The Link between FX Swaps and Currency Strength during the Credit Crisis of

Transcription

1 The Link beween FX Swaps and Currency Srengh during he Credi Crisis of Hans Genberg, Cho-Hoi Hui, Alfred Wong and Tsz-Kin Chung* Research Deparmen Hong Kong Moneary Auhoriy Absrac This paper analyses he impac of he global credi crisis on he money marke and discusses is poenial implicaions. The urbulence in money markes has spilled over o FX swap markes amid a reappraisal of counerpary risks during he recen financial urmoil. We examine he siuaions of six currencies: he euro, he Briish pound, he Ausralian dollar, he Japanese yen, he Hong Kong dollar and he Singapore dollar. We find ha (i) he risk premiums have indeed gone in andem wih he spreads of money marke raes over heir corresponding overnigh index swaps across he economies, a popular measure of poenial banking insolvency; and (ii) he risk premiums bear a negaive relaionship wih he srengh of he spo raes of he respecive currencies, which is consisen wih he increased pressure in he money and swap markes. JEL classificaion: F31; G15 Key words: FX swaps; covered ineres pariy; counerpary risk Auhor s Address: genberg@hkma.gov.hk; chhui@hkma.gov.hk; alfred_y_wong@hkma.gov.hk; bkchung@hkma.gov.hk * The auhors graefully acknowledge Yin-wong Cheung and he paricipans a he workshop on Exchange Rae Sysems and Currency Markes in Asia organised by Keio Universiy and Asian Developmen Bank Insiue for heir helpful suggesions and commens. The views and analysis expressed in his paper are hose of he auhors, and do no necessarily represen he views of he Hong Kong Moneary Auhoriy.

2 - 2 - I. Inroducion 1. The abrup escalaion of he recen global credi crisis las Sepember marked an imporan urning poin for many currencies. Those ha had exhibied remarkable srengh in early 2008 (e.g., he Briish pound, he euro, he Ausralia dollar) depreciaed abruply, while ohers (e.g., he Japanese yen, he Hong Kong dollar) rallied srongly. A he same ime, unusual pricing behaviour occurred in he money and foreign exchange (FX) swap markes, revealing significan and persisen deparure from covered ineres pariy, a well-esablished and well-esed heory in inernaional finance. Wha messages are hese aypical phenomena sending us? This paper aemps o shed ligh on hese quesions by looking a he FX swap marke more closely and finding ha he wo phenomena appear o be linked. II. Covered Ineres Pariy in Financial Turmoil rae F is: The ineres pariy heory saes ha he equilibrium forward exchange ( r) ( + q) S 1+ F = 1, (1) where S is he spo exchange rae (he domesic currency value of a uni of he US dollar), r and q are, respecively, he domesic and he US dollar raes of ineres on securiies ha idenical in all respecs excep for he currency of denominaion. The marke forward exchange rae F* gives a swap-implied US dollar ineres rae q*. Therefore, he reurn of invesing a sum of money in a domesic ineres-bearing asse for a cerain period of ime is he same as he reurn of invesing in a similar foreign ineres-bearing asse by convering he sum ino a foreign currency while simulaneously purchasing a fuures conrac o conver he invesmen back a he end of he period. If he reurns are differen, an arbirage ransacion could, in heory, produce a risk-free reurn. I is imporan o noe ha covered ineres pariy assumes ha asses denominaed in domesic and foreign currency are freely raded inernaionally (i.e., no capial conrols) and have negligible ransacion coss and similar risks. Given oday s echnology, hese assumpions normally hold in he inernaional financial markes, and so he pariy condiion is observed almos all he ime (excep for hose counries where capial conrols are sill in place). However, here are imes and siuaions in which he condiion breaks down. Taylor (1989) finds ha during he floaing of he serling in 1972 and he incepion of he European Moneary Sysem in 1979, significan deparure

3 - 3 - had occurred from covered ineres pariy for periods long enough o challenge he heory. 1 One possibiliy is ha in imes of financial urmoil, rue risks, or risks as perceived by marke paricipans, migh have changed, rendering he assumpions of covered ineres pariy inapplicable. Indeed, a recen BIS sudy by Baba and Packer (2008) finds ha in he recen global financial crisis he urbulence in money markes has spilled over o FX swap markes amid a reappraisal of counerpary risks. 2 The spillover occurred, as European banks needed o secure US dollar funding o suppor heir US conduis, while US banks also facing increased financing difficulies and having o preserve funds on hand became cauious in lending o heir European counerpars, forcing he laer o resor o convering euros ino dollars in he swap marke. As soon as European banks (borrowers), perhaps wih he excepion of he well-known members on he LIBOR panel, were perceived o be riskier by US banks (lenders), a risk premium quickly developed, adding o he dollar funding raes in he swap. In Char 1, he green line measures how much he implied 12-monh US dollar funding rae deviaes from he corresponding LIBOR he risk premium demanded by dollar lenders in he swap marke or he deparure from covered ineres pariy. 3,4 As can be seen, before he summer 2007 i oscillaed around 0% bu afer ha i sared o follow an upward rend. Around he beginning of Sepember 2008, i sho up and flucuaed widely. Baba and Packer focus on he shorer end of he marke in heir sudy, and heir findings yield similar resuls. Using CDS spreads and LIBORovernigh index swap (OIS) spreads as proxies, hey find evidence ha an increase in counerpary risks of European banks in general was he main driving force behind he deparure from he pariy condiion during he urmoil. 5 The LIBOR-OIS spread is commonly used as a measure of marke percepions of poenial insolvency of Oher sudies have aemped o raionalise hese deparures in erms of ransacions coss, e.g., Frenkel and Levich (1977) and Clinon (1988). The paper covers he sudy unil 12 Sepember 2008, i.e., before he Lehman defaul. The 3-monh conracs show similar resuls. The daa used in his paper are from Bloomberg. While inerbank ineres raes are in general bank specific, his sudy uses LIBORs which are he aggregae raes of panels of banks. An overnigh index swap (OIS) is a currency-specific ineres rae swap in which he floaing leg is linked o an index of daily overnigh raes. The wo paries agree o exchange a mauriy, on an agreed noional amoun, he difference beween ineres accrued a he agreed fixed rae and ineres accrued a he floaing index rae over he life of he swap. The fixed rae is a proxy for expeced fuure overnigh ineres raes. As overnigh lending generally bears lower credi and liquidiy risks, he credi risk and liquidiy risk premiums conained in he overnigh index swap raes should be small. Therefore, he spread of he 12-monh LIBOR relaive o 12-monh OIS rae generally reflecs he credi and liquidiy risks of he inerbank marke.

4 - 4 - counerparies, i.e., a higher LIBOR-OIS spread implies higher counerparies defaul risk. 6 Char 1: US dollar rae monh FX swap implied US dollar rae from euro (% poins) (% poins) EUR/USD swap-implied US dollar rae over US dollar LIBOR (rhs) US dollar LIBOR (lhs) EUR/USD swap-implied US dollar rae (lhs) Jan-07 Mar-07 Jun-07 Sep-07 Dec-07 Feb-08 May-08 Aug-08 Nov-08 Jan Implied US dollar rae over US dollar LIBOR III. Effecs of Counerpary Risks on Deviaions from Covered Ineres Pariy As compared o financial crises in he pas, one of he disinc characerisics of his crisis is how he resuling increase in counerpary risks paralyses he money marke. The significance of he Baba-Packer sudy lies in he uncovering of he mechanism in which he rise in counerpary risks among European banks as a resul of he urmoil feeds hrough from he money o he swap marke. Given ha European banks are no alone in his global crisis, he same should also be observed in economies where financial insiuions also suffered from a rise in counerpary risks. In Char 2, we plo he deviaions of he various currencies swap-implied US dollar funding raes from he corresponding LIBOR (or HIBOR). 6 See an aricle wrien by Alan Greenspan in he las 2008 issue of The Economis, advocaing he use of he LIBOR-OIS spread as a measure of marke percepions of exra capial needs for banks.

5 - 5 - Char 2: Deviaions of 12-monh FX swaps implied US dollar rae from dollar LIBOR 12 Sep - Lehman defaul GBP EUR AUD JPY - HKD SGD -1.5 Jan-07 May-07 Sep-07 Jan-08 May-08 Sep-08 Jan If Baba and Packer are righ in poining ou ha wha recenly happened in he swap marke reflecs essenially a rise in counerpary risks spilling over from he money marke, he amoun of premium or discoun as refleced in he swap-implied US dollar funding raes of differen FX swaps can be aken a measure of relaive risk of he banking secors of he economies concerned. The resuls of such measure should be suppored by he evidence in he LIBOR-OIS spread ha reflecs marke percepions of poenial banking insolvency. 7 The 12-monh LIBOR (or HIBOR)-OIS spreads of he respecive economies in Char 3 clearly show ha he banking secors in he US, he Euro area, he UK and Ausralia had a subsanial higher defaul risk han ha in Hong Kong, Japanese and Singapore in he las quarer of These observaions are consisen wih he premium or discoun as refleced in he swap-implied US dollar funding raes of differen FX swaps. 7 The CDS marke only covers a few banks in Hong Kong and Singapore. Therefore, heir CDS spreads do no represen he risk of he banking secors as a whole.

6 - 6 - Char 3: LIBOR-OIS spread (% poins) monh LIBOR(HIBOR)-OIS spreads USD LIBOR - OIS EUR LIBOR - OIS GBP LIBOR - OIS JPY LIBOR - OIS AUD LIBOR - OIS HKD LIBOR - OIS SGD LIBOR - OIS Ausralian dollar Briish pound Euro US dollar (% poins) LIBOR-OIS spread -0.5 Singapore dollar Japanese yen Hong Kong dollar Jan-07 Mar-07 Jun-07 Sep-07 Dec-07 Feb-08 May-08 Aug-08 Nov-08 Jan-09 - To es saisically wheher he changes of he differences beween he LIBOR-OIS spreads of he currencies concerned and he US dollar had effecs on he changes and volailiy of heir FX swap deviaions, we employ he EGARCH(1,1) model proposed by Nelson (1991) o es heir daa series from 9 Augus 2007 o 31 March Table 1 show ha he series are no saionary in level bu are saionary in he firs difference according o he augmened Dickey-Fuller es. The EGARCH(1,1) model in he firs difference we use can be wrien as Mean equaion: where dfxdev spread 2 = a + b d( LIBOR OIS) + ε ε ~ N( 0, σ ) (2) S FXdev = F * ( 1 + r ) (1 + q ) q q spread FC ( LIBOR OIS) = ( LIBOR OIS) ( LIBOR OIS) Variance equaion: ln 2 2 ε 1 ( ) = + ( ) + + ε 1 σ α β ln σ 1 γ η 2 / π σ 1 σ 1 USD (3) 8 This follows Taylor and Williams (2008a, b) o choose 9 Augus 2007 o mark he incepion of he urmoil, when BNP Paribas frozen redempions for hree of is invesmen funds.

7 - 7 - If he relaive risk of he banking sysems of he economies concerned is a deerminan of he premium or discoun as refleced in he swap-implied US dollar funding raes of differen FX swaps, he coefficien b in Eq.(2) should be posiive and saisically significan. Table 1: Augmened Dickey-Fuller es on he variables FXdev and ( LIBOR OIS) spread Sample period: 9 Augus 2007 o 31 March 2009 No. of observaion: 390 FXdev ( LIBOR OIS) spread Currency Level Daily change Level Daily change GBP * *** *** EUR *** *** AUD *** *** JPY *** *** HKD *** *** SGD *** *** 1.) ADF saisics are from he Augmened Dickey-Fuller uni roo es. The criical ADF values a he 10%, 5% and 1% significance level are -2.57, and respecively. 2.) *, ** and *** indicae significan a 10%, 5% and 1% levels respecively. 3.) The lag lengh of he ADF es is seleced according o he Schwarz Informaion Crieria wih maximum lags of 16. The resuls in Table 2 show he coefficiens b for he six currencies are posiive and significan a he 1% level. 9 This indicaes ha under he urmoil, FX swap deviaions in he euro, he pound and he Ausralian dollar ended o widen upward when counerpary risk was heighened for he financial insiuion in hese economies relaive o US counerpars. On he oher hand, he deviaions in he Hong Kong dollar, he yen and he Singapore dollar ended o go downward as counerpary risk of he financial insiuions in hese economies were perceived o low relaive o US counerpars. Regarding he causaliy beween he FX swap deviaions and he LIBOR-OIS spreads of he currencies, no evidence is found o suppor any causaliy beween hem based on he Granger causaliy es. 9 The sample is spli ino wo periods a 12 Sepember 2008, i.e. before and afer he Lehman inciden respecively. The esimaion resuls for b are posiive and significan for he wo periods. However, he adjused R-squares are higher for he pre-lehman inciden period.

8 - 8 - Table 2: Esimaion resuls of EGARCH(1,1) model Sample period: 9 Augus 2007 o 31 March 2009 No. of observaion = 389 afer adjusmen Currency Mean equaion Variance equaion Ljung-Box es up o lag 24 a b α β γ η Sandardized residuals (P-value) Squared sandardized residuals (P-value) Adj R-squared GBP 006 (02) 0.883** (23) ** (0.224) 0.953** (24) (86) 0.537** (0.134) (0.787) (0.867) EUR 02 (02) 0.853** (24) (0.202) 0.984** (18) (0.112) 0.398* (0.192) (0.886) (0.968) AUD 02 (03) 0.716** (34) ** (0.101) 0.970** (16) (66) 0.284** (71) (47) (0.796) JPY -02 (03) 0.869** (23) * (0.331) 0.940** (37) 08 (96) 0.524* (0.226) (0.154) (0.593) HKD 01 (02) 0.397** (27) (0.312) 0.929** (46) 15 (85) 0.405** (84) (09) (19) SGD -02 (03) 0.311** (29) (0.318) 0.931** (53) -79 (82) 0.315** (99) (05) (0.316) ) Numbers in parenheses are Bollerslev-Wooldrige robus sandard errors. * and ** indicae significan a 5% and 1% levels respecively. 2.) The Ljung-Box es (Q-saisics) idenifies wheher he auocorrelaions among daa are joinly zero up o a specified lag. Acceping he null hypohesis of he es means he daa is no serial correlaed. The sandardised residual is he residual divided by he esimaed volailiy. IV. Deviaions from Covered Ineres Pariy and Currency Srengh The financial urmoil in urn has had a significan impac on he foreign exchange marke a feaure of he crisis which so far has gone almos unnoiced by economiss and policymakers. This secion poins ou ha he financial urmoil has led o a reappraisal of counerpary risks of differen banking secors, which has in urn led o fundamenal changes in inernaional currency markes. As European banks resored o convering heir euros ino dollars o mee heir funding requiremens, he euro began o depreciae rapidly. The euro had been on a long-erm upward rend vis-à-vis he dollar before he news of he near-collapse of Bears Serns in March 2008 (Char 4). I had begun o sabilise

9 - 9 - since hen. And when severe financial difficulies of Freddie and Fannie surfaced in around Augus 2008, i fell sharply. Char 4: Deviaion of 12-monh FX swap implied US dollar rae from dollar LIBOR and EUR/USD exchange rae Sep - Lehman defaul (% poins) 1.8 Appreciaion agains US dollar 1.6 EUR/USD exchange rae EUR/USD spo exchange rae (lhs) Depreciaion agains US dollar EUR/USD swap-implied US dollar rae over US dollar LIBOR (rhs) Implied US dollar rae over US dollar LIBOR 0.85 Jan-07 Mar-07 Jun-07 Sep-07 Dec-07 Feb-08 May-08 Aug-08 Nov-08 Jan In his global crisis, where economies financial insiuions also suffered from a rise in counerpary risks, heir currencies should also experience sell-off o a similar exen. In Char 5, we plo he deviaions of he various currencies swap-implied US dollar funding raes from he corresponding LIBOR and heir exchange raes. As can be seen, some economies saw a significanly higher swap-implied funding rae over he US dollar LIBOR afer mid-sepember 2008 and, a he same ime, a sharply lower currency (i.e., he pound and he Ausralian dollar), while ohers enjoyed a swap-implied discoun from he US dollar LIBOR and, a he same ime, a much sronger currency (i.e., he yen and he Hong Kong dollar) The picure is less clear for he Singapore dollar. This may be due o he moneary policy for a sable Singapore dollar in effecive erms (see Secion III in Ma and McCauley (2009) for he discussion on he evoluion of he Singapore dollar).

10 Char 5: Deviaions of 12-monh FX swaps implied US dollar rae from dollar LIBOR and heir respecive exchange raes GBP/USD GBP Deviaion (rhs) GBP/USD spo exchange rae (lhs) Depreciaion agains USD 0.75 Jan-07 Jul-07 Feb-08 Sep EUR/USD 0.60 Deviaion (rhs) EUR EUR/USD spo exchange rae (lhs) Depreciaion agains USD 0.85 Jan-07 Jul-07 Feb-08 Sep AUD/USD AUD Deviaion (rhs) AUD/USD spo exchange rae (lhs) Depreciaion agains USD Jan-07 Jul-07 Feb-08 Sep JPY/USD 85 JPY HKD/USD 7.72 HKD SGD/USD 1.30 SGD Appreciaion agains USD Appreciaion agains USD Appreciaion agains USD Deviaion (rhs) 7.77 Deviaion (rhs) 1.40 Deviaion (rhs) 115 JPY/USD spo exchange rae (lhs) 125 Jan-07 Jul-07 Feb-08 Sep HKD/USD spo exchange rae (lhs) Jan-07 Jul-07 Feb-08 Sep SGD/USD spo exchange rae (lhs) 1.60 Jan-07 Jul-07 Feb-08 Sep Therefore, in conras o European and Ausralian banks (which need o conver heir currencies ino US dollars o mee heir funding requiremens), Hong Kong, Japanese and Singapore banks wih swap-implied discouns from he US dollar LIBOR (i.e., wih less defaul risk) arac financial insiuions (probably US banks) wih US dollars on hand o sell US dollars o hese Asian banks a spo in he FX swaps. 11 In his connecion, i becomes readily undersandable why he yen and he Hong Kong dollar appreciaed agains he US dollar afer he Lehman inciden. V. Conclusion This paper analyses he impac of he global credi crisis on he money marke and discusses is poenial implicaions. A recen BIS sudy finds ha he urbulence in money markes has spilled over o FX swap markes amid a reappraisal of counerpary risks during he recen financial urmoil. The spillover occurred, as European banks needed o secure US dollar funding o suppor heir US conduis while US banks were cauious in lending o hem. This forced he European banks o resor o convering heir euros ino dollars in 11 The counerpary risk a he forward legs of he swaps is also relaive low for hese Asian banks.

11 he swap marke, giving rise o a risk premium in he dollar funding raes in he swap. Our finding bears similariies wih he Japan premium episode in he lae 1990s when he crediworhiness of Japanese banks had subsanially deerioraed. Given he exreme difficuly of raising US dollars in money markes, Japanese banks urned o FX swap markes, which resuled in subsanial deviaions from covered ineres pariy (Covrig e al. (2004)). We exend he BIS analysis o examine he siuaions of five oher currencies. We find ha (i) he risk premiums have indeed gone in andem wih he spreads of money marke raes over heir corresponding overnigh index swaps across he economies, a popular measure of poenial banking insolvency; and (ii) he risk premiums bear a negaive relaionship wih he srengh of he spo raes of he respecive currencies, which is consisen wih he increased pressure in he swap markes. The implicaion of he analysis is ha he direcions of fund flows among differen economies may reflec he relaive safey and soundness of heir banking sysems during he credi crisis period.

12 References Baba N., Packer F Inerpreing Derivaions from Covered Ineres Pariy during he Financial Marke Turmoil of BIS Working Papers No Clinon K Transacion Coss and Covered Ineres Arbirage: Theory and Evidence. Journal of Poliical Economy 96, Covrig V., Low B. S., Melvin M A Yen Is No A Yen: TIBOR/LIBOR and he Deerminans of he Japan Premium. Journal of Financial and Quaniaive Analysis 39, Frenkel J. A., Levich R. M Transacion Coss and Ineres Arbirage: Tranquil versus Turbulen Periods. Journal of Poliical Economy 85, Ma G., McCauley R. N The Evolving Renminbi Regime and Implicaions for Asian Currency Sabiliy. Working Paper. BIS. Nelson B Condiional Heeroskedasiciy in Asse Reurns: A New Approach. Economerica 59, Taylor J. B., Williams J. C. 2008a. A Black Swan in he Money Marke. NBER Working Paper No Taylor J. B., Williams J. C. 2008b. Furher Resuls on a Black Swan in he Money Marke. manuscrip. Taylor M Covered Ineres Arbirage and Marke Turbulence. Economic Journal 99,

13 Appendix: Likelihood Raio Tes for Opimal Lag Lengh of VEC Model Pre-break Sample Pos-break Sample Lag Lengh LR saisic P-value LR saisic P-value 30 vs vs vs vs vs a 25 vs vs vs vs vs vs vs vs vs vs vs vs b 13 vs vs vs vs vs vs vs vs vs vs vs vs Noe: 1. a, b indicae ha he corresponding LR saisic is saisically significan a he 1% and 5% level respecively.