Accounting diversity and the value relevance of accounting earnings and book value in four countries-the United States, the United Kingdom, Canada and Japan.

Hsu, Kathy H.Y. ; Etheridge, Harlan L.

ABSTRACT

Diversities in accounting standards across different countries are of serious concern to the international investment community. Analysts, investors and security regulators have indicated that accounting differences can affect the usefulness of accounting numbers in assessing firm values (Choi and Mueller 1992). This paper investigates the impact of accounting diversity on the relative value relevance of accounting summary numbers in four countries: the United States, the United Kingdom, Canada and Japan. Accounting systems of these four countries are of interest because of the varying degree of perceived conservatism in accounting measurement practices across these four accounting systems and the size of the capital markets in these four countries.

To assess the differences in value relevance of two accounting summary numbers, earnings and book value, across different countries, we use empirical models and measurements that allow direct comparisons of the weights assigned to accounting earnings and book value by market participants of these four countries. To compare the relative value relevance of accounting earnings with that of accounting book value within each country, we use a standardized procedure to control the scale differences in earnings and book value so that the earnings valuation coefficient and the book value valuation coefficient can be compared. A ratio measure, which scales the valuation coefficients by the overall informativeness of earnings and book value of each country, is used to allow additional cross-country comparisons.

The results show that systematic differences in the importance of earnings and book value do exist across national boundaries. On average, earnings are more important than book value in explaining firm values in the U.S. and the U.K., while book value is more important than earnings in Canada and Japan.

In a model incorporating both earnings and book value to explain market values, accounting earnings and book value both have incremental value relevance over each other for the U.S. and Canadian samples. However, U.K. accounting book value and Japanese earnings do not consistently have incremental value relevance.

Our results also indicate trends of change in the overall, as well as the relative, importance of earnings and book value over the sample period. In particular, the overall value relevance of accounting earnings and book value increases over time for all four countries, and the value relevance of accounting book value relative to accounting earnings increases in both Canada and the U.K. over the sample period. Neither the U.S. or Japan show significant changes in the relative value relevance of the accounting numbers over the sample period.

INTRODUCTION

Diversities in accounting standards in different countries pose significant challenges to the international investment community. While current efforts to harmonize international accounting standards by the International Accounting Standards Committee (IASC) and the International Organization of Securities Commissions (IOSCO) are laudable, differences in accounting standards across national boundaries largely remain a fact of life. Although participants in the international investment community recognize that differences in accounting measurement and disclosure rules across countries can affect the usefulness of accounting numbers in accessing firm values and affect the flow of international capital, empirical evidence on the comparative impact of accounting diversity on firm valuation across various accounting regions is still limited.

This paper investigates the impact of accounting diversity on the relative value relevance of accounting summary numbers in four countries: the United States, the United Kingdom, Canada and Japan. These four countries are highly developed nations with historically different accounting concepts and accounting practices. Each of these four countries also has very active capital markets (1) and high levels of foreign direct investment flows. Most importantly, all four countries have representation on the IASC board, which approves the exposure drafts and final statements issued by the IASC. As a result, each of these four countries plays an important role in the development of international accounting standards.

While these four countries are important participants in the world economy and are instrumental in the accounting harmonization process, the domestic accounting standards and environments of these four countries are diverse. There are many facets to the accounting diversity across these four countries: differences in accounting measurement and disclosure rules (GAAP differences), in management behavior and in the degree of alignment between financial and tax accounting, etc. (Choi and Mueller 1992).

While the aggregate impacts of the aforementioned factors on the value relevance of accounting numbers are difficult to disentangle, we take a capital market approach to the cross-country comparison of accounting summary numbers that allows us to look at the outcome of these impacts from an equity valuation perspective (2). Using the Edwards-Bell-Ohlson (EBO) valuation model (Ohlson 1994) which expresses firm value as a function of book value and earnings, we address the following issues: (1) whether accounting diversity causes systematic differences in the joint value relevance of accounting earnings and book value across these four countries, and (2) whether the relative importance of earnings and book value differ systematically across these countries.

Our results indicate that systematic differences do exist in the value relevance of accounting numbers across these four countries. The joint value relevance of earnings and book value is the highest in the U.S. and Japan, and the lowest in U.K. Similarities in accounting practices and the business environment between the U.S. and Canada do not lead to similarity in the joint value relevance of accounting summary numbers.

In terms of the relative value relevance of accounting earnings and book value, our results suggest that accounting earnings are of greater importance to market participants than book value in the U.S. and in the U.K.; and accounting book value in Japan is more value-relevant than earnings. Accounting book value has higher value relevance than accounting earnings in Canada over our sample period, even though the U.S. and Canada are perceived as having similar accounting principles. We also find that in both the U.S. and Canada, earnings are incrementally value-relevant over book value, and book value has incremental value relevance over earnings. However, this mutual incremental value relevance of earnings and book value is weak and inconsistent over time in the U.K. and Japan. Additionally, using a time trend analysis, we find that the value relevance of accounting numbers in each of these four countries increases over our sample period. This would suggest that the accounting systems of all four countries included in this study have improved over time.

The next section discusses potential contributions to accounting diversity across the countries in this study and is followed by a review of prior literature and a discussion of the empirical models used in this study. Then the results and conclusions of the study are discussed.

ACCOUNTING DIVERSITY

As previously stated, accounting diversity across national borders probably is the result of numerous factors including differences in accounting measurement and disclosure rules (GAAP differences), differences in management behavior, and the degree of alignment between financial and tax accounting. From the perspective of GAAP differences, the accounting models adopted in the four countries in this study differ in the rigidity of their compliance with the historical cost principal. This leads to various degrees of conservatism in the earnings measure and the book value measure across countries.

In Japan, the earnings measure is more conservative than that of other countries. Historical cost accounting is required by the Commercial Code; among the few exceptions to historical cost is when the market value is significantly less than cost and is not expected to recover, or when an unexpected impairment in value occurs for fixed assets. Current cost accounting is not even allowed in supplementary disclosures.

In the U.S. and Canada, accounting measurements primarily follow historical cost accounting; only in limited circumstances (3) can the accrual concept be extended and depart from strict transaction cost measurements. The accounting standards of the U.S. and Canada are very similar and have only minor deviations (4).

In the U.K., however, revaluation of tangible fixed assets and investment is allowed and is shown under fixed assets. The effect of the revaluation is taken directly into a revaluation reserve without going through the income statement. After revaluation, depreciation must be based on the revalued amount; however, it is very common for an amount equal to the depreciation on the revaluation surplus be transferred from the revaluation reserve to the retained earnings account while only the historical cost portion is charged to the income statement (Woolf and Tanna 1988). It is believed that this non-conservative asset valuation convention might create opportunities for management manipulations of accounting numbers and cause book value to have smaller value relevance for market valuation. A prior empirical study by Weetman and Gray (1990) also reports a systematically more 'conservative', or understated, U.S. earnings measure when compared to the U.K. earnings measure using the same set of U.K. companies.

Another aspect of accounting diversity in these four countries involves the behavior of management in response to accounting numbers. For example, prior studies show that the information asymmetry problem is more severe in the U.S. than in Japan due to the close ties between Japanese market participants and the companies in which they invest. As a result, Japanese market participants are less responsive to short-term changes in profitability than U.S. market participants (Jacobson and Aaker 1993) which suggests that Japanese managers are less inclined than U.S. managers to manipulate short-term earnings at the expense of long-term profits.

Differences in the alignment of tax and financial accounting in these countries also contributes to accounting diversity. For example, the alignment of tax and financial accounting in Japan has long been attributed by researchers as one of the reasons that Japanese income numbers are artificially understated (e.g., French and Porterba 1991). The same level of alignment between financial accounting and tax accounting is not seen in the U.S., the U.K. or Canada. Consequently, Japanese accounting numbers are more conservative than those of the other three countries in the study (5).

PRIOR LITERATURE

Numerous studies over the past decade have examined the relation between earnings and stock prices. However, until recently, with the wider acceptance of the EBO valuation model, book value was not included in most of the studies comparing the relation between accounting numbers and stock prices across countries. While prior studies that examine the relationships between stock prices and accounting numbers across different countries provide valuable insight as to how diversity in accounting standards affects the value relevance of earnings, our study focuses on cross-country comparisons of the relationships between stock prices and accounting numbers using a model that incorporates both earnings and book value (6).

We review studies that either use methodology similar to that used in this study or have lines of inquiry similar to those in this study and broadly classify these studies into two categories: (1) studies that compare the value relevance of accounting earnings across countries and (2) studies that compare the joint and incremental value relevance of earnings and book value across countries.

Cross-country Comparison of Accounting Earnings

Prior research by Alford et al. (1993) provides comparisons of the timeliness and usefulness of accounting earnings of firms in the U.S. with that of firms in 17 different countries, including the U.K., Canada, and Japan. Their results indicate that U.K. earnings are more informative than U.S. earnings and that the explanatory power from the earnings-returns regression model is significantly lower in Japan (5.6%) than in the United Kingdom (26.1%), which would indicate that Japanese earnings are not as informative as the earnings of U.K. firms.

Weetman and Gray (1990) found that U.K. earnings measures tend to be systematically higher or less conservative than earnings measured under U.S. accounting principles. Aron (1987) and Cooke (1993) provide evidence that reported earnings of Japanese firms are significantly understated when compared to those of their U.S. counterparts. The empirical evidence suggests that other than quantitative differences in accounting earnings that can result from GAAP differences, the relative information content of accounting earnings also varies across different GAAP regions.

Cross-country Comparison of Accounting Earnings and Book Value

Joo and Lang (1994) use a price model to investigate the relationship between stock prices and accounting earnings in Germany, France and the U.K. both before and after implementation of the EU Fourth Directive. They report that significant differences in several financial ratios (ROE, E/P and B/M) remain in the post directive period. They also report that in a price regression model, firms in Germany and France tend to have larger coefficients estimates for shareholder's equity as well as for earnings than firms in the U.K. They conclude that the evidence supports the balance sheet focus and accounting conservatism found in Germany and France.

Harris, et al. (1994) compare the value relevance of accounting numbers between a matched sample of German firms and U.S. firms using a one variable model (price-to-earnings or price-to-book value). They find little difference in the overall value relevance (R2) of the accounting numbers of German and U.S. firms. However, the coefficients for book value and for earnings of German firms are greater than those of their U.S. counterparts. Hall, et al. (1994) also use a univariate model of price-to-earnings (book value) to examine and compare the relative value relevance of accounting earnings (book value) between the U.S. and Japan. They report lower (higher) association of accounting earnings (book value) to price in Japan as compared to that in the U.S.

More recently, King and Langli (1999) report that systematic differences in the value relevance of accounting numbers exist across Norway, German and the U.K. They also find that U.K. accounting numbers have the highest explanatory power when regressed on stock prices, followed by Norwegian, and then German, accounting numbers. However, they also find that book value is more value relevant in Germany and Norway than in the U.K.

Our study provides empirical evidence that corroborates with prior cross-country comparative studies and also extends prior empirical evidence in that it not only includes countries that have contrasting accounting standards and different degrees of conservatism bias in accounting measurements, but also includes countries that have similar accounting standards and business environments. This enables a fuller appreciation of the impact of accounting diversity on the relative value relevance of accounting measures. Additionally, our study also addresses some methodological issues not addressed in most of the prior cross-country comparative studies, including scale differences of earnings and book value and industry and size heterogeneity of cross-country samples.

MODEL AND HYPOTHESIS

We adopt the Edward-Bell-Ohlson valuation model (EBO, hereafter), which incorporates both accounting earning and book value in the same equity valuation model. The reformulated empirical earnings and book value model is as follows: Price-Earnings/Book Value Model:

[P.sub.it] = [w.sup.e.sub.t] [E.sub.it] + [w.sup.b.sub.t] [B.sub.it] + [[mu].sup.eb.sub.it] (1)

Where,

[P.sup.it] is the price per share of firm i at time t,

[E.sub.it] is the reported earnings per share of firm i at time t,

[B.sub.it] is the reported book value per share of firm i at time t,

[w.sup.e.sub.t] is the empirical Earnings Valuation Coefficient (EVC) (7) in the combined model,

[w.sup.b.sub.it] is the empirical Book-value Valuation Coefficient (BVC) in the combined model, and

[[mu].sup.eb.sub.it] is the error term.

Ohlson (1991, 1992 and 1995) and Feltham and Ohlson (1992) suggest that, theoretically, a firm's value can be determined by its earnings and book value, and that this valuation would be immune to differences in accounting methods across firms or countries given that clean surplus accounting is not systematically violated, i.e., changes in the book value of equity reflect only the income (loss) and owner's investment and dividends. We expect to see systematic differences in the value relevance of accounting measures across our sample countries due to various degrees of violation of the clean surplus assumption and conservatism bias that impede the ability of accounting measures to efficiently capture events that affect firms' value (see Appendix A for a list of clean surplus violations in the four countries).

Since many factors impact the value relevance of accounting numbers produced by each accounting system, e.g., capital providers; taxation and legal systems, etc., and the effect of these factors can not be easily isolated from each other; a capital market approach allows us to look at the outcome of these impacts from an equity valuation perspective. We take a capital market approach for cross-country comparison of accounting summary numbers and draw conclusions on the joint value relevance of accounting earnings and book value using the combined explanatory power of the two accounting numbers for market value (adjusted R2).

We anticipate that countries with similar accounting standards and business environments, i.e., the U.S. and Canada should have similar levels of joint value relevance of earnings and book value, while countries with different accounting standards and business environments, i.e., the U.K. and Japan, should have different levels of joint value relevance of earnings and book value.

The empirical model also enables us to make direct comparisons of the relative importance of earnings and book value in firm valuation as perceived by capital market participants. Prior studies by Collins et al. (1997) using U.S. samples indicate that U.S. accounting book value is more value-relevant when the value relevance of accounting earnings is low. Given this result and evidence of the differential earnings-to-price relation as reported by Alford et al. (1993), we anticipate that accounting book value should be more value-relevant in countries that have relatively low earnings value relevance. We also anticipate that due to the similarity of the accounting principles and business environments in the U.S. and Canada, accounting summary numbers from the U.S. and Canada should have similar levels of value relevance.

Since in most cases, a scale difference that can lead to biased coefficient estimates exists between accounting earnings and book value (Barth and Kallapur 1996), we use a standardized measure of the variables that allows direct contrasts of the earnings and the book value valuation coefficients. The standardization procedures include two steps: (1) all outlier observations that are outside of the three standard deviations range are deleted for every measure for each country from the entire sample (1983-1991); (2) after outliers are deleted, all variables used in the model are standardized by subtracting the new yearly means and dividing by the new yearly standard deviations. The standardized measures have a zero mean and a standard deviation of 1. Based on the standardized measure, the earning valuation coefficient for time t, and the book value valuation coefficient for time t, represent the effect of change in one standardized unit of earnings and book value, respectively, on one standardized unit of firm value. Since the coefficients of earnings and book value both are based on the standardized units, the magnitudes of the earnings and the book value coefficients can be compared directly.

Pooled regressions as well as yearly regression analyses are performed to analyze the joint and relative value relevance of earnings and book value within each country. Trend analyses of the regression coefficients are performed to investigate changes in the relative importance of earnings and book value over time. Additionally, size and industry variables are added individually, as well as collectively, into to the combined model to control for the impact of size and industry membership effects on the empirical results.

DATA AND SAMPLE SELECTION

Our data are extracted from the Global Vantage Industrial/Commercial and Issue Files for the period of 1983-1991. The sample is restricted to non-financial firms listed on the largest exchange in each country to increase the homogeneity of the sample, i.e., the New York Stock Exchange (U.S.), The Tokyo Stock Exchange (Japan), The Toronto Stock Exchange (Canada) and the London Exchange (U.K.).

Our primary focus is on the importance of accounting numbers prepared in accordance with the local accounting standards of each country; thus, only data prepared according to domestic accounting standards are used. A sample firm is considered as belonging to a certain country only if it is both incorporated and listed on the exchange of the same country. The purpose of this restriction is to eliminate the "foreign" listing firms from the samples of each country (8).

Observations are excluded if the firm had a change in fiscal year-end. A firm-year observation also is excluded from the sample if data are unavailable to calculate all the necessary variables. After deleting the outliers outside of the three standard deviations range for every measure for each country from the entire sample, the final sample consists of 4,187, 5,136, 2,224 and 3,645 firm-year observations for the U.S., the U.K., Canada and Japan respectively (9).

Table 1 provides the descriptive statistics of earnings per share, book value per share, price per share, total assets, and market value. Since the distributions of the variables are asymmetric, the median is used to compare the magnitude of the variables across countries. Among the four countries, the U.S. has the largest earnings per share as well as book value per share, followed by those of the Canadian, the U.K. and the Japanese samples. The U.S. sample also has the largest price per share followed by those of the Canadian, Japanese and U.K. samples. Regarding firm size (measured by total assets or total market value), the Japanese sample has the largest firm size followed by those of the U.S., Canadian and U.K. samples. While all variables are right-skewed, the Japanese sample has much larger earnings, book value and price per share than those of the samples of the other countries; the U.S. and U.K. firms have much larger total assets and total market value than those of the other countries.

Table 2 reports the pooled correlation analyses of standardized earnings, book value and price for each country. As expected, accounting earnings and book values in all four countries are positively correlated with price, indicating that accounting earnings and book value are important variables in explaining firm values in all four countries.

EMPIRICAL RESULTS

Cross-country Comparison of Value Relevance

Table 3, panel A reports the pooled and the yearly regression results of the combined earnings and book value regression model with standardized dependent and independent variables. As indicated before, the use of standardized variables in the regression analyses mitigates the scale differences and allows direct comparisons of the EVC (earnings valuation coefficient) and the BVC (book value valuation coefficient).

The pooled results show the cross-sectional average; however, to mitigate the potential impact of cross-sectional correlations as indicated by Bernard (1987), and our conclusions are drawn mainly from the mean analysis of the yearly regression results. The coefficients of our regression models are discussed first, followed by a discussion of explanatory power of the models.

The yearly regressions analyses show that the EVC and the BVC are both significant in the U.S. over the period encompassing 1982-1991. However, while the coefficients of earnings in the U.K. are significant in all nine years of our sample period, the coefficients of book value are only significant in four out of the nine years. Mean analysis of the yearly coefficients show that for the U.S. firms, both earnings and book value are value-relevant, while in the U.K. only earnings are value-relevant.

Contrary to the results of the U.S. and of the U.K. samples, the yearly regression coefficients of the book values of the Canadian and the Japanese samples are significant in all nine years, while the coefficients of earnings for these two countries are less stable over time. For the Canadian sample, the EVCs are significant in seven of the nine years (1985 and 1989 are insignificant); and for the Japanese sample, the earnings are significant only in five of the nine years (1983, 1985, 1986 and 1991 are insignificant). The mean analysis indicates that the Canadian EVC and BVC are both significant, while only the BVC is significant in Japan.

These results suggest that earnings and book value are both value-relevant for the U.S. market and the Canadian market; however, U.S. earnings is a more important variable in the pricing decision than book value while book value is more important than earnings in Canada. Either earnings or book value , not both, are important in the pricing decision in the U.K. and Japan. Earnings are important in U.K. while book value is not, while in Japan, book value is the dominating variable while earnings is not important. This also is consistent with our belief that earnings (book value) will be the more value-relevant of the two accounting numbers in countries whose financial accounting approach is earnings (book value) focused. However, Canadian accounting book value is more value-relevant than earnings during the sample period, which suggests that Canadian accounting standards that affect equity measures may be materially different from those of the U.S.

In terms of the ability of accounting numbers to explain stock prices, the explanatory power of the model (adjusted R2) in table 3 shows that both U.S. and Japanese earnings and book value are able to explain more than 60% of price variation (average of the yearly R2) while Canadian accounting summary numbers explain 57 % of price variation, and U.K. earnings and book value explain about 41% of stock price variation. This indicates that, on average, accounting numbers in the U.S. and Japan capture more information that is correlated with price variations (more value relevant) than Canadian and U.K. numbers, and that the overall information contained in the earnings and book value of the U.S. and the Japanese firms seem to be more aligned with the information market participants use in firm valuation compared to the accounting numbers in Canada and the U.K.

We perform detailed analyses of the earnings (EVC) and the book value (BVC) regression coefficients with the following measures: (1) the Sum, which is the sum of the earnings and book value coefficients in the combined model and represents the overall valuation weight market participants assign to each country's accounting summery numbers; (2) the Difference, which is the difference in magnitude of the valuation coefficients EVC and BVC and represents the difference in the valuation weights of earnings and book value; and (3) the Ratio, which is the difference between the EVC and BVC scaled by the magnitude of the Sum and enables cross-country comparison of the relative valuation importance of earnings and book value.

With the standardized earnings, book value and price, the regression coefficients do not suffer the bias cause by the scale differences between earnings and book value within each country and cross the four countries. We are able to make direct compare of the valuation weights (coefficients) market participants in each country assign to accounting variables. A larger valuation coefficient indicates a higher weight market participants assign to per standardized unit of earnings (book value). Table 3, panel B reports that, on average, the combined valuation weight of the earnings and of the book value regression coefficients (Sum) is the highest for the U.S. (0.86), followed closely by Japan (0.82) and Canada (0.80). The Sum for U.K. earnings and book value is the lowest at 0.64. In addition, the magnitude of the Sum increases over time for all four countries, which suggests that the valuation weights market participants place on the two accounting summary numbers for firm valuation purposes has increased over the sample time period for these four countries.

The Difference measure, which is a measure of the difference in the relative valuation weight of earnings and book value, is used to examine the relative importance of the earnings and book values. A positive (negative) measure of the Difference variable indicates that the earnings (book value) coefficient has a stronger response than the book value (earnings) coefficient, indicating that market participants perceive earnings (book value) to be relatively more important than accounting book value (earnings) in firm valuation. This measure provides an ordinal measure of the additional importance of one variable over another and can facilitate cross-country comparisons. The Difference measure from the pooled regression is positive for both the U.S. (0.19) and the U.K. (0.49), and negative for Canada (-0.33) and Japan (-0.31). The means of the yearly Difference measures are 0.21, 0.65, -0.26 and -0.37 for the U.S., the U.K., Canada and Japan, respectively. These results suggest that accounting earnings are more important to market participants in the U.S. and the U.K., while the Japanese and the Canadian market participants perceive book value to be more important.

This finding is inconsistent with our prior belief that countries with similar accounting standards should produce earnings and book value with similar valuation implications. In particular, the difference between the earnings valuation focus of the U.S. market and the book value valuation focus of the Canadian market is unexpected.

To enable cross-country comparisons of the extent that accounting earnings (book value) are more value-relevant than book value (earnings), the Ratio measure is used as reported in table 3, panel B. Since the magnitude of the Difference between the EVC and the BVC can be influenced by the overall magnitudes of these two variables across countries, scaling the Difference by the Sum enables unbiased comparisons across different countries.

On average (mean analysis), the U.S. has the smallest difference between the value relevance of earnings and book value (25%), while the U.K. (101%) exhibits the largest difference between the value relevance of earnings and book value. This result suggests that the U.S. market considers earnings and book value to be of somewhat equal importance in the valuation process, while U.K. market participants place much more importance on earnings compared to book value. Both Canadian and Japanese ratio measures are negative, and the results indicate that the difference in the market responses to book value and earnings is greater in Japan (-47%) compared to that in Canada (-32%). These results suggest that an even stronger book value focus exists in the capital markets in Japan compared to that in Canada.

Yearly comparison of the Ratio measure across the four countries is generally consistent with the ranking from the mean analyses. The U.K., in most years, has the largest Ratio among the four countries, followed by the U.S. However, during our sample period, the U.K.'s Ratio measure decreased substantially in the last two years (1990, 1991). This decrease is apparently due to the drastic increase of BVC in 1990 and 1991 in the U.K. while the EVC level remained stable. This change is interesting since it suggests that the value relevance of accounting book value increased after 1990 which coincides with the establishment of the Accounting Standards Board in the U.K. in response to criticism of the "creative accounting" practices of the 1980s.

The creation of the Accounting Standards Board (ASB) in the U.K. is a very significant event in the development of British accounting standards and practices because one of the primary foci of the ASB is to improve the quality of financial statements. While we cannot infer a causal relationship between that particular event and our empirical results, it is evident that beginning in 1990, British balance sheet numbers are more value-relevant than those of the previous seven years in our sample.

Canada and Japan Ratio measures are generally negative (BVC > EVC). Canadian Ratios become more negative toward the latter part of our sample period, while the Japanese Ratios become less negative (except in 1991) over our sample period. These results may indicate that changes in both the overall and the relative valuation weights of accounting earnings and accounting book value have occurred over time in these countries. A trend analysis is conducted to verify whether time trends do exist for the Sum, Difference and Ratio measures of each country.

Trend Analysis

Trend analysis is performed by regressing the time (years 1 through 9) on each of the variables of interest. Table 4 reports the results of trend analyses for the Sum, Difference and Ratio variables. The results confirm that over time, a significant positive trend exists for the combined valuation weights of earnings and book value (Sum) for the four countries. This trend is strongest for the U.K. (0.056), followed by Canada (0.027), Japan (0.023) and the U.S. (0.013), and it implies that reported earnings and book value are becoming more aligned over time with the underlying information used by market participants in the market valuation process. Collins, Maydew and Weiss (1997) report a similar finding with U.S. data over a 40-year sample period.

For the Difference measure, only the Canadian sample shows a significant negative time trend (-0.105), i.e., the Difference measure becomes more negative (book value becomes more value-relevant over time compared to earnings) over time. Since the time trend for the Difference measure can be affected by the time trend of the overall value relevance, we further analyze the time trend of the Ratio measure. The trend analysis of the Ratio measure shows that both the U.K. and Canadian ratios have a significantly negative time trend, which indicates that for these two countries, accounting book value becomes increasing more important to the market relative to earnings over the sample period. However, as shown by the Difference measure and the Ratio measure in table 3, panel B, earnings in the U.K. is still more important to the market than the book value, even though book value shows a significant increase in its importance over time. On the other hand, accounting book value in Canada remains as the most important valuation variable, and increasingly gains in importance over time.

Size and Industry Controls

To verify that the regression results are not driven by either size or industry factors, additional analyses, which control for size (total asset) and industry factors are performed. Table 5, panel A reports the pooled and the yearly regression results after adding size as an additional independent variable. The results shows that while size is a significant variable in both the pooled and the yearly regression analyses, the results remain qualitatively the same as those reported in Table 3, panel A. However, the regression model explanatory power (adjusted R2) improves after adding the size control.

Table 5, panel B reports the results of size control on the Sum, the Difference and the Ratio measures. The results also confirm that even though size is a significant factor in explaining firm value for our samples, our results are robust to firm size.

Since previous studies have found industry membership to be an important factor in firm valuation, additional analysis is performed to control for both industry membership and size. This additional analysis attempts to control for the differences in industry composition across the four countries by grouping observations of each country by their two-digit SIC codes. For each industry group of each country, we randomly select an equal number of observations. The number of observations selected for each industry in each country is based on the minimum total number of observations in that industry group across countries, to ensure that the number of observations in each industry are the same for every country.

The final industry controlled sample, as listed in appendix A, consists of 1,311 observations for each country across 38 industries. As can be seen from appendix A, since the Canadian (and in some cases, the Japanese) sample has the lowest number of observations in most industries, more observations per industry are removed from the U.S. and the U.K. samples.

Table 6, panels A and B reports the results of the regression analyses for the industry-controlled samples with size as an additional control variable. These results, in general, indicate that after controlling for both size and cross-country differences in industry composition, our primary conclusions remain unchanged: earnings in the U.S. and the U.K. are more relevant or firm valuation than book value, and book value in Canada and Japan are more relevant than earnings. However, the magnitude of the book value coefficients for the U.K. and Japanese samples after the industry compositions are changed. The U.K. book value coefficients are higher for the industry-controlled sample (0.35) while the magnitude of the earnings coefficients remains relatively unchanged. The Japanese book value coefficient also increases (0.87), while the valuation coefficient of Japanese earnings decreases (-0.11). The EVC and BVC for U.S. and Canada remain at levels that are similar to the original analysis. These differences in coefficient magnitude are caused by differences in sample compositions and suggest industry differences do exist in the magnitudes of the valuations coefficients; however, within each country, the relative value relevance of earnings and book value is unchanged by the industry controlled. Earnings remain to be more value relevant than book value for the U.S. and the U.K. and reverse is true for Canada and Japan. Yearly coefficients of earnings and book value also remain consistent with results reported for the original sample. The model explanatory power (adjusted R2) is higher for the U.K. industry-controlled sample compared to the original sample (0.54 vs. 0.41), but remains largely unchanged for the other countries. However, the explanatory power of the model for U.K. industry-controlled sample remains the lowest of the four countries. In fact, the ranking of the R2s across the four countries remains unchanged after controlling for both size and industry.

CONCLUSIONS

This study provides empirical evidence that systematic differences in the relative importance of earnings and book value do exist across national boundaries by using a standardized combined-earnings-and-book-value regression model. The standardized procedure controls for the scale difference of earnings and book value both within and across countries, thus allowing analyses of the relative value relevance of earnings and book value at various levels.

Our results show that both earnings and the book value are jointly value-relevant in the firm valuation processes of the U.S. and Canadian markets. However, accounting earnings appears to be more important than book value for the U.S. markets while book value is more important than earnings in Canadian markets. Given the similarity of the U.S. and Canadian accounting standards, these results are interesting and deserve future research.

While both earnings and book value are important in the U.S. and Canada, we provide evidence that the U.K. market focuses on earnings and that the Japanese market focuses on book value only. These results are robust when size and industry controls are added.

Comparison of the overall value relevance of earnings and book value among these four countries indicates that the joint value relevance of earnings and book value are at similar levels for U.S. and for Japan. These two countries also have the highest levels of joint value relevance of earnings and book value across the four countries while U.K. earnings and book value consistently have the lowest joint value relevance.

Our results also indicate that the U.K. market differentiates the two accounting summary numbers the most followed by Canada, Japan and the U.S. as indicated by the differences in the value relevance of earnings and book value across the four countries. Additional trend analyses reveals that time trends do exist over the sample period for the overall, as well as the relative, value relevance of accounting earnings and book value across these four countries. Our results show an overall increasing trend in the overall value relevance of earnings and book value over time for all four countries, which supports the notion that reported earnings and book value have become more aligned over time with the underlying information used by market participants in the market valuation process.

Assuming that the valuation relevance of accounting numbers is an important aspect of their qualities (10), these results indicate that there has been an improvement in the quality of accounting numbers over our sample period for these four countries. And of the four countries, the value relevance of the U.K. has the strongest increasing time trend followed by that of Canada, Japan and the U.S.

Our results also show a time trend in the relative importance of accounting earnings and book value for the Canadian and U.K. samples which indicates that during the time period of 1983 through 1991, both Canadian and U.K. book values significantly gained in importance. However, earnings is still the dominating valuation variable in the U.K. despite the increase in the value relevance of book value. The reasons for these time trends, and the underlying causes of the change in the relative importance of the two accounting summary numbers deserves further research.

This study provides systematic comparison as to the relative value relevance of the two most widely used accounting summary numbers. Evidence from this study enhances understanding of how the accrual accounting summary numbers can be perceived by capital market participants as having very different valuation importance when they are produced by different accounting systems. It also provides important insight as to how the relative valuation relevance of accounting earnings and book value has changed over time both within and across these four countries.

With the significant increase of direct foreign investments, the findings of our study have important implications for the use of accounting summary numbers in the international investment context. The investor today has the opportunity to invest not only in domestic firms, but also in firms trading in foreign markets. Since firms whose stock trades in these foreign markets use the domestic GAAP standards of their countries, it is important for investors to be aware of how foreign markets use the accounting numbers produced under domestic GAAP for firm valuation purposes. For example, this study shows that U.K. markets view the earnings of U.K. firms as an important valuation variable, while book value is not viewed as such. Consequently, investors desiring to invest directly in U.K. firms trading in U.K. markets need to be aware of this fact. Similarly, investors wishing to invest directly in Canadian or Japanese firms traded on their domestic stock markets need to be aware of the fact that their markets view book value as a more important valuation variable than earnings. And although Canadian markets do seem to use earnings in the firm valuation process, Japanese markets do not appear to do so.

Considering the intensification of international accounting standard setting activities, our study should also be of interest to the accounting standard setting bodies in the U.S., the U.K., Canada, and Japan. Our results provide base-line measures of value relevance that can be used by the standard setting bodies of these four countries to assess the impact of new accounting standards on the usefulness of accounting information developed under their GAAP systems. The results of our study also provide information regarding the evolution of the value relevance of the accounting numbers in the four countries in our study over time.

Limitations of our study include the fact that we only investigate the value relevance of accounting numbers from the perspective of the stock markets. Many different uses of accounting information exist and the relevance of accounting numbers probably differs for each use. Our study only examines the value relevance of earnings and book value for a limited period of time, nine years. The extension of this time frame may provide more insight into the evolution of the value relevance of these accounting numbers.

We examine the value relevance of the summary accounting numbers of four countries only. Examining the value relevance of the accounting numbers of other countries would provide valuable information not only to the international investment community, but also to the GAAP setting bodies of those countries.

REFERENCES

Alford, A., J. Jones, R. Leftwich, and M. Zimjewski (1993). The Relative Informativeness of Accounting Disclosures in Different Countries. Journal of Accounting Research, Supplement.: 183-223

Aron, Paul (1987) Japanese Price Earnings Multiples: Refined and Updated. Daiwa Securities American, Inc. Japanese Research Report #30.

Barth, Mary E. and Sanjay Kallapur (1996). The Effects of Cross-Sectional Scale Differences on Regression Results in Empirical Accounting Research Contemporary Accounting Research, Fall, Vol. 13, Iss. 2; pg. 527-568.

Bernard, V. L. (1987). Cross-Sectional Dependence and Problems in Inference in Market-Based Accounting Research. Journal of Accounting Research. (Spring): 1-48.

Bernard, V. L. (1993). Accounting-Based Valuation Methods: Evidence on the Market-to-Book Anomaly and Implications for Financial Statements Analysis. Working Paper.

Chan, L. C., Y. Hamao and L. Jesef (1991). Fundamentals and Stock Returns in Japan. Journal of Finance (December): 1739-1764.

Cooke, T.E. (1993). The Impact of Accounting Principles on Profits: The US versus Japan. Accounting and Business Research: 460-476.

Choi and Mueller (1992). International Accounting, 2nd Ed., Prentice-Hall.

Cheng, C. S. A.; W. S. Hopwood. and J. C. McKeown (1992). Non-Linearity and Specification Problems in Unexpected Earnings Response Regression Model. Accounting Review (July): 579-599.

Collins, D., Maydew, E. and I. Weiss (1997). Changes in the Value-Relevance of Earnings and Book Values over the Past Forty Years. Journal of Accounting and Economics 24: 39-67.

Easton, P. and T. Harris (1991). Earnings as and Explanatory Variable for Returns. Journal of Accounting Research (Spring): 19-36.

Fama, E. and K. French (1992). The Cross-Section of Expected Stock Returns. The Journal of Finance (June): 427-465.

Feltham, G. A. and J. A. Ohlson (1992). Valuation and Clean Surplus Accounting for Operating and Financial Activities. working paper, University of British Columbia (June).

French and Porterba (1991). Were Japanese Stock Prices Too High? Journal of Financial Economics 29 : 337-363. Gray, S. J. 1980. The Impact of International Accounting Differences from a Security-Analysis Perspective: Some European Evidence. Journal of Accounting Research (Spring): 64-76.

Hall, C., Y. Hamao and T. Harris (1994). A Comparison of Relations Between Security Market Prices, Returns and Accounting Measures in Japan and the United States. Journal of International Financial Management and Accounting: 47-73.

Harris, Travor S., Mark Lang, and Hans Peter Moller (1994). The Value Relevance of German Accounting Measures: An Empirical Analysis. Journal of Accounting Research, Autumn: 187-209.

Joos, P. and M. Lang (1994). The Effects of Accounting Diversity: Evidence from the European Union. Journal of Accounting Research (Supplement): 141-175.

King, Raymond D. and John Langli (1999). Accounting Diversity and Firm Valuation. International Journal of Accounting.

Meek, G. (1991). Capital Market Reactions to Accounting Earnings Announcements in an International Context. Journal of International Financial Management and Accounting: 93-109.

Modigliani, F. and M. Miller (1961). Dividend Policy, Growth and the Valuation of Shares. Journal of Business: 411433.

Mueller, Gerhard G. (1968). Accounting Principles Generally Accepted in the U.S. versus Those Accepted Elsewhere. The International Journal of Accounting, Spring: 91-104.

Nobes, Christopher (1994). International Guide to Interpreting Company Accounts--overcoming disparities in national accounting procedures. Financial Times Management Reports: FT Business Enterprise Ltd.

Ohlson, James A (1990). A Synthesis of Security Valuation Theory and the Role of Dividends, Cash Flows, and Earnings. Contemporary Accounting Research, Spring: 648-676.

Ohlson, J. (1991). The Theory of Value and Earnings, and an Introduction to the Ball-Brown Analysis. Contemporary Accounting Research, Fall: 1-19.

Ohlson, J. (1995). Earnings, Book Value and Dividends in Security Valuation. Contemporary Accounting Research, Spring: 661-687.

Ohlson, J. A., and S. Penman (1992). Disaggregated Accounting Data as Explanatory Variables for Returns. Journal of Accounting Auditing and Finance: 553-573.

Pope, P. and W. Ree (1993). International differences in GAAP and the Pricing of Earnings. Journal of International Financial Management and Accounting : 190-219.

Ryan, S. (1991). Historical Cost Accrual Methods and the Role of Book Value and Earnings in Linear Valuation Models. Journal of Accounting Auditing and Finance : 257-287.

Sakakibara, S., H. Ymaji, H. Sakurai and K. Shiroshita (1988). The Japanese Stock Market. New York.

Woolf, E. and S. Tanna (1988). Understanding Accounting Standards. The Institute of Chartered Accountants in England and Wales.

Weetman, P. and S. Gray (1990). International Financial Analysis and Comparative Corporate Performance: The Impact of U.K. vs. U.S. Accounting Principles on Earnings. Journal of International Financial Management and Accounting: 110-30.

ENDNOTES

(1) The capital markets in these four countries encompass more than 95% of the global market capitalization.

(2) A capital market approach investigating the value relevance of accounting summary numbers is a joint test of the value relevance of the accounting numbers and the efficiency of the information environment, especially in a cross-country comparative study, which would depend on the relative efficiency of the capital markets in responding to accounting information. Prior studies by Kamarotou and O'Hanlon (1989) and Chan et al. (1997) suggest that capital markets of these four countries are at least semi-strong form efficient.

(3) Examples of departures from historic cost include when present value techniques are used to measure certain long-term receivables and payables, when certain contingent liabilities are recognized, when post-retirement benefits of employees are brought to account, and when the "mark-to-market" rule is applied to certain investments.

(4) According to the Report on the Financial Accounting Practices in North America published by the U.S. Financial Accounting Standard Board, areas of differences between U.S. GAAP and Canadian GAAP include: (1) Foreign Currency Translation and Transactions: Canadian GAAP allows the impact of foreign exchange rate movements on non-monetary liabilities to be capitalized and amortized over future periods; (2) Extraordinary items: Canadian GAAP is more liberal regarding extraordinary items;(3) Interest Capitalization: Canadian GAAP does not require (it is an option) the capitalization of interest on certain qualifying fixed assets under construction and (4) Extractive Industries Accounting: Canadian GAAP accounts for exploration costs using the full cost method.

(5) Cooke (1993) reports evidence that the profits of companies measured in accordance with Japanese GAAP are considerably more conservative than if prepared in accordance with U.S. GAAP.

(6) Prior research suggests that there may be a differential and complementary relationship between earnings and book value in firm valuation. Specifically, when there is a decline in the value relevance in accounting earnings, accounting book value increases in its value relevance (Barth et al. 1997; Berger et al. 1996 and Collins et al. 1997).

(7) The term "earnings valuation coefficient" has not been formally used or defined in the accounting literature. Here, for simplicity, we use the terms EVC and BVC (book value valuation coefficient) to represent the regression coefficients.

(8) Due to data availability, multiple-listing firms, i.e., firms that are listed concurrently on several exchanges, are not identified. However, since the firms included in the sample are limited to those that use domestic accounting standards, the inclusion of multiple-listing firms should not contaminate the interpretation of the results.

(9) Parent-only numbers are used for the Japanese sample due to data limitations. A prior study has shown that consolidated earnings in Japan have different valuation implication than the parent-only results. The inferences that can be drawn from our results are limited to the data used.

(10) Lev (1989) used the R2 from earnings information content studies to infer the quality of accounting earnings.

Kathy H.Y. Hsu, University of Louisiana at Lafayette

Harlan L. Etheridge, University of Louisiana at Lafayette

Table 1: Descriptive Statistics

Varaibles MEAN STD MIN P1

US (total number of observations is 4,187 with 424, 383, 516,
458, 623, 523, 408, 403, and 449 for 1982 to 1991 respectively.)

Earnings 2.96 4.48 -53.23 -8.72
Per Share (-12.5) (-2.6)

Book Value 27.41 25.61 -120.50 -3.60
Per Share (-5.8) (-1.2)

Price Per 46.67 54.14 0.00 0.91
Share (-0.9) (-0.8)

Total Assets 2.41 7.24 0.00 0.02
 (-0.3) (-0.3)

Total Market 1.39 3.69 0.00 0.01
Value (-0.4) (-0.4)

UK (total number of observations is 5,136 with 503, 528, 550,
593, 624, 616, 591, 579, and 552 for 1982 to 1991 respectively.)

Earnings Per Sh 0.51 0.70 -7.48 -0.76
 (-11.4) (-1.8)

US (total number of observations is 4,187 with 424, 383, 516,
458, 623, 523, 408, 403, and 449 for 1982 to 1991 respectively.)

Book Value 3.78 4.93 -15.70 0.01
Per Share (-3.9) (-0.8)

Price 5.69 5.92 0.04 0.18
Per Share (-1.0) (-0.9)

Total Assets 0.88 3.07 0.00 0.00
 (-0.3) (-0.3)

Total Market 0.62 1.97 0.00 0.00
Value (-0.3) (-0.3)

Varaibles P10 MED P90 P99

US (total number of observations is 4,187 with 424, 383, 516,
458, 623, 523, 408, 403, and 449 for 1982 to 1991 respectively.)

Earnings -0.75 2.38 7.78 17.53
Per Share (-0.8) (-0.1) (1.1) (3.3)

Book Value 4.49 22.49 55.88 113.72
Per Share (-0.9) (-0.2) (1.1) (3.4)

Price Per 7.13 31.50 99.50 270.00
Share (-0.7) (-0.3) (1.0) (4.1)

Total Assets 0.08 0.52 5.45 26.73
 (-0.3) (-0.3) (0.4) (3.4)

Total Market 0.05 0.33 3.41 15.53
Value (-0.4) (-0.3) (0.5) (3.8)

UK (total number of observations is 5,136 with 503, 528, 550,
593, 624, 616, 591, 579, and 552 for 1982 to 1991 respectively.)

Earnings Per Sh 0.05 0.35 1.21 3.01
 (-0.7) (-0.2) (1.0) (3.6)

US (total number of observations is 4,187 with 424, 383, 516,
458, 623, 523, 408, 403, and 449 for 1982 to 1991 respectively.)

Book Value 0.55 2.34 8.25 22.88
Per Share (-0.7) (-0.3) (0.9) (3.9)

Price 0.92 3.78 12.79 29.07
Per Share (-0.8) (-0.3) (1.2) (4.0)

Total Assets 0.03 0.16 1.97 10.73
 (-0.3) (-0.2) (0.4) (3.2)

Total Market 0.02 0.12 1.41 9.16
Value (-0.3) (-0.3) (0.4) (4.3)

Varaibles MAX SKEW KURTO

US (total number of observations is 4,187 with 424, 383, 516,
458, 623, 523, 408, 403, and 449 for 1982 to 1991 respectively.)

Earnings 39.67 0.71 12.88
Per Share (8.2)

Book Value 463.38 3.24 32.11
Per Share (17.0)

Price Per 1012.50 4.50 40.96
Share (17.8)

Total Assets 168.26 10.26 157.31
 (22.9)

Total Market 75.61 9.61 139.97
Value (20.1)

UK (total number of observations is 5,136 with 503, 528, 550,
593, 624, 616, 591, 579, and 552 for 1982 to 1991 respectively.)

Earnings Per Sh 11.89 2.28 32.51
 (16.3)

US (total number of observations is 4,187 with 424, 383, 516,
458, 623, 523, 408, 403, and 449 for 1982 to 1991 respectively.)

Book Value 92.36 5.27 54.29
Per Share (18.0)

Price 55.91 2.58 9.79
Per Share (8.5)

Total Assets 56.77 10.75 146.61
 (18.2)

Total Market 37.31 8.75 104.17
Value (18.7)

Table 2: Correlation Analysis

 US (N=4187) UK (N=5136)

 Earnings Book Value Earnings Book Value

Panel A: Pearson Correlation Coefficients

Price 0.74 0.68 0.62 0.49
Earnings 0.66 0.74

Panel B: Spearman Correlation Coefficients

Price 0.78 0.77 0.80 0.69
Earnings 0.73 0.74

 Canada (N=2224) Japan (N=3645)

 Earnings Book Value Earnings Book Value

Panel A: Pearson Correlation Coefficients

Price 0.64 0.73 0.71 0.77
Earnings 0.72 0.82

Panel B: Spearman Correlation Coefficients

Price 0.65 0.74 0.69 0.77
Earnings 0.70 0.79

Note: The correlation analyses are based on the pooled observations
of the standardized measures. The standardized measures are
generated by substracting the means from the original measures and
divided by the standard deviation. The standardizing procedures are
applied for each year and for each country. All the correlation
coefficients are significant at the 1% level.

Table 3: Regression Results

Panel A: Adjusted [R.sup.2] and Coefficient Estimates

 US (N=4187)

 Adjusted
Year [R.sup.2] Earnings Book Value

Pooled 0.61 0.52 * 0.34 *
83 0.50 0.47 * 0.31 *
84 0.66 0.64 * 0.22 *
85 0.55 0.41 * 0.41 *
86 0.60 0.44 * 0.42 *
87 0.64 0.51 * 0.34 *
88 0.67 0.68 * 0.17 *
89 0.63 0.53 * 0.35 *
90 0.70 0.62 * 0.31 *
91 0.62 0.53 * 0.37 *
Mean Analysis (a) 0.62 0.54 * 0.32 *
Std 0.09 0.08
t-statistics 17.36 11.71

 UK (N=5136)

 Adjusted
Year [R.sup.2] Earnings Book Value

Pooled 0.38 0.56 * 0.07 *
83 0.23 0.57 * -0.11
84 0.31 0.84 * -0.36 *
85 0.20 0.36 * 0.12
86 0.34 0.73 * -0.18 *
87 0.46 0.88 * -0.27 *
88 0.56 0.74 * 0.01
89 0.50 0.69 * 0.03
90 0.53 0.45 * 0.38 *
91 0.57 0.53 * 0.34 *
Mean Analysis (a) 0.41 0.64 * 0.00
Std 0.18 0.255495
t-statistics 10.93 -0.052186

 Canada (N=2224)

 Adjusted
Year [R.sup.2] Earnings Book Value

Pooled 0.57 0.24 * 0.56 *
83 0.44 0.34 * 0.40 *
84 0.50 0.54 * 0.24 *
85 0.35 0.22 0.41 *
86 0.48 0.39 * 0.34 *
87 0.62 0.30 * 0.52 *
88 0.64 0.27 * 0.57 *
89 0.70 0.10 0.76 *
90 0.71 0.19 * 0.72 *
91 0.73 0.11 * 0.80 *
Mean Analysis (a) 0.57 0.27 * 0.53 *
Std 0.14 0.20
t-statistics 5.86 7.99

 Japan (N=3645)

 Adjusted
Year [R.sup.2] Earnings Book Value

Pooled 0.61 0.25 * 0.56 *
83 0.53 0.14 0.61 *
84 0.52 0.29 * 0.47 *
85 0.41 0.00 0.64 *
86 0.44 0.04 0.63 *
87 0.64 0.31 * 0.54 *
88 0.82 0.41 * 0.52 *
89 0.76 0.47 * 0.43 *
90 0.78 0.57 * 0.34 *
91 0.75 -0.32 * 1.09 *
Mean Analysis (a) 0.63 0.21 0.59
Std 0.28 0.213372
t-statistics 2.30 8.232867

* Implies that the coefficient value is significant at the 1%
significant level.

(a) Mean Analysis of the coefficients value are applied to the
simple mean of the coefficient, the t-tests are applied using
nine years of coefficient values.

Panel B: Sum and Difference of the Coefficient Estimations

 US (N=4187)

 Difference Ratio
Year Sum (a) (b) (c)

Pooled 0.86 0.19 22%
83 0.78 0.16 20%
84 0.86 0.42 49%
85 0.82 -0.01 -1%
86 0.87 0.02 2%
87 0.86 0.17 20%
88 0.85 0.50 59%
89 0.88 0.18 21%
90 0.93 0.31 33%
91 0.90 0.15 17%
Yearly Averages (d) 0.86 0.21 25%

 UK (N=5136)

Year Sum Difference Ratio

Pooled 0.64 0.49 78%
83 0.46 0.68 146%
84 0.48 1.21 252%
85 0.48 0.24 50%
86 0.55 0.92 165%
87 0.61 1.15 188%
88 0.75 0.73 97%
89 0.72 0.65 91%
90 0.83 0.08 9%
91 0.87 0.19 21%
Yearly Averages (d) 0.64 0.65 101%

 Canada (N=2224)

Year Sum Difference Ratio

Pooled 0.80 -0.33 -41%
83 0.74 -0.06 -8%
84 0.77 0.30 39%
85 0.62 -0.19 -31%
86 0.73 0.05 7%
87 0.82 -0.22 -26%
88 0.84 -0.31 -36%
89 0.85 -0.66 -78%
90 0.91 -0.53 -58%
91 0.90 -0.69 -77%
Yearly Averages (d) 0.80 -0.26 -32%

 Japan (N=3645)

Year Sum Difference Ratio

Pooled 0.82 -0.31 -38%
83 0.75 -0.47 -63%
84 0.76 -0.18 -23%
85 0.64 -0.64 -100%
86 0.67 -0.59 -87%
87 0.85 -0.23 -27%
88 0.93 -0.12 -12%
89 0.90 0.04 4%
90 0.91 0.23 25%
91 0.78 -1.41 -182%
Yearly Averages (d) 0.80 -0.37 -47%

(a) Sum is the sum of EVC and BVC for the yearly regressions.

(b) Difference is (EVC-BVC) for the yearly regressions.

(c) Ratio is computed by scaling the difference by the sum,
i.e. (EVC-BVC) / (EVC+BVC).

(d) Yearly averages are calculated based on the nine year
observations.

Table 4: Trend Analysis of the Explanatory Power of Earnings
and Book value Combined and the Sum, Difference, and the
Ratio of the Coefficient Estimations

Model: [y.sub.t] = a + b [year.sub.t] + [e.sub.t]

 [R.sup.2] Adjusted Sum Adjusted
 b [R.sup.2] b [R.sup.2]

U.S. 0.0138 0.285 0.013 0.645
 (0.0800) * (3.941) *

U.K. 0.0473 0.791 0.056 0.938
 (0.0008) * (11.073) *

CANADA 0.0441 0.763 0.027 0.598
 (0.0013) * (3.588) *

JAPAN 0.0457 0.585 0.023 0.249
 (0.0099) * (1.913) *

 Difference Adjusted Ratio Adjusted
 b [R.sup.2] b [R.sup.2]

U.S. 0.008 -0.123 0.007 -0.133
 (0.351) -0.252

U.K. (0.079) 0.172 -0.203 0.39
 (1.631) -(2.475) *

CANADA (0.105) 0.723 -0.117 0.646
 (4.672) * -(3.951) *

JAPAN (0.012) -0.138 -0.008 -0.142
 (0.176) -0.093

The trend analysis is performed by regressing the time on the
yearly sum (differences or ratio). There are nine observation
for each country.

* indicates that the coefficient is significant at 10%
significant level.

The number in the parentheses is the t-statistics.

[R.sup.2] is the adjusted [R.sup.2] of the earnings
and book value regress on price.

Sum is the sum of EVC and BVC for the yearly regressions.

Difference is (EVC-BVC) for the yearly regressions.

Ratio is computed by scaling the difference by the sum,
i.e. (EVC-BVC) / (EVC+BVC).

Table 5: Regression Results for the Prices Model with Size Control

Panel A: Adjusted [R.sup.2] and Coefficient Estimates

 US (N=4187)

Year Adjusted Earnings Book Size
 [R.sup.2] Value

Pooled 0.62 0.50 * 0.32 * 0.10 *
83 0.50 0.47 * 0.31 * 0.00
84 0.66 0.64 * 0.21 * 0.03
85 0.56 0.39 * 0.37 * 0.14
86 0.61 0.43 * 0.41 * 0.07 *
87 0.67 0.49 * 0.31 * 0.14 *
88 0.67 0.67 * 0.16 * 0.08
89 0.65 0.48 * 0.35 * 0.14
90 0.72 0.55 * 0.30 0.19 *
91 0.64 0.51 * 0.36 * 0.12 *
Mean 0.63 0.51 * 0.31 * 0.10 *
Analysis

 UK (N-=5136)

Year Adjusted Earnings Book Size
 [R.sup.2] Value

Pooled 0.41 0.53 * 0.08 * 0.17 *
83 0.29 0.53 * -0.11 0.25
84 0.35 0.77 * -0.33 0.22
85 0.22 0.33 0.13 * 0.15
86 0.37 0.71 * -0.17 0.16
87 0.47 0.85 * -0.25 0.11
88 0.57 0.71 * 0.02 * 0.11
89 0.52 0.64 * 0.05 0.15
90 0.56 0.43 * 0.37 * 0.16
91 0.59 0.50 * 0.34 * 0.16
Mean 0.44 0.61 * 0.00 0.16 *
Analysis

 Canada (N=2224)

Year Adjusted Earnings Book Size
 [R.sup.2] Value

 0.61 0.23 * 0.53 * 0.22 *
83 0.51 0.32 * 0.38 * 0.27 *
84 0.55 0.49 * 0.24 * 0.22 *
85 0.45 0.18 0.40 * 0.32 *
86 0.55 0.37 * 0.31 0.27 *
87 0.67 0.29 * 0.47 * 0.23 *
88 0.67 0.26 * 0.54 * 0.19 *
89 0.72 0.10 0.72 * 0.16 *
90 0.73 0.18 * 0.69 0.17 *
91 0.74 0.11 0.78 * 0.09
Mean 0.62 0.26 * 0.50 * 0.21 *
Analysis

 Japan (N-=3645)

Year Adjusted Earnings Book Size
 [R.sup.2] Value

Pooled 0.61 0.24 * 0.56 * 0.04
83 0.53 0.12 0.61 * 0.05
84 0.52 0.30 0.47 * -0.03
85 0.41 -0.01 * 0.64 * 0.04
86 0.46 -0.02 * 0.64 * 0.15 *
87 0.66 0.29 0.54 * 0.13 *
88 0.82 0.40 0.52 * 0.04 *
89 0.76 0.47 0.43 * 0.06
90 0.78 0.57 * 0.34 * -0.01
91 0.75 -0.32 1.10 * -0.02
Mean 0.63 -0.20 0.59 * 0.05 *
Analysis

* Implies that the coefficient value is significant at the 1%
significant level.

(a) Mean Analysis of the coefficients value are applied to the
simple mean of the coefficient, the t-tests are applied using
nine years of coefficient values.

Panel B: Sum and Difference of the Coefficient Estimations

 US (N=4187)

 Difference
Year Sum (a) (b) Ratio (c)

Pooled 0.86 0.19 0.22
83 0.78 0.15 0.20
84 0.84 0.43 0.51
85 0.76 0.02 0.02
86 0.84 0.03 0.03
87 0.80 0.18 0.22
88 0.82 0.51 0.62
89 0.83 0.14 0.16
90 0.95 0.25 0.29
91 0.87 0.14 0.16
Yearly 0.86 0.21 0.25
Averages (d)

 UK (N=5136)

Year Sum Difference Ratio

Pooled 0.61 0.45 0.74
83 0.42 0.64 1.54
84 0.44 1.10 2.49
85 0.46 0.20 0.44
86 0.54 0.88 1.63
87 0.60 1.10 1.85
88 0.73 0.69 0.94
89 0.70 0.59 0.85
90 0.80 0.06 0.07
91 0.84 0.16 0.19
Yearly 0.61 0.60 0.98
Averages (d)
 Canada (N=2224)

Year Sum Difference Ratio

Pooled 0.76 -0.31 -0.41
83 0.71 -0.06 -0.09
84 0.73 0.24 0.33
85 0.58 -0.22 -0.37
86 0.68 0.07 0.10
87 0.77 -0.18 -0.24
88 0.80 -0.28 -0.35
89 0.82 -0.63 -0.77
90 0.87 -0.51 -0.59
91 0.88 -0.67 -0.76
Yearly 0.76 -0.25 -0.33
Averages

 Japan (N=3645)

Year Sum Difference Ratio

Pooled 0.81 -0.31 -0.40
83 0.73 -0.47 -0.66
84 0.77 -0.18 -0.21
85 0.63 -0.64 -1.03
86 0.62 -0.59 -1.05
87 0.83 -0.23 -0.31
88 0.92 -0.12 -0.13
89 0.90 0.04 0.04
90 0.91 0.23 0.25
91 0.78 -1.41 -1.81
Yearly 0.79 -0.37 -0.49
Averages

(a) Sum is the sum of EVC and BVC for the yearly regressions.

(b) Difference is (EVC-BVC) for the yearly regressions.

(c) Ratio is computed by scaling the difference by the sum, i.e.
(EVC-BVC) / (EVC+BVC).

(d) Yearly averages are calculated based on the nine year
observations.

Table 6: Regression Results for the Prices Model
with Size Control, The Industry-Controlled Sample

 US (N=1131)

Year Adjusted Earnings Book Size
 [R.sup.2] Value

Pooled 0.63 0.50 * 0.34 * 0.14 *
83 0.49 0.40 * 0.37 * -0.04
84 0.67 0.52 * 0.26 * 0.02
85 0.63 0.51 * 0.36 * 0.16
86 0.68 0.38 * 0.37 * 0.31 *
87 0.68 0.53 * 0.36 * 0.14 *
88 0.67 0.55 * 0.35 * 0.13
89 0.43 0.25 * 0.62 * 0.05
90 0.83 0.83 * 0.14 0.17 *
91 0.66 0.47 * 0.27 * 0.44 *
Mean 0.64 0.49 * 0.34 * 0.15 *
Analysis

 UK (N-=1131)

Year Adjusted Earnings Book Size
 [R.sup.2] Value

Pooled 0.51 0.53 * 0.35 * 0.07 *
83 0.49 0.99 * -0.20 0.08
84 0.65 1.06 * 0.06 -0.03
85 0.32 0.18 0.62 * 0.04
86 0.60 1.20 * -0.11 0.05
87 0.54 0.76 * 0.13 0.06
88 0.62 0.64 * 0.28 * 0.09
89 0.53 0.70 * 0.10 0.12
90 0.60 0.70 * 0.62 * 0.10
91 0.46 0.39 * 0.39 * 0.15
Mean 0.54 0.70 * 0.21 0.07 *
Analysis

 Canada (N=1131)

Year Adjusted Earnings Book Size
 [R.sup.2] Value

 0.62 0.20 * 0.51 * 0.25 *
83 0.53 0.31 * 0.36 * 0.26 *
84 0.43 0.39 * 0.27 * 0.30 *
85 0.38 0.03 0.44 * 0.25 *
86 0.53 0.32 * 0.27 0.32 *
87 0.75 0.37 * 0.44 * 0.16 *
88 0.68 0.24 * 0.56 * 0.33 *
89 0.77 0.10 0.67 * 0.25 *
90 0.81 0.14 * 0.68 0.26 *
91 0.76 0.09 0.74 * 0.12
Mean 0.63 0.22 * 0.49 * 0.25 *
Analysis

 Japan (N-=1131)

Year Adjusted Earnings Book Size
 [R.sup.2] Value

 0.66 -0.11 * 0.87 * 0.09
83 0.59 -0.25 1.05 * 0.09
84 0.46 0.30 0.45 * 0.14
85 0.62 -0.43 * 0.99 * 0.01
86 0.58 -0.32 * 0.82 * 0.30 *
87 0.72 0.12 0.70 * 0.26 *
88 0.82 0.04 0.75 * 0.09 *
89 0.76 -0.16 0.98 * 0.05
90 0.70 0.47 * 0.44 * 0.04
91 0.62 -0.54 1.24 * -0.04
Mean 0.65 -0.08 0.82 * 0.10 *
Analysis

* Implies that the coefficient value is significant at the 1%
significant level.

(a) Mean Analysis of the coefficients value are applied to the
simple mean of the coefficient, the t-tests are applied using
nine years of coefficient values.