the Value Relevance of Disclosures of Liabilities

8/10/2019 the Value Relevance of Disclosures of Liabilities

1/19

1. IntroductionBy testing for their value relevance, this study ex-amines the usefulness to investors of mandateddisclosures by UK firms of the investor-firm shareof liabilities of equity-accounted associate and

joint venture investees. The study uses UK data forthe period from 1998 to 2003, immediately follow-ing the date of the introduction by the UKsAccounting Standards Board (ASB) of FRS 9:

Associates and Joint Ventures (ASB, 1997).Motivated in part by concern that single-line eq-

uity accounting was being used as an off-balance-sheet-financing device which had the effect of concealing the level of group gearing (ASB,

1996), FRS 9 introduced for the first time strictthresholds governing disclosures by UK firms of the investor-firm share of the underlying gross as-sets and liabilities of equity-accounted investees,and forced a substantial increase in such disclo-sures. On the grounds that joint ventures are sub-

ject to joint control by the investor firm whereasassociates are only subject to its significant influ-ence, FRS 9 also introduced a distinction betweenassociate investees and joint venture investees, re-quiring a higher and more prominent degree of dis-closure in respect of joint ventures. We examinethe value relevance of the FRS 9-mandated disclo-sures of the liabilities of equity-accounted in-vestees, which are the amounts by which the netinvestments in investees for which disclosures aremade would have to be grossed up to give the in-vestor-firm share of the gross assets and liabilitiesof those investees. In light of the concerns aboutconcealment of group gearing that partly motivat-ed the investee-liability disclosure requirements of FRS 9 and in light of the findings of previous USresearch, we predict that the mandated disclosuresare negatively associated with the market value of equity of the investor firm, and therefore have anegative coefficient in a value-relevance regres-sion. In light of the possibility that the creditors of j i t t ight b lik l th th f

Accounting and Business Research, Vol. 37. No. 4. pp. 267-284. 2007 267

The value relevance of disclosures of

liabilities of equity-accounted investees:UK evidenceJohn OHanlon and Paul Taylor*

Abstract This study examines the value relevance of mandated disclosures by UK firms of the investor-firmshare of liabilities of equity-accounted associate and joint venture investees. It does so for the six years followingthe introduction of FRS 9: Associates and Joint Ventures , which forced a substantial increase in such disclosuresby UK firms. Since the increased disclosure requirements were partly motivated by concern that single-line equityaccounting concealed the level of group gearing, and in light of previous US results, it is predicted that the man-

dated investee-liability disclosures have a negative coefficient in a value-relevance regression. The study also ex-amines whether value-relevance regression coefficients on investee-liability disclosures are more negative for jointventures than for associates and whether they are more negative in the presence of investor-firm guarantees of in-vestee-firm obligations than in the absence of such guarantees. The study reports that the coefficient on all investee-liability disclosures taken together has the predicted negative sign, and is significantly different from zero. It findslittle evidence that the negative valuation impact of liability disclosures is stronger for joint venture investees over-all than for associate investees overall, or stronger for guarantee cases overall than for non-guarantee cases overall.There is, however, some evidence that the impact for joint venture guarantee cases is stronger than that for jointventure non-guarantee cases and stronger than that for associate guarantee cases.

Key words: Equity accounting; joint ventures; associates; FRS 9; value relevance

*The authors are, respectively, Professor of Accounting andLecturer in Accounting in the Department of Accounting andFinance, Lancaster University Management School. The studyhas benefited from the helpful comments of Steven Young,Pelham Gore, Pauline Weetman (editor) and an anonymousreferee, and of participants at a workshop at the University of Melbourne. The authors thank the Centre for BusinessPerformance at the Institute of Chartered Accountants inEngland and Wales for financial support, and Victoria Wangand Noordad Aziz for assistance in the collection of data.Correspondence should be addressed to John OHanlon,Department of Accounting and Finance, Lancaster UniversityManagement School, Lancaster University, Lancaster, LA14YX, UK. Tel: 44 (0)1524 593631. Email: j.ohanlon@lancast-


2/19

the assets of the investor firm and in light of thegreater and more prominent disclosure that FRS 9required for joint ventures relative to associates,we also examine whether the value-relevance re-gression coefficient on investee-liability disclo-sures is more negative for joint ventures than for

associates. Furthermore, because of the recourse tothe assets of the investor firm conferred by an in-vestor-firm guarantee of investee-firm obligations,and in light of the findings of previous US re-search, we examine whether the value-relevanceregression coefficient on investee-liability disclo-sures is more negative in the presence of investor-firm guarantees than in the absence of suchguarantees.

The findings of the study indicate that disclo-sures of liabilities of equity-accounted investees,which enable equity-accounted net investments tobe grossed up by financial statement users, arenegatively associated with the market value of eq-uity of the investor firm. This is consistent with theconcerns about off-balance-sheet financing thathelped motivate the requirement for such disclo-sures under FRS 9. The study finds little evidencethat the negative valuation impact of liability dis-closures is stronger for joint venture investeesoverall than for associate investees overall, orstronger for guarantee cases overall than for non-guarantee cases overall. There is some evidencethat the impact for joint venture guarantee cases isstronger than that for joint venture non-guaranteecases and stronger than that for associate guaran-tee cases.

The remainder of the paper is organised as fol-lows. Section 2 gives the background to the study,outlining relevant issues relating to accounting forassociates and joint ventures and the results of re-lated research. Section 3 describes the research de-sign. Section 4 gives details of the data used in thestudy. Section 5 reports the results. Section 6 con-cludes.

2. BackgroundThis section provides background for the empiri-cal work reported in the study. Subsection 2.1 out-lines the history of the accounting for investmentsin associates and joint ventures in US, internation-al and UK generally accepted accounting practice(GAAP), with particular reference to the disclo-sure of the gross assets and liabilities of equity-ac-counted investees and to the provisions of the UKaccounting standard FRS 9. Subsection 2.2 out-lines related research. Subsection 2.3 summarisesthe section.

2.1. Accounting for associates and joint venturesAssociate and joint venture investees are each

subject to an important degree of influence by theinvestor firm, where the influence falls short of thecontrol that would classify the investee as a sub-sidiary of the investor firm. In the case of a joint

venture, the investor firm and one or more otherparties exercise joint control under a contractualarrangement; in the case of an associate, the in-vestor firm merely exercises significant influence. 1Some accounting regimes have drawn a distinctionbetween the two types of investee, and others havenot.

Until the early 1970s, an investment by one firmin another where the investee firm was not a sub-sidiary of the investor firm was generally account-ed for at cost in the balance sheet, and income wasrecognised when a dividend was received from the

investee. The inadequacies of such accounting forwhat might be vehicles for important activitiesmotivated the development of a more informativemethod of accounting for investee firms overwhich the investor firm had significant influenceshort of control. Under this method, termed the eq-uity method, the investee is accounted for in theinvestor firms balance sheet as a single-line itemcomprising cost plus the investor-firm share of thepost-acquisition change in the investees equity,and in its income statement as the investor-firmshare of aggregated components of investee net in-

come. Use of the method was pioneered by theRoyal Dutch Shell group in 1964 (Ernst & YoungLLP, 2001), and was introduced into US GAAPand UK GAAP in 1971. The relevant US account-ing standard was APB Opinion No. 18: The equitymethod of accounting for investments in commonstock (Accounting Principles Board, 1971), andthe relevant UK accounting standard was initiallySSAP 1: Accounting for Associated Companies(ASC, 1971). Subsequently, international account-ing standards incorporated the method in IAS 28:

Investments in Associates (IASB, 2003a) andIAS 31: Interests in Joint Ventures , (IASB, 2003b),which were first issued in 1988 and 1990, respec-tively.

Concern has been expressed in some quartersthat single-line equity accounting may reflect in-sufficient information about the activities of in-vestee firms, a particular concern being that suchaccounting could be used to conceal the level of investor-group debt (Bierman, 1992; ASB, 1994;ASB, 1996; Crichton, 1996; Johnson and Holgate,1996). This has led to the development of variousmethods for reflecting in the financial statementsof investor firms more detailed information fromthe underlying income statements and balancesheets of equity-accounted investees. These meth-d i l d i lid i h h

268 ACCOUNTING AND BUSINESS RESEARCH

1 See FRS 9: Associates and Joint Ventures (ASB, 1997),IAS 28: Investments in Associates (IASB, 2003a) and IAS 31:


3/19

statement line items is merged with the investor-firm line items, the expanded equity method,where the investor-firm share of investee-firm lineitems appears in the investor firms financial state-ments but is not merged with the investor-firm lineitems, and note disclosures. There has been much

debate about how far along the spectrum from sin-gle-line equity accounting to proportionate consol-idation regulators should go with regard toassociates and joint ventures, with single-line eq-uity accounting being seen by some as reflectingan inadequate amount of information about impor-tant activities and proportionate consolidationbeing seen by some as implying a greater degree of control than is actually exercised (ASB, 1994;Milburn and Chant, 1999). Differences of opinionhave been reflected in differences across regimesin accounting for associates and joint ventures.

With regard to disclosure of liabilities of equity-accounted investees, the USs APB Opinion No. 18states that, where material, it might be necessary todisclose information on the assets and liabilities of the investee firm. Materiality thresholds are notdefined by APB Opinion No. 18, but have been de-fined for SEC registrants by the SECs RegulationS-X. This requires note disclosure of summarisedinformation on assets, liabilities and results of op-erations of equity-accounted investees that exceedspecified materiality thresholds, and the filing of separate financial statements of equity-accountedinvestees that exceed higher materiality thresholds(Accounting Series Release 302: SeparateFinancial Statements Required by Regulation S-X ,SEC, 1981). 2 No distinction is drawn between as-sociates and joint ventures with regard to the re-quired method of accounting or the requireddisclosures.

International GAAP has distinguished betweenthe method of accounting for associates and themethod of accounting for joint ventures, includingwith regard to the way in which investee liabilitiesare reflected in investor-firm financial statements.When first issued in 1988, IAS 28: Investments in

Associates required associates to be accounted forby the equity method, with no requirement for dis-closure of liabilities of associates. However, thecurrent version of the standard, effective for ac-counting periods beginning on or after 1 January2005, requires disclosure of summarised financialinformation of all associates, including their ag-gregate assets, liabilities, revenue and net income(IASB, 2003a). IAS 31: Interests in Joint Venturesrequires that joint ventures be accounted for eitherby proportionate consolidation, which is the pre-ferred method, or by the equity method as for

associates. However, the IASB has recently sig-

nalled its intention to eliminate the main differencebetween IAS 31 and the corresponding US GAAPprovisions by removing the option to account for

joint ventures by proportionate consolidation andrequiring that they be accounted for by the equitymethod (IASB, 2007). This will remove a signifi-

cant distinction between the methods of account-ing for associates and for joint ventures, althoughdisclosures may continue to be made separately foreach class of investee.

Similar to APB Opinion No. 18, the UKs SSAP 1required that, if materially relevant from the per-spective of the investor firm, information on theassets and liabilities of an equity-accounted in-vestee should be disclosed in the investor firms fi-nancial statements. However, no materialitythresholds were defined and, for the period duringwhich SSAP 1 was in force, disclosure of such in-

formation by UK firms was rare (ASB, 1997, ap-pendix 3, paragraph 3). During the 1990s, therewas concern that this might be allowing equity-accounted investees to be used as a means of con-cealing debt (ASB, 1994; ASB, 1996; Crichton,1996; Johnson and Holgate, 1996). Furthermore,on the grounds that joint ventures and associatesare different in that the former is subject to jointcontrol by the investor firm whereas the latter ismerely subject to its significant influence, therewas some debate as to whether joint venturesshould be treated differently from associates.

These concerns were reflected in the developmentby the UKs ASB of FRS 9: Associates and Joint Ventures , which came into effect for accountingperiods ending on or after 23 June 1998. This de-fined for the first time strict thresholds governingthe disclosure by UK firms of the investor-firmshare of the liabilities of equity-accounted in-vestees, bringing about a significant increase insuch disclosure. It also introduced a clear distinc-tion between associates and joint ventures, requir-ing a higher and more prominent degree of disclosure for the latter.

The disclosures required by FRS 9 took threeprincipal forms. First, firms were required to ac-count for all joint ventures by the gross equitymethod, whereby the investor-firm share of the in-vestee firms gross assets and gross liabilities ap-peared on the face of the investor firms balancesheet and the investor-firm share of the investeefirms sales revenue appeared on the face of the in-vestor firms income statement. Second, if the in-vestor-firm share of one or more designated itemsfor aggregate associates or for aggregate joint ven-tures exceeded 15% of the corresponding item forthe investor firm, a 15% materiality disclosure forthat class of investee was required. The designateditems were (i) gross assets, (ii) gross liabilities,(iii) l d (i ) h h

Vol. 37 No. 4. 2007 269


4/19

ceeded, the investor firm had to disclose in aggre-gate its share of the following items of the in-vestees: sales revenue, fixed assets, current assets,liabilities due within one year, and liabilities dueafter one year. Third, if for any individual associ-ate or joint venture the investor-firm share of any

of the four designated items exceeded 25% of thecorresponding item for the investor firm, a 25%materiality disclosure was required. In this case,the investor firm had to disclose its share of thefollowing in respect of the individual investee:sales revenue, profit before tax, taxation, profitafter tax, fixed assets, current assets, liabilities duewithin one year, liabilities due after one year.

It is notable that, although it was issued at a timeof increasing support by the ASB for theInternational Accounting Standards Committee(IASC) harmonisation project, FRS 9 differedfrom the then current international accountingpractice, as reflected in IAS 28 and IAS 31, in anumber of significant respects. Areas of differenceincluded the requirement in FRS 9 to disclose in-formation about liabilities of associates, at a timewhen no such requirement existed in IAS 28, thelack of any provision within FRS 9 for proportion-ate consolidation, at a time when IAS 31 stronglyencouraged its use in accounting for joint ventures,the definition of significant influence, criteria forexclusion from the requirement to apply equity ac-counting, the reporting of components of equity-accounted income, the treatment of loss-makingequity-accounted investees, and the definition of types of joint venture (Ernst & Young LLP, 2001,ch. 7, Section 6). It is also notable that the revisionto IAS 28 that post-dated FRS 9, requiring the dis-closure of associate liabilities for accounting peri-ods beginning on or after 1 January 2005, and theproposed revision to IAS 31, requiring joint ven-tures to be accounted for by the equity method, aremoving international accounting practice with re-gard to the disclosure of the liabilities of equity-ac-counted investees closer to the FRS 9 position. 3However, it should also be noted that the proposedchange to IAS 31 runs counter to the path taken byFRS 9 in one respect, in that it substantially re-duces the difference between the method used toaccount for investments in joint ventures and thatused to account for investments in associates.

2.2. Related researchA number of previous studies examine the use-

fulness to investors of accounting disclosures re-

garding associates and joint ventures, focusing ontheir value relevance as measured by associationwith market value, on their risk relevance as meas-ured by association with risk measures, or on theirforecasting relevance as measured by their contri-bution to the forecasting of accounting numbers.

Motivated by the possibility that US GAAPslack of distinction between associates and jointventures might limit the usefulness of US account-ing data, Soonawalla (2006) examines the rele-vance to investors of associate and joint venturedisclosures in Canada and the UK for theperiods 19952000 and 19972000, respectively.Relevance is measured both by forecasting rele-vance and by value relevance. The focus is on thepotential incremental relevance of (i) equity-ac-counted income disaggregated into associate and

joint venture components, beyond equity-account-

ed income of associates and joint ventures takentogether, (ii) single-line equity-accounted net in-vestment disaggregated into associate and jointventure components, beyond the single-line equi-ty-accounted net investment in associates and jointventures taken together, and (iii) joint venture rev-enues and expenses, beyond joint venture earn-ings. The study finds evidence that disaggregatedassociate and joint venture information is forecast-ing-relevant and value-relevant beyond aggregateinformation, and therefore suggests that failure todistinguish between associates and joint ventures

may result in the concealment of valuable informa-tion. It does not consider the effect of investee-lia-bility disclosures or of investor-firm guarantees.

Loureno and Dias Curto (2006), using UK datafrom 1999 to 2004, report that investors value thegross assets of joint venture investees differentlyfrom investor-firm gross assets and that they valuethe liabilities of joint venture investees differentlyfrom investor-firm liabilities. The study does notconsider associates or the effect of investor-firmguarantees.

For Canadian firms for the period from 1995 to2000, Kothavala (2003) examines whether infor-mation on joint venture investees presented usingproportionate consolidation, which reflects the lia-bilities of joint venture investees, explains marketrisk measures better (is more risk-relevant) thaninformation based on single-line equity account-ing, which does not. The study is motivated in partby suggestions that the use of equity accountingrather than proportionate consolidation, by failingto reflect liabilities of investees, may allow thoseinvestees to be used as an off-balance-sheet-fi-nancing device (Bierman, 1992). The results of thestudy are mixed. Accounting numbers based onproportionate consolidation are more risk-relevantthan those based on single-line equity accounting

h i k i d i i l ili b


3 Unlike the international standards, FRS 9 made distinc-tions with regard to materiality, which resulted in liabilities of some non-material associates not being disclosed and detailsof particularly material associates and joint ventures being dis-closed individually. Also, it required that disclosures with re-


5/19

bond ratings. A related study by Stolzfus and Epps(2005) uses US data for the period from 1996 to1999 to examine whether accounting data con-structed as if joint ventures are accounted for byproportionate consolidation explain risk premiabetter than reported accounting data based on the

equity method. The study reports that, althoughproportionate consolidation does not dominate eq-uity accounting in explaining risk premia for all

joint ventures taken together, it does so for jointventures where obligations are guaranteed by theinvestor firm.

Graham, King and Morrill (2003), usingCanadian data for the period from 1995 to 2001,find that reported accounting data, in which jointventures are accounted for by proportionate con-solidation, have greater forecasting relevance thanaccounting data constructed as if joint ventures areaccounted for by the equity method.

Bauman (2003) uses a sample of 150 US firm-years for the years 2000 and 2001 to examine thevalue relevance to investors in the investor firm of disclosures of the liabilities of equity-accountedinvestees, which he terms off-balance-sheet activ-ities concealed by the equity method of account-ing. Because of the possibility that investors in theinvestor firm might place a higher weight on in-vestee-liability disclosures if there is an investor-firm guarantee of investee-firm obligations, adistinction is drawn between guarantee cases andnon-guarantee cases. No distinction is drawn be-tween joint ventures and other equity-accountedinvestees. In regression models in which the de-pendent variable is the market value of equity andexplanatory variables include investee-liabilitydisclosures, the coefficient on investee-liabilitydisclosures is negative both in guarantee cases andin non-guarantee cases, both in 2000 and 2001.Consistent with investors placing greater weighton investee-liability disclosures in the presence of an investor-firm guarantee, the coefficient forguarantee cases is significant in both years where-as that for non-guarantee cases is significant in

only one year, and the guarantee coefficient is

lower (more negative) than the non-guarantee co-efficient in both years.

2.3. SummaryThe issues of whether and how to reflect in the

financial statements of investor firms information

about equity-accounted investees have been debat-ed extensively and have generated a considerabledegree of regulatory activity, some of which isongoing at the time of writing. Prior research hasreported evidence on the value-relevance, risk-rel-evance and forecasting-relevance of such informa-tion, and has found that investors view informationabout equity-accounted investees differently de-pending upon whether it relates to associates or to

joint ventures and depending upon whether or notthere is an investor-firm guarantee of investee-firmobligations. A US study by Bauman (2003) has

found a negative association between the marketvalue of the investor firm and disclosures of liabil-ities of equity-accounted investees, with the valu-ation impact being more pronounced in thepresence of guarantees than in the absence of guar-antees.

In this study, we use a sample of UK firm-years,substantially larger than that used in Baumans USstudy and drawn from the six years immediatelyfollowing the major increase in disclosure require-ments brought about by FRS 9, to examine thevalue relevance to investors in the investor firm of

disclosures of liabilities of equity-accounted in-vestees. We also examine whether this differs be-tween associates and joint ventures, and whether itdiffers between guarantee cases and non-guaranteecases.

3. Research designThe main objective of this study is to examinewhether disclosures of the investor-firm share of the liabilities of equity-accounted investees arevalue-relevant to investors in the investor firm. Inorder to address this objective, the following re-

gression model is estimated:

Vol. 37 No. 4. 2007 271

4 Market value of equity is measured on an ex-dividendbasis, in order to be consistent with the balance sheet. The ad-

justment by the total return on the firms shares for the threemonths after the balance sheet date gives a market value meas-ure which (i) reflects information becoming available in thethree months following the balance sheet date, and (ii) is con-

(1)

where: MVE it is the ex-dividend market value of equity of the investor firm i at the balance sheetdate t , adjusted by the total return on the firmsshares for the three months after that date; 4 BVAi,t -1is the book value of assets of firm i at the previousb l h d 1 i l i f i i


6/19

ventures, and is used to scale all variables in themodel other than dummy variables; ILD it is theFRS 9-mandated investee-liability disclosure inrespect of material associates and all joint ven-tures, taken together; BVAit is the book value of as-sets of firm i at the balance sheet date t , inclusive

of its equity-accounted net investment; BVLit is thebook value of liabilities of firm i at the balancesheet date t , inclusive of any equity-accounted net-liability position recorded within provisions; NI it isthe net income of firm i for the period ended at thebalance sheet date t ; LOSS it is a dummy variablethat takes the value of 1 if firm i reports a loss inthe period ended at balance-sheet date t and zerootherwise; I j,it is a dummy variable that takes thevalue of 1 if firm i is from industry group j andzero otherwise, and allows for differences in theintercept term across five industry groups (seebelow); Y s,it is a dummy variable that takes thevalue of 1 if year t is equal to s and zero otherwise,and allows for differences in the intercept termacross six years (see below); the , , and terms are regression coefficients; and e1,it is a ran-dom error term. The first digit in each coefficientsubscript refers to the number of the regressionmodel, of which there are four in total (see below).

The explanatory variable of interest is ILD it / BVAi,t -1, which is the book-value-scaled in-vestee-liability disclosure for material associatesand all joint ventures. ILD it is the amount by whichthe equity-accounted net investments in the in-vestees for which disclosures are made would haveto be grossed up to give the investor-firm share of the gross assets and liabilities of those investees. Inthe majority of cases for which a disclosure ismade, the equity-accounted net investment appearsas an asset on the investor firms balance sheet. Insuch cases, ILD it comprises the investor-firm shareof the total liabilities of the relevant investees. Insome cases a disclosure is made where the equity-accounted investor-firm share of investee net liabil-ities is recorded within provisions by the investorfirm. In such cases, ILD it comprises the investor-

firm share of the total assets of the relevant in-vestees, equal to the investor-firm share of the totalliabilities of the relevant investees less the provi-

sion already recorded by the investor firm in re-spect of its share of those investeesnet liabilities. 5

ILD it does not reflect liabilities of non-material as-sociates that are not disclosed in the financial state-ments of the investor firm. In light of the fact thatthe requirement for UK firms to make investee-lia-

bility disclosures was partly motivated by concernthat single-line equity accounting was being usedas an off-balance-sheet financing device which hadthe effect of concealing the level of group gearing(ASB, 1996; Crichton, 1996; Johnson and Holgate,1996), and in light of the US findings of Bauman(2003), we predict that the coefficient on thesemandated disclosures is negative ( 11 < 0).

The other variables on the right-hand side of model (1) are control variables. We predict a posi-tive coefficient on book value of assets ( 12 > 0) anda negative coefficient on book value of liabilities(

13< 0). The dummy variable for loss cases is in-

cluded because of the evidence in Hayn (1995) thatlosses have a lesser impact on firms market valuethan profits. 14 is the coefficient on net income forprofit cases and 15 is the coefficient on net incomefor loss cases less the coefficient on net income forprofit cases; we predict 14 > 0 and 15 < 0. Thecontrol variables used in model (1) also appear inmodels (2), (3) and (4) described below, and theirpredicted signs in these models are as in model (1).

We also estimate two further regression models,each of which is an expansion of model (1), in orderto examine the value relevance of different classes of investee-liability disclosure. We test whether there isa difference in the value-relevance regression coeffi-cients (i) between investee-liability disclosures forassociates and investee-liability disclosures for jointventures and (ii) between investee-liability disclo-sures in the absence of investor-firm guarantees of investee-firm obligations and investee-liability dis-closures in the presence of such guarantees. Inmodel (2), investee-liability disclosures are dividedinto associate and joint venture components. Inmodel (3), they are divided according to whether ornot there is an investor-firm guarantee of investee-

firm obligations, without distinguishing between as-sociates and joint ventures, as in the model used byBauman (2003). Models (2) and (3) are as follows:


5 It is possible that an equity-accounted net-asset position and an equity-accounted net-liability position, recorded within pro-

(2)


7/19

where: ILDA it and ILDJ it are the separate investee-liability disclosures for firm is associates and jointventures, respectively; N it (G it ) is a dummy variablethat takes the value of 1 if there is not (if there is)an investor-firm guarantee of investee-firm obliga-tions and zero otherwise; and other notation is aspreviously defined. Our use of the dummy vari-ables N it and G it in model (3) is as in the US studyby Bauman (2003).

Regression coefficients in models (2) and (3)

can be interpreted as follows: 21 is the coefficienton associate-liability disclosures; 22 is the coeffi-cient on joint venture-liability disclosures; 31 isthe coefficient on all investee-liability disclosurestaken together where there is no investor-firmguarantee; 32 is the coefficient on all investee-liability disclosures taken together where there isa guarantee. Consistent with model (1), we predicta negative coefficient on each of the investee-liability disclosures in models (2) and (3) ( 21 < 0, 22 < 0, 31 < 0, 32 < 0). Because the creditors of

joint ventures might be more likely than those of

associates to have explicit or implicit recourse tothe assets of the investor firm, we predict thatthe coefficient on joint venture-liability disclo-sures is lower (more negative) than the correspon-ding coefficient on associate-liability disclosures( 22 21 < 0). Also, because of the recourse to theassets of the investor firm conferred by an in-vestor-firm guarantee of investee-firm obligationsand in light of the US findings of Bauman (2003),we predict that the coefficient on investee-liabilitydisclosures is lower (more negative) in the pres-ence of such guarantees than in the absence of

such guarantees ( 32 31 < 0).Finally, we estimate a fourth model that com-bines the two dimensions considered in (2) and (3).This allows us to test whether there is a differencein the value-relevance regression coefficients (i)between associate non-guarantee cases and associ-

ate guarantee cases and (ii) between joint venturenon-guarantee cases and joint venture guaranteecases. It also allows us to test whether there is a dif-ference in the value-relevance regression coeffi-cients (i) between associate non-guarantee casesand joint venture non-guarantee cases, and (ii) be-tween associate guarantee cases and joint ventureguarantee cases. In Model (4) below, AN it ( AG it ) isa dummy variable that takes the value of 1 if thereis not (if there is) an investor-firm guarantee of as-

sociate obligations and zero otherwise; JN it ( JG it )is a dummy variable that takes the value of 1 if there is not (if there is) an investor-firm guaranteeof joint venture obligations and zero otherwise;and other notation is as previously defined.

Regression coefficients in model (4) can be in-terpreted as follows: 41 is the coefficient on asso-ciate-liability disclosures where there is noassociate guarantee; 42 is the coefficient on asso-ciate-liability disclosures where there is an associ-ate guarantee; 43 is the coefficient on jointventure-liability disclosures where there is no joint

venture guarantee; 44 is the coefficient on jointventure-liability disclosures where there is a jointventure guarantee. Consistent with models (1), (2)and (3), we predict a negative coefficient oneach of the investee-liability disclosures ( 41 < 0, 42 < 0, 43 < 0, 44 < 0). Consistent with model(3), we predict that the coefficients on investee-liability disclosures are lower (more negative) inthe presence of an investor-firm guarantee thanin the absence of such a guarantee ( 42 41 < 0, 44 43 < 0). Also, consistent with model (2), wepredict that coefficients on joint venture-liability

disclosures are lower (more negative) than the cor-responding coefficients on associate-liability dis-closures ( 43 41 < 0, 44 42 < 0).

In light of the possibility that the relative com-plexity of model (4) might reduce its power to dis-tinguish between guaranteed and non-guaranteed

Vol. 37 No. 4. 2007 273

(3)

(4)


8/19

associates and between guaranteed and non-guar-anteed joint ventures, we also estimate the modelin two restricted forms: (i) including joint venturedisclosures, but not associate disclosures; (ii) in-cluding associate disclosures, but not joint venturedisclosures. The relevant coefficients from these

restricted forms of model (4) are very similar tothose from the full model (4), and are not reported.The restricted forms of the models are referred toin the account of sensitivity tests given in subsec-tion 5.2 below.

For the purposes of measuring differences be-tween coefficients on different classes of investee-liability disclosure within regression models (2),(3) and (4) as described above, we re-estimateeach model in an alternative incremental formula-tion in which each coefficient relating to investee-liability disclosures is equal to the differencebetween a pair of investee-liability-disclosure co-efficients from the relevant model. The coeffi-cients from these alternative incrementalformulations of models (2), (3) and (4), and t-sta-tistics thereon, are reported together with the re-sults for those models. 6

We also estimate each of models (1) to (4) sepa-rately for each of the six years covered by thestudy, exclusive of the dummy variable relating tothe year ( Y s,it ), and report the results from theseyearly regression models in summary form.

The overall data set is trimmed of outliers byremoving cases for which MVE it / BVA i,t -1 ,

ILD it / BVA i,t -1 , BVA it / BVA i,t -1 , BVL it / BVA i,t -1 and NI it / BVAi,t -1 fall in the most extreme 2% of the dis-tribution for that variable. The results from esti-mating each individual regression model arereported after deleting cases that meet the DFFITScutoff criterion of Belsey, Kuh and Welsch (1980,p. 28) for that model. 7

4. Data and descriptive statisticsThe data used in this study are drawn from the setof UK listed non-financial firms for which finan-

cial statement and market value data are availablefrom Extel and Datastream, respectively, for ac-counting year-ends from 23 June 1998, the manda-tory adoption date for FRS 9, to 31 December2003. We identify from balance-sheet data and in-come-statement data provided by Extel those firm-years where there is an investment in an associateand/or a joint venture, including cases where thefirms share of associate and/or joint venture net li-abilities is recorded within provisions. For thesefirm-years, we obtain published financial state-ments from the Thomson Research database orfrom the Perfect Information database. From thesefinancial statements, we identify the firm-years forwhich an FRS 9-mandated investee-liability dis-l i d d i h h

the amount corresponding to associates and theamount corresponding to joint ventures. Theseamounts are used to construct the investee-liabili-ty-disclosure regression variables ILD , ILDA , and

ILDJ . (From here on, notation referring to regres-sion variables omits the subscripts and the scaling

item.) In order to construct the dummy variables inrespect of guarantees ( N , G, AN , AG , JN and JG ),we also note from the financial statements whetherthere is an investor-firm guarantee of investee-firmobligations, whether there is a guarantee of associ-ate obligations and whether there is a guarantee of

joint venture obligations. 8 A guarantee is deemedto exist if the investor firm provides a guarantee inrespect of an investee firms liabilities, perform-ance or any of its contractual obligations. 9 In orderto conduct tests of the sensitivity of our results tothe use of more restrictive definitions of guaran-tees, we also note where there is a guaranteespecifically in respect of an investee firms liabili-ties and, where this is disclosed, we record infor-mation that is disclosed with regard to the amountof the guaranteed liabilities. For all firm-years forwhich an investee-liability disclosure is made, weobtain the following items from Extel (Extel itemnames in parentheses): book value of assets, de-noted BVA in regression models (TotalAssets);book value of liabilities, denoted BVL in regres-sion models (Debt + Creditors + OtherLiabilities +OtherLTLiabilities + TotalLiabMisc + (DeferredLiabilities DeferredLiabMinorityInterest)); netincome, denoted NI in regression models(ProfitAfterTax); creditor for dividends, used torestate the market value of equity to an ex-divi-dend basis (CreditorsDividendsDue). We also ob-tain the following items from DATASTREAM:market value of equity (item: HMV); return in-dices, used to adjust the market value of equity bythe total return for the three months after the bal-ance sheet date (data type: RI). For sensitivity testsreferred to below, we also collect from Extel andDatastream the available financial statement andmarket value data for all firms for which no


6 For each of the three models, all results other than thoserelating to investee-liability disclosures are identical in boththe initial formulation and the alternative incremental formu-lation.

7 As indicated below, we test the sensitivity of the results toan alternative treatment of outliers.

8 In all but one of the cases in which guarantees of equity-accounted investee-firm obligations are disclosed in the finan-cial statements, it is stated whether they relate to associates, to

joint ventures or to both. In the remaining case, it is not spec-ified whether guarantees relate to associates or joint ventures,and we assume that they relate to both classes of investee.

9 For accounting periods ending on or after 23 March 1999,the disclosure of guarantees in UK financial statementswas required by FRS 12: Provisions, Contingent Liabilitiesand Contingent Assets (ASB, 1998). For earlier accounting


9/19


10/19

investee-liability disclosure is made.Table 1 shows the breakdown of our data set by

year, broad industry group and class of FRS 9-mandated investee-liability disclosure. The yearand industry-group breakdowns correspond withthe year and industry-group dummy variables in-

cluded in regression models (1) to (4). There are7,211 firm-years for which accounting and marketvalue data are available from Extel andDatastream. In 2,184 of these firm-years (30% of 7,211), there are associate and/or joint venture in-vestments. It is notable that the proportion of firmswith investments in associates and/or joint ven-tures is particularly high in the utilities industrygroup. 10 In 1,220 of the 2,184 firm-years (56% of 2,184), there is an FRS 9-mandated associate-and/or joint venture-liability disclosure. Within the1,220 FRS 9 investee-liability disclosures, thereare 1,078 cases with joint venture gross equitymethod disclosures (88% of 1,220) and 246 caseswith associate materiality disclosures (20%). Wedo not distinguish between 15% and 25% materi-ality disclosures. The table also reports for eachFRS 9 disclosure category, the numbers of casesfor which an equity-accounted net liability is re-ported within the investor firms provisions. Of the1,220 cases for which there is an associate- and/or

joint venture-liability disclosure, there are 184such cases (15%); of the 1,078 cases for whichthere is a joint venture gross equity method disclo-sure, there are 177 such cases (16%); of the 246cases for which there is an associate materialitydisclosure, there are 11 such cases (4%). The finaltwo rows of the table report, respectively, the num-ber of firm-years in each disclosure category forwhich there is an investor-firm guarantee of in-vestee-firm obligations and, relevant to a sensitiv-ity test referred to below, the number of firm-yearsin each disclosure category for which there is aninvestor-firm guarantee of investee-firm liabilities.Of the 1,220 cases for which there is an associate-and/or joint venture-liability disclosure, there are232 cases (19%) where there is a guarantee of as-

sociate and/or joint venture obligations; of the1,078 cases for which there is a joint venture grossequity method disclosure, there are 191 caseswhere there is a guarantee of joint venture obliga-tions (18%); of the 246 cases for which there is anassociate materiality disclosure, there are 59 caseswhere there is a guarantee of associate obligations(24%). The corresponding figures for liabilityguarantees are 174 (14%), 146 (14%) and 43(17%), respectively.

Table 2 gives descriptive statistics for regressionvariables prior to outlier deletion, including non-zero cases of associate disclosures (ILDAs) andnon-zero cases of joint venture disclosures(ILDJs), both unscaled and scaled by lagged bookvalue of assets. The means for associate disclo-

sures are higher than those for joint venture disclo-sures, which is understandable in light of the factthat FRS 9 required disclosure of investee liabili-ties for all joint ventures, but only required disclo-sure for associates if they were material. Thestatistics for the scaled data reveal the existence of some extreme cases that suggest the need for theoutlier deletion referred to above. It is notable thatsome large losses cause the mean of scaled net in-come to be negative. Also, the median of book-value-scaled net income is only 4%, reflecting therelatively high proportion of loss cases in the peri-od covered by the study. 11 Table 3 gives correlationcoefficients between scaled variables used in re-gression model (1). These statistics are for the dataafter deletion of outliers, which reduces the dataset from 1,220 cases to 1,167 cases. It is notablethat the correlation between market value of equi-ty and investee-liability disclosures overall is veryclose to zero (0.01), and that the correlation be-tween ILDAs and ILDJs is low (0.12).

5. Results5.1. Results from regression models (1), (2), (3)

and (4)Table 4 reports the results of estimating regres-sion models (1), (2), (3) and (4). The first columnlists the explanatory variables for the regressionmodels, excluding the year and industry-groupdummy variables, the second column gives thepredicted signs of the regression coefficients, andthe remaining four columns report the regressioncoefficients and related information for each of thefour regression models. The table also reports re-sults from the alternative incremental formulationsof each of models (2), (3) and (4), which give the

differences between pairs of regression coeffi-cients on different classes of investee-liability dis-closure. In each case, the difference reported isequal to the coefficient on the first-named regres-sion variable less the coefficient on the second-named regression variable. Beneath each coefficientin Table 4, we report the t -statistic and the result of a significance test on the coefficient. For all coef-ficients, other than intercept terms, and for all dif-ferences between coefficients, significance testsare one-tailed. For the intercept terms, significancetests are two-tailed. For all regression coefficients,other than intercept terms, we also report the num-ber of years out of six for which the coefficient isof the predicted sign and the number of years outf i f hi h i i f h di d i d i


10 Outlier deletion did not result in significant variationfrom the distribution across time, industry group and disclo-sure category reported in Table 1.

11 A loss is reported in about 38% of the 7,211 firm-years


11/19

Vol. 37 No. 4. 2007 277

a b l e 2

e s c r

i p t i v e s t a t

i s t i c s

f o r r e g r e s s i o n v a r i a b

l e s

N u m b e r

S t a n d a r d

2 5 t h

5 0 t h

7 5 t h

o f c a s e s

M e a n

d e v i a t i o n

M i n i m u m

p e r c e n t i l e

p e r c e n t i l e

p e r c e n t i l e

M a x i m u m

r i a b l e s e x p r e s s e d i n m :

V E

1 , 2 2 0

2 , 6 7 0 . 5 7

1 1 , 1 7 1 . 7 2

0 . 2 2

4 7 . 0

5

2 1 7 . 5 5

1 , 1 6 4 . 3 2

1 8 7 , 2 8 7 . 7 9

D

1 , 2 2 0

2 1 8 . 6 4

8 1 0 . 3 4

< 0 . 0

1

1 . 3

2

1 0 . 2

3

9 1 . 8

0

1 0 , 0

4 2 . 0

0

D A ( n o n - z e r o c a s e s o n

l y )

2 4 6

3 9 8 . 7 1

1 , 3 7 0 . 8 0

< 0 . 0

1

3 . 8

9

2 0 . 0

0

1 5 6 . 0 0

9 , 9 5 4 . 0 0

D J ( n o n - z e r o c a s e s o n

l y )

1 , 0 7 8

1 5 6 . 4 6

5 2 0 . 6 7

< 0 . 0

1

1 . 2

0

8 . 3 5

8 6 . 0

0

6 , 2 1 2 . 0 0

V A

1 , 2 2 0

2 , 9 1 7 . 3 1

1 1 , 7 5 8 . 4 8

0 . 3 0

7 3 . 5

7

3 1 4 . 4 4

1 , 7 4 6 . 9 5

1 7 2 , 3 9 0 . 0 0

V L

1 , 2 2 0

1 , 5 3 6 . 7 7

4 , 6 0 3 . 5 5

0 . 1 0

3 3 . 4

4

1 9 9 . 9 0

1 , 1 3 1 . 9 5

5 5 , 3

3 1 . 3

9

I

1 , 2 2 0

6 2 . 3

6

7 9 4 . 9 5

1 5 , 6 7 9 . 0 0

0 . 3

3

9 . 1 6

5 0 . 7

0

8 , 0 0 7 . 7 7

r i a b l e s e x p r e s s e d a s a p r o p o r t i o n o f

g i n n i n g - o f - p e r i o d t o t a l a s s e t s :

V E

1 , 2 2 0

2 . 9 8

4 5 . 5

9

0 . 0 1

0 . 3

9

0 . 6 9

1 . 4 0

1 , 5 8 6 . 8 9

D

1 , 2 2 0

0 . 1 9

2 . 0 7

< 0 . 0

1

0 . 0

1

0 . 0 4

0 . 1 2

7 1 . 5

3

D A ( n o n - z e r o c a s e s o n

l y )

2 4 6

0 . 4 4

4 . 5 6

< 0 . 0

1

0 . 0

4

0 . 0 7

0 . 1 5

7 1 . 5

3

D J ( n o n - z e r o c a s e s o n

l y )

1 , 0 7 8

0 . 1 1

0 . 2 7

< 0 . 0

1

0 . 0

1

0 . 0 3

0 . 1 0

5 . 0 6

V A

1 , 2 2 0

1 . 4 2

3 . 4 2

0 . 0 4

0 . 9

6

1 . 0 6

1 . 2 1

8 5 . 5

5

V L

1 , 2 2 0

0 . 7 0

0 . 7 5

0 . 0 2

0 . 4

7

0 . 6 1

0 . 7 8

1 3 . 0

5

I

1 , 2 2 0

0 . 0

3

0 . 5 7

9 . 3

9

< 0 . 0

1

0 . 0 4

0 . 0 8

5 . 2 5

o t e s :

T h i s t a b l e r e p o r t s

d e s c r i p

t i v e s t a t

i s t i c s i n

m a n

d a s a p r o p o r

t i o n o f p r e v i o u s - p e r

i o d a s s e

t s , b

e f o r e

d e l e t i o n o

f o u

t l i e r s ,

f o r

t h e

1 , 2 2 0 c a s e s

f o r w

h i c h

t h e r e

i s a n

i n v e s t e e -

l i a b i l i t y

d i s c

l o s u r e

( s e e

T a b l e

1 ) .

T h e n o

t a t i o n u s e d

i n t h e

t a b l e

i s a s

f o l l o w s :

M V E i s t h e e x - d

i v i d e n

d m a r

k e t v a l u e o f e q u i

t y o f

t h e

i n v e s t o r

f i r m a t

i t s b a l a n c e s h e e

t d a t e , a d

j u s t e d

b y t h e t o t a l r e

t u r n o n

t h e

f i r m

s s h a r e s

f o r t h e

t h r e e m o n

t h s

a f t e r t

h a t d a t e ;

I L D

i s t h e

F R S 9 - m a n

d a t e d i n v e s t e e - l i a

b i l i t y

d i s c

l o s u r e

i n r e s p e c

t o f m a t e r i a

l a s s o c

i a t e s a n

d a l

l j o i n t v e n t u r e s , t

a k e n

t o g e

t h e r ;

I L D A i s t h e

i n v e s t e e - l i a b

i l i t y d i s c

l o s u r e

f o r a s s o c

i a t e s ;

I L D J i s t h e

i n v e s t e e - l i a b

i l i t y d i s c

l o s u r e

f o r j o i n t v e n t u r e s ;

B V A

i s t h e

b o o k v a

l u e o f a s s e

t s a t

t h e f i r m

s b a l a n c e s h e e

t d a t e ,

i n c l u s

i v e o f i t s e q u i

t y - a c c o u n t e d n e

t i n v e s

t m e n t

i n a s s o c i a t e s a n

d / o r

j o i n t v e n

t u r e s ;

B V L

i s t h e

b o o k v a

l u e o f

l i a b i l i t i e s a t t h e

f i r m

s b a l a n c e s h e e

t d a t e ,

i n c l u s i v e o f a n y e q u i

t y - a c c o u n t e d n e

t - l i a b i l i t y p o s i

t i o n r e c o r d e d w

i t h i n p r o v i s i o n s ;

N I

i s t h e n e

t i n c o m e o f

t h e

f i r m

f o r t h e a c c o u n

t i n g p e r i o

d e n

d e d a t

t h e b a

l a n c e s h e e

t d a t e .


12/19

For all four models, all control variables ( BVA, BVL, NI and NILOSS ) are of the predicted sign

and significantly different from zero at the 1%level, BVA, NI and NILOSS are each of the pre-dicted sign and significant in all six years, and

BVL is of the predicted sign in four years and sig-nificant in one year.

Model (1) examines the value relevance of in-vestee-liability disclosures ( ILDs ), without distin-guishing between associate and joint venture casesand without distinguishing between guarantee andnon-guarantee cases. The coefficient on the ILD isof the predicted negative sign and is significant.When the model is estimated separately for each of the six years from 1998 to 2003, the coefficient isof the predicted sign in all six years, although it issignificant in only two years. The result suggeststhat FRS 9-mandated disclosures, which facilitatedthe grossing up of equity-accounted net invest-ments in associates and joint ventures to reflect theinvestor-firm share of the liabilities of those in-vestees, gave investors a negative signal consistentwith the concerns about concealment of off-bal-ance-sheet financing reflected in ASB (1994),ASB (1996), Bierman (1992), Crichton (1996),Johnson and Holgate (1996) and Milburn andChant (1999). This finding is consistent with theUS finding of Bauman (2003).

M d l (2) h h th th i ffi

(more negative) for joint ventures, over which theinvestor firm exercises joint control (regression

variable: ILDJ ), than for associates, over which itmerely exercises significant influence (regressionvariable: ILDA ). The coefficients on associate dis-closures and on joint venture disclosures are eachof the predicted negative sign; the coefficient on

joint venture disclosures is significant whilst thatfor associate disclosures is not. When model (2) isestimated separately for each year, the coefficientson associate and joint venture disclosures are eachof the predicted negative sign in five of the sixyears, the coefficient on associate disclosures isnever significant, and the coefficient on joint ven-ture disclosures is significant in only one year. Inthat the coefficient for joint ventures is significantwhilst that for associates is not, the overall six-year findings are consistent with our predictionthat the negative impact of joint venture disclo-sures is stronger than that of associate disclosures.However, the coefficient for joint ventures is high-er (less negative) than that for associates, and inthe yearly regression models the difference be-tween the coefficients is of the predicted sign inonly two of the six years, and is never significant.

Model (3) shows whether the regression coeffi-cient on investee-liability disclosures is lower(more negative) in the presence of investor-firm

t f i t fi bli ti ( i


Table 3Pearson correlation coefficients for regression variables

MVE ILD ILDA ILDJ BVA BVL NI

MVE 1.00 ILD 0.01 1.00 ILDA 0.02 N/A 1.00 ILDJ 0.02 N/A 0.12 1.00 BVA 0.50 0.05 0.00 0.05 1.00 BVL 0.18 0.14 0.04 0.18 0.52 1.00 NI 0.17 0.08 0.06 0.05 0.02 0.11 1.00

Number of cases: 1,167

Note:This table reports the Pearson correlation coefficients for the following regression variables, scaled by previ-ous-period assets and after outlier deletion as described in the text:

MVE is the ex-dividend market value of equity of the investor firm at its balance sheet date, adjusted by thetotal return on the firms shares for the three months after that date;

ILD is the FRS 9-mandated investee-liability disclosure in respect of material associates and all joint ven-tures, taken together; ILDA is the investee-liability disclosure for associates; ILDJ is the investee-liability disclosure for joint ventures; BVA is the book value of assets at the firms balance sheet date, inclusive of its equity-accounted net invest-

ment in associates and/or joint ventures; BVL is the book value of liabilities at the firms balance sheet date, inclusive of any equity-accounted net-li-

ability position recorded within provisions; NI is the net income of the firm for the accounting period ended at the balance sheet date.


13/19

Vol. 37 No. 4. 2007 279

Table 4Regression results

Regression variable Predicted Model (1) Model (2) Model (3) Model (4)(scaling item and sign Coefficients Coefficients Coefficients Coefficientssubscripts are (t-statistics) (t-statistics) (t-statistics) (t-statistics)

suppressed in notation) [years: note 3] [years: note 3] [years: note 3] [years: note 3]Intercept ? 1.06 1.04 1.04 0.99

(7.46)** (7.52)** (7.37)** (6.86)**

ILD Negative 0.34(2.06)*

[6; 2]

ILDA Negative 0.56(1.43)[5; 0]

ILDJ Negative 0.33(2.07)*

[5; 1]

ILDN Negative 0.34(1.86)*

[5; 1]

ILDG Negative 0.37(1.47)[5; 0]

ILDAAN Negative 1.33(4.04)**

[5; 3]

ILDAAG Negative 0.37(0.65)[3; 0]

ILDJJN Negative 0.23(1.21)[5; 0]

ILDJJG Negative 0.73(3.01)**

[6; 0]

BVA Positive 1.50 1.53 1.50 1.46(9.52)** (10.03)** (9.38)** (8.66)**

[6; 6] [6; 6] [6; 6] [6; 6]

BVL Negative 0.33 0.37 0.36 0.35(2.44)** (2.70)** (2.57)** (2.49)**

[4; 1] [4; 1] [4; 1] [4; 1]

NI Positive 7.82 7.76 7.87 7.96(10.53)** (10.54)** (10.62)** (9.78)**

[6; 6] [6; 6] [6; 6] [6; 6]

NILOSS Negative 10.48 10.37 10.47 10.75(10.57)** (10.63)** (10.72)** (10.53)**

[6; 6] [6; 6] [6; 6] [6; 6]


14/19


Table 4Regression results (continued)

Regression variable Predicted Model (1) Model (2) Model (3) Model (4)(scaling item and sign Coefficients Coefficients Coefficients Coefficientssubscripts are (t-statistics) (t-statistics) (t-statistics) (t-statistics)

suppressed in notation) [years: note 3] [years: note 3] [years: note 3] [years: note 3] Differences (note 2): ILDJ less ILDA Negative 0.23

(0.57)[2; 0]

ILDG less ILDN Negative 0.03(0.12)[4; 0]

ILDAAG less ILDAAN Negative 1.70(2.70)[1; 0]

ILDJJG less ILDJJN Negative 0.50(1.90)*

[5; 0]

ILDJJN less ILDAAN Negative 1.10(3.04)[1; 0]

ILDJJG less ILDAAG Negative 1.10(1.72)*

[5; 1]

Adjusted R-squared 50.6% 50.8% 50.0% 48.5%Number of cases 1,167 1,167 1,167 1,167

Notes:1. Regression models are as follows:

(1)

(2)

(3)


15/19

Vol. 37 No. 4. 2007 281

Table 4Regression results (continued)

(4)

where:

MVE it is the ex-dividend market value of equity of the investor firm i at the balance sheet date t , adjust-ed by the total return on the firms shares for the three months after that date;

BVAi,t 1 is the book value of assets of firm i at the previous balance sheet date t -1, inclusive of the equity-

accounted net investment in associates and/or joint ventures, and is used to scale all variables inthe model other than dummy variables; ILD it is the FRS 9-mandated investee-liability disclosure of the investor firm i at the balance sheet date

t in respect of material associates and all joint ventures, taken together; ILDA it and are the investee-liability disclosures for associates and joint ventures, respectively; ILDJ it N it (G it ) is a dummy variable that takes the value of 1 if there is not (if there is) an investor-firm guarantee

of investee-firm obligations, and zero otherwise; AN it ( AG it ) is a dummy variable that takes the value of 1 if there is not (if there is) an investor-firm guarantee

of associate obligations, and zero otherwise; JN it ( JG it ) is a dummy variable that takes the value of 1 if there is not (if there is) an investor-firm guarantee

of joint venture obligations, and zero otherwise; BVAit is the book value of assets of firm i at the balance sheet date t , inclusive of the equity-accounted

net investment; BVLit is the book value of liabilities of firm i at the balance sheet date t , inclusive of any equity-

accounted net-liability position recorded within provisions; NI it is the net income of firm i for the period ended at the balance sheet date t ; LOSS it is a dummy variable that takes the value of 1 if firm i reports a loss in the period ended at

balance-sheet date t , and zero otherwise; I j,it is a dummy variable that takes the value of 1 if firm i is from industry group j , and zero otherwise;Y s,it is a dummy variable that takes the value of 1 if t = s, and zero otherwise;the , , , and terms are regression coefficients;the e terms are random error terms.2. For each regression variable, the table reports (i) the coefficient estimated from the overall six-year data set

from 1998 to 2003, (ii) the t -statistic thereon (in parentheses), (iii) the result of a significance test on the co-efficient, where * (**) indicates that the coefficient is significantly different from zero at the 5% (1%) leveland (iv) except for intercept coefficients, information on regression coefficients for models estimated sepa-rately for each year [in square parentheses]. See note 3 for details of the information reported in respect of yearly regression coefficients. Under the heading of Differences, the table reports coefficients and relatedstatistics from alternative incremental formulations of models (2), (3) and (4), where each coefficient relat-ing to investee-liability disclosures is equal to the difference between a pair of investee-liability-disclosurecoefficients reported above. For each of the alternative incremental formulations of models (2), (3) and (4),all other coefficients (and t -statistics thereon) are identical to those reported above, and are not separatelyreported. All t -statistics are calculated using the heteroskedasticity-consistent covariance matrix estimator(White, 1980). For the intercept terms, significance tests are two-tailed. For all other coefficients, includingthose from the alternative incremental formulations of models (2), (3) and (4), significance tests are one-tailed, with the predicted sign of each coefficient being given in the second column of the table.

3. The pairs of figures reported in square parentheses relate to the coefficients estimated when regression mod-

els are estimated separately for each of the six years from 1998 to 2003. The first figure of each pair indi-cates the number of years out of six for which the coefficient is of the predicted sign; the second figure of each pair indicates the number of years out of six for which the coefficient is of the predicted sign and sig-


16/19

antees (regression variable: ILDN ). Consistentwith models (1) and (2), both coefficients are of thepredicted negative sign; the coefficient for non-guarantee cases is significant, whilst that for guar-antee cases is not. When model (3) is estimatedseparately for each year, the coefficients for each

class are each of the predicted negative sign in fiveof the six years, the coefficient for non-guaranteecases is significant in only one year, and the coeffi-cient for guarantee cases is never significant.Although the coefficient for guarantee cases is, aspredicted, lower (more negative) than that for non-guarantee cases, the difference between coeffi-cients is not significant. In the yearly regressionmodels, the difference between coefficients is of the predicted sign in four out of six years, but isnever significant. The lack of evidence in our re-sults that the negative valuation impact of investee-liability disclosures is more pronounced in thepresence of investor-firm guarantees contrasts withthe conclusion of the US study by Bauman (2003).

Model (4) combines the two dimensions consid-ered separately in models (2) and (3), in that itcompares coefficients for investee-liability disclo-sures in respect of the following four cases: associ-ate cases in the absence of associate guarantees(regression variable: ILDAAN ); associate cases inthe presence of associate guarantees (regressionvariable: ILDAAG ); joint venture cases in the ab-sence of joint venture guarantees (regression vari-able: ILDJJN ); joint venture cases in the presenceof joint venture guarantees (regression variable:

ILDJJG ). The coefficients for associate non-guar-antee cases and for joint venture guarantee casesare of the predicted negative sign and are signifi-cant; the coefficient for joint venture non-guaranteecases is of the predicted negative sign, but is notsignificant; the coefficient for associate guaranteecases is positive, but would not be significantly dif-ferent from zero in a two-tailed hypothesis test.With the exception of the coefficient for associateguarantee cases, these results are consistent withthose for models (2) and (3) with regard to the neg-

ative sign of the coefficients on investee-liabilitydisclosures. When model (4) is estimated separate-ly for each year, the coefficients on investee-liabil-ity disclosures are of the predicted negative sign inthree years for associate guarantee cases, and in

five or six years for the other three classes, but onlyin the case of associate non-guarantee disclosuresare any of the yearly regression coefficients signif-icant (three out of six). With regard to differencesbetween coefficients, two of our predictions aresupported by the results. The coefficient for joint

venture guarantee cases is more negative than thatfor joint venture non-guarantee cases, and the dif-ference is significant; also, the coefficient for jointventure guarantee cases is more negative than thatfor associate guarantee cases, and the difference issignificant. In both cases, the difference betweencoefficients is of the predicted sign in five out of six years; the difference between coefficients for

joint venture guarantee cases and joint venture non-guarantee cases is not significant in any year; thedifference between coefficients for joint ventureguarantee cases and associate guarantee cases is

significant in one year only. Our prediction regard-ing the difference between the coefficients for asso-ciate guarantee cases and associate non-guaranteecases and our prediction regarding the differencebetween the coefficients for joint venture non-guar-antee cases and associate non-guarantee cases arenot supported: both differences are opposite to thepredicted sign, and would be significantly differentfrom zero in a two-tailed hypothesis test.

5.2. Sensitivity testsWe test the sensitivity of our results to a number

of alternative procedures.First, we test the sensitivity of our results to al-

ternative methods for distinguishing betweenguarantee cases and non-guarantee cases. For thispurpose, we repeat our analysis using data inwhich guarantee-related dummy variables are de-fined by reference to guarantees of liabilities,rather than by reference to guarantees of the broad-er category of obligations. We also repeat ouranalysis using data in which investee-liability dis-closures are divided into (i) amounts that we canidentify from notes to the financial statements aslikely to be viewed by readers as guaranteed by theinvestor firm (classified as guaranteed) and (ii)other amounts (classified as non-guaranteed).Furthermore, we allow for the possibility thatrecognition within the investor firms provisionsof a negative (net-liability) equity-accounted in-vestment might be interpreted as signalling the ex-istence of an implicit guarantee of the investeesobligations. In order to do this, we repeat ouranalysis using data in which investee-liability dis-closures in respect of all net-liability investees aretreated as guaranteed, regardless of whether aguarantee is explicitly mentioned in the financialstatements of the investor firm. 12

Second, because comparability of the value rele-f i li bili di l d j i


12 This treatment of disclosures in respect of net-liability in-vestees that are reflected within provisions will tend to over-state the extent of implicit guarantees under FRS 9. FRS 9required that equity-accounted net liabilities should only beomitted from the investor firms balance sheet if the investorfirm had irreversibly withdrawn from its relationship with theinvestee (ASB, 1997, paragraphs 4445). If the provisions hadbeen recorded under IAS 28, the treatment in our sensitivitytest would be clearly justified, since IAS 28 requires that net-liability cases should only appear on the investor firms bal-


17/19

to the fact that FRS 9 required disclosures for all joint ventures but only for material associates, theanalysis is repeated using only those disclosuresthat relate to material associates and material jointventures.

Third, the analysis is repeated using only those

cases for which there is both an associate disclo-sure and a joint venture disclosure.Fourth, the analysis is repeated including (i) all

cases for which there is an associate and/or jointventure investment, regardless of whether an in-vestee-liability disclosure is made, and (ii) allcases for which accounting and market value dataare available from Extel and Datastream, respec-tively, regardless of whether there is an equity-ac-counted investment.

Fifth, as mentioned earlier, in light of the rela-tively complex nature of model (4), the model is

estimated in two restricted forms: (i) including joint venture disclosures, but not associate disclo-sures; (ii) including associate disclosures, but not

joint venture disclosures.Finally, the analysis is repeated after deleting as

outliers 1% of extreme cases, instead of 2% of cases.

None of these alternative procedures gives re-sults that differ materially from those reported inTable 4.

5.3. Summary of resultsOverall, our results suggest that FRS9-mandated

investee-liability disclosures had a negative valua-tion impact. The value-relevance regression coef-ficient on all investee-liability disclosures takentogether is negative and significant. Seven of theeight value-relevance regression coefficients forparticular classes of investee-liability disclosuresare negative, although only four of these are sig-nificant. The negative sign of the coefficients oninvestee-liability disclosures is consistent with theUS findings of Bauman (2003). Contrary to ourpredictions, we do not find convincing evidencethat joint venture disclosures overall have astronger negative valuation impact than associatedisclosures overall or that disclosures in guaranteecases overall have a stronger negative valuationimpact than disclosures in non-guarantee casesoverall. However, consistent with our predictions,we do find evidence that disclosures for joint ven-ture guarantee cases have stronger negative valuerelevance than those for joint venture non-guaran-tee cases and those for associate guarantee cases.Our results should be viewed in light of the factthat, although our yearly value-relevance regres-sion coefficients on investee-liability disclosuresare of negative sign in 83% of cases (45 out of 54),they are significant in only 13% of cases (7 out of 54) b h k l f l

small sample sizes available for individual years.A feature of our results is the lack of convincing

evidence of a difference between the value rele-vance of investee-liability disclosures for jointventures and those for associates. This could indi-cate that there is limited economic difference be-

tween joint ventures and associates, at least withregard to information on investee indebtednessdisclosed in financial statements. We should alsobear in mind the possibility that the distinction be-tween the two types of investee entity may havebeen blurred because investor firms may havestructured investments as associates rather than as

joint ventures in order to avoid increased disclo-sure. This issue is not addressed in this study.However, the question of whether the existence of different accounting and disclosure requirementsfor different classes of investee may have influ-enced managers choices with regard to how in-vestments are structured would be a worthwhilesubject for further research.

6. ConclusionThe issues of whether and how to reflect in in-vestor-firm financial statements information aboutthe liabilities of equity-accounted investees havegiven rise to some debate and to some differencesin practice across GAAP regimes. This study ex-amines the value relevance of disclosures mandat-ed by the UK accounting standard FRS 9:

Associates and Joint Ventures (ASB, 1997) regard-ing the investor-firm share of liabilities of equity-accounted associates and joint ventures. FRS 9defined for the first time strict thresholds govern-ing such disclosure by UK investor firms, andbrought about a significant increase in disclosure.In light of the concerns about concealment of group gearing that partly motivated the investee-li-ability disclosure requirements of FRS 9, and thefindings of previous US research, we predict thatthe mandated disclosures are negatively associatedwith the market value of equity of the investor

firm. In light of the distinction that FRS 9 drew be-tween joint ventures and associates because theformer are subject to the investor firms joint con-trol whereas the latter are merely subject to its sig-nificant influence, and the possibility that thecreditors of joint ventures might be more likelythan those of associates to have explicit or implic-it recourse to the assets of the investor firm, thestudy also examines whether the negative valua-tion impact of investee-liability disclosures ismore pronounced for joint ventures than for asso-ciates. In light of findings of previous US research,it also examines whether the negative valuationimpact of investee-liability disclosures is morepronounced in the presence of investor-firm guar-

f i fi bli i h i h b

Vol. 37 No. 4. 2007 283


18/19

Subject to caveats regarding the significance of results from yearly regression models, our resultsconfirm the prior US finding of Bauman (2003)regarding the negative valuation impact of disclo-sures of the liabilities of equity-accounted in-vestees. From the perspective of the accounting

regulator, this evidence provides some confirma-tion of the value to investors of disclosures aboutliabilities of equity-accounted investees, and sug-gests that such disclosures are seen as negativesignals, consistent with the concerns about off-balance-sheet financing that partly motivated therequirements for such disclosures. These resultsare supportive of revisions made to IAS 28:

Investments in Associates (IASB, 2003a), whichpost-dated FRS 9 and which have required disclo-sure of associate liabilities for accounting periodsbeginning on or after 1 January 2005.

Our results relating to the distinction between joint ventures and associates have some relevancein light of the proposed revision of IAS 31:

Interests in Joint Ventures (IASB, 2003b), whichwould eliminate the use of proportionate consoli-dation and require that joint ventures be accountedfor by the equity method, as is required for associ-ates. There is some evidence in our study that thevalue-relevance of investee-liability disclosures ismore pronounced for guaranteed joint venture lia-bilities than for guaranteed associate liabilities.However, the lack of overall evidence of a differ-

ence between the value relevance of investee-liability disclosures for joint ventures and thosefor associates could indicate that there is limitedeconomic difference between joint ventures andassociates, at least with regard to information oninvestee indebtedness disclosed in financial state-ments, and we would not claim to report strong ev-idence of the need for different treatments for jointventure liabilities and associate liabilities. In anyevent, provided that the liabilities of equity-ac-counted joint ventures are required to be disclosedby way of note, separately from those of associ-

ates, the removal of the proportionate consolida-tion option in international GAAP would notsignificantly reduce the quality of information re-garding the liabilities of joint ventures.

ReferencesAccounting Principles Board (1971). APB Opinion No. 18:

The equity method of accounting for investments in com-mon stock . AICPA.

ASB (1994). Discussion Paper: Associates and Joint Ventures . UK: Accounting Standards Board.

ASB (1996). FRED 11: Associates and Joint Ventures . UK:

Accounting Standards Board.ASB (1997). FRS 9: Associates and Joint Ventures . UK:

Accounting Standards Board.ASB (1998). FRS 12: Provisions, Contingent Liabilities

and Contingent Assets . UK: Accounting Standards Board.ASC (1971). SSAP 1: Accounting for Associated

Companies . UK: Accounting Standards Committee.

ASC (1980). SSAP 18: Accounting for Contingencies . UK:Accounting Standards Committee.Bauman M. (2003). The impact and valuation of off-bal-

ance-sheet activities concealed by equity method account-ing. Accounting Horizons , 17(4): 303314.

Belsey, D., Kuh, E. and Welsch, R. (1980). Regression Diagnostics: Identifying Influential Data and Sources of Collinearity . Wiley.

Bierman, H. (1992). Proportionate consolidation and fi-nancial analysis. Accounting Horizons , 6 (4): 717.

Crichton, J. (1996). Finding a formula for togetherness. Accountancy , April, 1996: 93.

Ernst & Young LLP (2001). UK and International GAAP ,7th edition, Tolley.

Graham, R., King, R. and Morrill, K. (2003). Decisionusefulness of alternative joint venture reporting methods.

Accounting Horizons , 17 (3): 123137.Hayn, C. (1995). The information content of losses. Journal of Accounting and Economics , 20 (2): 125153.

IASB (2003a). IAS 28: Investments in Associates .International Accounting Standards Board.

IASB (2003b). IAS 31: Interests in Joint Ventures .International Accounting Standards Board.

IASB (2007). Amendments to IAS 31 Interests in Joint Ventures (March, 2007). International AccountingStandards Board.

Johnson, B. and Holgate, P. (1996). Associates and joint

ventures: The divorce. Accountancy , July, 1996: 95.Kothavala, K. (2003). Proportional consolidation versus

the equity method: A risk measurement perspective on re-porting interests in joint ventures. Journal of Accountingand Public Policy , 22 (6): 517538.

Loureno, I. and Dias Curto, J. (2006). The usefulness of alternative reporting methods for interests in jointly con-trolled entities: Evidence from UK listed companies.Working paper, ISCTE Business School.

Milburn, A. and Chant, P. (1999). Reporting Interests in Joint Ventures and Similar Arrangements . FinancialAccounting Standards Board.

SEC (1981). Accounting Series Release 302: SeparateFinancial Statements Required by Regulation S-X .Securities and Exchange Commission.

Soonawalla, K. (2006). Accounting for joint ventures andassociates in Canada, UK and US: Do US rules hide infor-mation?. Journal of Business Finance and Accounting , 33(3 & 4): 395: 417.

Stolzfus, R. and Epps, R. (2005), An empirical study of the value relevance of using proportionate consolidationaccounting for investments in joint ventures. AccountingForum , 29 (2): 169190.

White, H. (1980). A heteroskedasticity-consistent covari-ance matrix estimator and a direct test of heteroskedastic-ity. Econometrica 48 (4): 817838.



19/19

Reproduced withpermission of the copyright owner. Further reproductionprohibited without permission.

Documents

the Value Relevance of Disclosures of Liabilities