«Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.8, No5, November 2012 563 Why Did the Debt Maturity of the Japanese ...»
The “Aggregate” interest spread was at high level of more than 1.5% during the period of monetary easing in the year 1986-88. Then, it has dropped to the negative value temporarily during the period of monetary tightening by the Bank of Japan in the year 1990-911. Given the fact that the decline in short-rate in this period was fairly rapid, it is conceivable that the adjustment of long-term interest rates lagged behind, resulting in the rare negative long-short spread.
As the asset price bubble collapses, monetary policy had been loosened again and long-short spread went back to the level of over 0.5% by the year 1992. Then, after increasing to the level close to 0.8% in 1995, slowly it has started to decrease once again. Since the domestic financial crisis in the year 1997/98, it has remained at a level of around 0.5% (the numbers for the “domestic banks” spread after 2002).
The long-short spread data on new loans does exist for the years after 1993. It has started to decline since 1999, and remained at a very low level of less than 0.1% from the year 2001 until
2005. Although it had a slight increase during the year 2006 and 2008, the size of the spread was still only less than 0.2%. Therefore it is unlikely that the decrease in the level of interest rates had prompted the increase of debt maturity in the year 2000, when focused on the long-short spread of the new loan rates.
In the Panel B, the spread between loan and deposit rates are shown, using two different deposit rates, namely the long-term rate (more than 360 days) and short-term rate (150-180days). The short-term (loan - deposit) stands for the spread between the short-term loan and deposit rates and the long-term (loan-deposit) stands for the spread between long-term loan and deposit rates. “Long-term loan short-term deposit” is the difference between long-term loan rate and the short-term time deposit rate.
Although short-term interest rates of Bank of Japan rose at a considerably fast pace, initially the market participants seemed to beleive that the monetary tightening was only temporary, so the yield curve of government bonds at the time briefly went into a downward-slope.
T Iwaisako / Public Policy Review Figure 3: Various Interest Rate Spreads Panel A: Long-Short Loan Rate Spreads Panel B: Loan and Deposit Rate Spreads Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.8, No5, November 2012 571 For these loan-deposit spreads, the data are available only from the late 1980’s, so it is difficult to compare them in the pre-Bubble period and in the “Bubble economy” period. After monetary policy tightening around the year 1990, these spreads have significantly declined.
Then there was a large increase in 1992. After reaching their peaks in 1995, a gradual downward trend is observed for the long-short spread.
The differences between short-short and long-long “Loan-Deposit” spread was at around 0.7-1.0% in the 1990’s, but gradually declined to just under 0.5% by the late 2000’s. Judging from the loan and deposit rate spread movement, the long-term loan had become attractive for the financial institutions in the mid-1990s, when the increasing trend in debt maturity has slowed down. Then it has started to be less and less attractive in the period subsequent to the domestic financial crisis while the increasing trend of the debt maturity has become prominent again.
Overall, the evidences found from two graphs in Figure 3 seem to suggest that the benefits of increasing long-term loans for the financial institutions have declined over the years.
Therefore, it is unlikely that the increase in debt maturity in this period will be explained by the supply side factor. However, the discussions in this section are based on the evidence at the level of the average interest rate. I cannot be sure whether the conclusion will remain the same if the other risk factors such as default risks and the heterogeneity of the lender firms are taken into account.
4. Explanation based on the theory of financial intermediation.
During the period of rapid economic growth in 1950s and 60s, the Japanese companies had depended almost entirely on the bank loans for fundraising. Since the first oil shock in early 1970s, non-financial corporations have cut back their reliance on bank borrowings. It has been widely argued that borrower companies have tried to diversify their funding sources to increase the bargaining power in the debt contracts and mitigate managerial influence from the lender banks.
In this section, I discuss how the increase of average debt maturity of Japanese firms could be understand from the prevailing view in the recent years that firms have been trying to get rid of the influence of their main banks (such as Chapter 7 and 8 in Hoshi and Kashyap, 2001,).
(1) Choice of debt maturity and the reduction of debt as signaling devices.
Theoretically, the bank will have the strongest influence to the management of the borrowing company when the majority of existing loan contracts has reached their maturities and the firm has to refinance, i.e., when the firm wants to roll over its debts. If this argument was correct, the bank prefers short-term lending. If loan was carried out for the long-term, the T Iwaisako / Public Policy Review bank’s bargaining power will be weaker for the long-term, until the next timing of debt roll over comes to near future. Hence, the bank will offer the shorter debt maturity to the firm, leaving the borrower with the uncertainty regarding refinancing by repeating the short-term loan rather than the long-term loan contracts. In other words, in the long-term lender-borrower relationship, the implicit contract in the form of repeated short-term debt contract is a device for the financial institution to ensure its bargaining power after the loan has been made. Pioneering theoretical analyses of the debt maturity structure from such a view point include Flannery (1986) and Diamond (1991, 1993) who characterized the debt maturity length as the signaling device of the firms about their hidden characteristics.
However, applying the signaling model of Flannery=Diamond to the explanation of the increased corporate debt maturity in recent Japan appears to be difficult. In their model, two different qualities of the borrower firm groups are present. In order for the high quality group to distinguish themselves from bad quality group towards the investors/lenders, they choose to issue short-term debt because it is relatively more expensive for bad quality group to do so. So the shorter debt maturity serves as the signal of firms’ quality/profitability in the separating equilibrium. Therefore, this type of theoretical models suggests that better firms will issue more debts in shorter maturities. If we adopt to such a theoretical argument to explain the increased debt maturity of the Japanese firms, the story will go like as follows: as more and more firms became exposed to the threat of denied refinancing during the lingering recession in the late 1990’s, they gave up short-term debt financing and choose more stable financing options and this explains the increase of average debt maturity in 1990s and 2000s.
However, the increase of debt maturity has continued as a long-term trend since mid-1970s.
Unless we assume business environment surrounding the Japanese firms that have continued to deteriorate monotonically and persistently over 30 years, the above argument cannot explain such a long-term trend. In fact, the Japanese macroeconomic performance in mid-2000 has been relatively improved compared to in the end of the 1990’s. However, but the increasing trend has continued.
In addition, theoretically speaking, the cost incurred through the action (=signal) shall be sufficiently lower for the high-quality group than for the low-quality group. In that case, high-quality group may take action and distinguish themselves by sending a signal even by paying extra cost. However, if there were multiple choices or selections of actions that can serve as signals, what will be chosen and will behave as a signaling role cannot be determined by the theoretical argument alone 2. Practically, the choice of long-term and short-term As an example of signaling theory, the reason why the Azure-winged magpie has long tail is because even if they have a longer tail than other males, they would have enough strength to escape from the predators and so there are more possibilities to get attention from the female (Dawkins ). However, there are no reasons for the male to have a long tail in order to show its excellence, and such a behavior is not seen in the majority of animals other than azure-winged magpies.
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.8, No5, November 2012 573 borrowings is nothing more than one of candidates for corporate behaviors that serve as a signal.
At the same time, in the models of Flannery = Diamond, it is assumed that the amount of debt financing needs to be the same regardless of the quality of borrower firms. However, given that there has been a long-term downward trend in the debt/equity ratio, the action to cut back its debt more aggressively during the recession could have certainly work as the signal of the firm’s better health. Considering the circumstances surrounding the Japanese companies over the past quarter of a century, there were at least two candidates of firms’ actions that can serves signals, the selection of debt maturity length and (the reduction of) total debt amount.
(2) Empirical analysis on the relationship between the firm’s performance and its debt growth Given the theoretical argument in the previous subsection, I first examine the relationship between the performance of the firm and the growth rate of its debts/borrowings.
The data of thirty-four industries excluding agriculture, forestry and fishery industries from the FSSCI industry classifications is used in the following empirical analyses. The sample period is 1986-2008. For the start, I estimate the following equation for the growth of debt outstanding.
△debt = α + β∙ROA + β∙△invest + ε (1) where △debt is the growth rate of the debt amount (total borrowings, long-term borrowings, short-term borrowings) of industry i, △invest is the growth rate of the capital investment, ROA is the Return on Asset. It is difficult to consider this equation (1) as a causal relationship in economic model, so that the estimation result should be interpreted as descriptive rather than explanatory.
According to the discussion on the signaling model in the previous subsection, since ROA and other information of the certain fiscal year became fully revealed to the financial institutions or investors only at the beginning of next fiscal year, the growth rate of the debt in the same fiscal year can be the signal of the firm’s performance toward its lender. In that case, if it is only about the relationships between debt growth and the firm’s performance, ROA can be regressed on the growth rate of the debt too. However, even when the performances of two firms were identical, some exogenous factor might reduce/induce the equip investment of a particular firm so that the demand for the new fund of one firm can be different from another’s.
In that sense, it would be better to run the regression for the debt growth rather than for ROA.
T Iwaisako / Public Policy Review
Table 2 shows the estimation results for equation (1). The first column is the regression for the growth rate of total liabilities. There are some extreme observations in the service industries in 2000s for which the growth rates of debts exceed more than 100% per year and these outliers significantly affect empirical results. Hence, I exclude the observation with more than 200% of debt growth. This means that the observation of telecommunication industry in 2004 is take out from the sample. Unfortunately, explanatory power of the regression result without the outlier is not so high with R2 of 6%. While the capital investment is significantly positive, there is no influence of ROA on the growth rate of debt.
The regression results with the long-term borrowing as the dependent variable are reported in the 2nd and 3rd column in Table 2 and the short-term borrowing is used in the 4th and 5th column. Total debt growth in the same period has been added as an explanatory variable in 3rd and 5th column.
First, looking at the estimation result of the long-term debt in the 2nd column, and short-term debt in the 4th column, the coefficients of the capital investment are positive and significant in both cases. On the other hand, ROA is negatively significant in the regression for the long-term debt at 10% level, but is positive and insignificant in the regression for the short-term debt.
Next, in the 3rd and 5th column in which contemporaneous growth rate of the total debt is added as an explanatory variable, the explanatory power of the equation rise dramatically, from 5% to 6% for the long-term borrowing and 3% to 6% for the short-term borrowing. While the explanatory power of the capital investment for both long-term debt and short-term debt regressions have vanished, this is not surprising given the strong significance of the capital investment in explaining total debt in the 1st column.
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.8, No5, November 2012 575
Table 3: Corporate Performance and Firms’ Borrowings:
Manufacturing Industries vs. Non-manufacturing Industries
ROA’s influence on debt growth is also much clearer in the 3rd and 5th column. The coefficient of ROA is negatively significant in the long-term borrowing regression and positively significant in the short-term borrowing regression. These results are consistent with the implication of the Flannery = Diamond model that high quality firms are more likely to rely on short-term debt than low quality firms.
It should be noted, however, when we look at on the constant terms, the estimated coefficient for the long-term debt regression is significantly larger than the coefficient in the short-term debt regressions regardless total debt is included as an explanatory variable or not.
In other words, for the sample period considered in this paper, Japanese firms have been increasing their long-term borrowings relative to short-term borrowings for some reasons unrelated to their business performances measured by ROA.