Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 89 MACROECONOMIC VARIABLES AND STOCK MARKET INTERACTIONS: NEW ZEALAND EVIDENCE Chrisopher Gan *, Minsoo Lee **, Hua Hwa Au Yong ***, Jun Zhang **** Absrac In his paper, we examine he relaionships beween he New Zealand Soc Index and a se of seven macroeconomic variables from January 1990 o January 2003 using coinegraion ess. Specifically, we employ he Johansen Maximum Lielihood and Granger-causaliy ess o deermine wheher he New Zealand Soc Index is a leading indicaor for macroeconomic variables. In addiion, his paper also invesigaes he shor run dynamic linages beween NZSE40 and macroeconomic variables using innovaion accouning analyses. In general, he NZSE40 is consisenly deermined by he ineres rae, money supply and real GDP and here is no evidence ha he New Zealand Soc Index is a leading indicaor for changes in macroeconomic variables. Key words: share reurns, macroeconomic variables, coinegraion, Granger-causaliy. JEL Classificaion: G10, G15. 1. Inroducion Over he pas few decades, he ineracion of share reurns and he macroeconomic variables has been a subjec of ineres among academics and praciioners. I is ofen argued ha soc prices are deermined by some fundamenal macroeconomic variables such as he ineres rae, he exchange rae and he inflaion. Anecdoal evidence from he financial press indicaes ha invesors generally believe ha moneary policy and macroeconomic evens have a large influence on he volailiy of he soc price. This implies ha macroeconomic variables can influence invesors invesmen decision and moivaes many researchers o invesigae he relaionships beween share reurns and macroeconomic variables. Soc prices are generally believed o be deermined by some fundamenal macroeconomic variables such as ineres rae, exchange rae and inflaion raes. Several sudies have aemped o capure he effec of economic forces on soc reurns in differen counries. For example, using, he Arbirage Pricing Theory (APT), developed by Ross (1976), Chen e al. (1986) used some macroeconomic variables o explain soc reurns in he US soc mares. The auhors findings showed indusrial producion, changes in ris premiums, and changes in he erm srucure o be posiively relaed o he expeced soc reurns, while boh he anicipaed and unanicipaed inflaion raes were negaively relaed o he expeced soc reurns. The developmen of coinegraion analysis provided anoher approach o examine he relaionships beween he macroeconomic variables and soc reurns. For example, Muherjee and Naa (1995) employed he Johansen coinegraion es in he Vecor Error Correcion Model (VECM) and found ha he Japanese soc mare is coinegraed wih six macroeconomic variables namely, exchange rae, money supply, inflaion rae, indusrial producion, long erm governmen bond rae and he shor erm call money rae. The resuls of he long-erm coefficiens of he macroeconomic variables are consisen wih he hypohesized equilibrium relaionships. Furhermore, Mayasmai and Koh (2000) used he Johansen coinegraion es in he Vecor Error Correcion Model (VECM) and found ha he Singapore soc mare is coinegraed wih five macroeconomic variables. Kwon and Shin (1999) applied Engle-Granger coinegraion and he Granger-causaliy ess from he Vecor Error Correcion Model (VECM) and found ha he Korean soc mare is * Lincoln Universiy, New Zealand. ** American Universiy of Sharjah, Unied Arab Emiraes. *** Monash Universiy, Ausralia. **** Chrischurch, New Zealand. Chrisopher Gan, Minsoo Lee, Hua Hwa Au Yong, Jun Zhang, 2006
90 Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 coinegraed wih a se of macroeconomic variables. However, using he Granger-causaliy es on macroeconomic variables and he Korean soc index, he auhors found ha he Korean soc index is no a leading indicaor for economic variables. Cheung and Ng (1998), employed Johnson s coinegraion echnique wih quarerly daa from Canada, Germany, Ialy, Japan and US, and conclude ha here are long erm comovemens beween he naional soc index and some specific variables, such as real oil price, real consumpion, real money supply and real GNP oupu in hose five counries. Furhermore, he auhors found ha he real reurns on soc indexes are, generally, relaed o deviaions from empirical long-erm relaionships and o changes in macroeconomic variables. Many sudies have been published abou he relaionships beween soc reurns and macro variables in well-developed counries such as he US, Japan and European counries. However, regional soc mares such as New Zealand and Ausralia are less explored because of heir small sizes and geographic locaions. In his paper, we examine he relaionships beween he New Zealand Soc Exchange soc index and a se of seven macroeconomic variables from January 1990 o January 2003 using coinegraion ess. Specifically, we employ he Johansen Maximum Lielihood and Granger-causaliy ess from a Vecor Error Correcion Model (VECM) o deermine wheher he New Zealand Soc Index is a leading indicaor for he macroeconomic variables. In addiion, his paper also invesigaes he shor run dynamic linages beween NZSE40 and macroeconomic variables using innovaion accouning analyses. In general, he NZSE40 is consisenly deermined by he ineres rae, money supply and real GDP and here is no evidence ha he New Zealand Soc Index is a leading indicaor for changes in macroeconomic variables The balance of he paper is organized as follows. Secion 2 reviews previous lieraure on he relaionships beween macroeconomic variables and soc reurns. Secion 3 provides an overview of he New Zealand soc exchange mare and Secion 4 describes he daa used in he research. The economeric mehods and resuls are discussed in Secions 5 and 6, respecively. Secion 7 concludes he paper. 2. Relaionships beween Macroeconomic Variables and Soc Reurns The dynamic relaionships beween macroeconomic variables and share reurns have been widely discussed and debaed. The basis of hese sudies has been he use of models which sae ha share prices can be wrien as expeced discouned cash flow. Thus, he deerminans of share prices are he required rae of reurn and expeced cash flows (Elon and Gruber, 1991). Economic variables which impac fuure cash flows and required reurns can herefore be expeced o influence share prices. Fama and Gibbon (1982) examine he relaionship beween inflaion, real reurns and capial invesmen. Their resuls suppor Mundell (1963) and Tobin (1965) findings ha expeced real reurns on bills and expeced inflaion raes are negaively correlaed. The auhors sugges ha his relaionship arises wih share reurns due o a posiive relaionship beween expeced real reurns on financial asses and real aciviy. Fama (1991) argues early empirical wor showing ha expeced inflaion is negaively relaed o share prices implied he measured relaionship beween inflaion and share reurns is a spurious one. Gese and Roll (1983) found ha he US soc price is negaively relaed o he inflaion rae and posiively relaed o he real economic aciviy. The second relaionship is consisen wih Fama (1981), and Lee (1992) findings. Lee (1992) argues ha share reurns signal changes in expeced inflaion due o a lin beween money supply and expeced real aciviy. Darra (1990) examines he effecs of moneary and fiscal policy on share reurns in he Canadian share mare and concludes ha budge deficis, long-erm bond raes, ineres rae volailiy and indusrial producion deermine share reurns. In esing he validiy of he Arbirage Pricing Theory, Chen, Roll and Ross (1986) conclude macroeconomic variables are causally relaed o share reurns. Najand and Rahman (1991) applied he Schwer (1989) volailiy measure and found evidence of he exisence of a causal relaionship beween share reurns and inflaion. An increase in ineres rae would increase he required rae of reurn and he share price would decrease wih he increase in he ineres rae. An increase in ineres rae would raise he
Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 91 opporuniy coss of holding cash, and he rades off o holding oher ineres bearing securiies would lead o a decrease in share price. French e al. (1987) documened heoreically, ha soc reurns responded negaively o boh he long erm and shor erm ineres raes. However, Allen and Jagiani (1997) poined ou ha he ineres rae sensiiviy o soc reurns has decreased dramaically since he lae 80 s and he early 90 s because of he invenion of ineres rae derivaive conracs used for hedging purposes. Fuhermore, Bulmash and Trivoli (1991) found ha he US curren soc price is posiively correlaed wih he previous monh s soc price, money supply, recen federal deb, recen ax-exemp governmen deb, long-erm unemploymen, he broad money supply and he federal rae. However, here was a negaive relaionship beween soc prices and he Treasury bill rae, he inermediae lagged Treasury bond rae, he longer lagged federal deb, and he recen moneary base. When he domesic currency depreciaes agains foreign currencies, expor produc prices will decrease and, consequenly, he volume of he counry s expors will increase, assuming ha he demand for his produc is elasic. Muherjee and Naa (1995), Achsani and Srohe (2002) confirmed his posiive relaionship exised in Japan and Indonesia boh wo large expor counries. Ajayi and Mougoue (1996) also showed ha an increase in soc price has a negaive shor-erm effec on domesic currency values bu in he long erm his effec is posiive, while currency depreciaion has a negaive shor and long-erm effec on he soc mare. Chen (1991) sudied he relaionship beween changes in financial invesmen opporuniies and changes in he macroeconomy in he U.S poined ou ha he mare excess reurns can be forecased using macroeconomic variables such as he lagged producion growh rae, he erm srucure, he T-bill rae, he defaul spread and he dividend yield. The mare excess on reurns is negaively relaed o he economic growh variables (such as he T-bill rae, lagged producion growh rae, he defaul spread and erm srucure) and posiively relaed o expeced fuure economic growh facors (such as he mare dividend price raio and unexpeced fuure GNP growh). Chen, Roll and Ross (1986) suggesed he following macroeconomic variables were sysemaically affecing asse reurns: he spread beween long and shor-erm ineres raes; expeced and unexpeced inflaion; indusrial producion growh and he spread beween high and low-grade bonds. Indusrial producion growh is suggesed o proxy for real cash flows, inflaion affecs reurns as nominal cash flow growh raes are no equivalen o expeced inflaion raes, whils he spread beween long and shor-erm ineres raes and he high or low grade bond spread affec he choice of discoun rae. Similar o Chen, Roll and Ross (1986), Hamoa (1988) deermines wheher he observed relaionships beween macroeconomic variables and share reurns are sill applicable when he analysis is conduced in he Japanese mare. The auhor also includes inernaional rade variables. Apar from indusrial producion appearing insignifican in asse pricing, Hamao s findings are consisen wih Chen, Roll and Ross (1986) sudy. Poon and Taylor (1991) parallel he Chen, Roll and Ross (1986) sudy on he Unied Kingdom mare. Their resuls show ha macroeconomic variables do no appear o affec share reurns in he Unied Kingdom as hey do in he U.S. Poon and Taylor (1991) sugges ha eiher differen macroeconomic facors have an influence on share reurns in he Unied Kingdom or he mehodology employed by Chen, Roll and Ross (1986) is inefficien. The auhors reemphasize he imporance of represening only he unexpeced componen of share reurns and macroeconomic variables in he model and argue Chen, Roll and Ross (1986) findings may be an example of a spurious regression. The auhors use an ARIMA model o es heir daa and use he residuals from he model as innovaions. Theoreically, he money supply has a negaive impac on soc prices because, as money growh rae increases, he inflaion rae is also expeced o increase; consequenly he soc price should decrease. However, an increase in he money supply would also simulae he economy and corporae earnings would increase. This would liely resul in an increase in fuure cash flows and soc prices. Muherjee and Naa (1995), Maysami and Koh (2000), and Kwon and Shin (1999) found ha here is a posiive relaionship beween money supply and soc reurns. During he pas decade, researchers have exended he sudy of ineracion beween macroeconomic variables and share reurns o counries oher han he U.S. For example, Kwon and
92 Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 Shin (1999) examined he Korean mare and found he Korean soc mares are coinegraed wih he producion index, exchange rae, rade balance and money supply. The auhors did no find he soc price index o be a leading indicaor for macroeconomic variables. Leigh (1997) sudied he Singapore soc exchange (SSE) and found he Singapore soc index o be posiively relaed o money demand bu wih no relaionship o macroeconomic fundamenals. Similar resuls have been idenified by Fung and Lie (1990) in Taiwan. Gjerde and Saeem (1999), Achsani and Srohe (2002) examined small regional mares such as Norway and Indonesia and conclude ha soc reurns respond negaively o changes in ineres rae, bu posiively o oil prices (Norway being a ne oil exporing counry), and real economic aciviy. However, he relaionship beween soc price and inflaion rae is ambiguous. Achsani and Srohe s (2002) sudy showed negaive relaionships beween soc price and inflaion rae as well as call money raes. However, a posiive relaionship was idenified beween he soc price and gross domesic produc, money supply and exchange rae. Furhermore, he auhors failed o find any significan relaionship beween soc price and expor or long-erm ineres raes. Increases in oil price will be beneficial o hose counries whose expor producs are derived from crude oil or refined oil producs. Thus, here should be a posiive relaionship beween he oil price and soc prices in hose oil-exporing counries. Bu here should be a negaive relaionship in oil imporer counries. Increases in oil price would increase he cos of producion and, consequenly, he expeced cash flow would decrease. However, Chen e al. (1986) failed o find any relaionship beween he soc index and he oil price in US. 3. New Zealand Soc Exchange Mare The New Zealand Soc Exchange (NZSE) is one of he leas regulaed soc mares compared o oher soc mares in Asia such he Souh Korean Soc Exchange, which is monopolised by he cenral governmen (Kwon and Shin, 1999). Since he deregulaion of financial mares in 1984, he New Zealand Soc Exchange is self-regulaed wih minimal governmen inervenion. For example, New Zealand does no impose any sauory conrol on he Soc Exchange s lising rules while mos oher counries do. Unlie oher counries, insider rading in he NZSE is only a civil offence. While, in mos developed counries such as US and Japan, i is considered as a criminal offence (Yu, 2002). There were five main soc indexes published by he New Zealand Soc Exchange before May 2003, namely, he NZSE10, NZSE30, NZSE40, NZSESC and NZSEALL. The NZSE40 is he main public mare index used, and covered he op 40 larges and mos frequenly raded socs lised on he NZSE. On he oher hand, he NZSESC (small capial) is made up of all small companies ha are no included in he NZSE40 index. Since he NZSE40 is an official index for he NZSE, his research uses his index ogeher wih he NZSEALL as a proxy for he movemen of he New Zealand soc mare. Before 1992, he Barclay s Index was he major NZSE mare index unil is replacemen, in 1991, wih he NZSE40. The Barclay s Index comprises he op 40 socs raned by heir mare capializaion (Brailsford, 1995). The Barclay s Index is used as he proxy for NZSE soc index before 1992 in his research. From June 3, 2003 he NZSE changed o he NZX and he NZSE50 replaced he NZSE40 as he official published New Zealand Soc Index. However, his will no affec our research findings. The NZSE is sill used as New Zealand s Soc Exchange and he symbols of NZSEALL and NZSE40 will be used as New Zealand Soc Indexes hroughou his research. As of April 30, 2003 here were a oal of 196 companies lised on he NZSE mare and 213 securiies quoed. These securiies have a oal mare capializaion of NZ$42.3 billion. In he four monhs ended April 30, 2003, a oal of 2,494 million shares, wih a value of $6,131 million, were raded on he NZSE mare. In comparison, in he four monhs ended April 30, 2002, he NZSE processed rades oalling 2,862 million shares wih a oal value of $5,961 million. Thus, in general, he NZSE is no large and liquid compared wih oher soc mares, bu i is one of he leas inervened soc mares in he world.
Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 93 4. Daa A oal of seven macroeconomic variables and NZSE40 are used in he analyses. The definiions of each variable are described in Table 1. Table 1 Descripion of Macroeconomic Variables Variable Share price index (NZSE40) Inflaion rae (CPI) Exchange rae (EX) Gross domesic produc (GDP) Money supply (M1) Long erm ineres rae (LR) Shor erm ineres rae (SR) Domesic Reail Oil price (ROIL) Definiion Official published index of he mare-weighed value of closing price for 40 shares lised on he New Zealand Soc Exchange Consumer price index (Quarerly) End of monh price of domesic currency rae in erms of rade weighed index (TWI) Real producion based gross domesic produc a consan level Narrowly defined money supply in New Zealand End of monh average lending rae for loans in he money mare rae (Base lending rae) End of monh call deposi rae End of monh domesic reail oil price in erms of he NZ dollar The macroeconomic variables are monhly frequencies from January 1990 o January 2003 from DaaSream excep for he CPI, real GDP figures, domesic reail oil price (ROIL), which are obained from Saisics New Zealand. The reason monhly daa is chosen is because mos macroeconomic variables are available monhly in New Zealand. There is a oal of 157 monhly observaions for each variable excep for he CPI, real GDP and, domesic reail oil price (ROIL) ha are quarerly daa, each wih 39 observaions. Using he RAT program hese quarerly figures were successfully ransformed ino he monhly figures used in his research. The monhly daa for SR is only available from January 1991, and has 145 observaions. This ime period is chosen because New Zealand experienced financial reforms in 1984 and he New Zealand dollar was floaed in 1985. Any macroeconomic variables before 1985 are less reliable and maybe disored. 5. Mehodology This paper employs he Johansen mulivariae coinegraion es and Granger-causaliy es o deermine wheher seleced macroeconomic variables are coinegraed (hence possibly causally relaed) wih share prices in he New Zealand soc exchange. Furhermore, he impulse response and Error Variance Decomposiion analyses are used o examine he dynamic relaions beween soc indices and various macroeconomic variables. The Augmened Dicey-Fuller (ADF) and Philips-Perron (PP) approaches are used o pre-es he order of inegraion for all ime series variables 1. A visual inspecion of he ime series plo of he variables invesigaed suggess ha here are no significan brea poins during he sample period. The lag lengh for he ime series analysis is deermined by choosing he lag lengh given by he minimum Aaie Informaion Crieria and Schwarz Informaion Crieria. Lagrange Muliplier ess are run o ensure ha he residuals from he chosen lag lengh are serially uncorrelaed. 5.1. Johansen Mulivariae Coinegraion Tes To invesigae he long-run relaionship of he NZSE index and macroeconomy as a sysem of equaions, we employed he Johansen mulivariae coinegraion es. The relaionships among he variables are based on he following model: 1 All soc indexes and mos macroeconomic variables are esed in a uni roo model wih consan bu no rend. The CPI, GDP, M1 exhibi a ime rend, hus, a uni roo model wih ime rend is chosen in he uni roo esing of hese variables.
94 Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 X X X X X D 1 1 2 2, (1) 1 1 where i I 1 2 i for i=1,2,-1; I 1 2 I is an ideniy marix The marix i comprises he shor-erm adjusmen parameers, and marix conains he long-erm equilibrium relaionship informaion beween he X variables. The could be decomposed ino he produc of wo n by r marix and so ha = where he marix con- ' ains r coinegraion vecors and represens he speed of adjusmen parameers (Johansen, 1988). Johansen developed wo lielihood raio ess for esing he number of coinegraion vecors (r): he race es and he maximum Eigenvalue es. The race saisics ess he null hypohesis of r = 0 (i.e. no coinegraion) agains he alernaive ha r > 0 (i.e. here is one or more coinegraion vecor). The maximum Eigenvalue saisics es he null hypohesis ha he number of coinegraing vecors is r agains he specific alernaive of r + 1 coinegraing vecors. 5.2. Granger-causaliy Tes In order o examine wheher here are lead-lag relaionships beween NZSE reurns and various macroeconomic variables, we run he Granger-causaliy es. If he ime series of a variable is nonsaionary, I(1) and is no coinegraed, he variable is convered ino I(0) by firs differencing and he Granger-causaliy es can be applied as follows: X Y a x x, ix i x, iy i x, i1 i1 a y y, iy i y, ix i y, i1 i1, (2), (3) where X and Y are he firs difference of ime series variable while he series is nonsaionary. However, if a variable is nonsaionary and coinegraed, he Granger-causaliy es will be run based on he following equaions: Y X a ax x, ix i x, iy i x ECTx, i x, i1 i1 y y, iy i y, ix i y ECTy, i y, i1 i1, (4), (5) 0 y, i i 1 where x and y are he parameers of he ECT erm, measuring he error correcion mechanism ha drives he X and Y bac o heir long run equilibrium relaionship. The null hypohesis for he equaions (2) and (4) is H 0 : x, i 0, suggesing ha i 1 he lagged erms Y do no belong o he regression. Conversely, he null hypohesis for he equaions (3) and (5) is H : 0, ha is he lagged erms X do no belong o he regression. These hypoheses are esed using F-es. 5.3. Innovaion Accouning Innovaion accouning such as he impulse response funcion and he forecas error variance decomposiion (FEVD) is used in analysing he inerrelaionships among he variables chosen in he sysem. The impulse response funcions are responses of all variables in he model o a one uni srucural shoc o one variable in he model. The impulse responses are ploed on he Y-axis
Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 95 wih he periods from he iniial shoc on he X-axis. Formally, each j (i) is inerpreed as he ime specific parial derivaives of he VMA( ) funcion (Enders, 1995): X ji j ( i). (6) e h Equaion (6) measures he change in he j variable in period resuling from a uni shoc o he h variable in he presen period. The FEVD measures he proporion of movemen in a sequence aribued o is own shoc o disinguish i from movemens aribuable o shocs o anoher variable (Enders, 1995). In he FEVD analysis, he proporion of Y variance due o Z shoc can be expressed as: 2 2 2 2 z [ a12 (0) a12 (1)... a12 ( m 1) ]. (7) 2 y ( m) 2 One can see ha as m period increases he y (m) also increases. Furher, his variance can be separaed ino wo series: y and z series. Consequenly, he error variance for y can be composed of e y and e z. If e y approaches uniy i implies ha y series is independen of z series. I can be said ha y is exogenous relaive o z. On he oher hand, if e y approaches zero (indicaes ha (Enders, 1995). e z approaches uniy) he y is said o be endogenous wih respec o he z 6. Empirical Resuls 6.1. Uni Roo Tes Resuls Table 2 shows he ADF and PP uni roo ess resuls. The ADF uni roo es resuls indicae ha only he SR in level rejecs he null hypohesis of nonsaionary a he 5% significance level. The NZSE and oher macroeconomic variables are found o be non-saionary in level bu saionary in firs difference, I(1). The PP uni roo ess presen similar resuls: all macroeconomic variables and he soc index of he NZSE40 have uni roos (non-saionary) when esed in levels and have no uni roo (saionary) in he firs difference. These resuls are consisen wih previous lieraure ha found mos macroeconomic variables and soc indexes are non saionary and non-mean revering. Thus, all macroeconomic variables and soc indexes are regarded as I (1) in he subsequen ess. Uni Roo Tes Resuls (Macroeconomic Variables and NZSE Reurn) Table 2 Variables ADF Uni Roo Tes PP Uni Roo Tes Level Firs Difference Level Firs Difference NZSE40-1.82-13.47** -1.83-13.46** CPI -0.97-3.33* -1.05-4.00** EX -1.46-12.50** -1.51-12.51** GDP -3.29-3.71** -2.28-3.14* LR -2.58-7.43** -2.28-7.49** M1-2.30-7.04** -2.52-17.00** ROIL -1.46-3.87** -1.51-12.91** SR -3.01* -8.12** -2.81-8.69** Noes: * Significance a 5% level, ** Significance a 1% level.
96 Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 6.2. Johansen Mulivariae Coinegraion Tes Resuls The resuls of he Johansen s Trace and Max Eigenvalue ess are shown in Table 3. A he 5% significance level he Trace es suggess ha he variables are coinegraed wih r = 6 while he Max Eigenvalue es resuls sugges ha he variables are coinegraed wih r = 4 if model 2 or model 3 were chosen 1. I is common for he esimaed es saisics o show differen resuls (Harries, 1995). However, in he Max Eigenvalue es, boh he null and he alernaive hypoheses are more specific. Therefore, he ran will be dependen on he Max Eigenvalue es resuls, which implies ha here are a leas four coinegraion vecors (r = 4) in model 3. Johansen Coinegraion Tes Resuls Table 3 Maximum Eigenvalue Saisics Trace Saisics (Model 2) 1 Maximum Eigenvalue Saisics Trace Saisics (Model 3) 2 R=0 R<=1 R<=2 R<=3 R<=4 R<=5 R<=6 R<=7 74.683 (52.00) 69.552 (46.45) 61.971 (40.30) 52.006 (34.40) 27.252 (28.14) 19.395 (22.00) 12.505 (15.67) 4.371 (9.24) 321.736 (165.58) 247.053 (131.70) 177.501 (102.14) 115.530 (76.07) 63.523 (53.12) 36.271 (34.91) 16.876 (19.96) 4.337 (9.24) 74.281 (51.42) 63.030 (45.28) 56.142 (39.37) 42.008 (33.46) 24.397 (27.07) 19.252 (20.97) 12.276 (14.07) 2.617 (3.76) 294.003 (156.00) 219.722 (124.24) 156.692 (94.15) 100.549 (68.52) 58.541 (47.21) 34.145 (29.68) 14.893 (15.41) 2.617 (3.76) Noe: The values in braces show he 5% criical value due o McKinnon (1988). 1 Model 2 (wih inercep only) and 2 Model 3 (wih inercep and rend). 6.3. Granger-Causaliy Tes Resuls The Granger-causaliy es is conduced o sudy he lead-lag relaionships beween macroeconomic variables and he NZSE40. The resuls are repored in Table 4. Four macroeconomic variables, namely, EX, SR, ROIL and GDP are found o be he mos imporan variables in deermining he NZSE reurn when hey were considered in pairs wih he NZSE40 using he Grangecausaliy es. The resuls also indicae ha he NZSE does no Granger cause any macroeconomic variables in New Zealand in he sample period. This sugges ha he NZSE40 is no a leading indicaor for any macroeconomic variables in New Zealand, which is inconsisen wih empirical resuls in oher world dominan soc mares such as he US, and Japan (Fama, 1991; Gese and Roll, 1983). A raional explanaion is ha he raio of capializaion of he soc o GDP in New Zealand, compared wih oher inernaional soc mares, is relaively small. Therefore, he impac of capial mares on he whole economy is also low. Oher researchers found similar findings for small open mares. For example Kwon and Shin (1999) did no find he KSE soc index is a leading indicaor of macroeconomic variables in Korea. 1 When resricions are imposed on he deerminisic componens of he Johansen s mulivariae model, five possible models exis (Hansen and Juselius,1995). In his sudy, boh model 2 (wih inercep only) and model 3 (wih inercep and rend) resricions are analysed, since according o Hansen and Juselius (1995), he oher models ha are oo resricive or leas resricive are unliely o occur in pracice.
Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 97 The Granger-Causaliy Tes Resuls beween Variables Table 4 Variables DNZSE40 DCPI DEX DGDP DM1 DLR DSR DROIL DNZSE40 - - - - - - - DCPI - - - - - - - DEX 5% - - - - - 10% DGDP 10% - - - 5% 1% - DM1 - - - 10% - 1% - DLR - - - 1% 5% 5% - DSR 5% - - 5% - 10% - DROIL 1% 10% - - - - - Noe: - means variable in row does no Granger cause variables in column; The number indicaes how large a percenage he variables in a row Granger cause he variables in he column. 6.4. Innovaion Accouning Analysis The es resuls of he impulse response funcion of macroeconomic variables on he NZSE40 are shown in Figure 1. The impac of a shoc o share prices experienced a significan posiive effec, which weaened a he 6 monh horizon. I implies he impac of a shoc o share index reduced dramaically, he NZSE40 became more efficien and is less dependen on he previous share index. The effec of a shoc o real GDP on NZSE40 was posiive hroughou he nex 24 monhs ime horizon. This posiive impac reached a maximum in he enh monh and became quie sable afer 14 monhs. This resul implies ha unlie he previous research he shoc of Manufacuring Producion Index on he NZSE40 index was delayed for nearly one year. The shoc of Real GDP on he NZSE40 index was found o be quie direc and effecive in his research. In general, he impacs of a shoc on he EX, CPI, LR and GDP on he NZSE40 in his sudy are consisen wih oher soc mares empirical resuls (see Ajayi and Mougoue (1996), Chen e al. (1986), Muherjee and Naa (1995), Maysami and Koh (2000) and Kwon and Shin (1999)). In he long erm he shoc of an appreciaion of he EX in New Zealand would arac more invesors o inves in he soc mare alhough his impac migh be negaive in he shor erm. The shocs of CPI and LR always have negaive impacs on he soc index as idenified in many oher counries. For example, Chen e al. (1986), Muherjee and Naa (1995) confirmed hese negaive relaionships exised in many indusrialised counries. The posiive impac of a shoc of GDP on he NZSE40 is consisen wih he empirical resuls of Maysami and Koh (2000) and Kwon and Shin (1999) implying ha he soc index should reflec he real siuaion of he economy. Fama (1990) and Gese and Roll (1983) idenified his posiive relaionship in heir findings. The negaive impac of a shoc of M1 on he NZSE40 can be explained by he following facors: he money supply in New Zealand is influenced mainly by foreign invesors. If he ineres rae is high relaive o oher counries, he foreign invesors are liely o leave heir money in he ban raher han inves in he risy soc mare. If he ineres rae is low he invesors migh prefer o inves in oher mares. Hence, he shoc of M1 on he NZSE40 always resuls in a negaive impac during his research-esing period. The posiive impac of a shoc of reail oil price on he NZSE40 afer eigh monhs conradics our hypohesis. This is doubful, as New Zealand is a ne oil impor counry. The shoc of he reail oil price should have a negaive impac on he NZSE40. The possible explanaion of his posiive impac, in he long erm, is ha New Zealand is an agriculure-based counry; he NZSE40 may, herefore, be less influenced by he increase in he impored oil price. The resuls of forecas error variance decomposiion (FEVD) are shown in Table 5. The es resuls indicaed ha FEVD for he NZSE40 could be aribued o LR, SR and M1, afer wo years, which accouns for 21.1%, 18.3% and 18.1%, respecively. I is ineresing o noe ha he
98 Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 NZSE40 iself accouns for only 16.1% of is own innovaion accouning afer wo years. If considering only one year, he NZSE40 is sill he mos imporan variable o accoun for is own innovaion, which accouns for 27.8%. LR and GDP are he nex wo imporan variables o be considered in explaining he forecas error variance, which accouns for 25.8% and 20.4%, respecively. SR only conribues 6.5% o he forecas error variance while CPI accouns for 7.41% of he forecas error variance. This implies ha afer one year he NZSE40 is deermined more by GDP and LR, while a he end of wo years, SR replaced GDP as he dominan facor o deermine he NZSE40. However, FEVD resuls indicaed ha here is lile evidence o show ha he variance in share price can be accouned for by innovaions in he exchange rae over he 24 monhs. Response o Cholesy One S.D. Innovaions Response of NZSE40 o NZSE40 Response of NZSE40 o CPI Response of NZSE40 o EX - - - Response of NZSE40 o GDP Response of NZSE40 o M1 Response of NZSE40 o LR - - - Response of NZSE40 o SR Response of NZSE40 o ROIL - - Fig. 1. Impulse Response Funcion of NZSE40 o Shocs in Sysem Macroeconomic Variables
Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 99 Forecas Error Variance Decomposiion of NZSE40 Table 5 (NZSE40) 1 monh 6 12 18 24 SE NZSE40 CPI EX GDP M1 LR SR ROIL 0.016 0.028 0.039 0.051 0.060 100 54.310 27.824 19.285 16.066 0 13.346 7.415 5.299 4.165 0 2.733 3.151 5.563 5.71 0 8.999 20.401 15.563 12.871 0 1.870 6.927 13.898 18.089 0 15.745 25.779 22.663 21.098 0 1.577 6.470 15.532 18.313 0 1.420 2.034 2.197 3.685 7. Conclusions This sudy examines he relaionships beween he NZSE Index and a se of macroeconomic variables during he period of January 1990 o January 2003. The ime series daa se employed in his sudy comprises he monhly observaions of he New Zealand Soc Index (NZSE40), he inflaion rae (CPI), long erm ineres rae (LR), shor erm ineres rae (SR), he real rade weighed exchange rae index (EX), real gross domesic produc (GDP), narrowly defined money supply (M1) and domesic reail oil prices (ROIL). Using he Johansen mulivariae coinegraion ess, his sudy examines wheher he New Zealand Soc Index is coinegraed wih a group of macroeconomic variables in he long run. This sudy also examines wheher he New Zealand Soc Index is a leading indicaor for economic variables by employing Granger-causaliy ess. In addiion, using impulse response funcion and FEVD analysis, his sudy also invesigaes he shor run dynamic linages beween NZSE40 and macroeconomic variables. The impulse response funcion in Figure 1 shows ha he shoc of CPI has a negaive impac on he NZSE40 hroughou he esing period and his negaive impac reached a maximum in he fourh monh. These es resuls are similar o Chen e al. (1986) findings. The Johansen coinegraion es indicaes ha here exiss a long run relaionship beween NZSE40 and he macroeconomic variables esed. The Granger-causaliy es resul shows ha he NZSE40 is no a leading indicaor in New Zealand, possibly because he New Zealand soc mare is relaively small as compared o he soc mares of oher developed economics. Finally, using innovaion accouning, he IRF resuls indicae ha he impac of a shoc o EX, CPI, LR and GDP on he NZSE40 in his research was consisen wih oher soc mares empirical resuls. The FEVD es resuls indicae ha he NZSE40 could be explained by LR, SR, M1 and GDP. In general, he NZSE40 is consisenly deermined by he ineres rae, money supply and real GDP during 1990-2003. Our resuls sugges ha invesmen percepion of New Zealand is a mixure of oher maure soc mares, as was found in Korea, he US and Japan. Thus, invesors who are ineresed in invesing in New Zealand should pay more aenion o he above menioned macroeconomic variables raher han he exchange rae and inflaion rae index (CPI). Since he New Zealand soc mare is comparaively small relaive o he soc mare of oher developed counries, he New Zealand soc mare migh also be very sensiive o global macroeconomic facors or he macroeconomic facors of is major rading parner. Thus, fuure sudies can exend his sudy o include hose facors. References 1. Achsani, N. and H.G. Srohe. Soc Mare Reurns and Macroeconomic Facors, Evidence from Jaara Soc Exchange of Indonesia 1990-2001 // Universiä Posdam, Wirschafsund Sozialwissenschafliche Faulä, Discussion Paper, 2002. 2. Ajayi, R.A. and M. Mougoue. On he Dynamic Relaion beween Soc Prices and Exchange Raes // The Journal of Financial Research, 1996, No. 19, pp.193-207.
100 Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 3. Allen, L. and J. Jagiani. Ris and Mare Segmenaion in Financial Inermediaries Reurns // Journal of Financial Service Research, 1997, No. 12, pp. 159-173. 4. Brailsford T. Time Varying Volailiy and he Impac of Economic Reform on he New Zealand Soc Mare. // Woring Paper Series in Finance, The Ausralian Naional Universiy, Canberra, 1995. 5. Chen N.F. Financial Invesmen Opporuniies and he Macroeconomy // Journal of Finance, 1991, Vol. 16, No. 2, pp. 529-553. 6. Chen N.F., R. Roll and S.A. Ross. Economic Forces and he Soc Mare // Journal of Business, 1986, Vol. 59, No. 3, pp. 383-403. 7. Cheung YW and Ng. Inernaional Evidence on he Soc Mare and Aggregae Economic Aciviy // Journal of Empirical Finance, 1998, Vol. 5, pp. 281-296. 8. Darra, A.F. Soc Reurns, Money and Fiscal Policy // Journal of Financial and Quaniaive Analysis, 1990, Vol. 25, No. 3, Sepember, pp. 387-398. 9. Elon, E.J. and M. Gruber. Modern Porfolio Theory and Invesmen Analysis, Fourh Ediion, John Wiley & Sons, 1991. 10. Enders W. Applied Economeric Time Series. John Wiley & Sons Inc., Unied Saes, 1995. 11. Fama E.F. Soc Reurns, Real Aciviy, Inflaion and Money // American Economic Review, 1981, Vol. 71, No. 4, pp. 545-565. 12. Fama, E.F. and M. Gibbons. Inflaion, Real Reurns and Capial Invesmen // Journal of Moneary Economics, 1982, Vol. 9, No. 3, pp. 545-565. 13. Fama E.F. Soc Reurns, Expeced Reurns and Real Aciviy // Journal of Finance, 1990, Vol. 45, pp. 1089-1108. 14. Fama, E.F. Efficien Capial Mares: II // Journal of Finance, 1991, Vol. 46, No. 5, December, pp. 1575-1617. 15. Fung H.G. and C.J. Lie. Soc Mare and Economic Aciviies: A Casual Analysis. Pacific- Basin Capial Mares Research, Amserdam, 1990. 16. French, K.R., G.W. Schwer and R.F. Sanbaugh. Expeced Soc Reurns and Volailiy // Journal of Financial Economics, 1987, Vol. 19, pp. 3-29. 17. Gese R. and R. Roll. The Fiscal and Moneary Linage beween Soc Reurns and Inflaion // Journal of Finance, 1983, Vol. 38, No. 1, pp. 7-33. 18. Gjerde, O. and F. Saeem. Casual Relaions among Soc Reurns and Macroeconomic Variables in a Small, Open Economy // Journal of Inernaional Finance Mares Insiuions and Money, 1999, Vol. 9, pp. 61-74. 19. Hamao Y. An Empirical Invesigaion of he Arbirage Pricing Theory, in Elon E.J. and M.J. Gruber (eds), Japanese Capial Mares Analysis and Characerisics of Equiy, Deb and Financial Fuures Mares. Ballinger Publishing Company, Unied Saes, pp. 155-173, 1988. 20. Hansen H. and K. Juselius. CATS in RATS: Coinegraion Analysis of Time-Series. Esima, Unied Saes, 1995. 21. Harris, R.I.D. Using Coinegraion Analysis in Economeric Modelling. Prenice Hall, Unied Saes, 1995. 22. Johansen, S. Saisical Analysis of Coinegraion Vecors // Journal of Economic Dynamics and Conrol, 1988, Vol. 12, pp. 231-254. 23. Kwon, C.S. and T.S. Shin. Coinegraion and Causaliy beween Macroeconomic Variables and Soc Mare Reurns // Global Finance Journal, 1999, Vol. 10, No. 1, pp. 71-81. 24. Lee, B.S. Casual Relaions among Soc Reurns, Ineres Raes, Real Aciviy, and Inflaion // Journal of Finance, 1992, Vol. 47, No. 4, pp. 1591-1603. 25. Leigh, L. Soc Reurn Equilibrium and Macroeconomic Fundamenals // Inernaional Moneary Fund Woring Paper, 1997, No. 97/15, pp. 1-41. 26. Maysami, R.C. and T.S. Koh A. Vecor Error Correcion Model of he Singapore Soc Mare // Inernaional Review of Economics and Finance, 2000, Vol. 9, pp. 79-96. 27. Muherjee T.K. and A. Naa. Dynamic Relaions beween Macroeconomic Variables and he Japanese Soc Mare: An Applicaion of a Vecor Error Correcion Model // Journal of Financial Research, 1995, Vol. 18, No. 2, pp. 223-237.
Invesmen Managemen and Financial Innovaions, Volume 3, Issue 4, 2006 101 28. Mundell, R.A. Inflaion and Real Ineres // Journal of Poliical Economy, 1963, Vol. 71, No. 3, June, pp. 280-283. 29. Najand, M. and H. Rahman. Soc Mare Volailiy and Macroeconomic Variables: Inernaional Evidence // Journal of Mulinaional Financial Managemen, 1991, Vol. 1, No. 3. 30. Poon, S and S.J. Taylor. Macroeconomic Facors and he UK Soc Mare // Journal of Business and Accouning, 1991, Vol. 18, No. 5, pp. 619-636. 31. Ross, S.A. The Arbirage Theory of Capial Asse Pricing // Journal of Economic Theory, 1976, Vol. 13, No. 3, pp. 341-360. 32. Schwez, W. Tess for Uni Roos: A Mone Carlo Invesigaion // Journal of Business and Economeric Saisics, 1989, Vol. 7, No, 2, pp. 147-159. 33. Tobin, J. Money and Economic Growh // Economeric, 1965, Vol. 33, No. 4, Ocober, pp. 671-684. 34. Yu, Jun. Forecasing Volailiy in he New Zealand Soc Mare // Applied Financial Economics, 2002, No. 12, pp. 193-202.