Factors affecting stock market performance with special reference to market-to-book ratio in banking - the Israeli case

Transcription

1 Facors affecng sock marke performance wh specal reference o marke-o-book rao n bankng - he Israel case AUTHORS ARTICLE INFO JOURNAL FOUNDER Davd Ruhenberg Shaul Pearl Yoram Landskroner Davd Ruhenberg, Shaul Pearl and Yoram Landskroner (2). Facors affecng sock marke performance wh specal reference o marke-obook rao n bankng - he Israel case. Banks and Bank Sysems, 6() "Banks and Bank Sysems" LLC Consulng Publshng Company Busness Perspecves NUMBER OF REFERENCES NUMBER OF FIGURES NUMBER OF TABLES The auhor(s) 28. Ths publcaon s an open access arcle. busnessperspecves.org

2 Banks and Bank Sysems, Volume 6, Issue, 2 Davd Ruhenberg (Israel), Shaul Pearl (Israel), Yoram Landskroner (Israel) Facors affecng sock marke performance wh specal reference o marke-o-book rao n bankng he Israel case Absrac One of he mporan facors affecng sock reurns s he marke-o-book rao. In bankng, he rao has also been aken o be a proxy for he charer value of banks. The purpose of he paper s o derve and esmae emprcally he facors ha deermne relave values of equy n bankng. Our analycal framework s based on he dscouned cash flow (DCF) approach of valuaon. In our emprcal es, we use daa on Israel banks. The man resuls are ha rsk, reurn and marke and economc condons have a sgnfcan effec on he rao. Keywords: marke-o-book rao, Gordon growh model, charer value. JEL Classfcaon: G2. Inroducon One of he mos mporan facors affecng sock marke performance s he marke-o-book value rao. A large number of sudes, usng U.S. and nernaonal daa, have demonsraed ha hs rao has a sgnfcan explanaory power for cross-secon average sock reurns and ha hese reurns are hgher for socks wh hgh marke-o-book raos. See Fama and French (992) for he effec n he U.S., and Chan e al. (99), Fama and French (998) and more recenly Maroney and Proopapa-daks (22) for he effec n oher naonal markes. There are wo compeng explanaons for he above effec. One nerpreaon, conssen wh he effcen-marke hypohess, s ha he rao s a proxy for rsk and, hence, he posve relaonshp found beween hs rao and sock reurns (Fama and French, 992). Specfcally, Fama and French (996) and Vassalau and Xng (24) argue ha he rao s a proxy for fnancal dsress or defaul rsk. An alernave explanaon of hs effec s ha s a marke anomaly ha volaes he effcen marke hypohess. Lakonshok, Shlefer and Vshny (994) argue ha cognve bases and nvesors agency coss are he reasons for hs anomaly. There s exensve leraure n fnance on he relaonshp beween relave valuaon and sock reurns, and he facors ha deermne he marke-obook rao. In he me-seres, leraure papers have suded wheher he varaon of he rao reflecs varaon of expeced reurns (e.g., Kohar and Shanken, 997). In he cross-secon, leraure papers show how much of he rao s relaed o cashflow dfference (Fama and French, 995). However, he leraure on hs effec s que lmed n fnancal servce frms n general and n bankng n parcular. Davd Ruhenberg, Shaul Pearl, Yoram Landskroner, 2. The vews expressed n hs paper reflec he opnons of he auhors only. We hank Merav Kora for her excellen research asssance. Peerkor and Nelsen (25) develop a leverage-based valuaon model o nvesgae wheher he marke-o-book rao s a proxy for rsk. A number of aspecs make valuaon of fnancal servces frms unque. Frs and mos mporan, hese frms are hghly regulaed. Second, banks n some counres are consraned geographcally (e.g., lms on branchng and nersae bankng ha exsed unl recenly n he U.S.) or n erms of producs hey can sell (e.g., Glass-Seagall ac n he U.S. ha unl 2 separaed commercal from nvesmen bankng). Thrd, n mos counres here are resrcons on he enry of new banks or on mergers beween exsng banks. An mporan mplcaon of regulaon s ha resrcs compeon and, hus, affecs valuaon of banks. Sudes n bankng have examned he relave value of equy n he conex of he charer value of he bank and s relaonshp o regulaon and marke condons. Keeley (99) provdes emprcal suppor for he harer value hypohess : an ncrease of compeon n bankng reduces charer value and, hus, causes he marke value of he bank o declne relave o book value 2. A prme example s he U.S., as barrers, were removed compeon ncreased and consequenly charer value declned. Saunders and Wlson (2) argue ha regulaon ha resrcs compeon enhances charer value and ncreases self-regulaon by banks whle deregulaon does he oppose. They also found a posve relaonshp beween charer value and capal raos durng economc expanson and argue ha durng such perods bank charer value ncreases o reflec growh. In Israel he lberalzaon and prvazaon of money and capal markes has gahered momenum n he pas decade. As par of hs polcy, he governmen whdrew from he capal marke gradually and reduced s nvolvemen n fnancal nermedaon. Concurrenly, many publc companes, ncludng banks, have been prvazed. As par of he process, he shares of Israel s large banks were relsed for radng on he Tel Avv sock exchange afer havng been delsed snce he bank shares crss n The rao used n hs paper s marke value o book value of asses o oban hs rao, one adds he book value of deb o boh he marke and book values of equy. 87

3 Banks and Bank Sysems, Volume 6, Issue, 2 The objecve of hs paper s o denfy and esmae emprcally he facors ha deermne he marke-obook rao n bankng. In he paper, we es wo hypoheses: he effcen marke hypohess and he charer value hypohess n bankng. Our analyss s based on he dscouned cash flow (DCF) approach o he valuaon of equy. In order o derve a esable equaon, we make some smplfyng assumpons n he valuaon model. These assumpons, however, are relaxed n he emprcal ess. Ths approach o relave valuaon has been used exensvely (Fama and French, 995; and, recenly, Fama and French, 26). We use wo versons of a DCF model: () frs he dvdend valuaon model; and (2) he free cash flow o equy model (FCFE). The explanaory varables whch we derve and laer es emprcally are: profably of he bank; rsk (cred and marke); nvesmen polcy of he bank; macroeconomc and capal marke condons; bankng srucure (compeon) and regulaon of banks. Secon provdes he heorecal bass for he marke-o-book (MV/BV) rao, and he mehodology for dervng he facors ha affec ha rao for banks. Secon 2 presens an overvew of Israel bankng, summary sascs and emprcal fndngs wh regard o he MV/BV rao and s deermnans for he fve larges banks n Israel, coverng he perod of The fnal secon presens a summary and he conclusons ha arse from he emprcal fndngs.. Model and mehodology In hs secon, we generae he basc facors ha deermne he MV/BV rao. Our analycal framework s based on he valuaon of equy usng a dscouned cash flow approach. We assume a consan payou and nvesmen polcy of he bank ha mply a consan growh rae, hese assumpons are relaxed n he emprcal par of he paper. Assumng ha, he dvdend per share wll grow a a consan rae (g) he dvdend valuaon model can be wren as follows: P D g D, () k g k g where P s he curren prce of he share, D s he expeced dvdend per share n perod and k s rsk adjused cos of equy (nernal rae of reurn). Followng he well-known Gordon model we defne he varable b b as he consan rae of reaned earnngs or he reenon rao ha equals he renvesmen rae. The complemen fracon b s he proporon of dvdend pad ou from earnngs or he payou rao, hus, D E b, where E s he expeced earnngs per share n perod. We assume also a consan rae of reurn on equy E (), defned as follows:, where bv bv s book value per share a me. By subsung for E = bv and for g = b, dvdng boh sdes of equaon () by bv we oban he prce-o-book value rao (per share): P b (2) bv k b E By mulplyng he numeraor and he denomnaor on he lef sde of he equaon by he number of shares lsed for radng, we oban he marke value of equy of he frm dvded by s book value,.e., he marke-o-book rao MV /BV : MV b g BV k b k g (3) Thus, he MV/BV rao s deermned by he reurn on equy of he frm, he rsk of equy as refleced by he cos of equy capal k and s renvesmen polcy (b). Based on equaon (3), we oban he followng expeced relaonshps beween he MV/BV rao and he explanaory varables: MV / BV MV / BV R ; MV / BV MV / BV b k ; g k k (4). A second verson of a DCF approach we use here, s he free cash flow o equy dscoun model (FCFE). Ths model uses a more expansve defnon of cash flows o equy han ha used n he dvdend model. These cash flows are defned as hose lef over afer meeng all fnancal oblgaons and nvesmen expendures. The consan growh verson of he FCFE model can be expressed as: FCFE P, (5) k g where, FCFE s he expeced free cash flow o equy nex year, g = b s he consan growh rae of he cash flows, where b s he renvesmen rae. As before, equaon (5) can be rewren n relave erms,.e., as a marke-o-book rao. The cos of capal reflecs he reurn ha he nvesor requres or he long erm rae of reurn on equy. 88

4 In general, he renvesmens of he frms nclude capal expendure and workng capal needs. In he case of a fnancal servce frms, measurng eher of hese ems s problemac snce such frms nves mosly n nangble asses. Consequenly, followng Damodaran (22) we use operang expenses as a proxy for her nvesmens. Our esmaon equaon can be expressed n he followng general way: MV f, k, g BV or (6) MV BV f, k, b. The sgns over he explanaory varables express he expeced dreconaly of he effecs. In he emprcal par of he paper, we shall nroduce an addonal varable Z ha represens a vecor of facors ha affec he MV/BV rao of banks such as regulaon, he srucure of he bankng sysem prvazaon, and macroeconomc and capal marke condons, as a par of he charer value hypohess. Banks and Bank Sysems, Volume 6, Issue, 2 2. Marke value vs. book value n Israel bankng emprcal esmaes for Table. Israel s bankng sysem, fnancal nformaon n NIS bllon, December 29 Asses Percen of oal 2.. Overvew of he Israel s bankng sysem. The Israel bankng sysem s characerzed by a hgh degree of concenraon, refleced by several ndcaors. The larges fve bankng groups accoun for abou 95% of he sysem s asses, loans or deposs (see Table ). The banks n Israel operae as resrced unversal banks: n addon o classc bankng nermedaon, he large commercal banks headng he bankng groups have subsdares ha engage n several acves complemenary o commercal bankng. These acves nclude morgage bankng, cred cards, overseas bankng (va subsdares and branches), drec and ndrec ownershp of companes ha operae n he fnancal and capal markes (such as radng n secures, managemen of nvesmen companes and nvesmen bankng), nsurance companes. The Herfndhal-Hrschman ndex of concenraon (HHI) of he bankng sysem s.2 and s one of he hghes n he wesern world. The bankng sysem holds over half of he asses of he publc and exends more han 7% of oal loans n he economy. Cred o he publc Percen of oal Deposs from he publc Bank Hapoalm Bank Leum Israel Dscoun Bank Uned Mzrah Bank Frs Inernaonal Bank Oher commercal banks * Percen of oal Toal Bankng Sysem Noe: * Include Unon Bank of Israel, Bank of Jerusalem, Indusral Developmen Bank of Israel, Ozar Hahayal Bank Ld, Cbank N.A and HSBC Bank. Sourse: Bankng supervson deparmen, Bank of Israel. Followng he bank shares crss of 983, four of he larges fve banks were naonalzed and her shares delsed from radng on he Tel Avv sock exchange (TASE). As par of he lberalzaon and prvazaon of capal markes polces n Israel, he governmen relsed he shares of he banks for rade on he TASE over he perod of and sold s shares of he banks o he publc. For an updaed dealed analyss of he performance and srucure of he Israel s bankng sysem, see he annual survey of Israel's bankng sysem (29). The excepon was he Frs Inernaonal bank, whch dd no parcpae n he manpulaon of s share prces and, herefore, dd no suffer from he crss. Is shares raded connuously wh no nerrupon Daa summary sascs. In hs par of he paper, we calculae he marke-o-book rao (MV/BV) of Israel s larges fve bankng groups (Bank Hapoalm, Bank Leum, Israel Dscoun Bank, Frs Inernaonal Bank, and Uned Mzrah Bank) beween March 994 and Sepember 25. The bankng group n Israel has a s head a commercal bank (he paren bank) ha owns he subsdares and afflaed companes of he group 2. The shares raded on he TASE are hose of he paren 2 Unl December 3, 23, he repored, amouns n he fnancal saemens were adjused o changes n he CPI. Ths adjusmen of fnancal saemens for nflaon was dsconnued as of January, 24. The adjused amouns ncluded n he fnancal saemen as of December 3, 23 were used as he sarng pon for fnancal reporng henceforh. 89

5 Banks and Bank Sysems, Volume 6, Issue, 2 bank and, hus, reflec he performance of he enre group. The accounng daa used n hs paper s derved from he consoldaed fnancal saemen of he paren bank. The marke-value o book-value rao of bank durng me (MV/BV) s defned as follows: he marke value of bank s he value of he frm s lsed shares and warrans (excludng converble bonds); he marke value also ncludes unlsed shares, whch are valued a he marke prce of he lsed shares; he book value of equy ncludes share capal, premum and capal reserves and reaned earnngs on he consoldaed balance shees of he banks. I should be noed ha he daa n he quarerly fnancal saemens of he banks (as of oher publc frms) are released o he publc wh a lag of hree o four monhs. Therefore, marke value a quarer relaes o nformaon released o he publc a ha me and refers o busness resuls as of he end of he prevous quarer. MV Thus, he rao examned n hs paper s. BV Fgure plos he MV/BV rao of he larges fve banks beween March 994 and Decmber Mzrah Dscoun Hapoalm Leum Frs INTERNATIONAL 3/3/995 3/9/995 3/3/996 3/9/996 3/3/997 3/9/997 3/3/998 3/9/998 3/3/999 3/9/999 3/3/2 3/9/2 3/3/2 3/9/2 3/3/22 3/9/22 3/3/23 3/9/23 3/3/24 3/9/24 3/3/25 3/9/25 3/3/26 3/9/26 3/3/27 3/9/27 3/3/28 3/9/28 3/3/29 3/9/29 Fg.. MV/BV of he larges fve Commercal Banks, 3/994-2/29 As can be observed n Fgure he MV/BV raos of all he banks, whou excepon, have been rendng down durng he perod from January 2 o December 22. The average rao for he sysem fell below.. Noable n hs dagram he posve relaonshp beween MV/BV rao and he level of he frm s profably. In lae 2 and n December 22, reached of.74. I s concevable ha he seep decrease ha happened durng an economc recesson n Israel refleced he deeroraon n he performance of he banks and he expecaons ha commercal banks performance would no mprove n he near fuure. In examnng he developmens of hs rao n he pas sx years (par of he perod are no ncluded n he sample), we see ha from he begnnng of 23 afer hree years nvesors change her assessmen and expeced a subsanal mprovemen n he bank s favorable resuls n 24 and 25. Ths was reflecs by a connuous rse n he MV/BV rao for all banks durng hose years. The declne n he MV/BV rao durng 28 derved manly from he developmens and shocks n he global fnancal sysem n he real crss whch resuled from and mpaced he Israel economy as well. The average MV/BV rao of he fve large commercal banks ncreased n 29, crossng he hreshold of uny, and endng he year a.9 as agans.56 n Emprcal esmaon of he facors deermnng he MV/BV rao n bankng. In hs secon, we esmae he deermnans of he MV/BV rao usng a mulvarae regresson analyss. The heorecal bases for he emprcal analyss are presened n equaon (6). We performed a logarhmc ransformaon on all varables (x) wh he excepon of he varables ha can have a negave value, on whch we performed a ln(+ x) ransformaon. Followng are he specfcaon of he varables used n our analyss. The dependen varable s defned as follows: MV/BV he MV/BV rao of bank n quarer. Because, as we have already ndcaed, here s a lag of approxmaely hree monhs n he publcaon 9

6 Banks and Bank Sysems, Volume 6, Issue, 2 of he fnancal saemens whch are relaed o BV, we calculaed he rao ( MV /BV ). The ndependen varables and her hypoheszed effec on MV/BV are defned as follows: he reurn on equy of bank, durng he precedng four quarers (a year). We expec a posve relaonshp beween and he MV/BV rao of he bank (for llusraon, see Fgure 2). (MV/BV) % MV/BV Fg. 2. (MV/BV)B rao vs. annual reurn on equy () of he fve larges commercal banks, 6/995-2/29 The rsk of equy was esmaed usng wo approaches. Frs, overall, rsk s decomposed no wo major ypes of rsk: cred and marke (neres, nflaon, and exchange rae) rsks. The rsks were esmaed usng accounng daa. Second, by esmang he sysemac rsk based on he capal asse prcng model (CAPM): k j R f = j (R m R f ), where he excess of he requred rae of reurn on equy k over he rsk-free rae R f s deermned by he rsk premum whch equals he produc of he marke prce of rsk (R m R f ) and he sysemac rsk bea (), where R m s he expeced rae of reurn of he marke. For cred rsk, we used hree alernave measures: L / GDP he bank-cred o GDP rao. Ths rao s defned as oal cred o he publc exended by bank n quarer, dvded by he cumulave GDP n he precedng four quarers. As an alernave o hs varable, we examned wo addonal cred-rsk varables: he annual (he precedng four quarers) loan-loss provsons dvded by oal cred a quarer LLP / L and he rskweghed asses/oal asses rao n quarer RWA. All hese varables examne he qualy of bank cred, he hgher hey are, he lower s he qualy of cred (he lower s he repaymen ably of he borrowers or he hgher s he probably of defaul), and vce versa. Hence, hs group of varables s expeced o have a negave nfluence on he MV/BV rao. For marke rsk, we used hree alernave esmaes: he mpled sandard devaon of he NIS / $ NIS/dollar exchange rae. The sandard devaon was calculaed by usng he Black-Scholes formula for hree-monh call opons on he NIS/dollar exchange rae raded on he TASE; p e expeced nflaon based on he capal marke esmaon. The esmaon s done by akng he rao of yeld o maury of nonlnked governmen bonds (normally TB s) o he yeld o maury of CPI lnked governmen bonds wh equal me o maury. For ha purpose, we used 2 monhs nervals; p e sandard devaon of expeced nflaon, usng 2 monhs nervals. As explaned before he marke rsk varables are expeced o have a negave effec on he MV/BV rao. For he esmaon of sysemac rsk, we used hree alernave measures: bea () ha was esmaed by an OLS regresson of he monhly sock reurns (R ) on he monhly reurns on he TASE (value weghed ndex of he larges cap socks) (R m ) over a 24 monhs movng wndow: R = + R m + e ; he rsk premum j (R m R f ), where he marke prce of rsk (R m R f ) was esmaed as a 2 quarers' average (3 year); R m was calculaed as a geomerc average over 2 quarers and R f s he yeld o maury on a one-year reasury bll a each pon n me ; As s common n emprcal sudes of he CAPM he ex-ane expeced rae of reurn of he marke s replaced by an ex-pos average reurn. 9

7 Banks and Bank Sysems, Volume 6, Issue, 2 he requred rae of reurn k j = R f + j (R m -R f ), where each varable s defned above. As before, we expec all hree sysemac rsk varables o have a negave effec on he (MV/BV) rao. For renvesmen polcy, we used wo alernave varables:. b he reenon rao defned as one mnus he payou rao, he payou rao s measured as he average dvdends pad by bank n he calendar year dvded by he earnngs (avalable for dsrbuon) n ha quarer. Ths rao should reflec he dvdend polcy as declared by he bank for he upcomng calendar year. 2. b 2 renvesmen rao defned as he rao of operang expenses dvded by ne fnancal ncome plus operang ncome n quarer ; hs rao s also known as he effcency rao. The hypohess here s ha a hgher renvesmen resuls n greaer growh and, hence, a hgher MV/BV rao. For he economc facors, we used wo groups of varables:. To measure macroeconomc condons we used quarerly changes (n annual erms) n GDP (GDP), or busness secor GDP (GDP BS ) or changes n he compose sae-of he- economy ndex (CEI) ; 2. To measure capal marke condons we used he varable (MV/BV) NB defned as he MV/BV rao of all non-bank frms raded on he TASE n quarer. Ths varable allows us o neuralze he facors ha affec he sock marke n general. The hypohess s ha a rse n economc acvy and n he capal marke leads o a hgher MV/BV rao n bankng. The hypoheses rely on he assumpon ha durng economc expansons or alernavely durng bull markes he charer value of he banks ncreases. The compose sae-of-he-economy ndex s a synhec cyclcal ndcaor for examnng he drecon n whch real economc acvy s movng n real me. The ndex s calculaed from he monhly changes n fve componens ha reflec dfferen aspecs of real economc acvy: he ndex of manufacurng producon; mpors, excludng capal goods; rade and servces revenue; he number of employee poss n he busness secor; and goods expors (excludng agrculure, fuel, damonds and shps and arcraf). For he srucure of he bankng sysem, we used wo alernave varables:. HHI he Herfndhal-Hrschman ndex of concenraon of he bankng sysem, we measured hs ndex usng oal asses. 2. S he marke share of bank n he oal asses of he bankng sysem. The assumpon behnd hese wo varables s ha a hgher degree of concenraon leads o less compeon and, hence, an ncrease n he charer value of he bank and, subsequenly a rse n s MV/BV rao. For regulaory facors, we used wo dummy varables:. DR a dummy varable wh a value of for perods afer he mnmum requred capal rao has been rased by he Bank of Israel from 8% o 9% (n March 999), and zero oherwse. 2. DP a dummy varables for he prvazaon of he banks, wh a value of for he perods afer he prvazaon and zero oherwse. The regresson ha we ran for each bank was as follows: MV/BV marke rsk ( macroeconomc facors) ( MV/BV ) ( bankng srucure) ( cred rsk ( renvesmen polcy) 3 ( regulaon) 7 5 ) NB. Accordng o our analyss he hypoheszed sgns of he coeffcens n equaon (9) are: ; ; ; ; ; ; ; The second regresson where sysemac rsk was subsued for cred and marke rsk s: ( MV / BV) ( renvesmen 2 ( macroeconomc facors) 3 ( MV / BV) 4 ( bankng srucure) 5 4 ( sysemac rsk) NB polcy) ( regulaon) Herenafer, we analyze he regresson resuls obaned for each bank (each for he relevan perod durng whch s shares were raded n he TASE). 92

8 93 Independen varables Inercep (C) Reurn on equy () Cred rsk: Toal cred / GDP (L/GDP) Rsky asses / oal asses (RWA) Loan loss provson / oal cred (LLP/L) Marke Rsk: S.D of expeced nflaon (pe) Expeced nflaon (pe) S.D NIS / dollar ( NIS / $) Sysemac rsk k = Rf + (Rm-Rf) (Rm-Rf) Table 2. Regresson resuls of (MV/BV) for he larges fve banks (quarerly daa) Bank Hapoalm Bank Leum Israel Dscoun Bank Mzrah Bank Frs Inernaonal a b c a b c a b c a b c a b c -.46 (-3.7)* 7.98 (3.5)* -.72 (-.94)*** -.27 (-3.55)*.65 (2.5)** -.6 (-.78)*** -.7 (-.3) -.72 (-2.58)** 8.2 (.84)*** -.7 (-.97) 2.9 (.68) 7.5 (6.69)* (-2.73)** -.38 (-5.5)*.99 (6.29)* -.37 (-.69)*** -.4 (-.74)*** -.68 (-6.24)* 3.47 (7.34)* -.32 (-2.22)** -.6 (-.66)***.9 (.4).4 (3.76)* -.6 (-2.22)** -. (-2.34)** 4. (.53) 9.6 (7.58)* -2.5 (-.74)*** 2.39 (.3) 5.6 (2.28)** -2.5 (-4.65)* 2.72 (.3) 7.45 (3.44)* -.89 (-2.63)** -.7 (-.99)*** -.66 (-.52) 5.62 (.68)*** -.3 (-.9).79 (.25) 7.62 (3.36)* -3.5 (-2.59)** Noe: -values appear n parenheses under each coeffcen. * ndcaes % sgnfcance. ** 5 % sgnfcance. *** % sgnfcance. -.9 (-4.7)* 7.6 (.97)*** -.5 (-2.9)* -.7 (-2.34)** -.57 (-3.4)* 3.46 (3.8)* -.33 (-.88) -.3 (-4.4)* 9.83 (2.92)* -.8 (-2.73)** -.8 (-6.93)* 9.9 (3.25)* -. (-.59) (-5.55)* 8.4 (3.)* -.97 (-5.36)* -.8 (-2.23)** -.37 (-5.88)* 5.4 (.85)*** -.5 (-2.95)* -.23 (-2.)** -.7 (-3.43)* 4.74 (.99) -.7 (-2.46)** -.5 (-5.49)* 9.52 (4.5)* (-2.3)*** Banks and Bank Sysems, Volume 6, Issue, 2

9 94 Renvesmen: Reenon rao Effcency rao Independen varables Macroeconomc condons: Changes n GDPBS Changes n GDP Compose economc ndex Marke value / book value of non-bank frms (MV/BV)NB Bankng srucure facors: Herfndhal-Hrchman ndex (HHI) Marke share (S) Regulaon polcy: Dummy varable - DR Prvazaon polcy: Dummy varable DP MA() Table 2 (con.). Regresson resuls of (MV/BV) for he larges fve banks (quarerly daa) Bank Hapoalm Bank Leum Israel Dscoun Bank Mzrah Bank Frs Inernaonal a b c a b c a b c a b c a b c (.96) *** (2.6) ** (.74) *** (.97) *** (.98) *** (2.62) ** (3.3) * (2.64) **.22 (.87) *** (2.6) ** (2.) ***.24 (.73).36 (.86) *** (2.55) ** (2.95) * (3.5) * (2.7) ** (.56) (3.66) * (3.83) * (2.3) ** (2.) *** (2.29) ** (2.4) ** (2.97) * (2.5) ** (2.53) ** (3.5) * (2.98) *.7 (.84) ***.89 (.78) *** (8.) * (2.83) * (3.82) * (5.8) * (6.2) * (6.62) * (2.92) * (6.2) * (3.57) * (4.58) * (5.53) * (4.) * (4.3) * (5.4) * (5.69) * (2.58) ** (2.96) * (3.94) * (2.29) ** (2.86) * 2.54 (2.5) *** (.75) *** (.73) *** (.65) 3.95 (2.97) *.69 (2.8) **.8 (.37) (-2.79) * (-2.83) * (-4.99) * (-2.53) ** (-3.83) * (-3.7) * (-3.79) * (-4.63) * (-4.44)* (3.89) * (.7) *** (3.5) * (2.63) ** (2.79) *.97 (28.9) * (-2.97) * (-2.4) ** (-.92) ***.4 (3.8) *.57.3 (4.4) * (7.9) * Adjused R Banks and Bank Sysems, Volume 6, Issue, 2

10 Banks and Bank Sysems, Volume 6, Issue, 2 The man resuls, presened n Table 2 are:. was found, as expeced, o have a posve and sascally sgnfcan effec on MV/BV rao n all of he fve major banks. Ths fndng ndcaes ha nvesors n bank shares ake no accoun he profably of he bank when hey value he bank. 2. Specfcally, he coeffcens whch reflec elascy of (MV/BV) wh respec o (.e. ln( MV / BV ) ) vary beween 5.4% and 9.5% ln( R) n Frs Inernaonal. 3. Wh respec o cred rsk, whch s consdered o be he major ype of rsk faced by he banks, we used n equaon (7) hree alernave varables o represen hs rsk; L/GDP, RWA and LLP/L, all of whch were defned prevously. Generally, all of he hree varables were found, as expeced, o have a sgnfcan negave mpac on (MV/BV). The meanng of hese resuls s ha nvesors n bank shares beleve ha cred rsk aken by he bank reduces he aracveness of he shares whch s refleced n s MV/BV rao. Ths fndng s conssen wh he mplcaon of he effcen marke hypohess ha he nverse of he MV/BV rao s a proxy for rsk. 4. Conrary o cred rsk, marke rsk as refleced n p e, e or P NIS/$ proved o conrbue o MV/BV only n some banks and he sandard devaon of he foregn currency ( NIS/$ ) proved o be he bes of all (wh he excepon of Bank Hapoalm n whch dd no conrbue a all). A possble explanaon of hese resuls s ha he ne exposure (afer hedgng) of he banks o marke rsks s relavely small. 5. The sysemac rsk, whch replaced cred rsk and marke rsk and proved o be he mos sgnfcan, s he cos of equy (k). I proved o nfluence he MV/BV rao negavely n hree Because of negave values of durng some perods, we used ln(+), he elasces of MV/BV w.r. ( ) s calculaed as follows: ln( MV/BV) ln( MV/BV) ln ( ) ln( ) ln ( ) ( ) ( ) ln( ) ( ),, ( ) where he bar above he varable ndcaes he average of he varable durng he analyzed perod. Accordng o hs formulaon, he elasces vared beween.45 n Bank Dscoun o.46 n Bank Leum. banks (Hapoalm, Leum and Dscoun Bank). In one bank (Frs Inernaonal), he rsk premum ( (R m R f )) urned ou o be negavely sgnfcan. These fndngs are conssen wh he resuls obaned whle decomposng he rsk no cred o marke rsk. 6. The renvesmen varables proved o affec he MV/BV rao n all banks. Beween he wo varables used o represen hs group of varables, he reenon rao proved o be superor. Apparenly, he dvdends pad by he banks nfluenced he decsons by he nvesors n bank sock and, hus, effeced he MV/BV rao of he bank. As menoned earler n he paper, he renvesmen rao ogeher wh he reurn on equy of he bank deermne s growh paern (.e., g = b ). Thus, he fndng ha boh and b were found o be sgnfcan s of mporance and suppors he valdy of he dvdend growh model. 7. The nfluence of macroeconomc condons as refleced by changes n hree possble ndces of economc acves proves o be posve and sgnfcan. One canno say caegorcally ha one ndex s superor o he oher, snce all of hem had a posve and sascally sgnfcan effec on (MV/BV). Ths fndng suppors he hypohess ha he charer value of banks ncreases durng perods of economc expanson (Saunders and Wlson, 2). 8. As can be seen he compeon n he bankng ndusry as measured by several bankng srucure facors, had hardly any effec on (MV/BV). Ths s probably due o he hgh concenraon of he bankng ndusry and he small number of major banks n Israel. 9. In all banks, we found a sascally posve nfluence of capal marke condons, as measured by he rao (MV/BV) NB, on he MV/BV rao of he banks. The values of he coeffcens/elasces vary beween.7 (for he Frs Inernaonal Bank) o.74 (for Bank Hapoalm). The exsence of a coeffcen smaller hen one may ndcae ha he bank socks behave more conservavely han he res of he socks raded on he Tel Avv exchange.. The regulaory nfluence, represened by a dummy varable ha creaes dsncon beween he perod pror o he ncrease n he mnmum capal adequacy rao (March 999) from 8% o 9% and he perod afer, was found o have a sgnfcan negave mpac on MV/BV n four of he banks examned. Apparenly, he requremen by he bank of Israel o rase he mnmum capal adequacy rao was nerpreed as a measure desgned o cope wh a hgher rsk faced by he banks and mposed a hgher cos of equy fnancng on he banks. 95

11 Banks and Bank Sysems, Volume 6, Issue, 2. The varable ha s supposed o capure he effec of governmen polcy s measured by he prvazaon of banks durng he perod of The dummy varable chosen for ha purpose had proved o be sgnfcanly posve only for Bank Hapoalm for whch we had suffcen pre- and pos-prvazaon daa. Apparenly, he marke expeced ha prvazaon would mprove he performance of he bank n he fuure because of an ncrease n operang effcency and/or because of more pruden rsk managemen hen n he pas. 2. The adjused R 2 2 ( R ) were relavely hgh n all banks (beween 63% and 84%) and here was no evdence of seral correlaons n mos regressons. I should be noed ha whenever seral correlaon exsed n he regressons, we used a movng average or auoregressve error process of he frs degree o correc for. In hose cases, we added MA() or AR() erm o he regressons, he coeffcens of whch appear n Table 2. In he second sage of our emprcal analyss, we used a poolng mehod by combnng cross secon and me seres daa coverng he perod from he second quarer of 998 o he hrd quarer of 25. We use hs approach o augmen he above analyss ha reled on me seres for each bank separaely, n lgh of small number of banks n he Israel s bankng sysem. The resuls we obaned usng a poolng process s summarzed n Table 3. Table 3. Regresson resuls of (MV/BV), usng poolng process, he larges fve banks Independen varables Inercep (C) Profably Reurn on equy () Toal cred / GDP (L/GDP) Cred rsk Rsky asses / oal asses (RWA) Loan loss provson / oal cred (LLP/L) Marke rsk S.D NIS / dollar ( NIS / $) Sysemac rsk "Reenon rao" Renvesmen "Effcency rao" Changes n GDPBS Macroeconomc condons Changes n GDP Compose economc ndex Marke value / book value of non-bank frms Capal marke condons (MV/BV)NB Herfndhal-Hrchman ndex (HHI) Bankng srucure facors Marke share (S) Regulaon polcy Dummy varable DR Dummy varable for me DT: 22 DB: Bank Leum DB: Israel Dscoun Dummy varable for bank DB: Mzrah Bank DB: Frs Inernaonal (-.78) (-7.59) * (.3) -.38 (-6.52) * (7.94) * (6.52) * (5.42) * (8.8) * -. (-5.83) * -.26 (-4.94) * -. (-2.5) ** (-3.34) * (-3.46) * (-3.28) *.4 (.96) ***. (.3).22 (4.92) *.4 (.48).2 (3.3) * -. (-.66) *** (.98) ** (2.3) ** (3.33) * (6.29) * (6.) * (3.55) *.29 (4.34) * 2.5 (.95) *** (-.76) *** (-4.32) * (-4.56) * (-4.8) * (2.77) * (3.44) * (2.4) ** (2.2) ** (-4.88) * (-5.25) * (-2.54) ** (-2.4) ** -.26 (-.43).25 (.77).5 (.43) (-6.6) * (-2.46) ** (-.6) (-6.9) * (-3.89) * (-2.5) ** (-6.73) * (-3.92) * (-.97) *** 96

12 Banks and Bank Sysems, Volume 6, Issue, 2 Table 3 (con.). Regresson resuls of (MV/BV), usng poolng process, he larges fve banks AR() Independen varables (3.27) * (5.) * (4.96) * Adjused R Noe: -values appear n parenheses under each coeffcen. * ndcaes % sgnfcance. ** 5 % sgnfcance. *** % sgnfcance. The regresson resuls of hs process are smlar o he ones obaned usng me seres daa for each of he fve large banks separaely. The varables reurn on equy (); cred rsk (measured eher by he rao of loan-loss provsons o oal cred LLP/L or RWA or L/GDP); marke rsk (measured by he sandard devaon of NIS/$); sysemac rsk (represened by of each bank); economc facors (measured eher by changes n GDP or n GDP BS, or n CEI); capal marke facors (measured by (MV/BV) NB ); and he bankng marke srucure (measured by concenraon ndces HHI or S ). All obaned he hypoheszed sgns and were found o be sascally sgnfcan. Wh respec o he bankng srucure facor (measured by he H ndex of he ndusry), we found, as hypoheszed, ha he more concenraed he bankng sysem, he greaer he value of MV/BV rao, probably reflecng a hgher charer value of he banks. I should be noed ha n he prevous me seres analyss, usng each bank separaely, we dd no fnd a sgnfcan mpac of hs facor on he MV/BV rao. The possble explanaon for he dfferen resuls s he shorer me perod of he poolng daa (998-25) durng whch here was a sgnfcan ncrease n concenraon of he sysem. As can be seen, he varables represenng he renvesmen polcy (as measured eher by he effcency rao or by he reenon rao) were sgnfcan only n one regresson, whereas he dummy varable represenng he regulaory decson o rase he mnmum capal rao from 8% o 9% was found o be sgnfcan n all four regressons. In addon o he basc varables we added wo ses of dummy varables dsngushng beween he dfferen years as well as beween he dfferen banks. For hs purpose, we defned dsngushng beween wo years sub-perods: pror o 22, years characerzed by a declne n he MV/BV rao and afer. For he purpose of dsngushng beween banks, we dropped Bank Hapoalm and, hus, all oher banks are measured relave o ; n oher words, Bank Hapoalm served as a reference bank for ha purpose. The resuls obaned for he me dummy varables ndcae ha, ndeed, 22 was he lowes year n erms of he MV/BV rao of he sysem and he resuls of he cross secon dummy varables ndcae ha Bank Hapoalm had he hghes MV/BV rao among all he banks n he sample. These fndngs are suppored by he behavor of he MV/BV rao over me and among banks (see Fgure ). 2 Agan, here, we obaned a relavely hgh R and here was no seral correlaon n he daa (n some cases we had o add AR() n order o correc for seral correlaon). Conclusons and polcy mplcaons In hs paper, we developed, based on economc and fnance heory, a model for he deermnaon of MV/BV rao of a busness frm and adoped o a bankng frm. Usng Israel s bankng daa, we esed he relaonshp beween varous explanaory varables and he MV/BV raos of he fve larges banks n Israel durng he perod of usng me seres and pooled daa. The wo mos mporan groups of varables ha explaned ner varaon n MV/BV over me and across banks, are he rsk (measured by eher accounng daa or relyng on marke values based on he CAPM model) and he reurn on equy. The combned, opposng, effec of rsk and reurn on he MV/BV raos of banks, rases he mporance of rsk adjused performance measures of frms parcularly of banks, such as RAROC (see Landskroner, Ruhenberg and Zaken, 25). The fndngs of hs paper have polcy mplcaons. They can serve bank managemen and nvesors n bank socks n her decson-makng process, once hey know ha a combnaon of reurn and rsk deermnes he fuure values of he bank as refleced n s MV/BV rao. Our fndng suppors he effcen marke hypohess ha he MV/BV rao reflecs rsk of equy. We also fnd a posve relaonshp beween charer value and economc acvy, hs suppors he charer value hypohess n bankng and has mporan mplcaons for regulaon of banks. As deregulaon ncreases compeon and reduces charer value. On he oher hand, reduced governmen nvolvemen n capal markes and fnancal nermedaon, ncreases charer value. 97

13 Banks and Bank Sysems, Volume 6, Issue, 2 References. Bank of Israel, Bankng Supervson Deparmen, Israel s Bankng Sysem Annual Survey, Chan, L.K., Y. Hamao, and J. Lakonshok (99). Fundamenals and sock reurns n Japan, Journal of Fnance, 49, pp Damodaran, Aswah 22, Invesmen Valuaon, Wley Fnance, 2 nd ed., pp Fama, E.F., and K.R. French (992). The cross-secon of expeced sock reurns, Journal of Fnance, 47, pp Fama, E.F., and K.R. French (995). Sze and book-o-marke facors n earnngs and reurns, Journal of Fnance, 5, pp Fama, E.F., and K.R. French (996). Mulfacor explanaon of asse prcng anomales, Journal of Fnance, 5, pp Fama, E.F., and K.R. French (998). Taxes, fnancng decsons and frm value, Journal of Fnance, 53, pp Fama, E.F., and K.R. French (26). Profably, nvesmen and average reurns, Journal of Fnancal Economcs, 82, pp Keeley, M.C.(99). Depos nsurance, rsk and marke power n bankng, Amercan Economc Revew, 8 (5), pp Kohar and Shanken (997). Book-o-marke, dvdend yeld, and expeced marke reurns: a me-seres analyss, Journal of Fnancal Economcs, 44, pp Lakonshok, J., A. Shlefer and R. Vshny (994). Conraran nvesmen, exrapolaon and rsk, Journal of Fnance, 49, pp Landskroner, Y.D. Ruhenberg and D. Zaken (25). Dversfcaon and performance n bankng: he Israel Case, Journal of Fnancal Servces Research, 27, pp Maroney, N. and A. Proopaoadaks (22). The book-o-marke and sze effecs n a general asse prcng model: evdence from seven naonal markes, European Fnance Revew, 6, pp Peerkor, R., and J. Nelsen (25). Is he book-o-marke rao a measure of rsk?, Journal of Fnancal Research, 28, pp Saunders, A., and B. Wlson (2). An analyss of bank charer value and s rsk consranng ncenves, Journal of Fnancal Servces Research, 9 (2/3), pp Vassalou, M. and Y. Xng (24), Defaul rsk n equy reurns, Journal of Fnance, 59, pp