Systematic risk in the capital asset pricing model for australia a clinical death

Anoverviewofassetpricingmodel

Sincethe1950s,assetpricinghasseizedgreatattentionfrompolicymakers,academicsandp ractitionerswhichpushesittothe forefrontoffinance OnthegroundoftheModernPortfolioTh eory(MPT),Markowitz(1952)presentedtheefficientfrontiertodemonstratethetrade- offbetweenreturnandriskofaninvestmentportfolio.Fewyearslater,buildingontheearlierwork ofMarkowitz(1952),theCapitalAssetPricingModel(CAPM)wasdevelopedbySharpe(1964)andLint ner(1965).

The Capital Asset Pricing Model (CAPM) has been widely accepted by academics and practitioners until the introduction of the Fama-French three-factor model in 1992, which has been extensively used to explain stock returns Empirical studies indicate that CAPM tends to underestimate returns for low-beta assets and overestimate returns for high-beta assets, leading to mixed evidence regarding its validity for estimating expected equity returns Despite these criticisms, CAPM remains popular, with 74% of 392 U.S Chief Financial Officers (CFOs) using it to evaluate the cost of equity capital, as noted by Graham and Harvey (2001) Similarly, Brounen, Jong, and Koedijk (2004) found that 43% of 313 European CFOs relied on CAPM for the same purpose Reports by McKenzie and Partington (2014) revealed that regulators in Australia, Germany, New Zealand, and the United Kingdom primarily use CAPM to estimate the cost of equity, while U.S regulators prefer the Dividend Discount Model (DDM) as their first choice Vo (2015) argued against the application of the Fama-French model in Australian public policy, based on his analysis of weekly stock and market return data from listed Australian firms.

9 t o 31M a y 2 0 1 4 fromBloombergandutilizedFama-MacBeth(1973)’stwo- stageregressiontechnique.Hesuggestedthreedifferents c e n a r i o s t o c l a s s i f y rawd a t a ass u b samplesandf i v e differentportfolioformationstoputstockin.Theportfolioformationisinitiat edfromthreefundamentalideas:

The study explores portfolio formation based on three criteria: equal stock distribution within groups, firm market capitalization, and selection of top stocks, such as the top 50 or top 200 For each scenario, five approaches to portfolio formation were adopted, utilizing Fama-MacBeth's (1973) two-stage regression technique to assess the risk premium of various risk factors The findings indicate that the value risk factor is effectively priced, whereas the size risk factor is not, particularly in the context of Australian firms Consequently, the application of the Fama-French three-factor model is deemed inappropriate, aligning with the results of Brailsford, Gaunt, and O'Brien (2012) and Faff (2004).

The central focus of the Capital Asset Pricing Model (CAPM) is its beta, which is a crucial measure of systematic risk Savor and Wilson (2014) demonstrated that beta is strongly and positively correlated with average excess returns on days when significant economic announcements, such as inflation, employment, or Federal Open Market Committee interest rate decisions, occur Their study reinforces the relevance of beta, indicating that CAPM remains applicable, at least in the US market While both CAPM and the Fama-French three-factor model are regarded equally, the latter has shown effectiveness in the US but has not performed well in the Australian context (Vo, 2015) This raises the question of whether beta is still relevant in Australia, a topic that has not been recently explored, prompting this study to address the gap.

Duetotheforegoingdedicatedresearch,probably,apatternisobservedtohaveemergedt h a t diff erentassetpricingmodelsaresuitabletodifferentcountries.Therefore,thisresearchraiseupahypot hesisthatwhetherthesinglefactor assetpricingmodel-CAPMisusableornoti n calculationof areturnon equityinAsia-Pacificingeneralor in Australiainparticular.

LintnerversionofCAPMinthecontexto f Australia,thisstudybasesonthepioneeringworkbySav orandWilson(2014)fortheUS.Thisresearchutilizes dailydataformorethan 2,200Australianlisted firmsarecollectedfromBloombergf o r t h e p e r i o d from1 J a n u a r y 2 0 0 7 t o 3 1 December2 0

1 6 Daysw i t h announcements(thea- day)inrelationtogrowth,inflation,employment,centralbankannouncements,bonds,housi ng,consumersurveys,businesssurveysandspeechesfromtheP r i m e MinisterortheGovernor oftheReserveBankofAustraliascheduledtobeannouncedareallocatedintothegroupwhichisseparat edfromthen-day(non- announcementdays)group.Moreover,variousportfoliosareconsideredinthisstudyincluding:

(i)10beta-sortedportfolios;(ii)10idiosyncraticrisk-sortedportfolios(iii)25Fama- Frenchsizeandbook-to- marketportfolios;and(iv)industryportfolios.Portfolio’sreturnisconsideredintwo dimensions:value-weightandequal- weightbaseddirection.Inrelationtomethodology,thelinearregressionwithpanel- correctedstandarderrorsmethodandFama-

MacBethregressionareb o t h employed.T h e structureo f t h e s i s i s representedasf o l l o w :Chapter2 i s aboutLiteratureReview.DataandmethodologyarediscussedintheChapter3.Chapter4considersempericalresults.Concludingremarksandpolicyimplicationsareput in theChapter5.

Researchquestions

Itisnotedthatdifferentassetpricingmodelshavebeenappliedtodifferentcountriesorregionsa ndempiricalfindingsaregenerallymixed.Itisgenerallyagreedthatestimatingreturno n equityisstil l apuzzleregardlessofanumber ofNobel prizesh a v e beenawarded.Thisobservationleadsto followingresearchquestion:

 Dothefindingsabovestillholdwhenmacroeconomiceventsareclassifiedintovariousgroups,i n c l u d i n g ( i) ma cr o event-relatedgroup;( i i ) microe ve nt - r el at ed group;

( i i i ) financialevent-relatedgroup;and(iv)economicevent-relatedgroup?

Researchobjectives

 Aconfirmationo f t h e v a l i d i t y / n o n - v a l i d i t y ofemployingt h e CapitalA s s e t PricingModel( C A P M ) i n A u s t r a l i a o nt h e groundo f i ts Betaf o l l o w i n g Savora n d Wilson(2014)approach.

 TherobustnessofempiricalfindingsinrelationtothevalidityoftheCAPMusingSavorandW i l s o n (2014)approachw h e n v a r i o u s portfolioformationsa n d detailedclassificationof macroeconomiceventsareconsidered.

AchoiceofAustraliain this study

ItisoptimalifthisstudyisconductedusingdatafromVietnam.However,apreliminaryanalysisi ndicatesthata substantiallylargevolumeofdataisrequiredforthistypeofstudy.Inaddition,o n e o f t h e k e y corner stoneso f t h i s empiricals t u d y i s t h e a v a i l a b i l i t y ofvariousannouncementsin relation tomacroeconomicissues suchaseconomicgrowth,moneysupply,unemploymentandtheothers.Unfortunately,thistypeofdata isnotpubliclyandsubstantiallyavailableinVietnam.

From30countriesincludingin theAsiaPacificregion,Australiais thebestcandidateatleasto n t h e f o l l o w i n g aspects:

( i ) a s u b s t a n t i a l l y largev o l u m e o f dataf o r l i s t e d firmsare available(more than2,200listedfirmsfor morethan20yearsofdata);(ii)announcementsofmacroeconomicissuesarepubliclyavailableandtheyaretransparentlyrecorded;(iii)Australiai s byallmeansasmall,open,andadvancedeconomyintheregion;and(iv)supportfro mtheaccessofdataisavailableand confirmed.Assuch,Australiaisselectedfor thepurposeofthisstudy.

Theoreticalliterature

ModernPortfolioTheory

Markowitz(1952)suggestedModernPortfolioTheory(MPT)whichisoneofthetwostandar dassetpricingtheoriesasanexplanationofinvestmentbehavior.TheMPTisconstructedfrom therule(so-calledexpectedreturns-varianceofreturnsruleorE-

Vrule)thatt h e investorprobablydoesconsiderexpectedreturnsadesirablethingandvarianceofr eturnsanundesirablethingandfocusesonthestageoforiginatingtherelevantbeliefsandendin gw i t h thechoiceofportfolio 2In hisinterestingnote,hestatedthatthisrulehassomeadvantagest o shedl ightonrisk- averseinvestor’sbehavior(i.e.minimizingvarianceofreturnsforgivenexpectedreturnsandmaxi mizingexpectedreturnsforgivenvarianceofreturns)andtoimplydiversification.

A rational investor diversifies their funds across various securities to maximize expected returns rather than concentrating all investments in a single security with the highest discounted value Returns are not static and are influenced by both specific and non-specific firm characteristics, meaning investors must accept a degree of risk Importantly, the portfolio that offers the highest expected return does not always come with the lowest risk There exists a trade-off between expected return and risk, where a risk-averse investor aims to minimize risk for a given expected return while maximizing expected return for a specific level of risk This relationship is visually represented by a curve that illustrates the optimal combination of risk and expected return.

V r u l e alsosuggestsa guidet o t h e rightk i n d o f diversification.T h e diversificationprocessi s n o t ass i m p l e asincreasingt h e n u m b e r ofsecuritiesintheportfolio.Onedirectionofdiv ersificationofasetofsixtydifferentrailwaysecuritiesi s differentfromt h e sames i z e o n e w i t h railroad,p u b l i c utility,mining,manufacturing,constructionandreale s t a t e , …

O n e p l a u s i b l e reasoni s thatsecuritiesi n t h e sameindustryprobablytendstomovetogethergreatert hanthoseindifferentindustry.Anotherexplanationi s t h a t fromequation(2),t h e biggercovariancei s , t h e largert h e varianceo f portfoliois.Thatis,thelatterisbetterthantheformer.Finally,Mark owitzconcludedthatarisk- averseinvestorprobablyfollowsthestrategyofminimizingriskforgivenexpectedreturn

CapitalAllocationLine

Tobin(1958),inhisattractivepaper,developedhisSeparationTheoremtoinvestigatet h e o perationofthecapitalmarket.WhileMarkowitzfocusedontheriskyassetsanddiversification,h e t o o k o n e s t e p backf o r a b r o a d e r v i e w Hisidea,throughSeparationTheorem,statedthatanin vestorallocateshiswealthnotonlyontheriskyassetsbutalsother i s k l e s s one.Itissaidthattheri skyassetfeaturesfortheequitymarketandrisklessassetsdoesf o r thebondmarketsohisfindingisproba blyoneoftheconnectionsbetweenthestockmarketandb o n d market.Fort h e graphicalr e l a t i o n s h i p , t h e CapitalA l l o c a t i o n Line(CAL)wasintroducedasanappropriatenominator.

Giventheexpectedreturn of the riskyassetsandriskless assets areE(Rp) and

Rfw h il et h e i r risksaremeasuredbystandarddeviationdenotedbyσpandσf,respectively.Itiscouldbeinf erredthattheσfi szerobecausebydefinition,therisklessassetsproducesacertainfuturereturnandt h e i r covariance– cov(p,f ) - i s z e r o , t o o 3S u p p o s et h a t ani n v e s t o r placesa proportionofhiswealth(α))inthe riskyas setsthe remainder(1– α))int h e risklessassets.Accordingto Markowitz, hisexpectedreturnandriskareyielded:

Extractingα)fromequation(4)andsubstitutingfor it inequation(3).Thatyields:

CapitalAssetPricingModel

Sharpe( 1 9 6 4 ) re-employedt h e hypothesisofE - V r u l e o f risk- aversei n v e s t o r andportfolio’sexpectedreturnanditsriskmanipulation (Markowitz,1952 )andTobin(1958)’sf i n d i n g ofwealthallocationofaninvestorintoriskyassetandrisklessonein ordertoexaminefurthert h e operationo f capitalmarketasi n v e s t o r s physicalinteract.H i s inter estingn o t e p r o b a b l y couldbedividedintothreesubsections:

(i)theoptimalinvestmentpolicy;(ii)theequilibriumof thecapitalmarketand(iii) thecapitalassets’price.

Vrule,hepointedoutthatarationalinvestor is likelytop i c k upefficientportfolios whichare X,B,A , θandY inthe f o l l o w i n g figure.Remarkably,allthosepointslieonthesameline– theinvestmentopportunitycurve 4

Moreover,i n c o m b i n a t i o n w i t h wealthallocationo f r i s k y assetandr i s k l e s s one,h e demonstratedthatthereisalinearrelationshipbetweenportfolio’sexpectedreturnandits riski n t h e σ R ,ERplane 5

Thisrelationi s graphicallyrepresentedbyPB,P A andP θ l i n e i n theforegoi ngfigure.Hearguedthatalthoughaninvestorhasthreeoptions,theonewhoseslopeislowestw o u l d b e c h o s e n 6A s such,P θ i s t h e answer.Intuitively,t h i s decisioncouldbee x p l a i n e d byth eE-

Vrule.Fromtheverticalaxis,drawahorizontallinewhichintersectsPB,P A andPθatC,DandF,resp ectively.TheC,DandFportfolioallofferthesameriskbuttheirexpectedreturnsarenotequal.ThatofFp ortfolioishigherthanthatofDandthatofDishigherthan that ofC Thus, Fischosenorinvestor’sportfolioisreflectedbyPθ.

Next,intermsoftheequilibriumofthecapitalmarket,Sharpe(1964)statedthatitisaconseq uenceo f i n v e s t o r s ’ o p t i m a l investmentp o l i cyandt h e i r physicalinteraction.Particu larly,toaninvestor,acombinationofstocksintheportfolioFislikelytobringattractiveexpectedreturnasc omparedtoportfolioC,D.Therefore,eachinvestorwantstoownthatsuchportfolioFandrejectportfolio C,D.Asamatteroffact,tostocksintheportfolioF,higher

10 demandatanygivencurrentpriceleadstolessstocks’expectedreturn.Duetolowerstocks’expected returns,e x p e c t e d returno f t h e F p o r t f o l i o i s lowert o o T h i s leadst o p o r t f o l i o F becomei nefficientoritspositionshiftstotheleftwhileitsriskdoesnotvary.Similarly,bythesamearguments,por tfolioC,Dbecomeefficientoritspositionshiftstherightwhileitsriskdoesnotvary.Consequent ly,themovementofthoseportfoliosmakestheinvestmento p p o r t u n i t y curveto beflatter.Thisresultis represented in thefollowing figure.

Finally,regardingtothecapitalassets’price,theresearcherproposedtheterminologyofsystemat icriskdenotedbyβwhichwhichisdefinedastheresponseofstock’sreturnwithrespecttoreturnofefficientp ortfolioinordertoexplainexpectedstockreturninequilibrium.Itisalsoinferredthatβwhich,byitself,is acomponentoftheasset’stotalrisk.Onthegroundoftheideaofs l o p e ofthetangentlineofinvest mentopportunitycurveatθequaltotheslopeofPθyieldst h e followingequation:

 βwhichig: Theresponseofstock’sreturnwithrespectto return ofefficientportfolioAccordingt o t h e a b o v e equation,givent h e p u r e interestrateande x p e c t e d returno f efficientportfoliog,lowβwhichassetisprobablytoassociatewithlowexpectedreturnandvicever sa.T h i s f i n d i n g i s a l s o consistentw i t h risk- returnt r a d e o f f regimereflectedthroughMarkowitz(1952)’sefficientE-Vcombinationcurve.

TheDownsideoftheCAPM

In 2002, Estrada criticized the Capital Asset Pricing Model (CAPM) for its reliance on the assumption that asset returns follow symmetry and normality, suggesting that this may not be appropriate for manipulating asset returns He introduced downside beta as an alternative measure of risk, defined as the ratio of co-semi-variance between asset returns and market returns to the market's semi-variance of return Unlike traditional beta, which considers all returns, downside beta focuses solely on returns below the sample mean, as investors are primarily concerned with negative return fluctuations Utilizing the Morgan Stanley Capital Indices database of emerging markets, Estrada conducted regression analyses on monthly returns from 27 countries, revealing that downside beta is the only significant variable in explaining asset returns Further analysis, which categorized the countries into three groups based on beta rankings, confirmed that downside beta consistently outperforms traditional beta in explaining asset returns.

Fama-French’sThreefactorModel

FamaandFrench(1992),intheirremarkablepaper, showedavoice that themarketòasemployedaloneprobablyhasnoabilitytodescribestockreturnofnonfinancialfirmsliste dont h e NYSE,AMEXandNASDAQintheperiodof1963-

1990.Thisstudyattractsacademics’andpractitioners’attentionatthattimebecauseitsfindingcontradi ctstraditionalwisdomaboutt h e roleofmarketò.Indeed,followingBhandari(1988)andBanz(198 1)’sstudy,theystatedt h a t theexistingnegativerelationshipbetweensize(ME- equalstostockpricestimesnumbero f share)andaveragestockreturn.Moreover,theyalsofoundthat arobustpositiverelationshipbetweenbook-to-marketequity(BE/

ME)andaveragestockreturn(Stattman1980;Rosenberg,ReidandLanstein1985;Chan,HamaoandL akonishok1991).Onthatbasic,in1993,FamaandFrencharguedthattherearethreevariablesmake stockreturndeviatearounditsmean:(i)Rm-

The excess return between the market portfolio and the risk-free rate is referred to as the market risk premium Additionally, the return from high book-to-market ratio portfolios minus low book-to-market ratio portfolios is known as the value premium (HML), while the difference in returns between small capitalization portfolios and large capitalization portfolios is termed the size premium (SMB) These factors act as risk components, capturing variations in stock returns In essence, expected stock returns can be explained by the market risk premium, the high minus low returns, and the small minus big returns Based on this framework, Fama and French proposed an alternative asset pricing model.

Frenchthree-factormodel orFF3F,forshort.Themodel isexpressedasfollow:

E(Ri)=Rf+(E(Rm)–Rf)βwhichmkt+E(SMB)βwhichsmb+E(HML)βwhichhml

 βwhichsmb: Thefactorloadingof stock on SMBfactor.

 βwhichhml: Thefactorloadingof stock on HMLfactor.

 E(HML): Theexpectedr e t u r n o f highbook-to-marketr a t i o portfoliol e s s l o w book-to-marketratioportfolio.

Cahart’sFourfactorModel

In their 1993 study, Jegadeesh and Titman examined firms with significant profits listed on the NYSE and AMEX from 1965 to 1989 to explain stock returns They proposed a strategy that involves buying stocks that have performed well and selling those that have underperformed over the past six months, which can lead to substantial returns in the following six months This strategy is based on delayed price reactions to firm-specific information rather than systematic risk or common risk factors Four years later, Carhart (1997) built upon their work by developing the Carhart four-factor model (C4F) to describe mutual fund equity returns.

E(Ri)=Rf+(E(Rm)–Rf)βwhichmkt+E(SMB)βwhichsmb+E(HML)βwhichhml+E(WML)βwhichwml

 E(WML):T h e differenceinexpectedreturnbetweendiversifiedwinnerportfolioandl o o s e r portfolio.O t h e r factorsa r e defineds i m i l a r l y i n t h e FF3Fm o d e l

Fama-French’sFivefactorModel

FamaandFrench(2015)basedonthedividend- discountmodel(DDM)toarguethats o m e otherfactorsareable to explaintheshareprice.With abit ofmanipulation,theyexpresst h e relationbetweenexpectedreturnandexpectedinvestment,b o o k t o marketratioandexpectedinvestmentas follows:

FamaandFrenchstatedt h a t t h e aboveequationproducesthreeimplicationaboutexpectedr eturns:(i)keepotherthingsconstant,ahigherbook-to- marketequityratiopertainst o higherexpectedreturn;

(ii)keepotherthingsconstant,ahigherexpectedearningsyieldsahigherexpectedreturnand(iii)k eepotherthingsconstant,ahigherexpectedgrowthinbooke q u i t y relatedtolowerexpectedret urn.Duetothoseimplications,theauthorsconcludedthatinvestmentandprofitabilityarelikelyable todescribeexpectedstockreturn.Theirfindingisalsoconsistentw i t h t h e previousfindingssuchas Novy-

Marx(2013);HaugenandBaker(1996);Fairfield,WhisenantandYohn(2003);Titman,WeiandXie(20 04).TheworkofFamaandFrench(2015)probablycouldbeexpressedbythe followingequations:

Rit–Rft =αα i +b i (Rmt–Rft)+s i SMBt+h iHMLt+r i RMWt+c i CMAt

 HMLt: Thereturno f highbook-to-marketr a t i o p o r t f o l i o l e s s l o w b o o k - t o - marketratioportfolio.

Empiricalliterature

Empirical studies on the Capital Asset Pricing Model (CAPM) have prominently featured the works of Fama and MacBeth (1973) and Jensen, Black, & Scholes (1972) Fama and MacBeth examined the risk-return trade-off on the New York Stock Exchange, confirming a linear positive relationship between risk and return, and validating CAPM’s assertion that only beta systematically influences returns Shapiro and Lakonishok (1984) further explored CAPM by segmenting market excess returns into upmarket and downmarket groups, finding that high beta stocks outperform in upmarkets and underperform in downmarkets, consistent with CAPM predictions Subsequently, Pettengill, Sundaram, and Mathur (1995) utilized a similar data classification method, reinforcing these findings within the U.S market and contributing to the development of the conditional capital asset pricing model.

SincetheworkbyShapiroandLakonishok(1984)gotpublished,researchareaonassetpricingh asw i t n e s s e d a significantn u m b e r o f s t u d i e s t o havebeenc o n d u c t e d A m o n g t h e studi es,Lam(2001)forHongKongStockExchangeperhapsis probablyaprominentstudyinA s i a Pacific region.Thisstudydemonstratedthatbetaandreturnarepositivecorrelatedin theup- marketperiodwhilethisrelationshipturnsouttobenegativeinthedown- marketperiod.Inaddition,TangaandShumb(2003)investigatedtheconditionalcapitalassetpricing modelbyemployingdataatglobalscalewhichincludesmanycountriesinthePacificOceanregion

7Similar toShapiroandLakonishok(1984),theresultsconfirmedthatbetaandreturnmovedint h e same directionasmarketreturne x c e s s t h e risk- freer a t e Ins h o r t , findingsfromt h o s e studiessuggestedthatbetawasausefulriskmeasure.No tonlyadvocateoftheCAPMwasdiscoveredintheStockExchangebutalsoamajorityofregulatorsin Australia,Germany,NewZealand,US A , C a n a d a andU K st il l e m p l o y ittomeasuret h e costo f e q u i t y (Sudarsanam,Kaltenbronn,&Park,2011) 8

Incontrast,v a r i o u s studieshavedemonstratedt h e i r concernso f t h e C A P M whicho riginated fromitssimplicityandimplication.AL-

QudahandLaham(2013),intheirstudyont h e determinantso f s t o c k returno f 4 8 industrialco mpaniesl i s t e d i n theA m m a n S t o c k ExchangefromJanuary2000toDecember2009,arguedt hatthestatisticallysignificantimpacto f systematicriskonstockreturncouldnotbeestablishedinthes tudy.Moreover,employingdatagatheredfromFAMEdatabaseovertheperiodofApril2000toJune20 07,RamloganandBhatnagarconsideredthattheFama-Frenchthree- factormodel(FF3F)wassuperiortoexplaint h e stockreturnsthan theCAPM inUnitedKingdomstockmarket.

7 ThislistcontainsJapan,Canada,US,HongKong,SingaporeandTaiwan.

Daniel, Titman, and Wei (2001) analyzed monthly data from listed firms on the Tokyo Stock Exchange, which represents approximately 85% of Japan's total market size from 1971 to 1997, to highlight the significance of characteristic models over factor models in the Japanese context They categorized firms into 25 portfolios based on size and book-to-market ratio to evaluate the validity of the Capital Asset Pricing Model (CAPM) and the Fama-French Three-Factor Model (FF3F) Their findings indicated that CAPM may not be suitable for Japan, while FF3F appears to be more applicable Specifically, the OLS estimation method revealed that the intercepts for the 25 portfolios were jointly statistically significant, suggesting that certain firms within smaller portfolios and those with high book-to-market ratios achieved substantially high CAPM risk-adjusted abnormal returns, indicating that stock excess returns could not be fully explained by systematic risk.

Choudhary and Choudhary (2010) conducted an empirical study on the validity of the Capital Asset Pricing Model (CAPM) from 1996 to 2009 in the Bombay Stock Exchange, concluding that stock excess returns were not adequately explained by systematic risk Their findings indicated that higher risk did not correlate with higher expected returns, and the intercept was statistically significantly different from zero, contradicting traditional CAPM assumptions that the market risk premium should be positive and the intercept should be zero This research aligns with the conclusions of Daniel, Titman, and Wei (2001) and Bajpai and Sharma (2015) Furthermore, similar results were observed in the Malaysian Stock Exchange from 2001 to 2013 (Mollik, 2014) and the Nigerian stock exchange, highlighting the limitations of the traditional risk-return tradeoff suggested by CAPM.

Visually,thetraditionalCAPMislikelytobeaneleganteffectivemodeltoexpressreturni n termso frisk.Howerer,theanti-

The Capital Asset Pricing Model (CAPM) was developed by Sharpe (1964) and Lintner (1965) and has been foundational in finance An early exploration of stock returns was conducted by Basu (1977), who analyzed firms listed on the NYSE from 1956 to 1971, concluding that low price-to-earnings (P/E) ratios are likely associated with high returns, suggesting that P/E should be a determinant of stock returns However, Roll (1977) highlighted the challenges in forming a market portfolio, as it requires the inclusion of every asset, and noted that the linear relationship between expected return and beta can only be verified if the true market portfolio is utilized This raises questions about the validity of the CAPM equation, prompting further critiques from researchers such as Shanken (1985), Banz (1981), Reinganum (1981), and Amihud.

Inshort,forageneralviewoftheinvalidityoftheCAPMintherealworld,Fernandez(2015) criticizedthatisanabsurbmodelduetothefactthatitsassumptionsareunrealistic.Forexample,investors arehomogenous.InvestorsareguidedbyE-

Vmaxim.Moreover,YangandChae(2008)werealsointhesameside.Theyarguedthattransactioncost, investorirrationalityandmissingriskfactorswhichwerenotmentionedintheliteratureimplythefailureo fCAPM.A s such,CAPM has beenmirroringthecomplexityofrealworldbysenselesssimplification.

Intheassetpricingresearchfield,thesolutiontotheassetreturnisathornyandattractivesubject.Inre sponsetotheappeal,amajorityofresearcheshavebeenconductedtofigureoutt h e cross- sectionalpredictors ofstock return.Subrahmanyam(2010)contributed 50variables,McLean&Pontiff(2015)discovered82factors,Green,HandandZhang(2013)offer ed330firmspecificc h a r a c t e r i s t i c s Remarkably,Harvey,LiuandZhu( 2 0 1 5 ) u t i l i z e d m o r e s t r i c t statisticalcriterionthannormaltofiltertrulysignificantfactors.10Generally,theysegmente dfactorsasfollows:

Banz(1981),Reinganum(1981)andFama&French(1992)focusonfirmsizeelementwhileAmihud&Mendels on(1986)doesonbid-askspread.

Proxy for aggregate finanial marketmovement, includingmarket portfolio returns,volalitity,spuaredmarketreturns, 46 amongothers

Proxyformovementinmacroeconomicfund amentals,including consumption,investment,inflation,among 40 others

Proxyforaggregatemovementsinmarketmicrostructure orfinancial market frictions, including liquidity, 11

Microstructur e Behavioral behavior,sentimentorbehavior-driven 3 systematicmispricing

Proxyforaggregatemovementinfirm- levelaccountingvariables, includingpayoutyield,cashflow,among 8 others

Proxyforaggregatemovementsthatdonot fallintotheabovecategories,including momentum, investors’ beliefs, among 5 others

Proxy forfirm-levelidiosyncraticfinancial risks,includingvolatility, 61

Extremereturns,amongothers Proxyforfirm-levelfinancialmarket frictions,includingshortsalerestrictions, 28 transactioncosts,amongothers

Other includingPEratio, 87 debt-to-equityratio,amongothers Proxyforfirm- levelvariablesthatd o notfallintotheabovecat egories, includingpoliticalcampaigncontributions, 24 ranking- relatedfirmintangibles,am ongothers

In the development of asset pricing, the seminal work of Fama and French is crucial, as it encapsulated two decades of research on asset pricing models Their studies linked size, book-to-market ratio, earnings-price ratio, and leverage to stock returns In 1992, they demonstrated that variations in stock returns could be explained by size and book-to-market ratio, reducing the focus from four to two key variables The following year, they introduced the SMB (small minus big) and HML (high minus low) portfolios, suggesting that these portfolios, along with beta, serve as proxies for risk, leading to the proposal of the Fama-French three-factor model (FF3F) However, the FF3F has faced criticism for its data mining-based formation rather than a theoretical foundation Similarly, the C4F model is not recommended due to its bottom-up approach Research indicates that while the FF3F may outperform the Capital Asset Pricing Model (CAPM) in explaining stock returns, it is still not widely adopted in practice.

O'Brien,BrailsfordandGaunt(2008) statedthat theirfindingsadvocatedthesuperiorityo f FF3F.Theresearchutilizeddataintheperiodof1982-

In 2006, the Australian Stock Exchange dataset, encompassing approximately 98% of all listed Australian firms, provided a significant advantage over previous studies limited by time span and market coverage The research utilized the Fama and French method to create 25 portfolios based on size and book-to-market ratios Stocks were categorized into five segments according to book-to-market, each containing an equal number of observations Additionally, stocks were divided into five size portfolios, with the first portfolio representing the largest 75% of market capitalization, the second the next 15%, and the remaining portfolios comprising the largest 5%, 3%, and 2%, respectively This methodology resulted in 25 smaller portfolios derived from the combination of the book-to-market and size portfolios However, the CAPM regression was only able to explain returns for six of these portfolios, as the intercepts in the remaining portfolios were statistically insignificant.

20 significantdifferentfromzero.Putitdifferently,thepricingerrorsignalexistsintheCAPMo r bet acoefficient,byitself,cannotexplainstockreturn.Incontrast,nearlyallfactorloadingso f marketfac tor,s i z e f a c t o r andv a l u e - g r o w t h factorw e r e s t a t i s t i c a l l y differentfromz e r o Moreover,t he statisticso f interceptse x h i b i t oppositeresutswhichimpliedt h a t t h o s e threeforegoingfactorscapture stockreturn.Inshort,FF3FoutperformedCAPM.

Similarly,theresearchofPhongandHoang(2012)forlistedfirmsonVietnam’sstockmark etfromJan2007toDec2011alsoproducedthesameconclusion.Usingdifferentportfolioformationascomp aredtothoseofO'Brien,BrailsfordandGaunt(2008),theauthorsdivideddatasetintosix portfolio sindependently sortedonsize andbook-to- market ratioincluding:B/H,B/M,B/L;S / H , S/ MandS / L ande m b e d e d FF3FandC A PM i n them.Followingthemagnitudeo f a d j u s t e d -

R 2and levelo f t h e numbero f significantestimates,t h e researchersp o s i t e d thatCAPMwasli kelytobecrowdedoutbyFF3F.Atthesametime,theresearchofWasiuzzamanandMonfared(201 2)inMalaysiawhichemployedsimilardataclassificationandcomparablebenchmarksalsogenerate da consistentresult.

Gharghori, Lee, and Veeraraghavan (2009) found that the Fama-French Three-Factor model (FF3F) provides a superior explanation of stock returns compared to the Capital Asset Pricing Model (CAPM) for various factors, including size, book-to-market, earnings-to-price, and cash flow-to-price effects In their study, they allocated stocks into sextiles based on key variables and conducted CAPM and FF3F regressions using Generalized Method of Moments (GMM) on six portfolios Their findings indicated that the FF3F model outperformed CAPM in terms of estimates, t-statistics, adjusted R², and Wald tests However, the FF3F model struggled to explain returns for certain cash flow-to-price differentiated portfolios, suggesting it may not be as effective in Australia as it has been in the US.

FocusingontheinvalidityofFF3Fonpractice,Vo(2015)positedthatFF3Fprobablywasa nseminaleffortforacademicpurposes but theadoptationof themodel intopracticewasproblematic;andassuch,t h e m o d e l i s recommenededi n t h e c o n t e x t o f Australia.T h e conclusionextractedfromthenumericalresultsoffactorloadi ngsandinterceptsintheFF3Fm o d e l werenotinlinewiththeexpectation.Particularly,fromtheda tasetcoveredstockreturno f Australianlistedfirmsinthefive-yearperiod2009-

20 u i l d u p t e s t e d portfoliosf o r theverificationofFF3F.Moreimportantly,priortotheportfoliofor mationprocess,observations

21 weref i l t e r e d byt h r e e d i s t i n g u i s h scenarios.T h a t i s , f o r eachscenario,f i v e differentda taclassifcationapproacheswered e p l o y e d l a t e r U n d e r eacht e s t p o r t f o l i o , h e u t i l i z e d Fama-

MacBethregressiontofigureouttheestimatedcoefficientsofmarketfactor,sizefactorandvaluefa ctor.Inaddition,theresearcheralsoarguedthatthoseestimatedcoefficientsmustbep o s i t i v e numberandtheinterceptswerenotstatisticallysignificantdifferentfromzeroinallportfoliost o reflectt h e naturetradeoffbetweenreturno n e q u i t y andf a c t o r However,t h i s prerequisiteswere notsatisfiedbecauseoftheHML’sfactorloadingexhibitedinverserelationi n nearlyalltestedportfol ios.Inrelationtotwootherfactors,market’sfactorloadingdemonstratedthesamephenomenoninthe twofirstscenarios,whereasnegativeSMB’sfactorl o a d i n g wasobservedinthirdscenario.Hence,itislikelyreasonabletoarguethattheadoptiono f FF3Fmayleadtoseriouscauses.Thefollowingtabl econtainsvariousportfolioformationapproachessummariedbyVo(2015).

1 thesamenumbero f stocks.Stocksb y book-to- marketratios(lowesttohighest)ar e rankedandquintileportfoliosofequalnumbersofstocksa reformed.

Eachfirm(largesttosmallest)bymarketcapitalisationisrankedandthenassignedtoo n e offi vesizeportfolios.Thelargestsizeportfoliocontainsthefirstnnumberofstocksthatmakeup75

%oftotalmarketcapitalisation.Thesecondportfoliocontainsthenextn numberofstocksthat makeupthenext15%oftotalmarketcapitalisation.Thenext

2 3portfolioscontainthenext5%;3%;a n d 2 % o f totalmarketcapitalisation.T h e s e marketca pitalisationbreakpointsarearguedtoparallelthefindingsofFamaandFrench(2006).Forvalue factor,portfoliosareconstructedusingbook- tomarketbreakpointsdeterminedonthebasisofsortsonthetop200stocksandsubsequentlya ppliedtothefullsampleofstocks.

Forasizefactor,eachstockisfirstrankedbymarketcapitalisation(largesttosmallest).T h e lar gestsizeportfoliocontainsthelargest50stocks.Thesecondsizeportfolio containsthenext150stocks(i.e.stocks51–200).Thethirdandfourthsizeportfolios

3 containsthenext100and200stocks.Thefifthsizeportfoliocontainsallotherlisted stocks.Foravaluefactor,breakpointsforbook-to- marketvaluearedeterminedonthebasisofthetop200stocksandthenappliedtothefullsampleo fstocks.

5 sampleofstocks.Specifically,stockswithapriceoflessthan$0.20areexcludedfromt h e sam ple.

Followingthosededicatedresearches,fromthesetofsurveyedcountries,one patterneme rgest h a t t h e traditionalC A P M i s a p p l i c a b l e t o a n u m b e r o f n a t i o n s w h i l e i t i s n o t recommendedontheremains.Evenin thecontextofUSandJapan–twoleader nationsintheA s i a Pacific region,thevalidityofCAPMonpracticeisstillacontroversialtopic.Inaddition,t h e s t u d y ofFF3F’ sa d o p t i o n lefta confusingpictureatleasti n t h e c o n t e x t o f Australia.However,CAPMhaseq uallyb e e n proventonotworkwellintheAustraliancontext.Therefore,thisresearchrisesupahypoth esisthatwhetherthesinglefactorassetpricingmodel-

CAPMis usableornot incalculationofareturno n equity inAsia-Pacific ingeneralorinAustraliainparticular.

Abriefdescription of the method

Themainpurposeof thisresearchis toconsiderthe validityofCAPMunderthecontexto f Australia.Thisstudya d o p t s t h e approachfromSavorandW i l s o n ( 2 0 1 4 ) i s t h a t allobservationsareseparatedintoa- day(announcementdays)groupasitisrelatedtodaysonwhichgrowth,inflation,employment,ce ntralbankannouncements,bonds,housing,consumersurveys,businesssurveysandspeechesfromth ePrimeMinisterortheGovernoroftheReserveBankofAustraliascheduledtobeannouncedandn- day(non- announcementdays)groupasitpertainst o t h e o t h e r d a y s Ins h o r t , t h e C A P M equation undern e s t e d dataseparationi s representedasfollows:

Ri,t+1–Rf,t+1=α)0+γ1Dt+1+γ2(1- Dt+1)βwhich N, t+γ3Dt+1βwhich A, t+ài,t+1 i i

 βwhich A i,t Theresponseofstock’sreturnwithrespecttoreturnofmarketportfolio ina-day.

Thedailyreturnismeasuredbythedifferenceinnaturallogarithmicoftwocontinuouss t o c k closeprices.Mathematically,the dailyreturnisexpressedasfollows:

Ri,t =αln(ClosePricei,t)-ln(ClosePricei,t-1) Concerningregressiontechnique,thisstudyemployslinearregressionwithpanel- correctedstandarderrorsmethod.Therationalebehindisthatitfixesheteroscedasticity an dc o n t e m p o r a n e o u s l y correlatedacrosspanelsof disturbances.

Moreover,toverifythevalidityofCAPMinAustralia,thisresearchalsoemploystheFam aandMacBeth(1973)regressionforthreereasons:(i)itisapracticalwaytofigureouth o w t h e riskfactorsdescribetheassetreturn;(ii)itisdevelopedbyProfessorEugene

F.Famai s knownas“TheFatherofFinance”andinventor ofFama-Frenchthree- factormodeland(iii)t h i s regressiont e c h n i q u e i s alsoproposedbySavorandW i l s o n ( 2 0

1 4 ) , V o (2015)andBrailsford,GauntandO'Brien(2012).Technically,theFamaandMacBethr egression

N comprisestwostages.Inthefirst-stage,toeachsingletimeperiod,across- sectionalregressioni s performed.Next,inthesecond-stage,thosefirststage- estimatedcoefficientsareaveragedacrosstimetoobtainthefinalestimates.Inthescopeofthisresearc h,theFamaandMacBethregressionis utilized inthetwofollowingmodels: and i,t+1–R i,t+1–R f,t+1=α)0+γ0βwhich f,t+1=α)1+γ1βwhich i,t+ài,t+1 i,t+ài,t+1

WhereR A A f,t+1 isthedailyexpectedexcessreturnina-day,R N f,t+1is the dailyexpectedexcessreturninn-day.βwhich A isthe asset’smarket betaina-dayandβwhich N is theasset’smarket betainn-day(bothestimatedover thepreviousyearusingdailyreturns).

Datarequirementsanddatasources

Datacovers2,200AustralianlistedfirmsonAustralianSecuritiesExchangefromBloombergf rom1January2007to31December2016 11Two additionalrequiredinputsforC A P M , beingthe risk- freerateandthemarketreturns,arealsocollectedfromthesamesourceandt h e s a m e p e r i o d T h e risk-freer a t e ’sproxy-

CommonwealthGovernmentb o n d s fromReserveBankofAustraliawiththematurityof10year s-isadoptedinthisstudy.Themarketreturnis thereturnsummation ofalllistedAustralianfirmscollectedbythisresearch 12

Moreover,thedatasetalsorecordsnews(dayofissue,impact)inrelationtoeventtypessuchasgr owth,inflation,employment,centralbankannouncements,bonds,housing,consumersurveys,busines ssurveysandspeechesfromthePrimeMinisterortheGovernoroftheReserveBankofAustralia.Dayscon tainatleastoneofthoseeventtypesarecalledannouncementdayso r a- daysandviceversa.Moreover,theirimpactsonmarket- movingpotentialarecollectedandclassifiedintothreelevels:(i)highexpectedimpact;

(ii)mediumexpectedimpactand(iii)l o w expectedimpact.Inthecaseofmore thanoneimpactlevelsarepresented,thehighestonew o u l d b e chosen.T h o s e dataareavailable andjudgedbyForexF a c t o r y website(http:// www.forexfactory.com/).Inaddition,f o r consistency,t h i s segmento f datasets p a n s from1Janu ary2007 to31December2016.

11 From morethan2,200Australianlistedfirms,this researchremovedfirmswhoseindustrialsectorsarenotr e c o r d e d Moreover,research alsodidthesame tofirmswhi ch hadless than25%tototalobservations.Althoughthiscriterionislikelyarbitrary,anunbiasedview isthetoppriority.However,asmarketreturnisc a l c u l a t e d , allofthoserejectedfirmsareaccountedtokeepitasclos edaspossibletoitstheoreticalversion.

12 The AustralianSecuritiesExchange’sindexcouldbereferredtotheS&P/ASX20,S&P/ASX50,S&P/ASX

Portfolioconstructions

Tenbeta-sortedportfoliosandTenidiosyncraticrisk-sortedportfolios

In the analysis of 10 beta-sorted portfolios, each firm in the dataset is assigned to a corresponding portfolio based on a specific procedure For each year, stocks are evaluated separately for announcement days and non-announcement days by estimating individual market betas using one year of daily returns Firms are then categorized into deciles according to their estimated beta values, with the first portfolio representing the 10% of firms with the lowest beta and the tenth portfolio comprising the top 10% with the highest beta This sorting process is conducted annually, utilizing a one-year rolling beta calculated through ordinary least squares regression.

Rit–Rft =αα it +b i (Rmt–Rft)+àit

The idiosyncratic risk-sorted portfolio was constructed similarly, with the standard deviation of return residuals recorded each time an individual stock market beta was generated Firms were then sorted into deciles based on this statistical measure Specifically, the first portfolio comprises the top 10 percent of stocks with the lowest standard deviation of return residuals, while the materials in the tenth portfolio represent the top 10 percent with the highest standard deviation These idiosyncratic risk-sorted portfolios are rebalanced on an annual basis.

Accordingt o t h e c e n t r a l l i m i t a t i o n theorem,t h e oneyearrollingbet ai s es t i m at ed o n l y i f t he minimal o b s e r v a t i o n - 3 0 - i s satisfied.

Table3-1Summaryof thenumberoffirmsin 10beta-sortedportfolios and in 10idiosyncraticrisk-sortedportfolios

The 2016 analysis of beta-sorted and idiosyncratic risk portfolios reveals varying performance metrics across different time frames The average returns for the beta-sorted portfolios on an annual basis show consistent high values, with a peak of 115, while the idiosyncratic risk portfolios also maintain strong performance, reaching up to 115 In contrast, the n-day beta-sorted portfolios exhibit lower returns, with a maximum of 92, compared to the idiosyncratic risk portfolios that hover around 91 Notably, both portfolio types demonstrate stability, with most values clustered closely around their respective averages This data underscores the resilience of these portfolios, highlighting their potential for investors seeking reliable returns in varying market conditions.

The25Fama-Frenchsizeandbook-to-marketportfolios

Toconstructthe25FamaandFrenchsizeandbook-to- marketportfolio,thisresearchinheritedtheworkofFamaandFrench(1996).Particularly,fromtheto pview,thatapproacho f portfolioconstructionisrelatedtotwofinancialindicators:marketcapitaliz ationvalueandb o o k value.Togetherwithbeta-sortedportfolioandidiosyncraticrisk- sortedportfolio,the25FamaandFrenchsizeandbook-to- marketportfoliosarealsoupdatedeveryyear.Theadoptionoccurredasfollows:giventhestatisticaln umbersonmarketcapitalizationofallfirmsinthedataset,fourbreakpointsweremanipulated.

Each firm was categorized into one of five portfolios based on its market capitalization, with the first portfolio consisting of the lowest 20% of firms The second portfolio includes the next lowest 20%, followed by the third portfolio, which contains the subsequent 20% of firms The fourth portfolio encompasses the next 20% lowest market capitalization firms, while the final portfolio consists of the remaining firms in the dataset After this allocation, five segments were created, each containing an equal number of firms Additionally, based on the book-to-market ratio, four breakpoints were established, leading to a further classification into five portfolios within each segment, resulting in a total of 25 portfolios, each containing the same number of firms.

Table3-2Summaryof thenumberoffirmsin the 25Fama-Frenchsizeandbook-to-marketportfolios

Industryportfolios

Inrelationto industryportfolio, thisresearchadopts theGlobalIndustryClassificationStandard(GICS)toclassifyforallfirmsinthedatasetintosub- sectors.Thefollowingtablesummarizesthedistributionoffirmsacross sectors.

Calculationsofportfolio’sbetaandportfolio’sreturn

Pooledregression

Afterallocatingstockintodifferentportfoliosasdiscussedabove,thenextstepinthiss t u d y i s t o estimateportfolio’sbetasandt h e i r returns.Betacoefficientsaret h e n testedt o confirmwhe therornottheyare statisticallydifferentonthea-dayandn- day.Technically,thef o l l o w i n g panelregressionisconducted:

Ri,t+1–Rf,t+1=α)0+γ1Dt+1+γ2(1- Dt+1)βwhich N, t+γ3Dt+1βwhich A, t+ài,t+1 i i

WhereR i,t+1–Rf,t+1i s theexpectedreturninexcessoftherisk-freerateofportfolioi.βwhich A N i,tistheindividualportfoliomarketbetaina-day.βwhich i,tistheindividualportfoliomarketbeta inn- day.Dt+1i sadummyvariablewhosevalueisoneiftheexpectedexcessreturnatt+1representsfora- dayandviceversa.Intheaspectofeconometrics,to fixtheheteroskedasticityandcontemporaneouslycorrelatedacrosspanels,thepanel- correctedstandarderrorsestimationm e t h o d wasutilized.

Inrelationtotheexpectedexcessreturn,toeachyearasstockallocatedintoportfolio,t h e i r returninthenextyearisemployedtocalculatetheportfolioreturnseparatelyfora-dayandn- day.Particularly,toeacha-dayofthenextperiod,tovalue-

31 weightedreturnmanipulation,t h e dailyportfolioexpectedexcessreturnistheweightedaverageofitss tockexpectedexcess returnwhereindividualstockmarketcapitalizationwasutilizedasaweight.Then,thosedaily

N portfolio’sreturnsa r e averageda c r o s s t i m e Similarly,t o n- day,t h e s a m e procedurewasu t i l i z e d , t o o

Concerningaboutportfolio’smarketbeta,toeachyearasstockallocatedintoportfolio,separat elyfora-dayandn- day,bythesameprocess,thedailyportfolio’sreturnsismanipulatedasaweightedaverageofitsstockexces sreturnwhereindividualstockmarketcapitalizationwasemployedasa weight Noticeable,in co ntrast,the returnw i t h i n that yeari s employedinsteadofreturnofnextyear.Thosedailyportfolio’sre turnswhoseroleplayasanindependentvariablewereregressedagainthecorrespondingmarketriskpr emiumintheCAPMmodeltofigureout theportfolio’smarketbeta.

Fama-MacBethregression

Inthescopeofthisresearch,theFamaandMacBethregressionisutilizedinthetwof o l l o w i n g models: and i,t+1–R i,t+1–R f,t+1=α)0+γ0βwhich f,t+1=α)1+γ1βwhich i,t+ài,t+1 i,t+ài,t+1

WhereR A A f,t+1 isthedailyexpectedexcessreturnina-day,R N f,t+1is thedailyexpectedexcessreturninn-day.βwhich A istheasset’smarketbetaina-dayandβwhich N is theasset’smarketbetainn-day(bothestimatedoverthepreviousyearusingdailyreturns).

Therefore,needlesstosay,βwhich A andβwhich N aresimilartothoseutilizedin pooledregression. Nevertheless,inthelefthandsideoftwoforegoingmodels,thedependentvariablesdenoted a i l y expectedexcessreturnwhiledependentvariableemployedinpooledregressiondenotesaverage dailyexpectedexcessreturn.

Int h e samefashionwithpooled regression,two differentapproachesinreturnm a n i p u l a t i o n –valueweightedreturnmanipulationandequalweightedreturnmanipulation– arebothemployed. i,t i, i

Pooledregression’sresult

Rf,t+1=α)0+γ1Dt+1+γ2(1-Dt+1)βwhich N, t+γ3Dt+1βwhich A, t+ài,t+1usinglinearregressionwith panel- i i correctedstandarderrorsmethod.Inthemodel,βwhichi,tisrollingbetafromoneyearofdailyreturn.

Particularly,βwhich N isrollingbetafromoneyearofdailyreturnestimatedinn-dayonlywhileβwhich A tisrollingbetafromoneyearofdailyreturnestimatedina- dayonly.Dt+1isadummyvariablew h o s e valueisoneiftheexpectedexcessreturnatt+1representsf ora-dayandviceversa.Bothportfolio’sbetaandreturnarere-manipulatedeachyear.R i,t+1–

Rf,t+1i smeanofdailyt h e expectedreturninexcessoftherisk- freerate.ForeaseofexploitationthenatureofRi,t+1

–Rf,t+1,s u p p o s e t h a t i n a particularyeartherea r ema n n o u n c e m e n t - d a yandnnon-announcementday.WheneverDt+1takesone,Ri,t +1–

Rf,t+1representsforthemeanofmexpectedexcessreturnsinthatyear.Similarly,ifDt+1iszero,Ri,t+1–

Rf,t+1representsforthemeanofnexpectedexcessreturnsinthatyear.Tosumup,toeachportfolioinacerta inyear,i t i s describedbytwoexpectedexcessreturnsandtworollingbetas.

Fromt h e t o p v i e w , t h e focusnumericresulti n t h e foregoingt a b l e i s significanceo f Ada y*Beta’scoefficientandthatofempiricaltestingdemonstratingthe equalityincoefficiento fA d a y * B e t a ( o rt h e variabledenotedbyD t+1 β At )andNotAday*Bet a( o rt h e variabledenotedby(1-

To understand the role of beta as a measure of systematic risk, it is essential to examine how financial markets respond to macroeconomic news On days when such news is released, systematic risk significantly influences stock price fluctuations, indicating that beta can effectively explain stock prices Conversely, on days without major news, the impact of beta on stock prices appears diminished Therefore, it is crucial to investigate two specific criteria to clarify this relationship further.

(i)thesignificanceofAday*Beta’scoefficientand(ii) thedifferenceinAday*Beta’scoefficientandNotAday*Beta’scoefficient.

Intercept Aday*Beta NotAday*Beta R 2 Intercept Aday*Beta NotAday*Beta R 2

Aday*Beta=NotAday*Beta Aday*Beta=NotAday*Beta

Intercept Aday*Beta NotAday*Beta R 2 Intercept Aday*Beta NotAday*Beta R 2

Aday*Beta=NotAday*Beta Aday*Beta=NotAday*Beta

25FamaandFrenchsizeandbook-to-marketportfolio 25FamaandFrenchsizeandbook-to-marketportfolio

Intercept Aday*Beta NotAday*Beta R 2 Intercept Aday*Beta NotAday*Beta R 2

Aday*Beta=NotAday*Beta Aday*Beta=NotAday*Beta

Intercept Aday*Beta NotAday*Beta R 2 Intercept Aday*Beta NotAday*Beta R 2

Aday*Beta=NotAday*Beta Aday*Beta=NotAday*Beta

Note:Thep-valuesarereported inparentheses.*significant at 10%level,**significant at 5%level,***significantat 1%level

Inrelationtotheregressionresult,inthelefthandsideofTable4-1,forvalue- weightedreturnmanipulation,toten-betasortedportfolio,ona-day,theslopeofthesecuritymarketline –impliedriskpremium–equals-10.63bps(significantatlevelof1percent)butonn- day,theimpliedr i s k premiumis-

3.09andi s insignificant 14In addition,t h e hypothesist h a t thecoefficienton Aday*Betais thesameasthecoefficientonNotAday*Betaisrejectedatthe10percentlevel.Furthermore,thepictur econtinuestospreadon25FamaandFrenchsizeandbook-to-marketportfolio.Particularly,ona- day,theslopeofthesecuritymarketline–impliedr i s k premium–is-

58.85bps(significantatlevelof1percent)whileonn-day,theimpliedriskpremiumtakes-

18.2bpsandisinsignificant.ThetestindifferencerejectsthehypothesisofAday*Beta’scoeffic ientequalsthatofNotAday*Betaatlevelof 5 per cent.

Similarly,intherighthandsideofTable4-1, forequal- weightedreturnmanipulation,to2 5 FamaandFrenchsizeandbook-to-marketportfolio,ona- day,theimpliedriskpremiumequals-

22.95bpsandissignificantatlevelof5percent.Incontrast,onn-day,theimpliedriskpremiumis-

3.4bpsandisinsignificant.Moreover,thehypothesisthatcoefficientonAday*Betais thesameas thecoefficienton NotAday*Betaisrejectedatlevel of 10percent.

Amazingly,onecommonthingsharedinthoseportfoliosisthatbetaisnegativelyrelatedt o mean ofdailytheexpectedreturninexcessoftherisk-freerateina- day.Thisfindingisinterestingtointerpret.IncontrastwiththefindingsfortheUSwhereonannou ncementdayst h e relationbetweenaveragereturnsandbetais s t r o n g l y positive,thef i n d i n g forAustraliaaboutthisrelationshipisstatisticallysignificantandnegative.Thiscanbeexplainedbythef actt h a t Australiaisasmallopeneconomywhoseeconomicactivitiesareheavilyinfluencedb yo t h e r advancedandmajortrading partnerssuchastheUS.Assuch,announcementsbytheA ustralianauthoritiesappeartobringgoodnewsfortheAustralianeconomybecausethesenews areconsideredastherespondingfactorstowhathavebeendecidedandannouncedbyo t h e r ma jorcountriesincludingtheUS.

Remarkably,intheTable4-1,movingfromtoptodownandfromlefttoright,theR 2 soften- betasortedportfoliosinvalue-weightedreturnmanipulation and 25 FamaandFrenchsizeandbook- to-marketportfoliosinbothvalue-weightedreturnmanipulationandequal- weightedreturnm a n i p u l a t i o n a r e 1 4 8 percent,4 1 p e r centand3 5 percent,respectively.T h o s e numbersarequitelowbutasLewellen,Nagel,& Shanken(2010)stated thattests onas setpricingmodelfocusoninterceptand impliedriskpremiumnot onR 2

Nevertheless,thepictureisdifferenttootherportfolioconstructionsonbothtworeturnmanip ulationapproachesasa matter ofoneof twofollowing criteriaisnotsatisfied.More specially,t h e e q u a l i t y incoefficiento n Aday*Betaw i t h coefficientonNotAday*Betai s confirmedt o

14 bpsstandsforbasicpoint.Onebasicpointfaequals1/100 th of 1percent. i,t i,t manipulations.Inaddition,t o 1 0 beta- sortedportfoliost h e coefficient,t h e coefficiento f Aday*Betaisnotstatisticallysignificantdiff erentfromzerounderequalweightedapproachandthereisnoevidencedemonstratesthatstockbeta behavesdifferentlyina-dayascomparedt o n-day.

However,itisreasonabletoarguethattwocriteriaareinsufficienttoprovethatbetaas ameasureofsystematicriskiftheindependentvariables–βwhich N andβwhich A –aredifferentfrom eachotherinnature.Toexaminethishypothesis,theAppendix1offersdifferencetestingofbetain a-dayandbetainn- day.Fromthesuggestedresults,thisresearchfindslittleagainste v i d e n c e 15

Insummary,t h e betas’performancesi n t e n - betasortedp o r t f o l i o s and2 5 FamaandFrenchsizeandbook-to- marketportfoliosindicatethatbetaislikelytobeabletoexplaincrossvariationinstockreturnina- dayandthewayofexpressingstockreturnofbetaisactuallydifferentbetweena-dayandn- day.Thatis, ondayswhenmacrolevel– relatednewswhichisg e n e r a l l y consideredasa sourceofsystematicrisk is announced,stock marketbetademonstratesstrongrelationtoitsreturns.Incontrast,thatresultseemsnottobeacasetootherp ortfolioconstructions.Fromthatobservations,itturnsoutthatdifferentconclusionsofbeta’sr o l e a sameasureofsystematicriskareinheritedfromdifferentportfolioformationtechniques.

15 Due totheperformanceof testofdifferenceinbetaofannouncementdayandnon- announcementday,thisresultreminduswhynotestimatingbetabyfromdailyreturnregardlesstypeofday.Thef urtherresultispresentedintheAppendix4.

Fama-MacBethregression’sresult

Table 4-2 presents the estimates and R-squared values from the Fama-MacBeth regression model, focusing on the relationship between daily expected excess returns (R_A f,t+1 and R_N f,t+1) and market betas (β_A and β_N) The dependent and independent variables used in the analysis are consistent with those discussed earlier R_A f,t+1 represents the expected excess return over a day, while R_N f,t+1 indicates the expected excess return over n days The asset's market betas, β_A and β_N, are calculated based on daily returns from the previous year, and the portfolio's beta and return are recalibrated annually.

Becauseregressorsandregressandsremainstaythesameascomparedtothoseemployedi n Tab le4-1,therefore,tofigureoutrepresentativeofbetaasameasureofsystematicrisk, theidenticalfilterproceduresareconsidered.Theyare(i)thesignificanceofβwhichAcoefficientand(ii)t h e βwhich A ’sc oefficientand βwhich N ’scoefficientmust bedifferent.

Inr e l a t i o n t o FamaMacbethregression,i n t h e firsts t e p , f o r eachperiod( d a y int h i s stud y),thecross- sectionalregressionisimplemented 16M o r e specially,ineachperiod,thereare10observationstote n-betasortedportfolioandidiosyncraticrisk- sortedportfolio;thereare2 5 observationsto25 FamaandFrenchsizeandbook-to- marketportfolio andthereare11observationstoindustryportfolio.Then,inthesecondstep,th ebetacoefficient(impliedriskpremium)isobtainedastheaverageofthefirststepestimates.

Inthe Table4-2,generally,to valueweightedmanipulation,asallfour distinctportfolioconstructionsareconsidered,allportfolioshavenotdemonstratedthatbetaisstillalive Indeed,i t isnoticeablethatmainlythesecondcriterionisviolated.Putitdifferent,atgivennor malsignificantlevel, thecoefficientsof βwhich in a-dayarenotdifferentfrom that inn-day.

3alsostatedaboutt h e s a m e picturealthoughportfolio’sreturnismeanofitsallstockreturns.Asama tteroffact,thehypothesist h a t beta’scoefficientsofa-dayandn- dayequalisrejectedatanygivennormalsignificantlevel.

16 Thenumberofa-daysandn-daysislistedinTable1ofAppendix2.Assuch,toagivenyear,iftherearem a-day,there are totalmcross-sectional regressionstomodelR A –R A =α)+γβwhich A + à andif i,t+1 f,t+1 0 0i , t i,t+1 therearenn-day,thereare totalncross-sectionalregressions tomodelR N i, t+1 –R N f,t+1 =α) 0 + γ 0 βwhich N i,t + à i,t

Note:Thep-valuesarereported inparentheses.*significant at 10%level,**significant at5 % level,***significantat 1%level.

Note:Thep-valuesarereported inparentheses.*significant at 10%level,**significant at5 % level,***significantat 1%level

Result’sdiscussion

Usingtworegressionmethodstofourtypesofportfolioundertwodifferentapproaches:value- weightedreturnmanipulationandequal- weightedreturnmanipulation,itturnsoutthatt h e conclusionoftheroleofbetaasameasureofsyste maticriskvariesbyconsideredportfolioo r employedestimationtechnique,atleasti n t h i s study.A s a m a t t e r o f fact,acrossreturnm a n i p u l a t i o n , totheten- betasortedportfolio,theroleofbetawasonlydemonstratedinthevalue- weightedapproach(Table4-1).Similarly,tothelefthandsideofTable4-

The choice of portfolio formation techniques significantly influences investment conclusions, as seen in the comparison between ten-beta sorted portfolios and the 25 Fama and French size and book-to-market portfolios While the first technique supports the relevance of beta, the latter suggests a divergence in outcomes Additionally, manipulating returns—whether through value-weighted or equal-weighted approaches—yields varying results across different regression methods For instance, a linear regression with panel-corrected standard errors confirmed the validity of beta in the Australian market, whereas the Fama-Macbeth regression produced contrasting insights Therefore, selecting the right portfolio formation technique is crucial for achieving accurate expectations, as highlighted by Brailsford, Gaunt, & O'Brien (2012) and Vo (2015), who noted the performance of asset pricing tests is closely tied to the definition of portfolio formation.

Concludingremarks

The Lintner version of the Capital Asset Pricing Model (CAPM) has been widely utilized by economic regulators, policymakers, and financial practitioners globally for over 50 years, particularly for calculating the discount rate of risky projects However, studies indicate that CAPM often underestimates returns for low-beta assets and overestimates returns for high-beta assets, leading to biased estimations compared to market requirements In 1992, Fama and French advanced this discussion by introducing the Fama-French Three-Factor Model (FF3F), followed by the Five-Factor Model (FF5F) in 2013, which gained significant academic attention Despite this, Australian economic regulators, such as the Australian Energy Regulator and the Economic Regulation Authority, have not adopted FF3F in their regulatory decisions, raising concerns about its applicability in the Australian market (Vo, 2015).

Recently,aworkwasdonebySavorandWilson(2014)usingtheUSdataconfirmedthatbeta,t h e h e a r t o f t h e C A P M , i s afterallani m p o r t a n t measureo f s y s t e m a t i c r i s k Puti t different ly,betaisstillaliveintheUSmarket.Thisstudyisconductedtoconsiderthevalidityo f theSharpe- LintnerCAPMinthecontextofAustraliaonthegroundofSavorandWilson(2014),thefirstattem pt everforthe AsiaPacificregion.

This study analyzes over 2,200 Australian listed firms, utilizing approximately 2 million observations collected from Bloomberg between January 1, 2007, and December 31, 2016 It incorporates data from ForexFactory to categorize days with significant economic announcements—such as those related to growth, inflation, employment, and central bank communications—into an "announcement day" group, distinct from "non-announcement days." The research specifically investigates the relationship between beta and expected stock returns on both announcement and non-announcement days, employing four distinct portfolio formations, including ten beta-sorted portfolios.

(iii) 25Fama-Frenchsizeandbook-to-marketportfolios; and(iv)industryportfolios.

Using thelinearregressionwithpanel- correctedstandarderrors,thisstudyistoconfirmwhetherornotbetaisstill aliveintheAustralianc ontext Ifbetaisstillalive,impliedriskpremiumo n t h e announcementdays,t h ea- day,i ss t a t i s t i c a l l y differentfromt h a t o f n o n -

41 announcementdays,then-day;andbeta’scoefficientonthea- dayisstatisticallysignificantdifferentfromzero.UsingthepooledOLStechnique,for10beta- sortedportfoliosundervalue-weightedapproachand25FamaandFrenchsizeandbook-to- marketportfoliosundervalue-weightedandequal- weightedapproaches,theexpectationdoeshold.Inaddition,findingsfromt h i s studyalsodemonstr atethatbetaisnegativelyrelatedtodailyexpectedexcessreturnintheannouncementdays.However,th estudyfailstoproducethesamefindingswhentheFama-Macbethregressionis utilized.

It is argued that while certain announcements can significantly impact specific industries or firms, others may not experience the same effect This sensitivity analysis categorizes various announcements related to growth, inflation, employment, central bank actions, bonds, housing, consumer surveys, business surveys, and speeches from the Prime Minister or the Governor of the Reserve Bank of Australia into distinct subgroups These include a macro event-related group, which encompasses news about growth, inflation, employment, central bank policies, bonds, and speeches, and a micro event-related group, which focuses on housing, consumer surveys, and business surveys.

(iii)economicevent- relatedg r o u p i scharacterizedbynewsaboutgrowth,inflation,employment,housing,consumer surveys,businesssurveysandspeeches;and(iv)financialevent- relatedgroupcontainsnewsaboutCentralBankandbonds.Usingthesameportfolioconstructionpr ocedures,betaestimationapproaches,a n d regressionmethodology,keyfindingsfromt h i s t h i s se nsitivityanalysiscanbesummarizedasbelow.Detailsof theanalysiscanbefoundinAppendix5.

First,whenthemacroevent- relatedgroupisconsidered,inrelationtoconditionalbetaestimationapproach,most of theportfoliosstatedthatbetais not alive.The resultscomefromo n bothtworeturnmanipulationsandtworegressionmethodologies.Howeve r,the25FamaandFrenchsizeandbook-to- marketportfoliosofferanoppositeanswerifpooledregressionwasusedregardlessreturnmanipu lation.Intermsofcommonbeta,itseemstobeanextremecaseasaclear- cutconfirmationthatbetaisnotaliveisdrawn.Indeed,thereisnoevidencef o u n d inanypor tfolioconstructions,returnmanipulation andregressiontechniques.

Second,whenthemicroevent- relatedgroupisconsidered,toconditionalbeta,theclaimt e n d s toremainstaythesameascompar edtocommon betaestimationapproachdiscussed below.Needlesstosay,except25FamaandFr enchsizeandbook-to- marketportfoliosunderequalweightedreturnmanipulation,allportfolioshavedemonstratedthatbetais

42 unlikelyalive.Inaddition,theperformancecontinuestobeemphasizedifcommonbetaestimationapp roachi s accounted.

Third,whenthe financial event- relatedgroupi s considered,in relationtoconditionalbeta,ontheonehand,undervalueweightedreturn manipulation,usingpooledregression,only

25FamaandFrenchsizeandbook-to- marketportfoliosandindustryportfoliosupporttheroleo f betaasameasureofsystematicrisk,butonth eotherhandmostofportfoliosareunlikelytoagree.Intermsofcommonbeta,thepicturetendstobebetterb utitisstillfarfromaninversedsituationbecauseofthereare5outof16votes.Insummary,itcouldbe arguedthatbetaisu n l i k e l y aliveunder thisconsideration.

Finally,whentheeconomicevent- relatedgroupisconsidered,fromgeneralview,inbotht w o betaestimationapproaches,aquitesimilarc onclusionisrepeatedasmostofportfoliohavedemonstratedthatbetaisunlikelyalive.

Inanutshell,empiricalfindingsfromthisstudyhavedemonstratedthatbetamaybealivei n theA ustraliancontextontheconditionofportfolioconstructionsandeconometrictechniques.Assuch,f indingsfromthisstudyaresomewhatsimilarwiththosefromVo(2015)t h a t portfolioformationsdo matterwhenempirical studiesonassetpricing areconducted.

Policyimplications

Onthegroundofempiricalfindingsfromthisstudy,thefollowingrecommendationsareo n offer tovariousstakeholdersincluding academics,investors andpractitioners,andpolicymakers.

 First,itisanimplicationforacademicswhoaregoingto/ havebeenconductingstudieso n thesubject.Assetpricingisanextremelycomplicatedt ask.Ithoweverplaysanimportantrole inunderstandingandexplaining returnstotheassets.Assuch,seriousstudiesfromacademicsarealwaysr e q u i r e d t o p r o v i d e anunbiasedv i e w o n thiscomplicatedmatter.Inaddition,empiricalfindingsmayb esensitivetothedesignoft h e research.A s such,s e n s i t i v i t y analysisi s considere dimportantt o ensuret h e robustnessof theanalysis.

 Second,forpractitioners,thedebateonanappropriatenessof thecapitalassetpricingm o d e l s isnevereasyandended.Assuch,practitionersshouldb ecarefultoselectthem o s t appropriatemodel,whichmaynotbethebest,forthepurposeofth eexercise.Inaddition,recognizinga groupofdifferentm o d e l s o n t h e groundo f a fundam entalm o d e l appearsto provideastrongandrobustanchorforbusinessdecisions.

 Third,forpolicymakers,assetpricingisapuzzleindecisionmakingprocess.Assuch,i t isreco mmendedthedecisionmakersutilizevariousmodelsfordecisionmaking.However, asfindingscanbeanythingdependingonsample,timeperiod,economictechniquesandv ariouso t h e r factors,t h e r e sho ul d b e ananchor– t h e fundamentalm o d e l Asthenamerecommends,thefundamentalmodelshouldb edevelopedonas t r o n g theoreticalground.Itisimportanttonote thateconomictheoriesarerelativelys t a t i c whereasempiricalstudiesareevolvedm o r e oftent o reflectwhathasbeen

“observed”frompractice.Policyshouldbeformulatedwithstability.Assuch,withsuc haninterestingandfastchangingareaofassetpricing,conservatismappearstoberelevantand important tokeepin mind.

(2013).TheEffectofFinancialLeverage&SystematicRiskonS t o c k ReturnsintheAm manStockExchange(AnalyticalStudy–IndustrialSector).ResearchJournalof

Banz,R W (1981).T he relationshipbetweenr e t u r n andmarketv a l u e o f commons t o c k s

Basu,S.(1977).Investmentperformanceofcommonstocksinrelationtotheirprice- earningratio:Atestoftheefficientmarkethypothesis.TheJournalofFinance,32(3),663-682.

Fama,E.F.,&MacBeth,J.D.(1973).Risk,ReturnandEquilibrium:EmpiricalTests.Journalo f

Fama,E , & French,K (2015).A five- factora s s e t pricingm o d e l J o u r n a l o f FinancialEconomics,116(1),1-22.

(2016,December29).RetrievedfromWikipedia,TheFreeEncyclopedia.:https:// en.wikipedia.org/w/index.php?titlea

Fernandez,P.(2015).CAPM: Anabsurdmodel.Business ValuationReview,34(1),4-23.

Green,J , Hand,J.R , & Zhang,X F.( 2 0 1 3 ) T h e sup ra vi ew ofreturnpredictivesignals.

Harvey,C.R.,Liu,Y.,&Zhu,H.(2015,February3 ) RetrievedfromSSRN:https://ssrn.com/ abstract"49314

Henry, Ó T., & Street, C (2014) Retrieved fromhttps://www.aer.gov.au/system/files/D14%2052760%20%20Estimating

%20Beta_%20An%20update%20Olan%20T%20Henry%20April%202014.pdf

MarketModelBasedo n Semivariance.T h e J o u r n a l o f Financiala n d QuantitativeAn alysis,9(1),1-11.

(1972).T h e CapitalAssetPricingModel:S o m e EmpiricalTests.MichaelC.Jensen,STU

Model-Mining Experiment Retrieved from SSRN:http://dx.doi.org/10.2139/ssrn.2182139

Regulator Retrieved from https://www.aer.gov.au/system/files/Michael%20McKenzie%20%26%20Graham

%20Partington%20%E2%80%93%20Report%20to%20the%20AER,%20Part%20A- Return%20on%20Equity%20%E2%80%93%20October%202014_0.PDF

RetrievedfromSSRN:https://ssrn.com/abstract!56623

Mollik,A T (2014).T h e C A P M andr i s k - r e t u r n anomaliesi n Asianemergingmarkets: evidencefromBangladeshandMalaysia.T h e 1 1 t h A s i a n B u s i n e s s ResearchConference2014(pp 1- 24).Dhaka,Bangladesh:Australia: WorldBusinessInstitute.

O'Brien,M.A.,Brailsford,T.,&Gaunt,C.(2008,August5).SizeandBook-to-MarketFactorsi n

A u s t r a l i a RetrievedfromSSRN:http://dx.doi.org/10.2139/ssrn.1206542

(2013).CapitalAssetPricingModel(CAPM):EvidencefromNigeria.ResearchJourna lofFinanceandAccounting,4(9).

Olakojo,S.,&Ajide,K.(2010).TestingtheCapitalAssetPricingModel(CAPM):TheCaseo f t h e NigerianSecuritiesMarket.InternationalBusinessManagement,4(4),239-242.

(n.d.).TheCapitalAssetPricingModelversusTheThreeFactorModel:AUnitedKingdom Perspective.ThirdBiennialInternationalConference on Business,

Banking and Finance Retrieved fromhttps://sta.uwi.edu/conferences/09/finance/documents/Riad

(1981).Misspecificationofcapitalassetpricing:Empiricalanomaliesbasedo n earnings'yiel dsandmarketvalues.JournalofFinancialEconomics,9(1),19-46.

(1977).A c r i t i q u e o f t h e assetp r i c i n g theory’st e s t s PartI:O n pastandp o t e n t i a l t e s t a b i l i t y ofthetheory.JournalofFinancialEconomics,4(2),129-176.

(1984).StockReturns,Beta,VarianceandSize:AnEmpiricalAnalysis.FinancialAnalystsJou rnal,40(4),36-41.

(2011).CostofEquityforRegulatedCompanies:A n internationalComparisono f Regulatory Practices.CranfieldSchoolo f Management,UK,andMember,Competition Commission, UK.

Tanga,G.Y.,&Shumb,W C.(2003).The conditionalrelationshipbetweenbetaandreturns:recentevidencefrominternationalstockma rkets.InternationalBusinessReview,12(1),109-126.

1.0.PaperPresentedatAustralianEconometric Society Meeting.Hobart,Australia.

(2012).T e s t i n g t h e A p p l i c a b i l i t y o f T h e FamaandFrenchThree FactorM o d e l in Bursa Malaysia.T h e Global JournalofFinance andEconomics,9(1),33-43.

(2008,February28).FailureofAssetPricingModels:TransactionC o s t , Irrationality,o r M i s s i n g Factors.SocialScienceResearchN e t w o r k.Retrievedfromhttp://dx.doi.org/10.2139/ssrn.1100975

ThisAppendixistoassessdifferenceinestimatedbetaofannouncementdaysandnon- announcementdays.Thetwofirsttablesarededicatedt o tenbeta-sortedportfolios.Thetable3andtable4areabout10idiosyncraticrisk– sortedportfolios.Thenexttwotablesarerelatedto25FamaandFrenchsizeandbook-to- marketportfolios.Thelasttwotablesconcernabout11industryportfolios.Toeachtable,thenullhypothesisi s “βwhich annequals βwhichnon” andthe alternativehypothesis is“βwhichanni sdifferentfromβwhichnon”

Table1:Thistableobtainsthedifferenceintheestimatedtenbeta–sortedportfoliomarketbetaofannouncementdaysandnon- announcementdays.βwhich nonis chosenasreference.βwhichannandβwhichnonareportfoliomarketbetasinannouncementdayandnon-announcementday,respectively.Thep- valuesarereportedintheparentheses.Theportfolioreturnisweightedaveragebymarketvalueofitsallstockreturns.*significantat10%level,**sign ificantat5%level,***significantat 1%level

Beta Low 2 3 4 5 6 7 8 9 High βwhichnon -0.711 0.005 0.273 0.434 0.661 0.850 1.045 1.257 1.620 2.294 βwhichann-βwhichnon 0.024 0.031 -0.025 -0.012 -0.068 -0.090 -0.104 -0.118 -0.226 -0.198

Table2:Thistableobtainsthedifferenceintheestimatedtenbeta–sortedportfoliomarketbetaofannouncementdaysandnon- announcementdays.βwhichnonischosenasreference.βwhichannandβwhichnonareportfoliomarketbetasinannouncementdayandnon-announcementday,respectively.Thep- valuesarereportedintheparentheses.Theportfolioreturnismeanofitsallstockreturns.*significantat10%level,**significantat5%level,

Beta Low 2 3 4 5 6 7 8 9 High βwhichnon 0.939 1.073 1.168 1.328 1.141 1.274 1.041 1.016 0.905 1.186 βwhichann-βwhichnon 0.002 -0.017 0.058 -0.158 -0.041 -0.353 -0.160 -0.278 0.009 0.578

Table3:This tableobtains thedifferencein theestimated tenidiosyncraticrisk–sortedportfoliomarketbetaofannouncementdaysandnon- announcementdays.βwhichnonischosenasreference.βwhichannandβwhichnonareportfoliomarketbetasinannouncementdayandnon-announcementday,respectively.Thep- valuesarereportedin theparentheses.Theportfolioreturn isweightedaveragebymarketvalueofits all stock returns.

*significantat10% level, **significantat5% level, *** significant at 1%level

Beta Low 2 3 4 5 6 7 8 9 High βwhichnon -0.679 0.163 0.408 0.610 0.781 0.964 1.186 1.500 1.937 3.291 βwhichann-βwhichnon 0.185 0.053 0.030 0.012 0.028 0.031 0.023 -0.015 -0.061 -0.287

Table4:This tableobtains thedifferencein theestimated tenidiosyncraticrisk–sortedportfoliomarketbetaofannouncementdaysandnon- announcementdays.βwhichnonischosenasreference.βwhichannandβwhichnonareportfoliomarketbetasinannouncementdayandnon- announcement day,respectively.Thep- valuesarereportedintheparentheses.Theportfolioreturnismeanofitsallstockreturns.*significantat10%level,**significantat5%level,***signif icantat1%level

Beta Low 2 3 4 5 6 7 8 9 High βwhichnon 0.555 0.664 0.798 0.939 1.082 1.110 1.180 1.134 1.240 1.774 βwhichann-βwhichnon 0.029 0.068 0.089 0.093 0.034 0.021 -0.051 0.015 0.027 -0.063

Table5:Thistableobtainsthedifferenceintheestimated25FamaandFrenchsizeandbook-to- marketportfoliomarketbetaofannouncementdaysandnon- announcementdays.βwhichnoni schosenasreference.βwhichannandβwhichnonareportfoliomarketbetasinannouncementdayandnon- announcementday,respectively.Thep- valuesarereportedintheparentheses.Theportfolioreturnisweightedaveragebymarketvalueofitsalls t o c k returns.*significantat 10%level,**significantat 5%level,*** significant at 1%level

Beta Growth 2 3 4 Value βwhichnon Small 0.713 0.961 0.291 0.851 0.722 βwhichann-βwhichnon -0.054 -0.228 0.353 -0.077 -0.191

Table6:Thistableobtainsthedifferenceintheestimated25FamaandFrenchsizeandbook-to- marketportfoliomarketbetaofannouncementdaysandnon- announcementdays.βwhichnoni schosenasreference.βwhichannandβwhichnonareportfoliomarketbetasinannouncementdayandnon- announcementday,respectively.Thep- valuesarereportedintheparentheses.Theportfolioreturnismeanofitsallstockreturns.*significantat1 0 % lev el,**significantat5%level,*** significant at 1%level

Beta Growth 2 3 4 Value βwhichnon Small 0.753 0.963 0.486 0.678 0.921 βwhichann-βwhichnon -0.120 -0.089 0.056 0.193 -0.183

Table7:Thistableobtainsthedifferenceintheestimated11industryportfoliomarketbetaofannouncemen tdaysandnon- announcementdays.βwhichnonischosenasreference.βwhichannandβwhichnonareportfoliomarketbetasi n announcementdayandn o n - a n n o u n c e m e n t day,respectively.T h e p o r t f o l i o r e t u r n i s weightedaveragebymarketvalueofits allstockreturns.Thep-valuesarereportedintheparentheses.*significantat10%level,**significantat5% level, ***significantat1% level βwhichnon βwhichann-βwhichnon p-value ConsumerDiscretionary 0.769 -0.045 (0.301)

Table8:Thistableobtainsthedifferenceintheestimated11industryportfoliomarketbetaofannouncemen tdaysandnon- announcementdays.βwhichnonischosenasreference.βwhichannandβwhichnonareportfoliomarketbetasinannouncementdayandnon- announcementday,respectively.Theportfolioreturnismeano f itsallstockreturns.Thep- valuesarereportedintheparentheses.*significantat10%level,**significantat5%level,*** significant at 1%level βwhichnon βwhichann-βwhichnon p-value ConsumerDiscretionary 0.650 0.024 (0.469)

Table1:Thetablereportsthenumberofannouncement-daysaswellasnon- announcementdaysfrom2007to2016.DataisavailableandjudgedbyForexFactorywebsite (http://www.forexfactory.com)

Table2:Thetableillustrates9kindsofannouncementsinrelationtomacroeconomicissuesaswellastheirtypicalexamplesemployedi n t h i s study.Data is availableandjudgedbyForexFactorywebsite (http://www.forexfactory.com)

BusinessSurveys Measurementlevel of adiffusion indexbasedinsurveyedservice-basedcompanies

Australia Germany NewZealand USA Canada UK

NewYorkStateP ub l i c UtilitiesC ommission (NYSPUC)

G a s andEle c tri ci ty Mark ets (Ofgem)

RPM CAPM DDM RPM CAPM

Otheruse of DDM Cross- checkon MRP

*On theoverallCoE butnot forindividual firms.

Notes: CAPM: Sharpe-Lintner-Blackcapitalassetpricingmodel.

significant at the 1% level.

Ten-betasorted portfolio Ten-betasorted portfolio

Intercept Beta Aday*Beta R 2 Intercept Beta Aday*Beta R 2

Idiosyncratic risk-sortedportfolio Idiosyncraticrisk-sortedportfolio

Intercept Beta Aday*Beta R 2 Intercept Beta Aday*Beta R 2

25 FamaandFrenchsizeand book-to-marketportfolio 25 FamaandFrenchsizeand book-to-marketportfolio

Intercept Beta Aday*Beta R 2 Intercept Beta Aday*Beta R 2

Intercept Beta Aday*Beta R 2 Intercept Beta Aday*Beta R 2

D t+1 )β N i,t )intheempiricalresultsection,thisisdonebyconsideringbyverifyingthesignificanceofA day*BetaintheTable1.Thereasonbehindisthatbecauseoftheindependentvariable– betawasestimatedregardlesstypesofdaysoitissufficienttoanswerwhetherornotstockreturnisexplai nedbyi t s marketbeta.

=α)0+γ1βwhichi,t+ài,t+1usingFamaMacBethregresssion.Thetablecontainsresultsseparatelyonf o u r distinct typesofportfolios(betasortedportfolio,idiosyncraticrisk-sortedportfolio,25Fama-

Frenchsizeandbook-to- marketportfolioandindustryportfolio).Theportfolioreturnisweightedaveragebymarketvalueofitsa llstockreturns p-valuesarereportedinparentheses.

*significantat10%level,**significantat5% level, ***significantat 1%level

=α)0+γ1βwhichi,t+ài,t+1usingFamaMacBethregresssion.Thetablecontainsresultsseparatelyonf o u r distinct typesofportfolios(betasortedportfolio,idiosyncraticrisk-sortedportfolio,25Fama-

Frenchsizeandbook-to- marketportfolioandindustryportfolio).Theportfolioreturnismeanofitsallstockreturns.p- valuearereportedinparentheses.*significant at

10%level,**significantat5%level,***significantat1%level

This study categorizes announcements from the Prime Minister and the Governor of the Reserve Bank of Australia into four distinct groups: macroevent-related, microevent-related, financial event-related, and economic event-related The macroevent-related group includes news on growth, inflation, employment, the Central Bank, bonds, and speeches The microevent-related group focuses on housing, consumer surveys, and business surveys The financial event-related group encompasses news related to the Central Bank and bonds, while the economic event-related group covers topics such as growth, inflation, employment, housing, consumer surveys, business surveys, and speeches The findings for each group are presented in four tables.

Thefirsttwotablesdemonstrateresultin which beta isestimatedseparatelyfora-dayandn-day,so callconditionalbeta.

Aclaim“Yes”inthepooledregressionrowmeansthatcoefficientofβwhich N andβwhich A inthefollowingmodelaredifferentandβwhich A is significant

Ri,t+1–Rf,t+1=α)0+γ1Dt+1+γ2(1- Dt+1)βwhich N, t+γ3Dt+1βwhich A, t+ài,t+1 i i andviceversa

Aclaim“Yes”intheFama-Macbethregressionrowmeansthatcoefficientofβwhich N andβwhich A inthefollowingmodelaredifferentandβwhich A tissignificant.Althoughtheindependentvariables-βwhich-intwofollowingequationsaredenoteddifferentlyinthesurface,theyallequalinvalue. i,t+1–R f,t+1 =α)0+γ0βwhich i,t+ài,t+1 andviceve rsa i,t+1 –R f,t+1=α)0+γ0βwhich

Aclaim“Yes”in thepooledregressionrowmeansthatcoefficientγ3Dt+1isstatisticallysignificantdifferentfromzero.

Ri,t+1–Rf,t+1=α)0+γ1Dt+1+γ2βwhichi,t+γ3Dt+1βwhichi,t+ài,t+1 andviceversa

Aclaim“Yes”intheFama-Macbethregressionrowmeansthatcoefficientofβwhich N andβwhich A inthefollowingmodelaredifferentand i,tissignificant i,t+1–R f,t+1 =α)0+γ0βwhich i,t+ài,t+1 andviceversa i,t+1–R f,t+1=α)0+γ0βwhich i,t+ài,t+1

Ten-betasorted portfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression Yes No Yes No

Fama-Macbethregression No No No No

Ten-betasorted portfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression No No Yes No

Fama-Macbethregression No No No No

Ten-betasorted portfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression No No No No

Fama-Macbethregression No No No No

Ten-betasortedportfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression No No No No

Fama-Macbethregression No No No No

Ten-betasorted portfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression No No No No

Fama-Macbethregression No No No No

Ten-betasorted portfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression No No Yes No

Fama-Macbethregression No No No No

Ten-betasorted portfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression No No No No

Fama-Macbethregression No No No No

Ten-betasorted portfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression Yes No No No

Fama-Macbethregression No No No No

Ten-betasorted portfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression No No Yes Yes

Fama-Macbethregression No No No No

Ten-betasorted portfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression Yes No No No

Fama-Macbethregression No No No No

Ten-betasorted portfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression No Yes No No

Fama-Macbethregression No No No Yes

Ten-betasorted portfolio Idiosyncratic risk-sortedportfolio 25 FF portfolio Industryportfolio

Pooledregression Yes No No No

Fama-Macbethregression Yes No No Yes