The Study and Test Dividend Policy Based on Models Ohlson: Case Iran Market

Transcription

1 The Sudy and Tes Dividend Policy ased on Models Ohlson: Case Iran Marke Jalil Rahimi Shiraz Municipaliy ICT Organizaion Absrac This sudy invesigaes he effecs of dividend policy on marke value. empirically he research daa is colleced from 65 firms in he Tehran Securiy Exchange (TSE) during muliple regressions echnique is used for examining he saed hypiheses. I is used for, hree Ohlson equiy valuaion models. The firs model is based on permanen earning and he second model is based on expeced fuure earnings and, finally, he hird model is based on curren earnings.the relaionship beween earnings and dividends has conen informaion for evaluaion of he companies. In order o examine he hypoheses, daa is colleced from he annual repors of he companies using official bulleins of he Tehran sock exchange, mainly, hrough Novin sofware, Tadbir Pardaz sofware, and sock sies such as & The resuls indicae ha based on he models, deb, permanen income and, he ne operaing cash flows, deb and invesmen,, invesmens and deb are associaed wih sock prices, significanly and posiively and all samples are seleced,.i also confirms ha he resuls of he comparison of hree models, anaomical feaures and he abiliy of he models o predic he marke value are more imporan han he oher wo models.this research provided a good chance o examine he effecs of dividend policy on company's marke value. Meanwhile, i ook ino consideraion wheher oher conrol mechanisms such as firm s size have any effecs on his relaionship. Keywords:dividend policy, he company's marke value, equiy valuaion models Ohlson. INTRODUCTION Mos researches on securiies valuaion focus on ne income and book value (composiion of hem as he main facors deermining he value (price), less dividend policy and is relaion o he sock marke. Ohlson (995) model of equiy valuaion was expanded based on he heoreical framework of he dividend Modigliani and Miller (96). He claimed ha when dividends are paid, book value of equiy decreases. However, no impac is repored on curren earnings. Therefore, he curren marke value of he company's dividend policy is no affeced. Consequenly, recenly, equiy valuaion models based on Ohlson(995) ordered o invesigae he broader issues relaed o assess ineres rae, risk and he relaionship beween earnings and sock price. In his sudy, hree models of equiy valuaion have been saed based on Ohlson (995). The firs-ordered model, based on permanen earning, he second model is based on earnings from presen discouned value in expeced fuure. Finally, he hird model is based on curren earning. The permanen earning hypohesis uses dividend adjusmens. The permanen earning hypohesis componens form he presen discouned value of expeced fuure earnings, are represened by he permanen earning, wihou considering he volaile componens of earnings and dividends ha are used o adjus quickly. Gordon dividend growh model is based on accouning daa. He (96) believed, sock dividends affec recommendaions o increase he value of observaional managemen ha companies mus have for a high rae of dividend paymen. And he relaionship beween sock prices and earnings componens of he permanen, componens of he discouned value of expeced fuure earning and he curren earnings componens which are now being invesigaed, in he presen sudy, o deermine he predicabiliy and Anaomical Model ha are empirically examined. Feaures of his model are as follows: Model :I considers dividend as a fracion of he permanen componen of earnings. Model : I considers dividend as a fracion of he presen discouned value of expeced fuure earnings. Model :I considers dividend as a proporion of he curren earnings.. LITERATURE REVIEW Gordon (959) argued ha dividend policy affecs he company's cos of capial and his suppors he view ha he higher raes of paid dividends, cover more cos of capial Modigliani and Miller(96)Texed analysis of capial o draw he policy o pay dividends. They argued ha as long as he company's cash flow disribuion is sable and here is no ax effec, choosing a policy of paying dividends on he sock marke value has no effec. In heir analysis, here was he increased profis from he sale of shares of new sock supply.( Ohlson,995), There was also he residual income valuaion model based on earnings, book value of asses and developmen righs o owners of capial. Ohlson's model under he framework of he heory is developed in relaion o dividends. He claimed dividends are paid, curren book value of equiy will be reduced, bu i does no affec curren earnings. The claims concerning ha his heory of dividends Modigliani and Miller(96) is consisen. arclay,(995) afer numerous sudies, hree facors will be discussed as deerminans of corporae dividend policy; () he amoun of invesmen opporuniies, () he effecs of a 4

2 updaes and changes in cash dividends () firm size. In heir model, he cash dividend yield on he dividend policy is inended as a subsiue variable. ar&callen (996)' dynamic daa, he model Ohlson discovered and found ha earnings, book value of asses, equiy and dividend, in a period of ime, he dynamic informaion in connecion wih he evaluaion of he equiy are imporan. Lee& Frankel (998) found ha, if he esimaed marke value of companies is Ohlson evaluaion model and he invesmen decisions are aken, a higher efficiency can be occupied for periods of one, wo or hree more years. Lo& Lys(000)sudied Ohlson lierary model, and showed ha he evaluaion model developed by Ohlson on he accouning iems have an impac on financial informaion and marke research. Oher research and heories was conduced abou he deerminans of dividend policy by( einer, 00) o es his model ha had presened in previous research. His research makes clear ha he dividend yield on Swiss companies are associaed wih a dividend yield profis each year.this is consisen wih he heory of smoohing. Meanwhile, invesmen opporuniies and firm size, he Swiss company has an inverse relaionship wih income disribuion. Yun-sheng & e al (005) were invesigaed hree model of equiy valuaion models Ohlson, he relaionship beween earnings, dividends and dividend informaion conen in he evaluaion of he company. Experimenal resuls of heir research developed for he hree models shows ha, he Model anaomical feaures and he abiliy are beer o predic sock price. Also, all hree models, book value of equiy, deb and invesmens price of sock are relaed o each oher for he oal sample. alachandrak & &e al (007) Reducing he impac of dividends on he wealh of shareholders and he sock price reacion o dividend cus and he end of he period looked Ausralian companies. They found ha he marke reacion o dividend cus is more negaive during periods of declining profis sock a he end of he year. They also showed he amoun of dividends reduces he risk of company's profiabiliy and changes in corporae profiabiliy. (Fairchild,008) A sock dividend signals he complex relaionship beween a sock dividend, moivaed managers and i shows he value of he company.the research resuls show ha he sock dividend a signal of curren earning provides, and affecs he abiliy o paricipae in a new invesmen projec.. He also received signals ino higher dividends on invesmen because of he condiion and rea hair, and higher qualiy signal is lower han he dividend shares. Harris and e al (008) divided he discouned dividend model (DDM) or he Gordon growh model began. They also esimaed he inrinsic value of equiy model ha is used o esimae he reurn on equiy where reviewed. They showed he DDM model beween "! muliple in a marke wih marke growh rae relaionship exiss. Priesley & e al (009), smoohing i ino dividends and i were invesigaed predicabiliy. The resuls showed ha, if he sock splis is in a smoohing company here is predicabiliy in a sample of he limis. Also i is possible o predic, dividend growh socks in companies ha profi sharing is a smooh coa in hem o ge rid of ha less dividend smoohing. ooh & e al (009), examined how he imperaive of marke power on dividend policy affecs he marke measured by means of hree crieria:. Fayndl and Hrchymn he index. The basic firms of impored jackes..learner indices were used. Resuls showed ha he marke posiively on he abiliy o pay dividends and dividend decisions also affec he amoun of dividends. They found he impac on marke srucure decisions, where dividend socks, is a commercial risk. asel Al-Najjar & e al (00), examined he relaionship beween ime deposis and corporae dividend policy in a broad sample of 400 companies for he period Their invesigaion shows ha deposi, he dividend, leverage, growh; size, risk, and profi poenial are influenced. I also showed ha when boh variables are divided ino bonds and dividends hey are simulaneously moniored by he impac of any crieria. (Mazna Ramli, 00) examined he impac of policies on disribuion of profis and ownership srucure, in companies in Malaysia in a ime period His research shows ha companies ha benefi more from he shareholders have spli he shares of large and has increased; and he imporance of he dividend o shareholders will be higher in he company of grea imporance in he disribuion of profis. Dimiar Rafailo & e al (0), examined he impac of he company's characerizes on dividend decisions. The resuls showed ha dividend policy depends on he life cycle of a conflic of ineres beween minoriy and majoriy shareholders.. RESEARCH HYPOTHESES in fac This hypohesis, is he reconnaissance of hose financial variables ha affec he company's marke value. Corporae dividend policy has been used o reflec he same variables. These variables are permanen componens. Dividend discoun model 4

3 of earnings (book value of equiy, deb, invesmens, financial asses and permanen earning), he presen discouned value of he componens expeced fuure earnings (book value of equiy, deb and invesmen, and ne operaing cash flow) and componens curren earnings (book value of equiy, invesmen and financial asses and liabiliies). Hypoheses of his sudy are as follows:.. The firs of hypoheses: There is a significan relaionship beween he company's marke value and companies which considers dividend as a fracion of he permanen componen of earning... The second of hypoheses: There is a significan relaionship beween he company's marke value and companies which considers dividend as a fracion of he presen discouned value of expeced fuure earnings. The hird of hypoheses: There is a significan relaionship beween he company's marke value and companies which consider dividend as he proporion of he curren earnings. 4. VARIALES DEFINITIONS 4.. Dependen variable: Marke value of Sock: Las value raded (marke) is he common sock a he end of he fiscal period. The value of saisics and informaion is available on scholarships. Such as & 4.. Independen variables: The independen variables invesigaed financial asses, permanen earnings, book value of equiy, invesmens, deb and ne operaing cash flow of he balance shee daa, income saemen and operaing cash flows of he companies lised in Tehran Sock Exchange during Daa and informaion are colleced from he annual repors of he official bulleins of he Tehran sock exchange, marke informaion, sock organizaion library, Novin sofware, Tadbir Pardaz sofware, and sock sies suchaswww.rdis.ir& Independen variables of his sudy are calculaed: Financial asses: a simple measure of ne of financial obligaions (FA ) is he firs operaional asses, he operaing asses (OA ), subracing he book value of equiy, cash and shor-erm securiies, plus shor-erm and long-erm deb and minoriy righs are calculaed, Moron & schafere, (000). from he book value of equiy, he enire ne operaing asses (OA ) and ne of financial obligaions (FA ) is he ne resul of financial obligaions (FA ), he difference can be achieved from beween he ne book value of operaing asses of he owners of capial Felham & Ohlson(995). FA is calculaed and defined as follows: #$%&=%&'%(=)*+,'%(################################################() -. =Operaing asses OL =Operaing liabiliies A= asses Cash= Shor-erm invesmens + cash /& 0 =$%&' ####################################################################45 FA = financial asses 6# 7 = ook value of equiy Permanen income: Income before iems by disconinued operaions minus he unexpeced and he annual adjusmens of he ax Moron & Hong (999). This value can be easily calculaed based on daa and informaion exchange. ook value of equiy: shares of he Company's offices, is based on hisorical cos a he end of he fiscal period. The balance shee value of companies is available lised on he Sock Exchange Audied Financial Saemens. Invesmen: Is addiional invesmen is include long- erm and shor-erm invesmen. Liabiliies : liabiliies include curren liabiliies, long-erm liabiliies, oher liabiliies and a reverse is calculaed. Operaing ne cash flows: received minus oal cash paymens of cash from operaions or ne asses during a period o be deermined Felham & Ohlson (995). If he value of corporae cash flows, presened in he Audied Financial Saemens is available in sock. These Variables are summarized in he able ().. financial asses. Operaing asses 4

4 (Table ) Descripion of he variables Names of he Variables Proxies Calculaions FA 7 marke value of Sock Las value raded (marke) is he common sock a he end of he fiscal period. financial asses Ne of operaing asses- ook value of equiy Income before iems by disconinued < 8 9:; permanen earning operaions minus he unexpeced and he annual adjusmens of he ax 6 9:; ook value of equiy shares of he Company's offices a he end of he fiscal period, is based on hisorical > 9:; Invesmen Addiional Invesmen? 9:; Liabiliies liabiliies( curren liabiliies+longerm liabiliies+oher liabiliies+ 9:; Operaing ne cash flow received - oal cash paymens of cash 5. METHODS OF DATA ANALYSIS In his sudy, he muliple regressions are used for daa analysis. Iniial daa was insered in Excel spreadshee and SPSS sofware was applied o analyze he daa saisically. Also Rahavard Novin sofware, Tadbir Pardaz sofware, sock organizaion library and sock sies such as & were used. 6. RESEARCH METHOD AND REGRESSION MODEL Considering ha he aim of his sudy was o examine he relaionship beween sock prices and earning's componens of he permanen, componens of he discouned value of expeced fuure earning and he curren earning's componens ha as a share dividend policy is o replace he variables, and aenion o hree groups of variables and assumpions abou he research menioned in he same way i was prior o he model, and he overall sudy group for each of he following research was decisive. Firs of all variables in he model sudy in a muliple regression model esed general form i is as: follows In he Model : pi= h + hc + h Y + h + h 0 i, i, i, 4 i, i, In he Model : L + e (4) ) pi= k FA Y 0 i, i, i, 4 i, i, 7. SAMPLE SELECTION The sample was chosen from he firms lised on he Tehran sock exchange (TSE), from 00 o 0, using he following crieria: ). Firms were lised in TSE during ). Daa was available for all he years under he sudy. ). The companies didn have changed he fiscal year for he period sudied. 4). anks, Insurance and Invesmen firms were no considered in his sudy. The daa used in he analysis were colleced from he annual repors of he official bulleins of he Tehran sock exchange. The final sample conains 65 firms. 8. DATA ANALYSIS Pearson Correlaion Coefficien and Mulivariae were used o analyze daa. H 0=normal H = Daa is abnormal L + e (5) 44

5 Table () One-sample Kolmogorov-Smirnov Tes DIV N 650 Normal parameers a.b Mean Sd. Deviaion Mos Exreme Differences Absolue.074 Posiive.074 Negaive Kolmogorov-Smirnov Z. Asymp.. (-ailed).06 a. Tes disribuion is normal. b. Calculaed from daa. Following he able (II), = 0.06>0.05. Thus resuls show ha daa is normal. 8.. Companies which considers dividend as a fracion of he permanen componen of earning. Tesing Resuls of he firs group hypohesis: Table () Variables Enered Model Variables Enered Adjused R Square Durbin-Wason Mehod * +, FA +, /. +, A oal opimum model was used o predic he sock prices considering dividend as a fracion of he permanen componen of earning. We enered variables ino he model respecively. models were defined and finally he las model () including variables was defined as an opimum model for predicing he performance. As a resul, he regression model came as he followings: Table (4) Excluded Variables model Variable,! ",! ea ln Parial Correlaion VIf As i is seen,#$,% significance level is equal o 0.5> 0.05, herefore, his variable was no enering he model. Also&' (,) significance level is equal o 0.5> 0.05, herefore, his variable was no enering he model. Presening oal opimum model based on model (T-es) Opimum model was model, which had a more deerminaion coefficien han he previous ones. In fac, when mos variables were beside each oher, hey could presen a more precise predicion of he sock prices and in he firs group hypohesis, he opimum model was. Table (5) of model Model4 Consan * +,- FA +,- /. +, Unsandardized Sl. Erro Sandardized ea The opimal regression model was wrien as he following: Pi = i- 0.4FA+ 0.05XP (7) VIF 45

6 According o he saisical resuls of he firs group hypohesis o es he research, he firs group, debs and permanen earning are posiive and have a significan impac on he company's marke value while financial asses impac is negaive and significan. So if he dividend is based on a proporion of permanen earning, i can be argued ha increasing he debs and permanen earning, he company's marke value will increase.. Meanwhile, based on Table () he resuls of group ess, sugges ha, hree independen variables of he sudy have a significan relaionship wih he company's marke value /000), which ogeher offer a 80% (AdjR = (80) Explains he behavior of he dependen variable. 8.. Companies which considers dividend as a fracion of he presen discouned value of expeced fuure earnings. Tesing Resuls of he second group hypohesis: Table (6) Variables Enered Model Variables Enered Adjused R Square Durbin-wason Mehod # +, , A oal opimum model was used o predic he sock prices while considering dividend as a fracion of he presen discouned value of expeced fuure earnings. We enered variables ino he model respecively. models were defined and finally he las model () including variables was defined as an opimum model o predic he performance. As a resul, he regression model came as he followings: Table (7) Excluded Variables model Variable ",! ea ln Parial Correlaion As i is seen,",! significance level is equal o 0.9> 0.05, herefore, his variable was no enering he model. Presening oal opimum model based on model (T-es) Opimum model was model, which had a more deerminaion coefficien han he previous ones. In fac, when mos variables were beside each oher, hey could presen a more precise predicion of he sock prices and in he second group hypohesis, he opimum model was model. Table (8) of model Model Consan # +,- * +,- pi= h * +,- + h C h 0 i, i, 4 i, i, Unsandardized Sl. Erro Sandardized ea , The opimal regression model was wrien as he following: + h Pi = Li+ 0.i Ci (9) According o he resuls of saisics he second group hypohesis o es he research, he second group, Liabiliies, Invesmen and operaing ne cash flow have a posiive and significan impac on he company's marke value. So if he dividend is based on a proporion of he presen discouned value of expeced fuure earnings, i can be argued ha increasing he Liabiliies, Invesmen and Operaing ne cash flow, he company's marke value will increase.. Meanwhile, based on able (6) he second group es suggessha,hree independen variables have a significan relaionship wih he company's marke value /000), which ogeher offer a 9% (AdjR =.98) Explains he behavior of he dependen variable. L + e (8) VIf VIf 46

7 8.. Companies which consider dividend as a fracion of he curren earning Tesing Resuls of he hird group hypohesis: Table (9) Variables Enered Model Variables Enered Adjused R Square Durbin-wason!," # $,% Mehod & $,% A oal opimum model was used o predic he sock prices while considers dividend as a fracion of he presen discouned value of expeced fuure earnings. We enered variables ino he model respecively. models were defined and finally he las model () including variables was defined as an opimum model o predic he performance. As a resul, he regression model came as he followings: Table (0) Excluded Variables model Variable FA $,% ea ln Parial Correlaion 0.80 As i is seen, FA $,% significance level is equal o 0.09> 0.05, herefore, his variable was no enering he model. Presening oal opimum model based on model (T-es) Opimum model was model, which had a more deerminaion coefficien han he previous ones. In fac, when mos variables were beside each oher, hey could presen a more precise predicion of he sock prices and in he hird group hypohesis, he opimum model was model. Table () of model Model Consan!," # $,% pi= k Y i, i, i, i, Unsandardized Sl. Erro Sandardized ea & $,% The opimal regression model was wrien as he following: Pi = Yi + 0.i Li() (0) VIf According o he saisical resuls of he second group hypohesis o es he research, he second group, Liabiliies and Invesmen have a posiive and significan impac on he company's marke value while ook value of equiy 's impac is negaive and significan. So if he dividend is based on a proporion of he presen discouned value of expeced fuure earnings, i can be argued ha increasing he Liabiliies and Invesmen, he company's marke value will increase.. Meanwhile, based on Table (9) he resuls of ess group l he hird suggess ha he,hree independen variables have a significan relaionship wih he company's marke value /000), which ogeher offer an 8% (AdjR =.8) Explains he behavior of he dependen variable. 9. Conclusion In his sudy, he effec of dividend policy has been assessed on company marke value of companies lised in Tehran Sock Exchange. In his regard, hree models was reviewed of equiy valuaion models Ohlson, he relaionship beween earnings, dividends and dividend informaion conen of he evaluaion. According o he resuls of saisical models o es he research, he firs model, he Liabiliies and permanen earning sanding in he marke value of he company were exremely posiive and significan, while financial asses were negaive and significan. Thus, while he proporion of dividend earning is permanen, i can be claimed ha he increase in Liabiliies and permanen earning would increase he company's marke value. In he second model, Liabiliies,he ne operaing cash flow and Invesmen on marke value of he company were posiive and significan. So when he dividend is based on a proporion of he presen discouned value of expeced fuure earning, i confirms his L + e VIF 47

8 wih he increase in Liabiliies,he ne operaing cash flow and Invesmen, and he company's marke value will increase. And finally, he hird model, he relaionship beween he significan and posiive marke value and Invesmen Company, and Liabiliies showed ha he company's book value of equiy is negaive and significan. Thus i can be concluded ha when he dividend is based on he proporion of curren earning, wih he increase in Liabiliies invesmens he company's marke value will increase. These resuls are consisen wih he resuls of he, Yun-sheng & e al (005) sudy, ha reviewed ". The relaionship beween dividend policy and equiy valuaion models" were no experimenal resuls of heir research. I says ha for he hree models developed, all hree models, book value of equiy, Liabiliies and invesmen were mainly posiive and he sock price for he oal sample are relaed. And i also confirms he resuls of comparing he hree models show ha, he anaomical feaures of he second model's abiliy o predic marke value are more han he oher models. Such a conclusion is also consisen wih he resuls of he Yun-sheng & e al (005) sudy ha reviewed "The relaionship beween dividend policy and equiy valuaion models, also found, anaomical feaures and he abiliy of he model o predic he marke value o be more han he oher wo models. References -alachandrak&vidanapahirana (Augs 007) The Impac of Dividend Reducion on Shareholders Wealh,a: -arclay, M. J. e., ( Winer 995). The Deerminans of Corporae Leverage and Dividend Policies, Journal af Applied Corporae Finance,,07-4 -asil Al-Najjar& Yacine elghiar(december, 00 ) Corporae Cash Holdings and Dividend Paymens: Evidence from Simulaneous Analysis Managerial and Decision Economics,,a: 4-einer Sefan ( June 00 ), Theories and Deerminans af Dividend Policy available, a : 5-ooh Laurence &zhou jun (February, 009) marke power and dividend policy,a: 6-Chen long & priesley Richard (june, 009) dividend soohing and predic abiliy,a: 7-Dimiar Rafailov&Mirena Trifonova(Sepember, 0) The Effec of Firm Characerisics on Dividend Decisions of ulgarian Companies, a: 8-Fairchild Richard (Augus, 008) dividend policy signaling and free cash flow, a: 9-Franke,R., and M.CLee,(998) Accouning valuaion, Marke Expecaion, and cross-secional Sock Reurns, Journal of Accouning and Economics, 5, Gordon.M.J.,( 959). Dividends,Earnings, and Sock pricos, Review of Economic sudies, 4, Lo,K.,and T.Lys, (000), The ohlson Model: conribuion o valuaion Theory, Limiaions, and Epirical Applicaion, Journal of Accouning, Audiing &Finance,5,7-67 -Nahasa Mazna Ramli(February 00) Ownership Srucure and Dividend Policy: Evidence from Malaysian Companies Inernaional Review of usiness Research Papers, 6, Ohson,J.A,( 995), Earning, ook value and Dividend in Equiy valuaion conemporary Accouning Research,, Rober s. Harris &Kenneh M. Eades(ocober008) The Dividend Dicoun Model,a: 5-Yi-Mienlih, Yun-sheng HSU, Wood Liao,(005), The Relaionship beween Dividend policy and Equiy Valuaion Mode a: 48

9 APPENDIX Firs group hypohesis: Table () Model Model Summary d Adjused Sd. Error of R R Square R Square he Esimae.764 a b c a. Predicors: (Consan), i b. Predicors: (Consan), i, FA c. Predicors: (Consan), i, FA, XP d. Dependen Variable: Pi Durbin-Wason Model a. Predicors: (Consan), i b. Predicors: (Consan), i, FA c. Predicors: (Consan), i, FA, XP d. Dependen Variable: Pi ANOVA d Sum of Squares df Mean Square F. 8.7E a 5.8E E E b 4.9E E E c.7e E Table () Prediced Value Sd. Prediced Value Sd. a. Dependen Variable: Pi s Saisics a Minimum Maximum Mean Sd. Dev iaion N

10 Second group hypohesis: Table( 4) Model Summary d Model Adjused Sd. Error of R R Square R Square he Esimae.884 a b Durbin-Wason.964 c a. Predicors: (Consan), Li b. Predicors: (Consan), Li, i c. Predicors: (Consan), Li, i, Ci d. Dependen Variable: Pi Table( 5) s Saisics a Minimum Maximum Mean Sd. Dev iaion N Prediced Value Sd. Prediced Value Sd a. Dependen Variable: Pi hrid group hypohesis: Table( 6) Model Summary d Model Adjused Sd. Error of R R Square R Square he Esimae.89 a b Durbin-Wason.908 c a. Predicors: (Consan), Yi b. Predicors: (Consan), Yi, i c. Predicors: (Consan), Yi, i, Li d. Dependen Variable: Pi 50

11 Table( 7) ANOVA d Model a. Predicors: (Consan), Yi b. Predicors: (Consan), Yi, i Sum of Squares df Mean Square F..04E+0.07E a 9.0E E E+0.096E b 7.76E E E c 5.0E E c. Predicors: (Consan), Yi, i, Li d. Dependen Variable: Pi Table( 8) Prediced Value Sd. Prediced Value Sd. a. Dependen Variable: Pi s Saisics a Minimum Maximum Mean Sd. Dev iaion N