BROOKINGS DISCUSSION PAPERS IN INTERNATIONAL ECONOMICS

Transcription

1 BROOKINGS DISCUSSION PAPERS IN INTERNATIONAL ECONOMICS No. 55 CHANGES IN EQUITY RISK PERCEPTIONS: GLOBAL CONSEQUENCES AND POLICY RESPONSES Warwk J. MKbbn and Davd Vnes February Warwk J. MKbbn s a professor n the eonoms department of the Researh Shool of Paf and Asan Studes at the Australan Natonal Unversty (ANU), a non-resdent Senor Fellow n the Eonom Studes Program at the Brookngs Insttuton, and a Professoral Fellow at the Lowy Insttute for Internatonal Poly n Sydney. Davd Vnes s a professor at the Ballol College and Department of Eonoms at Oxford Unversty, the Researh Shool of Paf and Asan Studes at the Australan Natonal Unversty (ANU) and a researher at the Centre for Eonom Poly Researh. The vews expressed are those of the authors and should not be nterpreted as refletng the vews of other ndvduals or nsttutons named above. Brookngs Dsusson Papers n Internatonal Eonoms are rulated to stmulate dsusson and rtal omment. They have not been exposed to the regular Brookngs prepublaton revew and edtoral proess. Referenes n publatons to ths materal, other than aknowledgement by a wrter who has had aess to t, should be leared wth the author or authors.

2 Changes n Equty Rsk Pereptons: Global Consequenes and Poly Responses ABSTRACT The urrent weakness n the global eonomy has generated a debate on the lkely outlook for the world eonomy and the approprate response for monetary poles. The world eonomy s urrently beng buffeted by a number of major shoks. A partular feature has been the large fall n equty markets n many ountres. In ths paper we use the MSG3 global eonom model to assess the mpat on the global eonomy of a sharp rse n the equty rsk premum n a number of ountres. In partular we examne whether a rse n equty rsk prema (or fall n produtvty whh has many smlar mplatons) s a shok to aggregate supply or aggregate demand. We explore the dfferene n the transmsson mehansm f the shoks our just n one ountry (.e. the Unted States) versus aross the OECD generally. We then assess the approprate responses of monetary poly to shoks of ths type and explore whether there are gans to oordnatng the monetary poly responses of the G7. JEL Classfaton 3 Warwk J. MKbbn Researh Shool of Paf & Asan Studes Australan Natonal Unversty ACT, Australa Davd Vnes Ballol College and Department of Eonoms Oxford Unversty Manor Rd Oxford, OX 3UQ, UK and Eonoms Studes Program and Researh Shool of Paf and The Brookngs Insttuton Asan Studes 775 Massahusetts Ave NW Australan Natonal Unversty Washngton D.C. 36 USA ACT, Australa and Centre for Eonom Poly Researh

3 . Introduton The ontnung weakness n the global eonomy has led to a serous debate about auses and remedes. On the one hand have been those who seek to treat t as a downturn smlar to other reent reessons, n 975, 98-8, and These epsodes an be thought of as temporary demand shoks. Satsfatory management of these would requre a speedy ut n real nterest rates to stmulate demand and lead the eonomy bak lose to the NAIRU. On the other hand there have been those who treat the urrent slowdown as a longer term dsrupton of supply, aused by an over-aumulaton of partular knds of aptal, a reduton n the rate of produtvty growth, an nrease n the rsk premum appled to equty market nvestment or a burstng of an equty market bubble. These knds of shoks are muh harder to manage. The experene of the earler downturns may provde less gudane to poly makers than many ommentators urrently argue. In, the onsensus vew appeared to be one where the slowdown would be a relatvely shortlved V shaped one, wth a qute rapd reovery towards full employment, a return of nflaton to ts general target range, and a rse of global nterest rates towards long term levels. But ths poston depends at least n part on a vew that the slowdown has been aused manly by a short-term demand reduton, whh an be partly smoothed by monetary poly, and possbly also by fsal poly. The alternatve vew s that the downturn s nstead a onsequene of a reduton n the growth of supply potental, ether through a revson of produtvty growth, or through the adjustment to exess nvestment n nformaton tehnology or through great unertanty n the world n the wake of the September terrorst attaks, as well as a delne n demand through low nvestment and delnes n equty wealth. Suh redutons n demand, whh have an underlyng supply-sde ause, mght be more dffult to smooth. In other words, the urrent downturn mght be a shorter-term demand-sde feature, whh an be managed by poly adjustment, or a more strutural downturn emergng from longer-term busness yle features, whh s more dffult to deal wth through monetary poly response. Ths paper wll address ths dfferene of vew. We beleve that the latter vew s the orret nterpretaton of the auses of the downturn, so that the downturn has underlyng supply-sde auses due to the endng of the long boom of the 99s. In ths paper we summarze the supply sde shok through foussng on an nrease n the rsk premum on equtes (but the results are very smlar to a delne n eonomy wde TFP growth n that the attratveness of aptal aumulaton s dmnshed). But we want to onnet the supply shok to the arguments that the world s urrently experenng negatve demand shoks. The lnk between the two approahes an be demonstrated

4 through a model whh learly dentfes the demand effets of suh a supply sde shok, through ts effets n ausng a reduton the demand for aptal, and thus n nvestment. We want to see the extent to whh suh an outome, omng from the supply sde, mght nevertheless be possble to smooth wth approprate monetary poly on the demand sde. We wll fous not only on poly-makng n the US but also on the extent to whh poly-makng n Europe, and n the rest of the world, an omplement that n the US, and thus on the mplatons of the downturn for poly-makng n Europe, Canada and Japan. We wll also desrbe the mplatons of ths shok for the developng world, and n partular for regons where the shok does not our but whh are affeted by the transmsson of ts onsequenes. The paper uses the MSG3 mult-ountry model. The key feature of ths model s that t ombnes the modern ntertemporal optmzaton approah to modellng eonom behavour (as found n Blanhard and Fsher, 989, and Obstfeld and Rogoff, 996) wth short-run rule-of-thumb behavour. In dong ths t brngs together real busness yle models one the one hand, whh have a fully artulated analyss of forward-lookng produers and onsumers, and modern maroeonometr models on the other hand, whh desrbe the effets of demand downturns n the fae of wage (and pre) stkness. The key feature of the model s the way t models asset markets and the ntegraton of these markets wth real eonom atvty, n partular the meshng together of equty markets and bond markets, joned by hghly moble nternatonal flows of fnanal aptal, and short-run fxty of physal aptal. Other key features of the models are set out n seton 3. Our general fndngs are strkng and have sgnfant mplatons for present day debates on maroeonom poly. We fnd that n response to an nrease n global rsk (or delne n TFP growth), real wages must fall a great deal over tme, ompared wth what they would have been, beause of the reduton n the aptal stok per worker, whh ours beause of the nreased rskness of aptal. The real nterest rate must fall to ause demand to rse to make up for the very large reduton n nvestment whh ths nreased rskness provokes. In the short run, gven that aggregate supply falls slowly The MSG3 model s a setor verson of the G-Cubed model developed n MKbbn and Wloxen (998). It s a desendant of the approahes of the MSG model n MKbbn and Sahs (99) and the Jorgensen Wloxen (99) Model. See Bakus et al (99).

5 3 but nvestment falls sharply, onsumpton must rse to make up for the reduton n nvestment; the onsequene s surprsngly strong onsumpton, of the knd that has worred many ommentators as revealng an mbalane we see t as a natural onsequene of the knd of shok that we are studyng. Unemployment emerges, even when monetary poly s expansve enough to ndue a rse n onsumpton enough to use up some of the spare resoures. Furthermore a sgnfant rse n the pre of housng as a onsequene of the fall n nterest rates and a substtuton out of equtes nto other asset lasses s a key part of the way n whh onsumers wealth s rased so as to ndue the neessary nrease n onsumpton. A orollary of ths s that there s also a boom n the produton of housng. Agan these effets on the housng market have been seen by many ommentators as an mbalane we see them as a onsequene, too, of the shok whh we are studyng. We fnd that expansonary monetary poly by the monetary poly-maker n eah ountry an partly redue the output loss and resultng unemployment, but only by relatvely lttle. The hange n these varables affeted by poly aton, s small relatve to the underlyng adjustment n response to the shok, but they are nonetheless sgnfant n the very short term. Smlarly we fnd that nternatonal ooperaton over monetary poly between poly-makers an further assst wth redung the output loss and the resultng unemployment. Model Struture The MSG3 model s a dynam ntertemporal general equlbrum models that attempt to ntegrate the best features of tradtonal CGE models, real busness yle models and Keynesan maroeonometr models. A setor stylzed verson summarzng the key features are outlned n Appendx A. A summary of the key features of the model s ontaned n Table and the ountry and setor overage (as well as assumpton about exhange rate regmes) s gven n Table. The fve man features of the model are as follows: (a) The model s based on explt optmzaton by the agents (onsumers and frms) n eah eonomy. Where these models dffer from stat CGE models3 s n the assumpton of ntertemporal optmzaton by eonom agents, subjet to explt ntertemporal budget onstrants. Thus, n 3 See Hertel (997)

6 4 ontrast to stat CGE models, tme and dynams are of fundamental mportane n the MSG3 models. Ths makes ther ore theoretal strutures lke that of real busness yle models 4. (b) In order to trak the maro tme seres, the behavour of agents s modfed to allow for short-run devatons from optmal behavour, ether due to myopa or to restrtons on the ablty of households and frms to borrow at the rsk-free bond rate on government debt. For both households and frms, devatons from ntertemporal optmzng behavour take the form of rules of thumb, whh are onsstent wth an optmzng agent that does not update predtons based on new nformaton about future events. These rules of thumb are hosen to generate the same steady state behavour as optmzng agents so that n the long run there s only a sngle ntertemporal optmzng equlbrum of the model. In the short run, atual behavour s assumed to be a weghted average of the optmzng and the rule of thumb assumptons. Thus, aggregate onsumpton s a weghted average of onsumpton based on wealth (urrent asset valuaton and expeted future after tax labour nome) and onsumpton based on urrent dsposable nome. Ths s onsstent wth the eonometr results n Campbell and Mankw (987) and Hayash (98). Smlarly, aggregate nvestment s a weghted average of nvestment based on Tobn s q (a market valuaton of the expeted future hange n the margnal produt of aptal relatve to the ost) and nvestment based on urrent frm proft-nome. () As n all poly-relevant maroeonom models (but unlke n many CGE models), there s an explt treatment of the holdng of fnanal assets nludng money. Money s expltly ntrodued nto the model through a restrton that households requre money to purhase goods. Ths assumpton gves money an explt role. (d) The model allows for short-run nomnal wage rgdty (by dfferent degrees n dfferent ountres) and therefore allow for sgnfant perods of unemployment dependng on the labour market nsttutons n eah ountry. Ths assumpton, when taken together wth the explt role for money, s what gves the model ts maroeonom haratersts. (Here agan the model s assumptons dffer from the standard market learng assumpton n most CGE models.) (e) The model dstngushes between the stkness of physal aptal wthn setors and wthn ountres and the flexblty of fnanal aptal, whh mmedately flows to where expeted returns 4 Se for example Bakus et al (99)

7 5 are hghest. Ths mportant dstnton leads to a rtal dfferene between the quantty of physal aptal that s avalable at any tme to produe goods and serves, and the valuaton of that aptal as a result of desons about the alloaton of fnanal aptal. As a result of ths struture, the models ontan rh dynam behavour, drven on the one hand by asset aumulaton and, on the other hand, by wage adjustment to a neolassal steady state. A key pont to note s the way household durables, espeally housng are norporated nto the model. Ths s set out n detal n Appendx A. Households are assumed to onsume serves from a stok of durables, whh largely onssts of housng (from ths pont we wll refer to the entre stok of household durable as housng as a short-ut). Investment n housng s determned as part of the onsumer s optmzaton problem under the assumpton that housng aptal aumulaton s subjet to rsng margnal osts of adjustment (smlar to that for prvate aptal n produton). In other words t takes tme for the housng stok to hange n response to a hange n the demand for aptal serves. Thus surprse nrease n demand wll lead to aptal gans to exstng owners of housng but gradual adjustment on the supply sde. Households buy aptal goods from the aptal goods produton setor and add them to the exstng stok of aptal over tme. Households then onsume a flow of serves from the stok of housng over tme. In summary, the model embodes a wde range of assumptons about ndvdual behavour and empral regulartes n a general equlbrum framework. The omplex nterdependenes are then solved out numerally. It s mportant to stress that the term general equlbrum s used here to sgnfy that as many nteratons are possble are aptured, not that the eonomy s n a full market learng equlbrum at eah pont n tme. Although t s assumed that market fores eventually drve the world eonomy to a neolassal steady state growth- equlbrum, unemployment does emerge for long perods due to wage stkness, to an extent whh dffers between ountres due to dfferenes n labour market nsttutons. The equatons for a stylzed two-ountry/two-setor verson of the model are presented n Appendx A. The theoretal bass of the model an be found n MKbbn and Wloxen (998) and onlne at

8 6 3 The Shok(s) to be studed It s mportant to desrbe exatly what shok we are mplementng n the model. We defne the equty rsk premum as the term µ n equaton (ths s equaton n Appendx A). In ths equaton s the setor number and t s perod t. Equaton () shows that the hange n λ (whh s Tobn s Q adjusted by the relatve pre of nvestment goods see Appendx A) s related to the real nterest rates adjusted by depreaton and the rsk premum, the margnal produt of aptal and a term (quadrat n the rate of nvestment J/K) apturng the ontrbuton to the margnal ost of aptal of the adjustment ost of aptal. dλt dqt φ I J t () =(rt + δ + µ t ) λt ( τ ) P ( 4 ) t τ dt dk t K t Equaton () an be nterpreted as an arbtrage ondton relatng returns on equty to returns on other real and fnanal assets and the equty rsk premum. Thus we assume that the equty rsk premum s the exess return on aptal relatve to the real nterest rate adjusted for depreaton and allowng for adjustment osts, requred to ndue people to hold equtes relatve to government bonds wth a real nterest rate r. A rse n the equty rsk premum would mean that ether the margnal produt of aptal must rse n the long run or real nterest rates must fall or a ombnaton of both Equaton () an also be ntegrated and wrtten as the equal to the present value of the after-tax margnal produt of aptal n produton (the frst term n the ntegral) plus the savngs n subsequent adjustment osts t generates. () λ t = ( τ ) p t dq + ( τ ) p φ Jˆ k * I 4 dk ˆ Jˆ, kˆ e ( R( s) + δ + µ )( s t) ds As mentoned above, ths demonstrates that λ s smlar to Tobn s margnal Q onept (see Appendx A for the transformaton by the relatve pre of nvestment goods, requred to λ requred to generate Tobn s Q).

9 7 There are a number of ways to nterpret a rse n the equty rsk premum. The frst s a rse n the urrent and expeted rsk premum bak to longer-term values away from the low values reahed n the 99s at the peak of enthusasm of the new Eonomy. Ths an be nterpreted as the burstng of an equty pre bubble. Ths s potentally a permanent shok f t was expeted that the rsk premum n the 99s would ontnue beause of fundamental hanges n the world eonomy but now these expetatons have been fundamentally revsed. Another nterpretaton s that there has been a general rse n unertanty beause of September and other global events, whh wll not neessarly be permanent. To over these varous nterpretatons we onsder a persstent (but ultmately temporary) rsk shok as well as a permanent shok. The persstent temporary shok reflets our most lkely guess and s the fous of most of the smulaton analyss. We albrate the sze and duraton of the shok n our exerse as follows. Evdene suggests that the level of the rsk premum had been as hgh as 8 perent up untl the 98s, and that t steadly fell untl the late 99s onvergng towards zero. In the smulatons onsdered n ths paper, we take the nrease n the rsk premum as 5 perent (e 5 bass ponts), whh reverses 5/8 of the fall, whh happened n the 99s. In most of the smulatons we take ths as dyng out over ten years. The rsk premum s thus 5 bass ponts n the frst year, 45 bass ponts hgher n the seond year, 4 bass ponts hgher n the thrd year, and so on down to bass ponts n the th year. For purposes of omparson we also ompare the outomes wth what would happen f ths rse n the rsk premum were sustaned forever. The shoks ommene n the year. The reason for the hoe of a shok lastng ten years s that t lasts long enough to have a sgnfant effet on nvestment and the aptal stok, wth onsequental effets on the supply apaty of the eonomy. If the shok lasted for only a year or two then ths would not be the ase. And yet the shok gradually des out and so s not a permanent hange: we are not analysng results whh are the onsquenes of permanent hanges n the stok of aptal. But for omparson we study what would happen f there were a shok of ths sze whh was permanent. We apply ths shok to all developed OECD ountres to study the effets of a rse n the rsk premum n all developed ountres. We also study below what happens f ths shok happens only n the US. The OECD wde shok s symmetr wthn the OECD but asymmetr relatve to the developng ountres, whh are also n the model. Foussng on results for Chna gve some nsghts on the adjustment of ountres outsde the OECD. The results for a US only shok are

10 8 asymmetr wthn the OECD, whh gves addtonal nformaton on the nature of the transmsson of the shok globally. 4 An Outlne of Transmsson of the Permanent and the Temporary, Symmetr and Asymmetr Shok The relevant features of the model have been explaned above. We begn by dsussng the effets of a permanent rse n the equty rsk premum. Ths provdes the bakground for a dsusson at the end of ths seton of the temporary shok. In Seton 5 we dsuss the approprate maroeonom poly responses to a temporary rse n the equty rsk premum n detal. For both the permanent and temporary ases dsussed n ths Seton, monetary poly follows the smple rule of a fxed money supply n the US and all other ountres, exept n the EMU where there s a fxed money supply for EMU ountres and a monetary unon (e a fxed exhange rate) between the member ountres, and n Chna, OPEC and non-ol developng ountres whh are assumed to be peggng to the $US. 4. Permanent Shok We wll see that the mplatons of the permanent shok of 5 bass ponts are very sgnfant. These results demonstrate how mportant hanges n dsount rates and rsk prema an be for eonom atvty. It s mportant to stress that the sale of the permanent shok s not meant to apture what has happened n the world eonomy but nstead s desgned to reveal the underlyng fores at work n our study below of the temporary shok (whh s saled to represent what we beleve s a reasonable stylzaton of what has been happenng reently). Effets on the US We begn by dsussng the effets of a permanent symmetr shok appled to all OECD ountres. Results for a permanent rse n the equty rsk premum and a temporary rse n the equty rsk premum n all OECD ountres are ontaned n Fgure. Ths fgure shows the mpats on the Unted States (a ountry experenng the shok dretly) and Chna (a ountry not dretly experenng the shok) as examples. Chna as wth all developng ountres, s assumed not to dretly experene the shok.

11 9 In the fae of a permanent rsk shok n the ndustral ountres, frms gradually adjust the stok of physal aptal to ts lower desred level, whh depends on (a) the expeted produtvely of aptal and thus on the extent to whh t s expeted to yeld returns to aptal, and (b) on the rsk premum whh nvestors apply to these returns. The demand for aptal depends on Tobn s q, whh n turn depends on the rato of the future hanges n the margnal produt of aptal and real nterest rates. In the steady state, Tobn s Q relates margnal produt of aptal to the sum of the real nterest rate, the rate of depreaton of aptal, and the rsk premum. Our shok s to nrease the rsk premum n the way desrbed above, and so leads to a lower q n the short run dretly through a hgher rate for dsountng future proftablty of frms. As a result nvestment falls. The hange n the equty rsk premum wll also have dret effets on onsumpton. There s a delne n the value of aptal refleted n a delne n equty values. The expeted hange n the future aptal stok wll ause a reduton n the value of human aptal (the stok of future earnngs expeted to be derved from the applaton of aptal to labour, usng a redued amount of aptal). To the extent that onsumers are forward lookng wth respet to future labour nome and together wth the fall n the stok market value of wealth ths wll tend to depress onsumpton. However, there s a rse n the value of other assets suh as housng (dsussed n detal below). To the extent that households are lqudty onstraned they wll also tend to redue onsumpton beause nome has fallen as a result of lower nvestment. The mmedate mpats of ths shok wll thus have two omponents. (a) There wll be a reduton n aggregate demand durng the perod durng whh aptal s adjustng to the new lower level. (b) There s a gradual reduton n supply, aused by the reduton n the stok of aptal as a result of the lower level of nvestment. The reduton n the aptal stok s very gradual (although the valuaton through equty pre falls s sharp), and the resultng reduton n supply wll therefore be very gradual, too. How do these mpats mpnge on the eonomy? Frst, we fnd that the reduton n demand runs ahead of the reduton n supply. Ths means that there s a long-lastng need, by real nterest-rate falls, to promote demand to lmtng the reduton n nvestment, and to ause onsumpton to rse

12 (even although there are the above-desrbed mpat effets ausng t to fall).5 Seond, pres and wages are stky, beause wages are set one perod n advane. As a result, hanges n demand desrbed above, relatve to hanges n supply, move the eonomy away from the full utlsaton of resoures. In partular the demand for labour wll fall, and there wll be unemployment. Fgure presents some of the outomes for the US. (We do not show the outomes for any other developed ountry beause the results are very smlar we dsuss ths below). In the followng dsusson the sx harts wthn eah fgure are referred to as 6a through 6f wth 6a n the top left hand orner, 6b n the top rght hand orner and 6f the bottom rght hand orner of the fgure. Fgure e shows the large and long lastng falls n the real nterest rate that result from ths permanent rsk shok.6 The speed at whh aptal s adjusted down to the new lower level depends rually on osts of adjustment. Fgure b shows that, even wth suh a large reduton n the real nterest rate, a permanent nrease n the rsk premum of 5 perent would auses a very very large reduton n the aptal stok n the US namely 9 perent of GDP (whh s equvalent to approxmately 4 perent sne the aptal to output rato s approxmately.8) and that the reduton of the aptal stok to ts new lower level takes years. Ths large fall n the aptal stok s what we expet. Although we start out of steady state, wth the aptal stok gradually adjustng towards steady state, t was dong so along a path onsstent wth the expeted margnal produt of aptal, the long run real nterest rate (5%) and depreaton (%); now t must adjust to a new steady state n whh the rsk premum has rsen by 5% (5 bass ponts). That s an nrease n the margnal produt of aptal of one thrd, and t auses a reduton n the stok of aptal of 5 We wll dsuss the determnants of, and the behavour of, onsumpton below. 6 It s mportant to explan the effets on the nterest rate. The rsk shok wll lower the demand for nvestment and lead to a redued level of aptal, and a lower long run level of nvestment for replang the depreaton whh ours. In a smple Ramsay model somethng whh lowered the stok of aptal would lead to a lower human aptal and lower onsumpton, just suh that lower savngs mathed the lower nvestment, and just suh that the fall n nvestment and onsumpton together just mathed the fall n output, wthout any hange n the nterest rate. However thngs are dfferent here. Frst, the reduton n the stok of aptal s only gradual, so that n the short-to-medum run those onsumers who are forward-lookng see through to tme when the aptal stok, and output and wages, are already lower than they are now. Ths wll tend to redued onsumpton and nrease savngs relatve to urrent level of output, lowerng the short run. Seond, even n the long run the nterest rate s affeted by the rsk premum. At the ntal nterest rate, the rsk premum wll mean that the stok of aptal s lower than t was before, wth a lower aptal to output rato, and so a lower proporton of output beng requred to over depreaton. If the nterest rate stayed as before, the proporton of nome saved would reman as before [footnote - for forward-lookng onsumers that seems a orret lam about ntertemporal hoe], [n long run bakward lookng onsumers do not save, but the proporton of suh onsumers s unhanged]. So savngs would be hgher than nvestment. Hene proporton of the wedge mposed by the rsk premum must fall on the nterest rate.

13 more than twe that amount beause the produton funton s not Cobb Douglas and has an elastty of substtuton of less than unty. Over tme the hanges n aptal hange the supply potental of the eonomy. Fgure a shows what happens, as a result of the shok to employment. If wages and pres were perfetly flexble there would not be unemployment lke ths. Fgure a shows that, n response to ths shok, there would be a reduton of employment, whh reahes a maxmum of lose to perent, before real wage flexblty elmnates t. Eventually unemployment s removed by a fall n the real wage of lose to 9% (relatve to baselne). As we wll see later, there s a role for monetary poly n ntervenng to smooth the unemployment onsequenes of these shoks. Fgure shows the negatve effet on GDP n the US. Although n the short run ths s nfluened by unemployment, eventually t s determned by the reduton n supply aused by the reduton n the aptal stok. We see that, wthn years, the reduton n output s very large at 3 perent (relatve to base). It s useful to ompare ths wth the fall n the aptal stok, whh as noted above falls lose to 9 perent of GDP, whh s equvalent to a perentage fall of approxmately 5 perent. The rato of the proportonate fall n output to the proportonate fall n aptal s about 45 perent beause aptal s only one of the fators of produton, whh wll n equlbrum beome more labour ntensve. The rato s more than the aptal share of about one thrd, whh s what s what would be expeted f the produton funton was Cobb Douglas Overall we an see that suh a shok, f t were permanent, would have a very large effet on the US eonomy as well as on other OECD eonomes (not shown here). There would be a long-lastng fall n output. Ths would be as a result of a very long lastng reduton n the nvestment demand for aptal, and a permanent reduton n ts supply. Effets n other OECD ountres If suh a permanent shok were onfned to a sngle eonomy then ths reduton n the real nterest rate wll have sgnfant onsequenes for the exhange rate, and onsequently mplatons for the trade balane. If the OECD eonomes are suffently symmetral then there are no large movements of ths knd (see Fgures d and f). There are some nter-ountry dfferenes wthn Europe whh are dsussed below.

14 Effets n Developng Countres Fgure also plots the outomes for Chna, whh s not dretly experenng the OECD-wde shok. In the fae of the sustaned fall n world nterest rates, Chna experenes a sgnfant nrease n nvestment and rse n the aptal stok and a sgnfant nrease n GDP (5% ntally). Ths also auses a long-lastng boom to employment n Chna. The Chnese real exhange rate appreates as aptal flows nto Chna and real nterest rates fall sgnfantly (note than the nomnal exhange rate s pegged to the $US). The Chnese eonomy runs a large urrent aount deft wth advaned ountres refletng the large aptal nflow. However there are qukly dmnshng returns to the addtonal aptal beause the adjustment osts are proportonal to the rate of nvestment. The transmsson to Chna (and other developng ountres) shows that nreased growth n unaffeted eonomes s one way n whh the shoks to the world aused by rsk shoks are moderated. The real exhange rate of Chna eventually depreates sgnfantly, to prevent the urrent aount deft, aused by hgher growth, from ausng external asset labltes to explode wthout lmt. The real depreaton and hgher nome enable the external labltes to be serves through rsng exports relatve to mports over tme. 4. Temporary Shok Fgure also shows results for a temporary rsk shok, of the same sze (5 bass ponts), but whh gradually fades out over years. The real nterest rate falls n the frst three years by nearly three quarters of ts fall n the permanent ase. Although the aptal stok eventually returns to base, after more than years, t falls n the frst ten years by more than 5 perent of GDP. GDP s down by about 5 perent after ten years, before reoverng. Overall we an see that suh a shok, f t were temporary rather than permanent, would stll have onsderable effets on the global eonomy. The outomes shown would not atually ause a reesson (hange n growth rates an be alulated from the slopes of the GDP graph whh s devatons from baselne). But they would ause a very long-lastng reduton n the growth rate of output of approahng half a perent per year, for up to years. Ths s as a drawn out (over ten years) reduton n the nvestment demand for aptal, and a permanent reduton n the supply apaty of the eonomy for nearly years. Employment s lose to.5 perent lower durng the

15 3 transton. 4.3 US versus OECD-Wde Shok Fgure ontans the results for the Unted States and Chna omparng the results for the temporary shok n the OECD wth the US only shok. The results for the temporary OECD-wde shok are the same as that for the temporary shok n fgure, however removng the permanent shok from the graphs hanges the sale sgnfantly. The S next to eah ountry ndates a symmetr shok (OECD wde) whereas the A sgnfes an asymmetr shok (US only). The underlyng eonom story of the results has already been outlned above. The man pont of these results n Fgure s to show that the effet on the US s overwhelmngly the hange n the US rsk premum. The results dffer more for the US urrent aount, whh under a US only shok mproves by three tmes the amount ompared to the OECD wde shok. Smlarly the real exhange rate hanges are three tmes the magntude when the shok only ours n the Unted States ompared to the OECD-wde shok. It s not surprsng that the asymmetr shoks show up n the blateral varables between ountres. For Chna t s lear that the mpat on GDP and the aptal stok s more than double under the OECD wde shok than under the US only shok. The realloaton of global aptal s that muh larger when all ndustral eonomes experene the rsk shok ompared to when the US only experenes the shok. Also the movement n the Chnese real exhange rate s more than double for the OECD wde shok. 4.4 Detaled maroeonom adjustment wthn the Unted In fgure 3 we present more detaled results on the adjustments that our wthn the US eonomy for the OECD wde temporary rsk premum shok. We assume n ths and earler fgures that the smple monetary regme n plae s a fxed money supply rule. The remander of the results n ths paper wll examne optmal money rules but for dagnost purposes t s useful to begn wth the fxed money stok assumpton of the bas IS/LM framework. Fgure 3a shows the behavour of the major maroeonom aggregates, all for the US. Fgure 3b

16 4 shows the behavour of major assets value of aptal (.e. qk), government bonds, foregn assets, human aptal, and wealth. Fgure 3 shows what happens to employment and nflaton of both onsumer and produer pres and Fgure 3d shows outomes for the levels of wages and onsumer and produer pres; t thus depts a large fall n real wages of around 9 perent; durng the perod that aptal s down t s ths whh pres workers bak n to jobs and redues the resultng unemployment. Fgure 3e shows asset pres nludng the nterest rate and Tobn s Q for aptal used n produton and Tobn s Q n housng aptal. Fgure 3f shows the real exhange rate, all for the US. We have already explaned the overall maroeonom adjustment story to the hange n the equty rsk premum. The requred return to aptal rses due to the hgher rsk premum on equtes. Ths auses a large fall n the US share market, summarzed n the Asset Pre Chart under Tobn s Q. Note n ths hart that although the stok market falls sharply there s an equally sharp rse n the Tobn s Q for housng. People substtute out of equtes nto other assets suh as government debt drvng up the pre of debt and therefore drvng down the real nterest rate. They also substtute nto onsumer durables (prmarly housng) n ths model. The ombnaton of ths substtuton plus the fall n real nterest rates nreases the Q for housng further. Beause the stok of housng, lke the stok of physal aptal s ostly to adjust, the result s a rse n the pre of housng and a rse n the return to nvestng n housng. Ths lasts for 5 years even though the shok s temporary. The role of onsumpton and of housng Fgure 3a also shows that onsumpton atually rses slghtly as a result of the shok. But we have already noted that forward lookng onsumers wll see the future produtve power of labour as beng lower as a result of the fallng aptal stok, whh wll lead to a fall n the value of real wages. Those onsumers who are forward-lookng wll see the fall n future nome and the value of equtes and wll ut onsumpton aordngly. How then s the rse n onsumpton brought about n the fae of ths by the fallng rate of nterest? The fall n the real nterest rate has two onsequenes for wealth. () Human aptal rses (See Fgure 3b). Ths happens, even despte the reduton n the real wage (See Fgure 3d), beause a fall n the nterest rate nreases the dsounted value of any real nome.

17 5 () The value of exstng houses rses. Ths s refleted by the rse n the value of Tobn s q for housng (see Fgure 3e). It happens for two reasons. Frst the rse n the rsk premum through the fall n the real nterest rate, nreases the present dsounted value of housng serves, ausng the value of exstng houses to rse. Seond the relatve pre of produng new housng falls. Ths s beause the pre of nvestment goods falls (relatve to the pre of other produed goods), beause of the reduton n the demand for nvestment. That makes new housng heaper, beause t s produed n a way whh s ntensve n the use of nvestment goods, and that nreases the value of exstng housng. The rse n human aptal and n the value of housng, together mean that the value of overall wealth ntally goes up (See Fgure 3b). Ths happens even though the produtve potental of the eonomy has fallen, and even although the value of fnanal wealth wll have fallen sgnfantly wth the fall n the value of the stok market (see the fall n Tobn s q (non-energy) n Fgure 3e). The resultng rse n the value of wealth an then explan the rse n onsumpton on goods. Furthermore, beause of the nrease n Tobn s q for housng there s an nrease n expendture on housng tself, whh s part of onsumpton. Thus the fall n the nterest rate atually more than overturns the negatve effets, mentoned above, of fallng aptal stok leadng to fall n onsumpton (beause forward lookng onsumers foresee fallng output). That s why onsumpton goes up even although the produtvty of the eonomy at least for a perod of tme wll atually have fallen. The role of a rse n house pres, and a rse n expendture on housng, s essental n lmtng the negatve onsequenes of ths shok. In summary, Fgure 3a shows a sharp fall n real nvestment but a rse n prvate onsumpton. Ths helps dampen the GDP delne, whh s largely drven by fallng nvestment. The GDP delne s also partly offset beause of a real depreaton of the $US s effetve terms. Europe and the rest of the OECD are also experenng the shok thus most of the hange n real exhange rates s relatve to developng ountres. If the shok only ourred n the US the exhange rate hange would be larger and the trade balane mprovement s larger helpng to dampen the negatve effets of the shok.

18 6 An mportant nsght from these results s that the more the equty rsk shok s a global event the greater the negatve mpats on the affeted ountres beause there s less ablty to buffer the shok through balane of payments adjustment. Also the role of other assets n the eonomy, espeally housng s mportant n understandng the transmsson of the shok wthn an eonomy. 5 Maroeonom Poly and Outomes In ths seton nvestgate the effets of a polymaker pursung not a fxed money supply, but nstead adjustng nterest rates so that the outome optmses an ntertemporal loss funton7 shown n equaton (3). (3) W t = Σ t= δ t τ t 'Ω τ t Assume that a polymaker hooses a vetor of ontrol varables (Ut) to mnmze (3) subjet to the struture of the world eonomy as gven by the MSG3 model. Where τ s a vetor of targets (n the urrent ase ether nflaton alone or nflaton and employment) relatve to desred levels, Ω s a matrx of weghts on eah target nluded n the τ vetor, and the subsrpt refers to ountry. It s mportant to note that we assume the poly maker ares not about the gap between atual output and apaty output, but about the gap between atual and desred employment. We use employment rather than apaty output beause our shok s makng apaty output hange endogenously, and to fall. A poly maker who pursued the prevously projeted apaty level of output would reate an nflaton bas. Whereas employment wll return to ts natural rate wthn the model over tme and ths elmnates the nflaton bas. A seond, mportant aspet of the objetve funton s that we make the nflaton part of the objetve funton, produer pre nflaton rather than onsumer pre nflaton. We do ths beause

19 7 of the strong effets the hange n the pre of housng serves have on the measure of onsumer pre level n ths model. The produer pre ndex s loser to the onept of nflaton used n the onventonal maroeonoms lterature as t turns out the partular shok we are examnng n ths paper leads to a substantal hange n the pre of housng serves whh monetary authortes may not want to respond to. Housng serves and ther prng s somethng less than fully aptured n most atual measures of the CPI and we do not want to entangle our results wth ths effet. () Unted States Fgure 4 shows results for the Unted States under the ase where monetary authortes put all weght on nflaton (INFP) relatve to the ase where a weght of s put on nflaton a on employment devaton from desred (INFEMP). If the polymaker puts double the weght on nflaton as on employment and optmses monetary poly on ths bass, the real outomes for the US eonomy are very smlar over tme almost dental - to those wth a fxed money supply. Although n the ntal years there are sgnfant dfferenes. The sale of the graphs and the long tme perod tend to dsguse the monetary mpats n the very short term. Note n Fgure 4b that produer pre nflaton s targeted exatly n the INF regme but s allowed to flutuate relatve to desred when allowane s also made for devatons n employment from desred (fgure a). If the poly maker puts an nfnte relatve weght on nflaton (.e. two relatve to zero on employment) we see that output and employment outomes of the shok an be made worse.e. more severe. Employment falls more (.e. unemployment rses more), beause monetary poly s tghter. However note that nomnal nterest rates are lower under the nflaton target. Ths s beause when employment s gven some weght n the targetng, nflaton s hgher and thus the nflaton premum n the nomnal nterest rate s hgher. The key pont s that the real nterest rate s hgher under nflaton targetng and therefore monetary poly s tghter n ths ase. We see two thngs from ths seton. The frst of these s as expeted, and obvous. A looser monetary an redue the unemployment osts of the shok. In more detal, we see that ths s a shok whh, durng the perod n whh ts 7 Reent studes followng ths approah usng the MSG model to fous on poly ssues wthn Europe nlude Haber at al (), Haber et al () and Nek et al ()

20 8 negatve effets on the aptal stok reman present, requres a sgnfant fall n the real wage. To brng these about by wage uts, wthout any rse n the pre level, requres a sgnfant amount of unemployment. The unemployment requred to produe the wage uts an be redued by havng some nflaton. The seond result s less obvous, and mportant. Ths s that a more expansonary monetary poly, even a very expansonary monetary poly does not seem to be able to make very muh dfferene to the level of unemployment along the long perod of transton but does make a dfferene n the very short run. The downturn aused by the rsk shok seems to be n large part a downturn lower output and lower employment - that s hard to hange greatly by monetary poly. To a onsderable extent t s a supply sde phenomenon whh annot be hanged muh by monetary poly and omes from the reduton n supply potental aused by the reduton n the stok of aptal. Instead t s due to the fat that wages are suffently flexble n ths model that a hange n monetary poly, although t an have an effet for a few years before t s refleted n wages, annot do so for more than a few years, no matter how large t s beause any future hanges wll qukly feed nto wages. () Frane To llustrate the transmsson of the shok to other ountres we next present results for an llustratve European ountry, Frane. Fgure 5 ontans results when the shok s an OECD wde shok nludng n Frane but wth the optmal poly response (wth the European wde nflaton and employment n the European Central Bank s objetve funton). Ths an be ompared wth Fgure 6, whh also ontans results for Frane but n the ase where the shok only ours n the Unted States. The maroeonom adjustment n Frane when there s an OECD-wde rsk shok s very smlar to that for the Unted States that has already been analyzed n fgure 3. The major dfferene s the smaller hange n the Euro relatve to the dollar and therefore the less offset there s for Frenh GDP from an mprovement n net exports. Indeed exports fall muh more n Frane than n the US. The nomnal effetve exhange rate hanges lttle n Frane ompared to the US. Also there s a muh larger rse n pres n Frane sne the ECB s targetng European wde varables t does not plae a great deal of weght on Frenh nflaton. Ths mples a rse n nomnal wages n Frane

21 9 ompared to a fall n nomnal wages n the US. There s also a muh more persstent rse n pres n Frane beause of the ECB reaton. When the shok ours n the Unted States only, Frane faes a very dfferent type of shok. Reall that the US shok redued US real nterest rates and auses a global realloaton of aptal. Fnanal aptal flows nto Europe as a whole, whh appreates the Euro. Asset pres are also drven up n Europe and real nterest rates fall due to the nflow of aptal from the Unted States. Asset values also rse and there s a physal aumulaton of aptal. Employment rses above base as the European eonomy experenes a postve supply shok due to the aptal realloaton, but a negatve foregn demand shok due to the slowdown n the US eonomy. The net effet s postve for Europe both n terms of hgher employment as well as lower nflaton. Ths ontrasts wth the stagflatonary effet of the own ountry shok. If s worth pontng out that there s no dret ontrol over Frenh nflaton beause the ECB s targetng European wde nflaton and employment. Therefore pres and wages n Frane rse over tme. 7 Internatonal Cooperaton between Poly Makers We fnally study nternatonal ooperaton between polymakers. We solve for the nonooperatve (tme onsstent) Nash equlbrum n whh poly makers solve the objetve funton summarzed n the prevous seton takng as gven the behavour of other polymakers. We also solve the ooperatve equlbrum where a global polymaker weghts eah ountry s objetve funton by that ountry s GDP relatve to the group and sets poly approprately. Ths follows the tehnques outlned n MKbbn and Sahs (99) An mportant nnovaton here s, as n the prevous seton, makng the poly makers are not about the gap between atual output and apaty output, but nstead about employment. We dd ths n the prevous seton to avod the poly makers ntrodung an nflaton bas. To do so s doubly mportant n ths seton beause we know from Rogoff (985) that f polymakers have an nflaton bas then ooperaton between them an be welfare redung sne t enables them to avod depreatng the real exhange rate when they expand ndvdually and thus strengthens the temptaton to eah to explot nflaton bas by attemptng to over-expand.

22 In suh a setup we expet there to be a ooperaton ssue. In partular the attempts by the ndvdual polymakers to get demand bak to supply by expansonary monetary poly makers partly lmted by the fat that ths would ause the exhange rate to depreate and ause nflaton. Cooperaton wll enable them to avod ths and so to be more expansonary, (even although they do not have an nflaton bas and would not seek to expand as and when the eonomy had returned to full employment). Fgure 7 shows results, whh onform to ths expeted outome. Cooperaton gves a more expansonary response to the shok. It enables the poly makers to trade off lost output for hgher nflaton. These results mrror the results from the lterature of the 98s when poly ooperaton was shown to avod exessvely ontratonary responses to nflatonary shoks8. Here we show that ooperaton enables more expansonary (and more nflatonary) responses to negatve demand shoks. But the results, although of the expeted sgn, are hghly mportant n that they are small. They suggest, just lke the studes from the 98s, that the effets of ooperaton would be very lmted. Ths s beause a more expansonary monetary poly ndued by ooperaton, does not seem to be able to make very muh dfferene to the level of unemployment resultng from the shok that we are studyng. As already noted, the downturn aused by the rsk shok seems to be n large part a downturn lower output and lower employment - whh s hard to hange greatly by monetary poly. To a onsderable extent t s a supply sde phenomenon whh annot be hanged muh by monetary poly and omes from the reduton n supply potental aused by the reduton n the stok of aptal. Monetary ooperaton n these rumstanes (ompared to the ountry followng ther own optmal poles) seems unlkely to be able to yeld muh beneft. In realty however t s not lear that ountres are urrently followng optmal poles ndvdually. To the extent that ooperaton mght move them away from the urrently napproprate poles suh as n Japan and Europe, ooperaton mght have muh larger payoffs than smulated n the artfal world of ths

23 modellng exerse. 8 Conlusons. We fnd that the global adjustment to shfts n equty rsk prema s sgnfant. They omprse both demand and supply sde shoks and therefore are dffult to manage wth monetary poles. We fnd that what some ommentators regards as mbalanes suh as shfts n urrent aounts and hanges n housng pres an be explaned as optmal responses to the types of shoks beng experened (although the atual hanges observed mght be larger than optmal, the dreton of hange s onsstent wth the adjustment n ths paper). Adjustng monetary poles to deal wth these mbalanes would be ounterprodutve. There s a role for monetary poly n smoothng the adjustment on the demand sde relatve to the adjustment on the supply sde, however monetary poly annot offset the underlyng supply sde onsequenes of the shok. The transmsson of equty shoks has mportant mplaton for ountres not experenng the shok both wthn the OECD and wthn the developng world. The supply sde of ountres not experenng the equty rsk shoks wll tend to be postvely affeted but external demand delnes. The negatve external demand shok but postve domest supply shok ndued by foregn aptal nflow s very dfferent to the experene of the affeted ountres whh experene domest supply and demand ontratons. The optmal adjustment of poly n dfferent ountres s therefore lkely to dffer aross ountres. We have taken the hanges to equty rsk to be ompletely external to the model. Ths s mght be unsatsfatory sne rsk pereptons may n fat be endogenous to the poly responses. The determnatons of rsk assessment, s an mportant area of researh that needs a great deal of researh. We have also gnored the role of fsal poly, whh may have mportantly dfferent effets than monetary poly. Haber et al (,) have already demonstrated the mportane of ths ssue wthn Europe. Future work wll extend the urrent analyss to the top of ths paper. 8 see MKbbn (997) for a survey of these results.

24 Referenes Allsop C., W. MKbbn, and D. Vnes (999) Fsal Consoldaton n Europe: Some Empral Issues n Hughes-Hallett, A., Huthson, M. and S. Hougaard (999) Fsal Aspets of European Monetary Integraton Cambrdge Unversty Press pp (ISBN 5656X) Armngton P. (969) A theory of demand for Produts Dstngushed by Plae of Produton" IMF Staff Papers vol 6, pp Bakus D., P. Kehoe and F. Kydland (99) Internatonal Real Busness Cyles Journal of Poltal Eonomy,, August p Bagnol, P. MKbbn W. and P. Wloxen (996) Future Projetons and Strutural Change n N, Nakenov, W. Nordhaus, R. Rhels and F. Toth (ed) Clmate Change: Integratng Eonoms and Poly, CP 96-, Internatonal Insttute for Appled Systems Analyss (Austra), pp8-6. Blanhard O. () What do We Know About Maroeonoms that Fsher and Wksell Dd Not? NBER Workng Paper 755, Natonal Bureau of Eonom Researh, Boston. Blanhard O. and S. Fsher (989) Letures on Maroeonoms MIT Press, Cambrdge MA. Bryant R. Henderson, D, Holtham, G, Hooper P. and S. Symansky (eds)(988) Empral Maroeonoms for Interdependent Eonomes. Brookngs Insttuton. Bryant R., Hooper, P., and C. Mann (993) Evaluatng Poly Regmes: New Researh n Empral Maroeonoms, Brookngs Insttuton, Campbell J. And N.G. Mankw (987) "Permanent Inome, Current Inome and Consumpton" NBER Workng Paper 436. Dornbush R. (976) "Expetatons and Exhange Rate Dynams" Journal of Poltal Eonomy 84, pp Ersson N and J. Irons (995) "The Luas Crtque n Prate: Theory wthout Measurement," n K. Hoover (995) Maroeonometrs: Developments, Tensons, Prospets Haber, G., Nek, R., MKbbn, W. J., (). Monetary and Fsal Poly Rules n the European Eonom and Monetary Unon: A Smulaton Analyss. In: Cho, J. J., Wrase, J. (Eds), European Monetary Unon and Captal Markets, Elsever, pp95 7. Haber, G., Nek, R. and W. MKbbn () Global Implatons of Monetary and Fsal Poly Rules n the EMU Open Eonomes Revew vol 3, no 4, pp