Directory UMM :Data Elmu:jurnal:T:The International Journal Of Accounting:Vol35.Issue4.2000:
The International
Journal of
Accounting
Accounting Practices and the Market Valuation of
Accounting Numbers: Evidence from Indonesia, Korea,
Malaysia, the Philippines, Taiwan, and Thailand
Roger C. Graham* and Raymond D. Kingy
*Oregon State University, Corvallis, OR, USA and yUniversity of Oregon, Eugene, OR, USA
Key Words: International accounting practices; Valuation; Asia; Clean surplus; Conservatism
Abstract: This study examines the relation between stock prices and accounting earnings and
book values in six Asian countries: Indonesia, South Korea, Malaysia, the Philippines, Taiwan,
and Thailand. The analysis is based on a residual earnings model that expresses the value of the
firm in terms of book value and residual income. The model holds for any clean surplus
accounting system. However, for finite time horizons, biased accounting may affect model
estimates. The six countries examined in this study differ in faithfulness to clean surplus
accounting as well as bias (conservatism). The study addresses two questions. First, are there
systematic differences across countries in the value relevance of accounting, and are these
differences related to accounting differences? Second, are there systematic differences in the
incremental and relative information content of book value per share (BVPS) and abnormal
(residual) earnings per share (REPS) across the countries, and are such differences related to
accounting differences? We find differences across the six countries in the explanatory power of
BVPS and REPS for firm values. Explanatory power for Taiwan and Malaysia is relatively low
while that for Korea and the Philippines is relatively high. These differences are generally
consistent with differences in accounting practice; however, since Korean accounting practice is
strongly influenced by tax law, we did not expect the high association for Korea. Second, with
respect to the incremental and relative explanatory power of BVPS and REPS, we find BVPS to
have high explanatory power in the Philippines and Korea but little in Taiwan. In all six countries
REPS has less explanatory power than BVPS in most years. Again, the evidence may be
interpreted as suggesting accounting practice affects valuation (with Korea again as the
exception). Finally, we provide evidence on the sensitivity of the timing of comparisons of stock
prices and accounting values. We find that comparing prices at year-end (even though annual
accounting information has not been released at that time), in general, provides the highest
correlation between market and accounting numbers.
Differences in accounting practices across countries are a major concern to investors,
accounting standard setters, stock exchanges, and financial analysts. The International
Direct all correspondence to: Raymond D. King, Lundquist College of Business, University of Oregon, Eugene,
OR 97403, USA; E-mail: rking@oregon.uoregon.edu
The International Journal of Accounting, Vol. 35, No. 4, pp. 445±470
ISSN: 0020-7063.
All rights of reproduction in any form reserved.
Copyright # 2000 University of Illinois
446
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
Accounting Standards Committee (IASC) and the International Organization of Securities
Commissions (IOSCO) have devoted considerable effort to standardization or harmonization of accounting practices across countries. Investment professionals claim that accounting differences may impede international capital flows (Choi and Levich, 1991). This
study examines the relation between accounting numbers and firm market values in six
Asian countries with diverse accounting practices: Indonesia, Korea, Malaysia, the
Philippines, Taiwan, and Thailand. We focus on the incremental and relative explanatory
power of book value and residual earnings. Because accounting systems differ across the
six countries, we examine whether those differences are related to the valuation usefulness
of accounting measures. Our objective is to provide evidence on the value relevance of
accounting numbers from different accounting systems. Such evidence should inform the
current debate over international accounting standards and practices.
Our analysis follows a model developed by Preinreich (1938), Edwards and Bell (1961),
and Peasnell (1982) and formalized by Ohlson (1991, 1995) and Feltham and Ohlson
(1995) sometimes termed the Residual Earnings (Income) model. The model formally
states a simple concept: firm value is a function of book value and future residual earnings.
A key aspect of the model is that its valuation accuracy does not depend on a particular set
of ``good'' accounting procedures. The only requirement on accounting procedures is clean
surplus accounting, that is, book value of equity changes only with income or loss and net
capital investments and withdrawals (dividends) by owners. In addition, empirical
applications of the model to finite horizons are potentially affected by bias in the
accounting system. Therefore, comparisons across countries with different accounting
practices are one way to investigate the value relevance of different accounting practices.
Across the six countries, accounting systems vary in their faithfulness to clean surplus
accounting and in the extent to which they exhibit bias (conservatism). Hence, it is
possible that accounting values from some of the countries may provide better estimates of
firm value than accounting values from the other countries. Therefore, the usefulness of
accounting for firm valuation may differ across countries as well. On the other hand, the
accounting standards developed in these countries may be partly based on International
Accounting Standards (IAS) or US GAAP. This would tend to make accounting
procedures and their value relevance similar. Saudagaran and Diga (1997) report that of
our six countries, only Korea has not adopted some or all of IAS.
We investigate the value relevance of different accounting practices using an
empirical model that regresses current book value and current residual earnings on
market prices. In contrast, the residual income model is based on expected residual
earnings. Considerable prior research, as discussed in the next section, examines the
contemporaneous relation between accounting and market values. In this study, we
examine that relation for six Asian countries. However, our interest is in the relation
between accounting practices and the value relevance of accounting numbers. We
focus on differences in accounting procedures across the six countries that affect book
value and residual earnings.1 The accounting procedures selected: accounting for
goodwill, asset revaluations, leases, research and development (R&D) expenditures,
and the equity method of accounting for affiliated companies each may be categorized
in terms of faithfulness to clean surplus and extent of conservatism.
We address the implications of these accounting procedures for the value relevance of
accounting information. Philippine firms, for example, record goodwill and revalue assets,
Accounting Practices and the Market Valuation of Accounting Numbers
447
but firms in Taiwan do neither. This means that book values in the Philippines will reflect
market values of assets more closely than in Taiwan. Therefore, we expect the explanatory
power of book value will be greater for Philippine firms than for Taiwanese firms. As
another example, only Indonesian and Malaysian firms capitalize leases and R&D
expenditures and use the equity method for affiliated companies. These are less
conservative accounting practices than alternatives used in other counties.
We find accounting in Korea and Taiwan to be least faithful to clean surplus
accounting. Korea does not capitalize goodwill and asset revaluations are amortized to
equity according to tax law. Taiwan does not capitalize goodwill nor allow asset
revaluations. Korea is also the only country not to use the equity method for affiliated
companies. Thus, the earnings of Korean firms do not include the earnings of affiliated
firms. Philippine firms, however, amortize both goodwill and asset revaluations to income.
Recall that violations of clean surplus accounting occur when income does not reflect
changes in equity value. Thus, violations of clean surplus bias empirical calculations of
residual earnings. Therefore, we expect the explanatory power of residual earnings will be
highest for Philippine firms and least for Korean and Taiwan firms.
Overall, our results show significant differences across countries in the value relevance
of accounting earnings and book values. Explanatory power over all firm-years ranges
from R2 = .17 in Taiwan to R2 = .68 for Korea. The incremental explanatory power of book
value per share (BVPS) and residual earnings per share (REPS) is similarly diverse.
Incremental explanatory power of BVPS over all firm-years ranges from 7.2 percent
(Taiwan) to 65.3 percent (Philippines). For REPS, the incremental explanatory power over
all firm-years ranges from 1.4 percent (Korea) to 13.2 percent (Thailand).
Generally, we find differences in accounting appear to be related to differences in value
relevance. We find that the explanatory power of book value is highest in the Philippines
and lowest in Taiwan. This is consistent with our expectations based on the accounting
differences in the two countries. Indonesia and Malaysia have accounting systems that are
less conservative than other countries. However, we find the incremental explanatory
power of book value does not stand out as high in Indonesia or in Malaysia. This result is
only partly consistent with our expectations. We also expected that the relative explanatory
power of residual earnings would be high in the Philippines and low in Korea and Taiwan,
and the results support this prediction. Our comparisons across countries should be viewed
with caution because the number of years of data available ranges from only 2 years for the
Philippines to 10 years for Malaysia.
The next section of the article briefly reviews related research and this is followed by
the section discussing accounting differences in the six Asian countries. This is followed
by the description of the sample and development the study design. The section presenting
the analysis of our data and reporting the results of our tests follows. A final section
summarizes our findings.
RELATED RESEARCH
Research concerned with the relation of accounting numbers and stock prices covers
decades. In this brief review we summarize recent research with study designs and research
methods similar to ours. We see two principal strains: first, research focused on explaining
448
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
stock prices with accounting book value and earnings; and second, research examining the
incremental explanatory power of book value and earnings in the presence of the other.
Stock Prices Explained by Book Value and Earnings Per Share
Examining a large set of US firms, Bernard (1993) found that book values explain 55
percent of the cross-sectional variation in market prices. When current return on equity
(ROE; ranks) was added to the regression, these two accounting measures explain about
64 percent of the variation in market prices. Bernard (1994) finds that return on common
equity (ROE) is mean reverting over time so that firms with the highest (lowest) current
ROEs tend to have lower (higher) ROEs in later years.
King and Langli (1998) examine the explanatory power of BVPS and earnings per
share (EPS) for three European countries: Germany, Norway, and the UK. They find
significant differences in the valuation power of accounting book value and earnings
across the three countries, and they interpret some of the differences as consistent with
diversity in accounting practices. They also find future earnings realizations as proxies for
expected earnings do not have incremental explanatory power beyond that of current
earnings and book value.
Frankel and Lee (1999) look at the relation between accounting values, earnings
forecasts and market prices across 20 countries (including Korea and Thailand) for 8 years,
1987±1994. Sample sizes for Korea and Thailand are small with 3 to 8 observations per
year (33 total firm-years) for Korea and 1 to 40 observations per year (162 total firm-years)
for Thailand. They find that estimates of value based on the residual earnings model have
incremental explanatory power beyond book value and earnings in explaining market
value in all countries. In addition, they find evidence of superior returns to trading
strategies based on an estimate of value from a residual earnings model.
Joos and Lang (1994) relate book value and earnings to stock prices for France,
Germany, and the UK. Their sample covers 1982 to 1990, and they focus on the effects of
implementing the accounting related directives of the European Union. They find the
explanatory power of book value and earnings together ranges from 20 to 38 percent for
Germany, from 48 to 78 percent for France, and from 14 to 42 percent for the UK. They do
not examine incremental explanatory power. Evidence on changes over time is ambiguous,
probably because the time periods for the sample are relatively short.
Harris et al. (1994) examine the value relevance of accounting numbers for German
firms compared to that for a matched set of US firms for 1982±1991. They find little
difference in overall value relevance (R2) between German and US firms. However,
coefficients (multiples) on book value and on earnings for German firms are greater than
for matched US firms. Further, they find that consolidation increases the value relevance
of accounting numbers, and restatements of earnings to adjust for transitory elements in
German accounting also increases explanatory power.
Incremental Explanatory Power of Book Value and Earnings Per Share
Collins et al. (1997) examine the incremental explanatory power of book value and
earnings across a 41-year time period (1953±1993) for US firms. They find a decline
Accounting Practices and the Market Valuation of Accounting Numbers
449
in the ability of earnings to explain market prices over this period. But the explanatory
power of book values increases such that total explanatory power is actually higher in
more recent periods. Average adjusted R2 for a model regressing BVPS and EPS on
stock price for the first 10 years (1953±1962) was .50 increasing to .69 for their most
recent 10-year period (1984±1993). Collins et al. (1997) investigated possible reasons
for these changes. They find the reduced explanatory power of earnings is explained
by an increase in the incidence of one-time items and reported losses as well as a
decrease in the size of firms in the sample.
King and Langli (1998) examine a 15-year period (1982±1996) for Germany, Norway,
and the UK. They find that for Germany the incremental explanatory power of book value
increases significantly while that for EPS decreases. There is no significant change in their
common information. For Norway, there is no significant change in the incremental
explanatory power of book value or EPS over time. While for the UK, the incremental
explanatory power of book value increases and the incremental explanatory power of EPS
is unchanged over the time period.
Harris et al. (1994) also examine the separate explanatory power of book value and of
earnings using simple regressions with only one variable. They do not report the test
statistics. However, they say that while the explanatory power of EPS in Germany is
approximately equal to that in the US, the explanatory power of book value is much lower.
This contrasts sharply to the King and Langli (1998) results for a longer time period. The
Harris et al. (1994) results are not, however, tests of incremental explanatory power since
the simple regressions use only one variable.
This study extends the evidence summarized above. We examine the value relevance of
accounting numbers for companies in Asian countries. Prior financial reporting research in
English language journals has been limited.
ACCOUNTING DIVERSITY ACROSS THE SIX COUNTRIES
The accounting systems in all six countries have developed relatively recently. The six
accounting systems differ on some dimensions but are similar on others. Two dimensions
that we examine are (1) the model on which the accounting systems are based and (2) the
type of standard setting body. Table 1 shows these characteristics for the six countries. IAS
was the primary basis for accounting standards in Indonesia, Malaysia, and Thailand
(although Thailand has also been influenced by US GAAP). US GAAP, on the other hand,
was the primary basis in the Philippines and Taiwan (although Philippine GAAP is
secondarily based on tax law). Korean accounting standards are unique in that they are
based on Korean tax law that, like tax law in all countries, emphasizes cash realization.
Different accounting models may lead to differences in the value relevance of the resulting
accounting numbers. We have no prior expectations concerning the relative value
relevance of IAS versus US GAAP. It is likely, however, that tax law is more susceptible
to political influence than other accounting bases. To the extent that such political
influence might serve to make accounting less informative, Korean accounting may be
less value relevant because it is based on tax law.
The standard setting bodies in four of the six countries are independent of the
government. In Korea and Taiwan, however, standard setting is not independent. Where
450
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
Table 1. Accounting Standards and Standard Setting in the Six Countries
Country
Indonesia
Korea
Malaysia
Philippines
Taiwan
Thailand
Primary basis for
accounting standards
Independent accounting
standards setting body?
IAS
Tax Law
IAS
US GAAP
US GAAP
IAS
yes
no
yes
yes
no
yes
standard setting is influenced or controlled by government there is greater potential for
political influence in the standards setting process. As where accounting follows tax law,
this may lead to lower value relevance of accounting numbers.
We analyzed the accounting standards and practices for each country using a variety
of sources including Akathaporn (1995), Graham and Wang (1995), and publications
from the AICPA (1989, 1990, 1992), CIFAR (1995), Deloitte Touche Tohmatsu
International (1995a,b, 1996a,b, 1997), Price Waterhouse (1995a,b, 1996a,b,c,d), and
Mathew Bender & Co.(1996). Under the residual earnings model the only crucial
accounting characteristics are unbiased accounting and clean surplus accounting.
Conservative accounting practices are biased since value changes are reflected
asymmetrically, value declines are recognized more quickly than value increases.
The clean surplus relation (CSR) allows book value of equity to change only with
income or loss and net capital investments and withdrawals (dividends) by owners.
CSR is violated if changes in book value can by-pass income. We focus our analysis
on the effects of accounting differences on book value and on residual (abnormal)
earnings, the accounting arguments in the residual earnings model. However, accounting differences affect valuation only when they are violations of unbiased accounting
or clean surplus accounting.
Differences Across Countries Affecting Book Value
Revaluing assets is a violation of CSR if the accompanying credit is taken directly to
equity. Yet asset revaluations bring book value nearer to market value. Immediate write-off
of goodwill violates CSR and usually moves book values farther from market values. In
summary, both recognizing goodwill (consistent with CSR) and revaluing assets (violating
CSR) bring book value nearer to market value. Hence, developing predictions on the
effects of specific accounting treatments on value relevance is not always clear.
Conservative accounting (bias) is expected to generally reduce the value relevance of
both book value and earnings since the essence of conservatism is delay in reflecting
certain events in the accounting records.
Recording goodwill is common practice in Indonesia and the Philippines, and
uncommon but allowed in Malaysia and Thailand. Korea and Taiwan do not allow
goodwill to be recorded. Asset revaluations are common in all countries except Indonesia
and Taiwan where they are allowed but restricted in practice. Table 2 presents a summary
of our analysis of these accounting practices.
Accounting Practices and the Market Valuation of Accounting Numbers
451
Table 2. Goodwill and Asset Revaluations on the Balance Sheet
Country
Goodwill recorded
on balance sheet?
Asset revaluation
on balance sheet?
yes
no
some
yes
no
some
uncommon
common
common
common
uncommon
common
Indonesia
Korea
Malaysia
Philippines
Taiwan
Thailand
Table 3. Other Asset Values on the Balance Sheet
Country
Indonesia
Korea
Malaysia
Philippines
Taiwan
Thailand
Capital leases
on balance sheet?
yes
some
yes
no
some
not until 1996
R&D expenditures
on balance sheet?
yes
yes
yes
no
no
yes
Equity method
used for affiliates?
yes
no
yes
yes
yes
yes
Asset revaluation and goodwill are both recorded only in the Philippines, and neither is
recorded in Taiwan. The other four countries allow one procedure or the other but not
both. We expect the incremental explanatory power of book value to be high in the
Philippines and low in Taiwan relative to the other five countries. However, the effect of
these accounting practices on the incremental explanatory power of book value for the
other countries is ambiguous.
The six countries also differ in other accounting practices, including capitalizing leases,
capitalizing research development costs, and applying the equity method to affiliated
firms. Firms that capitalize and use the equity method are likely to have book values that
are closer to market values than firms that do not.2 Table 3 presents a summary of our
analysis of these accounting practices.
Only Indonesia and Malaysia allow or require all three accounting procedures. The
effect of these accounting differences on incremental explanatory power is ambiguous for
the other countries. However, we expect the explanatory power of book value in Indonesia
and Malaysia to be higher than in Thailand and Taiwan.
Differences Across Countries Affecting Residual Earnings
As explained in the previous section, conservatism is expected to reduce the value
relevance of both book value and residual earnings. In addition, we can make some
predictions about the effects of clean surplus accounting for goodwill and asset revaluations on the value relevance of book value. However, the effects of these accounting
practices on the value relevance of residual earnings are less clear.
The Philippines is the most faithful to clean surplus accounting as both goodwill
and asset revaluations are amortized to income over their useful lives. The other
452
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
Table 4. Goodwill and Asset Revaluation Amortization
Country
Indonesia
Korea
Malaysia
Philippines
Taiwan
Thailand
Goodwill
amortized to?
income
±
±
income
±
±
Amortization
period
useful life
±
immediate write-off
no more than 40 years
±
immediate write-off
Revaluations
amortized to?
Amortization
period
±
shareholder equity
shareholder equity
income
±
shareholder equity
±
per tax law
useful life
useful life
±
useful life
countries (1) do not record goodwill (Korea and Taiwan), (2) do not record asset
revaluations (Indonesia and Taiwan), (3) immediately write-off goodwill to equity
(Malaysia and Thailand), or (4) amortize asset revaluation increments to equity (Korea,
Malaysia, and Thailand). Korea and Taiwan are least faithful to clean surplus
accounting. Korea does not capitalize goodwill and asset revaluations are amortized
to equity according to schedules mandated by tax law. Taiwan does not capitalize
goodwill nor allow asset revaluations. Korea is also the only country that does not use
the equity method for affiliated companies; therefore the earnings of Korean firms do
not include the earnings of affiliated firms. We expect the relative explanatory power
of residual earnings to be high in the Philippines and low in Korea and Taiwan. The
effect on the explanatory power of residual earnings for the other countries is
ambiguous. Table 4 presents a summary of the amortization practices.
Summary of the Research Questions
Our examination of accounting practices reveals some systematic differences across the
six countries. While the differences appear to be substantial, it is an empirical question
whether they result in meaningful violations of the CSR or in significant accounting bias
(conservatism), the important factors for the residual earnings model. Accounting bias will
likely reduce the explanatory power of both book value and earnings. Furthermore,
violations of the CSR may either increase or decrease the explanatory power of book value
depending on whether the violation moves book value toward or away from market
values. Because countries differ on both dimensions, ex ante hypotheses on which effect
will dominate are problematic. Even so, we expect that bias and CSR violations will affect
the value relevance of accounting numbers in systematic ways. Particularly because
Philippine firms record both goodwill and asset revaluations and Taiwan firms do neither,
we expect the value relevance of book value to be greatest in the Philippines and least in
Taiwan. Because both Indonesian and Malaysian firms capitalize leases and R&D
expenditures and use the equity method we expect the value relevance of their book
values to exceed the value relevance of book value in Thailand and Taiwan. Because
Philippine firms amortize both goodwill and asset revaluations to income, we expect the
value relevance of residual earnings to be high in the Philippines.
In addition, we investigate changes in the value relevance of accounting numbers over
time. Our sample contains 3,655 firm-years across six countries. We have sufficient data
for only 2 years of yearly regressions for the Philippines but 10 years for Malaysia. We
Accounting Practices and the Market Valuation of Accounting Numbers
453
Table 5. Sample Countries and Firm-Years
Sample selection
Indonesia
Korea
Malaysia
Philippines
Taiwan
Thailand
Sample size
Firm-years
338
902
1,311
139
369
596
3,655
trace the total explanatory power of accounting earnings and book value and the
incremental explanatory power of each earnings and book value in the presence of the
other across time for each country.
SAMPLE AND STUDY DESIGN
Our sample covers publicly traded firms in Indonesia, Korea, Malaysia, the Philippines,
Taiwan, and Thailand across the period from 1987 to 1996. The stock prices and
accounting data for this study are from the Worldscope Global Researcher. The sample
selection criteria are:
1.
2.
3.
4.
5.
Accounting data is from consolidated financial statements.
Financial firms are excluded (insurance, banks, and other miscellaneous financial
firms). Accounting practices for these firms are so distinct that their valuation
parameters are likely to be substantially different from those for industrial firms.
Firms with negative book values are deleted. These firms are likely to be in
financial distress and may be interesting in their own right. However, the focus of
this study is the across country differences in value relevance of accounting
numbers derived under different accounting practices. Hence, restricting our
sample to firms with positive book values will allow us to focus on firms where
differences are mostly likely to reflect accounting differences.
Twelve firms with EPS greater than their BVPS are deleted since data on those firms
is likely to contain errors. These firms constituted less than 1 percent of the sample.
Twenty firm-years with excessive statistical influence in our regressions were
deleted. The firm-years showed undue influence by the diagnostics and cutoff rules
described in Belsley et al. (1980).
These restrictions on the sample will have several effects. First, the model will appear
to ``fit'' better than it would fit unrestricted data. That is, the explanatory power of book
value and of residual earnings information in the sample is likely to be greater than for an
unrestricted sample. Second, the samples across the six countries will be more homogeneous and the effects of different business cycles in the six countries will be reduced.
This should allow a better focus on the effects of accounting differences. Table 5 shows the
countries and number of firm-years.
454
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
Table 6. Descriptive Statistics on Variables for Six Southeast Asian Countries
Variable
N
Mean
Standard deviation
Panel A: Indonesia
Price
338
BVPS
338
EPS
338
REPS
338
ROE
340
2,813
1,668
242
ÿ 75
0.15
2,261
1,175
305
272
0.09
Panel B: Korea
Price
902
BVPS
902
EPS
902
REPS
902
ROE
902
25,629
21,019
1,353
ÿ 372
0.07
Panel C: Malaysia
Price
1,311
BVPS
1,311
EPS
1,311
REPS
1,311
ROE
1,311
5th Percentile
Median
95th Percentile
556
361
43
ÿ 474
0.03
2,236
1,374
154
ÿ 52
0.14
6,500
4,394
765
180
0.30
34,972
31,375
4,324
3,334
0.14
8,002
6,282
ÿ 979
ÿ 3,291
ÿ 0.09
18,100
14,104
786
ÿ 398
0.06
63,921
50,225
5,007
2,389
0.26
5.16
1.71
0.18
0.06
0.13
5.99
1.11
0.21
0.19
0.15
0.94
0.53
ÿ 0.03
ÿ 0.17
ÿ 0.04
3.84
1.44
0.15
0.04
0.12
13.96
3.72
0.56
0.37
0.38
Panel D: Philippines
Price
139
BVPS
139
EPS
139
REPS
139
ROE
139
19.06
7.91
0.97
ÿ 0.14
0.16
31.65
12.00
1.71
1.32
0.14
0.17
0.17
ÿ 0.08
ÿ 2.75
ÿ 0.04
6.37
2.54
0.37
ÿ 0.01
0.14
81.37
38.68
4.94
1.30
0.40
Panel E: Taiwan
Price
369
BVPS
369
EPS
369
REPS
369
ROE
369
30.55
12.43
1.20
ÿ 0.18
0.10
13.21
2.81
1.25
1.25
0.11
14.95
7.80
ÿ 0.79
ÿ 2.17
ÿ 0.06
27.89
12.19
1.16
ÿ 0.21
0.09
53.83
17.51
3.16
1.84
0.26
Panel F: Thailand
Price
596
BVPS
596
EPS
596
REPS
596
ROE
596
95.79
41.35
5.93
ÿ 0.23
0.12
102.39
35.05
8.32
5.10
0.14
18.25
11.05
ÿ 1.14
ÿ 6.41
ÿ 0.06
59.59
31.09
3.59
ÿ 0.78
0.13
330.00
117.83
21.24
7.48
0.34
Price = Stock price at the end of year t. BVPS = Book value of shareholders' equity at the end of year t. EPS = Earnings per share
for year t. REPS = Residual (abnormal) earnings per share = EPSt ÿ r*((BVPSt + BVPSt ÿ 1)/2) where r is the country average
lending rate in year t taken from the International Financial Statistics Yearbook. Price, book value, EPS, and REPS amounts are in
nominal local currency. ROE = Return on equity = EPSt /((BVt + BVt ÿ 1)/2).
Table 6 provides descriptive statistics for the sample. Per share values are in nominal
currency of the countries, so comparisons are difficult. However, we can compare ROE
(the ratio of EPS to average book value) across countries and the differences are
substantial. Median ROE ranges from 6 percent (over 8 years) in Korea to 14 percent
Accounting Practices and the Market Valuation of Accounting Numbers
455
Table 7. Correlation Statistics on Variables for Six Asian Countries ( p values in parentheses)
Spearman
Variable
N
Price
EPS
Panel A: Indonesia
Price
338
.6164 (.0001)
EPS
338 .5741 (.0001)
BV
338 .4777 (.0001) .6418 (.0001)
ROE
338 .3423 (.0001) .6118 (.0001)
REPS
338 .2498 (.0040) .6262 (.0001)
Pearson
Panel B: Korea
Price
902
EPS
902
BV
902
ROE
902
REPS
902
.6894
.8177
.1667
.3395
.5328 (.0001)
(.0001)
(.0001) .7438 (.0001)
(.0001) .4769 (.0001)
(.0001) .8529 (.0001)
Panel C: Malaysia
Price
1,311
.6224 (.0001)
EPS
1,311 .4171 (.0001)
BV
1,311 .5031 (.0001) .5552 (.0001)
ROE
1,311 .1093 (.0001) .6042 (.0001)
REPS
1,311 .2564 (.0001) .9093 (.0001)
Panel D: Philippines
PRICE
139
EPS
139 .7593
BV
139 .7906
ROE
139 .0169
REPS
139 .0727
Panel E: Taiwan
Price
369
EPS
369
BV
369
ROE
369
REPS
369
.8133 (.0001)
(.0001)
(.0001) .7790 (.0001)
(.8439) .2416 (.0042)
(.3949) .4301 (.0001)
.5043 (.0001)
.3618 (.0001)
.2639 (.0001) .2422 (.0001)
.3048 (.0001) .9353 (.0001)
.3116 (.0001) .9609 (.0001)
Panel F: Thailand
Price
596
.7308 (.0001)
EPS
596 .6284 (.0001)
BV
596 .5145 (.0013) .8121 (.0001)
ROE
596 .4415 (.0001) .5912 (.0001)
REPS
596 .5166 (.0001) .8068 (.0001)
BV
.5685 (.0001)
.6915 (.0001)
ÿ .0130 (.8119)
ÿ .1012 (.0631)
.6021 (.0001)
.5552 (.0001)
.0865 (.0093)
.2766 (.0001)
.5572 (.0001)
.5419 (.0001)
.0117 (.6730)
.2102 (.0001)
.8601 (.0001)
.7726 (.0001)
ÿ .1081 (.2052)
ÿ .1866 (.0278)
.3056 (.0001)
.2487 (.0001)
.0276 (.5972)
ÿ .0151 (.7731)
.5639 (.0001)
.7144 (.0001)
.2494 (.0001)
.3405 (.0001)
ROE
.2255 (.0001)
.5066 (.0001)
ÿ .1406 (.0097)
REPS
.0662
.2846
ÿ .3028
.8904
(.2252)
(.0001)
(.0001)
(.0001)
.1727
.6624
ÿ .0681
.7442
(.0012)
(.0001)
(.0408)
(.9190)
.4496
.8697
.1774
.9252
(.0001)
(.0001)
(.0001
(.0001)
.7768 (.0001)
.3095 (.0001)
.8285 (.0001)
.1374 (.0001)
.5963 (.0001)
.3272 (.0001)
.7678 (.0001)
.0160 (.5619)
.7218 (.0001)
.1725 (.0423)
.4577 (.0001)
ÿ .0654 (.4441)
.1177 (.1676)
.3776 (.0001)
ÿ .0715 (.4027)
.8089 (.0001)
.4931 (.0001)
.4399 (.0001)
.9499 (.0001)
ÿ .0010 (.9853)
.4431
.9291
ÿ .0390
.9804
(.0001)
(.0001)
(.4548)
(.0001)
.9909 (.0001)
.6144 (.0001)
.8625 (.0001)
.3443 (.0001)
.5523 (.0001)
.7757 (.0001)
.2411 (.0001)
.9362 (.0001)
.7237 (.0001)
Price = Stock price at the end of year t. EPS = Earnings per share for year t. BVPS = Book value of shareholders' equity at
the end of year t. ROE = Return on equity = EPSt / ((BVPSt + BVPSt ÿ 1)/2). REPS = Residual (abnormal) earnings per
share = EPSt ÿ r*((BVPSt + BVPSt ÿ 1)/2) where r is the country average lending rate in year t taken from the International
Financial Statistics Yearbook.
(over 5 and 2 years) in Indonesia and the Philippines. For comparison, King and Langli
(1998) find ROE over the 1980s and 1990s to be about 6 percent in Germany, 10 percent
in Norway, and 13 percent in the UK. In the US, this measure has averaged around 13
456
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
percent over the last 20 years. Prior research has speculated that such differences across
countries may reflect differences in conservatism of accounting methods. As noted above,
for example, Taiwan does not record either goodwill asset revaluations (most conservative)
while both are recognized in the Philippines (least conservative).
Table 7 reports the pair-wise correlation between stock price and accounting variables
for all countries. For all countries except Korea the rank (Spearman) correlations are
greater than the product±moment (Pearson) correlations. However, the patterns and
significance of the parametric (Spearman) and non-parametric (Pearson) correlations are
similar. Stock prices are strongly correlated with BVPS and with EPS for all countries. The
pair-wise correlations between price and EPS are approximately the same as between price
and BVPS. The correlation (Pearson) between BVPS and EPS is high in Thailand (.81),
the Philippines (.78), and Korea (.74), and relatively low for Taiwan (.24). If EPS is used
as a proxy for residual earnings, the high correlation between BVPS and EPS (they are
related by size) may make it difficult to partition value relevance between book value and
earnings. However, the pair-wise correlations between BVPS and REPS are far smaller,
significant in some cases, insignificant in others, and sometimes negative. This is
anticipated since there is no reason to expect residual (unexpected) per share earnings
to be related to book value. One important reason for using REPS in the empirical analysis
is to avoid the high correlation between BVPS and EPS.
TESTS AND ANALYSIS
Our analysis is based on contemporaneous cross-sectional regressions of accounting book
values and residual earnings on stock prices (dependent variable). We analyze both the
relative and the incremental explanatory power of book value and residual earnings using an
approach applied previously in accounting by Biddle et al. (1995) and Collins et al. (1997).
Empirical specification of the residual earnings model requires estimates of book value,
residual earnings, and the horizon for residual earnings. For residual earnings estimated to
terminate at time T, the model would be:
Priceit a0 a1 BVPSit a2 REit1 a3 REit2 a4 REit3 . . . ak REiT eit
1
where Priceit is the price per share of firm i at the end of period t, BVPSit is the book value
per share of firm i at the end of period t, and REit is the residual earnings per share of firm i
for year t + k.
The coefficient a1 would have an expected value of 1.0 while the coefficients a2 to a4
would have expected values of (1 + r) ÿ t. Finally, the expected value of coefficient ak
would be (1/r)*(1 + r) ÿ T.3 Residual earnings horizons will differ cross-sectionally; therefore, parsimonious cross-sectional representations of Equation (1) will have only a few
terms. For example Frankel and Lee (1999) use T = 2 because analysts' forecasts used to
predict future residual earnings were only available for 2 years.
Our first tests are concerned with the incremental explanatory power of book value and
residual earnings. As in Collins et al. (1997) we compare the results of three regression
equations to address the question of relative and incremental explanatory power. Equation
(2) below provides the most parsimonious empirical specification of the residual earnings
Accounting Practices and the Market Valuation of Accounting Numbers
457
model on a per share basis (the horizon is only one period). Current period residual
earnings is the proxy for future expected residual earnings. Residual earnings are estimated
by subtracting an estimate of normal (expected) earnings from reported earnings. Expected
earnings is the product of the estimated rate of return (r) and book value. Like Frankel and
Lee (1999), we derive the estimated rate of return from interest rates in the International
Financial Statistics Yearbook published by the International Monetary Fund (1997).
Frankel and Lee (1999), however, are able to calculate a risk-adjusted return by adding
a risk premium to long-term government bond rates Government bond rates are not
available for four of the countries; therefore, we use commercial lending rates. Conceptually, this rate is the sum of a riskless rate and the average commercial lending risk
premium. In the residual earnings model, book value and firm value are taken at time t
while future abnormal earnings are for periods after time t. In our empirical analysis
residual earnings (REt) are for the period ending at time t. Hence, as in Bernard (1994) and
Collins et al. (1997), current earnings is a proxy for expected future earnings.
Priceit b0 b1 BVPSit b2 REPSit eit
2
where Priceit is the stock price per share of firm i at the end of year t, BVPSit is the book
value of shareholders' equity of firm i at the end of year t, REPSit is the residual earnings per
share, which is equal to EPSit ÿ r*(BVPSt ÿ 1)4 (proxy for expected REPS in period t + 1),
EPSit is the earnings per share of firm i for year t, and r is the country's average commercial
lending rate in year t taken from the International Financial Statistics Yearbook.
Book values and earnings are, of course, unobservable until some weeks after the end of
the fiscal year. This raises the question of the timing of the market value measure to be
associated with the accounting variables. As discussed by Barth et al. (1996), choice of
contemporaneous versus lagged market values is a trade-off. The advantage to using a
lagged market price is that it may reasonably reflect the accounting results since sufficient
time has passed for these results to be public information. However, lagged market values
will include effects of information and events occurring after the end of the fiscal year.
Collins et al. (1997), examining associations between market and accounting numbers for
US firms, take prices 3 months after the end of the fiscal period. In cross-country studies,
however, this is problematic since the time lag between fiscal year-ends and report dates can
vary widely. For this reason, our tests examine the relation between accounting numbers
(book value and residual EPS for a fiscal year) and stock prices at the end of the fiscal year.
Later, we analyze the sensitivity of our results using stock prices lagged 0 to 6 months
following the end of the fiscal year.
Equation (2) expresses price as a function of book value and residual earnings.
Examining the relative and incremental explanatory power of book value and of residual
earnings requires two additional equations expressing price as a function of book value
alone, Equation (3), and residual earnings alone, Equation (4).5
Priceit c0 c1 BVPSit eit
3
Priceit d0 d1 REPSit eit
4
458
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
Following Theil (1971), we define the incremental explanatory power of the
book value and residual earnings variables in terms of differences in the coefficient
of determination (R2). These differences are sometimes called the semi-partial
coefficient of determination (Cohen and Cohen, 1975, pp. 79±84). They are a
measure of the incremental explanatory power of one variable given the remaining
independent variables.
2
, respectively.
Define the R2 statistics from Equations (2), (3), and (4) as Rb2, Rr2, and Rb,r
Then the incremental explanatory power is defined as:
2
R2bjr = Rb,r
ÿ R2r
R2rjb = R2b,r ÿ R2b
2
ÿ R2bjr ÿ R2rjb
R2com = Rb,r
The incremental explanatory power of book value is
the total explanatory power of book value and residual
earnings less the explanatory power of residual earnings alone.
The incremental explanatory power of residual earnings is
the total explanatory power of book value and residual
earnings less the explanatory power of book value alone.
The explanatory power common to book value and
residual earnings is the total explanatory power of book
value and residual earnings less the incremental explanatory power of book value and the incremental explanatory
power of residual earnings.6
We can also assess the relative explanatory power of book value and residual earnings
by comparing the conditional (incremental) power as shown above (Biddle et al., 1995).7
That is, we can also address the question of whether book values or residual earnings have
greater explanatory power for each country.
Explanatory Power of BVPS and REPS Across the Six Countries
Table 8 reports summaries of regressions (2), (3), and (4) as well as incremental R2 for
each year with 30 or more observations and for all years together for the six countries.
First, we focus on the coefficients and the significance of regressions (2), (3), and (4), and
then we analyze the relative and incremental explanatory power.8
Coefficients on BVPS are positive for all countries. They are significant overall and for
most years. BVPS coefficients are greater than 1.0 for Malaysia (2.75), the Philippines
(2.17), Taiwan (1.26), and Thailand (1.41) in regression (2). BVPS coefficients are less
than 1.0 for Indonesia (.82) and Korea (.89). Coefficients on REPS are positive and
significant for all countries except the Philippines. Coefficients on REPS for regression (4)
range from 1.26 for Malaysia to 10.90 for Thailand.9
We find significant differences in the value relevance of accounting across countries.
The explanatory power of book value and residual earnings is quite high for Korea and the
Philippines, near to what is found for Anglo-American markets. However, the explanatory
power for Taiwan is well below that found for most other countries. Table 9 presents a
Priceit b0 b1 BVPSt b2 REPS eit
R2b;r
Priceit c0 c1 BVPSt eit
R2b
Priceit d0 d1 REPS eit
R2r
where Priceit is the price per share of firm i at time t, BVPSt is the book value per share of firm i at the end of period t, and REPSt is the residual earnings per
share of firm i for year t.
R2bjr = R2b,r ÿ R2r
The incremental explanatory power of book value is the total explanatory power of book value and residual earnings less the
explanatory power of residual earnings alone.
The incremental explanatory power of residual earnings is the total explanatory power of book value and residual earnings less the
explanatory power of book value alone.
The explanatory power common to book value and residual earnings is the total explanatory power of book value and residual earnings
less the incremental explanatory power of book value and the incremental explanatory power of residual earnings alone.
R2rjb = R2b,r ÿ R2b
R2com = R2b,r ÿ R2bjr ÿ R2rjb
Year
N
Panel A: Indonesia
All
338
1991
46
1992
61
1993
63
1994
70
1995
80
Meana
64
b0
1,400
698
596
1,864
1,495
1,526
1,236
(7.83)
(2.12)
(1.82)
(6.19)
(3.41)
(3.61)
b1BV
0.96
1.62
1.75
1.04
0.79
0.63
1.17
(10.90)
(7.38)
(8.80)
(6.54)
(3.98)
(3.32)
b2REPS
2.49
6.08
5.66
0.86
3.07
3.41
3.82
(6.57)
(3.82)
(8.82)
(1.78)
(3.53)
(3.99)
2
Rb,r
.308
.559
.564
.419
.332
.266
.434
c1BV
0.90
1.16
1.32
0.99
0.90
0.66
1.01
(9.69)
(5.52)
(5.43)
(6.22)
(4.22)
(3.13)
Rb2
.219
.409
.358
.388
.208
.114
.295
d1REPS
2.07
ÿ 0.34
2.46
0.34
3.59
3.50
1.91
(4.73)
( ÿ 0.17)
(1.80)
(0.55)
(3.78)
(3.86)
Rr2
2
Rbjr
.062
.001
.052
.005
.173
.161
.078
.246
.558
.542
.414
.159
.105
.356
Rr2jb
.089
.150
.236
.031
.124
.152
.139
Accounting Practices and the Market Valuation of Accounting Numbers
Table 8. Incremental and Relative Information Content of Book Values and Residual Earnings: Regressions of Book Value and Residual Earnings on Price by
Year (t-statistics in parentheses)
2
Rcom
ÿ .027
ÿ .149
ÿ .184
ÿ .026
.049
.009
ÿ .060
459
(continued)
460
Table 8. (Continued)
Year
N
b0
b1BV
b2REPS
2
Rb,r
c1BV
Rb2
d1REPS
Rr2
2
Rbjr
Rr2jb
2
Rcom
0.87
0.21
0.31
0.24
0.27
0.50
0.74
0.81
0.92
0.51
(40.07)
(2.43)
(4.26)
(4.49)
(5.08)
(7.13)
(9.49)
(18.06)
(28.72)
1.29
1.94
1.44
0.89
0.73
1.99
3.26
1.61
0.32
1.52
(6.27)
(2.41)
(2.53)
(2.11)
(2.09)
(5.53)
(4.60)
(3.48)
(1.02)
.683
.182
.241
.240
.261
.584
.820
.816
.843
.498
0.91
0.18
0.22
0.18
0.26
0.67
1.05
0.91
0.91
0.55
(42.61)
(1.96)
(3.32)
(3.87)
(4.75)
(9.42)
(23.44)
(26.96)
(30.87)
.669
.077
.157
.187
.220
.461
.794
.803
.842
.442
3.56
1.62
0.25
ÿ 0.10
0.50
3.13
8.91
7.27
ÿ 3.22
2.92
(10.83)
(1.93)
(0.45)
( ÿ 0.26)
(1.25)
(7.97)
(18.64)
(12.73)
( ÿ 4.81)
.115
.075
.000
.001
.019
.379
.709
.477
.115
.222
.568
.107
.241
.239
.242
.205
.111
.339
.728
.276
.014
.105
.084
.053
.041
.123
.026
.013
.001
.056
.101
ÿ .030
ÿ .084
ÿ .052
ÿ .022
.256
.683
.464
.114
.166
Panel C: Malaysia
All
1,311
1987
42
1988
50
1989
59
1990
67
1991
102
1992
169
1993
172
1994
183
1995
258
1996
201
Meana
130
0.51 (2.00)
0.66 (1.77)
0.59 (1.89)
1.76 (4.54)
1.74 (4.27)
0.92 (3.69)
0.16 (0.66)
ÿ 1.11 (1.30)
0.86 (1.04)
0.80 (1.33)
4.21 (4.02)
1.06
2.53
1.38
1.24
0.75
0.85
0.99
1.72
4.31
3.03
2.38
1.37
1.80
(19.54)
(4.08)
(4.88)
(2.65)
(2.75)
(6.35)
(12.57)
(9.70)
(7.84)
(8.77)
(5.21)
5.07
2.57
2.33
4.28
6.00
9.33
8.30
5.09
2.41
3.25
7.64
5.12
(6.55)
(2.59)
(3.00)
(3.78)
(4.71)
(9.96)
(8.36)
(1.62)
(0.91)
(1.90)
(3.33)
.277
.300
.391
.322
.381
.637
.532
.371
.264
.263
.157
.362
2.71
0.82
1.18
0.98
1.24
1.32
1.45
4.37
3.07
2.48
1.97
1.89
(21.06)
(2.97)
(4.20)
(3.16)
(3.59)
(6.13)
(9.17)
(9.80)
(7.98)
(9.30)
(4.95)
.253
.180
.269
.149
.166
.273
.335
.361
.260
.253
.110
.236
8.25
ÿ 0.04
1.90
4.89
6.92
10.58
5.32
7.30
4.58
0.64
10.73
5.28
(9.60)
( ÿ 0.04)
(2.08)
(4.21)
(5.37)
(9.78)
(3.96)
(1.88)
(1.50)
(3.30)
(5.04)
.067
.000
.083
.237
.307
.489
.086
.020
.012
.041
.113
.139
.210
.300
.308
.085
.074
.148
.446
.351
.252
.222
.044
.223
.024
.120
.122
.173
.215
.364
.197
.010
.004
.010
.047
.126
.043
ÿ .120
ÿ .039
.064
.092
.125
ÿ .111
.010
.008
.031
.066
.013
Vol. 35, No. 4, 2000
7,746 (9.48)
15,983 (10.36)
19,043 (16.22)
13,791 (17.12)
9,826 (9.51)
9,383 (5.75)
11,318 (4.75)
13,979 (8.06)
3,266 (2.25)
12,074
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Panel B: Korea
All
901
1988
48
1989
61
1990
67
1991
82
1992
106
1993
145
1994
180
1995
182
Meana
109
2.44 (1.32)
0.45 (0.15)
ÿ 0.23 ( ÿ 0.16)
0.11
2.20 (16.75)
2.92 (14.63)
2.58 (22.99)
2.75
5.49 (4.60)
13.76 (6.67)
18.72 (10.09)
16.24
.680
.885
.921
.903
2.08 (1.30)
2.41 (8.27)
1.89 (10.88)
2.15
.625
.702
.716
.709
.169
.135
.272
.300
.236
1.24
2.19
1.52
1.41
1.71
(5.24)
(2.40)
(4.36)
(6.47)
.070
.126
.164
.191
.160
3.30
1.30
3.03
2.35
2.23
.397
.848
.727
.300
.512
.369
.551
1.50
2.06
1.93
1.33
1.53
1.38
1.65
(14.63)
(12.24)
(11.15)
(4.55)
(8.39)
(7.62)
.265
.798
.635
.139
.300
.251
.425
10.37
6.67
9.42
14.74
11.56
9.33
10.34
Panel E: Taiwan
All
369
1993
42
1994
98
1995
179
Meana
106
15.57
9.30
15.39
9.35
11.35
(5.43)
(0.84)
(3.82)
(3.43)
1.26
2.20
1.68
1.35
1.74
(5.64)
(2.38)
(5.08)
(6.63)
3.34
1.32
3.63
2.19
2.38
Panel F: Thailand
All
596
1991
40
1992
74
1993
130
1994
166
1995
175
Meana
117
51.28
18.63
27.59
102.78
47.66
38.21
46.97
(9.71)
(2.25)
(3.45)
(6.72)
(5.60)
(2.80)
1.28
1.81
1.68
0.57
1.10
1.12
1.26
(11.29)
(11.06)
(10.53)
(1.91)
(6.80)
(6.64)
7.76 (11.38)
2.84 (3.49)
5.55 (4.92)
12.64 (5.39)
9.18 (8.39)
7.14 (5.65)
7.47
a
(6.63)
(0.65)
(3.79)
(5.21)
.005
.010
.010
.010
.675
.875
.911
.893
.055
.183
.205
.194
ÿ .050
ÿ .173
ÿ .195
ÿ .184
(6.28)
(0.60)
(2.84)
(5.02)
.097
.009
.077
.125
.070
.072
.126
.195
.175
.166
.099
.009
.108
.109
.076
ÿ .002
.000
ÿ .031
.016
ÿ .006
(14.70)
(4.48)
(5.55)
(7.05)
(9.88)
(6.91)
.267
.345
.300
.280
.373
.216
.303
.130
.503
.427
.020
.139
.153
.248
.132
.050
.092
.161
.212
.118
.126
.135
.295
.208
.119
.161
.098
.177
1.74 (0.85)
2.27 (0.41)
ÿ 3.37 ( ÿ 0.68)
ÿ 0.55
Accounting Practices and the Market Valuation of Accounting
Journal of
Accounting
Accounting Practices and the Market Valuation of
Accounting Numbers: Evidence from Indonesia, Korea,
Malaysia, the Philippines, Taiwan, and Thailand
Roger C. Graham* and Raymond D. Kingy
*Oregon State University, Corvallis, OR, USA and yUniversity of Oregon, Eugene, OR, USA
Key Words: International accounting practices; Valuation; Asia; Clean surplus; Conservatism
Abstract: This study examines the relation between stock prices and accounting earnings and
book values in six Asian countries: Indonesia, South Korea, Malaysia, the Philippines, Taiwan,
and Thailand. The analysis is based on a residual earnings model that expresses the value of the
firm in terms of book value and residual income. The model holds for any clean surplus
accounting system. However, for finite time horizons, biased accounting may affect model
estimates. The six countries examined in this study differ in faithfulness to clean surplus
accounting as well as bias (conservatism). The study addresses two questions. First, are there
systematic differences across countries in the value relevance of accounting, and are these
differences related to accounting differences? Second, are there systematic differences in the
incremental and relative information content of book value per share (BVPS) and abnormal
(residual) earnings per share (REPS) across the countries, and are such differences related to
accounting differences? We find differences across the six countries in the explanatory power of
BVPS and REPS for firm values. Explanatory power for Taiwan and Malaysia is relatively low
while that for Korea and the Philippines is relatively high. These differences are generally
consistent with differences in accounting practice; however, since Korean accounting practice is
strongly influenced by tax law, we did not expect the high association for Korea. Second, with
respect to the incremental and relative explanatory power of BVPS and REPS, we find BVPS to
have high explanatory power in the Philippines and Korea but little in Taiwan. In all six countries
REPS has less explanatory power than BVPS in most years. Again, the evidence may be
interpreted as suggesting accounting practice affects valuation (with Korea again as the
exception). Finally, we provide evidence on the sensitivity of the timing of comparisons of stock
prices and accounting values. We find that comparing prices at year-end (even though annual
accounting information has not been released at that time), in general, provides the highest
correlation between market and accounting numbers.
Differences in accounting practices across countries are a major concern to investors,
accounting standard setters, stock exchanges, and financial analysts. The International
Direct all correspondence to: Raymond D. King, Lundquist College of Business, University of Oregon, Eugene,
OR 97403, USA; E-mail: rking@oregon.uoregon.edu
The International Journal of Accounting, Vol. 35, No. 4, pp. 445±470
ISSN: 0020-7063.
All rights of reproduction in any form reserved.
Copyright # 2000 University of Illinois
446
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
Accounting Standards Committee (IASC) and the International Organization of Securities
Commissions (IOSCO) have devoted considerable effort to standardization or harmonization of accounting practices across countries. Investment professionals claim that accounting differences may impede international capital flows (Choi and Levich, 1991). This
study examines the relation between accounting numbers and firm market values in six
Asian countries with diverse accounting practices: Indonesia, Korea, Malaysia, the
Philippines, Taiwan, and Thailand. We focus on the incremental and relative explanatory
power of book value and residual earnings. Because accounting systems differ across the
six countries, we examine whether those differences are related to the valuation usefulness
of accounting measures. Our objective is to provide evidence on the value relevance of
accounting numbers from different accounting systems. Such evidence should inform the
current debate over international accounting standards and practices.
Our analysis follows a model developed by Preinreich (1938), Edwards and Bell (1961),
and Peasnell (1982) and formalized by Ohlson (1991, 1995) and Feltham and Ohlson
(1995) sometimes termed the Residual Earnings (Income) model. The model formally
states a simple concept: firm value is a function of book value and future residual earnings.
A key aspect of the model is that its valuation accuracy does not depend on a particular set
of ``good'' accounting procedures. The only requirement on accounting procedures is clean
surplus accounting, that is, book value of equity changes only with income or loss and net
capital investments and withdrawals (dividends) by owners. In addition, empirical
applications of the model to finite horizons are potentially affected by bias in the
accounting system. Therefore, comparisons across countries with different accounting
practices are one way to investigate the value relevance of different accounting practices.
Across the six countries, accounting systems vary in their faithfulness to clean surplus
accounting and in the extent to which they exhibit bias (conservatism). Hence, it is
possible that accounting values from some of the countries may provide better estimates of
firm value than accounting values from the other countries. Therefore, the usefulness of
accounting for firm valuation may differ across countries as well. On the other hand, the
accounting standards developed in these countries may be partly based on International
Accounting Standards (IAS) or US GAAP. This would tend to make accounting
procedures and their value relevance similar. Saudagaran and Diga (1997) report that of
our six countries, only Korea has not adopted some or all of IAS.
We investigate the value relevance of different accounting practices using an
empirical model that regresses current book value and current residual earnings on
market prices. In contrast, the residual income model is based on expected residual
earnings. Considerable prior research, as discussed in the next section, examines the
contemporaneous relation between accounting and market values. In this study, we
examine that relation for six Asian countries. However, our interest is in the relation
between accounting practices and the value relevance of accounting numbers. We
focus on differences in accounting procedures across the six countries that affect book
value and residual earnings.1 The accounting procedures selected: accounting for
goodwill, asset revaluations, leases, research and development (R&D) expenditures,
and the equity method of accounting for affiliated companies each may be categorized
in terms of faithfulness to clean surplus and extent of conservatism.
We address the implications of these accounting procedures for the value relevance of
accounting information. Philippine firms, for example, record goodwill and revalue assets,
Accounting Practices and the Market Valuation of Accounting Numbers
447
but firms in Taiwan do neither. This means that book values in the Philippines will reflect
market values of assets more closely than in Taiwan. Therefore, we expect the explanatory
power of book value will be greater for Philippine firms than for Taiwanese firms. As
another example, only Indonesian and Malaysian firms capitalize leases and R&D
expenditures and use the equity method for affiliated companies. These are less
conservative accounting practices than alternatives used in other counties.
We find accounting in Korea and Taiwan to be least faithful to clean surplus
accounting. Korea does not capitalize goodwill and asset revaluations are amortized to
equity according to tax law. Taiwan does not capitalize goodwill nor allow asset
revaluations. Korea is also the only country not to use the equity method for affiliated
companies. Thus, the earnings of Korean firms do not include the earnings of affiliated
firms. Philippine firms, however, amortize both goodwill and asset revaluations to income.
Recall that violations of clean surplus accounting occur when income does not reflect
changes in equity value. Thus, violations of clean surplus bias empirical calculations of
residual earnings. Therefore, we expect the explanatory power of residual earnings will be
highest for Philippine firms and least for Korean and Taiwan firms.
Overall, our results show significant differences across countries in the value relevance
of accounting earnings and book values. Explanatory power over all firm-years ranges
from R2 = .17 in Taiwan to R2 = .68 for Korea. The incremental explanatory power of book
value per share (BVPS) and residual earnings per share (REPS) is similarly diverse.
Incremental explanatory power of BVPS over all firm-years ranges from 7.2 percent
(Taiwan) to 65.3 percent (Philippines). For REPS, the incremental explanatory power over
all firm-years ranges from 1.4 percent (Korea) to 13.2 percent (Thailand).
Generally, we find differences in accounting appear to be related to differences in value
relevance. We find that the explanatory power of book value is highest in the Philippines
and lowest in Taiwan. This is consistent with our expectations based on the accounting
differences in the two countries. Indonesia and Malaysia have accounting systems that are
less conservative than other countries. However, we find the incremental explanatory
power of book value does not stand out as high in Indonesia or in Malaysia. This result is
only partly consistent with our expectations. We also expected that the relative explanatory
power of residual earnings would be high in the Philippines and low in Korea and Taiwan,
and the results support this prediction. Our comparisons across countries should be viewed
with caution because the number of years of data available ranges from only 2 years for the
Philippines to 10 years for Malaysia.
The next section of the article briefly reviews related research and this is followed by
the section discussing accounting differences in the six Asian countries. This is followed
by the description of the sample and development the study design. The section presenting
the analysis of our data and reporting the results of our tests follows. A final section
summarizes our findings.
RELATED RESEARCH
Research concerned with the relation of accounting numbers and stock prices covers
decades. In this brief review we summarize recent research with study designs and research
methods similar to ours. We see two principal strains: first, research focused on explaining
448
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
stock prices with accounting book value and earnings; and second, research examining the
incremental explanatory power of book value and earnings in the presence of the other.
Stock Prices Explained by Book Value and Earnings Per Share
Examining a large set of US firms, Bernard (1993) found that book values explain 55
percent of the cross-sectional variation in market prices. When current return on equity
(ROE; ranks) was added to the regression, these two accounting measures explain about
64 percent of the variation in market prices. Bernard (1994) finds that return on common
equity (ROE) is mean reverting over time so that firms with the highest (lowest) current
ROEs tend to have lower (higher) ROEs in later years.
King and Langli (1998) examine the explanatory power of BVPS and earnings per
share (EPS) for three European countries: Germany, Norway, and the UK. They find
significant differences in the valuation power of accounting book value and earnings
across the three countries, and they interpret some of the differences as consistent with
diversity in accounting practices. They also find future earnings realizations as proxies for
expected earnings do not have incremental explanatory power beyond that of current
earnings and book value.
Frankel and Lee (1999) look at the relation between accounting values, earnings
forecasts and market prices across 20 countries (including Korea and Thailand) for 8 years,
1987±1994. Sample sizes for Korea and Thailand are small with 3 to 8 observations per
year (33 total firm-years) for Korea and 1 to 40 observations per year (162 total firm-years)
for Thailand. They find that estimates of value based on the residual earnings model have
incremental explanatory power beyond book value and earnings in explaining market
value in all countries. In addition, they find evidence of superior returns to trading
strategies based on an estimate of value from a residual earnings model.
Joos and Lang (1994) relate book value and earnings to stock prices for France,
Germany, and the UK. Their sample covers 1982 to 1990, and they focus on the effects of
implementing the accounting related directives of the European Union. They find the
explanatory power of book value and earnings together ranges from 20 to 38 percent for
Germany, from 48 to 78 percent for France, and from 14 to 42 percent for the UK. They do
not examine incremental explanatory power. Evidence on changes over time is ambiguous,
probably because the time periods for the sample are relatively short.
Harris et al. (1994) examine the value relevance of accounting numbers for German
firms compared to that for a matched set of US firms for 1982±1991. They find little
difference in overall value relevance (R2) between German and US firms. However,
coefficients (multiples) on book value and on earnings for German firms are greater than
for matched US firms. Further, they find that consolidation increases the value relevance
of accounting numbers, and restatements of earnings to adjust for transitory elements in
German accounting also increases explanatory power.
Incremental Explanatory Power of Book Value and Earnings Per Share
Collins et al. (1997) examine the incremental explanatory power of book value and
earnings across a 41-year time period (1953±1993) for US firms. They find a decline
Accounting Practices and the Market Valuation of Accounting Numbers
449
in the ability of earnings to explain market prices over this period. But the explanatory
power of book values increases such that total explanatory power is actually higher in
more recent periods. Average adjusted R2 for a model regressing BVPS and EPS on
stock price for the first 10 years (1953±1962) was .50 increasing to .69 for their most
recent 10-year period (1984±1993). Collins et al. (1997) investigated possible reasons
for these changes. They find the reduced explanatory power of earnings is explained
by an increase in the incidence of one-time items and reported losses as well as a
decrease in the size of firms in the sample.
King and Langli (1998) examine a 15-year period (1982±1996) for Germany, Norway,
and the UK. They find that for Germany the incremental explanatory power of book value
increases significantly while that for EPS decreases. There is no significant change in their
common information. For Norway, there is no significant change in the incremental
explanatory power of book value or EPS over time. While for the UK, the incremental
explanatory power of book value increases and the incremental explanatory power of EPS
is unchanged over the time period.
Harris et al. (1994) also examine the separate explanatory power of book value and of
earnings using simple regressions with only one variable. They do not report the test
statistics. However, they say that while the explanatory power of EPS in Germany is
approximately equal to that in the US, the explanatory power of book value is much lower.
This contrasts sharply to the King and Langli (1998) results for a longer time period. The
Harris et al. (1994) results are not, however, tests of incremental explanatory power since
the simple regressions use only one variable.
This study extends the evidence summarized above. We examine the value relevance of
accounting numbers for companies in Asian countries. Prior financial reporting research in
English language journals has been limited.
ACCOUNTING DIVERSITY ACROSS THE SIX COUNTRIES
The accounting systems in all six countries have developed relatively recently. The six
accounting systems differ on some dimensions but are similar on others. Two dimensions
that we examine are (1) the model on which the accounting systems are based and (2) the
type of standard setting body. Table 1 shows these characteristics for the six countries. IAS
was the primary basis for accounting standards in Indonesia, Malaysia, and Thailand
(although Thailand has also been influenced by US GAAP). US GAAP, on the other hand,
was the primary basis in the Philippines and Taiwan (although Philippine GAAP is
secondarily based on tax law). Korean accounting standards are unique in that they are
based on Korean tax law that, like tax law in all countries, emphasizes cash realization.
Different accounting models may lead to differences in the value relevance of the resulting
accounting numbers. We have no prior expectations concerning the relative value
relevance of IAS versus US GAAP. It is likely, however, that tax law is more susceptible
to political influence than other accounting bases. To the extent that such political
influence might serve to make accounting less informative, Korean accounting may be
less value relevant because it is based on tax law.
The standard setting bodies in four of the six countries are independent of the
government. In Korea and Taiwan, however, standard setting is not independent. Where
450
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
Table 1. Accounting Standards and Standard Setting in the Six Countries
Country
Indonesia
Korea
Malaysia
Philippines
Taiwan
Thailand
Primary basis for
accounting standards
Independent accounting
standards setting body?
IAS
Tax Law
IAS
US GAAP
US GAAP
IAS
yes
no
yes
yes
no
yes
standard setting is influenced or controlled by government there is greater potential for
political influence in the standards setting process. As where accounting follows tax law,
this may lead to lower value relevance of accounting numbers.
We analyzed the accounting standards and practices for each country using a variety
of sources including Akathaporn (1995), Graham and Wang (1995), and publications
from the AICPA (1989, 1990, 1992), CIFAR (1995), Deloitte Touche Tohmatsu
International (1995a,b, 1996a,b, 1997), Price Waterhouse (1995a,b, 1996a,b,c,d), and
Mathew Bender & Co.(1996). Under the residual earnings model the only crucial
accounting characteristics are unbiased accounting and clean surplus accounting.
Conservative accounting practices are biased since value changes are reflected
asymmetrically, value declines are recognized more quickly than value increases.
The clean surplus relation (CSR) allows book value of equity to change only with
income or loss and net capital investments and withdrawals (dividends) by owners.
CSR is violated if changes in book value can by-pass income. We focus our analysis
on the effects of accounting differences on book value and on residual (abnormal)
earnings, the accounting arguments in the residual earnings model. However, accounting differences affect valuation only when they are violations of unbiased accounting
or clean surplus accounting.
Differences Across Countries Affecting Book Value
Revaluing assets is a violation of CSR if the accompanying credit is taken directly to
equity. Yet asset revaluations bring book value nearer to market value. Immediate write-off
of goodwill violates CSR and usually moves book values farther from market values. In
summary, both recognizing goodwill (consistent with CSR) and revaluing assets (violating
CSR) bring book value nearer to market value. Hence, developing predictions on the
effects of specific accounting treatments on value relevance is not always clear.
Conservative accounting (bias) is expected to generally reduce the value relevance of
both book value and earnings since the essence of conservatism is delay in reflecting
certain events in the accounting records.
Recording goodwill is common practice in Indonesia and the Philippines, and
uncommon but allowed in Malaysia and Thailand. Korea and Taiwan do not allow
goodwill to be recorded. Asset revaluations are common in all countries except Indonesia
and Taiwan where they are allowed but restricted in practice. Table 2 presents a summary
of our analysis of these accounting practices.
Accounting Practices and the Market Valuation of Accounting Numbers
451
Table 2. Goodwill and Asset Revaluations on the Balance Sheet
Country
Goodwill recorded
on balance sheet?
Asset revaluation
on balance sheet?
yes
no
some
yes
no
some
uncommon
common
common
common
uncommon
common
Indonesia
Korea
Malaysia
Philippines
Taiwan
Thailand
Table 3. Other Asset Values on the Balance Sheet
Country
Indonesia
Korea
Malaysia
Philippines
Taiwan
Thailand
Capital leases
on balance sheet?
yes
some
yes
no
some
not until 1996
R&D expenditures
on balance sheet?
yes
yes
yes
no
no
yes
Equity method
used for affiliates?
yes
no
yes
yes
yes
yes
Asset revaluation and goodwill are both recorded only in the Philippines, and neither is
recorded in Taiwan. The other four countries allow one procedure or the other but not
both. We expect the incremental explanatory power of book value to be high in the
Philippines and low in Taiwan relative to the other five countries. However, the effect of
these accounting practices on the incremental explanatory power of book value for the
other countries is ambiguous.
The six countries also differ in other accounting practices, including capitalizing leases,
capitalizing research development costs, and applying the equity method to affiliated
firms. Firms that capitalize and use the equity method are likely to have book values that
are closer to market values than firms that do not.2 Table 3 presents a summary of our
analysis of these accounting practices.
Only Indonesia and Malaysia allow or require all three accounting procedures. The
effect of these accounting differences on incremental explanatory power is ambiguous for
the other countries. However, we expect the explanatory power of book value in Indonesia
and Malaysia to be higher than in Thailand and Taiwan.
Differences Across Countries Affecting Residual Earnings
As explained in the previous section, conservatism is expected to reduce the value
relevance of both book value and residual earnings. In addition, we can make some
predictions about the effects of clean surplus accounting for goodwill and asset revaluations on the value relevance of book value. However, the effects of these accounting
practices on the value relevance of residual earnings are less clear.
The Philippines is the most faithful to clean surplus accounting as both goodwill
and asset revaluations are amortized to income over their useful lives. The other
452
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
Table 4. Goodwill and Asset Revaluation Amortization
Country
Indonesia
Korea
Malaysia
Philippines
Taiwan
Thailand
Goodwill
amortized to?
income
±
±
income
±
±
Amortization
period
useful life
±
immediate write-off
no more than 40 years
±
immediate write-off
Revaluations
amortized to?
Amortization
period
±
shareholder equity
shareholder equity
income
±
shareholder equity
±
per tax law
useful life
useful life
±
useful life
countries (1) do not record goodwill (Korea and Taiwan), (2) do not record asset
revaluations (Indonesia and Taiwan), (3) immediately write-off goodwill to equity
(Malaysia and Thailand), or (4) amortize asset revaluation increments to equity (Korea,
Malaysia, and Thailand). Korea and Taiwan are least faithful to clean surplus
accounting. Korea does not capitalize goodwill and asset revaluations are amortized
to equity according to schedules mandated by tax law. Taiwan does not capitalize
goodwill nor allow asset revaluations. Korea is also the only country that does not use
the equity method for affiliated companies; therefore the earnings of Korean firms do
not include the earnings of affiliated firms. We expect the relative explanatory power
of residual earnings to be high in the Philippines and low in Korea and Taiwan. The
effect on the explanatory power of residual earnings for the other countries is
ambiguous. Table 4 presents a summary of the amortization practices.
Summary of the Research Questions
Our examination of accounting practices reveals some systematic differences across the
six countries. While the differences appear to be substantial, it is an empirical question
whether they result in meaningful violations of the CSR or in significant accounting bias
(conservatism), the important factors for the residual earnings model. Accounting bias will
likely reduce the explanatory power of both book value and earnings. Furthermore,
violations of the CSR may either increase or decrease the explanatory power of book value
depending on whether the violation moves book value toward or away from market
values. Because countries differ on both dimensions, ex ante hypotheses on which effect
will dominate are problematic. Even so, we expect that bias and CSR violations will affect
the value relevance of accounting numbers in systematic ways. Particularly because
Philippine firms record both goodwill and asset revaluations and Taiwan firms do neither,
we expect the value relevance of book value to be greatest in the Philippines and least in
Taiwan. Because both Indonesian and Malaysian firms capitalize leases and R&D
expenditures and use the equity method we expect the value relevance of their book
values to exceed the value relevance of book value in Thailand and Taiwan. Because
Philippine firms amortize both goodwill and asset revaluations to income, we expect the
value relevance of residual earnings to be high in the Philippines.
In addition, we investigate changes in the value relevance of accounting numbers over
time. Our sample contains 3,655 firm-years across six countries. We have sufficient data
for only 2 years of yearly regressions for the Philippines but 10 years for Malaysia. We
Accounting Practices and the Market Valuation of Accounting Numbers
453
Table 5. Sample Countries and Firm-Years
Sample selection
Indonesia
Korea
Malaysia
Philippines
Taiwan
Thailand
Sample size
Firm-years
338
902
1,311
139
369
596
3,655
trace the total explanatory power of accounting earnings and book value and the
incremental explanatory power of each earnings and book value in the presence of the
other across time for each country.
SAMPLE AND STUDY DESIGN
Our sample covers publicly traded firms in Indonesia, Korea, Malaysia, the Philippines,
Taiwan, and Thailand across the period from 1987 to 1996. The stock prices and
accounting data for this study are from the Worldscope Global Researcher. The sample
selection criteria are:
1.
2.
3.
4.
5.
Accounting data is from consolidated financial statements.
Financial firms are excluded (insurance, banks, and other miscellaneous financial
firms). Accounting practices for these firms are so distinct that their valuation
parameters are likely to be substantially different from those for industrial firms.
Firms with negative book values are deleted. These firms are likely to be in
financial distress and may be interesting in their own right. However, the focus of
this study is the across country differences in value relevance of accounting
numbers derived under different accounting practices. Hence, restricting our
sample to firms with positive book values will allow us to focus on firms where
differences are mostly likely to reflect accounting differences.
Twelve firms with EPS greater than their BVPS are deleted since data on those firms
is likely to contain errors. These firms constituted less than 1 percent of the sample.
Twenty firm-years with excessive statistical influence in our regressions were
deleted. The firm-years showed undue influence by the diagnostics and cutoff rules
described in Belsley et al. (1980).
These restrictions on the sample will have several effects. First, the model will appear
to ``fit'' better than it would fit unrestricted data. That is, the explanatory power of book
value and of residual earnings information in the sample is likely to be greater than for an
unrestricted sample. Second, the samples across the six countries will be more homogeneous and the effects of different business cycles in the six countries will be reduced.
This should allow a better focus on the effects of accounting differences. Table 5 shows the
countries and number of firm-years.
454
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
Table 6. Descriptive Statistics on Variables for Six Southeast Asian Countries
Variable
N
Mean
Standard deviation
Panel A: Indonesia
Price
338
BVPS
338
EPS
338
REPS
338
ROE
340
2,813
1,668
242
ÿ 75
0.15
2,261
1,175
305
272
0.09
Panel B: Korea
Price
902
BVPS
902
EPS
902
REPS
902
ROE
902
25,629
21,019
1,353
ÿ 372
0.07
Panel C: Malaysia
Price
1,311
BVPS
1,311
EPS
1,311
REPS
1,311
ROE
1,311
5th Percentile
Median
95th Percentile
556
361
43
ÿ 474
0.03
2,236
1,374
154
ÿ 52
0.14
6,500
4,394
765
180
0.30
34,972
31,375
4,324
3,334
0.14
8,002
6,282
ÿ 979
ÿ 3,291
ÿ 0.09
18,100
14,104
786
ÿ 398
0.06
63,921
50,225
5,007
2,389
0.26
5.16
1.71
0.18
0.06
0.13
5.99
1.11
0.21
0.19
0.15
0.94
0.53
ÿ 0.03
ÿ 0.17
ÿ 0.04
3.84
1.44
0.15
0.04
0.12
13.96
3.72
0.56
0.37
0.38
Panel D: Philippines
Price
139
BVPS
139
EPS
139
REPS
139
ROE
139
19.06
7.91
0.97
ÿ 0.14
0.16
31.65
12.00
1.71
1.32
0.14
0.17
0.17
ÿ 0.08
ÿ 2.75
ÿ 0.04
6.37
2.54
0.37
ÿ 0.01
0.14
81.37
38.68
4.94
1.30
0.40
Panel E: Taiwan
Price
369
BVPS
369
EPS
369
REPS
369
ROE
369
30.55
12.43
1.20
ÿ 0.18
0.10
13.21
2.81
1.25
1.25
0.11
14.95
7.80
ÿ 0.79
ÿ 2.17
ÿ 0.06
27.89
12.19
1.16
ÿ 0.21
0.09
53.83
17.51
3.16
1.84
0.26
Panel F: Thailand
Price
596
BVPS
596
EPS
596
REPS
596
ROE
596
95.79
41.35
5.93
ÿ 0.23
0.12
102.39
35.05
8.32
5.10
0.14
18.25
11.05
ÿ 1.14
ÿ 6.41
ÿ 0.06
59.59
31.09
3.59
ÿ 0.78
0.13
330.00
117.83
21.24
7.48
0.34
Price = Stock price at the end of year t. BVPS = Book value of shareholders' equity at the end of year t. EPS = Earnings per share
for year t. REPS = Residual (abnormal) earnings per share = EPSt ÿ r*((BVPSt + BVPSt ÿ 1)/2) where r is the country average
lending rate in year t taken from the International Financial Statistics Yearbook. Price, book value, EPS, and REPS amounts are in
nominal local currency. ROE = Return on equity = EPSt /((BVt + BVt ÿ 1)/2).
Table 6 provides descriptive statistics for the sample. Per share values are in nominal
currency of the countries, so comparisons are difficult. However, we can compare ROE
(the ratio of EPS to average book value) across countries and the differences are
substantial. Median ROE ranges from 6 percent (over 8 years) in Korea to 14 percent
Accounting Practices and the Market Valuation of Accounting Numbers
455
Table 7. Correlation Statistics on Variables for Six Asian Countries ( p values in parentheses)
Spearman
Variable
N
Price
EPS
Panel A: Indonesia
Price
338
.6164 (.0001)
EPS
338 .5741 (.0001)
BV
338 .4777 (.0001) .6418 (.0001)
ROE
338 .3423 (.0001) .6118 (.0001)
REPS
338 .2498 (.0040) .6262 (.0001)
Pearson
Panel B: Korea
Price
902
EPS
902
BV
902
ROE
902
REPS
902
.6894
.8177
.1667
.3395
.5328 (.0001)
(.0001)
(.0001) .7438 (.0001)
(.0001) .4769 (.0001)
(.0001) .8529 (.0001)
Panel C: Malaysia
Price
1,311
.6224 (.0001)
EPS
1,311 .4171 (.0001)
BV
1,311 .5031 (.0001) .5552 (.0001)
ROE
1,311 .1093 (.0001) .6042 (.0001)
REPS
1,311 .2564 (.0001) .9093 (.0001)
Panel D: Philippines
PRICE
139
EPS
139 .7593
BV
139 .7906
ROE
139 .0169
REPS
139 .0727
Panel E: Taiwan
Price
369
EPS
369
BV
369
ROE
369
REPS
369
.8133 (.0001)
(.0001)
(.0001) .7790 (.0001)
(.8439) .2416 (.0042)
(.3949) .4301 (.0001)
.5043 (.0001)
.3618 (.0001)
.2639 (.0001) .2422 (.0001)
.3048 (.0001) .9353 (.0001)
.3116 (.0001) .9609 (.0001)
Panel F: Thailand
Price
596
.7308 (.0001)
EPS
596 .6284 (.0001)
BV
596 .5145 (.0013) .8121 (.0001)
ROE
596 .4415 (.0001) .5912 (.0001)
REPS
596 .5166 (.0001) .8068 (.0001)
BV
.5685 (.0001)
.6915 (.0001)
ÿ .0130 (.8119)
ÿ .1012 (.0631)
.6021 (.0001)
.5552 (.0001)
.0865 (.0093)
.2766 (.0001)
.5572 (.0001)
.5419 (.0001)
.0117 (.6730)
.2102 (.0001)
.8601 (.0001)
.7726 (.0001)
ÿ .1081 (.2052)
ÿ .1866 (.0278)
.3056 (.0001)
.2487 (.0001)
.0276 (.5972)
ÿ .0151 (.7731)
.5639 (.0001)
.7144 (.0001)
.2494 (.0001)
.3405 (.0001)
ROE
.2255 (.0001)
.5066 (.0001)
ÿ .1406 (.0097)
REPS
.0662
.2846
ÿ .3028
.8904
(.2252)
(.0001)
(.0001)
(.0001)
.1727
.6624
ÿ .0681
.7442
(.0012)
(.0001)
(.0408)
(.9190)
.4496
.8697
.1774
.9252
(.0001)
(.0001)
(.0001
(.0001)
.7768 (.0001)
.3095 (.0001)
.8285 (.0001)
.1374 (.0001)
.5963 (.0001)
.3272 (.0001)
.7678 (.0001)
.0160 (.5619)
.7218 (.0001)
.1725 (.0423)
.4577 (.0001)
ÿ .0654 (.4441)
.1177 (.1676)
.3776 (.0001)
ÿ .0715 (.4027)
.8089 (.0001)
.4931 (.0001)
.4399 (.0001)
.9499 (.0001)
ÿ .0010 (.9853)
.4431
.9291
ÿ .0390
.9804
(.0001)
(.0001)
(.4548)
(.0001)
.9909 (.0001)
.6144 (.0001)
.8625 (.0001)
.3443 (.0001)
.5523 (.0001)
.7757 (.0001)
.2411 (.0001)
.9362 (.0001)
.7237 (.0001)
Price = Stock price at the end of year t. EPS = Earnings per share for year t. BVPS = Book value of shareholders' equity at
the end of year t. ROE = Return on equity = EPSt / ((BVPSt + BVPSt ÿ 1)/2). REPS = Residual (abnormal) earnings per
share = EPSt ÿ r*((BVPSt + BVPSt ÿ 1)/2) where r is the country average lending rate in year t taken from the International
Financial Statistics Yearbook.
(over 5 and 2 years) in Indonesia and the Philippines. For comparison, King and Langli
(1998) find ROE over the 1980s and 1990s to be about 6 percent in Germany, 10 percent
in Norway, and 13 percent in the UK. In the US, this measure has averaged around 13
456
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
percent over the last 20 years. Prior research has speculated that such differences across
countries may reflect differences in conservatism of accounting methods. As noted above,
for example, Taiwan does not record either goodwill asset revaluations (most conservative)
while both are recognized in the Philippines (least conservative).
Table 7 reports the pair-wise correlation between stock price and accounting variables
for all countries. For all countries except Korea the rank (Spearman) correlations are
greater than the product±moment (Pearson) correlations. However, the patterns and
significance of the parametric (Spearman) and non-parametric (Pearson) correlations are
similar. Stock prices are strongly correlated with BVPS and with EPS for all countries. The
pair-wise correlations between price and EPS are approximately the same as between price
and BVPS. The correlation (Pearson) between BVPS and EPS is high in Thailand (.81),
the Philippines (.78), and Korea (.74), and relatively low for Taiwan (.24). If EPS is used
as a proxy for residual earnings, the high correlation between BVPS and EPS (they are
related by size) may make it difficult to partition value relevance between book value and
earnings. However, the pair-wise correlations between BVPS and REPS are far smaller,
significant in some cases, insignificant in others, and sometimes negative. This is
anticipated since there is no reason to expect residual (unexpected) per share earnings
to be related to book value. One important reason for using REPS in the empirical analysis
is to avoid the high correlation between BVPS and EPS.
TESTS AND ANALYSIS
Our analysis is based on contemporaneous cross-sectional regressions of accounting book
values and residual earnings on stock prices (dependent variable). We analyze both the
relative and the incremental explanatory power of book value and residual earnings using an
approach applied previously in accounting by Biddle et al. (1995) and Collins et al. (1997).
Empirical specification of the residual earnings model requires estimates of book value,
residual earnings, and the horizon for residual earnings. For residual earnings estimated to
terminate at time T, the model would be:
Priceit a0 a1 BVPSit a2 REit1 a3 REit2 a4 REit3 . . . ak REiT eit
1
where Priceit is the price per share of firm i at the end of period t, BVPSit is the book value
per share of firm i at the end of period t, and REit is the residual earnings per share of firm i
for year t + k.
The coefficient a1 would have an expected value of 1.0 while the coefficients a2 to a4
would have expected values of (1 + r) ÿ t. Finally, the expected value of coefficient ak
would be (1/r)*(1 + r) ÿ T.3 Residual earnings horizons will differ cross-sectionally; therefore, parsimonious cross-sectional representations of Equation (1) will have only a few
terms. For example Frankel and Lee (1999) use T = 2 because analysts' forecasts used to
predict future residual earnings were only available for 2 years.
Our first tests are concerned with the incremental explanatory power of book value and
residual earnings. As in Collins et al. (1997) we compare the results of three regression
equations to address the question of relative and incremental explanatory power. Equation
(2) below provides the most parsimonious empirical specification of the residual earnings
Accounting Practices and the Market Valuation of Accounting Numbers
457
model on a per share basis (the horizon is only one period). Current period residual
earnings is the proxy for future expected residual earnings. Residual earnings are estimated
by subtracting an estimate of normal (expected) earnings from reported earnings. Expected
earnings is the product of the estimated rate of return (r) and book value. Like Frankel and
Lee (1999), we derive the estimated rate of return from interest rates in the International
Financial Statistics Yearbook published by the International Monetary Fund (1997).
Frankel and Lee (1999), however, are able to calculate a risk-adjusted return by adding
a risk premium to long-term government bond rates Government bond rates are not
available for four of the countries; therefore, we use commercial lending rates. Conceptually, this rate is the sum of a riskless rate and the average commercial lending risk
premium. In the residual earnings model, book value and firm value are taken at time t
while future abnormal earnings are for periods after time t. In our empirical analysis
residual earnings (REt) are for the period ending at time t. Hence, as in Bernard (1994) and
Collins et al. (1997), current earnings is a proxy for expected future earnings.
Priceit b0 b1 BVPSit b2 REPSit eit
2
where Priceit is the stock price per share of firm i at the end of year t, BVPSit is the book
value of shareholders' equity of firm i at the end of year t, REPSit is the residual earnings per
share, which is equal to EPSit ÿ r*(BVPSt ÿ 1)4 (proxy for expected REPS in period t + 1),
EPSit is the earnings per share of firm i for year t, and r is the country's average commercial
lending rate in year t taken from the International Financial Statistics Yearbook.
Book values and earnings are, of course, unobservable until some weeks after the end of
the fiscal year. This raises the question of the timing of the market value measure to be
associated with the accounting variables. As discussed by Barth et al. (1996), choice of
contemporaneous versus lagged market values is a trade-off. The advantage to using a
lagged market price is that it may reasonably reflect the accounting results since sufficient
time has passed for these results to be public information. However, lagged market values
will include effects of information and events occurring after the end of the fiscal year.
Collins et al. (1997), examining associations between market and accounting numbers for
US firms, take prices 3 months after the end of the fiscal period. In cross-country studies,
however, this is problematic since the time lag between fiscal year-ends and report dates can
vary widely. For this reason, our tests examine the relation between accounting numbers
(book value and residual EPS for a fiscal year) and stock prices at the end of the fiscal year.
Later, we analyze the sensitivity of our results using stock prices lagged 0 to 6 months
following the end of the fiscal year.
Equation (2) expresses price as a function of book value and residual earnings.
Examining the relative and incremental explanatory power of book value and of residual
earnings requires two additional equations expressing price as a function of book value
alone, Equation (3), and residual earnings alone, Equation (4).5
Priceit c0 c1 BVPSit eit
3
Priceit d0 d1 REPSit eit
4
458
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Vol. 35, No. 4, 2000
Following Theil (1971), we define the incremental explanatory power of the
book value and residual earnings variables in terms of differences in the coefficient
of determination (R2). These differences are sometimes called the semi-partial
coefficient of determination (Cohen and Cohen, 1975, pp. 79±84). They are a
measure of the incremental explanatory power of one variable given the remaining
independent variables.
2
, respectively.
Define the R2 statistics from Equations (2), (3), and (4) as Rb2, Rr2, and Rb,r
Then the incremental explanatory power is defined as:
2
R2bjr = Rb,r
ÿ R2r
R2rjb = R2b,r ÿ R2b
2
ÿ R2bjr ÿ R2rjb
R2com = Rb,r
The incremental explanatory power of book value is
the total explanatory power of book value and residual
earnings less the explanatory power of residual earnings alone.
The incremental explanatory power of residual earnings is
the total explanatory power of book value and residual
earnings less the explanatory power of book value alone.
The explanatory power common to book value and
residual earnings is the total explanatory power of book
value and residual earnings less the incremental explanatory power of book value and the incremental explanatory
power of residual earnings.6
We can also assess the relative explanatory power of book value and residual earnings
by comparing the conditional (incremental) power as shown above (Biddle et al., 1995).7
That is, we can also address the question of whether book values or residual earnings have
greater explanatory power for each country.
Explanatory Power of BVPS and REPS Across the Six Countries
Table 8 reports summaries of regressions (2), (3), and (4) as well as incremental R2 for
each year with 30 or more observations and for all years together for the six countries.
First, we focus on the coefficients and the significance of regressions (2), (3), and (4), and
then we analyze the relative and incremental explanatory power.8
Coefficients on BVPS are positive for all countries. They are significant overall and for
most years. BVPS coefficients are greater than 1.0 for Malaysia (2.75), the Philippines
(2.17), Taiwan (1.26), and Thailand (1.41) in regression (2). BVPS coefficients are less
than 1.0 for Indonesia (.82) and Korea (.89). Coefficients on REPS are positive and
significant for all countries except the Philippines. Coefficients on REPS for regression (4)
range from 1.26 for Malaysia to 10.90 for Thailand.9
We find significant differences in the value relevance of accounting across countries.
The explanatory power of book value and residual earnings is quite high for Korea and the
Philippines, near to what is found for Anglo-American markets. However, the explanatory
power for Taiwan is well below that found for most other countries. Table 9 presents a
Priceit b0 b1 BVPSt b2 REPS eit
R2b;r
Priceit c0 c1 BVPSt eit
R2b
Priceit d0 d1 REPS eit
R2r
where Priceit is the price per share of firm i at time t, BVPSt is the book value per share of firm i at the end of period t, and REPSt is the residual earnings per
share of firm i for year t.
R2bjr = R2b,r ÿ R2r
The incremental explanatory power of book value is the total explanatory power of book value and residual earnings less the
explanatory power of residual earnings alone.
The incremental explanatory power of residual earnings is the total explanatory power of book value and residual earnings less the
explanatory power of book value alone.
The explanatory power common to book value and residual earnings is the total explanatory power of book value and residual earnings
less the incremental explanatory power of book value and the incremental explanatory power of residual earnings alone.
R2rjb = R2b,r ÿ R2b
R2com = R2b,r ÿ R2bjr ÿ R2rjb
Year
N
Panel A: Indonesia
All
338
1991
46
1992
61
1993
63
1994
70
1995
80
Meana
64
b0
1,400
698
596
1,864
1,495
1,526
1,236
(7.83)
(2.12)
(1.82)
(6.19)
(3.41)
(3.61)
b1BV
0.96
1.62
1.75
1.04
0.79
0.63
1.17
(10.90)
(7.38)
(8.80)
(6.54)
(3.98)
(3.32)
b2REPS
2.49
6.08
5.66
0.86
3.07
3.41
3.82
(6.57)
(3.82)
(8.82)
(1.78)
(3.53)
(3.99)
2
Rb,r
.308
.559
.564
.419
.332
.266
.434
c1BV
0.90
1.16
1.32
0.99
0.90
0.66
1.01
(9.69)
(5.52)
(5.43)
(6.22)
(4.22)
(3.13)
Rb2
.219
.409
.358
.388
.208
.114
.295
d1REPS
2.07
ÿ 0.34
2.46
0.34
3.59
3.50
1.91
(4.73)
( ÿ 0.17)
(1.80)
(0.55)
(3.78)
(3.86)
Rr2
2
Rbjr
.062
.001
.052
.005
.173
.161
.078
.246
.558
.542
.414
.159
.105
.356
Rr2jb
.089
.150
.236
.031
.124
.152
.139
Accounting Practices and the Market Valuation of Accounting Numbers
Table 8. Incremental and Relative Information Content of Book Values and Residual Earnings: Regressions of Book Value and Residual Earnings on Price by
Year (t-statistics in parentheses)
2
Rcom
ÿ .027
ÿ .149
ÿ .184
ÿ .026
.049
.009
ÿ .060
459
(continued)
460
Table 8. (Continued)
Year
N
b0
b1BV
b2REPS
2
Rb,r
c1BV
Rb2
d1REPS
Rr2
2
Rbjr
Rr2jb
2
Rcom
0.87
0.21
0.31
0.24
0.27
0.50
0.74
0.81
0.92
0.51
(40.07)
(2.43)
(4.26)
(4.49)
(5.08)
(7.13)
(9.49)
(18.06)
(28.72)
1.29
1.94
1.44
0.89
0.73
1.99
3.26
1.61
0.32
1.52
(6.27)
(2.41)
(2.53)
(2.11)
(2.09)
(5.53)
(4.60)
(3.48)
(1.02)
.683
.182
.241
.240
.261
.584
.820
.816
.843
.498
0.91
0.18
0.22
0.18
0.26
0.67
1.05
0.91
0.91
0.55
(42.61)
(1.96)
(3.32)
(3.87)
(4.75)
(9.42)
(23.44)
(26.96)
(30.87)
.669
.077
.157
.187
.220
.461
.794
.803
.842
.442
3.56
1.62
0.25
ÿ 0.10
0.50
3.13
8.91
7.27
ÿ 3.22
2.92
(10.83)
(1.93)
(0.45)
( ÿ 0.26)
(1.25)
(7.97)
(18.64)
(12.73)
( ÿ 4.81)
.115
.075
.000
.001
.019
.379
.709
.477
.115
.222
.568
.107
.241
.239
.242
.205
.111
.339
.728
.276
.014
.105
.084
.053
.041
.123
.026
.013
.001
.056
.101
ÿ .030
ÿ .084
ÿ .052
ÿ .022
.256
.683
.464
.114
.166
Panel C: Malaysia
All
1,311
1987
42
1988
50
1989
59
1990
67
1991
102
1992
169
1993
172
1994
183
1995
258
1996
201
Meana
130
0.51 (2.00)
0.66 (1.77)
0.59 (1.89)
1.76 (4.54)
1.74 (4.27)
0.92 (3.69)
0.16 (0.66)
ÿ 1.11 (1.30)
0.86 (1.04)
0.80 (1.33)
4.21 (4.02)
1.06
2.53
1.38
1.24
0.75
0.85
0.99
1.72
4.31
3.03
2.38
1.37
1.80
(19.54)
(4.08)
(4.88)
(2.65)
(2.75)
(6.35)
(12.57)
(9.70)
(7.84)
(8.77)
(5.21)
5.07
2.57
2.33
4.28
6.00
9.33
8.30
5.09
2.41
3.25
7.64
5.12
(6.55)
(2.59)
(3.00)
(3.78)
(4.71)
(9.96)
(8.36)
(1.62)
(0.91)
(1.90)
(3.33)
.277
.300
.391
.322
.381
.637
.532
.371
.264
.263
.157
.362
2.71
0.82
1.18
0.98
1.24
1.32
1.45
4.37
3.07
2.48
1.97
1.89
(21.06)
(2.97)
(4.20)
(3.16)
(3.59)
(6.13)
(9.17)
(9.80)
(7.98)
(9.30)
(4.95)
.253
.180
.269
.149
.166
.273
.335
.361
.260
.253
.110
.236
8.25
ÿ 0.04
1.90
4.89
6.92
10.58
5.32
7.30
4.58
0.64
10.73
5.28
(9.60)
( ÿ 0.04)
(2.08)
(4.21)
(5.37)
(9.78)
(3.96)
(1.88)
(1.50)
(3.30)
(5.04)
.067
.000
.083
.237
.307
.489
.086
.020
.012
.041
.113
.139
.210
.300
.308
.085
.074
.148
.446
.351
.252
.222
.044
.223
.024
.120
.122
.173
.215
.364
.197
.010
.004
.010
.047
.126
.043
ÿ .120
ÿ .039
.064
.092
.125
ÿ .111
.010
.008
.031
.066
.013
Vol. 35, No. 4, 2000
7,746 (9.48)
15,983 (10.36)
19,043 (16.22)
13,791 (17.12)
9,826 (9.51)
9,383 (5.75)
11,318 (4.75)
13,979 (8.06)
3,266 (2.25)
12,074
THE INTERNATIONAL JOURNAL OF ACCOUNTING
Panel B: Korea
All
901
1988
48
1989
61
1990
67
1991
82
1992
106
1993
145
1994
180
1995
182
Meana
109
2.44 (1.32)
0.45 (0.15)
ÿ 0.23 ( ÿ 0.16)
0.11
2.20 (16.75)
2.92 (14.63)
2.58 (22.99)
2.75
5.49 (4.60)
13.76 (6.67)
18.72 (10.09)
16.24
.680
.885
.921
.903
2.08 (1.30)
2.41 (8.27)
1.89 (10.88)
2.15
.625
.702
.716
.709
.169
.135
.272
.300
.236
1.24
2.19
1.52
1.41
1.71
(5.24)
(2.40)
(4.36)
(6.47)
.070
.126
.164
.191
.160
3.30
1.30
3.03
2.35
2.23
.397
.848
.727
.300
.512
.369
.551
1.50
2.06
1.93
1.33
1.53
1.38
1.65
(14.63)
(12.24)
(11.15)
(4.55)
(8.39)
(7.62)
.265
.798
.635
.139
.300
.251
.425
10.37
6.67
9.42
14.74
11.56
9.33
10.34
Panel E: Taiwan
All
369
1993
42
1994
98
1995
179
Meana
106
15.57
9.30
15.39
9.35
11.35
(5.43)
(0.84)
(3.82)
(3.43)
1.26
2.20
1.68
1.35
1.74
(5.64)
(2.38)
(5.08)
(6.63)
3.34
1.32
3.63
2.19
2.38
Panel F: Thailand
All
596
1991
40
1992
74
1993
130
1994
166
1995
175
Meana
117
51.28
18.63
27.59
102.78
47.66
38.21
46.97
(9.71)
(2.25)
(3.45)
(6.72)
(5.60)
(2.80)
1.28
1.81
1.68
0.57
1.10
1.12
1.26
(11.29)
(11.06)
(10.53)
(1.91)
(6.80)
(6.64)
7.76 (11.38)
2.84 (3.49)
5.55 (4.92)
12.64 (5.39)
9.18 (8.39)
7.14 (5.65)
7.47
a
(6.63)
(0.65)
(3.79)
(5.21)
.005
.010
.010
.010
.675
.875
.911
.893
.055
.183
.205
.194
ÿ .050
ÿ .173
ÿ .195
ÿ .184
(6.28)
(0.60)
(2.84)
(5.02)
.097
.009
.077
.125
.070
.072
.126
.195
.175
.166
.099
.009
.108
.109
.076
ÿ .002
.000
ÿ .031
.016
ÿ .006
(14.70)
(4.48)
(5.55)
(7.05)
(9.88)
(6.91)
.267
.345
.300
.280
.373
.216
.303
.130
.503
.427
.020
.139
.153
.248
.132
.050
.092
.161
.212
.118
.126
.135
.295
.208
.119
.161
.098
.177
1.74 (0.85)
2.27 (0.41)
ÿ 3.37 ( ÿ 0.68)
ÿ 0.55
Accounting Practices and the Market Valuation of Accounting