Can Stocks Hedge against Inflation in the Long Run? Evidence from Ghana Stock Market

Transcription

1 Inernaional Journal of Business and Managemen Can Socks Hedge agains Inflaion in he Long Run? Evidence from Ghana Sock Marke Anokye Mohammed Adam School of Business, Universiy of Cape Coas, Cape Coas, Ghana Siaw Frimpong School of Business, Universiy of Cape Coas, Cape Coas, Ghana Absrac Based on Fisher (1930) hypohesis, we es wheher Ghana sock marke can provide hedge agains inflaion in he long run using coinegraion analysis. Using daa for he Daabank sock Index (DSI) from January 1991 o December 2007, he resuls give srong suppor for he hedge propery. Thus Ghana sock marke provides full hedge agains inflaion. The oucome of his sudy holds imporan lesson for he marke paricipans in developing marke (many of which have experienced decades of higher inflaion) ha curren inflaion may no necessarily be associaed wih expecaions of lower fuure reurns. Keywords: Sock Reurns, Inflaion, Hedging, Coinegraion JEL: G10, G15, C32 1. Inroducion The lieraure on he relaionship beween sock reurns and inflaion is one of he longes mos researched in economics and finance. Common socks are expeced o hedge agains inflaion; herefore, in a perfec marke, reurn on common equiy should keep pace wih he inflaion rae. Following he seminal work of Bodie (1976), his proposiion has been exensively esed in he conex of he Fisher hypohesis (Fisher, 1930), which originally posulaed ha he marke ineres rae comprises he expeced real ineres rae and expeced inflaion. This hypohesis, when applied o sock markes, posulaes a posiive one-o-one relaion beween sock reurns and inflaion. Therefore in a compeiive marke, equiy socks, which represen claims agains he real asses of a business, may serve as a hedge agains inflaion; hence reurn on common equiy should keep pace wih he inflaion rae. Tha is, socks mus compensae invesors compleely (and no by more) for increases in he general price level hrough corresponding increases in nominal sock reurns, hereby leaving real reurns unaffeced. The empirical evidence on he issue of wheher he Fisher hypohesis holds in sock markes from developed and emerging economies is far from conclusive. The cases of developing and small markes like Ghana have no been esed. Inflaion in Ghana following he economic resrucuring in he 1980s and financial reforms ha ensued, adherence o sric moneary and fiscal policies sill remains high and volaile compare o developed counries. Inflaion in Ghana has shown a general upward rend from 1990 in spie of he reforms (see Figure 1). The GSE capial appreciaed by 116% in 1993 and gained 124.3% in is index level in 1994 (GSE quarerly bullein, March 1995) adjourning GSE as sixh and bes performing emerging marke in 1993 and 1994 respecively. In 1995, he index grew 6.3%, his abysmal performance is parly aribue o high inflaion and ineres rae. A criical quesion worh addressing is wheher he Ghana sock marke is a good hedge agains inflaion. We provide evidence on he long-run relaion beween sock reurns and inflaion in a coinegraion framework. We make conribuion o he lieraure by bridging he gap in he lieraure hrough esing of wheher he common socks in developing marke offer a hedge agains inflaion. Our invesigaion reveals he exisence of long run posiive one-o-one relaionship beween sock reurns and inflaion, indicaing validiy of Fisher hypohesis. The empirical validiy of he generalized Fisher hypohesis has insighful implicaions on invesmen (see Shresha e al 2002). Also, evidence of he validiy of he hypohesis gives insigh of a measure of he relaive efficiency of he sock marke under inflaionary condiions. Ceeris 188

2 Inernaional Journal of Business and Managemen Vol. 5, No. 6; June 2010 paribus, sock prices of efficien sock marke would impound informaion conained in expeced inflaion concurrenly. The res of he paper is organised as follows: Secion 2 looks a he lieraure review, secion 3 oulines he model and secion 4 discusses he daa and is characerisics. Secion 5 presens he empirical resuls and discussions. The las secion concludes and summarises. 2. Lieraure review Though he heory is very sraigh forward regarding he naure of he relaionship beween sock reurns and inflaion, resuls of he lieraure are fairly mixed. However he general conclusion is ha socks do no hedge agains inflaion in he shor run. Bodie (1976), Fama and Schwer (1977) examined he case of USA and find an inverse relaionship beween sock reurns and inflaion. The pos war period sudy by Gulekin (1983) failed o find suppor for he Fisher hypohesis (excep in UK). According o Fama (1981), inflaion acs as a proxy for real-aciviy variables in models ha relae sock reurns o inflaion, and herefore he observed inverse relaionship beween real sock reurns and inflaion is spurious. There is some evidence of a significan posiive relaionship in he longer horizons (more han 2 years) bu ofen wih a coefficien differen from one (1) so ha he inflaion hedge is no perfec (See Anari and Kolari,2001). Luinel and Paudyal (2006) find posiive relaionship beween sock reurns and inflaion in he UK. Spyrou (2004) examine he relaionship beween inflaion and sock reurns for en imporan Emerging Sock Markes (ESM), namely, Chile, Mexico, Brazil, Argenina, Thailand, S. Korea, Malaysia, Hong Kong, Philippines and Turkey, during he 1990s. According o he resuls, he relaionship beween sock reurns and inflaion, for he whole sample period, is posiive and saisically significan for hree of he sample ESMs, while i is posiive (bu saisically insignifican) for a furher hree. Only for one ESM is he relaionship negaive and saisically significan. More recenly, Durai and Bhaduri (2009) esed Fama hypohesis for India in he pos-liberalized period from a developing counry perspecive. Examining his relaionship on he ime-scale decomposiion from a wavele muli-resoluion analysis suggess ha Fama's hypohesis holds only for he long ime scale and remains as a puzzle for he oher ime scales. 3. The Model The Fisher hypohesis saes ha he nominal ineres rae a ime, ( R ) comprises he ex-ane real ineres rae ( E 1[ r ]) and he expeced inflaion rae ( E 1[ ]) we can herefore wrie; R E denoes he condiional expecaion operaor. where Assuming ha under raional expecaions, he expeced and he acual inflaion and ineres rae may differ by a saionary zero mean forecas errors (Alagidede, 2008) ) and ) respecively. Therefore we wrie (3.1) as: ( E 1[ r ]) ( E 1[ ] R ( 1 r 1 2 (3.2) Thus he ex-pos real rae is r R (3.3) Where 1 2 Under fricionless sock marke condiions, he equaion (3.3) posulaes ha he expeced real reurn on common sock and he expeced inflaion rae vary independenly so ha, on average, invesors are compensaed for changes in purchasing power. The es of his join hypohesis can be obained from esimaes of he following regression p (3.4) where p is he nominal reurn on common socks and is he inflaion rae. The ime subscrip denoes reurns beween he end of ime period 1 and ime period. is he error correcion erm ( 2 (3.1) 189

3 Inernaional Journal of Business and Managemen We examine he long run and shor run relaionship using he coinegraion echniques, specifically Johansen 1, 1 (1995) As posulaed by he Fisher hypohesis, we expec he long-run vecor o be close o one, hus ' if common sock compleely hedge agains inflaion. 3.1 The Daa The daa used were monhly Daabank sock price index from January 1991 o December 2007 obained from Daabank Group research. The inflaion variable used is he monhly consumer price index (CPI) for Ghana obained from he Inernaional Financial Saisics (IFS of he IMF) covering he same period. The inflaion rae and he sock marke reurns were calculaed as he firs differences of he logarihmic price levels of he respecive series. Figure 1 shows ha he inflaion in Ghana is very volaile especially beween 1990 and 2000 while he sock price rended upward (see figure 2). Table 1 presens a summary of descripive saisics of he variable. Sample mean, sandard deviaion, skewness, and he Jacque-Bera saisic have been repored. The high sandard deviaion of P wih respec o he mean is an indicaion of high volailiy in he sock marke. From he Jarque-Bera saisics, he null hypohesis ha P and are normally disribued a 10% level of significance can no be rejeced. There is high correlaion beween Daabank sock prices and inflaion. Sock prices movemens are correlaed wih inflaion movemens, albei before a one-monh lag. The peak correlaion coefficiens are very large. Figure A1, in he Appendix, indicaes ha he cross-correlaion beween sock prices and inflaion movemens is posiive, peaking before a one-monh lead a Sock prices movemens are also correlaed wih lagged inflaion peaking a before a one-quarer lag (Figure A2 in Appendix). The high size of he correlaion coefficien is an indicaion ha greaer percenage of inflaion migh be passed hrough o sock prices 4. Empirical Resuls 4.1 Uni roo We employ wo uni roo ess (DF-GLS and ADF) and one saionariy es (KPSS). Ellio e al (1996) developed he DF-GLS es which is more efficien han he usual ADF. The KPSS ess he null of saionariy, whereas ADF and DF-GLS ess he null of a uni roo. If he KPSS es rejecs he null bu ADF and DF-GLS es does no, we can say ha all he hree ess suppor he same conclusion; ha is, he series in quesion is an I (1) process. The resuls are shown in Table 2. As indicaed by he ADF and DF-GLS, he null of uni roo in all he series canno be rejeced a he 1% level. 4.2 Coinegraion es Given our uni roo es our analysis proceeds wih he empirical examinaion of he long-run Fisher hypohesis. We employ he Johansen (1991) coinegraion es for a number of reasons: he echnique is powerful han he usual Engle-Granger approach, and i is robus o various deparures from normaliy in ha i allows any of he variables in he model o be used as dependen variable while mainaining he same coinegraion resuls. I also allows for hypohesis esing and we can generae various scenarios o analyze he shor erm dynamics versus he long-run relaionship beween nominal exchange raes and prices. Johansen s mehod however, does suffer from small sample bias (see Alagidede e al, 2008). The lag lengh is seleced o ensure here is no furher residual correlaion using he Akaike Informaion Crierion (AIC) (see resuls in Table 3). The resuls from Table 3 reveal ha, here is a leas one coinegraing vecor, indicaing ha sock prices and inflaion coinegraed. This also indicaes ha he sysem is saionary in one direcion. The resuls shown in Table 4 indicae ha he esimaed Fisher coefficien, is 1.32 and significan a 10% level. The coefficien of,, indicae he elasiciy of he changes in sock prices wih respec o corresponding changes in consumer prices. The resuls imply ha 1% increase in inflaion leads o 1.32% increase in sock prices. The esing of he null hypohesis ha sock prices and he price level move one-for-one ( 1) gives a es saisic of which, again, has o be compared o a 2 (1) -disribuion. The criical significance level is 35% which leads o he unambiguous es resul ha he null can no be rejeced. The conclusion is srong suppor for he long-run inflaion hedge hypohesis. To es for robusness, we conduced a number of diagnosic es and sabiliy es of he VECM and ineresingly, he model passes all he diagnosics es (see Table 5) and sabiliy es (see Figure 3) 5. Conclusion We examined wheher Ghana sock Marke provides a hedge agains inflaion, focusing explicily on he long-run horizon. We have esed he Fisher hypohesis based on he long-run relaion beween sock prices and he general price level, esimaed by coinegraion analysis. 190

4 Inernaional Journal of Business and Managemen Vol. 5, No. 6; June 2010 Using he Consumer Price Index as he relevan price measure, he resuls give srong suppor o he hedge hypohesis. The resul is robus over ime evidenced by he diagnosic es. The evidence confirms ha he Ghana marke is efficien in inflaionary environmens as invesors are compensaed in high sock reurns when prices of goods are on he rise. The oucome of his sudy hold imporan lesson for he marke paricipans in developing marke (many of which have experienced decades of higher inflaion) ha higher curren inflaion may no necessarily be associaed wih expecaions of lower fuure reurns. References Alagidede, P. (2008). How Inegraed are African Financial Markes wih he Res Of The World? EEFS Conference Paper. Alagidede,P., Tweneboah G., and Adam,A.M. (2008). Nominal Exchange Rae and Price Convergence in he Wes Africa Moneary Zone. Inernaional Journal of Business and Economics, Vol.7, No.3, Anari, A., and Kolari, J. (2001). Sock Prices and Inflaion. Journal of Financial Research, 24, Bodie, Z. (1976). Common Socks as a Hedge Agains Inflaion. Journal of Finance, 31, Boudoukh, J., Richardson, M., and Whielaw, R. F. (1994). Indusry Reurns and he Fisher Effec. Journal of Finance, 49, Durai, S. R. S., and Bhaduri S. N. (2009). Sock Prices, Inflaion and Oupu: Evidence from Wavele Analysis. Economic Modelling, 26 (2009) Fama, E. F. (1981). Sock reurns, real aciviy, inflaion, and money. American Economic Review, 71, Fama, E. F., and G. W. Schwer. (1977). Asse Reurns and Inflaion. Journal of Financial Economics, 5:2, Fama, E. F., and M. R. Gibbons. (1982). Inflaion, Real Reurns and Capial Invesmen. Journal of Moneary Economics, 9, Fisher, I. (1930). The Theory of Ineres, Macmillan. New York. Geske, R., and Roll, R. (1983). The fiscal and moneary linkage beween sock reurns and inflaion. Journal of Finance, 38, Ghana invesmen promoion Cenre quarerly repor. December 2007 Gulekin, N. B. (1983). Sock marke reurns and inflaion: evidence from oher counries. Journal of Finance, 38, Inernaional Moneary Fund. (2008). Inernaional Financial Saisics. Washingon, D.C.: Inernaional Moneary Fund. Johansen, S. (1995). Likelihood-Based Inference in Coinegraed Vecor Auoregressive Models. Oxford Universiy Press, Oxford. Kaul, G. (1987). Sock reurns and inflaion: he role of he moneary secor. Journal of Financial Economics, 18, Kaul, G. (1990). Moneary regimes and he relaion beween sock reurns and inflaionary expecaions. Journal of Financial and Quaniaive Analysis, 25, Kwiakowski, D., Phillips P. C. B., Schmid, P., and Shin, Y. (1992). Tesing he null hypohesis of saionariy agains he alernaive of a uni roo. Journal of Economerics, 54, Luinel, K.B., and Paudyal, K. (2006). Are Common Socks A Hedge agains Inflaion?. Journal of Financial Research, XXIX, (1), MacKinnon, J. (1991). Criical Values for Coinegraion Tess, in R.F Engle and C.W.J. Granger(eds). Long Economic Relaionships, Oxford, Clarendon Press. Shresha, K., Chen, S.S. and Lee, C. F. (2002). Are Expeced Inflaion Raes and Expeced Real Rae Negaively Correlaed? A Long-Run Tes of he Mundell-Tobin Hypohesis. Journal of Financial Research, 25, Spyrou, S. I. (2004). Are socks a good hedge agains inflaion? Evidence from emerging markes. Applied Economics, 36: 1,

5 Inernaional Journal of Business and Managemen Table 1. Summary of Descripive saisics Mean Median S. Dev. Skew JB Corr, ) P Table 2. Uni Roo Tes ( p P P DF_GLS ADF KPSS 1.69** 0.35* Noe: The null hypohesis for he DF-GLS and ADF es is ha he daa process under examinaion conains a uni roo. Criical values for he DF-GLS are 2.58 and 1.94 for he 1% and 5% levels respecively, given by Ellio e al. (1996). Criical values of ADF 1% and 5% are and respecively (see MacKinnon, 1991). KPSS criical values for 1% and 5% are 0.21 and respecively, from Kwiakowski e al (1992). **, * indicaes significance a he 1% and 5% levels respecively. Table 3. Trace and Maximum Eigenvalue Tess Trace Tes Max. Tes 0.05 Criical 0.05 Criical race Prob. Value max Prob. Value. Lags 2 2 r * * r Noe: The race and max give he race saisics, and he maximal-eigenvalue saisic respecively. The null hypohesis for hese wo ess here is ha he daa generaing processes under consideraion are no coinegraed. Criical values for boh race and maximum-eigenvalue saisics a he 5% level are given by MacKinnon-Haugh-Michelis(1999). * denoes he rejecion of he hypohesis a he 1% level Table 4. Long Run Relaionship beween P and Coinegraion Vecor Adjusmen Coefficien p ( ) [ ] * [ ] * Coinegraion Resricion 1 2 (1) [0.3531] Noe: 1is he resricion ha he Fisher coefficien is equal o 1.The formal es of his hypohesis is based on he likelihood raio saisic LR es of resricions: 2 (1 ). Probabiliy in he [ ] and sandard error in ( ). Table 5. VECM Model Diagnosic Tess Serial Correlaion F(4, 15)=0.406[0.666] Funcional Form F(1, 193)=1.398[0.238] Heeroskedasiciy F(12, 19)=1.3236[0.2832] 2 Normaliy (2)= [ ) ] 192

6 Inernaional Journal of Business and Managemen Vol. 5, No. 6; June Figure 1. Monhly inflaion rae DSI Figure 2. Daabank Sock Index CUSUM 5% Significance Figure 3. Plos of CUSUM of Recursive residuals 193

7 Inernaional Journal of Business and Managemen Cross-correlaion of Daabank sock prices and lag of inflaion Figure 4. Cross Correlaion Beween Daabank sock prices and Inflaion (Lead) Figure