Empirical Analysis of Liquidity-adjusted Capital Asset Pricing Model in the Banking Sector of Dhaka Stock Exchange

Transcription

1 IOSR Journal of Economics and Finance (IOSR-JEF) e-issn: , p-issn: Volume 8, Issue 5 Ver. III (Sep.- Oc.2017), PP Empirical Analysis of Liquidiy-adjused Capial Asse Pricing Model in he Banking Secor of Dhaka Sock Exchange Mony Adiya Lecurer, Deparmen of he Finance and Banking, Jaiya Kabi Kazi Nazrul Islam Universiy, Trishal,Mymensing,Bangladesh. adiya@jkkniu.edu.bd Absrac: This sudy endeavors o provide an empirical evidence of comparaively new model of asse pricing named Liquidiy-adjused Capial Asse Pricing Model. Tesing five years daa from 2011 o 2015 of he banking secor of Dhaka Sock Exchange his sudy reveals ha he explaining variables of LCAPM has failed o explain he reurn of he banking secor hus i has been concluded ha his new model is no good enough o analyze he pricing of asses in DSE. Finally, considering some limiaions of his analysis i is expeced o be useful for fuure researchers as i is supposed o provide new evidence regarding he asse pricing ools of DSE. Keywords: BLUE, LCAPM, liquidiy, marke reurn, rading volume Dae of Submission: Dae of accepance: I. Objecives Of The Sudy And Research Goal This sudy provides he empirical evidence from Bangladesh sock marke by examining how well he Liquidiy-adjused Capial Asse Pricing Model (LCAPM) work o quanify he risk and reurn relaionship in he banking secor of Dhaka Sock Exchange (DSE). LCAPM as proposed by Acharya and Pedersen (2005) [1] as well as by Liu (2006) [2] represen a new research direcion of asse pricing. Exising sudies show ha well known models of asse pricing canno explain expeced reurn whils LCAPM provides saisfacory explanaion for expeced reurn (Acharya and Pedersen, 2005 [1]; Liu, 2006 [2]; Tam, 2007 [3]). In his backdrop, his sudy is expeced o provide some addiional insighs ino he risk-reurn relaionship for he seleced DSE shares. So far as I know his is he firs aemp in Bangladesh o analyze he model of LCAPM in order o idenify he effeciveness of his asse pricing model for analyzing DSE insrumens. Specific research goal of he sudy is as follows: 1. Wheher he model of LCAPM is successful for analyzing he banking secor of Bangladesh sock marke? II. Lieraure Review And Needs Assessmen Of Presen Sudy The basic inuiion of his sudy is o provide empirical evidence o LCAPM proposiion base on he DSE daa of he banking secor. Tam (2007) [3] provides a review and comparison of hree models for asse pricing including classic CAPM, Fama-French (FF) Three-Facor Model and Liquidiy-adjused CAPM using UK sock marke daa. His research shows ha exising models such as CAPM and FF model canno effecively explain he abnormal reurns in he modern marke where newly proposed model LCAMP by Acharya and Pedersen (2005) [1]; Liu (2006) [2] aemp o challenge he old ones claiming ha i can provide beer explanaion. The evidence on he applicabiliy of CAPM and FF in DSE is no conclusive (see, for example, Rahman e al., 2006 [4]; Mollik and Bepar 2011 [5]; Azam and Ilyas, 2011 [6]; Hasan e al [7]; Chowdhury and Sharmin, 2013 [8], Alam e al [9]; Sayeed e al [10]). Therefore, his research examines he emerging role of LCAPM in he conex of he applicabiliy of new asse pricing model in DSE considering he banking secor. For he economic developmen of a counry sock marke plays an imporan role. While invesmen being a vial variable for developmen, sock marke s conribuion as a prime source of financing canno be ignored. This sudy has paricular relevance wih he economic developmen of he naion because he governmen effors o develop he marke. The sock marke crashes of 1996 and 2011 in Bangladesh indicae ha here exis much indiscreions regarding asse pricing of sock marke. Presen scenario of he sock marke is also no saisfacory. The paricipans of he sock marke, e.g. general invesors, insiuional invesors and fund managers may ge benefied using he findings of he sudy. Mos of he paricipans are seemed o vacillae beween hope and fear regarding invesmen for lack of knowledge. Usually he invesors from developing counries like Bangladesh manage heir porfolios and evaluae heir asses using single facor model (CAPM). Empirical evidence from his sudy is expeced o help invesors o choose he bes asse pricing model in making invesmen decision. Thus, i will be of grea advanage o he naion by DOI: / Page

2 Empirical Analysis of Liquidiy-adjused Capial Asse Pricing Model in he Banking Secor of.. conribuing o knowledge abou asse pricing model and by providing more accurae soluions o help reduce he invesmen risk levels in DSE. II. Mehodology In order o achieve he saed objecive, he sudy adoped a mehodology of wo-sage regressions which is suggesed by Fama and MacBeh (1973) [11]. The firs sage is a series of ime series regression for all sample shares o esimae he bea(s) of each facor. The process of daa rearrangemen, calculaion and analysis has been done using Excel spreadshee and STATA. The firs sage regression equaion for LCAPM is as follows: R R R R Liq i 1,2,..., n, 1,2,..., 60 (1 ) f, i m, f, i h Where subscrip indicaes in monh and i indicaes differen socks of banks; R i, is he monhly reurn of sock i; R f, is he risk free rae; R m, is he marke monhly reurn; is he coefficien o be esimaed by regression; Liq is he facor ha capures sock s liquidiy risk; i is he esimaed coefficien; is he residual erm and is he consan. Here, monhly share reurn of each Bank (R ) has been calculaed according o he following equaion: R Pi, P P 1 1 Where, R = reurn on share i on h monh P = price of share i on h monh P -1 = price of share i on day h -1 monh (2) R m, The marke reurn (R m, ) is calculaed by he following equaion: Index Index 1 Index 1 Where, Index is he DSE marke Index a he end of h monh (3) Anoher variable Liq is he variable ha measures liquidiy risk of holding a definie sock s and i is he esimaed coefficien. However, he LCAPM used in he proposed sudy is a sligh deviaion from he Acharyarya and Pedersen s (2005) [1], Liu s (2006) [2], and Tam s (2007) [3] models in he way ha he liquidiy facor is measured in a relaively less sophisicaed way due o difficuly in finding daa. For convenien sake liq in his paper is consruced as proxies for liquidiy risk by following he share urnover of each bank. oal monhly rading volumes liq average number of shares ous an ding To cerain exen, rading volumes can reflec liquidiy risk. Those socks wih less rading volumes imply ha hey are relaively difficul o be sold and hus indicae high liquidiy risk. In conrary, hose socks wih more rading volumes imply ha hey relaively easy o be sold and hus indicae low liquidiy risk. However, he absolue value of rading volumes needs o be adjused by number of ousanding shares o sandardize he measure. The adjusmen is necessary since low absolue rading volumes can sill has low liquidiy risk if he ousanding shares are also few. The second sage is he following cross secional regression on he beas and variance of residuals across all he banks enlised in DSE from he previous regression: Avg( R i ) 0 1i 2i 3 var( i ) i (5) The hypoheses for esing LCAMP esablished following Tam (2007) as, (4) H1: γ 0 =R f H2: γ 1 >0 H3: γ 2 0 DOI: / Page

3 H4: γ 3 =0 Empirical Analysis of Liquidiy-adjused Capial Asse Pricing Model in he Banking Secor of.. LCAPM is valid if four condiions are me. Firs, he consan erm equals risk free rae. Second, he coefficien of bea relaing o marke facor is greaer han zero. Third, he coefficien of bea relaing o liquidiy facor is differen from zero. Fourh, he coefficien of variance of residual equals zero. The robusness of he model has been deermined by he coefficiens, significance level and overall model fi. This mehodology examines he significance of he facors suggesed by his model. Furhermore, in he sudy hese following sandards in he Table 1 for a BLUE (Bes Linear Unbiased Esimaion) regression have also been considered as Tam (2007) [3] followed. III. Daa Collecion The daa used in his has been colleced mainly from DSE library excep he Treasury bill rae which colleced from he Bangladesh Bank. Sample selecion is a ough job in developing sock marke like Bangladesh where lo of anomalies and irregulariies exiss. However, he banking secor of DSE is chosen in his sudy because i is he main and counry s oldes sock exchange of Bangladesh where of banking secor has a disincive conribuion in he marke capializaion. Monhly sock prices of all banks lised in DSE for he period 2011 o 2015 has been included in he analysis. This sudy has seleced 60 monhs as an esimaion period because many sudies (see, for example, Fama and French, 1996 [12]) use an esimaion period of 60 monhs when employing monhly reurns. Therefore, here are in oal 60 (12 monhs x 5 years) ime series observaions. Monhly daa used in his analysis because he daily daa, hough beer for esimaing risk-reurn relaionship, is very noisy (Basu and Chawla, 2010 [13]). The All Share Price Index (DSI) and DSE Board Index (DSEX) are used as a proxy for he marke porfolio. DSI and DSEX are a marke value weighed index which is comprised of all lised companies of he exchange and reflecs general rends of he Bangladesh sock marke. Furhermore, cu off yield of Bangladesh governmen 91 days T-bill has been used as he proxy for he risk-free asse. IV. Daa Analysis And Resuls Inerpreaion Following he firs sage regression, base on he ime series daa of respecive banks, he coefficiens of each explaining variable has been esimaed (Appendix I). For beer analysis of resul, he firs sage regression model is recalled as below. R R R R Liq i 1,2,..., n, 1,2,..., 60 (1 ) f, i m, f, i Main saisics of firs sage regression has been presened in Table 2. From his analysis i can be said ha he marke reurn is a significan variable o explain he individual reurn of banks because he coefficiens of marke reurn β i has a p-value of less han 0.05 which indicaes ha marke reurn significanly posiively influence he share prices of banks lised in DSE. The coefficien of liquidiy λ i has a p-value of which indicaes ha he variable liquidiy is no significan o explain he reurn of banking secor. The value of R- square of his model is which indicaes ha here are many oher deerminans wihou marke reurn and liquidiy hose can explain he reurn of banks and his model has failed o capure hese deerminans. Table 2: Main saisics of 1 s sage regression of LCAPM Two chars in Fig.1 and Fig. 2 given below shows he disribuion of coefficiens of marke reurn (β i ) and liquidiy (λ i ) for series of he banks respecively. DOI: / Page

4 -2 b_liq b_rm_rf Empirical Analysis of Liquidiy-adjused Capial Asse Pricing Model in he Banking Secor of si Figure 1: disribuion of β i From Fig. 1 i can be said ha mos of he values of coefficien of marke reurn are posiive excep one value. From Fig. 2 i can be said ha mos of he values of coefficien of liquidiy are also posiive bu in his case nine values are negaive. Negaive values means ha hese values are negaively influencing he reurn of individual share si Figure 2: disribuion of λ i Deecion of regression error is very imporan o see he robusness of he model. According o he sandards of Table 1, heeroskedasiciy and auocorrelaion ess has been done for firs sage regression and he resuls of hese ess has been expressed in Table 3. From he Table 3 i can be concluded ha here is no problem of heeroskedasiciy as well as auocorrelaion. Therefore he resul of firs sage regression is accepable. Table 3: Heeroskedasiciy and Auocorrelaion ess resul of 1 s sage regression DOI: / Page

5 Empirical Analysis of Liquidiy-adjused Capial Asse Pricing Model in he Banking Secor of.. In he Table 4 he summery of second sage regression oupu has been shown. The second sage is simply a cross secional regression of he coefficiens esimaed in he firs sage. The second sage regression model is recalled as below. Avg( R i ) 0 1i 2i 3 var( i ) i (5) Table 4 shows ha R square of he model is which means ha explaining variables of his model can explain abou 39 percen of reurn of banks on average. Alhough he explaining variable of λ i is posiive bu i has a p-value of which is greaer han 0.05 on he conrary, he p-value of negaive β i is which is less han This express ha he explaining variable β i negaively bu significanly influencing he average reurn where he posiive influence of explaining variable of λ i is no significan. The coefficien of Variance of error erms and he consan erm are no also significan in he model since is coefficiens are negaive wih high p-value of and respecively. Table 4: Resul of 2 nd sage regression of LCAPM In he Table 5 he resul of hypoheses es of LCAPM model has been shown. The coefficien of he consan erm in he regression model is no equal o risk free rae. The firs hypohesis herefore should be rejeced. The coefficien of β i is negaive raher being posiive hus he second hypohesis is also rejeced. The coefficien of λ i is greaer han zero herefore he hird hypohesis is no rejeced. Finally he coefficien of Variance of error erms is less han zero and i is rejeced. Table 5: Resul of Hypoheses The VIF able and correlaion marix shown in he Table 6 and Table 7 confirm ha he es is mulicollineariy free. Table 6: VIF able of 2 nd sage regression Table 7: Correlaion marix of 2 nd sage regression From he Table 8 i can be concluded ha here is no problem of heeroskedasiciy since he saisics in he Breusch-Pagan es is greaer han is sandard value of Furhermore, since he series of daa of second sage regression model is no longer in ime series hence he es of auocorrelaion is avoided. DOI: / Page

6 Empirical Analysis of Liquidiy-adjused Capial Asse Pricing Model in he Banking Secor of.. Table 8: Resul of Breusch-Pagan es of 2 nd sage regression Following he summary of he es resul i can be seen ha LCAPM passes only 1 ou of 4 hypoheses which highligh poor robusness of he model. V. Conclusion There are many debaes regarding he robusness of differen models for asse pricing. The argumens are difficul o jusify because resuls of empirical ess may vary from differen imes, markes and mehodologies. As LCAMP has beer explanaory power (Acharya and Pedersen, 2005 [1]; Liu, 2006 [2]; Tam, 2007 [3]), i is expeced ha his sudy will provide addiional insigh ino he issue. Because his sudy conribues o lieraure by providing up-o-dae empirical evidence on LCAPM considering he daa of banks lised in he sock marke of Bangladesh. In his sudy LCAPM is rejeced because he coefficiens of explaining variables of LCAPM model hardly can explain he reurn of he banks and his model has failed o suppor mos of he hypoheses. Bu, he developmen of LCAPM is sill in he beginning sage. The fuure of LCAPM can be brigh when more and more empirical ess are presened o suppor i (Tam, 2007 [3]). Furhermore, given he relaively small sample size, only banking secor of DSE, he resuls should be inerpreed wih cauion in case of geing idea regarding he whole sock marke of DSE even regarding Bangladesh as well because in his sudy he daa of anoher sock marke of Bangladesh CSE (Chiagong Sock Exchange) has no been considered. Fuure research herefore migh be required o analyze a broader sample size in order o provide more comprehensive evidence. References [1] V. V. Acharya, and L. H. Pedersen, Asse pricng wih liquidiy risk, Journal of Financial Economics, 77, 2005, [2] W. Liu, A liquidiy-augmened capial asse pricing model, Journal of Financial Economics, 82(3), 2006, [3] K. T. Tam, Review and comparison of he models for asse pricing wih empirical evidence from UK sock marke, MA Finance and Invesmen Disseraion ( ), The Universiy of Noingham, Noinghamshire, England, UK, [4] M. Rahman, A. Baen, and A. U. Alam, An empirical esing of capial asse pricing model in Bangladesh, Journal of Applied Sciences, 6(3), 2006, [5] A. T. Mollik, and M. K. Bepar Risk-reurn radeoff in Dhaka Sock Exchange, Bangladesh: An emerging marke evidence, Ausralasian Accouning, Business and Finance Journal, 9(1), 2011, [6] M. Azam and J. Ilyas, An empirical comparison of CAPM and Fama-French model: A case sudy of KSE, Inerdisciplinary Journal of Conemporary Research in Business, 2(12), 2011, [7] M. Z. Hasan, A. A. Kamil, A. Musafa, and M. A. Baen, A validiy es of Capial Asse Pricing Model for Dhaka Sock Exchange, Journal of Applied Sciences, 11, 2011, [8] S. S. H. Chowdhury, and R. Sharmin, Relevan facors o explain cross-secion of expeced reurns of he firms lised in he Dhaka Sock Exchange, Inernaional Business Research, 6(3), 2013, [9] M. M. Alam, K. A. Alam and M.G.S. Uddin, Marke deph and risk reurn analysis of Dhaka Sock Exchange: An empirical es of marke efficiency, ASA Universiy Review, 1(1), 2007, [10] M. A. Sayeed, M. Khaun, and B. Chowdhury, Does Fama-French hree facor model ouweigh he CAPM model? Evidence from he Dhaka Sock Exchange, Rerieved May 18, 2016, from: hps://papers.ssrn.com/sol3/papers.cfm?absrac_id= [11] E. F. Fama, and J. D. MacBeh, Risk, reurn, and equilibrium: Empirical ess, The Journal of Poliical Economy, 81(3), 1973, [12] E. F. Fama, and K. R. French, Mulifacor explanaions of asse pricing anomalies, Journal of Finance, 51(1), 1996, [13] D. Basu, and D. Chawla, An empirical es of CAPM-he case of Indian sock marke, Global Business Review, 11(2), 2010, APPENDIX I: RESULT OF 1 ST STAGE REGRESSION Banks Bea p-value b_liq p-value Sd R DOI: / Page

7 Empirical Analysis of Liquidiy-adjused Capial Asse Pricing Model in he Banking Secor of APPENDIX II: BANKS LISTED IN DSE 1. AB Bank Limied 2. Ciy Bank Limied 3. Inernaional Finance Invesmen And Commerce Bank Limied 4. Islami Bank Limied 5. Naional Bank Limied 6. Pubali Bank Limied 7. Rupali Bank Limied 8. Unied Commercial Bank Limied 9. Uara Bank Limied 10. ICB Islamic Bank Limied 11. Easern Bank Limied 12. Al-Arafah Islami Bank Limied 13. Prime Bank Limied 14. Souheas Bank Limied 15. Dhaka Bank Limied 16. Naional Credi And Commerce Bank Limied 17. Social Invesmen Bank Limied 18. Duch-Bangla Bank Limied 19. Muual Trus Bank Limied 20. Sandard Bank Limied 21. One Bank Limied 22. Bank Asia Limied 23. Mercanile Bank Limied 24. Exim Bank Limied 25. Jamuna Bank Limied 26. Brac Bank Limied 27. Shahjalal Islami Bank Limied 28. Premier Bank Limied 29. Trus Bank Limied 30. Firs Securiy Islami Bank Limied Mony Adiya. Empirical Analysis of Liquidiy-Adjused Capial Asse Pricing Model in he Banking Secor of Dhaka Sock Exchange. IOSR Journal of Economics and Finance (IOSR- JEF), vol. 8, no. 5, 2017, pp DOI: / Page