The World Economy 2006TRADE and FINANCIAL INTEGRATION IN EAST ASIA KWANHO SHIN and CHAN-HYUN SOHN Trade and Financial Integration in East Asia: Effects on Co-movements Kwanho Shin1 and C
Trang 1The World Economy (2006)
TRADE and FINANCIAL INTEGRATION IN EAST ASIA
KWANHO SHIN and CHAN-HYUN SOHN
Trade and Financial Integration in East Asia: Effects on Co-movements
Kwanho Shin1 and Chan-Hyun Sohn2
1
Kangwon National University
In this paper we explore three important areas where deeper trade and financial integration in East Asia can influence: (1) business cycle co-movements in the region, (2) the extent of risk sharing across countries and (3) price co-movements across countries We find evidence that trade integration enhances co-movements of output but not of consumption across countries Especially the fact that trade enhancing co-movements across countries Deeper financial integration improves price co-movements weakly but does not enhance output or consumption co-movements at all However, since the current level of financial integration in East Asia is quite low, our evidence is too early to firmly determine the role of financial integration.
1 INTRODUCTION
A NUMBER of East Asian countries are seeking economic integration invarious ways Trade integration is one avenue For example, aside from thealready established ASEAN free trade arrangement, both China and Japan showmuch interest in forming free trade agreements with Korea as well as with ASEANcountries.1 The other avenue is financial integration After the sudden exchangecrisis of 1997, East Asian countries are also seeking deepening financial cooper-ation, as indicated by the discussions on the Chiang Mai Initiative and on theAsian bond market
What are the effects of trade and financial integration in East Asia? In thispaper, we explore three important areas where trade and financial integration canhave an influence First, we examine how trade and financial integration affectsbusiness cycle co-movements in the region Second, we investigate how tradeand financial integration affects the extent of risk sharing across countries bycomparing its impact on consumption co-movements with output co-movements.Finally, we examine how trade and financial integration affects price co-movementsacross countries
By analysing the changed patterns of various co-movements, we can alsogauge how they in turn influence the prospects of further integration in East Asia.For example, how synchronised business cycles of output have importantimplications for forming an extreme form of integration, a single market and singlecurrency area, namely a monetary union Since members of monetary union
This paper was prepared for the Joint YNU/KIEP International Conference on ‘Economic Integration and Structural Changes in East Asia’, held at Yokohama National University The authors appreciate comments provided by Paul De Grauwe and Etsuro Shioji and other conference participants The first author greatly appreciates the financial support by a Korea University Grant.
1
See Lee and Shin (2006, Table 3) for the movement towards regional trade agreements in East Asia.
Trang 21650 KWANHO SHIN AND CHAN-HYUN SOHN
sacrifice independent monetary policy, the cost of forming monetary union will
be lower if business cycles are synchronised so that the common monetary policyworks effectively for all member countries
While most studies focus on business cycles of output, we believe that sidering the extent of output co-movement is not enough to determine how costly
con-it is to form monetary union Since the eventual objective of monetary policy is
to maximise the welfare of the economy, which may be more closely related tosmoothing out consumption than output, if consumption does not move alongwith output, low co-movements of output itself may not necessarily be undesir-able for forming monetary union For example, if risk sharing is complete acrosscountries, despite any possible asymmetric movements of output, consumptionmovements will be perfectly correlated across countries.2 In this case it is notnecessary to implement independent monetary policy across countries becausethe common monetary policy can be effectively used to respond to the samemovement of consumption across countries Hence, the extent of financial inte-gration that is essentially expected to improve risk sharing should also be takeninto consideration in order to determine whether it is desirable to form monetaryunion or not
We also investigate how price co-movements are affected by deeper trade andfinancial integration A number of studies point out that prices across countriesare not converged because of so-called ‘border effects’ The high border effectsimply that resource allocation is not efficiently made across countries The degree
of integration between economies can be assessed by estimating the bordereffects As trade and financial integration deepen, however, the border effects areexpected to diminish We attempt to examine which integration is more effective
in reducing the border effects reflected in the price movements
The remainder of the paper follows in five sections In Section 2, we brieflyreview how trade and financial integration have advanced in East Asia InSection 3, we explain the data used in the empirical analyses Section 4 presentsour model and discusses the main empirical results on the impacts of trade andfinancial integration on output, consumption and price co-movements Conclud-ing remarks follow in Section 5
2 TRADE AND FINANCIAL INTEGRATION IN EAST ASIA
The export-led growth strategy in East Asia has provided impetus for theirrapid growth in the volume of trade in this area This is well illustrated by Table 1
2 This is true under an appropriate assumption on preference Mace (1991) showed that if the utility function takes a CRRA (constant relative risk aversion) form, complete risk sharing implies that the growth rate of consumption is equalised across countries.
Trang 3TRADE AND FINANCIAL INTEGRATION IN EAST ASIA 1651
which reports the share of trade (exports + imports) and GDPs of East Asiancountries and other areas in the world economy In the table, East Asian countriesare further divided into individual countries such as China, Japan and Korea, and
a group of remaining countries, ASEAN.3
According to Table 1, East Asia’s share in total global trade continuouslyincreased from 13.9 per cent in 1980 to 22.2 per cent in 2000 and then more orless stayed at around the same level until 2003 The share of GDP in East Asia alsoshows a similar pattern: East Asia’s share of GDP increased from 13.9 per cent
in 1980 to 22.6 per cent in 2000, but rather decreased a little since then.However, China’s share of trade or GDP has continuously increased WhileChina’s share in trade (one per cent) was far less than that of Japan in 1980(7.3 per cent), it has been increasing tremendously for the last 25 years, beingcomparable to Japan in 2003 China’s accomplishment in promoting trade isespecially remarkable since China’s share of GDP (3.9 per cent) is still far lessthan that of Japan (11.8 per cent) as of 2003
Due to the astonishing performance of China, the integration of trade amongEast Asian economies has also been steadily increasing According to Shin andWang (2005), the percentage of intra-regional exports in total exports increasedfrom 30.3 per cent in 1980 to 45.8 per cent in 2003 The corresponding percentage
3 ASEAN includes Myanmar, Cambodia, Indonesia, Malaysia, the Philippines, Singapore, Thailand and Vietnam We have added Hong Kong, Macau and Mongolia to ASEAN instead
of treating them separately.
TABLE 1 Trade Share of East Asia in the World
Trang 41652 KWANHO SHIN AND CHAN-HYUN SOHN
of intra-regional imports in total imports increased from 30.9 per cent in 1980 to49.2 per cent in 2003 Among the economies in East Asia, Japan had the lowestintra-regional share of trade at about 39.2 per cent in 2003
According to Lee and Shin (2006), the share of intra-regional trade in East Asiawas somewhat lower than the corresponding value for the EU area, which was 66per cent in 2000 They point out that one reason for relatively lower levels of intra-regional trade is a relatively larger share of trade with the United States Theshare of trade with the United States of total trade was about 14.1 per cent forEast Asian economies on average, contrasting to about eight per cent for Europeancountries in 2000 But, East Asia’s trade with the US tended to decline graduallyover the past decade and the same share amounts to 11.3 per cent in 2003 Asthis trend continues, the share of intra-regional trade is expected to grow further
In East Asia, there has also been a rapid increase in international capitalmobility, as East Asia has been deregulating its financial markets since the early1990s Bekaert and Harvey (1995), World Bank (1997) and Eichengreen andPark (2005a) pointed out that this continuous financial opening process hascontributed to the economies to become more integrated into global financialmarkets However, it is not clear that this process has also rendered the Asianeconomies to be financially more integrated within the region In general, whiletrade liberalisation tends to bring about trade integration more at the regionallevel, we may not expect that financial integration also takes place more intensely
at the regional level as well because financial assets are weightless In otherwords, since transaction costs are far less important for asset trade, there is noadvantage of financial integration among neighbouring countries
In fact, several studies claimed that the degree of financial market linkage inEast Asia remains still low and that, unlike trade integration, the integration offinancial markets in this region has been occurring more on a global level ratherthan on a regional level Park and Bae (2002) and Eichengreen and Park (2005b)pioneered this issue and found that East Asia has developed stronger financialties with the US and Western Europe than with one another Based on varioustests utilising cross-country interest rate and stock price data, Jeon et al (2005)and Keil et al (2004) also supported this finding By estimating the degree ofrisk sharing for East Asia, Kim et al (2006) also found supporting evidence thatthe degree of regional risk sharing within East Asia is quite low Using the mostrecent data, Kim et al (2006) confirm the above findings Hence, the majority ofempirical studies seem to suggest that the level of financial market integration inEast Asia is relatively lower.4
4 Despite this general tenor of existing research, some studies provide opposing evidence For instance, McCauley et al (2002) argued that the financial markets of East Asia are more integrated than is often suggested by investigating the international bond market and the international syndicated loan market.
Trang 5TRADE AND FINANCIAL INTEGRATION IN EAST ASIA 1653
3 THE DATA
We consider nine countries in East Asia: China, Hong Kong, Indonesia, Japan,Korea, Malaysia, Philippines, Singapore and Thailand The data for output,consumption, price and the interest rate are from the International Financial
con-stant local currency unit and price refers to the CPI index The interest rate data
on 90-day local money market rates are available at a monthly frequency Thebilateral trade data are collected from the Directions of Trade dataset Othervariables are obtained from the dataset provided by Rose (2004) that includescontrol variables related to various measures of distance and size used in astandard gravity equation Since most data are available from 1971 our samplestarts from 1971 Because of the financial crisis in 1997 in East Asia, we considertwo different sample periods: the first sample is up to 1996 excluding the crisisperiod, and the second sample is up to 2003 including the crisis period.5
In this paper, we have also added another important variable, the exchangerate regime, which is believed to play a crucial role in determining co-movementsacross countries.6 Based on the de facto classification of exchange rate regimesmade by Reinhart and Rogoff (2004), we reclassify exchange rate regimes intotwo broad groups: a peg and a float To define exchange rate regimes betweenEast Asian countries, we infer the exchange rate regime between any two coun-tries based on their relationship with an anchor currency If the two countrieshave their currencies pegged simultaneously to a common anchor currency, weclassify their bilateral exchange rate arrangement as a peg If one country pegsits currency and the other floats, their relationship is dominated by a float andclassified as a float
The dataset has a feature of panel structure consisting of 914 annual bilateralobservations clustered by 30 country pair groups over time for sample I (1971–1996) and 1,166 annual bilateral observations for sample II (1971–2003) Thenumber of observations varies per year Summary statistics for the data used inestimation is presented in Panel A for sample I and Panel B for sample II
4 THE IMPACTS OF TRADE AND FINANCIAL INTEGRATION ON CO-MOVEMENTS
As trade and financial integration deepen, the business cycle dynamics ofoutput, consumption and price are also affected In the literature, a number of
5 The interest rate data are used until 1999.
6 See Lee and Shin (2004) for the importance of exchange rate regimes in determining movements of output, consumption and price across countries Based on 186 countries, they find that exchange rate regimes are crucial in explaining the co-movements across countries.
Trang 6co-1654 KWANHO SHIN AND CHAN-HYUN SOHN
studies have produced various theoretical implications of trade and financial gration We will summarise the implications of trade and financial integrationfirst and then use them to construct an empirical model that will be implementedlater
inte-a Theoretical Foundation
Trade integration affects co-movements in various channels and therefore thetheory does not warrant an unambiguous guidance as to whether more trade willincrease the degree of output and consumption co-movements or not First, thespillover of aggregate demand shocks through trade tends to make businesscycles more correlated across countries For example, if one country is hit by apositive demand shock, increased income will generate higher demand forimports as well, acting as a positive demand shock for a trading partner Second,
as Eichengreen (1992) and Krugman (1993) argued, if an increase in tradelinkages encourages greater specialisation of production, it will result in lesssynchronisation of business cycles In this case, industry compositions are shapedquite asymmetrically across major trading partners, and if business cycles aredriven mainly by industry-specific shocks, different compositions of industrieswill contribute to less synchronisation
TABLE 2 Summary Statistics
Panel B: Sample Period: 1971–2003 (Number of Obs = 1,166)
Panel A: Sample Period: 1971–1996 (Number of Obs = 914)
Log of trade 16.68 1.68 Log of distance 7.38 0.46 Log of GDP in pairs 41.72 2.09 Log of per capita GDP in pairs 6.84 1.80 Log of area in pairs 23.64 3.97 Common land border dummy 0.085 0.28
Log of trade 17.02 1.71 Log of distance 7.38 0.46 Log of GDP in pairs 42.08 2.12 Log of per capita GDP in pairs 7.02 1.86 Log of area in pairs 23.66 3.98 Common land border dummy 0.085 0.28
Note:
These sample statistics are for country pairings in East Asia: China, Hong Kong, Indonesia, Japan, Korea, Malysia, Philippines, Singapore and Thailand.
Trang 7TRADE AND FINANCIAL INTEGRATION IN EAST ASIA 1655
Third, Frankel and Rose (1998) countered the above argument, insisting that
if intra-industry trade is more pronounced than inter-industry trade, businesscycles will become more positively correlated as trade integration strengthens.Based on 21 industrialised countries, they actually found that the more countriestrade with each other, the more highly correlated their business cycles are Whilethey conjectured that this positive correlation is due to intra-industry trade, actualconfirmation is made by Shin and Wang (2004) that explicitly find that intra-industry trade is a major source for generating higher co-movements Lastly,increased trade may create a greater need for more coordinated fiscal as well asmonetary policies, which synchronise policy shocks Then, business cyclesbecome more correlated as movements of outputs are also driven by coordinatedpolicy shocks
Financial integration can also affect business cycle co-movements First,Claessen et al (2001), Calvo and Reinhart (1996) and Cashin et al (1995) arguedthat capital flow can generate business cycle co-movement for the countries inthe same area that experience ebb and tide of capital at the same time Forexample, during the Asian crisis and the Latin American crises, a number ofcountries in the same area faced outflow of capital simultaneously, aggravatingtheir economies at the same time Second, as suggested by Kalemli-Ozcam et al.(2001), better risk sharing attained through greater financial market integrationmay induce higher specialisation of production and hence larger asymmetricshocks across countries In other words, better income insurance provided byrisk sharing across countries enables each country to take more risk byspecialising more in industries, which leads to less synchronisation ofbusiness cycles
Third, better risk sharing due to deeper financial integration also has importantimplications for co-movements of consumption across countries as well Forexample, an influential paper by Backus et al (1992) showed that if internationalcapital markets are complete, country-specific technology shocks lead to equilib-rium consumption paths that are both less variable and less closely related todomestic output than they are in closed-economy real business cycle models.While quantitative properties of the theoretical economy depend to a large extent
on the specification and the parameter values of the model, the theory suggeststhat the consumption growth correlation across countries should be higherthan output growth correlation.7 Hence, financial integration may increase
7 Backus et al (1992), however, found using data for 11 OECD economies that the consumption growth correlation is actually lower than the output growth correlation This is referred to as one
of the six major puzzles in international economics and termed as the international consumption correlation puzzle by Obstfeld and Rogoff (2001) Recently, Hess and Shin (1998) and Crucini (1999) extended the analysis to intra-national data based on state-level regional data in the US and found that the puzzle is preserved even within a country.
Trang 81656 KWANHO SHIN AND CHAN-HYUN SOHN
or at least does not decrease consumption co-movement as much as output
co-movement does
While there have been various models developed to demonstrate how trade
and financial integration affect output and consumption co-movements across
countries, less attention has been made on the effects of trade and financial
integration on price co-movements We expect, however, that both types of
inte-gration enhance price co-movements Especially, deeper financial inteinte-gration
implies that the arbitrage opportunity of trading financial assets weakens,
imply-ing quicker convergence of prices of assets As trade increases, the arbitrage
opportunity of trading goods also disappears, suggesting that the price of real
goods converges more quickly
b The Empirical Model
Since theoretical predictions are varied and often conflicting in some cases,
the answer to the impacts of trade and financial integration on output,
consump-tion and price co-movements lies in the empirical analyses To implement the
empirical analyses, we need to construct co-movement measures and the indices
of trade and financial integration
We compute co-movements of each variable empirically by following the
same approach to Lee and Shin (2004) that extends Alesina et al (2002) and
Tenreyro and Barro (2002) For output co-movement, we calculate relative output
movements between countries i and j by subtracting output growth for country
j from that for country i: ∆ ln(Y it) − ∆ ln(Y jt) Then for every pair of countries,
(i, j), we compute the second-order auto-regression of the annual time series:
∆ ln(Y it) −∆ ln(Y jt) =c0+c1(∆ ln(Y it−1) −∆ ln(Y jt−1))
+c2(∆ ln(Y it−2) −∆ ln(Y jt−2)) + (1)
We use the negative of the absolute value of the estimated residual multiplied by
100 as the extent of output co-movement at each point of time:
(2)
We also measure the extent of co-movements for the entire sample period by
computing the negative of the root-mean-squared error multiplied by 100.8
In the same way, we use relative consumption and price movements,
∆ ln(C it) − ∆ ln(C jt) and ∆ ln(P it) − ∆ ln(P jt ), between countries i and j, and
compute the co-movement measures of consumption and price:
8 See Lee and Shin (2004) for a detailed derivation of the co-movement measures.
u ijt Y
CoY ijt = −|u ijt Y | 100 ×
Trang 9TRADE AND FINANCIAL INTEGRATION IN EAST ASIA 1657
(3)(4)where and are the residuals estimated by the second-order auto regression
of relative consumption and price movements between countries i and j respectively Trade integration between a pair of countries, (i, j ) is defined by normalising
trade (exports + imports) between the pair by the sum of world trade made bythe pair as follows:9
where x ijt (x jit ) denotes total nominal exports from country i ( j ) to country j (i ) during period t; m ijt (m jit ) denotes total nominal imports from country j (i) to country i ( j ) during period t; and X it (X jt ) and M it (M jt) denote total global exports
and imports for country i ( j ) during period t.
While trade integration measure is quite straightforward, a measure of cial integration is generally hard to obtain In the literature, some studies useddirect measures of bilateral capital flows for a subset of countries However, suchmeasures are not available for the countries considered in this paper Henceforth
finan-we decide to use an indirect measure based on the returns on financial assets.Relying on the high-frequency movements of the short-term interest rate, wederive an index of financial integration Namely, we calculate the correlation ofthe monthly interest rates during the corresponding year and use it as a measure
of financial integration
Unlike the measure of trade integration, a caution is warranted to draw themeasure of financial integration from the co-movements of the returns on finan-cial assets such as the interest rate That is, we cannot conclude that the financialintegration is deeper simply because the interest rates move more closelytogether For example, if each country is strongly integrated to a third country,despite no actual integration between the two countries, the interest rates in thetwo countries can move together closely This is a very realistic scenario for EastAsian countries because a number of countries in this area are expected to have
a strong connection to the global financial markets such as the US market
In order to isolate the bilateral integration between any two countries, weeliminate the connection of each country to the global market by regressing theinterest rate of each country on the interest rate of the US and use the residuals
9 An alternative way is to normalise trade by the sum of total trade made by the pair of countries The main results do not change if this alternative measure is used.
Trang 101658 KWANHO SHIN AND CHAN-HYUN SOHN
For example, for each year t, we regress the monthly interest rates of Korea and
Japan on the monthly interest rate of the US respectively:
where , and are the monthly interest rates for Korea, Japan and the
US, respectively, for year t.10 Then we use the residuals, and to calculate
the correlation for each year t that will act as a measure of financial integration
between Korea and Japan for the corresponding year In general we define the
degree of financial integration between countries, (i, j ), as follows:
where and are the monthly residuals calculated from the regression ofeach country’s monthly interest rate on the monthly US interest rate for each year
In the main equation that investigates how trade and financial integrationaffect output co-movements, we employ two types of estimation based onthe panel regression and cross-section regression respectively The first type ofequation for the panel analyses is as follows:
Co_Y ijt = β0 + β1tradeintijt + β1exchangeijt + δYEARt + εijt
(5)
Co_Y ijt = β0 + β1financeintijt + β1exchangeijt + δYEARt + εijt,
where Co_Y ijt is the extent of output co-movement between country (i, j ) at each
point of time, and exchangeijt is the regime classification dummy Equation (5)enables us to utilise information in (1) at each time of the period, and hence toadopt a panel regression approach which allows us to eliminate unobserved,country-specific effects
While the first type of equation has its advantage of adopting panel regression,the residual term may not reflect the degree of co-movement at every period oftime Instead it is more likely that the degree of co-movement is measured by thesum of the residuals for the entire sample period The second type of equationhinges on this idea and forms a cross-section regression as follows:
10 In order to control the global market connection, instead of using the nominal interest rate of the US itself, we have also considered the nominal interest rate adjusted by the exchange rate changes, which is defined as , to take into consideration the exchange rate move- ments However, the main results are robust against the modification of the definition.