
New Zealand 1932
Argentina 1931
Uruguay 1929
Hungary 1932
India 1932
Poland None
Australia 1929
Belgium 1932
Greece 1931
United Kingdom None
Korea None
France None
Canada 1933
Portugal 1930
Sweden None
Netherlands 1934
Italy 1929
United States 1933
Median
Standard deviation
Sources: Mitchell (2003a, 20036, 2003c); Reinhart and Rogoff (2009)
and sources cited therein; and authors' calculations.
(a.) Central government only.
Table 5. Volatility of and Reversals in Real Government Spending
in Selected Countries, 1929-39
Standard
deviation Amplitude
of annual of largest
changes in real fiscal
government reversal (a)
spending (percentage
Country (percent) points)
Italy 105.4 227.7
Greece 61.3 159.2
Peru 45.0 64.8
United States 33.8 47.5
Brazil 31.5 55.0
Finland 30.6 33.0
Portugal 29.7 66.6
Japan 28.9 42.3
Chile 27.6 51.5
Colombia 27.6 68.0
Venezuela 24.7 46.8
France 21.5 15.1
Germany 21.2 21.5
Canada 16.6 39.9
Sweden 16.5 31.4
Austria 16.4 18.5
Uruguay 16.4 24.1
Argentina 15.2 21.2
Netherlands 12.5 20.7
Mexico 12.0 24.4
Korea 11.9 16.6
Netherlands East Indies 11.7 30.3
India 10.0 13.5
Hungary 9.6 20.1
United Kingdom 9.0 12.8
Norway 8.6 11.9
Poland 7.8 12.8
Australia 7.2 11.8
New Zealand 6.5 8.9
No. of
reversals with
Year of amplitude > 10
largest percentage
Country reversal points
Italy 1937 3
Greece 1931 3
Peru 1929 5
United States 1933, 1935 (b) 3
Brazil 1933 5
Finland 1929 4
Portugal 1937 2
Japan 1937 4
Chile 1929 4
Colombia 1929 3
Venezuela 1929 4
France 1936 2
Germany 1929 2
Canada 1933 2
Sweden 1934 2
Austria 1931, 1932 (b) 3
Uruguay 1929 2
Argentina 1935 3
Netherlands 1932 3
Mexico 1937 1
Korea 1932 3
Netherlands East Indies 1938 1
India 1931 2
Hungary 1932 1
United Kingdom 1933 1
Norway 1933 1
Poland 1937 1
Australia 1932 1
New Zealand 1932 0
Sources: Mitchell (2003a, 2003b, and 2003c); Reinhart and Rogoff
(2009) and sources cited therein; authors' calculations.
(a.) A reversal is defined as a year of rising followed by a year of
declining government spending; the amplitude of a reversal is
calculated as growth in year t minus growth in year t + 1. For
instance, if real spending rose by 15 percent in year t and declined
by 12 percent in the following year, the amplitude would be 15-(-12)
= 27 percentage points.
(b.) Reversals in the two years were comparable in amplitude.
Table 6. Regressions Explaining the Depth and Duration of the
Great Depression by Volatility of Government Spending, 1929-39 (a)
Dependent
variable: Dependent
Peak-to-trough variable:
decline in Years until
real GDP return to
Independent variable (percent) precrisis level
Constant 14.22 25.15 10.51 9.88
(11.42) (11.26) 6.56 (7.15)
Annual change in real 1.49 -9.50 -0.83 -1.58
government spending (7.71) (7.90) 4.42 (5.01)
divided by its standard
deviation (b)
Dummy for Axis power 3.38 -0.75
(5.70) (3.62)
Dummy for Latin America 15.54 3.61
(4.91) (3.11)
Dummy for British 3.05 5.19
Commonwealth (5.21) (3.30)
Adjusted [R.sup.2] 0.00 0.29 0.00 0.14
Standard error of the 11.04 9.83 6.33 6.24
regression
Dependent
variable:
Growth in real
GDP per capita,
Independent variable 1929-37
Constant -2.32 -7.51
14.05 (15.83)
Annual change in real 5.64 11.86
government spending 9.48 (11.10)
divided by its standard
deviation (b)
Dummy for Axis power -4.72
(8.01)
Dummy for Latin America -9.63
(6.90)
Dummy for British -4.14
Commonwealth (7.32)
Adjusted [R.sup.2] 0.01 0.09
Standard error of the 13.57 13.82
regression
Source: Authors' regressions.
(a.) Sample consists of 30 countries in all regressions. Numbers
in parentheses are standard errors.
(b.) Sample covers 1929-39.
