Hedgers, Speculators and Forward Markets: Evidence From Currency Markets

Transcription

1 Hedgers, Speculaors and Forward Markes: Evidence From Currency Markes K.F. Radalj Deparmen of Economics, Universiy of Wesern Ausralia, Absrac: Since Keynes (1930) and Hicks (1939) propounded heir heory of normal backwardaion, he issue of wheher hedgers mus pay speculaors an insurance premium has remained conroversial. Recen heoreical developmens incorporaing he exisence of marke imperfecions have validaed he exisence of an insurance premium charged o hedgers by speculaors. Owing o differences in daa ses and economeric mehods, a consensus has no ye been reached. Drawing inspiraion from asse pricing heory a model of currency reurns is used, similar o ha in Mark (1988) and he imporance of speculaive influences is esed. The purpose of he paper is o highligh he heoreical and saisical deficiencies of he exan lieraure and o examine he robusness of previous empirical resuls o changes in specificaion. Applicaions o risk managemen and forecasing are immediae, as knowledge of any insurance premium is crucial in formulaing an opimal hedging sraegy and an opimal forecasing model. Keywords: Speculaors, normal backwardaion. 1. INTRODUCTION The issue of wheher fuures prices exhibi a bias ha compensaes speculaors for risk daes back o Keynes (1930 and Hicks (1939). They purpored ha because speculaors provide hedgers wih he abiliy o manage risk, hey charge a premium for heir services. In conras, modern porfolio heory assers ha in perfec, fricionless markes, only risks ha invesors canno diversify will enile hem o earn a premium for bearing risk. However, recen heoreical work ha incorporaes imperfecions such as rading coss and non-markeable posiions, allows risk arising from hedging pressure o co-exis wih radiional sources of sysemaic risk, such as marke risk see Hirshleifer (1990) for example. The quesion is imporan for numerous reasons. If premia exis wihin prices, agens looking o forward prices o form expecaions mus incorporae premiums ino heir analysis. In formulaing hedging sraegies, opimal hedging sraegies may depend upon he size of any exan premium. Moreover, an overwhelming body of lieraure has documened he seemingly biased naure of he forward rae as a predicor of fuure spo reurns. The presence of ime-varying risk premia in exchange rae markes has been suggesed as a possible source of his bias. 1 Therefore, i is no surprising ha a grea deal of 1 For a review of his lieraure see Engel, C., (1996). research on he exisence of fuures premia has been conduced. Despie his hough, no consensus appears in sigh. The work of Carer e al. (1983), Chang (1985) and Bessembinder (1992) seems o suppor he noion ha speculaors can charge hedgers a premium for bearing he risks ha hey are rying o offse. Kolb (1992) and Charah e al. (1997) provide conflicing evidence. One reason we may never be cerain abou wheher a speculaive premium exiss because i is consisen wih oher reasons, such as superior forecasing abiliy on he par of agens. These sudies analyse simply fuures reurns, hus wheher speculaors influence spo and forward marke reurns remains o be seen. A recen paper by de Roon e al. (2000) proposes a novel idea whereby no only he paricular commodiy s speculaors charge a premium, bu hose in like asse classes also conribue o he exen of hedging pressure. They find also ha hedging pressure variables affec he underlying asse. However, he resuls are no compleely saisfacory. They measure marke reurns by he S&P 500. While he marke index will never be compleely observable, i seems ha a more reliable and relevan measure could be obained by insering world equiy reurns, as measured by he Morgan Sanley Capial Index (MSCI). We address his concern in his paper. Moreover, 2. EMPIRICAL SPECIFICATION Under risk-neuraliy, he sandard no-arbirage assumpion dicaes ha forward raes should be

2 an unbiased predicor of fuure spo exchange raes such ha: = F S ε where ε is a zero-mean, serially uncorrelaed process. If a risk premium exiss, his relaionship mus be exended o include he premium, ρ: S = F + ρ + ε + 1 Following Mark (1988), we analyse wha are noionally called forward reurns, ha is he log difference beween realized spo raes and forward raes S +1 F. 2 The original hypohesis proposed by Keynes was couched in erms of forward raes and realized spo raes, so his appears o be a more relevan es of he heory han merely analysing fuures reurns as in Bessembinder (1992) and de Roon e al. (2000). The model above does no sipulae wha sources he premium are derived from. Following he heoreical research of Hirshleifer (1990), and empirical papers such as Mark (1988), we include world equiy reurns and ne currency speculaors, scaled by he amoun of open ineres. Therefore, he model is: 5 ( S F 1 ) = α + βxsr + γ i SPECOI i, 1 + ε i= 1 where XSR = [(MSCI -MSCI -1 )/MSCI -1 ] - r US, SPECOI = (longspec-shorspec)/open ineres and ε is a zero mean, serially uncorrelaed process. 3. DATA In measuring forward raes we consruced he heoreical forward rae under covered ineres pariy. The daa was obained from Daasream and sampled on a monhly basis o avoid he complicaions of overlapping observaions. World excess reurns are measured as he monhly reurns on he MSCI index in excess of he U.S. dollar ineres rae. Speculaion is measured as he difference beween long and shor speculaors as repored in he Commodiy Fuures Trading Commission s (CFTC) Commimen of Traders repors. The CFTC requires ha large raders disclose heir purpose for rading fuures and is 2 The erm, reurn is no sricly rue as no capial is invesed in enering a forward conrac. he mos common way of gauging speculaive ineres applied in he lieraure. The ne amoun is scaled by oal open ineres o accoun for possible paerns in he amoun of fuures being raded. The sample spans Sepember 1992 o Ocober 2002 for 130 observaions. Tables 1 and 2 presen summary saisics of he variables used in he regression analysis along wih augmened Dickey-Fuller ess o esablish wheher he daa displays non-saionary behaviour. Consisen wih prior sudies, he ADF rejecs he null hypohesis of a uni roo in reurns a he one percen significance level. Furhermore, he measure of speculaion is also seen o be saionary. This suggess OLS mehods are accepable procedures o esimae he model. Over he period speculaors ended o be negaive across all currencies, perhaps reflecing he fac ha he USD appreciaed considerably over he period as i has become he sore of value for cenral banks and he uni of accoun in which global commerce is conduced. Speculaors appear o be mos volaile in he Ausralian dollar when compared o he mean amoun of conracs, perhaps reflecing ha he Ausralian dollar is ofen viewed as being a vehicle for speculaion. Table 2 presens summary saisics on world equiy reurns and he reurns o holding forward posiions. The null hypohesis of a uni roo was rejeced a he one percen level in all forward reurns and world equiy reurns. On average he difference beween realised spo raes and forward raes was slighly posiive, indicaing ha he forward rae under-prediced realised spo raes on average. World excess reurns were slighly negaive over he period, in large par due o he recen poor performance of equiy markes. 4. RESULTS Generally speaking, none of he models exhibied serially correlaed residuals, which is consisen wih weak-form efficiency as pas price movemens canno be used o predic fuure reurns. Moreover, he consan erms were no significanly differen from zero, suggesing ha excess reurns or anoher source of risk premium did no exis. 3 Apar from he Canadian dollar, he residuals did no display significan ARCH 3 I may be possible ha a ime-varying premium exiss ha flucuaes around zero giving an insignifican consan erm. However, recursive esimaion did no show parameer sabiliy o be an issue.

3 effecs, a resul consisen wih Baillie and Bollerslev (1989) who find ha as he sampling inerval is lenghened he presence of ARCH effecs weakens. In he Canadian dollar case he model was re-esimaed using maximum likelihood echniques, Bollerslev-Wooldridge sandard errors and employed he BHHH algorihm. Changing esimaion mehods impaced he resuls minimally, and when esimaed, he ARCH effecs were no significan. The Ausralian dollar is ofen characerised as a commodiy currency. Through he Ausralian economy s reliance upon mineral expors, is value is radiionally ied closely o commodiy prices, which in urn are linked o world economic growh. This relaionship may reflec he srong relaionship beween world equiy reurns and he Ausralian dollar and is refleced in he negaive relaionship beween Ausralian dollar movemens and world equiy reurns because he currency is quoed in domesic unis per uni of foreign exchange. The Ausralian dollar is also hough o arac a large degree of speculaion. However, from he resuls in Table 3 i seems ha speculaors are no capable of earning a premium on Ausralian dollar forwards. A similar sory is old in Table 4, which indicaes ha world excess reurns also explained a significan porion of Canadian dollar forward reurns. TABLE 1: SUMMARY STATISTICS OF SPECULATIVE INTEREST Sample based on 131 observaions over he period Sepember 1992 o Ocober ASPECOI CSPECOI GSPECOI JSPECOI SSPECOI Mean Sd. Dev CofV Skewness Kurosis Jarque-Bera Probabiliy ADF ** ** ** ** ** ** denoes significan a he one percen level. TABLE 2: FORWARD MARKET AND WORLD EQUITY RETURNS Sample based on 130 observaions over he period Sepember 1992 o Ocober AFWDRET CFWDRET GFWDRET JFWDRET SFWDRET XSRET Mean Sd. Dev CofV Skewness Kurosis JB ** ** ** ADF ** ** ** ** ** ** ** denoes significan a he one percen level.

4 TABLE 3 AUSTRALIAN DOLLAR C XSRET ** ASPECOI(-1) CSPECOI(-1) GSPECOI(-1) JSPECOI(-1) SSPECOI(-1) Adj R-squared D-W ARCH ** denoes significan a he one percen level. As he world s second larges economy, i is lile surprise ha movemens in he Yen are correlaed wih world equiy reurns. However, Table 6 indicaes ha here was no sign ha speculaion influenced Japanese forward reurns. Ineresingly, Table 7 shows ha he Swiss franc does no exhibi a srong correlaion wih world equiy reurns. The franc is probably viewed as a sore of value and as such would no be expeced o exhibi a srong correlaion wih global equiy reurns. Ausralian dollar speculaors were marginally significan in explaining Swiss franc forward reurns over he period. Why his arises is no immediaely eviden given ha he Swiss and Ausralian economies have very lile associaion. TABLE 4 CANADIAN DOLLAR C XSRET ** ASPECOI(-1) CSPECOI(-1) GSPECOI(-1) JSPECOI(-1) SSPECOI(-1) Adj R-squared D-W ARCH ** ** denoes significan a he one percen level. TABLE 5 BRITISH POUND C XSRET ASPECOI(-1) CSPECOI(-1) GSPECOI(-1) JSPECOI(-1) SSPECOI(-1)

5 Adj R-squared D-W ARCH ** denoes significan a he one percen level TABLE 6 JAPANESE YEN C XSRET * ASPECOI(-1) CSPECOI(-1) GSPECOI(-1) JSPECOI(-1) SSPECOI(-1) Adj R-squared D-W ARCH *denoes significan a he en percen level. TABLE 7 SWISS FRANC C XSRET ASPECOI(-1) * CSPECOI(-1) GSPECOI(-1) JSPECOI(-1) SSPECOI(-1) Adj. R-squared D-W ARCH *denoes significan a he en percen level. Comparing across currencies, our resuls differ somewha from de Roon e al. (2000). They found he Deuschemark o significanly influence European currencies. Unforunaely due o he inroducion of he Euro, including he DM in our sample would have grealy reduced he number of observaions. Also hey found ha each currency s speculaors significanly affec fuures reurns, in conras o he above resuls which indicae ha his is seldom he case. I may be ha fuures markes have become more efficien in incorporaing his informaion as our sample is more recen, or ha speculaive premiums have araced more speculaors such ha he speculaive premium has been bid owards zero. Also i shows ha using world equiy reurns o condiion for sysemaic risk may be an imporan consideraion. Furhermore, i seems ha global economic growh prospecs are imporan facors for economies ha depend upon world economic performance, such as Ausralia. This is a very ineresing resul, as i indicaes ha for foreign invesors, invesmen in such currencies enails added risk, due o he presence of correlaion beween he currency and he world equiy marke in general. This is in conras o sudies ino Arbirae Pricing Theory, which ofen fail o find significan currency effecs. Thus, if fund managers wish o pick socks based upon individual meri, i may pay hem o hedge heir

6 foreign currency exposure o eliminae his componen of risk wihin such asses. CONCLUSIONS The issue of wheher speculaors earn a premium for absorbing risk from hedgers has been exensively debaed, wih conflicing empirical evidence. Following de Roon e al. (2002), we include currency speculaors across differen currencies in our empirical model. The es is also a more direc es of he normal backwardaion heory as we analyse how forward raes relae o realised spo raes, as opposed o analysing fuures reurns, which is commonly applied in he lieraure. We find no evidence ha speculaors are capable of charging hedgers a premium for bearing risk. However, we do find significan evidence of sysemaic risk facors in currency markes from he perspecive of a U.S. invesor, using he MSCI as he benchmark marke index. 5. ACKNOWLEDGEMENTS The auhor would like o acknowledge he financial assisance of he Hacke Posgraduae Scholarship. I would like o hank Michael McAleer for helpful commens and suggesions. 6. BIBLIOGRAPHY Baillie, R.T., Bollerslev, T., The Message in Daily Exchange Raes: A Condiional Variance Tale, Journal of Business and Economic Saisics, 7, pp Bessembinder, H., (1992), Sysemaic Risk, Hedging Pressure and Risk Premiums in Fuures Markes, Review of Financial Sudies, 5, pp Carer, C.A., Rausser, G.C., Schmiz, A., (1983), Efficien Asse Porfolios and he Theory of Normal Backwardaion, Journal of Poliical Economy, 91, pp Chang, E.C., (1985), Reurns o Speculaors and he Theory of Normal Backwardaion, Journal of Finance, 40, pp Charah, A., Lian, Y., Song, F., (1997), Commimen of Traders, Basis Behaviour and he Issue of Risk Premia in Fuures Markes, Journal of Fuures Markes, 17, pp de Roon, F.A., Nijman, T.E., Veld, C., (2000), Hedging Pressure Effecs in Fuures Markes, Journal of Finance, 55, pp Engel, C.E., (1996), The Forward Discoun Anomaly and he Risk Premium: A Survey of Recen Evidence, Journal of Empirical Finance, 3, pp Hicks, J.R, (1939), Value and Capial, Cambridge: Oxfor Universiy Press. Hirshleifer, D., (1990), Hedging Pressure and Fuures Price Movemens in a General Equilibrium Model, Economerica, 58, pp Keynes, J.M., (1930), A Treaise on Money, Vol. II, Macmillan, London. Kolb, R.W., (1992), Is Normal Backwardaion Normal?, Journal of Fuures Markes, 12, pp Mark, N.C., (1988), Time-Varying Bea and Risk Premia in he Pricing of Forward Foreign Exchange Conracs, Journal of Financial Economics, 22, pp