Macroeconomic News Surprises, Business Cycles, and Interest Rate Swap Spreads

Transcription

1 Macroeconomic News Surprises, Business Cycles, and Ineres Rae Swap Spreads Fang, V. 1, C.T. Lin 2, and L. Roadcap 1 1 Deparmen of Accouning and Finance, Monash Universiy, Vicoria 2 School of Commerce, Universiy of Adelaide, Souh Ausralia Vicor.fang@buseco.monash.edu.au Keywords: Ineres rae swap spreads, macroeconomic news, business cycles, and probabiliy of defaul. EXTENDED ABSTRACT This sudy examines he response of Ausralian ineres rae swap spreads o he arrival of macroeconomic news informaion during he economic expansion and conracion periods. We find ha he impac of news announcemens on swap spread change differs and largely depends on he sae of he economy. The unexpeced inflaion rae is he only news released ha has significan impac on swap spreads across all mauriies during conracions and remains he imporan news announcemen hroughou he business cycles. The unanicipaed unemploymen rae ends o be more relevan o he 10-year swap and he unanicipaed change in money supply ends o be more relevan o he 4-and 7-year swaps during expansions. We also find shocks from hese news surprises appear o have significan impac on he condiional volailiy of he swap spread change during boh economic phases. The macroeconomic shocks in general are negaively relaed o he condiional volailiy of he swap spread change, suggesing ha he news worhy announcemens end o reduce uncerainy on he news announcemen days in he swap marke during expansion and conracion periods. 1. INTRODUCTION Macroeconomic news plays an essenial role in financial markes by revealing new informaion abou he fundamenals of he economy. Periodic news announcemens such as unemploymen rae and consumer price index (CPI) are herefore closely wached by marke paricipans who respond quickly and accordingly o unexpeced elemen of he news. These news surprises herefore form an imporan par of he price discovery process in financial asses. An exensive lieraure examining he impac of macroeconomic news surprises on financial asses has well been documened, mosly on bonds, socks, and foreign exchange. For example, Fleming and Remolona (1997), Bollerslev e al. (2000), and Green (2004) find ha news surprises from GDP, inflaion rae, unemploymen rae, or consumer confidence are significanly relaed o changes in Treasury yields especially around he ime of he announcemens. Similarly, Chen e al. (1986), and Brenner e al. (2005) repor ha he same economic surprises affec sock prices albei hrough a more complicaed mechanism due o poenial changes in expeced cash flows, he discoun rae, he risk premium, or a combinaion of hese hree pricing facors. In addiion, Anderson e al. (2003) and Simpson e al. (2005) show ha announcemens relaed o ineres rae and inflaion have significan impac on he exchange rae. In addiion, a number of empirical sudies sugges ha macroeconomic fundamenals play an imporan role in deermining credi spread dynamics. For examples, Fama and French (1989) show ha credi spreads (he difference beween corporae bond yield and Treasury bond yield wih an equivalen mauriy) are significanly relaed o macroeconomic condiions and widen during recession periods. Ewing (2003) furher shows ha he defaul risk premium is relaively higher during recessionary condiions. Duffie e al. (2006) show ha macroeconomic fundamenals have an impac on defaul raes or yield spread changes. Korajczyjk and Levy (2003) find ha macroeconomic condiions accoun for 12% o 51% of he variaion in firms leverage beween 1984 and 1998, and Elon e al. (2001) find ha firms leverage has a significan explanaory power for yield spread changes. A recen sudy by Tang and Yan (2006) show ha firm characerisics have significan effecs on credi spreads and hese effecs vary wih economic condiions. Lile empirical work has been done o examine he impac of macroeconomic variables on swap 2153

2 spreads ouside U.S. In his sudy, we aemp o examine he effecs of macroeconomic fundamenals on swap spreads under differen sae of economy. Specifically, our sudy addresses he following ses of quesions. Firs, how do unanicipaed macroeconomic news announcemens impac on he Ausralian dollar ineres rae swap spreads? Second, do hese unanicipaed macroeconomic announcemens affec swap spreads differenly across differen sae of he economy? Third, are hese influences consisen wih exising heories on ineres rae deerminaion? The swap spread is he difference beween he fixed rae of a plainvanilla swap and he yield of a governmen bond of similar mauriy. Since an ineres rae swap is an OTC derivaive and is no exchange raded, i does no enjoy a paymen guaranee by a clearing organizaion or exchange. Consequenly, he pricing of swaps and he corresponding swap spreads reflec credi risk across differen mauriies. Changes in he swap spreads due o news surprises, he unexpeced componen of he news announcemens, should herefore conain new informaion abou defaul risk and he overall credi worhiness of he Ausralian corporae secor in addiion o ineres rae risk. To conduc empirical analysis on he effec of news announcemens, we examine news of money supply growh, unemploymen rae and consumer price index (CPI). We choose hese macroeconomic announcemens because hey are he mos closely wached economic indicaors and are well known o offer insigh ino he inrinsic healh of he economy, he fuure direcion of ineres raes, and he performance of financial markes. A broad consensus has also been reached ha only a small number of macroeconomic facors have a significan impac on pricing and reurn. For ineres rae swaps in paricular, Fornari (2004) finds ha only six macroeconomic variables are influenial on US implied volailiies ha exraced from swapion prices namely US non-farm payrolls, he US Insiue for Supply Managemen (ISM) index, jobless claims, he Chicago Purchasing Managers (CPM) index, durable goods orders and reail sales announcemens. Because some of he news announcemens are no available in Ausralia and hey end o provide similar informaion on he economy, hence, we resric our analysis o he 3 ypes of macroeconomic announcemens. 2. DATA For he 2, 4, 7 and 10-year ineres swap raes, we use he daily closing mid-raes from Daasream over he period from January 3, 1995 o December 31, The Treasury bond yields of corresponding mauriies are obained from he Reserve bank of Ausralia (RBA). Swap spreads for he 6 mauriies are hen calculaed by subracing he swap raes from he Treasury bond yields of he same mauriy, giving a oal sample size of 2533 observaions for each swap mauriy. Table 1 provides some descripive saisics for he swap spreads. I shows ha he average swap spread rises as he mauriy of he swap increases, capuring he risk premium embedded in he swap raes. The volailiy of he swap spread however appears o be fairly consan across all mauriies. Across all mauriies, he skewness and he kurosis of he swap spread show non-normaliy wih some skewness and hin ails. Each of he swap spreads also exhibis non-saionariy based on he Augmened Dickey-Fuller (ADF) es, which fails o rejec he null hypohesis a he 1 percen significan level. We also collec 3 differen scheduled macroeconomic announcemens over he same 10- year period as he swap spreads. Monhly unemploymen rae is compiled by he Ausralian Bureau of Saisics and released a 11:30am on he second Thursday of he monh. Informaion on money supply growh is announced a 11:30 am on he firs Friday of he monh by he Reserve Bank of Ausralia. Finally, quarerly consumer price index is made public a 11:30 am on he las Wednesday of he monh following every quarer (for example, March quarer CPI is released on he las Wednesday of April). In order o examine he exen o which hese economic fundamenals affec he swap marke, i is imporan o undersand and properly model he news informaion arrival. Releases of macroeconomic announcemens are parly anicipaed by he marke. A he poin of announcemen release, he marke only reacs o he unexpeced componen of he news i.e., o he news informaion ha deviaes from marke expecaions. While he announcemen of expeced componen of news informaion has lile influence since i has largely been absorbed in he curren price by he marke prior o he release. Table 1. Descripive saisics of he ineres rae swap spreads. This able repors he summary saisics of he daily ineres rae swaps from January 1995 o December The criical value for -raio a 1% and 5% significan level are and respecively. 2154

3 2-year 4-year 7-year 10-year Mean Medium Sd. Dev Skewness Kurosis ADF -raio N Since economic aciviies end o vary widely over business cycles, one primary quesion of ineres in his sudy is wheher he response of swap spreads o changes in macroeconomic fundamenals varies sysemaically over ime. In oher words, could he same informaion be inerpreed differenly depending on he sae of he economy as he news arrives? To es his hypohesis, we firs need o classify he level of economic aciviies ino wo differen saes - expansions and conracions over he business cycles. Table 2. Ausralian Business Cycles. This Table repors he Ausralian business cycles from January 1990 o December There are 123 expansion and 57 conracion monhs. Turning Poin Peak Trough Conracion (Peak Trough) Duraion in Monhs Expansion (Trough Peak) We measure expansions and conracions using he local maxima and minima of he sample pah of Gross Domesic Produc (GDP), he naural measuremen of he level of economic aciviies. Wih he business cycle reference daes of he Ausralian Bureau of Saisics (1992), we are able o idenify peaks and roughs in Ausralia s business cycle from 1990 o We denoe he period from rough o peak as expansions and from peak o rough as conracions. Table 2 shows he daes of urning poins in he Ausralian business cycles ogeher wih he duraion of each phase of cycle. As expeced, we find ha here are more expansions (123 monhs) han conracions (57 monhs). The Ausralian economy had paricularly been doing well from 1995 o 2000 during which expansions ook place for mos of he sample period. 3. EMPIRICAL MODELS In our subsequen analysis, we firs examine he surprise effecs of he news arrivals on he swap spread change and is volailiy over he full sample period, and hen proceed o invesigae he effecs in each sae of economy. On he full sample analysis, we run he following MA-EGARCH model, ΔSwapsprd ε = z h ~ (0, hi, ), z ~ iid(0,1) ε ln( hi, ) = β0 + β1 ln( hi, 1) + β2 h ε β 3 h 1 1 = α + ε + c q k= β D π N e, α. ε k k 1 1 NEWS + α N, N + D e, NEWS N, (1) Where: Δ Swapsprd i, = Swap spread change for mauriy I (i=1, 2, 3, 4, 5 and 6 for 2, 3, 4, 5, 7 and 10-year swaps respecively) ε = The error erm is assumed ~ (0, h ); h = Condiional swap spread volailiy for mauriy i; q = Number of moving average erms included in he condiional mean equaion o remove serial correlaion in he esimaed sandardized errors, z ; β 2 = Measures he sign effec where a negaive shock increases volailiy if β 2 < -1, and a posiive shock reduces volailiy if β 2 > 1; NEWS,N,, = The unexpeced componen of each macroeconomic announcemen (N = UE, MS, CPI) as measured by he difference beween he. To es wheher he sae of economy changes he news announcemen effecs, we muliply he unexpeced news NEWS N,, wih a dummy variable D e ha capures he sae of he economy. 4. EMPIRICAL RESULTS Table 3 repors he full sample resuls of he effecs of news surprises on swap spread changes. Among he 3 news announcemens, we find ha only unexpeced inflaion rae is significan in explaining he swap spread changes across all mauriies excep 7-year swap. The inverse relaionship beween inflaion surprises and changes in he swap spread indicaes ha a higher (lower) han expeced inflaion rae conribues o a reducion (widening) in he swap spreads. The changes of he swap 2155

4 spread however could come from changes in Treasury yield or changes in swap rae. In oher words, a reducion of he swap spread may indicae ha he swap rae and Treasury yield move in opposie direcions, or in he same direcion bu differen magniudes. Hence, while a higher han expeced inflaion rae (indicaes a srong economic growh) increases inflaion premium on ineres rae, i also reduces he defaul risk premium of he swap rae. Consisen wih previous sudies (see Roley (1983), Ederingon and Lee (1993), he 4- and 7-year swap spread change respond negaively o unexpeced change in money supply In heir sudy, Urich and Wachel (1981) find ha ineres rae levels have a posiive relaionship wih he unexpeced change in money supply. They inerpre his as an inflaionary effec. Tha is unexpeced change in money supply may exer an upward pressure on ineres rae as he cenral bank may engage in an open marke operaion ha ighens he supply of reserves o offse he unexpeced change. Hence, causing he swap spread change o decrease (as ineres raes are expeced o rise). The 10-year swap spread change is he only swap ha responds negaively and significanly o unanicipaed unemploymen rae change. This suggess ha he arrivals of unemploymen news have he greaes effec on he long erm ineres rae swap spread in boh economic and saisical sense. On he condiional volailiy of he swap spread changes, he esimaes of he lagged volailiy coefficien β 1 are highly saisically significan and fall wihin he range This implies ha he swap spread volailiies are highly persisen across all mauriies. The asymmeric impac (as measured by β 2 ) is no significan across all swap mauriies which means ha negaive shocks have greaer impac on swap marke han he posiive shocks is no eviden. Finally, he coefficiens ( β 3 ) of lagged innovaions are posiive and saisically significan. Table 3. The Impac of macroeconomic News Surprises on Ausralian Ineres Rae Swap Spreads: Full Sample. This able repors quasi-maximum likelihood (QML) esimae resuls for he EGARCH (1,1) model. The sample period is from January, 1995 o December, A *, ** and ** * indicae saisical significance a he 1%, 5% and 10% levels respecively. 2-year swap spread 4-year swap spread 7-year swap spread 10-year swap spread Panel A: Condiional Mean α c ** α i, MS ** * α i, UE ** α i, CPI * ** * Panel B: Condiional Variance Β * * * * β * * * * β β * * * * β i, MS * β i, UE β i, CPI * * In relaion o he unanicipaed macroeconomic announcemens on he condiional volailiy of swap spread change, he unanicipaed news on inflaion rae are found o have negaive effecs on he condiional volailiy of 2-year and 10-year swap spread change and he unexpeced componen of he change in money supply also has a significanly negaive effec on he 10-year condiional volailiy of swap spread change. Overall, he resuls seem o be consisen wih 2156

5 Ederingon and Lee (1996) and Brenner, e al. (2005). Ederingon and Lee (1996) show ha implied volailiy ends o rise in he days before news announcemen release. They also find ha here is a sharp fall in implied volailiies jus afer announcemens. They argue ha he announcemen iself helps marke o resolve uncerainy. Furher ino our analysis, we examine swap spread responses o macroeconomic announcemen surprises under differen sae of economy. Recen sudies (see Boyd e al. (2005) and Anderson e al. (2005)) sugges ha he impac of some macroeconomic variables ends o dominae in periods of economic growh while oher variables may be more influenial in periods of economic slowdowns. Tables 4 and 5 repor he impac of macroeconomic surprises on swap spreads during expansions and conracions respecively. I is ineresing o noe ha in he economic expansion periods (see Table 4), swap marke paricipans seem o ignore he inflaion rae surprises as none of he coefficiens of inflaion rae surprises (measured byα CPI ) in he mean equaion is saisically significan. Oher macroeconomic surprises have he same significan effecs on swap spread change as discussed in he full sample periods. In conras o he findings during economic expansions, inflaion rae surprises appear o be he mos influenial news informaion during conracions. Table 5 shows ha swap spread change across all mauriy is highly responsive o he CPI surprises. The coefficien α for CPI inflaion rae surprises is negaively significan a he 5 percen level for all swap spread change. One explanaion for he significan relaionship during conracions is ha swap marke paricipans pay more aenion o inflaion figure because an increase in inflaion (indicaes fuure ineres raes are expeced o increase) is associaed wih an improving business condiion, which in urn improves credi qualiy and subsequenly causes he swap spread o decline. Unlike during expansions however, news surprises on inflaion rae are no imporan on he swap spread changes. Table 4. The Impac of Macroeconomic News Surprises on Ausralia Ineres Rae Swap Spreads: Expansion Periods. This able repors quasi-maximum likelihood (QML) esimae resuls. The sample period is from January, 1995 o December, A *, ** and ** * indicae saisical significance a he 1%, 5% and 10% levels respecively. 2-year swap spread 4-year swap spread 7-year swap spread 10-year swap spread Panel A: Condiional Mean α c ** α i, MS * * α i, UE ** α i, CPI Panel B: Condiional Variance Β * * * β * * * * β β * * * * β i, MS ** β i, UE β i, CPI * * Finally, resuls in Panel B of Tables 4 and 5 show ha he macroeconomic shocks in general are negaively relaed o changes in condiional volailiy of he swap spreads, suggesing ha 2157

6 hese news worhy announcemens end o reduce uncerainy on he announcemen days in he swap marke during expansion and conracion periods. While here are some variaions on he effecs of he news surprises in differen phase of he business cycles, he impacs appear o be quie consisen. For example, shocks generaed from inflaion rae announcemens are generally negaively significan a he 1 percen level across all mauriies in boh saes of economy. They also end o dominae oher news announcemens in he swap marke. The shocks from he inflaion rae appear o dampen he condiional volailiy of he swap spread. On he oher hand, he shocks from money supply changes are resriced o he condiional volailiy of 10-year swap spread during boh expansions and conracions. The effec from he money supply shocks ends o calm down he condiional volailiy of he swap spread. Table 5. The Impac of Macroeconomic News Surprises on Ausralia Ineres Rae Swap Spreads: Conracion Periods. This able repors quasi-maximum likelihood (QML) esimae resuls for he EGARCH (1,1) model The sample period is from January, 1995 o December, A *, ** and ** * indicae saisical significance a he 1%, 5% and 10% levels respecively. 2-year swap spread 4-year swap spread 7-year swap spread 10-year swap spread Panel A: Condiional Mean α c * α i, MS ** α i, UE α i, CPI * ** ** * Panel B: Condiional Variance Β * * * * β * * * * β β * * * * β i, MS * β i, UE ** *** β i, CPI ** * ** * 5. CONCLUSIONS This sudy invesigaes which macroeconomic news announcemens are more influenial on he behaviour of Ausralian ineres rae swap spreads and o he exen ha hey affec he changes in he swap spread during economic expansions and conracions. Using news releases from money supply growh, unemploymen rae and inflaion rae, we find ha news informaion provided by inflaion rae announcemens is he only imporan news hroughou he business cycles bu mos dominan during economic conracions. Whils he unanicipaed unemploymen rae ends o be more relevan o 10-year swap and he unanicipaed change in money supply ends o be more relevan o 4-and 7-year swaps during expansions. Our findings sugges ha marke paricipans pay aenion o differen macroeconomic news announcemens depending on he sae of he economy. Informaion revealed by hese news releases herefore may vary in heir relevance in each phase of he business cycles. We also find shocks from hese news surprises appear o have significan impac on he condiional volailiy of he swap spreads during boh economic phases. The macroeconomic shocks in general are negaively relaed o changes in condiional volailiy of he swap spreads, suggesing ha hese news worhy announcemens end o reduce uncerainy on he news announcemen days in he swap marke during expansion and conracion periods. While here are some variaions on he effecs of he news surprises in differen phase of he business cycles, he impacs appear o be quie consisen 2158

7 6. REFERENCES Anderson, T.G., Bollerslev, T., Diebold, F.X., and Vega, C. (2003), Micro Effecs of Macro Announcemens: Real-Time Price Discovery in Foreign Exchange, American Economic Review, 93, Anderson, T.G., Bollerslev, T., Diebold, F.X., and Vega, C. (2005), Real-Time Discovery in Sock, Bond and Foreign Exchange Marke Working Paper, Norhwesern Universiy Ausralian Bureau of Saisics (1992), The Business Cycle in Ausralia: 1959 o 1992, Ausralian Economic Indicaors. Bollerslev T. P. and Wooldridge, J. M. (1990). Quasi-maximum likelihood esimaion and inference in dynamic models wih imevarying covariances, Economeric Reviews 11, Bollerslev, T., Ca J. and Song, F.M. (2000), Inraday Periodiciy, Long Memory Volailiy, and Macroeconomic Announcemen Effecs in he U.S. Treasury Bond Marke, Journal of Empirical Finance, 7, Boyd, J.H., Hu, J., and Jagannahan, R. (2005), The Sock Marke s Reacion o Unemploymen News: Why Bad News is Usually Good for Socks, The Journal of Finance, Vol. LX, Brenner, M., Pasquariello, and Subrahmanyam, M. (2005), Financial Markes and he (Macro) Economy, Working Paper: Deparmen of Finance, Universiy of Michigan Business School. Chen, N., Roll, R., and Ross, S. (1986), Economic Forces and he Sock Marke, Journal of Business, 59, Duffie D, Saia L, Wang, K, (2006), Muliperiod Corporae Defaul Predicion wih Sochasic Covariaes, Journal of Financial Economics, forhcoming. Ederingon, L.H. and J.H. Lee (1993), How Marke Process Informaion: News Release and Volailiy. Journal of Finance 48(4): Ederingon, L.H. and J.H. Lee (1996), The Creaion and Resoluion of Marke Uncerainy: The Impac of Informaion Release on Implied Volailiy, Journal of Financial and Quaniaive Analysis 31, pp Elon EJ, Gruber MJ, Agrawal D, Mannc (2001) Explaining he Rae Spread on Corporae Bonds, Journal of Finance 56: Ewing, B.T., (2003), The Response of he Defaul Risk premium o Macroeconomic Shock, Quarerly Review of Economic and Finance 43, Fama, E.F., French, K.R., (1989), Business Condiions and Expeced Reurns on Socks and Bonds, Journal of Financial Economics 25, Fleming, M. J. and Remolona, E. M. (1997), Wha Moves he Bond Marke, Economic Policy Review, Federal Reserve Bank of New York, 3, Fornar F. (2004), Macroeconomic Announcemens and Implied Volailiies in Swapion Markes, BIS Quarerly Review, Sepember, Green, C.T. (2004), Economic News and he Impac of Trading on Bond Prices, Journal of Finance, 59, Korajczyk RA, Levy A, (2003), Capial Srucure Choice: Macroeconomic Condiions and Financial Consrains. Journal of Financial Economics 68: Roley, V.V. (1983), The Response of Shor Term Ineres Rae o Weekly Money Announcemen, Journal of Money, Credi and Banking, 15(3): Simpson, M. W., Ramchander, S., and Chaudhry, M. (2005), The Impac of Macroeconomic Surprises on Spo and Forward Foreign Exchange Markes, Journal of Inernaional Money and Finance, 24, Tang Y. and Hong Y.(2006), Macroeconomic Condiions, Firm Characerisics and Credi Spreads, Journal of Financial Services Research, 29: Urich, T. and Wachel, P. (1981), Marke Response o he Weekly Money Supply Announcemens in he 1970s, Journal of Finance, 36,