Factors affecting price earnings ratios and market values of Japanes firms.
There are a number of possible explanations for the rapid rise in PE ratios and market values of Japanese firms. [sup.1] These explanations include changes in firm characteristics such as ownership structure, rates of growth in earnings, risk, the use of special reserves, and the market value of hidden asset investments. This study examines the link between these factors and changes in PE ratios and market values for a sample of Japanese firms over the 1979 to 1989 period.
A number of authors examine the relation between PE ratios, risk, and returns in the domestic market (BASU , Reinganum , , , Cook and Roseff , and Keown, Pinkerton, and Chen ). Their work suggests that the level of PE ratios is related to firm size and unsystematic risk. Other authors examine Japanese stock returns. Aggarwal, Rao, and Hiraki  examine departures from normality for the statistical distributions of Japanese stock returns. Jaffe and Westerfield , .
Exhibit 1. Yearly rice/Earnings Ratios from 1979 to 1989 for Japanese and U.S. Companies Year Japan U.S. 1979 16.6 7.0 1980 17.9 8.1 1981 24.9 8.2 1982 23.7 9.7 1983 29.4 11.1 1984 26.3 9.6 1985 29.4 12.9 1986 58.6 15.5 1987 50.4 17.2 1988 54.3 11.3 1989 53.7 10.6 Source: Japan price/earnings ratios for index of nonfinancial firms from Nomura Research Institute 350. United States price/earnings ratios based on Standard and Poor's S&P 500 index.
and Kato and Schallheim  examine Japanese stock returns and confirm the existence of market anomalies that have been identified in domestic markets. Hamao  compares Japanese stock returns to domestic stock returns, finding that Japanese returns are higher. While these authors examine the characteristics of returns, none attempt to identify the factors that impact prices or market values.
French and Poterba  and Frankel  specifically investigate possible reasons for the increasing disparity in PE ratios and market capitalization between Japan and the United States. French and Poterba examine factors influencing market aggregate PE ratios and attempt to use differences in the financial accounting practices observed between Japan and U.S. as a basis for the adjustment of Japanese PE ratios in order to see if these adjusted Japanese ratios are similar to U.S. ratios. They find that these accounting-based adjustments are able to explain, at most, only half of the differences in PE ratios across countries and that these accounting differences cannot explain the dramatic increase in PE ratios during the 1980's or the decline in 1990. Furthermore, their results indicate that differences in discount rates and expected growth rates are also unable to explain this divergence in PE ratios. They state that changes in land value may be an important factor, although they do not specifically investigate this issue. Frankel surmises that possible explanations for high PE ratios in Japan include: lower interest rates, higher growth rates, and high land prices. In addition, he notes that PE ratios were relatively high in the 1960's and 1970's when Japanese interest rates were low and economic growth rates were high. He explains that this is due to the institutional aspects of the Japanese financial system.
These previous papers suggest that there are still many unanswered questions about the factors related to changes in PE ratios and market values in Japanese firms. This study investigates cross-sectional explanations of Japanese PE ratios at the firm level, which contributes to the studies of Frankel  and French and Poterba . We focus on two dependent variables, the percentage change in the Pe ratio (PE) and the percentage change in the market value of equity (MV), and examine whether these variables are related to changes in the value of corporate land holdings, ownership structure, risk, earnings growth, dividend payout, divident growth, and the use of special reserves. Results of this study suggest that changes in the value of hidden assets such as land, changes in the patterns of ownership by corporations, Japanese individuals and foreign investors, changes in expected earnings growth and earnings risk, and changes in dividends per share and dividend payout are significantly related to changes in PE ratios and market values. Our results, therefore, support the contentional of French and Poterba  that changes in land values appear to explain at least a portion of changes in Japanese PE ratios.
This paper is divided into four sections. The first section models factors affecting stock price and PE ratios, and discusses the factors that may be responsible for increases in PE ratios and market capitalization. The second section discusses the data and empirical methodology used to examine the relations between these factors and the market variables. Section III presents results of various empirical tests. Conclusions are presented in Section IV.
PE ratio is a common measure used to indicate market assessment of a company's performance and can be stated as PE = stock price ([P.sub.0])/earnings per share ([EPS.sub.1)), where [P.sub.0] may be represented by a constant dividend growth model, [P.sub.0] = [D.sub.1]/(r - g), and where [D.sub.1] measures the expected dividend next year, r is the investor's required rate of return, and g is the expected growth rate in dividends. Following Brealey and Myers , PE may be related as:
[Mathematical Expression Omitted]
Thus, a company's PE ratio depends on the expected growth rate in dividends, the risk of the company, which corresponds directly with the required rate of return, and the dividend payout ([D.sub.1]/[EPS.sub.1]) the company can achieve while still maintaining the growth rate indefinitely. The factors we hypothesize to influence changes in PE ratios and market values, through their effect on expected growth, risk or payout, are discussed below.
A. Dividend Growth Model Related Proxies
Earnings Growth. Growth in earnings should influence both a firm's PE ratio and market value of equity. If stock prices reflect a capitalization of expected future earnings, increasing earnings should be associated with an increase in share price and in total market capitalization. If investors believe that this higher rate of growth in earnings is sustainable, the PE ratio should rise.
Since Japanese firms rely heavily on nonoperating income from investments in securities of affiliates and other assets, we use the total of operating and nonoperating earnings, as reported in the fiscal year-end financial statements, in the calculation of the earnings growth proxy. This measure of earnings growth reflects the performance of the firm before interest expenses, set asides for special reserves, and tax payments. The proxy EG is the percentage change in the expected growth rate of earnings from 1979 to 1989. 
Dividends Growth. Increases in dividends per share over the period should be associated with both increases in earnings and increase in stock prices. It could be argued that an increase in dividends per share could be perceived as a proxy of the permanent component of the increase in earnings per share. We measure the change in dividends per share as the percentage increase over the ten-year period.
Risk. In the PE model discussed earlier, the risk associated with the earnings stream is captured by the discount rate (r), which is a function of systematic, market-based risk. Other things held constant, there is an inverse relation between risk and PE. Fiscal year-end data from 1975 through 1979 are used to calculate the beginning of the period standard deviation and fiscal year-end data from 1985 through 1989 are used to calculate the ending standard deviation. (3) The change in standard deviation is calculated using these beginning and ending standard deviations.
Dividend Payout. This study examines the relation between the percentage change in the dividend payout ratio over the 1979 to 1989 period to changes in the PE ratio and market value. Since investors are assumed to be willing to pay more for a share of stock that pays higher dividends, other things held constant, an increased payout suggests a higher PE ratio.
B. Additional Variables
The uniqueness of Japanese firms necessitates several additions to our set of explanatory variables. First, land holdings and special accounting reserves are included, since factors related to these variables may result in earnings or dividend payout being understated. Second, differences in Japanese organizational structure may result in less agency/information problems and thus, lower risk. Therefore, we include various ownership variables to control for the distinctive nature of Japanese organizational structure.
Land Holdings. The price level of real estate in Japan has increased dramatically since 1979. Exhibit 2 shows property value index increases for two categories of Japanese commercial property over the 1979 to 1989 period. During this ten-year period, the price index for industrial property increased by approximately 164%, while the price index of office property increased by approximately 421%. (4) French and Poterba  postulate the run-up in Japanese stock market prices during the 1980's is related to the run-up in Japanese land prices, since many companies have large land holdings. Frankel  documents this relationship between Japanese stock and land price run-ups.
Accounting rules in Japan, like accounting rules in the U.S., require that unrealized gains in capital assets, such as land, not be reported until a transaction takes place. Therefore, although the land holdings of corporations in Japan have made substantial contributions to the economic value of these corporations, this increase in land value is not reflected in either the corporate balance sheet or income statement, but instead, is carried on the books
Exhibit 2. Property Index for Japanese Industry. Office and Major City Property (a) (index for March of each year, based on the March 1980 index equal to 100) Year Industrial % Change Office % Change 1979 92.0 90.3 1980 100.0 8.7 100.0 10.7 1981 106.7 6.7 108.1 8.1 1982 112.6 5.5 116.7 8.0 1983 116.7 3.6 124.6 6.8 1984 120.6 3.3 135.7 8.9 1985 125.0 3.6 153.6 13.2 1986 131.2 5.0 197.9 28.8 1987 153.6 17.1 264.7 33.8 1988 183.2 19.3 375.2 41.7 1989 243.7 33.0 469.5 25.1 Source: Keizai Tokei Nenkan: Economic Statistical Yearbook 1989, Tokyo, Toyo Keizai Shinposha, 1989. (a) Property index based on average property values of six largest Japanese largest Japanese metropolitan areas.
at a historic book value.  In an efficient market, investors realize that the true value of a firm with large investments in real estate is greater than the reported book value and that economic earnings are greater than reported earnings. As property values rise, the firms' market value and PE ratio should reflect this change in true value. We expect that firms with large increases in the market value of their real estate holdings over this ten-year period will have a corresponding increase in the firm's market value and PE ratio.
This study examines the link between increases in land values and the dependent variables, using an independent proxy variable (LAND) that tries to capture the percentage increase in the market value of land. The calculation of the LAND proxy begins with the total (balance sheet) book values of all land owned by sample firms at the end of their 1979 and 1989 fiscal years. These land categories include land under office and production facilities, land held for investment purposs, producing agricultural and forest lands, and raw land. These book value amounts for land are then weighted by a land price index for the respective years and used as the beginning and ending values to calculate the percentage change in estimated market value of land over the period.
Special Accounting Reserves. Freitas  discusses the use of special accounting reserves by Japanese firms and notes that these reserves allow a reduction in taxable income and an increase in after-tax cash flows while under-reporting after-tax earnings. Furthermore, he states that in many cases these special reserves must be reversed after a certain period of time has passed. If reserves are increasing overtime, reported after-tax earnings will be increasingly understated. If markets are efficient, the use of special reserves will not impact share price or market value. However, if investors make decisions based on reported earnings and not actual earnings, a reduction in earnings through the use of special reserves may depress share price and lower market value. If the use of special reserves is reduced, reported after-tax earnings and taxes will rise.  A proxy variable representing the percentage change in the proportion of total assets held as special reserves (RES) is used to examine this relation.
Ownership Structure. Different corporate environments within the U.S. and Japan result in different corporate ownership structures. While U.S. corporations are characterized by highly diffuse ownership structures, cross-ownership among Japanese corporations and financial institutions is common (Demsetz and Lehn , Nishiyama , and Okumura ). In Japan, it is often the case that the motivation underlying the cross-ownership of stock is associated with membership in groups of affiliated firms, called keiretsu. 
Kester . McDonald , and Hodder and Tschoegl  discuss the pattern of cross-ownership in Japanese financial institutions and corporations that characterizes the keiretsu. The interlocking ownership and management of these keiretsu members affect market measures, such as price and total capitalization, in a number of ways. First, keiretsu relations reduce many agency-related costs of operation. High levels of cross-firm ownership within the keiretsu act to increase the percentage of total outstanding shares of stock that are controlled by affiliated firms. Since there are strong ties between these firms, the management of affiliated firms influences the decision-making process of other affiliated firms as if they were insider shareholders. This, in turn, reduces the agency costs identified by Jensen and Meckling . Furthermore, as noted by Freitas and Otsuki  and Hodder and Tschoegl , in the event of a group member experiencing financial distress, other keiretsu members will assist that troubled firm through favorable interfirm pricing arrangements, timing of interfirm payments, and extension of credit, thus reducing corporate financial risk. Firms within the keiretsu group operate to the betterent of the group as a whole, even though there may be some short-run costs to the firm extending support.
The extent of the assistance within these corporate groups can range to making loans or extending credit to members of the keiretsu by the lead bank and other financial institutions. Although recently, the links between banks and affiliated firms have been weakened by changes in the regulatory structure, the share of total external financing provided by banks is still relatively high when compared to other industriliazed countries (Frankel  and Meerschwam ). When financial institutions play the role of both shareholder and creditor, debt-related agency costs identified by Myers  and Jensen and Meckling  are lower. These reduced agency costs suggest that an increase in the financial institutions' cross-ownership of a Japanese company should lead to increases in the share price, PE ratio, and market value of the firm, if all other factors are held constant.
Increases in cross-ownership may also impact share prices and market value through a supply constraint effect. If the percentage ownership held by affiliated firms increases without issuing additional shares of stock, the supply of shares available for public trading decreases and, given a constant demand, share price and market value of equity may increase. This supply effect may also affect PE ratios and market values, even if the cross-ownership activity is driven by investment motives rather than keiretsu affiliation motives.
Because of the reasons discussed above, cross-ownership of corporate stock may effect PE ratios and firm market value in three ways. First, we expect firms with large cross-holdings or substantially increasing cross-holdings to demonstrate greater increases in PE ratios and market values due to the current value of these assets being reflected in market prices, while the historic costs are reflected in the accounting statements.  Second, due to decreased financial risk, we expect an increase in PE ratios and market values for firms whose stock is increasingly being held in friendly hands. Third, a supply effect may cause share price to increase for reasons not corresponding to corporate earnings.
There is another possible factor that may influence the relation between changes in financial institution ownership and PE ratios and market values during the time period examined. In 1977, the Japaese Fair Trade Commission passed a ruling that forced Japanese financial institutions to reduce their holdings of stock in other firms to five percent or less of the other firms' equity by December 1, 1987. Compliance with this new rule prior to the 1987 deadline should result in a decrease in financial institution ownership regardless of the direction of the PE ratios and market values of the corporations in which banks held more than five percent of the outstanding stock.
We examine the relation of cross-ownership to PE ratio and market value by focusing on proxies representing the percentage increase in share ownership by either financial instititions (FIN) or other corporations (CORP). Unfortunately, due to data limitations, we cannot separate out the different motives responsible for cross-ownership patterns or identify whether any obsersved relationship is due to a reduction in agency cost or a supply effect.
Another ownership structure proxy examined is the percentage change in the share of ownership held by individual Japanese investors (INDV). If individual investors exhibit a strong demand for stock whose shares are closely held by affiliated firms, the supply constraint may influence the dependent variables. Alternatively, individual investors may be attracted to investments in the stock of firms that experience strong performance. In either case, a positive relation is expected.
Foreign Investment. Over the last decade, Japanese stock markets have become increasingly open to individual and institutional foreign investors (see Frankel  for a discussion of Japanese domestic market innovation and deregulation, and international market liberalization). This increased foreign demand for Japanese securities, coupled with the high degree of corporate and institutional ownership, may result in demand for Japanese securities that outpaces supply. These supply-and-demand forces may result in increasingly share prices and market value. On the other hand, foreign investors may have been drawn to invest in the firms that experienced rapid increases in earnings, PE ratios, and market values. Unfortunately, our methodology does not allow us to address the direction of causality. Regardless of the direction of the linkage, we hypothesize a direct relation between the change in foreign ownership and the dependent variables. We use a proxy to represent the percentage change in foreign ownership (FOR) in order to examine these relations.
C. Other Factors Not Examined
The difference in PE ratios and market values between Japan and the U.S. may be explained by a lower discount rate in Japan, based on arguments that the most of equity capital is lower in Japan. Friend and Tokutsu  find that, for the period 1962-1984, the cost of capital was lower in Japan than in the U.S., but the difference appears too small to explain higher PE ratios and market values in Japan. Frankel  also finds lower rates in Japan and believes that these lower real interest rates, along with higher expected growth rates, explain much of the high equity and land prices, but do not explain the large increases in equity and land prices during the 1980's.
Changes in accounting procedures may impact reported earnings and firm value. Freitas  and Michel and Shaked  note that Japanese firms are required to use the same depreciation method for both tax reporting and public financial statements. Because of this, few Japanese firms change their depreciation method. Examination of the firms in the sample reveals that none of the firms report a change in depreciation methods. Therefore, we do not examine depreciation method changes in this study.
In Japan, the basic requirement for reporting consolidated financial statements is the presence of majority ownership in a subsidiary, although consolidated reporting is optional if the subsidiary provides less than 15% of sales of assets. Freitas  notes that there exists a loophole in this requirement and that most Japanese firms report unconsolidated earnings by choice. If a firm shifts from reporting on an unconsolidated basis to a consolidated basis, reported earnings will rise. However, since the data used in this study are all drawn from unconsolidated financial statements, the consolidation issue is not a factor.
II. Data and Methodology
Our analysis employs fiscal year-end financial statement data for 1975 through 1989 obtained from the Nikkei Annual, Unconsolidated, Industrial Financial Data File (Nikkei). These data include: the total book value of land adjusted by the property price index for commercial office property, (9) ownership by nonfinancial corporations, ownership by financial institutions and trust companies, ownership by individual Japanese investors, ownership by foreign investors (individuals, corporations and mutual funds), total operating and nonoperating earnings, and special reserves. These data allows us to calculate changes in these variables over a ten-year period. When calculating percentage changes over the period, the beginning values are from 1979 fiscal year-end data and ending values are from 1989 fiscal year-end data. For the ownership change proxies (corporate ownership, financial institution ownership, individual ownership, foreign ownership), the beginning and ending values used in the percentage change calculation are the proportion of outstanding shares of common stock owned by the various groups. For the risk and earnings growth proxies, five years of annual data are used to calculate both the beginning and ending values for the period. (10) Month-end closing stock prices for the last month of the fiscal year are collected from various issues of the Kabuka Soran Stock Price Review. Stock prices for many of the stocks traded on the second sections of the exchanges in 1979 are not available for inclusion in the calculations of the dependent variables. This acted as the major constraint on our sample.
Preliminary examination of the sample reveals that many of our empirical proxies possess nonnormal distributions due to the presence of extreme outliers. We adjusted for this nonnormality using a method based on Chebyshev's inequality that serves to identify firms with severe outlying data observations. (11) The outlier adjustment is accomplished as follows. First, the mean and standard deviation of each of the variables is calculated for the entire sample. Next, any set of firm observations containing proxy variables whose values lay outside of plus or minus three standard deviations from their respective mean are removed from the sample. The empirical analysis is performed on the remaining 475 firms.
Summary statistics are presented for the entire sample in Exhibit 3. On average, earnings-to-price multiples decreased by 52.0%, PE ratios increased by 231.2%, similar in magnitude to the increase shown in Exhibit 1, and market values increased by 504.9% from 1979 to 1989. The percentage change in the value of land (LAND) increased, on average, by 129.4% percent over the period. (12) Share ownership by nonfinancial and financial corporations increased by 20.7% and 31.3%, respectively. Individual ownership of Japanese corporate stock decreased by 31.2%, while foreign ownership increased dramatically
[TABULAR DATA OMITTED]
(by approximately 3,000%) during the decade. This large percentage increase in foreign ownership may be attributed to the opening of Japanese financial markets, a strong Japanese economy attracting an increased amount of investment funds, and a general increase in foreign investment by large institutional investors. This large percentage increase may also reflect the extremely low level of foreign ownership at the beginning of the period examined. The change in risk, measured as the change in standard deviation of earnings, decreased by 18.8%. The expected growth rate in earnings increased by 1.6%, the growth in dividends increased by 84.4%, dividend payout increased by 3.5%, and the use of special reserves decreased 94.3% over the period.
Two different models are examined in our analysis. The first model allows us to test the relationship between our explanatory proxies and the change in a company's PE ratio from 1979 to 1989. (13) However, to avoid extreme values of PE and diminish kurtosis within the distribution of our dependent variable, we measure the dependent variable as the reciprocal to PE, the change in the earnings to price multiple, EP. (14) Since we define the dependent variable in our analysis ([Delta]EP) as the percentage change in EP from one period to the next, we leave firms with negative earnings in the analysis to avoid truncation of our distribution. Therefore, [Delta]EP may decrease as well as increase by a factor of greater than 100%. (15) The model is as follows:
[Mathematical Equation Omitted]
[[Delta]EP.sub.i] = the percentage change in EP ratio from 1979 to 1989 for company i;
[LAND.sub.i] = the percentage change in the yen-donominated value of land scaled by total assets from 1979 to 1989 for company i;
[CORP.sub.i] = the percentage change in the share of ownership held by nonfinancial corporations from 1979 to 1989 for company i;
[FIN.sub.i] = the percentage change in the share of ownership held by financial institutions and trust companies from 1979 to 1989 for company i;
[INDV.sub.i] = the percentage change in the share of ownership held by individual Japanese investors from 1979 to 1989 for company i;
[FOR.sub.i] = the percentage change in the share of ownership held by foreign individuals, mutual funds, and corporations from 1979 to 1989 for company i;
[SD.sub.i] = the percentage change in standard deviation of earnings from the 1975 through 1979 period to the 1985 through 1989 period for company i;
[EG.sub.i] = the percentage change in the expected growth rate in earnings from 1979 to 1989 for company i;
[DG.sub.i] = the percentage change in the dividends per share from 1979 to 1989 for company i;
[DPO.sub.i] = the percentage change in the dividend payout ratio from 1979 to 1989 for company i; and
[RES.sub.i] = the percentage change in special reserves from 1979 to 1989 for company i.
A second model is included to test the relations between our proxy variables and the change in firm market values over the ten-year period. This form of the model is as follows:
[Mathematical Equations Omitted]
[[Delta]MV.sub.i] = the percentage change in the market value of equity from 1979 to 1989 for company i.
This analysis allows us to test the effect of our proposed explanatory variables on a measure of market value that is not dependent on an accounting measure. A common criticism of PE ratios (EP) is that the denominator (numerator) is reported earnings for the previous year-end and may have little relation to future prospects of the firm. Furthermore, some accounting practices, such as the use of special reserves, may distort reported earnings relative to actual earnings. A total market value measure avoids this distortion.
The regression results for Equations (4) and (5) are presented in Exhibit 4. The regression of the explanatory variables against the change in EP is significant at the 0.01 level, as indicated by an F-statistic of 14.78. Individual significant coefficient are found for the variables LAND, SD, EG, DG, and DPO.
The relationship between changes in land values (LAND) and EP is found to be significant at the five percent level ([[Beta].sub.1] = 0.43 (t = 2.19)). This provides evidence that investments in land are related to the change in PE ratios over time. (16) These results, reported in Exhibit 4
[TABULAR DATA OMITTED]
and footnote 16, provide support for the contention of French and Poterba  that increases in land values may be an important factor in the increase in PE ratios during the 1980's. (17) However, results reported in footnote 16 must be viewed with caution since they are based on land proxies that do not represent scaled changes, whereas other proxies used in the regression are scaled.
Our change in risk measure, SD, has a significantly positive coefficient ([[Beta].sub.6] = 0.53 (t = 2.01)). This positive relation is expected, since an increase in risk will increase an investor's required rate of return and bring about a resulting increase in EP. Our proxy for the change in expected earnings growth is significant and positive ([[Beta.sub.7] = 112.55 (t = 7.09)), revealing a direct relation between increases in earnings growth and EP. This result indicates that increased risk associated with the earnings increase dominates the effect of higher growth on EP. The coefficient for the change in dividends per share is significantly positive ([[Beta.sub.8] = 1.11 (t = 4.30)), suggesting an inverse relation between increases in dividends per share as a proxy for lack of growth opportunities and PE ratio. The coefficient for the change in dividend payout, DPO, is significantly negative ([[Beta].sub.9] = -0.98 (t = 4.32)) as expected. The coefficients associated with the ownership proxies (CORP, FIN, INDV, and FOR) and the RES proxy are not found to be significantly different from zero. (18)
The second stage of our analysis examines the variables that are related to changes in the market value of the firm. These results are also presented in Exhibit 4. The explanatory power of the regression in significant at the 0.01 level, as indicated by an F-statistic of 11.99. The results are similar to regression Equation (4) showing significant coefficients for the change in expected earnings growth ([[Beta].sub.7]=-17.89 (t=2.10)), the change in dividends growth ([[Beta].sub.8=0.59 (t+4.24)), and the change in dividend payout coefficient ([[Beta].sub.9]=-0.6 (t=-5.12)). Additionally, we find significant coefficients for changes in ownership by Japanese individuals ([[Beta].sub.4]=-3.03 (t=-5.17)), changes in foreign ownership ([[Beta].sub.5]=0.01 (t=4.38)), and changes in special reserves ([[Beta].sub.10=5.97 (t=3.75)).
However, we do not find a significant relation between land value and MV as expected. This lack of results may be the result of a misspecified proxy. Exhibit 5 shows results of another regression of the absolute change in estimated yen-denominated market value of land (LAND2) against the absolute change in the yen-denominated market value of the firm ([Delta]YenMV), which is significant at the 0.01 level (F-statistic = 589.50 [R.sup.2] = 55.4%). The coefficient for the land change proxy is significant at the 0.01 level, providing strong evidence that aggregate changes in land value are directly related to aggregate changes in the market value of the firm. Further, our summary statistics show that our land variable had a large increase over the period. However, we feel that a large portion of the increase in land prices is due to the fixed supply of land and restrictive features in Japanese real estate markets, such as a high level of capital gains tax at sale, building restrictions, and special protection for rice paddies (see Frankel  for further discussion).
The direct relation between dividend growth and market value and the indirect relation between dividend payout and market value indicate that, although dividend payout lagged behind increases in earnings, Japanese firms were still able to increase dividends during this period of rising stock prices and market values.
The strong negative relation between changes in individual ownership and changes in market value is hard to explain. A possible explanation is linked to the risk preference of individual investors. Although the Japanese population has among the highest national household savings rates (23.0% compared to 11.5% in the U.S.), the high price of land and home ownership represents a large portion of this savings (65% of Japanese savings) (Frankel ). Since individuals have such a large part of their savings invested in illiquid real assets, they may be
[TABULAR DATA OMITTED]
attracted to relatively low risk (and low performance) stock investments. Coefficients for changes in foreign ownership correspond with conclusions of Frankel . International liberalization of the Japanese stock market has allowed a large increase in direct foreign investment that corresponds to the rise in market value. This provides some support for a supply effect hypothesis, but since the corporate and financial institutional ownership proxies are found to be insignificant and the foreign holdings are such a small proportion of total ownership, the support is weak. It may be the case that foreign investors are attracted to firms with rapidly rising market values. The coefficient for special reserves is positive as expected. This indicates that an increase (decrease) in the ability to shelter income from income taxes through the use of special reserves corresponds to an increase (decrease) in firm market value.
The results of this study suggest that changes in EP (PE ratios) are positively (negatively) related to proxies representing the changes in earnings growth and dividends growth, and negatively (positively) related to changes in land value and dividend payout ratios. Changes in firm market values are positively related to changes in foreign ownership, changes in dividend growth, and changes in special reserves, and are negatively related to changes in individual ownership and dividend payout.
Results associated with the land proxies reported in Exhibits 4 and 5 and in footnote 16 suggest that changes in land value are related to firm value. These results support the contention of French and Poterba  that land-price appreciation does explain a portion of the growth of Japanese PE ratios in the 1980's. Changes in some types of ownership do appear to be related to changes in PE ratio and market value. Changes in corporate ownership are found to be significantly related to changes in PE, when PE is used as a dependent variable rather than the EP inverse transformation. This suggests that increases in corporate cross-ownership are related to increases in PE ratios. Furthermore, it suggests that taking the inverse of the PE ratio obscures this relationship. The lack of an observed relation between cross-ownership and market value is puzzling and suggests that cross-ownership does not affect firm value. The change in financial institution ownership is not found to be related to the change in EP, market values, or PE ratios. This suggests that cross-ownership by financial institutions and the phase-in of the five percent ownership limit imposed by the Japanese Fair Trade Commission either had little effect on PE ratios and market value or the effects canceled. Foreign ownership is found to be related to market value and PE ratio changes, but the results do not allow a clear statement of the reasons for the observed relation.
Our results suggest that variables such as risk, earnings and dividend growth, and dividend payout, that are normally associated with textbook models of PE ratio, are also significantly related to changes in both PE ratio and market value changes in Japanese firms. Furthermore, the rise in PE ratios and market values are not a consequence of a decrease in the required rate of return in Japan over this period. Our evidence indicates that the required rate of return rose during the period, since we show an increase in risk, and we expect financial liberalization in Japan to increase the cost of capital to Japanese firms and to narrow cost-of-capital differences between Japan and other industrialized nations. The results associated with the change in special reserves support an earnings-based influence on PE ratios and market values.
We summarize our results as follows. The primary factors behind the dramatic increase in PE ratios and market values during the 1980's appear to be related to the increase in earnings. This includes the change in reported earnings and the change in unreported earnings from hidden sources such as investments in land. These results support the contention of Frankel  that the dramatic rise in Japanese stock prices during the 1980's may not be a speculative bubble, but instead, is related to economic factors.
This study has attempted to identify factors that are related to the rise in PE ratios and market values of Japanese firms during the 1979 to 1989 period. The results suggest a number of areas where further research is needed. First, the relation between hidden assets such as land and market values of Japanese firms needs to be explored more rigorously. Second, additional research investigating the link between keiretsu membership, corporate cross-ownership, and firm performance is needed to better decipher results found in this study. Finally, the link between financial institution ownership and performance needs to be examined with a methodology that controls for the changing regulatory structure over time.
(1) The recent drop in Japanese PE ratios and market values in the 1990's suggest that their dramatic rise during the 1980's may have been a speculative bubble. However, Frankel  finds that the increase in the real interest rate in Japan does appear to be capable of explaining most of the 1990 drop in the Japanese stock market, suggesting the 1980 rise may not be a speculative bubble.
(2) The beginning and ending growth rates are separately estimated using a log linear regression approach in which the natural log of total earnings is regressed against time for the five years prior to 1979 (or 1989) and the estimated slope coefficient (b=log(1 + g)) is solved for g by taking the antilog and subtracting one.
(3) We assume that the systematic risk is highly correlated with a firm's total risk (systematic and unsystematic) as measured by earnings variability. We proxy risk using changes in the standard deviation of earnings, where earnings is defined as the combination of both operating and nonoperating earnings. Preliminary analysis of the data revealed that many firms have substantial nonoperating income. Therefore, we decided to include nonoperating income in the earnings definition.
(4) In addition, Frankel  provides numerous citations documenting the "soaring" price of land in Japan in the 1980's.
(5) Kane and Unal  develop a model estimating hidden value from balance sheet sources. Land exists as hidden capital since the accounting measure used to estimate its value causes the firm's reported net worth to understate its true economic value.
(6) Kester  discusses special reserves and notes that reserve provisions are instituted annually and are applicable to specific industries.
(7) The keiretsu form of corporate grouping is the successor to the zaibatsu organization which was broken up after World War II by the occupation forces.
(8) For example, McDonald  finds that double counting due to cross-ownership accounts for at least 24% of Japan's reported market capitalization.
(9) We also adjusted book value of land by the price index for industrial property and in a separate set of regressions obtained similar results to those reported here.
(10) Using only five years of data may be a potential weakness in the methodology, but data limitations prevented the use of annual earnings data prior to 1975.
(11) See Ezzamel and Mar-Molinero  for a discussion of the validity of this method of dealing with outliers.
(12) An alternative proxy for land, LAND5, measured as the percentage change in the book value of land weighted by the appropriate property land index shows an average increase of 1,436.2% from 1979 to 1989.
(13) The question of whether institutional and related stock holdings are simultaneously determined with PE ratios is not directly addressed in this study. A possible way to answer this question may be through using a two-stage least square regression, estimating the institutional holdings as a function of all other variables and then using the predicted holdings from these equations as the independent variable.
(14) See Litzenberger and Rao  for a discussion of the reasons behind this E/P transformation. In addition to the E/P regression, a regression of the variables on PE is also performed. This is done because it is felt that given the outlier adjustment described earlier, the additional adjustment associated with inverting the PE may have actually caused a loss of information about some of the existing relationships.
(15) For example, if a firm has earnings of $1.00 and price of $20.00 per share (EP = 0.05) in 1979 and earnings of -$1.00 with a price of $20.00 per share (EP = -0.05) in 1989, [Delta]EP will be a negative 200% for the period. Conversely, if earnings increase by the same amount to $3.00 per share in 1989, while price remains constant (EP = 0.15), then [Delta]EP will be a positive 200%.
(16) In order to examine the sensitivity of the LAND related results to the specification of the proxy, we test alternative land proxies in our regressions. The first, LAND3, is calculated as the natural log of the total area of land owned at the beginning of the ten-year period. While this alternative land area proxy does not represent a change in land ownership or value over the period, it does capture the relative size, across different firms, of the hidden land asset prior to the increase in land prices during the 1980's. The regression results with this proxy reveal that land is positively related to MV at the 0.01 level of significance and negatively related to EP at the 0.05 level. The second proxy, LAND4, is calculated as the natural log of the yen amount of change in land value over the period. this proxy does represent a change in land value over the period but is not scaled, as is the proxy used in Exhibit 4. Regression results reveal land is positively related to MV and negatively related to EP at the 0.01 level of significance. The third proxy, LAND5, is calculated as the percentage change in the book value of land weighted by the appropriate property value index. Regression results reveal land is not significantly related to MV or EP. The results for other proxies in these alternative regressions are similar to those reported in Exhibit 4. These results may reflect a firm-size effect that is related to land value. Unfortunately, a close examination of the relations between firm size, land values, and firm performance is beyond the scope of this paper.
(17) The question of whether both PE ratios (market value) and land are driven by the same underlying factors, such as low interest rates, is beyond the scope of this paper. To address this issue and remove the effect of interest costs, the dependent variables (PE ratio and market value) and our land variable can be restated by regressing the absolute and relative treasury interest rates against these variables and then using the residuals from these equations as the dependent (PE ratio residual, market value residual) or independent (land residual) variables.
(18) The results for the PE regression described in footnote 14 are as follows: [[Beta].sub.0]=4.87 (t=3.71), [[Beta].sub.1]=-0.02 (t=0.26), [[beta].sub.2]=0.54 (t=3.12), [[Beta].sub.3]=-0.11 (t=0.48), [[Beta].sub.4]=1.23 (t=2.46), [[Beta].sub.5]=0.01, (t=1.34), [[beta].sub.6]=0.06 (t=0.47), [[beta].sub.7]=-34.13 (t=4.65), [[Beta].sub.8]=-0.53 (t=4.35), [[beta].sub.9]=1.40 (t=13.37), [[beta].sub.10]=2.27 (t=1.65) with a significant F-statistic of 28.13 and an [R.sup.2] of 36.40%. The results for the PE regression model are stronger than those associated with the EP model. Not only is the overall model performance better (adjusted [R.sup.2] of 36% as opposed to 21%), but two of the ownership proxies (CORP and INDV) are found to have significant relations to PE. This suggests that the Litzenberger and Rao E/P transformation may obscure some relationships and that the outlier adjustment alone may alleviate most of the outlier problems associated with using PE as proxy variable.
 R. Aggarwal, R.P. Rao, and T. Hiraki, "Skewnes and Kurtosis in Japanese Equity Returns: Empirical Evidence," Journal of Financial Research (FAll 1989), pp. 253-260.
 S. Basu, "Investment Performance of Common Stock in Relation to Their Price-Earnings Ratios: A Test of the Efficient Markets Hypothesis," Journal of Finance (June 1977), pp. 663-682.
 R.A. Brealey and S.C. Myers, Principles of Corporate Finance, New York, McGraw-Hill, Inc., fourth edition, 1991.
 T.J. Cook and M.S. Rozeff, "Size and Earnings/Price Ratio Anomalies: One Effect of Two?" Journal of Financial and Quantitative Analysis (December 1984), pp. 449-466.
 H. Demsetz and K. Lehn, "The Structure of Corporate Ownership: Causes and Consequences," Journal of Political Economy (1985), pp. 1155-1177.
 M. Ezzamel and C. Mar-Molinero, "The Distributional Properties of Financial Ratios in UK Manufacturing Companies," Journal of Business Finance and Accounting (Spring 1990), pp. 1-29.
 J.A. Frankel, "The Japanese Cost of Finance: A Survey," Financial Management (Spring 1991), pp. 95-127.
 L.P. Freitas, "Views from Abroad: Japan," Journal of Accounting, Auditing and Finance (Fall 1980), pp. 70-76.
 L.P. Freitas and T. Otsuki, "Debt Management Practices and Their Effect on Cost of Capital: Japan and U.S.," in Miyasato, Seigen and T. Otsuki (eds.), Special Project on U.S.-Japan Economic and Trade Relations, Japan, International University of Japan, 1988.
 K.R. French and J.M. Poterba, "Were Japanese Stock Prices Too High?" Journal of Financial Economics (October 1991), pp. 337-363.
 I. Friend and I. Tokutsu, "The Cost of Capital to Corporations in Japan and the U.S.A.; Journal of Banking and Finance (1987), pp. 313-327.
 Y. Hamao, "Japanese Stocks, Bonds, Bills, and Inflation, 1973-87," Journal of Portfolio Management (Winter 1989), pp.20-26.
 J.E. Hodder and A.E. Tschoegl, "Some Aspects of Japanese Corporate Finance," Journal of Financial and Quantitative Analysis (June 1985), pp. 173-191.
 J. Jaffe and R. Westerfield, "Patterns in Japanese Common Stock Returns: Day of the Week and Turn of the Year Effects," Journal of Financial and Quantitative Analysis (June 1985), pp. 261-272.
 J. Jaffe and R. Westerfield, "Is There a Monthly Effect in Stock Market Returns? Evidence from Foreign Countries," Journal of Banking and Finance (1989), pp. 237-244.
 M. Jensen and W. Meckling, "Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure," Journal of Financial Economics (October 1976), pp. 305-360.
 E.J. Kane and H. Unal, "Modeling Structural and Temporal Variation in the Market's Valuation of Banking Firms," Journal of Finance (March 1990), pp. 113-136.
 K. Kato and J.S. Schallheim, "Seasonal and Size Anomalies in the Japanese Stock Market," Journal of Financial and Quantitative Analysis (June 1985), pp. 243-260.
 A.J. Keown, J.M. Pinkerton, and S.N. Chen, "Portfolio Selection Based Upon P/E Patios: Diversification, Risk Decomposition and Implications," Journal of Business Finance and Accounting (Summer 1987), pp. 187-198.
 C.W. Kester, "Capital and Ownership Structure: A Comparison of United States and Japanese Manufacturing Corporations," Financial Management (Spring 1986), pp. 5-16.
 R.H. Litzenberger and C.U. Rao, "Estimates of the Marginal Rate of Time Preference and Average Risk Aversion of Investors in Electric Utility Shares: 1960-1966," Bell Journal of Economics (Spring 1971), pp. 265-279.
 J. McDonald, "The Mochiai Effect: Japanese Corporate Cross-Holdings," Journal of Portfolio Management (Fall 1989), pp. 90-95.
 D. Meerschwam, "The United States and Japan's Dangerous Liaison and the Irrelevance of the Cost of Capital," Harvard Business School, 1990.
 A. Michela and I. Shaked, "Japanese Leverage: Myth or Reality?," Financial Analyst Journal (July/August 1985), pp. 61-67.
 S.C. Myers, "Determinants of Corporate Borrowing," Journal of Financial Economics (1977), pp. 147-175.
 T. Nishiyama, "The Structure of Managerial Control: Who Owns and Controls Japanese Business?" Japanese Economic Studies (Fall 1982), pp. 33-37.
 H. Okumura, "Interfirm Relations in an Enterprise Group: The Case of Mitsubishi," Japanese Economic Studies (Summer 1982), pp. 53-82.
 M.R. Reinganum, "Misspecification of Capital Asset Pricing, Empirical Anomalies Based on Earnings' Yields and Market Values," Journal of Financial Economics (1981), pp. 19-46.
 M.R. Reinganum, "A Direct Test of Roll's Conjecture on the Firm's Size Effect," Journal of Finance (March 1982), pp. 27-35.
 M.R. Reinganum, "The Anomalous Stock Market Behavior of Small Firms in January: Empirical Test for Tax-Loss Selling Effects," Journal of Financial Economics (June 1983), pp. 89-104.
Richard L. Constand is an Assistant Professor of Finance and Lewis P. Freitas is a Professor of Finance at the University of Hawaii, Honolulu. Michael J. Sullivan is an Assistant Professor of Finance at the University of Nevada, Las Vegas.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||International Finance Special Issue|
|Author:||Constand, Richard L.; Freitas, Lewis P.; Sullivan, Michael J.|
|Date:||Dec 22, 1991|
|Previous Article:||Cross-border liability, border taxes, and capital structure.|
|Next Article:||Management ownership and firm compensation policy: evidence from converting savings and loan associations.|