The Co-movement Between Returns of Foreign Exchange Rates in the Central European Countries

A C T A U N I V E R S I T A T I S L O D Z I E N S I S FOLIA OECONOMICA 192, 2005

M a ł g o r z a t a D o m a n *

TH E C O -M O V EM EN T BETW EEN R E T U R N S O F FO REIG N E XCH ANG E RATES IN TH E CENTRAL E U R O PE A N CO U N T R IE S

Abstract. The analysis of conditional correlations between returns o f foreign exchange rates gives us significant information about co-movement between different currency markets. In the paper, we model this kind o f dependency in the case o f currency markets in Central European countries using Engle’s DCC models. We investigate the changes in the level of conditional correlations during stability and crisis periods. In this context we try to find the evidence o f the contagion effect in the considered region.

Keywords: currency market, co-movement, dynamic correlations, exchange rates, contagion. JEL Classification: F31, C32.

1. INTRODUCTION

T h e questio n a b o u t co-m ovem ent o f security prices is one o f the m o st im p o rta n t problem s in finance theory an d applications. A n answ er to this question affects such fields as portfolio selection, risk m anag em ent or security pricing. T h e analysis o f conditional correlations betw een financial returns gives a possibility to describe dependencies changing in tim e an d to get a deeper insight into the financial m ark ets dynam ics.

T h e currencies are an im p o rtan t class o f assets. T h e m echanism o f currency co-m ovem ent is essential for diversifiability o f currency exchange risk, w hich occurs d u rin g cross-border investm ents an d trad e. H ow ever, the analysis o f co-m ovem ent in the case o f currency m ark ets is m o re com plicated th a n it is, fo r exam ple, in the case o f capital m ark e t. T he specifics o f this problem are connected with the fact th a t exam ining the linkages between currency m ark e ts, we use exchange rates w hich are alw ays against som e

th ird currency (usually US d o llar or euro). T he results o f analysis essentially depend on the choice o f this third currency.

In this p ap e r we use Engle’s dynam ic co nditio nal co rrelatio n (D C C ) m odel to investigate the dynam ic conditional co rrelatio n betw een currencies o f C en tral E u ro p e an countries (Czech R epublic, H u n g ary , P o la n d , Slovakia) and R u ssian ruble. O u r analysis is based on the re tu rn s o f exchange rates against th e U S d o llar an d the euro. As b y prod uct, we get som e results concerning the co n tag io n effect in this region.

2. CO-MOVEMENT OF CURRENCY MARKETS

In view o f financial m ark et globalization, th e currency m ark e t co­ m ovem ent is w orth the careful analysis because o f its im p o rtan ce for risk m anag em ent. T h e currency co-m ovem ent is essential for the diversification o f currency exchange risk. T ho u g h the exchange rates fluctuations are subject to different shocks, it is possible to find som e p a tte rn o f dependencies suggesting th a t som e currencies can co-m ove in a p red ictab le m a n n e r (cf. F ig u re 2).

T h e cro ss-m ark et linkages can be m easured by a n u m b e r o f different statistics such as the co rrelation in asset return s, the p ro b a b ility o f a spe­ culative attac k , o r the transm ission o f shocks o r volatility. H ow ever the analysis o f co nditional correlations between the retu rn s o f interest seems to be the m o st p o p u lar m ethod. Such app ro ach is very co m m o n in literature concerning the capital m ark e t co-m ovem ent. T he suggestion th a t one should m easure contagion as a significant increase in th e co rrelatio n betw een asset re tu rn s, com es from K ing and W adhw ani (1990).

T h e investigation o f currency m ark ets linkages is m o re com plicated th an it is in the capital m a rk e t case. T h e reason for this is th a t th e analysis o f co-m ovem ent betw een tw o currencies requires the co n sid eratio n o f exchange rates which are alw ays calculated against a third currency. T h e choice o f this th ird currency significantly influences the results o f com pariso n. T he plots o f exchange rates presented in F igures 1 and 2 show different behavior o f the retu rn s against the US d o llar and euro. T h e diverse dependencies betw een the retu rn s o f exchange rates calculated again st different currencies m ake the inference rath er sophisticated. As m entioned above, the exam ination o f cross-m arket correlations is usually based on som e co rrelatio n tests. In this p ap e r, we p ropose to analyze the dependencies betw een som e currency m ark e ts by directly m odeling the dynam ic co n ditio nal co rrelatio n between exchange ra te returns. T he to o l is E ngle’s D C C m odel presented in the next section.

... EUR/CZK ---(EUR/HUF)/10 --- (E UR/PLN)-10 --- EUR/RUB --- EUR/SKK

Fig. I. The plots o f EUR/CZK, (EU R/HUF)/10, (E U R /P L N )• 10, EUR /R UB and EUR/SKK

Fig. 2. The plots o f USD/CZK, (USD/H U F)/10, (U SD /PLN )- 10, USD/RU B and USD/SKK

T h e subject o f special interest o f researchers and p ractitioners is th e effect o f co n tag io n occurring w hen a financial crisis starts as country-specific event b u t quickly spreads in to o th er countries. T here is n o ag reem ent on the definition o f co ntagion betw een the financial m arkets. In this p ap e r we apply F o rb e s’ and R ig o b o n ’s (2001) definition o f contag io n. It m ean s th a t we characterize th e contag io n as a significant increase in cro ss-m ark et linkages after a crisis occurring in one country. T he term interdependence is used in the situ atio n w here the strong linkages between the tw o m ark e ts are perm anent.

A s concerns the question ab o u t the evidence o f co n tag io n , th e analysis o f direct m odeling o f correlations seems to have som e adv an tag es over the co rrelatio n tests. T h ere is no need to divide the sam ple accordin g to the crisis and non-crisis periods. T h e crisis period is easy to identify as a period

o f increased volatility. T h e breakpoints are detected by m eans o f an estim ated volatility level. T h e m odel we used takes into account the co nd ition al hetero­ scedasticity o f re tu rn series. It is very im p o rtan t because o f the fact th a t the m ajo rity o f co rrelatio n tests occur to be inaccurate du e to heteroscedasticity. F o rb es an d R igobon (2001, 2002) claim th a t cro ss-m ark et co rrelatio n coeffi­ cients are co n d itio n al on m ark e t volatility. D u ring crises, m ark e ts arc m ore volatile and estim ates of correlation coefficients tend to increase and be biased upw ard which results in finding spurious evidence o f co ntagion .

3. THE DCC MODEL

Let y { and y 2 be tw o ran dom variables with th e co n d itio n al m eans equal to zero. T h e co nditional correlatio n betw een an d y 2 is defined by

Е(У1У2\П, i)

w here Q t_! is the set o f info rm ation available on the daily re tu rn process r, up to tim e £ — 1.

It is reasonable to assum e th a t conditional co rrelatio n s change over tim e. T h e problem o f estim ation o f the co n ditio nal co rrelatio n between financial variables is still widely discussed in the econom etric literature. In this p ap e r, we propose to use E ngle’s (2000) D C C m odel to describe the con ditio n al co rrelatio n s betw een returns and volum e.

Let r( = ( r ..., rkl)' be a m ultidim ensional series o f retu rns. T h e r, can be w ritten as the sum r, = ц, + y„ where y, = (y u , ..., y kt)' an d ц, = E (r(|Q ,-x ) is the co n d itio n al m ean o f the vector rt up on th e in fo rm atio n set Q ,_ i.

In the D C C m odel, as in m ultivariate G A R C H m odels, we assum e th a t

y t satisfies the eq uation

yf = H P £ t,

where e, is a /с-dim ensional process o f norm ally d istrib u ted independent ra n d o m variables w ith unit m ean and the covariance m atrix I*(e, ~ iid(0, I*)). T h en ~Л Г(0,Н ,). T he m atrix H , is specified as

H , = D,RfDt,

where D ( = diag(V f»u„ •••, \ f h a t) and the con ditio n al variances hiit can be m odeled by any univariate G A R C H m odel, fo r exam ple,

ч P

hllt = o)t + Ž<xuy l t - j + Z ß i M i - j , * = 1, •••, к

]= i j=i

T h e m atrix R, is defined by

R, = (diag(Q () ) = - 1/2 Q t(diag (Q t) ) ~ 1/2.

T h e positive definite sym m etric (k x k) m atrices Q , are assum ed to satisfy the eq u a tio n

Qt = (i -

Ž

ä

W Í

w

- .

u

;-»

i

Z

äq

,-..

\ m - l в - l / m- 1 я = 1

T h e coefficients o f /с-dim ensional vectors u( are o f the form uit = y it/ s / h iit. T h e ( k x к) m atrix Q is the unconditional covariance m atrix o f the variables u(. T h e assum ptions a b o u t the p aram eters a m, ß n are as follows: a m, ß n > 0 and Z “» .+ Z A < 1'

m = 1 n = 1

T h e special case o f the above m odel (c o n stan t m atrix R, = R) is the c o n sta n t co n d itio n al correlatio n m odel introduced by Bollerslev (1990).

A n estim atio n procedure for D C C m odel proceeds in tw o steps where in the first one the univariate G A R C H m odels are estim ated for each residuals series y it, and in the second one, the residuals divided by their conditional stan d ard deviation estim ated in the first step, are used to estim ate the p aram eters o f the dynam ic correlation. T h e likelihood fun ctio n used in the first step is the sum o f the likelihoods o f the considered univariate G A R C H m odels and so it is given by

Q L ,( < p |y ,)= Z fc ln (2 ^ ) + 2 1 n (|D f|) + y ; D r 1I fcD r 1y«) = A = i

= ^ZÍrin(2*) + (Z(l*№) + f j j

-G iven the p a ra m e te r vector (p, the consistent estim ato r o f co rrelation p aram eters vector \|/ can be determ ined by m eans o f the lik eliho od function

Q L 2(4/1 <p,y() = - * £ ( ln IR < I + uíR <“ 1

Zt=l

T h e softw are we used for the calculations presented in this pap er is the U C S D _ G A R C II T oo lb o x v. 2.0 by K . S heppard.

4. THE DATA

T h e d a ta consist o f daily exchange rates o f C zech k o ru n a (C Z K ), H u n g a rian fo rin t (H U F ), Polish zloty (P L N ), R ussian ruble (R U B ) and Slovak k o ru n a (S K Jt) against the US d o llar (U S D ) and eu ro (E U R ). In each case the elem ents o f tim e series under scrutiny are the prices o f 1 U SD in th e c o rre sp o n d in g d om estic currency. T h e period u n d e r analysis is 14.11.1996 - 27.02.2004. It gives 1870 observations in each tim e series. T he analysis is based on the percentage logarithm ic returns r„ given by the form ula

r, = 100(ln(p,) ln(pt_ j) ) ,

w here p, m ean s the value o f exchange rate on day t.

T h e descriptive statistics o f the re tu rn s are presented in T ab le 1.

Table 1. Descriptive statistics o f returns

Data Min Max Mean Std Kurtosis Skewness

EUR/CZK. -2.7919 8.8903 -0.0028 0.5339 46.603 2.5741 E U R /H U F -2.6219 6.3644 0.0136 0.4753 33.191 2.4671 EUR/PLN -3.1100 5.5271 0.0170 0.7003 7.4788 0.7580 EUR/RUB -33.0391 30.1915 0.0872 2.1445 96.316 1.5236 EUR/SKK -3.4653 6.2110 0.0019 0.4764 34.225 1.9180 USD/CZK -2.9196 9.5904 -0.0019 0.7794 15.9428 0.8829 U SD/H U F -3.2520 5.7415 0.0145 0.7042 9.1738 0.5592 USD/PLN -3.4125 4.2208 0.0179 0.6304 6.8919 0.3112 USD/RU B -33.163 28.959 0.0881 2.0534 107.97 1.3638 USD/SKK -3.5921 6.0160 0.0028 0.7055 9.6405 0.4639

A n im p o rta n t p a rt o f o u r analysis concerns th e co n tag io n effect between the m ark e ts u n d er scrutiny. F o r this reason, we p resent som e in fo rm ation a b o u t the financial crises th a t occurred in th e investigated perio d. O ur ev alu atio n o f the crisis periods is strongly influenced by the results o f Serwa and Bohl (2003). F ro m o u r point o f view, the m o st im p o rta n t are the situ atio n s w hen a crisis started in the co u n try from the co nsidered group.

T h ere arc th ree such currency crises: the Czech k o ru n a in M ay 1997, the R ussian ruble in 1998 and the H u n g arian fo rin t crisis in 2003.

Table 2. Rough information about the crisis periods and corresponding observations in presented data set

Crisis Period (approximately) Observations

Czech koruna crisis May 1997 115-131

Asian flu July - December 1997 157-282

Russian cold June - September 1998 386-470 Brazilian fever November 1998 - March 1999 491-581 Turkish collapse December 2000 - March 2001 1029-1110

September 11 September 2001 1231-1254

Argentina January - February 2002 1300-1353

Forint crisis June 2003 1679-1700

5. EMPIRICAL RESULTS

O u r analysis o f co-m ovem ent betw een the C en tral E u ro p e an currency m ark e ts is based on the tw o groups o f five exchange rates presented in T ab le 1. A ccording to the assum ption o f D C C m odel, we fit the VAR m odel to each g ro u p o f tim e series. T hen we apply the Engle and S heppard (2001) c o n sta n t co rrelatio n test to the V A R residuals. T h e null hypothesis o f c o n sta n t co rrelatio n is strongly rejected in b o th cases (T able 3). These results justify the m odeling o f dynam ic correlations.

Table 3. Results from Engle-Sheppard’s constant conditional correlation test

Reference currency Test statistics p-value

EUR 16.0521 0.0003

USD 24.0262 0.0000

T h e p aram eters o f univariate G A R C H and D C C m odels are presented in T ables 4 and 7. A p a rt from the dynam ic co n ditio nal correlation s, we estim ate the c o n sta n t conditional correlations for exchange rates in each g roup, using B ollerslev’s C C C m odel. T he values presented in T ab les 5 and 8 can be treated as representing the average level o f the dynam ic conditional correlations.

la b ie 4. Parameters o f fitted GARCH models and DCC model. The analysis of dynamic correlations between EUR/CZK, EU R /H U F, EUR/PLN, EUR /R U B and EUR/SKK

GARCH CZK HUF PLN RUB SKK

0) 0.0168 (0.0043) 0.0446 (0.1651) 0.0459 (0.0093) 0.0258 (0.0068) 0.0117 (0.0058) «1 0.1882 (0.0477) 0.2868 (0.0928) 0.1359 (0.0241) 0.1435 (0.0289) 0.3074 (0.1074) A 0.7677 (0.0243) 0.5390 (0.0957) 0.7745 (0.0118) 0.8372 (0.0099) 0.6926 (0.0741) Parameters o f DCC model 01 0.0166 (0.0056) ß 0.9623 (0.0164)

Table 5. Constant conditional correlations between EUR /C ZK , E U R /H U F , EUR /PLN, EUR/RUB and EUR/SKK

SKK RUB PLN HUF

CZK 0.3138 (0.0347) 0.1290 (0.0290) 0.2310 (0.0291) 0.1549 (0.0749) HUF 0.2485 (0.0680) 0.1860 (0.0445) 0.3198 (0.0477)

PLN 0.2026 (0.0369) 0.4775 (0.0312) RUB 0.1655 (0.0345)

T h e p lots in F igures 1-10 show the estim ates o f co n ditio nal co rrelatio ns for the exchange rates against euro. T h e co n stan t co nd itio n al co rrelatio ns have ra th e r low, positive values (0.1 -0 .2 ), except for the co rrelatio n between E U R /P L N and E U R /R U B - a b o u t 0.5. T h e flu c tu a tio n s o f d yn am ic co rrelatio n s are qu ite strong. A s concerns th e co n tag io n effect, d u rin g the tim e o f alm ost all crises one can observe a significant ju m p to high positive level in d y n a m ic c o rre la tio n s in th e case o f E U R /C Z K -E U R /P L N , E U R /H U F -E U R /P L N and E U R /P L N - E U R /S K K . A t the sam e tim e, there is surely no increase in correlations betw een E U R /C Z K an d E U R /H U F . T h e results concerning the rem aining pairs o f exchange rates are rath er am bigu ous (T able 6).

Fig. 3. Conditional dynamic correlations between EUR/CZK and E U R /H U F with the level of constant conditional correlation (CCC)

Fig. 4. Conditional dynamic correlations between EUR/CZK. and EUR/PLN with the level o f constant conditional correlation (CCC)

Fig. 6. Conditional dynamie correlations between EUR/CZK and EUR /SK K with the level o f constant conditional correlation (CCC)

Fig. 7. Conditional dynamic correlations between EUR /H U F and EUR /PLN with the level o f constant conditional correlation (CCC)

Fig. 9. Conditional dynamic correlations between E U R /H U F and EUR/SK K with the level o f constant conditional correlation (CCC)

Fig. 10. Conditional dynamic correlations between EUR/PLN and EUR/SK K with the level o f constant conditional correlation (CCC)

•0.1 -0 .2 •0.3 -0.4 — R U B -S K K — CCC I S S S S S S S S S S S S K S S S S g S S S S K S S S S g S S S S K g S

Fig. 12. Conditional dynamie correlations between EUR/PLN and EUR/RUD with the level of constant conditional correlation (CCC)

Table 6. Exchange rates against euro

EUR Czech Asia Rusia Brazil Turkey September 11 Argentína Hungary

C Z K -H U F 1 Í t 1 i i t T C Z K -P L N «—f ₁ «-+ _t «-♦ 4—* _t *—¥ C Z K .-R U B 1 Í i i t 1 i C Z K -S K K _T 1 1 t i i t 1 H U F -P L N _{i T Í} Í i T 1 t H U F -R U B _{T T t i i Í 1 1} H U F -S K K _{i T Í T i i} * - * _Í P L N -R U B Í T i 1 i T I T P L N -S K K _{T T t t i T i T} R U B -S K K _{Í t 1} 4—► «—> «—► _{1 T}

Note: Behavior o f dynamic correlations during period of crises presented in Table 2: f - significant increase; *-*- increase, but not very high; | - no increase or even decrease in dynamic conditional correlation level.

T ables 7 and 8 show the p aram eters o f the D C C m odels and co n stan t co n d itio n al correlations estim ated for exchange rates ag ain st U SD . C orres­ p o n d in g plots o f conditio n al co rrelatio ns are presented in F igu res 13-22.

Table 7. Parameters of fitted GARCH models and DCC model. The analysis o f dynamic correlations between USD/CZK, U SD /H U F, USD/PLN, U SD /R U B and U SD/SKK

GARCH CZK HUF PLN RUB SKK

Ш 0.0801 (0.0146) 0.0229 (0.0062) 0.0512 (0.0097) 0.0025 (0.0008) 0.0632 (0.0188) “ i 0.1136 (0.0486) 0.0672 (0.0158) 0.1984 (0.0416) 0.2211 (0.0350) 0.0598 (0.0253) 0.7571 (0.0136) 0.8875 (0.0020) 0.6780 (0.0315) 0.7789 (0.0185) 0.8127 (0.0088) Parameters of DCC model a 0.0373 (0.0070) ß 0.9205 (0.0242)

Table 8. Constant conditional correlations between U SD /C ZK , U S D /H U F , USD/PLN , USD/RUB and USD/SKK

SKK RUB PLN HUF

CZK 0.7101 (0.0340) -0.0003 (0.9901) 0.4100 (0.0279) 0.6428 (0.0485) HUF 0.6659 (0.0471) -0.0052 (0.8212) 0.4408 (0.0228)

PLN 0.3689 (0.0352) 0.0393 (0.0221) RUB -0.0079 (0.7212)

T h e linkages betw een analyzed currencies m odeled o n the basis o f exchange rates against U S D are m ostly strong er (m ean level a b o u t 0.4-0.7) th a n those ag ain st the euro. T he correlations betw een the C en tral E u ro pean and R ussian exchange rates change the signs and th eir m ean level is a b o u t zero

Fig. 14. Conditional dynamie correlations between USD/CZK and U SD/PLN with the level of constant conditional correlation (CCC)

Fig. 15. Conditional dynamic correlations between USD/CZK and U SD /R U B with the level of constant conditional correlation (CCC)

0.9 08 0.7 0 6 0.5 0.4 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3

Fig. 17. Conditional dynamic correlations between U SD /H U F and U SD/PLN with the level o f constant conditional correlation (CCC)

Fig. 18. Conditional dynamic correlations between U SD /H U F and U SD /R U B with the level o f constant conditional correlation (CCC)

Fig. 20. Conditional dynamie correlations between U SD/PLN and U SD /R U B with the level o f constant conditional correlation (CCC)

Fig. 21. Conditional dynamic correlations between USD/PLN and U SD/SKK with the level o f constant conditional correlation (CCC)

T a b le 9 describes the chang es o f d y n am ic c o n d itio n a l c o rrelatio n s betw een th e exchange rates against the U SD . T h e shadow ed row s present pairs o f exchange rates for which the co rrelatio n level m ostly increase d u rin g the crises period. H ere we have six pairs o f currencies presenting such behavior: C Z K -P L N , C Z K -R U B , H U F -P L N , H U F -R U B , P L N -R U B and P L N -S K K .

Table 9. Exchange rales against USD

EUR C z e c h Asia Rusia Brazil Turkey September 11 Argentina Hungary

C Z K -H U F _{i t I} i i i 1 T C Z K -P L N _t

T

r

Note: Behavior o f dynamic correlations during period of crises presented in Table 2: t - significant increase; «-* - increase, but not very high; | - no increase or even decrease in dynamic conditional correlation level

T h e analysis o f dynam ic conditional correlations leads us to a conclusion th a t C Z K , H U F , PLN and SK K co-m ove m uch m o re ag ain st U SD than against the euro. A ccording to E un and Lai (2004), it reflects the fact th at all above m en tio n ed currencies are strongly connected w ith euro, which is the d o m in atin g currency in the considered region. O ne can suppose th a t the co rrelatio n level will even decrease after jo in in g the E U by Czech R epublic, H u n g a ry , P o land and Slovakia. T h e sign o f dynam ic correlation betw een exchange rates o f C en tral E u ro p e an cou ntries an d R u ssia changes over time. In such situ atio n the lim itation of analysis to c o n sta n t conditional c o rrelatio n w ould result in conclusion o f no interdependence.

T h e fact th a t the degree o f co-m ovem ent o f exchange rates depends on the choice o f the reference currency m akes investigation o f co n tag io n effect very sophisticated. I t is n o t o u r goal to study this p ro blem in general in this paper. A ctually, we only p o in t the situations w hen there is no contagion because the co rrelatio n level d uring crisis does n o t increase. O u r results

allow us to claim th a t there is no contagion in th e case o f pairs: H U F -C Z K and R U B -C Z K . W e can observe an increase in positive correlatio n o f exchange rates against U S D and E U R between the Polish zloty and the H u g a rian fo rin t as well as between the Polish zloty and the Slovak k o ru n a and the Polish zloty and the Czech koruna. T h u s th ere is a possibility th at co n tag io n cfTcct can exist in the case o f these pairs o f currency m arkets. In the case o f the rem aining pairs the dependencies are n o t so clear. D uring som e crises the dynam ic conditional correlation level increases while during the o th er it decreases.

6. CONCLUSIONS

T h e m echanism o f currency co-m ovem ent is essential fo r diversifiability o f currency exchange risk, which occurs d u rin g cro ss-b o rd er investm ents and trad e, an d so it is a subject o f special interests o f financial researchers and practitioners. T h e cross-m arket linkages can be m easured by a num ber o f different statistics such as the correlation in asset returns, th e transm ission o f shocks o r volatility, o r the prob ability o f a speculative attack . In this paper we describe the co-movem ent of Central European and Russian financial m ark e ts by directly m odeling dynam ic conditional co rrelatio n s between the retu rn s o f exchange rates. T h e m ain problem occurring d u rin g th e analysis o f currency m a rk e t linkages is th a t the co-m ovem ent o f currencies is always against a third currency. T he choice o f this third currency essentially influenced the results o f investigation. O ur results show the significant differences in behav ior o f analyzed exchange rates against th e US d o llar an d against the euro. C Z K , H U F PLN and S K K co-m ove m uch m o re ag ain st the U SD th an against the euro. T his p henom eno n is a result o f the do m inating positio n o f eu ro in considered region. T he financial m ark e ts o f Czech R epublic, H un g ary , Poland and Slovakia show the high degree o f in ter­ dependence. T h e R ussian currency does n o t indicate any p erm a n en t linkages w ith the o th er currencies un d er scrutiny.

REFERENCES

Bollerslev, T. (1990), “ Modeling the Coherence in Short-run Nominal Exchange Rates: A Multivariate Generalized ARCH Model”, Review o f Economics and Statistics, 72, 498-505. Engle, R. F. (2002), “Dynamic Conditional Correlation - A Simple Class o f Multivariate

GARCH M odels”, Journal o f Business and Economic Statistics, 20, 339-350.

Engle, R. F. and Sheppard, К (2001), Theoretical and Empirical Properties o f Dynamic Conditional

Eun, C. S. and Lai, S.-C. S. (2004), “The Currency Comovement”, Paper presented at the FM A European Conference, Zurich.

Forbes, K. and Rigobon, R. (2001), “Measuring Contagion: Conceptual and Empirical Issues”, ln: Claessens, S. and Forbes, K. i. (eds.), international Financial Contagion, Boston, London: Kluwer Academic Publishers.

Forbes, K. and Rigobon, R. (2002), “N o Contagion, Only Interdependence: Measuring Stock Market Co-Movements”, Journal o f Finance, 57, 2223-2261.

King, M. A. and Wadhwani, S. (1990), “Transmission o f Volatility Between Stock Markets”,

Review o f Financial Studies, 3, 5-33.

Serwa, D . and Bohl, M. T. (2003), “Financial Contagion: Empirical Evidence on Emerging European Capital Market” , EUV: Working Paper Series, 5.

Małgorzata Doman

W SPÓŁZALEŻNOŚCI MIĘDZY STOPAMI ZWROTU KURSÓW WALUTOWYCH W KRAJACH EUROPY CENTRALNEJ

(Streszczenie)

Analiza korelacji warunkowych między zwrotami kursów walutowych daje nam istotną informację na temat współzależności pomiędzy rynkami walutowymi. W niniejszym artykule opisujemy ten rodzaj zależności w przypadku rynków walutowych w krajach Europy Centralnej, posługując się modelem dynamicznych korelacji warunkowych (DCC), wprowadzonym przez Engle’a. Badamy zmiany w poziomie korelacji warunkowych między kursami analizowanych walut względem euro i dolara amerykańskiego, w okresach stabilności rynków i w okresach kryzysów. Uzyskane wyniki wskazują, że analizowane kursy walutowe podążają za od­ powiadającymi kursami euro. Otrzymujemy również pewne wyniki dotyczące efektu zarażania pomiędzy rozważanymi rynkami.