Monetary Shocks and REIT Returns *

Transcription

1 Moneary Shocks and REIT Reurns * Don Bredin Universiy College Dublin School of Business, Universiy College Dublin, Blackrock, Couny Dublin, Ireland. don.bredin@ucd.ie Gerard O Reilly Cenral Bank and Financial Services Auhoriy of Ireland Economic Analysis and Research Deparmen, Cenral Bank and Financial Services Auhoriy of Ireland, PO Box 559, Dame Sree, Dublin, Ireland. gerard.oreilly@cenralbank.ie and Simon Sevenson Cass Business School, Ciy Universiy Faculy of Finance, Cass Business School, Ciy Universiy, 06 Bunhill Row, London, ECY 8TZ, UK. Tel: , Fax: , s.sevenson-@ciy.ac.uk * The auhors would like o hank seminar paricipans a Cass Business School, Ciy Universiy and he Universiy of Amserdam and paricipans a he Pacific-Rim Real Esae Sociey conference, he join conference of he Asian Real Esae Sociey and he American Real Esae & Urban Economics Associaion and he 006 Hong Kong-Singapore research symposium for commens on previous drafs of his paper. Auhor o whom correspondence should be addressed.

2 Moneary Shocks and REIT Reurns Absrac We invesigae he influence of unanicipaed changes in US moneary policy on Equiy Real Esae Invesmen Truss (REIT s). Alhough a number of sudies have invesigaed he issue of ineres rae changes, he effec of unanicipaed changes has no previously been addressed in erms of possible effecs on boh REIT s reurns and volailiy. The resuls show a srong response in boh he firs and second momens of REIT reurns o unexpeced policy rae changes. The resuls for he impac of he shock on boh mean and volailiy of reurns is consisen wih resuls from sudies addressing broader equiy markes. However, we find evidence boh agains behavioral changes in volailiy coinciden o US moneary policy decisions and asymmeric responses o he moneary policy shock.

3 Moneary Shocks and REIT Reurns : Inroducion The ineres in he recen appoinmen of Ben Bernanke as he Chairman of he Federal Reserve is a furher indicaion of he primacy of moneary policy as a main ool used by US policy makers in he sabilizaion of inflaion and oupu. The imporance of moneary policy changes and he ransmission of informaion conained herein o asse markes has been a subjec of ineres in a number of papers in recen years. This paper examines he impac of changes in he main moneary policy insrumen in he Unied Saes, he Federal Funds Rae, on Equiy Real Esae Invesmen Truss (REITs). The raionale behind he examinaion of REITs is due o heir unique srucure in comparison o mainsream equiies. In paricular, he requiremen ha a leas 75% of heir asses be invesed in real esae and he minimum 90% payou of heir axable earnings as dividends may lead o a differen response in REIT prices in comparison wih equiies o changes in moneary policy. The impac of policy rae changes on he general equiy marke can be viewed as occurring hrough hree channels. Firsly, he impac on he expeced level of fuure dividends of he firms, secondly, any associaed change in he real ineres rae used o discoun hese dividends and hirdly changes in he equiy risk premium. Given he characerisics of no only REITs bu also he underlying privae real esae marke, a number of aspecs of hese linkages may ake on addiional imporance in he conex of he raded real esae secor. Wih respec o dividends, he 90% dividend payou requiremen will lead o more subsanial income flows from REITs han common socks. Moneary policy changes will have an influence on general economic aciviy ha feeds hrough o occupaional demand in he underlying real esae marke. This will impac upon rens obainable by he REIT from he underlying propery porfolio and hence will direcly affec he dividend paymens of he firm. In addiion, rae changes will also influence he value of he underlying porfolio. No only will changes in renal income impac on propery values bu furhermore, given he linkages beween he space and capial markes (DiPasquale & Wheaon, 99 and Fisher, 99), here is an impac hrough cap raes on he value of he underlying porfolio. These effecs 3

4 mean ha REITs are far more heavily ied o heir underlying asse base han boh equiies generally and oher forms of real esae securiies, such as corporae based vehicles in markes such as he UK and Hong Kong. I also means ha he response of REITs o changes in moneary policy may differ from he general evidence regarding he sock marke. A furher facor ha may also lead o differences in he resuls for REITs in comparison o he overall equiy marke is he relaive size and mauriy of he secor. While REITs were esablished by Congress in 960 heir growh has largely occurred since he early nineies. In 99, for example, he oal marke capializaion of he equiy REIT secor was $8,785m according o NAREIT (Naional Associaion of Real Esae Invesmen Truss). A he end of 005 his had increased o over $300bn, while he number of Equiy REITs had increased from 86 o 5. Amongs oher papers, Coer & Sevenson (006) noe in heir examinaion of REIT volailiy ha his growh in he secor has led, paricularly in recen years, o changes in he dynamics in he secor. However, while subsanial growh has occurred REIT s sill mainly comprise of small and mid cap firms, wih an average size of jus under $bn. Our mehodological approach draws on he recen work of Bomfim (003), Jones e al. (998) and Anderson and Bollerslev (998). The ransmission of moneary policy informaion is assessed hrough an analysis of meeings of he Federal Open Marke Commiee (FOMC). As in papers such as Bomfim (003) we proxy marke expecaions concerning changes in he Fed Funds Rae hrough changes in he Fed Funds Fuures Conrac. Our paper examines hree key hypoheses. Firsly, we invesigae he impac of unanicipaed changes in moneary policy on boh he reurns and volailiy of he REIT secor. Secondly, we es for asymmery in he response wih respec o a posiive or negaive unanicipaed change in he policy rae. Finally, behavioral changes in REIT volailiy around he ime of FOMC meeings are considered. In paricular, we invesigae wha is commonly referred o as he calm before he sorm effec. This refers o he fac ha volailiy ends o fall immediaely prior o an announcemen. This effec has been noed by Jones e al. (998), Li & Engle (998) and Bomfim (003) for he Treasury Bond, Treasury Bill and Sock markes respecively. 4

5 The remainder of he paper is laid ou as follows. The following secion briefly reviews he exising lieraure in relaion o ineres rae changes on REIT s, he effec of anicipaed versus unanicipaed changes and finally behavioral changes around even days. Secion 3 deails he mehodological approach and he daa requiremens. Secion 4 conains he main empirical findings and he concluding commens are provided in he final secion. : Lieraure Review. Ineres Rae Changes and REIT s While here have been a number of recen sudies ha have analyzed he relaionship beween REITs and ineres rae changes, no accoun has been explicily aken for he degree o which hese changes have been anicipaed. A number of papers have shown ha he sensiiviy of REITs o ineres raes is boh ime-varying and also dependen on he rae used. Boh Chen & Tzang (988) and Liang e al. (995) find evidence of insabiliy in heir findings dependen on he exac ime period examined. This resul has been corroboraed by Devaney (00) and He e al. (003) 3. He e al. (003) also confirm previous findings showing ha REITs are mos sensiive o changes in long-erm yields and low-grade corporae bonds alhough, as wih oher proxies used, hese findings are also ime-varying. This is a finding ha is consisen wih he lieraure o have examined financial insiuions (e.g. Kane & Unal, 988). Devaney (00) uilizes a GARCH-M model similar o ha used in he broader ineres rae sensiiviy lieraure such as Elyasiani & Mansur (998). This is one of he few papers o have exended he analysis o examine he impac of ineres raes on REIT volailiy. The resuls illusrae he difference in focus beween he Equiy and Morgage REIT sub-secors. While highly significan findings are repored wih regard o he morgage secor he resuls for Equiy REITs differ. While he coefficiens are largely of he anicipaed sign hey are generally insignifican. I should be noed however ha Devaney (00) analyzed monhly daa 4. A recen paper by Coer & Sevenson (006) examines daily REIT volailiy. While he focus of ha paper is no concerned wih ineres rae sensiiviy, Treasury Bills are incorporaed ino he mulivariae GARCH model used o examine he underlying volailiy 5

6 dynamics of REITs. The resuls show ha Treasury Bill movemens are significan in erms of boh reurns and volailiy for Equiy REITs.. Anicipaed versus Unanicipaed Changes Given ha he curren paper analyses he impac of official rae changes and decisions of he FOMC, he analysis here links in wih he broader lieraure ha has examined he impac of macroeconomic variables on equiy markes. Flannery & Proopapadkis (00) examine he effec of 7 differen macroeconomic announcemens on equiy reurns. They find evidence ha six (CPI, PPI, a moneary aggregae, balance of rade, employmen and housing sars) are priced. However, only unanicipaed money supply announcemens influence boh he firs and second momen of sock reurns. Connolly & Wang (003) examine he impac of moneary announcemens in an inernaional environmen looking a he US, UK and Japan 5. One ineresing resul from his sudy is evidence supporive of an asymmeric response in erms of wheher he announcemen conained good or bad news. This is a similar finding o ha repored by Bomfim (003) and is consisen wih he leverage effec noed by Black (976). A large lieraure has examined he specific impac of US moneary policy on equiy markes 6. An imporan issue in any examinaion of rae changes by he Federal Reserve is ha of echnical and non-echnical rae changes. Prior o 979 he Federal Reserve effecively changed he discoun rae o bring i ino line wih marke raes 7. Boh Smirlock & Yawiz (985) and Pearce & Roley (985) provide evidence on he impac of rae changes on he sock marke. Pearce & Roley (985) is one of he firs sudies o spli he rae change ino is expeced and unexpeced componen, hrough he use of survey daa. Pos 979 and he change in rae change policy he auhors show ha sock prices reac significanly o unanicipaed changes in he discoun rae. A furher change in he operaion of he Federal Reserve occurred in 994. Prior o February 994 he Federal Reserve would effecively release informaion on rae changes he day afer a FOMC meeing hrough he Open Marke Desk. However, pos February 994 rae changes have been publicly announced direcly afer each FOMC meeing. 6

7 Thorbecke (997) provides empirical evidence concerning he influence of moneary policy on socks. The paper uilizes he Federal Funds Rae and non-borrowed reserves. While he general resuls highligh ha an expansionary moneary policy increases ex-pos reurns, an ineresing elemen of he analysis is ha asymmeries in he responses may also help o explain he findings of Fama & French (995). The auhors find ha moneary shocks affec smaller firms o a greaer exen han large firms. I is hypohesized ha his is due o he impac on credi availabiliy noed by Gerler & Gilchris (994). Given he relaive size of REITs i may herefore be expeced ha his would lead o an enhanced sensiiviy in comparison o he overall marke. However, i should be remembered ha he ax saus of REITs also brings ino quesion he ax advanages of deb issuance. Kuner (00) assesses he influence of policy based rae changes by he Federal Reserve on marke raes. Marke raes are proxied by Treasury bill, noe and bond yields. The resuls highligh he imporance of decomposing expeced and unexpeced componens of moneary policy changes. While expeced rae changes are no saisically significan, unexpeced rae changes resul in a large and significan response in marke raes. Paelis (997) noes ha moneary policy changes can also provide valuable predicive informaion on fuure sock marke movemens. Furhermore, Rigobon & Sack (003) find ha he relaionship beween ineres raes and sock prices is a bilaeral one, reporing evidence ha sock marke behavior influences fuure ineres rae movemens. A recen paper by Bernanke & Kuner (005) adops boh an even sudy mehodology and a VAR model of he ype proposed by Campbell (99). The even sudy resuls show a significan response o unanicipaed changes in he rae. The VAR analysis finds ha he primary impac of rae changes ono prices is hrough heir impac on expeced fuure excess reurns 8..3 Behavioral Effecs around Even Days The likely behavioral changes in he volailiy of asse reurns around paricular even windows and ess of he calm before he sorm effec have has been invesigaed by Jones e al. (998). The paper examines he impac of employmen and PPI announcemens on bond reurns. They also examine wha may cause volailiy persisence. They find no evidence of persisence in volailiy following a moneary announcemen, concluding insead ha volailiy persisence may be a resul of he 7

8 clusering of news announcemens. Using he Michell & Mulherin (994) daabase of news evens hey illusrae ha informaion announcemens are posiively auocorrelaed a significan levels a a daily inerval 9. Bomfim (003) examines he S&P 500 Composie and is response o Fed Funds Rae changes. The auhor iniially finds no evidence of he calm before he sorm effec. While volailiy is higher on he day of he announcemen, here is no reducion in volailiy on he day prior o he FOMC meeing. However, he sample examined in his paper exends back o 989 and herefore pre-daes he change in Federal Reserve policy in February 994. As noed by Bomfim (003) beween 989 and 994 only 4% of rae changes were aken a scheduled meeings of he FOMC. However, since he policy change in 994 he vas majoriy of rae changes have coincided wih FOMC meeings. In he Bonfim (003) sample, ending in 998, 85% of all rae changes occurred and were announced on meeing days. Once he auhor akes accoun of pre and pos 994, he finds significan evidence of a calm before he sorm effec. Finally, Bonfim (003) also finds evidence ha he effec of he shock on volailiy is asymmeric. 3: Daa and Mehodological Framework Our mehodology draws on he recen work of Bomfim (003), Jones e al. (998) and Anderson and Bollerslev (998). Firsly, we examine he impac of FOMC announcemens on boh he reurns and volailiy of he REIT secor. Specifically, by spliing he rae change ino is anicipaed and unanicipaed componens he analysis allows an examinaion of he impac of unexpeced rae changes. Secondly, he behaviour of REIT reurns around he ime of FOMC meeings is considered and we invesigae he calm before he sorm effec. The daa is his paper is daily and exends from 3 s January 996 hrough o March s 005. A poenial issue wih he use of daily daa is ha i may mask he exac impac. In paricular, i is hard o isolae he impac of Federal Funds Rae changes as oher announcemens may be made ha day. However, as Bomfim (003) noes, FOMC meeings do no sysemaically coincide wih any one economic dae release. The REIT marke is proxied by he Dow Jones-Wilshire Equiy REIT Index. As noed 8

9 in he inroducion, his paper solely examines he Equiy REIT secor and does no examine, eiher in aggregae or in isolaion, he Morgage REIT secor. The change in he Federal Funds Targe Rae was obained from he Federal Reserve Board of Governors. The proxy used for he unanicipaed change in he arge rae is he -day change in he price of he -monh ahead 30-day Federal Funds Fuures Conrac raded on he Chicago Board of Trade (CBOT). Previous papers o have uilized such a proxy for moneary policy changes include: Bomfim & Reinhar (000), Kuner (00), Poole & Rasche (000), Reinhar & Simin (997), Roley & Sellon (998) and Thornon (998). Previous empirical work in he field such as Connolly & Wang (003), Flannery & Proopapadkis (00) and Li & Engle (998) use alernaive measures of expecaions. These alernaives include he growh rae of money supply and survey daa, however, Gurkaynak e al. (00) show ha he fed funds fuures conrac provides he bes available forecas of he Feds Fund Rae 0. The modeling approach adoped here is based on ha used by Bomfim (003) and Jones e al. (998). The GARCH model can be specified as follows: REIT = β 0 + β FFF + β REIT + β 3SP + β 4Mon + β5tue + β 6Thu + β 7Fri + µ () µ = e s () e E E = υ h (3) ( e Ω ) = 0 ( e Ω ) = h ( Ω ) = sh E u (4) h α α α (5) = + 0 h + e The REIT series is he dependen variable in he condiional mean equaion. The independen variables comprise of he -day change in he fed funds fuures (FFF), he lagged one-day REIT reurn and he S&P 500. Dummy variables for days of he week are also incorporaed ino he specificaion. The unexplained componen (µ + ) comprises of a non-normal sochasic elemen (e + ) whose condiional variance is 9

10 ime-varying and a variable (s + ). The variable indicaes he impac of paricular day effecs and can be expressed as: s δ δ δ δ δ ( F ) ( F ) ( F ) = + 0 I + I + I + + 3Mon + 4Tue + 5Thu + 6Fri + FFF (6) δ δ φ Where I (F) is a dummy se o uniy when here is FOMC meeing and zero oherwise. The model is esimaed using he quasi-maximum likelihood procedure proposed by Bollerslev & Wooldridge (99). As previously menioned, hree key hypoheses are esed. The firs hypohesis relaes o a news effec and wheher an unanicipaed change in he fed funds rae has any effec on he REIT secor. This is examined hrough he condiional mean equaion. The hypohesis would be suppored if β is negaive and saisically significan. We also address wheher he shock o moneary policy has any effec on he second momens, which would be highlighed by he saisical significance of φ in Equaion (6). The possibiliy ha here may be an asymmeric volailiy effec (ha higher han expeced changes in raes will lead o grea volailiy) will also be considered. The final hypohesis relaes o he calm before he sorm effec. This refers o a hypohesized lower level of volailiy on he day before FOMC meeings and higher on he day of he announcemen iself. This is esed based he resuls from Equaion (6). The hypohesis is confirmed if coefficien δ 0 is posiive and significan and δ is negaive and significan a convenional levels. 4: Empirical Resuls The model is esimaed under a variey of differen scenarios. The iniial examinaion concenraes upon changes in he Fed Fund Fuures on FOMC meeing days. We hen exend his o also incorporae unscheduled rae changes ha ake place ouside of scheduled FOMC meeings. The iniial analysis is repored in Table. From hese resuls i can be seen ha he change in Fed Funds Fuures impacs significanly on boh he mean and volailiy of he Equiy REIT secor. Furhermore, he signs of he coefficiens in relaion o he mean and volailiy equaions have he anicipaed sign, i.e., negaive and posiive respecively. This alone is ineresing given he frequen 0

11 lack of consisen findings in previous sudies of REIT sensiiviy o ineres rae movemens. The Devaney (00) paper adops he mos similar mehodological approach, in ha a GARCH based model, in his case a GARCH-in-Mean specificaion, is used. However, he analysis on marke raes generally finds an insignifican response in eiher he mean or variance equaions. Only when he Morgage REIT secor is examined are significan coefficiens repored. This divergence in findings highlighs he imporance of aking ino accoun marke expecaions and incorporaing ino he model specificaion he unanicipaed naure of he rae change. In addiion, i should also be reieraed ha he Devaney (00) paper examined monhly no daily daa. As would also be expeced, he coefficiens relaing o he lagged REIT secor and he marke index, as proxied by he S&P 500, are posiive and significan a convenional levels. There is also evidence of GARCH effecs in he model, jusifying he use of his form of specificaion. One issue relaing o he day of he week dummies ha deserves menion is ha in boh he mean and variance equaions he coefficiens referring o Friday are posiively signed and significan a convenional levels. This indicaes a Friday effec in boh he firs and second momens of daily REIT index daa. The second issue relaes o he hypohesized calm before he sorm. As noed, his is esed hrough an examinaion of coefficiens δ 0 and δ in he variance equaion. Unlike previous empirical evidence such as Bomfim (003) we find no evidence of such an effec. For he hypohesis o be suppored, δ 0 should be posiive and significan and δ negaive and significan. The resuls show ha neiher coefficien is saisically significan. Hence, our resuls indicae ha while here is a reacion in erms of boh REIT s reurns and heir volailiy o new informaion, we find no behavioral change around he ime scheduled ineres rae changes. The behavioral insignificance may occur for a number of reasons including issues relaed o REIT s reurns or he use of scheduled ineres rae changes. I is possible ha non-synchronous and hin rading leads o he behavioral insignificance in he Wilshire REIT index hereby conribuing o hese findings. As noed previously,

12 despie he increase in boh he size of he REIT secor and he corresponding increase in rading volume in recen years, he secor is relaively small. While he average marke cap in he secor was jus under $bn as of he end of 005, 46% of he firms had a marke value less han $bn. The use of individual REIT reurns and he separae examinaion of REITs of differing levels of boh marke value and rading volume may produce more conclusive findings in his regard. In comparison o he findings of Bomfim (003) in relaion o he S&P 500, i should be emphasized ha his iniial lack of significan evidence was in relaion o he sample pre-daing he change in Fed policy in 994. Once his was accouned for in he analysis significan resuls were repored. As our sample daes only from 996 he change in Fed policy canno be a possible reason behind he lack of significan evidence. In order o furher examine possible causes behind he differences in hese findings we re-esimae he model under wo alernaive scenarios. The firs is o use a porfolio of large cap heavily raded REITs o check for possible non-synchronous rading effecs. The porfolio is creaed using he larges percenile of Equiy REITs in each year during our sample. As wih he Dow Jones-Wilshire index he porfolio is valueweighed. The second scenario is o re-esimae he ess using he S&P 500 o check for consisency in he Bomfim (003) resuls and o exacly mach he sample period. The resuls for boh he large-cap REIT porfolio and he S&P are conained in Table. The resuls are ineresing in a number of respecs. Firsly, wih regard o he large cap REIT porfolio, here is no discernable difference in he resuls, indicaing ha he previously repored findings for he overall secor index were no unduly influenced by non-synchronous rading effecs. In general here are no subsanial changes in he coefficiens repored wih significan impacs on REIT volailiy recorded, he significance of coefficiens relaing o he lagged REIT and S&P reurns in he mean equaion and he lack of significance in relaion o he calm before he sorm. One difference is ha changes in he fed funds fuures rae does no significanly impac upon REIT reurns. This view is o some exen also suppored in wih respec o he resuls for he S&P 500. These findings, using a maching sample period, differ subsanially from hose conained in he Bomfim (003) paper and sugges ha his findings are sensiive o he exac sample used. As previously noed, Bomfim (003) argues ha he significan

13 resuls obained followed he change in Federal Reserve operaing procedure in 994, using a sample of daa ha ends in December 998. The change in he fed funds fuures does no impac significanly on eiher reurns or volailiy in he S&P 3. Furhermore, here is no evidence of a significan calm before he sorm effec in his sample period. The lack of significance in he mean equaion and he similar finding in relaion o large cap REITs would sugges ha he findings in he iniial analysis are indeed driven by he enhanced exposure of smaller REITs wih regard o changes in he fed funds rae. In relaion o he variance equaion, he fac ha boh REIT series repored significan responses in volailiy whereas he S&P did no would sugges ha i is more of a REIT specific issue raher han merely one of firm size. As noed in he inroducion here is a srong inuiive argumen as o why Real Esae Invesmen Truss may display a higher degree of sensiiviy o ineres rae movemens in comparison o he general sock marke. This comes from he muliple impacs of rae changes wih regard o issues such as he effec on cap raes and he poenial impac on underlying occupaional demand in he propery marke. The analysis conained in Table is solely concerned wih rae changes announced a scheduled meeings of he FOMC. In order o consider he sensiiviy of our resuls we invesigae he impac of rae changes on all announcemen days, boh scheduled and unscheduled. While he number of unscheduled announcemens has fallen dramaically in recen years i is sill an imporan issue o consider. This is paricularly he case for he evens of 00. During he firs half of 00 here were wo unscheduled rae changes (ineres rae reducions), 3 rd January and 8 h April. These wo paricular unscheduled rae changes are noeworhy given he Fed s preference for scheduled rae changes in recen years and he fac ha hey were boh 50 basis poin reducions. In addiion, he impac of 9/ was also a major facor on he markes ineres rae expecaions and he acions of he Federal Reserve. For his reason, we exend he analysis, as repored in Table 3, o include rae changes ha occurred ouside of he auspices of a scheduled FOMC meeing 4. There are relaively few changes in he resuls afer his exension of he analysis. As wih he original specificaion, GARCH effecs are eviden, here is he anicipaed 3

14 influence of boh lagged REIT reurns and he conemporaneous S&P 500 in he mean equaion and evidence of a Friday effec on boh reurns and volailiy. In addiion, as wih he resuls previously discussed here is no significan evidence of a calm before he sorm effec. Finally, here is evidence ha here is a significan response o he unanicipaed componen of he rae change in erms of boh he mean and variance equaion. As wih he iniial analysis we also re-run he analysis using he large-cap REIT porfolio and he S&P 500. The paerns in he findings are broadly similar o hose in he iniial ess wih a lack of significance in he mean equaion for he large cap REIT porfolio. One difference is however observed in relaion o he S&P in ha while volailiy does no significanly fall he day preceding an ineres rae announcemen i does rise o a significan exen on he day of he announcemen. This makes inuiive sense in ha he sole difference in his analysis is ha i incorporaes unscheduled announcemens and herefore any rae changes will no have been anicipaed o he same exen as wih hose made a FOMC meeings 5. Given he consisen saisically significan effec of he shock on boh he mean and he variance, in he final par of he analysis we invesigae wheher his effec is asymmeric. This analysis is based on he leverage effec noed by Black (976) and he volailiy feedback hypohesis of French e al. (989). This has been suppored empirically in papers such as French e al. (989) and Nelson (99) while asymmery has also been repored in papers closely relaed o he curren sudy such as Bomfim (003) and Connolly & Wang (998). To examine his issue he variance equaion is adjused o ake he following form. s = + δ I + φ FFF + + δ I + φ FFF + + δ Mon + δ Tue + δ Thu + δ Fri (7) where posiive and negaive unexpeced changes in he fed funds fuures rae are separaed. 6 7 The resuls, conained in Table 4, show no evidence of any asymmery wih respec o he shock on REIT s volailiy. The exising evidence on mainsream equiy reurns has largely found evidence of an asymmerical response, wih an enhanced rise in volailiy following a negaive shock, i.e. a higher han anicipaed rise in raes, in comparison o posiive shocks. However, in he case of REITs boh 4

15 coefficiens are of he same sign and are no saisically differen, wih a p-value from he Wald saisic of 0.3. The resuls for he large-cap REIT porfolio and he S&P 500 also fail o repor significan findings 8. Given he lack of significan findings in relaion o eiher large-cap REIT porfolio or he S&P 500, i is impossible o sae wheher he overall REIT resuls are due o non-synchronous or hin rading issues or o a REIT specific issue. However, hey do however indicae ha previous general empirical evidence on asymmery is perhaps sensiive o he sample period used. 5: Conclusions This sudy offers a number of imporan conribuions in he analysis of he sensiiviy of REITs o changes in ineres raes. Firsly, i has, hrough he use of he fed funds fuures, indicaed he effec of unanicipaed changes in ineres raes on boh REIT s reurns and heir volailiy. Alhough here is an esablished lieraure addressing he influence of ineres rae changes on REITS reurns, his is he firs paper o explicily ake accoun of he impac of unanicipaed changes. Secondly, i has specifically esed for boh asymmeric responses in volailiy o ineres rae movemens and he calm before he sorm effec. The analysis provides a number of ineresing resuls, relaive o previous sudies on REIT s reurns and volailiy, bu also relaive o mainsream equiies. In comparison o previous sudies of REIT ineres rae sensiiviy he main resuls do show significan responses in boh reurns and volailiy o unanicipaed rae changes. The imporance of he specificaion of unanicipaed changes in ineres raes is criical in considering he previous resuls where ineres rae changes were adoped. Alhough he effec of he shock is significan on boh reurns and volailiy, here is no evidence of asymmery. There is also no evidence of changing volailiy behavior coinciden o Federal Reserve announcemens, calm before he sorm. While he lack of any behavioral effec may appear inconsisen wih he previous resuls repored for equiies we also repor resuls for boh large cap REIT s and equiies ha are consisen wih an absence of a significan calm before he sorm effec, which would indicae ha previously repored resuls are sensiive o he exac sample analyzed. 5

16 References Allen, M., Madura, J. & Springer, T. (000). REIT Characerisics and he Sensiiviy of REIT Reurns, Journal of Real Esae Finance and Economics,, 4-5. Andersen, T. & Bollerslev, T. (997). Inraday Periodiciy and Volailiy Persisence in Financial Markes, Journal of Empirical Finance, 4, Bae, S.C. (990). Ineres Rae Changes and Common Sock Reurns of Financial Inermediaries, Journal of Financial Research, 3, Bernanke, B.S. & Kuner, K.N. (005). Wha Explains he Sock Marke s Reacion o Federal Reserve Policy?, Journal of Finance, 60, -57. Black, F. (976). Sudies in Sock Price Volailiy, Proceedings of he Meeings of he American Saisical Associaion, Business and Economics Secion, Bollerslev, T. & Wooldridge, J. (99). Quasi Maximum Likelihood Esimaion and Inference in Dynamic Models wih Time-Varying Variances, Economeric Review,, Bomfim, A. (003). Pre-announcemen Effecs, News Effecs and Volailiy, Journal of Banking & Finance, 7, Bomfim, A. & Reinhar, V. (000). Making News: Financial Marke Effecs of Federal Reserve Disclosure Pracices, Manuscrip, Federal Reserve Board. Campbell, J. (99). A Variance Decomposiion for Sock Reurns, The Economic Journal, 0, Casanias, R. (979). Macroinformaion asnd he Variabiliy of Sock Marke Prices, Journal of Finance, 34, Chen, C.R., Mohan, N. & Seiner, T.L. (999). Discoun Rae Changes, Sock Marke Reurns, Volailiy and Trading Volume: Evidence form Inraday Daa and Implicaions for Marke Efficiency, Journal of Banking & Finance, 3, Chen, S., Hsieh, C. & Jordan, B. (997). Real Esae and he Arbirage Pricing Theory: Macroeconomic Variables vs. Derived Facors, Real Esae Economics, 5, Chen, S., Hsieh, C., Vines, T. & Chiou, S. (998). Macroeconomic Variables, Firm Specific Variables and Reurns o Equiy REITs, Journal of Real Esae Research, 6, Chen, K. & Tzang, D. (988). Ineres-Rae Sensiiviy of Real Esae Invesmen Truss, Journal of Real Esae Research, 3, 3-. Connolly, R. & Wang, F. (003). Economic Equiy Marke Comovemens: Economic News or Conagion, Pacific Basin Finance Journal,, Conover, C.M., Jensen, G.R. & Johnson, R.R. (999). Moneary Environmens and Inernaional Sock Reurns, Journal of Banking & Finance, 3, Cook, T. & Hahn, T. (988). The Informaion Conen of Discoun Rae Announcemens and Their Effec on Marke Ineres Raes, Journal of Money, Credi & Banking, 0,

17 Coer, J. & Sevenson, S. (006). Mulivariae Modelling of Daily REIT Volailiy, Journal of Real Esae Finance & Economics, 3, Dara, A. & Glascock, J. (989). Real Esae Reurns, Money and fiscal Deficis: Is he Real Esae Marke Efficien?, Journal of Real Esae Finance & Economics,, Devaney, M. (00). Time-Varying Risk Premia for Real Esae Invesmen Truss: A GARCH-M Model, Quarerly Review of Economics & Finance, 4, DiPasquale, D. & Wheaon, W. (99). The Markes for Real Esae Asses and Space: A Concepual Framework, Journal of he American Real Esae & Urban Economics Associaion, 0, Dueker, M. (99). The Response of Marke Ineres Raes To Discoun Rae Changes, Federal Reserve Bank of S. Louis Review, 74, Elyasiani, E. & Mansur, I. (998). Sensiiviy of he Bank Sock Reurns Disribuion o Changes in he Level and Volailiy of Ineres Raes: A GARCH-M model, Journal of Banking & Finance, : Fama, E.F. & French, K.R. (995). Size and Book-o-Marke Facors in Earnings and Reurns, Journal of Finance, 50, Fisher, J. (99). Inegraing Research on Markes for Space and Capial, Journal of he American Real Esae & Urban Economics Associaion, 0, Flannery, M.J. & James, C.M. (984). The Effec of Ineres Rae Changes in The Common Sock Reurns of Financial Insiuions, Journal of Finance, 39, Flannery, M.J., Hameed, A.S. and Harjes, R.H. (997). Asse Pricing, Time-Varying Risk Premia and Ineres Rae Risk, Journal of Banking & Finance, : Flannery, M.J. & Proopapadakis, A.A. (00). Macroeconomic Facors Do Influence Aggregae Sock Reurns, Review of Financial Sudies, 5, French, K., Schwer, W. & Samburgh, R. (989). Expeced Sock Reurns and Volailiy, Journal of Financial Economics, 9, 3-9. Gerler, M. & Gilchris, S. (994). Moneary Policy, Business Cycles and he Behavior of Small Manufacuring Firms, Quarerly Journal of Economics, 09, Gurkaynak, R., Sack, B. & Swanson, E. (00). Marke Based Measures of Moneary Policy Expansion, working paper, Board of Governors of he Federal Reserve Sysem. He, L.T., Webb, J.R. & Myer, F.C.N. (003). Ineres Rae Sensiiviies of REIT Reurns, Inernaional Real Esae Review, 6, -. Jensen, G.R. & Johnson, R.R. (995). Discoun Rae Changes and Securiy Reurns in he US, 96-99, Journal of Banking and Finance, 9, Jensen, G.R., Mercer, J.M. & Johnson, R.R. (996). Business Condiions, Moneary Policy and Expeced Securiy Reurns, Journal of Financial Economics, 40, Jones, C.M., Lamon, O. & Lumsdaine, R.L. (998). Macroeconomic News and Bond Marke Volailiy, Journal of Financial Economics, 47,

18 Kane, E.J. & Unal, H. (988). Change in Marke Assessmen of Deposi Insiuion Riskiness, Journal of Financial Services Research,, 0-9. Kuner, K.N. (00). Moneary Policy Surprises and Ineres Raes: Evidence from he Feds Funds Fuures Marke, Journal of Moneary Economics, 47, Kwan, S.H. (99). Reexaminaion of Ineres Rae Sensiiviy of Commercial Bank Sock Reurns using a Random Coefficien Model, Journal of Financial Services Research, 5, Lasrapes, W.D. (998). Inernaional Evidence on Equiy Prices, Ineres Raes and Money, Journal of Inernaional Money and Finance, 7, Li, K. & Engle, R. (998). Macroeconomic Announcemens and he Volailiy of he Treasury Marke, Deparmen of Economics, Universiy of California, San Diego, Working Paper Liang, Y., McInosh, W. & Webb, J. (995). Ineremporal Changes in he Riskiness of REITs, Journal of Real Esae Research, 0, Liang, Y. & Webb, J. (995). Pricing of Ineres Rae Risk for Morgage REITs, Journal of Real Esae Research, 0, Ling, D. & Naranjo, N. (997). Economic Risk Facors and Commercial Real Esae Reurns, Journal of Real Esae Finance & Economics, 4, Lynge, M.J. & Zumwal, J.K. (980). An Empirical Sudy of he Ineres Rae Sensiiviy of Commercial Bank Reurns: A Muli-Index Approach, Journal of Financial and Quaniaive Analysis, 5, McCue, T. & Kling, J. (994). Real Esae Reurn and he Macroeconomy: Some Empirical Evidence from Real Esae Invesmen Trus Daa, 97-99, Journal of Propery Research, 9, Michell, M. & Mulherin, J. (994). The Impac of Public Informaion on he Sock Marke, Journal of Finance, 49, Mueller, G. & Pauley, K. (995). The Effec of Ineres Rae Movemens on Real Esae Invesmen Truss, Journal of Real Esae Research, 0, Nelson, D. (99). Condiional Heeroscedasiciy in Asse Reurns: A New Approach, Economerica, 59, Paelis, A.D. (997). Sock Reurn Predicabiliy and he Role of Moneary Policy, Journal of Finance, 5, Pearce, D.K. & Roley, V.V. (985). Sock Prices and Economic News, Journal of Business, 58, Poole, W. & Rasche, R.H. (000). Perfecing he Marke s Knowledge of Moneary Policy, Journal of Financial Services Research, 8, Reinhar, V. & Simin, T. (997). The Marke Reacion o Federal Reserve Policy Acion from 989 o 99, Journal of Economics and Business, 49,

19 Rigobon, R. & Sack, B. (003). Measuring he Reacion of Moneary Policy o he Sock Marke, Quarerly Journal of Economics, 8, Rigobon, R. & Sack, B. (004). The Impac of Moneary Policy on Asse Prices, Journal of Moneary Economics, 5, Roley, V. & Sellon, G. (998). Marke Reacion o Moneary Policy Non-Announcemens, Federal Reserve Bank of Kansas Ciy, working paper Roley, V. & Troll, R. (984). The Impac of Discoun Rae Changes on Marke Ineres Raes, Federal Reserve Bank of Kansas Ciy Economic Review, January, Smirlock, M. & Yawiz, J. (985). Asse Reurns, Discoun Rae Changes, and Marke Efficiency, Journal of Finance, 40, Sevenson, S., Wilson, P. & Zurbrugg, R. (005). Assessing he Time-Varying Ineres Rae Sensiiviy of Real Esae Securiies, working paper, Cenre for Real Esae Research, Smurfi School of Business, Universiy College Dublin. Sone, B.K. (974). Sysemaic Ineres Rae Risk in a Two-Index Model of Reurns, Journal of Financial and Quaniaive Analysis, 9, Swanson, Z., Theis, J. & Casey, K. (00). REIT Risk Premium Sensiiviy and Ineres Raes, Journal of Real Esae Finance & Economics, 4, Sweeney, R.J. & Warga, A.D. (986). The Pricing of Ineres Rae Risk: Evidence from he Sock Marke, Journal of Finance, 4, Thorbecke, W. (997). On Sock Marke Reurns and Moneary Policy, Journal of Finance, 5, Thornon, D. (998). Does he Fed s New Policy of Immediae Disclosure Affec he Marke, working paper, Federal Reserve Bank of S. Louis. Waud, R. (970). Public Inerpreaion of Federal Reserve Discoun Rae Changes: Evidence on he Announcemen Effec', Economerica, 38, Yourougou, P. (990). Ineres-Rae Risk and he Pricing of Deposiory Financial Inermediary Common Sock, Journal of Banking & Finance, 4,

20 Tables Table : Impac of US Moneary Policy Shocks on he Mean & Volailiy of REIT s (Scheduled Announcemens) REIT= β + β FFF + β REIT + β S& P + β Mon+ β Tue+ β Thu+ β Fri 0 0 s = + δ I h = α + α h 0 + δ I + α e + δ I + + δ Mon+ δ Tue+ δ Thu+ δ Fri + φ FFF Variable Coefficien -saisic Panel A: Mean Equaion β β -0.84* β 0.8* 9.80 β 3 0.4* 8.70 β β * 3. β β 7 0.* 3.37 Panel B: Variance Equaion α 0 0.0* 4.9 α 0.3* 9.06 α 0.80* 4.8 δ δ δ δ δ δ δ 6 0.*.35 φ 0.47*.04 Using one day change in monh ahead federal funds fuure conrac as unanicipaed change. The saisics are robus using he procedure from Bollerslev & Wooldridge (99). * indicaes saisical significance a convenional levels

21 Table : Impac of US Moneary Policy Shocks on he Mean & Volailiy of Large Cap REIT s and S&P 500 (Scheduled Announcemens) LREIT= β + β FFF + β LREIT + β S & P + β Mon+ β Tue+ β Thu+ β Fri 0 s = + δ I 0 h = α + αh s = + δ I 0 + α e + δ I + δ I + S & P = β + β FFF + β S & P 0 0 h = α + α h 0 + α e + δ I 3 + δ I + + δ Mon+ δ Tue+ δ Thu+ δ Fri + φ FFF β Mon+ β Tue+ β Thu+ β Fri + δ Mon+ δ Tue+ δ Thu+ δ Fri + φ FFF Variable Large Cap REITs S&P 500 Coefficien -saisic Coefficien -saisic Panel A: Mean Equaion β β β 0.* 6.49 β 3 0.6* β * β * β β 7 0.* Panel B: Variance Equaion α 0 0.0* * 3.38 α 0.0* * 7.45 α 0.89* *.60 δ δ δ δ *.49 δ δ δ 6 0.7* *.5 φ 0.84* Using one day change in monh ahead federal funds fuure conrac as unanicipaed change. The saisics are robus using he procedure from Bollerslev & Wooldridge (99). * indicaes saisical significance a convenional levels

22 Table 3: Impac of US Moneary Policy Shocks on he Mean & Volailiy of REIT s (Toal Announcemens) REIT = β + β FFF + β REIT 0 0 h = α + α h s = + δ I ( TA) 0 + α e + δ I ( TA) + δ I ( TA) + + β S & P + β Mon+ β Tue+ β Thu+ β Fri+ β 00 + δ Mon+ δ Tue+ δ Thu+ δ Fri + δ 00+ φ FFF Variable Coefficien -saisic Panel A: Mean Equaion β β -0.7* -5.0 β 0.8* 9.89 β 3 0.4* 8.76 β β * 3.0 β β 7 0.* 3.35 β Panel B: Variance Equaion α 0 0.0* 4.78 α 0.3* 9.05 α 0.80* 4.70 δ δ δ δ δ δ δ 6 0.3*.36 δ φ 0.57*.0 Using one day change in monh ahead federal funds fuure conrac as unanicipaed change. The saisics are robus using he procedure from Bollerslev & Wooldridge (99). * indicaes saisical significance a convenional levels

23 Table 4: Tess for Asymmery in he Response of REITs o Moneary Policy Shocks REIT = β + β FFF h = α + α h 0 0 s = + δ I + α e + δ I + + β REIT + β S & P + β Mon + β Tue + β Thu + β Fri + δ Mon + δ Tue + δ Thu + δ Fri + φ FFF φ FFF Variable Coefficien -saisic Panel A: Mean Equaion β β -0.84* β 0.8* 9.80 β 3 0.4* 8.6 β β * 3.5 β β 7 0.* 3.37 Panel B: Variance Equaion α 0 0.0* 4.73 α 0.3* 9.06 α 0.8* 5.4 δ δ δ δ δ δ 6 0.*.9 φ φ Hypohesis Tes (p-values for Wald Saisic) φ =φ =0 0.3 Using one day change in monh ahead federal funds fuure conrac as unanicipaed change. The saisics are robus using he procedure from Bollerslev & Wooldridge (99). * indicaes saisical significance a convenional levels 3

24 Endnoes: Noe ha his paper solely examines he response of Equiy REITs and does no consider he Morgage REIT secor. See Allen e al. (000), Chen e al. (997), Chen & Tzang (988), Devaney (00), Liang & Webb (995), Ling & Naranjo (997), McCue & Kling (994), Mueller & Pauley (995) and Swanson e al. (00). 3 He e al. (003) highligh he imporance of proxies by illusraing he sensiiviy of resuls o he ineres rae proxy used. They also find furher evidence concerning he ime-varying naure of he linkages beween ineres raes and real esae securiies. Using a Flexible Leas Squares approach he paper highlighs ha all of he proxies esed have ime-varying characerisics. 4 Sevenson e al. (005) adop a similar mehodological approach in heir analysis of UK propery companies and do find significan sensiiviy in boh he mean and variance equaions using daily daa. 5 Conover e al. (999) also noe he imporance and influence of US moneary policy in an inernaional conex, while Lasrapes (998) provides furher inernaional empirical evidence on he influence of moneary policy on equiy markes. 6 An early paper o examine his is Waud (970). 7 Roley & Troll (984), Cook & Hahn (988) and Duecker (99) examine he issue of echnical and non-echnical rae changes in he conex of he impac of policy rae changes on marke ineres raes. 8 Furher papers o have examined issues concerned wih macroeconomic daa and sock movemens include Berry & Howe (994), Michell & Mulherin (994), Ederingon & Lee (993), Culer e al., (989) and Roll (988). 9 Casanias (979) provides an early sudy on he volailiy of he markes surrounding he release of economic daa. 0 We follow he approach used in Poole and Rasche (000) and use he monh ahead conrac (raher han he curren monh) o derive our surprise and so avoid making he adjusmen as in Kuner (00). Noe ha given he daa period examined ( ), i is no necessary o ake ino accoun he change in he operaions of he Federal Reserve in 994. One possible reason behind his is ha he impac of fed funds rae changes is more pronounced in smaller REITs and ha his is driving he overall findings. However, ess on a corresponding small cap REIT porfolio also find insignifican resuls. 3 Our resuls may be considered o be consisen wih he underlying hesis of he Bonfim (003) sudy, i.e. ha moneary policy ransparency has increased dramaically pos Given he evens of he firs nine monhs of 00, he unusually large changes in moneary policy and he evens of 9/, we also incorporae a dummy variable ino boh he mean and variance equaions. As can be seen from Table 3, he dummy variable is no saisically significan. 5 The deailed resuls for he large-cap REIT porfolio and he S&P 500 are available from he auhors. 6 Tess of any possible asymmery in he impac on reurns was esed, bu found no o exis. 7 When esing for possible asymmery in he volailiy response o shocks, we isolae posiive and negaive surprises on scheduled announcemen days. In order o avoid poenial mulicollineariy in our resuls, we omi I (sa) from he regression. 4

25 8 The deailed resuls for he large-cap REIT porfolio and he S&P 500 are available from he auhors. 5