Feb. 5, 1991 (rev. 3/4/91) The Cost of Capital in Japan: Update Jeffrey A. Frankel Professor of Economics University of California Berkeley, CA 94720 The author is also a Research Associate of the National Bureau of Economin Research and a Visiting Scholar of the Institute of International Economics. This article is forthcoming in Business Economics, April 1991. Many economists were skeptical, ten years ago, when George Hatsopoulos began arguing that Japanese corporations had an unfair advantage over their American competitors in the form of a lower cost of capital.1 In the course of the 1980s, however, the theory gained credibility. In the view of most mainstream economists, such as Martin Feldstein, the origin of the record U.S. trade deficits that emerged after 1982 lay in the fall in national saving. While this school of thought was not consistent with the viewpoint that blamed the trade deficit on unfair trade policies, it did fit in neatly with the argument that U.S. corporations faced a shortage of capital, as reflected in real interest rates that were higher domestically than in Japan, and that productivity growth was suffering as a result.2 A survey of the now-voluminous literature on the cost of capital in Japan and the United States in the 1980s reveals that many different factors are potentially involved in evaluating the possible differential.3 Nevertheless, a robust conclusion emerges. Throughout the 1980s, up to 1989, the cost of capital was indeed -by a variety of measures -- lower in Japan than in the United States. Since 1989, there have been dramatic developments in Japanese financial markets. A new Bank of Japan governor, less enthusiastic about buying dollars to support the U.S. currency than some others in the Japanese government and more intent on fighting inflation, has tightened Japanese monetary policy, raising real interest rates steeply in 1989 and 1990. The Japanese stock market fell sharply at the beginning of 1990, presumably as a result of the increase in interest rates, and fell again in the Fall, presumably as a result of the beginning of the Kuwait crisis. This article reviews findings on the Japanese cost of capital in the 1980s. These findings include evidence of liberalization and other structural trends that have been underway in Japanese financial markets for some years, accelerated to a certain extent by pressure from the U.S. Treasury. It then seeks to update the analysis to reflect the developments of 1990. The conclusion is that the increase in real interest rates and the accompanying fall in the stock market in Japan have been great enough virtually to eliminate the difference in the cost of capital vis-a-vis the United States. This still leaves open the question whether or not this recent development is "good news," and what American policy should be in talks with the Japanese government on financial policy. 1. Japan's Weighted-Average Market Cost of Capital in the 1980s A standard measure of the corporate cost of capital is the weighted average of the cost of debt and the cost of equity finance. Many would argue that this standard definition is not relevant for Japan, because corporations have traditionally obtained little of their capital by issuing securities on the open market, but rather have relied on bank loans and retained earnings (e.g., Meerschwam, 1989). This point is addressed in the fourth section of this article. For the moment, the standard measure will serve well to illustrate three of the arguments most commonly made to show that Japan had a lower cost of capital in the 1980s. Under the standard definition, the claim can be broken down into some combination of 1 the following three possibilities: (a) the cost of borrowing was lower in Japan, (b) the cost of equity was lower in Japan, or (c) the weight on debt-financing (versus equity-financing) was higher in Japan. All three statements contain some truth. Calculations using 10-year government bond yields suggest that Japanese real interest rates were below U.S. real rates virtually continuously from 1967 to 1988. The evident reason for low Japanese real interest rates was a high saving rate. The real differential narrowed after 1984, presumably due to the influence of increasingly-tightly integrated international financial markets, but had not disappeared by 1988. In the past, Japanese corporations have had a much higher ratio of debt to equity than U.S. corporations, that is, they have been much more highly leveraged. In the period 1970-72, for example, debt/equity ratios in Japan were four times as high as in the United States. This commonly-observed characteristic of the Japanese system is one major reason why calculations often show a lower overall cost of capital in Japan than in the United States; equityfinancing is known to be more expensive than debt-financing in any market, presumably because portfolio investors demand a higher expected return on equity to compensate them for higher risk. How have Japanese firms been able to rely so heavily on debt? As a number of authors have pointed out, a particular debt/equity ratio that would be very risky for a U.S. firm would have been less risky for a Japanese firm. There are several reasons for this. Much of the borrowing, particularly for members of a keiretsu, was from the firm's main bank. A main bank would not cut off lending in time of financial difficulty; to the contrary it would do all it could to see the company through. Another reason is that, until recently, all loans had to be collateralized. In any case, it is important to note that the seemingly robust regularity that "Japanese firms are highly leveraged" now appears to be a thing of the past. The debt/equity ratio declined throughout most of the 1970s and 1980s, and by one measure had by 1986 fallen to the level in the United States.4 The third of the standard components of the overall cost of capital, the cost of equity financing, is the most ambiguous of the components to measure. One approach has been to use the realized market rate of return on equity, i.e., the dividend/price ratio plus the rate of increase of equity prices. The problem with this approach is that stockholders' realized rate of return on equity is a very noisy indicator of ex ante expectations. In the absence of a speculative bubble, stock prices can be thought of as the present discounted value of expected future dividends or free cash flow. The equity cost of capital is the implicit discount rate, and thus can potentially be inferred from the ratio of stock prices to current dividends or earnings -whichever denominator can more reliably be used to estimate future prospects. We consider the subject of the price/dividend ratio first, and then turn to the price/earnings ratio. There has been no upward trend in Japanese dividends per share over the last 20 years. This makes it especially difficult to explain the high level of Japanese stock prices, if one follows the 2 common approach of choosing the present-discounted-value-of-futuredividends formula and estimating expected dividends from actual realized dividends. On the other hand, the observed high level of prices relative to dividends would be perfectly understandable if the increase in dividends were thought still to lie in the future. One possible explanation as to why Japanese dividends have been low in the past is that Japanese corporations have over the postwar period had many profitable investment opportunities, but until the late 1980s have not had sufficiently free access to securities markets to drive their cost of capital into equality with the rate of return on these investments. For this reason, they have used retained earnings to finance investment. Such considerations suggest that looking at the past history of dividends may not be a very useful way to estimate future dividends. An alternative approach is to look at the amount of earnings the firm is required to generate per unit of equity, that is, the inverse of the price/earnings ratio P/E. The P/E ratio (like the price/dividend ratio) has been observed to be higher in Japan than in the United States ever since the early 1970s. Because this difference could be explained by a lower discount rate in Japan, it is often the basis of arguments that the cost of equity capital is lower in Japan. 2. Price/earnings ratios Some, such as Ando and Auerbach, have looked at the priceearnings ratio because they are interested in the cost-of-capital question, and they consider P/E to be inversely related to the required rate of return re. Others, such as French and Poterba (1989) are interested in the price-earnings ratio to evaluate whether the late-1980s increase in prices was excessive. They note that, as of the end of 1988, Japan's reported P/E was more than four times that in the United States. So big an apparent discrepancy would be difficult to explain. If earnings (as a commonly-used proxy for free cash flow) were expected to grow at rate ge, then the earnings/price ratio should equal re - ge in order for the return to investment to be equalized to the opportunity cost of funds. The end-1988 differential between reported earnings/price ratios in the United States and Japan was .06 [=.078-.018]. A 6-point difference in the required rate of return on capital re would support the cost-of-capital-advantage school, but seems too large to be plausible. French and Poterba (1989), Ando and Auerbach (1988), and others have emphasized the importance of correcting earnings for a number of measurement problems. Ando and Auerbach focussed on three distortions related to inflation: depreciation accounting, inventory accounting, and accounting for nominal liabilities.5 French and Poterba pointed out some additional corrections to make to reported earnings. First, earnings reported by U.S. corporations include the profits of subsidiaries, while those reported by Japanese firms do not (only actual dividends received from subsidiaries). Second, reported Japanese earnings deduct (both on the firms' tax returns and on their financial statements) generous allowances for special reserves for such possible future contingencies as product returns, repairs, and retirement benefits. 3 Third, Japanese firms often take greater depreciation allowances. All three factors work to make reported earnings look smaller in Japan, and P/E ratios larger. French and Poterba found that correcting for all three accounting differences in earnings explained about half of the difference between Japanese and U.S. ratios, but still left corrected end-1988 Japanese P/E ratios at about twice U.S. levels. The difference between the U.S. and Japanese E/P ratios is about 5.4 per centage points after the correction (averaging two methods). A more rapid expected growth rate ge in Japan could explain only part of this difference. The real growth rate of the Japanese economy had averaged 1.6 per cent faster than the U.S. economy over 1980-88. There was no particular reason to expect the real growth rate of the economy to increase in the future, or to expect the growth rate of earnings or cash flow to be higher than the growth rate of GNP. It seems that a cost of equity capital re which was lower in Japan than in the United States is needed to explain the rest of the difference. This leaves the question of why the cost of equity capital was lower. We already observed that the real interest rate was lower in Japan in the 1980s. In addition, it is possible that the equity premium, the compensation for holding risky stocks beyond the return on holding bonds, was also lower in Japan, though several studies on the subject have failed to find evidence of the lower level of riskiness in the Japanese market that would merit a lower risk premium. It has also been suggested that the lower cost of equity capital in Japan is due to more favorable tax treatment of investment. It would presumably be more convenient for any American businessman who wished to claim that Japanese industry had an "unfair advantage" in the form of a low cost of capital, if the source of the advantage were more favorable tax treatment by the Japanese government. Ando and Auerbach (1988), Shoven (1989), and many others have looked at the corporate tax question, from a number of different angles. As many studies have found that the tax system in the 1980s was less favorable to investment in Japan as the other way around. The overall finding seems to be that differences in tax treatment tend to be dwarfed by other factors like differences in real interest rates. Furthermore, tax reforms that took effect in Japan in April of 1988 and April of 1989 have now narrowed some of the existing differences. The lower cost of both debt and equity finance in Japan appears to stem primarily from a higher private saving rate, and not from tax differences. Some observers have explained the tremendous run-up in stock prices in the late 1980s by the appreciation in the land held by corporations. This hypothesis, even if accurate, only pushes the question back one stage. If rents are expected to grow at rate gr, then the price/rental ratio should be given by 1/(r - gr). Thus the same possible explanations arise for high land prices as arise for high equity prices: a low discount rate r or a high growth rate gr.6 It appears likely that each of these factors explains part of the gap between land price/rental ratios in Japan and the United States as of the end of the 1980s. 4 But there remains a puzzle as to why land price/rental ratios in Japan increased so much in the 1980s. Real interest rates were even lower and growth rates even higher in earlier decades. Why did price/rental ratios not reach their high levels until the late 1980s? The same question applies to the price/dividend and price/earnings ratios in the stock market: if the lofty levels of the end of the 1980s were due to a low real interest rate and high growth rate, then why were these ratios lower, rather than higher, in earlier decades? Were the late-1980s run-ups in land and equity markets due to speculative bubbles? 3. Update for the Events of 1990 We have seen that the required rate of return appeared to be lower in Japan in the 1980s than in the United States. Although a lower real interest rate probably could not in itself explain the sky-high P/E ratios in the Japanese stock market at the end of the decade, it is possible that a combination of the low discount rate, the high expected growth rate, and the need to correct accounting differences in the measurement of earnings, could together explain the end-1988 P/E level. But we have not yet addressed the possibility of a speculative bubble. In 1990, the Japanese stock market lost almost half its value. At first glance, this plunge could be interpreted as clear evidence that the run-up of prices in the late 1980s was indeed a speculative bubble. Unfortunately for this view, the macroeconomic fundamentals changed dramatically at the same time. The Bank of Japan began to tighten Japanese monetary policy in 1989, raising real interest rates steeply in 1990. In this section we try some simple calculations to see if the changes in macroeconomic fundamentals can explain the decline of the Japanese stock market between late 1989 and the end of 1990. The calculations use monthly survey data collected from a sample of banks, multinational corporations and other forecasters by Alan Teck's Currency Forecasters' Digest of White Plains, New York. Table 1 estimates the Japanese 10-year real interest rate to have been 2.61 per cent in September 1989, and the 10-year expected rate of economic growth to have been 3.75 per cent. One is tempted to take the difference r - g as an estimate of re - ge and see if it equals the ratio of earnings to prices. But the difference, -1.14, is less than zero, and would thus apparently be capable of explaining any P/E ratio, no matter how high. Clearly the real interest rate underestimates the required rate of return on capital -- presumably due to a risk premium of the sort discussed above in the section on price/earnings ratios -- or else the GNP growth rate overestimates the rate of growth of earnings.7 French and Poterba's figures, after adjustment of earnings, show that the U.S. E/P ratio exceeded the Japanese E/P ratio by about 5.4 per cent in late 1988. P/E ratios were almost the same at the end of 1989 as they were at the end of 1988. Table 1 shows an apparent "overvaluation" of the Japanese stock and land markets at the end of the 1980s, beyond what could be attributed to real interest rates or expected growth rates in Japan and the United States. This overvaluation could be attributed either to (1) a speculative bubble, or (2) a higher equity risk premium for the 5 United States than for Japan, or some other source of bias in r - g (as an estimate of the difference between the relevant discount and growth rates) that is greater for the United States than for Japan. Let us consider the second hypothesis, and assume that the difference in risk premiums (or other source of bias) between the two countries remained the same at the end of 1990 as in late 1989: about 3 per cent if unadjusted earnings figures are used. Is the 1990 increase in Japanese real interest rates capable of explaining the collapse of the Japanese market? As of December 1990, the real interest rate in Japan was up by 1.29 points. As a result the difference between the U.S. and Japanese estimates of r - g was down by 0.57 points from September 1989. Over the same period, the difference in (unadjusted) E/P ratios fell by 0.61 points. Thus the increase in real interest rates in Japan can explain most of the decline in the stock market in 1990 vis-a-vis the United States, and there is no obvious need for recourse to the hypothesis of a burst speculative bubble. What are the implications of the 1990 developments for the cost of capital question? The difference in real interest rates between the United States and Japan has disappeared completely. There is probably now little left also of the difference in P/E ratios once the accounting adjustments are made to Japanese earnings.8 In short, though by most measures the cost of finance in the 1980s was cheaper in Japan than in the United States, this appears no longer to be the case. Whether this is cause for rejoicing among American businessmen is another question. Given the high degree of international integration that has taken place over the last ten years, fluctuations in saving are reflected in capital flows between Japan and the rest of the world as easily as in domestic investment. In other words, corporate borrowers in Japan are not the only ones to feel the effect of a decreased availability of Japanese savings; borrowers in the United States and elsewhere in the 1990s will feel it as well. 4. The Increasing Relevance of "The Market" in the Japanese Financial System It is true that in the past, the standardly-computed weightedaverage cost of issuing securities in the market has not been entirely relevant in Japan. The fourth respect in which the cost of capital can be said to have been lower in Japan in the past involves the benefits of internal and main-bank financing. These benefits have been persuasively described as mechanisms for avoiding problems of imperfect information or incentives, where such problems can complicate the financing of investment projects by issuing securities on the market.9 Such benefits are less easily quantifiable than the first three factors (real interest rate, cost of equity finance, and debt/equity ratio), but the direction has been clear: for any given level of the measured cost of capital, many Japanese corporations, certainly the members of keiretsu, have faced a lower "effective" cost of capital, implying they have been more inclined to undertake an investment with given prospects, than an American firm would be in the same situation. The market is becoming more and more relevant in Japan. Established banking relationships began to break down in the 1980s 6 and the market began to take their place, as corporations began to use banks less and bond markets more. This process accelerated as the result of international liberalization as well as domestic deregulation. This increasing market-orientation provides the explanation for one of the greatest puzzles to emerge from the literature on Japanese finance: the increase in stock and land prices in the 1980s. As mentioned above, if some combination of low real interest rate and a high expected growth rate explains the high levels of Japanese stock and land prices in the late 1980s, these prices should have been even higher in the 1970s, when real interest rates were even lower and expected growth rates even higher. The answer is that the arbitrage between the interest rate and real assets that we take for granted in a market-oriented system was not entirely relevant in the earlier period. In the 1980s the increased availability of funds that could be used for asset-market arbitrage allowed the great run-up in equity prices and land prices. But we are left with another, even greater puzzle: if "main bank" relationships were indeed so advantageous to Japanese corporations, why have they now been abandoning these relationships in favor of the market? It is possible that "insiders," those corporations with access to preferentially priced funds, had an advantage over "outsiders," and that this advantage was lost recently when the latter gained access to the escape route of borrowing abroad. If the outsiders had previously been subsidizing the insiders, their escape from the closed system may have driven up the cost of capital for the former. It may not be possible for trust and long-term relationships to survive in an environment where new-comers deal only in explicit contracts. 5. Conclusions for Policy To summarize, all four respects in which Japan could be said to have a lower cost of capital ten years ago have recently either diminished or vanished altogether. Real interest rates have risen to U.S. levels, partly as a result of the removal of Japanese barriers to capital outflow. The required rate of return to equity capital has risen in tandem. The ability of Japanese corporations to lever themselves highly has diminished, as the Japanese system gradually moves from one of stable long-term relationships to one that is more market-oriented. Finally, this same trend of increasing market-orientation has undermined the previous ability of keiretsu-affiliated corporations to obtain loans from their main banks on terms that were more favorable than can be obtained in American-style securities markets. The structural trends of international and domestic liberalization in Japanese markets have, to a significant extent, been accelerated over the past ten years by pressure from the U. S. Government. The usual skeptical American view underlying negotiations with Japan, that significant demands will either not be agreed to speedily or not be implemented faithfully, does not fit the events well in the case of financial matters. The U.S. Treasury during the first Reagan Administration asked Japan's Ministry of Finance to remove its remaining controls on international capital flows and otherwise to liberalize its financial markets. The talks 7 culminated in the Yen/Dollar Agreement of May 1984 in which the Ministry of Finance acceded to the U.S. demands.10 Notwithstanding the rapid pace of Japanese liberalization, American pressure has continued. One of the U.S. requests in the Structural Impediments Initiative, which produced an agreement in June 1990, was a loosening of the keiretsu relationships. But the logic underlying U.S. demands on Japan is confused. Three rationales have been explictly offered by the U.S.Treasury. One rationale has been to increase the flow of capital from Japan to the United States, to give American corporations access to the same cheap capital as their Japanese rivals, thereby allowing them to increase investment and long-term productivity growth. Another rationale was to appreciate the yen against the dollar and reduce the U.S. trade deficit. (This original motivation was explicit on the part of Treasury Secretary Don Regan, though it contradicts the first because the trade balance and the capital flow are one-andthe-same.) A third rationale on the part of the Treasury has been to promote the efficiency of the Japanese economy. But American businessmen and Congressmen, whose concerns with the bilateral trade imbalance constituted the political impetus behind the Administration campaigns, wish to improve the competitiveness of the American economy, not of the Japanese. Perhaps cozy Japanese banking relationships have been an effective way of avoiding information and incentive problems, in which case the rationale for recent U.S. action would seem clear: to drag Japanese efficiency down to the level of the United States. Or, on the other hand, perhaps the Japanese financial system will benefit from its increasing market-orientation. Either way, valid American concerns that U.S. productivity growth has been hampered by inadequate investment in plant and equipment, R and D, and education, should be addressed by looking to our own economy, rather than making demands on Japan that are conflicting, changeable, and poorly thought out. * * * * 8 References Ando, Albert, and Alan Auerbach, 1988, "The Cost of Capital in the U.S. and Japan: A Comparison," NBER Working Paper no. 2286. Journal of the Japanese and International Economies, vol.2, pp. 134-158. Frankel, Jeffrey, 1984, "The Yen/Dollar Agreement: Liberalizing Japanese Capital Markets," Policy Analyses in International Economics no. 9, Washington, D.C.: Institute for International Economics. Frankel, Jeffrey, 1991, "The Japanese Cost of Finance: A Survey," Forthcoming in Financial Management, Spring. [Parts of this paper draw on National Bureau of Economic Research Working Paper No. 3156.] French, Kenneth and James Poterba, 1989, "Are Japanese Stock Prices Too High?" revised, National Bureau of Economic Research, Aug. Hale, David, 1990, "Economic Consequences of the Tokyo Stock Market Crash," U.S.-Japan Consultative Group on International Monetary Affairs, Washington, D.C., July 23. Hatsopoulos, George, 1983, "High Cost of Capital: Handicap of American Industry," Study sponsored by the American Business Conference and Thermo Electron Corp., April. Hatsopoulos, George, and Stephen Brooks, 1986, "The Gap in the Cost of Capital: Causes, Effects, and Remedies," in Ralph Landau and Dale Jorgenson, eds., Technology and Economic Policy, Cambridge, MA: Ballinger 221-280. Hatsopoulos, George, Paul Krugman, and Larry Summers, 1988, "U.S. Competitiveness: Beyond the Trade Deficit," Science, July 15, 299-307. Hoshi, Takeo, Anil Kashyap, and David Sharfstein, 1990, "Corporate Structure, Liquidity, and Investment: Evidence from Japanese Panel Data," Quarterly Journal of Economics, forthcoming. Meerschwam, David, 1989, "The Japanese Financial System and the Cost of Capital," NBER conference on The US and Japan: Trade and Investment, Paul Krugman, ed., forthcoming, University of Chicago Press, Chicago. Shoven, John, 1989, "The Japanese Tax Reform and the Effective Rate of Tax on Japanese Corporate Investments," NBER Working Paper no. 2791; in L.Summers, Tax Policy and the Economy, M.I.T. Press: Cambridge, MA. 9 Table 1 Can macroeconomic fundamentals explain the 1990 P/E decline? (figures in per cent) Sept. 1989 Dec. 1990 Japan1 interest rate (10-year) expected inflation rate (CPI, 1989-1998) real interest rate r ____ 2.61 ____ 3.90 expected real growth rate (1989-1998) g 3.75 4.05 -1.14 -0.15 4.86 2.25 r - g 6.60 2.70 United States2 interest rate (10-year) expected inflation rate (CPI, 1989-1998) real interest rate r ____ 3.40 ____ 3.57 expected real growth rate (1989-1998) g 2.70 2.45 0.70 1.12 8.15 4.75 r - g 7.97 4.40 change Differential (r-g)US - (r-g)J 0.57 1.84 1.27 Japan E/P (unadjusted)3 +0.87 U.S. E/P (unadjusted)4 +0.26 1.60 2.47 6.47 6.73 - Differential (E/P)US - (E/P)J 4.87 4.26 0.61 .. Hatsopoulos (1983) and Hatsopoulos and Brooks (1986). .. In 1988, the different strands -- the businessman's concerns 1 2 3 4 From Currency Forecasters' Digest. From Currency Forecasters' Digest. From Nikkei data bank. From Economic Indicators, Council of Economic Advisers. 10 about American competitiveness and the macroeconomist's concerns about national saving and (as well as possible inefficiencies in American financial markets) -- were elegantly fused together in an article co-authored by Hatsopoulos, Krugman and Summers. .. Frankel (1991). .. French and Poterba (1989). .. Most importantly, Japanese firms have relied more on debt than equity (see above), so the fact that inflation reduces the real value of outstanding liabilities has been more important for them. .. Inferring the discount rate from observed price/rental ratios for land has an advantage over looking at price/earnings ratios for equities: in the case of land each period's rent is a conceptually correct measure of the return on holding land, whereas in the case of equities only that portion of the earnings that is not reinvested is a conceptually correct measure of the return. In the case of Japan, the true discount rate should be even lower than that inferred from the P/E ratio, because a higher proportion of earnings are reinvested than in the United States. .. As noted earlier, the capitalization formula does not strictly apply to P/E ratios, because the portion of earnings that are reinvested are not available as returns to the stockholder. (This just makes the gap between the discount rate growth rate differential and the E/P ratio that much harder to explain.) One would be on firmer theoretical foundations to match up the calculations reported in this section to observations on the price/dividend ratio, for the case of stocks, or the price/rental ratio, for the case of land. Unfortunately, reliable recent data for land are not available. .. Some may continue to believe that the standard weighted average of debt and equity is not relevant for Japan because many corpora-tions still get much of their financing from main banks. Even in the case of bank borrowing, however, there is reason to think that the era of cheap finance is over. Japanese banking was itself the industry hardest-hit in the 1990 stock market collapse, and is now under pressure to restrict lending in order to meet stringent new international standards for capital adequacy. Hale (1990). .. Hoshi, Kashyap and Sharfstein (1990) study the investment patterns of a sample of Japanese firms, segregated into those that have affiliations with a large bank through their keiretsu and those that do not belong to such a group; one possible conclusion is that "the institutional arrangements in Japan may offer Japanese firms an important competitive advantage (p.24)." See Part IV of Frankel (1991) for a review of the literature. ... Frankel (1984). 11