- Email: [email protected]
Financial stress and economic activity in some emerging Asian economies
Accepted Manuscript Title: Financial Stress And Economic Activity In Some Emerging Asian Economies Author: Emrah I. Cevik Sel Dibooglu PII: DOI: Reference:
S0275-5319(15)30022-2 http://dx.doi.org/doi:10.1016/j.ribaf.2015.09.017 RIBAF 398
To appear in:
Research in International Business and Finance
Received date: Accepted date:
23-2-2015 9-9-2015
Please cite this article as: Cevik, E.I., Dibooglu, S.,Financial Stress And Economic Activity In Some Emerging Asian Economies, Research in International Business and Finance (2015), http://dx.doi.org/10.1016/j.ribaf.2015.09.017 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
FINANCIAL STRESS AND ECONOMIC ACTIVITY IN SOME EMERGING ASIAN ECONOMIES Emrah I. Cevik, Sel Dibooglu*[email protected] University of Missouri St. Louis, 408 SSB, St Louis 63121, US Phone: +13145165530 1. Introduction
ip t
The global financial crisis that started in the US spread worldwide and sent shockwaves through the global financial system. Although policy makers have conducted
cr
several monetary and fiscal stimulus packages, credit conditions were tightened, the risk
us
premium on interbank borrowing significantly increased and trade credit decreased with falling demand, specifically for the capital goods and the manufacturing sectors in general. As
an
a result, the global financial crisis caused a significant decline in global economic activity and the developed as well as emerging economies suffered a significant slump not seen since the
M
1930s.
The emerging economies were affected by the global financial crisis through various
d
channels such as contagion, a decline in capital flows, and by trade channels. As a result,
te
financial stress (or financial instability) started to increase and output sharply decreased in a large numbers of emerging economies even in those that lacked a serious fiscal and financial
Ac ce p
imbalances. These developments in the global economy have emphasized the importance of identifying and assessing the linkages between financial stress and the real economy. In this regard, Lo Duca and Peltonen (2010) and others suggested that financial stress can affect economic activity through various channels. The first channel is called the financial accelerator in which shocks that affect the creditworthiness of borrowers tend to amplify output fluctuations where credit conditions of the financial system affect the willingness to provide credit to the economy. Second, factors that impact lenders’ balance sheets can magnify economic downturns mimicking weak bank capital; banks may become more reluctant to provide capital to the real sector, may be forced to deleverage leading to sharper
1
Page 1 of 26
economic downturns. Moreover, the structure and the development of the financial system affects how large is the interconnection between real and financial sectors in the economy. The onslaught of the financial crisis of 2007-2008 and the economic downturn that
ip t
followed highlighted the importance of the link between the financial sector and real economic activity in an interconnected world. In this regard, it is very important to measure
cr
financial stress in the economy by extracting signals from variables that that are thought to
capture some aspect of financial stress. How can measuring and monitoring financial stress
us
contribute to the design and implementation of proper macroeconomic policies? While in
an
normal times, the standard evaluation of macroeconomic prospects (maintaining full employment and price stability) is adequate and there are useful policy benchmark rules (such
M
as the Taylor rule), heightened periods of financial stress may call for policy responses that are different than the usual prescriptions. That is because a period of excessive financial stress
d
may produce substantial spillovers that constrain the credit intermediation capacity of the
te
financial sector and hence require policy to be recalibrated. A financial stress index not only is useful in evaluating macroeconomic prospects and designing monetary and fiscal policy
Ac ce p
measures, it is also useful in assessing financial conditions and fragility of the financial sector. Doing so contributes to a smoothly functioning financial system. For example, in periods of heightened financial stress it may not be sufficient to adjust short term interest rates. When markets suffer from illiquidity, there is increased uncertainty about asset values and lenders are unwilling to accept these assets as collateral; as such, credit intermediation declines and real economic activity is adversely affected. Under these circumstances, policymakers may have to resort to unconventional policy measures to deal with liquidity problems. Therefore measuring financial stress not only is important from the design and implementation of macroeconomic policy but also contributes indirectly to a smooth, robust and more resilient financial system.
2
Page 2 of 26
Therefore, the identification of states of financial stress is important for optimal policy design and hence a financial stress index can provide valuable benefits for policymakers. Louzis and Vouldis (2012) suggested that a composite financial stress index provides insights
ip t
into the propagation channels of specific events and the extent to which a financial crisis affects segments of the financial system. Grimaldi (2010) emphasizes attractive features of a
cr
financial stress index as being based on real time-high frequency data; broadly assessing the
level of stress of the overall financial system; and being based on a small group of indicators.
us
Perhaps due to the 2007-2009 global financial crisis, financial stress has been studied
an
extensively with a wide range of different components of financial stress. For instance, Hanschel and Monnin (2005) developed a financial stress index to measure the degree of
M
stress for the Swiss banking system. Illing and Liu (2006) proposed a financial stress index for the Canadian financial system. Hakkio and Keaton (2009) suggested a comprehensive
d
financial stress index to determine episodes of financial stress in the U.S. Cardarelli et al.
te
(2009) developed a financial stress index for 17 advanced economies. Grimaldi (2010) constructed a financial stress index to determine episodes of financial stress in the Euro area.
Ac ce p
Louzis and Vouldis (2012) suggested a financial systemic stress index for the Greek economy. Hollo et al. (2012) constructed a financial stress index for the Euro area and they called the index a composite indicator of systemic stress. Even before the global financial crisis, empirical studies in the literature specifically
focused on predicting financial crises in emerging economies by means of early warning indicators (Kaminsky et al. 1998, Demirgüç-Kunt and Detragiache 1998, Beckman et al. 2006, Davis and Karim 2008). However, Balakrishnan et al. (2011) argued that these studies are not appropriate to study episodes of financial stress. In this sense, there is growing literature that focuses gauging financial stress for emerging economies. Balakrishnan et al. (2011) developed a financial stress index where developments in the banking sector, stock
3
Page 3 of 26
exchange market, sovereign debt spreads, and exchange market were considered as components of financial stress. Cevik et al. (2013a) modified and extended the index proposed by Balakrishnan et al. (2011) with specific considerations for the Turkish economy.
ip t
Similarly, Cevik et al. (2013b) proposed a comprehensive index of financial stress for Bulgaria, Czech Republic, Hungary, Poland and Russia. These studies considered risks in the
cr
banking sector, securities and money market, external debt, sovereign spreads and trade finance as indicators of financial stress.
us
The literature that focuses on financial stress in Asian countries is more recent. The
an
Bank of Thailand (2010) proposed a monthly financial stress index for 1996-2009. Their financial stress index reflects risk to the functioning of the financial system in six areas: the
M
bank index beta, inverted term spreads, corporate bond spreads, stock index volatility, volatility of the government bond price index, and exchange rate volatility. Osorio et al.
d
(2011) proposed a quarterly financial conditions index for 2005-2010 for Asian countries.1
te
They considered interest rate market, exchange rate market, domestic credit market and equity market to calculate a financial stress index. Hwa et al. (2012) suggested a monthly financial
Ac ce p
stress index for the ASEAN-5 countries namely Indonesia, Malaysia, Philippines, Singapore and Thailand for 1997-2009. They emphasize four essential components to gauge financial stress: banking sector, equity markets, foreign exchange market and domestic bond markets. Lee et al. (2013) analyzed the financial stability of Korean banking system for the 2003 -2011 period by developing a composite financial stability index. The main objective of this study is to contribute to the literature by constructing a
financial stress index, studying episodes of elevated stress for Indonesia, South Korea, Malaysia, the Philippines, and Thailand. We also examine relationship between financial 1
Their country selection is based on data availability and the sample includes Australia, China, Hong Kong
SAR, India, Indonesia, Japan, Korea, Malaysia, New Zealand, Philippines, Singapore, Thailand and Taiwan Province of China. 4
Page 4 of 26
stress and economic activity for the sample countries. Our sample countries were chosen based on data availability. Our paper has several innovations: First, we consider additional factors such as risks in the banking sector and sovereign risk as additional indicators of
ip t
financial stress. Second, we use a contingent claim analysis and option pricing methods to calculate sovereign risk for the sample countries. In examining the relationship between
cr
financial stress and economic activity, we determine recessions according to the Harding and Pagan (2002) algorithm. As another contribution, we use a dynamic factor model to aggregate
us
the factors contributing to financial stress. The link between financial stress and the real
an
sector is an important one. An increase in financial stress can potentially produce substantial spillovers and systemic risks that constrain the credit intermediation capacity of the financial
M
sector. With financial stress, reduced asset values decrease the collateral that provides credit to the economy. At the same time, with deleveraging, banks lose capital and become reluctant
d
to lend to businesses. Moreover, strict credit standards and the accompanying credit crunch
te
with a negative outlook affects economic activity adversely. In those times, policymaking should include not only ordinary measures of monetary and fiscal policy, but policymakers
Ac ce p
have to be extra vigilant. In emerging markets dealing with financial stress and the specific policy actions depend on the source of stress. If financial stress is due to banking sector problems, policy actions need to focus on strengthening the banking sector. However if the source of financial stress is an external, conventional policy measures will be inadequate. In that case, international policy coordination, access to currency swap lines, and contingency funding facilities become very important in dealing with financial stress. Measuring financial stress not only provides a quantitative benchmark to assess the intensity of stress, but also gives an idea about the relative contribution of each financial indicator to the overall measure of stress and hence helps in formulating the policy response. In this paper we identify the
5
Page 5 of 26
specific factors that have contributed to elevated financial stress in sample countries in known historical crisis periods. A financial stress index would provide valuable information as a heightened index
ip t
helps fine tune economic policy. This is particularly important for emerging Asian countries because Gupta and Miniane (2009) showed that recessions accompanied by financial stress
cr
are substantially longer and deeper than others in Asian countries. In addition, deep recessions in Asia have resulted in substantial declines in potential output growth and this may point to
us
permanent effects of high financial stress. Therefore, we modify and extend the index
an
proposed by Balakrishnan et al. (2011) for South Asian countries with specific considerations for Indonesia, South Korea, Malaysia, the Philippines, and Thailand. Section 2 represents the
M
effect of the global crisis on the Asian countries in question. Section 3 explains the components of the financial stress index. Section 4 elaborates on the construction of the index
d
and shows our indicators capture key aspects of financial stress. Section 5 examines the
te
relationship between financial stress and economic activity and shows how an elevated index tended to coincide with known crisis episodes. Finally, we show the constructed index
Ac ce p
provides valuable information on economic activity. 2. Some Key Economic Variables during the Global Financial Crisis
As useful background, we briefly summarize recent developments on economic
growth, inflation rates, foreign trade, equity prices, and international reserves for the sample countries.
Figure 1 shows economic growth during the global financial crisis for individual
economies. As seen in Figure 1, economic growth significantly decreased in all countries
except for Indonesia between 2008 and 2009 due to the global financial crisis. Specifically, negative economic growth was observed in South Korea, Malaysia and Thailand in 2009. Fig. 1 Yearly Growth Rate of Reel GDP Fig. 2 Key Economic Variables
6
Page 6 of 26
We also present other key economic variables during the global financial crisis in Figure 2. Panel (a) in Figure 2 shows annual inflation based on consumer prices. Inflation rates for selected Asian countries display a similar pattern where we observe an increase in inflation
ip t
rates in the middle of 2007. Then, there was a substantial decrease in inflation rates in all countries and inflation rates have reached the pre-crisis levels at beginning of 2010.
cr
Panel (b) shows the evolution of foreign international reserves (excluding gold) during the crisis. All countries have experienced a decline in foreign reserves due to capital outflows
us
during the crisis. Also, South Korea and Malaysia showed the largest decline in reserves.
an
Panel (c) presents movements in equity market returns where equity markets exhibit a similar pattern during the crisis. Falling equity returns are evident in all countries in 2007 and
M
2008 in which the largest stock market declines have occurred in Indonesia (–109%) and South Korea (–101%),
d
Finally, we show behavior of foreign trade (sum of merchandise exports and imports)
te
during this crisis in Panel (d) of Figure 2. As can be seen in the Figure, total merchandise trade in these countries collapsed at the end of 2008 due to a sharp decline in economic
Ac ce p
activity of trading partners and perhaps lowers availability of trade credit.
The figure presented above shows that the crisis affected Asian countries similarly and
the recovery seems to have been rapid as most of the variables reached pre-crisis levels at the beginning of 2010. IMF (2010) emphasized on three factors for a fast recovery in Asian countries. First regional heavyweights such as China was not affected much by the crisis and showed the fastest recovery which helped other Asian countries. The second factor is related to trade and finance that started to normalize in February 2009 and hence overall economic activity came back to pre-crisis levels very quickly as compared to other emerging countries. The last factor is related to the region’s aggressive countercyclical response that caused a speedy recovery in Southeast Asian countries. 7
Page 7 of 26
3. Financial Stress and its Indicators
Even though episodes of financial stress are defined in various ways in the literature, studies in this area have generally emphasized two main underlying stress phenomena: increase in uncertainty and changing expectations. Grimaldi (2010) defined the level of stress
ip t
as being determined by the interaction between financial vulnerabilities and the size of
shocks. Balakrishnan et al. (2011) emphasized four fundamental characteristics of financial
cr
stress: large shifts in asset prices, an abrupt increase in risk and/or uncertainty, liquidity
us
droughts, and concerns about the health of the banking system. Similarly, Hakkio and Keeton (2009) argue that episodes of financial stress must involve at least one of the following:
an
increased uncertainty about fundamental value of assets, increased asymmetry of information, decreased willingness to hold risky assets, and decreased willingness to hold illiquid assets.
M
Studies on financial stress in the literature generally consider five essential components: the financial intermediaries sector, money markets, equity markets, bond
d
markets, and foreign exchange markets. However, financial stress in emerging economies is
te
not only related to financial markets but also some external flow variables. Therefore,
Ac ce p
focusing only on financial markets may not be adequate for developing a financial stress index as there are additional sources of financial stress such as external debt, current account deficits and sovereign risk.
In order to account for different aspects of financial stress in South Asian economies,
we augment conventional factors with measures of external debt and as well as specific measures of sovereign risk. As a result, our financial stress index reflects the developments in the riskiness in the banking sector, security market risk, currency risk, external debt, and sovereign risk.
3.1. Riskiness of the Banking Sector
A sound financial sector is an important element for the real economy. The financial sector is instrumental in providing credit, payment mechanisms, and fund transfer services to 8
Page 8 of 26
households and businesses alike and in managing potential risks. Therefore, a healthy financial sector should lower the cost and risk of producing and trading of goods and services among market participants. Moreover, a large number studies documented the effects of a
ip t
weak banking system as an important factor behind the Asian crisis in 1998. For instance, Turner (2007) showed that banking system weakness, poor macroeconomic policies, and a
cr
massive shock all contributed to the Asian crisis. Hence riskiness of the banking sector is an important component of a financial stress index. In this paper, we consider the volatility of
us
banking sector index in the stock market and use the following EGARCH model that is
an
proposed by Nelson (1991) to obtain banking sector index returns volatility:2,3,4,5
(
( )
)
M
(1) rB ,t = X t′θ + ε t log σ t2 = ω + β log σ t2− j + α
ε t −1 ε + γ t −1 σ t −1 σ t −1
te
d
where rB,t indicates banking sector index returns, X t′ includes a constant and autoregressive variables of the banking sector index returns, εt is an error term and σt is the conditional
Ac ce p
variance. There are two features of EGARCH model compared to a GARCH model. First; because the left-hand side is the log of the conditional variance, the leverage effect is exponential, rather than quadratic, and that forecasts of the conditional variance are
2
Hanschel and Monnin (2005) suggested using both market and balance sheet data of banking sector to gauge
financial stress. We consider such balance sheet variables (e.g. total deposit and total loan of banking sector) but these variables did not improve the performance of our financial stress index in identifying specific events. 3
Note that although several volatility models are proposed in the literature, in order to account for different
specifications in asset returns, Andersen and Bollerslev (2001) found that GARCH and stochastic volatility models provide good volatility forecasts particularly in high frequency data. 4
Due to banking sector stock index data availability, we use the financial services sector stock index for
Indonesia, Malaysia and Philippines. 5
We consider other types of GARCH specifications, but the EGARCH (1,1) model has better fit according to the
model selection information criteria. 9
Page 9 of 26
guaranteed to be nonnegative. The other advantage is that the EGARCH model allows for the presence of leverage effects in the volatility. 3.2. Security Market Risk
It is well known that the stock markets are important indicators of a healthy financial
ip t
system and studies in the literature use securities market as a component of the financial stress index. Therefore an EGARCH model can be employed to measure stock market volatility as
(
( )
)
ε t −1 ε + γ t −1 σ t −1 σ t −1
an
(2) rS ,t = X t′θ + ε t log σ t2 = ω + β log σ t2− j + α
us
cr
follows:
where rS,t indicates stock market index returns, X t′ includes a constant and autoregressive
M
variables of stock market returns, εt is an error term and σt is the conditional variance. 3.2. Currency Risk
d
Currency risk is another important component of financial stress for developing
te
economies. We consider Exchange Market Pressure Index (EMPI) proposed by Girton and Roper (1977) as a component of our financial stress index. Therefore, we construct the EMPI
Ac ce p
by combining exchange rate movements, changes in international reserves, and changes in the overnight interest rate relative to the US as follows: (3) EMPI t =
∆et
σ ∆e
−
∆ Re st
σ ∆ Re s
+
(
∆ it − iUS ,t
σ ∆ (i − i ) t
)
US , t
where ∆et and ∆Rest are 12-month percent changes in the exchange rate and total foreign international reserves minus gold, and it and iUS represent the overnight interest rate for the home country and the US, respectively. Similarly σx denotes the standard deviation of variable x, (x = the changes in the exchange rate, total reserves, and overnight interest rate). Even
though EMBI is used to identify crisis periods in the literature, Bussierre and Fratzscher
10
Page 10 of 26
(2006) emphasized the advantage of EMPI in capturing both successful and unsuccessful speculative attacks. 3.3. External Debt
Although external debt is important for sustainable economic growth in emerging
ip t
economies, short term external debt played a significant role in the Asian and Russian crises
and hence excessive increase in the short term external debt can cast a doubt on sustainability
cr
of such debt. Hence, we consider the ratio of short term external debt to GDP that is widely
us
examined as a potential indicator of financial crises for developing countries (see Sachs et al. (1996); Abiad (2003); Bussiere and Fratzscher (2006)). As such, an increase in the ratio of
an
short term external debt to GDP may be expected to cast a doubt on the sustainability of external debt.6
M
3.4. Sovereign Risk
Another component of our financial stress index is sovereign risk that is specifically
d
important for emerging markets. Because interest rate spreads between home country and the
te
US can be used as an indicator of risk perception in the home country, sovereign bond spreads (the difference between home country’s Emerging Market Bond Index (EMBI) and 10 year
Ac ce p
US Treasury yield) are used in constructing the financial stress index in the literature (see Balakrishnan et al. (2011) and Cevik et al. (2013a)). Due to EMBI spreads data unavailability for our sample countries, we employ the so called Contingent Claim Analysis (CCA) to compute sovereign risk for the countries in question. The CCA is a framework that combines balance sheet information with commonly
used risk measurement tools to construct a marked-to-market balance sheet to identify and quantify risks. In essence, the CCA approach models firm equity as a contingent claim on a firm’s assets. It is a contingent claim because the value of firm equity depends on the value of
6
As external debt is measured quarterly in these countries, we use the cubic spline method to obtain monthly
external debt series. Even though there are several methods to interpolate data in the literature, we use the cubic spline method because it is simple, fast, efficient and stable. 11
Page 11 of 26
the firm’s assets and the default-free value of the firm’s liability at a particular point in time (Geuorguiev et al., 2009). The firm equity can be written as a call option as follows: (4) E = max[ A − DB, 0]
ip t
where A is firm assets and DB is the default barrier. In Equation (6), when the firm can generate enough cash to cover its current debt obligations (A ≥ DB), the value of firm’s equity
cr
is equal to difference between the value of firm’s assets (A) and the current debt obligations of
us
the firm or default barrier (DB). On the other hand, when the firm’s assets do not cover the current debt obligations (A ≤ DB), firm equity equals zero.
an
While the CCA analysis is commonly used to calculate default probability of firms in the finance literature, there have been studies that measure the sovereign risk profile of
M
countries based on the CCA approach. For instance, Gapen et al. (2008) and 2008) employed the CCA analysis to determine the sovereign risk profile for 12 developing economies. They
d
showed that the CCA generates similar results to EMBI and CDS spreads. Also Keller et al.
te
(2007) examined changes in Turkey’s sovereign risk profile by using the CCA and found a strong correlation between sovereign risk indicators based on the CCA and EMBI spreads.
Ac ce p
According to Gapen et al. (2008) and Keller et al. (2007), the sovereign balance sheet
of a country can be represented as in Table 1.
In order to derive default probabilities, some assumptions must be made about the
seniority structure of a sovereign country’s liabilities. To derive external default probabilities, external debt is assumed to be the more senior liability, whereas domestic debt and base money are assumed to represent the equity portion of the sovereign balance sheet and thus can be viewed as a contingent claim on the residual value of sovereign assets. The sovereign is assumed to default whenever the value of its implied assets falls below a distress barrier. The difference between the asset value and the distress barrier, scaled by the asset volatility, is
12
Page 12 of 26
referred to as the distance-to-distress, while the area of the distribution that falls below the distress barrier represents the sovereign’s default probability (Keller et al., 2007). In this context the market value of equity can be modeled by using Black and Scholes
ip t
(1973) and Merton (1974) formula for call options: (5) E = AN ( d1 ) − DBe − rt N ( d 2 )
σ A T , d 2 = d1 − σ A T , r is the risk
us
term external debt,7 d1 = ln ( A DB ) + ( r + 0.5 × σ 2A ) T
cr
where A is the assets, DB, the default barrier, is equal to the sum of short term and half of long
free interest rate8, T is the time to maturity of the default barrier,9 N(d) is a cumulative
an
probability distribution function for a standard normal variable, and σA is the standard deviation of assets. Under the Merton (1974) bond pricing model’s assumptions, the volatility
Aσ A N ( d1 ) E
d
(6) σ E =
M
of the equity is:
(
)
ln ( A DB ) + r − 0.5 × σ 2A T
σA T
Ac ce p
(7) DD =
te
Distance to default (DD) can be written as:
Under the normal distribution which is implied by Merton (1974), the default
probability or the risk-neutral default probability (RNDP) can be calculated as N(-DD). Although Equations (7) and (8) can be solved simultaneously, Crosbie and Bohn
(2003) argue that market leverage moves around far too much for Equation (8) to provide reasonable results. In order to overcome this problem we used an iterative procedure. First, we set an initial value of A = E + DB and compute the standard deviation of the log asset returns.
7
External debt that matures within one year is defined as short term with any other liability deemed long term.
8
The one year US Treasury Bill rate is taken to be the risk free interest rate.
9
Although the time horizon of the estimate of default risk can vary in the literature, Gapen et al. (2008) and
Keller et al. (2007) considered one year as the time horizon. Therefore we use t=1 in the model estimation. 13
Page 13 of 26
Then we insert A and σA into Equation (7) and compute new values of A and σA. The procedure is repeated until convergence where the sum of squared differences between consecutive asset values is less than 10-10.
ip t
4. Empirical Results 4.1. Data
We construct a financial stress index to gauge the degree of the financial stress in
cr
Indonesia, Malaysia, Philippines, South Korea and Thailand for the periods of 1995-2013.
us
Data on overnight interest rate, and base money were taken from the IMF-IFS database. Banking sector stock market data were collected from Bloomberg. Data on the stock market
an
index, the foreign exchange rate, international reserves, and real GDP were obtained from World Bank-GEM database. Finally, data on domestic currency debt and external debt were
M
obtained from national central banks. Before the aggregation of the components, all of the components are standardized by subtracting their means and dividing by their standard
te
4.2. Aggregation of the Components
d
deviations.
Constructing of a comprehensive index depends on aggregation of the variables and
Ac ce p
there are various aggregation methods in the literature that can be classified into varianceequal weight method (e.g. Balakrishnan et al. 2011; Cardarelli et al. 2009; Hanschel and Monnin, 2005), principal components analysis (Hakkio and Keeton, 2009; Illing and Liu, 2006; Cevik et al. 2013a; Cevik et al. 2013b; the Federal Reserve Bank of Kansas City and the Federal Reserve Bank of St. Louis), and, recently, portfolio theory based aggregation schemes that take into account the correlation structure of stress indicators in order to quantify the level of systemic stress (Hollo et al. 2012, and Louzis and Vouldis, 2012). In this paper, we employ a dynamic factor model to construct financial stress index for sample countries. We assume that each of the variables above captures some aspect of financial stress and hence all variables are likely to move together according to the level of
14
Page 14 of 26
financial stress in the economy. In this context, the common component for these variables represents financial stress in an economy. Dynamic factor model can be defined as follows: (8) yt = χ t + ε t , ε t ∼ N ( 0, ψ )
ip t
where yt is indicator of financial stress index, χ t = ΛFSI t and ψ is assumed to be diagonal.
cr
Note that the idiosyncratic components of the observable variables are assumed to follow
AR(1) process. The common factor or financial stress index can be estimated by using the
us
entire set of indicators. Dynamics of financial stress can be captured by the following
(9) FSI t =
p
ρi FSI t −i + υt , υt ∼ N ( 0, Σ )
M
i =1
an
autoregressive process:10
In periods of high financial stress, the pressure on the value of local currencies
d
increases, stock market becomes more volatile, sovereign risks rise and doubts on
te
sustainability of external debt can be expected to rise. Therefore we expect that an increase in the EMPI, stock market volatility, banking sector volatility, the ratio of total external debt to
Ac ce p
foreign reserves, and sovereign risk raise financial stress. 4.3. Extraction of Financial Stress via a Dynamic Factor Model
The dynamic factor model results for Indonesia, Malaysia, Philippines, South Korea
and Thailand are given in Table 2. It should be noted that all variables are standardized before the aggregation and hence coefficients for the variables represent the effect of one-standarddeviation change in the respective variable on the index. According to the coefficients of components, an increase in the EMPI, stock market volatility, banking sector volatility, sovereign risk, and short term external debt- GDP ratio raises financial stress in all countries. Results in Table 2 show that EMPI is not statistically significant in Thailand. In addition, short term external debt- GDP ratio is not statistically significant for Malaysia and South 10
We select lag lengths by using Schwarz Bayesian information criterion (BIC). 15
Page 15 of 26
Korea. The Bayesian Information Criterion, BIC, suggests a first degree autoregressive process for Indonesia and Thailand. On the other hand, BIC suggest two lags for Malaysia and Philippines and three lags for the South Korea.11
ip t
The estimated coefficients in Table 2 indicated that a one-standard-deviation increase in EMPI has similar effects on financial stress in Indonesia, Malaysia and Philippines.
cr
Volatility in the stock market and banking sector index seems to be an important source of
us
financial stress in all countries. In this context, banking sector volatility seems to be a modest factor contributing to financial stress in Malaysia. Short term external debt component has
an
similar effects on the financial stress index in Indonesia, Philippines and Thailand but it is found to be statistically insignificant for Malaysia and the South Korea. Sovereign risk and
M
EMPI seem to contribute to financial stress differently across countries.
4.4. The Evolution of Financial Stress
d
In this section, we examine the evolution of financial stress for the Southeast Asian countries.
te
We are particularly interested in whether heightened episodes of financial stress correspond to
Ac ce p
known episodes of downturns in the real economy and the behavior of financial stress over the business cycle. The behavior of financial stress in the sample countries is given in Figure 1 where the shaded areas correspond to recessions that are determined according to the Harding and Pagan (2002) algorithm. Panel a) of Figure 3 indicates that the Indonesia Financial Stress Index (I-FSI) tracks
recessions closely in the sample.12 Specifically the I-FSI increased at the beginning of the 1997 and remained above zero until end of 2001. Although there were some internal problems (such as under-supervised banks, extensive crony capitalism, corruption, monopoly power,
11
The sum of autoregressive parameters is found to be higher than 0.9 for all countries except for Indonesia and
the result is consistent with empirical results found in Carpenter et al. (2014). 12
Due to banking sector index data availability, the Indonesia Financial Stress Index starts from January 1996. 16
Page 16 of 26
and growing short term debt) in the Indonesian economy, it was the second in terms of export growth in 1996 among the Asiean-5 countries before the Asian crisis. Moreover, the budget surplus averaged over one percent in the previous four years, while credit growth was modest.
ip t
Hence, it can be said that the crisis in 1998 appeared in Indonesia due to weak financial sector, political uncertainty and contagion effect rather than poor traditional economic
cr
fundamentals (Cerra and Saxena, 2000). Therefore, the index has reached its highest level in
January 1998 due to Asian crisis in 1997-98. During the 2000-2002 period, because of several
us
financial and economic crises in the world (such as the stock market crash in 2000, the
an
Turkish crisis in 2001, the Argentina crisis in 2002, Enron scandal and 9/11 attacks in 2001), the financial stress index remained above zero until the first half of 2002. Then the I-FSI
M
decreased and remained under zero. Finally, financial stress started to rise again in the beginning of 2008 with the start of the global financial crisis.
d
The Malaysia Financial Stress Index (M-FSI) is plotted in Panel b of Figure 3.13 As
te
seen in the figure, the M-FSI captures successfully all periods of recessions in Malaysia.
Ac ce p
Financial stress index remained under zero until the beginning of the 1998 in Malaysia. Due to the Asian crisis, financial stress index started to increase at the beginning of 1998 and it has reached the highest value at May 1998. After that, financial stress in Malaysia decreased but remained below zero until end of 2001 as a result of crises in developed and emerging economies. Financial stress in Malaysia started to decrease at the beginning of 2002, and then the M-FSI was below zero until end of 2008. As in other emerging economies, the global financial crisis caused an increase in the financial stress index in Malaysia but the effect of global financial crisis on financial stress seems to be fairly limited as there is no substantial increase in financial stress during the financial crisis in Malaysia.
13
Due to EMBI data availability, the Malaysia Financial Stress Index starts from October 1996. 17
Page 17 of 26
According to Panel c of Figure 3, the financial stress in the Philippines was high at the beginning of the 1998 and it remained above zero until first half of 2003 except for the beginning of 2002.14 As in other countries in the region, financial stress rose during the Asian
ip t
crisis in 1997-98 in the Philippines and the index reached the highest value in September 1998. Although financial stress was below zero between 2004 and 2008, it started to increase
cr
at the end of 2007 due to the global financial crisis.
us
The South Korean Financial Stress Index (SK-FSI) remained above zero at the end of 1996 ant it started to increase at the beginning of 1997. Although there were no serious
an
macroeconomic imbalances in the Korean economy before the Asian crisis, Korea was hit by the Asian crisis during the fall of 1997, just a few months after other countries of the region.
M
For instance, the annual growth rate of GDP was still above 5% in the second quarter of 1997. During the summer of 1997, the Korean Won was not under pressure because it was the only
d
currency in the region to have depreciated in real terms in previous years. On the other hand,
te
the financial situation started to deteriorate before the financial crisis in Korea. During the first quarter of 1997, a series of bankruptcies hit some of the big conglomerates due to the
Ac ce p
combination of their huge rate of indebtedness and the increase in foreign interest rates. By the end of October, the Won was overvalued in comparison with the other Asian currencies that lost some value since the summer. On October 21st, Korea had to request the assistance of the IMF. The first of the five IMF rescue plans for Korea was launched on December 5th. The government gave up the peg on December 16th which precipitated a depreciation of the Korean Won by 43% (Robert, 2005: 35-36). The financial stress index reached the highest level in January 1998 due to the Asian crisis. After that financial stress started to decrease in South Korea and the index remained below zero between 2003 and 2007. As in other Asian countries, the global financial crisis led to elevated financial stress in South Korea. 14
Due to EMBI spreads data availability, the Philippines Financial Stress Index starts from January 1998. 18
Page 18 of 26
The Thailand Financial Stress Index (T-FSI) is plotted in panel e of Figure 3. At the beginning of the sample, financial stress seems to be very low in Thailand. However, financial stress in Thailand started to rise earlier than in other emerging Asian countries before the
ip t
Asian crisis. This result is consistent with our expectations as the Asian crisis first appeared in Thailand. As seen in Figure 3, the financial stress index started to increase at the beginning of
cr
1996 in Thailand and the index has reached the highest level in September 1998. When we look at the chronology of the crisis, pressure on the Thai currency started to rise at the
us
beginning of 1996 due to widening external deficit, slowing economy, emerging strains in the
an
financial system, and growing political instability, which, taken together, started to weaken investor confidence. The vulnerability in Thailand stemmed from Thai finance companies
M
with their heavy short term borrowing abroad. After the attempted rescue of the largest Thai finance company failed in May 1997, the weakening of investors’ confidence intensified in
d
Thailand. On 2 July 1997 the Bank of Thailand abandoned the exchange rate parity which had
te
fixed the Thai baht to the US dollar and allowed the currency to float. In late July, the Thai government requested financial support from the IMF, resulting in an international rescue
Ac ce p
package totaling US$ 17 billion (Fries et al., 1999). Due to the turmoil that followed, the financial stress index remained above zero between 2000 and 2001. Then, the index decreased until 2008 when the global financial crisis led to a rise in financial stress in Thailand as well. Overall not only does our financial stress index track known episodes of financial turmoil well in the sample, but it seems that financial stress precedes economic downturns in sample countries. In that regard, financial stress seems to be a leading economic indicator in the sample. 5. The Relationship between Financial Stress and Economic Activity
In this section, we examine the relation between financial stress and economic activity. The transmission channels between financial stress and economic activity have been widely
19
Page 19 of 26
argued in the literature where the role of financial leverage and bank capital channel are emphasized. Moreover, Hakkio and Keaton (2009) argued the role of uncertainty about the price of financial assets and the economic outlook. Davig and Hakkio (2010) emphasized
ip t
“real options” and financial accelerator” for the transmission channels between financial stress and economic activity. Due to the global financial crisis, there have been a growing
cr
literature that focus on the relationship between financial stress and economic activity (Li and
us
St-Amant, 2007; Claessens et al., 2008; Hakkio and Keeton, 2009; Cardarelli et al., 2010). We examine the relationship between the financial stress index and economic activity
an
using a bivariate vector autoregression (VAR) model. Economic activity is tracked via some economic indicators. For example, Hakkio and Keaton (2009), and Davig and Hakkio (2010)
M
employed the Chicago Fed National Activity Index, which combines 85 macroeconomic time series as a measure for economic activity. Cevik et al. (2013a) and Cevik et al. (2013b)
d
considered the yearly growth rate in the industrial production, foreign trade and gross fixed
te
capital formation as a proxy for economic activity. Louzis and Vouldis (2012) calculated an index of economic activity as the first principal component of five variables namely,
Ac ce p
unemployment, manufacturing index, exports, imports and producers price index. Hollo et al. (2012) used industrial production index as a proxy for economic activity. As in Hollo et al. (2012), we consider yearly growth rate of the industrial production index (GIP) as proxy for economic activity. Then, we employ a bivariate VAR model to examine dynamic relation between financial stress and economic activity in all countries.15 We orthogonolize the innovations using a Cholesky decomposition where the financial stress shock is ranked first: it is assumed that real economic activity shocks have no contemporaneous effects on financial
15
In all cases the lag length is determined by the Schwarz Bayesian Information Criterion (BIC). 20
Page 20 of 26
stress and they may affect financial stress after a month. The impulse response functions for the financial stress index and GIP are presented in Figure 4 for each country in our sample16. The results in Figure 4 indicate that the responses of growth in industrial production to
ip t
a shock in financial stress are negative and significant for all countries. Note that the greatest impact of financial stress on economic activity as measured by the industrial production
cr
growth seems to be in Malaysia. The effect of financial stress on economic activity is nearly
us
statistically insignificant in the Philippines. However, the responses of the financial stress index to an industrial production shock are not statistically significant and industrial
an
production seems to be exogenous to financial stress in all sample countries. Overall our results indicate that the financial stress causes significant economic slowdowns in Southeast
M
Asia in our sample. 5. Conclusions
d
This paper constructed a financial stress index using riskiness in the banking sector,
te
security market risk, currency risk, external debt and sovereign risk. Using data from Indonesia, South Korea, Malaysia, the Philippines, and Thailand we employ a dynamic factor
Ac ce p
model to aggregate the aforementioned factors into a single financial stress index and examine the relationship between financial stress and economic activity. Indeed our results show that when foreign exchange risks rise, stock market becomes more volatile, sovereign risks rise and when there are doubts on sustainability of external debt, financial stress rise substantially. Not only does our financial stress index track known episodes of financial turmoil well in the sample, but it seems that financial stress precedes economic downturns in sample countries. As such, financial stress seems to be a leading economic indicator in the sample. In that regard, the financial stress index contains valuable information for policy makers.
16
We also derive impulse response functions with the order of the shocks reversed for all countries and get
qualitatively similar results. These are available upon request. 21
Page 21 of 26
Our results show that the financial stress index has important bearings on economic activity in our sample. A bivariate VAR model of financial stress and industrial production shows that financial stress causes significant economic slowdowns. References
ip t
Abiad, Abdul (2003). Early Warning Systems: A Survey and Regime Switching Approach. IMF Working Paper, No: 03/32.
cr
Balakrishnan, Ravi, Stephan Danninger, Selim Elekdag and Rina Tytell (2011). The
Finance and Trade, Vol. 47, Supplement 2, pp. 40-68.
us
Transmission of Financial Stress from Advanced to Emerging Economies. Emerging Markets
an
Bank of Thailand (2010). Thailand’s Financial Stress Index.
http://www.bot.or.th/English/MonetaryPolicy/Inflation/PaperInFrame/100_Chapter_4_E_Apr
M
10.pdf
Beckmann, Daniela, Lukas Menkhoff and Katja Sawischlewski (2006). Robust Lesson
d
about Practical Early Warning Systems. Journal of Policy Modeling, Vol. 28, pp. 163-193.
te
Black, Fischer and Myron Scholes (1973). The Pricing of Options and Corporate
Ac ce p
Liabilities. Journal of Political Economy, Vol. 81, pp. 637-659. Bussiere, Matthiue and Marcel Fratzscher (2006). Towards a New Early Warning
System of Financial Crises. Journal of International Money and Finance, Vol. 25, pp. 953973.
Cardarelli, Roberto, Selim Elekdag and Subir Lall (2009). Financial Stress,
Downturns, and Recoveries. IMF Working Paper, No 09/100. Carpenter, Seth, Selva Demiralp, Bernd Schlusche and Zeynep Senyuz (2014). Measuring Stress in Money Markets: A Dynamic Factor Approach. Economics Letters, Vol. 125, pp. 101-106.
22
Page 22 of 26
Cerra, Valerie and Sweta Chaman Saxena (2000). Contagion, Monsoons, and Domestic Turmoil in Indonesia: A Case Study Asian Currency Crisis,” IMF Working Paper, WP/00/60.
in Turkey. Journal of Policy Modeling, Vol. 35, pp 370-383.
ip t
Cevik, Emrah I., Sel Dibooğlu and Turalay Kenc (2013a). Measuring Financial Stress
cr
Cevik, Emrah I., Sel Dibooğlu and Ali M. Kutan (2013b). Measuring Financial Stress in Transition Economies. Journal of Financial Stability, Vol. 9, pp 597-611.
us
Claessens, Stijn, M. Ayhan Köse and Marco E. Terrones (2008). Financial Stress and
an
Economic Activity. Journal of BRSA Banking and Financial Markets, Vol. 2, pp. 11-24. Crosbie, Peter and Jeff Bohn (2003). Modeling Default Risk. http://www.ma.hw.ac.uk
M
/~mcneil/F79CR/Crosbie_Bohn.pdf
Davig, Troy and Craig Hakkio (2010). What Is the Effect of Financial Stress on
te
Quarter, 35-62.
d
Economic Activity?, Federal Reserve Bank of Kansas City Economic Review, Second
Davis, E. Philip and Dilruba Karim (2008). Comparing Early Warning Systems for
Ac ce p
Banking Crises. Journal of Financial Stability, Vol. 4, pp. 89-120. Demirgüç-Kunt, Aslı and Enrica Detragiache (1998). The Determinants of Banking
Crises in Developing and Developed Countries. IMF Staff Papers, Vol. 45 (1), pp. 81-109. Fries, Steven, Martin Raiser and Nicholas Stern (1999). Stress Test for Reforms.
Economics of Transition, Vol 7, pp. 535-567. Gapen, Michael T, Dale F. Gray, Cheng Hoon Lim and Yingbin Xiao (2008).
Measuring and Analyzing Sovereign Risk with Contingent Claims, IMF Staff Papers, Vol. 55 (1), pp. 109-148. Geuorguiev, Nikolay, Rodney Ramcharan, Alison Stuart and Burcu Aydin (2009). South Africa: Selected Issues. IMF Country Report No. 09/276.
23
Page 23 of 26
Girton, L. and D. E. Roper (1977). A Monetary Model of Exchange Market Pressure Applied to the Postwar Canadian Experience, American Economic Review, Vol. 67, pp. 537548.
Euro Area. Working Paper Series No 1214. European Central Bank.
ip t
Grimaldi, Marianna Blix (2010). Detecting and Interpreting Financial Stress in the
cr
Gupta, S., and J. Miniane (2009). Recessions and Recoveries in Asia: What Can the Past Teach Us About the Present Recession? ADBI Working Paper 150. Tokyo: Asian
us
Development Bank Institute. Available:
an
http://www.adbi.org/workingpaper/2009/09/02/3324.recessions.recoveries.asia/ Hakkio, S. Craig and William R. Keeton (2009). Financial Stress: What Is It, How
Economic Review, Second Quarter, 5-50.
M
Can It Be Measured, and Why Does It Matter?. Federal Reserve Bank of Kansas City
d
Hanschel, Elke and Pierre Monnin (2005). Measuring and Forecasting Stress in the
te
Banking Sector: Evidence From Switzerland. BIS Working Paper, No. 22. Harding, Don and Adrian Pagan (2002). Dissecting the Cycle: A Methodological
Ac ce p
Investigation, Journal of Monetary Economics, Vol. 49, pp 365-381. Hollo, Daniel, Manfred Kremer and Marco Lo Duca (2012): CISS - A Composite
Indicator of Systemic Stress in the Financial System. Working Paper Series No 1426. European Central Bank.
Hwa, Tng Boon, Kwek Kian Teng and Sheng (2012). Financial Stress in Asean-5
Economies from the Asian Crisis to the Global Crisis. The Singapore Economic Review, Vol. 57 (2), pp. 1-24. Illing, Mark and Ying Liu (2006). Measuring Financial Stress in a Developed Country: An Application to Canada. Journal of Financial Stability, Vol. 2, pp. 243-265. IMF (2010). World Economic Outlook. Rebalancing Growth April 2010.
24
Page 24 of 26
Kaminsky, Graciel A., Saul Lizondo and Carmen M. Reinhart (1998). Leading Indicators of Currency Crises. IMF Staff Papers, Vol. 45, No. 1, pp. 1-48. Keller, Christian, Peter Kunzel and Marcos Souto (2007). Measuring Sovereign Risk
ip t
in Turkey: an Application of the Contingent Claims Approach, IMF Working Paper, No 07/233.
cr
Lee, Jong Han, Jaemin Ryu and Dimitrios P. Tsomocos (2013). Measures of Systemic Risk and Financial Fragility in Korea. Annals of Finance, Vol. 9, pp. 757-786.
us
Li, Fuchun and Pierre St-Amant (2007). Financial Stress, Monetary Policy, and
an
Economic Activity. 42nd Annual Meeting of the Canadian Economics Association, Vancouver, B.C.
M
Lo Duca, M., Peltonen, T. (2010). Macro-financial Vulnerabilities and Future Financial Stress - Assessing Systemic Risks and Predicting Systemic Events. Working Paper
d
Series No 1311. European Central Bank.
te
Louzis, P. Dimitrios and Angelos T. Vouldis (2012). A Methodology for Constructing a Financial Systemic Stress Index: An Application to Greece. Economic Modelling. Vol. 29,
Ac ce p
pp. 1228-1241.
Merton, Robert C. (1974). On the Pricing of Corporate Debt: The Risk Structure of
Interest Rates. Journal of Finance, Vol. 29, pp. 449-470. Nelson, D. B. (1991). Conditional Heteroskedasticity in Asset Returns: A New
Approach, Econometrica, Vol. 59, pp. 347-370. Osorio, Carolina, Runchana Pongsaparn and D. Filiz Unsal (2011). A quantitative
Assessment of Financial Conditions in Asia. IMF Working Paper, WP/11/173. Robert, Marc Y. (2005). The 1997-1998 Korean Crisis: Domestic or External Causes?. Journal of Policy Modeling, Vol. 27, pp. 33-53.
25
Page 25 of 26
Sachs, J., A.Tornell and A. Velasco (1996). Financial Crises Emerging Markets: The Lesson from 1995. NBER Working Paper, No:W5576, pp.147-198. Turner, Philip (2007). Are Banking Systems in East Asia Stronger?. Asian Economic
ip t
Policy Review, Vol.2 (1), pp.75-95.
cr
Fig. 3 Financial Stress for South Asian Countries
us
Fig. 4 Impulse Response Functions
Table 1. The Balance Sheet of a Sovereign Country
Liabilities External Debt Equity
an
Assets International Reserves Net Financial Assets (Discounted Value of Primary Fiscal Surpluses) Value of Monopoly over Issue of Money Other Assets less Guarantees
M
Domestic Debt Base money
Ac ce p
te
d
Table 2 Dynamic Factor Model Results Variables Indonesia Malaysia Philippines S. Korea Thailand EMPI 0.072*** 0.052*** 0.085** 0.161*** 0.032 Stock market volatility 0.433*** 0.055*** 0.200*** 0.185*** 0.253*** Banking sector volatility 0.552*** 0.060*** 0.260*** 0.150*** 0.256*** Sovereign risk 0.014** 0.047*** 0.182*** 0.143*** 0.019*** Short term external debt / GDP 0.030* -0.002 0.075*** 0.013 0.025* FSIt-1 0.845*** 1.843*** 1.381*** 1.792*** 0.962*** FSIt-2 -0.858*** -0.440*** -1.191*** FSIt-3 0.357*** Note: ***, ** and * indicates statistically significant coefficients at 1%, 5% and 10% level respectively.
26
Page 26 of 26
S0275-5319(15)30022-2 http://dx.doi.org/doi:10.1016/j.ribaf.2015.09.017 RIBAF 398
To appear in:
Research in International Business and Finance
Received date: Accepted date:
23-2-2015 9-9-2015
Please cite this article as: Cevik, E.I., Dibooglu, S.,Financial Stress And Economic Activity In Some Emerging Asian Economies, Research in International Business and Finance (2015), http://dx.doi.org/10.1016/j.ribaf.2015.09.017 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
FINANCIAL STRESS AND ECONOMIC ACTIVITY IN SOME EMERGING ASIAN ECONOMIES Emrah I. Cevik, Sel Dibooglu*[email protected] University of Missouri St. Louis, 408 SSB, St Louis 63121, US Phone: +13145165530 1. Introduction
ip t
The global financial crisis that started in the US spread worldwide and sent shockwaves through the global financial system. Although policy makers have conducted
cr
several monetary and fiscal stimulus packages, credit conditions were tightened, the risk
us
premium on interbank borrowing significantly increased and trade credit decreased with falling demand, specifically for the capital goods and the manufacturing sectors in general. As
an
a result, the global financial crisis caused a significant decline in global economic activity and the developed as well as emerging economies suffered a significant slump not seen since the
M
1930s.
The emerging economies were affected by the global financial crisis through various
d
channels such as contagion, a decline in capital flows, and by trade channels. As a result,
te
financial stress (or financial instability) started to increase and output sharply decreased in a large numbers of emerging economies even in those that lacked a serious fiscal and financial
Ac ce p
imbalances. These developments in the global economy have emphasized the importance of identifying and assessing the linkages between financial stress and the real economy. In this regard, Lo Duca and Peltonen (2010) and others suggested that financial stress can affect economic activity through various channels. The first channel is called the financial accelerator in which shocks that affect the creditworthiness of borrowers tend to amplify output fluctuations where credit conditions of the financial system affect the willingness to provide credit to the economy. Second, factors that impact lenders’ balance sheets can magnify economic downturns mimicking weak bank capital; banks may become more reluctant to provide capital to the real sector, may be forced to deleverage leading to sharper
1
Page 1 of 26
economic downturns. Moreover, the structure and the development of the financial system affects how large is the interconnection between real and financial sectors in the economy. The onslaught of the financial crisis of 2007-2008 and the economic downturn that
ip t
followed highlighted the importance of the link between the financial sector and real economic activity in an interconnected world. In this regard, it is very important to measure
cr
financial stress in the economy by extracting signals from variables that that are thought to
capture some aspect of financial stress. How can measuring and monitoring financial stress
us
contribute to the design and implementation of proper macroeconomic policies? While in
an
normal times, the standard evaluation of macroeconomic prospects (maintaining full employment and price stability) is adequate and there are useful policy benchmark rules (such
M
as the Taylor rule), heightened periods of financial stress may call for policy responses that are different than the usual prescriptions. That is because a period of excessive financial stress
d
may produce substantial spillovers that constrain the credit intermediation capacity of the
te
financial sector and hence require policy to be recalibrated. A financial stress index not only is useful in evaluating macroeconomic prospects and designing monetary and fiscal policy
Ac ce p
measures, it is also useful in assessing financial conditions and fragility of the financial sector. Doing so contributes to a smoothly functioning financial system. For example, in periods of heightened financial stress it may not be sufficient to adjust short term interest rates. When markets suffer from illiquidity, there is increased uncertainty about asset values and lenders are unwilling to accept these assets as collateral; as such, credit intermediation declines and real economic activity is adversely affected. Under these circumstances, policymakers may have to resort to unconventional policy measures to deal with liquidity problems. Therefore measuring financial stress not only is important from the design and implementation of macroeconomic policy but also contributes indirectly to a smooth, robust and more resilient financial system.
2
Page 2 of 26
Therefore, the identification of states of financial stress is important for optimal policy design and hence a financial stress index can provide valuable benefits for policymakers. Louzis and Vouldis (2012) suggested that a composite financial stress index provides insights
ip t
into the propagation channels of specific events and the extent to which a financial crisis affects segments of the financial system. Grimaldi (2010) emphasizes attractive features of a
cr
financial stress index as being based on real time-high frequency data; broadly assessing the
level of stress of the overall financial system; and being based on a small group of indicators.
us
Perhaps due to the 2007-2009 global financial crisis, financial stress has been studied
an
extensively with a wide range of different components of financial stress. For instance, Hanschel and Monnin (2005) developed a financial stress index to measure the degree of
M
stress for the Swiss banking system. Illing and Liu (2006) proposed a financial stress index for the Canadian financial system. Hakkio and Keaton (2009) suggested a comprehensive
d
financial stress index to determine episodes of financial stress in the U.S. Cardarelli et al.
te
(2009) developed a financial stress index for 17 advanced economies. Grimaldi (2010) constructed a financial stress index to determine episodes of financial stress in the Euro area.
Ac ce p
Louzis and Vouldis (2012) suggested a financial systemic stress index for the Greek economy. Hollo et al. (2012) constructed a financial stress index for the Euro area and they called the index a composite indicator of systemic stress. Even before the global financial crisis, empirical studies in the literature specifically
focused on predicting financial crises in emerging economies by means of early warning indicators (Kaminsky et al. 1998, Demirgüç-Kunt and Detragiache 1998, Beckman et al. 2006, Davis and Karim 2008). However, Balakrishnan et al. (2011) argued that these studies are not appropriate to study episodes of financial stress. In this sense, there is growing literature that focuses gauging financial stress for emerging economies. Balakrishnan et al. (2011) developed a financial stress index where developments in the banking sector, stock
3
Page 3 of 26
exchange market, sovereign debt spreads, and exchange market were considered as components of financial stress. Cevik et al. (2013a) modified and extended the index proposed by Balakrishnan et al. (2011) with specific considerations for the Turkish economy.
ip t
Similarly, Cevik et al. (2013b) proposed a comprehensive index of financial stress for Bulgaria, Czech Republic, Hungary, Poland and Russia. These studies considered risks in the
cr
banking sector, securities and money market, external debt, sovereign spreads and trade finance as indicators of financial stress.
us
The literature that focuses on financial stress in Asian countries is more recent. The
an
Bank of Thailand (2010) proposed a monthly financial stress index for 1996-2009. Their financial stress index reflects risk to the functioning of the financial system in six areas: the
M
bank index beta, inverted term spreads, corporate bond spreads, stock index volatility, volatility of the government bond price index, and exchange rate volatility. Osorio et al.
d
(2011) proposed a quarterly financial conditions index for 2005-2010 for Asian countries.1
te
They considered interest rate market, exchange rate market, domestic credit market and equity market to calculate a financial stress index. Hwa et al. (2012) suggested a monthly financial
Ac ce p
stress index for the ASEAN-5 countries namely Indonesia, Malaysia, Philippines, Singapore and Thailand for 1997-2009. They emphasize four essential components to gauge financial stress: banking sector, equity markets, foreign exchange market and domestic bond markets. Lee et al. (2013) analyzed the financial stability of Korean banking system for the 2003 -2011 period by developing a composite financial stability index. The main objective of this study is to contribute to the literature by constructing a
financial stress index, studying episodes of elevated stress for Indonesia, South Korea, Malaysia, the Philippines, and Thailand. We also examine relationship between financial 1
Their country selection is based on data availability and the sample includes Australia, China, Hong Kong
SAR, India, Indonesia, Japan, Korea, Malaysia, New Zealand, Philippines, Singapore, Thailand and Taiwan Province of China. 4
Page 4 of 26
stress and economic activity for the sample countries. Our sample countries were chosen based on data availability. Our paper has several innovations: First, we consider additional factors such as risks in the banking sector and sovereign risk as additional indicators of
ip t
financial stress. Second, we use a contingent claim analysis and option pricing methods to calculate sovereign risk for the sample countries. In examining the relationship between
cr
financial stress and economic activity, we determine recessions according to the Harding and Pagan (2002) algorithm. As another contribution, we use a dynamic factor model to aggregate
us
the factors contributing to financial stress. The link between financial stress and the real
an
sector is an important one. An increase in financial stress can potentially produce substantial spillovers and systemic risks that constrain the credit intermediation capacity of the financial
M
sector. With financial stress, reduced asset values decrease the collateral that provides credit to the economy. At the same time, with deleveraging, banks lose capital and become reluctant
d
to lend to businesses. Moreover, strict credit standards and the accompanying credit crunch
te
with a negative outlook affects economic activity adversely. In those times, policymaking should include not only ordinary measures of monetary and fiscal policy, but policymakers
Ac ce p
have to be extra vigilant. In emerging markets dealing with financial stress and the specific policy actions depend on the source of stress. If financial stress is due to banking sector problems, policy actions need to focus on strengthening the banking sector. However if the source of financial stress is an external, conventional policy measures will be inadequate. In that case, international policy coordination, access to currency swap lines, and contingency funding facilities become very important in dealing with financial stress. Measuring financial stress not only provides a quantitative benchmark to assess the intensity of stress, but also gives an idea about the relative contribution of each financial indicator to the overall measure of stress and hence helps in formulating the policy response. In this paper we identify the
5
Page 5 of 26
specific factors that have contributed to elevated financial stress in sample countries in known historical crisis periods. A financial stress index would provide valuable information as a heightened index
ip t
helps fine tune economic policy. This is particularly important for emerging Asian countries because Gupta and Miniane (2009) showed that recessions accompanied by financial stress
cr
are substantially longer and deeper than others in Asian countries. In addition, deep recessions in Asia have resulted in substantial declines in potential output growth and this may point to
us
permanent effects of high financial stress. Therefore, we modify and extend the index
an
proposed by Balakrishnan et al. (2011) for South Asian countries with specific considerations for Indonesia, South Korea, Malaysia, the Philippines, and Thailand. Section 2 represents the
M
effect of the global crisis on the Asian countries in question. Section 3 explains the components of the financial stress index. Section 4 elaborates on the construction of the index
d
and shows our indicators capture key aspects of financial stress. Section 5 examines the
te
relationship between financial stress and economic activity and shows how an elevated index tended to coincide with known crisis episodes. Finally, we show the constructed index
Ac ce p
provides valuable information on economic activity. 2. Some Key Economic Variables during the Global Financial Crisis
As useful background, we briefly summarize recent developments on economic
growth, inflation rates, foreign trade, equity prices, and international reserves for the sample countries.
Figure 1 shows economic growth during the global financial crisis for individual
economies. As seen in Figure 1, economic growth significantly decreased in all countries
except for Indonesia between 2008 and 2009 due to the global financial crisis. Specifically, negative economic growth was observed in South Korea, Malaysia and Thailand in 2009. Fig. 1 Yearly Growth Rate of Reel GDP Fig. 2 Key Economic Variables
6
Page 6 of 26
We also present other key economic variables during the global financial crisis in Figure 2. Panel (a) in Figure 2 shows annual inflation based on consumer prices. Inflation rates for selected Asian countries display a similar pattern where we observe an increase in inflation
ip t
rates in the middle of 2007. Then, there was a substantial decrease in inflation rates in all countries and inflation rates have reached the pre-crisis levels at beginning of 2010.
cr
Panel (b) shows the evolution of foreign international reserves (excluding gold) during the crisis. All countries have experienced a decline in foreign reserves due to capital outflows
us
during the crisis. Also, South Korea and Malaysia showed the largest decline in reserves.
an
Panel (c) presents movements in equity market returns where equity markets exhibit a similar pattern during the crisis. Falling equity returns are evident in all countries in 2007 and
M
2008 in which the largest stock market declines have occurred in Indonesia (–109%) and South Korea (–101%),
d
Finally, we show behavior of foreign trade (sum of merchandise exports and imports)
te
during this crisis in Panel (d) of Figure 2. As can be seen in the Figure, total merchandise trade in these countries collapsed at the end of 2008 due to a sharp decline in economic
Ac ce p
activity of trading partners and perhaps lowers availability of trade credit.
The figure presented above shows that the crisis affected Asian countries similarly and
the recovery seems to have been rapid as most of the variables reached pre-crisis levels at the beginning of 2010. IMF (2010) emphasized on three factors for a fast recovery in Asian countries. First regional heavyweights such as China was not affected much by the crisis and showed the fastest recovery which helped other Asian countries. The second factor is related to trade and finance that started to normalize in February 2009 and hence overall economic activity came back to pre-crisis levels very quickly as compared to other emerging countries. The last factor is related to the region’s aggressive countercyclical response that caused a speedy recovery in Southeast Asian countries. 7
Page 7 of 26
3. Financial Stress and its Indicators
Even though episodes of financial stress are defined in various ways in the literature, studies in this area have generally emphasized two main underlying stress phenomena: increase in uncertainty and changing expectations. Grimaldi (2010) defined the level of stress
ip t
as being determined by the interaction between financial vulnerabilities and the size of
shocks. Balakrishnan et al. (2011) emphasized four fundamental characteristics of financial
cr
stress: large shifts in asset prices, an abrupt increase in risk and/or uncertainty, liquidity
us
droughts, and concerns about the health of the banking system. Similarly, Hakkio and Keeton (2009) argue that episodes of financial stress must involve at least one of the following:
an
increased uncertainty about fundamental value of assets, increased asymmetry of information, decreased willingness to hold risky assets, and decreased willingness to hold illiquid assets.
M
Studies on financial stress in the literature generally consider five essential components: the financial intermediaries sector, money markets, equity markets, bond
d
markets, and foreign exchange markets. However, financial stress in emerging economies is
te
not only related to financial markets but also some external flow variables. Therefore,
Ac ce p
focusing only on financial markets may not be adequate for developing a financial stress index as there are additional sources of financial stress such as external debt, current account deficits and sovereign risk.
In order to account for different aspects of financial stress in South Asian economies,
we augment conventional factors with measures of external debt and as well as specific measures of sovereign risk. As a result, our financial stress index reflects the developments in the riskiness in the banking sector, security market risk, currency risk, external debt, and sovereign risk.
3.1. Riskiness of the Banking Sector
A sound financial sector is an important element for the real economy. The financial sector is instrumental in providing credit, payment mechanisms, and fund transfer services to 8
Page 8 of 26
households and businesses alike and in managing potential risks. Therefore, a healthy financial sector should lower the cost and risk of producing and trading of goods and services among market participants. Moreover, a large number studies documented the effects of a
ip t
weak banking system as an important factor behind the Asian crisis in 1998. For instance, Turner (2007) showed that banking system weakness, poor macroeconomic policies, and a
cr
massive shock all contributed to the Asian crisis. Hence riskiness of the banking sector is an important component of a financial stress index. In this paper, we consider the volatility of
us
banking sector index in the stock market and use the following EGARCH model that is
an
proposed by Nelson (1991) to obtain banking sector index returns volatility:2,3,4,5
(
( )
)
M
(1) rB ,t = X t′θ + ε t log σ t2 = ω + β log σ t2− j + α
ε t −1 ε + γ t −1 σ t −1 σ t −1
te
d
where rB,t indicates banking sector index returns, X t′ includes a constant and autoregressive variables of the banking sector index returns, εt is an error term and σt is the conditional
Ac ce p
variance. There are two features of EGARCH model compared to a GARCH model. First; because the left-hand side is the log of the conditional variance, the leverage effect is exponential, rather than quadratic, and that forecasts of the conditional variance are
2
Hanschel and Monnin (2005) suggested using both market and balance sheet data of banking sector to gauge
financial stress. We consider such balance sheet variables (e.g. total deposit and total loan of banking sector) but these variables did not improve the performance of our financial stress index in identifying specific events. 3
Note that although several volatility models are proposed in the literature, in order to account for different
specifications in asset returns, Andersen and Bollerslev (2001) found that GARCH and stochastic volatility models provide good volatility forecasts particularly in high frequency data. 4
Due to banking sector stock index data availability, we use the financial services sector stock index for
Indonesia, Malaysia and Philippines. 5
We consider other types of GARCH specifications, but the EGARCH (1,1) model has better fit according to the
model selection information criteria. 9
Page 9 of 26
guaranteed to be nonnegative. The other advantage is that the EGARCH model allows for the presence of leverage effects in the volatility. 3.2. Security Market Risk
It is well known that the stock markets are important indicators of a healthy financial
ip t
system and studies in the literature use securities market as a component of the financial stress index. Therefore an EGARCH model can be employed to measure stock market volatility as
(
( )
)
ε t −1 ε + γ t −1 σ t −1 σ t −1
an
(2) rS ,t = X t′θ + ε t log σ t2 = ω + β log σ t2− j + α
us
cr
follows:
where rS,t indicates stock market index returns, X t′ includes a constant and autoregressive
M
variables of stock market returns, εt is an error term and σt is the conditional variance. 3.2. Currency Risk
d
Currency risk is another important component of financial stress for developing
te
economies. We consider Exchange Market Pressure Index (EMPI) proposed by Girton and Roper (1977) as a component of our financial stress index. Therefore, we construct the EMPI
Ac ce p
by combining exchange rate movements, changes in international reserves, and changes in the overnight interest rate relative to the US as follows: (3) EMPI t =
∆et
σ ∆e
−
∆ Re st
σ ∆ Re s
+
(
∆ it − iUS ,t
σ ∆ (i − i ) t
)
US , t
where ∆et and ∆Rest are 12-month percent changes in the exchange rate and total foreign international reserves minus gold, and it and iUS represent the overnight interest rate for the home country and the US, respectively. Similarly σx denotes the standard deviation of variable x, (x = the changes in the exchange rate, total reserves, and overnight interest rate). Even
though EMBI is used to identify crisis periods in the literature, Bussierre and Fratzscher
10
Page 10 of 26
(2006) emphasized the advantage of EMPI in capturing both successful and unsuccessful speculative attacks. 3.3. External Debt
Although external debt is important for sustainable economic growth in emerging
ip t
economies, short term external debt played a significant role in the Asian and Russian crises
and hence excessive increase in the short term external debt can cast a doubt on sustainability
cr
of such debt. Hence, we consider the ratio of short term external debt to GDP that is widely
us
examined as a potential indicator of financial crises for developing countries (see Sachs et al. (1996); Abiad (2003); Bussiere and Fratzscher (2006)). As such, an increase in the ratio of
an
short term external debt to GDP may be expected to cast a doubt on the sustainability of external debt.6
M
3.4. Sovereign Risk
Another component of our financial stress index is sovereign risk that is specifically
d
important for emerging markets. Because interest rate spreads between home country and the
te
US can be used as an indicator of risk perception in the home country, sovereign bond spreads (the difference between home country’s Emerging Market Bond Index (EMBI) and 10 year
Ac ce p
US Treasury yield) are used in constructing the financial stress index in the literature (see Balakrishnan et al. (2011) and Cevik et al. (2013a)). Due to EMBI spreads data unavailability for our sample countries, we employ the so called Contingent Claim Analysis (CCA) to compute sovereign risk for the countries in question. The CCA is a framework that combines balance sheet information with commonly
used risk measurement tools to construct a marked-to-market balance sheet to identify and quantify risks. In essence, the CCA approach models firm equity as a contingent claim on a firm’s assets. It is a contingent claim because the value of firm equity depends on the value of
6
As external debt is measured quarterly in these countries, we use the cubic spline method to obtain monthly
external debt series. Even though there are several methods to interpolate data in the literature, we use the cubic spline method because it is simple, fast, efficient and stable. 11
Page 11 of 26
the firm’s assets and the default-free value of the firm’s liability at a particular point in time (Geuorguiev et al., 2009). The firm equity can be written as a call option as follows: (4) E = max[ A − DB, 0]
ip t
where A is firm assets and DB is the default barrier. In Equation (6), when the firm can generate enough cash to cover its current debt obligations (A ≥ DB), the value of firm’s equity
cr
is equal to difference between the value of firm’s assets (A) and the current debt obligations of
us
the firm or default barrier (DB). On the other hand, when the firm’s assets do not cover the current debt obligations (A ≤ DB), firm equity equals zero.
an
While the CCA analysis is commonly used to calculate default probability of firms in the finance literature, there have been studies that measure the sovereign risk profile of
M
countries based on the CCA approach. For instance, Gapen et al. (2008) and 2008) employed the CCA analysis to determine the sovereign risk profile for 12 developing economies. They
d
showed that the CCA generates similar results to EMBI and CDS spreads. Also Keller et al.
te
(2007) examined changes in Turkey’s sovereign risk profile by using the CCA and found a strong correlation between sovereign risk indicators based on the CCA and EMBI spreads.
Ac ce p
According to Gapen et al. (2008) and Keller et al. (2007), the sovereign balance sheet
of a country can be represented as in Table 1.
In order to derive default probabilities, some assumptions must be made about the
seniority structure of a sovereign country’s liabilities. To derive external default probabilities, external debt is assumed to be the more senior liability, whereas domestic debt and base money are assumed to represent the equity portion of the sovereign balance sheet and thus can be viewed as a contingent claim on the residual value of sovereign assets. The sovereign is assumed to default whenever the value of its implied assets falls below a distress barrier. The difference between the asset value and the distress barrier, scaled by the asset volatility, is
12
Page 12 of 26
referred to as the distance-to-distress, while the area of the distribution that falls below the distress barrier represents the sovereign’s default probability (Keller et al., 2007). In this context the market value of equity can be modeled by using Black and Scholes
ip t
(1973) and Merton (1974) formula for call options: (5) E = AN ( d1 ) − DBe − rt N ( d 2 )
σ A T , d 2 = d1 − σ A T , r is the risk
us
term external debt,7 d1 = ln ( A DB ) + ( r + 0.5 × σ 2A ) T
cr
where A is the assets, DB, the default barrier, is equal to the sum of short term and half of long
free interest rate8, T is the time to maturity of the default barrier,9 N(d) is a cumulative
an
probability distribution function for a standard normal variable, and σA is the standard deviation of assets. Under the Merton (1974) bond pricing model’s assumptions, the volatility
Aσ A N ( d1 ) E
d
(6) σ E =
M
of the equity is:
(
)
ln ( A DB ) + r − 0.5 × σ 2A T
σA T
Ac ce p
(7) DD =
te
Distance to default (DD) can be written as:
Under the normal distribution which is implied by Merton (1974), the default
probability or the risk-neutral default probability (RNDP) can be calculated as N(-DD). Although Equations (7) and (8) can be solved simultaneously, Crosbie and Bohn
(2003) argue that market leverage moves around far too much for Equation (8) to provide reasonable results. In order to overcome this problem we used an iterative procedure. First, we set an initial value of A = E + DB and compute the standard deviation of the log asset returns.
7
External debt that matures within one year is defined as short term with any other liability deemed long term.
8
The one year US Treasury Bill rate is taken to be the risk free interest rate.
9
Although the time horizon of the estimate of default risk can vary in the literature, Gapen et al. (2008) and
Keller et al. (2007) considered one year as the time horizon. Therefore we use t=1 in the model estimation. 13
Page 13 of 26
Then we insert A and σA into Equation (7) and compute new values of A and σA. The procedure is repeated until convergence where the sum of squared differences between consecutive asset values is less than 10-10.
ip t
4. Empirical Results 4.1. Data
We construct a financial stress index to gauge the degree of the financial stress in
cr
Indonesia, Malaysia, Philippines, South Korea and Thailand for the periods of 1995-2013.
us
Data on overnight interest rate, and base money were taken from the IMF-IFS database. Banking sector stock market data were collected from Bloomberg. Data on the stock market
an
index, the foreign exchange rate, international reserves, and real GDP were obtained from World Bank-GEM database. Finally, data on domestic currency debt and external debt were
M
obtained from national central banks. Before the aggregation of the components, all of the components are standardized by subtracting their means and dividing by their standard
te
4.2. Aggregation of the Components
d
deviations.
Constructing of a comprehensive index depends on aggregation of the variables and
Ac ce p
there are various aggregation methods in the literature that can be classified into varianceequal weight method (e.g. Balakrishnan et al. 2011; Cardarelli et al. 2009; Hanschel and Monnin, 2005), principal components analysis (Hakkio and Keeton, 2009; Illing and Liu, 2006; Cevik et al. 2013a; Cevik et al. 2013b; the Federal Reserve Bank of Kansas City and the Federal Reserve Bank of St. Louis), and, recently, portfolio theory based aggregation schemes that take into account the correlation structure of stress indicators in order to quantify the level of systemic stress (Hollo et al. 2012, and Louzis and Vouldis, 2012). In this paper, we employ a dynamic factor model to construct financial stress index for sample countries. We assume that each of the variables above captures some aspect of financial stress and hence all variables are likely to move together according to the level of
14
Page 14 of 26
financial stress in the economy. In this context, the common component for these variables represents financial stress in an economy. Dynamic factor model can be defined as follows: (8) yt = χ t + ε t , ε t ∼ N ( 0, ψ )
ip t
where yt is indicator of financial stress index, χ t = ΛFSI t and ψ is assumed to be diagonal.
cr
Note that the idiosyncratic components of the observable variables are assumed to follow
AR(1) process. The common factor or financial stress index can be estimated by using the
us
entire set of indicators. Dynamics of financial stress can be captured by the following
(9) FSI t =
p
ρi FSI t −i + υt , υt ∼ N ( 0, Σ )
M
i =1
an
autoregressive process:10
In periods of high financial stress, the pressure on the value of local currencies
d
increases, stock market becomes more volatile, sovereign risks rise and doubts on
te
sustainability of external debt can be expected to rise. Therefore we expect that an increase in the EMPI, stock market volatility, banking sector volatility, the ratio of total external debt to
Ac ce p
foreign reserves, and sovereign risk raise financial stress. 4.3. Extraction of Financial Stress via a Dynamic Factor Model
The dynamic factor model results for Indonesia, Malaysia, Philippines, South Korea
and Thailand are given in Table 2. It should be noted that all variables are standardized before the aggregation and hence coefficients for the variables represent the effect of one-standarddeviation change in the respective variable on the index. According to the coefficients of components, an increase in the EMPI, stock market volatility, banking sector volatility, sovereign risk, and short term external debt- GDP ratio raises financial stress in all countries. Results in Table 2 show that EMPI is not statistically significant in Thailand. In addition, short term external debt- GDP ratio is not statistically significant for Malaysia and South 10
We select lag lengths by using Schwarz Bayesian information criterion (BIC). 15
Page 15 of 26
Korea. The Bayesian Information Criterion, BIC, suggests a first degree autoregressive process for Indonesia and Thailand. On the other hand, BIC suggest two lags for Malaysia and Philippines and three lags for the South Korea.11
ip t
The estimated coefficients in Table 2 indicated that a one-standard-deviation increase in EMPI has similar effects on financial stress in Indonesia, Malaysia and Philippines.
cr
Volatility in the stock market and banking sector index seems to be an important source of
us
financial stress in all countries. In this context, banking sector volatility seems to be a modest factor contributing to financial stress in Malaysia. Short term external debt component has
an
similar effects on the financial stress index in Indonesia, Philippines and Thailand but it is found to be statistically insignificant for Malaysia and the South Korea. Sovereign risk and
M
EMPI seem to contribute to financial stress differently across countries.
4.4. The Evolution of Financial Stress
d
In this section, we examine the evolution of financial stress for the Southeast Asian countries.
te
We are particularly interested in whether heightened episodes of financial stress correspond to
Ac ce p
known episodes of downturns in the real economy and the behavior of financial stress over the business cycle. The behavior of financial stress in the sample countries is given in Figure 1 where the shaded areas correspond to recessions that are determined according to the Harding and Pagan (2002) algorithm. Panel a) of Figure 3 indicates that the Indonesia Financial Stress Index (I-FSI) tracks
recessions closely in the sample.12 Specifically the I-FSI increased at the beginning of the 1997 and remained above zero until end of 2001. Although there were some internal problems (such as under-supervised banks, extensive crony capitalism, corruption, monopoly power,
11
The sum of autoregressive parameters is found to be higher than 0.9 for all countries except for Indonesia and
the result is consistent with empirical results found in Carpenter et al. (2014). 12
Due to banking sector index data availability, the Indonesia Financial Stress Index starts from January 1996. 16
Page 16 of 26
and growing short term debt) in the Indonesian economy, it was the second in terms of export growth in 1996 among the Asiean-5 countries before the Asian crisis. Moreover, the budget surplus averaged over one percent in the previous four years, while credit growth was modest.
ip t
Hence, it can be said that the crisis in 1998 appeared in Indonesia due to weak financial sector, political uncertainty and contagion effect rather than poor traditional economic
cr
fundamentals (Cerra and Saxena, 2000). Therefore, the index has reached its highest level in
January 1998 due to Asian crisis in 1997-98. During the 2000-2002 period, because of several
us
financial and economic crises in the world (such as the stock market crash in 2000, the
an
Turkish crisis in 2001, the Argentina crisis in 2002, Enron scandal and 9/11 attacks in 2001), the financial stress index remained above zero until the first half of 2002. Then the I-FSI
M
decreased and remained under zero. Finally, financial stress started to rise again in the beginning of 2008 with the start of the global financial crisis.
d
The Malaysia Financial Stress Index (M-FSI) is plotted in Panel b of Figure 3.13 As
te
seen in the figure, the M-FSI captures successfully all periods of recessions in Malaysia.
Ac ce p
Financial stress index remained under zero until the beginning of the 1998 in Malaysia. Due to the Asian crisis, financial stress index started to increase at the beginning of 1998 and it has reached the highest value at May 1998. After that, financial stress in Malaysia decreased but remained below zero until end of 2001 as a result of crises in developed and emerging economies. Financial stress in Malaysia started to decrease at the beginning of 2002, and then the M-FSI was below zero until end of 2008. As in other emerging economies, the global financial crisis caused an increase in the financial stress index in Malaysia but the effect of global financial crisis on financial stress seems to be fairly limited as there is no substantial increase in financial stress during the financial crisis in Malaysia.
13
Due to EMBI data availability, the Malaysia Financial Stress Index starts from October 1996. 17
Page 17 of 26
According to Panel c of Figure 3, the financial stress in the Philippines was high at the beginning of the 1998 and it remained above zero until first half of 2003 except for the beginning of 2002.14 As in other countries in the region, financial stress rose during the Asian
ip t
crisis in 1997-98 in the Philippines and the index reached the highest value in September 1998. Although financial stress was below zero between 2004 and 2008, it started to increase
cr
at the end of 2007 due to the global financial crisis.
us
The South Korean Financial Stress Index (SK-FSI) remained above zero at the end of 1996 ant it started to increase at the beginning of 1997. Although there were no serious
an
macroeconomic imbalances in the Korean economy before the Asian crisis, Korea was hit by the Asian crisis during the fall of 1997, just a few months after other countries of the region.
M
For instance, the annual growth rate of GDP was still above 5% in the second quarter of 1997. During the summer of 1997, the Korean Won was not under pressure because it was the only
d
currency in the region to have depreciated in real terms in previous years. On the other hand,
te
the financial situation started to deteriorate before the financial crisis in Korea. During the first quarter of 1997, a series of bankruptcies hit some of the big conglomerates due to the
Ac ce p
combination of their huge rate of indebtedness and the increase in foreign interest rates. By the end of October, the Won was overvalued in comparison with the other Asian currencies that lost some value since the summer. On October 21st, Korea had to request the assistance of the IMF. The first of the five IMF rescue plans for Korea was launched on December 5th. The government gave up the peg on December 16th which precipitated a depreciation of the Korean Won by 43% (Robert, 2005: 35-36). The financial stress index reached the highest level in January 1998 due to the Asian crisis. After that financial stress started to decrease in South Korea and the index remained below zero between 2003 and 2007. As in other Asian countries, the global financial crisis led to elevated financial stress in South Korea. 14
Due to EMBI spreads data availability, the Philippines Financial Stress Index starts from January 1998. 18
Page 18 of 26
The Thailand Financial Stress Index (T-FSI) is plotted in panel e of Figure 3. At the beginning of the sample, financial stress seems to be very low in Thailand. However, financial stress in Thailand started to rise earlier than in other emerging Asian countries before the
ip t
Asian crisis. This result is consistent with our expectations as the Asian crisis first appeared in Thailand. As seen in Figure 3, the financial stress index started to increase at the beginning of
cr
1996 in Thailand and the index has reached the highest level in September 1998. When we look at the chronology of the crisis, pressure on the Thai currency started to rise at the
us
beginning of 1996 due to widening external deficit, slowing economy, emerging strains in the
an
financial system, and growing political instability, which, taken together, started to weaken investor confidence. The vulnerability in Thailand stemmed from Thai finance companies
M
with their heavy short term borrowing abroad. After the attempted rescue of the largest Thai finance company failed in May 1997, the weakening of investors’ confidence intensified in
d
Thailand. On 2 July 1997 the Bank of Thailand abandoned the exchange rate parity which had
te
fixed the Thai baht to the US dollar and allowed the currency to float. In late July, the Thai government requested financial support from the IMF, resulting in an international rescue
Ac ce p
package totaling US$ 17 billion (Fries et al., 1999). Due to the turmoil that followed, the financial stress index remained above zero between 2000 and 2001. Then, the index decreased until 2008 when the global financial crisis led to a rise in financial stress in Thailand as well. Overall not only does our financial stress index track known episodes of financial turmoil well in the sample, but it seems that financial stress precedes economic downturns in sample countries. In that regard, financial stress seems to be a leading economic indicator in the sample. 5. The Relationship between Financial Stress and Economic Activity
In this section, we examine the relation between financial stress and economic activity. The transmission channels between financial stress and economic activity have been widely
19
Page 19 of 26
argued in the literature where the role of financial leverage and bank capital channel are emphasized. Moreover, Hakkio and Keaton (2009) argued the role of uncertainty about the price of financial assets and the economic outlook. Davig and Hakkio (2010) emphasized
ip t
“real options” and financial accelerator” for the transmission channels between financial stress and economic activity. Due to the global financial crisis, there have been a growing
cr
literature that focus on the relationship between financial stress and economic activity (Li and
us
St-Amant, 2007; Claessens et al., 2008; Hakkio and Keeton, 2009; Cardarelli et al., 2010). We examine the relationship between the financial stress index and economic activity
an
using a bivariate vector autoregression (VAR) model. Economic activity is tracked via some economic indicators. For example, Hakkio and Keaton (2009), and Davig and Hakkio (2010)
M
employed the Chicago Fed National Activity Index, which combines 85 macroeconomic time series as a measure for economic activity. Cevik et al. (2013a) and Cevik et al. (2013b)
d
considered the yearly growth rate in the industrial production, foreign trade and gross fixed
te
capital formation as a proxy for economic activity. Louzis and Vouldis (2012) calculated an index of economic activity as the first principal component of five variables namely,
Ac ce p
unemployment, manufacturing index, exports, imports and producers price index. Hollo et al. (2012) used industrial production index as a proxy for economic activity. As in Hollo et al. (2012), we consider yearly growth rate of the industrial production index (GIP) as proxy for economic activity. Then, we employ a bivariate VAR model to examine dynamic relation between financial stress and economic activity in all countries.15 We orthogonolize the innovations using a Cholesky decomposition where the financial stress shock is ranked first: it is assumed that real economic activity shocks have no contemporaneous effects on financial
15
In all cases the lag length is determined by the Schwarz Bayesian Information Criterion (BIC). 20
Page 20 of 26
stress and they may affect financial stress after a month. The impulse response functions for the financial stress index and GIP are presented in Figure 4 for each country in our sample16. The results in Figure 4 indicate that the responses of growth in industrial production to
ip t
a shock in financial stress are negative and significant for all countries. Note that the greatest impact of financial stress on economic activity as measured by the industrial production
cr
growth seems to be in Malaysia. The effect of financial stress on economic activity is nearly
us
statistically insignificant in the Philippines. However, the responses of the financial stress index to an industrial production shock are not statistically significant and industrial
an
production seems to be exogenous to financial stress in all sample countries. Overall our results indicate that the financial stress causes significant economic slowdowns in Southeast
M
Asia in our sample. 5. Conclusions
d
This paper constructed a financial stress index using riskiness in the banking sector,
te
security market risk, currency risk, external debt and sovereign risk. Using data from Indonesia, South Korea, Malaysia, the Philippines, and Thailand we employ a dynamic factor
Ac ce p
model to aggregate the aforementioned factors into a single financial stress index and examine the relationship between financial stress and economic activity. Indeed our results show that when foreign exchange risks rise, stock market becomes more volatile, sovereign risks rise and when there are doubts on sustainability of external debt, financial stress rise substantially. Not only does our financial stress index track known episodes of financial turmoil well in the sample, but it seems that financial stress precedes economic downturns in sample countries. As such, financial stress seems to be a leading economic indicator in the sample. In that regard, the financial stress index contains valuable information for policy makers.
16
We also derive impulse response functions with the order of the shocks reversed for all countries and get
qualitatively similar results. These are available upon request. 21
Page 21 of 26
Our results show that the financial stress index has important bearings on economic activity in our sample. A bivariate VAR model of financial stress and industrial production shows that financial stress causes significant economic slowdowns. References
ip t
Abiad, Abdul (2003). Early Warning Systems: A Survey and Regime Switching Approach. IMF Working Paper, No: 03/32.
cr
Balakrishnan, Ravi, Stephan Danninger, Selim Elekdag and Rina Tytell (2011). The
Finance and Trade, Vol. 47, Supplement 2, pp. 40-68.
us
Transmission of Financial Stress from Advanced to Emerging Economies. Emerging Markets
an
Bank of Thailand (2010). Thailand’s Financial Stress Index.
http://www.bot.or.th/English/MonetaryPolicy/Inflation/PaperInFrame/100_Chapter_4_E_Apr
M
10.pdf
Beckmann, Daniela, Lukas Menkhoff and Katja Sawischlewski (2006). Robust Lesson
d
about Practical Early Warning Systems. Journal of Policy Modeling, Vol. 28, pp. 163-193.
te
Black, Fischer and Myron Scholes (1973). The Pricing of Options and Corporate
Ac ce p
Liabilities. Journal of Political Economy, Vol. 81, pp. 637-659. Bussiere, Matthiue and Marcel Fratzscher (2006). Towards a New Early Warning
System of Financial Crises. Journal of International Money and Finance, Vol. 25, pp. 953973.
Cardarelli, Roberto, Selim Elekdag and Subir Lall (2009). Financial Stress,
Downturns, and Recoveries. IMF Working Paper, No 09/100. Carpenter, Seth, Selva Demiralp, Bernd Schlusche and Zeynep Senyuz (2014). Measuring Stress in Money Markets: A Dynamic Factor Approach. Economics Letters, Vol. 125, pp. 101-106.
22
Page 22 of 26
Cerra, Valerie and Sweta Chaman Saxena (2000). Contagion, Monsoons, and Domestic Turmoil in Indonesia: A Case Study Asian Currency Crisis,” IMF Working Paper, WP/00/60.
in Turkey. Journal of Policy Modeling, Vol. 35, pp 370-383.
ip t
Cevik, Emrah I., Sel Dibooğlu and Turalay Kenc (2013a). Measuring Financial Stress
cr
Cevik, Emrah I., Sel Dibooğlu and Ali M. Kutan (2013b). Measuring Financial Stress in Transition Economies. Journal of Financial Stability, Vol. 9, pp 597-611.
us
Claessens, Stijn, M. Ayhan Köse and Marco E. Terrones (2008). Financial Stress and
an
Economic Activity. Journal of BRSA Banking and Financial Markets, Vol. 2, pp. 11-24. Crosbie, Peter and Jeff Bohn (2003). Modeling Default Risk. http://www.ma.hw.ac.uk
M
/~mcneil/F79CR/Crosbie_Bohn.pdf
Davig, Troy and Craig Hakkio (2010). What Is the Effect of Financial Stress on
te
Quarter, 35-62.
d
Economic Activity?, Federal Reserve Bank of Kansas City Economic Review, Second
Davis, E. Philip and Dilruba Karim (2008). Comparing Early Warning Systems for
Ac ce p
Banking Crises. Journal of Financial Stability, Vol. 4, pp. 89-120. Demirgüç-Kunt, Aslı and Enrica Detragiache (1998). The Determinants of Banking
Crises in Developing and Developed Countries. IMF Staff Papers, Vol. 45 (1), pp. 81-109. Fries, Steven, Martin Raiser and Nicholas Stern (1999). Stress Test for Reforms.
Economics of Transition, Vol 7, pp. 535-567. Gapen, Michael T, Dale F. Gray, Cheng Hoon Lim and Yingbin Xiao (2008).
Measuring and Analyzing Sovereign Risk with Contingent Claims, IMF Staff Papers, Vol. 55 (1), pp. 109-148. Geuorguiev, Nikolay, Rodney Ramcharan, Alison Stuart and Burcu Aydin (2009). South Africa: Selected Issues. IMF Country Report No. 09/276.
23
Page 23 of 26
Girton, L. and D. E. Roper (1977). A Monetary Model of Exchange Market Pressure Applied to the Postwar Canadian Experience, American Economic Review, Vol. 67, pp. 537548.
Euro Area. Working Paper Series No 1214. European Central Bank.
ip t
Grimaldi, Marianna Blix (2010). Detecting and Interpreting Financial Stress in the
cr
Gupta, S., and J. Miniane (2009). Recessions and Recoveries in Asia: What Can the Past Teach Us About the Present Recession? ADBI Working Paper 150. Tokyo: Asian
us
Development Bank Institute. Available:
an
http://www.adbi.org/workingpaper/2009/09/02/3324.recessions.recoveries.asia/ Hakkio, S. Craig and William R. Keeton (2009). Financial Stress: What Is It, How
Economic Review, Second Quarter, 5-50.
M
Can It Be Measured, and Why Does It Matter?. Federal Reserve Bank of Kansas City
d
Hanschel, Elke and Pierre Monnin (2005). Measuring and Forecasting Stress in the
te
Banking Sector: Evidence From Switzerland. BIS Working Paper, No. 22. Harding, Don and Adrian Pagan (2002). Dissecting the Cycle: A Methodological
Ac ce p
Investigation, Journal of Monetary Economics, Vol. 49, pp 365-381. Hollo, Daniel, Manfred Kremer and Marco Lo Duca (2012): CISS - A Composite
Indicator of Systemic Stress in the Financial System. Working Paper Series No 1426. European Central Bank.
Hwa, Tng Boon, Kwek Kian Teng and Sheng (2012). Financial Stress in Asean-5
Economies from the Asian Crisis to the Global Crisis. The Singapore Economic Review, Vol. 57 (2), pp. 1-24. Illing, Mark and Ying Liu (2006). Measuring Financial Stress in a Developed Country: An Application to Canada. Journal of Financial Stability, Vol. 2, pp. 243-265. IMF (2010). World Economic Outlook. Rebalancing Growth April 2010.
24
Page 24 of 26
Kaminsky, Graciel A., Saul Lizondo and Carmen M. Reinhart (1998). Leading Indicators of Currency Crises. IMF Staff Papers, Vol. 45, No. 1, pp. 1-48. Keller, Christian, Peter Kunzel and Marcos Souto (2007). Measuring Sovereign Risk
ip t
in Turkey: an Application of the Contingent Claims Approach, IMF Working Paper, No 07/233.
cr
Lee, Jong Han, Jaemin Ryu and Dimitrios P. Tsomocos (2013). Measures of Systemic Risk and Financial Fragility in Korea. Annals of Finance, Vol. 9, pp. 757-786.
us
Li, Fuchun and Pierre St-Amant (2007). Financial Stress, Monetary Policy, and
an
Economic Activity. 42nd Annual Meeting of the Canadian Economics Association, Vancouver, B.C.
M
Lo Duca, M., Peltonen, T. (2010). Macro-financial Vulnerabilities and Future Financial Stress - Assessing Systemic Risks and Predicting Systemic Events. Working Paper
d
Series No 1311. European Central Bank.
te
Louzis, P. Dimitrios and Angelos T. Vouldis (2012). A Methodology for Constructing a Financial Systemic Stress Index: An Application to Greece. Economic Modelling. Vol. 29,
Ac ce p
pp. 1228-1241.
Merton, Robert C. (1974). On the Pricing of Corporate Debt: The Risk Structure of
Interest Rates. Journal of Finance, Vol. 29, pp. 449-470. Nelson, D. B. (1991). Conditional Heteroskedasticity in Asset Returns: A New
Approach, Econometrica, Vol. 59, pp. 347-370. Osorio, Carolina, Runchana Pongsaparn and D. Filiz Unsal (2011). A quantitative
Assessment of Financial Conditions in Asia. IMF Working Paper, WP/11/173. Robert, Marc Y. (2005). The 1997-1998 Korean Crisis: Domestic or External Causes?. Journal of Policy Modeling, Vol. 27, pp. 33-53.
25
Page 25 of 26
Sachs, J., A.Tornell and A. Velasco (1996). Financial Crises Emerging Markets: The Lesson from 1995. NBER Working Paper, No:W5576, pp.147-198. Turner, Philip (2007). Are Banking Systems in East Asia Stronger?. Asian Economic
ip t
Policy Review, Vol.2 (1), pp.75-95.
cr
Fig. 3 Financial Stress for South Asian Countries
us
Fig. 4 Impulse Response Functions
Table 1. The Balance Sheet of a Sovereign Country
Liabilities External Debt Equity
an
Assets International Reserves Net Financial Assets (Discounted Value of Primary Fiscal Surpluses) Value of Monopoly over Issue of Money Other Assets less Guarantees
M
Domestic Debt Base money
Ac ce p
te
d
Table 2 Dynamic Factor Model Results Variables Indonesia Malaysia Philippines S. Korea Thailand EMPI 0.072*** 0.052*** 0.085** 0.161*** 0.032 Stock market volatility 0.433*** 0.055*** 0.200*** 0.185*** 0.253*** Banking sector volatility 0.552*** 0.060*** 0.260*** 0.150*** 0.256*** Sovereign risk 0.014** 0.047*** 0.182*** 0.143*** 0.019*** Short term external debt / GDP 0.030* -0.002 0.075*** 0.013 0.025* FSIt-1 0.845*** 1.843*** 1.381*** 1.792*** 0.962*** FSIt-2 -0.858*** -0.440*** -1.191*** FSIt-3 0.357*** Note: ***, ** and * indicates statistically significant coefficients at 1%, 5% and 10% level respectively.
26
Page 26 of 26