A Preliminary Analysis for Measuring Operating Performance of Real Estate Investment Trusts in Taiwan: Net Income vs.

Transcription

1 Conemporary Managemen Research Pages , Vol. 7, No. 4, December 20 A Preliminary Analysis for Measuring Operaing Performance of Real Esae Invesmen Truss in Taiwan: Ne Income vs. Operaion Funds Ming-Chi Chen Naional Sun Ya-sen Universiy mcchen@finance.nsysu.edu.w I-Chun Tsai Naional Universiy of Kaohsiung icsai@nuk.edu.w Kuan Liang Chen PKF Taiwan gary@pkf.com.w Henry Y. Lo Nan Kai Universiy of Technology ylo@nku.edu.w ABSTRACT The Naional Associaion of Real Esae Invesmen Trus (NAREIT) has suggesed ha Ne Income (NI) migh no be he bes measure of operaing performance for income-producing real esae, since depreciaion measuremen based on hisorical cos is inappropriae for income-producing real esae. Thus, he NAREIT purposes Funds From Operaion (FFO) as a supplemenal measure for operaing performance of income-producing real esae. In Taiwan, Real Esae Invesmen Truss (REITs) are a relaively new ype of income-producing real esae invesmen. Since he daa on REITs in Taiwan is no ye sufficien, his work uses hoel, ourism and deparmen sore indusries, or real esae operaing firms (REOCs), as a proxy for a preliminary sudy of REITs in Taiwan. The empirical resuls show ha using a

2 Conemporary Managemen Research 272 model incorporaing boh NI and FFO undersaes he explanaory power of wo variables, since mulicollineariy occurs. Therefore, he curren sudy separaely esimaes he informaion conen of NI and FFO, finding ha he explanaory power of NI is higher han ha of FFO, hough he forecasing abiliy of FFO is higher han ha of NI. Bu he differences beween he explanaory power and forecasing abiliy of NI and FFO are very small. The resuls show ha NI and FFO are boh informaive for income-producing real esae in Taiwan. Keywords: Ne Income, Fund From Operaion, Real Esae Invesmen Truss INTRODUCTION Alhough real esae securiizaion has jus sared o develop in Taiwan, financial markes in various oher counries have adoped i for many years. Among inernaional financial markes, Real Esae Invesmen Truss (REITs) in he Unied Saes are he mos successful example of real esae securiizaion. The recenly passed Taiwan Real Esae Securiizaion Ac was parially srucured afer he U.S. REITs. REITs were esablished by he U.S. Congress in 960 o allow small invesors o paricipae in diversified, large-scale, income-producing real esae enerprises. They are corporaions or business russ managed by boards of direcors or rusees. According o he Naional Associaion of Real Esae Invesmen Truss (NAREIT), he number of equiy REITs rose from fify-six o 58, and oal marke capializaion increased from $6.8 billion o $34 billion during he period from 989 o REITs have received increased aenion in he financial lieraure due parially o heir unique characerisics and increasing populariy among invesors. REITs are one of he few publicly raded secors ha use an alernaive form of performance measuremen. The main indusry rade associaion, NAREIT has suggesed ha Ne Income is no a meaningful measure of operaing performance for income-producing real esae. Wih REITs, financial performance is universally deermined by funds from operaions, or FFO, which adds back depreciaion when calculaing cash flow. NAREIT claims ha FFO is a beer measure of operaing performance because hisorical cos depreciaion is inappropriae for income-producing real esae and does no correlae wih changes in he value of hese asses. Alhough he indusry believes FFO provides beer informaion han Ne Income, he Securiy Exchange Commiee (SEC) has hisorically no favored seing up unique accouning rules and measures for specific indusries. Unil recenly, he SEC had no sancioned FFO as a supplemenal performance measuremen o Ne

3 Conemporary Managemen Research 273 Income. Real esae securiizaion has been an imporan issue in Taiwan for many years. Afer he Execuive Yuan made he firs draf of he Real Esae Securizaion Ac in 2002, real esae securiizaion has been gaining increasing aenion from Taiwan s financial indusry. Under he Taiwan Real Esae Securiizaion Ac, professional real esae invesmen russ purchase, manage and sell income-producing real properies and pass he renal income and capial gains on o invesors. Therefore, NAREIT s asserion ha FFO is a more meaningful performance measuremen for income-producing real esae should apply o equiy REITs or real esae operaing corporaions (REOCs) in Taiwan. This paper examines he informaion conen of he variables, Ne Income and FFO, o es which one ells more abou operaing performance for income-producing real esae. Since daa on REITs in Taiwan is sill insufficien, his work uses he hoel, ourism and deparmen sore indusries ogeher as a proxy, o sudy REOCs for a preliminary invesigaion of REITs in Taiwan. The paper srucure is as follows. Secion 2 describes he operaing measuremen of US REITs. Secion 3 reviews previous sudies. Secion 4 discusses he mehodology and empirical models his paper uses. Secion 5 repors and discusses esimaion resuls, while he las secion provides a summary of he main findings and presens some conclusions. OPERATING MEASUREMENT OF US REITS A REIT is a company ha owns, and in mos cases, operaes income-producing real esae such as aparmens, shopping ceners, offices, hoels and warehouses. Some REITs also engage in financing real esae. The shares of mos REITs are freely raded, usually on a major sock exchange. The asse srucure of real esae invesmen russ is composed almos enirely of fixed asses, and consequenly, hey are subjec o a large depreciaion and amorizaion expense. Due o his uniqueness, REITs are one of he few publicly raded secors ha use an alernaive form of performance measuremen. For example, NAREIT saes, mos real esae indusry invesors have concluded ha operaing resuls for real esae companies based on hisorical cos accouning are misleading or uninformaive. Therefore, REITs financial performance, in addiion o ne income, is deermined by funds from operaions, or FFO, which add back depreciaion and amorizaion expense and exclude gains or losses from sales of propery o ne income. NAREIT has promoed he concep of Funds From Operaions as an indusry-wide

4 Conemporary Managemen Research 274 measure of REIT operaing performance ha does no include some iems in radiional ne income compued according o GAAP. Alhough FFO has been repored by some REITs since he early 980 s, NAREIT firs provided a formal definiion of FFO in 99. In he Whie Paper on Funds From Operaions published in April 2002, NAREIT adoped he following definiion of FFO: FUNDS FROM OPERATIONS means ne income (compued according o generally acceped accouning principles), excluding gains (or losses) from sales of propery, plus depreciaion and amorizaion, afer adjusing for unconsolidaed parnerships and join venures. Adjusmen calculaions for unconsolidaed parnerships and join venures reflec funds from operaions on he same basis. To promoe FFO as an alernaive performance measure, NAREIT argues ha he income measuremen model currenly prescribed by GAAP does no accuraely reflec he economic performance of REITs. In is published Whie Paper on Funds From Operaions, NAREIT saes ha Hisorical cos accouning for real esae asses implicily assumes ha he value of real esae asses diminishes predicably over ime. Since real esae values rise or fall wih marke condiions, many indusry invesors consider presenaions of operaing resuls for real esae companies ha use hisorical cos accouning as insufficien by hemselves. Therefore, Funds From Operaions was developed o address his problem and serve as a sandard supplemenal measure of operaing performance ha excludes hisorical cos depreciaion on income-producing real esae from he GAAP-based ne income. Furhermore, because FFO is a measure of recurring operaions, GAAP s exraordinary or unusual iems and significan non-recurring iems should be excluded from FFO. In 998, NAREIT formed he Bes Financial Pracices Council in an effor o enhance he qualiy and effeciveness of indusry financial pracices. These indusry financial pracices include financial reporing and disclosure pracices required by generally acceped accouning principles, as well as pracices relaed o he indusry s supplemenal performance measure, Funds From Operaions. The Council recommends ha REITs should provide he following reconciliaion in heir financial saemens: The above reconciliaion shows ha FFO is adjused from Ne Income by adding back depreciaion of real esae asses and deducing gains or losses from he sale of depreciable real esae and gains or losses from sales of oher real esae and securiies. The adjusmens for minoriy ineress and for unconsolidaed affiliaes are

5 Conemporary Managemen Research 275 adjusmens ha a REIT has o make if i invess in oher REITs. This is because he invesor REIT migh include he invesed REIT s profi o which he invesor REIT is eniled o as he reurn of invesmen in is financial saemens. Basically, he adjusmens for minoriy ineress and for unconsolidaed affiliaes are made as if he invesed REIT reconciled Ne Income ino FFO, and he invesor REIT recorded he reurn of invesmen based on he invesmen s FFO, no Ne Income. FFO/GAAP Ne Income Reconciliaion Ne income aribuable o common shareholders.. $x,xxx,xxx Adjusmens: Depreciaion of real esae asses...x,xxx,xxx Gain/losses from sales of depreciable real esae..xxx,xxx Gain/losses from sales of oher real esae and securiies..xxx,xxx Oher iems: Adjusmens for minoriy ineress.xxx,xxx Adjusmens for unconsolidaed affiliaes.. xxx,xxx FFO $x,xxx,xxx Weighed average shares and unis: Basic.x,xxx,xxx Dilued..x,xxx,xxx Source: NAREIT s Naional Policy Bullein on FFO Whie Paper Disclosures. One of he advanages of FFO is is reamen of depreciaion and amorizaion expense. GAAP ne income rules sae ha, as wih facory machines and compuers, depreciaing asses on a hisorical cos basis wear ou and become obsolee wih he passage of ime. REIT execuives have argued ha his ype of deprecaion is no value relevan for he ype of income-producing properies ha hey own and operae. The buildings hey own do no decline in value over ime in he same manner as oher ypes of asses. In fac, hey may become more valuable over ime, once locaion and enancies have been esablished. Real esae values rise and fall, someimes jus as suddenly as he sock marke. The erm FFO was developed as a performance measure wih a proper adjusmen for inaccuracies associaed wih GAAP accouning. However, using an alernaive financial measuremen is obviously difficul when comparing REIT socks wih hose of oher companies. Since he boom line ne income is he GAAP sandardized operaing performance measure of all companies, i is difficul for alernaive performance measures o replace ne

6 Conemporary Managemen Research 276 income. Therefore, alhough FFO is a widely used operaing performance measuremen in he REIT indusry, i is no likely ha FFO will replace Ne Income in he fuure. Tradiionally, hree accouning-based performance measures include: ne income, cash flow from operaions (CFO) and earning before ineres, ax, depreciaion and amorizaion (EBITDA). Table shows he direc calculaion of hese performance measures from cash operaing revenue and he indirec calculaion of each performance measure from GAAP ne income. Boh earnings and CFO are required GAAP disclosures, and EBITDA, compued from audied numbers, can generally be obained from published financial saemens. FFO is a widely used performance measure in he REIT indusry. I assumes ha hisorical cos depreciaion expense is no relevan in valuing income-producing real esae. However, hisorical cos based depreciaion expense was never inended o adjus for changes in he fair value of asses. In addiion, deparmen sores, office buildings and oher real esae asses ofen require significan capial improvemens in order o remain compeiive. These capial expendiures are no all expensed; some of hem may be capialized o increase asse value. The problem of FFO is ha i does no consider he effec of capializing hese expendiures. Therefore, he empirical quesion is wheher FFO is more value-relevan o invesors han Ne Income. LITERATURE REVIEW Previous Sudies on Tradiional Accouning-Based Informaion In an earlier sudy, Ball and Brown (968) examined he informaion conen of cash flows using operaing income as a proxy for cash flows. They found ha cash flows performed poorly relaive o ne income and earnings per share in predicing abnormal sock reurns. Alhough he cash flow proxy used by Ball and Brown was raher unsophisicaed, heir finding ha he marke responds more favorably o earnings han o cash flows is consisen wih more recen evidence. Since he sudy of Ball and Brown, many oher sudies have been conduced based on oher accouning-based informaion. Two recen sudies include Kallunki, Marikainen and Marikainen (998), and Dhaliwal, Subramanyam and Trezevan (999). Of hese, Kallunki, Marikainen and Marikainen (998) use esing daa from Finland o invesigae which income levels and componens relae o marke-o-book equiy raios. The resuls of heir empirical esing sugges ha income saemen iems, such as sales and operaing income, oher han boom-line earnings, provide useful

7 Conemporary Managemen Research 277 informaion for invesors. The sudy by Dhaliwal, Subramanyam and Trezevan (999) evaluaes which of he wo performance measures, comprehensive income or ne income, beer summarize firm performance as refleced in sock price. Their resul suggess no evidence supporing he conenion ha comprehensive income is more srongly associaed wih reurn or ha i is beer in predicing fuure cash flows han ne income. Furhermore, he only componen of comprehensive income ha improves he associaion beween income and reurns is markeable securiies adjusmen. For cash flows and accruals, previous sudies have generally focused on wo main opics. The firs opic is he incremenal informaion conen of accruals over cash flows. Though accruals are subjec o possible managemen manipulaion, hey may sill conain informaion regarding fuure earnings and cash flows. The second opic is wheher informaion regarding separae cash flow and accrual componens of earnings has value relevance. Wilson (986) invesigaes he incremenal informaion conen of accruals over cash flow for he period. His sudy updaes cash flow expecaions for he year-end earnings announcemen and measures he cross-secional marke response o he incremenal cash flow informaion release dae using daily reurns. He measures cash flow using daa from he Saemen of Changes in Financial Posiion and adjuss for curren accruals in his definiion of cash flows. As a resul, he finds incremenal informaive conen for cash flow and curren accruals bu no for non-curren accruals. In his laer sudy, Wilson (987) invesigaes he exen o which cash flow and accrual componens of earning conain informaion ha is incremenal o ha conained in earnings alone. To es his, he uses he reurn reacing over a nine-day inerval cenered on he dae a which he annual repor arrives a he SEC, and defines cash flow as boh working capial from operaions and cash operaions. The resuls show ha cash from operaions and oal accruals, ogeher, have greaer abiliy o explain sock reurns han boom-line earnings and here is no associaion beween working capial and sock reurn. By expanding Wilson (987) s sudy o hiry-wo quarers, wih 976 o 984 daa, Bernard and Sober (989) sugges ha here is no evidence ha he associaion beween sock marke and operaing cash flow is greaer han he associaion beween sock marke and curren accruals. In a subsequen sudy, Livna and Zarowin (990) examine he informaion conen of hree major componens of saemen of cash flows. In heir sudy, hey conduc separae ess on each of he hree major componens - cash from operaing aciviies, cash from invesing aciviies and cash from financing aciviies. Their resuls suggess ha cash from operaions and accruals has no associaion wih abnormal sock reurn and if cash from operaions and cash from financing aciviies are regressed ogeher,

8 Conemporary Managemen Research 278 hey are boh significanly associaed wih abnormal reurns. However, here is no evidence o show any associaion beween cash flow from invesing aciviies and abnormal reurns.

9 Conemporary Managemen Research 279 Previous Sudies on Funds From Operaions The NAREIT has suggesed ha Ne Income is a misleading measure of operaing performance for he real esae indusry, and his conenion has led o fuure sudies examining FFO informaion conen. The sudy by Gore and So (998), examines he informaion conen of ne income and FFO ogeher wih he individual componens (depreciaion and gain/loss from sale of properies) of ne income ha differ from FFO o assess heir value relevance. Their finding suggess ha FFO is more relaed o sock reurn han ne income. They also find ha gain/loss from he sale of properies is significanly associaed wih sock reurn bu depreciaion expense is no. Vincen (999) compares he informaion conen of FFO, earnings per share (EPS) and oher alernaive performance measures (cash from operaions CFO and earnings before ineres, axes, depreciaion and amorizaion EBITDA), using sock reurns as he benchmark. The resuls indicae ha boh FFO and earnings per share provide incremenal informaion conen. The resuls also sugges ha, wih weak evidence, EPS has greaer relaive informaion conen over oher performance measures. According o Vincen s resuls, Skinner (999) menions ha i is seemingly easy o adjus back and forh beween NI and FFO based on published informaion, so i is no clear why invesors should care which of hese alernaive measures are repored as he primary performance measure in financial saemens. In Skinner s opinion, Vincen s sudy does no clearly spell ou he economic argumen for FFO as a superior measure of firm performance. Skinner suggess ha Vincen s sudy use fuure realized measures of performance o provide a benchmark for assessing how well curren period variables measure firm performance. Graham and Knigh (2000) use hree alernaive marke-based accouning research models o examine relaive and incremenal informaion conen of ne income and FFO. Their resuls are consisen wih Gore and So (998) in suggesing ha FFO is relaively and incremenally informaive in predicing sock reurns. In a recen sudy, Sunda and Typpo (2004) examine he role ha earning ransiiviy has on value-relevance of FFO relaive o earnings. The findings indicae ha as earnings become more ransiive, FFO gains value-relevance, whereas earnings lose value-relevance. Addiionally, boh FFO and earning provide incremenal informaion in he presence of he oher. Previous sudies on cash flow are generally consisen in showing ha cash flow provides valuable informaion conen in evaluaing firm performance, hough sudies on accrual sugges ha non-curren accrual iems are no informaive in evaluaing firm performance. These findings provide suppor for NAREIT s claim ha FFO provides relevan informaion for measuring real esae firm performance. Following

10 Conemporary Managemen Research 280 NAREIT s claim, several sudies have been done o deermine he informaion relevance of FFO using REIT firm sock reurns as he benchmark. Their findings are consisen in suggesing ha FFO provides relevan informaion for evaluaing firm performance. However, since NAREIT suggess ha FFO performs beer han Ne Income in measuring operaing performance of income producing real esae, a sudy on a radiional non-reit real esae indusry can provide useful informaion. METHODOLOGY This research uses hree mehodologies o capure and compare he informaion conen of wo imporan variables, NI and FFO, o effecively evaluae he sock value of real esae operaing firms. The curren work firs uses panel daa regression o compare he abiliy of he wo independen variables, NI and FFO, for explaining he reurns of relaed asses. Second, his sudy uses he marke model o esimae he abnormal reurn of real esae operaing firms, and ess wheher or no NI and FFO are informed, for esing which variable provides more incremenal informaion. Thirdly, his work compares he forecasing abiliy of NI and FFO. Condiional on he resuls of hese hree mehods, his invesigaion compleely analyzes he informaion conen, he incremen informaion and he forecasing abiliy of NI and FFO. The empirical models used in his sudy are simply shown as follows. Panel Daa Regression To simply compare he abiliy of NI and FFO for explaining he reurns of relaed asses, he curren sudy uses a regression incorporaed cross secional and ime series daa, or panel daa regression, shown in he following. R i, a0 anii, a2ffoi, i, () where R i, is he log reurn of company i on ime, NI i, is he per share ne income of company i on ime, FFO i, is he per share funds from operaions of company i on ime, a 0 is he consan erm, a is he coefficien showing he relaionship beween ne income and he relaed sock reurn, and a 2 is he coefficien showing he relaionship beween operaion funds and he relaed sock reurn. Following he panel daa regression, his sudy compares he informaion conen by observing wheher a or a 2 is more significanly nonzero. i, is he error erm.

11 Conemporary Managemen Research 28 The Tes of Incremen Informaion Alhough NI and FFO can be informaive, he informaion conen sill needs o be analyzed o es wheher or no i is incremenal, since he same informaion migh be obained from oher variables, in which case he informaion from NI or FFO would be useless. This invesigaion uses a wo-sep mehodology o es wheher or no he informaion of NI and FFO is incremen and useful. This work firs esimaes he marke model o obain innovaions, which are par of reurns, and canno be explained by marke reurns; and hen separaely ess he explanaory power of NI and FFO on he innovaions. The models used here are shown as follows. R i, b0 b Rm ui, u i, 0 NIi,, i, u i, 0 FFOi, 2, i, (2) (3) (4) where R m is he reurn of he whole sock marke ime, and oher variables are defined as above. From equaion 2, we find he innovaions, which are he par of reurns and canno be explained by he marke reurns, ha is u i,. Then his work separaely ess he explanaory power of NI and FFO on he innovaions using equaion 3 and 4. If ( ) is significanly nonzero, hen here is incremenal informaion in NI (FFO), which provides addiional informaion for he unexpeced reurns. Besides,,i, (,i, ) is he error erm in equaion 3 (4). Esimaing Forecasing Abiliy Following esimaing equaions 3 and 4, he curren sudy uses daa from 998Q4 o 2004Q4 o esimae he coefficiens, and hen uses he esimaed coefficiens o forecas innovaions of real esae operaing firms, o compare he forecasing abiliy of NI and FFO. The mehodology is simply shown as follows. u ˆi, ˆ 0 ˆ NIi, ˆ ˆ u i, 0 NIi, ˆ (5) (6) By equaions 5 and equaion 6, he coefficiens ( 0,, 0 and ), esimaed

12 Conemporary Managemen Research 282 by he daa from 998Q4 o 2004Q4, are used o forecas he innovaions from 2005Q o 2006Q4. Then he forecasing error, calculaed by he difference beween he rue and he forecasing value of he innovaions, can be used o calculae he saisics, Roo Mean Squared Error (RMSE), Mean Absolue Error (MAE) and Mean Absolue Percen Error (MAPE), o compare he forecasing abiliy of NI and FFO. DATA DESCRIPTION AND EMPIRICAL RESULTS Daa Because hoels and deparmen sores fi he crieria of holding and operaing real esae as a primary operaion, he curren sudy draws empirical daa from he hoel and deparmen sore secor of he Taiwan Sock Exchange Corporaion and OTC marke. The daa se used in his analysis was obained from he TEJ daa bank, providing a sample of 393 observaions during 998Q4 o 2006Q4 (he 998Q4 daa is he beginning of 999 daa) for welve publicly raded Taiwan lised firms. This sudy calculaed he daa, which canno be found direcly from TEJ, wih he defined financial formulae. The definiions of variables used in his research are shown as follows.. Index: Taiwan Sock Exchange Capializaion Weighed Sock Index. 2. Sock Price (PRICE): Share price of sample company socks raded in he Taiwan Sock Exchange and OTC marke. 3. Earnings per Share (NI): Earnings from coninuing operaions before ineres and axes. This is obained from he income saemens of each sample company. 4. Depreciaion Expenses per Share (DEPR): Depreciaion is a cos allocaion of capial asses for he periods receiving benefis from asses. This is esimaed based on asses hisorical price. 5. Gain/Loss from he Sale of Properies per Share (GAIN): Normally, a he end of an asse s life, he asse is sold. If he corporaion has accuraely projeced he residual value, here is no gain or loss on he ransacion. If he residual value was no esimaed accuraely, eiher a gain or a loss resuls from his ransacion. 6. Funds from Operaions per Share (FFO): FFO is a widely used operaing performance measure in he REIT indusry. The curren research calculaes FFO as follows. FFO = Ne Income - Gain/Loss from he sale of properies + Depreciaion Expenses Table 2 presens a summary of he descripive saisics for variables, and Table 3 shows he correlaion marix of variables.

13 Conemporary Managemen Research 283 Table 2 Descripive Saisics of Variables Saisics INDEX PRICE NI DEPR GAIN FFO Mean Maximize Minimize Sd. Dev Skewness Kurosis ADF-Fisher PP -Fisher (0.0) 28.9 (0.22) 36.7 (0.07) (0.02) 3.46 (0.00) (0.00) 0.87 (0.00) (0.00) 8.0 (0.00) (0.00) (0.00) 99. (0.00) Noes: Probabiliies for Fisher ess are compued using an asympoic Chi-square disribuion. P-values are repored in parenheses. Table 3 Correlaion Marix of Variables Correlaion RM RI NI FFO GAIN DEPR RM RI NI FFO GAIN DEPR Table 2 also repors he oucome of ess for saionariy. This paper uses he augmened Dickey-Fuller es and he PP es. Two variables, sock marke index and sock price, canno significanly rejec he null hypohesis which assumes ha hese variables have a uni roo. To avoid he problem of spurious regression, his paper uses reurn daa, sock marke reurns and individual sock reurn, o esimae he empirical models. The very high correlaion of 0.94 beween NI and FFO, means ha using hese wo variables in a regression may resul in mulicollineariy, causing indeerminae regression coefficiens and infinie sandard errors wih misleading esimaed resuls. Therefore, grea care is essenial in dealing wih he problem of mulicollineariy, o ensure ha he resuls of regression incorporae boh variables, NI and FFO.

14 Conemporary Managemen Research 284 Empirical Resuls. Panel Daa Regression To simply compare he abiliy of NI and FFO for explaining he reurns of relaed asses, his paper uses a panel daa regression, wih he following resuls. Table 4 shows ha he coefficiens are all insignifican, and since he correlaion beween NI and FFO is very high, he resuls migh indicae ha mulicollineariy is occurring. Therefore, his work revises he regression model o reduce he mulicollineariy problem, and allows only one independen variable for each regression. The resuls are shown as follows. Table 4 shows ha he coefficiens are all insignifican, and since he correlaion beween NI and FFO is very high, he resuls migh indicae ha mulicollineariy is occurring. Therefore, his work revises he regression model o reduce he mulicollineariy problem, and allows only one independen variable for each regression. The resuls are shown as follows. Model: Table 4 Resuls of Panel Daa Regression R i, a0 anii, a2ffoi, i, Variable Coefficien Sd. Error -Saisic P-value a a a Adjused R-squared AIC Log likelihood SBC Noes: where R i, is he reurn of company i on ime, NI i, is he per share ne income of company i on ime, FFO i, is he per share funds from operaions of company i on ime, a 0 is he consan erm, a is he coefficien shown he relaionship beween he ne income and he relaed sock reurn, and a 2 is he coefficien shown he relaionship beween he funds from operaion and he relaed sock reurn. Table 5 resuls confirm our suspicion ha he coefficiens of independen variables are all significan, indicaing ha he informaion of NI and FFO explains varying reurns of asses. Furhermore, he resuls of Table 4 and Table 5 are similar o hose of Graham and Knigh (2000), which show he coefficiens are insignifican in he model conaining boh NI and FFO, and he coefficiens are all significan in simple regressions.

15 Conemporary Managemen Research 285 Regression : R a Table 5 Resuls of Simple Regression a NI i, 0 i, i, Variable Coefficien Sd. Error -Saisic P-value a a Adjused R-squared AIC Log likelihood SBC Regression 2: R a a FFO i, 0 i, i, Variable Coefficien Sd. Error -Saisic P-value 2 a a Adjused R-squared AIC Log likelihood SBC Noes: where R i, is he reurn of company i on ime, NI i, is he per share ne income of company i on ime, FFO i, is he per share funds from operaions of company i on ime. 2. The Tes of Incremen Informaion Alhough NI and FFO are all informaive in simple regressions, he informaion conen sill mus be analyzed o es wheher or no i is incremenal, using a wo-sep mehodology. This sudy firs esimaes he marke model o obain innovaions, which are he par of reurns ha canno be explained by he marke reurns. Then i separaely ess he explanaory power of NI and FFO on he innovaions. The resuls are as follows. Table 6 shows ha NI and FFO all significanly have incremened informaion, which can explain he innovaions of reurns. The explanaory power of NI is also higher han ha of FFO, since he p-value of is higher han ha of. Moreover, he values of R-squared and log likelihood in he firs regression are also higher han hose of he second, alhough he difference beween he explanaory power of NI and FFO is acually very small. 3. Esimaing Forecasing Abiliy This research uses forecasing error, calculaed by he difference beween he rue value and he forecasing value of innovaions, o calculae he saisics of Roo Mean

16 Conemporary Managemen Research 286 Squared Error (RMSE), Mean Absolue Error (MAE) and Mean Absolue Percen Error (MAPE). This compares he forecasing abiliy of NI and FFO, wih he following resuls. Table 7 shows ha he forecasing abiliy of FFO is beer han ha of NI, since he forecasing errors, ha is RMSE, MAE, and MAPE of FFO, are all less han ha of NI, alhough he difference beween he forecasing abiliy of NI and FFO here is also very small. These resuls indicae ha NI and FFO are boh informaive for income-producing real esae in Taiwan. Hence, he invesors of hese relaed asses should boh care he informaion of a company s NI and FFO. Sep : Table 6 The Resuls of Incremen Informaion Tes R i, b0 b Rm ui, Variable Coefficien Sd. Error -Saisic P-value b b Adjused R-squared AIC Log likelihood SBC Sep 2: Regression : u NI i, 0 i, i, Variable Coefficien Sd. Error -Saisic P-value Adjused R-squared AIC Log likelihood SBC

17 Conemporary Managemen Research 287 Regression 2: u FFO 2 i, 0 i, i, Variable Coefficien Sd. Error -Saisic P-value Adjused R-squared AIC Log likelihood SBC Noes: Where R i, is he reurn of company i on ime, R m is he sock marke reurn on ime, NI i, is he per share ne income of company i on ime, FFO i, is he per share funds from operaions of company i on ime. Table 7 The Resuls of he Forecasing Abiliy Saisics RMSE MAE MAPE NI FFO CONCLUSION The Naional Associaion of Real Esae Invesmen Trus (NAREIT) has suggesed ha Ne Income (NI) migh no be he bes measure of operaing performance for income-producing real esae, since he depreciaion measuremen based on hisorical cos is inappropriae for his ype of asse. Thus, he NAREIT purposes Funds From Operaion (FFO) as a supplemenal measure for operaing performance of income-producing real esae. To consider his quesion wih daa from a non-wesern marke, his work uses Real Esae Invesmen Truss (REITs) in Taiwan. Since his is a relaively new ype of invesmen, daa for REITs in Taiwan is insufficien. Therefore his research uses hoel, ourism and deparmen sore indusries as he proxy for a preliminary sudy of REITs in Taiwan. To capure and compare he informaion conen of wo imporan variables, NI and FFO, o effecively examine he sock value of real esae operaing firms, he curren research builds hree mehodologies. Firs, panel daa regression compares he abiliy of wo independen variables, NI and FFO, for explaining he reurns of relaed asses. Second, o es which variable has more incremenal informaion, he marke model esimaes he abnormal reurn of real esae operaing firms, and ess wheher or no NI and FFO are informed. Thirdly, his work compares he forecasing abiliy of

18 Conemporary Managemen Research 288 NI and FFO. Based on he resuls of hese hree mehods, we horoughly analyze informaion conen, incremenal informaion and forecasing abiliy of NI and FFO. The empirical resuls show ha using a model incorporaing boh NI and FFO, undersaes he explanaory power of wo variables since mulicollineariy occurs. Therefore, we separaely esimae he informaion conen of NI and FFO, finding ha he explanaory power of NI is higher han ha of FFO, hough he forecasing abiliy of FFO is higher han ha of NI. However, he differences beween he explanaory power and forecasing abiliy of NI and FFO are very small. These resuls indicae ha NI and FFO are boh informaive for income-producing real esae in Taiwan. Hence, he invesors of hese relaed asses should consider he informaion of a company s NI and FFO. This research did no accoun for ransiory earnings ino FFO due o no ransiory earnings daa available in Taiwan. Fuure research should consider his aspec since he explanaory power of FFO may be greaer when considering he daa of ransiory earnings. REFERENCES Ball, R., & Brown, P. (968). An empirical evaluaion of accouning income numbers. Journal of Accouning Research, 6(2), Bernard, V. L., & Sober, T. L. (989). The naure and amoun of informaion refleced in cash flows and accruals. The Accouning Review, 62(4), Dhaliwal, D., Subramanyam, K. R., & Trezevan, R. (999). Is comprehensive income superior o ne income as a measure of firm performance. Journal of Accouning and Economics, 26(-3), Gore, R., & So, D. M. (998). Toward a more informaive measure of operaing performance in he REIT indusry: ne income vs. funds from operaions. Accouning Horizons, 2(4), Graham, C. M., & Knigh, J. R. (2000). Cash flows vs. earnings in he valuaion of equiy REITs. Journal of Real Esae Porfolio Managemen, 6(), Kallunki, J. P., Marikainen, M., & Marikaninen, T. (998). Accouning income, income componens and marke-o book equiy raios: finnish evidence. The Inernaional Journal of Accouning, 33(3), Livna, J., & Zarowin, P. (990). The incremenal informaion conen of cash flow componens. Journal of Accouning and Economics, 3(), NAREIT (2002). NAREIT whie paper on funds from operaions (Naional Associaion of Real Esae Invesmen Truss). Washingon, D. C.: NAREIT, Inc.

19 Conemporary Managemen Research 289 Skinner, D. J. (999). How well does ne income measure firm performance? A discussion of wo sudies. Journal of Accouning and Economics, 26(-3), 05-. Sunda, R. A., & Typpo, E. (2004). The relevance of earnings and funds flow from operaions in he presence of ransiory earnings. Journal of Real Esae Porfolio Managemen, 0(), Vincen, L. (999). The informaion conen of funds from operaions (FFO) for Real Esae Invesmen Truss (REITs). Journal of Accouning and Economics, 26(-3), Wilson, G. P. (986). The relaive informaion conen of accruals and cash flows: combined evidence a he earnings announcemen and annual repor release dae. Journal of Accouning Research, 24(3), Wilson, G. P. (987). The incremenal informaion conen of he accruals and funds componens of earning afer conrolling for earnings. The Accouning Review, 62(2),

20 Conemporary Managemen Research 290 APPENDIX: LIST OF SAMPLE COMPANIES Code Company Name in English Marke of Lising 270 WANHWA ENTERPRISE COMPANY Taiwan Sock Exchange 2702 HOTEL HOLIDAY GARDEN Taiwan Sock Exchange 2704 THE AMBASSADOR HOTEL, LTD. Taiwan Sock Exchange 2705 THE LEOFOO DEVELOPMENT CO., LTD. Taiwan Sock Exchange 2706 FIRST HOTEL COMPANY LTD. Taiwan Sock Exchange 2707 FORMOSA INTERNATIONAL HOTELS CORPORATION Taiwan Sock Exchange 570 JANFUSUN FANCY WORLD OTC 5703 THE LANDIS TAIPEI HOTEL CO., LTD OTC 5704 HOTEL ROYAL CHIHPEN OTC 290 SHIN SHIN CO LTD. Taiwan Sock Exchange 2903 FAR EASTERN DEPARTMENT STORES LTD. Taiwan Sock Exchange 290 TONLIN DEPARTMENT STORE CO.,LTD. Taiwan Sock Exchange