Discussion of empirical evidence on the evolution of international earnings.

I. INTRODUCTION

In "Empirical Evidence on the Evolution of International Earnings" Land and Lang (2002) (hereafter the paper) investigate a question related to international accounting differences: Has there been convergence of earnings multiples over time, supporting an inference

that financial accounting measures used in different countries become more similar over time? The authors make a convincing argument that anecdotal evidence would lead us to expect such a convergence in accounting measures; and although their research design does not allow them to make a causal link between their empirical evidence and accounting changes across countries, they do a thorough job of attempting to control for alternate explanations. After reading the paper the reader is left with two strong impressions. First, average earnings/price (E/P) ratios across the countries analyzed in this paper (Australia, Canada, France, Germany, Japan, the U.K., and the U.S.) have moved closer together from the 1987-1992 time period to the 1994-1999 time period. Second, this convergence of E/P ratios seems to be related to accounting accruals.

Whether E/P ratios have converged over time is an empirical question, as the authors point out, and it seems appropriate that the question be studied by examining the data as the authors have done. The authors use different types of tests to try to role out alternate explanations for the convergence of E/P ratios, and the results of these tests seem consistent with expectations given the anecdotal evidence cited by the authors (i.e., the results are not surprising).

As noted by some of the Conference participants, the apparent convergence of financial reporting across countries is evidence only of the convergence of accounting measurement since it is based on reported net income. Convergence of accounting measurement would suggest that net income is being computed using similar accounting methods across countries. An example of convergence in measurement is the use of the comprehensive (rather than the partial) method of recording deferred tax liabilities, since this affects reported deferred tax expense and, hence, reported net income. Another question of equal interest to researchers is whether there has been a convergence in accounting disclosure over this same time period. An example of convergence in disclosure is disclosing the market value of all financial assets and liabilities, since this does not affect reported net income. If capital markets are efficient with respect to accounting information, then disclosure of value-relevant information will affect stock prices, and therefore the convergence of accounting disclosure should also be of interest to accounting researchers.

Unfortunately the research design of this study cannot address the question of whether accounting disclosures have also converged. Also, the results of this study cannot tell us anything about what has happened to accounting standards in different countries over this time period, since the net income measure used in the study may reflect choices made by firms to report using either U.S. GAAP or International Accounting Standards (IAS) rather than the GAAP of their home country; that is, the convergence of E/P ratios reflects a convergence toward using either U.S. GAAP or IAS and not a convergence of standards, per se. The authors report that the proportion of sample firms reporting using IAS or U.S. GAAP was 0.8 percent in the first period and 2.1 percent in the second period, and that the conclusions are not sensitive to excluding these observations. Therefore, although a shift to IAS or U.S. GAAP appears to have occurred during the sample period, such a shift does not affect the paper's results.

II. IMPACT OF STOCK MARKET INDEX INCREASES

One alternative explanation for the convergence in E/P ratios documented by the paper is that the convergence is due to denominator effects, not numerator effects. That is, the results might reflect the increases in stock market indexes that occurred in all of the sample countries (except Japan) over the period 1987-1992 to 1994-1999. Since the dependent variable in these tests is the ratio of earnings to stock price, systematic increases in stock prices in the sample countries would decrease average E/P ratios. Similarly, systematic decreases in stock prices in Japan over this same time period would increase average E/P ratios for the Japanese firms in the sample.

Although the authors argue that their results suggest E/P ratios converge over their sample time period, their results can also be interpreted as demonstrating that E/P ratios decreased over the time period in five of the seven sample countries (Australia, U.K., France, U.S., and Canada), while E/P ratios increased in two of the sample countries (Germany and Japan). Changes in stock market indexes over the same time period can explain these changes in six out of the seven sample countries, with only the results for Germany being inconsistent with a stock market-based explanation. To test this explanation, for each of the sample countries I compute the percentage change in the country's major stock market index from the January 1, 1990 level to the January 1, 1997 level. I also used data from Land and Lang (2002, Table 1) to compute the percentage change in mean E/P ratios across the two time periods. These results are shown in Table 1.

When Germany is excluded from the sample countries, the Pearson correlation between the percentage change in E/P ratios and the percentage change in market indexes is -0.83. A regression of the change in E/P ratio on the change in market index has an [R.sup.2] of 0.70. Thus it appears that, except for Germany, a large part of the change in E/P ratios observed over this time period may be related to changes in stock prices in the sample countries.

Table 1 also shows the mean cash flow/price ratios for the sample countries for the two subperiods (taken from Land and Lang [2002, Table 3]), as well as the percentage changes in these ratios. It is apparent from these data that the pattern observed for the E/P ratios (i.e., decreases in ratios for five of the sample countries and increases in ratios for Germany and Japan) is also present in the cash flow/price ratios. The Pearson correlation between percentage changes in E/P ratios and percentage changes in cash flow/price ratios is 0.97. This suggests that the pattern observed in the E/P ratio changes is also present in the cash flow/price ratio changes, and that, except for Germany, this pattern can be explained by changes in stock market indexes in the sample countries.

Obviously, the results for Germany are not consistent with the stock index explanation, since the German stock market index increased over the sample period, while both E/P ratios and cash flow/price ratios for German firms also increased. However, regardless of the cause of the increase in E/P ratios for German firms over the sample period, the same increase appears in the cash flow/price ratios as well. The difference is that the increase in E/P ratios causes German firms to move closer to the mean, while the increase in cash flow/price ratios causes German firms to move further from the mean.

III. ALTERNATE EXPLANATIONS

Although the authors have been careful to design tests to rule out several competing explanations for their results, some alternate explanations suggested by conference participants warrant mention in this discussion. First, the way investors use accounting information to set stock prices may be changing over time. For example, increasing sophistication on the part of investors in the sample firms would be expected to change the observed relation between reported accounting earnings and stock prices. This might occur if the sample firms are large multinationals of the type that mutual fund managers are likely to follow, and follow increasingly over the sample period considered. Second, the integration of global capital markets over this time period could lead to consistent pricing of risks in all markets. Third, several Conference participants thought that the results were simply an example of mean reversion over time.

The authors have done a good job of addressing alternate explanations by including several control variables in their tests. However, the use of these control variables also led to comments by Conference participants. The way control variables are used in the paper assumes that the effect of a particular control variable remains constant over time, but the level of the control variable changes. Conference participants suggested that an alternative approach would be to allow the effect of the control variable (as measured by the regression coefficient) to change over time as well. This would require the inclusion of interactions between the time period indicator variables and the control variables. The authors have included such an interaction between the cash flow/price ratios and the country and subperiod indicator variables in the paper's Table 5 and Table 6 regressions.

IV. CONCLUSION

The results reported in the study are difficult to interpret for two reasons. First, both changes in E/P ratios and changes in cash flow/price ratios exhibit a similar pattern over time, in that both sets of ratios are decreasing for five of the sample countries (Australia, Canada, France, U.K., and U.S.), and both sets of ratios are increasing for Germany and Japan. For all of the countries except Germany, this result can be explained by changes in stock market indexes over the same time period. I interpret this as evidence that whatever is causing increases or decreases in E/P ratios is also causing similar increases and decreases in cash flow/price ratios. In my view, the reader is left to wonder whether the apparent convergence reported in the paper can be explained by stock price movements for six of the sample countries, and an anomalous result for the German firms.

The authors take a different view of the cash flow/price results, maintaining that, although there are significant changes in some cash flow/price ratios over the period, there is not a consistent pattern, since four countries' multiples move away from the average. In reaching their conclusion that there is little similarity between the movements of E/P and cash flow/price ratios the authors are comparing movement toward or away from the mean. However, if the comparison is between increases or decreases in ratios, then the pattern is nearly identical between the two sets of measures.

The second reason the results are difficult to interpret is that, even if the authors are correct and there has actually been convergence in E/P ratios across the sample countries over time, what should we make of this result? Absent specific knowledge of the cause of the convergence, it is difficult to draw public policy or regulatory implications. We do not know if organizations such as the IASC have been successful in their drive to harmonize accounting standards, whether behavior of managers has changed, or whether demand by capital markets has led to the use of similar accounting measures. Without more understanding of what is behind the convergence of E/P ratios, the results in and of themselves leave the reader wondering what conclusions to draw from the paper.

TABLE 1
Changes in E/P Ratios, Stock Market Indexes, and Cash
Flow/Price Ratios from 1987-1992 to 1994-1999

             (1)     (2)      (3)      (4)      (5)
             E/P     E/P
            Ratio   Ratio             Market   Market
            1987-   1994-             Index    Index
            1992    1999    (2)/(1)    1990     1997

Australia   0.095   0.072    0.753     1,655    2,411
Canada      0.070   0.066    0.938     4,004    5,904
Germany     0.042   0.068    1.611     1,443    2,849
France      0.080   0.066    0.826     1,832    2,257
U.K.        0.089   0.077    0.871     2,434    4,057
Japan       0.019   0.038    1.938    38,713   19,446
U.S.        0.073   0.061    0.841       360      737

              (6)      (7)     (8)      (9)
                      Cash    Cash
                      Flow/   Flow/
                      Price   Price
                      Ratio   Ratio
                      1987-   1994-
            (5)/(4)   1992    1999    (8)/(7)

Australia    1.457    0.143   0.121    0.842
Canada       1.475    0.118   0.115    0.971
Germany      1.974    0.090   0.117    1.297
France       1.232    0.128   0.115    0.899
U.K.         1.667    0.137   0.126    0.923
Japan        0.502    0.067   0.087    1.285
U.S.         2.047    0.121   0.110    0.912

Stock market indexes are All Ordinaries (Australia), TSE 300 Composite
(Canada), DAX (Germany), CAC 40 (France), FTSE 100 (U.K.), Nikkei 225
(Japan), and S&P 500 (U.S.).

E/P ratios are from Land and Lang (2002, Table 1). Market indexes are
from Yahoo! Finance. Market indexes for Germany and France were not
available as of January 1, 1990, so the earliest date in 1990 for which
data were available was used for those countries. Cash flow/price
ratios are from Land and Lang (2002, Table 3).

Helpful comments were received from Katherine Schipper and Mark H. Lang.

REFERENCE

Land, J., and M. H. Lang. 2002. Empirical evidence on the evolution of international earnings. The Accounting Review (Supplement): 115-133.

David A. Guenther
University of Colorado at Boulder