The Macrotheme Review A multidisciplinary journal of global macro trends

Transcription

1 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 The Macroheme Review A mulidisciplinary journal of global macro rends The Balance of Paymens Dynamics in he of Irina Khvosova, Alexander Larin, Anna Novak*, and Andrei Shulgin Naional Research Universiy Higher School of Economics, Russia aenovak@hse.ru* Absrac The paper analyses he key facors of balance of paymens dynamics for counries wih differen exchange rae regimes. We consider he differences in approaches o he analysis of balance of paymens effecs, and provide an overview of recen sudies on curren accoun and capial accoun dynamics. We presen an esimaes based on quarerly daa on 40 counries wih floaing exchange rae regime and 38 counries wih inermediae exchange rae regimes from 2006 o In he period of crisis, he response of rade balance is opposie depending on exchange rae regime. Daa also suppor he hypohesis of reversal effec of BOP. The hypohesis abou ineres rae o be a policy insrumen in crisis period is no suppored. Keywords: balance of paymens, curren accoun, capial accoun, exchange rae regime 1. Inroducion Moneary policy is closely conneced wih he dynamics of balance of paymens (BOP). If a counry chooses a floaing exchange rae regime, he BOP deermines he dynamics of he naional currency exchange rae, which, in urn, influence he macroeconomic performance of he counry. In his case he dynamics of BOP is an imporan indicaor of moneary policy. In case of inermediae exchange rae regimes he siuaion is more complicaed. On he one hand, moneary auhoriies preven significan exchange rae volailiy, ha is, i becomes possible o have shor-erm paymen imbalance. On he oher hand, he moneary auhoriies have o hink abou he fuure moneary sabiliy, for which hey need o adjus moneary policy o reduce he imbalance (Summers, 1996; Taylor, 2002). In his case, BOP is no only an indicaor, bu also he arge of moneary policy. The described effec was observed in developing counries during he global financial crisis in For example, several CIS counries (Ukraine, Belarus, Tajikisan, Kyrgyzsan, and ohers) were forced o devalue heir currencies only o reduce he emerging negaive paymen imbalances. Under floaing exchange rae regime, he shocks of curren and capial accouns are linked wih flucuaions in he equilibrium exchange rae. On he conrary, under inermediae exchange rae regime, Cenral Bank sabilizes boh BOP and exchange rae. As a resul, he exogenous shocks of curren and capial accouns are more closely conneced wih he exchange rae dynamics, han in inermediae regime (Kharel and Marin, 2010). Consequenly, in counries wih floaing and 82

2 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 inermediae exchange rae regimes under he same environmenal condiions in he shor erm, he join dynamics of curren accoun and real exchange rae may have a differen direcion due o he correcion of he moneary auhoriies. Capial accoun of BOP plays an imporan role in periods of problems wih liquidiy and insolvency (Calvo, 1996; Chang and Turnovsky, 2009). In counries wih floaing exchange regime, BOP imbalances acivae he marke mechanism, which leads o changes in he level of ineres raes. If moneary auhoriies prefer o inerfere in he esablishmen of BOP equilibrium, he negaive shock of capial ouflows can be compensaed by adjusing ineres raes. High ineres raes in his case are working in wo direcions: a) increase he demand for liquid asses in he domesic counry; b) increase he cos of servicing he naional deb. Therefore, in counries wih inermediae exchange rae regime his effec is uncerain during he crisis period. The aim of he paper is wofold. Firsly, he purpose is o discuss heoreical aspecs of he BOP effecs on he moneary policy. The paper examines he differences in approaches o he analysis of he BOP effecs; provides an overview of heoreical and empirical researches devoed o he analysis of BOP dynamics. The grea aenion is paid o he peculiariies of using he moneary insrumens o sabilize BOP, depending on exchange rae regime. Secondly, he aim is o invesigae he moneary dynamics of counries wih differen exchange rae regimes in he crisis period of and o reveal differences in he sabilizaion behavior of moneary auhoriies. The res of he paper has he following srucure. In Secion 2 we describe heoreical aspecs of BOP dynamics. Firsly, we consider rade balance as an elemen of he moneary ransmission mechanism. We summarize sudies ha consider he response of he curren accoun of BOP o he price level and he nominal exchange rae. We discuss here he exisence of he J-curve effec of he Marshall-Lerner condiion. Then, we consider BOP as an inermediae arge of moneary policy. Sudies on his problem are based on he assumpion of he exisence of he susainable level of curren accoun balance and on he idea of reversal dynamic of rade balance. Depending on exchange rae regime his process may differ significanly. Finally, we shif our focus o he role of capial accoun in he analysis of he balance of paymens. We consider sudies ha esimae moneary policy reacion o capial accoun shocks and conclude ha capial accoun is an inermediae goal of moneary policy in he period of problems wih liquidiy and solvency. In Secion 3 we provide he resuls of economeric esing of he BOP effecs in crisis period for counries wih floaing and inermediae exchange rae regimes. We show he differences in he sabilizaion policy of hese wo groups of counries. Firsly, we describe he daa used and he procedure of choosing he crisis period ha are based on Bai-Perron es. To analyze he relaion beween exchange rae and curren accoun, we esimae simple linear models using weighed ordinary leas squares. We esimae he coefficien a exchange rae o find how i affecs rade balance for differen group of counries for crisis and non-crisis periods. We also es he hypohesis of reversal effec of BOP for boh groups of counries. To reveal he relaion beween ineres rae and capial accoun, we esimae linear regression as well. We esimae, wheher ineres rae is considered as policy insrumen in crisis and non-crisis period or no. We also presen a series of ess o reveal he role of he difference in average values of shocks. We conclude in Secion 4. 83

3 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer Facors of BOP dynamics Exchange rae regime is a way for moneary auhoriies o esablish exchange rae relaions beween naional currency and foreign currencies. There are fixed, floaing and various ypes of inermediae exchange rae regimes 1. A key feaure for counries wih inermediae and fixed exchange rae is he addiional componen in BOP he change in inernaional reserves. This componen makes i possible o smooh flucuaions or compleely fix he exchange rae in case of BOP imbalance. Among many papers ha analyze he dynamics of BOP, here are several areas of research. The firs group of papers is devoed o he sudy of reacion of rade balance o changes in exchange rae. The second group pays aenion o he sudy of BOP reversal effec. The hird one sudies he dynamics of capial accoun, depending on a variey of financial facors Trade balance as an elemen of moneary ransmission mechanism In an open economy, moneary ransmission mechanism largely depends on he erms of rade of he counry, ha is, he curren accoun of BOP (Gali and Monaselli, 2008; Svensson, 2001, 2003). Trade flows are changing under influence of he relaive changes in he prices of radable goods and he dynamics of capial flows. The radiional approach o modeling he dynamics of BOP in reduced form is (Rose, 1991; Lee and Chinn, 1998; Boyd a al., 2001; Gomez and Paz, 2005): b * x m ( s p p ) x m e, where b rade balance, p price level in he counry, p* foreign price level, m volume of impors, s nominal exchange rae, x expors, e real effecive exchange rae. All he variables are log ransformed. Devaluaion is usually associaed wih he improvemen of rade balance. However, here is no consensus on how he effecs of expors and impors are disribued over ime. Orku (1950) argued ha he rade flow responds differenly o small, emporary shocks and large, permanen changes (e.g. devaluaion). This means ha he adjusmen of rade balance o large-scale changes in price level or o changes of nominal exchange rae is faser han he adjusmen o small changes. Tha is why he response of curren accoun of BOP in crisis period (which implies a subsanial devaluaion) may differ from is reacion in non crisis period. Laer his effec has been sudied for counries wih differen exchange rae regimes. Wilson and Takacs (1979), Janz and Rhomberg (1973) have sudied he difference in responses o changes in raes for counries wih fixed exchange rae regime. Wilson and Takacs have shown he same response of rade balance o nominal exchange rae and price level changes. Janz and Rhomberg subjeced hese resuls o he criicism and demonsraed ha he reacion ime o changes in nominal exchange rae is smaller han he response ime o change in price levels. This idea was furher developed by Bahmani-Osco (1986), Bahmani-Osco and Kara (2003), Hacker and Haami (2004), Boyd a al.(2001), Gomez and Paz (2005). In addiion, hese sudies also observe he J-curve effec. This phenomenon describes he fac ha he iniial impac of devaluaion is negaive for rade balance (reducion of impor exceeds expor (1) 1 IMF De Faco Classificaion of Exchange Rae Regimes April 31,

4 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 growh), bu over ime expors increases due o compeiive prices and, ulimaely, rade balance is growing significanly. Janz and Rhomberg explain he J-curve effec by differen lags, including ime lag, decision-making lag, delivery lag and produc replacemen lag. The resuls of Wilson and Takacs also have been expanded. In 1991 Tegene, using vecor auoregression approach, came o he conclusion ha expor and impor funcions are equally responsive o relaive price changes and changes in nominal exchange rae. The nex area of BOP researches is devoed o he devaluaion impac on BOP dynamics. The devaluaion of naional currency is usually associaed wih he improvemen of rade balance, bu is accompanied by an adjusmen o changes in real exchange rae wih a cerain ime lag. The size of he lag is deermined by individual characerisics of each counry. I may even be zero for some counries. The sudy of his effec has also been associaed wih he problem of he Marshall-Lerner condiion: devaluaion is accompanied by he growh of rade balance if price elasiciy of expor and impor is greaer han one. Empirical sudies of rade balance reacion o exchange rae flucuaions have found ha he condiion is no saisfied in he shor erm (see, for example, Gomez and Paz, 2005). However, he reacion may depend on he exchange rae regime, he size of he flucuaions, and may be differen for he cases of real and nominal devaluaion. The oher line of researches links BOP and exchange rae by analyzing how BOP imbalance affecs foreign exchange rae over ime. Mos heoreical works use dynamic approach o he analysis of inernaional paymens according o he join dynamic behavior of he various componens of BOP. The paper of Muller-Planenberg (2010) summarizes he resuls of heoreical sudies of dynamics of exchange rae and provides models ha explain fundamenal differences beween ineracion of exchange rae and inernaional paymens depending on imposed resricions on capial flows and exchange rae regime of he counry. Muller-Planenberg paper is a synhesis of heoreical models of previous sudies. Theoreical models of Muller-Planenberg have an empirical suppor. Bussiere and Mulder (1999), Eichengreen (2003), Ponines and Siregar (2008) have shown ha variables such as he curren accoun, expor growh, inernaional reserves and shor-erm inernaional deb are good indicaors o predic currency crises The balance of paymens as an inermediae arge of moneary policy BOP is an inermediae arge of moneary policy when he sabiliy of exchange rae regime is under he hrea. Sudies on his problem, based on he asserion of exisence of susainable level of curren accoun balance he level a which he counry is able o mee all is obligaions o foreign loans due o he curren and fuure savings. In he ligh of his assumpion, i is considered ha rade defici is a problem for moneary auhoriies, hreaening he sabiliy of moneary sphere, when he level exceeds susainable limi. Thus, for he U.S. i is abou 5% of GDP (Mann, 2002, Freund, 2000,2005) for New Zealand, Porugal, Singapore - 10% (Summers, 1996), France, Ialy, Spain - 3 % (Taylor, 2002). Change in he direcion of BOP dynamics is called he "reversal". Mos of he papers devoed o he reversal of BOP, sugges he effec of marke adjusmen mechanisms in he case of floaing exchange rae regimes (Obsfeld and Rogoff, 2005; Freund, 2000; Mann, 2002). In he case of counries wih inermediae exchange rae regimes reversal is conrolled by moneary auhoriies, 85

5 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 who are rying o correc he effec of marke facors. The effec of reversal hen arises in he form of: (a) delayed reacion of componens of BOP o he ongoing correcion of exchange rae, (b) smoohed response of he Cenral Bank o paymens imbalance. The imbalance is hreaens exchange rae regime in he counry, while he reversal is accompanied by depreciaion of naional currency. The overall effec of devaluaion of developed counries was esimaed o be 20% (Freund, 2005). Using he definiion of currency crisis proposed by Frankel and Rose (1996), he devaluaion of he naional currency by 25% or more in nominal erms is considered o be a currency crisis, respecively, we can say ha he imbalances in BOP can lead o currency crisis (Edwards, 2001). The hrea of crisis simulaes moneary auhoriies of inermediae exchange rae regimes o inervene in he process of reversal. There are sudies for developing counries, showing ha he effec of reversal does no necessarily enail a currency crisis. In Milesi-Ferrei and Razin (1998) less han a hird of cases in a sample of 105 counries sysemaically accompanied by a currency crisis. This effec shows ha under he same environmenal condiions in he counries in ransiion mode and floaing exchange rae devaluaion will occur a differen imes and a differen speeds. Moreover, in he shor erm, he join dynamics of curren accoun and real exchange rae may have a differen direcion due o correcions of moneary auhoriies The role of capial accoun in he analysis of he BOP Capial accoun of BOP comes o he fore during he problems wih liquidiy and insolvency. If moneary auhoriies prefer o influence he esablishmen of BOP, negaive shock of capial ouflows is associaed wih wo alernaives: a) he reducion of producion in response o decline in invesmen and b) he use of inernaional reserves o miigae he impac on domesic demand (Ranciere and Jeanne, 2006). Using, for example, ineres rae in order o regulae capial flows can lead o negaive consequences (Lahiri, Vegh, 2003; Pak-Hung, 2009) and only ighen he problems in he financial secor. High ineres raes in his case, firs, increase he demand for ineres-bearing liquid asses in he counry, and secondly, increase he cos of servicing he public deb. Thus, he effec is uncerain during he crisis, and is characerisic of counries wih inermediae exchange rae regimes. The relaionship beween financial variables looks ambiguous. For example, relaively small negaive shock (ineres rae change) may lead o radical changes in he dynamics of he capial accoun and have serious consequences for he social sphere (Calvo, 1996; Chang, 2009). Wih he example of he financial crisis in Mexico in 1994 Calvo showed ha in a world where inernaional relaions are well developed, he reacion of invesors o financial shocks can be disproporionaely high, a leas in he iniial response. The reason for his may be financial vulnerabiliy of he counry or expecaions of invesors. Financial facors in crisis and non-crisis period may have a compleely differen impac on capial accoun of BOP. This conclusion is mos ypical for counries proecing heir exchange rae regime. 86

6 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer Empirical analysis of he BOP effecs on moneary policy 3.1. The daa In his paper we invesigae he join dynamics of BOP and exchange rae of 78 counries which produce more han 93% of he World Gross Domesic Produc. We divide he enire sample ino wo groups. The firs group includes counries ha manage he exchange rae of he naional currency, and he second group involves counries ha do no conrol he exchange rae. Groups were formed on he basis of he Inernaional Moneary Fund (IMF) classificaion (Table 1). The firs group includes 40 counries wih a floaing exchange rae regime and wo counries wih currency board exchange raes regime. We add hese wo counries because in case of Currency Board he naional currency is pegged o free floaing currencies ha makes differences beween hese wo regimes irrelevan in his paricular sudy. We also rea counries of Euroarea as one observaion. The second group consiss of 38 counries wih inermediae exchange rae regimes (managed floaing, crawling peg and currency band). Daa for each counry cover he period from he hird quarer of 2006 (2006Q3) o he firs quarer of 2010 (2010Q1). Economeric esing is carried ou on quarerly daa forming he panel dimension in 78 counries and 15 quarers. The analysis is based on he dynamics of BOP, balance of rade, ineres raes, real effecive exchange raes. 87

7 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 Table 1. Classificaion of he sample counries according o he exchange rae regime Free floaing (38) Ausralia Brazil Canada Chile Czech Republic Euroarea Hungary Iceland Israel Mexico New Zealand Norway Philippines Poland Souh Africa Souh Korea Sweden Swizerland Turkey Unied Kingdom USA Japan Group I (40) Group II (38) Currency Convenional Crawling Board fixed peg peg (2) (17) (3) Bulgaria Hong Kong Argenina Belarus Belize Croaia Denmark Fiji Kazakhsan Lavia Lesoho Macedonia Morocco Russian Federaion Samoa Solomon Islands Trinidad and Tobago Tunisia Venezuela Bolivia China Nicaragua Noe: The number of counries included in each group is in parenheses. BOP saisics Currency band (1) Cosa Rica Managed floaing (17) Armenia Colombia Georgia India Indonesia Kyrgyzsan Malaysia Moldova Pakisan Paraguay Peru Romania Singapore Thailand Uganda Ukraine Uruguay We use he IMF quarerly saisics on capial, financial and curren accoun. I is imporan o noe ha he mehodology of compiling BOP have significan differences among counries we consider. For example, some counries do no separae financial accoun ino a paricular group. Therefore, we consider he sum of capial and financial accouns o avoid problems wih he comparabiliy of mehodologies. Trade Balance Quarerly rade balance daa are also aken from he IMF saisics. To fill he gap in case when here were no daa on rade balance, we use he following calculaions: 88

8 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 TB i, EX i, i,, (1) i, er IM where TB, quarerly rade balance of counry i in period (millions US dollars), i EX i, IM i, quarerly expor (impor) of counry i in naional currency, er i, average exchange rae for he quarer. Ineres rae Proxy for he ineres rae is he annual discoun rae a he end of he quarer. In he absence of discoun rae for any paricular counry, we use refinancing rae (percen per annum); in he absence of refinancing rae, we use repo rae (percen per annum). Real effecive exchange rae For counries where his indicaor was no availiable in he IMF daabase, we calculae he rae on he basis of counry s impor srucure. For calculaions we use he srucure of impor ino he counry in We consider his srucure o be consan during he period under consideraion. We calculae he real exchange rae of domesic currency and he currencies of counriesimporers: NEER REER * where P, (2) P REER real effecive exchange rae (domesic currency agains he uni of foreign currency), NEER N e j 1 j, * level in he domesic counry, P N CPI * j N j 1 j * j P CPI nominal effecive exchange rae, price j 1 j, * j average price level of imporers; j share of he imporer j based on he impor srucure; e j, nominal exchange rae (domesic currency agains he uni of foreign currency). Economeric es requires saionariy of he considered series. Dickey-Fuller es indicae he presence of a uni roo in all of he series, which demonsraes ha hey are no saionary. In addiion, he graphs of he dynamics of indices also show signs of he presence of a uni roo - here is no reurn o he average level. Tha is he reason why we use firs differences of he daa (quarerly changes). Descripion and symbols of all he variables which we use in he economeric analysis are described in Table 2. 89

9 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 Table 2. Variables descipion Idenificaion of he variable Descripion of he variable TB Quarerly change in rade balance of he counry (KA+FA ) Quarerly change in capial and financial accoun of he counry CA Quarerly change in curren accoun of he counry R Change in ineres rae, percenage poins REER Logarihm of changes in he real effecive exchange rae GDPj Gross domesic produc of counry j in Idenificaion of he crisis period One of he main issues of his paper is o describe feaures of moneary policy in counries wih differen exchange rae regimes. Special aenion is paid o he period of crisis. To do his we firs need o deermine ime bounds of he crisis when i sared and when i ended. In order o deec sar and end poins of crisis period, le us suppose ha hese poins refer o srucural breaks in relaions beween key macroeconomic variables and here were no oher srucural breaks. Then we need o break he sample ino hree homogeneous periods, when hese relaions sayed he same period before crisis, crisis period and period afer crisis. We use he idea of Bai-Perron es o pick hese periods ou. The idea is simple one jus needs o look over all possible ways of breaking ino periods, run regression and choose he way which gives he highes likelihood. We run his procedure for several specificaions of regression and wih differen variables. All he specificaions give he similar daes of he crisis i sared in he fourh quarer of 2008 and ended in he second quarer of The corresponding subsamples we use for inference are presened in Table 3. Table 3. Subsamples of homogeneous periods Name Daes Lengh of he period 2006 Q Q3 9 quarers 2008 Q Q2 3 quarers Afer 2009 Q Q1 3 quarers A quesion arises if hese periods are ruly characerized by differen join dynamics of key macroeconomic variables. We run Wald es and Likelihood Raio (LR) es o check his ou. For simpliciy we presen es saisics only for he rade balance regression: 90

10 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 TB Cons. where reer REER TB TB 1, (3) TB is a quarer rade balance change, adjused o he GDP, REER is a logarihm of he real effecive exchange rae growh, TB 1 is lagged value of rade balance, adjused o GDP, reer and TB are parameers of ineres, is macroeconomic shock in period. Since oher specificaions give he similar es saisics, presened resuls are robus o model specificaion. Table 4. Tes saics for srucural break ess (Н 0 : no srucural breaks) s o compare Wald sa. LR sa. vs. vs. Afer Afer vs. Noe: p-values in parenheses (0.000) (0.004) (0.000) (0.000) (0.006) (0.000) The null hypohesis of he srucural break ess saes ha coefficiens in equaion (3) are he same for wo periods. The null hypoheses are rejeced for all reasonable significance levels (Table 3). Thus he dynamics of variables differ for hese hree periods and we canno join hem o esimae parameers of ineres. However criical values of he ess are underesimaed since we choose crisis period by maximizing likelihood. In addiion, es saisics for he srucural break beween non-crisis periods are low enough. This resul may say in favor of he hypohesis ha parameers are he same for non-crisis periods. Tha is why furher, we also presen resuls for non-crisis periods, which consis of period before crisis and period afer crisis. Average values of key variables are presened in Table 5. The comparison of non-crisis and crisis periods saes ha here was a significan capial ouflow in a period of crisis. On average rade balance change was negaive in non-crisis period, and posiive in crisis period. Real effecive exchange rae came down in crises and rose in non-crisis period. I is ineresing o noe ha afer he crisis, real effecive exchange rae growh differ for before crisis and afer crisis periods on average i was 0.58% and 0.81% respecively. Ineres raes dynamics were negaive (he average decrease is 0.56% per quarer), while in non-crisis period, heir change was close o zero. 91

11 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 Table 5. Average values of variables (all counries) Variable Afer Non- All s TB KA+ FA CA R (%) REER (%) Average values of variables for wo groups of counries wih floaing exchange rae regime and wih inermediae exchange rae regime are presened in Tables 6 and 7 respecively. The average dynamics of rade balance, as well as he dynamics of curren accoun, are almos he same for hese wo groups, irrespecively of he period we consider. The disinguishing feaure of mos counries wih one of inermediae exchange rae regimes is is high dependence on expor. Tables 6 and 7 indicae high capial ouflow in crisis period and especially for counries wih floaing exchange rae regime. This fac says in favor of hypohesis ha reacion o shocks is higher for counries wih floaing regime, han for counries wih inermediae regime. The dynamics of real effecive exchange rae for hese wo groups of counries are differen as well. In pre-crisis period, counries wih inermediae regime demonsrae higher growh of he exchange rae 0.99% while for counries wih floaing regime average growh were 0.51%. In crisis, exchange rae growh fell down o -0.43% and -1.91% for inermediae and floaing regimes respecively. In pos-crisis period, exchange rae dynamics were resored for counries wih floaing regime is growh came o 1.66%. A he same ime, in counries wih inermediae regime, exchange rae coninued o fall down by 0.09% per quarer. Table 6. Average values of variables (counries wih floaing exchange rae regime) Variable Afer Non- All s TB KA+ FA CA R (%) REER (%)

12 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 Table 7. Average values of variables (counries wih inermediae exchange rae regime) Variable Afer Non- All s TB KA+ FA CA R (%) REER (%) Analyzing he dynamics of ineres raes, one may poin ou he endency o decrease for boh groups of counries. However, in crisis, he decrease of ineres rae came o 1.06% per quarer for counries wih floaing regime. This decrease is four imes as high as he average by all periods we consider. For counries wih inermediae regime, he decrease of ineres raes is no so dramaic. This may be explained by he fac ha he sample consiss of developing counries whose goal is o arac capial. Tha is why hey ry o avoid ineres raes decrease, which may negaively affec invesmen image of he counry Tesing he BOP Effecs To deec differences in macroeconomic relaions in crisis and non-crisis periods, we use simple linear models of rade balance and esimae hem wih OLS. Esimaion resuls help o explain rade balance reacion o shocks of expor, impor and capial accoun. We concenrae our aenion o wo inerrelaions. Firs, we esimae how exchange rae affecs he dynamics of curren accoun. And second, we esimae how ineres rae affecs he dynamics of capial accoun. Dependen variables in wo corresponding regressions are rade balance change ( TB) and a sum of capial and financial accouns changes (KA+FA) respecively. We include logarihm of real effecive exchange rae growh ( REER) as an explanaory variable in he firs regression and ineres rae change ( R) in he second regression. I is obvious ha economies of he counries in sample differ in scale. As a consequence, shocks boh of capial and curren accoun also differ in scale, or in oher words have differen variance. When esimaing equaions wih OLS, his leads o he problem of heeroskedasiciy. The problem of heeroskedasiciy is caused by differences in scales of economies and/or counries foreign rading aciviy. Tha is why o solve he problem and o obain more effecive esimaes, we use weighed ordinary leas squares. We use wo ways of weighing on GDP and on sum of expor and impor. Bu in he second case sandard errors become higher (esimaes are less precise). Tha is why we presen esimaes, obained by weighing all observaions on inverse o nominal GDP of Trade balance and Exchange Rae To reveal he influence of exchange rae on rade balance, we esimae he following linear model: TB Cons. reer REER TB TB 1. (5) 93

13 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 The model has an auoregressive componen rade balance of he previous quarer ha is needed o ake he so-called reversal effec ino accoun. The reversal effec is a propery of BOP dynamics ha consiss in is reversion o zero (or some average) level when i goes oo far from his level. If his effec exiss, we will see a negaive esimae of he parameer. A problem may arise if he rade balance series are no saionary. Formal ess sae ha rade balance series have uni roo. Bu his resul may be caused by a shor lengh of he series ha do no allow Dickey-Fuller es o rejec null hypohesis. In his paper we consider rade balance series as saionary. Panel characer of he daa implies he exisence of individual effecs of counries. In oher words, we may assume ha each counry has is own saionary level of rade balance ha may differ from oher counries. In ha case, aking firs difference removes hese individual effecs from he regression equaion, and he models we esimae are equal o he models wih individual effecs of rade balance. We do no presen esimaes of individual effecs since hey are no informaive in he conex of his research. Tables 8, 9 and 10 presen esimaion resuls for he whole sample and for wo groups of counries separaely. These resuls allow us o say ha: a) he whole sample resuls confirm he hypohesis ha exchange rae affecs rade balance posiively; b) for counries wih inermediae exchange rae regime, he esimae of coefficien a real effecive exchange rae is significanly less han zero in crisis period, bu larger han zero for noncrisis period; c) for counries wih floaing exchange rae regime, exchange rae affecs rade balance posiively for all periods. TB Table 8. Esimaion resuls, rade balance regression (all counries) Cons. ΔREER TB -1 Dependen Variable: ΔTB, Weighs: 1/GDP 2009 Afer Non- All s * *** [0.0003] [0.0006] [0.0003] [0.0008] [0.0003] ** *** *** ** *** [0.0169] [0.0222] [0.0129] [0.0137] [0.0090] *** *** *** [0.0088] [0.0205] [0.0079] [0.0208] [0.0082] R Log likelihood Number of observaions Noe: *, **, *** significance a 10%, 5% and 1% levels respecively. Sandard errors in square brackes. 94

14 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 The esimaes allow us o conclude ha for counries wih inermediae regime, changes in rade balance may push Cenral Bank o correc exchange rae. I is worh noing ha daa suppor he hypohesis of reversal effec of BOP esimaes of he coefficien a lagged rade balance is significanly negaive in boh crisis and non-crisis periods. Thus, if rade balance goes oo far from is average level, we may expec is reversal dynamics. And he more his deviaion is, he more likely reversal will happen. The fac ha he esimaes are insignifican for some subsamples and some periods (pre-crisis and pos-crisis) may be caused by small number of observaions ha leads o high sandard errors. Table 9. Esimaion resuls, rade balance regression (counries wih inermediae regime) Dependen Variable: ΔTB, Weighs: 1/GDP 2009 Afer Non- All s Cons * * *** [0.0017] [0.0029] [0.0015] [0.0037] [0.0015] ΔREER *** * *** *** [0.0627] [0.0782] [0.0446] [0.0511] [0.0359] TB *** *** [0.0299] [0.0614] [0.0286] [0.0492] [0.0266] R Log likelihood Number of observaions Noe: *, **, *** significance a 10%, 5% and 1% levels respecively. Sandard errors in square brackes. 95

15 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 Table 10. Esimaion resuls, rade balance regression (counries wih floaing regime) Dependen Variable: ΔTB, Weighs: 1/GDP 2009 Afer Non- All s Cons *** *** [0.0005] [0.0005] [0.0004] [0.0011] [0.0004] ΔREER *** *** *** [0.0211] [0.0174] [0.0150] [0.0159] [0.0110] TB ** *** *** *** [0.0124] [0.0186] [0.0103] [0.0330] [0.0118] R Log likelihood Number of observaions Noe: *, **, *** significance a 10%, 5% and 1% levels respecively. Sandard errors in square brackes. When analyzing hese resuls, one may sugges a hypohesis ha differen reacion of rade balance o exchange rae may be caused by differen shocks (posiive or negaive), which came o hese wo groups of counries. In oher words, if counries wih inermediae exchange rae regime suffered from large posiive shocks of rade balance and counries wih floaing exchange rae from large negaive shocks, we would expec he resuls, obained above. In ha case he negaive impac of exchange rae is no necessary. In order o verify his hypohesis, we include dummy variable for exchange rae regime floaing or inermediae and esimae he equaion on he whole sample. If he reacion o exchange rae in hese wo groups of counries is he same and differences are jus a resul of dissimilar shocks, hen we may expec he significance of dummy variable. In his regression he coefficien a exchange rae regime denoes he difference in average values of shocks beween counries wih floaing regime and wih inermediae regime. Zero value of his coefficien is consisen wih he hypohesis ha average shocks are he same and moneary policy were ruly differen for hese wo groups of counries. In oher words Cenral Banks reac differenly o he same shocks, depending on exchange rae regime. The equaion we esimae is: TB Cons. reer REER fl floaer, (6) where floaer is a dummy variable ha akes zero value for counries wih inermediae regime and uniy for counries wih floaing regime. In he period of crisis, exchange rae regime is insignifican. Thus he hypohesis abou difference in average values of shocks is rejeced and shocks are no he reason why reacion o exchange rae is differen (Table 11). 96

16 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 Table 11. Esimaion resuls, rade balance regression wih dummy for exchange rae regime Dependen Variable: ΔTB, Weighs: 1/GDP 2009 Afer Non- All s Cons ** *** ** [0.0012] [0.0021] [0.0010] [0.0033] [0.0011] ΔREER *** *** *** *** [0.0156] [0.0222] [0.0125] [0.0159] [0.0093] Floaer *** *** *** [0.0012] [0.0022] [0.0011] [0.0034] [0.0012] R Log likelihood Number of observaions Noe: *, **, *** significance a 10%, 5% and 1% levels respecively. Sandard errors in square brackes. Capial Accoun and Ineres Rae To analyze he relaion beween ineres rae and capial/financial accoun, we esimae following equaion: ( KA FA) Cons. R ( KA FA). (7) 1 Esimaes, obained on he whole sample, suppor he hypohesis ha Cenral Banks use ineres rae as policy insrumen in boh crisis and non-crisis periods. Moreover in he period of crisis, he esimae of he coefficien a ineres rae is wice as high as in non-crisis period i equals and respecively (Table 12). When considering counries by exchange rae regime, esimaes says abou posiive relaion for counries wih inermediae regime (Table 13), bu do no reveal any dependence for counries wih floaing regime (Table 14). 97

17 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 Table 12. Esimaion resuls, capial accoun regression (all counries) Dependen Variable: Δ(KA+FA), Weighs: 1/GDP 2009 Afer Non- All s Cons * *** [0.0003] [0.0006] [0.0003] [0.0008] [0.0003] ** *** *** ** *** ΔR [0.0169] [0.0222] [0.0129] [0.0137] [0.0090] Δ(KA+FA) *** *** *** [0.0088] [0.0205] [0.0079] [0.0208] [0.0082] R Log likelihood Number of observaions Noe: *, **, *** significance a 10%, 5% and 1% levels respecively. Sandard errors in square brackes. The esimaes allow us o conclude ha for counries wih inermediae exchange rae regime, a shock of capial accoun may be a signal for Cenral Bank o correc ineres rae in economy. A he same ime, for counries wih floaing exchange rae regime here is no such a phenomenon. In oher words, he hypohesis abou ineres rae o be a policy insrumen in crisis period is no suppored by he daa. Table 13. Esimaion resuls, capial accoun regression (counries wih inermediae regime) Dependen Variable: Δ(KA+FA), Weighs: 1/GDP 2009 Afer Non- All s Cons ** *** *** [0.0044] [0.0071] [0.0035] [0.0080] [0.0032] *** *** ΔR [0.0130] [0.0171] [0.0105] [0.0095] [0.0063] Δ(KA+FA) ** *** *** *** [0.0702] [0.1667] [0.0632] [0.0810] [0.0489] R Log likelihood Number of observaions Noe: *, **, *** significance a 10%, 5% and 1% levels respecively. Sandard errors in square brackes. 98

18 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 Table 14. Esimaion resuls, capial accoun regression (counries wih floaing regime) Dependen Variable: Δ(KA+FA), Weighs: 1/GDP 2009 Afer Non- All s Cons *** *** [0.0013] [0.0023] [0.0012] [0.0039] [0.0012] *** ** ΔR [0.0022] [0.0177] [0.0023] [0.0033] [0.0017] Δ(KA+FA) *** *** *** *** *** [0.0725] [0.1230] [0.0658] [0.1335] [0.0587] R Log likelihood Number of observaions Noe: *, **, *** significance a 10%, 5% and 1% levels respecively. Sandard errors in square brackes. Here we also check he hypohesis abou differen shocks of capial accoun for counries wih differen exchange rae regimes. As we do above for rade balance regression, we esimae an equaion wih dummy variable for exchange rae regime: ( KA FA) Cons. R floaer (8) Esimaion resuls are in Table 15. Table 15. Esimaion resuls, capial accoun regression wih dummy for exchange rae regime Dependen Variable: Δ(KA+FA), Weighs: 1/GDP 2009 Afer Non- All s Cons *** [0.0041] [0.0069] [0.0035] [0.0079] [0.0033] ΔR *** [0.0020] [0.0129] [0.0020] [0.0027] [0.0014] Floaer *** [0.0043] [0.0072] [0.0037] [0.0082] [0.0034] R Log likelihood Number of observaions Noe: *, **, *** significance a 10%, 5% and 1% levels respecively. Sandard errors in square brackes. 99

19 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 In crisis period, dummy for exchange rae regime is insignifican. Thus, differen shocks are no a reason why reacion o ineres rae was differen. I suppors he opinion ha Cenral Bank reacion o incoming shocks was differen and sabilizaion policy was differen depending on exchange rae regime. 4. Concluding Remarks This paper invesigaes BOP effecs in he period of crisis of Inference is based on wo subsamples. Firs subsample consiss of 40 counries wih floaing exchange rae regime. Second subsample consiss of 38 counries wih inermediae exchange rae regimes. The period of crisis from IV quarer 2008 o II quarer of 2009 is chosen as period wih unusual dynamics of invesigaed variables. Esimaion resuls allow us o ell abou differences in Cenral Bank policy for counries wih differen exchange rae regimes. For counries wih one of inermediae regimes, daa suppor he hypohesis abou negaive relaion beween rade balance and real effecive exchange rae in he period of crisis. Afer crisis, his negaive relaion became even clearer. A he same ime, for counries wih floaing exchange rae, his relaion sayed posiive. The hypohesis of reversal effec is suppored for boh groups of counries if rade balance goes oo far from is average level, we may expec is reversal dynamics. The resuls for capial and financial accoun confirm he main conclusion Cenral Bank policy is highly influenced by exchange rae regime of he counry. In crisis, saisically significan relaion beween capial accoun and ineres rae is observed only for counries wih inermediae exchange rae regime. References Bahmani-Oskooee, M. (1986) Deerminans of Inernaional Trade Flows: he Case of Developing Counries // Journal of Developmen Economics, 20, Bahmani-Oskooee, M., Kara, O. (2003) Relaive Responsiveness of Trade Flows o a Change in Prices and Exchange Rae // Inernaional Review of Applied Economics, 17(3), Bussiere, M., & Mulder, C. (1999) Exernal Vulnerabiliy in Emerging Marke Economies: How High Liquidiy Can Offse Weak Fundamenals and he Effecs of Conagion // IMF Working Paper 88. Boyd, D., Caporale, G.M., Smih, R. (2001) Real Exchange Rae Effecs on he Balance of Trade: Coinegraion and he Marshall Lerner Condiion // Inernaional Journal of Finance and Economics, 6, Calvo, G. A. (1996) Capial Flows and Macroeconomic Managemen: Tequila Lessons // Inernaional Journal of Finance Economics, 1, Chung K., Turnovsky S. (2009.) Foreign Deb Supply in an Imperfec Inernaional Capial Marke: Theory and Evidence. // Journal of Inernaional Money and Finance, 29, Edwards S. (2001) Does he Curren Accoun Maer? // Naional Bureau of Economic Research, Inc. WP 8275 Eichengreen, B. (2003). Capial Flows and Crises // Cambridge, Massachuses: The MIT Press. Frankel J., Rose A. (1996) Currency Crashes in Emerging Markes: An Empirical Treamen // Journal of Inernaional Economics, 41(3 4),

20 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 Freund, C.L. (2005) Curren Accoun Adjusmens in Indusrialized Counries // Journal of Inernaional Money and Finance, 24 (8), Galí J., Monacelli T. (2008) Opimal Moneary and Fiscal Policy in a Currency Union // Journal of Inernaional Economics, 76(1), Gomes F., Paz L. (2005) Can Real Exchange Rae Devaluaion Improve he Trade Balance? The Brazilian case // Applied Economics Leers, 12, Hacker, R., Haemi-J, A. (2004) The Effec of Exchange Rae Changes on Trade Balances in he Shor and Long Run: Evidence from German Trade wih Transiional Cenral European Economies. // Economics of Transiion, 12(4), Junz, H., Rhomberg R. R. (1973) Price Compeiiveness in Expor Trade Among Indusrial Counries // American Economic Review, Papers and Proceedings, 63, Kharel R., Marin C., Milas C. (2010) The Complex Response of Moneary Policy o he Exchange Rae // Scoish Journal of Poliical Economy, 57, Lahiri A., Vegh C. (2003) Delaying he Ineviable: Ineres Rae Defense and BOP Crises // Journal of Poliical Economy, 111(2), Lee, J., Chinn, M. D. (1998) The Curren accoun and he Real Exchange Rae: a Srucural VAR Analysis of Major Currencies // Naional Bureau of Economic Research, Inc. WP6495. Mann, C.L. (2002) Perspecives on he US Curren Accoun Defici and Susainabiliy // Journal of Economic Perspecives, 16 (3), Milesi-Ferrei, G.M., Razin A. (1998) Curren Accoun Reversals and Currency Crises: Empirical Regulariies // IMF Working Papers 98/89. Muller-Planenberg N. (2010) BOP Accouning and Exchange Rae Dynamics // Inernaional Review of Economics and Finance, 19, Obsfeld, M., Rogoff, K. (2005) The Unsusainable US Curren Accoun Posiion Revisied // Universiy of California Berkeley, Mimeo, WP Orcu, G. H. (1950) Measuremen of Price Elasiciies in Inernaional Trade // Review of Economics and Saisics, May, Ponines, V., Siregar, R. (2008) Fundamenal Pifalls of Exchange Marke Pressure-based Approaches o Idenificaion of Currency Crises // Inernaional Review of Economics and Finance, 17(3), Pak-Hung, M. (2009) Impossible Triniy, Capial Flow Marke and Financial Sabiliy // Kyklos, Blackwell Publishing, 62(4), Ranciere R., Jeanne O. (2006) The Opimal Level of Inernaional Reserves for Emerging Marke Counries: Formulas and Applicaions // IMF Working Papers 06/229. Rose, A. K. (1991) The Role of Exchange Raes in a Popular Model of Inernaional Trade: Does he Marshall Lerner s Condiion Hold? // Journal of Inernaional Economics, 30, Summers, L. C. (1996) Volaile Capial Flows: Taming Their Impac on Lain America // Iner-American Developmen Bank and Johns Hopkins Universiy Press, Balimore, Svensson, L.E.O. (2001) Transparency and Credibiliy: Moneary Policy wih Unobservable Goals. // Inernaional Economic Review, 42 (2), Svensson, L.E.O. (2003) Commen on: The Fuure of Moneary Aggregaes in Moneary Policy Analysis // Journal of Moneary Economics, 50/5,

21 Irina Khvosova, Alexander Larin, Anna Novak, and Andrei Shulgin, The Macroheme Review 2(7), Winer 2013 Taylor, A.M. (2002) A Cenury of Curren Accoun Dynamics // Journal of Inernaional Money and Finance, 21 (6), Tegene, A. (1991) Trade Flows, Relaive Prices, and Effecive Exchange Raes: a VAR on Ehiopian Daa // Applied Economics, 23, Wilson, J.F., Takacs, W.E. (1979) Differenial Responses o Price and Exchange Rae Influences in he Foreign Trade of Seleced Indusrial Counries // Review of Economics and Saisics, 61(2),