The Day of the Week effect on stock market returns and volatility: Evidence from Nigeria and South Africa.

Transcription

1 MPRA Munich Personal RePEc Archive The Day of he Week effec on sock marke reurns and volailiy: Evidence from Nigeria and Souh Africa. Umar Bida Ndako Deparmen of Economics Bayero Universiy Kano P.M.B.3011 Kano, Nigeria 5. July 013 Online a hp://mpra.ub.uni-muenchen.de/48076/ MPRA Paper No , posed 10. July :17 UTC

2 The Day of he Week effec on sock marke reurns and volailiy: Evidence from Nigeria and Souh Africa. BY UMAR BIDA NDAKO Deparmen of Economics Bayero Universiy Kano P.M.B.3011 Kano, Nigeria Absrac: This paper examines he day of he week effec for he Nigerian and Souh African equiy markes over pre-liberalisaion and pos-liberalisaion periods. The paper uses Exponenial Generalized Auoregressive Condiional Heroskedasiciy (EGARCH) model o esimae he day of he week effec boh in he mean and variance equaions. The pos-liberalisaion period for he Nigerian equiy marke exhibis day of he week effec on Fridays only in he mean equaion. While in he variance equaion, here is evidence of day of he week effec on Tuesdays and Thursdays respecively. In Souh Africa, here is significan evidence of he day of he week effec on Mondays and Fridays during he pre-liberalizaion period. During he pos-liberalisaion period, here is evidence of day of he week effec on Thursdays in he mean equaion and Fridays only in he variance equaion. JEL Classificaion: C, G10, G1 Keywords: day of he week effec; volailiy; E-GARCH model

3 1 Inroducion The issue of he day of he week or anomalies in sock reurns, foreign exchange and T-bill markes are well documened and widely examined in he lieraure of financial markes. As observed by Basher and Sadorsky (006) examining day of he week effec is very crucial o invesors because i provides he necessary informaion o invesors o adjus heir porfolio based on he abnormaliy of he calendar effec. For example, invesors could buy socks wih days of abnormal reurns and sells socks wih days of abnormally high reurns. Earlier works in his area include Cross (1973), French (1980), Gibbons and Hess (1981), Keim and Sambaugh (1984) and Aggarwal and Rivoli (1989). However, all of he above works focus only on he mean equaions of he sock marke reurns and use Ordinary Leas Squares (OLS) esimaion mehod o regress reurns on five daily dummy variables. Recen empirical sudies are now considering no only he mean reurns bu also he variance equaions. This can be found in he sudies by Hsieh (1988), Berumen and Kiymaz (001), Kiymaz and Berumen (003) and Yalcin and Yucel (006). Considering day of he week in he variance equaion also has he suppor of Engle (1993) as he poins ou ha hose risk-averse invesors would reduce heir invesmen on hose asses which are likely o increase in volailiy. Therefore, consideraion of boh reurns and volailiy are criical o invesors. Kiyamaz and Berumen (003) clearly show ha invesors are no only ineresed abou a day reurns bu also abou he corresponding volailiy of he day and having such knowledge allow invesors o adjus heir porfolios by aking ino accoun day of he week variaion in volailiy. Many sudies on he day of he week boh on developed and oher emerging markes have used GARCH (1, 1) model o examine he effec on he day of he week. This sudy, however, uses he Exponenial Generalized Auoregressive Condiional Heeroskedasiciy (EGARCH) model. The main essence of using he EGARCH model is o capure he possible asymmeric effec which is no possible wih GARCH (1, 1) models. I shows he leverage effec, ha is he reacion of he invesors o he news in he markes. When volailiy rises i means invesors are responding o bad news and when i falls i means hey are responding o good news. Also wih he EGARCH model, we do no require he assumpion of nonnegaiviy resricion, which ensures ha all coefficiens are posiive under GARCH (1, 1) model. Alhough here are several works on Sub-Saharan Africa counries on day of he week effec, here has no been much effor in examining he asymmeric effec of he daa. This sudy aemps o fill his gap by examining day of he week effec for Nigerian and Souh African equiy markes using he EGARCH model. The sudy follows Kiyamaz and Berumen (003) by considering day of he week in boh reurns and variance specificaion. Whils Kiyamaz and Berumen (003) use GARCH (1, 1) model, his sudy uses EGARCH model. The moivaion for his work is based on he fac ha boh Nigeria and Souh Africa liberalise heir financial markes in 1995 and as expeced, i has provided opporuniies for foreign invesors o acively paricipae in hese markes which in urn increases he level of liquidiy, saving and growh of hese economies. However, experience has shown paricularly from he Souh Eas Asia ha hese huge poenials are ofen characerised by a high level of uncerainy or volailiy as explained by Singh (1997), Sigliz (000), Allen and Gale (000). This parly explains why we esimae he day of he week effec boh in he mean and in variance equaions. Evidence from figure 1 below indicaes he daily sock reurns for Nigeria and Souh Africa respecively. Boh reurns clearly exhibi volailiy clusering, which is high period is followed by low period volailiy. I evolves over ime in a coninuous manner and i is asymmeric, responding o increasing and decreasing in prices.

4 The objecive of his sudy is herefore o examine he day of he week effec boh in he mean and variance equaions for Nigerian and Souh African equiy markes respecively. The choice of Nigeria and Souh Africa for his sudy is due o heir imporan economic role in sub-saharan Africa; hey are he wo bigges economies in Sub-Saharan Africa. The wo economies also have rich experiences in economic and financial secor reforms over he pas wo decades. The sudy covers wo periods: he pre-liberalisaion and pos-liberalisaion periods excep for he Nigerian daily reurns, which cover only he pos - liberalisaion period due o lack of daa for he pre-liberalisaion period. The pre-liberalisaion period is simply he period before he domesic markes open up o foreign invesors while pos-liberalisaion is afer he markes have become inernaionalised. The main finding from he sudy is ha for he Nigerian equiy marke, here is evidence of day of he week effec on Fridays only in he mean equaion. While in he variance equaion, here is evidence of day of he week effec only on Tuesdays and Thursdays respecively. In Souh Africa, here is significan evidence of he day of he week on Mondays and Fridays during he pre-liberalisaion period. During he pos-liberalizaion period, here is evidence of day of he week effec on Thursdays in he mean equaion and Fridays only in he variance equaion. Following his inroducion, Secion wo comprises a brief review of he Nigerian and Souh Africa capial marke developmen. Secion hree is he lieraure review and secion four consiss of daa source and model specificaion. Secion five gives he analysis of he empirical resuls and secion six concludes he paper. Figure 1: Sock daily reurns for Nigeria and Souh Africa Sock daily reurns for Nigeria Sock daily reurns for Souh Africa 8 RT.08 SRT Brief review of Nigerian and Souh African Equiy marke developmen The Nigerian Sock Exchange (NSE) is relaively young when compared wih esablishmen of Nigerian commercial banks. I sared operaions in mid-1961 wih eigh socks and equiies; here were also abou seven UK firms quoed on NSE which had, a he same ime, dual quoaions on he London Sock Exchange. A he commencemen of operaions, he NSE sared wih 0.3 million shares worh N1.5 m in 334 deals and he value coninued o grow seadily o N16.6m in 634 deals by 1970 (CBN 004).

5 Since he 1960s up o he lae 1980s, rading a he NSE was dominaed by governmen securiies and his was parly explained by he implemenaion of he Nigerian Enerprises Promoion Decree of 197 and 1977 which allowed a high level of public paricipaion in he capial marke. Also prior o he deregulaion of he Nigerian capial marke in 1995, he pricing of new issues was conrolled by he SEC as agains firms preferences for a marke deermined pricing sysem (CBN 004). In 1995 he Federal Governmen liberalised he capial marke wih he abrogaion of Laws ha preven foreign invesors from paricipaing in he domesic capial marke. This includes: The Foreign Exchange (Monioring and Miscellaneous Provision Decree No: 17, 1995; Nigerian Invesmen Promoion Commission Decree No: 16, 1995; Companies and Allied Maers Decree of 1990 and Securiies and Invesmen Ac (ISA) 45 of These legislaions have accorded Nigerians and foreign invesors he same righ, privileges and opporuniies for invesmen in securiies in he Nigerian capial markes. Oher key measures include: The Cenral Securiy Clearing Sysem (CSCS) which commenced operaions in April I is a cenral deposiory for all he share cerificaes of quoed securiies including new issues. The coming of he CSCS has made he rading on he NSE o be carried ou on an Auomaed Trading Sysem (ATS), which enable a marke order o be carried ou in ransacion days T+3 days (NSE 009). Also in July 00, he sock exchange inroduced he e-business plaform, which makes i possible for invesors in he Exchange Marke o access heir CSCS daabase from he Exchange Websie for he purpose of monioring movemen in heir sock accouns. In 007, he SEC approved he esablishmen of he Invesors Proecion Fund (IPF) wih he objecive of compensaing invesors who suffer losses as a resul of insolvency, bankrupcy or negligence on he par of a dealing member firm of a securiies exchange. Also in 007, o furher deepen he marke, he NSE launched new producs, including morgagebacked securiies, asse-backed securiies, derivaives and exchange- raded funds. In 008, he NSE crashed and suffered he heavies loss in is hisory; alhough he global financial crisis parly conribued o his, he main source of he crisis could be raced o insider abuse, share manipulaion, margin loan scandals and oher negaive aciviies perpeuaed by operaors of he marke. The sock marke had less han N1rillion marke capialisaion in 1999 bu his jumped o N15.3 rillion in he firs quarer of 008, however, his amoun plummeed o N7.53 rillion in he firs week of November 008 (NSE 009). The Johannesburg Sock Exchange (JSE) is he oldes in Sub-Saharan Africa esablished in The JSE is now he mos developed sock marke in he Africa. I has undergone for he pas wo decades a series of reforms paricularly since he inroducion of he consiuional democracy and he end of aparheid in This has helped o increase is performance and compeiiveness a he global sage. World Bank (007) indicaes ha he Johannesburg Sock Marke is ranked he fourh larges among he emerging markes. The IMF (008) also confirms ha Souh Africa s financial sysem is he mos robus in erms of financial infrasrucure and pruden macroeconomic managemen. In May 00, JSE replaced is rading sysem wih he securiies rading sysem of he London Sock Exchange. This furher helps in increasing he level of confidence of he invesors in he JSE. I also helps in reducing he hinness level of he rading and improving he level of marke liquidiy. In 004, he JSE esablished he Alernaive Exchange (ALTx) which is a plaform ha 1 The Decree No: 16 and 17 replaced he abrogaed Nigerian Enerprises Promoion Decree of 1989 and Exchange Conrol Ac of 196 respecively. However, he Invesmen and Securiies Ac (ISA) 007 repealed he ISA 45 of 1999 and subsequenly hose earlier decrees. This paper is an abridged version of a secion of he empirical chaper of my PhD hesis a Universiy of Leiceser.

6 promoes small and medium-sized companies from all secors of he economy especially hose ha have he poenial o grow are encouraged o ge enlised. Currency derivaives were also launched in 007 by he JSE and hese provide opporuniies for invesors o hedge agains major currency flucuaions. (Ndako, 010) 3 Lieraure Review Many empirical works on he day of he week effec in sock reurns have been conduced boh in developed and emerging markes. Earlier sudies include Osborne (196), Fama (1965), Cross (1973), and French (1980). Ohers include Gibbons and Hess (1981), Keim and Sambaugh (1984), Jukuns (1986), and Lakonishen and Smid (1988). All hese sudies arrived a differen conclusions on he effec on he day of he week. Ahanassakos and Robinson (1994) and Dubois and Louve (1996) boh examine day of he week effec for boh developed and emerging markes in heir sudies. They obain negaive Monday sock reurns for he Unied Saes, European and Hong-Kong markes and negaive Tuesday sock reurns for Ausralia, Japan and Souh Korea respecively. Meanwhile, earlier sudies by Kao (1990) also obained low Tuesday and high Wednesday for Japan. Sudies by Poshakwale and Murinde (001) esablish a significan negaive day of he week on Monday and posiive on Friday for Hungary and Poland sock markes. Brooks and Persand (001) evaluae he evidence of he day of he week for five Souheas Asia: Malaysia, Souh Korea, The Philippines, Taiwan and Thailand. They find a significan day of he week effec in hree of he five sock markes sudied. However, hey furher conclude ha marke risk alone may be insufficien in capuring he calendar anomalies. Hui (005) using non-parameric es, examines day of he week effec for four Asian Pacific markes and wo developed markes. He esablishes ha Hong-Kong, Taiwan and Singapore show higher average reurns on Fridays and lower average reurns on Mondays bu he Unied Saes, Japan and Souh Korea show a mixed paern. In he overall, i is only Singapore ha shows a significan day of he week effec. In heir sudies of Chinese sock markes, Cai e al. (006) esablish he presence of he day of he week effec wih negaive reurns on Mondays and Tuesdays afer conrolling for he effec of spill over and auocorrelaion. Recen empirical sudies also consider day of he week effec on volailiy. Berumen and Kiyamaz (001) in addiion o sock reurns also consider day of he week effec on volailiy. They examine he S&P 500 sock index for he period and find a significan day of he week effec boh in reurns and volailiy equaions. They esablish a significan highes and lowes day of he week effec on Wednesday and Monday for reurns equaions, and highes and lowes volailiy on Friday and Wednesday respecively. Kiyamaz and Berumen (003) also find a significan day of he week boh in reurns and volailiy equaions using daily sock indexes for Canada, Germany, Japan, Unied Kingdom, and he Unied Saes. They find highes volailiy on Mondays for Germany and Japan, Fridays for Canada and Unied Saes and Thursdays for he Unied Kingdom. Basher and Sadorsky (006) using boh uncondiional and condiional risk analysis, examine day of he week effec for 1 emerging markes. They find no significan day of he week effec for he majoriy of emerging markes sudied excep he Philippines, Pakisan and Taiwan, which exhibi day of he week afer adjusing for marke risk. Charles (010) examines he asymmeric effec on he day of he week effec for five major inernaional sock markes: France, Germany, Unied Saes, Unied Kingdom and Japan. Using daily sock reurns, he esablishes ha i is possible ha day of he week may creae asymmery in he series bu here is no sufficien evidence ha i influences he seasonal effecs. Hogholn and Knif (010) use differen levels of porfolio aggregaion examine

7 wheher i is possible o examine he aggregaion level effec of hose facors driving he day of he effec. They conclude ha he days of he week srucure are more pronounced in he condiional volailiy in he mean reurns. Some of he sudies in Africa include Aly e al. (004), Agahee (008), Chukwogur (008) and Tachiwon (010). Aly e al. (004) examine day of he week effec using daily sock reurns for Egyp. On he average, he sudy indicaes a posiive and significan Mondays reurns bu here is no indicaion of he day of he week. Tachiwon (010) give an analysis of day of he week effec of he Wes African regional sock marke for he period He observes lowes reurns on Tuesdays and Wednesdays and highes reurns on Fridays. 4 Economeric mehodology In his sudy, we use he exponenial generalized auoregressive condiional heeroskedasiciy (EGARCH) model of Nelson (1991). He has shown ha he EGARCH model performs beer han he GARCH (1, 1) model. This is because EGARCH model does no require he condiion of non negaiviy. Tha is i does no require a siuaion ha ensure coefficiens o be posiive. The EGARCH model also allows us o capure he asymmeric characerisics of daa. This is agains he GARCH (1, 1) model which is based on symmeric assumpion. Therefore, EGARCH model provides he opporuniy for he leverage effec, which usually indicaes he level of response of he invesors o marke news. Volailiy ends o rise in response o bad news (i.e. lower unexpeced sock reurns) and fall in response o good news (i.e. higher unexpeced sock reurns). This idea is moivaed by he earlier empirical work of Black (1976), Chrisie (198) and French, Schwer, and Sambaugh (1987). Therefore, good news and bad news have differen predicabiliy for fuure volailiy as agains oher GARCH models which show symmeric effecs. This paper employs he EGARCH model and i is specified under wo equaions: he mean and condiional variance equaions. Mean equaion This is specified as follows: Y (1) Variance equaion 1 1 log( ).log( 1) () 1 1 Where,, and are consan parameers o be esimaed. Condiional variance log( ) indicaes ha even if he parameers are quie negaive, log( ) will sill be posiive. Is he ARCH erm ha measures he magniude of he shock of he news abou he volailiy. on he oher hand is he GARCH erm and i capures he persisence in he condiional variance. A large posiive indicaes ha i will ake a long ime for he persisence o die ou following a shock he sock markes. The measures he leverage effec; he idea is expeced o be negaive showing ha he bad news has a bigger impac on volailiy han he good news of he same magniude. If 0 indicaes symmeric 0 indicaes posiive news and if 0 his indicaes bad news. The news impac curve (NIC) relaes 1 o in such a way ha pas sock reurns are relaed o curren volailiy. This sudy herefore Follows Engle and Ng (1993) o plo he news impac curves from he EGARCH model. The curve enables us o measure he relaionship beween he news and fuure volailiy. The values of he residuals usually increase exponenially in wo direcions of eiher posiive or

8 negaive parameers. If he parameer is posiive, i is an indicaion of good news arrival in he marke and as expeced invesors will reac more o posiive news han negaive news. However, if he parameer is more negaive, ha means bad news arrival o he marke and invesors will reac accordingly. ( ) h Aexp 1 for 1 0 (3) ( ) h Aexp 1 for 1 0 A exp w /, is he sandard deviaion and w is he consan erm, is Where he parameer of he log ( ( h 1) erm, while is he parameer for he 1 / h 1 erm and is he parameer for he 1/ h 1 erm in he EGARCH log-variance equaion The model is specified in wo ways: firs is he specificaion of he mean reurns only and he second specificaion comprises boh days of he week effec in reurn and volailiy equaions. R o M M T log( ).log( T TH 1 TH ) 1 F 1 F 1 n I 1 1 R i i (4) R o M M Log ( ) V 1 1 C V 1 1 T M T M TH TH V T T V TH F TH F V F n I 1 R i i F log( ).log( 1) + (5) Where R is he reurn and M, T, TH and F are he dummy variables for Mondays, Tuesdays, Thursdays and Fridays a ime respecively. The small n is he lag order which is included o eliminae he possible auocorrelaion in he series. Each day akes he value of 1 and 0 oherwise and Wednesday s dummy is excluded in order avoid he dummy variable rap. 4.1 Daa and official Daes for Sock Marke Liberalizaion This paper uses daily sock price indexes of he Nigeria Sock Exchange (NSE) and Johannesburg Sock Exchange (JSE). The Nigerian daily daa are obained from Daa sream Inernaional (NSE all shares index, S&P/IFCG index). The daily daa for Souh Africa are obained from he JSE and Daa sream Inernaional (JSE/Acuaries and FTSE/JSE). The reurns are obained from he log difference change in he price index R log P log P 1 The oal sample of he Nigerian daily reurn series covers Augus 1 s, 1995 o November 30 h, 010 [4001 observaions] For Souh African daily reurn series, i covers he period January 1 s November 30 h, 010, [5456 observaions]

9 Table 1 The official liberalizaion daes for he wo counries are summarised in he able below: Counry Bekaere al (003) Fuchs-Schundeln and Funke (003) Bekaer e al (005) Nigeria 08/ / Souh Africa / Empirical Resuls The empirical resuls are presened in hree sages: he firs sage is he presenaion of summary saisics; he second sage is he day of he week effec. The hird sage is he presenaion of he news impac curves. Table Summary Saisics for Sock Reurns in Nigeria and Souh Africa Daily daa Nigeria Full-sample (Aug 95- Nov 010) Pre-liberalisaion Pos-liberalisaion (Aug 95 Nov 010 Mean S.D Skewness Kurosis J. Bera Observaion *** *** 4001 Souh Africa Mean S.D Skewness Kurosis J. Bera Observaion Full-sample (Jan 90- Nov 010) *** 5456 Pre- liberalisaion (Jan 90 Feb 95) *** 1346 Pos-liberalisaion (Mar 95 Nov *** 4110 Noe: *** 1% level of significance Table gives he summary saisics for Nigerian and Souh African daily daa. The resuls essenially cover for wo periods: pre-liberalisaion and pos-liberalisaion periods respecively. Meanwhile he Nigerian daily reurn series covers only he pos-liberalisaion period as here is no available daa for he pre-liberalisaion period. For Nigeria, he daily mean reurn is posiive wih an average reurn of 5.5% for he pos liberalisaion period. I indicaes a negaive skewness and a kurosis ha is above 3. I also fails he Jarque-Bera normal disribuion hypohesis a he 1% level of significance. For Souh Africa, he daily mean reurns are greaer han zero for all he hree sample periods wih he daily average reurn of 0.04%, 0.04% and.96 % for he full sample, preliberalisaion and pos liberalisaion periods respecively. Volailiy measured by sandard deviaion indicaes a high level of volailiy in he pos liberalisaion period han in he pre -

10 liberalisaion period. The resuls also show a negaive skewness and a kurosis ha is quie above 3 for all he hree samples, clearly indicaing a lepokuric disribuion. Jarque Bera also rejecs he null hypohesis of a normal disribuion. Table (3) presens he Nigerian pos-liberalisaion period. Esimaes from he mean equaion shows ha he reurns for he oher days of he week are no significanly differen from Wednesday. This means here is no day of he week effec in reurn equaion. The highes reurn is recorded on Fridays wih 1.% and his is followed by Thursdays and Tuesdays wih 1% and 0.6% respecively. However, evidence from he esimaes of boh reurns and variance equaions show ha in he mean equaion, here is a significan day of he week effec on Fridays, which is saisically significan a he 5% level. From he condiional variance equaion, i can be observed ha he volailiy of Tuesdays and Thursdays is differen from Wednesdays and saisically significan a 5% and 10 % level respecively. Fridays volailiy is lower han Wednesdays and he high volailiy in he week is recorded on Tuesdays. Therefore, for he Nigerian pos-liberalisaion period, here is Friday effec in he mean equaion and boh Tuesday and Thursday effecs in he variance equaion. The diagnosic ess indicae absence of auocorrelaion and no ARCH effec a lag (1) from he daa esimaion Tables (4) and (5) presen he esimaes of he day of he week effec for pre- and pos-liberalisaion periods in Souh Africa. The resuls are presened in reurn and also boh in reurn and in volailiy equaions. In he pre-liberalisaion period, he reurns are no significanly differen from Wednesdays excep for Mondays which is differen from Wednesdays a 5% level of significance. This suggess ha he reurns for Wednesdays are higher han Mondays. Thursdays produce he highes reurns during his period while he lowes reurns are recorded on Mondays. In he reurn and volailiy equaions, he resuls of he reurns do no significanly change. Mondays reurns are sill he only reurns ha are significanly differen from Wednesdays a 5% level of significance wih he highes reurns sill on Thursdays bu lowes on Fridays. For he reurn and condiional variance equaion, he esimaed coefficiens indicae evidence of he day of he week effecs on Mondays and Thursdays a 5% and 10 % level of significan respecively. For he condiional variance, he evidence shows ha he volailiy of Fridays is significanly differen from he Wednesdays volailiy. The highes volailiy occurs on Fridays while he lowes occurs on Mondays indicaing ha Wednesdays volailiy is higher han Fridays. Therefore, for Souh Africa, here is a Monday and Thursday effec in he mean equaion only for he pre and pos-liberalisaion periods. There is a Friday effec on he esimaes of he variance equaion. The diagnosics saisics of boh Ljung-Box Q-saisics a levels and a squares indicae no serial correlaion a lag 1 respecively. The ARCH-LM es also indicaes no ARCH effec. In he pos-liberalisaion period, here is no day of he week effec. All he days reurns are no significanly differen from Wednesdays wih he highes reurns in he week being Thursdays followed by Mondays, and he lowes reurns in he week being Tuesdays. The condiional variance equaion indicaes ha he variances for he day of he week are no significanly differen from Wednesdays for Souh Africa. Mondays have lower volailiy han Wednesdays while higher volailiy of he week is observed on Fridays. The auocorrelaion ess show absence of auocorrelaion a lag 1 and no ARCH effec in he daa.

11 Table3 Day of he week effec (pos-liberalizaion period) for Nigeria Esimaes of reurn equaion Esimaes of reurn and volailiy equaions Consan (3.359) Monday (-1.537) Tuesday (0.504) Thursday (0.776) Friday (0.899) Consan (1.307) Monday (-1.638) Tuesday (0.78) Thursday (1.511) Friday *** (4.16) R(-1) *** (1.07) R(-1) *** (.41) R(-1) *** (7.89) R(-) 0.114*** (7.111) Volailiy (-15.36)*** (15.00)*** (.917)*** (35.9)*** Diagnosic es Ljung Box (1) Q sa. - levels (0.476) Ljung Box (1) Q sa.-squares (0.895) ARCH LM Tes Volailiy *** (-5.66) *** (4.95) 0.083*** (.90) *** (347.3) Monday (-0.961) Tuesday 0.403*** (.804) Thursday * (-1.8) Friday (0.64) Diagnosic es Ljung Box (1) Q sa.- levels (0.153) Ljung Box (1) Q sa.-squares (0.317) ARCH LM Tes Z-saisics and probabiliies are in parenheses for he reurn and volailiy esimaion and diagnosics ess respecively. (***) (**) (*) indicae he saisical level of significance a 1%, 5% and 10% respecively. R (-1) and R (-) indicae auocorrelaion lags.

12 Table4 Day of he week effec (pre-liberalizaion period) for Souh Africa Esimaes of reurn equaion Esimaes of reurn and volailiy equaions Consan ** (.1715) Monday ** (.134) Tuesday (1.574) Thursday (0.1) Friday (-1.504) R(-1) (-6.687)*** Volailiy (-38.35)*** 0.53 (6.775)*** (-.334)** (6.57)*** Diagnosic es Ljung Box (1) Q sa.- levels (0.30) Ljung Box (1) Q sa.-squares 9.94 (0.678) ARCH LM Tes Consan ** (.1699) Monday ** (-.079) Tuesday (-1.359) Thursday (-0.147) Friday (-1.476) R(-1) (7.705)*** Volailiy (-0.446) 0.80*** (3.436) (0.016) (14.18)*** Monday (1.1634) Tuesday (-1.544) Thursday -0.0 ( ) Friday (.644)** Diagnosic es Ljung Box (1) Q sa.- levels (0.695) Ljung Box (1) Q sa.-squares (0.83) ARCH LM Tes Z-saisics and probabiliies are in parenheses for he reurn and volailiy esimaion and diagnosics ess respecively. (***) (**) (*) indicae he saisical level of significance a 1%, 5% and 10% respecively. R (-1) indicaes auocorrelaion lags.

13 Table5 Day of he week effec (pos-liberalizaion period) for Souh Africa Esimaes of reurn equaion Esimaes of reurn and volailiy equaions Consan (1.10) Monday 0.001** (.804) Tuesday (0.018) Thursday (1.638) Friday (-0.135) Consan * (1.80) Monday (1.465) Tuesday (-1.049) Thursday * (1.769) Friday (-0.006) R(-1) *** (5.0) R(-1) *** (5.368) Volailiy *** (-7.785) *** (10.6) *** (-5.941) 0.980*** (66.) Diagnosic es Ljung Box (1) Q sa. - levels (0.575) Ljung Box (1) Q sa. -squares (0.999) ARCH LM Tes Volailiy *** (-8.695) *** (19.0) *** (-16.15) *** (71.36) Monday (-1.053) Tuesday (-.606) Thursday (0.153) Friday ** (1.99) Diagnosics ess Ljung Box (1) Q sa.- levels (0.749) Ljung Box (1) Q sa.-squares.03 (0.991) ARCH LM Tes Z-saisics and probabiliies are in parenheses for he reurn and volailiy esimaion and diagnosics ess respecively. (***) (**) (*) indicae he saisical level of significance a 1%, 5% and 10% respecively.. R (-1) indicaes auocorrelaion lags.

14 SIG SIG SIG News Impac Curves Figure () Souh Africa: pre-liberalisaion (Daily) -A Souh Africa: pos-liberalisaion (Daily) -B Z Z Nigeria: pos-liberalisaion (Daily) -C Z Figures (a) o (c) presen he news impac curves (NICs) for daily sock reurns for Souh Africa and Nigeria respecively. I usually plos he nex period volailiy (sig) ha would arise from various posiive and negaive values of he lagged residuals or shocks (Z) The verical axis represens he level of curren volailiy while he horizonal axis represens he lagged residuals from he EGARCH model. The curves confirm ha he bad news of he same magniude resuled in more volailiy in he wo sample periods. The Nigerian daily daa however, exhibis posiive impac curve. Figure (A) is he Souh African daily reurn series (pre-liberalisaion period) and i indicaes ha he lagged value of he shock or error ranging from -1 o 1. I furher shows ha he value of condiional variance is for a shock of -1 and 0.00 for a shock of 1. This clearly indicaes ha invesors reac more o negaive news han posiive news. Figure (B) on he oher hand is he Souh African daily reurn series (pos-liberalisaion) and i shows a condiional variance of for a shock of -1 and 0.00 for a shock of 1. This also shows ha invesors reac more o negaive news han o posiive news. Figure (C) depics he Nigerian daily reurn series for pos-liberalisaion period wih condiional variance of for a shock of 1. This clearly indicaes ha invesors in he Nigerian equiy marke reac more o posiive news han he negaive news and his is a odds wih mos empirical findings.

15 6 Conclusions This sudy examines he day of he week effec for Nigerian and Souh African equiy markes. I covers wo periods: pre-liberalisaion and pos liberalisaion, however, for Nigeria, he sudy covers only pos-liberalisaion period as here is no available daa for he pre - liberalisaion period. The paper uses an EGARCH model o esimae he day of he week effec in he mean and variance equaions. In Souh Africa, here is significan evidence of he day of he week on Mondays and Fridays during he pre-liberalisaion period bu here is no day of week effec in he pos-liberalisaion period. The pos liberalisaion period of he Nigerian equiy marke also exhibis day of he week effec on Mondays and Tuesdays respecively. References Agahee, U. S. (008) Day of he Week Effecs: Evidence from he Sock Exchange of Mauriius (SEM) Inernaional Research Journal of Finance and Economics 17, 8-14 Aggarwal, R. and Rivoli (1989) Seasonal and day-of-he week effec in four emerging sock markes Financial Review, 4, pp Allen, G and D. Gale (000), Bubbles and Crises, The Economic Journal, 110, pp Aly, H., Mehdian, S. and Perry, M. (004) An Analysis of he Day-of-he-Week Effecs in he Egypian Sock Marke, Inernaional Journal of Business, 9, pp: Ahanassakos, G., and Robinson, M. J. (1994). The day-of-he-week anomaly: The Torono sock exchange experience Journal of Business Finance and Accouning, 1, pp Basher, S. A. and Sadorsk, P. (006) Day-of-he-week effecs in emerging sock markes, Applied Economics Leers, 13, pp Bekaer, G. and C. Harvey (000), Foreign Speculaors and Emerging Equiy Markes, The Journal of Finance, 55, pp Bekaer, G; C.R. Harvey, and C.T. Lundblad (003), Equiy Marke Liberalizaion in Emerging Markes, Journal of Financial Research, 6, PP Bekaer, G. C.R. Harvey, and C. Lundblad (005), Does financial liberalizaion spur growh? Journal of Financial Economics, 77, pp.3-55 Berumen, H. and Kiymaz, H (001) The day of he week effec on sock marke volailiy Journal of Economics and Finance, 5, pp Bhaacharya, U. and Hazem D. (00) The World Price of Insider Trading The Journal of Finance, 57, pp Black, F. (1976), Sudies of Sock Price Volailiy Changes, Proceedings of he American Saisical Associaion, Business and Economic Saisics Secion,

16 Brooks, C. and Persand G. (001) Seasonaliy in Souheas Asian sock markes: Some new evidence on day-of-he-week effecs Applied Economics Leers, 8, pp Cai, J., Li, Y., and Qi, Y. (006). The day-of-he-week effec: New evidence from he Chinese sock marke The Chinese Economy, 39, pp Charles, A. (010) The day-of-he-week effecs on he volailiy: The role of he asymmery European Journal of Operaional Research. 0, pp Chrisie, A.A. (198), The Sochasic Behaviour of Common Sock Variances: Value, Leverage and Ineres Rae Effecs, Journal of Financial Economics, 10, Chukwuogor, C. (008) An Economeric Analysis of African Sock Marke: Annual Reurns Analysis, Day-of-he-Week Effec and Volailiy of Reurns Inernaional Research Journal of Finance and Economics 14, Cenral Bank of Nigeria, (004), Nigeria: Major Economies, Finance and Banking Indicaors Cross, F (1973) The behaviour of sock prices on Fridays and Mondays Financial Analys Journal, 9, pp Dubois M, Louve P. (1996) The day-of-he-week effec: he inernaional evidence Journal of Banking and Finance, 0, pp Engle, R. (1993). Saisical models for financial volailiy Financial Analyss Journal, Fama, E. F. (1965) The behaviour of sock marke prices Journal of Business, 38, pp French, K. (1980) Sock reurns and he weekend effec Journal of Financial Economics, 8, pp French. K.. Schwer. G. and Sambaugh. R. (1987). Expeced sock reurns and volailiy Journal of Financial Economics. 19, pp.3-30 Fuchs-Schundeln, N. and N. Funke (003) Sock marke liberalizaions: Financial and macroeconomic implicaions Review of World Economics, 139, pp Gibbons, M. and Hess, P (1981) Day of he week effecs and asse reurns Journal of Business, 54, pp Hsieh, D. A. (1988) The saisical properies of daily foreign exchange raes: Journal of Inernaional Economics, 4, pp Hogholm, K. and Knif J. (009) The impac of porfolio aggregaion on day-of-he-week effec: Evidence from Finland Global Finance Journal 0, No.1, pp Hui, T. K. (005). Day-of-he-week-effecs in US and Asia-Pacific Sock markes during he Asian financial crisis: a non-parameric approach Omega, 33, pp

17 Inernaional Moneary Fund (008), Souh Africa: Financial Sabiliy Assessmen, Including Repor on he Observance of Sandards and Codes on he following opic: Securiies Regulaion, Publicaion Services, Washingon, D.C Johannesburg Sock Exchange: Available a: hp//ww.jse.co.za [accessed January 010] Junkus, J. C. (1986) Weekend and day of he week effecs in reurns on sock index Fuures The Journal of Fuures Markes, 6, pp Kao, K. (1990). Weekday paerns in Japanese sock reurns Managemen Science, 36, pp Kiyamaz, H. and Berumen, H. (003) The day of he week effec on sock marke volailiy and volume: Inernaional evidence Review of Financial Economics, 1, pp Keim, B and Sambaugh, F., (1984) A furher invesigaion of he weekend effec in sock reurns Journal of Finance, 39, pp Lakonishok, J. and Smid, S. (1988). Are seasonal anomalies real? A niney-year Perspecive Review of Financial Sudies 1, pp Ndako U.B. (010) Financial developmen, economic growh and marke volailiy: Evidence from Nigeria and Souh Africa PhD Thesis, Universiy of Leiceser, Unied Kingdom. Ndako U. B. (010) Sock markes, Banks and Economic growh: Time series evidence from Souh Africa The African Finance Journal, PP. 7-9 Nelson, D.B. (1991). Condiional Heeroskedasiciy in Asse Reurns: A New Approach Economerica, 59, pp Nigerian Sock Exchange (009). Available a: hp:// [Accessed January 010] Osborne, M. F. M. (196) Periodic srucure in he Brownian moion of he sock marke Operaions Research, 10, pp Poshakwale, S., and Murinde, V. (001). Modeling he volailiy in Eas European emerging sock markes: Evidence on Hungary and Poland Applied Financial Economics, Vol.11, pp. No.4, Sigliz, J.E. (000), Capial Marke Liberalizaion, Economic Growh, and Insabiliy, World Developmen, 8, pp Singh, A. (1997), Financial Liberalizaion, Sock markes and Economic Developmen, The Economic Journal, 107, pp Tachiwou, A. M. (010) Day-of-he-Week-Effecs in Wes African Regional Sock Marke Inernaional Journal of Economics and Finance 4,

18 World Bank (007) Making Finance Work for Africa The Inernaional Bank for Reconsrucion and Developmen/The World Bank. Available a: [Accessed Augus, 010] Yalcin, Y. and E.M. Yucel (006) The day-of he week effec on sock marke volailiy and reurn: Evidence from Emerging Markes Czech journal of Economics and Finance, 56, pp. 58-7