36
Journal of Accounting Research Vol. 36 Supplement 1998 Printed in U.S.A. Revalued Financial, Tangible, and Intangible Assets: Associations with Share Prices and Non-Market-Based Value Estimates MAR Y E. BARTH* AN D GRE G CLINCHf 1 . Introduction This study investigates whether relevance, reliability, and timeliness of Australian asset revaluations diflFer across types of assets, including in- vestments, property, plant and equipment, and intangibles. We also in- vestigate whether they differ if the valuation amount is determined by the firm's Board of Directors or an independent appraiser, for more ver- sus less timely valuations, and for revalued amounts that are above or below historical cost.^ We base our inferences on the association between •Stanford University; fAustralian Graduate School of Management. We appreciate helpful com ments and suggestions by workshop participants at the X'i'i?, Journal of Account- in g Research Conference, the NYU Intangibles Research Conference, especially discussant Jon Low, Massey University, the University of Sydney, the University of Tasmania, the 1997 Australian Graduate School of Management Finance and Accounting Research Camp, and the 1997 American Accounting Association Financial Accounting and Reporting Section conference, especially discussants Mark Lang and Jim Leisenring, and an anonymous re- viewer. We also appreciate the research assistance of Kazbi Kothavala and Kerry Pat- tenden and funding by the Class of 1969 Faculty Fellowship and Financial Research Initiative of the Stanford University Graduate School of Business, and New York Univer- sity's Leonard N. Stern School of Business. We also thank IIBIEIS for permitting use of their analyst forecast data and New York University's Leonard N. Stern School of Business for enabling access to the IIBIEIS data. ' Throughout, we use the term "revaluations" to refer to recognized revalued amounts associated with assets that have been revalued. We use the phrase "current-year revalua- tions" to refer to revaluations made in the current year. 199 Copyright © , Institute of Professional Accounting, 1999

AFR L3 12 Barth Clinch 1988

Embed Size (px)

Citation preview

Page 1: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 1/36

Journal of Accounting ResearchVol. 36 Supplement 1998

Printed in U.S.A.

Revalued Financial, Tangible,

and Intangible Assets:

Associations with Share Prices and

Non-Market-Based Value EstimatesMARY E. BARTH* AND GREG CLINCHf

1. Introduction

This study investigates wh ethe r relevance , reliability, and timeliness ofAustralian asset revaluations diflFer across types of assets, including in-vestments, property, plant and equipment, and intangibles. We also in-vestigate whether they differ if the valuation amount is determined bythe firm's Board of Directors or an independent appraiser, for more ver-sus less timely valuations, and for revalued amounts that are above orbelow historical cost.^ We base our inferences on the association between

•Stanford University; fAustralian Graduate School of Management. We appreciatehelpful com me nts and suggestions by workshop participan ts at the X'i'i?, Journal of Account-

ing Research Conference, the NYU Intangibles Research Conference, especially discussantJon Low, Massey University, the University of Sydney, the University of Tasmania, the 1997Australian G raduate School of Manag ement Finance and Accounting Research Cam p, and

the 1997 American Accounting Association Financial Accounting and Reporting Sectionconference, especially discussants Mark Lang and Jim Leisenring, and an anonymous re-viewer. We also appreciate the research assistance of Kazbi Kothavala and Kerry Pat-tenden and funding by the Class of 1969 Faculty Fellowship and Financial ResearchInitiative of the Stanford University Graduate School of Business, and New York Univer-

sity's Leonard N. Stern School of Business. We also thankIIBIEIS

for permitting use oftheir analyst forecast d ata and New York University's Leon ard N. Stern School of Businessfor enabling access to the IIBIEIS data.

Page 2: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 2/36

20 0 ENHANCING THE FINANCIAL REPORTING MOD EL: 19 98

the recognized amounts for various categories of revalued assets of Aus-tralian firms and share prices and a non-market-based estimate of firmvalue, which is based on analysts' earnings forecasts.

Our evidence bears directly on revaluation practices under AustralianGAAP. However, our evidence also bears indirectly on current issues fac-ing the Financial Accounting Standards Board (FASB) in the UnitedStates. U.S. CAAP requires disclosure, and the FASB is considering re-quiring recognition, of fair values of all financial instruments. Currently,there is no U.S. proposal to disclose or recognize nonfinancial assets atfair value (FASB [1991; 1996]). Yet, fair values of all assets likely are rel-evant to financial statement users. One reason the FASB distinguishesfinancial and nonfinancial assets is the belief that fair values are not re-

liably estimable for nonfinancial assets, especially intangible assets andtangible assets whose value in use varies from exit or entry value (Barthand L andsm an [19 95]). Because Australian GAAP perm its revaluing alllong-lived assets at fair value (and many Australian firms do so), exam-ining Australian revaluations by asset class permits us to test this belief.

Also, Australian GAAP permits revaluations based on independent ap-praisers ' or direc tors' value estimates, which may differ in reliability, anddoes n ot re qu ire revaluations every year, possibly affecting relevanceand timeliness.2 We investigate all of these possibilities.

Although revaluations of appreciated assets are discretionary under

Australian GAAP, revaluations of impaired assets are required, as theyare under U.S. GAAP However, determining whether a long-term assetis impaired and the amount of the impairment requires considerablejudgment.^ The most notable change that would result from adoptingfair value accounting for long-lived assets under U.S. GAAP is recogni-tion of such assets at amounts in excess of depreciated historical cost.Australian firms afTord us an opportunity to provide evidence on this di-mension of the fair value accounting debate. Specifically, we investigatewhether revalued amounts in excess of historical cost are value relevantand investigate the relation between share returns and revaluations thatwould not be permitted under U.S. GAAP. *

Our primary findings are based on estimating relations between shareprice, or a firm value estimate based on analysts' earnings forecasts,and operating earnings, book value of equity minus the book values for

2 Australian GAAP permits considerable discretion regarding asset revaluations, includ-ing whether and when to revalue upward appreciated assets. Thus, effects of discretioncan affect our inferences. Although we interpret our findings with this possibility in mind,we leave to future research a comprehensive study of the effects of discretion.

^ Because of the diverse measurement and disclosure practices relating to asset impair-ment , t he EASB issued Statement of Einancial Accounting Standards No. 121 (EASB [1995]) in

1995, which clarifies GAA Prelating to impairme nt of long-lived assets (see, e.g., SEAS No. 121

Page 3: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 3/36

REVALUED FINANCIAL, TANG IBLE, AND INTANGIBLE ASSETS 2 0 1

the asset classes we investigate separately, and am ounts recognized at costand at revalued amounts for those asset classes. The sample comprises350 publicly traded Australian firms for 1991 through 1995: the largest100 firms and a random sample of 250 smaller firms. Because the rela-tions we estimate likely difFer across industries, we pr ese nt sepa rate find-ings for firms in the nonfinancial, mining, and financial industries.

Regarding asset class, we find that, as expected based on prior researchon fair values of financial assets, revalued investments are consistentlysignificantly associated with share prices, except for investments of non-financial firms in associated companies. We also find that revalued in-tangible assets are consistently significantly positively associated withshare prices, contradicting the view that such estimates are unreliable. In

fact, there is little evidence to indicate that investors distinguish recog-nized cost and revalued amounts for investments and intangible assets.^

The findings regarding revalued property, plant, and equipment (PPF)are less consistent. We find that revalued aggregate PPF is significantlypositively associated with share prices for firms in all three industries.However, whereas revalued plant and equipment is value relevant formining firms, it is insignificantly related to share prices for non financialfirms and significantly negatively related to share prices for financialfirms. Revalued property is not significantly associated with share pricesfor any industry, although it is for nonbank financial firms.

Regarding source of revaluation value estimates, we find little evi-dence indicating independent appraiser-based revaluation amounts arevalue relevant more often than director-based amounts. These findingssuggest that the relevance of directors' private information about assetvalues can outweigh potential effects of directors' self-interest on the es-timates. Regarding age of revaluations, we find that revalued intangibleassets are value relevant regardless of age, and that investment revalua-tions from other than the current year are value relevant for nonfinan-cial firms. Surprisingly, revaluations of PPEm ore tha n three years old arevalue relevant more consistently than more timely revaluations. Takentogether, these findings suggest that lack of revaluation timeliness doesnot eliminate the revalued amounts' relevance.

Finding that revalued amounts are significantly associated with shareprices suggests they have impUcations for firms' future profitability. AsBernard [1993] suggests, we investigate this directly by estimating the re-lation between the revalued amounts and a non-market-based estimateof firm value, based on the present value of analysts' forecasts of futureearnings. Findings using this firm value estimate generally corroborate

^ Under Australian GAAP, investments are not recognized based on the equity method.

Page 4: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 4/36

20 2 MARY E. BARTH AND GREG CLINCH

those using share prices, increasing our confidence that revalued am oun tsreflect asset value estimates.

We next estimate the relation between share returns and current-year

revaluations, pa rtitioned by wheth er the revaluation affects n et incom eor is recognized directly in shareholders' equity. We find that aggregaterevaluations recognized directly in earnings are significantly positivelyrelated to share returns for nonfinancial firms, but those recognized inequity are not. The reverse is true for mining and financial firms. Furtheranalyses reveal that the positive earn ings re lation primarily is attribu tableto PPE and intangible asset revaluations, whereas the positive equity re-lation primarily is attribu table to investmen ts. We find little difference inthe relation with returns for director- or independent appraiser-based

revaluations.Relating to am ounts no t observed un de r U.S. GAAP, we first estimatethe price specifications after partitioning revalued asset amounts intothose stated above cost (not permitted under U.S. CAAP) and those statedbelow cost (required under U.S. GAAP). We find revalued amounts forinvestments and PPE generally are value relevant regardless of this par-titioning; data limitations preclude conducting this analysis for intangi-ble assets. Second, we find that upward revaluations recognized in equityare positively related to returns, although not significantly so for non-financial firms. However, revaluations recognized in earnings, regardless

of sign, are negatively associated with returns, with the exception of down-ward revaluations for nonfinancial firms. Downward revaluations recog-nized in earnings, i.e., those required by Australian GAAP and U.S. GAAP,are significantly positively related to returns for nonfinancial firms. Ourfinding that upward revaluations recognized in earnings have a negativerelation with returns suggests the market discounts discretionary earn-ings increases, because upward revaluations are discretionary. We findlittle difference between the returns findings for director- and indepen-dent appraiser-based revaluations.

Finally, because our sample comprises firms of disparate size, we alsoestimate the price and returns regressions separately for two size-parti-tioned samples. We find little difference between them for revalued in-vestments and intangible assets, but revalued PPE is more consistentlyvalue relevant for smaller nonfinancial and financial firms. Also, revalu-ations by smaller firms are somewhat more consistently significantly as-sociated with returns than those of larger firms.

Section 2 describes Australian GAAP for revaluations and reviews re-lated research. Section 3 develops the estimation equations and section4 describes the data and descriptive statistics. Sections 5 and 6 present

findings from the price and non-market-based value estimate regressionsand returns specifications, respectively. Section 7 presents findings from

Page 5: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 5/36

REVALUED FINANC L\L, TANGIBLE, AND INTANGIBLE ASSETS 2 0 3

2. Australian GA AP for Revaluations and Related Research

2. 1 AUSTRALIAN GAAP FOR REVALUATIONS

Australian GAAP (ASRB 1010 [1987], as amended by AASB 1010[1993]) permits firms to revalue noncurrent assets upward when the as-set's recoverable a m ou nt exceeds its carrying am ou nt an d requires firmsto revalue noncurrent assets downward when the asset's recoverableamount falls below its carrying amount. The standard defines recoverableamount as the net amount expected to be recovered through net cashinflows arising from the asset's continued use and subsequent disposal.^

Asset revaluation increments can be recognized only if all assets in anasset class are revalued, i.e., selective upward revaluations are not per-

mitted. However, revaluation decrements, when required, are permittedfor individual assets. Revaluation increments are credited directly to anequity asset revaluation reserve, unless the increment reverses a previ-ous decrement for the same class of assets that was recognized in earn-ings. In that case, the increment is recognized in earnings to the extentof the previous decrement recognized in earnings. Analogously, revalu-ation decrements are recognized as expenses unless the decrement re-verses an existing previous increment for the same class of assets thatwas credited directly to an equity revaluation reserve. In that case, thedecrement is recognized as an adjustment to the revaluation reserve, to

the extent of the previous increment recognized in the revaluation re-serve.' The standard applies to all noncurrent assets other than invento-ries or foreign currency monetary assets, except that goodwill can berevalued only downward.

Under Australian GAAP, required disclosures for revalued assets in-clude, for each class of revalued asset, the year of the revaluation andwhether the carrying amounts were determined by an independent val-uation. For revaluations based on independent valuations, the revaluedamount equals the amount to which the independent valuer certifies andthe valuer's name is disclosed in the annual report. Disclosures also in-clude, by class of asset, the amount before accumulated depreciation of

^ Recoverable amount can be calculated based on the present value or nominal value ofthese cash flows, at the discretion of management, and the calculation method must bedisclosed. Approximately 30% of our sample firms discount future cash flows in determin-ing recoverable amount. For upward revaluations, the revalued carrying amount need notequal, but must not exceed, recoverable amount, although it is rare for upward revalua-tions to be based on recoverable amount. In contrast, downward revaluations must be torecoverable amount.

'When depreciable assets are revalued, any balances in accumulated depreciation are

credited to the asset account to which they relate and subsequent depreciation is based onthe revalued amount. As revalued assets are depreciated, depreciation of the revaluedamount is consid ered realized. Any gain or loss on revalued assets is the difference between

Page 6: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 6/36

2 04 MARY E. BARTH AND GREG GLINGH

assets carried at revalued amounts and at cost. We exploit these disclo-sures to provide evidence on value relevance of revalued assets by class

of asset and hy source and age of revaluation.2. 2 RELATED RESEARCH

2.2.1. Asset Revaluations. This study directly contributes to the litera-ture investigating asset revaluations of Australian and U.K. firms. Somestudies (e.g.,Amir, Harris, and Venuti [1993] and Barth and Clinch [1996])investigate asset revaluations using reconciliations between domestic andU.S. GAAP earnings and shareholders' equity, which the U.S. Securitiesand Exchange Commission (SFC) requires for foreign firms trading equityshares in U.S. markets. Although the reconciliations permit direct com-

parisons of cost and revalued amounts for the same assets, they do notperm it investigating disaggregated asset revaluations. M oreover, most cross-listed revaluation firms are U.K. firms, which g enerally only revalue PPE.

For a combined U.K. and Australian sample. Amir, Harris, and Venuti[1993] find some evidence of value relevance for revaluation-relatedreco nciling items, whereas Barth an d Clinch [1996] find that neither U.K.nor Australian asset revaluations are positively correlated with informationinvestors use in setting share prices.

Other studies investigate asset revaluations by firms that do not nec-essarily trade shares in U.S. securities markets. Investigating Australianfirms. Brown, Izan, and Loh [1992], Henderson and Goodwin [1992],Whittred and Chan [1992], and Cotter [1997] focus on managements'motivations for revaluing assets, whereas Sharpe and Walker [1975], Brownand Finn [1980], Standish and Ung [1982], and Emanuel [1989] inves-tigate the impact of revaluation announcements on share prices. Inves-tigating U.K. firms, Aboody, Barth, and Kasznik [forthcoming] find thatPPF revaluations have predictive power rega rding future profitability.

The most closely related study is Easton, Eddey, and Harris [1993](henceforth EEH), which investigates value relevance of Australian asset

revaluations for 72 industrial firms from 1981 to 1990. EEH find thataggregate revaluation reserve increments have significant explanatorypower for retu rns over earnings and ea rnings chang es, and that the levelof the aggregate revaluation reserve has significant explanatory powerfor price-to-book ratios. They also find that including the revaluation re-serve in book value results in price-to-book ratios closer to one and withlower variance than those obtained when excluding the revaluation re-serve. EEH interpret their findings as indicating asset revaluations helpalign market and book values of equity, although revaluations are nottimely. Bernard [1993] notes that EEH's finding of value relevance for

revaluations is particularly interesting because property, i.e., land andbuildings, is the primary target of revaluations for EEH's sample firms,

Page 7: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 7/36

REVALUED FINANCIAL, TANG IBLE, AND INTANGIBLE ASSETS 2 0 5

an indirect measure of expected future operations. We provide directevidence on these two issues by investigating separately property revalu-

ations and by investigating the association between revaluations and firmvalue based on forecasts of future earnings.

This study extends FEH by partitioning assets into major asset classesand, as Bernard [1993] suggests, by estimating the relation between re-valued amounts and an estimate of firm value based on analysts' earningsforecasts (Frankel and Lee [1996]). Also, we investigate separately non-financial, mining, and financial services firms, provide evidence on valuerelevance for three major classes of assets: investments, PPE, and intan-gibles, and subclasses within these three, and investigate whether cap-ital market participants interpret differently director- and independent

appra iser-determ ined value estimates and wh ether the age of revaluationsaffects their value relevan ce. We also explore w heth er value relevance ofasset revaluations differs for revalued am oun ts above and below h istoricalcost and investigate whether the relation between returns and revalua-tions differs for revaluations initially recognized directly in equity and inearnings. Because our 1991-95 sample period includes growth and re-cession years, we investigate separately upward and downward revalua-tions. Finally, we investigate differences associated with firm size, analystfollowing, and asset turnover in revaluations' value relevance an d relationwith returns.^

2.2.2. F inancial Assets. By investigating revalued investments, this studyalso con tribute s to th e financial assets fair value literature (e.g.. Landsm an[1986], Barth [1991], Bartb, Beaver, and L andsman [1992], Barth [1994],Bernard, Merton, and Palepu [1995], Barth, Beaver, and Landsman[1996], Fcche r, Ramesh, and Thiagarajan [1996], Nelson [1996], andVenkatachalam [1996]). These studies' findings indicate that fair valueestimates are value relevant for at least some financial assets, particularlybanks' investment securities, which are listed investments. The invest-ments asset class we investigate is composed primarily of associated com-

pany and listed investments. Also, we report evidence for three majorindustry groups, including financial firms. In Australia the financial firmgroup includes four large banks, and a few smaller and regional banks,plus nonbanking financial sector firms, so it is m ore diverse than the U.S.bank samples in the cited studies.

2.2.3. Nonfinancial Assets. This study also contributes to the litera-ture investigating asset value-related estimates for nonfinancial assetsof U.S. firms. Although the estimates these studies investigate are not

8 Easton and Eddey [1997] extend EEH to 1991-93. They find that the revaluation re-

serve increment (change in revaluation reserve increment) individually provides significantexplanatory power for returns incremental to earnings and changes in earnings for 1991(1992). In 1993, the revaluation reserve increment and change in revaluation reserve in-

Page 8: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 8/36

2 0 6 MARY E. BARTH AND GREG CLINCH

estimates of fair value per se, the estimates likely are more closely relatedto fair values than to historical costs. The studies provide m ixed evidenceregarding the estimates' value relevance. For exam ple. Bell [1983], Mag-liolo [1986], and Harris and Ohlson [1987] present mixed findings re-garding the incremental explanatory power of oil and gas disclosuresunder SFAS No. 19 (FASB [1977]), which on e can view as fair value disclo-sures. W hen investigating cur ren t cost and constant dollar disclosures re-lated to inventory and PPE under SFAS No. 33 (FASB [1979]), Beaver andLandsman [1983], Beaver and Ryan [1985], and Bernard and Ruland[1987] find no evidence of value relevance of the current cost or constantdollar amounts incremental to book values. Other studies (e.g., Bublitz,Frecka, and McKeown [1985], Murdoch [1986], Haw and Lustgarten

[1988], Hopwood and Schafer [1989], and Lobo and Song [1989]) findincremental explanatory power for current costs in particular settings.O ne exp lanation these studies offer for lack of value relevance is tha t thedisclosed am ounts con tain nontrivial estimation erro r.

3. Research D esign

3.1 PRICE REGRESSIONS

We seek to assess the value relevance of revaluations as a recognitionbasis for assets, where value relevance refers to th e ability of revalued assetamounts to reflect information relevant to investors. We begin by usingshare price as a summary measure of information relevant to investorsand investigate the ability of recognized financial statement amounts toexplain this measure, based on (1).

Pn = Wo + w\BVEn + w^NIn + (Hit (1)

where P is share price of firm i at time t, EVE is book value of equityper share, and NI is net operating income per share. We include BVEand M i n (1) as summ ary m easures of information reflected in finan-cial statement accounting numbers. We include WQ ^"^d ^it ^'^ capturethe portion of price unexplained by BVE and NI. EEH also use (1),without the intercept, as the basis for their regression equations. Equa-tion (1) also is consistent with the theoretical model in Ohlson [1995].

Book value of equity and earnings are the explanatory variables in(1), yet we seek to determine whether revalued assets' value relevancevaries across asset classes, source of value estimates, or age of revaluedam oun ts. To this en d, we partition BVE as follows:

BVE = BV+ INVEST_COST+ INVEST_REVAL + PPE_COST + PPFLREVAL

+ INTAN_COST + INTAN_REVAL (2)

where BV is book value of equity after subtracting recognized amounts

Page 9: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 9/36

REVALUED FINANGIAL, TANG IBLE, AND INTANGIBLE ASSETS 2 0 7

cost or at revalued amounts. When investigating subclasses of assets andthe source or age of revaluations, we use alternative partitions of BVF.

Substituting (2) for BVE in (1), and estimating a separate coefficientfor each variable in (2), results in our primary regression equation:

P =aQ+ a^BV+ a^NI* a^DISC + a^INVFST_COST + a^INWST_REVAL

+ a^PPF_COST+ ajPPF^RFVAL + agINTAN_COST

+ agINTAN_RFVAL + e (3)

where P is share price as of fiscal year-end. We also include an addi-tional variable in (3), DISC, which represents valuation increments ordecrements relating to investments, property, plant, and equipment,

and intangibles disclosed in footnotes but not recognized in the finan-cial statements, e is the regression error term.^ All variables are deflatedby number of shares outstanding and firm and time subscripts are sup-pressed.^" We predict all coefiicients in (3) to be positive; a coefficientindistinguishable from zero indicates the associated variable is not valuerelevant. Because (3) includes components of book value of equity andnet income, we cannot predict coefficient magnitudes (Ohlson [1995]).However, we report tests of equality of various combinations of coeffi-cients in (5) to test whether cost or revalued amounts are priced by in-vestors difi^erently from each other or from other assets.^^

3.2 NON-MARKET-BASED FIRM VALUE REGRESSIONS

Bernard [1993] suggests using benchmarks based on estimated futureoperating profitability to investigate whether revalued asset amounts arevalue relevant. Using such a benchmark provides evidence on whether

^ EEH also include revaluation increm ents in their price regressions. However, becauseof severe multicollinearity, they report regression summary statistics from estimating equa-tions that include only the aggregate revaluation reserve or the revaluation increments.Thus, we include in (3) only asset revaluations, which is the analogue of EEH's revaluation

reserve variable. In section 6, we report findings relating to revaluation increments. Also,one can interpret EEH's price regression as permitting the coefficient on current-yearrevaluations, i.e., the revaluation increment, to differ from that on revaluations from prioryears, i.e., the aggregate revaluation reserve. In section 5.4, we permit the asset revaluationcoefficients to vary for three age groups, including the current year.

'"Deflation mitigates potential effects of scale on our inferences. Untabulated findingsreveal our inferences are insensitive to using book value of equity (as in EEH) or sales asalternative deflators and estimating the equations in undeflated form, but including sales,number of shares, or V, defined below, as additional independent variables (Barth andKallapur [1996]). Our inferences also are unaffected by using prices in (3), analyst earn -ings forecasts in (4), and returns in (8) as of three months after year-end.

"These tests assume that within an asset class revalued and nonrevalued assets areeconomically similar. Because upward revaluations are discretionary, revalued and non-revalued assets could be economically different. To the extent they differ, our tests of

Page 10: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 10/36

2 0 8 MARY E. BARTH AND GREG GLINCH

the link between asset values and future operations is important forvalue relevance. Thus, we estimate versions of (3) where the de pe nd en tvariable is an estimate of firm value based on analysts' earnings forecasts:

V = bQ+ biBV+ b^NI* b^DISC+ b^INVEST_COST

+ br,INVEST_REVAL + b^PPFLCOST + b'jPPE^REVAL

+ bglNTAN.COST + bglNTAN^REVAL + u. (4)

Variables other than V are as previously def ined. Following Ohlson

[1995] and Frank el an d Lee [1996] , Vis based on the fol lowing equ atio n:

1 - rBVE t)/{l + r) + {FEt.,^ ' rBV Ei^i)/{r{l + r)) (5 )

where FJE^^-I is the I/B/E/S median earnings forecast as of the end of thecurrent fiscal year, t, for following year, t+ 1; FEf.^.^ i* "^^e IIBIEIS medianearnings forecast at year t for year t + 2; and r is tbe d iscount ra te, whichwe assume equals 10%. EVE, is shareho lders' equity at year t and £VE( + iis calculated usitig "clean surplus," which requires a forecast of the yeart + 1 dividend. We use the year t dividend multiplied by the dividendgrowth rate over the past five years as the dividend forecast. Thus, V isthe present value of forecasted abnormal earnings for two years, plusdiscounted abnormal earnings for the remaining years to infinity, as-suming abnormal earnings for the remaining years equal year t + 2 ab-

normal earnings. We use analysts' forecasts for two years because IIBIEIStypically does not include forecasts beyond two years for Australianfirms. To the extent revaluations of long-term assets affect forecasts ofearnings beyond two years, our ability to detect a significant relation isimpaired.

Equation (5) simplifies to:

V, = FE,^^l{r{l + r)) + {dtg,)l{l + r) (6)

where g, is the time t dividend growth rate (Penman [1996]). Thus, V

reflects the present value of analysts' expected future earnings plus thepresent value of dividends between time t and the earnings forecast pe-riod. Representing firm value as the present value of future earningsis standard in the valuation literature (see, e.g.. Miller and Modigliani[1966]). However, rather than basing the value estimate on, e.g., pastearnin gs as an estimate of future earn ings . Vis based on analysts' forecastsof future earnin gs. ^

3.3 RETURNS REGRESSIONS

Findings from estimating (3) and (4) provide evidence on whether

revalued assets are value relevant, whether their implicit valuation co-

Page 11: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 11/36

REVALUED FINANCIAL, TANGIBLE, AND INTANGIBLE ASSETS 209

efficients differ from those of assets recognized at cost, and whether es-

timation error in revalued amounts is sufficient to eliminate their value

relevance. Because we cannot disaggregate revalued amounts into their

cost and revaluation components, we cannot establish the incremental

value of revalued amounts, given cost data. Also, we cannot distinguish

whether cost-based amounts are value relevant because they are corre-

lated with assets' values and/or whether revalued amounts are value rel-

evant because they are correlated with assets' costs.

To provide some evidence on this question and to investigate the time-

liness of current revaluations, we investigate whether annual share re-

turns are associated with current-year revaluations. As before, (1)

provides the basis for the estimating equation. As in EEH, note that:

ABVE = NI-DIV+ RRI+ other (7)

where A^VEis the change in book value of equity in year t, DTV is divi-

dends, and RRI is revaluation reserve increment. That is, RRI is the

amount of upward or downward asset revaluation for the year recog-

nized directly in equity, other represents changes to equity other than

from earnings, dividends, and increments to revaluation reserves. Thus,

first-difTerencing (1), substituting (7) for ABVE, and deflating all vari-

ables by beginning-of-year price yields:

RET = k^ + kiNI + k^ANI + k^RRI + V (8)

where RET is raw return, i.e., P^ + DT V^ - P .j / P^.j, v includes other,

and ^0 is W Q I - WQM- ^ denotes annual change.

As in (3) and (4), we disaggregate RRIinto revaluation increments by

asset class. Also, recall that some revaluations, e.g., write-downs of assets

not previously written up, are recognized in net income, RRIPL. Thus,

we also partition earnings into operating earnings before revaluations

and RRIPLby asset class, and estimate versions of the following returns

equation:

RET = Co + q M + C 2 A M + c^INVEST_RRI + c^PPE_RR[

C Q I N V E S T _ R R I P L + cyPPE_RRIPL

(|) (9)

where RET is the firm's 12-month raw share return ending at fiscal

year-end, RRI denotes the revaluation reserve increment for year t, and

RRIPL denotes revaluations recognized in earnings. Other variables are

as defined previously.^^

include ARRIin their returns specification, although, as with their price regres-sions, they report findings only for regressions that include either RRI or ARRI, but not

Page 12: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 12/36

2 1 0 MARY E. BARTH AND GREG GLINGH

4. Data and Descriptive Statistics

4. 1 SAMPLE FIRMS AND DATA

The sample is composed of the 100 largest companies listed on theAustralian Stock Exchange {ASX), as measured by m arket value of equityas of Ju ne 30,19 96, and a rando m sample of 250 firms selected from theremaining Australian firms traded on the ASX with market value of eq-uity greater than A$10 million. Seven hundred seventy-six of the 1,171com panies with Jun e 30, 1996 ASX share prices meet our market capital-ization criterion, indicating that one-third of traded AustraUan firmshave market value of equity less than A$10 million. Nonetheless, oursample firms represent 8 1% of the total market capitalization of the ASX

dom estic stocks. We exclude 18 and 3 foreign-domiciled firms from thetop 100 and random sample firms, respectively, because they do not fol-low Australian GAAP. We include a firm in our sample for the years ithas data the equations require. We select the 100 largest firms to facili-tate comparisons with EEH and extend the sample to smaller firms tofacilitate generalization of our findings. The sample period is 1991-95.We obtain financial statement data from firms' annual reports to share-holders and market data from the Australian Gradu ate School of M anage-m ent's C entre for R esearch in Finance share price file. We obtain analysts'earnings forecasts from IIBlEtS.

4.2 DESGRIPTIVE STATISTIGS

Table 1, panel A, presents industry and calendar year breakdowns ofthe sample firms. It reveals that n o single indu stry dom inates the sam ple,except gold m ining firms, which comprise a large fraction of the minin gindustry sample. We prese nt separate findings for firms in the nonfinan -cial, mining, and financial industries because the relation between shareprices, or the non-market-based measure of firm value, and revaluedam oun ts likely differs across industries. Data limitations p reclude us from

using more refined industry classifications. Because we select our samplebased on 1996 market value of equity and do not require firms to haveavailable data for all sample years, panel A also reveals that the samplesize increases over time.

Table 1, panel B, reveals small variations in m arke t capitalization acrossindustries but large variations in total assets and sales. Financial firms,on average, have large total assets, with a skewed distribution, reflect-ing their typical asset structure and the presence of some firms withextremely large total assets. Mining firms, on average, have the smallestsales, partially reflecting the fact that some mining firms are in the ex-

ploration stage and thus have large assets and/or market capitalizationbut small sales. These industry differences in accou nting am ounts refiect

Page 13: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 13/36

T A B L E 1

Descriptive Statistics Relating to Sample of Publicly T raded Australian Firms from 1991 to 1995

Panel A: Industry and Calendar-Year Sample Com position

Industry 1995 1994 1993 1992

Total

1991 Observations Companies

NonfinancialDevelopers and Contractors

Building Materials

Alcohol and Tobacco

Food and Household Goods

Chemicals

Engineering

Paper and Packaging

Retail

Transport

Media

Miscellaneous Services

Miscellaneous Industrials

Diversified Indtistrials

Tourism and Leisure

Total Nonfinancial

Mining

Gold

Other Metals

Solid Fuels

Oil and Gas

Diversified Resources

Total Mining

Financial

Banks

Insurance

Entrepreneurial Investors

Investment and Financial Services

Property Trusts

Total Financial

Total

78

5

6

3

3

25

5

12

13

13

6

8

96

39

15

211

2

69

8

5

4

24

15

56

22 1

6

6

4

6

23

23

3

9

6

9

6

6

71

25

13

3

12

255

8

5

2

17

13

45

17 1

77

4

5

2

3

25

3

8

9

8

6

5

74

27

123

13

2

57

8

4

3

19

9

43

17 4

6

7

3

5

2

21

4

2

7

8

9

5

4

65

2610

3

13

2

54

8

23

16

5

34

153

4

6

24

2

21

3

23

8

8

5

3

53

24

7

213

1

47

7

0

1

14

5

27

12 7

30

34

18

26

11

13

8

2015

39

44

47

28

26

359

141

57

13

62

9

282

39

16

13

90

47

205

846

5

6

3

3

2

6

5

12

14

14

6

8

100

42

15

3

14

2

76

5

4

26

15

58

234

Panel B: Size of Sample Firms

Variable Mean Median Standard Deviation

Market Capitalization

Nonfinancial

Mining

Financial

Total Assets

Nonfinancial

Mining

Financial

Sales

Nonfinancial

Mining

Financial

1,085.00

1,301.50

1,171.40

1,439.80

1,026.90

9,243.60

1,439.80

560.40

761.20

256.80

133.90

194.60

228.00

90.40

280.10

228.00

30.6045.80

1,926.50

4,343.50

3,046.30

2,826.30

3,640.00

29,943.20

2,826.30

2,154.202,104.70

Page 14: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 14/36

212 MARY E. BARTH AND GREG GLINCH

'C

J

PS

1.If

11s s

< S,

• 2

a

t

C i l > Ol O CO(?) CO O) O t ^

C) CM CTO m i-Ht o a; 1-H CM - CN

Tt CO in d eo

_ . O " CO ^H £>•O i O 00 ( i-H Od CM CO -- d d

1

in

lO

00

IO

<M

toIOCM

I-Hin

o

Ol HH HH O O

Ol C

O r.H HH

s

.CO

in

.o

to

Ol

.o

to

.o

to

d

t^ Ol r Ol oin to q q qH.; r-; d d -

OlCO

d

lOi>

d

lOOl

d

oo

d

O (N lO Ol Tf •*CO Tf Tf 00 IM CM

^H OO Tf CO O '-t

04 IM Ot t COCM (M H

CO O O H CM IO

^H to 00 to Ol Ol

1 > ;O -—I ,—« CO I-H t—ITj* O C l CO i-H I-H i-JCsi CM C d d "- i I-H rt O O O

Ol CM l> OCO CO rH T-H

to eo O4 H o O

. o Ol in Tf ^

to CM O> I> O O

' ^ ^ 6 6 S

in

Tf_

< 6

CM -^ l > O CM I-H CMI-H O O O O O O

2

2

OlOl

1

0

§

Tf

0

i>

o

to

eo

0

00

o

^ eo 00 t~ tn oiq in to Ol q qeo oi oi —^ d d

CM O O O I O O J ^ H C M ^ H

CO C O C O O I - H O O O

to O COCO Ol Ol

e

dian

u

3

gin

Tf

3

in

CO

00

CM

1

to

CM

CO

3

oo

to

3

o• *

o

to

o

CM

OO

to

CM

o

o

o

60S

H ^

CM

O

117

to

o

00

°

toto

oo00

o

to

00CM

o

CM

o

o

00

TfTf

CM

O

Ol

.

CM ^^

to

oo

00 oo

CM OO

3

CM

OO

o

00

.o

.

.51

o

.o

00

.o

,16

o

^H CM CO o into to Tf Ol ^H

oi oo oo oi d d

Tf

d

00Ol

Page 15: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 15/36

REVALUED FINANGIAL, TANGIBLE, AND INTANGIBLE ASSETS 213

rH OO rH r H rH

O M OO ^H O r^

tO Ol to CO ^~* ^O i n Tj< <M IM O

<o S r^ d d d

oo oo in in

IM CN O • *

d d d d

o in to o r

00 00 in (N (M

00 I—< o in(M IM <M • *

d d d d

soo o in in to<M rH in o o o o o o o o

T)< to O Tf rt I>

d d d d d d

c^ Ol Oi T f.. r^" O O O O O

d S d d d d

o o o o o o

to rH O (N T(> O

O <N

o o

0 0 " ^CM • ^

CO

o o

0 0

o

0 0

oo

o

oo

inO rH O rt rH

.S "

TJtC U

\f\

I S s

S o

Page 16: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 16/36

2 1 4 MARY E. BARTH AND GREG CLINGH

each asset category. It reveals that the mean (median) market-to-book ra-tio ranges from 1.30 (0.99) for financial firms to 3.01 (1.67) for miningfirms. These statistics indicate that, despite asset revaluations, on average,sample firms have net off-balance-sheet assets. Untabulated statisticsreveal that for the three samples together, share price, P, and the non-market-based estimate of firm value, V, are highly positively correlated(Pearson correlation = 0.854). Although one would expect a significantpositive correlation because both are measures of firm value, the differ-ence in the source of the estimates makes the correlation striking.

Table 2 reveals that revaluations comprise a large fraction of recog-nized amounts for all asset categories and most investment revaluationsare based on directors' valuations. Revalued investment amounts most

frequently are based on current valuations, although many mining andnonfinancial firms' investment revaluations are more than three yearsold.^'* In contrast, there are many PPE revaluations in all age categoriesfor all three industries. Regarding timing of revaluations, untabulatedstatistics reveal that there are many PPE revaluations in each year sincebefore 1981. Although some investment revaluations also predate 1981,there are almost non e between 1981 and 1985. Almost no intangible assetrevaluations predate 1988.

Table 2 also reveals that many nonfinancial firms revalue intangibleassets, but no mining and few financial firms do so.^^ However, becausemining firms include intangible mineral rights in PPE, some revaluedPPE for mining firms relates to intangible assets. Untabulated statisticsindicate th at the most comm only revalued intangible asset for nonfinan-cial firms is brands. For these firms, revalued intangibles include brandsidentified separately: 32%, brands, patents, and licenses identified as agroup: 15%, licenses identified separately: 13.5%, goodwill: 13.5%, tech-nology assets: 12%, and other, including executory contracts: 13.5%.'^Finally, table 2 reveals that most revaluations are based on directors' val-uations, except for PPE revaluations for nonfinancial and financial firms,

which often are based on independent valuations.!^

' ' ' Because m ost financial firms' investment revaluations th at are based on directo rs' val-uations also are current, our findings in table 6 below relating to age of financial firms'possessive investment revaluations could be attributable to the same revaluations as thefindings relating to director-valuation-based financial firms' investment revaluations in ta-ble 5. The statistics in table 2 suggest that other revaluation partitions we investigate are un-likely to be driven by the same observations.

'^Review of annual reports reveals that intangible assets revaluations do not relate tothe purchase of intangibles, e.g., through a purchase-business combination. Rather, intan-gible asset revaluations relate to revaluing intangibles either by recognizing internally gen-

erated intangibles or changing the carrying amount of a purchased intangible asset.'^ The distributions of revalued intangibles are similar for the large- and small-firm sub-samples, although large firms have a greater percentag e of revalued goodwill and the small

Page 17: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 17/36

REVALUED FINANCIAL, TANG IBLE, AND INTANGIBLE ASSETS 2 1 5

5. Findings from Price and N on-Market-Based Value

Regressions

5. 1 MAJOR ASSET CLASSES

Table 3, panels A and B, present summary statistics from estimating(3) and (4) together with p-values associated with tests of equality ofcoefficients. To mitigate effects of extreme observations, we exclude ob-servations with estimated residuals greater than three standard devia-tions from zero. Also, we only estimate a coefficient for a particular assetpartition if the numher of firms with nonzero observations for that par-tition exceeds five. Following Barth and Kallapur [1996], we hase all teststatistics on White [1980] standard errors. The term significant refers to

statistical significance at a levels less than 5%, using a one-sided alterna-tive, because we predict all coefficients to he positive. Tahulated findingsare hased on estimating all equations pooled cross-sectionally and inter-temporally; untahulated findings from separate-year estimations result insimilar inferences.

Looking first at the price regression findings in tahle 3, panel A, we seethat revalued amounts in all three asset classes are value relevant; thecoefficients on INVEST_REVAL, PPE_REVAL, a n d INTAN_REVAL are con-

sistently positive, as predicted, and significandy dilTerent from zero.^^ As

expected, coefficients on most other independent variahles are signifi-cantly positive. The exceptions are those on intangihle asset cost-hasedamounts, INTAN_COST, for mining firms, and other disclosed hut not rec-ognized value estimates, DISC, for nonfinancial firms.

The /j-values associated with tests of coefficient equality reveal thatthe null hypothesis that all halance sheet coefficients are equal is re-jected, except for mining firms, indicating that not all components ofhook value of equity have the same value multiple. Similarly, tests gen-erally reject the null hypotheses that all _COST coefficients are equal,whereas equality of all _REVAL coefficients is rejected only for non-

financial firms. Tests comparing coefficients on cost-based and revalua-tion-based amounts reject equality only for nonfinancial firms' PPE andintangible assets, suggesting that in most cases investors do not dis-tinguish cost- and revaluation-based amounts for investments or PPE.

the variables are similar to the Pearson correlations. The statistics also indicate that thelevel of correlation is not extreme. The average absolute correlation (rank correlation)between each of the asset cost and revaluation amounts is 0.20 (0.18), 0.31 (0.32), and0.45 (0.45), for nonfinancial, mining, and financial firms, respectively.

'*To investigate whether the significant relation is attributable to the act of revaluationrather than to the revalued amounts, we reestimated (3) after including indicator vari-ables that equal one if the firm had nonzero assets in that class and cost basis, and zero

Page 18: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 18/36

TABLE 3

Sum mary Statistics from Regression of Price, P, and Non-Market-Based Estimate ofValue, V, on

Income and Book Value of Equity Partitioned by Asset Class andValuation Basis

Sam ple of Publicly Traded Australian Firms from 1991 to 1995

Panel A: Price as Dependent Variable

Variable

Intercept

BV

NI

DISC

COST

_INVEST

_PPE

_INTAN

REVALUATION

_INVEST

_PPE

_INTAN

n

Adj. ft2

Nonfinancial

Coef. (

0.40 3.16

1.34 12.20

2.65 3.83

-0.37 -0.70

1.03 5.13

1.45 13.15

1.18 7.17

1.22 5.52

0.59 2.71

0.65 7.85

347

0.808

p-Values for Tests of Coefficient Equality

Coefficient Test

All Balance Sheet

All _COST

All _REVAL

INVEST_COST and _REVAL

PPE_COST and _REVAL

INTAN_COST and _REVAL

Nonfinancial

0.00

0.06

0.03

0.52

0.00

0.00

Mining

Coef. t

0.19 2.32

1.24 6.48

1.95 2.66

3.39 3.84

1.40 3.46

1.37 9.77

-0.24 -0.23

1.37 3.03

1.42 5.82

268

0.872

Mining

0.55

0.31

0.92

0.96

0.84

Panel B: Non-Market-Based Value Estimate as Dependent Variable

Variable

Intercept

BV

NI

DISC

COST

_INVEST

_PPE

_INTANREVALUATION

_INVEST

_PPE

_INTAN

n

Adj R2

Nonfinancial

Coef. (

1.86 7.07

0.61 4.51

2.46 3.36

-0.19 -0.56

-0.39 -1.06

0.94 7.96

-0.27 -0.96

0.82 2.82

-0.64 -2.58

0.35 2.19

184

0.684

/)-Values for Tests of Coefficient Equality

Coefficient Test

All Balance Sheet

All _COST

All _REVALINVEST_COSTand _REVAL

PPE-COST and _REVAL

Nonfinancial

0.00

0.00

0.00

0.04

0.00

Mining

Coef. t

0.28 2.73

0.43 4.07

1.75 3.92

0.45 0.83

0.42 1.95

0.66 7.91

-0.35 -0.64

3.34 3.69

1.09 7.22

12 9

0.909

Mining

0.00

0.07

0.01

0.00

0.01

Financial

Coef. t

0.81 7.09

0.68 5.27

3.85 5.68

0.78 7.32

0.26 2.33

0.41 2.15

1.41 2.45

0.33 6.23

0.55 1.91

19 5

0.932

Financial

0.00

0.00

0.44

0.44

0.74

Financial

Coef. (

1.83 3.95

0.52 2.08

1.81 2.43

-3.65 -1.69

0.22 1.03

1.47 1.97

3.07 3.42

-0.07 -0.36

-0.17 -0.26

67

0.748

Financial

0.00

0.01

0.88

0.20

0.20

Page 19: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 19/36

REVALUED FINANCIAL, TANGIBLE, AND INTANGIBLE ASSETS

TABLE 4Sum mary Statistics from Regression of Price, P, on Income andBook Value

of Equity Pa rtitioned by Asset Class and Valuation Basis

Samp le of Publicly Traded Australian Firms from 1991 to 1995

217

Variable

InterceptBV

NI

DISC

COST

^SSOdJNVEST

_LISTED_INVEST

^ROP_PPE

^&E_PPE.^GDWL_INTAN

-VARIED_mTAN

REVALUATION

-ASSOC^INVEST

-LISTED_INVEST

^ROP_PPE

.-P&FWPE

^VAR1ED_INTAN

n

Adj. /J2

Nonfinancial

Coef.

0.37

1.35 ]

2.45

-0.27 -

0.00

1.31

1.32

1.50 ]

1.86

1.10

0.46

1.22

0.38

0.91

0.85

346

0.832

/(-Values for Tests ofCoefficient EqualityCoefficient Test

All Balance Sheet

ASSOC INVEST_COST and _REVAL

USTED INVEST_COST and _REVAL

PROP PPFLCOST and _REVAL

P&E PPE^COST and _REVAL

VARIED INTAN_COST and _REVAL

t

3.08

13.25

3.98

-0.47

0.00

6.08

5.01

13.01

4.90

5.85

0.92

5.61

1.69

1.40

5.28

Nonfinancial

0.00

0.60

0.78

0.00

0.38

0.35

Mining

Coef.

0.19

1.20

1.90

3.65

1.97

1.39

1.31

1.44

-4.83 -

2.27

1.59

0.99

-1.51 -

2.19

268

0.884

t

2.19

5.86

2.68

4.56

3.07

3.02

7.77

9.11

-3.50

3.92

3.18

2.23

-1.52

5.32

Mining

0.00

0.68

0.53

0.00

0.08

Financial

Coef.

0.86

0.23

5.12

0.50

0.73

0.52

-0.40

0.85

-0.10

2.08

0.26

-0.52

-6.20

19 7

0.947

t

7.55

2.21

8.25

4.94

2.16

5.07

-0.92

5.56

-0.13

7.09

5.32

-1.78

-4.93

Financial

0.00

0.00

0,

0,

,82

,00

u. . .1 u. viv iiuLt u UJ iiuiiit^ i Ul siiaics uuLSLdiiuiiig. unin iiiiiiuiuons preciuae estimation ot coetnc

each asset partition. Coefficients are only estimated if there are nonzero observations for more than fiive firms.

Findings reported in section 5.2 below, w here we partition by asset sub-

class, indicate that the rejections of coefficient equality in table 3 are at-tributable to property, in the case of PPE, and goodwill, in the case ofintangible assets.

If we turn next to the findings from thenon-market-based estimate of

firm value, V, regression, table 3,panel B, reveals that, because calculat-ing V requires analyst coverage and earnings forecasts, the sample is

Page 20: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 20/36

2 1 8 MARY E. BARTH AND GREG GLINGH

firms, revalued PPE is significantly negatively associated with V hut sig-nificantly positively associated with price. For financial firms, revaluedinvestments and revalued PPE are insignificantly associated with V hut

significantly associated with price. Tests comparing coefficients in panelB reject equality more frequently than is the case for the price regres-sions in panel A.

In secdon 8.1 we report findings from estimadng (3) separately forthe large and smaller sample firms. Firms with analyst coverage, whichcomprise the sample in tahle 3, panel B, are comparahle to the largerfirm sample in section 8.1. Findings reported there for the larger firmsare consistent with those reported in tahle 3, panel B. This suggests thatthe inconsistencies hetween the table 3, panel A and panel B findings

are attrihutahle to differences in firm size across the two samples, not todifferences in the dependent variahles, Pand V Thus, the tahle 3, panelB findings indicate that implications for future profitahility of revaluedassets are reflected in non-market-hased estimates of firm value and inshare prices.

5. 2 ASSET SUBGLASSES

In this section, we investigate further the p oten tial for differentialvalue relevance across asset classes hy disaggregating each major assetclass into two suhclasses. EEH's findings also mo tivate this secdon's anal-

ysis. Because more of EEH's sample firms revalue property than otherassets, one interpretation of their findings is that their value relevancefinding for revaluations primarily is attrihutahle to property. Our find-ings indicate this is not the case for our sample, in that investments' andintang ihle assets' revalued am oun ts are significantly associated with shareprices and estimates of future profitahility. However, in section 5.1 we ag-gregate prop erty and plant and equipm ent; in this section, we investigateseparately its two suhclasses.^^

Tahle 4 presents results from estimating (3) hut disaggregating in-vestments into investments in associated companies and other, primarilylisted, investments; PPE into property, primarily land and buildings, andplant and equipment; and intangihle assets into goodwill and other in-tangibles. Bernard [1993] leads us to expect that investments in associ-ated companies, plant and equipment, and other intangihle assets aremore likely to be associated with firms' future profitability and, thus,share prices, than listed investments, property, and goodwill.

Findings in tahle 4 reveal that, as expected, revalued investments inlisted companies are significantly associated with prices for all three in-dustries. However, revalued investments in associated companies are

only value relevant for m ining firms. Also, revalued pro pe rty is only m ar-

Page 21: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 21/36

REVALUED FINANCIAL, TANG IBLE, AND INTANGIBLE ASSETS 2 1 9

ginally significantly positively associated with price for nonfinancial firmsand it is insignificantly associated with price for mining and financialfirms.2O Revalued plant and equipment is significantly related to price,

but only for mining firms. Unexpectedly, revalued plant and equipmentis significantly negatively associated with price for financial firms. Thereare insufficient observations to estimate a coefficient for revalued good-will, perhaps because goodwill can only be revalued downward. However,consistent with table 3, table 4 reveals that revalued intangible assetsother than goodwill are value relevant.

Probability values from tests of equality of coefficients reject, amongothers, the null hypothesis that all balance sheet coefficients are equalfor all specifications. However, the statistics fail to reject the null hy-

potheses of equal coefficients for cost and revalued amounts for severalasset subclasses. For example, for nonfinancial and mining firms, onlyequality of cost and revalued property amounts is rejected. Failure toreject these null hypotheses suggests investors often do not distinguishvaluation implications of cost and revalued amounts.

We do not tabulate the summary statistics from regressions in which V

is the dependent variable. However, the evidence from those estimationslargely is consistent with that from the price regressions, although sev-eral coefficients are no longer significantly diiTerent from zero, likelyreflecting the substantially smaller sample size. The most notable differ-

ences relate to nonfinancial and financial firms for which the coeffi-cients on revalued property are significantly negatively related to V.

5.3 DIRECTOR S VERSUS INDEPE NDE NT VALUERS

Australian firms can base revaluations on director or independent ap-praiser value estimates. We seek to determine whether value relevanceof revalued am ou nts differs by valuation sou rce. If ind ep en de nt ap-praiser estimates are more reliable than those of directors, either becauseof their asset value estimation expertise or because they lack motives tomanage financial statement amounts, then one would expect indepen-dent appraiser-based revaluations to be more value relevant than thosebased on director valuations. If directors have private information aboutasset values and reflect that information in their value estimates, thenone would expect director-based revaluations to be more value relevant.In many cases, firms disclose that the directors considered independentappraisals in arriving at their valuation.

Table 5 presents the results. It reveals that, in most cases, revalued as-sets based on director and independent appraiser valuations are sig-nificantly positively associated with price, although, again, results for PPE

2" Untabulated findings reveal that the negative, although not significatit, coefficient on

Page 22: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 22/36

220 MARY E. BARTH AND GREG CLINGH

TABLE 5

Sum mary Statistics from Regression of Share Price, P, on Income and Book Value of Equity

Partitioned by Asset Class, Valuation Basis, and Source of Valuation

Sam ple of Publicly Traded Australian Firms from 1991 to 1995

Variable

Intercept

BV

NI

DISC

COST

_INVEST

_PPE

_INTAN

REVALUATION_INVFST_DIRECTOR

_INVEST_INDEPENDENT

_PPEJOIRECTOR

_FPE_INDEPENDENT

_INTAN_DIRECTOR

_INTAN^NDEPFNDENT

n

Adj. fi2

Nonfinancial

Coef.

0.58

1.17

2.92

-0.32

1.19

1.41

0.99

1.03

-0.07

0.49

0.48

0.61

347

0.767

/(-Values for Tests of Coefficient Equality

Coefficient Test

All Balance SheetA ll INVEST_REVAL

A ll PPE^REVAL

A ll INTAN_REVAL

t

3.60

7.14

3.50

-0.64

4.34

9.35

4.73

3.39

-0.21

2.01

3.75

2.33

Nonfinancial

0.00

0,

0,

,06

,60

Mining

Coef.

0.17

1.31

2.07

3.60

1.35

1.39

0.23

1.30

1.93

0.63

268

0.875

t

2.08

6.61

2.90

4.33

3.40

9.80

0.23

3.18

5.15

1.58

Mining

0.09

0.01

Financial

Coef.

0.75

0.70

3.43

0.73

0.37

0.52

1.22

0.37

0.25

0.85

0.17

195

0.933

t

7.05

5.71

5.52

7.54

3.46

2.46

2.77

6.59

3.47

2.63

0.44

Financial

0.00

0,

0,

,17

,08

Pis share price asof fiscal year-end. BVis book value of equity after su btracting investments (INVEST),property, plant, and equipment (PPE), and intangible assets (INTAN). Mis operating income. DISC isdisclosed, but not recognized, asset value estimates. COST and ftEVAL deno te recognized am ounts basedon historical cost and revaluations. DIRECTOR an d INDEPENDENT denote revaluation based ondirec-tors' and independent appraisers' valuation. All variables are deflated by number of shares o utstanding.Data limitations preclude estimation of coefficients for each asset partition. Coefficients are only esti-mated if there are n onzero observations for more thanfivefirms.

are not consistent across the three industries.^' Interestingly, in several

cases, coefficients on director-based revalued amounts are significandylarger than those on independent appraiser-based revalued amounts, sug-gesting the director-based amounts do not have more estimation errorthan the independent appraiser-based amounts. Untabulated findingsfrom the Vregressions generally corroborate those from the price regres-sions, although some coefficients for nonfinancial and financial firms are

no longer significant. Taken together, the table 5 findings indicate direc-tors' valuations result invalue-relevant revalued amounts as often as do

independent appraisers' valuations, suggesting the relevance of direc-

tors' private information about asset values counterbalances the value

Page 23: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 23/36

REVALUED FINANCIAL, TANGIBLE, AND INTANGIBLE ASSETS 221

TABLE 6

Summ ary Statistics from Regression of Share Price, P, on Income andBook Value of Equity

Partitioned by Asset Class, Valuation Basis, andAge of Revaluation

Sam ple of Publicly Traded Australian Eirms from 1991 to 1995

Variable

Nonfinancial

Coef. t

Mining

Coef. t

Financial

Coef. T

I n t e r c e p t

BV

NI

DISC

COST

.JNVEST

..INTAN

REVALUATION

^INVEST_CURR

..INVEST_PREV2

_INVEST_OLD

.J'PE_CURR

0.37 2.90

1.38 13.90

2.18 3.75

-0.47 -0.86

1.04 4.92

1.56 14.23

1.19 7.52

.SPE^OLD

^INTAN_CURR

.JNTAN_PREV2

Adj.

1.40

1.44

0.50

0.54

0.35

0.60

1.05

0.65

0.90

348

0.807

3.28

5.31

0.88

1.58

1.22

1.55

4.93

8.52

6.13

0.26

1.22

1.73

3.19

1.54

1.35

-0.64

1.00

0.02

0.94

0.19

1.89

268

0.872

3.19

6.37

2.35

3.40

3.61

9.70

-0.57

2.36

0.05

2.03

0.33

4.86

0.79

0.59

4.91

0.65

0.26

0.34

0.78

7.23

4.96

7.72

7.36

2.42

1.93

1.42

0.30 5.99

0.21

-0.07

4.90

196

0.947

0.76

-0.14

2.13

/(-Values for Tests of Coefficient Equality

Coefficient Test

All Balance Sheet

All INVEST_REVAL

All PPELREVAL

All INTAN.REVAL

Nonfinancial

0.00

0.05

0.79

0.03

Mining

0.00

0.06

0.04

Financial

0.00

0.00

Pis share price asof fiscal year-end. BVis book value of equity after subtracting investments (INVEST),property, plant, and equipment (PPE), and intangible assets (INTAN). Nlis operating income. DISCisdisclosed, but not recognized, asset value estimates. COST and REVAL denote recognized amounts basedon historical cost and revaluations. CURR, PREV2, andOLD denote whether the recognized revaluedamo unt is based on a revaluation in the cu rren t year, the previous two years, or earlier. All variables are

deflated by number of shares outstanding. Data limitations preclude estimation of coefficients for eachasset partition. Coefficients are only estimated if there are nonzero observations for more thanfivefirms.

relevance enhancing eflFects of independent appraisers' expertise andlack of self-interest.

5.4 AGE OF REVALUED AMOUNT

Although current revaluations likely reflect current value, Australianfirms need not revalue assets every year and, thus, some revaluedamounts are hased on out-of-date, and perhaps stale, valuadons. To in-

vestigate effects onvalue relevance of revaluadon age, we reestimate (3)and (4) partitioning revalued amounts into three age categories: cur-

Page 24: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 24/36

2 2 2 MARY E. BARTH AND GREG CLINGH

age category is significandy related to price for nonfinancial firms; formining firms, current and old PPE revalued amounts are value relevant,

but not those in the PREV2 category; and for financial firms, only old re-valued amounts are value relevant.^^ Tests of coefficient estimates rejectthe null hypothesis that different age category coefficients are equal inmost cases. Untabulated findings from the V regressions generally cor-roborate those from the price regressions.

6. Findings from Returns Specifications

Thus far, the empirical analyses focus on associations between shareprice, or a non-market-based estimate of firm value, and assets recog-

nized at revalued amounts. In this section, we estimate the relation be-tween current-year revaluations and share returns. Finding a significantrelation between revaluations and returns is strong evidence that revalu-ations are value relevant and timely. However, because firms do not re-value assets every year, there are substantially fewer observations for thisanalysis and revalued amounts likely reflect value changes over severalyears, resulting in the associated valuation effects being reflected at leastpartially in share prices before the returns window. Both of these effectsreduce our ability to detect a significant relation between revaluationsand returns.

Table 7, pane l A, presen ts summary statistics from estimating (8), whichincludes revaluation reserve increments, RRI, and revaluations includedas part of earnings, RRIPL. It reveals that for nonfinancial firms, RRIPLis significantly positively related to returns, but RRI is not. The reverse istrue for mining and financial firms. Panel B of table 7 presents resultsfrom estimating (9) and indicates that RRIfor investments is significandypositively related to retu rns for bo th applicable samples, whereas RRI forPPE is significantly positively related to returns only for mining firms.Contrary to predictions and expectations based on the price regressions,

RRI for intang ible assets is significantly negatively related to ret urn s. T hefindings for RRIPL are noticeably different. For investments, RRIPL is in -significantly related to retu rns for all industries, whereas, consistent withpredictions, for PPE and intangibles, RRIPL is significandy positively re-lated to returns for all applicable samples. The asset class is "unknown"for some assets because disclosures do not always identify the asset classto which each revaluation relates.

Table 7, panel C, presents results from esdm ating (8) bu t pa rdd on ingrevaluations by source, i.e., director- or ind ep en de nt appraiser-based val-uations. Consistent with the price regressions, the findings suggest direc-

tor-based revaluations are value relevant at least as often as independent

Page 25: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 25/36

REVALUED FINANCIAL, TANGIBLE, AND INTANGIBLE ASSETS 223

T A B L E 7Summ ary Statistics from Regression of Returns on Income, Change in Income, and Current-Year

Revaluations, Partitioned by Recognition in Earnings or Directly in Equity

Samp le of Publicly Traded Australian Firms from 1991 to 1995

Panel A: Aggregate Revaluations

Variable

Intercept

NI

AN I

RR I

RRIPL

n

Adj.i{2

Nonfinancial

Coef.

0.17

-0.20

0.310.300.79

303

0.144

t

8.29

-7.25

7.400.842.17

Panel B: Revaluations Partitioned by Asset C lass

Variable

InterceptNI

AN I

RRI_INVESTRRI_PPE

RRI^INTANRRI^unknown

RmPL_INVESTRRIPL_PPERRIPL_INTANRRIPL^unknownn

Adj. fl2

Nonfinancial

Coef.

0.17

-0.200.31I.Ol0.24

-1.61-5.83

-5.141.32

1.84

-2.71

304

0.143

t

8.11

-7.577.741.85

0.57-4.81-1.19

-1.383.311.91

-0.46

Mining

Coef.

0.23

0.720.31

0.250.11

249

0.092

t

5.09

1.73

2.422.200.28

Mining

Coef.

0.230.680.31

0.50

-0.581.19

249

0.096

Panel C: Revaluations Partitioned by Valuation Source

Variable

Intercept

NI

ANI

RRI_DIRECTORRR1_INDEPENDENTRRI^unknown

RRIPL^DIRECTORRRIPL_INDEPENDENTRRIPL_unknown

n

Adj. /?2

Nonfinancial

Coef.

0.17

-0.200.310.780.870.081.43

4.03

-0.12303

0.137

(

8.17

-7.237.372.101.25

0.163.462.37

-0.15

t

5.261.61

2.39

2.11

-0.932.24

Mining

Coef.

0.23

0.720.31

0.27

0.19-0.61

0.12249

0.084

t

5.01

1.58

2.401.33

0.24-0.09

0.30

Financial

Coef.

0.13

-0.23

0.080.19

-0.31166

0.578

t

4.98

-2.42

2.921.92

-1.40

Financial

Coef.

0.13-0.24

0.090.23

-0.53

-0.28

166

0.575

t

4.87

-2.383.272.43

-0.68

-1.35

Financial

Coef.

0.13

-0.210.06

0.140.260.20

-0.74

-0.08166

0.580

t

5.01

-1.881.32

0.140.771.05

-1.43

-0.65

^ „^ ^^«. ^nd {pncef + dividends^— pnce^^^) I pricei j . NI is o p e r a t -

ing income. A denotes annual change. RRIis revaluation reserve increm ent, i.e., current-year revalua-tion amount recognized directly in equity. RRIPL is revaluation reserve increment in profit and loss.I.e., current-year revaluation amount recognized in earnings. INVESTis investments. PPEis propertyplant, and equipment. INTAN is intangible assets. DIRECTOR and INDEPENDENT denote source of

revaluation amount, unknoim denotes revaluations where disclosures do not identify asset partition.

Page 26: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 26/36

2 24 MARY E. BARTH AND GREG CLINCH

appraiser-hased valuadons. D irector-hased RRI and RRIPL and indepen-dent appraiser-hased RRIPL are significandy positively related to returnsfor no nfinancial firms; no o the r relations are significantly d ifferent from

7. Distinguishing Asset Impa irment from O ther Revalua tions

This secdon reports findings from two analyses focused on the valua-tion effects of Australian accounting amounts that are not observedunder U.S. GAAP. Australian GAAP requires recognizing revaluadons inearnings or equity depending on whether the revaluation is up or downand on prior upward or downward revaluations of the same assets. In

particular, RRI is positive only when an asset is written up from its carry-ing value (either hased on previous upward revaluations or historicalcost); RRI is negative only when a previously upward-revalued asset isrevalued downward. RRI amounts would not he observed under U.S.GAAR24 Analogously, positive RRIPL would not he observed under U.S.GAAP because RRIPL is positive only for upward revaluadons of a previ-ously downward-revalued asset, which is not permitted under U.S. GAAP.

The only revaluadons permitted under U.S. GAAP are those recognizedas negative RRIPL , i.e., a downward revaluation of an asset that bad notpreviously been written up.

We reestimate (3) and (4) after pardtioning revalued amounts intothose more likely to be stated above or below historical cost. Because theinformation necessary for a completely accurate partition is not avail-able, we rely on a classificadon algoritbm and include an "unknown"category for each asset class. Our algorithm is as follows. First, we obtainthe revaluation date from annual report footnote disclosures. Becauseour revaluation data b egin in 1991, we classify revaluations as unknow nif the date is before 1991. Second, if tbe date is between 1991 and 1995,we determine whetber there is a revaluadon increment in tbat year forthe same asset class or, if there are no revaluation increments identified

for that class, we determine whether there is an unclassified revaluadonincrement.25 If we are unable to identify a revaluation increment in thatyear, we classify the revaluation as unknown. Third, if we identify a reval-uation increment that matches by year and asset class, or is unclassifiedbut matches by year, we classify the carrying value as above cost if the

23 In all three panels of table 7, nonfinancial and financial firms exhibit a significantlynegative association between returns and net income. Untabulated findings indicate thisresult is attributable to the smaller sample firms. Om itting these firms results in a signifi-

cantly positive association between returns and net income, without affecting our infer-ences relating to revaluation increments.

2* Although asset write-downs are required under U.S. GAAP, negative i{R/would not be

Page 27: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 27/36

REVALUED FINANCIAL, TANGIBLE, AND INTANGIBLE ASSETS

TABLE 8

Summ ary Statistics from Regression Price, P, on Income and Book Value

of Equity, Partitioned by Revalued Assets Above and Below Cost

Samp le of Publicly Tra ded Australian Firms from 1991 to 1995

225

Variable

Intercept

BV

NI

DISC

COST

INVEST

-PPE

^INTAN

REVALUATION^INVEST_ABOVE

^INVEST_BELOW

^INVEST_unknown

-PPFUS.BOVE

^PE_BELOW

^PFL.unknoxun

.JNTAN_BELOW

^INTAN_unknown

n

Adj. R^

Nonfinancial

Coef.

0.43

1.39

2.53-0.33

0.921.40

1.25

2.11

1.02

0.990.47

6.26

0.83- 2 . 2 4

0.72

3450.822

t

3.5013.48

4.18-0.61

3.97

13.467.66

11.64

1.95

10.86

2.032.95

3.57-1.298.21

/(-Values for Tests of Coefficient EqualityCoefficient Test

All Balance Sh eetAll INVEST._REVAL

All PPFLREVAL

All INTAN_REVAL

Nonfinancial

0,0,

00

00

0.02

0.09

Mining

Coef.

0.251.23

2.17

2.89

1.54

1.32

- 0 . 6 5 •

5.901.76

0.501.46

1.55

268

0.879

t

2.936.193.04

3.69

3.74

8.79

-0.59

4.65

2.700.83

1.88

6.25

Mining

0.00

0.000.23

Financial

Coef.

0.730.793.53

0.76

0.320.431.17

0.39

0.180.36

0.55

1.18

19 4

0.940

t

6.926.705.457.64

3.322.532.85

6.46

2.83

3.522.08

3.25

Financial

0.000.030.00

Pis share price as of fiscal year-end. BVis book value of equity after su btracting investments (INVEST),property, plant, and equipment (PPE), and intangible assets (INTAN). Mis operating income. DISCisdisclosed, but not recognized , asset value estimates. CO^Tand REVAL deno te recognized am ounts basedon historical cost and revaluations. ABOVE, BELOW, and unknown denote w hether the recognized reval-ued amount is likely above or below historical cost, or whether it is not possible to make an assessment.All variables are deflated by number of shares outstanding. Data limitations preclude estimation of

coefficientsfor

each asset partition. Coefficientsare

only estimatedif

thereare

nonzero observationsfor

more thanfivefirms.

revaluation is to the reva luation reserv e. We classify the carrying value as

below cost if the revaluation is taken to earnings.^^Table 8 presents findings relating to (3) and reveals that revalued in-

vestments and PPE are significantly related to share prices regardless of

2 Obviously, our algorithm is imperfect. For example, all revaluations prior to 1991 are

classified as unknow n. Also, firms revalue assets by class, so it is possible that a revaluation

we classify asabove-cost actually is the net of an upward and downward revaluation of twoassets within a single asset class. Moreover, it is sometimes difficult to identify the asset classto which a revaluation relates, even if a description of the asset is disclosed, e.g., some val-

Page 28: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 28/36

2 2 6 MARY E. BARTH AND GREG GLINCH

whether they are above or below historical cost, with the single exceptionof above-cost PPE revaluations for mining firms. These findings suggest

that upward a nd downward revaluations of investments an d PPE are valuerelevant. We are un able to classify m ost intangib le revaluations, altho ughthose we are able to classify as below cost are insignificantly related toprice. Tests of equality of coefficients reject the null, except for those onPPE for mining firms and intangible assets.

Untab ulated findings from the V regressions reveal some different in-ferences from the price regressions. For nonfinancial firms, below-costrevalued investments and all revalued PPE amounts are insignificantlypositively related to V Contrary to predictions, above-cost revalued PPE

for nonfinancial firms is significantly negatively related to V For mining

firms, above-cost revalued PPE is significantly related to V; it is insig-nificantly related to price. For financial firms, no revalued amount issignificantly related to V.

Table 9 presents results from estimating the returns regressions, afterpartitioning RRI and RRIPL by sign. Panel A presents results for aggre-gate revaluations and indicates that, for nonfinancial firms, only negativeRRIPL, which would be observed under U.S. as well as Australian GAAP,is significantly related to ret urn s, and its relation is positive, as pred icted .This finding indicates that asset write-downs from historical cost are rel-evant to investors and timely. For mining and financial firms, only posi-tive RFil is significantly positively related to returns, as predicted. Contraryto pred ictions, positive RRIPL is significantly negatively related to returnsfor both samples. It is also negatively related to returns for nonfinan cialfirms, although not significantly so. These findings indicate that upwardrevaluations recognized in earnings are valued negatively by investors,suggesting investors discount discretionary earnings increases. For finan-cial firms, investors also discoun t negative RRIPL; its coefficient is signifi-cantly negative.^^

Table 9, panel B, presents results from estimating (9) partitioning RRI

and RRIPL by sign. It reveals that investments' RRI are positively relatedto returns, regardless of sign, but only positive RRI are significantly re-lated. Find ings for PPE RRI are mixed. Negative PPE RRI for nonfinancialfirms are significantly positively related to returns, as predicted, but therelation is insignificant for the o the r two industries. Positive PPE RRI aresignificantly positively related to returns for financial firms but aresignificantly negatively (insignificantly) related for mining (nonfinancial)firms. Regarding RRIPL, in contrast to predictions, investment RRIPL arenegatively related to returns, regardless of sign or industry, altho ugh onlysignificantly so for positive RRIPL for financial firms. Negative PPE and

Page 29: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 29/36

REVALUED FINANCIAL, TANGIBLE, ANr> INTANGIBLE ASSETS 227

T A B L E 9

Summ ary Statistics from Regression of Returns on Income, Change in Income, and Current-Year

Revaluations, Partitioned by Recognition in Earnings or Directly in Equity and Sign

Samp le of Publicly Traded Australian Eirms from 1991 to 1995

Panel A: Aggregate Revaluations

Variable

InterceptNI

ANI

RRI.J'OSITIVE

RRI_NEGATrVE

RRIPL_POSmVE

RRIPL_NEGATIVEn

Adj. /?2

Panel B: Revaluations Partitioned

InterceptNI

&NI

RRI_INVEST_POSmVE

RRI-INVEST_NEGATIVE

RRI_PPEWOSITP/E

RRI_PPE_NEGATrVE

RRI_unknown_POSITIVERRIPL_INVEST_POSrTIVE

RRIPL_INVEST_NEGATIVE

RRIPL^PE^MEGATIVERRIPL_INTAN_NEGATIVE

RRIPL_unknown^NEGATIVEn

Adj. fl2

Panel C: Revaluations Partitioned

InterceptNI

ANI

RRI_DIRECTOR_POSnTVERRI_DIRECTOR_NEGATTVE

RRI_INDEPENDENT_POSITIVE

RRI_INDEPENDENT_NEGATIVERRI_unknown_POSITrVE

RRI_unknown.jmGATIVE

RRIPL.MIRECTOR.J'OSrnVERRIPLJDIRECTOR^NEGATIVE

RRIPL_unknown_POSrTIVE

RRIPL_unknown_NEGATIVE

n

Adj. /J2

Nonfinancial

Coef.

0.17

-0.200.310.030.76

-2.60

0.40303

0.140

t

8.20

-7.247.380.07

1.21

-1.34

4.28

by Asset Class

0.17-0.20

0.310.85

14.80

-0.280.80

-6.17-1.71

-5.820.79

1.60

-3.40304

0.144

8.02

-7.59

7.751.91

1.27

-0.611.65

-1.19-0.83

-1.422.311.67

-0.58

by Valuation Source

0.18-0.20

0.310.55

19.66

13.20-0.34

0.54

1.02

1.69

-0.47303

0.144

8.24

-7.277.412.932.17

2.82-0.65

0.78

2.720.37

-0.60

Mining

Coef.

0.230.710.31

0.530.54

-12.88

-0.18249

0.087

0.230.680.31

0.490.50

-0.91

0.88

249

0.094

0.240.77

0.320.181.73

-0.14

15.44

-0.19249

0.088

t

5.13

1.63

2.423.24

0.82-2.46

-1.05

5.221.53

2.40

2.490.77

-1.421.59

5.141.69

2.462.616.54

-0.20

1.51

-0.48

Financial

Coef.

0.12-0.23

0.080.230.07

-0.24

-0.42166

0.573

0.13-0.24

0.080.26

0.12-1.70-0.11

-0.25-0.40

166

0.568

0.13-0.20

0.050.23

-0.060.05

0.280.21

-0.11-0.33-0.81

-0.11166

0.569

t

4.07-2.36

2.581.74

0.26-2.28

-7.36

4.04-2.37

2.921.96

0.48-2.15-0.10

-2.33

-1.72

4.10-1.831.26

1.35

-0.19

0.060.701.20

-0.11-0.29-1.66

-0.84

Returns are 12-month returns end ing at year-end (/m(re, + <iiuidmrfi,-/>ric«,.i)//)n(;e,-i.fi/y is revaluation reserveincrement, i.e., current-year revaluation amount recognized directly in equity. RRIPL is revaluation reserve incre-

Page 30: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 30/36

2 2 8 MARY E. BARTH AND GREG GLINGH

intangibles' RRIPL are positively related to returns, as predicted. Thefindings indicate that returns generally reflect negative revaluations rec-ognized in earnings as predicted. They also indicate that positive revalu-ations recogn ized in earning s are viewed negatively by investors, althoughthe evidence relates only to investments.

Table 9, panel C, reports findings from partitioning RRI and RRIPLby sign and source of valuation and reveals no particular pattern ofsignificance for coefficients on director- and independent appraiser-based valuations; we have insufficient observations to estimate coeffi-cients for independent-based valuations recognized in earnings.

8. Additional Analyses

8.1 FIRM SIZE

Sample firms comprise two disparate size groups, the top 100 firmsand 250 smaller firms. It is possible that large and small firms have differ-ent economic characteristics, which could affect our inferences regard-ing revaluations. In particular, the availability of information sources toassist investors in their assessment of revaluation disclosures likely variesby firm size. Thus, we present in table 10 findings from price regressionsanalogous to those in table 3 separately for the top 100 firms and the 250smaller firms.

Table 10 reveals that investment and intangible revalued amounts aresignificantly positively related to price, regardless of firm size, with thesingle exception of investment revalued amounts for financial firms. Incontrast, PPE revalued amounts are only significantly positively relatedto price for small nonfinancial, large mining, and small financial firms.Untabulated findings based on partitioning PPE into subclasses revealthat the significance of PPE for small firms is attributable to plant andequipment, not property. Untabulated findings also reveal little differ-ence in significance between director- and independent appraiser-based

valuations or age-partitioned revaluations, although director-based reval-ued amounts are somewhat more strongly associated v«th price for smallfirms than independent appraiser-based amounts. Untabulated resultsfrom returns regressions indicate no systematic differences between largeand small firms.

8.2 ANALYST FOLLO WING AND ASSET TURNOVE R

Comparing the number of observations for the price and V regres-sions in table 3 reveals that many sam ple firms have n o analyst following.Firms' information environments, such as reflected in analyst following,

might affect how investors view revaluations. Also, because lack of ana-lyst following is concentrated in the small firm sample, it is possible that

Page 31: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 31/36

REVALUED FINANGIAL, TANGIBLE, AND INTANGIBLE ASSETS

T A B L E 1 0

Summ ary Statistics from Regression of Price, P, on Income

and Book Value of Equity Partitioned by Asset Class

Samp le of Publicly Traded Australian Firms from 1991 to 1995 P artitioned by SiuCoefficients (t-Statistics)

229

Variable

Intercept

BV

NI

DISC

COST

..INVEST

^P E

..INTAN

REVALUATION

^INVEST

^P E

^INTAN

n

Adj. ft2

Nonfinancial

Large

1.80

(4.73)

0.75

(5.49)

5.49

(4.43)

-0.63

(-1.41)

-0.05

(-0.17)

1.19

(6.56)

-0.30(-1.30)

0.66

(2.96)

-0.89

(-1.88)

0.69(3.74)16 0

0.794

Small

0.08(0.84)

1.49(13.70)

1.16

(3.29)

1.31(1.89)

0.96

(3.38)

1.25

(12.22)

1.43

(9.76)

1.87

(3.75)

1.04

(7.45)

1.78(8.34)188

0.830

Mining

Large

0.66(2.29)

1.23(4.89)

1.95(1.46)

2.86(3.13)

1.08(1.93)

1.31

(6.12)

-0.55(-0.55)

3.47(3.81)

1.64

(5.97)

105

0.814

Small

0.22(4.15)

0.78(3.51)

2.28(4.31)

2.24(1.15)

2.15

(5.12)

0.96(5.64)

0.41

(4.26)

-0.18(-0.56)

161

0.659

Financial

Large

4.11

(10.31)

-0.15(-1.34)

2.89(4.23)

0.39(2.48)

-0.29(-2.97)

-^.04

(-0.38)

3.71(6.76)

-0.75

(-4.88)

-0.53

(-1.60)

62

0.959

Small

0.38(4.79)

0.68(6.24)

2.83(4.09)

0.62(6.04)

0.70

(6.57)

0.55(2.79)

0.03(0.05)

0.41

(8.29)

1.12

(3.02)

13 5

0.905

Pi s share price as of fiscal year-end. BVis book value of equity after subtracting investments (INVEST),property, plant, and equipment (PPE), and intangible assets (INTAN). NI is operating income. DISCisdisclosed, but not recognized, asset value estimates. CO.ST and REVAL deno te recognized am ounts basedon historical cost and revaluations. All variables are deflated by num ber of shares outstan ding . Data lim-itations preclude estimation of coefficients for each asset partition. Coefficients are only estimated ifthere are non zero observations for mo re thanfivefirms.

firm sample based on analyst following and repeat our price and returnsanalyses. Untabulated results reveal few systematic differences in signifi-cance of revalued amounts between the analyst and no-analyst samples.We observe only that investment revalued amounts are more strongly

associated with price for financial firms not followed by analysts thanfor financial firms followed by analysts, and revaluations recognized in

Page 32: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 32/36

2 3 0 MARY E. BARTH AND GREG GLINCH

Finally, detecting value relevance, and differences in value relevance,could be difficult for revalued amounts of firms with cost-based amountsclose to current value. To investigate this possibility we reestimate the

price and returns regressions, permitting the coefficient on PPE revalua-tions to differ for firms with low asset turnover. We define asset turnoveras capital expenditures divided by the carrying amount of PPE, and setan indicator variable equal to one for firms with asset turnover belowthe sample median, and zero otherwise. Untabulated results reveal nosignificant incremental coefficient for nonfinancial firms. For miningfirms, the incremental coefficient is significantly negative for aggregatePPE and plant and equipment revalued amounts; that on property reval-ued am ounts is insignificantly positive. For financial firms, the incre me n-tal coefficient is insignificantly negative for PPE, reflecting a significantlynegative (positive) incremental coefficient for property (plant and equip-ment) . Only the p lant and equ ipm ent finding is consistent with expecta-tions. Untabulated results from the returns regressions indicate that,contrary to expectations, the incremental coefficient for mining firms issignificantly negative. However, it is insignificantly different from zerofor nonfinancial and financial firms. Thus, we find little evidence sup-porting the conjecture and none of these findings resolves unexpectedfindings in tables 3, 4, and 7.

9. Summary and Concluding Remarks

This study investigates the extent to which different types of revaluedassets of Australian firms are associated with share prices and non-mar-ket-based estimates of firm value, which are based on the present valueof analysts' forecasts of future earnings. We con tribute to extant researchon aggregate revaluation data by investigating whether relevance, reli-ability, and timeliness of revalued assets vary systematically by asset classor by source or age of the revalued amount, and whether price and re-turn associations mirror the ability of revalued amounts to refiect antic-

ipated future profitability. We also investigate whether asset impairments,a type of revaluation permitted under U.S. GAAP, exhibit different rela-tions with firm value from other asset revaluations, which are not per-mitted under U.S. GAAP

Taken together, our findings suggest revalued financial, tangible, andintangib le assets are value relevant. Although the financial assets findingsare not surprising based on prior research, the intangible assets findingsare striking in their strength and consistency. Findings for PPE are lessconsistent, although the stronger value relevance for plant and equip-ment than for property suggests revalued operating assets are more valuerelevant than assets less directly related to operations. Perhaps surpris-ingly, there is little evidence to indicate that director- and independent

Page 33: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 33/36

REVALUED FINANGIAL, TANGIBLE, AND INTANGIBLE ASSETS 231

directors' private information enha nces value estimates despite their po-

tential self-interested financial statement management incentives. Also,several year-old revalued amounts are value relevant, suggesting timeli-

ness is not critical for long-term asset revaluations. Finally, the evidencesuggests that both upward and downward revaluations arevalue relevant,although the discretionary nature of asset write-ups through earningscan affect their value relevance.

REFERENCES

ABOODY, D.; M. E. BARTH; AND R. KASZNIK. "Revaluations of Fixed Assets and Future Firm

Performance: Evidence from the U.K." foumal ofAccounting and Economics (forthcoming).

AMIR, E.; T S. HARRIS; AND E. K. VENUTI . "A Comparison of the Vahie-Relevance of U.S.

versus Non-U.S. GAAP Accounting Measures Using Form 20-F Reconciliations."/ourreaiof Accounting Research (Supplement 1993): 230-6 4.

ACCOUNTING STANDARDS REVIEW BOARD. Approved Accounting Standard—ASRB 1010: Ac-

counting for th e Revaluation of Non-Current Assets. M elbourne: ASRB, 1987.

AUSTRALIAN ACCOUNTING STANDARDS BOARD. AASB 1010: Accounting for theRevaluation of

Non-Current Assets. M elbourne: AASB, 1993.

BARTH, M. E. "Relative Measurement Errors among Alternative Pension Asset and LiabilityMeasures." Th e Accounting Review (July 1991): 433-63.

• "Fail" Value Accounting: E vidence from Investm ent Securities and the Market Val-uation of Banks." The Accounting Review (January 1994): 1-25.

BARTH, M. E., AND G. GLINCH. "International Differences in Accounting Standards: Evi-

dence from U.K., Australian, and Ganadian Firms." Contemporary Accounting Research(Spring 1996): 135-70.

BARTH, M. E., AND S. KALLAPUR. "The Effects of Gross-Sectional Scale Differences on

Regression Results in Empirical A ccounting Research." Contemporary Accounting Research(Fall 1996): 527 -67 .

BARTH, M . E.,AND W. R. LANDSMAN. "Fundamental Issues Related to Using Fair ValueAccounting for Financial Reporting." Accounting Horizons (December 1995): 97-107.

BARTH, M. E.; W. H. BEAVER; AND W. R. LANDSMAN. "The Market Valuation Implications

of Net Periodic Pension Cost." Journal of Accounting and Economics (March 1992): 27-62.• "Value-Relevance of Banks' Fair Value Disclosures under SFAS No. 107."The Ac-

counting Review (October 1996): 513-37 .

BEAVER, W. H. , AN D W. R. LANDSMAN. Incremental Information Content ofStatement 33 Disclo-

sures. Stamford, Conn.: FASB, 1983.BEAVER, W. H. , AND S. RYAN. "HOW Well Do Statement No. 33Earnings Explain Stock

Returns}" Financial Analysts Journal (September/October 1985): 66-71.

BELL, T B. "Market Reaction to Reserve Recognition Accounting."/owma/ of AccountingResearch (Spring 1983): 1-17.

BERNARD, V. L. "Discussion of An Investigation of Revaluations of Tangible Long-LivedAssets." Journal of Accounting Research (Supplement 1993): 39-45.

BERNARD, V L., AND R. RULAND. "The Incremental Information Content of Historical

Gost and Gurrent Gost Numbers: Time Series Analysis." The Accounting Review (October1987): 701-22.

BERNARD, V L.; R. G. ME RT O N ; AND K. G. PALEPU. "Mark-to-Market Accounting for U.S.

Banks and Thrifts: Lessons from the Danish Experience."/ourrea/ of Accounting Research(Spring 1995): 1-32.

BROWN, P. D, AND E J. FINN. "Asset Revaluations and Stock Prices: Alternative Interpreta-

Page 34: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 34/36

232 MARY E. BARTH AND GREG CLINCH

BRO W N, P. D.; H. Y IZAN; AND A. L. LOH. "Fixed Asset Revaluations and Managerial Incen-tives." Abacus (March 1992): 36-57.

BuBLiTZ, B.; T. J. FRECKA; AND J. C. MCKEOWN. "Market Association Tests and FASB State-ment No. 33 Disclosures: A Reexamination."/oumflZ of Accounting Research (Supplement1985): 1-23.

COTTER, J. "Asset Revaluations and Debt Contracting." W orking pape r. U niversity of

Southern Queensland, 1997.

EASTON, P. D., AND P. H. EDDEY. "The Relevance of Asset Revaluations O ver an EconomicCycle." Australian Accounting Review (May 1997): 22 -30 .

EASTON, P. D.; P. H. EDDEY; AND T. S. HARRIS. "An Investigation of Revaluations of Tangi-

ble Long-Lived Assets." foumal of Accounting Research (Supplement 1993): 1-38.

ECCHER, A.; K . RAMESH; AND S. R. THIAGARAJAN. "Fair Value Disclosures of Bank HoldingCom panies." foumal of Accounting and Economics (August-December 1996): 79-117.

EMANUEL, D. "Asset Revaluations and Share P rice Revisions. "/ouma/ of Business Finance and

Accounting (Spring 1989): 213-27 .

FINANCIAL ACCO UNTIN G STANDARDS BO ARD. Statement of Financial Accounting Standards

No. 19: Financial Accounting and Reporting by Oil an d Cas Producing Companies. Stamford,Con n.: FASB, 1977.

. Statement ofFinancial Accounting Standards No. 33: Financial Accounting and Changing

Prices. Stamford, Conn.: FASB, 1979.. Statement of Financial Accounting Standards No. 107: Disclosures about Fair Value of

Financial Instruments. Norwalk, Conn.: FASB, 1991.. Statement of Financial Accounting Standards No. 121: Accounting for the Impairment of

Long-Lived Assets and for Long-Lived Assets to Be Disposed O f Norwalk, Conn.: FASB, 1995._ . Proposed Statement ofFinancial Accounting Standard, Accounting for Derivative and Sim-

ilar Financial Instruments and for Hedging Activities. Norwalk, Conn.: FASB, 1996.FRANKEL, R., AND C. M. C. LEE. "Accounting Diversity an d Intern ational Valuation." W ork-

ing paper 96-01, New York Stock Exchange, 1996.HARRIS, X S., AND J. A. OHLSON. "Accounting Disclosures and the Market's Valuation of

O il and Gas P roperties." The Accounting Review (O ctober 1987): 651-7 0.

HAW, I. M., AND S. LUSTGARTEN. "Evidence on Income Measurement Properties of ASRNo. 190 andSFAS No. 33 Data." Journal of Accounting Research (Autumn 1988): 331-52.

HENDERSO N, S. , AND J. GOODW IN. "The Case against Asset Revaluations." Abacus (March1992): 75-87.

HOPWOOD, W, AND T. SCHAEFER. "Firm-Specific Responsiveness to Input Price Changesan d the Incremental Information Content in Current Cost Income." T h e AccountingReview (April 1989): 312 -38 .

LANDSMAN, W . "An Em pirical Investigation of P ension Fund P roperty Rights." The Account-

ingReview (O ctober 1986): 662-91.LO BO, G. J., AND I. M. SONG. "The Incremental Information in SFAS No. 33 Income Dis-

closures over Historical Cost Income and Its Cash and Accrual Components." The

Accounting Review (April 1989): 329-43.MAGLIOLO, J. "Capital Market Analysis of Reserve Recognition Accounting." foumal of

Accounting Research (Supplement 1986): 69-10 8.MILLER, M. H., AND E MO DIGLIANI. "Some Estimates of the Cost of Capital to the Electric

Utility Industry, 1954-57." American Economic Review (June 1966): 333-91.

MURDOCH, B. "The Information Content of FAS 33 Returns on Equity." T he Accounting Re-

view (April 1986): 27 3-8 7.NELSON, K. "Fair Value Accounting for Commercial Banks: An Empirical Analysis of SFAS

No. 107." The Accounting Review (April 1996): 161-82.OHLSON, J. "Earnings, Book Values and Dividends in Security Valuation." Contemporary

Accounting Research (Spring 1995): 661-87.

Page 35: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 35/36

REVALUED FINANCIAL, TANG IBLE, AND INTANGIBLE ASSETS 2 3 3

SHARPE, I. G., AND R. G. W ALK ER. "Asset Revaluations and Stock Market Prices." fouma t of

Accounting Research (Autumn 1975): 293-310.

STANDISH , P., AND S. UNG. "Corporate Signalling, Asset Revaluations and the Stock P rices

of British Companies." The Accounting Review (October 1982): 701-1 5.VENKATACHALAM, M . "Value-Relevance of Banks ' Derivatives Disclosures." foumal of A c-

counting and Economics (August-December 1996): 327-5 5.

WHITE, H . "A H eteroskedasticity-Consistent Covariance M atrix Estimator and a Direct Testfor H eteroskedasticity." Econometrica (May 1980): 817 -38 .

W H ITTRED, G ., AND Y. K , CHAN. "Asset Revaluations and the Mitigation of Underinvest-ment." Abacus (March 1992): 3-35.

Page 36: AFR L3 12 Barth Clinch 1988

8/3/2019 AFR L3 12 Barth Clinch 1988

http://slidepdf.com/reader/full/afr-l3-12-barth-clinch-1988 36/36