Understanding the Cash Flow-Fundamental Ratio

Transcription

1 Inernaional Journal of Economics and Financial Issues Vol. 5, No., 05, pp ISSN: Undersanding he Cash Flow-Fundamenal Raio Chyi-Lun Chiou Deparmen of Business Adminisraion, Fu Jen Caholic Universiy, No. 50, Zhongzheng Rd., Xinzhuang Dis., New Taipei Ciy, 405, Taiwan, R.O.C. TEL: ABSTRACT: This aricle invesigaes he use of cash flow-fundamenal raio in forecasing sock marke reurn and examines implicaions behind his raio. By presuming he dynamics of cash flow-fundamenal raio I idenify he relaionship beween economic uncerainy and risk premium. The evidence shows ha cash flow-fundamenal raio is procyclical and is a predicor of cash flow growh and excess reurns. The cash flow-fundamenal raio is proved o be negaively associaed wih risk premium. I also examine ha he mean-reversion propery of cash flow-fundamenal raio is riggered by profiabiliy. In conras o he assumpion of saionary in sock price, mean reversion in profiabiliy is more reasonable and has been proved by Fama and French (000). Keywords: Predicabiliy of sock reurn; Cash flow-fundamenal raio. JEL Classificaions: G. Inroducion The predicabiliy of sock reurns is one of he core issue in financial economics. Previous lieraure (e.g., Campbell and Shiller, 989, 00; Fama and French, 988; Lamon, 998; Vuoleenaho, 000; Lewellen, 004) shows ha financial raios based on raional expecaions such as dividend yields, earning yields, and book-o-marke raios have forecasing power for he sock reurns. Several oher sudies aribue he predicabiliy of reurns o irraional movemens in sock prices (e.g., DeBond and Thaler, 985). These argumens share one common feaure ha predicabiliy is driven by he mean-reversion of sock prices. However, wha causes his mean reversion is ambiguous. Furhermore, he sabiliy of a financial raio iself is ofen challenged. To fill his gap Jiang and Lee (007) propose a new indicaor based on a loglinear combinaion of dividend yields and book-o-marke raios in deecing reurns profiabiliy. They demonsrae ha his new indicaor no only is saionary bu also is able o predic fuure sock reurns. Alhough hey show his indicaor is superior o dividend yields or book-o-marke raios alone in forecasing ess, he explici informaion or inuiion behind his indicaor is unclear. In his paper I propose a valuaion framework based on he cash flow-fundamenal raio o invesigae he new indicaor proposed by Jiang and Lee (007). I presume ha he cash flow-fundamenal raio is mean-revering and governed by economic uncerainy. Accordingly, I find ha he cash flow-fundamenal raio is associaed wih fuure profiabiliy and excess sock reurns. I invesigae ha a non-price financial raio is capable of predicing sock reurns when i has close relaionship wih economic uncerainy. There are wo feaures embodied in he cash flow-fundamenal raio. Firs, he cash flow-fundamenal raio forecass reurns because i conains informaion abou ime-varying economic uncerainy (condiional volailiy of consumpion). The evidence shows ha realized consumpion volailiy is prediced by he cash flow-fundamenal raio for long horizons. A higher cash flow-fundamenal raio implies less uncerainy in he fuure, and hence, a lower risk premium is compensaed. Second, he cash flow-fundamenal raio forecass reurns because cash flow growh While he former inerpreers ha he mean-reversion is due o raional ime-varying discoun raes, he laer saes his mean-reversion is caused by he correcion of marke overreacion. 48

2 Inernaional Journal of Economics and Financial Issues, Vol. 5, No., 05, pp is associaed wih his raio. Evidence shows ha expeced cash flow growh can be well prediced by he cash flow-fundamenal raio. A lower cash flow-fundamenal raio follows a higher cash flow growh rae. Alhough cash flow growh can be well prediced by cash flow-fundamenal raio a shor and medium horizons, his raio has very lile predicable power in book equiy growh a he same horizons. I is argued ha he predicive power of financial raios relies on he mean-reversion propery hey hold. Valuaion raios based on prices such as dividend yield are capable of predicing fuure reurns when sock prices are predicable. Referring o Campbell and Shiller (00), he sabiliy of a valuaion raio implies somehing mus be predicable based on he indicaor, eiher he numeraor or he denominaor. The evidence shows ha he cash flow-fundamenal raio can predic fuure cash flow growh in a significan way. Based on he equilibrium in he framework, my resuls are also consisen wih he raional pricing sory ha he cash flow-fundamenal raio capures informaion abou he risk premium. My conribuion is wofold. Firs, my framework idenifies he relaionship beween dividend changes and risk changes in aggregae level. In he radiional valuaion model, he fundamenal news abou a sock valuaion has wo componens: cash flows and discoun raes (risk characerisics). Curren dividend-signaling models sugges ha dividend increases reveal good news abou fuure cash flows. However, empirical evidence provides lile implicaions abou such a predicion (e.g., Allen and Michaely, 00). An alernaive explanaion proposed by Grullon e al. (00), named he mauriy hypohesis, hen argues ha dividends convey informaion abou changes in risks, raher han abou cash flow growh. I examine ha when he cash flow-fundamenal raio rises because of he decrease in uncerainy, expeced excess reurns and cash flow growh decline. My framework verifies ha if good news abou dividend increases is no abou an increase in fuure cash flows, hen i migh relae o a decline in sysemaic risks. However, he similar logic canno be applied o book-o-marke raios or dividend yields. 3 Hence, i is no appropriae o examine he relaionship beween dividends and reurns by means of dividend yields or book-o-marke raios. Second, I show ha he predicabiliy of cash flow-fundamenal raio is irrelevan o he mean-reversion of sock prices as argued in dividend yields and book-o-marke raios. In general, changes in dividend yields or book-o-marke raios can reflec boh changes in discoun raes and cash flow growh; however, he relaive imporance of hese wo componens is no quie clear. As documened by Bansal and Yarson (004), an increase in economic uncerainy raises risk premium, as well as dividend yield. Since cash flow growh is hard o be prediced by his raio (e.g., Menzly e al., 004; Leau and Wacher, 007), he predicive power of dividend yield ies o he mean-reversion in sock prices. 4 By conras, book-o-marke raio predics fuure profiabiliy wih a negaive relaion and forecass excess sock reurns wih a posiive relaion (Vuoleenaho, 000). Alhough book-o-marke raio can forecas profiabiliy and sock reurns, he relaionship beween economic uncerainy and book-o-marke raios is no clear. The superioriy of he cash flow-fundamenal raio relies on he feaure ha i can be relaed no only o fuure profiabiliy and excess sock reurns bu also o economic uncerainy. This connecion is criical because i is useful o relae profiabiliy and sock reurns wih business condiions. In brief, my work is consisen wih raional expecaions, as shown in Campbell and Shiller (989). The remainder of he paper is organized as follows. In he nex secion, I propose he valuaion framework based on he cash flow-fundamenal raio and he corresponding equilibrium in our model. Secion 3 elaboraes my empirical seing, including he descripion of daa and empirical resuls o suppor asse pricing implicaions of our framework. The las secion provides concluding commens. Campbell and Vuoleenaho (004) propose a wo-bea model ha splis a sock s risks in wo loadings, cash flows and discoun raes. 3 See Campbell and Shiller (989) and Vuoleenaho (000). 4 Some lieraure finds oher significan indicaors o predic dividend growh. For example, Menzly e al. (004) propose ha he relaive share, he share of consumpion each asse produces, is a good predicor of dividend growh a he aggregae level. Leau and Wacher (007) oherwise idenify ha consumpiondividend raio has good predicive power for dividend growh especially a long horizons. 49

3 Undersanding he Cash Flow-Fundamenal Raio. The Valuaion Framework In his secion, I consider a represenaive agen who chooses consumpion level o maximize an expeced power uiliy funcion in which an explici cash flows dynamics, a dividend yield wih respec o book value, is inroduced. I find his indicaor has some criical asse pricing implicaions. I consider an economy populaed by a represenaive agen who maximizes an expeced power uiliy funcion of he form: s C s e ds, () where C is he aggregae consumpion level a ime, is he coefficien of consan relaive risk aversion, and is he subjecive discoun rae. The aggregae wealh he agen holds is consruced by financial and nonfinancial asses. Hence, he aggregae consumpion is financed by hese wo sources of income. I assume he agen is iniially endowed wih one share of a sock which pay dividends D in he form of he consumpion good and is book equiy is worh B. The ime-varying insananeous cash flows generaed is hen defined as dividends wih respec o book equiy: D F exp X, () B where X is assumed o follow a simple mean-reversion process:, dx X X d X db (3) in which is he speed of mean reversion, is some posiive, and B is a sandard Brownian moion. Given ha and X are posiive, he value of X is nonnegaive. This nonnegaive propery hen ensures ha he cash flow-fundamenal raio is always beween zero and one. Because of he mean-reversion naure of he dynamics, he cash flow-fundamenal raio converges o a seady-sae value as. Nex, I se he book equiy-consumpion raio, one par of aggregae wealh-consumpion raio exogenously as, B Y, C (4) where Y evolves as follows, dy X db. (5) Y Here, is some posiive and B is anoher sandard Brownian moion independen o B. In addiion, we specify he aggregae consumpion follows he process: dc gd X C db3, (6) where 0 and B 3 is also a sandard Brownian moion independen o B. I presume, however, he correlaion beween he wo Brownian moions, B and B 3, is d. Before moving on, I have o sae why he dividend-book equiy raio is relaed o he dynamic of consumpion. I assume consumpion is composed of labor income and financial income. Furhermore, I assume ha he former sands for expeced consumpion, while he laer relaes o unexpeced consumpion change. My inuiion is ha he payou level, proxy by dividend-book raio in his work, is posiively relaed o unexpeced consumpion growh. Bansal and Yaron (004) assume ha boh he volailiy of consumpion growh and he volailiy of dividend growh are ime-varying and governed by he same economic uncerainy. Insead of direcly modeling dividend process, I examine he dividend-book raio process o help us o idenify he dividend payou policy. When he economy conains more uncerainy, boh consumpion and dividend will be expeced o grow a higher speed in he near fuure; however, his uncerainy also makes he dividend payou

4 Inernaional Journal of Economics and Financial Issues, Vol. 5, No., 05, pp say a a lower level now. Besides, I assume ha 0. The inuiion is as follows. When a shock forces he economy become less uncerainy oday, his shock also commis a posiive innovaion o consumpion and dividend a he same ime. Less uncerainy follows a higher dividend payou level, dividend level and consumpion. However, he expeced dividend growh rae becomes less aracive because of he lower reained earnings. Meanwhile, people also expec a lower excess reurn in he near fuure because of fewer growh opporuniies. An applicaion of Io s lemma implies he expeced growh rae of book value and dividends, db E gd, (7) B dd E g X X d. D (8) From equaion (8) i is clear ha he cash flow-fundamenal raio is negaively relaed o he cash flow growh if 0. This implies ha if corporae dividend is procyclical, expeced cash flow growh decreases wih he cash flow-fundamenal raio. In addiion, from equaion (6) and (7), we noe ha expeced consumpion growh and book equiy growh are boh consan. I implies ha boh consumpion and book equiy are no predicable in my framework. Appling he Euler equaion, in equilibrium he price of he sock saisfies, s Cs P E e D sds C Here, and a, s is given by, and where, s Cs E e C sfsy sds C s a, s 0,. B e a s F ds s a0, s exp g Xa, u du, (0) a, s, s e., By simple calculaion he process for excess sock reurns implied by he model can be expressed as, H X, () R where H is he elasiciy of marke-o-book raio wih respec o F. Equaion () shows ha if he corporae dividend is procyclical, equiy risk premium is negaively relaed o he cash flow-fundamenal raio, as documened by Jiang and Lee (007). This resul is conrary o prior evidences ha high dividend yields predic high excess reurns (e.g., Campbell and Shiller, 989; Fama and French, 988; Bansal and Yaron, 004). I will discuss heir difference explicily in he nex secion. Equaion () also shows ha he equiy premium specified in my work is composed of wo ingrediens. The firs erm reflecs he covariance beween consumpion growh and percenage changes in he cash flow-fundamenal raio. This erm implies ha he ime-varying propery of he (9) () 5

5 Undersanding he Cash Flow-Fundamenal Raio risk premium is conribued o he ime variaion in he cash flow-fundamenal raio, given ha he correlaion coefficien is no zero. The second erm represens he consumpion risk premium as described in many consumpion-based pricing models. This erm also idenifies he ime-varying propery of discoun raes. I shows ha he cash flow-fundamenal raio is negaively relaed o discoun raes. 3. Daa and Model Implicaions In his secion, I provide evidences abou he asse marke implicaions of my framework described in he las secion. In his work, I employ he dividend-o-book equiy raio as he proxy for cash flow-fundamenal raio. Firs I have o idenify his raio in an appropriae way and hen provide some implicaions behind i. In heory, i is sraighforward o define he dividend-o-book equiy raio as aggregae dividends divided by aggregae book equiy. However, i may no be appropriae o esimae his raio by using he repored daa, for reasons described below. Boh aggregae dividends and book equiy, in pracice, should be modified. 3.. Daa Descripion I collec daa for he S&P 500 index from Shiller (989) and CRSP for he sample period of , including book equiy, prices, earnings, dividends, and sock reurns. The risk free rae is he rerun on 30 days Treasury bill rae from CRSP. All nominal quaniies are deflaed using he CPI. 5 For S&P 500, high qualiy book equiy daa is generally unavailable prior o 977. I is, herefore, crucial o choose an appropriae proxy for he book equiy. As we know, he common feaure of book equiy is well-described by he clean-surplus relaion. 6 This accouning ideniy has been widely applied in many sudies, such as Vuoleenaho (000) and Pásor and Veronesi (003). Nex, I need o specify a good proxy for dividends because he repored dividend level is arificial and involves some mis-measuremen. Firs of all, previous lieraure has noed ha firms ry o smooh dividends over ime; hence, he repored dividends may no reflec he rue value of firms. For example, Linner (956) documens ha managers have a arge level of dividends equal o a fracion of curren earnings. Recenly, Marsh and Meron (987), Wu and Wang (000), and Allen and Michaely (00) also idenify a similar endency among firms. In addiion o earnings, dividend policy iself may be used o reveal oher informaion o he marke or o resolve agency problems (e.g., Allen and Michaely, 00). Second, recen evidence suggess ha repurchases have subsiued for cash dividends over he pas 5 o 0 years (e.g., Fama and French, 00). Boudoukh e al. (007) furher poin ou ha payou (dividends plus repurchases) yields have beer forecasing power for sock reurns han cash dividends alone. All hese consideraions sugges ha he repored dividends alone are no a suiable reflecion of he rue cash flows generaed by firms. To acquire more reliable informaion abou he rue value of firms we assume ha aggregae dividends are equal o a consan fracion of aggregae earnings, suggesed by Lee e al. (999) and Longsaff and Piazzesi (004). We assume he dividend payou raio is 50 percen in accordance wih he hisorical average for his sample period. 7 However, i should be noed ha alering he payou raio is irrelevan o our resuls. Finally, I use annual consumpion daa from he Naional Income and Produc Accouns repored by he Bureau of Economic Analysis (BEA) for he period Consumpion is defined as he sum of aggregae nondurables and services consumpion. Nominal consumpion is also deflaed by realized inflaion using he Consumer Price Index. The esimaed populaion a each year end, aken from he Census Bureau, is used o calculae consumpion in per capia. To perceive how he book equiy repored by he S&P 500 is close o he clean surplus book equiy, I compare hese wo series and he corresponding dividend-o-book equiy raios for he period Table repors he comparison s resuls. Throughou my sample period, book equiy calculaed by he clean surplus relaion is always higher han he book equiy repored by he 5 The consumer price index is obained from he FRED daabase. 6 I should be noed ha alhough we apply he clean surplus relaion o consruc a proxy for book equiy, in heoreical framework, we do no assume he process for book equiy is governed by his relaion as assumed by Pásor and Veronesi (003). 7 The idenical dividend-o-earning raio is applied by Liner (956), and Longsaff and Piazzesi (004). The average dividend-o-earning raio in our sample is 56.8 percen, and he median is 54.7 percen.

6 Inernaional Journal of Economics and Financial Issues, Vol. 5, No., 05, pp S&P 500. However, he correlaion beween hese wo book equiy series is prey high (99.69%). In addiion, he wo log dividend-o-book equiy series consruced from hese wo approaches have a 98.5% correlaion. And he correlaion beween changes in he wo raios is 99.08%. In summary, alhough book equiy esimaed by he clean surplus relaion is higher han he repored level of S&P, he corresponding dividend-o-book equiy raios esimaed by hese wo mehods share a similar ime series propery. Table. Comparison of repored book equiy and book equiy from clean surplus relaion, This able compares dividend-o-book equiy raios for book equiy based on he repored value and clean surplus relaion. Dividends, D, is proxy by a consan payou raio imes he repored earnings, in which he consan payou is se o 0.5. Daa of book equiy, dividends, and earnings are from he S&P 500. Panel shows he descripive saisics of he cash flow-fundamenal raio for sample period Level Correlaion Book Equiy Clean Surplus Repored by S&P Book Equiy 99.69% Correlaion of Levels of Book Equiy Clean Surplus log Dividend-o-Book Equiy Repored by S&P Book Equiy 98.5% Correlaion of Changes of Book Equiy Clean Surplus log Dividend-o-Book Equiy Repored by S&P Book Equiy 99.08% Descripive of Book Equiy Clean Surplus log Dividend-o-Book Equiy Repored by S&P Book Equiy Mean Sandard Deviaion Auocorrelaion In our sample period, he correlaion beween percenage changes in he cash flow-fundamenal raio and consumpion growh is We find ha he ime varying propery of he cash flow-fundamenal raio is relaed o business cycles. Figure plos he ime series of he cash flow-fundamenal raio and he marke-o-book raio for As shown, hese wo raios decline a lo a recessions, especially a he Grea Depression. Afer war, hey boh swich o a higher sable level. Besides, hey end o be more volaile afer 990. The -saisic for log cash flow-fundamenal raio from he regression of log marke-o-book raio on log cash flow-fundamenal raio is Evidence suggess ha he cash flow-fundamenal raio and he marke-o-book raio are posiively relaed and boh of hem are procyclical. Based on our framework, his procyclicaliy in urn implies ha he excess reurns are counercyclical. Mos imporanly, his procyclical feaure is criical o verify he negaive relaionship beween he cash flow-fundamenal raio and cash flow growh as documened by our framework and evidences. 3.. Asse Pricing Implicaions Table provides evidence ha fuure realized consumpion volailiy is prediced by he cash flow-fundamenal raio. The dividend-o-book equiy raio, proxy of cash flow-fundamenal raio, predics fuure realized volailiy wih negaive coefficiens. The -saisics are all above 3 and he R s are around 5% for horizons of up o 0 years. If consumpion volailiy were no ime-varying, he slope on he dividend-o-book raio would be zero. As argued by my framework, his evidence indicaes ha corporae cash flows conain informaion regarding persisen flucuaions in economic uncerainy. Bansal and Yarson (004) also documen ime-varying consumpion volailiy in erms of price-dividend raio. However, hey show ha higher dividend yields follow greaer flucuaions in economic uncerainy. 53

7 Undersanding he Cash Flow-Fundamenal Raio Figure. The cash flow-fundamenal raio and marke-o-book raio, The solid line is he sandardized log cash flow-fundamenal raio. The dash line marked wih + is he sandardized log marke-o-book raio. Shaded areas indicae recessions as deermined by he NBER. The cash flow-fundamenal raio is he raio of earnings per share o book equiy per share imes he payou raio which is se o 0.5. Book equiy, B, is consruced by clean surplus relaion. Daa of book equiy, prices, and earnings are from he S&P. log(d/b) log(m/b) Sandardized log(d/b) and log(m/b) Table. Properies of consumpion volailiy The enries provide regression resuls for 0 log D / B g a, H, where H indicaes he H forecas horizon in years and g 5 c c igi a g, i represens he absolue value of he residual from he regression a g, c, in which g denoes annual consumpion growh rae. D / B is proxy of he cash flow-fundamenal raio. The saisics are relied on annual observaions of real nondurables and services consumpion from BEA for sample period Dividends, D, is proxy by a consan payou raio imes he repored earnings, in which he consan payou is se o 0.5. Book equiy, B, is consruced by clean surplus relaion. Daa of book equiy, dividends, and earnings are from he S&P 500. For each daa regression, -sa is calculaed by sandard error which is correced by Newey and Wes (987) using 0 lags. And R denoes he adjused-r. Horizon in Years sa R

8 Inernaional Journal of Economics and Financial Issues, Vol. 5, No., 05, pp Panel A of Table 3 repors he resuls of long-horizon regression of excess reurns on he cash flow-fundamenal raio or simplified by cash flow-o-book equiy raio for our sample. In my daa, high cash flow-o-book equiy raio significanly predics low excess reurns in he medium horizon. Significan coefficiens for absolue -saisics above happen a a horizon up o four years. The R reaches 0.08 a a horizon of four years. However, he coefficiens and he R s rise slowly wih he horizon. This negaive correlaion provides a good explanaion for resuls found in Jiang and Lee (007) ha higher dividend-o-book raio predics lower sock reurns. Compared wih he resuls of dividend yields and book-o-marke raios from previous sudies, I find ha he cash flow-o-book equiy raio has less predicive power for sock reurns in he long horizon. Panel B of Table 3 shows regression resuls where he dependen variable is he sum of annual consumpion growh raes. A shorer horizons he cash flow-fundamenal raio has lile predicive power. When considering longer horizons of up o eigh years, however, consumpion growh raes can be forecased by his raio wih negaive coefficiens. I find he coefficiens and R increase as forecasing horizons become longer. My cash flows model implies ha he cash flow-fundamenal raio forecass fuure cash flow growh. I furher deec wheher he cash flow-fundamenal is able o predic fuure cash flow growh rae. Panel C of Table 3 shows he regression resul of regressing he sum of annual dividend growh rae on cash flow-fundamenal raio. The evidence shows ha is significanly negaive for all horizons implying he higher in cash flow-fundamenal he lower in cash flow growh. The cash flow-fundamenal raio is a significan predicor of cash flow growh a he aggregae level wih R equal o 3%, %, and 8% for he -, 4-, and 0-year horizons respecively. This evidence is also consisen wih my framework ha he cash flow-fundamenal raio and cash flow growh rae are negaively correlaed if he corporae dividend is procyclical. Panel D of Table 3 repors he resul of long-horizon regression of book equiy growh on he cash flow-fundamenal raio. As we expeced, he cash flow-fundamenal raio has lile predicive power in book equiy growh. Combining informaion revealed in Panel C and Panel D of Table 3, I briefly inerpre he driving force of he mean-reversion propery behind he cash flow-fundamenal raio. As documened by Campbell and Shiller (00), he sabiliy of a financial raio implies ha eiher he numeraor or he denominaor mus be predicable based on he raio. Resuls from Panel C and Panel D verify ha he cash flow-fundamenal raio is saionary as here is subsanial predicabiliy of cash flow growh a aggregae level. Evidence shows ha i is he cash flows raher han book equiy ha faciliaes he cash flow-fundamenal raio back o is mean value. In brief, referring o he resuls from Table and Table 3, we demonsrae ha he increase in cash flow-fundamenal raio is relaed o he decline in economic uncerainy and he predicabiliy power of he cash flow-fundamenal raio on excess reurn is guided by his connecion. Nex, I invesigae why he dividend yield and he cash flow-fundamenal raio predic excess reurns wih an opposie sign. Firs, I find ha hese wo financial raios conain informaion abou fuure economic uncerainy. Bansal and Yaron (004) provide evidence ha condiional volailiy of consumpion can be prediced by he price-dividend raio wih negaive coefficiens. In conras, we find a higher cash flow-fundamenal raio predics lower consumpion volailiy as shown in Table. As a high risk premium is expeced when he economy is more uncerain, i is reasonable o expec ha a lower price-dividend raio or a lower cash flow-fundamenal raio predics a higher reurn. Second, he driving force of mean-reversion behind he price-dividend raio and he cash flow-fundamenal raio is disinc. I is well known ha he predicive power of any financial raio relies on he sabiliy of a valuaion raio. Previous sudies have found ha he price-dividend raio is a poor predicor of dividend growh (e.g., Campbell and Shiller, 00; Menzly e al., 004; Leau and Wacher, 007). Therefore, lieraure proposes ha i is sock price ha resores he price-dividend raio back o is mean value. In oher words, he price growh should be forecas by he price-dividend raio. When he growh in sock price is limied, a higher price-dividend raio would resul in a lower excess reurn. In conras, in my framework i is he cash flows ha governs he mean-reversion propery of he cash flow-fundamenal raio. A higher cash flow-fundamenal raio predics a lower cash flow growh rae and hence a lower excess reurn. 55

9 Undersanding he Cash Flow-Fundamenal Raio Table 3. Long-horizon regressions predicabiliy of excess reurns, growh raes in consumpion, cash flows, and book equiy. This able repors evidences on predicabiliy of fuure excess reurns and growh raes by he lagged dividend-o-book equiy raio. The corresponding regression model in Panel A is H m f m f r i r i 0 log D / B, where r H i r i i is he excess reurn and H denoes he forecas horizon in years. The corresponding regression models in Panel B, Panel C, and Panel D are H j gi 0 log D / B, where j,, H i c c d b. g is he annualized consumpion growh rae. is he annualized cash flow growh rae. And b g is he annualized book equiy growh rae. D / B is proxy of he cash flow-fundamenal raio. The saisics are relied on annual observaions of real nondurable and service consumpion from BEA for sample period Dividends, D, is proxy by a consan payou raio imes he repored earnings, in which he consan payou is se o 0.5. Book equiy, B, is consruced by clean surplus relaion. Daa of book equiy, dividends, and earnings are from he S&P 500. For each daa regression, -sa is calculaed by sandard errors which are correced by Newey andwes (987) using 0 lags. And R denoes he adjused-r. Horizon in Years Panel A: Excess Reurns sa R Panel B: Growh Raes in Consumpion sa R Panel C: Growh Raes in Cash Flows sa R Panel D: Growh Raes in Book Equiy sa R In summary, I inerpre why a high cash flow-fundamenal raio predics a lower excess reurn. On he one hand, high cash flow-fundamenal raios can predic lower consumpion volailiy, which implies less economic uncerainy. On he oher hand, a lower cash flow growh is expeced when he cash flow-fundamenal raio is high, which in urn implies a lower excess reurn. Besides, hese wo properies explain why dividend yields and cash flow-fundamenal raios predic excess reurns wih an opposie sign. 4. Conclusion Previous lieraure has widely applied financial raios such as dividend yields and book-o-marke raios o forecas asse reurns. Jiang and Lee (007) documen ha linear combinaion of log dividend yields and log book-o-marke raios have beer performance han individual financial raios in some aspecs. However, he explici informaion behind his loglinear model is sill unknown. In his paper, I invesigae he characerisics of cash flow-fundamenal raio o verify he raionale behind his loglinear model. Two piece of informaion is found. Firs, he realized consumpion volailiy, which represens flucuaing economic uncerainy, is prediced by he cash flow-fundamenal raio. Second, he cash flow-fundamenal raio can forecas fuure cash flow growh in all horizons. Evidence shows ha higher cash flow-fundamenal raios predic lower consumpion volailiy in he fuure as well as d g

10 Inernaional Journal of Economics and Financial Issues, Vol. 5, No., 05, pp lower cash flow growh. Boh feaures indicae ha higher equiy premium is compensaed for a lower cash flow-fundamenal raio because of he increase in uncerainy. Moreover, for he cash flow-fundamenal raio, he abiliy o predic fuure cash flow growh ensures he saionary propery needed by any financial raio ha possesses forecasing power. References Allen, F., Michaely, R. (00), Payou policy. Working paper, Wharon School, Universiy of Pennsylvania. Bansal, R., Yaron, A. (004), Risks for he long run: A poenial resoluion of asse pricing puzzles. Journal of Finance, 59, Boudoukh, J., Michaely, R., Richardson, M., Robers, M.R. (007), On he imporance of measuring payou yield: Implicaions for empirical asse pricing. Journal of Finance, 6, Campbell, J.Y., Shiller, R.J. (989), The dividend-price raio and expecaions of fuure dividends and discoun facors. Review of Financial Sudies, (3), Campbell, J.Y., Shiller, R.J. (00), Valuaion raios and he long-run sock marke oulook: An updae (No. w8). Naional Bureau of Economic Research. DeBond, W.F.M., Thaler, R. (985), Does he sock marke overreac? Journal of Finance, 40, Fama, E.F., French, K.R. (988), Dividend yields and expeced sock reurns. Journal of Financial Economics,, 3-5. Fama, E.F., French, K.R., (000), Forecasing profiabiliy and earnings. Journal of Business, 73, Fama, E.F., French, K.R. (00), Disappearing dividends: Changing firm characerisics or lower propensiy o pay? Journal of Financial Economics, 60, Grullon, G., Michaely, R., Swaminahan, B. (00), Are dividend changes a sign of firm mauriy? Journal of Business, 75, Jiang, X., Lee, B.S. (007), Sock reurns, dividend yield, and book-o-marke raio. Journal of Banking & Finance, 3(), Lamon, O. (998), Earnings and expeced reurns. Journal of Finance, 53, Lee, C., Myers, J., Swaminahan, B. (999), Wha is he inrinsic value of he Dow? Journal of Finance, 54, Leau, M., Wacher, J.A. (007), Why is long-horizon equiy less risky? A duraion-based explanaion of he value premium. Journal of Finance, 6, Lewellen, J. (004), Predicing reurns wih financial raios. Journal of Financial Economics, 74, Linner, J. (956), Disribuion of incomes of corporaions among dividends, reained earnings, and axes. The American Economic Review, Longsaff, F.A., Piazzesi, M. (004), Corporae earnings and he equiy premium. Journal of Financial Economics, 74, Marsh, T.A., Meron, R.C. (987), Dividend behavior for he aggregae sock marke. Journal of Business, 60, -40. Menzly, L., Sanos, T., Veronesi, P. (004), Undersanding predicabiliy. Journal of Poliical Economy,, -47. Pásor, L., Veronesi, P. (003), Sock valuaion and learning abou profiabiliy. Journal of Finance, 58(5), Shiller, R.J. (989), Marke volailiy. The MIT Press, Cambridge, Massachuses. Vuoleenaho, T. (000), Undersanding he aggregae book-o-marke raio and is implicaions o curren equiy-premium expecaions. Working paper, Harvard Universiy. Wu, C., Wang, X.M. (000), The Predicive Abiliy of Dividend and Earnings Yields for Long Term Sock Reurns. Financial Review, 35(),