Stock Market Reaction to Debt-Based Securities: Empirical Evidence. Mohammad Elian 1. Tai Young-Taft 2

Transcription

1 Stock Market Reacton to Debt-Based Securtes: Emprcal Evdence Mohammad Elan Ta Young-Taft Abstract Ths paper tests for the stock market reacton to corporate debt-based securtes offerng. A standard event study framework s appled to calculate and test Average Abnormal Returns (AARs) and Cumulatve Average Abnormal Returns (CAARs), both on and surroundng the announcement date. A noteworthy concluson s the absenteesm of sgnfcant postve abnormal returns for events surroundng the ssue date. Conversely, negatve abnormal returns are reported where the AAR s negatve on the announcement ssue date and remans negatve three days thereafter. As for the GCC case, such market behavor reflects the controversy about debt-based securtes (Sukuk) and ther full complance wth Islamc Law. In lne wth CAARs, the tests ndcate nsgnfcant results, ncludng bounded asymmetrc and symmetrc event wndows for perods before and after the announcement date, concludng nsgnfcant CAARs, hence, no wealth effect. Wthn the equty marketplace, these results wll be of practcal relevance to corporatons, practtoners and decson-makers. JEL Classfcaton: G, G4, G3 Key words: Market reacton, Debt-based securtes, Sukuk, Event studes Correspondng author. Department of Economcs and Fnance, Gulf Unversty for Scence and Technology (GUST), Kuwat, e-mal:elan.m@gust.edu.kw. Department of Economcs and Fnance, Gulf Unversty for Scence and Technology (GUST), Kuwat. e-mal: young-taft.j@gust.edu.kw We are ndebted to Professor Dlp k. Ghosh for hs gudance and valuable comments. Any errors that may reman are our responsblty. 46

2 - Introducton Fnancng decsons and factors affectng corporatons captal structure, together wth ther concomtant mpact on shareholder wealth, have been heavly debated fnancal ssues for years. In terms of captal structure formulaton, corporatons can buld ther own structure by ssung pure equty, debt, and/or convertble securtes that have a hybrd nature of equty and debt nstruments. However, t seems that the stock market reacts dfferently to such fnancng source decsons, causng a postve or negatve nfluence on wealth, as evdenced by the lterature. For nstance, one part of the lterature shows that the stock market does not react sgnfcantly to the ssue of new debt. In addton, the ssuance of new equty leads to a sgnfcant negatve stock prce reacton, whle the ssuance of convertble bonds leads to a sgnfcant stock prce reducton that s smaller than the reducton assocated wth the ssuance of pure equty. The pragmatc evdence advocates that pure equty fnancng has a relatvely large negatve effect, whle ssues of debt have a small nonnegatve effect on the value of the ssung corporaton (see Roon and Veld, 995; Abhyankar and Dunnng, 999). Nevertheless, the most recent fnancal crses, manly the Asan fnancal crss n 997 and the world fnancal crss n 8, hghlghted the connotaton between captal structure decsons and the massve market reacton to the falure of many corporatons worldwde. These crses suggested the need for rsk dversfcaton n terms of fnancng decsons made by corporatons as well as by governments. The ndcators showed that less credt dversfcaton and relance on drect borrowng from banks would create lqudty problems, thereby causng a negatve mpact on most economes. Hence, the low records of lqudty ratos, together wth other volatle market condtons, reflected relatvely hgh asset-lablty maturty msmatches for most corporatons, thus callng for them to swtch ther fnancng strateges to bond ssung. Ths was essental n order to replace borrowng from banks and/or to have another alternatve to equty fnancng, whch seemed costly. Gven ths foremost corporate concern, Sukuk (as debtbased securtes) were proposed as a means of sourcng funds va the captal markets. 3 3 As a concept, Sukuk, the plural form of the Arabc word Sakk, meanng certfcate, reflects ownershp or partcpaton rghts n underlyng assets. Actually, Sukuk means certfcates, whle techncally t refers to certfcates representng fnancal oblgatons arsng from trade or 47

3 In terms of growth, wthout a doubt, the last decade has seen a dramatc growth n Sukuk securtes, whch have typfed the fastest growng debt nstrument n the Islamc captal market. Notably, snce, captal markets have experenced substantal expanson towards sustanng most Sukuk structures, by offerng nvestors the chance to add portfolo dversfcaton advantages and allocate funds to nvestment alternatves n the form of new asset classes. 4 Issuers can take advantage of ths ncreased lqudty by tappng nto a growng demand from hgh net worth nvestors who are dggng for Islamc law (Shar a)-complant nvestment ventures and/or products. Furthermore, t s obvous that the growth of Sukuk s supported by ther many descrptons wthn the feld of debt-based fnancng. For example, asde from other features, Sukuk products are asset-backed securtes that grant value to nvestors and ssuers and hence trgger supply and demand actvtes from Muslm and non-muslm economes, wthout excepton. 5 Other renforcement factors nclude the growng sophstcaton of structures, the clarty of regulatory treatment, and the strategc focus necessary to develop comprehensve Islamc fnancal systems (see ISRA, ). To sum up, the growng trend of the Islamc debt market, and the emergence of new types of debt-based fnancng nstruments, provde the ground to undertake dynamc market tests to provde market nsghts for commercal busnesses. Accordng to Shar a standard 7 (), AAOFIFI (8) defnes Sukuk (Sukuk-stthmar) nvestment as certfcates of equal value representng undvded shares n ownershp of tangble assets, usufructs and servces, assets of partcular projects or specal nvestment actvty.. Gven ts adequacy standard (IFSB ), the Islamc Fnancal Servces Board (IFSB) defnes Sukuk as certfcates that represent the holders proportonate ownershp n undvded part of an underlyng asset where the holder assumes all rghts and oblgatons to such asset, (IFSB, 5). The Securtes Commsson Malaysa (SC) defnes Sukuk as a document or certfcate whch represents the value of an asset. 4 As far as Sukuk structure s concerned, they can be structured nto varous types based on many Islamc fnancng models. The AAOIFI has ssued gudelne standards for 4 dfferent types of Sukuk that can be tradable or non-tradable development and ndustral fnancng. As broad categores, there are three major clusters, namely () sale-based Sukuk, ncludng murabahah (cost-plus sale), bay bthaman ajl (deferred payment sale), bay al-nah, stsna, and salam; () lease-based Sukuk, ncludng jarah (leasng) and a mxture of jarah and stsna; and (3) equty-based Sukuk, ncludng musharakah (proft and loss-sharng), mudarabah (costplus-proft-sharng) and wakalah b stthmar (ISRA, ). From a fnancal perspectve, there are three common types of Sukuk, namely fxed ncome Sukuk, asset-backed Sukuk (ABS), and hybrd Sukuk. 5 See the followng secton for a comparson between Sukuk and conventonal bonds. 48

4 corporatons and governments. Ths paper tests for the stock exchange market reacton to corporate Sukuk offerngs, usng data for the GCC regon. More precsely, we test whether the announcement of corporate Sukuk ssues wll cause a sgnfcant abnormal return for ssung corporatons. Abnormal as well as cumulatve average abnormal returns are calculated and tested usng the standard event study framework, whch allows for assessng the mpact of a corporate event on stock returns. 6 As per the GCC context, we nclude frms ssung corporate Sukuk for the perod spannng 4-, where 5 lsted corporatons are ncluded. In addton, we argue that as the bulk of Sukuk s made up of corporate debt, testng for the stock exchange market reacton to such debt s extremely valuable. In order to motvate ths study, t s drven by the followng attrbutes. Frst, we examne market reacton to the ssuance of debt (leverage) securtes n vared nsttutonal and regulatory settngs wthn the GCC regon. Second, we seek to understand how the ssung of debt-based securtes nfluences the value of a corporaton. In terms of captal structure, ths would be reasonably relevant for managers who are n charge of takng fnancng decsons. Thrd, we ascertan the nature of the nformaton and understand the nformatonal content of bond (Sukuk) ssuance, whch wll provde the chance to explore whether the ssuance of Sukuk would delver new nformaton about future cash flows to securty holders, namely stockholders and bondholders, n the marketplace. Keepng ths n mnd, ths paper contrbutes to the lterature through the followng: Frst, based on a dataset of Sukuk securtes ssued between 98 and by GCC corporatons, we examne the market reacton to Sukuk ssues. Prevous studes have concentrated on conventonal bonds, but less or even no attenton has been drected towards Sukuk as an alternatve to conventonal bonds. Second, we examne the market reacton to corporate Sukuk offerngs ssued by a varety of corporatons n domestc GCC markets. Ths s qute relevant to managers and regulators n the feld of fund sourcng. Thrd, the paper explores dfferent hypothetcal model forecasts, by relatng ssue-perod abnormal returns and the cross-sectonal corporate characterstcs of Sukuk ssuers. 6 The event study methodology s used earler by Godlewsk et al. () and Modrzadehbam and Mansourfar (). The former analyzed the stock market reacton to Sukuk and conventonal bonds ssuance wthn the Malaysan context, whle the latter tested for the mpact of debt announcements on prvate frms stock returns. 49

5 The focus on the GCC regon s justfed by the followng. Frst, gven the tghtened lqudty condtons begnnng n 8, regulatory polcy ntatves reflected an ncreasng relance on external debt fnancng. Second are unstable and volatle lqudty condtons n the regon, stmulated manly by fluctuatons n ol prces, therefore reflectng volatlty n credt growth rates for banks and causng a slowdown n the economy, along wth worsenng bank asset qualty (see Al-Hassan et al., ). Thrd, statstcally, the GCC has % of the global Sukuk market, wth total ssuances worth US$6.6 bllon Saud Araba leads the GCC regon wth US$5.3 bllon, followed by UAE (US$65 mllon) and Bahran (US$ mllon). Furthermore, whle domestc GCC ssuances have been lmted, the regon has moved toward the nternatonal market, reflectng more confdence n the structure of the global Sukuk market. For example, the GCC nternatonal Sukuk ncreased from n H to US$75 mllon n H and US$65 mllon n H. Also, n terms of currency, for H, whle the Malaysan rnggt enjoyed the largest share of ssuance wth 7% of the market, GCC nternatonal dollar Sukuk ssuance came second wth a total of USA$7.7 bllon, reflectng that n the GCC most ssues are n US dollars, aganst whch these currences are pegged. The fndngs conclude the absenteesm of sgnfcant postve abnormal returns for events surroundng the Sukuk ssue date wndow [-3,3] (for {-3, 3} days after the ssue date, here and throughout); however, negatve abnormal returns are reported where the AAR s negatve on the announcement ssue date and remans negatve over [,3]. The fndngs show sgnfcant negatve abnormal returns on day [] before the announcement date [-] and nsgnfcant negatve AARs over [,3], reflectng negatve market reacton on the ssue date and three days thereafter. As per the CAARs, the p-values for both the z-test and t-test ndcate nsgnfcant results, ncludng asymmetrc and symmetrc event wndows for perods before and after the announcement date. Negatve CAARs surroundng the ssue dates [-,], [-,], and [-3,3], and the CAARs for event wndows [-,] and [,], ndcate that such a type of debt does not lead to sgnfcant cumulatve abnormal returns, therefore concludng that no wealth effect s found durng the event wndow surroundng the announcement date for Sukuk ssuance. The rest of the paper s organzed as follows. Secton outlnes dfferences between Sukuk and conventonal bonds, whle secton 3 revews the related lterature. Secton 4 outlnes the standard event study methodology 5

6 used. Fndngs and dscussons are gven n secton 5, whle summary and concludng remarks are provded n secton 6. - Comparson between Sukuk and conventonal bonds One queston from an emprcal-theologcal framework pertans to the dfference between Sukuk and conventonal bond ssuance. In fnancal termnology, both are equvalent; however, Sukuk are generally structured as debt nstruments approved by most Shar a scholars based on complance wth Shar a prncples (the Islamc legal code). Nevertheless, whle Sukuk allow governments (va ssung soveregn Sukuk) and corporatons (va ssung corporate Sukuk) to rase funds n captal markets, there s a current debate as to whether these debt nstruments dffer sgnfcantly from conventonal bonds. In possble response to ths noton, Mller et al. (7) and Wlson (8) focused on Sukuk returns and contended that they are desgned to replcate features of standard bonds. Other common features nclude late payment penaltes upon default, purchase undertakng n equtybased structures, and ownershp status n asset-based transactons (Cakr and Rae, 7; Lahsasna and Ln, ). Addtonal smlartes may nclude the followng: () smlar to conventonal bonds, Sukuk provde a rsk reducton beneft n terms of nvestment dversfcaton, partcularly when nvestng n fxed ncome portfolos; () Sukuk are consdered to serve as securty nstruments that generate predctable fxed or floatng levels of return; (3) Sukuk are traded n the secondary market, albet less than conventonal bonds; and (4) Sukuk are rated by nternatonal ratng agences (see Cakr and Rae, 7; Lahsasna and Ln, ). Conversely, unlke Sukuk, conventonal bonds represent ssues of pure debt, whle Sukuk typfy an ownershp stake n the underlyng asset or securty. For example, many Sukuk contracts generate a lessee/lessor agreement rather than a lender/borrower relatonshp (Mrakhor, 7). In addton, Sukuk securtes must be backed by a real asset or project, whereas conventonal bonds are based solely on cash flows, ndcatng that conventonal bonds seem to be rsker than Sukuk to many nvestors n the marketplace. In terms of structure and thus valuaton Sukuk do not engender the exchange of paper or money; however, the structure s smlar to asset securtzaton (see Ayob, 8). As such, the asset becomes the focal pont of tests and analyss, unlke conventonal bonds where the ssue remans the focus. Unlke Sukuk, conventonal bonds usually represent non-asset- 5

7 backed nterest-based fundng for general corporate purposes. Accordngly, as tradng n all-purpose ndebtedness s prohbted by Shar a prncples, the ssung and tradng of conventonal bonds hence do not seem complant wth Shar a prncples ether (Howladar, 9). In general, as a dstnctve feature of Sukuk, asset-backed Sukuk perform dfferently to conventonal asset-based securtes, where Sukuk should grant the nvestor a share of an asset or busness/project venture along wth the cash flows and rsk commensurate wth ths ownershp (Howladar, 9). 7 In other words, as tradng n debt securtes s not permtted under Shar a prncples, Sukuk must be asset-backed, whereby what s beng bought and sold n the case of tradng Sukuk s a rght to the underlyng asset or securty. Investors n Sukuk cannot receve nterest but rather permssble returns such as rents n the case of jara (lease) Sukuk or proft shares n the case of Sukuk mudarabah (proft sharng) and/or Sukuk musharakah (proftloss sharng). Furthermore, Dusuk and Mokhtar () dentfed that Sukuk are frequently referred to as Islamc bonds, nstgatng the msunderstandng that they are just lke bonds; however, the two debt nstruments are dametrcally opposed n nature. The authors reported that Sukuk should be n the form of an nvestment certfcate representng ownershp of an asset or venture, whle conventonal bonds are generally ssued to evdence debt. They added that, unlke Sukuk, conventonal bonds do not typfy ownershp but nstead document the nterest-bearng debt owed to the bondholder by the bond ssuer who owned the commercal or ndustral enterprse. Sukuk are unlke conventonal bonds, whch are structured as debt nstruments wth fxed nterest and cannot be structured to evdence a loan. Coupon nterest payments n the case of conventonal bonds are determned as a fxed percentage of the bond par value, not as a percentage of actual profts. Unlke Sukuk, conventonal bonds guarantee the bond prncpal (bond par value) when redeemed at maturty. Under Sukuk, holdng nvestors receve a passthrough of ncome, whch represents ther proportonate benefcal ownershp n the underlyng asset. The return may be fxed or floatng, but t cannot be by any means n the form of nterest for lendng pure money (the usage of money as a commodty) based on Shar a prohbton of nterest. 7 However, from a rsk-return perspectve, t seems that most current Sukuk structures (unsecured asset-based structures) have much n common wth conventonal fxed ncome or debt nstruments, and unsecure fxed ncome securtes n partcular, Howladar (9). 5

8 To conclude, Sukuk presents an alternatve debt nstrument to conventonal bonds and uses assets and many contracts n order to comply wth Shar a and provde an alternatve fnancng nstrument. Furthermore, Sukuk holders are enttled to share revenues generated by the Sukuk assets and may be enttled to a share of the proceeds from realzng the Sukuk assets. The sharng proporton of the proceeds s not fxed upfront; rather, t s based on the actual proft realzed from the project or the venture (Usman, 7; Mokhtar and Abdulkader, 9; Dusuk and Mokhtar, ). It seems that Sukuk are an alternatve to asset monetzaton, syndcate project fnancng, fnancng through asset-backed securtzaton, and publc fnancng, and therefore, gven the many forms of Sukuk used n practce, they are nstruments used n Islamc fnance to tap captal markets. 3 - Lterature revew Emprcal evdence on stock market reactons to Sukuk ssuance are lmted and provde ambguous results. Cakr and Rae (7) examne the rsk-reducton advantages of ssung soveregn Sukuk by consderng them and two Eurobond soveregn ssues va the same source and by computng VaR for Eurobond ssue-only portfolos and Eurobond ssues along wth the Sukuk ssues, fndng reduced VaR for the second case. By usng data for frms lsted n Bursa Malaysa, Ashhar et al. (9) nvestgate the mpact of Sukuk and conventonal bond announcements on shareholder wealth for the perod to 6. They use 6 days (3 days before the announcement day and 3 days thereafter) to undertake an event wndow analyss n ther emergng market. Ther results ndcate that, on average, nvestors n the Malaysan market do ndeed react to Sukuk announcements, thus sgnalng a wealth effect through the announcement of Sukuk ssues; however, nvestors do not react to conventonal bond announcements. The bond offerng sze appears to have a mldly negatve nfluence on the cumulatve average abnormal return. Ibrahm and Mna (9) examne the wealth effects of Sukuk offerngs and ther determnants. For the perod -6 n Malaysa, they fnd that the market reacton s sgnfcantly postve durng event wndows [- 3,] and [-3,3] surroundng the announcements of Sukuk ssuance. In terms of determnants, they conclude that the wealth effect of Sukuk ssuance announcements s postvely affected by the ssuers nvestment opportunty 53

9 and negatvely assocated wth the ssue sze, frm sze, and whether the announcement s accompaned by Securty Commsson approval. Ameer and Othman () fnd sgnfcant negatve abnormal returns around the announcement days (the responses are asymmetrcal) to dfferent types of bond ssuance announcements n Malaysa for the perod -7. Practcally, the negatve reacton observed s not consstent wth the optmal captal structure hypothess, whch states that the ssung frm trades off nterest tax sheld and bankruptcy costs to maxmze frms value va stock apprecaton. The theory predcts a non-negatve change n the frms value regardless of leverage change sgn (Abhyankar and Dunnng, 999). Godlewsk et al. () test 7 ssuances (77 of whch are Sukuk and 93 conventonal bonds) for dfferences between Sukuk and standard bond announcements among Malaysan lsted frms that ssued both wthn the to 9 wndow and wth an OLS market model. They use 9 days, from days to days, pror to the ssuance, to estmate the market model, and they report that extendng the perod to 5 days before the announcement or stoppng t up to 3 days pror to the announcement does not affect the sgnfcance or drecton of the effect. Addtonally, they report to have vared the procedure across four stock ndces, resultng n concordant effects. They consder event wndows comprsng () the day of announcement, () the day of the announcement, one day pror and one day after, and (3) the day of the announcement, two days pror and two days after. They fnd that whle there s no reacton to the announcement of regular bonds or Sukuk ssuances over the one- or three-day event wndows, there s a negatve reacton to the announcement of Sukuk bonds over the fve-day event wndow, whch they suggest may be due to an adverse selecton mechansm whereby Sukuk are ssued by fnancally less competent frms. Addtonally, not ncludng a number of days after the announcement may also bas nterpretaton of the effects. Fnally, not reportng sets of event wndows other than the three they select may gve nterpreter-lmted nformaton from whch to form an apprasal of the effect. They pont out the reasons why they do not consder GCC-lsted Sukuk are that () the majorty of ssues are soveregn and () there s no actve secondary market for them. We do not see a majortaran ssuance of soveregn bonds among the Sukuk class as a reason not to study corporate bonds, however, and nether do we consder the lack of an actve secondary market as a reason not to do the same. We agree that the thnness of markets may nterestngly affect our 54

10 concerns, but ths presumably can be tested wth data and a consderate modelng approach. Modrzadehbam and Mansourfar () examne the mpact of Sukuk announcements on prvate frms stock returns, n order to determne the nformaton content of Islamc prvate debt offerngs. Ther sample conssts of 45 lsted frms on Bursa Malaysa nvolved n ssung Islamc debt for the perod 5 to 8. After calculatng abnormal and cumulatve average abnormal returns, ther results show that sgnfcant (negatve) abnormal returns occur one day before announcement day, representng market nvestors adverse behavor towards an Islamc prvate debt announcement. As for the cumulatve average abnormal returns (CAARs), ther results reveal nsgnfcant negatve CAARs, rejectng ther hypothess regardng postve market reactons to an Islamc bond announcement. Ahmad and Rahm (3) nvestgate whether markets react asymmetrcally to the ssuance of selected Sukuk structures, namely jarah (leasng) and musharakah (proft-loss sharng), n Malaysa for the perod 8-. They use cumulatve average abnormal returns on symmetrc and asymmetrc events based on the reacton of the FTSE Kuala Lumpur Composton Index to the announcement of Sukuk ssuance. They conclude wth sgnfcant, postve symmetrc and asymmetrc market reactons to selected Sukuk structure ssuances. 4 - Methodology specfcaton 4. Aggregate cumulatve abnormal returns We follow sources such as Gasbarro et al. (4), Lummer and McConnell (989), and Preece and Mullneaux (7), and n partcular we adopt a great deal of the notaton and exposton of MacKnlay (997). Frst, we compute the stock s pre-event return, or normal return. Then, we compute the abnormal return: The actual ex-post return over the event wndow mnus the normal return over the event wndow, wth the normal return beng the expected return wthout condtonng on the event takng place. Notatonally, the abnormal return can be expressed as: AR = R E( R X ), () t t t t 55

11 wth ndexng the frm, t the event date (n place of the tme ndex t ), R t the actual return, X t condtonng nformaton for the normal return model, and E ( R t X t ) normal returns. In practce the event date s expanded to an event wndow over whch perod the event effect s consdered (that s presumng X s gven pror to the event date but not t updated over the event wndow), and X t s estmated usng a perod (mmedately or shortly) pror to the event date. The normal return s typcally modeled wth the constant mean return model, wth X a constant, or the market model, where t X s the market return. Here, the market return model assumes a lnear relatonshp between the market return and the frm s stock return. It s possble to use a method that s consstent, gven the autocorrelaton and heteroskedastcty of abnormal returns, for example by usng a GMM estmaton, though authors such as Brown and Warner (98,985) fnd the constant mean return model can be comparable to such methods. One can measure real returns or excess returns (the return mnus the nomnal rsk-free return of a T-bll wth one month left to maturty) as well as nomnal returns. We consder nomnal returns n ths paper, argung that our tme horzon s suffcently short to do so wthout sgnfcant bas. By removng the porton of the return related to varaton n market return, abnormal return varance s reduced, and so the market model may mprove upon the constant mean return model. Its beneft depends on the R of the market model regresson, wth a hgher R suggestng greater varance reducton. One can addtonally nclude other factors such as ndustry ndexes n the model, as dscussed n Sharpe (97) and Sharpe et al. (999). 8 Addtonally, one can select frms of smlar sze relatve to the market value of ther equty as a samplng method to reduce possble nose. We consder the market model. In partcular, we estmate: R R, () t = mt t t 8 For more nformaton on portfolo selecton, captal markets and fnancal assets, see Markowtz (959) and Ghosh (). 56

12 for R mt perod- t return on the relevant market ndex, and t the error term, wth E = ) and var ( ) =, usng OLS, wth a wndow of days ( t t before the event, that s for t = to t =, wth t =. Wth our estmates for and, ˆ and compute our estmate for the abnormal returns: AR ˆ ˆ t R R, = t mt ˆ, respectvely, we then for the 3 days after and ncludng the event (the event wndow),.e. for t t = to t = 9. Ths gves us N = returns whch can be used to = estmate t X and N = 3 abnormal returns to consder changes n nformaton as a result of the event. Gven a null hypothess of no effect, abnormal returns wll be jontly normally dstrbuted wth zero mean and: ( R ˆ mt m) ( ARt ) = [ ], (4) N ˆ m for ˆ = R. The second addtve term, n addton to, s due m t= N mt to a samplng error n and across event wndow observatons, whch leads to the seral correlaton of abnormal returns. We remnd the reader of our dsparate sample szes noted above and that ths ssue exacerbates the ( R ˆ ) problem. However, as N, [ mt m ], as wll the seral N ˆ m correlaton of abnormal returns. We consder our estmaton wndow to be of suffcent length to neglect ths term, whch mples: ARt : N(, ( ARt )). (5) We then aggregate our returns by securty through tme. To acheve ths we compute the cumulatve abnormal returns: (3) 57

13 CAR( t ) = t AR, (6) t t = t = for each t {,,..., 9}, whch gves us cumulatve abnormal returns up to and ncludng t. As N, var( CAR ( t )) = ( t ) ( t t ). (7) Ths asymptote s often used as an estmate of true varance, and we use t n ths case as well. The null hypothess of no effect across the frm s cumulatve abnormal returns by tme perod become:s CAR ( t) : N(, ( t)). (8) We then aggregate our estmates for the ndvdual frms cumulatve abnormal returns seres across frms, to derve a sample-wde tme seres of sample cumulatve abnormal returns. For each tme perod we have: N CAR( t ) = CAR ( t ), (9) N = for N = 5 the number of frms. In order for ths to be a consstent estmator, we need to have no clusterng, n that none of the event wndows overlaps wth another, and there s no correlaton across exchanges or frms. In our case, the Kuwat frm (whch has only 55 days over whch to compute the normal return) overlaps n ts entrety wth a Bahran frm (the one wth only 6 days after and ncludng the announcement wth whch to consder changes n 58 X t ), wth ts event date beng November st 7, and that of the Bahran frm beng November 5 th 7. However, our assumpton that the dfferent exchanges provde ndependent samples also apples to ndependence relatve to tme perods. If ths hypothess does not hold, t wll ntroduce a clusterng bas that would addtonally translate nto a jont seral correlaton bas relatve to the varyng tme perod of estmate bas, though ths would go to zero, as N. We

14 fnally remnd the reader of our dscusson regardng clusterng relatve to country and frm class n ths regard. Gven the above dscusson: N ( t = var CAR( t )) = N ( ), () wth correspondent null hypothess of no effect of announcement CAR t ) : N(, var( CAR( ))). () ( t model: Usng the consstent sample varance estmate of from the market ˆ = N t= ( R t ˆ ˆ R mt ), () In place of the known varance, we can consder our fnal null hypothess by usng the asymptote as N, N }. { 4. Aggregate abnormal returns We are addtonally nterested n aggregate abnormal returns across securtes by tme perod: N ARt = ARt, (3) N wth: = lm N var( ARt ) N N =. (4) 59

15 5 - Data collecton Data were extracted from the Islamc Fnance Informaton Servce (IFIS) database, and we ncluded all frm-ssued corporate Sukuk for the perod 4-. The net sample sze s controlled by nformaton avalablty for varables, namely closng stock prces for corporatons ssung debt-based Sukuk and synchronzed market ndex closng prces, for a varety of GCC stock exchanges. The flterng process accounts for, out of a gross 39 lsted corporate ssues, a net fgure of 5 corporate Sukuk ssues ncluded n the test. Ths cohort s large enough to apply the market model and to test for the null hypothess of no abnormal daly returns under the event study framework, where non-standardzed and standardzed average abnormal daly returns (AARs) and cumulatve average abnormal daly returns (CAARs) are calculated. In ths paper, we consder corporate Sukuk ssuance announcements and ther effect on stock returns for four GCC exchanges: Saud Araba, Kuwat, the Unted Arab Emrates, and Bahran. In our sample of 5 frms we nclude from Saud Araba (three of whch are subsdares of the same company, SABIC), three from Bahran, one from the Unted Arab Emrates (UAE), and one from Kuwat. Here, we assume all exchanges GCC exchanges are equvalent across countres, whle subsdares of SABIC functon autonomously, so there s no clusterng across countres or frms. Addtonally, three of our frms do not have complete preannouncement data, and two dfferent frms do not have complete postannouncement data. In partcular, whereas the other frms have stock prce data for each of the days pror to the date of the announcement, one of the SABIC subsdares only has 9, one of the frms from Bahran has 35, and another frm from Saud Araba s mssng sx days. The two frms wth less than the standard 3 days of returns (ncludng announcement date) are the other Bahran frms, wth 6 and sx days after the returns each. Ths may pose sgnfcant volatons of the hypotheses mentoned earler, n addton to autocorrelaton across frms and countres, though all of the concerns can be annulled f we consder our sample to represent suffcently the asymptotc results of the standard lnear theory. 6

16 6 - Fndngs and dscusson: Event study analyss 6. Aggregate abnormal returns Table presents the fndngs of the event study as per average abnormal returns (AARs). Gven space lmtatons, we restrcted our analyss to the event perod comprsng a total of 3 days, estmated from [-5] to day [5] after the Sukuk ssue announcement. 9 The results show that, out of 3 days, the AARs were negatve for about out of the 5 days before the Sukuk announcement, whle they were negatve for about four out of the 5 days after announcng the ssue date. In addton, the results show that the AARs were postve for about fve out of the 5 days before the Sukuk announcement, whle they were postve for about out of the 5 days after announcng the ssue date. The AARs were negatve on the Sukuk ssue date and remaned negatve three days after [,3], followed by postve days thereafter, except for day. The market reacton s revealed to be sgnfcantly postve for day [- ] and day [4], wth average announcement perod abnormal returns of.% and.6%, respectvely. No sgnfcant postve abnormal returns are concluded for other perods; however, postve average abnormal returns are reported for day [-3] and day [-]. In addton, postve average abnormal returns are revealed from [4] to [5] (except []), thereby ndcatng that the stock market reacts postvely to late Sukuk ssuance announced by frms, partcularly to those whose shares are traded more frequently. On the contrary, the Sukuk announcement does not result n a postve mpact on stock returns durng the event perod surroundng the ssue date [-,]. The negatve abnormal return on [-] s sgnfcant at the % level, reflectng negatve market reacton the day before the ssue date, whle the announcement day ([]) and one day thereafter ([]) are nsgnfcant. Market reacton s revealed to be sgnfcantly negatve for the [-6] and [-] perods, wth average announcement perod abnormal returns of.9% and.7%, respectvely. As per our case, regardless of market reacton behavor (postve or negatve), a possble justfcaton for the early market response s attrbuted to market neffcency, where offerng nformaton s leaked before the announcement. 9 Accordng to Panaydes and Gong (), an -day event wndow can fully capture the effects of an event of nterest. 6

17 Furthermore, a noteworthy concluson at ths stage s the absenteesm of any sgnfcant market reacton to Sukuk ssuance wthn the GCC regon, partcularly for events surroundng the Sukuk ssue date [-,], therefore suggestng that ssung addtonal debt may be nterpreted as a negatve sgnal n many condtons. In other words, t seems that debt announcements may send negatve credblty sgnals regardng the qualty of the ssung frm, thus causng adverse selecton problems (asymmetrc nformaton) between market partcpants and leadng to hgher agency costs and hence adverse stock market reactons. Overall, our conclusons algn wth earler studes whch provde evdence that stock markets do not react, or even react negatvely, to debt announcements. For example, Akhgbe et al. (997) show that stock prces react negatvely to new debt ssuance announcements, when ssung debt s encouraged by an unexpected cash shortfall. Also, as suggested by Godlewsk et al. (), negatve market reactons to debt announcements may be attrbuted to the belef that more debt would cause greater moral hazard that may exst under two scenaros: () more debt would cause hgher levels of credt as well as bankruptcy rsks and () more debt upsurges agency costs that exst due to conflcts of nterest between debt holders and shareholders. For the GCC case, we frst argue that the controversy regardng Sukuk and ts complance wth Shar a typfes the crucal reason behnd the lmted nvestor base for Islamc Sukuk, justfyng, n turn, the lmted market response to the Sukuk ssue. Second, smlartes between Sukuk and conventonal bonds reduce nvestor belef n Sukuk as a pure Islamc fnance securty. As a pont of llustraton, Usman (9), as cted by Shakh and Saeed (), stated that 85% of the Sukuk ssued worldwde dd not comply wth Shar a prncples, n addton to other smlartes such as late payment See Eckbo (986) and Mkkelson and Partch (986) for conventonal bond ssues and Ameer and Othman (), Godlewsk et al. (), and Modrzadehbam and Mansourfar () for Sukuk ssues. They justfed negatve market reacton to the Sukuk announcement by usng the adverse selecton mechansm argument, whereby borrowers wth low credt worthness, low return antcpaton, less proftablty, and hgh leverage have the motve to offer proft and losssharng fnancng securtes (Sukuk) schemes, n order to control for rsk n case of busness falure. These less healthy borrowers, and those wth lower expected earnngs (lower fnancal results), wll prefer to ssue Sukuk, n order to reduce ther share n case of loss, rather than fxed ncome securtes schemes, whch normally shoulder a hgher fnancal burden. Gven such a scenaro, ssung Sukuk s gong to be consdered a negatve sgnal to market partcpants, causng n turn lower stock prces that adversely affect a frm s value. 6

18 penaltes upon default. In addton, Herwany and Febran () dentfed that due to the absence of relevant Sukuk prcng models, the same prcng benchmarkng s used for both of Sukuk and conventonal bonds,.e. Lbor. In fact, most of these factors caused less deepenng of fath n Sukuk wthn the context of GCC countres. However, such atttudes may change future-wse, gven the good ndcators of Sukuk performance n global markets. For nstance, accordng to S&P, t s expected that the GCC regon wll ssue more Sukuk as a result of beneftng from hgh ol prces and ncreasng hydrocarbon projects that need massve fundng. Table. AARs of Sukuk ssuance over the perod 4- Days AAR AAR AAR-z AAR-t Stat. (p-value) (p-value) [-5] [-4] [-3] [-]..3.**.3** [-] [-] [-9] [-8] [-7] [-6] **.57** [-5] [-4] [-3] [-] [-] ***.*** [] [+] [+] [+3] [+4] [+5] [+6] [+7] [+8] [+9] [+] [+]

19 Table. AARs of Sukuk ssuance over the perod 4- (cont.) Days AAR AAR AAR-z AAR-t Stat. (p-value) (p-value) [+] [+3] [+4].6.83.*.8* [+5] Fgure shows AARs durng the 5-day event wndow. They show a sharp drop over days -9 to -4, reflectng negatve AAR values, followed by an upward trend for days -3 and -, showng slghtly postve AAR values. The postve trend s reversed agan to negatve one day before the Sukuk ssue date [-], assgnng the possblty of market anomaly. The AARs experenced negatve value on the ssue date, followed by three days thereafter, namely days,, and 3. The negatve trend adjusts gradually durng days 4 to 5, ndcatng fluctuatng postve AARs, excludng day. The AARs conclude at a postve peak two weeks after ssuance, on day [4]. Fgure. AARs of Sukuk durng the 5-days event wndow 6. Cumulatve aggregate abnormal returns Table presents the cumulatve average abnormal returns (CAARs) and reports postve CAAR results for events over 3 days, estmated from 5 days before to 5 days after the Sukuk ssue announcement. Lke the AAR, CAAR s consdered a helpful statstcal analyss that asssts n ganng 64

20 awareness of the aggregate mpact of the AAR, partcularly f the effect of the event durng the event wndow s not restrcted purely to the event date. Out of 3 days, event wndows show postve CAARs, namely nne event wndows before the event day ([-5] to [-7]) and three thereafter, namely [], [4], and [5]. Conversely, 8 event wndows show negatve CAARs. The CAARs are negatve for the sx days before the Sukuk announcement ([-6] to [-]) and are negatve for days after announcng the ssue date ([] to [3]), except for day []. As for the ssue date, t shows negatve CAARs. To provde more nsght, we extend the analyss perod to a total of 6 days, estmated from [-3] to day [3], where postve CAARs are reported for the perod ([-3,6] to [-3,9]). Overall, the CAAR p-values for both the z-test and t-test ndcate nsgnfcant results, ncludng asymmetrc and symmetrc event wndows for perods before and after the announcement date. The nsgnfcant and negatve CAARs surroundng the ssue date [-,], [-,] and [-3,3] ndcate that there s no wealth effect for the shareholders of frms-ssung Islamc bonds (Sukuk). Meanwhle, CAARs for event wndows [-,] and [,] ndcate that such a type of debt does not lead to sgnfcant cumulatve abnormal returns, concludng no wealth effect s found durng the Sukuk event wndow. However, when the event wndow s extended to day [,4], postve and nsgnfcant CAARs are reported and reman untl day [,9]. Also, when the event wndow s extended to [,9], postve and sgnfcant CAARs are concluded at the 5% and % levels. The wealth effect s small, rangng from.5% for wndow [,4] and.% for wndow [,9], wth the maxmum wealth effects reported to wndows [,9] and [,], wth 3.59% and 3.37%, respectvely. Both are sgnfcant at the 5% level. Table. CAARs of Sukuk ssuance over the perod 4- Days CAAR CAAR CAAR-z CAAR-t Stat. (p-value) (p-value) [-5] [-4] [-3] [-] [-] [-] [-9]

21 Table. CAARs of Sukuk ssuance over the perod 4- (cont.) Days CAAR CAAR CAAR-z CAAR-t Stat. (p-value) (p-value) [-8] [-7] [-6] [-5] [-4] [-3] [-] [-] [] [+] [+] [+3] [+4] [+5] [+6] [+7] [+8] [+9] [+] [+] [+] [+3] [+4] [+5] The cumulatve abnormal returns trend can be seen clearly n Fgure. The CAARs show postve/negatve values, thus suggestng a postve/negatve nterpretaton of Sukuk announcements by market partcpants. Except for day, whch shows postve CAARs, downward CAAR value trend are reported for day -6 to day 3, after whch CAAR values are adjusted to have postve values for days 4 and 5. It appears that nvestors need tme two weeks after the ssue date n our case to dgest the nformaton content of a Sukuk announcement. The postve but nsgnfcant CAAR values llustrate that there s no announcement effect, and hence no wealth effect, lnked to Islamc Sukuk offerngs. 66

22 Fgure. CAARs of Sukuk durng the 5-day event wndow 7 - Summary and concludng remarks An extensve body of lterature has examned ssues related to the concept and structure of Sukuk, the extent of ther complance wth Islamc (Shar a) law, and the dfferences between Sukuk and conventonal bonds, whle lmted contrbutons have been made n the lterature regardng the corporate fnance perspectve of Sukuk ssues. Ths paper brdges ths gap by testng for the stock exchange market reacton to corporate Sukuk offerngs, usng data for the GCC regon over the perod 4-. To capture fully the effects of our event of nterest, we used an event perod of 3days to consder event wndows estmated from -5 days pror to and 5 post a Sukuk ssue announcement. As for aggregate abnormal returns, t s concluded that the market reacts postvely as well as negatvely to a Sukuk offerng. The market reacton s revealed to be sgnfcantly postve at the 5% level for the [-] and the [4] ssuance perods, mplyng a lmted postve nfluence on shareholder wealth. Correspondngly, whle no sgnfcant postve abnormal returns are concluded for other perods, postve and nsgnfcant average abnormal returns are reported for the [-3] and [-] perods. The AAR s negatve on the Sukuk ssue dates and remans negatve three days thereafter ([], [], and [3]), followed by postve days, except for day, thereby ndcatng that the stock market reacts postvely and late to Sukuk ssuances announced by frms. The announcement of Sukuk results n a negatve mpact on stock returns durng the event perod surroundng the ssue 67

23 date {[-],[],[]}. Furthermore, at the % level, the market reacton s revealed to be sgnfcantly negatve for the [-6] and [-] perods, mplyng a negatve nfluence on shareholder wealth. Overall, the postve or negatve market reactons pror and post the ssue date ndcate that ether nvestors antcpate the announcement or there s market neffcency whereby offerng nformaton s leaked out to the market before the announcement date. However, gven the features of the GCC stock markets and ther mcrostructure, the possblty of nformaton leakage mght be hgher than nvestors antcpaton of Sukuk ssuance. In addton, gven the lmted sgnfcant values revealed, t seems that GCC stock markets do not consder ssung Sukuk as good news whch may postvely affect the stock prces of the ssung frms, thus reflectng low confdence n Sukuk as a wealth maxmzaton nstrument and low nvestor apprecaton for the nformaton content of Sukuk ssuance. In terms of ther synchronzed nature and tmng, Sukuk ssuances do not seem to be crucal n transferrng nformaton regardng frms future earnngs, and hence ther future performance and captal sourcng, thus suggestng low, f not negatve, mpacts on frms stock prces. Ths provdes lttle support for the sgnalng hypothess n terms of Sukuk ssuance. Our concluson regardng the aggregate abnormal returns of Sukuk ssuance may mply that they are not partcularly relevant for future stock prce predctons, demonstratng that stock prce reactons may not fully reflect the mpact of Sukuk announcements, partcularly for emergng markets. Gven ths possblty, future research, to test for postve abnormal returns for frms wth more actvely traded stocks relatve to less actvely traded stock, would be helpful. For cumulatve abnormal returns aggregated across securtes, we test the hypothess as to whether or not they are dfferent from a zero mean normal dstrbuton wth equal varance (asymptotcally) to the estmated dstrbuton. As pror and post event perods, out of 3, event wndows show postve CAARs whle 9 show negatve CAARs, ncludng the ssue date. The CAAR p-values for both the z-test and t-test ndcate nsgnfcant results, ncludng asymmetrc and symmetrc event wndows for perods before and after the announcement date. The nsgnfcant and negatve CAARs surroundng the ssue date [-,], [-,] and [-3,3] ndcate that such a type of debt does not lead to sgnfcant cumulatve abnormal returns, concludng that there s no wealth effect assocated wth shareholders of frms ssung Islamc bonds 68

24 (Sukuk). When the event wndow s extended to nclude days [,4] to [,9], postve and sgnfcant CAARs are found at the 5% and % levels. As per the average cumulatve abnormal returns, t s obvous that none of the aggregate cumulatve abnormal returns s sgnfcant. However, when we nstead test for an ncrease n estmator pont mass,.e. perform a- one taled test on our dstrbuton, all returns after and ncludng 5 days after the announcement are sgnfcant at the % level. One possble reason for ths may be that the frms are credt-starved, a common feature of developng countres, and when they do acqure the requred captal they are able to put t to use n a manner to whch the market responds. It s nterestng that t takes about two weeks to regster the effect, and that the effect s consstent for the next two weeks. Agan, varabltes n the market reacton to Sukuk ssues, and the concluded zero wealth effect of Sukuk ssuance, are justfed by low fnancal market effcency and legal dsclosure requrements wthn the framework of emergng and developng economes, where there s a chance of leakage of nformaton when new Sukuk offerngs are on the cards. Ths suggests that nvestors receve nformaton about Sukuk ssues on the grapevne, before the news becomes publc knowledge. As such, t s possble that abnormal returns are realzed pror to the announcement date. For our case, ths perod s expected to be one to two weeks pror to the announcement date. A second potental justfcaton for the low wealth effect of Sukuk ssuance announcements s that there are smaller nvestor bases for Sukuk wthn the GCC regon, sgnalng lower cost advantages and thus causng hgher captal cost. Ths establshes that Sukuk n the GCC regon are stll controversal n regard to legal acceptablty or complance wth Shar a, due to the belef that Sukuk features n general are not sgnfcantly dfferent from conventonal bonds. In other words, Sukuk offerngs do not contrbute postvely towards enhancng the Shar a-complance status of the ssung frms, thereby causng a lower nvestor base for Sukuk, leadng to lower CAARs, and hence, no sgnfcant wealth effect. A thrd possble reason may be attrbuted to frms tax consderatons whle sourcng by debt va Sukuk, whch mght not gan much on tax subsdes or tax deductblty, snce Sukuk do not have coupon characterstcs. Thus, Sukuk ssues are not consdered examples of low-cost captal nvestment, causng the hgher possblty of default rsk n case of more debt ssues typfed by Sukuk. In ths regard, the massve ssues surroundng Sukuk le outsde the GCC regon, n order to beneft from tax consderatons and hence 69

25 tax subsdes; the massve amounts of Sukuk ssues are offcally regstered for Malaysa and Indonesa rather than GCC. References AAOIFI-Accountng and audtng organzaton of Islamc fnancal nsttutons, 8. Shar a standards, Manama, Bahran, avalable at Abhyankar A. and A. Dunnng, 999. Wealth effects of convertble bond and convertble preference share ssues: an emprcal analyss of the UK market. Journal of Bankng and Fnance 3, Ahmad, N. and S. Rahm, 3. Post-Crss Stock Market Reactons Followng Sukuk Issuance n Malaysa, Proceedng of the nd Internatonal Conference on Arts, Socal Scences & Technology (ICAST ), organzed by the Research and Industral Lnkages Department of Unversty Teknolog MARA (UITM) Kedah, Park Royal Hotel, Penang, Malaysa, 3-5 March : I5--I5-. Akhgbe, A., J. Easterwood, and R.Pettt, 997. Wealth effects of corporate debt ssues: The mpact of ssuer motvatons. Fnancal Management, Al-Hassan, A., M. Khams, and N. Ould,. The GCC Bankng Sector: Topography and Analyss. IMF Workng Paper, /87, -37. Ameer, R. and R. Othman,. Stock Market Reacton to Bonds Issuance: Evdence from Malaysan Bankng Sector. Internatonal Research Journal of Fnance and Economcs, ISSN Issue 45. Ashhar, Z., L. Chun, and A. Nassr, 9. Conventonal vs Islamc Bond Announcements: The Effects on Shareholders Wealth. Internatonal Journal of Busness and Management, Vol. 4 (6). Ayub, M. 8. Securtzaton, Sukuk and Fund Management Potental to be Realzed by Islamc Fnancal Insttutons. NIBAF, State Bank of Pakstan, Karach, -6. Brown, S. and J. Warner, 98. Measurng Securty Prce Performance, Journal of Fnancal Economcs, 8(3), Brown, S. and J. Warner, 985. Usng Daly Stock Returns: The Case of Event Studes, Journal of Fnancal Economcs, 4(), 3-3. Cakr, S. and F. Rae, 7. Sukuk vs. Eurobonds: Is there a dfference n value-at- rsk? Internatonal Monetary Fund Workng Paper, WP/7/37. 7

26 Dasuk, A. and S. Mokhtar,. Crtcal Apprsal of Shar ah Issues on Oownershp n Asset-Based Sukuk as Implmented n the Islamc Debt Market. Internatonal Shar ah Research Academy for Islamc Fnance (ISRA), Research Papers (8), -35. Eckbo, B Valuaton Effects of Corporate Debt Offerng. Journal of Fanncal Economcs, 5, 9-5. Gasbarro, D., K., Le, R. Schwebach, and J. Zumwalt, 4. Syndcated Loan Announcements and Borrower Value. Journal of Fnancal Research, 7, Ghosh, D.. Captal Markets and Fnancal Assets: Decsons on Acquston and Issuance of Securtes. Lambert Academc Publshng, Germany. Godlewsk, C., R. Turk-Arss and L. Well,. Are Islamc Investment Certfcates Specal? Evdence on the Post-Announcement Performance of Sukuk Issues. LaRGE Workng Paper No. -5. Avalable at SSRN: or Herwany, A. and E. Febran,. A Value at Rsk Analyss on the Performance of Sukuk and Conventonal Bonds: The Case of Asan Markets. Avalable at SSRN: Howladar, K. 9. The Future of Sukuk: Substance over Form? Understandng Islamc Securtzaton, Asset-Backed and AAOIFI Prncples, Moody s Investosr, 6 May. Ibrahm, Y. and M. Mna, 9. Islamc Bonds and the Wealth Effects: Evdence from Malaysa. Investment Management and Fnancal Innovatons, Vol. 6, Issue. ISRA-Internatonal Shar ah Research Academy for Islamc Fnance.. Internatonal Islamc Fnancal System: Prncples & Operatons. Edton, Kuala Lumpur, Malaysa. Lahsasna, A. and L. Ln,. Issues n Islamc Captal Markets: Islamc Bond/Sukuk. Thrd Internatonal Conference on Busness and Economc Research, Bandung, Indonesa, March -3. Lummer, S. and J. McConnell, 989. Further Evdence on the Bank Lendng Process and the Captal-Market Response to Bank Loan Agreements, Journal of Fnancal Economcs, 5, 99-. MacKnlay, A Event Studes n Economcs and Fnance, Journal of Economc Lterature, 35,