Stock Selection Based on Fundamental Analysis Approach by Grey Relational Analysis: A Case of Turkey

Transcription

1 Internatonal Journal of Economcs and Fnance; Vol. 8, No. 7; 216 ISSN X E-ISSN Publshed by Canadan Center of Scence and Educaton Stock Selecton Based on Fundamental Analyss Approach by Grey Relatonal Analyss: A Case of Turkey Mehmet Fath Bayramoglu 1,2 & Coskun Hamzaceb 3 1 Department of Busness Admnstraton, Bulent Ecevt Unversty, Zonguldak, Turkey 2 Vstng Professor, Department of Economcs and Fnance, Lamar Unversty, TX, USA 3 Department of Industral Engneerng, Karadenz Techncal Unversty, Trabzon, Turkey Correspondence: Mehmet Fath Bayramoglu, Department of Busness Admnstraton, Bulent Ecevt Unversty, Zonguldak 671, Turkey. Vstng Professor, Department of Economcs and Fnance, Lamar Unversty, TX 7771, USA. Tel: E-mal: fath.bayramoglu@beun.edu.tr Receved: Aprl 11, 216 Accepted: May 11, 216 Onlne Publshed: June 25, 216 do:1.5539/jef.v8n7p178 URL: Abstract Determnng fnancal assets for nvestment s a sgnfcant problem for nvestors. Especally, snce t s rsker n comparson wth other fnancal assets, selectng common stock s not only mportant for nvestors but also, t s a complcated decson-makng process, because, although they agree to accept a rsk, common stock nvestors tend to keep the rsk they undertake at a gven level. Wth ths purpose, nvestors prefer to conduct varous analyss and predctons to decde whch common stocks they wll nvest. In ths study, by usng mcro and macro varables, whch have been determned wth a fundamental analyss approach, a common stock selecton s conducted for nne dfferent corporatons dong actvtes n ten lnes of busness n Borsa Istanbul. In the study, Grey Relatonal Analyss (GRA), whch s developed n the framework of Grey System Theory, s used as a mathematcal model. Therefore, a concluson s obtaned about whch parameters are more mportant n selectng a common stock, and the effcency of GRA Method s tested. Keywords: stock selecton, fundamental analyss, grey relatonal analyss, beta coeffcent, Borsa Istanbul 1. Introducton There are much multple-crtera decson makng (MCDM) n our daly lfe (Kou et al., 28, Huang et al., 24). Dfferent from sngle-crtera decson-makng problem, MCDM ams to select the best from the exstng alternatves, polces, actons, or canddates, by consderng multple attrbutes, goals, or crtera, whch are frequently n conflct wth each other. Therefore, how to make a trade-off between these conflctng attrbutes and then make a decson could pose a severe problem (Kuo et al., 28). In ther daly decsons, people mplctly consder the rsks they face. In an ncreasngly complex world, the resultng decsons are not always approprate because the lmts of the human mnd do not allow for an mplct consderaton of a vast number of dfferent factors. In the past, many lessons had to be learned by tral and error. In a smlar manner, a formal analyss may contrbute to the decson-makng process n complex stuatons (Bohnenblust & Slovc, 1998). Many decsons are made under uncertan and/or ncomplete nformaton. The analyss of such decson-makng s an nterestng topc, and many researchers have been engaged wth these decson-makng problems (Huang et al., 24). There are several common methodologes for MCDM are developed such as Analytc Herarchy Process (AHP), Analytc Network Process (ANP), Data Envelopment Analyss (DEA), Elmnaton et Chox Tradusant la Realte (ELECTRE), Preference Rankng Organzaton Method for Enrchment Evaluaton (PROMETHEE), Smple Addtve Weghtng (SAW) Technque for Order Preference by Smlarty to Ideal Soluton (TOPSIS), Grey Relatonal Analyss (GRA), etc. Stock nvestment s accepted as a decson-makng problem n the scence of fnance. The crtcal ssue for decson-makng n stock tradng s a selecton of the rght stock at the rght tme. There are many analytcal approaches for decson makng n a stock exchange, whch are categorzed n two groups of techncal analyss 178

2 Internatonal Journal of Economcs and Fnance Vol. 8, No. 7; 216 and fundamental analyss (Albadv et al., 27). Techncal analyss nvolves the use of hstorcal market data, such as prce, volume, and other observables, to predct future returns on fnancal assets (Yamamoto, 212). Fundamental analyss nvolves audt reports, fnancal statements, management capabltes and stock s rate of return, beta coeffcent and other return and rsk parameters (Albadv et al., 27, Dhatt et al., 1999, Mukherj et al., 1997, Markowtz, 1952). In ths study, the purpose s to determne macroeconomc and ndvdual mcro varables, whch can contrbute to the decson-makng process n a stock selecton applcaton wthn the framework of Grey Relatonal Analyss Method and basc analyss approach. The reason why such a base has been chosen for the study s that the stock selecton process n a portfolo nvestment plays an mportant role as much as a determnaton of stock nvestment rates to be made snce the step of determnaton of captal allocaton rates s the succeedng step after the stock selecton process. However, nether Tradtonal Portfolo Approaches nor Modern Portfolo Theory presents a proactve proposton for the ssue of stock selecton. Markowtz OV Model, as one of the leadng modern approaches, makes a stock selecton n the optmzaton phase. For the prevous phases, the only proposton of the Model s selectng stocks from dfferent lnes of busness. However, t s also possble to select best stocks among the set of assets thanks to a stock analyss to be conducted wth mathematcal models before the optmzaton phase. In ths way, a more effcent optmzaton process can be performed. As such, the prmary purpose of ths study measures the effcences of Beta coeffcent, the rate of return, standard devaton, and coeffcent of varaton as the varables of fundamental analyss, whch are used n stock selecton, through GRA method. 2. Grey Relatonal Analyss (GRA) In grey system theory, accordng to the degree of nformaton, f the system nformaton s fully known, the system s called a whte system; f the nformaton s unknown, t s referred to as a black system. A system wth nformaton known partally s known as a grey system. The grey system theory ncludes fve major parts: () grey predcton, () grey relatonal analyss (GRA), () grey decson, (v) grey programmng and (v) grey control. GRA s part of Grey System Theory, whch s sutable for solvng problems wth complcated nterrelatonshps between multple factors and varables (We, 211). GRA s a tool of grey system theory for analyzng the relatonshp between a reference seres and other seres. Grey system theory was developed by Julong Deng n The GRA s a quanttatve analyss to explore the smlarty and dssmlarty among factors n developng a dynamc process (Deng, 1989). More clearly, GRA s a method used to determne the relatonal degree between each factor n a grey system and compared factor seres. Every factor s defned as a sequence. The degree of nter-factoral nfluence s called as a grey relatonal degree. In other words, GRA ams to measure the smlarty between the compared seres. The GRA methodology s as follows (Hamzaçeb & Pekkaya, 211): Step 1. Defne the problem: alternatves 1,...,m j 1,...,n,, crtera 1, 2, 3 n,..., (1) Step 2. Determne the reference seres: reference seres may come nto exstence va the mnmum or the maxmum f the crtera requre the maxmzaton (mnmzaton) the reference sequence value of the related crtera s the maxmum (mnmum) value of the alternatve seres values of the alternatve set or nomnal value. 1, 2, 3 n,..., (2) Step 3. Normalzaton: n order the make the values free of a unt the normalzaton process s done. Ths process s called grey relatonal generatng. The normalzaton process can occur n three types:. Hgher s better:. Lower s better:. Nomnal s better: k k mn k max mn k k max k max mn k k k k (3) (4) 179

3 Internatonal Journal of Economcs and Fnance Vol. 8, No. 7; 216 k k max k (5) where k s the value after the normalzaton, k s the value before the normalzaton, and k mn, th max k are the smallest and largest values of the k response before the normalzaton respectvely. Step 4. Calculate the grey relatonal coeffcent: Grey Relatonal Coeffcent (GRC) s an ndcator of the smlarty between the reference seres and alternatve seres. k, k mn k Step 5. Calculate the Grey Relatonal Grade (GRG): GRG s used for overall evaluaton of alternatves dependng on all crtera. If all crtera have equal mportance, the GRG can be calculated by Eq. (9), for dfferent mportance degree of the crtera, the GRG can be calculated by Eq. (1). The GRG values are used to rank the alternatves accordng to the smlarty to reference seres. The hgher GRG value ndcates the hgher smlarty. 1 n, k k (7), n k 1 n max max, w k k k (8), k1 It s evdent to see that the GRA method s beng used n varous decson-makng processes under the framework of the scence of fnance when one analyzes the current related lterature. The Grey Relatonal Analyss Method was employed by Fang-Mn and by Wang-Chng (21) and by Ho (26) n rato analyss, by Lu et al. (215) n analyss of house-purchase behavors, by Huang and Jane (29) n the determnaton of nvestment rates of stocks, by Chen et al. (214) n forecastng of REIT returns, by Hamzaçeb and Pekkaya (211) n decson makng on stock selecton, by Ln and Wu (211) and by Doğan (213) n measurng the performances of banks, by Kung and Wen (27) and by Lee et al. (212) n measurng company performances, by Wang et al. (215) n capturng of customer requrements, by Guo et al. (215) n forecastng of new product dffuson, by Brgun and Gungör (214) n selecton of call center stes, by Pan and Leu (216) n analyss of bank servce satsfacton, by Camela et al. (213) n analyss of fnancal sector n Europe, by Wang et al. (214) n measurng toursm companes n Tawan, by Kaygsz et al. (215) n evaluaton of banks commercal credt applcatons, by Sabau-Popa and Bolos (214) n nvestgatng the effects of macroeconomc varables on Bucharest Stock Exchange, and by Wang et al. (214) n nvestment decson makng process. 3. Data In ths study, a stock selecton process n the Borsa Istanbul s conducted as a decson-makng process. The study ncludes nne dfferent companes conductng actvtes n ten lnes of busness. These companes are ALCTL, ANELT, ARENA, ESCOM, INDES, NETAS (Technology Informaton Technology), RYSAS (Transportaton, Communcaton and Storage Transportaton Other Servces Related Transport), INTEM (Wholesale and Retal Trade Wholesale and Retal Trade, Hotels and Restaurants Lumber and Buldng Materal Wholesale Trade) and SELEC (Textle, Apparel, Furnture Wholesale Trade Wholesale and Retal Trade, Hotels and Restaurants Wholesale Trade). In the developed model, four dfferent ndependent varables as Beta coeffcent (BC), the rate of return (RoR), standard devaton (STD) and coeffcent of varaton (CV) are used. Independent varables are calculated separately for each of the stocks. The dependent varables are the rates of returns of stocks n the followng month. Why the dependent and ndependent varables have been determned by the rsk - return perspectve s the theory tself s based on rsk and ncome (Markowtz, 1952). As such, t s assumed that these parameters are mportant decson factors n stock selecton. Besdes, snce busness lne dversfcaton s another consttuent of the Modern Portfolo Theory, the assets, whch are used n ths emprcal study, are selected from dfferent ndustres. The study covers a perod of 13 months from December 211 to December 212 as the term of monthly analyss. (6) 18

4 Internatonal Journal of Economcs and Fnance Vol. 8, No. 7; 216 In the modelng phase, the tme seres are establshed wth an assumpton that the values of ndependent varables at the date of t 1 wll affect the value of the dependent varable at the date of t. Consequently, the values of the perod between December 211 and November 212 are used for ndependent varables. Provdng a more relable calculaton of these varables, the tme of the data s expanded between January 27 and November 212. When the values of the perod between January 212 and December 212 are used as the dependent varable. The data are obtaned from DrectFN TM. Table 1. Results of smulatons SIMULATION 1 SIMULATION 2 Months SCC for Modelng SCC for Test Months SCC for Modelng SCC for Test Weghts 1 Weghts SIMULATION 3 SIMULATION 4 Months SCC for Modelng SCC for Test Months SCC for Modelng SCC for Test Weghts Weghts SIMULATION 5 SIMULATION 6 Months SCC for Modelng SCC for Test Months SCC for Modelng SCC for Test Weghts.95.5 Weghts SIMULATION 7 SIMULATION 8 Months SCC for Modelng SCC for Test Months SCC for Modelng SCC for Test Weghts Weghts Results The data was dvded nto two parts as modelng and testng. The modelng part was used to determne best weght values of crtera (Beta coeffcent BC, rate of return RoR, standard devaton STD, coeffcent of varaton CoV ) and the testng part was used for to test these weghts. The results of smulatons as shown n 181

5 Internatonal Journal of Economcs and Fnance Vol. 8, No. 7; 216 Table 1. Spearman Correlaton Coeffcent (SCC) was used as performance tool whle measurng the ftness between the actual and predcted return orders. The Spearman correlaton coeffcents were gven the Table 1, whch presented above. These results showed the good ftness between the actual return order and predcted return order by GRA model. These results help to determne the best values of the crtera weghts. Accordng to these results most mportant crtera s the Beta coeffcent. When the results of the smulaton are analyzed, t s evdent to see that SSC test results obtaned wth the GRA Method postvely ncrease when weght value gven to the Beta coeffcent s ncreased. In other words, the farest results to the actual ones are obtaned by gvng weght to the Beta coeffcent. For nstance, when weght s gven to the Beta coeffcent by 1% (1.) n the Smulaton-1, the SSC between the actual results and test results ncreases and when weght s gven to the Beta coeffcent by 6% (.6) n the Smulaton-2, the SSC decreases. These results ndcate that although the ndependent varable, whch has the strongest correlaton wth the return of the stocks n the followng month, s the Beta coeffcent among the varables used n ths study, ths correlaton s not so much strong. When the results are evaluated, n general, t s presented that although there s an explanatory correlaton between these four varables and the return of the stocks n the followng month; however ths correlaton s nadequate to make a stock selecton. 5. Concluson Portfolo optmzaton operatons wth the rsk return perspectve may produce favorable results have been presented by Harry M. Markowtz ntally and other ndvdual scentsts. In ths study, the effcences of these varables n the stock selecton phase as the prevous stage before optmzaton are analyzed. However, the results obtaned ndcate that these varables are not suffcent determners n the stock selecton process as much as they are n the optmzaton phase. The results of the study are n parallel wth the emprcal studes carred by Mukherj et al. (1997) and Dhatt et al. (1999) on Korean stock market. Mukherj et al. (1997) and Dhatt et al. (1999) offer to use rates, whch can be calculated wth the data from the fnancal statements of the companes such as especally Market Value / Book Value ratos or Sales/Prce ratos and Debt/Equty ratos, n stock selecton nstead of varables as Beta coeffcent. In ths study, varables determned wth return rsk perspectve s used nstead of fnancal statement ratos. However, t s presented that there s no strong relaton between these varables and stocks. By the results of the study, even the Beta coeffcent as the varable that has the most substantal relatonshp s not adequate for decson makng. As such, n the followng studes, emprcal analyss s gong to be conducted to measure the effectveness of fnancal statement ratos n the stock selecton wth GRA method and determne whether the related lterature apples to the Borsa Istanbul. References Albadv, A., Chaharsoogh, S. K., & Esfahanpour, A. (27). Decson makng n stock tradng: An applcaton of PROMETHEE. European Journal of Operatonal Research, 177, Brgun, S., & Gungor, C. (214). A mult-crtera call center ste selecton by Herarchy Grey Relatonal Analyss. Journal of Aeronautcs and Space Technologes, 7(1), Bohnenblust, H., & Slovc, P. (1998). Integratng techncal analyss and publc values n rsk-based decson makng. Relablty Engneerng and System Safety, 59, Camela, D., Eml, S., & Lvu-Adran, C. (21). Grey Relatonal Analyss of the fnancal sector n Europe. The Journal of Grey System, 25(4), Chen, J. H., Cheng, T. T., Ho, C. R., & Daz, J. F. (214). Grey Relatonal Analyss and Neural Network Forecastng of REIT returns. Quanttatve Fnance, 14(11), Deng, J. (1989). Introducton to Grey System Theory. The Journal of Grey System, 1, Dhatt, M. S., Km, Y. H., & Mukherj, S. (1999). Relatons between stock returns and fundamental varables: Evdence from a segmented market. Asa-Pacfc Fnancal Markets, 6,

6 Internatonal Journal of Economcs and Fnance Vol. 8, No. 7; Doğan, M. (213). Measurng bank performance wth Gray Relatonal Analyss: The case of Turkey. Ege Academc Revew, 13(2), Fang-Mn, L., & Wang-Chng, C (21). A precauton dagnoss of fnancal dstress va Grey Stuaton Decson. The Journal of Grey System, 4, Guo, H., Xao, X., & Forrest, J. (215). The forecastng of new product dffuson by Grey Model. The Journal of Grey System, 27(2), Hamzaceb, C., & Pekkaya, M. (211). Determnng of stock nvestments wth Grey Relatonal Analyss. Expert Systems wth Applcatons, 38, Ho, C. T. (26). Measurng bank operatons performance: An approach based on Grey Relaton Analyss. The Journal of the Operatonal Research Socety, 57(4), Huang, K. Y., & Jane, C. J. (29). A hybrd model for stock market forecastng and portfolo selecton based on ARX, Grey System, and RR theores. Expert Systems wth Applcatons, 36, Huang, W., Goto, S., & Nakamura, M. (24). Decson-makng for stock tradng based on probablty by consderng whole market movement. European Journal of Operatonal Research, 157, Kaygısız Ertuğ, Z., & Grgner, N. (213). Evaluaton of banks commercal credt applcatons usng the Analytc Herarchy Process and Grey Relatonal Analyss: A comparson between publc and prvate banks. South Afrcan Journal of Economc and Management Scences, 18(3), Kung, C. Y., & Wen, K. L. (27). Applyng Grey Relatonal Analyss and Grey Decson-Makng to evaluate the relatonshp between company attrbutes and ts fnancal performance-a case study of venture captal enterprses n Tawan. Decson Support Systems, 43, Kuo, Y., Yang, T., & Huang, G. W. (28). The use of Grey Relatonal Analyss n solvng multple attrbute decson-makng problems. Computers & Industral Engneerng, 55, Lee, P. T. W., Lna, C. W., & Shn, S. H. (212). Comparatve study on fnancal postons of shppng companes n Tawan and Korea usng Entropy and Grey Relaton Analyss. Expert Systems wth Applcatons, 39, Ln, S. L., & Wu, S. J. (211). Is Grey Relatonal Analyss superor to the conventonal technques n predctng fnancal crss? Expert Systems wth Applcatons, 38, Lu, J., Lu, X., Lu, Y., & Lu, S. (215). A new decson process algorthm for MCDM problems wth nterval grey numbers va Decson Target Adjustment. The Journal of Grey System, 27(2), Markowtz, H. M. (1952). Portfolo selecton. The Journal of Fnance, 7(1), Mukherj, S., Dhatt, M. S., & Km, Y. H. (1997). A Fundamental Analyss of Korean stock returns. Fnancal Analysts Journal, 53(3), Pan, W. T., & Leu, P. (216). An analyss of bank servce satsfacton based on Quantle Regresson and Grey Relatonal Analyss. Mathematcal Problems n Engneerng, Sabau-Popa, D., Bolos, M., Scarlat, E., Delcea, C., & Bradea, I. (214). Effects of macroeconomc varables on stock prces of the Bucharest Stock Exchange (BSE). Academy of Economc Studes, 48(4), Wang, F., L, H., & Dong, M. (215). Capturng the key customer requrements for complex equpment desgn usng Grey Relatonal Analyss. The Journal of Grey System, 27(3), Wang, S. M., Hseh, C. H., & Se, P. R. (215). A grey relaton analyss of the performance of lsted hosptalty companes n Tawan. Internatonal Journal of Organzatonal Innovaton, 7(3), Wang, Y., Sh, X. Sun, J., & Qan, W. (214). A grey nterval relatonal degree-based dynamc multattrbute decson makng method and ts applcaton n nvestment decson makng. Mathematcal Problems n 183

7 Internatonal Journal of Economcs and Fnance Vol. 8, No. 7; 216 Engneerng, We, G. (211). Grey Relatonal Analyss model for dynamc hybrd multple attrbute decson makng. Knowledge-Based Systems, 55, Yamamoto, R. (212). Intraday Techncal Analyss of ndvdual stocks on the Tokyo Stock Exchange. Journal of Bankng & Fnance, 36(11), Copyrghts Copyrght for ths artcle s retaned by the author(s), wth frst publcaton rghts granted to the journal. Ths s an open-access artcle dstrbuted under the terms and condtons of the Creatve Commons Attrbuton Lcense ( 184