Day-of-the-week effects in selected East Asian stock markets

Transcription

1 MPRA Munich Personal RePEc Archive Day-of-he-week effecs in seleced Eas Asian sock markes Ricky Chee-Jiun Chia and Venus Khim-Sen Liew and Syed Azizi Wafa Syed Khalid Wafa Labuan School of Inernaional Business and Finance, Universii Malaysia Sabah. 007 Online a hp://mpra.ub.uni-muenchen.de/799/ MPRA Paper No. 799, posed. February :39 UTC

2 Day-of-he-week effecs in Seleced Eas Asian sock markes Ricky Chee-Jiun Chia Venus Khim-Sen Liew * Syed Azizi Wafa Syed Khalid Wafa Labuan School of Inernaional Business and Finance, Universii Malaysia Sabah, Jalan Sungai Pagar, Labuan, Malaysia. Absrac This sudy examines he day-of-he-week effecs in he Taiwan, Singapore, Hong Kong and Souh Korea sock markes. Various significan day-of-he-week effecs, including he ypical negaive Monday and posiive Friday effecs are deeced in he sock markes of Taiwan, Singapore and Hong Kong. Furher analysis shows ha only Friday effec in Taiwan is susainable while all oher effecs disappeared compleely afer accouning for equiy risks. Besides, his sudy also finds evidences of risk and reurn radeoff as well as asymmerical marke effecs. Corresponding auhor: Venus Khim-Sen Liew Labuan School of Inernaional Business and Finance, Universii Malaysia Sabah, Jalan Sungai Pagar, Labuan, Malaysia Tel: venusliew@ums.edu.my, venusliew@yahoo.com Fax: * Corresponding auhor.

3 Day-of-he-week effecs in Seleced Eas Asian sock markes I. Inroducion The day-of-he-week effec, in which he mean reurns are observed o be differen on each day of he week, is one of he mos well-known sock marke anomalies. The presence or absence of his effec has imporan implicaions for invesors. Their invesmen sraegies, porfolio selecion and managemen will vary in accord o differen effecs, o reap maximum profis. Due o is imporance, his issue has received exensive research in he pas. Mos empirical sudies repored significan negaive average daily reurn on Monday, alhough here is also some evidence of he lowes mean reurns on oher days of he week (see, Brooks and Persand, 00; Basher and Sadorsky, 006). Ineresingly, mos of he previous findings are obained by esimaion procedures ha do no allow for he consideraion of asymmeric behavior in sock reurns. However, here is no reason o pre-suppose ha he reurn in he sock marke is symmeric is naure. Indeed, Alexakis and Xanhaki (995), among he few, are able o find evidence of he asymmeric behavior in day-of-he-week effec in he Greek sock marke. Similar evidence from he sock markes of 9 developed counries is found in Balaban e al. (00). Working along his line of research, he major objecive of his paper is o sudy he asymmeric behavior in he sock markes of Hong Kong, Taiwan, Singapore and Souh Korea. The four economies are collecively known as Eas Asian One excepional case is he work of Basher and Sadorsky (006), who esimaed condiional models ha allow for asymmeric marke effecs. Few researchers like Engle and Ng (993), for insance, have poined ou ha he marke reacion on bad and good news appears o be asymmery in naure. Besides, Engle (00) argued ha marke paricipan reac differenly wih negaive and posiive reurn. Indeed, here are empirical findings o sugges asymmerical sock marke behavior. For insance, Nelson (99) found ha negaive reurns are followed by a higher volailiy han he posiive reurns. Glosen e al. (993) show ha posiive (negaive) unanicipaed reurns end o resul in downward (upward) marke revision.

4 Tigers for mainaining high growh raes and rapid indusrializaion beween he since early 960s 3. Today, hese markes are he main plaforms for foreign invesmens in he Eas Asian region 4, oher han he Tokyo Sock Exchange marke. Thus, i is worh o scruinize wheher here is any calendar anomaly (in he presen conex, day-of-he-week effec) in hese markes. Previously, some evidence of day-of-he-week effecs in hese four markes has been provided by Wong e al. (99), Choudhry (000), Brooks and Persand (00), jus o name a few, using various mehodologies 5. Noneheless, he above menioned sudies have no accouned for he asymmerical marke behavior. Hence, his sudy exends he sudy of day-of-he-week effecs in hese markes, by incorporaing asymmerical behavior. II. Daa and Mehodology The daa for his sudy consiss of daily closing composie indices on he Eas Asian Tigers sock markes, covering he period from s January 000 o 3 s December 006. This sudy adops he following commonly used model o examine he day-of-he-week effec: R = a + 5 k a δ + b R ε () i i i= i= i i + 3 See hp:// [Rerieved: ]. 4 According o he World Federaion of Exchange (007), as of he end of 006, he oal marke capializaion of hese four sock markes is USD 3.53 rillions, which is abou one-hird (one-half if Japan is excluded) of he oal marke capializaion in he Eas Asian region. 5 These auhors have included a leas one of he four Tigers markes in heir sudies, bu none of hese sudies are solely devoed o he sudy of he Tiger markes as a whole using comparable mehodology. 3

5 where R = 00 ln( I / I ) is he rae of reurn in period. I is he sock index a he end of period. a o a 5 represen he dummy variables for Tuesday o Friday 6. ε denoes he disurbance erm wih zero mean and consan variance. Besides, he following Exponenial Generalized Auoregressive Condiional Heeroscedasiciy in mean (EGARCH-M) model, which allows one o incorporae volailiy effec, risk premium as well as asymmerical behavior in esimaion, is also included in his sudy: R 5 = k i i i= i= i i d + d δ + e R + fσ + ξ () where p q 5 ξ i ξ i logσ = g + γ j logσ j + βi + ψi + h iδi (3) j= i= σ i π σ i i= where f measures he reward o risk raio, ξ is an error erm wih zero mean and condiional variance σ (see Nelson, 99 for deails). The signs of he esimaed f and ψ i are of paricular ineres. If f < 0, i indicaes ha here is a radeoff relaionship beween reurn and risk. If ψ i 0, i can be inerpreed as evidence of sock marke asymmeric behavior. 6 Monday dummy variable is excluded o avoid he dummy variable rap. 4

6 III. The Resuls Table conains summary saisics for he daily reurns for all he Eas Asian Tigers sock indices. Among ohers, he mos ineresing feaures revealed by his able include: Firs, Monday reurns are consisenly negaive while he Friday reurns are always posiive in hese markes. Second, all reurns have asymmerical disribuion as hey have non-zero skewness. These wo feaures sum up o sugges he presence of asymmerical day-of-he-week effecs in hese markes, which is confirmed by he succeeding formal analysis. [Inser Table here] Table presens he OLS resuls for he day-of-he-week effecs in his sudy. The resuls show ha he coefficien of inercep erm ( a ) ha represens he average daily reurn on he benchmark day of Monday is significanly negaive in he Taiwan (-0.6%) and Singapore (-0.3%) sock markes. This observaion is consisen wih previous finding of significan negaive Monday and posiive Friday effecs in mos sock markes. In he case of Hong Kong, here is a Friday effec, alhough negaive Monday effec is no presen. Besides, Tuesday and Thursday effecs in also found in Singapore. In addiion, Taiwan has significanly differen effecs hroughou he 5 rading days, while here is no effec a all in Souh Korea, however. I is noeworhy ha Brooks and Persand (00) also repored no day-of-he-week effec in Souh Korea. [Inser Table here] 5

7 In furher analysis, he significan mean reurns repored in Table are included as par of he explanaory variables in he EGARCH - M models, o deermine wheher he effecs are due o he equiy risks (Lucey, 000). The esimaed resuls are summarized in Table 3. Noe ha, if he included dummy variables (for days significan in OLS esimaion) are sill significan in he mean equaion of he EGARCH M model, i may be concluded ha he calendar effec is no due o he variaion in he equiy risk. By his principle, Table 3 reveals ha equiy risk can accoun for he all he idenified day-of he-week effecs in Taiwan, Hong Kong and Singapore sock markes, wih he excepion of Friday effec in Taiwan. In oher words, only he Friday effec in Taiwan is susainable afer adjusing for equiy risks 7. Besides, wo oher sylized feaures are observed in Table 3: Firs, he risk premium has negaive impac (indicaed by he sign of f ) on sock reurns in Taiwan (significan a 0% level), Singapore and Hong Kong. This implies a radeoff beween reurn and risk. Second, he hree marke reurns exhibi asymmerical behavior, since a leas one ψ is saisically differen from zero. Moreover, he saisically significan posiive values of ψ s in mos cases reveals ha posiive reurn will induce higher volailiy in reurn han negaive reurn in hese markes excep Souh Korea. [Inser Table 3 here] IV. Conclusion This sudy examines he day-of-he-week effecs in he Taiwan, Singapore, Hong Kong and Souh Korea sock markes. Among ohers, his sudy, in line wih mos previous sudies, finds he exisence of various significan day-of-he-week effecs, including he 7 This finding is consisen wih Basher and Sardorsky (006), which show ha day-of-he-week effecs are presen in, among ohers, Taiwan even afer adjusing for marke risk. 6

8 ypical negaive Monday and posiive Friday effecs, in all hese markes excep he Souh Korea sock marke. However, afer adjusing for equiy risks, only he Friday effec (posiive reurn) in Taiwan is susainable while all oher effecs disappeared compleely. Besides, his sudy also finds evidences of risk and reurn radeoff as well as asymmerical marke effecs. One major implicaion of hese findings is ha invesors in Taiwan, Singapore and Hong Kong may consider buying shares on Monday and selling hem on Friday, or conversely shor-selling on Friday and buying back on Monday. As for he Souh Korea sock marke, furher researches may be conduced o see wheher here are oher forms of anomalies o be exploied. References: Alexakis, P. and Xanhakis, M. (995) Day of he week effec on he Greek sock marke, Applied Financial Economics, 5, Balaban, E., Bayar, A. and Kan, O. Z. (00) Sock reurns, seasonaliy and asymmeric condiional volailiy in world equiy markes, Applied Economics Leers, 8, Basher, S. A. and Sadorsky, P. (006) Day-of-he-week effecs in emerging sock markes, Applied Economics Leers, 3, Brooks, C. and Persand, G. (00) Seasonaliy in Souheas Asian sock markes: some new evidence on day-of-he-week effecs, Applied Economics Leers, 8, Choudhry, T. (000) Day of he week effec in emerging Asian sock markes: evidence from he GARCH model, Applied Economics Leers, 0, Engle, R. (00) GARCH 0: The use of ARCH/GARCH models in Applied Economics, Journal of Economics Perspecive, 5, Engle, R. F. and Ng, V. K. (993) Meauring and Tesing he Impac of News on volailiy, Journal of Finance, 48,

9 Glosen, L. R., Jagannahan, R. and Runkle, D. E. (993) On he Relaion beween he Expeced Value and he Volailiy of he Nominal Excess Reurn on Sock, The Journal of Finance, 5, Lucey, B. M. (000) Anomalous daily seasonaliy in Ireland? Applied Economics Leers, 7, Nelson, D. B. (99) Condional heeroskedasiciy in asse reurns: A new approach, Economerica, 59, Wong, K.A., Hui, T.K. and Chan, C.Y. (99) Day-of-he-week effecs: evidence from developing sock markes, Applied Financial Economics,, World Federaion of Exchange (007) Equiy: Domesic Marke Capializaion. Available:hp:// [Rerieved: ]. 8

10 Table. Summary saisics on he day-of-he-week effecs in he Eas Asian Tigers sock markes Mean Sd. Dev Skewness Kurosis Jarque-Bera (P - value) Mean Reurn per uni of Risk Taiwan Monday (0.0000) Tuesday (0.0000) Wednesday (0.0000) Thursday (0.0000) Friday (0.0000) 0.76 Singapore Monday (0.0000) Tuesday (0.0000) Wednesday (0.0000) Thursday (0.0000) 0.09 Friday (0.0000) 0.8 Hong Kong Monday (0.0000) Tuesday (0.0000) Wednesday (0.0000) Thursday (0.0000) Friday (0.0000) Souh Korea Monday (0.0000) Tuesday (0.0000) Wednesday (0.0000) Thursday (0.0000) 0.06 Friday (0.0000)

11 Table. OLS resuls for day-of-he-week effecs Parameer Taiwan Singapore Hong Kong Souh Korea Consan * -0.54** a (0.003) (0.08) (0.4738) (0.55) Tuesday 0.49** 0.745** a (0.0369) (0.030) (0.398) (0.8) Wednesday * a 3 (0.008) (0.43) (0.6348) (0.05) Thursday a ** (0.063) 0.586** (0.0495) (0.7656) 0.73 (0.0) Friday 0.405* 0.53* 0.54*** 0.74 a 5 (0.0004) (0.007) (0.0987) (0.055) Reurn (-) b (0.7) (0.3540) (0.46) (0.887) Reurn (-) 0.055** b (0.049) (0.9076) (0.48) (0.09) Reurn (-3) ** b 3 (0.74) (0.0354) (0.89) (0.8004) Reurn (-4) * * b 4 (0.000) (0.0056) (0.3757) (0.936) ARCH-LM Saisic (p-value) a 5 lags lags Ljung-Box Q Saisic (p-value) b 5 lags lags Wald Tes (p-value) c F-saisic Chi square *, ** and *** denoe significan a, 5 and 0% level respecively. Numbers in parenheses depic p- values. a Tes for remaining ARCH effec. b Tes for serial correlaion. c Null hypohesis is: H = a = a = a = (same average daily reurn for he week) a5 0

12 Table 3. Esimaed EGARCH - Mean model Parameer Taiwan Singapore Hong Kong (p,q) (,4) (4,3) (3,3) Consan, d (0.5559) (0.4759) (0.67) Tuesday, d (0.564) (0.8408) -- Wednesday, d (0.864) Thursday, d (0.5430) (0.765) -- Friday, d (0.0508)*** (0.907) (0.47) Reurn (-), e (0.740) (0.7097) (0.480) Reurn (-), e (0.7890) (0.4550) (0.3) Reurn (-3), e (0.969) (0.8634) (0.060)*** Reurn (-4), e (0.0068)* (0.6) (0.734) f (0.095)*** (0.6305) (0.986)

13 Table 3. Esimaed EGARCH - Mean model (coninued) Variance Equaion Parameer Taiwan Singapore Hong Kong g (0.0000)* -0.9 (0.0000)* 0.8 (0.0000)* γ (0.0000)* (0.0000)* (0.0000)* γ (0.0000)* (0.983) γ (0.0000)* (0.0000)* 3 γ (0.0000)* -- 4 γ β.378 (0.0000)* (0.0000)* (0.0000)* β (0.0000)* (0.0000)* (0.0000)* β (0.0000)* 0.97 (0.0000)* (0.0000)* 3 β (0.0000)* β ψ 0.07 (0.0000)* (0.0000)* (0.0000)* ψ (0.0000)* (0.80) ψ (0.0000)* 0.4 (0.0000)* 3 ψ (0.0000)* -- Tuesday, h (0.0000)* (0.4350) -- Wednesday, h (0.0000)* Thursday, h (0.086)*** (0.0000)* -- Friday, h (0.0000)* (0.0000)* (0.0000)* ARCH-LM Saisic (p-value) a 5 lags lags Ljung-Box Q Saisic (p-value) b 5 lags lags Wald Tes (p-value) c F-saisic Chi square Noes: SIC is used o selec he bes fi GARCH (p, q) models, ou of he various combinaion of p and q, which range from o 5 in boh cases, see Lucey (000). *, ** and *** denoe significan a, 5 and 0% level respecively. Numbers in parenheses depic p-values. a Tes for remaining ARCH effec. b Tes for serial correlaion. c Tes he significance of he variance equaion.