International Capital Flows and Boom-Bust Cycles in the Asia Pacific Region +
This paper documents evidence of business cycle synchronization in selected Asia Pacific countries in the 1990s. We explain business cycle synchronization by the channel of international capital flows. Using the VAR method, we find that most Asian countries experience boom-bust cycles following capital inflows, where the boom in output is mostly driven by consumption and investment. Empirical evidence shows that capital flows in the region are highly correlated, which supports the conclusion that capital market liberalization has contributed to business cycle synchronization in Asia. We also find that business cycles in the Asian crisis countries are...
…………..o0o…………..
International Capital Flows and Boom-
Bust Cycles in the Asia Pacific Region
+
International Capital Flows and Boom-Bust Cycles
in the Asia Pacific Region +
Soyoung Kim*
University of Illinois at Urbana-Champaign and Korea University
Sunghyun H. Kim**
Tufts University
Yunjong Wang***
SK Research Institute
Abstract
This paper documents evidence of business cycle synchronization in selected Asia Pacific
countries in the 1990s. We explain business cycle synchronization by the channel of international
capital flows. Using the VAR method, we find that most Asian countries experience boom-bust
cycles following capital inflows, where the boom in output is mostly driven by consumption and
investment. Empirical evidence shows that capital flows in the region are highly correlated,
which supports the conclusion that capital market liberalization has contributed to business cycle
synchronization in Asia. We also find that business cycles in the Asian crisis countries are highly
synchronized with those in Japan.
JEL Classification: F02, F36, F41
Key words: business cycle synchronization, capital flows, boom-bust cycles, financial integration.
+
We are grateful to Gordon de Brouwer, Barry Eichengreen, Takeo Hoshi, Takatoshi Ito, Eiji Ogawa, and
Yung Chul Park for their helpful comments and suggestions. This research was kindly supported by a Ford
Foundation grant.
*
Department of Economics, University of Illinois at Urbana-Champaign, DKH, 225b, 1407 W. Gregory
Drive, Urbana, IL 61801.
**
Corresponding Author. Department of Economics, Tufts University, Medford MA 02155. Tel: 617-627-
3662, Fax: 617-627-3917, E-mail: [email protected].
***
SK Research Institute, 14th Floor, Seoul Finance Center, 84 Taepyungro 1-ga, Seoul 100-101, Korea.
1
1. Introduction
Over the past decade, a number of Asia Pacific countries have liberalized their financial
markets to foreign capital by reducing restrictions on inward and outward capital flows. Increased
capital flows due to financial integration can generate substantial effects on business cycles.
Large capital inflows following financial market liberalization can generate an initial surge in
investment and asset price bubbles followed by capital outflows and recession, the so-called
boom-bust cycles. In worst cases, the boom-bust cycles can end with a sudden reversal of capital
flows and financial crises. 1 On the other hand, by allowing domestic residents to engage in
international financial asset transactions, financial market opening can reduce the volatility of
some macroeconomic variables such as consumption through risk-sharing.2
What are the macroeconomic effects of capital flows, in particular on business cycle
fluctuations? Do business cycles become less volatile and more synchronized across countries as
the degree of financial integration increases? Understanding the business cycle implications of
capital flows is important as it can also reveal a great deal about the welfare implications of
financial market liberalization policies as well as international monetary arrangements.
This paper focuses on the effects of capital flows due to financial market liberalization on
business cycles, in particular co-movements across countries.3 We aim to shed some light on this
issue by providing detailed stylized facts on capital flows and business cycles in the Asia Pacific
region and by empirically analyzing the relationship between capital flows and business cycles.
For empirical analysis, we adopt the VAR (Vector Auto-regression) method. We, first, identify
the capital flow shocks and then examine their effects on cyclical movements of key
macroeconomic variables in each country. We also examine whether these effects are consistent
with the boom-bust cycle theory. By further analyzing the cross-country correlation of capital
flow shocks, we try to infer the role of capital flows in explaining business cycle synchronization.
Economic theory does not provide a unanimous prediction on the effects of capital flows
on co-movements of business cycles. Financial market integration can increase business cycle co-
movements as macroeconomic effects of capital flows in different countries follow similar
1
Although other fundamental domestic problems contribute to financial crises, capital account
liberalization and the resulting lending booms sometimes end in twin currency and banking crises.
2
Domestic residents can reduce fluctuations in income stream and consumption by borrowing from abroad
during recessions or lending to foreign countries during booms. International portfolio diversification
enables consumers and firms to achieve risk-sharing gains by diversifying risks associated with country-
specific shocks.
3
We do not focus on the effects of capital flows on business cycle volatility. See Buch, Dopke and
Pierdzioch (2002) and Kose, Prasad, and Terrones (2003a, 2003b) on this issue.
2
patterns through various channels of contagion and common shocks.4 However, co-movements
of output can decrease as allocation of capital becomes more efficient, allowing production to
become more specialized.5 Other variables also affect the relationship between capital flows and
business cycles, including monetary and fiscal policies, the nature of underlying shocks in the
economy, etc.6
Using the data of twelve Asia Pacific countries, we find the following stylized facts of
business cycles. First, business cycles in the five Asian crisis countries are highly synchronized
and follow business cycles in Japan, while they differ from cycles in Australia and New Zealand.
On the other hand, greater China, including Hong Kong and Taiwan, show similar cyclical
movements. Second, in general, business cycles in the 1990s are more synchronized across
countries than those in the 1980s, which supports the view that financial and trade integration
increases business cycle synchronization in Asia.
Using the VAR method, we find empirical evidence that positive capital flow shocks
(capital inflows) affect output, consumption, and investment positively in most countries, which
is consistent with the story of boom-bust cycles. In addition, capital flow shocks are highly
correlated across the crisis countries. These two results imply that capital flow shocks can explain
business cycle synchronization among the crisis countries to some extent.
The remaining sections of this paper are organized as follows. Section 2 provides
literature survey on the relationship between financial integration and business cycles. In section
3, we analyze trends and stylized facts of business cycles in the region. In particular, we
investigate how the volatility of business cycles in each country has changed over time and
whether we can find any evidence of business cycle synchronization in the region. We examine
the following twelve countries in the Asia Pacific region: five Asian crisis countries (Indonesia,
Korea, Malaysia, the Philippines, and Thailand), China, Singapore, Taiwan, Hong Kong, Japan,
Australia and New Zealand. Section 4 provides an empirical analysis of the relationship between
capital flows and business cycles. We use the VAR method to analyze how capital flow shocks
affect various macroeconomic variables and investigate whether capital flow shocks generate
boom-bust cycles in the region. We also analyze the properties of capital flow shocks identified
4
See Kim, Kose and Plummer (2001) for a detailed explanation on financial contagion.
5
See Heathcote and Perri (2002), Imbs (2003), and Kalemli-Ozcan et al. (2001).
6
Another important issue in the literature is trade integration and its impact on business cycles. Trade
integration can generate synchronized business cycles if countries mostly engage in intra-industry trade,
while trade integration can decrease the degree of co-movements if trade promotes inter-industry
specialization and countries are subject to industry-specific shocks. See Frankel and Rose (1998), and Shin
and Wang (2004).
3
in our models. In particular, we investigate whether the estimated capital flow shocks are driven
by exogenous economic events and correlated across countries. Section 5 concludes the paper.
2. Theoretical Overview
This section explains different theories on the effects of economic integration on the
symmetry of business cycles and documents empirical studies on this issue.7 Financial market
integration can decrease co-movements of output by increasing industrial specialization (Kalemli-
Ozcan et al. 2001). Countries with integrated international financial markets can ensure against
country-specific shocks through portfolio diversification; therefore such countries can afford to
have a specialized production structure. That is, financial market integration allows firms to take
full advantage of comparative advantage and engage in production specialization, which in turn
increases the asymmetry of output as long as industry-specific shocks exist.
Heathcote and Perri (2002) analyzed the same issue from a different angle. They noted a
significant drop in the cross-country correlation of output in the 1990s and argued that the drop
was due to a decrease in cross-country correlation of productivity shocks combined with
increased financial market integration. Degree of financial market integration endogenously and
positively responds to the correlation of shocks. That is, as productivity shocks become less
correlated, potential welfare gains from portfolio diversification increase, as does the degree of
financial market integration.
However, countries with liberalized capital accounts can be significantly more
synchronized, even though they are more specialized (Imbs, 2003). A large body of literature on
contagion argues that capital flows in different countries, in particular developing countries in the
same region, are synchronized through various channels of financial contagion including herd
behavior, information asymmetry, etc. (Calvo and Mendoza, 2000; Mendoza, 2001).
International investors may classify different countries in a single group and make region-based
investment decisions. In addition, capital flows can be highly synchronized if shocks that
determine capital flows are positively correlated or spill over across countries, or if developing
countries go through a financial liberalization process at the same time. Since capital inflows
have significant effects on business cycles (so-called “boom-bust” cycles), if capital flows are
7
Note that we focus on the effects of financial market integration on output co-movements, not cross-
country consumption correlation which is expected to increase as consumers in different countries receive a
similar income stream through portfolio diversification and consumption smoothing.
4
highly correlated and have similar effects on business cycles, then financial integration can
contribute to synchronization of business cycles.
3. Trends and Stylized Facts of Business Cycles
In this section, we document the main characteristics of business cycles of the selected
countries in the Asia Pacific region.8 We use the data from the International Financial Statistics
(IFS) and examine volatility (measured by standard deviation) and co-movements (measured by
cross-country correlation) of output, consumption and investment in these countries. The sample
period is from 1980 to 2001 and all the data are Hodrick-Prescott filtered (with filtering
parameter = 100). Since we are interested in changes in business cycle statistics as financial
markets liberalize, we examine business cycles in different sub-sample periods: 1980-1989 and
1990-2001. For the second period, we use the data with and without the Asian crisis period
because the data for that period may distort the statistics.
We focus on two aspects of business cycles related to financial market liberalization and
examine whether the stylized facts derived from the data support the theoretical predictions
studied in the previous section. First, we investigate how much the volatility of business cycles
has changed over time. As financial markets develop over time, volatility of consumption is
likely to decrease through consumption smoothing and risk sharing channels unless output
volatility increases substantially. However, the impact on volatility of output is more ambiguous
as argued in the previous section. Second, we focus on the degree to which business cycles in the
region are synchronized and the changes in the degree of business cycle synchronization over
time. We expect that business cycles in this region become more synchronized due to the
region’s trade integration and high portion of intra-industry trade. However, the effects of
financial integration on business cycle co-movements are ambiguous as argued in the previous
section.
3.1. Volatility of Business Cycles
Table 1 presents volatility of output, relative volatility of consumption and investment in
four different periods - the whole period, the 1980s, and the 1990s with and without the Asian
crisis period. The output volatility is relatively low with a standard deviation ranging from 1.93 to
8
See Kim, Kose and Plummer (2003) for a detailed analysis of stylized facts of business cycles in Asia and
the G-7 countries.
5
2.46 in more developed countries in the region: Japan, Australia and New Zealand. On the other
hand, less developed countries in the region exhibit higher volatility: 5.60 in Thailand, 4.69 in
Indonesia and 4.71 in Malaysia. Developed countries tend to have more stable industrial
structures and output streams. Small countries that depend on natural resources for their main
products tend to have volatile output streams due to volatile prices (terms of trade) of primary
goods. Moreover, the share of agricultural activity is higher and the shares of the industry and
service sectors are lower in the less developed countries. The agricultural sector output is highly
variable since it is heavily affected by extremely volatile productivity and price shocks.
Comparing output volatility in the two periods, the results are mixed. Five countries show
significant increases (Korea, Indonesia, Malaysia, Thailand and Japan), one country shows a
significant decrease (the Philippines), and the remaining countries do not experience significant
changes over time. Except for the Philippines, the five Asian crisis countries show higher
volatility of output in the 1990s compared to the 1980s. This result is consistent even when the
crisis period is excluded. On the other hand, greater China (China, Hong Kong, and Taiwan) and
Singapore do not experience a rise in output volatility in the 1990s, as well as Australia and New
Zealand.
According to the consumption smoothing property in the inter-temporal current account
model, consumption should be less volatile than output (Obstfeld and Rogoff, 1996). Countries,
when facing positive shocks, lend to foreign countries in order to smooth the consumption stream
over time, and vice versa. However, in the table, we observe that this is not the case in many
countries.9 The table shows that consumption volatility is significantly less than output volatility
in only five countries including more developed countries (Japan, Australia, and New Zealand) in
the region. Developed countries can smooth their consumption by using various risk-sharing
instruments. As financial markets develop, developing countries should be able to gain access to
these risk-sharing instruments and reduce the volatility of their consumption stream. There is no
significant change over time in consumption volatility and no explicit pattern is detected in the
table.
Investment is three to four times more volatile than output in the table, which is the
typical result in other empirical and simulation studies (Baxter and Crucini 1995; Kim, Kose and
Plummer 2001). Investment volatility in China, Singapore and Japan is among the lowest with a
relative standard deviation of less than or around three, while investment in the five Asian crisis
9
We should note that the volatility of consumption changes depending on the specific consumption data. It
is known that the volatility of durable goods consumption is two to four times higher than that of
nondurables consumption (see Backus, Kehoe and Kydland, 1995).
6
countries is quite volatile with a relative standard deviation higher than four. There are no
significant patterns of change in investment volatility in the 1980s and 1990s. For some countries
(Indonesia and Japan), it significantly decreases, while other countries do not display any notable
pattern.
Including the crisis period in the data for the 1990s does not significantly change the
statistics for all three variables. No systematic patterns of change in volatility result from
including or excluding this period in the data. In sum, we found that output volatility increases in
the 1990s in many countries and consumption smoothing is not realized as consumption volatility
is higher than output volatility in most countries.
3.2. Co-movements of Business Cycles
Table 2 shows cross-country correlation of output to illustrate the degree to which
business cycles are synchronized across countries. The first panel shows the results from the
entire sample period. A significant and positive correlation is exhibited across most countries,
except for Australia, New Zealand and China. The business cycles of Australia and New Zealand
are negatively correlated with those of most other Asian countries: specifically 7 and 5 cases of
negative correlation, respectively. Australia and New Zealand each have a positive (but not
strongly positive) output correlation with China, Hong Kong and Taiwan. This is no surprise
because the industrial structures of those two countries are totally different from the typical
structure in Asian countries. China’s business cycles are also negatively correlated with other
economies except Taiwan and Hong Kong. This can be explained by the fact that the three
economies—China, Hong Kong and Taiwan, known together as Greater China—are in the same
economic zone.10 A high correlation between Malaysia and Singapore can be explained in the
same context.
The seven Asian crisis countries (including Singapore and Hong Kong) show positive
correlation with each other and they are positively correlated with business cycles in Japan as
10
Since its recent economic reform, China has embarked upon a process of financial and real integration
with Hong Kong and Taiwan. Even before Hong Kong’s return to China’s sovereignty in 1997, it had
achieved a high degree of integration with the mainland. With respect to trade, for instance, Hong Kong
intermediates a lion’s share of China’s external trade via re-exports and offshore trade. Regarding financial
activity, a substantial amount of the international capital (in the forms of foreign direct investment, equity
and bond financing and syndicated loans) financing China’s economic expansion is raised via Hong Kong.
Economic links between China and Taiwan have also proliferated since the 1990s. According to official
statistics (although the official statistics under-represent the overall economic interest of Taiwan in China),
China is the largest recipient of Taiwan’s overseas investment and Taiwan is China’s third-largest source of
foreign direct investment (Cheung, Chinn and Fujii, 2002).
7
well. This indicates that Japan has been leading business cycles in the region. McKinnon and
Schnabl (2002) showed that the yen/dollar exchange rate significantly affects business cycles in
the East Asian countries through trade and FDI channels. For example, depreciation of the yen in
1995 slowed East Asian export expansion significantly, while yen appreciation accelerates
Japanese FDI into the East Asian countries. Bayoumi and Eichengreen (1999) find that the
correlation of supply shocks in the region is especially high for two groups, with Japan and Korea
in one group and Indonesia, Malaysia, and Singapore in the other. Loayza, Lopez and Ubide
(2001) examine common patterns in aggregate demand and supply shocks with a different
methodology. They find strong co-movements for two groups: Japan, Korea and Singapore make
up one group, and Indonesia, Malaysia and Thailand, another group. These results indicate that
there are two different business cycles in the region, even though the East Asian countries show
relatively strong co-movements as a whole.
Comparing the data of the 1980s and 1990s proves that business cycles are more
synchronized in the 1990s. We examine this property by comparing the number of negative cross-
country correlations of output in the two periods. We observe a negative correlation in 17 country
pairs during the 1980s, while the number decreases to 10 in the 1990s. Moreover, in the 1990s,
without Australia, only two country pairs display a negative correlation. Out of a total of 66 pairs,
41 cases show that correlation increases from the 1980s to the 1990s. 11 In fact, correlation
coefficients are significantly positive in most of the 41 cases; only four pairs exhibit a correlation
coefficient of less than 0.4.
The empirical results for this region support the view that business cycles become more
synchronized as financial markets liberalize. Empirical results on business cycle co-movements in
previous studies are mixed, depending on sample countries and periods. Some document that the
correlation of output decreases over time, in particular in the 1990s. Heathcote and Perri (2002)
showed that output correlation among the U.S., Europe, Canada and Japan dropped from 0.76 to
0.26. On the other hand, Kose et al. (2003a), using the data for 21 industrial and 55 developing
countries, showed that output correlation in general increased in the 1990s from the previous
periods. This is mostly due to the industrial countries in the sample.
In conclusion, we can summarize the main characteristics of the business cycle co-
movements as follows. First, business cycles in Australia and New Zealand are different from
those in the East Asian countries. Second, business cycles in the five Asian crisis countries are
highly synchronized and follow business cycles in Japan. Third, the countries in Greater China,
11
This case is indicated by bold and italic numbers in the table. We do not report the case excluding the
crisis period but the results are similar.
8
which encompasses Hong Kong and Taiwan, show similar cyclical movements. Finally, business
cycles in general are more synchronized across countries in the 1990s than in the 1980s, which
supports the view that financial integration increases business cycle synchronization.
4. Capital Flows and Business Cycles: Empirical Studies
In this section, we investigate how capital flow shocks affect the business cycle dynamics
of the Asia Pacific countries, for example, whether capital flows generate boom-bust cycles, and
whether capital flows help explain the synchronization of the business cycles in the Asian
countries. Capital flows, especially after the financial market liberalization, may increase the
volatility of business cycles by creating boom-bust cycles, in particular fluctuations in investment,
consumption, exchange rate, and other asset prices. Further, if capital flows are positively
correlated across countries, due to simultaneous capital market liberalization in Asian countries or
due to the herd behavior of international investors or due to common shocks, the boom-bust
cycles in each country may imply the synchronization of the business cycles.
For empirical methodology, we adopt the VAR estimation method to extract the shocks
to capital flows, to analyze how shocks to capital flows affect the various macroeconomic
variables in each country, and to examine how the shocks to capital flows are correlated across
countries.12
4.1. Vector Auto-Regression Model
We assume that the economy is described by a structural form equation
G(L)yt = et (1)
where G(L) is a matrix polynomial in the lag operator L, yt is an n×1 data vector, and et is an n×1
structural disturbance vector.13 We assume that et is serially uncorrelated and var(et)=Λ, which is
a diagonal matrix where the diagonal elements are the variances of structural disturbances. That
is, structural disturbances are assumed to be mutually uncorrelated.
12
A similar empirical methodology was used in Kim, Kim and Wang (2002) to analyze the boom-bust
cycles in Korea. Tornell and Westermann (2002) also examined the boom-bust cycles by using a sample of
39 countries.
13
For simplicity, we present the model without the vector of constants. Alternatively, we can regard each
variable as a deviation from its steady state.
9
We can estimate a reduced form equation (VAR)
yt = B(L)yt-1 + ut, (2)
where B(L) is a matrix polynomial in lag operator L and var(ut)= Σ.
There are several ways of recovering the parameters in the structural-form equation from
the estimated parameters in the reduced-form equation. The identification schemes under
consideration impose restrictions on contemporaneous structural parameters only. Let G0 be the
contemporaneous coefficient matrix in the structural form, and let G0(L) be the coefficient matrix
in G(L) without the contemporaneous coefficient G0. That is,
G(L) = G0+ G0(L). (3)
Then, the parameters in the structural-form equation and those in the reduced-form
equation are related by
B(L) = - G0-1 G0 (L). (4)
In addition, the structural disturbances and the reduced-form residuals are related by
et= G0ut, (5)
which implies
Σ=G0-1ΛG0-1. (6)
In the method proposed by Sims (1980), identification is achieved by Cholesky
decomposition of the reduced-form residuals, Λ. In this case, G0 becomes triangular so that a
recursive structure, that is, the Wold-causal chain, is assumed. In a general non-recursive
modeling strategy suggested by Blanchard and Watson (1986) and Sims (1986), maximum
likelihood estimates of Λ and G0 can be obtained only through the sample estimate of Σ. The
right-hand side of the equation (6) has n×(n+1) free parameters to be estimated. Since Σ contains
n×(n+1)/2 parameters, by normalizing n diagonal elements of G0 to 1’s, we need at least n×(n-
10
1)/2 restrictions on G0 to achieve identification. In this generalized structural VAR approach, G0
can be any structure (non-recursive). In this paper, recursive modeling is used.
4.2. Basic Model and Effects on Output
We construct a basic model to examine the effects of capital flow shocks on output. The
basic model includes three variables, {CUR, RGDP, CAP}, where CUR is the current account (as
the ratio to the trend GDP), RGDP is the log of real GDP, and CAP is the capital account (as the
ratio to the trend GDP).14 A constant term and complete seasonal dummies are included. Four
lags are assumed.15 CAP and RGDP are included in the model since they are primary variables of
interest; we examine the effects of capital flows or capital account on the real GDP. CUR is
included to control the capital account movements that depend on current account movements
since some capital account movements are often related to the financing of current account
imbalances and we are interested in extracting autonomous capital flows.
The basic model uses a recursive structure, in which the ordering of the variables is
{CUR, RGDP, CAP}, where the contemporaneously exogenous variables are ordered first. With
this ordering, the shocks to capital flows are extracted by conditioning on the current and lagged
CUR and RGDP, in addition to their own lagged variables. We condition on the current (and
lagged) CUR since current account imbalances are often financed by capital account. We exclude
such endogenous movements of capital flows from the shocks to capital flows. In addition, we
condition on the current (and lagged) real GDP since changes in the real GDP may affect the
capital account. For example, an increase in the real GDP may attract more capital, and improve
the capital account. We exclude the endogenous movements of capital flows due to the real GDP
changes from the shocks to capital flows since we would like to infer the effects of capital flow
shocks to real GDP.16
The sample period is 1990-2001, during which capital account was liberalized in these
Asian-Pacific countries (Grenville 1998; de Brouwer 1999, 2001). We consider two samples, one
14
We use an exponential trend on the GDP level (or a linear trend on the log level of GDP). When
constructing the ratio, we use all variables in terms of U.S. dollars.
15
We adopt the Bayesian inference, which is not subject to conventional criticism in the presence of unit
root and co-integration. See Sims (1988) and Sims and Uhlig (1991). We also experimented with the log
level of the variables but results were qualitatively unchanged.
16
Note that the effects of CAP shocks on CUR and RGDP are invariant to the ordering between CUR and
RGDP. On the other hand, capital flows might affect CUR and RGDP within a quarter, and the CUR and
RGDP shocks may reflect some part of (exogenous) CAP shocks. However, even in such cases, CAP
shocks still represent the shocks to CAP that are not endogenous to CUR and RGDP changes since they do
11
with the crisis period and the other without it (dropping 1997:3-1998:2). We relate the capital
flow shocks identified in the model to the financial market liberalization and the global common
shocks under a more liberalized financial market. If the capital account had been tightly
controlled (i.e., China), the shocks to capital flows in our model or autonomous capital flows
would have been very small since the capital account should have been directed to finance the
current account imbalances (note that our model identifies capital flow shocks, by controlling for
the current account movement). Therefore, by examining the effects of autonomous capital
account shocks during the sample period, we can infer the consequences of capital account
liberalization.
We use quarterly data for the estimation since monthly data is not available for most
countries. We consider nine countries for which quarterly data series are available for most of the
sample period. They are Korea, Japan, Indonesia, Thailand, the Philippines, Singapore, Taiwan,
Australia, and New Zealand.17 Data sources are International Financial Statistics, ADB Database,
and Bloomberg.
The impulse responses to CAP shocks over three years are reported in Figure 1 for the
sample including the crisis period and Figure 2 for the sample dropping the crisis period. Dotted
lines are one standard error bands. The scale represents percentage changes. At the top of each
column, the country names are denoted. At the far left of each row, the name of each responding
variable is reported.
First, we explain the results for the sample including the crisis period. In response to
positive CAP shocks, the real GDP tends to increase in all countries, except for Singapore. In
Singapore, capital inflows did not generate a boom in the economy. This can be explained by the
fact that Singapore serves as an intermediary of international capital flows, not as a final
destination of foreign capital, which means that real economic activities in Singapore have little
relationship to capital flows in and out of the country.18 The positive effect of capital inflows is
significant in most countries, including all crisis countries under consideration, and quite
persistent in many countries. The positive effects last for more than three years in most countries.
not result from endogenous responses to CUR and RGDP, although CUR and RGDP shocks may include
(exogenous) shocks to CAP in addition to shocks to CUR and RGDP.
17
The estimation period for Thailand is from 1993 since the data series are available only from 1993.
18
Although Singapore as a regional financial center has relatively more open financial markets vis-à-vis
other East Asian economies, it maintained strong economic fundamentals and well-functioning financial
systems. Singapore was a creditor country before the crisis, having no external debt. Furthermore, when
neighboring countries were hit, Singapore was able to manage the contagion by floating its currency. Like
Singapore, Hong Kong had financially sound and economically healthy fundamentals as well as mature
institutions, but it still became a victim of the crisis because its firm commitment to the pegged exchange
12
For example, in New Zealand and the Philippines, the positive effects are different from zero with
more than 68 percent probability at least for two and a half years. Although the positive effects
after two years are less significant in most other countries, the point estimates show that the
effects are positive for more than three years in all countries but Korea, Thailand, and Singapore.
The results for the sample excluding the crisis period, reported in Figure 2, are not much different
except for Indonesia. The negative effects of capital outflows during the crisis period were so
dominant in Indonesia that the boom-bust cycles disappear when this period is excluded.
4.3. Effects on Other Macro Variables
We modify the basic model to examine the effects of capital flow shocks on other
macroeconomic variables. The modified model uses a recursive structure, in which the ordering
of the variables is {CUR, X, CAP}, where X denotes the variable in interest. With this ordering,
the shocks to capital flows are extracted by conditioning on the current and lagged CUR and X, in
addition to their own lagged variables. We condition on the current (and lagged) CUR and X as
before. First, the current account imbalances are often financed by capital account, and we would
like to exclude such endogenous movements of capital flows from the shocks to capital flows.
Second, we condition on the current (and lagged) X since changes in X may affect the capital
account.19
We include (real) consumption, (real) investment, the price level, and the real exchange
rate as X. Each variable is used as a log form. To construct real consumption and real
investment, nominal data are deflated by using a GDP deflator. As the price level, we used the
GDP deflator. The real exchange rate is constructed by a nominal exchange rate against the U.S.
dollar and the GDP deflators of each country and the U.S. Note that an increase in the real
exchange rate is a real exchange rate appreciation.20
Figures 3 and 4 report the results. We did not report the results for consumption and
investment for Taiwan and consumption for Singapore since quarterly data series are not
rate system invited speculative attacks. Hong Kong weathered a series of attacks at the expense of its
overall macroeconomic performance.
19
As in the basic model, we order X before CAP. By doing so, CAP shocks represent the shocks to CAP
that are not endogenous to CUR and RGDP changes since they do not result from endogenous responses to
CUR and X, although CUR and X shocks may include (exogenous) shocks to CAP, in addition to shocks to
CUR and RGDP.
20
For Taiwan, CPI is used since a GDP deflator is not available.
13
available.21 The first two rows report the responses of consumption (“CONS”) and investment
(“INV”); consumption and investment increase in almost all countries. The increase in
consumption and investment is especially significant in all the Asian crisis countries. When we
exclude the crisis period, the positive effects of capital inflows on consumption and investment
become weaker in the Asian crisis countries, especially in Indonesia. This is because, among the
crisis countries, Indonesia experienced the most serious and prolonged damage. From this
analysis, we can easily infer that the increase in output following capital flow shocks is mostly
due to the increase in consumption and investment because the current account negatively
responds to capital flow shocks (Figures 1 and 2).
The third and the fourth rows report the responses of the price level (“PGDP”) and the
real exchange rate (“RER”). The price level responses are mixed, depending on the country and
the sample. For real exchange rate, we expect to observe real appreciation following capital
inflows. The graphs show that real exchange rate appreciates in most countries except for
Thailand. This is actually due to the inclusion of the crisis period, as Figure 4 without the crisis
period shows a real appreciation in Thailand as well. For Indonesia and Korea, the exchange rate
initially depreciates and starts to appreciate with some time lag (2 quarters).
4.4. Properties of Estimated Capital Flow Shocks
The validity of the VAR results in the previous section depends on the identification of
shocks, whether capital account shocks represent exogenous changes in capital flows, for
example, due to capital account liberalization or due to abrupt changes in the behavior of
international investors as in the financial crisis or due to global common shocks. In this part, we
examine whether the estimated capital flow shocks actually represent such shocks by plotting
cumulative capital flow shocks for each country and relating them to economic events occurred.
Figure 5 plots identified cumulative capital account shocks in each country.22 For Asian
crisis countries (Korea, Indonesia, the Philippines and Thailand), we observe positive capital flow
shocks in 1994-96 period when these countries actively embarked on financial market
deregulation and opening (Furman and Stiglitz 1998, de Brouwer 1999, Kim, Kim and Wang
2002). For example, Korea allowed nonresidents to directly purchase stocks of Korean companies
21
Note that the data for Indonesian investment and consumption are only available from 1993, so the
results are for the period of 1993-2001.
22
We plot cumulative capital flows shocks because capital account shocks themselves are very volatile.
14
up to 3% per individual in 1992 and this share increased to 23% in May 1997. As a result, the
external debt in these crisis-hit countries increased dramatically for three years from 1994 to 1996.
This period also coincides with low world interest rate and the appreciation of Yen. Yen
appreciation increased Japanese overseas direct investment in East Asia. Low interest rates in the
industrial countries including Japan produced the portfolio flows to the East Asian economies. On
the other hand, the graphs show negative capital flow shocks during the crisis period 1997-98 as
large current account deficits turned into surpluses.
Australia and New Zealand recorded persistent current account deficits throughout the
1990s. For Australia, we observe positive capital flow shocks from the mid-1990s when the
country persistently marked current account deficits. For New Zealand, the capital inflows
continued until 1997 and the capital account reversed into deficits during 1998-2000. In contrast,
Taiwan experienced current account surpluses and net capital outflows before the Asian crisis.
Thus, for Taiwan, we observe negative capital flow shocks in 1995-96.
4.5. Synchronization of Capital Flows and Business Cycles
In the previous parts, we show that a positive shock to capital flows increases output in
most countries, and the increase in output is mostly due to a boom in consumption and investment.
The findings, especially for the case of the full sample including the crisis period, is consistent
with the “boom-bust” cycle following the financial market liberalization. In our model, a big
surge in capital inflows after the financial market liberalization can be captured as a positive
shock to capital flows, and such a positive shock leads to a boom. Later, when capital flows are
reversed, capital outflows can be captured as a negative shock to capital flows in our model, and
such a negative shock leads to a bust stage.
However, the evidence alone is not enough to support the hypothesis that capital flow
shocks or the financial market liberalization process increases business cycle synchronization in
the Asia Pacific region. Only when capital flow shocks are highly correlated across countries in
the region, can they increase co-movements of business cycles. Otherwise, capital flow shocks
may not contribute to business cycle synchronization.
In this regard, we calculate the cross-country correlations of the capital flow shocks
identified in our model. We use two measures. First, we use the capital flow shocks themselves.
Second, we use the cumulative capital flow shocks. The capital flows shocks in our model
typically have a persistent effect on output, so the cross-country correlation of cumulative capital
15
shocks may be more relevant measures regarding their effects on synchronization of business
cycles.
Tables 3 reports the results for cumulative capital flow shocks, for the period with and
without the crisis. For both sample periods, we find positive correlations of capital flow shocks
among the crisis countries. As shown in the previous section, since capital flow shocks have
similar effects on business cycles, we can conclude that capital flow shocks contribute to business
cycle synchronization among the crisis countries. Table 4 reports the results for (non-cumulative)
capital flow shocks. Positive correlations among crisis countries are found in most cases.
We suggest two possible reasons to explain why capital flow shocks among the crisis
countries are positively correlated. First, the timing of financial market liberalization in those
countries was similar, and each country experienced a boom-bust cycle after the liberalization.
Thus, the financial market liberalization process itself contributes to the synchronization of the
business cycles. Second, given some extent of openness in the financial markets, contagion
through financial channels contributed to similar capital flows in these countries. Due to
information cascade, international investors classify these countries in the same group and apply a
single investment decision for the whole group. Combined with herd behavior, financial
contagion contributed to the synchronization of capital flows and eventually of business cycles.
We also find two interesting observations. First, there is a positive correlation of capital
flow shocks between the crisis countries and Japan. All correlations for capital flow shocks and
cumulative capital flow shocks, with and without the crisis period, are positive, except for only
two cases. This result suggests that capital flow shocks can explain the synchronization of the
business cycles of Japan and the crisis countries. Second, we may not observe synchronized
business cycles among the crisis countries in the future. Since the Asian Crisis, foreign investors
have started to differentiate Korea from the other four Asian crisis countries. Korea is the only
country that has net capital inflows in the post-crisis period. Therefore, considering that capital
flows have been generating similar boom-bust cycles in the crisis countries, business cycles in
Korea may follow a different path from the other four countries in the future.
5. Conclusion
The relationship between financial integration and co-movements of business cycles is
not unambiguous, both theoretically and empirically. In this paper, we first document business
cycle synchronization in a number of the Asia Pacific countries and try to explain the
phenomenon by examining financial market liberalization and capital flows. We find that
16
business cycle synchronization among the Asian crisis countries in the 1990s can be at least
partially explained by synchronization of capital flows and the ensuing boom-bust cycles after the
financial market liberalization. Therefore, the results imply that financial market liberalization is
likely to synchronize business cycles across a group of countries. This is an interesting finding
since recent studies using data from developed countries often conclude the opposite.
Understanding the effects of capital flows on business cycle co-movements has
important implications for various issues. First, potential welfare gains from international risk
sharing highly depend on the degree of business cycle synchronization across countries. When
countries follow similar business cycles, it is less efficient to share risks across countries. If
financial market liberalization and capital flows increase business cycle co-movements, then
potential welfare gains from financial market liberalization would be lower than the level
calculated from the existing level of business cycle co-movements. Therefore, potential welfare
gains from financial market liberalization might be over-estimated.
Second, the findings of this paper can have implications for financial market
liberalization policies. In implementing financial market liberalization policies, policymakers
should consider the effects of the speed and sequencing of such policies on business cycles and
eventually on welfare. Finally, our results have implications for regional monetary and financial
integration in terms of optimum currency area criteria. For example, one of the conditions for an
optimum currency area is the presence of similar business cycle movements in the potential
candidate countries.
When most emerging East Asian countries started to liberalize their financial markets in
the early 1990s, no regional risk-sharing mechanism existed. Although Japan still remained an
important source country for external financing before the crisis, Western investors outside the
region also played an important role. Since the crisis, however, most East Asian countries have
become net providers of international capital due to their current account surpluses. While
receiving inflows of foreign direct and portfolio investment on a net basis, these countries have
repaid large sums of bank loans for the past several years. Looking to the future, whether
countries in the Asia Pacific region have similar patterns of capital flows will be an empirical
question. However, until a regional risk-sharing mechanism for integrating the financial markets
in the region is fully developed, most East Asian countries are likely to become more integrated
into the global financial markets.
17
References
Backus, D, P Kehoe, and F. Kydland (1995), “International Business Cycles: Theory and
Evidence,” in T. Cooley, ed., Frontier of Business Cycle Research, Princeton University Press.
Baxter, M. and M. Crucini (1995), “Business Cycles and Asset Structure of Foreign Trade,”
International Economic Review 36, 821-54.
Bayoumi, T. and B. Eichengreen (1999), “Is Asia an Optimum Currency Area? Can It Become
One?” in S. Collinon, J. Pisani-Ferry and Y.C. Park, eds., Exchange Rate Policies in Emerging
Asian Countries, Routledge Studies in the Growth Economies of Asia, Routledge: London and
New York.
Blanchard, O.J. and M.W. Watson (1986), “Are Business Cycles All Alike?” in R. Gordon, ed.,
The American Business Cycle: Continuity and Change, University of Chicago Press, Chicago,
123-156.
Buch, C., J. Dopke, and C. Pierdzioch (2002), “Financial Openness and Business Cycle
Volatility,” mimeo.
Calvo, G., and E. Mendoza (2000), “Rational Contagion and the Globalization of Securities
Markets,” Journal of International Economics 51, 79-113.
Cheung, Y., M.D. Chinn, and E. Fujii (2002), “China, Hong Kong, and Taiwan: A Quantitative
Assessment of Real and Financial Integration,” mimeo.
de Brouwer, G.J. (1999) Financial Integration in East Asia, Cambridge: Cambridge University
Press.
de Brouwer, G.J. (2001) “East Asian Financial Markets: Current Status and Future
Development,” in Y. H. Kim and Y. Wang, eds., Regional Financial Arrangements in
East Asia, Korea Institute for International Economic Policy.
Frankel, J. A., and A. Rose (1998), “The Endogeneity of the Optimum Currency Area Criteria,”
The Economic Journal 108, 1009-1025.
Furman, J. and J.E. Stiglitz (1998), “Economic Crises: Evidence and Insights from East Asia,”
Brookings Papers on Economic Activity 2, 1-135.
Grenville, S.A. (1998) “Capital Flows and Crises,” Reserve Bank of Australia Bulletin
(December), pp. 16-31.
Heathcote, J. and F. Perri (2002), “Financial Globalization and Real Regionalization,” NBER
Working Paper No. 9292.
Imbs, J. (2003), “Trade, Finance, Specialization and Synchronization,” London School of
Business, mimeo.
18
Kalemli-Ozcan, S., B. Sorensen, and O. Yosha (2001), “Economic Integration, Industrial
Specialization, and the Asymmetry of Macroeconomic Fluctuations,” Journal of International
Economics 55, 107-137.
Kim, S., S.H. Kim and Y. Wang (2002), “Macroeconomic Effects of Capital Account Liberalization:
The Case of Korea,” Review of Development Economics, forthcoming.
Kim, S.H., A. Kose, and M. Plummer (2001), "Understanding the Asian Contagion: An
International Business Cycle Perspective," Asian Economic Journal 15, 111-38.
Kim, S.H., A. Kose, and M. Plummer (2003), “Dynamics of Business Cycles in Asia: Differences
and Similarities,” Review of Development Economics 7, pp 462-477.
Kose, A., E. Prasad, and M. Terrones (2003a), “How Does Globalization Affect the
Synchronization of Business Cycles?” IMF Working Paper No. 03-27, International Monetary
Fund.
Kose, A., E. Prasad, and M. Terrones (2003b), “Financial Integration and Macroeconomic
Volatility,” IMF Working Paper No. 03-50, International Monetary Fund.
Loayza, N., H. Lopez, and A. Ubide (2001), “Co-movements and Sectoral Interdependence:
Evidence for Latin America, East Asia and Europe,” IMF Staff Papers 48, 367-395.
McKinnon, R., and G. Schnabl (2002), “Synchronized Business Cycles in East Asia: Fluctuations
in the Yen/Dollar Exchange Rate and China’s Stabilizing Role,” IMES Discussion Paper Series
No. 2002-E-13, Institute for Monetary and Economic Studies, Bank of Japan.
Mendoza, E. (2001), “Credit, Prices and Crashes: Business Cycles with a Sudden Stop,” NBER
Working Paper No. 8338.
Obstfeld, M. and K. Rogoff (1996), Foundation of International Macroeconomics, Cambridge,
MA: MIT Press.
Shin, K. and Y. Wang (2004), “Trade Integration and Business Cycle Co-movements: The Case
of Korea with Other Asian Countries,” Japan and the World Economy 16:2, 213-230.
Sims, C.A. (1980), “Macroeconomics and Reality,” Econometrica 48, 1-48.
Sims, C.A. (1986), “Are Forecasting Models Usable for Policy Analysis?” Federal Reserve Bank
of Minneapolis Quarterly Review, Winter, 2-16.
Sims, C.A. (1988) “Bayesian Skepticism on Unit Root Econometrics.” Journal of Economic
Dynamics and Control 12, 463-474.
Sims, C.A. and H. Uhlig (1991), “Understanding Unit Rooters: A Helicopter Tour,”
Econometrica 59, 1591-1599.
Tornell, A. and F. Westermann (2002), “Boom-Bust Cycles in Middle Income Countries: Facts
and Explanation,” NBER Working Paper No. 9219.
19