Asset Prices and the Conduct of Monetary Policy

Transcription

1 Asse Prices and he Conduc of Moneary Policy Charles Goodhar and Boris Hofmann * ABSTRACT In simple backward-looking srucural models of he economy he opimal moneary policy rule is given by a Taylor-ype ineres rae rule, wih he ineres rae being a funcion of curren and lagged inflaion raes and he curren and lagged oupu gap. Such a rule is opimal because curren and pas inflaion raes and oupu gaps are sufficien saisics for fuure inflaion and demand condiions, which are argeed by he cenral bank. We show ha fuure demand condiions and CPI inflaion in he G7 counries are also deermined by he exchange rae and propery and share prices. Taking he UK as an example we discuss he implicaions of his finding for he conduc of moneary policy and show ha disregarding asse price movemens leads o a sub-opimal oucome for he economy in erms of inflaion and oupu gap variabiliy. This resul no only obains because he informaion conained in asse prices abou fuure demand condiions is ignored, bu also because heir omission from he model inroduces considerable biases, so ha moneary policy would be based on a mis-specified model of he economy. We also show how a Financial Condiions Index (FCI), a weighed average of he shor-erm real ineres rae, he real exchange rae, real propery and real share prices can be derived based on he esimaed models. The derived FCI appears o be a useful predicor of fuure CPI inflaion. * The auhors are respecively a he Financial Markes Group, London School of Economics, and he Zenrum für Europäische Inegraionsforschung, Universiy of Bonn.

2 . Inroducion Since he seminal work by Taylor (993), he analysis of moneary policy rules has received considerable aenion in he heoreical and empirical lieraure. In simple backward-looking srucural models of he economy (see e.g. Svenssson, 997, Rudebusch and Svensson, 998), he opimal ineres rae rule relaes he policy rae o curren and pas inflaion raes and curren and pas oupu gaps. Such a policy rule is opimal, because curren and lagged inflaion raes and oupu gaps are sufficien saisics for fuure inflaion raes and oupu gaps, which are argeed by he cenral bank. Recenly here has been an increasing ineres in he role of asse prices for he conduc of moneary policy. The consensus view a he momen seems o be ha cenral banks should only respond o asse price movemens if hey are expeced o affec fuure CPI inflaion and he oupu gap (Bernanke and Gerler, 999). Besides he ineres rae, he exchange rae is usually considered o be he mos imporan deerminan of aggregae demand and channel of moneary policy ransmission in open economies. Several cenral banks adoped a Moneary Condiions Index (MCI), a weighed average of he shor-erm ineres rae and he exchange rae as an operaing arge (Bank of Canada, Reserve Bank of New Zealand) or an indicaor (Swedish Riksbank, Bank of Norway, Bank of Finland, Bank of Iceland) for moneary policy in he early-mid 990s. A more recen developmen is he ineres in he role of housing and equiy prices for he design of moneary policy. Housing and equiy prices may affec demand via direc and indirec wealh effecs. A change in propery and equiy prices affecs consumer wealh, which may induce consumers o change heir consumpion plans (Modigliani, 97). Recen evidence repored in Case, Quigley and Shiller (00) suggess ha propery prices have a sronger effec on household consumpion han equiy prices. A more indirec wealh effec of asse price movemens operaes via households and firms balance-shees. Households and firms may be borrowing consrained due o asymmeric informaion in he credi marke, Formal reamens of he heoreical underpinnings of MCIs can be found in Gerlach and Smes (998), Ball, (998) and Svensson (000). A discussion of he problems enailed wih consrucing and inerpreing MCIs can be found in Eika, Ericsson and Nymoen (996).

3 which gives rise o adverse selecion and moral hazard problems. As a resul, households and firms can only borrow when hey offer collaeral, so ha heir borrowing capaciy is a funcion of heir ne worh, which in urn depends on asse valuaions. Share prices may provide a proxy for he ne worh of lised companies. However, he balance shee posiion of lised companies can, bu does no necessarily have o be closely correlaed wih he balance shee posiion of non-lised companies and of households. Propery prices are likely o be a more useful indicaor for he borrowing capaciy of he privae secor, since a large par of privae secor credi is secured by real esae collaeral 3. Thus, from a heoreical poin of view here seems o be a srong case also o consider propery and share prices as deerminans of aggregae demand, which would imply a direc reacion of moneary policy o movemens in hese asse prices. This issue has proven o be highly conroversial. Cecchei, Genberg, Lipsky and Wadwhani (000) and Goodhar (00) argue in favour of a direc response of moneary policy o asse price movemens which are no in line wih perceived fundamenals, while Bernanke and Gerler (999) and Gerler, Goodfriend, Issing and Spavena (998) are more scepical. The consequences of a direc response of moneary policy o asse prices is usually analysed based on calibraed models (see e.g. Bernanke and Gerler, 999). Here we adop a differen approach and assess he role of asse prices in he G7 counries based on esimaed small srucural models in he spiri of Rudebusch and Svensson (998). The model consiss of a backward-looking Phillips Curve, relaing CPI inflaion o is own lags and he lagged oupu gap, and a backward looking IS Curve, relaing he oupu gap o is own lags and lags of he real ineres rae, he real exchange rae, real propery prices and real share prices. This framework is cerainly oo simple o derive any firm recommendaions for he conduc of moneary policy. Our aim is raher o show ha asse prices conain useful informaion abou fuure demand condiions and ha ignoring asse prices no only means loss of his informaion bu may also give rise o considerable biases in empirical models used for he analysis of moneary policy. The plan of he paper is as follows: In he nex secion we presen esimaes of his simple model. In Secion 3 we discuss he implicaions of he resuls presened in Secion for he Basic references for his lieraure are Bernanke and Gerler (989) and Kiyoaki and Moore (997). For a survey see Bernanke, Gerler and Gilchris (998). 3

4 conduc of moneary. Taking he UK as an example we explicily calculae he opimal policy rule and show wha happens if he effec of propery and share prices on aggregae demand is ignored. In Secion 4 we show how a Financial Condiions Index (FCI), a weighed average of he ineres rae, he exchange rae, propery prices and share prices, can be derived based on he esimaed models. Such an index can be used as a summary indicaor for aggregae demand condiions. The derived FCI is in fac found o be a useful predicor of fuure CPI inflaion. Secion 5 concludes.. Asse Prices in a Simple Backward Looking Model of he Economy In order o assess he poenial role of asse prices for he conduc of moneary policy, we esimae an exended version of he simple srucural model proposed by Rudebusch and Svensson (998), which is an empirical applicaion of he Svensson (997) model, for he G7 counries over he sample period using quarerly daa 4. In his framework he economy is modelled by a backward-looking supply or Phillips Curve and a backwardlooking demand or IS curve. The Phillips Curve relaes inflaion o lagged inflaion raes, lagged oupu gaps and he curren and lagged changes in he world price of oil as a proxy for supply shocks 5. The developmen of he oupu gap is explained by he IS Curve. We model he oupu gap as a funcion of is own lags and lags of he real ineres rae, he real effecive exchange rae, real house prices, real share prices and of exernal demand condiions proxied by he OECD oupu gap: 3 In fac, credi aggregaes are closely correlaed wih propery prices (Borio, Kennedy and Prowse, 994, IMF, 000, BIS, 00). See Hofmann (00) for a survey on his issue and formal evidence. 4 In Goodhar and Hofmann (000) we do a similar exercise for a larger sample of indusrialised counries. 5 The change in he world price of oil was a firs also included in he IS equaion. Bu since i was always insignifican and did also no help wih ouliers we eliminaed i from he IS equaions. We also ried o include asse prices in he Phillips Curve, since especially exchange raes and house prices could well have a direc effec on consumer prices, bu we were unable o deec any significan direc effec of asse prices on CPI inflaion. 4

5 Phillips Curve n n () π = β iπ i + β j y j + 3 β 3kdpo k + ε i= j= n k = 0 IS Curve m m m3 oecd () y = γ i y i + λ jrir j + λ3k rex k + λ4k rhp k + λ5krsp k + λq y q + η i= j= k = m4 k = m5 k = m6 q= π is quarerly inflaion in he consumer price index, measured as he quarer-o-quarer percen change in he CPI, y is he percen gap beween real GDP and poenial real GDP, where poenial real GDP is calculaed using a Hodrick-Presco filer wih a smoohing parameer of 600. dpo is he quarer-o-quarer percen change in he world price of oil. This variable acs as a proxy for supply shocks and helped o eliminae heeroskedasiciy. Rir, rex, rhp and rsp are respecively he percen gap beween he ex-pos real shor-erm ineres rae, measured as he shor-erm money marke rae less quarerly inflaion, he real effecive exchange rae 6, real residenial propery prices and real share prices and heir respecive longrun rend levels, calculaed using a Hodrick-Presco filer wih a smoohing parameer of 600. In order o conrol for exernal demand effecs we also include lags of he OECDoupu gap ( daa-appendix. oecd y ). A deailed descripion of he daa and heir sources is provided in he The modelling approach adoped here enails several poins which call for some deeper discussion before we presen he esimaion resuls. The use of he Hodrick-Presco filer as a rend filer is fairly sandard for real GDP, bu no for ineres raes and asse prices. Real ineres raes are usually assumed o be mean revering, so ha he long-run level of he real ineres rae is consan. However, real ineres raes are on average much higher since he 980s han hey were in he 970s, a developmen which is usually explained by rising governmen deb, he liberalisaion of capial markes and disinflaionary moneary policy (Group of Ten, 995). As a resul, mean reversion of real ineres raes is ofen rejeced by sandard uni roo ess. 6 The exchange rae is measured as unis of home currency per uni of foreign currency, so ha an increase in he real exchange rae is a real depreciaion. 5

6 Uni roo ess usually also sugges ha real exchange raes, real house prices and real share prices are no mean revering. Non-consan long-run levels of he real exchange rae may obain because he real exchange is a funcion of oher real variables. For example, he Balassa-Samuelson effec implies ha he equilibrium level of he real exchange rae of a counry depends on he produciviy growh in is radable goods secor relaive o he produciviy growh in he radable goods secor of is rading parners. Alernaive equilibrium conceps sugges ha equilibrium real exchange raes are deermined by relaive real aciviy, ne foreign asse posiions or real ineres rae differenials 7. The long-run level of real house prices is explained by changes in consrucion coss, real income, real financing coss and demographic facors (Mankiw and Weil, 989, Poerba, 99). The long-run level of real equiy prices depends, according o sandard asse pricing formulae (Gordon, 96), on he expeced long-run growh rae of dividends and he discoun rae. All hese variables are non-consan, which implies a ime-variable long-run level of real house prices and real equiy prices. Thus, here are many facors poenially deermining he long-run levels of ineres raes, exchange raes and house and equiy prices. An explici modelling of real ineres raes and real asse prices is clearly beyond he scope of his paper, and i is also no clear wheher such an aemp would have proven o be successful. For his reason we decided o filer ou he long-run rend movemens of he ineres rae and asse prices by using he same variable rend filer ha was applied o real GDP 8. Oher problems may arise from he simple specificaion of our model. Smes (997) shows ha he opimal ineres rae response o asse price movemens depends on wheher asse prices are driven by supply (fundamenal) or demand (non-fundamenal) facors. If asse prices are mainly driven by supply facors, hen heir effec on aggregae demand helps o equilibrae oupu demand and supply, so ha no policy response is needed. If demand facors are he main driving force of asse prices, hen asse price movemens end o cause disequilibria in he goods marke and should herefore be neuralised by ineres rae 7 MacDonald (000) provides a comprehensive survey of all he conceps for equilibrium real exchange raes. 6

7 reacions. How o inerpre movemens of an asse price may be enaively inferred from is pas correlaion wih he oupu gap. If movemens of an asse price mainly conain informaion abou aggregae demand raher han supply condiions, we would expec his o be refleced in a significan posiive ex-pos correlaion beween he asse price and he oupu gap. If he asse price is also driven by supply or fundamenal facors, is correlaion wih he oupu gap would be expeced no o be significanly differen from zero or o be even negaive. Adding asse prices o he analysis may also inroduce a simulaneiy problem. Propery and share prices are ofen characerised as forward looking variables, so ha including hem as regressors in a backward-looking model in he spiri of Rudebusch and Svensson (998), as we do here, may inroduce a simulaneiy bias in he esimaing equaions. However, similar simulaneiy problems may already arise from including ineres raes and exchange raes in he analysis. The cenral bank may raise ineres raes and exchange raes may appreciae in anicipaion of fuure oupu gaps. Thus, including propery and share prices in he analysis does no inroduce a problem ha has no poenially been here before. In he lieraure on moneary policy ransmission and moneary policy rules i is always assumed ha he informaion se of he cenral bank can be characerised by lags of endogenous and exogenous variables, so ha no simulaneiy problem arises. Bu why should asse markes have informaion abou fuure oupu and inflaion ha is superior o ha of he cenral bank? There is no reason why his should be he case, even if sock marke invesors are always fully raional. All hese poenial problems should cerainly be kep in mind, bu we believe ha hey are no so severe o render our esimaes invalid. Esimaes of he model are presened in Table. The equaions were esimaed separaely by OLS 9. In order o obain well behaved residuals a number of impulse dummies, which are mainly relaed o he oil price shocks, have also been included. The lag orders were chosen by a general-o-specific modelling sraegy, eliminaing all variables which were no significan a leas a he 0% level. Underneah he coefficien esimaes we repor -saisics in parenheses. For each equaion we also repor he adjused 8 We also esimaed a specificaion using he level of he real ineres rae and firs differences of he real exchange rae, real house prices and real equiy prices. The resuls are qualiaively no very differen from he ones repored below and are available upon reques. 7

8 R, Whie s (98) es for heeroskedasiciy (H) and a Lagrange-Muliplier es for serial correlaion up o order five (LM). The diagnosics sugges ha here is no evidence of misspecificaion in he esimaed equaions. The excepion is he Ialian Phillips Curve, where here is some evidence of heeroskedasiciy. Recursive Chow breakpoin ess, which we do no repor, did no indicae sub-sample insabiliy in any of he esimaed equaions. For he Phillips Curves we find ha he oupu gap is significan a leas a he 5% level in all counries. The coefficien esimaes sugges ha an increase in he oupu gap by one percenage poin leads o an increase of he inflaion rae of beween 0.3 percenage poins (Germany) and 0.5 percenage poins (UK) in he following quarer. Excep for Japan, he change in he world price of oil significanly affecs consumer price inflaion, eiher immediaely or wih a lag. The effec is paricularly srong in Ialy, where he coefficiens of he curren and lagged change in he oil price sum o 0.9. In he oher counries he sum of he oil price coefficiens never exceeds The resuls for he IS Curve sugges ha he real ineres rae, he real exchange rae, real house prices and real equiy prices all have a significan effec on he oupu gap, alhough he iming of he impac someimes differs considerably. The esimaed coefficiens for he real ineres rae are smaller han he usual esimaes of above -0. and vary beween 0. (Germany) and 0.05 (UK). The real exchange rae has a paricularly srong effec on oupu gaps in Germany and Ialy (0.06). In he oher counries he exchange rae coefficien varies beween 0.0 and The esimaed coefficien for real share prices in he US and he UK are hree imes larger han he coefficiens esimaed for he oher counries (0.0). This finding may reflec he higher share of equiy in privae secor wealh in hese wo counries (OECD, 000). The effec of house prices on he oupu gap is larger han he effec of share prices and in mos cases also larger han he effec of he exchange rae. The esimaed house price coefficien ranges beween 0.06 (France) and 0.03 (Ialy and UK). Excep for Japan and he UK we also find a significan effec of exernal demand condiion, proxied by he OECD oupu gap. 9 SUR esimaes gave almos idenical resuls. 8

9 Table : Regression Resuls π = 0.45π - 0.0π π 3-0.π π y DPO + 0.0DPO + 0.0DPO + 0.0DPO 3 USA Japan Germany France Ialy UK Canada (4.47) (-.9) (4.40) (-.8) (.6) (4.09) (5.53) (.6) (.48) (.4) R = 0.83 H = 4.7 (0.3) LM= 8.63 (0.) y = 0.87 y y - 0. y RIR REX RHP RSP +.88D78-.36D80 (9.98) (-.93) (-.89) (-.86) (.80) (.) (5.9) (4.70) (-3.73) R = 0.88 H = 3.7 (3.7) LM = 0.. (0.8) π = 0.5π + 0.7π + 0.0π y + 8.4D D74 (3.84) (4.4) (3.4) (3.77) (3.83) (9.48) R = 0.85 H = 4.6 (0.5) LM = 0.3 (0.69) y = 0.83 y - 0.3π RIR REXG RHP + 0.0RSP (.4) (-5.6) (-.5) (.64) (.64) (.84) R = 0.79 H = 3.0 (0.37) LM =.50 (0.) π = 0.46π + 0.0π π y + 0.0DPO (5.6) (.9) (.0) (.68) (.5) R = 0.63 H = 0.75 (0.38) LM =.8 (0.8) y = 0.3 y - 0.0RIR REX RHP RSP oecd y (3.05) (-.09) (.0) (.76) (3.63) (3.49) R =0.67 H = 5.58 (0.) LM =.67 (0.75) π = 0.53π π y DPO + 0.0DPO (7.) (5.33) (.05) (4.0) (.78) R = 0.85 H = 7.73 (0.06) LM = 7.5 (0.9) y = 0.90 y y RIR REX RHP RSP + 0. y oecd (.53) (-4.8) (-.0) (.65) (.49) (.64) (.69) R =0.77 H = 4.43 (0.4) LM = 4.8 (0.5) π = 0.37π + 0.6π + 0.π y DPO DPO DPO DPO D76 (4.39) (3.9) (3.76) (.00) (4.75) (.98) (.84) (3.43) (4.4) R = 0.86 H = 33.3 (0.0) LM =.6 (0.07) y = 0.56 y y RIR REX RHP RSP y oecd -.03D744 (8.43) (-3.4) (-.45) (4.9) (.77) (4.05) (4.9) (-3.39) R =0.86 H = 6.6 (0.36) LM = 0.5 (0.) π = 0.38π + 0.6π + 0.5π y DPO D D793 (5.9) (3.68) (.50) (.95) (.9) (4.4) (6.) R = 0.79 H = 7.54 (0.3) LM = 5.46 (0.36) y = 0.66 y y 3-0. y RIR REX RHP RSP -.9D74-.79D D79 (9.43) (.9) (-.80) (-.35) (.3) (.8) (4.9) (-3.69) (-.3) (4.87) R = 0.8 H = 7.6 (0.06) LM = 5.43 (0.37) π = 0.58π + 0.0π 3-0.3π π y + 0.0DPO (7.49) (.) (-.) (.85) (3.57) (.49) R = 0.79 H = 0.0 (0.06) LM = 8.37 (0.4) y = 0.6 y y RIR REX RHP + 0.0RSP y oecd y oecd (7.36) (-.59) (-.93) (.65) (.57) (.67) (4.75) (-.40) R = 0.84 H =.49 (0.3) LM = 6.0 (0.30) Noe: The able repors he resuls of esimaing equaions () and (). Coefficien esimaes are repored wih -saisics in parenheses. R is he adjused coefficien of deerminaion, H is Whie s (98) es for heeroskedasiciy and LM is a Lagrange-Muliplier es for serial correlaion up o order five. In parenheses we repor probabiliy values for he diagnosic ess. 9

10 3. Implicaions for he Conduc of Moneary Policy The resuls presened in he previous secion sugges ha asse prices significanly affec fuure demand condiions. Wha does his finding imply for he conduc of moneary policy? In he following we will only consider he case of he UK. Le us assume ha he cenral bank cares abou boh inflaion and oupu variabiliy, so ha he ineremporal loss of he cenral bank is given by (3) L = E τ δ ( π + τ + y+ τ ), τ = 0 where δ is he cenral bank s discoun rae. For simpliciy we have assumed ha he inflaion arge rae is zero and ha he cenral bank gives equal weigh o inflaion and oupu sabilisaion. Furhermore, we do no consider any addiional ineres rae smoohing objecive 0. Following Rudebusch and Svensson (998) we consider he limiing case δ =. The ineremporal loss funcion of he cenral bank is hen given by he sum of he uncondiional sum of he variances of he goal variables. The opimal ineres rae rule for he UK which minimises his loss funcion is given by: (4) i = r * +.78π + 0.3rex π + 0.3rhp π rsp + 0.4π dpo +.34y + 0.5y dpo y 3 The ineres rae responds o curren and lagged values of CPI inflaion and he oupu gap, bu also o he real exchange rae, real house prices, real share prices and he change in he world oil price. The quesion is now wha would happen if he cenral bank would ignore he 0 An objecive of he cenral bank o smooh shor-erm ineres raes may be moivaed by concerns abou financial marke sabiliy (Goodfriend, 989), uncerainy abou he economic environmen (Blinder, 998) or he aim o influence long-erm ineres raes (Goodfriend, 99). Inroducing an addiional ineres rae smoohing objecive helps o obain smaller and hus more realisic reacion coefficiens for he opimal ineres rae rule. The qualiaive resuls would, however, no be affeced. 0

11 effec of propery prices and share prices on aggregae demand, and insead assume ha oupu is affeced only by he ineres rae and he exchange rae, he sandard open economy deerminans of aggregae demand? Dropping real propery and real equiy prices from he model and re-esimaing he IS Curve yields (5) y = 0.86y y y rir rex - -.9D D D79 (.68) (.90) (-.86) (-.3) (.7) (-.59) (-.83) (4.87) We see ha dropping propery and equiy prices resuls in subsanially lower esimaed coefficiens for he ineres rae and he exchange rae. This finding is consisen wih he resuls repored in Goodhar and Hofmann (000). There we esimae backward-looking IS Curves for a sample of 7 indusrialised counries and show ha he real ineres rae coefficien is generally much smaller and insignifican when asse prices are omied from he model. This finding can be explained by a posiive correlaion beween real ineres raes and propery and equiy prices, which may arise if he cenral bank reacs o flucuaions in asse prices in order o sabilise he oupu gap. The ineres rae will hen proxy he effec of he omied asse prices on he oupu gap. Since asse prices are posiively correlaed wih he oupu gap, he esimaed ineres rae effec will be biased downwards. The lower esimaed coefficien of he exchange rae may also be explained by proxy effecs giving rise o a downward bias. Rising equiy and propery prices may rigger capial inflows, which give rise o an appreciaion of he currency. The correlaion beween equiy and propery prices and he exchange rae is herefore likely o be negaive. Thus, since equiy and propery are posiively correlaed wih he oupu gap, heir omission will give rise o a downward bias in he esimaed exchange rae coefficien. Dropping propery and equiy prices from he model also implies a differen opimal moneary policy rule. The opimal rule is now given by See Rudebusch and Svensson (998), p. 0.

12 (6) i = r * π + 0.6rex + 0.7π + 0.3dpo +.06π dpo + 0.4π y y.y 3 We see ha he opimal rule has changed considerably. No only is here now no direc response o propery and equiy prices, bu also he response o he oher variables has changed. The opimal response of moneary policy o curren and lagged inflaion raes and oupu gaps, and also o he change in he oil price is much higher now because of he lower esimaed ineres rae elasiciy. In order o assess he welfare effecs of ignoring he effecs of propery and equiy prices, we assume ha he rue model of he economy is he one esimaed in he previous secion, comprising propery and equiy prices. We simulae he model once using he opimal rule (4) corresponding o he rue model and once wih he (sub-) opimal rule (6) corresponding o he sandard open economy model wihou propery and equiy prices. Table displays he derived variances of he goal variables and he corresponding loss given by he unweighed sum of he variances. Table : Resuls on Volailiy and Loss Variance π Variance y Loss Opimal Rule Open Economy Rule We find ha using he rule based on he model wihou propery and equiy prices leads o an increase in he loss of more han 60%. This increase in he loss is no only caused by ignoring he effec of propery and equiy prices on oupu, bu also by using he wrong response parameers for all oher variables, which is a resul of he biases arising from excluding propery and share prices from he model.

13 4. Moneary and Financial Condiions Indices for he UK Moneary Condiions Indices (MCIs), weighed averages of he shor-erm ineres rae and he exchange rae, are commonly used by cenral banks, financial insiuions and nongovernmenal insiuions as indicaors for aggregae demand condiions. The weighs depend on he respecive effec of he ineres rae and he exchange rae on aggregae demand. Based on he models esimaed in he previous secion we can obain an exended MCI, or FCI (Financial Condiions Index) for he UK, comprising in addiion o he ineres rae and he exchange rae also propery and share prices. From he coefficien esimaes of he UK IS Curve in Table we obain a FCI weigh of 0.37 for he real ineres rae, of 0.3 for he real exchange rae and real propery prices and of 0.7 for real share prices. From equaion (6) we can derive, for comparison, a sandard MCI. The respecive weighs are 0.66 for he real ineres rae and 0.34 for he real exchange rae. Figure shows he derived FCI and he MCI for he UK. The indices are consruced so ha a higher level of he index reflecs expansionary moneary or financial condiions. Figure : An FCI and an MCI for he UK FCI MCI For a survey on MCIs for he UK see Baini and Turnbull (000). 3

14 No surprisingly, we find ha he FCI is more volaile han he MCI, and ofen imes he indicaors give conflicing signals abou fuure demand condiions 3. Thus, he FCI would ofen ell policy makers a differen sory han he MCI. The quesion is, which sory is more informaive and more reliable? In order o assess he informaion conen of he FCI and he MCI we esimae simple forecasing regressions for he CPI inflaion rae of he form (7) π = α + β π 4 + β I 4 + ε, where I is eiher he FCI or he MCI. In Table 3 we repor he esimaed forecasing equaions. In parenheses we repor -saisics calculaed based on Newey-Wes heeroskedasiciy and auocorrelaion consisen sandard errors. Table 3: Forecasing regressions for UK CPI Inflaion π = (5.5) (5.8) π FCI -4 R = 0.45 π = (4.4) (.75) π MCI -4 R = 0.34 Noe: T-saisics are in parenheses and are based on Newey-Wes heeroskedasiciy and auocorrelaion consisen sandard errors. The coefficien of he FCI is significan a he % level, ha of he MCI only a he 0% level. Also, he fi of he FCI forecasing regression is clearly beer han he fi of he MCI equaion. Thus, he FCI seems o conain valuable informaion abou fuure inflaion four quarers ahead. The MCI also conains informaion abou fuure inflaion, bu appears o be less informaive han he FCI. 3 The correlaion beween he FCI and he MCI is

15 5. Conclusions In simple backward-looking srucural models of he economy opimal moneary policy is given by Taylor-ype ineres rae rules, wih he ineres rae being a funcion of curren and lagged inflaion raes and he curren and lagged oupu gap. In open economy exensions of hese models he cenral bank addiionally reacs o he exchange rae. However, oher asse prices, propery and share prices, may also affec aggregae demand. In order o conrol fuure inflaion, he cenral bank may also have o reac o movemens of hese asse prices if hey have a significan effec on oupu. Based on an esimaed backward-looking srucural model of he economy we show ha besides he real ineres rae and he real exchange rae also propery and equiy prices have a significan effec on G7 oupu gaps. This finding implies ha moneary policy should also respond o propery and equiy price movemens in order o offse heir effec on he oupu gap. Taking he UK as an example we show ha ignoring propery and equiy prices in fac leads o a sub-opimal oucome for he economy in erms of inflaion and oupu gap variabiliy. This resul no only obains because he informaion conained in asse prices is ignored, bu also because heir omission from he model inroduces subsanial biases, so ha moneary policy would be based on a mis-specified model of he economy. The evidence presened in his paper seems o suppor an acive response of moneary policy o asse price movemens. However, we do no wan o advocae a mechanical policy response o asse prices. Any policy response o asse price movemens mus be preceded by a horough analysis of he causes of hese movemens and ake ino accoun he endogeneiy of asse valuaions. The chosen framework is cerainly much oo simple o derive any firm policy conclusions. Our aim is raher o show ha asse prices conain useful informaion abou fuure demand condiions and ha ignoring asse prices no only means loss of his informaion bu may also give rise o considerable biases in empirical models used for he analysis of moneary policy. 5

16 References Ball, L. (998), Policy Rules for Open Economies, NBER Working Paper No Baini, N. and K. Turnbull (000), Moneary Condiions Indices for he UK: A Survey, Exernal MPC Uni Discussion Paper No., Bank of England Bernanke, B. and M. Gerler (989), Agency Coss, Collaeral and Business Flucuaions, American Economic Review, 79, pp 4-3 Bernanke, B., Gerler, M. and S. Gilchris (998), The Financial Acceleraor in a Quaniaive Business Cycle Framework, NBER WP 6455 Bernanke, B. and M. Gerler (999), Moneary Policy and Asse Price Volailiy, Federal Reserve Bank of Kansas Ciy Economic Review, Fourh Quarer 999, pp 7-5 BIS (00), 7 s Annual Repor. Blinder, A. (998), Cenral Banking in Theory and Pracice, MIT Press, Cambridge, MA. Borio, C., Kennedy, N. and S. Prowse (994), Exploring Aggregae Asse Price Flucuaions across Counries: Measuremen, Deerminans and Moneary Policy Implicaions, Bank for Inernaional Selemens, BIS Working Paper No. 40 Case, K., Quigley, J. and R. Shiller (00), Comparing Wealh Effecs: The Sock Marke versus he Housing Marke, Cowles Foundaion Discussion Paper No. 335, Yale Universiy Cecchei, S., Genberg, H., Lipsky, J. and S. Wadhwani (000), Asse Prices and Cenral Bank Policy, Geneva Repors on he World Economy Eika, K., Ericsson, N. and R. Nymoen (996) Hazards in Implemening a Moneary Condiions Index, Oxford Bullein of Economics and Saisics, 58 (4), November, pp

17 Gerlach, S. and F. Smes (996), MCIs and Moneary Policy under Floaing Raes, unpublished manuscrip, Bank for Inernaional Selemens Gerler, M., Goodfriend, M., Issing, O. and L. Spavena (998), Asse Prices and Moneary Policy: Four Views, Cenre for Economic Policy Research (CEPR) and Bank for Inernaional Selemens (BIS), CEPR: London Goodfriend, M. (989), Ineres Rae Smoohing and Price Level Trend-Saionariy, Journal of Moneary Economics, 9, pp Goodfriend, M. (99), Ineres Raes and he Conduc of Moneary Policy. Carnegie Rocheser Series on Public Policy, 34, pp 7-30 Goodhar, C. (00), Wha Weigh Should be Given o Asse Prices in he Measuremen of Inflaion?, The Economic Journal,, pp Goodhar, C. and B. Hofmann (000), Financial Variables and he Conduc of moneary Policy, Sveriges Riksbank Working Paper No. Gordon, M. (96). The Invesmen, Financing and Valuaion of he Corporaion, Irwin, Homewood, IL. Group of Ten (995). Saving, Invesmen and Real Ineres Raes Hofmann, B. (00), The Deerminans of Privae Secor Credi in Indusrialised Counries: Do Propery Prices Maer?, BIS Working Paper, forhcoming IMF (000), World Economic Oulook, May 000 Kiyoaki, N. and J. Moore (997), Credi Cycles, Journal of Poliical Economy, 05, pp -48 MacDonald, R. (000), Conceps o Calculae Equilibrium Exchange Raes: An Overview, Economic Research Group of he Deusche Bundesbank Discussion Paper No. 3/00 Mankiw, N. and D. Weil (989), The Baby Boom, he Baby Bus, and he Housing Marke, Regional Science and Urban Economics, 9, pp

18 Modigliani, F. (97), Moneary Policy and Consumpion, in: Federal Reserve Bank of Boson, Consumer Spending and Moneary Policy: he Linkages OECD (000), Economic Oulook 68, December 000 Poerba, J. (99), Housing Price Dynamics: The Role of Tax Policy and Demography, Brooking Papers on Economic Aciviy, pp Rudebusch, G. and L. Svensson (998), Policy Rules for Inflaion Targeing, in: J. B. Taylor (ed.), Moneary Policy Rules, Universiy of Chicago Press for NBER Smes, F. (997), Financial Asse Prices and Moneary Policy: Theory and Evidence, CEPR Discussion Paper No. 75 Svensson, L. (997), Inflaion Forecas Targeing: Implemening and Monioring Inflaion Targes, European Economic Review, 4, pp -46 Svensson, L. (000), Open-Economy Inflaion Targeing, Journal of Inernaional Economics, 50, pp Whie, H. (980), A Heroskedasiciy-Consisen Covariance Marix and a Direc Tes for Heeroskedasiciy, Economerica, 48, pp

19 Daa Appendix Consumer Prices Consumer Price Index, IMF Inernaional Financial Saisics, series code 64; OECD Main Economic Indicaors for Germany. Gross Domesic Produc (GDP) IMF Inernaional Financial Saisics, series code 99b.p.; OECD-GDP: OECD Main Economic Indicaors World price index of peroleum IMF Inernaional Financial Saisics Real Exchange Raes Effecive real exchange rae, OECD Main Economic Indicaors. Shor-erm ineres raes Overnigh money marke raes for France, Germany, Ialy, Japan and he USA, IMF Inernaional Financial Saisics, series code 60b; hree monhs commercial paper rae for Canada (IMF, series code 60bc) and hree monhs inerbank rae for he UK (BIS). Share prices Share price index, IMF Inernaional Financial Saisics, series code 6 Propery Prices Residenial propery price index from naional sources as shown in Appendix-Table. Annual daa from he firs quarer of each year for Germany and semi-annual daa for Ialy and Japan were convered o quarerly frequency by linear inerpolaion. 9

20 Appendix-Table : Residenial propery prices Unied Saes Series Median sales prices of second hand one-family houses Source Naional Associaion of Realors Japan Naion-wide land price index Japan Real Esae Insiue Germany Average sales price of dwellings in Frankfur, Munich, Hamburg and Berlin Ring Deuscher Makler France Residenial house price index Bank of France Ialy Unied Kingdom Canada House price index for he whole counry Naionwide Anglia House Price Index Muliple lising service price index of exising homes Bank of Ialy/ Il consulene immobiliare Daasream Bank of Canada 0