THE TRANSMISSION OF MONETARY POLICY IN THE EURO AREA: IMPLICATIONS FOR THE EUROPEAN CENTRAL BANK

Transcription

1 Deparmen of Economics and Business Adminisraion Academic year 2-21 THE TRANSMISSION OF MONETARY POLICY IN THE EURO AREA: IMPLICATIONS FOR THE EUROPEAN CENTRAL BANK Ger Peersman Promoor : Prof. Dr. Rudi Vander Venne Co-promoor: Prof. Dr. Frank Smes Disseraion submied o obain he degree of Ph.D. in Economics March 21

2 PREFACE In wriing his disseraion, I owe a heavy inellecual deb o he many people ha have simulaed my ineres and guided me hrough he fields of economics over he years. I hank my promoor, Rudi Vander Venne, for his enhusiasic and excellen suppor during he developmen of his disseraion. I gained enormously from a virually coninuous conversaion abou economics and oher social opics. I has been a pleasure o be his research and eaching assisan for almos six years. I would like o hank my co-promoor and co-auhor of several papers, Frank Smes. He inroduced me in he academic world of moneary economiss and cenral bankers. He has guided my work very closely, and he augh me o do applied research and o wrie publishable papers. I wan o address a special hanks o he oher members of my Ph.D. commiee, Freddy Heylen, Paul De Grauwe and Jan Smes. They provided deailed, houghful and consrucive commens on almos every aspec of he hesis. Feedback from hese people on he preliminary ediion has been graefully received and incorporaed where appropriae ino his ediion. All remaining errors are my own responsibiliy. In alphabeical order, I have also received criical, subsanive suggesions from Ignazio Angeloni, Annick Bruggeman, Richard Clarida, Geer Dhaene, Shamik Dhar, Anil Kashyap, Lavan Mahadeva, Ben McCallum, Benoi Mojon, Gabriel Perez-Quiros, Peer Sinclair, Lars Svensson, John Taylor and Anders Vredin. They made valuable commens and suggesions concerning some chapers. I also wan o acknowledge seminar and conference paricipans a he European Cenral Bank, Bank of England, Sveriges Riksbank, De Nederlandsche Bank, Ghen Universiy, he Banca d Ialia and Bocconi Universiy conference on Moneary policy of he ESCB: Sraegic and Implemenaion Issues, and he Federal Reserve Bank of San Francisco and Sanford Universiy conference on Moneary policy and Moneary Insiuions. I am also graeful o he Direcorae General Research from he European Cenral Bank. I worked a lo on his hesis while being in he ECB s Graduae Research Programme. I benefied from alking wih he permanen saff and from paricipaing in he meeings of he Moneary Transmission Mechanism Nework. Bu I would be remiss no o menion he subsanial deb of graiude I owe o he secrearial saff, Olivia and Nahalie, my many friends and colleagues, especially I

3 John and Benoi, from he ime I have been a Ghen Universiy. I hank also my parens because hey gave me he opporuniy o sudy in he firs place, and los of ohers. Finally, I would like o hank Nahalie, for providing me wih a warm and happy environmen, for her coninuous moral suppor and los of paience. Ger Peersman Ghen, April 25, 21 II

4 CONTENTS Preface Conens 1. The ransmission of moneary policy in he euro area: an overview 1.1. Inroducion 1.2. The channels of moneary ransmission and heir implicaions for EMU Models wih flexible prices Imperfec informaion Limied paricipaion models Models wih sicky prices The ineres rae channel Overview Moneary policy and he erm srucure of ineres raes Implicaions for EMU The asse prices channel Overview Implicaions for EMU The credi channels Bank lending channel Balance shee channel Financial acceleraor Empirical evidence: Is here a credi channel of moneary policy ransmission? Implicaions for EMU Influence of changes in oupu on inflaion Summary 1.3. Empirical evidence for he euro area Large economeric models Small economeric models 1.4. Conclusions Research conclusions Policy implicaions Furher research 1.5. Appendix Appendix 1.1. Moneary policy and long erm ineres raes in Germany Appendix 1.2. Some more evidence on he ransmission mechanism in he euro area 1.6. References I III III

5 2. The moneary ransmission mechanism in he euro area: more evidence from VAR analysis 2.1. Inroducion 2.2. A VAR-model for he euro area The specificaion of he benchmark model Basic esimaion resuls A comparison wih he Unied Saes 2.3. A robusness analysis The inclusion of M3 in he VAR Using he German ineres rae insead of he area-wide ineres rae Using he inflaion rae insead of he price level Alernaive idenificaion schemes An ineracion beween shor-erm ineres rae, exchange rae and he money aggregae A mixure of shor and long-run resricions 2.4. The effec of moneary policy on oher macro variables The exended model The resuls 2.5. Concluding remarks 2.6. References 3. A VAR descripion of he effecs of moneary policy in he individual counries of he euro area 3.1. Inroducion 3.2. The effecs of moneary policy in he counries of he euro area 3.3. VAR models for he individual counries in he euro area The specificaion of he benchmark VAR-model for he euro area VAR models for he individual counries in he euro area Germany France, Ialy, Spain, Finland, Ireland, Porugal and Belgium Ausria, he Neherlands (and Belgium) 3.4. Esimaion resuls 3.5. Conclusions 3.6. References IV

6 4. Are he effecs of moneary policy in he euro area greaer in recessions han in booms? 4.1. Inroducion 4.2. Is here a common cycle in he euro area? 4.3. The asymmeric effecs of area-wide moneary policy shocks A moneary policy VAR for he euro area Moneary policy shocks in he mulivariae MSM model 4.4. Are he moneary policy effecs differen across counries? 4.5. Does moneary policy change he likelihood of a recession? 4.6. Conclusions 4.7. References 5. The indusry effecs of moneary policy in he euro area 5.1. Inroducion 5.2. The indusry effecs of moneary policy Mehodology Esimaion resuls 5.3. Indusry characerisics and he moneary policy effecs Indusry characerisics The convenional ineres rae channel The financial acceleraor channel Specificaion and resuls One-sep esimaion 5.4. Conclusions 5.5. Appendix Appendix 5.1. Daa sources and definiions 5.6. References 6. The Taylor rule: a useful moneary policy benchmark for he euro area? 6.1. Inroducion 6.2. Taylor rules from he pas 6.3. The Taylor rule in an aggregae model for he EU An esimaed aggregae model for he EU How well does he Taylor rule perform? Insrumen rules and he loss funcion Resuls 6.4. Uncerainy and he robusness of simple Taylor rules The effec of esimaion error in he oupu gap The effec of uncerainy abou he ransmission mechanism 6.5. Conclusions V

7 6.6. Appendix Appendix 6.1. Some more evidence on he ransmission mechanism in Europe Appendix 6.2. Esimaion and opimal conrol of he EU5 model A Esimaion of he EU5 model A Opimal conrol of he EU5 model 6.7. References VI

8 CHAPTER 1 THE TRANSMISSION OF MONETARY POLICY IN THE EURO AREA: AN OVERVIEW Overview In his chaper, a general overview of he ransmission mechanism of moneary policy in he euro area is provided. We give a brief overview of he differen channels of moneary ransmission and heir implicaions for he EMU. Furher, he exising empirical evidence on moneary ransmission in he euro area is discussed. Also, he new empirical evidence provided in his hesis, is discussed in he conex of he lieraure. Finally, he general conclusions of his hesis are repored.

9 1.1. INTRODUCTION The Eurosysem sill faces some imporan challenges since he birh of he euro on January 1, The main ask of he European Cenral Bank (ECB) is o conduc moneary policy for he euro area. Following he Maasrich Treay, he primary objecive of he ECB is o ensure price sabiliy. Moreover, wihou prejudice o he objecive of price sabiliy, he ECB shall suppor he general economic condiions in he area, such as a susainable and non-inflaionary growh. In order o achieve hese finals goals, he ECB has some insrumens under conrol o conduc moneary policy. Primary among hese is he conrol of he shor-erm ineres rae and he moneary base. There is however sill a lo of uncerainy regarding he impac and iming of hese insrumens on he final objecives. More generally, moneary policy has become he cenral ool of macroeconomic sabilisaion over recen years, bu one ha has someimes unexpeced and unwaned consequences. To be successful in conducing moneary policy, he moneary auhoriies mus have an accurae assessmen of he effecs of heir policy on he economy. This requires a good undersanding of he ransmission mechanisms hrough which moneary policy affecs he economy. This challenge is even more complicaed for he euro area because while he ECB may conduc moneary policy based on union-wide aggregaes, he impac of is policy can be differen across he member counries. Specifically, under he EMU, member counries will be subjec o common moneary policy shocks. Given he diversiy in he economic and financial srucures across he economies, hese common moneary shocks can be expeced o have a differen impac, in erms of iming, magniude and disribuional effecs. Lile is known abou wha differences migh arise, given he absence of any hisorical experience in Europe wih a common moneary policy. Moreover, he creaion of he Eurosysem consiues an enormous regime shif, which can change he ransmission mechanism, making he job of he European Cenral Bank even more difficul. In his hesis, we discuss he ransmission mechanism of moneary policy. More specifically, empirical evidence is provided for he euro area. There has been already a large body of lieraure concerning he ransmission mechanism. In his firs chaper, we ry o give a brief overview of his exising lieraure. An undersanding of he ransmission channels is essenial for he design and implemenaion of moneary policy. These channels are discussed in he nex secion. Disincion is made beween he radiional money view (we make a disincion beween models wih flexible prices and models wih sicky prices) and he more recen credi view. We also documen in some deail differences in he srucures of he member counries ha may be relevan for poenial asymmeries of he moneary policy ransmission in he euro area. Besides he EMU members, wo possible fuure candidaes, Sweden and he UK, are also included. Since he 2

10 inroducion of he euro causes a regime shif, a discussion of he consequences of EMU on he differen channels is repored. The quesion is asked wheher he poenial asymmeric effecs will disappear or become sronger wih he inroducion of he euro. Secion 1.3 gives an overview of he exising empirical evidence, and secion 1.4 discusses he general conclusions. New empirical evidence is provided in chapers 2 ill 6. These chapers can be viewed as separae and independen papers. Mos of hese chapers conain work wih co-auhors, and some of hem are already published or forhcoming in books or journals. Chaper 2, The moneary ransmission mechanism in he euro area: more evidence from VAR analysis is join work wih Frank Smes, and is forhcoming as a chaper of an ECB book concerning he moneary ransmission mechanism. An exensive version of Chaper 3, A VAR descripion of he effecs of moneary policy in he individual counries of he euro area, ogeher wih Benoi Mojon is also forhcoming in he same book. Chapers 4, 5 and 6 are again join work wih Frank Smes. Chaper 4, Are he effecs of moneary policy in he euro area greaer in recessions han in booms? is currenly an ECB Working Paper, and will be published in a Bank of England CCBS book on he moneary policy process from Cambridge Universiy Press. Chaper 6, The Taylor rule; A useful moneary policy benchmark for he euro area?, has been published in Inernaional Finance, 1999, No. 1. This hesis conains he working paper version. In chaper 2, we invesigae he effecs of an unanicipaed change in moneary policy using he vecor auoregression (VAR) mehodology. We focus on he area as a whole, and esimae he model using synheic euro area daa from 198 ill We show ha using a sandard idenificaion scheme delivers plausible effecs of moneary policy. An unexpeced, emporary rise in he shor-erm ineres rae ends o be followed by a real appreciaion of he exchange rae and a emporary fall in oupu afer wo quarers. The effec on oupu reaches a peak afer five quarers, o slowly reurn o baseline afer weny quarers. Prices are more sluggish and only sar o fall significanly below zero some quarers afer GDP. The effec on prices is also more persisen. We invesigae also he reacion of oher macro variables, and he GDP componens o a moneary policy shock. As expeced, invesmen reacs more, and privae consumpion reacs less han GDP. We also find a srong and immediae liquidiy effec on M1, and a slow decrease of M3 and oher credi aggregaes. One criicism of he area-wide approach is, of course, ha because here was no single moneary policy regime, a model based on synheic euro area daa is likely o be misspecified. In order o check he robusness of he area-wide esimaes, in he hird chaper, we herefore modify he basic euro area model and esimae i for each of he counries in he euro area separaely. In doing so, we minimise he changes in he specificaion. We show ha hree relaively simple idenificaion 3

11 schemes, depending on he moneary policy decision process in he ERM, obain similar and consisen effecs for he individual counries as in he area-wide model. In chaper 4, we invesigae wheher here is a differen impac of moneary policy in recessions and expansions for eigh counries of he euro area (Germany, France, Ialy, Spain, Finland, Belgium, Ausria, and he Neherlands). To do so, we ake a euro area-wide approach. We proceed in various seps. In he firs sep, we es wheher here has indeed been a common cycle in hese counries using a mulivariae exension of Hamilon s (1989) wo-sae Markov Swiching Model (MSM). In oher words, we ask wheher before he sar of EMU heir business cycles were sufficienly synchronised. This is of independen ineres because i allows us o see wheher differences in cyclical siuaion are likely o complicae moneary policy. We he excepion of Finland, we find ha we canno rejec he hypohesis ha here was a common business cycle in hese counries. We hen exend he mulivariae MSM o es wheher he area-wide moneary policy impulses, esimaed in chaper 2, have differen effecs on oupu in booms versus recessions. We find ha moneary policy has considerably larger effecs on aciviy when he economy is in a recession. These asymmeries are mos pronounced in Germany, France, Spain, Ialy, and Belgium. We also find weak evidence ha a ighening of moneary policy reduces he probabiliy of saying in a boom and no evidence ha an expansionary moneary policy increases he probabiliy of going from a recession o a boom. In chaper 5, we swich from he counry level o he indusry level. The effecs of he area-wide moneary policy shocks on oupu growh, obained in chaper 2, are invesigaed for eleven indusries of seven euro area counries. On average, he negaive effec of an ineres rae ighening on oupu is significanly greaer in recessions han in booms. There is, however, considerable cross-indusry heerogeneiy in boh he overall policy effecs and he degree of asymmery across he wo business cycle phases. We hen explore which indusry characerisics can accoun for his heerogeneiy. Differences in he overall policy effecs can mainly be explained by he durabiliy of he goods produced in he secor, he capial inensiy of producion and he degree of openness. This can be regarded as evidence for he convenional ineres rae/cos-of-capial channel. In conras, differences in he degree of asymmery of policy effecs seem o be relaed o differences in financial srucure, in paricular he mauriy srucure of deb and he need for working capial. This suggess ha financial acceleraor mechanisms can parly explain cross-indusry differences in asymmery. Finally, he conens of chaper 6 are somewha differen from he oher chapers. This chaper explores he Taylor rule as a benchmark for analysing moneary policy in he euro area. The focus is mainly on he endogenous par of moneary policy, i.e. he reacion funcion of he cenral bank. Firs, i presens evidence ha ineres 4

12 raes in Germany and he euro area can be described by a Taylor rule wih ineres rae smoohing. Second, i shows ha an opimised Taylor rule performs quie well in a closed economy model of he euro area. Finally, he robusness of hese resuls o esimaion errors in he oupu gap and he effec of uncerainy abou he ransmission mechanism are examined. 5

13 1.2. THE CHANNELS OF MONETARY TRANSMISSION AND THEIR IMPLICATIONS FOR EMU The money view has been he radiional way of explaining he moneary ransmission mechanism. I consiss of several channels, all of hem inerlinked and each of hem having several sages. We divide hem ino wo groups: one group assumes flexible prices, while he oher assumes ha prices are sicky. The models of he money view do no make a disincion beween he differen sources of invesmen financing. In line wih he Modigliani and Miller (1958) heorem, hese models assume ha he manner of financing invesmen is irrelevan for he invesmen decision aken by he enerprises. In conras, models of he credi view, also discussed in his secion, explicily disinguish beween bank loans and bond issuing. The differen channels, wih heir implicaions for EMU are discussed in he nex subsecions. The explanaion of he differen channels, and heir inerlinkages, is based on Graph 1.1. Graph 1.1 The channels of moneary ransmission Money supply Official ineres raes (13) bank deposis (2) marke ineres raes (16) (8) (4) cash flow equiy prices exchange rae (17) (18) (11) (9) (5) ne worh wealh Tobin's q ne expor (1) (14) (19) (7) exernal finance premium (3) (2) (12) (1) (6) bank loans (15) invesmen / consumpion / oupu (21) inflaion 6

14 MODELS WITH FLEXIBLE PRICES There are wo classes of models ha can explain shor-run real effecs of moneary policy wih flexible prices and wages. The firs is based on imperfec informaion, and he second on rading resricions in financial markes Imperfec informaion The models based on imperfec informaion (channel 1 in Graph 1.1) focuse on mispercepions of he public abou aggregae economic condiions, originally developed by Friedman (1968) and exended by Lucas (1972). 1 They believe ha prices and wages are fully flexible in he shor-erm, bu price rigidiies arise as a resul of limis on informaion. The mechanism for limiing informaion is ha rading akes place in several disinc markes wih individuals who have imperfec informaion abou aggregae economic variables. Then, an increase in he price in one of he markes can signal eiher ha he money sock has increased, in which case here is a general increase in all prices and no need o change producion, or ha he relaive price has increased, in which case i makes sense o increase producion. This implies ha, wih limied informaion abou he source of he price rise, suppliers mus solve a signal exracion problem. The soluion o his problem is o increase supply when he price rises, bu by an amoun ha depends on he relaive variabiliy of money supply changes and marke specific shocks in he pas. The Lucas model has he following policy implicaions. Firs, only moneary policy acions ha are no sysemaic have a shor-run influence on oupu. Sysemaic policy changes are fully refleced in he price level. Secondly, he greaer he unpredicabiliy of moneary growh (i.e. he less credible he moneary auhoriy), he less he influence of moneary policy shocks on real oupu. A hird influenial implicaion of Lucas model is known as he policy irrelevance hypohesis and was demonsraed by Sargen and Wallace (1975) and Barro (1976). If sysemaic moneary policy has no real effecs, he choice of he moneary policy rule by he cenral bank is of no consequence for oupu developmens. The early empirical evidence was in favour of Lucas model. Lucas (1973) finds ha nominal shocks have smaller real effecs when aggregae demand is more volaile, as his model predics. Barro (1977, 1978) and Barro and Rush (198) find ha only he unpredicable componen of moneary growh affecs real oupu and unemploymen. Anicipaed moneary policy was irrelevan. However, subsequen empirical work has criicised his. Gordon (1982) and Mishkin (1982) show ha also 1 Varians of he Lucas (1972) model are Barro (1976), Romer (1996), Mahews (1997), and Walsh (1998). 7

15 anicipaed changes in he money supply have imporan oupu and employmen effecs, and herefore he choice of policy rule is no irrelevan for he behaviour of real economic aciviy. Pesaran (1982) and Ball e al. (1988) es he Lucas model agains respecively an alernaive Keynesian and new-keynesian model and find ha hey can rejec he former, bu no he laer. As a consequence, hese models are no longer viewed as an adequae explanaion for he shor-run real effecs of moneary policy. However, he disincion beween anicipaed and unanicipaed moneary policy is sill very imporan in recen empirical and heoreical work. For example, we will recognise he influence of an unanicipaed moneary policy shock, modified for sysemaic moneary policy, on real oupu in secion 1.3.2, where we discuss he lieraure of VARs. Also, i is likely ha a complee heory of price rigidiies, as he foundaion of he ineres rae channel, will evenually involve elemens of he limied informaion heory. Individuals seing prices do no change hem frequenly, or in response o every change in marke condiions, because hey have many hings o hink abou and limied ime o do differen asks of daa-gahering and analysis (Sims, 1998). Moreover, he microeconomic foundaions of he saggered conrac models involve imperfec informaion abou wheher a shock is emporary or permanen or local or economy-wide (Taylor, 1998). Since his channel depends on he relaive variabiliy of aggregae and relaive prices, nohing from he pas can be used o assess he srengh of his channel in he Eurozone. This is because he ECB is a new insiuion. Asymmeric effecs beween he member counries, due o he direc moneary channel, are also impossible since here is only one single moneary policy. We can expec ha he imporance of his channel is decreasing over ime because i is much easier o have access o informaion abou evoluions of prices wih he developmens of echnology Limied paricipaion models Shor-erm effecs of moneary policy on he economy in an environmen of fully flexible prices and wages can also be generaed wih limied paricipaion models, developed by Grossman and Weiss (1983), Roemberg (1984), Lucas (199), Chrisiano and Eichenbaum (1992), and Chrisiano, Eichenbaum and Evans (1996). Saring from a basic cash-in-advance framework, hese models generae real effecs of moneary policy because cerain resricions limi he abiliy of economic agens o make financial ransacions. So, mos people do no paricipae in financial markes coninuously and have o make heir cash decisions for more han one period. Generally, hree differen represenaive agens are assumed: households, enerprises and financial inermediaries. An unanicipaed increase of money is injeced ino he economy hrough he financial inermediaries. Because households are saggered in heir visis o financial markes, hey can no adjus heir money 8

16 balances. The addiional liquidiy is offered o he enerprises who will only accep his a a lower ineres rae. This lower ineres rae will simulae invesmen and oupu. 2 If he households were no resriced o paricipae in financial markes, his lower ineres rae would induce lower household savings and hereby offse he ineres rae effec. The relevance of he channels emphasized in limied-paricipaion models for undersanding he broader effecs of moneary policy on he aggregae economy remains an open debae. Dosey and Ireland (1995), and King and Wason (1996) show ha he predicions of his class of models do no accoun he magniude acually found in he daa. However, Chrisiano, Eichenbaum and Evans (1996) show ha heir limied paricipaion model is able o mach he evidence on he effecs of a moneary policy shock if he labour supply wage elasiciy is assumed o be very high. Again, we could no expec asymmeric effecs of his channel across counries in he euro area because financial markes are equally developed. However, he impac of his channel can change over ime (and probably has changed he las decade) as financial secors evolve and he cos of ransacions falls (Walsh, 1998) MODELS WITH STICKY PRICES The ineres rae channel Overview The ineres rae channel is he sandard feaure of he radiional exbook IS/LM model. 3 Roughly spoken, he ransmission works along he following lines: The cenral bank raises he real money balances of households. For a given ineres rae, his means an excess of cash in he percepion of households. When households sar o reduce heir real balances hrough he acquisiion of bonds hey hereby also reduce he ineres rae (sep 2 in Graph 1.1) and hereby he cos of capial, which leads o an increase in invesmen and oupu (sep 3). 4 This is he convenional ineres rae channel. Policy-induced changes in open marke ineres raes discourage or encourage invesmen spending, and hereby affec aggregae demand and oupu. The inflexibiliy of he price level is he crucial assumpion in 2 Remark ha he effec of moneary policy on invesmen and oupu goes hrough a liquidiy effec: An increase of money resuls in a decrease of he real and nominal ineres rae. However, we do no call his he ineres rae channel (as in he nex subsecion) because he laer is radiionally based on models wih sicky prices, while we sill have flexible prices in he limied paricipaion models. 3 See e.g. Hall and Taylor (1988) and Romer (1996). 4 This cos of capial effec also plays a role for durable consumer goods of households, such as houses and cars. 9

17 his framework. As long as he price level is fixed, he cenral bank can change he amoun of real balances and herefore influence real economic aciviy. Once nominal rigidiies are inroduced, i follows ha boh sysemaic and unexpeced moneary policy acions have an effec on oupu. This implies ha he choice of policy rule by he moneary auhoriy is relevan for he shor-run movemens of boh prices and oupu. The implicaion of sicky prices and wages is also invesigaed in general equilibrium models (e.g. Blanchard and Kiyoaki, 1987, and Chari, Kehoe and McGraan, 1996). These models sar from explicily specified marke imperfecions and examine opimal behaviour under such imperfecions. They also find nonneuraliy of moneary policy. Mos of hese models also explicily model he form of price sickiness. Examples of models ha explain he rigidiy of prices are he saggered prices model of Taylor (1979, 198), he predeermined prices model of Fischer (1977), and Phelps and Taylor (1977), Calvo (1983) price seing behaviour models, or he models wih endogenous price seing behaviour of Sheshinski and Weiss (1977, 1983) and Caplin and Spulber (1987). 5 6 Moneary policy and he erm srucure of ineres raes Aggregae spending decisions are generally viewed as more closely relaed o longerm ineres raes and bank lending raes, while he moneary policy insrumen is bes represened by he shor-erm ineres rae. Changes in he shor-erm ineres rae will only affec aggregae spending decisions if longer-erm raes and lending raes are affeced. The impac of shor-erm raes on long-erm raes is difficul o assess. This is because he relaionship beween shor raes and long raes is believed o be based on he expecaions heory of ineres raes. In his heory, long-erm raes are an average of curren shor-erm raes and expeced fuure shor-erm raes. So, moneary policy affecs long-erm raes o he exen ha i influences curren and expeced shor-erm raes. This effec on long raes of a change in he sance of moneary policy will parly depend on he impac of he policy change on inflaion expecaions. The influence of moneary policy on he ineres raes of differen mauriies is an imporan issue for he European Cenral Bank, because he reacion of he economies of he euro area o a moneary policy impulse could be differen due o differences in he reacion of longer erm ineres raes and bank lending raes o a moneary policy shock. 7 5 On he oher hand, we have also he New Keynesian economics models wih small menu coss (Mankiw, 1985, and Akerlof and Yellen, 1985). 6 For an overview of he empirical evidence of price and wage sickiness, boh a he macroeconomic and microeconomic level, see Taylor (1998). 7 See he discussion in he nex subsecion. 1

18 The empirical evidence on he quaniaive effec of moneary policy on he longerm ineres raes finds on average a posiive relaionship: an increase in he cenral bank rae leads o an increase in ineres raes of all mauriies. 8 Romer and Romer (1996) find his posiive movemen in he long rae inconsisen wih sandard moneary heory, because an increase in shor raes should reduce inflaion, and hence reduce he level of sufficienly long raes. Also, here are many excepions o his empirical phenomenon. Examples are declines of long-erm raes in response o a moneary ighening. Romer and Romer (1996) explain his empirical finding by arguing ha when he Cenral Bank ighens moneary policy, marke paricipans infer ha i has unfavourable privae informaion abou he likely behaviour of inflaion, and hey herefore revise heir expecaions of inflaion upward. Ellingsen and Södersröm (21) consruc a model for he US wihin which he mechanism of Romer and Romer fis. The model hey use is aken from Svensson (1997), and is a dynamic version of a simple aggregae supply-aggregae demand model, where hey add an equaion for he erm srucure of ineres raes. They presume ha a change in moneary policy can come abou for wo reasons: eiher he moneary auhoriies respond o new and possibly privae knowledge abou he economy (such as supply and demand shocks), or heir policy preferences change (which is a moneary policy shock). In he firs case, policy is essenially endogenous, reflecing new inpu ino a given objecive funcion. In he second case, policy is exogenous, in he sense ha he inpu is he same bu he objecive funcion has changed. Afer an endogenous policy acion, heir model predics ha ineres raes of all mauriies move in he same direcion as he policy innovaion. On he oher hand, afer an exogenous policy acion, shor and long ineres raes should move in he opposie direcions. Ellingsen and Södersröm (21) find posiive empirical evidence for heir proposiions by inerpreing newspaper repors from he Wall Sree Journal immediaely before and afer each meeing of he FOMC. In appendix 1.1, we esimae a simple srucural VAR for Germany o es he model of Ellingsen and Södersröm. Firs, we look a he impulse responses of he ineres raes of all mauriies o supply and demand shocks. These shocks are symmerically observed by all agens or he cenral bank may have privae (advanced) informaion abou hese shocks. In boh cases, we find ha ineres raes of all mauriies move in he same direcion. Second, he impac of pure exogenous moneary policy shocks is invesigaed. In ha case, he preferences of he cenral bank have changed and sufficienly long ineres raes move in he opposie direcion, because expeced inflaion is reduced. 8 For example: Cook and Hahn (1989), Dale (1993), Roley and Sellon (1995), Buiglione e al (1997), and Mehra (1996). 11

19 Implicaions for EMU A necessary condiion for he ineres rae channel o be operaive is ha price adjusmen does no occur insananeously. In he radiional lieraure, 9 firms price-seing decisions largely depend on he level of he average inflaion rae. Again, he disribuional consequences of his behaviour have disappeared since he sar of he moneary union. However, asymmeric effecs are sill possible because of differences in, e.g. he labour marke. In he laer, wage seing behaviour plays a crucial role. Table 1.1 presens some indicaors of price flexibiliy. The employmen proecion variable is a weighed average of indicaors peraining o regular labour conracs, emporary conracs, and collecive dismissals calculaed by he OECD. Proecion is very low, and hus labour markes are more flexible in he UK and Ireland. Employmen proecion is more similar in he oher counries, and raher high in he souhern counries (Ialy, Porugal and Spain). The second indicaor is a measure of real wage rigidiy obained from Berhold e al. (1999). They esimae he effec of a wage-push shock on unemploymen using SVAR. Low values indicae ha real wages are very flexible. The counries can be pu ino hree groups. Sweden, Finland and Porugal have highly flexible real wages. Germany, he Neherlands and he UK exhibi a medium level of real wage flexibiliy, whereas Ialy, Ausria, Belgium, Spain, Ireland and France have highly rigid real wages. The las column presens he price response of firms o a change in heir capaciy uilisaion, esimaed by Roeger and in Veld (1999). Prices are relaive flexible in Porugal, he Nerherlands, Ausria and Ireland, and rigid in Spain en Finland. We have o remark ha he shif o he EMU could considerably change he working of he labour marke. The resisance of policymakers o labour marke reforms migh decline because hey realize ha moneary policy is no longer available o accommodae asymmeric shocks. Moreover, he bargaining power of he unions is expeced o decline, leading o more flexible labour markes, and hus convergence across counries owards more flexible wages and prices. However, here is considerable disagreemen in he lieraure ha his will indeed be he case. 1 One moivaion for labour marke reforms is he reducion of equilibrium inflaion. The higher equilibrium unemploymen is, he higher is he credibiliy problem of moneary policy and he higher is herefore he equilibrium inflaion rae. The benefis of labour marke reforms in erms of a lower equilibrium inflaion rae are higher under moneary auonomy han in EMU. In he laer, labour marke reforms produce posiive exernaliies as he inflaion bias is also reduced for he oher member counries. As is well known from sandard welfare heory, he amoun of aciviies producing posiive exernaliies is below he social opimum. 9 See he previous subsecions for he references. 1 For an overview, see Bean (1998), Calmfors (1998) or Berhold e al. (1999). 12

20 Anoher reason why EMU migh lead o less labour marke reforms is ha i akes a long ime for supply side policies on heir own o resul in employmen gains. The governmen herefore runs he risk ha i akes oo long before he posiive effecs of he reforms underaken are fel. A parallel demand side policy, such as moneary policy, speeds up he process of employmen gains and is hus helpful o enforce labour marke reforms. The laer is, however, only possible if labour marke policies and moneary policies are a he naional level. I is also unlikely ha labour marke reforms will be realised quickly because hese reforms are mosly implemened in acue crisis siuaions, when unemploymen rises sharply due o negaive shocks. We can a leas expec ha i will ake some ime before labour marke reforms will ake place, and ha hese differences across counries are sill imporan for he ransmission of moneary policy in he shor o medium erm. Table 1.1 Price flexibiliy indicaors Employmen proecion a Measure for real wage rigidiy b Price response o capaciy uilisaion changes c Germany 2.6 (2) Neherlands 2.2 (13) Belgium 2.5 (16) France 2.8 (21) Spain 3.1 (22) Ialy 3.4 (23) Ausria 2.3 (15) Finland 2.1 (11) Ireland 1.1 (5) Porugal 3.7 (26) Sweden 2.6 (18) UK.9 (2) a: summary indicaor of sricness of employmen proecion, OECD, Ranking of all OECD counries beween parenheses. All rankings increase wih he sricness of employmen proecion. b: esimaed by Berhold, Fehn and Thode (1999). Indicaor is he effec of a wage-push shock on unemploymen (using SVAR). c: esimaed by Roeger and in Veld (1999). The impac and iming of moneary policy acions can differ across indusries, because of varying ineres sensiiviies in he demand for producs. Durable manufacured goods, such as houses and cars are expeced o be more responsive o ineres raes han non-durable goods and services. Carlino and Defina (1998) find for he US ha saes wih a higher share of manufacuring and a higher share of consrucion are more sensiive o a moneary policy shock. We also find in chaper 5 ha indusries producing durable goods respond significanly more o a moneary policy shock. 13

21 For he EMU, he share of manufacuring and consrucion in oal GDP is comparable across he member counries (Table 1.2). The excepions are Germany, wih a higher share, and he Neherlands wih a lower share han average. Asymmerical responses of hese counries are possible. The expecaions abou he influence of he moneary union on indusrial srucure are mixed. The European Commission (1997) argues ha he European inegraion will promoe inraindusry rade, leading o less cross-counry differences in indusrial srucure. On he oher hand, Krugman (1993) claims ha EMU is likely o increase regional specialisaion in Europe, implying divergence in indusrial srucure. So, if he inroducion of he euro leads o concenraion of he manufacuring and consrucion secors in some counries, hese counries will be more vulnerable o moneary policy shocks. Table 1.2 Indusrial srucure in European counries Share of manufacuring and consrucion in oal GDP a Gross fixed capial formaion as a % of GDP b Germany Neherlands Belgium France Spain Ialy Ausria Finland Ireland 16.8 Porugal Sweden 15.2 UK 15.6 a: Carlino and Defina (1998). b: average of , European Economy (1998). Capial-inensive indusries, which need more invesmen, are also expeced o be affeced more srongly han less capial-inensive indusries (he cos-of-capial channel is more poen for he former). This is confirmed for he euro area by he evidence provided in chaper 5. Indusries wih higher invesmen inensiy are more vulnerable o area-wide moneary policy shocks. In chaper 2, we find economeric evidence for he euro area ha invesmen, which is likely o be much more sensiive o a change in he cos of capial, responds sronger o a moneary policy shock han consumpion. The response of invesmen is hree imes as large as he response of oal GDP, while he response of consumpion is less han GDP. This is illusraed in Graph 1.2, obained from chaper 2. A comparison of he euro area and he US (Table 1.3) ells us ha gross fixed capial formaion is similar in boh economies. However, he las column of Table 1.2 shows ha here is a considerable difference across counries. Capial-inensiy is high for Porugal, 14

22 Ausria, and Germany, and raher low for Sweden and he UK. The oher counries are more in beween. Graph 1.2 Impulse response of GDP, invesmen and consumpion o a moneary policy shock (Esimaion period: ) Horizon (quarers) GDP CONS INV Table 1.3 Key characerisics of he euro area and he Unied Saes (percenages of GDP in 1999) GDP componens Privae consumpion Governmen consumpion Gross fixed capial formaion Expors Impors Financial markes Euro area Unied Saes Sock marke capialisaion Deb securiies Of which issued by: Governmen Banks Corporae secor a Bank asses Of which: Loans o governmen Loans o he corporae secor Loans for house purchase Oher loans o households Source: ECB (2). a: This figure includes deb securiies issued by governmen-sponsored enerprises and federally relaed morgage pools (42% of GDP). 15

23 Wih respec o he influence of moneary policy, a difference should be made beween households, he governmen and firms, because he euro area is characerised by ne credior households, while he public and corporae secors are ne debors (ECB, 2). On he one hand, an increase of he ineres rae modifies he cos of capial leading o a change in invesmen by firms and durable consumpion by households, and an increase of he burden of he governmen deficis. The laer is a very imporan borrower on he financial markes. Due o he Sabiliy and Growh Pac, his will enail a cu in governmen expendiures and hampers he working of he auomaic sabilisers. In ha case, he fiscal policy reacion o a conracionary moneary policy shock will be more resricive in Belgium and Ialy, wih heir large socks of governmen deb (firs column of Table 1.4). Table 1.4 provides also an overview of credi o firms and households. Nex o Germany and he Neherlands, credi is wice as imporan in Sweden and he UK (he poenial members) han in Ialy. Table 1.4 Composiion of credi Gross nom. consolidaed deb of he governmen Credi o he non-governmen secor as a % of GDP (1998) a Toal Hous. Firms Share hous./ firms in oal credi Ne. ineres received by hous. /o. liab b Hous. fin. liab. /disposable income b Germany / Neherlands / Belgium / France / Spain / Ialy / Ausria /68 Finland 49.6 Ireland 59.5 Porugal 6. Sweden / UK / c 12. Noe: all figures are calculaed from Borio (1996). a: As a percenage of GDP, European Economy (1998). b: Kneeshaw (1995). c: The UK number includes dividend income. On he oher hand, he impac on non-durable consumpion of households goes hrough income and subsiuion effecs (Favero and Giavazzi, 1999). The subsiuion effec reduces consumpion when here is a rise in he real ineres rae because i is more profiable o save. The income effec will depend on he ne credi posiion of he consumer. An increase in ineres raes raises disposable income primarily in Ialy and he Neherlands (Table 1.4) while wealh decreases in Sweden. The share of financial liabiliies in disposable income is also low in Ialy 16

24 and very high in Sweden and he UK, making he negaive consumpion effec of he laer wo sronger. The figures and poenial asymmeric effecs should, however, be inerpreed by making a disincion depending on he mauriy of households and firms' deb. As explained in he previous subsecion, a shif in he sance of moneary policy is immediaely ransmied o shor-erm money marke raes and floaing raes linked o shor-erm raes. The impac on longer-erm raes, however, is more difficul o assess. Graph 1.3 shows he response of he shor, and long-erm ineres rae o a moneary policy shock in he euro area (see chaper 2 for a descripion of he VAR). The shor rae increases wih 3 basis poins, while he long rae increase is only 8 basis poins. 11 Since a large proporion of households ineres income comes from asses a a long erm rae, he income effec is likely o be raher small and will adjus only gradually o changes in he ineres rae (ECB, 2). Graph 1.3 Impulse response of shor and long-erm ineres rae o a moneary policy shock (Esimaion period: ) Horizon (quarers) long shor The share of credi a shor-erm or adjusable medium and long-erm raes is very differen across European counries (Table 1.5). Ialy, Ausria and he UK are very shor-erm oriened, whereas credi in he Neherlands is mainly a fixed raes. However, when we relae hese figures o he counries' GDP, only he UK, and o a lesser exen Ausria, is significanly more sensiive o a change in shor-erm ineres raes, because morgage deb of households in he UK is primarily relaed o he shor-erm ineres rae. Alhough he share of adjusable ineres raes is very large in Ialy, he proporion o oal GDP lies close o he EU-average, because he 11 Noe ha we do no find a negaive correlaion beween he shor and long-erm ineres rae afer a moneary policy shock, as we did in he German case (see appendix 1.1). 17

25 oal volume of non-governmen credi is very low in Ialy (specifically he households credi). The governmen s need o finance large budge deficis can be a reason for prevening he growh of he consumer-deb indusry in Ialy (Favero and Giavazzi, 1999). The Neherlands sill remain weakly sensiive o changes in shor-erm ineres raes, despie heir large raio of credi o GDP from Table 1.4. One can argue ha he shif o EMU can significanly change he composiion of Ialian credi owards less adjusable forms. The reason is ha, in he pas, Ialy was characerised by high and variable inflaion raes. The shif o a more credible ECB lowers he uncerainy abou fuure inflaion, making long-erm credi more aracive. More generally, he single moneary policy in Euroland may lead o less divergence of he credi srucure across members. Table 1.5 The erm srucure of credi o he privae secor S.T. + Adj. M.T. & L.T. / Toal credi Shor erm + adjus. medium and long erm credi as a % of GDP Effec on he lending rae of changes in money marke rae a Toal Households Firms 1 quarer Long run Germany Neherlands Belgium France b.74 b Spain Ialy /13 34/ Ausria Finland.2.6 Ireland Porugal Sweden b.92 b UK Noe: all figures are calculaed from Borio (1996). a: Coarelli and Kourelis (1994). b: Borio (1995). The reacion of he economies o a moneary policy impulse could be differen due o discrepancies in he response, boh in magniude and iming, of bank lending raes o a change in he ineres rae conrolled by he moneary auhoriy. Banks wih a close relaionship wih heir cusomers will (emporarily) absorb a par of he ineres rae change. Compeiion in he banking secor is also an imporan facor of he pass-hrough from marke raes o reail bank ineres raes. Coarelli and Kourelis (1994) find ha he adjusmen of he bank lending raes is insananeous and complee in he Neherlands and he UK. The response is significanly slower in Ialy, France and Finland. For he laer wo, he response is even incomplee in he long run (see he las wo columns in Table 1.5). They also find ha he degree 18

26 of sickiness is higher when compeiion in he banking secor is lower for a larger sample of counries. We can conclude ha he ineres rae channel is a very imporan channel for moneary policy ransmission in he euro area. The srengh of his channel will persis afer he inroducion of he euro. Table 1.6 provides an overview of he main deerminans of his channel across he individual counries based on he discussion in his secion. We assigned each individual counry a grade from A o E for each deerminan. Grade A means very sensiive o moneary policy compared o he oher counries and E means less sensiive o moneary policy. An overall grade based on a subjecive weighing of he individual grades is provided in secion 1.2.5, when we compare all individual ransmission channels. The real wage rigidiy measure from Table 1.1 is considered as he mos imporan deerminan of price rigidiies, unless here is a srong deviaion of he oher price flexibiliy indicaors. The grade of households consumpion is based on oal credi o households, correced for he adjusabiliy of households credi and he ne ineres received. A similar correcion for he adjusabiliy of credi and he effec of moneary policy on lending raes is done for firms. A firs inspecion of he able suggess ha i is likely ha many differences across counries offse each oher. We expec, however, a sronger han average ineres rae channel for Germany, Spain and Ausria and a weaker han average ineres rae channel for he Neherlands. 12 Table 1.6 Overview of he ineres rae channel Price rigidiies Indusrial srucure Governmen deb Households consumpion Firms Germany C A C B C Neherlands D D C D C Belgium B D A C C France C C C C C Spain A B C C B Ialy A C A E B Ausria C B C C A Finland D C D Ireland C D C Porugal D A C Sweden D D C B B UK D D D A C Noe: A and B indicae a higher han average sensiiviy of he counry o a moneary policy shock, C an average and, D and E a lower han average sensiiviy. 12 See also secion for a more exensive discussion of he subjecive weighing of he individual deerminans and a comparison of he differen channels across counries. 19

27 THE ASSET PRICES CHANNEL Overview Whenever a cenral bank alers he ineres rae, i simulaneously alers he araciveness of he own currency. An increase means ha foreign invesors shif par of heir invesmen o financial asses denominaed in his currency. As a resul, he currency appreciaes and lowers he compeiiveness of he domesic indusry (sep 4 of Graph 1.1). The lowering of he compeiiveness resuls in a deerioraion of ne expors (sep 5). Moneary conracion hus decreases naional oupu hrough an addiional channel (sep 6). In sum, moneary policy is more poen because i involves he exchange rae. This channel is ofen called he exchange rae channel. 13 A change in he exchange rae has also a direc effec on inflaion hrough he change in impor prices (sep 7). Changes in he moneary policy sance also affec he marke value of securiies, such as bonds and equiies (sep 8). The price of bonds is inversely relaed o he ineres rae, so a rise in ineres raes lowers bond prices, and vice versa for a fall in ineres raes. Higher ineres raes also lower oher securiies prices, such as equiies and houses. This is because expeced fuure reurns are discouned by a larger facor, so he presen value of any given fuure income falls. 14 On he one hand, he fall in he prices of hose asses reduces consumpion expendiure hrough wealh effecs (sep 11 and 12). On he oher hand, he fall in he prices lowers Tobin's q (defined as he marke value of firms divided by he replacemen cos of physical capial). The crucial poin of his discussion is ha here exiss a link beween Tobin's q and invesmen spending (Tobin, 1969). When q is low, he firm will inves less because he marke value of he firm is low relaive o he cos of capial. If companies wan o acquire capial when q is low, hey can buy anoher firm more cheaply. So, invesmen spending and oupu will be lower (sep 1). Implicaions for EMU The members of he euro area are all open economies. I is, however, only he exra- EMU rade of goods and services ha is imporan, because an aleraion of he euro doesn' affec inra-emu rade. We can consider he euro area as a large closed economy wih a weak exchange rae channel, comparable o he US (see Table 13 For he heoreical models underpinning his exchange rae channel, we refer o Obsfeld and Rogoff (1995, 1996) and Wickens (1985). 14 A more Keynesian view comes o a similar conclusion because i supposes ha a rise in he ineres rae makes bonds more aracive relaive o equiies, hereby causing he price of equiies o decrease (Mishkin, 1996). 2