Jump to ContentJump to Main Navigation
Japan's Bubble, Deflation, and Long-term Stagnation$

Koichi Hamada, Anil K Kashyap, and David E. Weinstein

Print publication date: 2010

Print ISBN-13: 9780262014892

Published to MIT Press Scholarship Online: August 2013

DOI: 10.7551/mitpress/9780262014892.001.0001

Show Summary Details
Page of

PRINTED FROM MIT PRESS SCHOLARSHIP ONLINE (www.mitpress.universitypressscholarship.com). (c) Copyright The MIT Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in MITSO for personal use (for details see http://www.mitpress.universitypressscholarship.com/page/privacy-policy). Subscriber: null; date: 21 June 2018

Corporate Restructuring in Japan during the Lost Decade

Corporate Restructuring in Japan during the Lost Decade

Chapter:
(p.343) 10 Corporate Restructuring in Japan during the Lost Decade
Source:
Japan's Bubble, Deflation, and Long-term Stagnation
Author(s):

Takeo Hoshi

Satoshi Koibuchi

Ulrike Schaede

Publisher:
The MIT Press
DOI:10.7551/mitpress/9780262014892.003.0010

Abstract and Keywords

This chapter examines corporate restructuring in Japan between 1981 and 2007 in relation to financial deregulation and the banking crisis. Drawing on data on the balance sheets and income statements of publicly traded companies that were restructured, it analyzes different types of restructuring, the characteristics of the restructured companies, and how they changed over time, especially during the lost decade. Historically, larger firms, distressed firms, and firms that are highly dependent on bank loans were more likely to be restructured. In addition, firms that undergo restructuring experience a reduction in employment growth, capital, and total debt, but not bank loans. During the lost decade, however, these patterns were not observed. The likelihood of distressed firms undergoing restructuring has decreased over time, high main bank dependence had no impact on the likelihood of restructuring, and restructured firms did not reduce investments as much as they did in the 1980s.

Keywords:   restructuring, Japan, deregulation, banking crisis, publicly traded companies, lost decade, distressed firms, bank loans, debt, investments

This chapter looks at the effects of financial deregulation, the banking crisis, and corporate reforms on the incidence and processes of corporate restructuring in Japan.

During Japan’s period of rapid growth from the 1950s into the 1970s, an institutionalized mechanism of corporate restructuring developed, led by the distressed company’s main bank. In the typical rescue event, the main bank, being a large shareholder and usually the largest lender, intervened by dispatching executives and restructuring debt (often with the help of other lenders) so that the company could recover and resume debt repayment. Most bail-outs were informal; courts were rarely involved. By the time Japan entered the 1980s, such rescue intervention had become quasi-automated, and the few exceptional cases where a main bank refused to bail out a large firm were well publicized and only proved the standard rule.

Given its large exposure in both debt and equity at the time, the main bank had an obvious economic interest in the recovery of the customer. Moreover, it also strove to maintain its reputation as a dependable monitor on behalf of other lenders. Sheard (1989) argues that this type of reciprocal delegated monitoring reduced costs by eliminating duplication of monitoring effort. Under delegated monitoring, the main bank was responsible for rescuing a client in financial trouble. Politics may occasionally have played a role, as the government had set strong priorities toward supporting Japan’s largest companies, in order to uphold employment. (There is a large literature on Japan’s main bank system in general, and main bank interventions in particular. See, for example, Aoki and Patrick 1994; Hoshi, Kashyap, and Scharfstein 1990; Hoshi and Kashyap 2001; Hirota and Miyajima 2001; Kaplan and Minton 1994; Kang and Shivdasani 1995; Morck and Nakamura 1999; and Sheard 1989, 1994.)

(p.344) In chapter 9, Peek reviews the literature on main bank rescues and provides evidence that increases in main bank lending were beneficial to distressed firms in the 1980s in terms of performance, measured either as earnings before interest and taxes (EBIT) divided by total assets or by operating income divided by total assets. These beneficial aspects cannot be found for the 1990s, which leads Peek to conclude that the effectiveness of increased main bank lending has declined. In this chapter, we examine a broader set of episodes of restructuring, where main bank rescues are just one subset of the data. Thus, we address the larger question of the overall incidence of restructuring. During the 1990s, did the nature of restructuring change, and did restructuring lose or gain effectiveness?

Japan’s political economy underwent great change during the 1990s. The combination of financial deregulation that began in the late 1970s, the subsequent disintermediation between large firms and their banks, the banking crisis and large-scale bank mergers, and vast-ranging reforms of both the banking and corporate sector beginning in 1998 may have altered the economic calculations of both banks and companies. As bank-company relations have become less tight, and as banks fought to survive a severe non-performing loan crisis, are banks still inclined to lead restructuring efforts for their clients? Has the introduction of new bankruptcy laws and procedures altered the choices made by managers or creditors of troubled companies? How have new laws and rules on corporate reorganization—such as easier processes of spin-offs or lay-offs—changed the content of restructuring plans? And have new accounting rules and increased transparency on corporate financials invited new players in addition to the main bank?

Based on a unique database of major corporate restructuring cases between 1981 and 2007, combined with financial data for large Japanese firms, we study the incidence and content of restructuring. We identify changes over time in what triggered the onset of corporate restructuring, who led corporate restructuring, and how restructuring affected employment, capital growth, bank loan growth, and total debt growth of the distressed company.

We find that the likelihood that a large distressed firm undergoes restructuring has decreased over time, especially during the 1990s. Moreover, firms that announce a “restructuring” effort show greater downward adjustments in the growth rates of employment, capital, and debt than other distressed firms, meaning that these announcements are not just a publicity effort but have true economic implications. (p.345) Moreover, our data present evidence that high reliance on main bank lending is no longer a reliable predictor that a troubled firm will undergo restructuring. In other words, main banks no longer push distressed clients into restructuring more than other creditors. That said, we also find that financial restructuring measures are more pronounced when a bank leads the effort. Thus, the main bank is no longer the automatic last resort for a troubled firm, but when it steps in, it is more aggressive in relieving the debt burden of the company than other leaders of restructuring.

This chapter is organized as follows. Section 1 provides details on financial deregulation and corporate reforms to explain how the environment and motivations for undergoing corporate restructuring in Japan have changed. Section 2 introduces the database and provides an overview of the data. Section 3 contains the data analysis and findings.

1 Changes in Corporate Restructuring

Financial deregulation in Japan began in the late 1970s and occurred in a piecemeal and skewed process that lifted restrictions on large firms’ financing first (Hoshi and Kashyap 1999, 2001). As deregulation of financing options for large companies proceeded more quickly than deregulation aimed at banks or savers, banks started to lose corporate customers while savers continued to hold large portions of their assets in bank deposits. The banks searched for new clients among small and medium companies, especially those in real estate. But whereas those new clients had looked promising during the land price boom of the 1980s, loans turned sour as land prices dropped in the 1990s. Non-performing loans were a major problem for Japanese banks throughout the decade, and the non-performing loan problem escalated into a fullblown crisis in the late 1990s, and continued into the early 2000s.

The banking crisis led to great changes in Japan’s banking industry. Some banks went bankrupt, others merged. The previous 13 large city banks reorganized into four mega-financial groups, all straddling previous business-group boundaries (such as Sumitomo-Mitsui). This muddled the previously clearly defined bank-company relations, and left the new banks with competing clients. For example, Mizuho Financial Group, formed in 2002, found itself with six large construction companies as clients, all in dire straits at the time. The long-term credit banks were nationalized and sold to foreign financial interests to be turned into commercial banks (Shinsei, Aozora).

(p.346) For our study of corporate restructuring, the changes brought about by financial deregulation and the banking crisis raise the question of whether the new business environment and clientele of Japanese banks may have altered bank willingness and effectiveness in bailing out troubled firms.

The Big Bang reforms of 1998 marked a final stage of two decades of gradual financial deregulation (Hoshi and Kashyap 2001). One important aspect of these reforms is the introduction of new accounting and disclosure rules, including consolidated balance sheets and mark-to-market of shareholdings. This had two main ramifications. First, corporate cross-shareholdings—many of which suffered losses with falling stock prices during the 1990s—were suddenly expensive to carry, which triggered a sell-off of some of the holdings. Whereas in 1987, large banks owned 14.9% of shares (in market value) listed on the Tokyo Stock Exchange, and corporations 30.1%, by 2001 these had dropped to 10.1% and 21.8%, respectively. A 2001 law limited bank shareholdings to its Tier 1 (own) capital, and by 2008 large banks owned only 4.9% TSE market value. New owners emerged in foreigners and institutional investors who, by 2007, combined to hold almost 50% of TSE market value (TSE 2007; Schaede 2008).

The new transparency made it easier to assess Japanese investments, which brought in new investors such as equity funds. This raises the question whether active investors have emerged as new players that guide troubled firms into restructuring, and if so, how their methods of restructuring differ from those of main banks.

The period 1998–2006 also brought great changes in corporate strategymaking, so much so that Schaede (2008) has labeled it a “strategic inflection point.” It coincides with political leadership of Prime Ministers Ryūtarō Hashimoto and Junichirō Koizumi, who shifted Japan toward “leave it to the market.” Rather than react to the severe banking crisis of 1998 with increased regulation, Koizumi and his Minister of Financial Affairs, Heizo Takenaka, further pushed deregulation between 2001 and 2006. Their 2002 Financial Revival Program put great emphasis on non-performing loan clean-up and was associated with a political push toward direct loan write-offs and drastic restructuring of troubled debtors, even those previously considered “too big to fail.” As the government allowed several large financial institutions to fail and raised pressure on banks to reduce their non-performing loans, it became clear that it was no longer blindly supporting bail-outs of large firms. By withdrawing the previous mandate to save employment by (p.347) helping troubled firms, the government freed banks from earlier obligations to bail-out large clients.

Partly in response to this shift, large companies called for more options in corporate reorganization of their own. To provide more flexibility, the Commercial Code was revised every year beginning in 1997, and was eventually replaced by a new Corporation Law in 2006. The gist of these revisions was to facilitate reorganization through spin-offs, spin-outs, different types of stock, stock swaps, mergers, and acquisitions (Yanagawa 2006, Schaede 2008). For example, one important change was to revoke the veto right of employees in a spin-out. Previously, corporate unions had rejected such moves on the grounds that smaller firms traditionally paid lower salaries. The revocation of this rule greatly reduced the voice of labor in restructuring decisions.

One big impact of these reforms was an increase in hostile takeover attempts, which became a new cause of concern for executives at underperforming firms. A wave of proactive restructuring set in, whereby large firms refocused on core businesses, spun off non-core operations, and restructured their finances. Importantly for our study, these legal reforms also afforded leaders of forced restructuring—the main bank, a lead company, or a consortium of both—more flexibility in how to restructure a distressed company, such as through labor adjustment. A first easing of lay-off rules came through court rulings in 2000, followed by a revision of the Labor Standard Law, effective in 2004.

Another major change was the complete revision of bankruptcy laws. In their old versions, these had rarely been employed because they were cumbersome, expensive, and often perceived as less efficient than an informal workout with the main bank. However, in 2000 the new Civil Rehabilitation Law (Minji-saisei-hō) replaced the old, clumsy Composition Law (Wagi-hō) to design new court-based bankruptcy procedures. Another legal-based procedure was “Corporate Reordering” (kaisha-seiri) based on the Commercial Code. This was also cumbersome, and it was discontinued with the Corporation Law of 2006. (See Schaede 2008.)

In 2003 the Corporate Reorganization Law (Kaisha-kōsei-hō) was revised to allow for Chapter 11-type turnarounds. In 2004, the Liquidation Law (Hasan-hō) was revised to simplify legal procedures for a shutdown and distribution of assets. Moreover, the 2001 “Guideline for Out-of-Court Workouts” added to this a new structure for bank-led workouts by stipulating how debt forgiveness should be organized in cases with multiple lenders but uncertain claims. While Koibuchi (2008) shows (p.348) that this Guideline has so far rarely been used, it represents yet another alternative for companies in distress.

For our study, the important question is whether these new options of turnaround, bankruptcy, and exit have influenced the choices taken by troubled firms, between undergoing informal restructuring led by a main bank or the formal, court-based processes of bankruptcy.

2 Database and Overview Statistics

We have constructed a database of major episodes of corporate restructuring for the odd years from 1981 to 2007, a set of 14 years of observation, from newspaper articles. The Japanese for restructuring is saiken Corporate Restructuring in Japan during the Lost Decade. We searched for this word in Nikkei Telecom 21, an electronic database that includes Japan’s leading economic and financial newspapers published by Nihon Keizai Shinbun-sha (Nihon Keizai Shinbun, morning and evening editions and including the economic sections of regional editions, Nihon Kin’yū Shinbun, Nihon Sangyō Shinbun, and Nihon Ryūtsū Shinbun). Because restructuring typically lasts longer than one year, we are confident the search picked up most episodes over the 27-year period. We assume the newspapers have been consistent in their reporting and choice on saiken over time. (Japanese is much more consistent in word use than English. Where an English-language article may refer to “firms,” “companies,” and “corporations” interchangeably and all with the same meaning—presumably to avoid tediousness from repetition—a Japanese-language article would use the same word, even if repeatedly. Word choice consistency is helpful for our purpose.)

Because the newspapers may fail to report all restructuring cases, the number of episodes in our database is the lower bound of the true frequency. However, we are unaware of any major cases not included in our database, and thus are confident that it includes all major cases of restructuring.

We selected for our sample only publicly listed companies, because this allows us to combine our newspaper database with accounting and financial data. Before beginning our analysis, we conducted one more robustness check to account for possible selection bias. To check for this independently from the Nikkei newspapers, we picked a random period (1997–98) and identified all listed companies that had a new bank director on their board in 1998 as compared to 1997, based on Tōyō Keizai’s Kigyō Keiretsu Sōran. Empirical research has suggested that the dispatch of a board member is often the first sign of a bank intervention. (p.349) We compared this list to our database, and found six cases where a company had new directors and had incurred losses, but had not announced a “restructuring.” The six companies are smaller than the average companies identified as undergoing restructuring. They are also more profitable, and none can be considered “distressed” according to our measures as described below. Our database for 1997 included 139 cases of restructuring, suggesting an omission rate well below 5%.

In the next step, we coded the information contained in each newspaper article to build a database that contains important characteristics of each restructuring episode. This included: timing; the identity of the restructuring leader (such as the main bank or an affiliated company); any reliance on bankruptcy laws; financial restructuring (such as debt forgiveness, interest concession); restructuring on the asset side of the balance sheet (asset sales or stock sales); managerial change (dispatch of directors, turnover of incumbent managers); corporate reorganization (spin-offs or exit from a business); labor adjustment (lay-offs, early retirement, hiring freeze or employee dispatches); and salary adjustments (wage or bonus cuts, executive compensation reduction).

Table 10.1 categorizes the informational items we collected from the newspaper articles into five groups: intervention, capital changes, business plan changes, corporate finance reforms, and labor adjustments; the last three rows summarize three main groups of measures: corporate organization, employment reduction, and salary reduction. The 27-year period is divided into three sub-periods: (1) pre-bubble and bubble (1981–91); (2) the lost decade (1993–2003); and (3) strategic inflection (2005–07). Thus, the first two periods each cover six observations, the last, two. The main emphasis is on comparing the first two periods, that is the pre-bubble and bubble years, with the lost decade.

For the entire period, we have 1,756 episodes of corporate restructuring. The number increased from 572 during the pre-bubble and bubble years to 929 during the lost decade. In terms of content, beginning with the last three measures, we see an increase in the portion of companies under restructuring that reorganize. We attribute this to legal reforms that have facilitated reorganizat ion through sell-offs and exits from certain businesses beginning in 1998. In contrast, employment and salary adjustments do not fluctuate much across the three periods, although employment reduction was adopted by fewer firms under restructuring in the most-recent period. Two explanations stand out for this result: reducing the work force had become progressively easier since 2000, so (p.350)

Table 10.1 Summary Statistics for 1,756 Restructuring Cases, 1981–2007

Period

1981–1991

1993–2003

2005–2007

Number of restructuring cases

572

929

255

Intervention

37.94%

37.03%

41.96%

  Director dispatch

23.08%

17.98%

20.39%

  Managerial turnover

22.20%

26.26%

31.76%

  Change of restructuring plan

2.62%

3.34%

3.92%

  Executive bonus cut

4.20%

6.24%

4.31%

Capital Changes

25.87%

22.17%

23.53%

  Asset sale

23.95%

17.87%

20.00%

  Sold shares held in other companies

6.64%

8.29%

8.63%

  Business Plan Changes

55.77%

57.05%

60.00%

  Size reduction

12.24%

12.81%

21.18%

  Cost reduction

19.76%

18.30%

25.10%

  Sales promotion

19.76%

12.49%

30.98%

  Exit from a business line

9.44%

15.72%

23.92%

  New entry

9.97%

3.66%

8.63%

  Spin off

17.13%

10.23%

14.12%

  Liquidation of affiliated companies

8.92%

16.58%

14.12%

Corporate Finance Changes

16.61%

19.16%

30.20%

  New loans

3.32%

3.44%

5.49%

  Interest reduction

5.59%

3.23%

0.39%

  Debt forgiveness

2.27%

5.71%

5.88%

  Debt-equity swap

0.17%

1.18%

3.53%

  New equity issue

8.39%

10.76%

23.92%

  Equity reduction

2.80%

3.55%

6.27%

Labor Adjustment

26.22%

28.63%

21.57%

  Relocation of labor (tenseki)

8.04%

5.38%

5.88%

  Furlough

0.70%

0.32%

0.00%

  Stop new hires

5.94%

7.75%

1.96%

  Early retirement

10.14%

16.25%

13.33%

  Lay-offs

4.72%

3.66%

2.35%

  Dispatching workers (shukko¯)

7.34%

6.14%

2.75%

  Wage reduction

2.62%

3.01%

3.92%

  Bonus cut

1.05%

1.40%

1.18%

Reorganization (exits, spin-offs, and size reduction)

32.69%

42.63%

41.18%

Employment adjustment (relocation, furlough, hire stop, retirement, and lay-offs)

25.17%

27.56%

20.78%

Salary adjustment (wage or salary cuts and executive bonus reduction)

5.77%

7.97%

6.67%

(p.351) that by 2005 many companies had already restructured labor. Moreover, Japanese companies were expecting a labor shortage beginning around 2010, which may have made them cautious in laying off workers.

Looking at the more-detailed level reported in the upper half of table 10.1, management turnover increased. While changes in capital structure were similar over the three periods, the proportion of firms under restructuring that sold assets declined, whereas those that sold shares held in other companies rose during the lost decade. For all three periods, the majority of restructured firms changed their business plan in one way or another. Scaling down the business has become an increasingly important tool, as have cost reductions and sales promotions, especially in the 2000s. An interesting finding is the substantial decline in the proportion of restructured companies that “entered a new business” during the lost decade: from almost 1 in 10 during the 1980s, the share fell to less than 1 in 25. Such strategic repositioning occurred much less frequently during the lost decade, perhaps because the continued stagnation of the 1990s reduced optimism regarding expansion. A similar activity drop is observed for sales promotion during the lost decade.

Under corporate finance, we observe changes tied to the reforms of corporate and financial laws. Interest rate concession became a less-used tool as interest rates dropped during the 1990s. In contrast, debt forgiveness became more popular. Debt-equity swaps were conducted more than previously. The share of restructuring companies issuing new equity, in contrast, climbed substantially, perhaps reflecting the recovery of the stock market between 2004 and 2007.

The incidence of labor adjustments remained stable during the lost decade but declined in the 2000s. Dispatching (shukkō) and relocation (tenseki) of workers to affiliated companies have often been mentioned in the literature as important adjustment valves for large companies. However, even in the 1980s only about 7% to 8% of restructuring firms announced these measures, and the rates fell in the mid 2000s. One reason may be that by the time a company undergoes restructuring, it has already dispatched and relocated employees in large numbers. Indeed, it used to be the rule that companies could not lay-off lifetime employees until they had relocated or dispatched workers and initiated early retirement programs. Early retirement continues to be the most-frequent means of labor adjustments.

Table 10.2 offers an overview of the data by type of restructuring leader. Most cases were informal; court-supervised cases were only (p.352) (p.353)

Table 10.2 Overview Data: Differences in Restructuring, by Restructuring Leader

Self-direct or not available

Main bank

Industrial firm

Group of banks

Group of firms

Banks and firms

Investment funds

Total

Total

1,058

147

321

50

42

108

30

1,756

Bankruptcy Laws

  Liquidation

0

1

1

1

0

2

0

5

  Corp. reorganization (rev. 2003)

*26

1

24

4

4

1

0

60

  Commercial code (until 2006)

0

1

2

0

0

0

0

3

  Composition (until 2000)

1

0

0

0

0

0

0

1

  Civil rehabilitation (after 2000)

*12

2

10

2

1

4

3

34

  Total court supervised

39

5

37

7

5

7

3

103

Number of Executives Dispatched

  0

1,045

74

169

33

22

41

21

1,405

  1

8

52

97

7

12

25

5

206

  2

4

11

30

6

2

15

3

71

  3

0

4

6

1

3

5

0

19

  4

0

1

6

2

1

10

0

20

  5–13

1

5

13

1

2

12

1

35

Replacement of Executives

  No replacement

851

95

209

34

27

69

19

1,304

  Yes

207

52

112

16

15

39

11

452

Revisions of Reorganization Plan

  None

1,033

136

317

43

41

101

29

1,700

  Once

24

11

3

6

1

6

1

52

  Twice

0

0

1

1

0

0

0

2

  Three times

1

0

0

0

0

1

0

2

Involved Exit, Spinoff or Liquidation of Affiliates

  No

643

75

240

20

19

51

20

1,068

  Yes

415

72

81

30

23

57

10

688

Involved Labor Adjustments

  No

764

109

264

33

32

71

30

1,303

  Yes

294

38

57

17

10

37

0

453

Involved Salary Adjustments

  No

989

131

311

38

40

94

29

1,632

  Yes

69

16

10

12

2

14

1

124

(*) The sponsors for these cases could not be identified from the newspaper articles.

(p.354) 5.8% of the total. When opting for the Corporate Reorganization or Civil Rehabilitation Law (equivalent to Chapter 11) a distressed company needs to identify a sponsor. Bank-led restructurings were almost always conducted out-of-court. All leaders appear roughly equally likely to dispatch directors and replace the management of the distressed firm. Episodes led by a group of banks or companies tend to have a higher incidence of corporate reorganization, labor adjustments, and salary adjustments.

For our analysis of these differences in process and leadership of corporate restructuring, we also look at the financial conditions of the

Table 10.3 Summary Statistics for Financial Data for All Listed Firms, 1981–2007 (averages for each period; in million yen)

Period

1981–1991

1993–2003

2005–2007

% of firms under restructuring

4.14%

5.28%

4.31%

% of firms with negative net profits for 2 years in a row

1.86%

6.33%

5.08%

% of firms with interest coverage ratio 〈1 for 2 years in a row

12.32%

17.39%

12.69%

% of firms with negative operating income for 2 years in a row

1.67%

4.43%

3.52%

Total assets

130,496

166,887

169,573

Total bank loans

39,364

39,812

32,101

Total debt

52,441

63,110

51,297

Bank debt to total assets ratio

25.61%

24.34%

18.29%

Total debt to total assets ratio

29.44%

29.79%

21.61%

Operating income / total assets

6.85%

4.74%

9.09%

Operating income

6,258

5,560

7,140

Net profit / total assets

2.95%

1.05%

4.64%

Net profit

2,376

147

3,562

Interest coverage ratio

0.65

4.67

3.38

Main bank dependence (proportion of loans from the largest lender)

26.46%

28.46%

NA

Amount of loan from the largest lender

3,514

5,492

NA

Employment growth

0.85%

−1.34%

1.18%

Growth of depreciable assets

8.16%

1.28%

1.58%

Bank borrowing growth

3.97%

0.41%

−1.37%

Total debt growth

5.57%

−0.15%

−1.72%

The last four rows are calculated for the sub-sample that we use for regression analyses in tables 10.6 through 10.9, and excludes outliers of those variables.

(p.355) companies undergoing restructuring. We source this information from the Nikkei Financial Database, which contains accounting data for all listed firms (including firms that used to be listed). Fewer listed firms underwent restructuring during the 1980s than during the lost decade. Summary statistics of financial variables are presented in table 10.3.

3 Data Analysis and Findings

For our analysis, we have three sample sets. The first is all companies listed at some point between 1980 and 2007. Of these, we identify two subgroups: companies that are in distress due to poor financial results (for example, as expressed in operating income), and those we identified through our newspaper search as having undergone restructuring. We begin our analysis with a few graphs that differentiate distressed companies that undergo restructuring from distressed firms that do not. We consider three alternative measures for “distress”: (1) negative operating income during the previous two years; (2) negative net profits for the previous two years; and (3) an interest coverage ratio below one for the previous two years (this is a cash flow measure calculated by dividing a company’s earnings before interest and taxes by its interest expenses).

Figure 10.1 shows the ratio of firms under restructuring to number of distressed firms (defined as those with negative operating income during the previous two years). There appears to be a downward trend in distressed firms restructuring. We find a similar downward trend when we use negative net profits to define “distress.” In contrast, when we use the interest coverage ratio to identify firms in “distress,” no trend is obvious, as shown in figure 10.2.

Figure 10.3 shows differences across firms under distress that undergo restructuring from distressed firms that do not, in terms of employment adjustment (using negative operating income as the distress marker). It appears that distressed firms under restructuring reduce their labor force more than those not undergoing restructuring. Figure 10.4 shows that distressed firms under restructuring reduce their investments more than other distressed firms.

The figures suggest that undergoing “restructuring” results in real adjustments. They also provide evidence that the likelihood of a distressed firm restructuring has changed over time, as may have the measures adopted during restructuring. These figures do not consider factors other than financial distress, such as industrial composition or (p.356)

Corporate Restructuring in Japan during the Lost Decade

Figure 10.1 Restructuring and Distress: Ratio of Firms Under Restructuring to Number of Firms with Negative Operating Income During the Previous Two Years.

Corporate Restructuring in Japan during the Lost Decade

Figure 10.2 Restructuring and Distress: Ratio of Firms Under Restructuring to Number of Firms with Interest Coverage Ratio Ã1 for the Previous Two Years.

(p.357)
Corporate Restructuring in Japan during the Lost Decade

Figure 10.3 Employment Adjustment by Firms Under Distress, Comparing Firms Under Restructuring (Solid Line) with Firms Not Under Restructuring (Dotted Line).

Corporate Restructuring in Japan during the Lost Decade

Figure 10.4 Growth of Depreciable Assets by Firms Under Distress, Comparing Firms Under Restructuring (Solid Line) with Firms Not Under Restructuring (Dotted Line).

(p.358) size-distribution of restructuring companies. Thus, we next conduct regression analysis to control for some obvious factors that may influence the probability of undergoing restructuring or restructuring content.

3.1 Changes in the Likelihood of Undergoing Restructuring

To understand the apparent decline in the incidence of restructuring by distressed firms, we estimate several probit regressions. The dependent variable is “saiken,” which takes the value 1 if a firm is identified as restructuring in that year, and 0 if not. We examine whether the likelihood of undergoing restructuring is determined by whether the company is in financial distress (determined in three different ways: operating income, interest coverage, and net profits), and by the company’s dependence on bank borrowing (measured by the ratio of total bank debt to total assets). Moreover, we look at how relations have changed over time.

We report estimation results that use negative operating income as the distress measure in table 10.4. The first column contains the minimal specification that includes: the ratio of bank debt to total assets; (the natural logarithm of) total assets (both measured at the beginning of the period); and “distress,” which takes the value 1 if the firm has negative operating income over the previous two years. The coefficient estimates on these variables are all positive and statistically significant. Thus, not surprisingly, companies in distress are more likely to be restructured. Companies that depend more on bank debt are also more likely to be restructured, as are larger companies. We repeated these estimations using our alternative definitions of distress (interest coverage ratio, net profits). The results are qualitatively similar, and are not detailed here.

In the specification in the second column, we add a two-year lag of the saiken variable as an explanatory variable, to see if a company under restructuring is likely to stay under restructuring for more than two years. We can also think of this as an attempt to control for the existence of chronically depressed and restructured companies. Because the lag is not available for the first year of our sample (1981), the number of observation drops. The coefficient estimate on the lagged saiken is positive and statistically significant, suggesting that a company under restructuring is indeed more likely to continue being restructured, even after two years. The coefficient estimates on the other variables are now a bit smaller, but still positive and statistically significant. (p.359)

Table 10.4 Regression Results: Determinants of Restructuring

Variables

(1)

(2)

(3)

(4)

Bank debt / total assets

1.197

1.025

1.381

1.484

(0.055)

(0.059)

(0.070)

(0.093)

DISTRESS

0.991

0.787

0.764

0.693

(0.042)

(0.045)

(0.046)

(0.057)

Log(total assets)

0.196

0.168

0.206

0.201

(0.007)

(0.008)

(0.009)

(0.013)

saiken 2 years ago

1.074

0.968

0.877

(0.039)

(0.040)

(0.049)

Main bank dependence

0.179

(0.113)

Year dummies

No

No

No

No

Industry dummies

No

No

Yes

Yes

Number of observations

36,488

33,640

33,308

17,868

The dependent variable is saiken, which takes the value 1 if the firm was under restructuring during the year, and 0 otherwise. DISTRESS takes 1 if the firm experienced negative operating income in the previous two years. Each column reports the coefficient estimates and standard errors (in parentheses) for a probit model. The sample period is every odd year from 1981 to 2007. The model also includes a constant term. The rows “Year dummies” and “Industry dummies” indicate inclusion of these dummies (yes or no). The coefficient estimates for the year dummies, industry dummies, and the constant term are not reported.

The main bank dependence variable in column 4 was originally calculated by one of the authors for a different project that covered fewer industries and ended in 2002.

To interpret the economic importance of these findings, we calculated the marginal change in the probability of restructuring when each left-side variable changes by an infinitesimal amount (or from 0 to 1 for a dummy variable). (The calculation is done using the “dprobit” command of the Stata 10.0.) For the specification in column 2, a distressed firm is 10.0% more likely to be under restructuring than a firm not in distress. A firm undergoing restructuring two years earlier is 16.8% more likely to be still under restructuring. When the bank debt to asset ratio increases by 0.20 (the standard deviation for the sample, with a mean for the period 1981–2007 of 0.24), the probability of restructuring increases by 1.4%. Finally, for a firm that is larger by one standard deviation, the likelihood of restructuring is larger by 1.9%.

The third specification controls for industry effects by adding industry dummies. All the coefficient estimates we report remain positive (p.360) and statistically significant. While there is variation across industries, no industries stand out as driving the results. Calculating industry differences by using the food processing industry as the reference point, we find that firms in shipbuilding and precision machinery face a probability of restructuring that is at least 1% higher. For the non-bank financial, ground transportation, warehouse, and electric utility industries this probability is lower by at least 2%. In general, non-manufacturing industries seem to be less-likely to be restructured. Moreover, when we estimate the model with just one industry dummy variable that distinguishes manufacturing from non-manufacturing, the coefficient on the non-manufacturing dummy is negative. The coefficient estimates imply that the probability of restructuring for non-manufacturing firms is lower by 1.2%.

In the last column, we add “main bank dependence” as an independent variable. This is measured as the amount of loans from the largest lender divided by total bank loans to the company, and the data sample extends only to 2002. We find main bank dependence has a positive association, but the finding is not statistically significant. The likelihood that a company will undergo restructuring increases with the company’s dependence on its main bank, but the evidence is statistically weak.

To identify possible change over time, we now look at the three sub-periods by adding period dummy variables and interacting them with some key explanatory variables. Table 10.5 reports the results.

The specification in the first column adds the interaction between the period dummy and distress (operating income). The estimated coefficients on the interaction terms suggest that the impact of distress on the likelihood of restructuring has faded over time. Compared with the 1983–91 period, during the 1993–2003 lost decade, the likelihood that a distressed company will undergo restructuring has declined. When calculating the marginal impact on the probability of restructuring, we find that a distressed firm was 18.0% more likely to be restructured during 1983–91 than a non-distressed firm. This declined to 16.0% in 1993–2003, but saw an uptick to 16.5% for 2005–07.

In the second column, we add an interaction term between the bank debt to total assets ratio and the period dummies. The impact of bank debt of a distressed company on the likelihood of restructuring declined during the 1990s. Even though firms that relied more on bank debt continued to be more likely to be restructured during the lost decade, the magnitude of the difference declined, suggesting a reduced (p.361)

Table 10.5 Determinants of Restructuring: Changes Over Time

Variables

(1)

(2)

(3)

Bank debt / total assets

1.382

1.555

1.543

(0.072)

(0.122)

(0.096)

Bank debt / total assets X

−0.293

(1993–2003 dummy)

(0.144)

Bank debt / total assets X

−0.041

(2005–2007 dummy)

(0.210)

DISTRESS

1.188

1.176

1.085

(0.086)

(0.087)

(0.102)

DISTRESS X (1993–2003 dummy)

−0.596

−0.572

−0.533

(0.103)

(0.104)

(0.120)

DISTRESS X (2005–2007 dummy)

−0.376

−0.355

(0.152)

(0.152)

Log(total assets)

0.207

0.207

0.202

(0.009)

(0.009)

(0.013)

saiken 2 years ago

0.993

0.994

0.896

(0.041)

(0.041)

(0.049)

Main bank dependence

0.765

(0.177)

Main bank dependence X

−0.852

(1993–2003 dummy)

(0.212)

Year dummies

Yes

Yes

Yes

Industry dummies

Yes

Yes

Yes

Number of observations

33,308

33,308

17,868

See notes to table 10.4.

impact of bank financing, as compared to other forms of financing, in triggering corporate restructuring.

Next, we examine changes in the importance of main bank dependence over time. Again, data constraints reduce the number of observations. Recall that in table 10.4 we estimated the impact of main bank dependence without considering possible changes over time, and found the coefficient estimate to be positive but statistically insignificant. When we allow the coefficient estimate to be different for the two sub-periods, we find a positive, statistically significant impact for 1983–91. Thus, in the 1980s, the higher the loans outstanding from the main bank as a percentage of total loans, the more likely a company was to undergo restructuring. Calculating the marginal impact of main bank dependence on the probability of restructuring using the point estimates of the (p.362) coefficients, we find that when the main bank dependence increases by 0.17 (the standard deviation of the sample; the mean is 0.28), the restructuring probability increased by 1.0% during 1983–91. This is consistent with research on this period that main banks engaged in corporate rescue operations in an almost automated fashion. However, during the lost decade, this was no longer so. The sum of the last two coefficient estimates is −0.087 for the lost decade. In terms of the marginal impact on the likelihood of restructuring, this implies that an increase in main bank dependence by one standard deviation reduces the probability by 0.1% (essentially zero).

Taken together, these results provide new insights into the changes in determinants for restructuring events over time. Clearly, distressed firms with more bank debt are more likely to be restructured, but over time this tendency has declined. Especially striking is the disappearance of any impact of main bank dependence on the likelihood of restructuring. During the 1980s, a company that depended highly on its main bank was likely to be restructured. There was no such association during the lost decade.

3.2 Differences in the Processes of Restructuring

The next question is whether restructuring content has changed over time. (To mitigate problems caused by a few extreme observations, in all following regressions that use growth variables as the dependent variable, we drop all observations where either the growth rate or the lagged growth rate is below −50% or above 100%.)

Employment adjustment, measured as growth in the number of employees over the previous year, is examined first. From figure 10.3, we already know that, on average, distressed companies undergoing restructuring reduce their labor force more than distressed companies not under restructuring. Regression analysis confirms this result, as shown in table 10.6.

Here we use the saiken variable as an explanatory variable for employment growth. The regression also includes a one-year lag of employment growth, to capture persistence at the firm level. We also include year dummies and industry dummies to control for any time-specific or industry-specific fixed effects. Finally, all regression specifications also include the distress dummy as an explanatory variable.

The basic specification in the first column shows that distressed companies in general have lower employment growth, which is not surprising. When we control for the general effect of distress on employment (p.363)

Table 10.6 Restructuring and Employment Growth

Variables

(1)

(2)

(3)

Lagged employment growth

0.303

0.303

0.303

(0.011)

(0.011)

(0.011)

DISTRESS

−0.032

−0.029

−0.032

(0.004)

(0.004)

(0.004)

saiken

−0.033

−0.031

−0.032

(0.003)

(0.003)

(0.004)

DISTRESS X saiken

−0.016

(0.011)

saiken x period 3 dummy

−0.0040

(1993–2003)

(0.0060)

saiken x period 4 dummy

0.0078

(2005–2007)

(0.0091)

Year dummies

Yes

Yes

Yes

Industry dummies

Yes

Yes

Yes

Number of observations

26,750

26,750

26,750

R-squared

.192

.192

.192

The dependent variable is the growth rate of number of employees for the firm.

Observations where either the growth rate or the lagged growth rate is below −50% or above 100% are excluded. DISTRESS takes 1 if the firm experienced negative operating income in the previous two years. Each column reports the result for a regression model. The models are estimated using ordinary least squares (OLS). Each cell shows the coefficient estimate and the White (1980) standard error. The sample period is every odd year from 1981 to 2007. The model also includes a constant term. The rows “Year dummies” and “Industry dummies” indicate inclusion of these dummies (yes or no). The coefficient estimates for the year dummies, industry dummies, and the constant term are not reported.

growth, the regression result shows that companies under restructuring reduce employment even more. The point estimate suggests that employment growth in firms undergoing restructuring is 3.5% lower.

The specification in the second column adds the interaction term between the distress variable and saiken to examine whether the impact of restructuring on employment adjustment is more pronounced for distressed firms or non-distressed firms. While the result is not statistically significant, the coefficient estimate on the interaction term suggests that impact on employment adjustment is stronger for firms in distress. The point estimate suggests that when a firm is not in distress, restructuring lowers employment growth by 3.1%, but when the firm is in distress, restructuring lowers employment growth (p.364) by 4.7% (on top of the 2.9% employment growth reductions due to distress).

In the third column we introduce interaction terms between saiken and the period dummies to see whether the employment impact of restructuring changed over time. The estimated coefficients on all interaction terms are tiny and not significantly different from zero. Thus, we conclude that restructuring consistently led to lower employment growth throughout the entire period.

Table 10.7 reports a similar regression analysis with capital growth (growth in depreciable assets, a proxy for investment rate) as the dependent variable. Similar to employment growth, distressed firms display reduced capital growth. Figure 10.4 suggested that distressed firms under restructuring reduce capital growth more than other distressed firms. Regression analysis confirms this finding and shows that firms under restructuring reduced investments by 4.6% more than other firms in the basic specification (column 1). In column 2, we add the interaction term to allow the impact of restructuring to differ between distressed firms and non-distressed firms. For non-distressed firms,

Table 10.7 Restructuring and Capital Growth

Variables

(1)

(2)

(3)

Lagged capital growth

0.138

0.138

0.138

(0.008)

(0.008)

(0.08)

DISTRESS

−0.051

−0.056

−0.051

(0.005)

(0.005)

(0.005)

saiken

−0.046

−0.045

−0.054

(0.005)

(0.005)

(0.008)

DISTRESS X saiken

0.031

(0.016)

saiken X (1993–2003 dummy)

0.025

(0.010)

saiken X (2005–2007 dummy)

−0.0046

(0.017)

Year dummies

Yes

Yes

Yes

Industry dummies

Yes

Yes

Yes

Number of observations

26750

26750

26750

R-squared

.100

.101

.101

The dependent variable is the growth rate of depreciable assets for the firm. See table 10.6 for other notes.

(p.365) undergoing restructuring meant reduced investment by 4.5% (nearly identical to the column 1 result). For firms in distress, however, undergoing restructuring meant a reduction of 1.4% compared to a firm in distress but not under restructuring.

The period-distress interaction term in table 10.7 is positive and statistically significant for the lost decade, suggesting that during the 1990s firms under restructuring did not reduce capital growth as much as they used to, or have done in the 2000s. Recall from table 10.3 that the average growth rate of capital for all firms during the first period was 8.2%. The point estimates of our regression suggest that during the 1980s a firm undergoing restructuring reduced capital growth to 2.8%, a reduction of 5.4 percentage points. During the lost decade, in contrast, the average capital growth rate was 1.3%, but firms under restructuring reduced capital by 1.6%. While this reduction by 2.9 percentage points is still substantial, it is less than in the 1980s, and it is also less than in the mid-2000s, when the reduction is 4.3% during restructuring compared to an average capital growth of 1.6%, a 5.9 percentage point difference.

Taken these findings together and combining them with insights drawn from table 10.5, it appears that, during the lost decade, fewer firms in distress underwent restructuring, and when they did, the magnitude of their investment adjustments was smaller. This is consistent with the notion that corporate restructuring during the lost decade was at times tentative.

Table 10.8 examines the growth of bank loans during restructuring. Research has shown that there is no stereotypical modus operandi in this regard: In some cases, the main bank may increase loans to help a troubled client, while in others it reduces the debt burden, with debt forgiveness being the extreme example. There is no empirical evidence that the average firm under main bank rescue increased its loans, as pointed out by Miwa (1985) and Hoshi, Kashyap, and Scharfstein (1990). However, our results suggest that the lost decade was different. Whereas we find that growth rates of bank loans to restructuring firms were not different from those to non-restructuring firms during the 1980s and the 2000s, in the 1993–2003 period companies under restructuring seem to have received additional loans. The point estimates suggest that an average increase in bank borrowings by 0.4% during 1993–2003 (as shown in table 10.3) compared to a 2.1% increase in bank loans by firms under restructuring. Ironically, Peek’s findings (in chapter 9) suggest that this is exactly the period when additional loans from (p.366)

Table 10.8 Restructuring and Bank Loan Growth

Variables

(1)

(2)

(3)

Lagged bank loan growth

0.149

0.149

0.148

(0.008)

(0.008)

(0.008)

DISTRESS

−0.0049

−0.0011

−0.0047

(0.0068)

(0.0072)

(0.0068)

saiken

−0.0009

0.0021

−0.017

(0.0064)

(0.0068)

(0.011)

DISTRESS X saiken

−0.023

(0.020)

saiken X (1993–2003 dummy)

0.034

(0.014)

saiken X (2005–2007 dummy)

−0.020

(0.020)

Year dummies

Yes

Yes

Yes

Industry dummies

Yes

Yes

Yes

Number of observations

26750

26750

26750

R-squared

.045

.045

.046

The dependent variable is the growth rate of total bank borrowings by the firm. See table 10.6 for other notes.

the main bank lost effectiveness, as there is no evidence that increased main bank loans during this period helped distressed companies improve either in terms of EBIT divided by total assets or operating income.

Table 10.9 shows regression results for the growth rate of total debt. In contrast to the previous finding, distressed firms tend to have lower total debt growth in all specifications. It is not clear whether this is demand-driven (the company refrains from taking on more debt) or supply-driven (creditors forgive the existing debt or refuse to lend more). The result for the basic specification suggests that firms under restructuring reduce debt by 1.6%. One might think that bank efforts to remove non-performing loans from their balance sheets (for example, by selling loans off to the Cooperative Credit Purchase Corporation or other investors) could lead to this finding. However, we are using balance sheet data from companies for our analysis, and while a removal of non-performing loans may reduce loan assets on the bank’s balance sheet, it will not change the company’s balance sheet unless or until the loans are completely forgiven. (p.367)

Table 10.9 Restructuring and Growth of Total Debt

Variables

(1)

(2)

(3)

Lagged total debt growth

0.118

0.118

0.117

(0.008)

(0.008)

(0.008)

DISTRESS

−0.017

−0.015

−0.017

(0.006)

(0.007)

(0.006)

saiken

−0.016

−0.015

−0.033

(0.006)

(0.006)

(0.011)

DISTRESS X saiken

−0.010

(0.019)

saiken X (1993–2003 dummy)

0.033

(0.013)

saiken X (2005–2007 dummy)

−0.004

(0.019)

Year dummies

Yes

Yes

Yes

Industry dummies

Yes

Yes

Yes

Number of observations

26750

26750

26750

R-squared

.048

.048

.048

The dependent variable is the growth rate of total debt of the firm. See table 10.6 for other notes.

The interaction term between DISTRESS and saiken is not statistically significant, meaning there is no difference in the impact of restructuring between firms in distress and firms not in distress. Moreover, the coefficient estimate on the interaction between saiken and the lost decade dummy is positive and offsets the negative coefficient on the saiken variable. From this we can see that during the lost decade, restructuring did not affect the growth rate of debt. In other words, while restructuring meant debt reduction in the 1980s, it did not in the 1990s. The point estimates suggest that while the average debt growth in the 1980s was 5.6% (table 10.3), firms undergoing restructuring showed debt growth of only 2.3%. In the 1990s the average firm reduced debt by 0.15% whether or not it was undergoing restructuring. This is consistent with the previous finding, in that restructuring efforts during the lost decade often remained tentative.

Overall, we find that firms under restructuring tended to reduce their growth of employment, capital, and total debt, though not bank loans. However, the lost decade was different in that companies under restructuring did not stand out for lower debt growth, and in fact may (p.368) have seen higher bank loan growth than companies not under restructuring. Moreover, even though companies under restructuring cut employment growth and capital growth during the lost decade, the extent to which they reduced capital growth was significantly less.

3.3 The Role of the Restructuring Leader

We now consider whether there are differences in process, depending on who leads the restructuring episode. Table 10.2 showed preliminary data on such differences, as well as six categories of leaders of restructuring: the main bank, a group of banks, a company or group of companies, a combination thereof, or an equity fund.

First, we examine whether growth of employment, capital, bank loans or total debt differ depending on the identity of the leaders. To do this, we regress the growth variable on its lag and six dummy variables, each taking the value 1 for a particular type of leadership. Because our regression includes a constant term, the excluded category is self-leadership (including cases where the restructuring leader is unidentified).

Results reported in table 10.10 column 1 show no differences across restructuring leadership in terms of employment growth. Likewise, capital growth rates do not differ greatly, with one exception: when a group of companies leads the restructuring, capital growth is reduced more than with other leaders (column 2). In contrast, in terms of bank loan growth we find some interesting differences (column 3). Loans are reduced more when one bank leads the restructuring effort. This result is statistically significant, suggesting that a bank as rescue leader seems to ask for more aggressive measures of financial restructuring. A similar finding is observed for total debt growth.

The apparent differences between bank leadership and other types of leadership prompt us to look at the factors that make a bank-led restructuring more likely, and whether those factors have changed over time. To examine this issue, we estimate a series of probit models similar to those above, regarding the determinants of restructuring events and changes over time. Here, the dependent variable, “bank-led,” takes the value 1 if a restructuring episode is led by a single bank (probably the main bank), and 0 otherwise. Note that the bank-led variable is defined only for those firms under restructuring that we coded for our database, and the sample size for this analysis is at most 1,756. Industry dummies are included in all specifications, as are year dummies when we add interaction terms with periods. In general, restructuring firms tend to be larger and have a higher dependence on bank debt. (p.369)

Table 10.10 Restructuring by Identity of Leader

Dependent variable →

Employment growth

Capital growth

Bank loan growth

Total debt growth

Explanatory variables ↓

(1)

(2)

(3)

(4)

Lagged dependent variable

0.228

0.193

0.163

0.104

(0.045)

(0.041)

(0.038)

(0.036)

DISTRESS

−0.052

−0.018

−0.025

−0.028

(0.011)

(0.015)

(0.019)

(0.017)

Single bank-led

−0.009

−0.015

−0.059

−0.054

(0.009)

(0.019)

(0.019)

(0.019)

Led by a single industrial

0.007

−0.018

−0.010

0.002

firm

(0.007)

(0.012)

(0.019)

(0.018)

Led by a group of banks

−0.020

0.0004

0.055

0.006

(0.020)

(0.036)

(0.045)

(0.028)

Led by a group of industrial

−0.041

−0.068

−0.054

−0.091

firms

(0.020)

(0.032)

(0.044)

(0.036)

Led by a group of banks

0.006

−0.031

0.006

−0.002

and industrial firms

(0.015)

(0.019)

(0.025)

(0.023)

Private equity fund-led

−0.018

−0.075

−0.020

−0.007

(0.023)

(0.047)

(0.045)

(0.045)

Year dummies

Yes

Yes

Yes

Yes

Industry dummies

Yes

Yes

Yes

Yes

Number of observations

1260

1260

1260

1260

R-squared

.175

.125

.101

.100

The dependent variable is specified in the first row of the table. The sample includes only those firms that are under restructuring. See table 10.6 for other notes.

The results in table 10.11 indicate that large size and a high ratio of bank debt to assets increase the probability that restructuring is led by a bank. The distress dummy does not seem to matter, suggesting that the likelihood of bank-led restructuring (as opposed to other leaders) does not depend on whether a company is in distress or not, at least as measured by our classifications of negative operating income, interest coverage ratio, or negative net profits. The specification in column 2 includes main bank dependence as an additional explanatory variable. As one might expect, we find that higher dependence on a main bank increases the probability of bank-led restructuring. The last two columns report estimation results with various interaction terms with periods. None of these is statistically significant. Thus, we do not find any significant change over time in the determinants of bank-led restructuring. (p.370)

Table 10.11 Determinants of Bank-led Restructurign

Variables

(1)

(2)

(3)

(4)

Bank debt / total assets

1.143

1.370

0.911

1.312

(0.237)

(0.307)

(0.383)

(0.311)

Bank debt / total assets X

0.462

(1993–2003 dummy)

(0.489)

Bank debt / total assets X

−0.690

(2005–2007 dummy)

(0.744)

DISTRESS

−0.040

−0.219

0.099

−0.190

(0.141)

(0.182)

(0.160)

(0.205)

DISTRESS X (1993–2003

−0.083

−0.013

dummy)

(0.180)

(0.205)

DISTRESS X (2005–2007

−0.430

dummy)

(0.477)

Log (total assets)

0.081

0.058

0.108

0.092

(0.033)

(0.045)

(0.035)

(0.048)

Main bank dependence

1.065

0.678

(0.375)

(0.541)

Main bank dependence X

0.888

(1993–2003 dummy)

(0.663)

Year dummies

No

No

Yes

Yes

Industry dummies

Yes

Yes

Yes

Yes

Number of observations

1446

966

1446

966

The dependent variable takes the value 1 if the firm’s restructuring is led by a single bank during the year, and 0 otherwise. The sample includes only those firms under restructuring. Thus, the 0 value for the dependent variable means that the restructuring is led by entities other than a single bank or the leader is not clear. See table 10.6 for other notes.

We conclude from table 10.11 that, conditional on a restructuring event, the likelihood that the restructuring is led by a single bank is positively influenced by the firm’s size, dependence on bank debt in general, and dependence on the main bank in particular. We do not find any significant changes over time.

4 Conclusions

Using a unique database on major corporate restructuring cases in Japan, this chapter has examined possible changes in rescue mechanisms over time and the role of the main bank. We have looked at what type of firms are more likely to be restructured, what restructuring implies (p.371) for the company’s employment, investment, and debt, and how those relations have changed over time, and in particular during the lost decade.

Our analysis offers four important findings. First, the likelihood for firms in distress to undergo restructuring has decreased over time. Throughout the period 1981–2007, the larger the company in distress and the higher its dependence on bank debt, the more likely it was to undergo restructuring. However, this likelihood has declined over time.

Second, during the 1980s a high dependence on loans from a main bank meant a higher likelihood that a company would undergo restructuring. However, this was not true during the lost decade. During the 1990s, high main bank dependence had no impact on the likelihood that a firm would undergo restructuring.

Third, the lost decade was also different in terms of content of restructuring. During the 1990s firms undergoing restructuring, compared with firms not restructuring, did not reduce investments as much as they did in the 1980s. Further, these firms did not reduce debt as much, and may have increased bank borrowing. Even though restructuring firms reduced employment more than other firms, overall it appears that restructuring efforts during the lost decade were less aggressive than in the 1980s, and as they appear to be in the 2000s.

Finally, we find that even though main bank-led restructuring appeared to happen less frequently over time, when it did happen, it seemed to be effective in terms of pushing financial adjustments. When a bank took charge, the distressed firm reduced debt and bank loans more aggressively.

Overall, the results are consistent with the notion that corporate restructuring was less frequent and less decisive during the 1990s, and that the lack of vigorous restructuring slowed the Japanese economy. Our results show that there were some large and highly bank-dependent companies in distress in the 1990s that would have been restructured in the 1980s but were not during the lost decade. This decline in restructuring activity was apparently linked to a reluctance by main banks to intervene in troubled companies, as firms with high main-bank dependence were more likely to be restructured in the 1980s than in the 1990s. It is possible that banks were themselves in too much trouble to bail out their failing clients, or that financial deregulation and a changing bank-firm relationship, in combination with the reforms of the 1990s, have altered the behavior of banks. Either way, supporting arguments proposed (p.372) by Caballero, Hoshi, and Kashyap (2008) and others, the lessened role of the main banks in pushing corporate restructuring during the 1990s, and the slow emergence of alternative mechanisms and players only in the 2000s, may have contributed to the 1990s becoming a lost decade.

At the same time, main banks continue to lead some episodes of corporate restructuring, and when they do, we find the restructuring effort to reduce the growth of bank borrowing and debt more drastic. Thus, it is possible that the banks are still willing and able to restructure troubled clients as well as, or even better than, they used to do, but that they have become more selective in doing so. This interesting question is left for future research.

Acknowledgments

We are grateful for helpful comments by Anil Kashyap and the editorial board, participants of the ESRI conferences in New York and San Francisco, and seminar participants at Hitotsubashi University and the Development Bank of Japan. We thank Emi Fukuda, Kanako Hotta, Masafumi Iino, Akifumi Irie, Yuichiro Kawai, Yoshikazu Kuki, Kuniaki Nemoto, Masashi Osakada, Mary Shiratori, Christopher Syling, Kunio Takeda, and Koki Yoshida for research assistance.

References

Bibliography references:

Aoki, Masahiko and Hugh Patrick (Eds.). 1994. The Japanese Main Bank System. Oxford, UK: Oxford University Press.

Caballero, Ricardo, Takeo Hoshi, and Anil Kashyap (2008). “Zombie Lending and Depressed Restructuring in Japan.” American Economic Review 98: 1943–77.

Hirota, Shinichi, and Hideaki Miyajima (2001). “Mein Banku Kainyu-gata Gabanansu ha Henka Shitaka? (Has Main Bank-led Governance Changed?),” Gendai Fainansu, 10: 35–61.

Hoshi, Takeo, and Anil Kashyap (2001). Corporate Financing and Governance in Japan: Road to the Future. Cambridge, MA: MIT Press.

Hoshi, Takeo, Anil Kashyap and David Scharfstein (1990). “The Role of Banks in Reducing the Costs of Financial Distress in Japan,” Journal of Financial Economics, 27, 67–88.

Iwaisako, Tokuo (2005). “Corporate Investment and Restructuring,” in Takatoshi Ito, Hugh Patrick, and David Weinstein (Eds.) Reviving Japan’s Economy. Cambridge, MA: MIT Press, 275–310.

Kang, Jun-Koo & Shivdasani, Anil. 1995. “Firm Performance, Corporate Governance, and Top Executive Turnover in Japan,” Journal of Financial Economics, 38:29–58.

(p.373) Kaplan, Steven N. & Minton, Bernadette A. 1994. “Appointments of Outsiders to Japanese Boards: Determinants and Implications for Managers,” Journal of Financial Economics, 36:225–258.

Koibuchi, Satoshi (2008). Ushinawareta junen no saiken hoki: Sangyo saiei kiko no inpakuto (”Debt Forgiveness during the ‘Lost Decade’: Impacts of the Industrial Revitalization Corporation of Japan”), Kinyu keizai kenkyu (Review of Monetary and Financial Studies), 27: 1–24.

Morck, Randall & Nakamura, Masao. 1999. “Banks and Corporate Control in Japan,” The Journal of Finance, 54 1: 319–339.

Peek, Joe and Eric S. Rosengren, (2005). “Unnatural Selection: Perverse Incentives and the Misallocation of Credit in Japan,” American Economic Review, 95, 1144–1166.

Schaede, Ulrike (2008). Choose and Focus: Japanese Business Strategies for the 21st Century. Ithaca, NY: Cornell University Press.

Sheard, Paul. 1989. “The Main Bank System and Corporate Monitoring and Control in Japan,” Journal of Economic Behavior and Organization, 11:399–422.

Sheard, Paul (1994). “Main Banks and the Governance of Financial Distress,” in Masahiko Aoki and Hugh Patrick (Eds.) The Japanese Main Bank System. Oxford, UK: Oxford University Press, pp. 188–230.

TSE (Tokyo Stock Exchange). 2007. Heisei 18nendo kabushiki bunpu jokyo no chosa kekka ni tsuite (Results of the 2006 Survey on Shareholdings). http://www.tse.or.jp/data/examination/distribute/h18/distribute_h18a.pdf. Tokyo.

White, Halbert (1980). “A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity,” Econometrica, 48: 817–838.

Yanagawa, Noriyuki (2006). Ho to KigyO Kodo no Keizai Bunseki (Economic Analysis of Law and Corporate Behavior). Tokyo, Japan: Nihon Keizai Shimbun-sha.