Sturzenegger, A. & Moya, R. (2003) - Economic Cycles
Sturzenegger, A. & Moya, R. (2003) - Economic Cycles
Sturzenegger, A. & Moya, R. (2003) - Economic Cycles
Economic cycles
ADOLFO STURZENEGGER
Universidad Nacional de La Plata and Universidad Austral
RAMIRO MOYA
Fundación Investigaciones Económicas Latinoamericanas
Argentina’s economic history is rich in contrast. It was once one of the most
open in the world in terms of the flows of goods, services, and factors; soon
thereafter it became almost closed to international trade for many decades. It
was also one of the most promising countries in terms of economic growth
by the end of the nineteenth century, comparable only to the United States,
Canada, and Australia; eighty or so years later, however, it had became apparent
that initial conditions are not enough to guarantee continued development. In
macroeconomic terms, Argentina was one of the most stable and conservative
countries until the Great Depression, after which it turned into one of the most
unstable, and experienced one of the highest inflation rates.
In this chapter we study the business cycle of Argentina, or the short-run
macroeconomic fluctuations of the economy, over the period 1884 to 1990. We
start in 1884 because data is limited for earlier years, and we end in 1990 since
a major macroeconomic reform in that year suggests a natural break.
This period of study provides a laboratory for reviewing the cyclical be
havior of the economy in long-run series. In this work it will be shown that,
just like the experience of other countries after the Second World War when
the economic environment changed, increases in the volatility and persistence
of major nominal variables (monetary aggregates, prices, inflation, nominal
exchange rate) became noticeable. We also see in the Argentine data other
typical changes: some reduction in the volatility of GDP and consumption, and
anomalous investment behavior over historically exceptional periods.
Still, maybe the most substantive and striking fact is seen in the behavior
of nominal variables that cast doubts on monetary neutrality in the short term.
Specifically, the nominal exchange rate seems to move pari passu with the real
exchange rate, as in other countries, challenging the theories of the real business
We especially thank Pablo Alvarez for comments and statistical support, and conference partic
ipants for their suggestions.
87
88 Adolfo Sturzenegger and Ramiro Moya
cycle (RBC) school, which state that the real exchange rate should move only
on account of any changes in the fundamental real conditions in the economy.
However, the reason that changes in the real exchange rate scarcely show any
relation with the GDP cycle is also a puzzle, leaving open the question of what
might be behind this asynchronicity. Other facts tend to agree with certain
tenets of the RBC school, such as the countercyclical pattern of prices and the
strong procyclical pattern of the real wage. Is there a basis here for a simple,
general theory of cycles for Argentina? Obviously, the behavior of the real
exchange rate makes doubtful even the more basic formulations of the RBC
theory for the Argentine economy.
We claim that some of these unusual findings can be reconciled with a the
ory of “stop-and-go” cycles based on strong, contractionary devaluations and
recoveries via expansive demand-driven policies.1 For example, history shows
that devaluations were used to solve balance of payments problems through an
increase in tradable prices that, in turn, caused a fall in real wages and in the
real money supply. This policy of exchange rate devaluation was against all
priors in that it was decidedly contractionary and not at all expansionary, as in
the simple Mundell-Fleming theory. Some of these patterns will be noted in
this chapter.
But first, we will begin by trying to simply identify the business cycle. The
next section starts with a more traditional approach to business cycle study,
letting the data speak through a simple analysis. We will then use the more
modern method of detrending each economic series (with the bandpass filter)
to isolate trend and cyclical components in macroeconomic time series. This
will be helpful for our goal of identifying important cyclical regularities in
Argentine macroeconomic history.
(i) Over the period of 106 years that are covered, positive GDP growth
rates are evident in 80 years (or 75 percent of the sample), implying
that negative growth rates in the same variable occur in the remaining
26 years.
(ii) Twenty GDP expansion phases and nineteen contraction phases oc
curred during the period from 1885 to 1990
(iii) Expansions averaged 4.3 years in duration, while contractions lasted
2 As detailed below, the filter used corresponds to that originally used by Baxter and King (1999).
3 Additionally, even when the economic cycle was not the main issue, almost every historical
study of Argentina has attempted to define cycles and crises in terms of variations in GDP. See,
for example, Di Telia and Dombusch (1989) and Gerchunoff and Llach (1998).
4 For more details, see also Table 4.10 in the Appendix.
90 Adolfo Sturzenegger and Ramiro Moya
5 By filtering the GDP series - that is, by isolating the cycle from the long-term trend - the duration
of this economic expansion period is reduced.
Economic cycles 91
GDP growth rate was 5.4 percent, for a cumulative increase of 107.6
percent. Another significant period of expansion was from 1903 to 1913
when the average annual growth was 7 percent and the cumulative figure
was 111.2 percent.
(viii) The longest cycle (from one peak to the next) lasted from 1962 until
1977 (16 years) and achieved 103.4 percent cumulative growth. The
shortest contraction and expansion periods lasted only one year and
occurred in 1900-01 and 1943-44.
(ix) The most serious crisis (contraction) in terms of the year-to-year fall
in the average annual GDP growth rate occurred during 1914 when the
rate declined 10.3 percent. The most intense expansion, also in terms
of the annual growth rate, occurred between 1887 and 1889 when GDP
increased 13.1 percent annually and totaled 44 percent in only three
years.
Table 4.2 summarizes the main contraction and expansion periods. If cycles
are outlined by contractions - what we will call crises - the next logical question
is: What brought about these collapses in output?6
6 This does not mean that it should be accepted that the economy has a natural tendency to grow
- as unfortunately was seen in Argentina in the 1980s.
92 Adolfo Sturzenegger and Ramiro Moya
that some of the less important periods of contraction have not been included.
There are some common patterns among the different crises, and in fact, external
factors (e.g., the balance of payments) were the main causes of these crises in
all cases, with recent exceptions in 1978,1985, and 1988-89.
Crises brought about largely by problems with the current account occurred
in 1897,1952, and 1959; those related primarily to the capital account occurred
in 1890-91 and 1981-82; and those related to problems on both sides of the
balance of payments occurred in 1914, 1916-17,1930-32, and 1962-63.
To overcome almost all crises it proved necessary to induce some explicit
readjustment to the nominal exchange rate or to the monetary arrangements
in force (e.g., the annulment of the currency board), with exceptions in 1897,
1916-17, and 1952.7
From 1952 onward crises also originated in problems with inflation. This
fact is key to an understanding of what happened during postwar cycles in
Argentina, a point that will be discussed later. As a consequence, from that year
on all crises, without exception, were overcome in conjunction with stabilization
7
The 1952 stabilization plan did not include an explicit devaluation, although the change in export
taxes involved an improvement of the prices received by exporters.
Economic cycles 93
plans. The 1978 crisis was the only one in which a fall in GDP was intentionally
implemented by the government through a monetary contraction that was meant
to stop inflation.
In more general terms, we can also observe remarkable differences among
cycles, both before and after World War II. The first major difference is that
crises before the Great War were more serious. In this regard, the most important
collapses in GDP in Argentine history correspond to the 1890-91 crisis if the
fall is calculated in terms of the cumulative rate of decline, and the 1914 crisis
if measured in annual rates.
A second difference between both periods is of a more qualitative nature.
Before World War II, external factors prevailed as the main causes of crisis,
while after World War II causes were almost exclusively domestic. Perhaps
this distinction is suggestive of an explanation for the varying degree in the
severity of crises over time. A third difference, which might supplement any
explanation for differences in severity between the pre- and post-war periods, is
that during the latter period there was greater macroeconomic activism through
fiscal, monetary, or exchange rate adjustments.
These numerous observations can now be contrasted with a more modem
approach to economic cycles, which in fact leads to similar conclusions. It
will show that GDP tended to be less volatile after 1945 than before, and that
the increase in volatility and correlation between nominal variables (e.g., the
nominal exchange rate) and GDP indicates a higher level of policy activism
after World War II.
o
The parameters used are the same as those suggested by the authors for annual series: the high
frequency threshold is two years, the low frequency threshold is eight years and the moving
average is three years, so the filter is BP(2,8,3).
9 The figures in Appendix 2 show GDP at an annual frequency filtered by the above three method
ologies. The series derived from the bandpass filter stand out in relation to the other two in a
number of ways, namely their lower volatility (with this contrast being greater before 1945),
greater persistence vis-a-vis the first difference filter, and lower persistence vis-a-vis the Hodrick-
Prescott series.
Economic cycles 97
percent annual rate of decline. Other years that saw significant declines in GDP
were 1897 (-9.8 percent), 1902 (-8 percent), 1914 (-9.1 percent), 1945 (-8.1
percent), and 1959 (-9.7 percent). Expansions also follow almost the same
pattern as detected in the unfiltered GDP analysis, though again the ordering of
expansions by GDP growth rate is different. In fact, the most vigorous growth
corresponded not to the one detected in 1887-89, but the one of 1918 with a
13.3 percent rate of increase. Other pronounced periods of expansion occurred
in the years 1944 (7 percent) and 1887-89 (6.9 percent annually).
The two methods of characterizing cycles are therefore in general agreement.
The cyclical regularities are discussed in greater detail in the following two
sections using the filtered GDP series.
98 Adolfo Sturzenegger and Ramiro Moya
Some studies consider that Argentine policy choices initiated the import substitution process
during the years of the Great Depression (Diaz Alejandro 1970). It has also been claimed,
however, that external conditions (the disruption of international trade caused by the Great
Depression and World War II) generated this process “naturally”. What is somewhat agreed
upon among different authors is that the process of import substitution and greater autarky of
the Argentine economy was explicitly encouraged starting with the policies applied by General
Perón (1946-55).
Economic cycles 99
in which cyclical variables are analyzed by separating the postwar period from
the prewar period. Thus, for example, Basu and Taylor (1999) envision four
stages according to the monetary arrangement of countries from 1870, one of
them being the era of the Bretton Woods agreement (1945-70). Backus and
Kehoe (1992) divide their cross-country study into the prewar (before 1914),
interwar (1920-39), and postwar (after 1945) periods.11
11 Appendix 3 also examines the same variables considered in this section, but according to a
different division of history based on the periods 1884-1914, 1914-19, 1919-39, 1939-45, and
1945-90.
19
See Appendix 2 for the details.
1^
It should be noted that the GDP series for the earliest years of the sample naturally shows a
higher volatility as a result of the large share of agricultural and primary products in output.
Indeed, there could be a sectoral bias in the calculated GDP before World War II that would
make the series show a figmentary change in volatility. For the United States, this effect was
highlighted by Romer (1986).
100 Adolfo Sturzenegger and Ramiro Moya
Argentina, Kydland and Zarazaga (1997) found the same result with quarterly
data.
This high volatility of consumption in relation to GDP could be explained
by the construction of the consumption aggregate: it includes durable goods
and changes in stocks, which implies higher volatility. In fact, for the United
States, Stock and Watson (1998) reported that the consumption of services was
considerably less volatile (and also less procyclical) than the consumption of
durable goods - a finding consistent with the smoothing of consumption implied
by the permanent-income hypothesis. The procyclicality of consumption has
also increased during the postwar period, with less smoothing of consumption
evident in more recent years.14
Investment shows a volatility reduction of one-third for the postwar period.
Another finding, which is also remarked upon in other cycle studies (e.g., Basu
and Taylor 1999), is that investment volatility is about 3.5 times the size of that
for GDP. Even though the standard deviation between the two selected periods
shows some stability, the widening in a larger number of substages demonstrates
great variability in this indicator. What should be expected according to the
economic theory of investment behavior? If there is a possibility to distribute
risk, consumption stability should be higher in times of greater openness to
capital flows, but in such a case, investment stability will be adversely affected
since a positive productivity shock would attract external savings and magnify
the investment boom.
In contrast, in an autarkic economy, since there must be equality between sav
ing and investment, there should be more stability in investment vis-a-vis GDP.
However, as noted in the experience of other countries, the question remains:
Why was investment more volatile in closed-economy periods such as 1919-39
and 1939-45, when its standard deviation was between 4.5 and 5 times that of
GDP? As noted by Basu and Taylor (1999), it may be that the Great Depression
experience is not fully explained by conventional business cycle theories, and
such an explanation would have to consider neglected factors such as deflation
and its effect on debts. It should also be stated that investment volatility was
slightly higher before 1914 than in the postwar years, in accordance with what
should be expected in a more open economy versus a more closed one (al
though the difference is not substantial). The cross-correlation of investment
14 This result is also consistent with income changes having a permanent nature and, therefore,
consumption responding more closely to income. Thus, it must be assumed that a correlation
increase means that income shocks have taken on more of a permanent nature during the postwar
period than in the previous years, though inherently this is difficult to demonstrate.
102 Adolfo Sturzenegger and Ramiro Moya
with GDP, which is relatively high, would also indicate behavior of a quite
autarkic economy in both periods, but even more so after 1945.15
As regards imports, there is a slight volatility increase after 1945. Exports,
in contrast, show a lower volatility during the postwar period.16 Given the
decreasing weight of agricultural products in total exports, volatility should
be lower in external sales. Indeed, what is remarkable in foreign trade is the
substantial increase in import procyclicality after the war and the increasing
acyclicality of exports after 1945. In the case of imports, it should be noted that
the increasing share of capital and intermediate goods, and the decreasing share
of consumption goods, in total imports can explain the increase in procyclicality
after the Second World War. Finally, the persistence measure has declined
during the postwar period for both imports and exports.17
As a consequence of this import and export behavior, the balance of trade
shows a higher volatility after World War II, though not substantially so. The
most significant change is in its relationship with the cycle: having been acycli-
cal before 1945, it became highly countercyclical in the postwar years. The
main explanation lies in a higher import correlation with the GDP cycle and the
acyclicality of exports.
Employment behavior is procyclical throughout the whole sample, although
the correlation is very low. Employment volatility is lower than that of GDP
(after 1945 it represents half its standard deviation), which would indicate rel
atively low variability throughout the entire economic cycle.18
Real wages show a significant volatility increase after World War II, and
a standard deviation indicator four times as great as previously. In turn, this
made real wage volatility turn from half to twice that of GDP. Procyclicality
and persistence also increased substantially. The high correlation to the cycle is
remarkable in relation to other countries’ experiences. This procyclical behav
ior of real wages is consistent with most theories of economic cycles in which
consumers optimize their choice between leisure and consumption.19 How
ever, for Keynesian theories, the real wage should be countercyclical because
^ However, it should be pointed out that if consumption was more volatile than the GDP because
it includes durables goods consumption, then investment should be less volatile relative to
GDP, ceteris paribus. The procyclicality of comovements would be changed too - rising for
consumption and falling for investment.
16 These results should be read carefully as both series begin in 1916 (after being filtered) and
results are strongly influenced by the period between the wars.
17 The high import procyclicality and the low connection of exports with the cycle are also con
sistent with the results found by Stock and Watson (1998) for the United States.
18 It is important to point out that the employment series is a poor proxy: it originally was con
structed based on population growth, so this could attenuate the cyclical behavior of the series
artificially.
19 A high cyclical correlation was also found for postwar Argentina by Basu and Taylor (1999).
Economic cycles 103
the nominal wage is fixed (following Keynes’ own theory in 1936) or, at the
best, acyclical (as in the theory of efficient wages).20
21 The exchange rate is defined as pesos per U.S. dollar. An increase in the exchange rate corre
sponds to a depreciation, and a decrease to an appreciation.
Economic cycles 105
is the same as that noted for price levels (see below), showing that devaluations
tended to be contemporaneous with the more inflationary periods.
International reserves, in turn, were also volatile in the postwar period. They
had no relation with the cycle after 1945, but they did before that year when they
showed a slightly procyclical behavior. This last fact is evidence of an economy
that experienced booms and busts with classical gold standard adjustment at
the turn of century and before the 1930s. After those years, reserves tended
to be acyclical because capital and goods flows tended to move with no clear
pattern; for example, sometimes the start of recessions were coincidental with
balance of trade deficits (implying depleted reserves) that would be mitigated
by foreign aid (rebuilding the stock of reserves).
Table 4.6. Cyclical behavior o f inflation rates, import and export prices, and
terms o f trade, 1887-1990
Deviation from trend
1887-1990 1887-1945 1945-1990
log dWPI Standard Dev. 0.280 0.096 0.407
Obs. Number 103 58 46
Output corr. -0.333 -0.137 -0.538
Persistence 0.256 -0.266 0.295
log dCPI Standard Dev. 0.288 0.064 0.395
Obs. Number 88 43 46
Output corr. -0.355 0.122 -0.540
Persistence 0.327 -0.389 0.345
log WPI Standard Dev. 0.266 0.073 0.394
Obs. Number 104 59 46
Output corr. -0.354 -0.333 -0.525
Persistence 0.348 0.154 0.357
log CPI Standard Dev. 0.271 0.047 0.376
Obs. Number 89 44 46
Output corr. -0.308 0.157 -0.480
Persistence 0.341 0.092 0.342
log IMPPRIC Standard Dev. 0.245 0.385 0.059
Obs. Number 75 30 46
Output corr. 0.046 0.034 0.135
Persistence -0.285 -0.299 0.110
log EXPPRIC Standard Dev. 0.098 0.123 0.082
Obs. Number 75 30 46
Output corr. 0.255 0.251 0.315
Persistence 0.332 0.386 0.279
log TOT Standard Dev. 0.077 0.096 0.071
Obs. Number 75 30 46
Output corr. 0.172 0.174 0.252
Persistence 0.000 0.003 0.067
Notes: The variables dWPI and dCPI show inflation rates (log differences) while log WPI and log
CPI show price levels. IMPPRIC and EXPPRIC are import and export prices respectively. TOT is
the terms of trade. All variables were computed using the Baxter-King bandpass filter.
Sources: See Appendix 1.
balances. The economy was then put in recession because aggregate demand
had fallen, but the improved balance of trade set the stage for recovery.
Argentina is a small country compared to the rest of the world and was,
before World War II, very open to the flow of goods and services from other
countries. Over the 1913-39 period, the trade ratio (measured by the sum of
exports and imports as a share of GDP) was 55 percent. The policy of import
substitution implemented later decreased this ratio to average levels of only
21 percent in the period 1945-90. The data also show that import and export
prices were relatively less volatile after 1945 than before that year. As they are
Economic cycles 107
implicit prices of actual traded amounts (i.e., total value divided by quantity),
there could be some bias if external sales and purchases changed substantially
from one period to another.22
The Argentine terms of trade, like those of the OECD countries studied by
Backus, Kehoe, and Kydland (1995), showed a relatively high volatility with
respect to GDP, and a ratio of around 2.5 in both periods.23 It may be more
interesting to note that the terms of trade show a very low correlation with
the economic cycle in the two periods under consideration. With the prewar
economy so open to trade with the rest of the world, it should be expected that
changes in its terms of trade would have effects on the GDP cycle by increasing
the level of activity when the terms of trade are high and by reducing them
when the terms are low. This result could imply that the terms of trade did
not influence the cyclical evolution of the economy; or, that attempts made to
isolate the economy from changes in the terms of trade were successful. That is,
if government policy was to reduce the effect on domestic prices of an increase
in grain prices in foreign markets and local supply could have not responded to
that increase, the effect on production could have been slight. In Sturzenegger
(1990) it was noted that government trade policy intended to isolate the effect
of changes in export prices on the income received by farmers, which could
resolve this puzzling disconnect between terms of trade and real output.
4A. 1.4 Behavior o f real monetary aggregates, interest rate, and real
exchange rate
We next examine the behavior of various monetary variables, interest rates, and
real and nominal exchange rates in Table 4.7. The real monetary base shows a
low correlation with GDP. The volatility of both the real monetary base and the
real M2 aggregate increases after 1945. The real M2 aggregate shows a high
correlation with the cycle. This result is not in itself compelling evidence of the
nonneutrality of money since - as is pointed out insistently by RBC theorists
- it does not imply causality. The real interest rate also shows a sustained
increase in volatility and a low correlation with GDP. Again, it can be asked
if this supports money nonneutrality, or whether should it be understood as an
endogenous response to changes in the conditions of the real economy.
90
In the chapter by Berlinsky in this volume we find that for imports the change has been substantial.
In fact, consumption goods that in 1900 accounted for almost 40 percent of all imports, by the
1950s and later, except in a few years, never accounted for more than 10 percent of total imports.
Offsetting this change, intermediate goods (excluding fuels) saw a rise in their share of imports.
Also, although they are not strictly comparable under the different filters used, the volatility of
the terms of trade in developed countries was, for the 1970-90 period, about half that found in
this study for Argentina, except for Japan, where the standard deviation of the terms of trade
was very similar.
108 Adolfo Sturzenegger and Ramiro Moya
Table 4.7. Cyclical behavior o f real exchange rate, real monetary aggregates,
and real interest rate, 1887-1990
Deviation from trend ______
1887-1990 1887-1945 1945-1990
log ERR Standard Dev. 0.151 0.068 0.201
Obs. Number 89 44 46
Output corr. -0.172 -0.167 -0.204
Persistence 0.147 -0.161 0.179
log MOR Standard Dev. 0.151 0.073 0.199
Obs. Number 89 44 46
Output corr. 0.079 0.128 0.092
Persistence 0.076 -0.012 0.079
log M2R Standard Dev. 0.093 0.055 0.118
Obs. Number 89 44 46
Output corr. 0.500 0.456 0.589
Persistence 0.058 0.058 0.052
log ERR1 Standard Dev. 0.116 0.059 0.162
Obs. Number 104 59 46
Output corr. -0.033 0.080 -0.104
Persistence 0.128 0.033 0.145
log MORI Standard Dev. 0.128 0.084 0.168
Obs. Number 104 59 46
Output corr. 0.264 0.363 0.253
Persistence 0.009 -0.100 0.048
log M2R1 Standard Dev. 0.125 0.112 0.141
Obs. Number 104 59 46
Output corr. 0.618 0.604 0.666
Persistence 0.084 0.039 0.123
IRR Standard Dev. 0.336 0.132 0.449
Obs. Number 88 43 46
Output corr. -0.004 -0.057 0.044
Persistence 0.150 -0.014 0.151
IRR1 Standard Dev. 0.290 0.125 0.414
Obs. Number 104 59 46
Output corr. 0.072 0.124 0.106
Persistence 0.124 0.185 0.109
Notes: The variables ERR, MOR, M2R, and IRR show real exchange rate (log(ER) + log(GDP
deflator for USA) - log (CPI)), real monetary base, real M2, and real interest rate (paid on deposits).
All variables were adjusted by CPI except ERR1, MORI, M2R1, and IRR1, where WPI was used
instead. All variables were computed using the Baxter-King bandpass filter.
Sources: See Appendix 1.
This impasse has led some authors to look for more convincing evidence
of money neutrality in the behavior of real and nominal exchange rates. We
also believe that the questions must be resolved in a wider context that also
explains other behaviors, such as that noted in the real exchange rate. The
cyclical behavior of the real exchange rate also shows a substantial increase
Economic cycles 109
Fig. 4.3. Cyclical behavior of the real exchange rate (RER) and nominal exchange
rate (ER)
in volatility after the war. This behavior is matched by higher volatility in the
nominal exchange rate. In fact, as noted in Figure 4.3, the variables tend to
move together, a common observation for open economies.
Is this evidence of money nonneutrality? If money were neutral, the real
exchange rate would move in response to changes in the fundamental condi
tions of economy; that is, it would be determined on the “real side” (e.g., as a
result of changes in relative productivities). In contrast, the nominal exchange
rate should move in response to the conditions on the “monetary side” of the
economy. What can be seen from the evidence on Argentine history is that the
real exchange rate moves in step with fluctuations in the nominal exchange rate;
that is, a priori there seems to be evidence of nonneutrality in the short run.24
These results, partially confirming money nonneutrality, are shown for a group of countries,
including Argentina, in Basu and Taylor (1999). For more evidence on the connection between
nominal and real exchange rate see Frankel and Rose (1995).
110 Adolfo Sturzenegger and Ramiro Moya
This result does not seem to be consistent with the predictions of simple new
classical or RBC theories in which the real exchange rate is entirely determined
on the real side of the economy. Instead, it seems to be more consistent with
theories at the Keynesian-monetarist nexus (e.g., Friedman 1968) that posit
money nonneutrality in the short run.
However, we must also note that the very low contemporary correlation
between the real exchange rate and the GDP cycle seems to show that changes
in relative prices do not affect the real economy in any important way.25 There
is some indication that changes in the real exchange rate have a positive effect
on the balance of trade surplus with a one period lag, but not on the level of
economic activity. This problem might be resolved by the fact that an increase
in economic activity leads, at a later stage, to a balance of trade deterioration
that eventually must be corrected by a contraction in the economy and a change
in the exchange rate.26 This scenario fits well with our stop-and-go view and is
also manifest in the slight negative correlation between the real exchange rate
and GDP. However, the hypothesis deserves even further research because the
low correlation is still striking.
25 This result was also noticed by Baxter and Stockman (1989) for a number of countries.
26 An approximation with 4 lags resulted from the Granger causality test, which (at the 5 percent
significance level) rejected the hypothesis that the real exchange rate does not cause the balance
of trade; we also rejected the hypothesis that the balance of trade is not caused by the GDP.
However, the hypothesis that GDP is not caused by the real exchange rate is only rejected at a
28 percent significance level.
27 Data from all theses countries have been taken from Maddison (1995) and filtered by the bandpass
method. Except for Brazil, whose series began in 1900, the other countries have a span of data
comparable to the Argentine series (1887-1990).
Economic cycles 111
Table 4.8. GDP correlations with foreign countries, deviation from trend,
1887-1990
U.S.A. U.K. Canada Australia Brazil
1887-1990 0.079 0.161 0.071 0.074 0.201
1887-1945 0.143 0.210 0.105 0.142 0.261
1945-1990 -0.223 0.044 -0.044 -0.110 0.172
1887-1914 -0.006 0.273 -0.016 0.222 0.091
1914-1919 0.305 0.044 -0.542 -0.573 -0.844
1919-1939 0.648 0.620 0.676 0.218 0.640
1939-1945 -0.215 0.030 0.082 -0.025 0.449
1945-1990 -0.223 0.044 -0.044 -0.110 0.172
Notes and sources: See text and Appendix 1.
that of the United States and opposite to that of Brazil, Australia, and Canada.
During the years between the wars (1919-39), Argentine GDP followed the
cyclical behavior of all countries (except Australia). World War II drove the
Argentine cycle in a similar way to that of Brazil, and slightly counter to that
of the United States. It was in the postwar period, however, that the Argentine
cycle found itself in a pattern unlike any other, although it did again behave
counter to that of the United States. In these cross-correlations, it is clear that
the Argentine cyclical pattern, except for certain stages (namely, the world wars
and the Great Depression), had a tendency to move quite independently, even
during periods of relative openness to external flows of goods, services, and
factors.
It can also be noted that the Argentine cycle’s volatility was higher than that
of almost all the other countries, especially after World War II, as shown in
Table 4.9. Substantial differences exist between Argentine cycles and those of
the United Kingdom, Australia, and Brazil, whereas the volatility in Argentina
is similar to that in the United States. This result is surprising as there are
no strong correlations with any other country’s cycle; thus, it is important to
determine which factors explain this similarity between Argentine and U.S.
volatility. Further investigation of this result is beyond the scope of this work.
4.5 Conclusions
In this paper we have tried to analyze how business cycles behaved throughout
Argentine history. Our first approach, a more traditional one, showed that major
crises happened in early years (1914, 1890, and the 1930s) and that they were
related mainly to external factors. The import substitution process undertaken
after the 1930s was associated with a quite different cycle, one more related to
activist macroeconomic policies, and coterminous with higher (and sometimes
112 Adolfo Sturzenegger and Ramiro Moya
Table 4.9. GDP relative volatility versus foreign countries, deviation from
trend, 1887-1990
Argentina = 1
U.S.A. U.K. Canada Australia Brazil
1887-1990 1.089 0.602 0.910 0.749 0.643
1887-1945 1.194 0.638 1.044 0.833 0.652
1945-1990 0.923 0.498 0.500 0.514 0.656
1887-1914 0.799 0.389 0.705 0.799 0.591
1914-1919 0.921 0.645 1.073 0.650 0.456
1919-1939 1.554 0.927 1.670 0.870 0.910
1939-1945 2.820 1.119 1.514 1.554 0.766
1945-1990 0.923 0.498 0.500 0.514 0.656
Notes and sources: See text and Appendix 1.
unusually high) inflation rates. This last factor is crucial to our understanding
of how the Argentine economy performed subsequently.
Our second approach, using more formal quantitative techniques, exposed
certain regularities that are explained by no single economic theory. For ex
ample, RBC theory can explain some of the countercyclical price patterns, but
Keynesian and monetarist theories fit better the real and nominal exchange rate
movements. Even so, we are hard pressed to explain the unusually low cor
relation between the real exchange rate and GDR For now, it should be noted
that we did not set out to promote any one theory, but simply to report the main
empirical results. Future research will be directed to resolving these puzzles.
We do claim that, in the postwar period, many of the stylized facts seem to
fit quite well with what we call a stop-and-go cycle story in which devaluations
are contractionary in countries such as Argentina because both export supply
and import demand appear unresponsive to movements in the exchange rate. If
we assume that this is the case and take into account the facts that imports grew
faster than GDP and that both were highly correlated, we find that booms are as
sociated with a negative balance of trade and stability in prices, while recessions
involve high inflation and an improving balance of trade as large devaluations
set in at the end of a booming phase. This type of story, though conjectural, can
perhaps better explain some of the unusual and persistent features of postwar
Argentine macroeconomic fluctuations that we have uncovered.
Table 4.13. Cyclical behavior o f inflation rates, import and export prices, and
terms o f trade, 1887-1990
Deviation from trend
1887- 1914- 1919- 1939- 1945-
1914 1919 1939 1945 1990
log dWPI Standard Dev. 0.122 0.049 0.069 0.066 0.407
Obs. Number 27 6 21 7 46
Output corr. -0.245 -0.790 0.456 -0.097 -0.538
Persistence -0.282 0.123 -0.183 -0.665 0.295
log dCPI Standard Dev. 0.031 0.088 0.079 0.049 0.395
Obs. Number 12 6 21 7 46
Output corr. 0.497 -0.114 0.322 -0.597 -0.540
Persistence -0.221 -0.465 -0.407 -0.435 0.346
log WPI Standard Dev. 0.091 0.068 0.056 0.045 0.394
Obs. Number 28 6 21 7 46
Output corr. -0.563 -0.140 0.363 -0.286 -0.525
Persistence 0.121 0.756 0.157 -0.176 0.357
log CPI Standard Dev. 0.023 0.067 0.053 0.037 0.376
Obs. Number 13 6 21 7 46
Output corr. -0.066 0.307 0.289 -0.374 -0.480
Persistence -0.109 0.143 0.038 0.118 0.342
log IMPPRIC Standard Dev. 0.139 0.444 0.777 0.059
Obs. Number 0 4 21 7 46
Output corr. 0.612 0.045 -0.102 0.135
Persistence 0.157 -0.503 -0.707 0.111
log EXPPRIC Standard Dev. 0.122 0.132 0.088 0.082
Obs. Number 0 4 21 7 46
Output corr. -0.067 0.523 0.100 0.315
Persistence 0.203 0.336 0.396 0.252
log TOT Standard Dev. 0.118 0.092 0.099 0.071
Obs. Number 0 4 21 7 46
Output corr. -0.788 0.528 0.358 0.252
Persistence -0.789 0.105 0.377 -0.008
Notes: The variables dWPI and dCPI show inflation rates (log differences) while log WPI and log
CPI show price levels. IMPPRIC and EXPPRIC are import and export prices respectively. TOT is
the terms of trade. All variables were computed using the Baxter-King bandpass filter.
Sources: See Appendix 1.
118 Adolfo Sturzenegger and Ramiro Moya
Notes and sources: Hodrick-Prescott penalty parameter k = 100. See text and Appendix 1.
120 Adolfo Sturzenegger and Ramiro Moya
References
Akerlof, George A., and Janet L. Yellen. 1985. A Near-Rational Model of the Business
Cycle, with Wage and Price Intertia. Quarterly Journal o f Economics 100 (5): 823-
38.
Backus, David K., and Patrick J. Kehoe. 1992. International Evidence on the Historical
Properties of Business Cycles. American Economic Review 82 (4): 864-88.
Backus, David K., Patrick J. Kehoe, and Finn E. Kydland. 1995. International Business
Cycles: Theory and Evidence. In Frontiers o f Business Cycle Research, edited by T.
F. Cooley. Princeton: Princeton University Press.
Basu, Susanto, and Alan M. Taylor. 1999. International Business Cycles in Historical
Perspective. Journal o f Economic Perspectives 13 (2): 45-68.
Baxter, Marianne, and Robert G. King. 1999. Measuring Business Cycles: Approxi
mate Band-Pass Filters for Economic Time Series. Review o f Economics and Statis
tics 81 (4): 575-93.
Baxter, Marianne, and Alan C. Stockman. 1989. Business Cycles and the Exchange-
Rate Regime: Some International Evidence. Journal o f Monetary Economics 23 (3):
377^100.
Brodersohn, Mario S. 1974. Política económica de corto plazo, crecimiento e inflación
en la Argentina, 1950-1972. In Problemas económicos argentinos, edited by M. S.
Brodersohn. Buenos Aires: Ediciones Macchi.
Burns, Arthur F., and Wesley Clair Mitchell. 1946. Measuring Business Cycles. New
York: National Bureau of Economic Research.
Carrera, Jorge, Mariano Féliz, and Demian Panigo. 1996. Ciclo económico en Argentina
y Brasil. Anales de la Asociación Argentina de Economía Política, XXXI Reunión
Anual, vol. 6.
Cooley, Thomas F., ed. 1995. Frontiers o f Business Cycle Research. Princeton: Prince
ton University Press.
Cortés Conde, Roberto. 1989. Dinero, deuda y crisis: Evolución fiscal y monetaria en
la Argentina, 1862-1890. Buenos Aires: Editorial Sudamericana.
della Paolera, Gerardo, and Javier Ortiz. 1995. Dinero, intermediación financiera y nivel
de actividad en 110 años de historia económica argentina. Documentos de Trabajo
36, Universidad Torcuato Di Telia (December).
Díaz Alejandro, Carlos F. 1963. A Note on the Impact of Devaluation and the Redis
tributive Effect. Journal o f Political Economy 71 (6): 577-80.
---------- . 1966. Devaluación de la tasa de cambio en un país semi-industrializado: La
experiencia de la Argentina, 1955-1961. Buenos Aires: Editorial del Instituto.
---------- . 1970. Essays on the Economic History o f the Argentine Republic. New Haven,
Conn.: Yale University Press.
Di Telia, Guido, and Rudiger Dombusch. 1989. The Political Economy o f Argentina,
1946-83. Basingstoke: Macmillan.
Di Telia, Guido, and Manuel Zymelman. 1967. Las etapas del desarrollo económico
argentino. Buenos Aires: Editorial Universitaria de Buenos Aires.
---------- . 1973. Loe ciclos económicos argentinos. Buenos Aires: Editorial Paidós.
Frankel, Jeffrey A., and Andrew K. Rose. 1995. Empirical Research on Nominal
Exchange Rates. In Handbook o f International Economics, edited by G. Grossman
and K. Rogoff. Amsterdam: North Holland.
Friedman, Milton. 1968. The Role of Monetary Policy. American Economic Review
58 (1): 1-17.
Gerchunoff, Pablo, and Lucas Llach. 1998. El ciclo de la ilusión y el desencanto: Un
siglo de políticas económicas argentinas. Buenos Aires: Ariel.
Economic cycles 121
Hodrick, Robert J., and Edward C. Prescott. 1997. Postwar U.S. Business Cycles: An
Empirical Investigation. Journal o f Money, Credit, and Banking 29 (1): 1-16.
IEERAL (Instituto de Estudios Económicos sobre la Realidad Argentina y Latinoamer
icana). 1986. Estadísticas de la evolución económica de Argentina 1913-1984.
Estudios 9 (39): 103-84.
Koopmans, Tjailing C. 1947. Measurement Without Theory. Review o f Economic
Statistics 29 (3): 161-72.
Kydland, Finn E., and Edward C. Prescott. 1990. Business Cycles: Real Facts and a
Monetary Myth. Federal Reserve Bank o f Minneapolis Quarterly Review 14: 3-18.
Kydland, Finn E., and Carlos E. J. M. Zarazaga. 1997. Is the Business Cycle o f Argentina
“Different”? Federal Reserve Bank o f Dallas Economic Review, 4th Quarter, 21-36.
Lucas, Jr., Robert E. 1977. Understanding Business Cycles. Journal o f Monetary
Economics 5 (Supplement): 7-29.
Maddison, Angus. 1995. Monitoring the World Economy. Paris: OECD.
Nelson, Charles R., and Charles I. Plosser. 1982. Trends and Random Walks in Macroe
conomic Time Series: Some Evidence and Implications. Journal o f Monetary Eco
nomics 10 (2): 139-62.
Romer, Christina D. 1986. The Prewar Business Cycle Reconsidered: New Estimates
of Gross National Product, 1869-1908. Journal o f Political Economy 97 (1): 1-37.
Stock, James H., and Mark W. Watson. 1998. Business Cycle Fluctuations in U.S.
Macroeconomic Time Series. Working Paper Series no. 6528 (April), National
Bureau of Economic Research.
Sturzenegger, Adolfo. 1990. El caso argentino. In Economía política de las inter
venciones de precios agrícolas en América Latina, edited by A. O. Krueger, M. W.
Schiff, and A. Valdés. San Francisco: Banco Mundial, Centro Internacional para el
Desarrollo Económico afiliado al Instituto de Estudios Contemporáneos.
Zamowitz, Victor. 1992. Business Cycles: Theory, History, Indicators, and Forecast
ing. Chicago: University of Chicago Press.