Kiel Institute for World Economics

Transcription

1 Kiel Insiue for World Economics Duesernbrooker Weg Kiel (Germany) Kiel Working Paper No The Inegraion of Imperfec Financial Markes: Implicaions for Business Cycle Volailiy by Claudia M. Buch and Chrisian Pierdzioch April 2003 The responsibiliy for he conens of he working papers ress wih he auhors, no he Insiue. Since working papers are of a preliminary naure, i may be useful o conac he auhors of a paricular working paper abou resuls or caveas before referring o, or quoing, a paper. Any commens on working papers should be sen direcly o he auhors.

2 The Inegraion of Imperfec Financial Markes: Implicaions for Business Cycle Volailiy April 2003 Claudia M. Buch and Chrisian Pierdzioch Kiel Insiue for World Economics, Duesernbrooker Weg 120, Kiel, Germany Absrac During he las wo decades, he degree of openness of naional financial sysems has increased subsanially. A he same ime, asymmeries in informaion and oher financial marke fricions have remain prevalen. We sudy boh empirically and heoreically he implicaions of he opening up of naional financial sysems in he presence of financial marke fricions for business cycle volailiy. In our empirical analysis, we demonsrae ha sylised facs sugges ha counries wih more developed financial sysems have lower business cycle volailiy. Financial openness has no srong impac on business cycle volailiy, in conras. In our heoreical analysis, we use a dynamic general equilibrium model o sudy he implicaions of he opening up of naional financial markes and of financial marke fricions for business cycle volailiy. We find ha he implicaions of opening up naional financial markes for business cycle volailiy are largely unaffeced by he presence of financial marke fricions. Keywords: Business cycle volailiy; Financial fricions; Financial marke inegraion JEL classificaion: F31, F32, F36, F41, E44 Acknowledgmens Par of his paper was wrien during a research visi of Chrisian Pierdzioch a he Naional Bureau of Economic Research (NBER), Cambridge MA. The hospialiy of he NBER is graefully acknowledged. This paper is par of a research projec on Volailiy in he Global Economy: The Role of Financial Markes. The auhors hank he Friz-Thyssen Foundaion for financial suppor.

3 1 1 Moivaion During he las decades, inernaional financial markes have become more inegraed, and inernaional capial flows have grown rapidly. A he same ime, financial marke fricions in he form of informaion and ransacion coss remain prevalen, as evidenced, for insance, by recen debaes on weaknesses in corporae governance sysems. 1 In his environmen, inernaional financial marke inegraion may no only bring benefis by allowing for a beer risk sharing among households and a beer allocaion of capial across counries. Raher, financial marke inegraion may also increase business cycle volailiy by magnifying he effecs of exising disorions beleaguering naional financial markes. If inernaional financial marke inegraion leads o increased insabiliy in he form of greaer business cycle volailiy, policy makers may hus be emped o slow down he process of inernaional financial marke inegraion. Finding an answer o he quesion of how inernaional financial marke inegraion and financial marke fricions inerac in shaping business cycle volailiy is herefore imporan. In his paper, we seek o answer his quesion by sudying he ineracion beween inernaional financial marke inegraion, naional financial marke fricions, and business cycle volailiy in he framework of a New Open Economy Macroeconomics (NOEM) model of he ype recenly developed by Obsfeld and Rogoff (1995). The model we use in our analysis merges elemens of he lieraure emphasizing he role of financial marke inegraion for business cycle volailiy (Suherland 1996, Senay 1998) wih elemens of recen lieraure sressing he role of financial marke fricions and financial acceleraor 1 The financial crises of Asia in he laer 1990s, for insance, have been aribued o weak corporae governance sysems by many observers. Also, fraud in US companies has

4 2 effecs for macroeconomic flucuaions (Aghion e al. 1999, Bacchea and Caminal 2000, Faia 2001a, 2001b, Gilchris e al. 2002). The basic building blocks of our model are he same as hose in he NOEM model recenly developed by Faia (2001a). Faia has buil a financial acceleraor mechanism à la Bernanke e al. (2000) ino an oherwise sandard NOEM model. Her NOEM model combines many of he elemens ha have been sressed in recen lieraure o be imporan for he general equilibrium modeling of he implicaions of financial marke fricions for macroeconomic flucuaions. In addiion, he srucure of her model closely corresponds o oher recen conribuions analyzing he financial acceleraor in open economies (see, among ohers, Gilchris e al. 2002). Thus, using he model developed by Faia as a saring poin for our analysis guaranees ha we use a model wih many feaures ha are commonly used in he lieraure, implying ha he resuls we derive do no hinge upon uncommon and arbirary assumpions. Our model differs from he model developed by Faia (2001a) wih regard o he srucure of inernaional financial markes. Faia assumes ha inernaional financial markes are complee in an Arrow-Debreu sense and ha here are no coss of rading financial asses inernaionally. In conras, we model a bond economy wih incomplee financial markes as in Obsfeld and Rogoff (1996). Moreover, we follow Suherland (1996) and Senay (1998) in assuming ha households incur ransacion coss when underaking posiions in inernaional financial markes. This implies ha inernaionally raded domesic and foreign financial asses become imperfec subsiues. By varying he ransacion coss for underaking crossborder financial ransacions, we can analyze how he inegraion of inernaional financial markes affecs macroeconomic flucuaions in he presence of financial marke fricions. spurred debaes on reforms of corporae governance sysems. See Bech e al. (2002) for an exensive survey.

5 3 By combining he assumpions of incomplee financial marke inegraion and financial marke fricions, our model addresses wo key feaures of real-world financial markes. Firs, our model assumes ha inernaional bond markes are inegraed o a greaer degree han inernaional bank lending markes, which we consider o be perfecly segmened. Second, we focus on deb markes. This acknowledges he fac ha inernaional capial flows are dominaed by deb raher han equiy (see, e.g. Kraay e al. 2000). We calibrae our model o sudy how financial marke inegraion and financial marke fricions affec he impac of exogenous moneary policy and produciviy shocks on business cycle volailiy. We find ha he implicaions of financial marke inegraion for business cycle volailiy are largely unaffeced by he presence of financial marke fricions. We organize he remainder of his paper as follows. In Secion 2, we discuss some sylized facs of inernaional financial marke inegraion and presen empirical evidence on he degree of financial fricions on naional financial markes. Using cross-secion regressions, we sudy how financial marke inegraion and financial marke fricions affec business cycle volailiy. In Secion 3, we provide a non-echnical overview of he NOEM model we use o derive he resuls we repor in his paper. In Secion 4, we calibrae and simulae our NOEM model in order o analyze he implicaions for business cycle volailiy of he opening up of naional financial markes in he presence credi marke fricions. In Secion 4, we conclude and discuss he policy implicaions of our research. We provide echnical deails concerning he srucure of our model a he end of he paper (Technical Appendix). 2 Financial Inegraion and Financial Fricions: Empirical Evidence Because he model we presen in Secion 3 will focus on he impac of he inegraion of inernaional financial markes and of credi marke fricions on business cycle volailiy, his secion presens descripive saisics as well as regression-based evidence on hese variables.

6 4 2.1 Sylised Facs The heoreical model ha we use in Secion 3 makes cerain assumpions on he degree of inernaional inegraion of differen segmens of financial markes and on he imporance of financial fricions. More specifically, he model focuses on deb markes, and i assumes ha bond markes are inegraed o a greaer degree han bank lending markes. Banking markes are perfecly segmened across counries whereas rading in bonds is possible inernaionally, albei a cerain coss. Moreover, domesic credi markes are assumed o be characerized by asymmeries in informaion, which make he coss of exernal deb finance a funcion of firms ne worh. Before we urn o a more deailed analysis of he link beween hese financial marke srucures and business cycle volailiy, i is useful o review he relevan empirical evidence. We consider he degree of inernaional inegraion of differen financial marke segmens firs. Generally, barriers o inegraion can be eiher direc, aking he form of ourigh resricions on capial flows, or indirec, aking he form of ransacion and informaion coss. Because hese indirec barriers o inegraion are difficul o measure, he empirical analysis below will use informaion on he presence of capial conrols as proxies for he openness of financial sysems. We use a 0/1-dummy ha capures he fac wheher counries mainain conrols on cross-border capial flows. We measure capial conrols in he mid-1990s, using a daa se kindly provided by Gian-Maria Milesi-Ferrei. This daa se is based on he IMF s surveys of exchange rae resricions. The summary saisics provided in Table 1 show ha OECD counries have lower capial conrols han non-oecd counries. This confirms ha quie significan advances have been made owards increased financial inegraion, in paricular in OECD counries. Inser Table 1 abou here.

7 5 A he same ime, indirec barriers o inegraion in he form of ransacion and informaion coss differ considerably among differen financial marke segmens. While comprehensive and comparable indicaors on he degree of financial inegraion of individual marke segmens across counries are difficul o obain, he inroducion of he euro has spurred research ino he degree of inegraion of differen financial marke segmens across Europe. Hence, he evidence obained for European markes can serve as a case sudy for inegraion rends in OECD counries. Wih regard o he degree of financial inegraion in general, evidence suggess an increasing openness of European financial markes ha has been promoed by he Single Marke program in 1992 as well as by he inroducion of he euro in 1999 (Frazscher 2002, Lemmen 1998). A he same ime, he degree of inegraion of differen financial marke segmens differs considerably. Wih regard o banking markes, sudies find a general rend owards ineres rae convergence across Europe, in paricular for money markes, while reail ineres raes sill show relaively high degrees of dispersion across counries (Kleimeier and Sander 2000, Ceneno and Mello 1999). Essenially, he inerbank and he corporae bond marke in Europe show a relaively large degree of inegraion already while he collaerized money marke and equiy markes are sill largely naional in scope (BIS 2001). The model we use in Secion 3 no only assumes ha he degrees of inegraion of bond and banking markes differ, i also assumes ha credi marke fricions in he form of asymmeries in informaion beween firms and banks persis on a domesic level. Again, i is difficul o obain direc measures of he severiy of informaion problems. Therefore, our empirical model will use he acual imporance of credi markes in an economy, measured as he raio of credi graned by commercial banks and by oher financial insiuions over GDP, as a

8 6 proxy. We have obained hese daa from he daabase on financial srucures around he world, which has been compiled by he World Bank. 2 We again use daa for he mid-1990s. As evidenced by Table 1, credi markes end o be more imporan in OECD han in non- OECD counries. Similar cross-counries differences can be found for oher financial marke segmens. In he EU, for insance, he imporance of sock markes, measured in erms of marke capializaion over GDP, varies beween 184% in Luxembourg and 33% in Ialy (Giannei e al. 2002). 2.2 Regression Resuls To wha exen does he degree of financial openness and of financial marke fricions have an impac on business cycle volailiy across counries? Before we use our model o sudy his quesion, we provide a firs, regression-based answer o his quesion. To seup our regression-based analysis, we proceed as follows. We use he sandard deviaion of he growh of real GNP (measured in 1995 US-Dollars) as a proxy for business cycle volailiy. 3 Daa are available for a sample of 76 counries. To analyze he deerminans of business-cycle volailiy in our cross-secion of counries, we regress our measure of business cycle volailiy on he volailiy (i.e., sandard deviaion) of a number of conrol variables. To compue he sandard deviaions of he conrol variables, we use annual daa for he 1990s. We have obained he daa from he CD-Rom World Developmen Indicaors 2002 of he World Bank. In addiion o he volailiy of he main variables of ineres, i.e., he financial marke variables ha have been described above, we conrol for he volailiy of governmen spending 2 The daa can be downloaded from hp://econ.worldbank.org/programs/finance/opic/finsrucures/.

9 7 (sandard deviaion of he growh of general governmen final consumpion expendiure in consan (1995) US-Dollars), and he volailiy of moneary policy (coefficien of variaion, i.e., sandard deviaion divided by he mean, of lending raes). We also use he volailiy of real exchange raes as a proxy for produciviy shocks (sandard deviaion of he growh of he real effecive exchange raes index, 1990 = 100). Table 1 gives he summary saisics of he daa ha we use. Generally, business cycle volailiy and he volailiy of our conrol variables are higher in non-oecd han in OECD counries. This holds for he volailiy of GDP as well as for governmen spending. Ineresingly hough, he volailiy of ineres raes in OECD and non-oecd counries is fairly similar, he non-oecd counries being slighly more heerogeneous. Equipped wih hese conrol variables, we esimae an equaion of he form m conrol 0 + β i iσ cycle σ i = α + u =1 i, cycle where business cycle volailiy ( ) depends on a consan inercep erm, α, he volailiy σ i 0 conrol of he conrol variables ( σ ), and on a counry-specific sochasic disurbance erm, u. i, i Table 2 presens he regression resuls. To capure differences beween OECD and non- OECD counries in business cycle volailiy, we include a dummy variable which is se equal o one for OECD counries. The negaive and significan sign ha we obain for his dummy variable suggess ha economic developmen lowers business cycle volailiy. We obain qualiaively similar resuls when we use GDP per capia as an alernaive proxy for economic developmen (resuls no repored). However, in conras o he OECD-dummy, GDP per capia is less significan. 3 While i would be preferable o use more sophisicaed filering echniques such as Band- Pass- or HP-filers o isolae he cyclical componen of he daa, sample sizes have been oo small for he developing counries o implemen hese mehods.

10 8 Inser Table 2 abou here. Generally, higher volailiy of lending raes and of governmen spending increases business cycle volailiy. While volailiy of governmen spending is generally significan a leas a he 10 percen level, ineres rae volailiy becomes insignifican if credi over GDP is enered as an addiional conrol variable. Moreover, resuls in column (7) of Table 2 show ha higher volailiy of he real exchange raes, which we use as a proxy for erms of rade and, hus, produciviy shocks, has no significan impac on business cycle volailiy in our crosscounry regression. As regards he effecs of financial marke condiions on business cycle volailiy, our resuls sugges ha larger credi markes are associaed wih lower business cycle volailiy. However, his effec is no significan unless we ener he credi marke variable separaely for counries a differen saes of developmen (column 4). This specificaion shows ha a higher volume of credi over GDP is associaed wih lower volailiy of GDP in high-income counries only. Because, in his specificaion, he OECD dummy becomes insignifican, we canno isolae he effec of economic developmen in general from he effecs of he developmen of he financial sysem. Openness for foreign capial, as measured hrough he inensiy of capial conrols, has no significan impac on business cycle volailiy. This resul does no change if we spli up he sample by income level. Because work by, for insance, Aghion e al. (1999) suggess ha he link beween financial openness and business cycle volailiy depends on he degree of developmen of he domesic financial sysem, we addiionally include an ineracion erm beween he size of credi markes and openness (see column 6 of Table 2). This ineracion erm is insignifican hough. If we addiionally include he proxies for openness and developmen (no repored), all variables become insignifican, and here is subsanial mulicollineariy in he daa.

11 9 One ineresing quesion is wheher OECD and non-oecd counries differ only wih regard o he level of business cycle volailiy or also wih regard o he impac of he conrol variables we use. We use wo differen mehodologies o es for srucural differences beween hese groups of counries (resuls no repored). Firs, we inerac he OECD dummy wih he remaining regressors. Second, we spli he sample ino he wo sub-groups and esimae he equaions separaely. Boh mehods yield he same qualiaive resuls. The ineracion erms are insignifican. This suggess ha differences beween he OECD and he non-oecd counries do no work hrough he channels ha we conrol for in our regressions, i.e., hey are no sysemaically linked o differences in moneary and fiscal policy volailiy or differences in financial sysems. Moreover, some of he variables ha we find o be significan deerminans of business cycle volailiy in he full sample are insignifican in he sub-samples (OECD versus non-oecd). Hence, he resuls ha we obain parly sem from he heerogeneiy beween hese wo counry groups. To sum up, in regression-based empirical analysis we have esablished he following hree main sylized facs ha our heoreical model mus be compaible wih: Business cycle volailiy ends o be lower in more developed counries wih more developed domesic credi markes. There is no evidence ha business cycle volailiy is linked significanly o he openness of he financial sysem. The implicaions of financial marke openness and of credi marke fricions for business cycle volailiy end o be largely independen of each oher. 3 Overview of he Model We now lay ou a heoreical model ha helps explain he resuls of our regression-based empirical analysis repored in Secion 2. Here, we provide an overview of he srucure of he

12 10 model. The main building blocks of he model are commonly used in he open-economy macroeconomics and inernaional finance lieraure, so our discussion can be relaively brief. We give deails concerning he mahemaical specificaion of he various building blocks of he model a he end of he paper (Technical Appendix). 3.1 The Household and Producion Secors Our model is a dynamic sochasic general equilibrium New Open Economy Macroeconmic (NOEM) model. The basic srucure of our model is idenical o he srucure of he model developed by Obsfeld and Rogoff (1996). As in heir model, he world is made up of wo counries. The counries are of equal size. Each counry is inhabied by infiniely-lived idenical households. Households form raional expecaions and maximize heir expeced lifeime uiliy. Households save, inves in domesic and foreign nominal one-period bonds, make deposis a domesic financial inermediaries, and supply labor. Hence, wih respec o he asse allocaion choices of households, we depar from he workhorse NOEM model by allowing for he possibiliy ha households hold deposis in addiion o bonds. Also noe ha, while households may inves in eiher domesic or foreign bonds, hey can only hold deposis wih domesic financial inermediaries. As argued in Secion 2, his assumpion is consisen wih he sylized facs on he differen degrees of inernaional inegraion of financial marke segmens. In addiion, each counry feaures a producion secor. As in Bernanke e al. (2000) and Faia (2001a), he producion secor is made up of hree ypes of firms: enrepreneurs, capial producers, and reailers. The enrepreneurs ac in a compeiive environmen. Using a consan-reurns-o-scale producion funcion, hey combine he labor supplied by households wih physical capial o produce a wholesale good. The enrepreneurs hire labor in a perfecly compeiive labor marke. They buy physical capial from capial producers who ac in a

13 11 compeiive environmen as well. Capial producers, in urn, use a producion echnology ha embeds adjusmen coss o produce capial goods (see Kiyoaki and Moore, 1997). The assumpion ha he producion of capial goods involves adjusmen coss adds variabiliy o he real of price of capial (Tobins Q). This, in urn, conribues o he variabiliy of he ne worh of enrepreneurs and, as discussed below, reinforces he effecs of he financial fricion on macroeconomic flucuaions and, hereby, gives rise o a so-called financial acceleraor mechanism. Reailers buy he wholesale good and differeniae i a no coss. They sell he differeniaed good hey produce in a monopolisically compeiive goods marke. The goods produced by reailers can be used as consumpion goods and as invesmen goods. Each reailer has monopoly power on he marke for is differeniaed produc. The reailers, herefore, rea he prices hey charge for heir producs as a choice variable. In consequence, one has o specify a price seing mechanism. We follow McCallum and Nelson (2000) and Bernanke and Gerler (1999) in assuming ha reailers behave according o a price-seing mechanism similar o he one inroduced by Fuhrer and Moore (1995). This price-seing mechanism renders i possible ha firms combine forward-looking and backward-looking elemens when seing he prices of heir producs. Such a behavior is in line wih recen empirical evidence (Gali e al. 2001). 3.2 The Domesic Credi Marke Following he lieraure, we buil a financial acceleraor ino he model by assuming ha he risk-neural enrepreneurs finance he acquisiion of physical capial by borrowing from a compeiive domesic financial inermediaion secor. The financial inermediaries ge he money hey need o finance hese loans by collecing deposis from domesic households. Due o asymmeric informaion problems, financial inermediaries do no supply loans a he riskfree nominal ineres rae. Raher, enrepreneurs encouner an exernal finance premium (defined as he coss of exernal funds minus he opporuniy coss of inernal funds) when

14 12 borrowing from financial inermediaries. The exernal finance premium makes uncollaeralized exernal finance more expensive hen inernal finance and, hereby, affecs he invesmen decisions of firms. To moivae he exisence of he exernal finance premium as a reflecion of credi marke fricions, we follow Bernanke e al. (2000). In heir model, he credi marke fricion arises in a world in which reurns on invesmen are sochasic and are a funcion of boh idiosyncraic and aggregae risk. The core credi marke fricion ha gives rise o he exernal finance premium is ha he financial inermediaries, in conras o enrepreneurs, can only observe he aggregae shock. This implies ha, in he case of defaul of he enrepreneur, he financial inermediaries mus pay a fixed audiing cos in order o observe he realized reurns of he enrepreneur. In consequence, an agency problem arises due o he exisence of a cosly sae verificaion problem as in Townsend (1979). The opporuniy cos of he financial inermediaries when supplying loans o firms is he risk-less ineres rae. The reason is ha, in equilibrium, he financial inermediaries can diversify away all idiosyncraic risk of lending o enrepreneurs by holding a perfecly diversified porfolio. Because households are risk averse and enrepreneurs are risk neural, he laer bear he enire aggregae risk of heir business. The exisence of aggregae risk implies ha he loan rae enrepreneurs mus pay when borrowing from financial inermediaries is a funcion of he expeced reurn on capial and, hus, of macroeconomic condiions. The reason for his is ha he lower is he realized aggregae shock o he reurn on capial he higher is he defaul probabiliy of enrepreneurs and, as a resul, he higher is he realizaion of he idiosyncraic shock required such ha he enrepreneur is able o repay he loan. Because enrepreneurs always have o offer he financial inermediaries a conrac such ha he expeced reurn on he loan is equal o he risk-free ineres rae, a higher required realizaion of he idiosyncraic shock implies ha he exernal finance premium and, hus, he

15 13 coss of exernal finance increase. If enrepreneurs face a negaive aggregae shock and heir balance shee worsens, hey mus compensae financial inermediaries for he increased probabiliy of defaul. Because he enrepreneur balances in equilibrium he reurns o capial and he marginal cos of exernal finance, his implies ha, when choosing he amoun of invesmen, he enrepreneur is consrained by he exisence of he higher exernal finance premium. The exernal finance premium is inversely linked o he srengh of he ne worh of enrepreneurs. This assumpion inroduces he financial acceleraor ino he model. The ne worh of enrepreneurs is defined as he sum of heir liquid asses and he collaeral value of illiquid asses minus ousanding obligaions. Because enrepreneurs ne worh is procyclical, he exernal finance premium is couner-cyclical. Due o he couner-cyclical behavior of he exernal finance premium, he credi marke fricion magnifies cyclical flucuaions. This is wha Bernanke e al. (2000) call he financial acceleraor mechanism. Finally, o preven enrepreneurs from accumulaing enough wealh o become fully selffinancing, he assumpion ha enrepreneurs have finie lives is needed. Thus, in every period a cerain number of enrepreneurs leave he model and new enrepreneurs ener he marke for he producion of he wholesale good. 3.3 Inernaional Financial Markes The markes for deposis and loans are no he only financial markes in he model. In addiion o he deposi and loans markes, which are perfecly segmened across counries, here is also a marke for domesic and foreign nominal one-period bonds. Thus, households allocae heir financial wealh across domesic deposis and domesic bonds as well as foreign bonds. While domesic bonds and deposis are perfec subsiues, domesic and foreign bonds are imperfec subsiues.

16 14 We inroduce his feaure ino he model by assuming ha households incur ransacion coss when underaking posiions in he inernaional bond marke. Thus, when choosing he opimal allocaion of heir wealh, households have o ake ino accoun ha inernaional bond markes are no perfecly inegraed. Whereas domesic (foreign) households have free access o he home (foreign) bond marke, hey incur ransacion coss when underaking posiions in he inernaional bond marke. To model he ransacion coss for underaking posiions in he inernaional bond marke, we follow Suherland (1996), Senay (1998), and Benigno (2000) in assuming ha he real ransacion coss for cross-border capial movemens are a convex funcion of he flow of funds ransferred from he domesic o he foreign bond marke. The inroducion of he assumpion ha domesic and foreign bonds are imperfec subsiues implies ha he no-arbirage condiion of uncovered ineres rae pariy does no hold anymore in is mos basic form. In is basic form, he condiion of uncovered ineres rae pariy sipulaes ha, in a fricionless economy in which domesic and foreign ineres bearing securiies are perfec subsiues, he ineres accrued from holding domesic securiies mus be idenical o he expeced reurn from holding foreign securiies. The expeced reurn from holding foreign securiies is given by he sum of he yield on foreign securiies and he expeced rae of change of he nominal exchange rae. In our model, his basic version of he condiion of uncovered ineres rae pariy does no hold. Raher, a modified version of his no-arbirage condiion for inernaional bonds markes applies. The modificaion of he condiion of uncovered ineres rae pariy we have o make reflecs ha he ransacion coss for underaking posiions in he inernaional bond marke drive a wedge beween he reurns on domesic and he reurns on foreign bonds. Because he ransacion coss are a funcion of he flow of funds involved in cross-border financial ransacions (i.e., inernaional capial flows), he wedge beween he reurn on domesic and

17 15 he reurn on foreign bonds is a funcion of his variable as well. Only in he limiing case of no ransacions coss, in which inernaional bond markes become fully inegraed, does he basic form of he condiion of uncovered ineres rae pariy hold in our model. (See also he Appendix, Equaion (A.8).) 3.4 The Governmen Secor Finally, in order o close our model, we have o specify he governmen secor. We absrac from governmen purchases of consumpion goods. This implies ha he inegral of lump-sum ransfers aken over all households in he domesic and foreign economy, respecively, is zero. Wih respec o moneary policy, we assume ha he shor-erm ineres rae is he insrumen used by he cenral bank. To formalize his noion, we model he ineres-rae seing of he cenral bank by means of a simple cenral bank reacion funcion similar o he one used by Taylor (1993). The version of he Taylor rule we use has also been used by Bernanke e al. (2000) and Faia (2001). I sipulaes ha he cenral bank ses he nominal ineres rae in response o deviaions of inflaion from a arge level of inflaion. In addiion, he cenral bank reacion funcion we use capures he ineres-rae smoohing objecive of cenral banks (Goodfriend 1991). 4 Financial Marke Inegraion, Financial Marke Fricions, and Business Cycle Volailiy In his secion, we use our model o analyze how he inegraion of inernaional financial markes affecs business cycle volailiy in he presence of a financial acceleraor mechanism. In conras o he previous NOEM lieraure sudying he implicaions of financial marke inegraion for business cycle volailiy (Suherland 1996, Senay 1998), we ask wheher, and if so, how he effecs of exogenous shocks differ in a world wih financial inegraion from hose in a world wihou financial inegraion if we addiionally consider he

18 16 effecs of fricions in domesic financial markes ha give rise o a financial acceleraor mechanism. To answer his quesion, we simulae our model numerically. To code up our numerical simulaions, we ake hree seps. (1) We follow Obsfeld and Rogoff (1995) and many ohers and log-linearize he model around a symmeric monopolisic compeiion flexible-price seady sae equilibrium in which he domesic and foreign asse posiions are zero. (2) We hen calibrae he model. The calibraion of he model is given in Table 1 and closely follows Bernanke e al. (2000) and Faia (2001a) who use calibraed parameers ha are consisen wih U.S. and European daa. We ake he parameer describing he ransacions coss for aking posiion in inernaional bond markes from Suherland (1996) and Senay (1998). Using he parameer values ha are fairly sandard in he lieraure assures ha our resuls can be compared o hose repored in he previous lieraure. I also assures ha he resuls ha drop ou of our simulaions do no depend upon arbirary and empirically unreasonable numerical values for he srucural parameers of our model. (3) We use he soluion algorihm developed by Klein (2000) o solve our log-linearized model numerically. 4 The remainder of his secion comes in wo pars. In he firs par, we use impulse response funcions o discuss how financial marke inegraion affecs he way exogenous shocks propagae hrough he economy. To illusrae how he model works, we discuss in some deail how a moneary policy shock propagaes hrough he economy. We also analyze how financial marke inegraion affecs he macro-dynamic consequences of produciviy shocks. In he second par, we analyze in deail how he presence of he financial acceleraor 4 We use Paul Klein s algorihm solve.k in Malab o solve he model numerically. In our simulaions of he model, we neglec he variaion in erms ha have no percepible impac on dynamics (e.g., he erms capuring he consumpion of enrepreneurs). See also he discussion in Bernanke e al. (1999).

19 17 mechanism affecs he implicaions of he inegraion of financial markes for business cycle volailiy. 4.1 Financial Marke Inegraion and Macroeconomic Flucuaions To analyze how he inegraion of inernaional bond markes affecs he way in which exogenous shocks propagae hrough he economy, we plo in Figures 1 and 2 impulse response funcions o visualize he dynamic response of a number of key domesic variables o such exogenous shocks. The ime unis on he figures are quarers. The unanicipaed, emporary exogenous shocks we consider are moneary policy and produciviy shocks. For boh moneary policy and produciviy shocks we plo impulse response funcions for a version of our model in which inernaional capial mobiliy is low (dashed lines) and a version of our model in which inernaional capial mobiliy is high (solid lines). If capial mobiliy is low, he ransacion coss for underaking posiions in inernaional financial markes are relaively high. To analyze how he model works, consider he impulse response funcions ha summarize he macro-dynamic implicaions of a uni domesic moneary policy shock (Figure 1). The moneary policy shock implies ha he domesic cenral bank raises he shor-erm nominal ineres rae. Because he resul is a rise in he domesic real ineres rae, consumpion decreases. Also, he domesic moneary policy shock implies ha domesic bonds become more aracive relaive o foreign bonds and he nominal exchange rae appreciaes. Because he prices of he differeniaed goods produced in he reail secor adjus sluggishly, he nominal appreciaion resuls in an appreciaion of he domesic erms of rade, defined as he domesic currency price of foreign goods in erms of he home currency price of domesic goods. The appreciaion of he erms of rade, in urn, makes domesic goods more expensive relaive o foreign goods and leads, hereby, o a decline in home oupu and

20 18 o a deerioraion of he rade balance. The deerioraion of he rade balance implies ha he foreign asse posiion of he domesic economy sars declining. Inser Figure 1 abou here. Why does he degree of inernaional financial marke inegraion maer for he dynamics of he model in he afermah of a moneary policy shock? Wih inernaional financial markes being imperfecly inegraed, he impac of he moneary policy shock on he dynamics of he foreign asse posiion is direcly refleced in he condiion of uncovered ineres rae pariy. As deailed in he Appendix (see equaion (A.8)), his condiion sipulaes ha, a any poin in ime, he inernaional nominal ineres rae differenial is proporional o he sum of he expeced rae of change of he nominal exchange rae and he expeced rae of change of he cross-border flow of funds. This direc effec of he change in he foreign asse posiion on he inernaional nominal yield differenial does no arise in a world of perfec inernaional capial mobiliy. I follows from he dynamics of he rade balance discussed above ha he expeced rae of change of he cross-border flow of funds is posiive in he afermah of a moneary policy shock, i.e., he domesic counry sars exporing financial capial. To see his, noe ha he rade balance defici realized in he immediae afermah of he moneary policy shock gradually urns ino a surplus as he domesic currency sars depreciaing again. From his i immediaely follows ha he expeced rae of change of he cross-border flow of funds is posiive. This, in urn, implies ha, for any given ineres rae differenial, he expeced rae of depreciaion of he domesic currency mus be smaller wih segmened inernaional financial markes. As a resul, he iniial appreciaion of he Home currency is less pronounced when inernaional financial markes are segmened.

21 19 From his argumen i follows ha he oupu effec of he moneary policy shock is larger in he case of high capial mobiliy han in he case of low capial mobiliy. Thus, as in workhorse model of inernaional macroeconomiss, he by now classic Mundell-Fleming model (Fleming 1962, Mundell 1963), swiching from a world of low o a world of high inernaional capial mobiliy increases business cycle volailiy. This resul is in line wih he resuls derived in a similar NOEM model by Suherland (1996). The radiional Mundell-Fleming mechanism is no he only mechanism hrough which moneary policy shocks affec macroeconomic dynamics. A furher mechanism ha has o be aken ino consideraion when analyzing how a moneary policy shock propagaes hrough our model economy is he financial acceleraor mechanism. To see more clearly how he financial acceleraor works, i is useful o realize ha he increase in he nominal and he real ineres rae brough abou by he moneary policy shock deerioraes he financing condiions for invesmen and, hus, slows down capial accumulaion. As invesmen declines, he price of capial (Tobin s Q) declines as well. The decline in he price of capial worsens enrepeneurs balance shees and gives rise o an increase in he exernal finance premium. Any rise in he exernal finance premium increases enrepeneurs coss of loans and his, in urn, reinforces he decrease in he demand for invesmen goods. This spiral of a deerioraion of invesmen condiions, a decrease in he ne worh of enrepeneurs, and a rise in he exernal finance premium is wha is known in he lieraure as he financial acceleraor mechanism (see Bernanke e al. 2000). The financial acceleraor mechanism is also a work if a domesic produciviy shock his he economy. As revealed by Figure 2, a produciviy shock resuls in an oupu boom and, as refleced in he increase in consumpion, leads o a decline in he real ineres rae. The decline in he real ineres rae, in urn, spurs invesmen and resuls in an increase in he real price of

22 20 capial. This improves enrepreneurs balance shees and reduces he exernal finance premium, which, in urn, provides a furher simulus for invesmen. Inser Figure 2 abou here. I is also worh noing ha he effec of he produciviy shock on he real price of capial and on invesmen is less pronounced if capial mobiliy is high. 5 The reason is ha, if capial mobiliy is high, he produciviy shock exers a relaively srong effec on he nominal exchange rae and, hus, on he erms of rade. Movemens in he erms of rade, in urn, have a direc effec on he real value of he ne worh of enrepreneurs and, hus, on he exernal finance premium. From his argumen i follows ha he invesmen boom riggered by he produciviy shock mus be less srong if capial mobiliy is high. 4.2 Financial Marke Inegraion, he Financial Acceleraor, and Business Cycle Volailiy So far, our discussion has cenered on he quesion how he inegraion of inernaional financial markes may effec he propagaion of a given exogenous (moneary policy or produciviy) shock. In his discussion, we implicily assumed ha he parameer capuring he srengh of he financial acceleraor mechanism in credi markes (he elasiciy of he exernal finance premium wih respec o he leverage raio) is fixed a a cerain level. We now urn o he quesion how he inegraion of inernaional financial markes shapes he effecs of exogenous shocks under alernaive assumpions regarding he numerical value of he parameer capuring he prevalence of he financial acceleraor mechanism. This discussion allows racing ou wheher here are significan inerdependencies beween 5 Similar argumens apply in he case of a moneary policy shock.

23 21 financial marke inegraion and financial marke fricions in erms of heir effecs on business cycle volailiy. To sar our discussion, we plo in Figures 3 and 4 he ampliude of he impulse response funcions as observed during a ime inerval of weny periods following a uni domesic moneary policy shock and a uni domesic produciviy shock, respecively, on he verical axis. The elasiciy of he exernal finance premium wih respec o he leverage raio is ploed on he horizonal axis. 6 Thus, when moving from he lef o he righ on he horizonal axis, he severiy of he fricion in he credi marke increases. Inser Figures 3 and 4 abou here. Figures 3 and 4 show ha he magniude of he effecs of domesic moneary policy and produciviy shocks depends upon he degree of inernaional capial mobiliy and on he severiy of he fricion in he credi marke. For example, he figures reveal ha he impac of moneary policy and produciviy shocks on oupu is increasing in he severiy of he fricion in he credi marke as measured in erms of he elasiciy of he exernal finance premium wih respec o he leverage raio. This resul is in line wih he resuls of he regression-based analysis presened in Secion 2, which have shown ha business cycle volailiy ends o be an inverse funcion of he degree of developmen of he domesic credi marke. The figures also depic ha swiching from low o high capial mobiliy in general resuls in an increase in he effec of moneary and produciviy shocks on oupu and he nominal exchange rae. Thus, business cycle volailiy ends o be higher he higher is he degree of 6 We use he ampliude of he impulse response funcion raher han is variance o measure he magniude of he effecs of he exogenous shock because he foreign bond holdings of privae households are no saionary. I would be sraighforward o add o our model a modelling device ha would make he foreign asse posiion of privae households and, hus, he iniial seady sae around which he model is log-linearized saionary (Schmi- Grohe und Uribe 2002).

24 22 capial mobiliy. In his respec i is also worh noing from Figures 3 and 4, bu also from Figures 1 and 2, ha his effec ends o be relaively small in quaniaive erms. This resul is also in line wih he resul of he regression-based analysis presened in Secion 2 ha he link beween business cycle volailiy and financial openness ends o be no very srong. A furher resul depiced in Figures 3 and 4 is ha he changes in he magniude of he oupu effecs of moneary policy and produciviy shocks and, hus, changes in business cycle volailiy resuling from swiching from low o high capial mobiliy are largely independen of he severiy of he fricion in he credi marke. This resul follows from he fac ha in he case of oupu he dashed lines ploed in he figures more or less parallel he solid lines. Thus, in line wih he empirical evidence repored in Secion 2, he implicaions of financial marke inegraion and of financial marke fricions for business cycle volailiy are largely independen of each oher. 7 5 Concluding Remarks In his paper, we documened hree sylized fac concerning he linkage beween inernaional financial marke inegraion, financial marke fricions, and business cycle volailiy and have suggesed a heoreical model ha helps explaining hese sylized facs. The hree sylized facs we esablished are: (i) Empirically, he link beween business cycle volailiy and he openness of he financial sysem is no very srong. (ii) Business cycle volailiy ends o be lower in more developed counries wih more develop domesic credi markes. (iii) The implicaions of financial marke openness and of credi marke fricions for business cycle volailiy end o be largely independen of each oher. To explain hese facs, we used a 7 This resul indicaes ha he resuls repored by Suherland (1996) and Senay (1998), who do no consider he effecs of financial marke fricions when analysing he link beween

25 23 dynamic, sochasic NOEM model ha feaures credi marke fricions and ransacion coss for underaking posiions in inernaional bond markes. Our simulaions of he model revealed ha his model is broadly consisen wih he hree sylized facs our empirical analysis has revealed: financial openness has only a small impac on business cycle volailiy, a reducion of domesic financial marke fricions lowers business cycle volailiy, and he impac of financial developmen and financial openness is largely unrelaed. The resul of his paper ha he impac of financial opening on business cycle volailiy seems o be largely independen from he sae of developmen of he domesic financial sysem seems o be a odds boh wih recen heoreical lieraure and wih curren policy discussions. Recen heoreical work by Aghion e al. (1999), for insance, shows ha financial opening migh be harmful for counries a medium levels of financial marke developmen as i may aggravae insabiliies. Recen episodes of currency and banking secor crises seem o have shown ha hese ineracion effecs are in fac empirically imporan. In many insances, he severiy of currency crises has been linked o weaknesses in domesic financial sysems. These experiences have revived an old debae on he appropriae sequencing of inernal and exernal financial liberalizaion and, in fac, many observers conclude ha financial sysems should open up for foreign capial only if domesic financial sysems have been reformed sufficienly. The focus of his paper differed from he discussion on he appropriae sequencing of financial liberalizaion in wo main regards. The firs main difference beween our paper and he lieraure dealing wih he opimal sequencing of exernal and inernal financial liberalizaion is ha we did no analyse counries ha are facing he risk of an acue financial crisis. Raher, we described he longer-erm linkages beween financial openness and financial developmen, on he one hand, and he volailiy of he real economy, on he oher hand. Mos financial marke inegraion and macroeconomic volailiy, end o be robus wih respec

26 24 imporanly, we ruled ou fixed exchange raes or oher forms of exchange rae managemen. Therefore, we canno use he paper o draw conclusions wih regard o policy responses ha are appropriae if speculaive pressure on a currency is building up. The second main difference beween his paper and he sequencing lieraure is ha we considered counries ha can use heir own currency when conracing wih foreign couner paries. We did no, in oher words, sudy cases in which he currency of denominaion of domesic ransacions is ha of he foreign counry, as is he case for many developing counries and emerging markes. Hence, our analysis is applicable in a sric sense only o relaively maure marke economies. Despie hese obvious differences o he lieraure on he opimal sequencing of inernal and exernal financial liberalizaion, our paper holds some ineresing implicaions. Perhaps he mos imporan message is ha ofen ariculaed fears ha he ongoing inernaional inegraion of financial markes may amplify he effecs of disorions in naional financial markes and may, hereby, lead o greaer economic insabiliy may no be jusified. While boh he empirical and heoreical framework in principle allow for he ineracion beween fricions on domesic financial markes and financial openness, hese effecs do no seem o be imporan in quaniaive erms. In our heoreical model, feedback effecs beween financial openness and financial developmen would work hrough he impac of financial openness on Tobin s Q and he real exchange raes. However, under plausible parameer consellaions, hese feedback channels do no seem o be imporan quaniaively. In our empirical model, we likewise did no deec significan ineracion effecs beween financial openness (i.e., he presence of conrols on cross-border financial credis) and he size of he domesic credi marke, which we used as a proxy for he sae of developmen of he domesic financial sysem. Ye, while he degree of openness of counries does no seem o be o he specificaion of he credi marke.

27 25 of empirical significance in explaining volailiy, a higher degree of developmen of he domesic financial sysem seems o be associaed wih lower volailiy. Through his channel, financial openness migh indirecly impac upon volailiy. This is because here is evidence ha he sae of financial developmen is endogenously relaed o he degree of financial openness of counries (see, e.g, Kaminsky and Schmukler 2001, Rajan and Zingales 2001). The expeced enlargemen of he European Union is one policy area for which he resuls of his paper are imporan. The accession saes of Cenral and Easern Europe are expeced o join he EU in he year Iner alia, accession will imply he full aboliion of remaining conrols on he free flow of capial across borders as well as furher developmen of he domesic financial sysem hrough paricipaion in he Single Marke for capial. Our resuls sugges ha he benefis of financial inegraion are likely o ouweigh he risks in erms of increased insabiliies. However, more research ino he links beween financial openness, financial secor developmen, and business cycle volailiy will be required o es he robusness of hese resuls. This research should address wo main issues. Firs, he accession saes of Cenral and Easern Europe are required o paricipae in he European Exchange Rae Mechanism (ERM2) before being considered o paricipae in he euro. Fixing he exchange rae, in urn, is likely o inroduce insabiliies and speculaive dynamics, an issue we did no address in his paper. Second, fuure research should pay more aenion o he poenial causes of differences beween he business cycle characerisics of emerging markes and indusrialized counries. Research ino he growh effecs of capial accoun liberalizaion shows ha counries ha are a differen sages of heir developmen process benefi from financial liberalizaion o quie differen degrees (for recen evidence see Klein 2003), and here is also evidence ha business cycle volailiy in emerging markes shows paerns differen from business cycle volailiy in

28 26 developed counries (Kose e al. 2002). Exploring he causes for hese differences boh in heoreical and empirical work will be a challenging ask for fuure research. 6 References Aghion, P., P. Bacchea, and A. Banerjee (1999). Capial Markes and he Insabiliy of Open Economics. In: P.R. Agenor, M. Miller, D. Vines, and A Weber (eds.). The Asian Financial Crisis: Causes, Conagion, and Consequences. Cambridge Universiy Press. Forhcoming. Bacchea, P., and R. Caminal (2000). Do capial marke imperfecions exacerbae oupu flucuaions? European Economic Review 44 (3): Bank for Inernaional Selemens (BIS) (2001). 70 h Annual Repor. Basle. Bech, M., P. Bolon, and A. Röell (2002). Corporae Governance and Conrol. Naional Bureau of Economic Research (NBER). Working Paper Cambridge, MA. Bernanke, B.S., and A.S. Blinder (1988). Credi, Money, and Aggregae Demand. American Economic Review 78(2): Bernanke, B.S., M. Gerler, and S. Gilchris (2000). The Financial Acceleraor in a Quaniaive Business Cycle Framework. In: Taylor, J., and M. Woodford (eds.), Handbook of Macroeconomics, Amserdam: Norh Holland, chaper 21. Ceneno, M., and A.S. Mello (1999). How Inegraed are he Money Marke and he Bank Loans Marke Wihin he European Union? Journal of Inernaional Money and Finance 18: Faia, E. (2001a). Sabilizaion Policy in a Two Counry Model and he Role of Financial Fricions. Working Paper No. 56. European Cenral Bank. Frankfur am Main, Germany. Faia, E. (2001b). Moneary Policy in a World wih Differen Financial Sysems. New York Universiy. Ocober. mimeo. Fleming, J.M. (1962). Domesic financial policies under fixed and under floaing exchange raes. I.M.F. Saff Papers 9: Frazscher, M. (2002). Financial Marke Inegraion in Europe: On he Effecs of EMU on Sock Markes. Inernaional Journal of Finance and Economics 7: Fuhrer, Jeffrey C. and George R. Moore (1995). Inflaion Persisence. Quarerly Journal of Economics 110: Gali, Jordi, Mark Gerler, and David Lopez-Salido (2001). European Inflaion Dynamics. European Economic Review 45: Giannei, M, L. Guiso, T. Jappelli, M. Padula, and M. Pagano (2002). Financial Marke Inegraion, Corporae Financing and Economic Growh. European Commission. Economic Papers 179. November. Brussels. Gilchris, Haircaul, and Kempf (2002). Financial Inegraion in a Moneary Union. Working Paper No European Cenral Bank. Frankfur am Main, Germany. Goodfriend, M.S. (1991). Ineres raes and he conduc of moneary policy. Carnegie- Rocheser Conference Series on Public Policy 34: 7-30.