ELSEVIER Journal of Accounting and Economics 30 (2001) 401--420

JOURNAL OF Accounting &Economics

www .elsev ier.com/loca te/ec on base

Accounting standards and value relevance of financial statements: An international

analysis *

Mingyi Hung* Leventhal Schuul uf"Accuuntin!J and M arshall Sdwul uf"Business, University uj"Suuthem Califumia,

Los Angeles, CA 90089-0441, USA

Accepted 19 April 200 I

Abslract

Using 17,743 finn -year observations of industrial companies in 21 countries from 199 1 to 1997, th is paper finds t hat the use of accrual account ing (versus cash accounting) negatively affects the value relevance of fina ncia l statements in countries with weak shareholder protection. This negative effect, however, does not exist in countries with strong sha reholder protection. These fin dings are co nsistent with the belief that shareholder protection improves the effectiveness of accrual accounting, and suggest the importance of considering shareholder protection when formulating accounting policies related to accruals. (G! 2001 Elsevier Science B. V. All rights reserved.

JEL classification: GIS; i\141

Keyword 1.· Internationa l financial reporting; Accounting standards; Accrual accoun ting

* I am grateful to Jennifer Babcock, who not only provided indexes of global accounting standards bul a lso shared many important insights lhat substamially contribuled to the paper. I thank Paul Asquith, Paul Healy, S.P. Kothari, and G. Peter Wilson for their encouragement and guidance. T also thank C harles Chen (the referee), M a rk DeFond, M ary Ellen Ca rter, K.R. Suhrama nyam, Rohert Trezevan t, Rehecca Tsui, \Vim Va n der Stede, E ric \Vo lff, Jerry Zimmerman (the edi tor), and workshop participants at Bank of .Japan, Boston College, Emory Universily, Massachusetts Institule of Technology, T ulane University. University of British Columbia, Universily of Southern California. Waseda University, and the 2000 American Accounting Association Annual Meeting .

*Tel. : - 1-2 13-740-7377; fax: - 1 -2 1 3-747-2~ 15.

E-1nail address: mingyih@ usc.cdu (lvl. Hung).

Ol fiS-4101 /01 /$ - see front matter (Q 200 1 E lsevier Science B.V. A ll rights reserved. PTT: S 0 1 fi 5 - 4 1 0 1 ( 0 1 ) 0 0 0 1 1 - R

402 M. Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200 / j 401- 420

1. Introduction

This paper investigates the relation between accrual accounting and the value relevance of accounting measures in co un tries wi th different levels of shareholder protection. This issue is important because accrual accounting, which transforms cash flows into earnings, is a key feature of any accounting system. The study finds that stronger shareholder protection, an insti tutional factor characterizing a country's corporate governance environment (La Porta ct al. , 2000), improves the effectiveness of the accrual system. T his finding suggests the importance of considering shareholder protection when formulat­ing accounting policies related to accruals.

Accrual acco un ting provides hcttcr matching of revenues and expenses than cash accounting and therefore makes accounting information more value relevant. However, accrual accounting a lso presents more opportunities for managers to manipulate accruals for personal gain and hence may cau se accounting information to be less value relevant. Consequently, an accounting system mandating more accrual accounting (hereafter, referred to as 'a higher use of accrual accounting') has offsetting effects on the value relevance of accounting information.

I argue that managers are more likely to behave opportunistically in an environment with weak shareholder protection (La Porta et al. , 1997). Since a higher usc of accrual accounting provides managers with more opportunities to manage earnings and poor shareholder protection exacerbates this managerial propensity, I predict that a higher use of accrual acco unting negatively affects the value relevance of accounting information in markets with weak shareholder protection. In addition, since strong shareholder protection deters managers from manipulating accruals, I predict that strong shareholder protection will attenuate this negative impact.

Using 17,743 firm-year observations of ind ustrial companies in 21 countries from 1991 to 1997, I test the impact of accrual accounting on the value relevance of accounting numhcrs for countries with di fferent levels of shareholder protection. I measure a country's use of accrual accounting by the frequency of accrual-related accounting standards and evaluate the level of shareholder protection hy antidircctor rights and legal system (La Porta et a!., 1996; Ball et a!., 2000a,b). Following prior studies, I define the value relevance of financia l sta tements as the ahility o f accounting data to summarize information impounded in market prices (Chang, 1998; Francis and Schipper, 1999). As in Chang (1998), I focus on two summary accounting performance measures from financia l sta tements: earnings and return on equity (R0 £). 1

1 This study docs no t address operating cash flows because the usc of accru al accounting only aft'ccts earnings, not cash flows.

M. Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200 / j 401- 420 401

Controlling for tax and financial reporting conformity, the study has three primary findings. First, a higher usc of accrual accounting negatively a!Tccts the value relevance of accounting measures for countries with weak shareholder protection. Second, strong shareholder protection a ttenuates accrua l account­ing's negative effect on the value relevance of earnings. Finally, accrual accounting does not negatively affect earnings' value relevance for countries with strong shareholder protection.

My interpretation of the results depends on the assumption that price formation is roughly the same across countries. However, price formation likely varies across countries and is likely correlated with shareholder protection. A plausible scenario is that countries with strong shareholder protection arc likely to have more private information production and mandated disclosures, which ceteris paribus decrease earnings' value relevance by causing prices to lead earnings more. In this scenario, countries with strong shareholder protection would display a relatively lower association between earnings and stock returns because prices would lead earnings to a greater exten t. Therefore, high shareholder protection wo uld appear to reduce the value relevance of earnings. However, this scenario works against finding support for my hypotheses.2

My findings contribute to the literature on earnings' properties in two ways. First, since a country has a fairly constant accrual system, previous within-country studies generally compared the value relevance of earnings with that of cash flows (e.g., Ball and Brown, 1968; Bowen et a l. , 1987; Dec how, 1994; C heng et a l. , 1996). My study complements these within-country studies by comparing the value relevance of earnings across di!Terent countries' accrua l systems. Second, prior cross-country studies have mainly examined the relation between country-specific factors and the value relevance of accounting numbers (Alford et a!., 1993; Ali and Hwang, 2000; Ball ct a l. , 2000a, b).3 This study adds to these cross-country studies by addressing the impact of shareholder protection on the relation between the usc of accrual accounting and the value relevance of accounting information.

The paper is structured as follows. Section 2 develops the research hypotheses. Section 3 presents the research methodology. Section 4 dcscrihcs the sample selection and variable definitions. Section 5 shows the empirical results. Section 6 summarizes the findings and concl udcs the study.

2 Tl is possi hle tha t the re a re a lso oth er unspecified scenarios, where shareholder protection affects not jusl earnings but also price formalion via non-accouming-based information. lha t could provide allernative explanations for the resulls.

3 Allhough La Porta el al. (1996. 1997, 2000) discuss the relalion belween accounting standards and institutional factors across countries, their proxy for accounting standards is the disclosure level based on annual reports rather than the a ttribute of the accounting standards.

404 M. Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200 / j 401- 420

2. Hypothesis development

Accrual accounting systems are expected to generate more value relevant accounting performance measures (i.e., earnings and ROE) than cash nows because accrual systems are better at matching revenues and expenses (Ball and Brown, 1968; Dechow, 1994; and Cheng eta!. , 1996). Thus, accrual accounting performance measures help investors better assess firm values and operating performance than operating cash flows.

However, accrual systems a lso a llow managers to opportunistica lly manipulate accruals. Because managers make estimates for the accrual system and are often evaluated and rewarded based on accounting performance measures, managers might manipula te accruals for personal gain (Hea ly, 1985; DeAngelo, 1988; McNichols and Wilson, 1988) and thus cause accounting measures to he less value relevant.

Consequently, movement toward a higher use of accrual accounting has offsetting effects on the value relevance of accounting measures. For example, consider accounting for pensions. U.S. standards require that pension costs be recognized (i.e., accrued) in the balance sheet and charged to earnings when the costs are incurred . Recognizing pension costs when incurred, rather than when paid, better matches revenues and expenses and thus generates more value relevant performance measures for U .S. companies. However, a U.S. manager might hi as the estimate of a company's pension expense downward (upward) to prevent a negative earnings surprise (take a bath) and thus reduce the value relevance of earnings to investors. As a result, the net impact of adopting accrual pension accounting on the value relevance of accounting performance measures is unclear.

However, strong shareholder protection in the marketplace should attenuate management opportunism (Jensen and Meckling, 1976; Holmstrom, 1979). Alternatively, >veak shareholder protection in the marketplace will exacerbate the opportunism. Therefore, I argue that managers are more likely to manipula te accruals in weak shareholder protection environments than in strong shareholder protection environments. For example, the U.S. has many mechanisms for oppressed shareholders to make legal claims against directors, but Germany has few such mechanisms (La Porta ct a l. , 1996). While U .S. managers who materially misrepresent earnings generally face shareholders' class action suits and securities regula tors' inves tigations, German managers rarely face such consequences. Due to the higher cost of opportunistic behavior, U.S. managers, relative to German managers, are less likely to ex hibit such behavior.

The preceding discussion leads to two hypotheses. First , since a higher usc of accrual accounting provides more opportunities for earnings management and inadequate shareholder protection exacerbates this manage­rial propensity, I predict that the use of accrual accounting will negatively

M. Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200 / j 401- 420 405

aflect the value relevance of earnings in countries with weak shareholder protection.

H~pothesis I. The usc o f accrua l accounting negatively reduces the value relevance of accounting performance measures (i.e., earnings and ROE) in countries with weak shareholder protection.

Second, since shareholder protection deters managers from manipulating accruals, r predict tha t strong shareholder protection will reduce the ncgati vc impact of accrual accounting on value relevance.

H~pothesis 2. Strong sha reholder protection reduces the negative impact of accrual accounting on the value relevance of accounting performance measures.

3. Research design

This section discusses the method for measuring the use of accrual accounting, evaluating shareholder protection, calculating the value re levance of accounting performance measures, and assessing the link between tax reporting and financial accounting.

3.1. Use of accrual accounting

The use of accrual accounting represents the extent that the accounting system deviates from a cash method of accounting. To measure the use of accrual accounting, I create an accrual index from the data in the 1993 In ternational Accounting Summaries by Coopers and Lybrand (Coopers & Lybrand, 1993).4 I assume the accounting standard s in 1993 are representative of my sample period, 1991- 1997. One concern is that several countries in my sample began modifying their national accounting standards to conform to Internationa l Accounting Standards (lAS) during the period analyzed. However, Joos and Lang (1 994) suggest that harmonizing national accounting standards to achieve greater conformity is a slow process.

I compute the accrual index hy equally weighting 11 accrual-rela ted accounting standards for each country. Among the accounting standards summarized in Coopers and Lybrand (1993), I select 11 standards that arc directly related to the timing differences between cash receipt/disbursement and revenue/expense recognition. For example, one accrual-related accounting standard is accounting for research and development (R&D). All else equa l, a

4 1 thank Jennifer Babcock for graciously providing the accrual indexes for the sample countries.

40fi M. Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200 / j 401- 420

country that requires R&D expenditures to be capitalized and amortized, such as Finland, will have a higher accrua l index than a country that requires R&D expenditures to be immediately expensed, such as the U.S. The accrual index excludes accounting treatment of measurement issues, such as asset revaluation and inflation adjustment, because such treatment does not involve direct cash receipt/disbursement. 5

An accounting standard is assigned a weight of one if it applies accrual methods.6 For example, in the U.S., accounting for other post-retirement benefits is assigned a weight o f one since it requires the benefits to be accrued , but in Japan, it is assigned a zero because it does not require this. Panel A of Table l further explains the assignment of weights. Table 2 lists the resulting accrual indexes for the sample countries. Table 2 shows that the U.S. requi res the most accrual accounting, 0.86, and Switzerland the least, 0.32.

Table 2 also compares the accrual indexes among the M uellcr et a l. (1994) accounting clusters. These clusters are based on the overall similarity of countries' accounting practices. Mueller et al. classify countries into four clusters: British-American, Continental, South American, and Mixed Econo­my. The sample countries fall into either the British- American or the Continental clusters. Table 2 shows that the accrual indexes are similar for countries in the same accounting cluster, which is expected since the accrual indexes are based on accounting standards. For example, Australia, Canada , the U.K., and the U.S. a ll belong to the British- American cluster and have accrual indexes around 0.8 . The average accrual index is 0.75 for countries in the Briti sh-American cluster and 0.57 for those in the Continental cluster, with the difference significant at p-value less than 0.01.

3.2. Proxies for shareholder protection: Antidireclor rights indexes and legal systems

I use two alternative proxies for the level of shareholder protection in a country, antidirector rights and legal systems, based on La Porta et al. (1 996) and Ball et al. (2000a), because there is no single agreed­upon measure for shareholder protection. 7 The first proxy for shar e­holder protection is antidirector rights. Since shareholders exercise

5 Since the impact of assets revalua tion on value relevance of earnings has generated great interest among researchers (Faston e l a l. , 1993; Barth and C linch , 199R), T rerun the tests after includi ng accoun ti ng standards on the revaluation of properly, p lant, and equipment, a nd on the revaluation of in tangibles in the accrua l index . The results a re simi lar.

6 1 use an equal weighting method because there is no well-defined theory for other weighting methods. I note that the importance or an accounting standard varies across countries but see no reason that the equal weighting will b ias the results.

7 1 also combine antidircctor righ ts and legal systems to form a single proxy for shareholder protcct.ion. The results are q ualitatively the same.

M. Hung i Jnw·nal nf Accounting and Rconnmics 30 (200 / j 401- 420 407

Table l Procedures for calculating accrual and tax-hook conformi ty indexes." The accrual index measures the use of accrual accounting and the tax-book confom1ity index measures the link between tax reporting and financial accounting

Accounting standards

Panel A: Accrual index

Goodwiii/Ts it capitalized?

Equity method/Is it req uired? Depreciation/Is additional accelerated depreciation allowcd'lb Purchased intangiblc"j ls it capita lized? Developed intangible0 /Ts it capitalized? R&D expenditurejl s it capitalized? Interest capitalization/Is it capitalized? Finance lease/Is it capitalized? Percentage of completion/Is it allowed0

Pension/Arc future pension costs accrued'' Other post-retirement benefits/Are they accrued?

Panel B: Tax-book cvnfurmity index

A vcragc consensus cstima tc of the relation between tax and financial reporting. Do d eferred taxes exist?

Does legal form dominate substance? Ts ad ditional accelerated depreciation allowed? Do amortization periods depend on tax laws? Docs lease capitalization depend on tax law0

Rating

1-Yes 0.5-Yes ( < = 5 Years) 0-May expense !-Yes 0-No l-No 0-Yes

!-Required 0.5-Pcrmittcd 0-Not permitted 1-Permilled 0.5-L imited 0-Not t-x:mli tted 1-Permilled 0.5-L imited 0-Nott-x:mlitted !-Permitted 0-Not permitted !-Yes 0.5-0ptional/limited 0-No !-Required 0.5-Either 0-Not permitted !-Yes 0-No 1-Yes 0-No

- l if the resulting number from the calculation below is greater than 0 = 0 o thcrwisc !-Strong 0.5-Modcratc/significant 0-Weak

l-No d eferred tax 0.5-L im ited 0-Y csjrecognize 1-Yes 0.5-Sometimes 0-No 1-Yes 0.5-Limitt:d 0- o

!-Yes 0.5-Limited 0-No

!-Yes 0.5-Limited 0-No

Weight

1/1 1

lfl l lfl l

lfl l 1/1 1 1/1 1 lfl l lfl l lfl l lfl l 1/1 1

60%

20%

5%

5%

" Indexes are provided by Jennifer Babcock . Data sources are Coopers and Lybrand (1 993) and Alexander an d Archer ( 1995).

b Additional accelera ted deprecia tion refers to accelerated depreciation methods other than declining balance, double-declining balance, and sum-of-the-years'-digits methods.

c Excluding goodwill and R&D costs.

their rights by voting fo r directors, La Porta et a!. (1996) measured worldwide antidircctor rights by the case with which shareholders exercise their right to vote. They found that countries with strong anti­director rights have larger and more liquid capital markets, which implies

40R M . Hung i Jnw·nal nf Accounting and Rconnmics 30 (200 / j 401- 420

Table 2 Accrual index, accounting duster, antidi reclor rights ind ex, legal system, and lax-hook conformity index hy country"

Country Accrual index Acco unting Antidirector Legal system Tax-book cluster rights index conformity ind ex

Australia 0))2 British-A.merican 4 l 0 Belgium 0.6~ Continental 0 0 l Canada O.R2 Rri tish- A merican 4 I 0 Denmark 0.55 Continental ] 0 0 Finland 0.55 Continental 2 0 I France 0.64 Continenta l 2 0 Germany 0.41 Continental 0 I Hong Kong 0.64 British-American 4 0 Treland O.R2 Rri tish- A merican ] I 0 Ttaly 0.45 Continental 0 0 Japan 0.55 Continenta l 3 0 Netherlands 0.73 British American 2 0 0 New Zealand 0.73 British American 4 I 0 Norway 0))2 Continental 3 0 0 Singapore 0.64 Rri tish- A merican ] 0 South Africa 0.6R Rri tish- A merican 4 I 0 Spain 0.77 Continental 2 0 Sweden 0.59 Continenta l 2 0 Switzerland 0.32 Continenta l 0 U.K. 0.!!2 British-American 4 0 u.s. 0.!!6 British-American 5 0

a Dejiniliuns: Accrual index represents the degree to which the accounting system moves away from a cash method measure of per formance. A higher index value indicates higher use of accrual accounting. The index is constructed as reported in Table l. Accounting clusta refers to the cluster classification assigned according to the country's accounting practices by Mueller et al. (1994). Antidirectnr rig/us index indicates how easy it is for shareholders to exercise their voting ri ghts. This index is the antidirector ri gh t index constructed hy La Porta et al. (1996) . This index ranges from 0 to 5. It aggregates the following components of shareholder rights: (I) the ability to vote by mail, (2) the ability to gain comrol of shares during the shareholders' meeting, (3) the possibility of cumulative voting for directors, (4) the case of calling an extraordinary shareholders mee ting, and (5) th e availability of mechanisms allowing minority sh areholders to make legal claims against th e directors . Legalsyswm equals I if common law and equals 0 if code law. ?ax-bonk cnnfnrmily index shows the convergence between tax reporting and fin ancial accoun ting. Tt equals I for countries with high tax-book conformity and eq uals 0 for coumries with low conformity. The index is constructed as reported in Table I.

that antidirector rights and stimulate outside markcls.8

discourage opportunism by incumbent managers investors' willingness to participate m capital

~ Another implication for their results is that coumries with strong antidirector rights probably have more information production (i.e., analysts, news releases, mandated disclosures, conference calls, etc.). I thank th e editor for providing this insigh t.

M. Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200 / j 401- 420 409

The antidirector rights index, drawn from La Porta et al. (1996), ranges from zero to five. Each country star ts with an index value of zero and receives an additional point for each of the following:

1. The country allows shareholders to vote by mail. 2. The country does not require shareholders to deposit their shares before the

shareholders' meeting. 3. The government allows cumulative voting for directors. 4. T he minimum percentage of share capital that entitles a shareholder to call

an extraordinary shareholder meeting is less than 5%. 5. Minority shareholders are allowed to make legal claims against the directors.

The second proxy for shareholder protection is a country's legal system, generally classiAed as common law or code law. I assign the legal system a value of one if it is common law and zero if code law. Common-law countries are likely to exhibit greater shareholder protection than code-law countries because their public shareholders are more willing to provide funding to companies. Common law originated in England and was established chiefly by judges who resolved speciAc factual disputes. Code law (or civil law) originated in ancient Rome and was instituted as rules of conduct linked to concepts of justice and morality. Ballet al. (2000a) suggest that common laws are adapted to contracting in open, public markets, and code laws are appropriate for contracting between a small number of parties. Thus, in common-law countries, such as the U.K. and the U.S., companies rely heavily on public shareholders and creditors as sources of capital. In contrast, in code-law countries, such as France and Germany, companies typically rely on employees, managers, banks, and governments.

Table 2 presents the antidirector rights indexes and legal systems for the sample countries. A high association should exist between antidirector rights and legal system because both factors are proxies for shareholder protection. As expected, the data indicate tha t common-law countries have higher antidirector rights than do the code-law countries. For example, the U.S., a common-law country, has a score of Ave, the highest antidirector rights score. In contrast, Belgium and Italy, both code-law countries, have a score of zero, the lowest antidirector rights level. The mean (median) antidircctor rights is 3.89 (4.00) for the common-law countries and 1.75 (2.00) for the code-law countries, with both differences significant at p-values less than 0.0 1.

3.3. Value relevance of accounting pe1j'ormance measures

Following recent U.S. studies on value relevance, I define value relevance as the ability of an accounting measure to capture or summarize information that

4 10 M. Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200 / j 401- 420

aflects firm value. Y Using this definition, researchers often measure value relevance as the association between an accounting measure and stock returns and operationalize the value relevance in two ways: a portfolio-returns approach and a regression-variations approach. I usc the portfolio-returns approach to operationalize the value relevance of accounting measures becau se of its statistical superiority over the regression-variations approach (Kothari and Zimmerman, 1995; Francis and Schipper, 1999).

The portfolio-returns approach defines the value relevance of accounting measures as the proportion of information in security re turns captured hy the accounting measures (Alford et al. , 1993; Chang, 1998; Francis and Schipper, 1999). This approach measures value relevance as the total return that could be earned from a portfolio based on perfect foresight of earnings. Value relevance is scaled by the total return earned on a portfolio based on advance knowledge of market prices.

I use the following procedure to calculate the value relevance of accounting performance measures. First, at the end of each year, 1 rank firms in each country-specific sample hy change in net income (f../'1/f), change in ROE (f...ROE), and market-adjusted return (AdjRel). Next, following Alford et al. (1993) and Francis and Schipper (1999), 1 compute 15-month cumulative market-adjusted returns ending three months after the fiscal year, for three hedge portfolios:

/'... N I portfolio refers to the cq ually weighted hedge portfolio formed on the basis of /'...N I and scaled by beginning-of-year market value. I take long positions in stocks with the highest 40% of f.. IV I and short positions in the lowest 40% of each year.

/'..ROE portfolio refers to the equally weighted hedge portfo lio formed on the basis of f..ROE. I take long positions in stocks with the highest 40% of f...ROE and short positions in the lowest 40% of each year.

AdjRet portfolio refers to the equally weighted hedge portfolio formed on the basis of A((jRet. 1 take long positions in stocks with the highest 40% of AdjRet and short positions in the lowest 40% of each year.

Finally, I calculate the value relevance of earnings and ROE as the ratio of the corresponding return earned from the f...JVI and f..ROE portfolios divided by the return earned from the Adj Ret portfolio .

3.4. Tax-book conformity

Previous studies (Joos and Lang, 1994; Ali and Hwang, 2000) document that accounting information in countries with a strong link between tax and

~ See Francis and Schipper (1999) for discussions of alternalive de(initions of value relevance. Note that the definition of value relevance also depends on the information production system (or price formation process).

M. Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200/ j 401- 420 4 11

financial accounting is less value relevant. Thus, I control fo r this link by using a tax-book conformity index that shows the convergence between tax reporting and financial accounting. Based on Coopers and Lybrand (1993) and the 1995 European Accounting Guide, edited by Alexander and Archer (Alexander a nd Archer, 1995), I aggregate the average consensus estimate of the proximity of tax and financial accounting systems with other tax- related indicato rs shown in Panel B of Table 1. The higher the n um bcr, the stronger the link between tax and financial reporting.

To be consistent with prior studies (Alford ct a l., 1993; Ali and Hwan g, 2000), I use a high/low coding scheme to classify the resulting numbers from the tax-book conformity index calculation. 10 I assign the index a one if the resulting number is greater than zero and assign a zero otherwise (i.e., one designates high tax-book conformity, and zero designates low tax-book conformity). T he final numbers, provided in Table 2, a rc similar to those in prior studies. 11

4. Sample selection, variable definitions and descriptive statistics

4.1. Sample

I select the sample from the intersection of the Global Vantage Industrial; Commercial and Issue Files. I initially include countries with more than 100 total firm-year observations, if they have at least one observation each year. I restrict the sample to industria l firms (S IC codes 2000-3999 or 5000-5999), as in Alford et al. (1 993). I use this restriction to increase the homogeneity of the sample and the comparability of the results across countries.

In addition, each firm-year o bserva tion must satisfy four requirements. First , firm-year observations must have sufficient data to calculate ch ange in net income (!J.Nf) , cha nge in ROE (!J. ROE), and returns ( Ret) . Second, firm-year observations must not include the highest or lowest 1% values of each variable (NI , !J.N I, R OE, !J.ROE, and R et) within each country. Third, firms must not cha nge their fiscal year-end during the sample period. 1-ior example, if a firm changes its fiscal year-end in 1994 from June to December, its 1994 net income might cover only J unc 1994 to Decem bcr 1994. This would make the re lation bet>veen net income and a nnual returns in 1994 inconsistent >vith the relation in

10The resulting numbers fall toward extreme values. Therefore, it seems reasonable to assign the tax-hook conformity index as a hi nary variable rather than a continuous one. The results are qualitatively the same regardless of the specification of the tax-book conformity index (i.e., cominuous versus binary).

11 I have identical lax-book conformity classifications for the 16 coun tries, except Norway, in Alford et a l. (1993). I repeat the analyses after changing the dassification for Norway. The qualitative results do not change.

4 12 M . Hung i Jow·nal of Accounting a11d Rconomics 30 (200/ j 401- 420

Table 3 Distribution of finn -year observations hy country and fiscal year"

Country Total 1991 1992 1993 1994 1995 1996 1997

A ustralia 336 42 44 45 48 48 64 45 Belgium 87 8 12 l l l l 12 14 19 Canada 93g 130 125 l2g 125 140 147 143 Denmark 151 12 l l 12 13 32 35 36 Finland 90 10 12 10 10 14 15 19 France 442 49 54 49 50 57 89 94 Germany 869 100 I l l 113 116 138 157 134 Hong Kong 105 12 13 12 14 18 17 19 Ireland 124 13 17 l g lg 2 1 20 17 Italy 109 12 14 13 13 13 19 25 J apan 1,865 23 1 236 213 235 232 259 439 Netherlands 309 JO 40 4 1 44 44 49 55 New Zealand 43 5 5 6 6 8 8 5 Norway 107 15 15 15 16 17 16 13 Singapore 2 18 23 28 27 29 4 1 39 31 South Africa 90 l l 13 13 13 13 12 15 Spain 146 6 8 2 1 21 27 32 31 Sweden 94 15 17 10 10 11 16 15 Switzerland 138 18 16 11 14 24 26 29 U.K. 2,410 313 326 34 1 372 376 388 294 U.S. 9,072 1,273 1,283 1,321 1,347 1,307 I ,295 1,246

Total 17,743 2,334 2,400 2,450 2,525 2,593 2,717 2,724

" Sample: T select 17,743 firm-year observations from th e Global Va11tage flldzmrial/Commercial a nd Issue Files from 1991 to 1997. Tn selecti ng, T use the fo llowi ng cri teria: (1) each country needs to have more than 100 initial total firm-year observations, provided that it has a t least one ohservat ion each year, (2) only industrial firms (SIC codes 2000 3999 or 5000 5999) a re included, (3) each fi rm-year observa tion needs to have sufficient data to compute change in net income, change in ROE, and returns (4) firms do not change their fiscal year-end (5) firm-year observations do not have th e highest or lowest l % values of net income, change in net income, ROE, change in ROE, and returns within each coun try, and (o) firm-year data a re prepared under domestic accounti ng standards.

prior years. Fourth, finn-year data must be prepared under domestic accounting standards, ra ther than other standards such as lAS or modined U.S. standards.

The sample selection procedures yield 17,743 nm1-year ohservations from 1991 to 1997 for the 21 countries listed in Table 2.12 Table 3 lists the distribution of finn-year observations by country and fiscal year. The size of nm1-years for these sample coun tries ranges from 43 (New Zealand) to 9,072

12 I have three mor e sample coumries than Alford et al. ( 1993): Finla nd, New Zealand. and South Africa. I usc a more recen t time period so these countries pass the requirement of initial 100 firm­year observations.

M . Hung i Jnw·nal nf Accounting and Rconnmics 30 (200 / j 401- 420 411

(the U.S.). The number of firms grows over time mainly because of the increasing coverage of interna tiona l companies in the Global Vantage Files.

4.2. Variable defmitions

The earnings variable (.1\/ f) is net income before extraordinary items scaled by the beginning-of-year market value. The ROE variable (ROE) is earnings divided by beginning-of-year equity book values. The market value variable (M V) is price multiplied by num bcr of shares outstanding a t fiscal year-end. If a firm has multiple issues, I use the oldest issue because the identification of the primary issue is not available. The return variable (Ret) is the 15-month compound return (adjusted for dividends and splits) ending three months after the fiscal year-end. Market-adjusted return (Ar(jRet) is return minus the return on the equa lly weighted market portfolio in the firm 's country.

Table 4 shows the descriptive statistics of Ret, NI, and ROE across countries. Mean returns fall between - 8.0% (Japan) and 41.4% (Spain). The standard deviations of returns range from 31.7% (Japan) to 60.5% (Norway). This wide range shows the importance of controlling for market volatility across countries, and therefore supports my usc o f the portfolio-returns test instead of the regression-variations test. Returns are also positively skewed (medians lower than means) across all countries except New Zealand. The mean (median) N I fall between 1.5% (2.0%) in Japan and 146.9% (1 3.3%) in Sweden. 13 Finally, the mean (median) ROE fall between 3.1 % (3.8%) in Japan and 17.9% (17.7%) in South Africa.

5. Empirical results

Table 5 provides the market-adjusted returns for the /I,N l , /I,ROE, and AdjRet portfolios and value relevance measures of IV / and ROE for the sample countries. For example, the U.S.'s market-adj usted returns are 30.6% for the /I,N l portfolio and 80.4% for the AdjRet portfolio. Consequently, the value relevance of l'.lJ, the returns for the /l, l'·.lJ portfolio divided by the returns for the AdjRet portfolio, is 38.0%. This number indicates that about 40% of perfect foresight return s a rc available to U.S. investors with advance knowledge

13My sample descriptive statistics o r R ei and i'>'l ar e comparable with Ba ll e t al. (1998) for the seven countries in their study. Other multicountry s tudies (Alford et al. , 1993; Ali and H wang, 2000) do not provide descriptive statistics for comparison. T note th at there a re ex treme observations in :VI for Sweden, even though T have deleted observations with extreme 2% values. T ran domly check an nual reports for the companies with extreme values. The observations appear to be data error s in Global Van/aye Files. T herefor e. I delete two Swedish companies with NI greater than 300% and repeat the ana lyses. The revised standard devia t ion of 11/I for Sweden red uccs to under 60% from over ~00%. The qualita t ive results do no t change .

4 14 M. Hung i Jnw·nal nf Accounting and Rconnmics 30 (200/ j 401- 420

Table 4 Descriptive statistics of return , net income, and return on equity (HOP-)''

Ret(%) :VI(%) H()f; ( % )

Country N Mean Med. Std. Mean Med. Std. Mean Med. Std. dev . dev . d ev.

Australia 336 26.3 lH 39.3 5.4 7.1 ll.l 9.5 11.0 12.1 Belgium ~7 27.3 2 1.0 36.2 l:LO 7.0 9.!:l 14.7 14.0 9.6 Canada 93R 21.6 LU 46.9 3.] 6.1 16.4 5.9 R.6 2 1. 1 Denmark 15 1 16.9 11. 1 35.5 R.7 7.R 7.9 11.4 11.7 7.5 Finland 90 30.0 22.8 49 .6 16.2 11.4 37.2 9.7 13.3 37.1 France 442 28.0 22.0 39.8 5.2 6.2 11.9 9.2 10.3 12.7 Germany 869 11.6 6.2 33 .2 3.5 4.6 11.3 6.1 7.8 16.9 Hong Kong 105 l2.!:l 4.2 53 .2 7.0 7.3 lH 14. ~ 14.9 17.7 Treland 124 31.7 23.8 43 .2 R.6 9.4 10.1 15.9 14.6 17.2 Ttaly 109 37.3 23.6 59.0 4.1 7.0 17.7 5.1 7.2 12.4 Japan 1,865 - 8.0 - 12.3 31.7 1.5 2.0 3.2 3.1 3.8 6.9 . etherlands 309 28.0 23.5 34.5 9.1 9.1 5.8 17.9 16.6 12.4 New Zealand 43 14.1 15.0 41.3 6.0 7.2 12.4 11.6 l l.l 13.2 Norway 107 34.0 l !:l .2 60.5 9.!:l 9.9 10.9 17.9 l5. !:l 22.7 Singapore 21R 3.1 - U 36.8 3.8 4.3 4.R 7.8 7.3 9. 1 South Africa 90 24.7 14.4 45.8 10.7 8.9 9.7 17.9 17.7 7.9 Spain 146 41.4 30.7 55.4 4.8 7.6 19.5 10.4 10.5 16.3 Sweden 94 35.4 29.0 40 .3 146.9 13.3 866.1 15.9 16.0 16.3 Switzerland l3 !:l 25 .!:l 20.6 40.3 15.5 11.7 42.9 9.9 9.4 13.0 U.K. 2,410 19.2 16.0 3!:l .O 6.1 7.1 10.1 14.7 14.3 l !:l.9 U.S . 9,072 20.5 15.2 45 .9 2.2 6.0 18.0 8.0 11.7 28.1

a Sample: I select 17,743 fi rm-year observations from the Glubal Van/aye /ndustrialfCummercial and Issue Fih•s from 199 1 to 1997. I usc the following criteria: ( l) each country needs to have more than 100 initial total firm-year observations, provided that it has at least one observation each year, (2) only industrial fi rms (SIC codes 200Q-3999 or 500Q-5999) arc included , (3) each finn-year observation needs to have suffici ent data to calculate change in net income, change in RO 1-:, and returns (4) firms do not change their fisca l year-end (5) firm-year ohservati ons do not have the highest or lowest I% values of ne t income. chan ge in net income, ROE. change in ROE, and re turns within each country. and (6) firm-year da ta are prepared under domestic accoumin g standards.

Variable definitions: lV denotes the number of firm-year observations. Ret d enotes 15-month returns ending three months after the fisc.a l year-end, adjusted for dividends and stock splits. NI denotes net income before extraordinary items, scaled hy heginning-of-year market value. RO/<,· denotes return on equity.

of earnings. This percentage 1s compara ble to the findings in Alford et aL (1993).

Table 6 reports the Pearson and Spearma n correla tions among the value relevance of .NI, value relevance of ROE, accrual index, tax-book conformity index, and the proxies for shareholder protection (antidirector rights index and legal system). The upper (lower) tria ngle of T able 6 shows the Pearson (Spearman) correlation coefficients. Table 6 shows that the accrual index is

M . Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200 / j 401- 420 4 15

Table 5 Cumulative market-adjusted returns to hedge portfol ios based on perfect kn owledge of net income, ROf:, and stock price, 1991- 1997 (15-month period ending three months a fter the fisca l year-end)"

!lNI portfolio D.ROE portfolio AdjRet portfolio

Country Ad) Ret Proportion Ad) Ret Proportion Ad) Ret (%) of AdjRer (%) of Ad} Ret (%)

portfol io(%) portfolio (%)

Australia 22 .fi 33.9 31.2 46.R 6fi.(i Belgium 2.7 4.7 21.0 36.7 57.4 Canada 24.5 30.7 31.6 39.7 79.5 Denmark 14.3 24.0 25.2 42 .5 59.4 Finland !U 12.0 21.0 31.3 67 .0 France 22.fi 33.6 32.7 4R.7 67.1 Germany 14.9 2R.5 14.2 27. 1 52.2 Hong K ong 19.9 26.2 21.6 28.4 76.1 Ireland 22.3 32.7 12.9 18.9 68.3 Italy 19.4 30.1 26.5 41.3 64.2 Japan ~ . 3 22.6 10.3 2~.1 36.6 Netherlands 15.4 27.4 23.4 41.6 5fi.3 New Zealand 36.fi 55.7 24. 1 36.R 65.6 Norway 4.fi 5.3 17.1 19.R Rfi.5 Singapore 18.3 36.2 16.9 33.4 50.5 South Africa 22.4 30.4 3ll.O 51.6 73.ll Spain 16.4 21.4 lU 15.4 76.6 Sweden 10.5 17.9 16. 1 27.5 5R.6 Switzerland 28.7 48.6 39.4 66.7 59.0 U.K. 22.8 34.1 28.6 42.9 66.8 u.s. 30.6 3ll.O 34.6 43 .0 goA

Average 1 R.4 2R.3 23.7 36.6 65.2

" DPjill itinns: !'J.Nt denotes change in net income scaled by beginning-of-year market value. !'J. NJ portfolio r efers to the equally weighted hedge portfol io that takes long (short) positions in stocks with the highest (lowest) 40% of t.NI. t.ROE denotes changes in ROE. D.ROE portfolio refers to the equally weighted hedge portfolio that takes long (short) positions in stocks with the highest (lowest) 40% of t.ROE. AdjRet denotes market-adjusted return. AdjRet portfolio refers to th e equally weighted hedge portfolio that takes long (short) posi tio ns in stocks with the highest (lowest) 40% of AdjHet.

positively associated >vith the antidirecto r rights index and legal system a nd negatively associated with the tax-book conformity index (correlation coeffi cients arc a ll signi fi cant at the 0.0 1 level). T he results suggest tha t countries with a higher use of accrual accounting have st ronger shareholder pro tection and weaker a lignment hetween their tax and financial reporting. Additionally, the association bet>veen the value relevance of NI and legal system is significantly positive at the 0.01 level.

4 16 M. Hung i Jow·nal of Accounting and &anomies 30 (200/ j 401- 420

Table 6 Correlation coefficients of value relevance o f earnings, value rel evance o f HOI:', accrual index, antidirector ri ghts index, legal system, and tax-hook conformity index; Pearson (Spearman) correlation coelftcients in the upper (lower) triangle; two-tailed p-values in parentheses"

Variable Vai_Nf Vai_ROE Accrual index Anlidirector Legal system Tax-book rigllls litdex conformity

index

Val_Nl 0.5 1 -0.06 0.33 0 .51 - 0.2R (0.02) (O.RO) (0.1 4) (0.01) (0.2 1)

Vai_ROE 0.56 - 0.28 0.01 0.10 - 0.05 (0.01) (0.22) (0.98) (0.67) (0.83)

Accrual index 0.20 - O.ot 0.65 0.57 - 0.64 (0.3g) (0.9g) ( <0.01) (0.01) ( <0.0 1)

A 111 idirectm· 0.43 0.21 0.64 0 .7R - 0.7R rights index (0.05) (0.36) (<0.0 1) ( <0.01 ) (<0.0 1) Legal system 0.62 0.21 0.59 0.83 0 .75

(<0.01) (0.37) ( <0.0 l ) ( <0.01) (<0.0 1) Tax-book - 0.38 - 0.19 - 0.65 - 0.80 - 0.75 conformity index (0.09) (0.41) ( <0.0 1) ( <0.01) (<0.01)

" Definitions: I d efin e Vai_:VI, value relevance of earnings, as the market-adjusted return for the ;LV/ portfolio scaled hy the mar ket-adjusted return for the AdjHet portfolio, as summarized in column 3, laheled "Proportion of Ad} Net portfolio," ofTahle 5. Val_ROf:, value relevance of return on eq uity, is the market-adjusted return for the !lROE portfolio scaled by the market-adjusted return for the Ad)Ret portfolio, as summarized in column 5. labeled " Proportion o f Ad)Ret portfolio," of Table 5. Accrual index represents the degree that the accounting system deviates from a cash measure of performance. A higher index indicates higher accrual accounting standards. The index is constructed as reported in Table I. Anlidirec/Vr riyhts index indicates how easy it is for shareholders w exercise their voting rights. This index is the antidirecwr rights index constructed by La Porta et at. ( 1996). This index ranges from 0 to 5. It aggregates the following components of shareholder rights: (l) the ability to vote by mail, (2) the ability to gain control of shares during th e shareholders' meeti ng, (3) the possibility of cumulative voti ng for directors, (4) the ease of calli ng an extraordinary shareho lder meeting, and (5) the availahility of mechanisms of allowing minority shareholders to make legal claims against the directors. Leyal s_rstem eq uals I if common law and 0 if code law. Tax-book Wl1/urmily index shows the con vergence between tax reporting and fi nancia l accounting. It equals l for countries with high tax-book conformity and equals 0 for countries with low confo rmity. The index is constructed as reported in Table l .

Table 7 presents a series of country- level o rdinary least squares regre­ssions for testing the hypotheses. Models 1 and 2 (Models 3 and 4) u se value relevance of N I (value relevance of ROE) as the dependant variable. Each regression model inc I udcs Lhc following independent varia bles: Lhc accrual index (Accrual), the interaction term between the accrual index a nd the shareholder protccLion varia ble (Accruai*Antidirector rights in M odels 1 and 3 a nd Accruar,L egal system in M odels 2 a nd 4), the shareholder protection variable (Antidirector rights in Models 1 and 3 and Legal

M . Hung i Jnw·nal nf Accounting and Rconnmics 30 (200/ j 401- 420 4 17

Table 7 Ordinary least squares regression o f value relevance of earnings and ROF; on accrual, anti director rights, legal system , and tax-book conformity. T define value relevance of earnin gs and ROF: as the return for the D.1V/ and .6.ROE portfolios scaled by the retLLrn for the AdjRet portfolio, as summarized in columns 3 and 5 labeled " Proportion of AdjRet portfolio" of Table 5 (two-tailed p­valucs in paren theses)"

Vai_NI Vai_ROE

Independent vari a hle Pred. Model I Model 2 Model 3 Model 4 s1gn

fnlercepl 0 .8 1( <0.01) 0 .60( <0.01) 0 .96( <0.0 1) 0 .80( <0.01) Accrual - 0 .97( <0.01) - 0.59(0.02) - 0 .93(0.02) - 0.65(0.02) Acauai•Anlidirector + 0 .26(0.05) 0 .23(0.1 0) rights A 111 idirectnr· rights - 0. 11(0.1 4) - 0 .1 4(0.15) A ccrual• Legal system 0.6fi(O.I fi) O.R2(0.14) Legal system - 0 .30(0.39) - 0 .55(0.1 9) Tax-book wnfvrmily - 0.05(0.48) - 0.03(0.67) - 0.07(0.40) - 0.08(0.34)

1V 21 21 21 21 Adj . R2 0.29 0.39 0.13 0.13

" D~/inilivns: Accrual represents the degree to which the accounting system deviates from a cash measure of performance./\ higher number indica tes higher use of accrual accoLLnting. The variable is constructed as reported in Table l. Anlidireclvr ri!Jhls indicates how easy it is for sharehold ers to exercise their voting rights. This varia ble is the antidircctor ri ght index constructed by La Porta et al. (1990). This index ranges from 0 to 5. ll aggregates the fo llowing components o f share ho lder rights: (l) the a bility to vote by ma il, (2) the ability to gain control o r shares during the shareholders' meet.ing, (3) the possibility of cumulative vot.ing for directors, (4) the case of calling an extraordinary shareholder mcet.ing, and (5) the availability of mech anisms of allowing minority shareholders to make legal claims against the directors. Li'gal sysh•m equals l if common law and 0 if code law. Fax-bonk conformity shows the convergence between tax reporting and fi nancia l accoun ti ng. It equals 1 for countries with high tax-hook confonn ity and equals 0 for countries with low conformity. The variable is constructed as reported in Ta ble l.

system in Models 2 and 4), a nd the tax-book conformity index (Tax-book conformity) .14

The inclusion of the interac tion tem1 in the regression models tests the eiTect of accrual accounting on the value relevance of accounting numbers for countries with diiTerent levels of shareholder protection (M adda la, 1992). For example, in Model 2, the coefficient of Accrual represents this effect for countries with weak shar eholder protection. The coefficient of Accruaf*Legal

14 T o reduce the infiuence of outliers and ensure a no1mal distribution of the error terms for such a small sample, I transform the value r elevance measur es and accrua l indexes based on rank transforma tion (Conover and !man, 198 1) and repeat the regression analyses. The results are similar to those discussed below.

4 1R M. Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200 / j 401- 420

system shows the incremental effect when moving from an environment with weak shareholder protection to one with strong sha reholder protection. T he sum of the coefficients of Accrual and Accrual* Legal system represents this cfTcct for country with strong shareholder protection.

Hypothesis I predicts that Accrual will have a negative coefficient. As predicted, the coefficients of Accrual are significantly negative across all the models at the 0.01 level based on a one- tailed test. T he adj usted R2 s range from 0.1 3 to 0.39. This finding indicates that a higher use of accrual accounting lowers the value relevance o f accounting performance measures for countries with weak shareholder protection .

Hypothesis 2 predicts that Accruaf*Antidirector rights and Accruaf*Legal system will have positive coefficients. As predicted, the coefficient of Accrual* Anlidirector rights in M odels 1 and 3 are positively significant at the 0.05 level based on a one-tailed test. T he coefficient o f Acn·ual*Legal system in M odels 2 and 4 are positively significant at the 0.10 level based on a one-tailed test.

A further analysis testing the sum of the coefficients of Accrual and Accrual* Antidirector rights (or Accrual and Accrual* Legal system) indicates that the usc of accrual accounting does not negatively affect the value relevance of accounting information for countries with strong shareholder protection. For example, the two-tailed p-value from testing the sum of coefficients of Accrual and Accrual*Legal system equals 0.85. Finally, although the control variable Tax-book conformity has negati ve signs in all models, as expected, the coefficients are not significant at any conventional level.

6. Conclusions

This paper investigates tl1e effect of accrual accounting on the value relevance of financia l sta tements across countries. I ana lyze 17,743 fi rm-year observations for industrial firms in 21 countries during the period 199 1 1997. I show that the usc of accrua l accounting negatively a fTccts the value relevance of accounting performance measures (earnings and ROE) for countries with >veak shareholder protection (i.e. , countries with low antidirector rights or a code­law system). In addition, strong shareho lder protecti on a ttenuates the negati ve impact a nd increases the value relevance. Finally, accrual accounting does not negatively a fTcct the value relevance for countries with strong shareholder protection (i.e., countries with high antidirector rights or a common-law system).

There arc severa l limitations to interpreting the results. J-iirst, the study only includes industrial companies and nearly all of the sample countries are developed coun tries. T he same results may not apply to non-manufacturing companies or to emerging countries. Second, the study defines the value relevance of financial statements as the ability of accounting numbers to

M. Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200 / j 401- 420 4 19

summarize the information that causes contemporaneous p rice changes. One caveat of this definition is that i t says nothing a bout whether investors usc accounting numbers to set market prices. Third, the study focuses on a composite attribute of national accounting standards (i.e., usc of accrual accounting). Therefore, the findings do not infer how a particular standard aflects the value relevance of earnings.

Overall , the results arc consistent with the belief that shareholder protection improves the eflectiveness of accrual accounting. The findings suggest the importance of considering institutional factors such as shareholder protection when formulating accounting policies related to accruals.

References

Alexander, D., Archer, S., 1995. European Accounting Guide, 2nd Edition. Harcourt Brace & Company, New York, NY.

Alford, A. , Jones, J., Leftwich, R., Zmijewski , M., 1993. The relative in fomuttiveness o f accounting disclosures in different countries. Journal of Accounting R esearch 31 (Suppl.), 183 223.

Ali, A., Hwang, L , 2000. Country-specific factors rela ted to financial reporting and the value relevance of accounting data . Journal of Accounting Research 3~ , 1- 22.

Ball, R. , Brown, P. , 196~. An empirical evaluation of acco unting income numbers. Journal of Accounting Research fi, 159- 178.

Hall, R ., Kothari, S.P. , Ro hin, A., 2000a. The effect of international instillltiona l factors on properties of accounting earnings. Journal of Accounting and Economics 29, I 51.

Ball, R., Robin, A., \\' u, J., 2000b. Incentives versttS standards: Properties of accotmting income in four F.ast Asian countries, and implications fo r acceptance of TAS. \Vorking paper, Un iversity of Rochester .

Barth. M., Clinch, G .. 1998. Revalued financial. tangible, and intangible assets: Associa tions with share prices and non-market-based value estimates. Journal of Accouming Research 36 (Suppl.), 199-233.

Bowen, R. , Burgstahlcr, D. , Daley, L. , 1 9~7. The incremental information content of accrual versus cash flows . The Accoun ting R eview fi1 , 723- 747.

Chang, J., 1998. The decline in value relevance of earnin gs and hook values. \Vorking paper, Harvard University.

Cheng, A .. Liu, C.S .. Schaefer, T .. 1996. Earnings permanence and the incrememal info1mation content of cash flows from operations. Journal of Accounting Research 34, 173-1 81.

Conover, W.J. , !man, R .L. , 198 1. Rank transformation as a bridge between parametric and nonparametric statistics. The American Sta tistici an 35, 124-129.

Coopers & Lybrand, 1993. International Accounting Summaries: A Guide for In terpretation and Comparison. 2nd Edition. Wiley, New York, N Y.

DeAngelo. L., 1988. Management competition. information costs, and corporate governance: The use of accounting performance measures in proxy contests. Journal of Accounting and Economics 10. 3--36.

Dechow, P., 1994. Accounting earnings and cash flows as measures of firm performance: The role of accounting accruals. Journal of Accounting and Economics 18, 3- 42.

Easton. P., Eddey. P. , Harris, T. , 1993. An in vestigation of revalua tions of tangible long-lived assets. Journal of Accounting Research 31 (Suppl.), I 38.

Francis, J. , Schipper, K. , 1999. Have financial statements lost their relevance'! Journal of Accounting Research 37, 319-352.

420 M. Hung i Jnw·nal nf Accounting a11d Rconnmics 30 (200/ j 401- 420

H ealy, P. , 19!:l5. The effect of bonus schemes on accounting decisio ns. Journa l of Accounting and Economics 7, RS-1 07.

Holmstrom, R. , 1979. Moral hazard and observahility. Be ll Journa l of Economics I 0, 74-9 1. J ensen, M ., Meckling, W., 1976. Theory of the iirm: Management behavior, agency costs and

ownership structure. Jo urnal of F inancia l Economics 3, 305 360. J oos. P .. Lang, M .. 1994. The effects of accounting diversity: Evidence from the European Union.

Jo urna l of Accounting Research 32 (Suppl.), 1 4 1 -1 6~ .

Kothari , S.P., Zimmem1an, J. , 1995. Price and return models. Jo urna l of Accoun ting and Economics 20, 155-192.

La Porta, R., Lopez-de-Silanes, F. , Shleifer, /\., Vishny, R., 1996. Law and iinance. NDER "V or king paper 566 1.

La Porta, R., Lopcz-dc-Silancs, F., Shlcifcr. A ., Vishny, R .. 1997. Legal determinants of external finance. T he J o urna l of F inance 52, 1131- 11 50.

La Porta, R. , Lopez-de-Silanes, F ., Shleifer, A., Vishny, R. , 2000. Investor protection and corporate governance. Journa l of Financia l Economics 5R, :l- 37.

Madda la, G .S., 1992. Introduction to Econometrics, 2nd Edition. Prentice-Ha ll, Englewood Cliffs, NJ.

McNicho ls, 1v1. , Wilson, G.P. , 19S!:L Evidence o f earnings management from the provisions for bank debts. J o urna l of Accounting Research 26 (Suppl.). 1- 3 1.

M ueller, G. , vernon, H. , Meek, G., 1994. Accounting: An Interna tional Perspective. I rwin lvfcGraw-Hi ll, Roston, lvfA.