- Email: [email protected]
The effects of bank relationships on firm private debt restructuring: Evidence from an emerging market
Research in International Business and Finance 25 (2011) 113–125
Contents lists available at ScienceDirect
Research in International Business and Finance j o ur na l ho me pa ge : w w w . e l s e v i e r . c o m / l o c a t e / r i b a f
The effects of bank relationships on firm private debt restructuring: Evidence from an emerging market Jiang-Chuan Huang a,∗, Chin-Sheng Huang b a
Department of Finance, Transworld University, 1221, Jen Nang Rd., Douliu, Yunlin 640, Taiwan, ROC Department of Finance, National Yunlin University of Science & Technology, Room MB401, No. 123, University Road, Sec.3, Douliu, Yunlin 64002, Taiwan, ROC
b
a r t i c l e
i n f o
Article history: Received 30 November 2008 Received in revised form 8 September 2010 Accepted 9 September 2010 Available online 17 September 2010 JEL classification: G21 G33 G34 Keywords: Bank relationship Debt restructuring Private renegotiation Financial distress
a b s t r a c t Our paper seeks to examine the direct benefit of bank relationships for a distressed borrower by assessing its influence on the success of firm private debt restructuring. We find that a distressed firm with a stronger bank relationship has a greater probability to successfully restructure its debt through private renegotiation. Accordingly, an analysis of credit rating recovery provides complementary evidence on the factors of successful debt restructuring. A duration analysis of the length of time needed for a debt restructuring to be completed is fully consistent with our documented results. We conclude that in a bank dominated financial system like Taiwan’s where firms are heavily bank-dependent, the bank–firm relationship is of crucial importance to the success of financially distressed firms in private debt restructuring. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Extant literature on financial intermediation (see, e.g., Diamond, 1984; Ramakrishnan and Thakor, 1984) emphasizes the role of banks in generating information, for instance, through screening (Diamond, 1991) and monitoring (Rajan and Winton, 1995). This access to information is especially relevant if the borrowing firm is in financial distress. A bank’s evaluation of the borrowers based on their inside information affects the bank’s decision to renegotiate the debt or force a firm into
∗ Corresponding author. Tel.: +886 5 534201#5223; fax: +886 5 5312079. E-mail addresses: [email protected] (J.-C. Huang), [email protected] (C.-S. Huang). 0275-5319/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.ribaf.2010.09.001
114
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
bankruptcy (Chemmanur and Fulghieri, 1994). While the borrowing firm encounters financial distress, previous lending relationships with the distressed firm create a significant negative effect on the lending bank (Dahiya et al., 2003). Consequently, when the borrowing firm is in financial distress, the lending banks have two options. First, prudent banking norms immediately require the firm to repay the debt regardless of the possibility of its recovery (i.e., an instant termination of bank relationships). Second, banks are insiders, with significant information advantages. They decide to extensively involve in the firm’s private renegotiation and debt restructuring to recover losses (i.e., a continuation of bank relationships). Therefore, bank relationships may or may not benefit distressed corporate borrowers encountering financial distress. Relatively few studies in emerging economies investigate this issue. Thus, this study examines how bank relationships affect the probability of successful private debt restructuring for financially distressed borrowers. Recent research has shown how bank relationships affect private debt restructuring of financially distressed firms. Brunner and Krahnen (2008) used the number of banks as a proxy for bank relationships and observed an extensive involvement of banks in their borrowers’ debt restructuring and private workout activities. Brunner and Krahnen found that the probability of recovery from a distressed situation is negatively related to the number of banks. Couwenberg and Jong (2006) used bank debt as a proxy for the effect of bank relationships. Using this proxy, they studied the private restructuring processes in Dutch distressed firms. They found that bank debt has a significantly positive effect on the likelihood of restructuring success. While these studies show that bank relationships increase the value of distressed firms, bank relationship measures are susceptible to potential weakness, such as the degree of bank relationships measured by a single proxy. Indeed, if a firm requires fewer external funds, it would finance fewer loans with a smaller number of banks. In this situation, there is not a close or strong relationship between bank and firm. Moreover, the Taiwan financial structure is typically a bank-based financial system in which firms tend to be heavily dependent upon bank loans.1 It seems inappropriate to employ a single proxy, the size of the bank debt ratio, for bank relationships because most firms have high bank debt ratios. Hence, in contrast to the previous literature which used only a single proxy for bank relationships, the conclusions of the present research may be robust to the different bank relationship measures. The concept of a “bank relationship” is quite elusive in banking theory. There is no uniformly accepted methodology for measuring the presence and strength of bank relationships (Bharath et al., 2007). If the precise point of the start of a bank relationship is available, researchers often use the length of a relationship as a proxy for its strength (see, e.g. Petersen and Rajan, 1994; Berger and Udell, 1995; Elsas and Krahnen, 1998). In cases in which this information is not available, the existence of a prior bank relationship is used as a proxy (see, for example, Dahiya et al., 2003; Schenone, 2004; Bharath et al., 2007). Not having the precise records of bank relationships in the Taiwan banking industry, this present study adopts the existence of a prior bank relationship as a proxy for the presence of bank relationships. Moreover, this study looks at the size of bank relationships and the number of bank relationships by searching the borrower’s previous borrowing records. In short, this study simultaneously discusses three measures of bank relationships to examine the effect of the success of private debt restructuring for financially distressed firms. These measures include the existence, size, and number of bank relationships. The early empirical studies mainly have been concerned with the argument of formal bankruptcy procedures and have paid relatively less attention to out-of-court debt restructuring. Jensen (1989) is one of a few works that advocated that private contractual arrangements for resolving default represent a viable and less costly alternative to the legal remedies provided by Chapter 11. Gilson et al. (1990) further examined the determinants of 169 financially distressed firms’ choice between formal bankruptcy and out-of-court restructuring. Gilson et al. found that about half (80 firms) successfully restructured their debt through out-of-court renegotiations and these firms have more intangible assets, a larger percentage of debt owed to banks, and fewer lenders. 1 Shen and Wang (2005) reported that the external funding sources of Taiwanese firms account for 60% of the total funds. Among the entirely external funds, the largest shares are loans from financial institutions, particularly from commercial banks. Hence, bank relationships are of crucial importance for Taiwan’s firms. Suggestively, the financial structure in Taiwan is much like a bank-based financial market.
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
115
Moreover, the financial literature has identified firm size (Couwenberg and Jong, 2006), firm age (Demine and Carvalho, 2006), leverage (Brunner and Krahnen, 2008; Couwenberg and Jong, 2006), and return on assets (Claessens et al., 2003; Couwenberg and Jong, 2006; Franks and Sussman, 2005) as the critical factors and investigated the effects of firm characteristics upon the probability of firm survival. However, regarding the possible impacts of debt structure on the success of firms’ private restructuring, the extant literature produces mixed results. Financial distress is more likely to be resolved through private renegotiation when both relatively less debt is owed to trade creditors (Gilson et al., 1990) and more is owed to bank lenders (Couwenberg and Jong, 2006; Gilson et al., 1990). Franks and Sussman (2005) showed that the number and value of collateral over bank debt do not significantly affect the probability of firm survival. Consequently, this study also investigates the impacts of firm characteristics and debt structure upon the success of private debt restructuring. The aim of this study is to investigate the determinants of successful private debt restructuring for financially distressed firms using Taiwan data. In particular, the primary goal of the present research is to examine the effects of bank relationships upon the success of private debt restructuring of distressed firms. This study hypothesizes that bank relationships significantly increase the success of private debt restructuring for financially distressed firms. This hypothesis predicts that stronger bank relationships increase the probability that a distressed firm can successfully restructure its debt through private renegotiation. Debt restructuring seldom begins or ends with a formal public announcement, and outsiders do not generally have a clear picture of a firm’s private debt restructuring plan. This makes it difficult for researchers to collect data. This study presents a well-defined beginning and ending date of firm debt restructurings to identify the success or failure of a distressed firm’s private debt restructurings without depending on a bank’s internal information using a credit rating index. Consequently, this study contributes to the firm’s private debt restructuring research by providing a helpful methodology for the relevant studies in the financial market where few data are publicly available. The remainder of the paper is organized as follows. Section 2 describes the data and sample selection. The methodology and empirical results are presented in Section 3. The final section contains the concluding remarks. 2. Data and sample selection The sample in this study consists of companies listed in Taiwan that encountered financial distress from 1995 to 2003. These companies either successfully recovered or failed in private debt restructuring. In addition, general characteristics of the firms, the firm’s income statement and balance sheet data, and an assessment of the corporate credit rating are retrieved from the Taiwan Economic Journal (TEJ) database. This study proposes an alternative procedure to identity the duration of private debt restructuring. A distress event is defined in this research as the point at which the Taiwan Corporate Credit Risk Index (TCRI)2 assigns a firm a distress rating of 9 or 10. Likewise, this study shows that firm’s private debt restructuring tends to be as successful when a distressed firm’s TCRI rating improves from a distressed rating of 9 or 10 to a rating of 6 or better. This proposed empirical method will significantly mitigate the data impediments confined in the research field of debt restructuring among distressed firms. The sample in this study includes 302 firms that either was distressed or had defaulted on their obligations.3 Among these selected firms, 10 firms did not have enough financial data4 to trace their
2 The TCRI evaluation system was developed by the TEJ in August, 1991. The standard methodology of the rating relies on a scoring system according to 10 financial criteria, including profitability, safety, activity and size on corporate performance and prospects, and a linear weighting system with the fixed weighting factors. TCRI employs a 1–10 rating scale (best to worst), in which notches 1–4 are categorized as high investment grade, notches 5–6 are categorized as medium investment grade, notches 7–9 represents speculative grade, and the notch 10 is reserved for a default case (extremely high risk). 3 In the category of firms with multiple defaults during the 1995–2003 period, we have only retained the first default case in order to avoid several defaults linked to the same firm which might have biased the econometric analysis. 4 Some of the firms merged or were taken over the period 1995–2003. Thus, those firms did not have available financial data anymore.
116
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
debt restructurings, and 15 firms were eliminated due to judicial restructurings. Ultimately, this study uses a minimum total bank debt of 15 million (NTD) as the selection criteria to ensure accurate information from previous bank relationships. The sample excluded 5 firms with small bank debt, yielding 272 distressed or default firms of which 63 (23.16%) restructured successfully and 209 (76.84%) were not successful during the period, 1995–2006. 3. Empirical results 3.1. Descriptive statistics of the sample Table 1 presents the descriptive statistics of the sample. Panel A of Table 1 is the distribution of the sample by year of debt restructurings. In 272 financially distressed firms, over half (i.e., 51.84%) are clustered in the years 1998–2000 due to the timing of the domestic financial crisis (from July 1998 to January 1999) and the economic recession (from July 2000 to September 2001). Additionally, the average length of 63 successfully restructured firms is 43.22 months; in particular, the first 2 years (1995–1996) show higher average length of successful restructurings than in other periods. The restructuring firms are categorized into 16 industries based on the security code of the Taiwan Security Exchange Council (TSEC) as reported in Panel B. Particularly, the Electronics industry (34.19% and 57.14%) displays the highest percentage of overall distressed firms and successful firms, and the shortest average length of successful restructuring (34.81 months) among all the industries; the average assets (NTD 3353 million) and the average age (12.31 years) are less than the average values of overall successful firms. The descriptive statistics of the firm characteristics are presented in Panels C of Table 1. The firm size (as measured by the annual sales and the total assets), total liabilities, and return on assets (ROA) have consistently displayed considerable volatility. Moreover, the average length of successful debt restructuring period is 43.22 months (median is 36 months) which is apparently longer than those reported for US firms of 15.4 months (Gilson et al., 1990), small UK firms of 9.2 months (Franks and Sussman, 2005), and Dutch firms of 24 months (Couwenberg and Jong, 2006). One main reason is that a definition of a successful debt restructuring in this study requires firms succeeding in credit ratings upgrading to indeed lower their risk category (i.e., TCRI rating 6 or better). The lending bank information and firm’s debt structure are presented in Panel D of Table 1. The average number of banks amounts to 10.42 (with a median of 8) ranging from single banking to 41 banks. This figure is apparently higher than those reported for the Taiwan non-financially distressed firms of 9.47 banks by Fok et al. (2004) and 8.355 banks by Shen and Wang (2005). The evidence agrees with the presumed argument that the distressed firms usually maintain more banking relationships to finance extra external funds in the cases of insufficient cash flows. The size of bank relationships is a proxy as a measure for the exposure of the main bank, i.e., total amount of loans to the distressed borrower by the main bank in the last 5 years divided by total amount of loans from all lending banks. The average bank relationship size is 0.38, with the median 0.33, ranging from 0.02 to 0.97. 3.2. Univariate analysis The sample of 272 debt-restructuring attempts is further partitioned into two subsamples of successful firms (63 firms) and unsuccessful firms (209 firms). Table 2 presents the results of differentiate analysis of firm characteristics, bank characteristics, and debt structure for these two subsamples. In successful firms, the selected firm characteristics of average total assets, firm age, and leverage ratio are significantly (at the 1% level) smaller than those in unsuccessful firms are. For the ROA, however, the difference between in these two types of firms is insignificant. The proxies of bank relationships, including the existence of prior bank relationships, the size of bank relationships, and the number of banks, are distinct between the two subsamples (at the 1% level). Successful firms have a significantly greater average account-payable ratio (at the 5% level), but the bank debt ratio and the secured debt ratio exhibit insignificant differences between the two groups. The government rescue dummy shows no significant difference between two groups. In the group of successful firms, the industry dummy and real GDP growth rate are significantly greater than those in the unsuccessful group are (at the 1%
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
117
Table 1 Descriptive statistics of the sample. Panel A: Distribution of the sample by year of debt restructuring Year
Total firms (and %)
1995 1996 1997 1998 1999 2000 2001 2002 2003 Total
23(8.46) 16(5.88) 15(5.51) 60(22.06) 37(13.60) 44(16.18) 29(10.66) 30(11.03) 18(6.22) 272(100.00)
Successful firms (and %) 11(17.46) 8(12.70) 9(14.29) 6(9.52) 8(12.70) 10(15.87) 5(7.94) 3(4.76) 3(4.76) 63(100.00)
Percentage of successful firms (%)
Average length for successful restructuring (month)
47.83 50.00 60.00 10.00 21.62 22.73 17.24 10.00 16.67 23.16
64.00 60.38 40.67 29.50 34.88 38.10 25.20 39.00 30.00 43.22
Panel B: Distribution of the sample by industry of debt restructuring Successful debt restructuring a
Code
Industry
12 13 14 15 16 17 18
Food Plastics Textiles Electrical Machinery Wire and Cable Chemical Ceramics and glass products Pulp and paper Steel Rubber Electronics Construction Transportation Tourism Department stores Other Over sample
19 20 21 23, 24b 25 26 27 29 99
Total firms (and %)
Successful firms (and %)
Successful percentage (%)
Average length (month)
Average asset (NTD mil.)
Average age (year)
15(5.51) 6(2.21) 21(7.72) 13(4.78) 5(1.84) 9(3.31) 5(1.84)
2(3.17) 0(0.00) 0(0.00) 2(3.17) 0(0.00) 6(9.52) 0(0.00)
13.33 N/A N/A 15.38 N/A 66.67 N/A
72.00 N/A N/A 39.00 N/A 59.50 N/A
7751 N/A N/A 2188 N/A 5508 N/A
27.50 N/A N/A 21.50 N/A 30.17 N/A
1(0.37) 25(9.19) 1(0.37) 93(34.19) 48(17.65) 9(3.31) 2(0.74) 4(1.47) 15(5.51) 272(100.00)
0(0.00) 6(9.52) 0(0.00) 36(57.14) 7(11.11) 2(3.17) 0(0.00) 1(1.59) 1(1.59) 63(100.00)
N/A 24.00 N/A 38.71 14.58 22.22 N/A 25.00 6.67 23.16
N/A 48.50 N/A 34.81 53.14 66.00 N/A 48.00 48.00 43.22
N/A 17,868 N/A 3353 7261 1704 N/A 1342 1823 5369
N/A 18.17 N/A 12.31 10.57 15.00 N/A 8.00 3.00 15.02
Panel C: Firm characteristics and restructuring information (N = 272)
Total sales (NTD in million) Total assets (NTD in million) Total liabilities (NTD in million) Return on assets (%) Firm’s age (year) Length for successful restructuring (month)
Mean
Median
2301 7405 4463 0.17 20 43.22
1058 3244 1803 1.79 18 36
Standard deviation 3752 10,224 6441 8.55 11.80 32.52
Min.
Max.
0.54 114 33 -33.38 1 6
37,867 52,631 37,081 30.42 51 108
118
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
Table 1 (Continued) Panel D: Lending bank information and firm’s debt structure (N = 272)
Number of banks Size of bank relationships Total account-payable (NTD in million) Total bank debt (NTD in million) Total secured debt (NTD in million)
Mean
Median
Standard deviation
10.42 0.38 412 2534 1902
8 0.33 176 982 682
7 0.21 711 3903 3201
Min.
Max.
1 0.02 1.46 16 0
41 0.97 6424 22,820 21,435
a The restructuring firms are categorized into 16 industries based on the security code of the Taiwan Security Exchange Council (TSEC). In Panel B, the industry code is represented by the first two digits of the security code. b The Electronics industry is divided into two industry sectors. One industry sector includes semiconductor, computer and peripheral equipment, electronic parts/components, and other electronics (the first two digits of the security code is ‘23’). The other industry sector contains communications and internet, electronic products distribution, and information service (the first two digits of the security code is ‘24’).
Table 2 Differences in mean (median) of firm characteristics, bank characteristics and debt structure. Mean (Median)
Total assets (NTD in million) Government rescuea Industrya Real GDP growth rate Existence of bank relationshipsa Size of bank relationships Number of banks Firm age (year) Leverage ratio Return on assets (%) Account-payable ratio Bank debt ratio Secured debt ratio Samples a ** ***
Differences of successful and unsuccessful firms (p-value)
Successful firms
Unsuccessful firms
Differences
t-test
Wilcoxon test (Rank-sum)
5369(1522) 0.36(0.00) 0.68(1.00) 5.11(5.77) 0.76(1.00) 0.50(0.53) 8.33(7.00) 15.02(11.00) 0.60(0.61) 0.11(1.26) 0.15(0.12) 0.63(0.63) 0.70(0.78) 63
8018(4174) 0.42(0.00) 0.46(0.00) 4.25(4.55) 0.53(1.00) 0.33(0.31) 11.05(9.00) 21.51(21.00) 0.66(0.66) 0.18(1.84) 0.11(0.10) 0.59(0.60) 0.74(0.81) 209
−2649(−2652) −0.06(0.00) 0.22(1.00) 0.86(1.22) 0.23(0.00) 0.17(0.22) −2.72(−2.00) −6.49(−10.00) −0.06(−0.05) −0.07(−0.58) 0.04(0.02) 0.04(0.03) −0.04(−0.03)
0.01*** 0.43 0.00*** 0.01** 0.00*** 0.00*** 0.00*** 0.00*** 0.00*** 0.95 0.02** 0.31 0.80
0.00*** 0.38 0.00*** 0.00*** 0.00*** 0.00*** 0.00*** 0.00*** 0.00*** 0.94 0.05** 0.81 0.77
Government rescue, Industry, and existence of bank relationships are represented by dummies. Significant at the 5%. Significant at the 1%.
level and 5% level, respectively). This study also performed a nonparametric Wilcoxon Rank-sum test for median differences between the two subsamples. In virtually all cases, the direction and the level of statistical significance of the results of this test are consistent with those obtained from the t-tests. 3.3. Pearson correlation analysis The correlations between the influencing factors of the regression analysis of the following section are shown in Table 3. Particularly, there is a strong positive correlation of 0.674 between firm size and the number of banks because relatively larger firms typically use bank loans from multiple bank lenders. This result is affirmative with Houston and James (1996) that multiple-bank firms tend to be larger than single-bank firms and Shen and Wang (2005) that the optimal number of bank relationships is not constant but possibly increases with the size of a firm. In addition, two potentially influencing factors for the strength of the bank relationships, the number of banks versus the size of bank relationships, are negatively correlated (−0.549) because for a specific firm, the loan shares of a main lending bank decrease with the number of lending banks. All other variables show moderately Pearson correlations ranging from −0.453 to 0.428.
1. Dependent variable 2. Firm size 3. Government rescue 4. Industry 5. Real GDP growth rate 6. Existence of bank relationships 7. Size of bank relationships 8. Number of banks 9. Firm age 10.Leverage ratio 11.Return on assets 12.Account-payable ratio 13.Bank debt ratio 14.Secured debt ratio
1
2
3
4
5
6
7
8
9
10
11
12
13
14
1.000 −0.228 −0.048 0.184 0.149 0.201 0.349 −0.329 −0.284 −0.206 0.003 0.150 0.062 −0.016
1.000 0.175 −0.314 −0.006 −0.016 −0.431 0.674 0.428 0.191 0.178 −0.297 −0.145 −0.070
1.000 −0.167 0.331 0.038 −0.125 0.158 0.176 −0.080 0.062 −0.069 −0.028 −0.080
1.000 −0.090 −0.005 0.141 −0.310 −0.453 −0.178 −0.085 0.248 0.027 0.132
1.000 −0.105 0.045 −0.065 −0.042 0.004 0.112 −0.008 −0.075 −0.125
1.000 0.304 −0.068 0.031 −0.012 −0.057 −0.018 0.157 0.114
1.000 −0.549 −0.301 −0.186 −0.136 0.081 0.019 0.122
1.000 0.369 0.288 0.231 −0.208 −0.002 −0.120
1.000 0.149 0.257 −0.032 −0.233 −0.057
1.000 0.071 −0.052 −0.024 −0.135
1.000 0.098 −0.114 0.060
1.000 −0.205 −0.072
1.000 0.039
1.000
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
Table 3 Description of the correlations between the factors of potential influences (N = 272).
119
120
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
3.4. Determinants of successful debt restructurings for financially distressed firms Three distinct estimation methodologies are employed in analyzing the determinants of restructured success. First, this study uses a logistic regression model to illustrate how potent influencing variables affect the probability of restructured success. Next, an ordered-probit model is employed to examine the effects of potential factors upon the probability of the corporate rating improvement. Furthermore, a duration model is used to ascertain how variables might carry weight on the time length of successfully completing private debt restructuring. 3.4.1. Prediction of successful restructuring In the analysis of Section 3.2, this study has documented several differences between successful and unsuccessful firms in univariate comparisons. Table 4 further reports the empirical results of logistic regressions. The logistic regression model contains two explanatory variables: control variables and theory-based variables. The control variables consist of firm size (log of total assets), a government rescue dummy5 , an industry dummy6 , and real GDP growth rate (counted at the onset of the firm’s debt-restructuring year). The theory-based variables in the regression models come from bank relationships, firm characteristics, and debt structure. Specifically, this study constructs three alternative bank relationship measures by analyzing a distressed firm’s previous borrowing records. Specifically, this study looks back over a period of 5 years for each loan by a typically distressed firm.7 The first bank relationship measure is specified as a dummy variable taking one if the main bank8 of the distressed firm had served as a main bank in lending (making previous loans) to the distressed borrower before its default. The second bank relationship variable is the exposure of the main bank to the distressed borrower which captures the size of past bank relationships. The last bank relationship variable is the number of banks which represents the strength of bank relationships. This study predicts that the existence of prior relationships and size of bank relationships have positive effects on the probability of restructuring success. However, the number of banks has the opposite effect on the restructuring success. The firm characteristics of this analysis include firm age (log of the firm’s age), leverage ratio (total liabilities over total assets), and ROA (in the year prior to the onset debt restructuring). The firm’s debt structure is proxied by three variables: relative size of account-payable debt (account-payable debt over total liabilities), bank debt (bank debt over total liabilities) and secured debt (secured debt over bank debt). All the debt structure variables are computed at the onset of the firm’s debt-restructuring year. As the results in Panel A of Table 4 indicate, regressions (1)–(6) show that all of the coefficients of the three bank relationship variables are highly significant and with correct signs. Since there is a high correlation (−0.549) between the number of banks and the size of bank relationships, this study built up a principal component, named intensity of bank relationships, in the regressions (6)–(9).9 The negative coefficient of the intensity of bank relationships in regressions (6)–(9) implies that the negative
5 Inoue et al. (2008) documented that out-of-court restructuring has a positive effect on the market value of the distressed firms only when monitored by bank supervisors. Claessens et al. (2003) ascribed the high likelihood of out-of-court renegotiations in the East Asian countries to poorer credit rights and an inefficient judicial system in the region. In the Asian financial crisis, 1997–1998, the Ministry of Finance in Taiwan set up a “Committee for Composition Corporate Operational Funds” for loan relief during the period, November 1998 to December 2001. Hence, this study defines a government rescue effort dummy to examine the influence of government involvement on the success of firm private debt restructuring. 6 The industry dummy is unity if a distressed firm belongs to the “new” economy (e.g., the electronics and high-tech industries) and zero if it belongs to the “conventional” economy (e.g., the construction, paper and pulp, food, and steel industries). 7 Bharath et al. (2007) reported that they chose the 5 year window as approximately 75% of the loan facilities in their sample having maturities of less than or equal to 5 years. Hence, most of the borrowers in the studied sample would need to refinance their debt within five years. 8 This study focuses on the main bank of a particular distressed firm, as the information-intensive role that the hypothesis in this study is most appropriate for the main bank, which typically holds the largest share of a firm’s bank debt (see as, Gilson et al., 1990; Franks and Sussman, 2005; Bharath et al., 2007; Inoue et al., 2008). Thus, the responsibilities of a main bank best fit the description of a relationship lender. 9 In regressions (6)–(9), this study employs the intensity of bank relationship variable to replace the number of banks and the size of bank relationships in order to avoid the statistical multicollinearity problem.
Table 4 The determinants of successful restructuring for financially distressed firms—Logistic regression. Panel A: Bank relationships Variables
(1)
(2)
1.231 (1.452) −0.176* (0.092) −0.231 (0.223) 0.541** (0.239) 0.187*** (0.062) 0.887*** (0.244)
−1.319 (1.527) −0.095 (0.092) −0.056 (0.215) 0.571** (0.227) 0.150*** (0.058)
(3)
(4)
−0.201 (0.016) −0.158 (0.217) 0.455** (0.229) 0.150*** (0.059)
2.361*** (0.517)
(5)
−1.162 (1.524) −0.121 (0.092) −0.118 (0.214) 0.566** (0.226) 0.163*** (0.057) 0.624*** (0.237) 1.913*** (0.517)
272 0.209
272 0.191
−2.159 (0.433)
−0.207 (0.212) 0.450** (0.223) 0.163*** (0.056) 0.786*** (0.233)
−0.139 (0.209) 0.496** (0.214) 0.152*** (0.055) 0.628*** (0.229)
−0.637*** (0.154)
−0.709*** (0.161) 272 0.162
(6)
−0.854 (0.540)
272 0.232
272 0.214
−0.387*** (0.081) 272 0.244
Panel B: Firm characteristics and debt structure Variables Intercept Government rescue Industry Real GDP growth rate Existence of bank relationships Intensity of bank relationshipsa Firm age Leverage ratio Return on assets Account-payable ratio Bank debt ratio Secured debt ratio Sample size Pseudo R-sq
(7) −0.107 (0.712) −0.127 (0.211) 0.242 (0.216) 0.122** (0.053) 0.685*** (0.228) −0.296*** (0.086) −0.355** (0.151) −1.293* (0.689) 0.020* (0.012)
272 0.250
(8) −2.147*** (0.560) −0.161 (0.206) 0.429** (0.218) 0.150*** (0.055) 0.613*** (0.227) −0.382*** (0.080)
1.068 (0.885) 0.362 (0.310) −0.393 (0.348) 272 0.250
(9) 0.143 (0.806) −0.153 (0.207) 0.174 (0.221) 0.115** (0.053) 0.693*** (0.225) −0.295*** (0.085) −0.370** (0.154) −1.287* (0.664) 0.021* (0.012) 1.090 (0.872) 0.255 (0.301) −0.560 (0.342) 272 0.270
Numbers in the parentheses are standard errors. a We apply principal component analysis to retrieve a new variable of the intensity of bank relationships which is composed from the number of banks and the size of bank relationships. In regressions (6)–(9), we employ the intensity of bank relationship variable to replace the number of banks and the size of bank relationships in order to avoid the statistical multicollinearity problem. * Significant at the 10% level. **
Significant at the 5%. Significant at the 1%.
121
***
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
Intercept Firm size Government rescue Industry Real GDP growth rate Existence of bank relationships Size of bank relationships Number of banks Intensity of bank relationshipsa Sample size Pseudo R-sq
122
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
Table 5 The distribution of cumulative corporate rating improvement. The length of restructuring after post-distress 12-Month
24-Month
The ratings of cumulative recovery
Firms (%)
Firms (%)
0 rating 1 rating 2 ratings 3 ratings 4 ratings (and more) Samples (N)
88.56 3.69 4.43 2.58 0.74 272
72.48 10.85 8.14 6.59 1.94 258
36-Month Firms (%) 62.14 13.58 9.88 10.29 4.11 243
48-Month Firms (%) 56.05 13.90 10.31 12.56 7.18 223
effects of number of banks dominates the effect of size of bank relationships. Overall, the empirical results indicate that the existence of prior bank relationships, the larger size of bank relationships and stronger degree of bank relationships (less number of banks) significantly increase the probability of successful private debt-restructuring. Moreover, it is clear that a strong bank relationship alleviates information asymmetry of banks and benefits the private debt restructuring of distressed firms. Among the control variables, the Panel A reports that the firm size is negatively related to the success of firm restructuring only at marginal significance.10 This result may be imputed to the structure of Taiwanese industry where a large proportion of distressed firms belong to the “new economy” as reported previously in Panel B of Table 1 and in Table 2. Due to the absence of a consummate debt renegotiation mechanism in most emerging markets, the government rescue dummy is insignificant. Consistently with extant literature, the industry dummy and the growth rate of real GDP are statistically significant and positively related to the likelihood of successful private debt restructuring. As the results in Panel B of Table 4 show, the coefficient of the firm age is significantly negative at least at the 5% level which indicates that the younger distressed firms are more likely to be successful in restructuring. The negative effect of firm age in this study may be ascribed to a high proportion of distressed firms in the “new economy.” Incidentally, firm age shows a high negative correlation (−0.453) with the industry dummy. In regression (7), the industry dummy turns insignificant which may indicate that the firm age effect overshadow the industry effect on the successful restructuring. Furthermore, the coefficient of leverage is significantly negative which suggests that a higher leverage ratio before restructuring is less likely to be successful in restructuring. The significantly positive results (at the 10% level) for ROA show that firms with a solid operating performance before restructuring have a higher incidence of surviving. In regression (8), all measures of debt structure yield insignificant coefficients although having the correct signs. Ultimately, regression (9) contains the control variables and theoretical explanatory variables. The results show that the coefficients of overall variables have similar effects as those illustrated previously in this section. Overall, this empirical evidence indicates three bank relationship proxies capture more information concerning the efficiency of firm private debt-restructuring than those investigated in single proxy setting. Two arguments are worth noting here. First, the information advantage provided by prior bank relationships allows banks to accurately discriminate between viable and non-viable firms and to provide greater assistance to potentially successful firms. Second, the severity of the holdout problem depends on the number of lending banks participating in the firm’s private debt-restructuring. 3.4.2. Corporate rating recovery analysis In this section, this study focuses upon determinants of corporate credit rating improvement in the sample of successful debt restructuring and delivers the relevant evidence for the previous analysis. Table 5 documents the distributions of cumulative corporate rating recovery in terms of restructuring durations for 4 years annually. It is clear that the cumulative improvements increase with the lengths
10 The number of banks has a correlation of 0.674 with firm size. In unreported additional regressions we included these two variables and found that the number of banks turns insignificant, while the firm size is still not significant and has a positive effect. Therefore, we separated these two variables and report the results in Table 4, Panel A.
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
123
of restructuring. Specifically, the proportions of cumulative recovery of the 4 ratings (or more) elevate from 0.74% for the horizon of 12-month, up to 1.94%, 4.11%, and reach at 7.18% for the horizons of 24-month, 36-month, and 48-month, respectively. The same phenomenon occurs for the cumulative improvement of 3 ratings which surges to 2.58%, 6.59%, 10.29%, and 12.56%, respectively. Following the same line of reasoning, the proportion of no improvement has declined from 88.56% to 72.48%, 62.14%, and 56.05% for the horizons of the 4 years annually. This study further employs an order-probit regression for ordered response data to measure the factors determining cumulative rating improvement of distressed firms after post-distress. The model consists of the control variables and explanatory variables that are similar to the logistics regression employed earlier. However, the firm size and government rescue variables are eliminated due to their insignificant effect on firm restructuring by the previous analysis. Moreover, the real GDP growth rate is also excluded because the effects of macro-economic conditions are difficult to estimate year-by-year. Table 6 documents the regression results for each observation window after post-distress. First, the empirical evidence illustrates that the existence of prior relationships (and the intensity of bank relationships) has a statistically significant positive (negative) impact on the firm recovery at all horizons. This confirms the sample univariate results and the logistic regression results reported earlier. Second, as expected, the firm age, ROA and leverage variables all show statistically significant effects on the firm recovery across all horizons and in agreement with those findings in previous analysis. Third, an alternative finding is that the account-payable also indicates a significantly positive effect on the firm’s recovery during all horizons. These results suggest that trade creditors might provide additional financing to restructured firms due to their information asymmetry (Couwenberg and Jong, 2006; Franks and Sussman, 2005). 3.4.3. Duration analysis This study takes a further step to investigate the determinants of the duration of firm debt restructuring. Since the data involves right censored observations that bias the OLS estimates, this study employs the duration models to deal with right censoring (Ongena and Smith, 2001). The dependent variable is the duration of a distress episode (in months) that ends with a distress event of the rating improving to 6 or better. This variable makes it possible to analyze the time required for successfully completing private debt restructuring. The independent variables coincide with those of the preceding logistic regressions of Table 4. The coefficients are expected to have opposite signs compared to the logistic regressions since the dependent variable (i.e., the log of the length of restructuring completion period by months) is an inverse measure of restructuring success: the higher restructuring performance the shorter time for the firm to recover. The specifications of the duration model are
Table 6 The determinants of corporate rating recovery—Ordered-probit regression. Variables
12-Month
24-Month
36-Month
48-Month
Existence of bank relationships Intensity of bank relationshipsa Industry Firm age Leverage ratio Return on assets Account-payable ratio Bank debt ratio Secured debt ratio Sample size Pseudo R-sq
0.585** (0.302) −0.249** (0.123) 0.085 (0.286) −0.457 (0.309) −2.697*** (0.847) 0.104*** (0.025) 2.632*** (0.969) 0.576 (0.773) −0.859 (0.439) 272 0.346
0.406** (0.197) −0.282** (0.141) −0.055 (0.217) −0.383* (0.211) −2.868*** (0.612) 0.057*** (0.012) 3.195*** (0.589) 0.589 (0.613) −0.195 (0.376) 258 0.300
0.571*** (0.184) −0.215*** (0.082) −0.295 (0.204) −0.479** (0.193) −1.654** (0.525) 0.061*** (0.011) 1.803*** (0.677) 0.492 (0.520) −0.307 (0.304) 243 0.270
0.624*** (0.184) −0.203*** (0.073) 0.010 (0.204) −0.500*** (0.195) −1.932*** (0.503) 0.058*** (0.011) 1.520** (0.689) 0.195 (0.545) −0.446 (0.301) 223 0.285
Numbers in parentheses are standard errors. a We apply principal component analysis to retrieve a new variable of the intensity of bank relationships which is composed from the number of banks and the size of bank relationships. * Significant at the 10% level. ** Significant at the 5%. *** Significant at the 1%.
124
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
Table 7 The determinants of the duration of firm restructuring—Weibull distribution. Variables Intercept Firm size Government rescue Industry Real GDP growth rate Existence of bank relationships Size of bank relationships Number of banks Intensity of bank relationshipsa Firm age Leverage ratio Return on assets Account-payable ratio Bank debt ratio Secured debt ratio Sigma () Sample size
(1) 1.620*** (0.453) 0.006 (0.027) 0.040 (0.057 −0.116* (0.060) −0.020 (0.014) −0.202*** (0.063) −0.463*** (0.131)
(2)
(3)
1.192*** (0.221)
1.550*** (0.212)
0.033 (0.057) −0.067 (0.061) −0.014 (0.014) −0.249*** (0.064)
0.024 (0.055) −0.095 (0.059) 0.004 (0.014) −0.207*** (0.063)
0.125*** (0.048) 0.096** (0.040) 0.473** (0.188) −0.003 (0.003) −0.515** (0.220) −0.072 (0.084) 0.132 (0.092) 1.651*** (0.112) 272
0.112*** (0.042) 0.404** (0.204) −0.003 (0.003) −0.437** (0.204) −0.045 (0.084) 0.161* (0.094) 1.625*** (0.112) 272
0.088*** (0.024) 0.090** (0.041) 0.398** (0.186) −0.004 (0.003) −0.448** (0.203) −0.067 (0.077) 0.148 (0.090) 1.655*** (0.212) 272
Numbers in parentheses are standard errors. a We apply the principal component analysis to retrieve a new variable of the intensity of bank relationships which is composed from the number of banks and the size of bank relationships. * Significant at the 10% level. ** Significant at the 5%. *** Significant at the 1%.
conducted by a parametric survival model under the Weibull distribution (same as Ongena and Smith, 2001; Brunner and Krahnen, 2008). The results drawn from Table 7 are in line with those of the logistic estimations. The evidence shows that the existence of prior bank relationships and size of bank relationships reduce the time needed to achieve success. However, an increasing number of banks and an increasing intensity of bank relationships significantly lengthen the time required to terminate a restructuring completion. Moreover, the severity of the distress shock (leverage ratio) extends the time of the distress episode and delays an eventual completion as indicated in regressions (1)–(3). However, ROA and the real GDP growth rate turn out to be insignificant in this duration analysis. Overall, the duration analysis supports our earlier findings that bank relationships significantly account for the success of restructuring and the time needed for recovery in private debt restructuring pursued by distressed firms. 4. Conclusions This paper investigates the determinants of successful private debt-restructuring pursued by financially distressed Taiwanese firms during the period 1995–2003. Several salient features in this study will shed light on the problem of bank relationships and private debt restructuring. Firstly, this study presents a credit rating index to identify the success or failure of firm private debt restructuring. The proposed methodology provides a potential method for future research in the field of firm private debt restructuring, which frequently is confronted with data impediments. Because bank loans are private instruments, the data relevant to debt restructuring in emerging markets is rarely available to the public. Secondly, in terms of contributions to the literature, this study is one of only a few attempts to carefully document the effects of bank relationships on firm private debt restructuring. Specifically, this study employs three proxies for completely measuring the degree of bank relationships and identifies the various effects of these proxies on the success of private debt restructuring. Results indicate that bank relationships significantly affect the success of firm private debt restructuring. A financially distressed firm with a stronger bank relationship is more likely to successfully restructure its debt. Therefore, the empirical results in this study fully support the hypotheses defined by this study. Empir-
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
125
ical results also show that firms with younger age, higher profitability, less severity of distress, more account-payable debt and better macroeconomic conditions exhibit a significantly higher likelihood of successful restructuring. Finally, two supplementary pieces of evidence drawn from an analysis of the credit rating recovery and a duration analysis of the length of time needed for a private debt restructuring simultaneously confirm the findings reported in the logistic regression analysis. In brief, bank–firm relationships are crucial to the success of private debt restructuring undertaken by financially distressed firms, particularly in a bank dominated financial system where firms are heavily bank-dependent as evidenced in Taiwan. References Berger, A.N., Udell, G.F., 1995. Relationships lending and lines of credit in small firm finance. J. Bus. 68, 351–381. Bharath, S., Dahiya, S., Saunders, A., Srinivasan, A., 2007. So what do I get? The bank’s view of lending relationships. J. Financ. Econ. 85, 368–419. Brunner, A., Krahnen, J.P., 2008. Multiple lenders and corporate distress: evidence on debt restructuring. Rev. Econ. Stud. 75, 415–442. Chemmanur, T.J., Fulghieri, P., 1994. Reputation, renegotiation, and the choice between bank loans and publicly traded debt. Rev. Financ. Stud. 7, 475–506. Claessens, S., Djankov, S., Klapper, L., 2003. Resolution of corporate distress in East Asia. J. Empirical Financ. 10, 199–216. Couwenberg, O., Jong, A., 2006. It takes two to tango: an empirical tale of distressed firms and assisting banks. Int. Rev. Law Econ. 26, 429–454. Dahiya, S., Saunders, A., Srinivasan, A., 2003. Financial distress and bank lending relationships. J. Financ. 58, 375–399. Demine, J., Carvalho, C.N., 2006. Bank loan losses-given-default: a case study. J. Bank. Financ. 30, 1219–1243. Diamond, D., 1984. Financial intermediation and delegated monitoring. Rev. Econ. Stud. 51, 393–414. Diamond, D., 1991. Monitoring and reputation: the choice between bank loans and directly placed debt. J. Polit. Econ. 99, 689–721. Elsas, R., Krahnen, J.P., 1998. Is relationship lending special? Evidence from credit-file data in Germany. J. Bank. Financ. 22, 1283–1316. Fok, R.C.W., Chang, Y.C., Lee, W.T., 2004. Bank relationships and their effects on firm performance around the Asian financial crisis: evidence from Taiwan. Financ. Manage. (summer), 89–112. Franks, J., Sussman, O., 2005. Financial distress and bank restructuring of small to medium size UK companies. Rev. Financ. 9, 65–96. Gilson, S.C., John, K., Lang, L.H.P., 1990. Troubled debt restructurings: an empirical study of private reorganization of firms in default. J. Financ. Econ. 27, 315–354. Houston, J.F., James, C.M., 1996. Bank information monopolies and the choice between public and private debt claims. J. Financ. 51, 1863–1889. Inoue, K., Kato, H.K., Bremer, M., 2008. Corporate restructuring in Japan: who monitors the monitor? J. Bank. Financ. 32, 2628–2635. Jensen, M.C., 1989. Active investors, LOBs, and the privatization of bankruptcy. J. Appl. Corp. Financ. 2, 35–44. Ongena, S, Smith, C., 2001. The duration of bank relationships. J. Financ. Econ. 61, 449–475. Petersen, M., Rajan, R., 1994. The benefits of lending relationships: evidence from small business data. J. Financ. 49, 3–37. Rajan, R., Winton, A., 1995. Covenants and collateral as incentives to monitor. J. Financ. 50, 1113–1146. Ramakrishnan, R., Thakor, A., 1984. Information reliability and a theory of financial intermediation. Rev. Econ. Stud. 51, 415–432. Schenone, C., 2004. The effect of banking relationships on the firm’s IPO underpricing. J. Financ. 59, 2903–2958. Shen, C.H., Wang, C.A., 2005. Does bank relationship matter for a firm’s investment and financial constraints? The case of Taiwan. Pacific-Basin Financ. J. 13, 163–184.
Contents lists available at ScienceDirect
Research in International Business and Finance j o ur na l ho me pa ge : w w w . e l s e v i e r . c o m / l o c a t e / r i b a f
The effects of bank relationships on firm private debt restructuring: Evidence from an emerging market Jiang-Chuan Huang a,∗, Chin-Sheng Huang b a
Department of Finance, Transworld University, 1221, Jen Nang Rd., Douliu, Yunlin 640, Taiwan, ROC Department of Finance, National Yunlin University of Science & Technology, Room MB401, No. 123, University Road, Sec.3, Douliu, Yunlin 64002, Taiwan, ROC
b
a r t i c l e
i n f o
Article history: Received 30 November 2008 Received in revised form 8 September 2010 Accepted 9 September 2010 Available online 17 September 2010 JEL classification: G21 G33 G34 Keywords: Bank relationship Debt restructuring Private renegotiation Financial distress
a b s t r a c t Our paper seeks to examine the direct benefit of bank relationships for a distressed borrower by assessing its influence on the success of firm private debt restructuring. We find that a distressed firm with a stronger bank relationship has a greater probability to successfully restructure its debt through private renegotiation. Accordingly, an analysis of credit rating recovery provides complementary evidence on the factors of successful debt restructuring. A duration analysis of the length of time needed for a debt restructuring to be completed is fully consistent with our documented results. We conclude that in a bank dominated financial system like Taiwan’s where firms are heavily bank-dependent, the bank–firm relationship is of crucial importance to the success of financially distressed firms in private debt restructuring. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Extant literature on financial intermediation (see, e.g., Diamond, 1984; Ramakrishnan and Thakor, 1984) emphasizes the role of banks in generating information, for instance, through screening (Diamond, 1991) and monitoring (Rajan and Winton, 1995). This access to information is especially relevant if the borrowing firm is in financial distress. A bank’s evaluation of the borrowers based on their inside information affects the bank’s decision to renegotiate the debt or force a firm into
∗ Corresponding author. Tel.: +886 5 534201#5223; fax: +886 5 5312079. E-mail addresses: [email protected] (J.-C. Huang), [email protected] (C.-S. Huang). 0275-5319/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.ribaf.2010.09.001
114
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
bankruptcy (Chemmanur and Fulghieri, 1994). While the borrowing firm encounters financial distress, previous lending relationships with the distressed firm create a significant negative effect on the lending bank (Dahiya et al., 2003). Consequently, when the borrowing firm is in financial distress, the lending banks have two options. First, prudent banking norms immediately require the firm to repay the debt regardless of the possibility of its recovery (i.e., an instant termination of bank relationships). Second, banks are insiders, with significant information advantages. They decide to extensively involve in the firm’s private renegotiation and debt restructuring to recover losses (i.e., a continuation of bank relationships). Therefore, bank relationships may or may not benefit distressed corporate borrowers encountering financial distress. Relatively few studies in emerging economies investigate this issue. Thus, this study examines how bank relationships affect the probability of successful private debt restructuring for financially distressed borrowers. Recent research has shown how bank relationships affect private debt restructuring of financially distressed firms. Brunner and Krahnen (2008) used the number of banks as a proxy for bank relationships and observed an extensive involvement of banks in their borrowers’ debt restructuring and private workout activities. Brunner and Krahnen found that the probability of recovery from a distressed situation is negatively related to the number of banks. Couwenberg and Jong (2006) used bank debt as a proxy for the effect of bank relationships. Using this proxy, they studied the private restructuring processes in Dutch distressed firms. They found that bank debt has a significantly positive effect on the likelihood of restructuring success. While these studies show that bank relationships increase the value of distressed firms, bank relationship measures are susceptible to potential weakness, such as the degree of bank relationships measured by a single proxy. Indeed, if a firm requires fewer external funds, it would finance fewer loans with a smaller number of banks. In this situation, there is not a close or strong relationship between bank and firm. Moreover, the Taiwan financial structure is typically a bank-based financial system in which firms tend to be heavily dependent upon bank loans.1 It seems inappropriate to employ a single proxy, the size of the bank debt ratio, for bank relationships because most firms have high bank debt ratios. Hence, in contrast to the previous literature which used only a single proxy for bank relationships, the conclusions of the present research may be robust to the different bank relationship measures. The concept of a “bank relationship” is quite elusive in banking theory. There is no uniformly accepted methodology for measuring the presence and strength of bank relationships (Bharath et al., 2007). If the precise point of the start of a bank relationship is available, researchers often use the length of a relationship as a proxy for its strength (see, e.g. Petersen and Rajan, 1994; Berger and Udell, 1995; Elsas and Krahnen, 1998). In cases in which this information is not available, the existence of a prior bank relationship is used as a proxy (see, for example, Dahiya et al., 2003; Schenone, 2004; Bharath et al., 2007). Not having the precise records of bank relationships in the Taiwan banking industry, this present study adopts the existence of a prior bank relationship as a proxy for the presence of bank relationships. Moreover, this study looks at the size of bank relationships and the number of bank relationships by searching the borrower’s previous borrowing records. In short, this study simultaneously discusses three measures of bank relationships to examine the effect of the success of private debt restructuring for financially distressed firms. These measures include the existence, size, and number of bank relationships. The early empirical studies mainly have been concerned with the argument of formal bankruptcy procedures and have paid relatively less attention to out-of-court debt restructuring. Jensen (1989) is one of a few works that advocated that private contractual arrangements for resolving default represent a viable and less costly alternative to the legal remedies provided by Chapter 11. Gilson et al. (1990) further examined the determinants of 169 financially distressed firms’ choice between formal bankruptcy and out-of-court restructuring. Gilson et al. found that about half (80 firms) successfully restructured their debt through out-of-court renegotiations and these firms have more intangible assets, a larger percentage of debt owed to banks, and fewer lenders. 1 Shen and Wang (2005) reported that the external funding sources of Taiwanese firms account for 60% of the total funds. Among the entirely external funds, the largest shares are loans from financial institutions, particularly from commercial banks. Hence, bank relationships are of crucial importance for Taiwan’s firms. Suggestively, the financial structure in Taiwan is much like a bank-based financial market.
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
115
Moreover, the financial literature has identified firm size (Couwenberg and Jong, 2006), firm age (Demine and Carvalho, 2006), leverage (Brunner and Krahnen, 2008; Couwenberg and Jong, 2006), and return on assets (Claessens et al., 2003; Couwenberg and Jong, 2006; Franks and Sussman, 2005) as the critical factors and investigated the effects of firm characteristics upon the probability of firm survival. However, regarding the possible impacts of debt structure on the success of firms’ private restructuring, the extant literature produces mixed results. Financial distress is more likely to be resolved through private renegotiation when both relatively less debt is owed to trade creditors (Gilson et al., 1990) and more is owed to bank lenders (Couwenberg and Jong, 2006; Gilson et al., 1990). Franks and Sussman (2005) showed that the number and value of collateral over bank debt do not significantly affect the probability of firm survival. Consequently, this study also investigates the impacts of firm characteristics and debt structure upon the success of private debt restructuring. The aim of this study is to investigate the determinants of successful private debt restructuring for financially distressed firms using Taiwan data. In particular, the primary goal of the present research is to examine the effects of bank relationships upon the success of private debt restructuring of distressed firms. This study hypothesizes that bank relationships significantly increase the success of private debt restructuring for financially distressed firms. This hypothesis predicts that stronger bank relationships increase the probability that a distressed firm can successfully restructure its debt through private renegotiation. Debt restructuring seldom begins or ends with a formal public announcement, and outsiders do not generally have a clear picture of a firm’s private debt restructuring plan. This makes it difficult for researchers to collect data. This study presents a well-defined beginning and ending date of firm debt restructurings to identify the success or failure of a distressed firm’s private debt restructurings without depending on a bank’s internal information using a credit rating index. Consequently, this study contributes to the firm’s private debt restructuring research by providing a helpful methodology for the relevant studies in the financial market where few data are publicly available. The remainder of the paper is organized as follows. Section 2 describes the data and sample selection. The methodology and empirical results are presented in Section 3. The final section contains the concluding remarks. 2. Data and sample selection The sample in this study consists of companies listed in Taiwan that encountered financial distress from 1995 to 2003. These companies either successfully recovered or failed in private debt restructuring. In addition, general characteristics of the firms, the firm’s income statement and balance sheet data, and an assessment of the corporate credit rating are retrieved from the Taiwan Economic Journal (TEJ) database. This study proposes an alternative procedure to identity the duration of private debt restructuring. A distress event is defined in this research as the point at which the Taiwan Corporate Credit Risk Index (TCRI)2 assigns a firm a distress rating of 9 or 10. Likewise, this study shows that firm’s private debt restructuring tends to be as successful when a distressed firm’s TCRI rating improves from a distressed rating of 9 or 10 to a rating of 6 or better. This proposed empirical method will significantly mitigate the data impediments confined in the research field of debt restructuring among distressed firms. The sample in this study includes 302 firms that either was distressed or had defaulted on their obligations.3 Among these selected firms, 10 firms did not have enough financial data4 to trace their
2 The TCRI evaluation system was developed by the TEJ in August, 1991. The standard methodology of the rating relies on a scoring system according to 10 financial criteria, including profitability, safety, activity and size on corporate performance and prospects, and a linear weighting system with the fixed weighting factors. TCRI employs a 1–10 rating scale (best to worst), in which notches 1–4 are categorized as high investment grade, notches 5–6 are categorized as medium investment grade, notches 7–9 represents speculative grade, and the notch 10 is reserved for a default case (extremely high risk). 3 In the category of firms with multiple defaults during the 1995–2003 period, we have only retained the first default case in order to avoid several defaults linked to the same firm which might have biased the econometric analysis. 4 Some of the firms merged or were taken over the period 1995–2003. Thus, those firms did not have available financial data anymore.
116
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
debt restructurings, and 15 firms were eliminated due to judicial restructurings. Ultimately, this study uses a minimum total bank debt of 15 million (NTD) as the selection criteria to ensure accurate information from previous bank relationships. The sample excluded 5 firms with small bank debt, yielding 272 distressed or default firms of which 63 (23.16%) restructured successfully and 209 (76.84%) were not successful during the period, 1995–2006. 3. Empirical results 3.1. Descriptive statistics of the sample Table 1 presents the descriptive statistics of the sample. Panel A of Table 1 is the distribution of the sample by year of debt restructurings. In 272 financially distressed firms, over half (i.e., 51.84%) are clustered in the years 1998–2000 due to the timing of the domestic financial crisis (from July 1998 to January 1999) and the economic recession (from July 2000 to September 2001). Additionally, the average length of 63 successfully restructured firms is 43.22 months; in particular, the first 2 years (1995–1996) show higher average length of successful restructurings than in other periods. The restructuring firms are categorized into 16 industries based on the security code of the Taiwan Security Exchange Council (TSEC) as reported in Panel B. Particularly, the Electronics industry (34.19% and 57.14%) displays the highest percentage of overall distressed firms and successful firms, and the shortest average length of successful restructuring (34.81 months) among all the industries; the average assets (NTD 3353 million) and the average age (12.31 years) are less than the average values of overall successful firms. The descriptive statistics of the firm characteristics are presented in Panels C of Table 1. The firm size (as measured by the annual sales and the total assets), total liabilities, and return on assets (ROA) have consistently displayed considerable volatility. Moreover, the average length of successful debt restructuring period is 43.22 months (median is 36 months) which is apparently longer than those reported for US firms of 15.4 months (Gilson et al., 1990), small UK firms of 9.2 months (Franks and Sussman, 2005), and Dutch firms of 24 months (Couwenberg and Jong, 2006). One main reason is that a definition of a successful debt restructuring in this study requires firms succeeding in credit ratings upgrading to indeed lower their risk category (i.e., TCRI rating 6 or better). The lending bank information and firm’s debt structure are presented in Panel D of Table 1. The average number of banks amounts to 10.42 (with a median of 8) ranging from single banking to 41 banks. This figure is apparently higher than those reported for the Taiwan non-financially distressed firms of 9.47 banks by Fok et al. (2004) and 8.355 banks by Shen and Wang (2005). The evidence agrees with the presumed argument that the distressed firms usually maintain more banking relationships to finance extra external funds in the cases of insufficient cash flows. The size of bank relationships is a proxy as a measure for the exposure of the main bank, i.e., total amount of loans to the distressed borrower by the main bank in the last 5 years divided by total amount of loans from all lending banks. The average bank relationship size is 0.38, with the median 0.33, ranging from 0.02 to 0.97. 3.2. Univariate analysis The sample of 272 debt-restructuring attempts is further partitioned into two subsamples of successful firms (63 firms) and unsuccessful firms (209 firms). Table 2 presents the results of differentiate analysis of firm characteristics, bank characteristics, and debt structure for these two subsamples. In successful firms, the selected firm characteristics of average total assets, firm age, and leverage ratio are significantly (at the 1% level) smaller than those in unsuccessful firms are. For the ROA, however, the difference between in these two types of firms is insignificant. The proxies of bank relationships, including the existence of prior bank relationships, the size of bank relationships, and the number of banks, are distinct between the two subsamples (at the 1% level). Successful firms have a significantly greater average account-payable ratio (at the 5% level), but the bank debt ratio and the secured debt ratio exhibit insignificant differences between the two groups. The government rescue dummy shows no significant difference between two groups. In the group of successful firms, the industry dummy and real GDP growth rate are significantly greater than those in the unsuccessful group are (at the 1%
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
117
Table 1 Descriptive statistics of the sample. Panel A: Distribution of the sample by year of debt restructuring Year
Total firms (and %)
1995 1996 1997 1998 1999 2000 2001 2002 2003 Total
23(8.46) 16(5.88) 15(5.51) 60(22.06) 37(13.60) 44(16.18) 29(10.66) 30(11.03) 18(6.22) 272(100.00)
Successful firms (and %) 11(17.46) 8(12.70) 9(14.29) 6(9.52) 8(12.70) 10(15.87) 5(7.94) 3(4.76) 3(4.76) 63(100.00)
Percentage of successful firms (%)
Average length for successful restructuring (month)
47.83 50.00 60.00 10.00 21.62 22.73 17.24 10.00 16.67 23.16
64.00 60.38 40.67 29.50 34.88 38.10 25.20 39.00 30.00 43.22
Panel B: Distribution of the sample by industry of debt restructuring Successful debt restructuring a
Code
Industry
12 13 14 15 16 17 18
Food Plastics Textiles Electrical Machinery Wire and Cable Chemical Ceramics and glass products Pulp and paper Steel Rubber Electronics Construction Transportation Tourism Department stores Other Over sample
19 20 21 23, 24b 25 26 27 29 99
Total firms (and %)
Successful firms (and %)
Successful percentage (%)
Average length (month)
Average asset (NTD mil.)
Average age (year)
15(5.51) 6(2.21) 21(7.72) 13(4.78) 5(1.84) 9(3.31) 5(1.84)
2(3.17) 0(0.00) 0(0.00) 2(3.17) 0(0.00) 6(9.52) 0(0.00)
13.33 N/A N/A 15.38 N/A 66.67 N/A
72.00 N/A N/A 39.00 N/A 59.50 N/A
7751 N/A N/A 2188 N/A 5508 N/A
27.50 N/A N/A 21.50 N/A 30.17 N/A
1(0.37) 25(9.19) 1(0.37) 93(34.19) 48(17.65) 9(3.31) 2(0.74) 4(1.47) 15(5.51) 272(100.00)
0(0.00) 6(9.52) 0(0.00) 36(57.14) 7(11.11) 2(3.17) 0(0.00) 1(1.59) 1(1.59) 63(100.00)
N/A 24.00 N/A 38.71 14.58 22.22 N/A 25.00 6.67 23.16
N/A 48.50 N/A 34.81 53.14 66.00 N/A 48.00 48.00 43.22
N/A 17,868 N/A 3353 7261 1704 N/A 1342 1823 5369
N/A 18.17 N/A 12.31 10.57 15.00 N/A 8.00 3.00 15.02
Panel C: Firm characteristics and restructuring information (N = 272)
Total sales (NTD in million) Total assets (NTD in million) Total liabilities (NTD in million) Return on assets (%) Firm’s age (year) Length for successful restructuring (month)
Mean
Median
2301 7405 4463 0.17 20 43.22
1058 3244 1803 1.79 18 36
Standard deviation 3752 10,224 6441 8.55 11.80 32.52
Min.
Max.
0.54 114 33 -33.38 1 6
37,867 52,631 37,081 30.42 51 108
118
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
Table 1 (Continued) Panel D: Lending bank information and firm’s debt structure (N = 272)
Number of banks Size of bank relationships Total account-payable (NTD in million) Total bank debt (NTD in million) Total secured debt (NTD in million)
Mean
Median
Standard deviation
10.42 0.38 412 2534 1902
8 0.33 176 982 682
7 0.21 711 3903 3201
Min.
Max.
1 0.02 1.46 16 0
41 0.97 6424 22,820 21,435
a The restructuring firms are categorized into 16 industries based on the security code of the Taiwan Security Exchange Council (TSEC). In Panel B, the industry code is represented by the first two digits of the security code. b The Electronics industry is divided into two industry sectors. One industry sector includes semiconductor, computer and peripheral equipment, electronic parts/components, and other electronics (the first two digits of the security code is ‘23’). The other industry sector contains communications and internet, electronic products distribution, and information service (the first two digits of the security code is ‘24’).
Table 2 Differences in mean (median) of firm characteristics, bank characteristics and debt structure. Mean (Median)
Total assets (NTD in million) Government rescuea Industrya Real GDP growth rate Existence of bank relationshipsa Size of bank relationships Number of banks Firm age (year) Leverage ratio Return on assets (%) Account-payable ratio Bank debt ratio Secured debt ratio Samples a ** ***
Differences of successful and unsuccessful firms (p-value)
Successful firms
Unsuccessful firms
Differences
t-test
Wilcoxon test (Rank-sum)
5369(1522) 0.36(0.00) 0.68(1.00) 5.11(5.77) 0.76(1.00) 0.50(0.53) 8.33(7.00) 15.02(11.00) 0.60(0.61) 0.11(1.26) 0.15(0.12) 0.63(0.63) 0.70(0.78) 63
8018(4174) 0.42(0.00) 0.46(0.00) 4.25(4.55) 0.53(1.00) 0.33(0.31) 11.05(9.00) 21.51(21.00) 0.66(0.66) 0.18(1.84) 0.11(0.10) 0.59(0.60) 0.74(0.81) 209
−2649(−2652) −0.06(0.00) 0.22(1.00) 0.86(1.22) 0.23(0.00) 0.17(0.22) −2.72(−2.00) −6.49(−10.00) −0.06(−0.05) −0.07(−0.58) 0.04(0.02) 0.04(0.03) −0.04(−0.03)
0.01*** 0.43 0.00*** 0.01** 0.00*** 0.00*** 0.00*** 0.00*** 0.00*** 0.95 0.02** 0.31 0.80
0.00*** 0.38 0.00*** 0.00*** 0.00*** 0.00*** 0.00*** 0.00*** 0.00*** 0.94 0.05** 0.81 0.77
Government rescue, Industry, and existence of bank relationships are represented by dummies. Significant at the 5%. Significant at the 1%.
level and 5% level, respectively). This study also performed a nonparametric Wilcoxon Rank-sum test for median differences between the two subsamples. In virtually all cases, the direction and the level of statistical significance of the results of this test are consistent with those obtained from the t-tests. 3.3. Pearson correlation analysis The correlations between the influencing factors of the regression analysis of the following section are shown in Table 3. Particularly, there is a strong positive correlation of 0.674 between firm size and the number of banks because relatively larger firms typically use bank loans from multiple bank lenders. This result is affirmative with Houston and James (1996) that multiple-bank firms tend to be larger than single-bank firms and Shen and Wang (2005) that the optimal number of bank relationships is not constant but possibly increases with the size of a firm. In addition, two potentially influencing factors for the strength of the bank relationships, the number of banks versus the size of bank relationships, are negatively correlated (−0.549) because for a specific firm, the loan shares of a main lending bank decrease with the number of lending banks. All other variables show moderately Pearson correlations ranging from −0.453 to 0.428.
1. Dependent variable 2. Firm size 3. Government rescue 4. Industry 5. Real GDP growth rate 6. Existence of bank relationships 7. Size of bank relationships 8. Number of banks 9. Firm age 10.Leverage ratio 11.Return on assets 12.Account-payable ratio 13.Bank debt ratio 14.Secured debt ratio
1
2
3
4
5
6
7
8
9
10
11
12
13
14
1.000 −0.228 −0.048 0.184 0.149 0.201 0.349 −0.329 −0.284 −0.206 0.003 0.150 0.062 −0.016
1.000 0.175 −0.314 −0.006 −0.016 −0.431 0.674 0.428 0.191 0.178 −0.297 −0.145 −0.070
1.000 −0.167 0.331 0.038 −0.125 0.158 0.176 −0.080 0.062 −0.069 −0.028 −0.080
1.000 −0.090 −0.005 0.141 −0.310 −0.453 −0.178 −0.085 0.248 0.027 0.132
1.000 −0.105 0.045 −0.065 −0.042 0.004 0.112 −0.008 −0.075 −0.125
1.000 0.304 −0.068 0.031 −0.012 −0.057 −0.018 0.157 0.114
1.000 −0.549 −0.301 −0.186 −0.136 0.081 0.019 0.122
1.000 0.369 0.288 0.231 −0.208 −0.002 −0.120
1.000 0.149 0.257 −0.032 −0.233 −0.057
1.000 0.071 −0.052 −0.024 −0.135
1.000 0.098 −0.114 0.060
1.000 −0.205 −0.072
1.000 0.039
1.000
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
Table 3 Description of the correlations between the factors of potential influences (N = 272).
119
120
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
3.4. Determinants of successful debt restructurings for financially distressed firms Three distinct estimation methodologies are employed in analyzing the determinants of restructured success. First, this study uses a logistic regression model to illustrate how potent influencing variables affect the probability of restructured success. Next, an ordered-probit model is employed to examine the effects of potential factors upon the probability of the corporate rating improvement. Furthermore, a duration model is used to ascertain how variables might carry weight on the time length of successfully completing private debt restructuring. 3.4.1. Prediction of successful restructuring In the analysis of Section 3.2, this study has documented several differences between successful and unsuccessful firms in univariate comparisons. Table 4 further reports the empirical results of logistic regressions. The logistic regression model contains two explanatory variables: control variables and theory-based variables. The control variables consist of firm size (log of total assets), a government rescue dummy5 , an industry dummy6 , and real GDP growth rate (counted at the onset of the firm’s debt-restructuring year). The theory-based variables in the regression models come from bank relationships, firm characteristics, and debt structure. Specifically, this study constructs three alternative bank relationship measures by analyzing a distressed firm’s previous borrowing records. Specifically, this study looks back over a period of 5 years for each loan by a typically distressed firm.7 The first bank relationship measure is specified as a dummy variable taking one if the main bank8 of the distressed firm had served as a main bank in lending (making previous loans) to the distressed borrower before its default. The second bank relationship variable is the exposure of the main bank to the distressed borrower which captures the size of past bank relationships. The last bank relationship variable is the number of banks which represents the strength of bank relationships. This study predicts that the existence of prior relationships and size of bank relationships have positive effects on the probability of restructuring success. However, the number of banks has the opposite effect on the restructuring success. The firm characteristics of this analysis include firm age (log of the firm’s age), leverage ratio (total liabilities over total assets), and ROA (in the year prior to the onset debt restructuring). The firm’s debt structure is proxied by three variables: relative size of account-payable debt (account-payable debt over total liabilities), bank debt (bank debt over total liabilities) and secured debt (secured debt over bank debt). All the debt structure variables are computed at the onset of the firm’s debt-restructuring year. As the results in Panel A of Table 4 indicate, regressions (1)–(6) show that all of the coefficients of the three bank relationship variables are highly significant and with correct signs. Since there is a high correlation (−0.549) between the number of banks and the size of bank relationships, this study built up a principal component, named intensity of bank relationships, in the regressions (6)–(9).9 The negative coefficient of the intensity of bank relationships in regressions (6)–(9) implies that the negative
5 Inoue et al. (2008) documented that out-of-court restructuring has a positive effect on the market value of the distressed firms only when monitored by bank supervisors. Claessens et al. (2003) ascribed the high likelihood of out-of-court renegotiations in the East Asian countries to poorer credit rights and an inefficient judicial system in the region. In the Asian financial crisis, 1997–1998, the Ministry of Finance in Taiwan set up a “Committee for Composition Corporate Operational Funds” for loan relief during the period, November 1998 to December 2001. Hence, this study defines a government rescue effort dummy to examine the influence of government involvement on the success of firm private debt restructuring. 6 The industry dummy is unity if a distressed firm belongs to the “new” economy (e.g., the electronics and high-tech industries) and zero if it belongs to the “conventional” economy (e.g., the construction, paper and pulp, food, and steel industries). 7 Bharath et al. (2007) reported that they chose the 5 year window as approximately 75% of the loan facilities in their sample having maturities of less than or equal to 5 years. Hence, most of the borrowers in the studied sample would need to refinance their debt within five years. 8 This study focuses on the main bank of a particular distressed firm, as the information-intensive role that the hypothesis in this study is most appropriate for the main bank, which typically holds the largest share of a firm’s bank debt (see as, Gilson et al., 1990; Franks and Sussman, 2005; Bharath et al., 2007; Inoue et al., 2008). Thus, the responsibilities of a main bank best fit the description of a relationship lender. 9 In regressions (6)–(9), this study employs the intensity of bank relationship variable to replace the number of banks and the size of bank relationships in order to avoid the statistical multicollinearity problem.
Table 4 The determinants of successful restructuring for financially distressed firms—Logistic regression. Panel A: Bank relationships Variables
(1)
(2)
1.231 (1.452) −0.176* (0.092) −0.231 (0.223) 0.541** (0.239) 0.187*** (0.062) 0.887*** (0.244)
−1.319 (1.527) −0.095 (0.092) −0.056 (0.215) 0.571** (0.227) 0.150*** (0.058)
(3)
(4)
−0.201 (0.016) −0.158 (0.217) 0.455** (0.229) 0.150*** (0.059)
2.361*** (0.517)
(5)
−1.162 (1.524) −0.121 (0.092) −0.118 (0.214) 0.566** (0.226) 0.163*** (0.057) 0.624*** (0.237) 1.913*** (0.517)
272 0.209
272 0.191
−2.159 (0.433)
−0.207 (0.212) 0.450** (0.223) 0.163*** (0.056) 0.786*** (0.233)
−0.139 (0.209) 0.496** (0.214) 0.152*** (0.055) 0.628*** (0.229)
−0.637*** (0.154)
−0.709*** (0.161) 272 0.162
(6)
−0.854 (0.540)
272 0.232
272 0.214
−0.387*** (0.081) 272 0.244
Panel B: Firm characteristics and debt structure Variables Intercept Government rescue Industry Real GDP growth rate Existence of bank relationships Intensity of bank relationshipsa Firm age Leverage ratio Return on assets Account-payable ratio Bank debt ratio Secured debt ratio Sample size Pseudo R-sq
(7) −0.107 (0.712) −0.127 (0.211) 0.242 (0.216) 0.122** (0.053) 0.685*** (0.228) −0.296*** (0.086) −0.355** (0.151) −1.293* (0.689) 0.020* (0.012)
272 0.250
(8) −2.147*** (0.560) −0.161 (0.206) 0.429** (0.218) 0.150*** (0.055) 0.613*** (0.227) −0.382*** (0.080)
1.068 (0.885) 0.362 (0.310) −0.393 (0.348) 272 0.250
(9) 0.143 (0.806) −0.153 (0.207) 0.174 (0.221) 0.115** (0.053) 0.693*** (0.225) −0.295*** (0.085) −0.370** (0.154) −1.287* (0.664) 0.021* (0.012) 1.090 (0.872) 0.255 (0.301) −0.560 (0.342) 272 0.270
Numbers in the parentheses are standard errors. a We apply principal component analysis to retrieve a new variable of the intensity of bank relationships which is composed from the number of banks and the size of bank relationships. In regressions (6)–(9), we employ the intensity of bank relationship variable to replace the number of banks and the size of bank relationships in order to avoid the statistical multicollinearity problem. * Significant at the 10% level. **
Significant at the 5%. Significant at the 1%.
121
***
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
Intercept Firm size Government rescue Industry Real GDP growth rate Existence of bank relationships Size of bank relationships Number of banks Intensity of bank relationshipsa Sample size Pseudo R-sq
122
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
Table 5 The distribution of cumulative corporate rating improvement. The length of restructuring after post-distress 12-Month
24-Month
The ratings of cumulative recovery
Firms (%)
Firms (%)
0 rating 1 rating 2 ratings 3 ratings 4 ratings (and more) Samples (N)
88.56 3.69 4.43 2.58 0.74 272
72.48 10.85 8.14 6.59 1.94 258
36-Month Firms (%) 62.14 13.58 9.88 10.29 4.11 243
48-Month Firms (%) 56.05 13.90 10.31 12.56 7.18 223
effects of number of banks dominates the effect of size of bank relationships. Overall, the empirical results indicate that the existence of prior bank relationships, the larger size of bank relationships and stronger degree of bank relationships (less number of banks) significantly increase the probability of successful private debt-restructuring. Moreover, it is clear that a strong bank relationship alleviates information asymmetry of banks and benefits the private debt restructuring of distressed firms. Among the control variables, the Panel A reports that the firm size is negatively related to the success of firm restructuring only at marginal significance.10 This result may be imputed to the structure of Taiwanese industry where a large proportion of distressed firms belong to the “new economy” as reported previously in Panel B of Table 1 and in Table 2. Due to the absence of a consummate debt renegotiation mechanism in most emerging markets, the government rescue dummy is insignificant. Consistently with extant literature, the industry dummy and the growth rate of real GDP are statistically significant and positively related to the likelihood of successful private debt restructuring. As the results in Panel B of Table 4 show, the coefficient of the firm age is significantly negative at least at the 5% level which indicates that the younger distressed firms are more likely to be successful in restructuring. The negative effect of firm age in this study may be ascribed to a high proportion of distressed firms in the “new economy.” Incidentally, firm age shows a high negative correlation (−0.453) with the industry dummy. In regression (7), the industry dummy turns insignificant which may indicate that the firm age effect overshadow the industry effect on the successful restructuring. Furthermore, the coefficient of leverage is significantly negative which suggests that a higher leverage ratio before restructuring is less likely to be successful in restructuring. The significantly positive results (at the 10% level) for ROA show that firms with a solid operating performance before restructuring have a higher incidence of surviving. In regression (8), all measures of debt structure yield insignificant coefficients although having the correct signs. Ultimately, regression (9) contains the control variables and theoretical explanatory variables. The results show that the coefficients of overall variables have similar effects as those illustrated previously in this section. Overall, this empirical evidence indicates three bank relationship proxies capture more information concerning the efficiency of firm private debt-restructuring than those investigated in single proxy setting. Two arguments are worth noting here. First, the information advantage provided by prior bank relationships allows banks to accurately discriminate between viable and non-viable firms and to provide greater assistance to potentially successful firms. Second, the severity of the holdout problem depends on the number of lending banks participating in the firm’s private debt-restructuring. 3.4.2. Corporate rating recovery analysis In this section, this study focuses upon determinants of corporate credit rating improvement in the sample of successful debt restructuring and delivers the relevant evidence for the previous analysis. Table 5 documents the distributions of cumulative corporate rating recovery in terms of restructuring durations for 4 years annually. It is clear that the cumulative improvements increase with the lengths
10 The number of banks has a correlation of 0.674 with firm size. In unreported additional regressions we included these two variables and found that the number of banks turns insignificant, while the firm size is still not significant and has a positive effect. Therefore, we separated these two variables and report the results in Table 4, Panel A.
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
123
of restructuring. Specifically, the proportions of cumulative recovery of the 4 ratings (or more) elevate from 0.74% for the horizon of 12-month, up to 1.94%, 4.11%, and reach at 7.18% for the horizons of 24-month, 36-month, and 48-month, respectively. The same phenomenon occurs for the cumulative improvement of 3 ratings which surges to 2.58%, 6.59%, 10.29%, and 12.56%, respectively. Following the same line of reasoning, the proportion of no improvement has declined from 88.56% to 72.48%, 62.14%, and 56.05% for the horizons of the 4 years annually. This study further employs an order-probit regression for ordered response data to measure the factors determining cumulative rating improvement of distressed firms after post-distress. The model consists of the control variables and explanatory variables that are similar to the logistics regression employed earlier. However, the firm size and government rescue variables are eliminated due to their insignificant effect on firm restructuring by the previous analysis. Moreover, the real GDP growth rate is also excluded because the effects of macro-economic conditions are difficult to estimate year-by-year. Table 6 documents the regression results for each observation window after post-distress. First, the empirical evidence illustrates that the existence of prior relationships (and the intensity of bank relationships) has a statistically significant positive (negative) impact on the firm recovery at all horizons. This confirms the sample univariate results and the logistic regression results reported earlier. Second, as expected, the firm age, ROA and leverage variables all show statistically significant effects on the firm recovery across all horizons and in agreement with those findings in previous analysis. Third, an alternative finding is that the account-payable also indicates a significantly positive effect on the firm’s recovery during all horizons. These results suggest that trade creditors might provide additional financing to restructured firms due to their information asymmetry (Couwenberg and Jong, 2006; Franks and Sussman, 2005). 3.4.3. Duration analysis This study takes a further step to investigate the determinants of the duration of firm debt restructuring. Since the data involves right censored observations that bias the OLS estimates, this study employs the duration models to deal with right censoring (Ongena and Smith, 2001). The dependent variable is the duration of a distress episode (in months) that ends with a distress event of the rating improving to 6 or better. This variable makes it possible to analyze the time required for successfully completing private debt restructuring. The independent variables coincide with those of the preceding logistic regressions of Table 4. The coefficients are expected to have opposite signs compared to the logistic regressions since the dependent variable (i.e., the log of the length of restructuring completion period by months) is an inverse measure of restructuring success: the higher restructuring performance the shorter time for the firm to recover. The specifications of the duration model are
Table 6 The determinants of corporate rating recovery—Ordered-probit regression. Variables
12-Month
24-Month
36-Month
48-Month
Existence of bank relationships Intensity of bank relationshipsa Industry Firm age Leverage ratio Return on assets Account-payable ratio Bank debt ratio Secured debt ratio Sample size Pseudo R-sq
0.585** (0.302) −0.249** (0.123) 0.085 (0.286) −0.457 (0.309) −2.697*** (0.847) 0.104*** (0.025) 2.632*** (0.969) 0.576 (0.773) −0.859 (0.439) 272 0.346
0.406** (0.197) −0.282** (0.141) −0.055 (0.217) −0.383* (0.211) −2.868*** (0.612) 0.057*** (0.012) 3.195*** (0.589) 0.589 (0.613) −0.195 (0.376) 258 0.300
0.571*** (0.184) −0.215*** (0.082) −0.295 (0.204) −0.479** (0.193) −1.654** (0.525) 0.061*** (0.011) 1.803*** (0.677) 0.492 (0.520) −0.307 (0.304) 243 0.270
0.624*** (0.184) −0.203*** (0.073) 0.010 (0.204) −0.500*** (0.195) −1.932*** (0.503) 0.058*** (0.011) 1.520** (0.689) 0.195 (0.545) −0.446 (0.301) 223 0.285
Numbers in parentheses are standard errors. a We apply principal component analysis to retrieve a new variable of the intensity of bank relationships which is composed from the number of banks and the size of bank relationships. * Significant at the 10% level. ** Significant at the 5%. *** Significant at the 1%.
124
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
Table 7 The determinants of the duration of firm restructuring—Weibull distribution. Variables Intercept Firm size Government rescue Industry Real GDP growth rate Existence of bank relationships Size of bank relationships Number of banks Intensity of bank relationshipsa Firm age Leverage ratio Return on assets Account-payable ratio Bank debt ratio Secured debt ratio Sigma () Sample size
(1) 1.620*** (0.453) 0.006 (0.027) 0.040 (0.057 −0.116* (0.060) −0.020 (0.014) −0.202*** (0.063) −0.463*** (0.131)
(2)
(3)
1.192*** (0.221)
1.550*** (0.212)
0.033 (0.057) −0.067 (0.061) −0.014 (0.014) −0.249*** (0.064)
0.024 (0.055) −0.095 (0.059) 0.004 (0.014) −0.207*** (0.063)
0.125*** (0.048) 0.096** (0.040) 0.473** (0.188) −0.003 (0.003) −0.515** (0.220) −0.072 (0.084) 0.132 (0.092) 1.651*** (0.112) 272
0.112*** (0.042) 0.404** (0.204) −0.003 (0.003) −0.437** (0.204) −0.045 (0.084) 0.161* (0.094) 1.625*** (0.112) 272
0.088*** (0.024) 0.090** (0.041) 0.398** (0.186) −0.004 (0.003) −0.448** (0.203) −0.067 (0.077) 0.148 (0.090) 1.655*** (0.212) 272
Numbers in parentheses are standard errors. a We apply the principal component analysis to retrieve a new variable of the intensity of bank relationships which is composed from the number of banks and the size of bank relationships. * Significant at the 10% level. ** Significant at the 5%. *** Significant at the 1%.
conducted by a parametric survival model under the Weibull distribution (same as Ongena and Smith, 2001; Brunner and Krahnen, 2008). The results drawn from Table 7 are in line with those of the logistic estimations. The evidence shows that the existence of prior bank relationships and size of bank relationships reduce the time needed to achieve success. However, an increasing number of banks and an increasing intensity of bank relationships significantly lengthen the time required to terminate a restructuring completion. Moreover, the severity of the distress shock (leverage ratio) extends the time of the distress episode and delays an eventual completion as indicated in regressions (1)–(3). However, ROA and the real GDP growth rate turn out to be insignificant in this duration analysis. Overall, the duration analysis supports our earlier findings that bank relationships significantly account for the success of restructuring and the time needed for recovery in private debt restructuring pursued by distressed firms. 4. Conclusions This paper investigates the determinants of successful private debt-restructuring pursued by financially distressed Taiwanese firms during the period 1995–2003. Several salient features in this study will shed light on the problem of bank relationships and private debt restructuring. Firstly, this study presents a credit rating index to identify the success or failure of firm private debt restructuring. The proposed methodology provides a potential method for future research in the field of firm private debt restructuring, which frequently is confronted with data impediments. Because bank loans are private instruments, the data relevant to debt restructuring in emerging markets is rarely available to the public. Secondly, in terms of contributions to the literature, this study is one of only a few attempts to carefully document the effects of bank relationships on firm private debt restructuring. Specifically, this study employs three proxies for completely measuring the degree of bank relationships and identifies the various effects of these proxies on the success of private debt restructuring. Results indicate that bank relationships significantly affect the success of firm private debt restructuring. A financially distressed firm with a stronger bank relationship is more likely to successfully restructure its debt. Therefore, the empirical results in this study fully support the hypotheses defined by this study. Empir-
J.-C. Huang, C.-S. Huang / Research in International Business and Finance 25 (2011) 113–125
125
ical results also show that firms with younger age, higher profitability, less severity of distress, more account-payable debt and better macroeconomic conditions exhibit a significantly higher likelihood of successful restructuring. Finally, two supplementary pieces of evidence drawn from an analysis of the credit rating recovery and a duration analysis of the length of time needed for a private debt restructuring simultaneously confirm the findings reported in the logistic regression analysis. In brief, bank–firm relationships are crucial to the success of private debt restructuring undertaken by financially distressed firms, particularly in a bank dominated financial system where firms are heavily bank-dependent as evidenced in Taiwan. References Berger, A.N., Udell, G.F., 1995. Relationships lending and lines of credit in small firm finance. J. Bus. 68, 351–381. Bharath, S., Dahiya, S., Saunders, A., Srinivasan, A., 2007. So what do I get? The bank’s view of lending relationships. J. Financ. Econ. 85, 368–419. Brunner, A., Krahnen, J.P., 2008. Multiple lenders and corporate distress: evidence on debt restructuring. Rev. Econ. Stud. 75, 415–442. Chemmanur, T.J., Fulghieri, P., 1994. Reputation, renegotiation, and the choice between bank loans and publicly traded debt. Rev. Financ. Stud. 7, 475–506. Claessens, S., Djankov, S., Klapper, L., 2003. Resolution of corporate distress in East Asia. J. Empirical Financ. 10, 199–216. Couwenberg, O., Jong, A., 2006. It takes two to tango: an empirical tale of distressed firms and assisting banks. Int. Rev. Law Econ. 26, 429–454. Dahiya, S., Saunders, A., Srinivasan, A., 2003. Financial distress and bank lending relationships. J. Financ. 58, 375–399. Demine, J., Carvalho, C.N., 2006. Bank loan losses-given-default: a case study. J. Bank. Financ. 30, 1219–1243. Diamond, D., 1984. Financial intermediation and delegated monitoring. Rev. Econ. Stud. 51, 393–414. Diamond, D., 1991. Monitoring and reputation: the choice between bank loans and directly placed debt. J. Polit. Econ. 99, 689–721. Elsas, R., Krahnen, J.P., 1998. Is relationship lending special? Evidence from credit-file data in Germany. J. Bank. Financ. 22, 1283–1316. Fok, R.C.W., Chang, Y.C., Lee, W.T., 2004. Bank relationships and their effects on firm performance around the Asian financial crisis: evidence from Taiwan. Financ. Manage. (summer), 89–112. Franks, J., Sussman, O., 2005. Financial distress and bank restructuring of small to medium size UK companies. Rev. Financ. 9, 65–96. Gilson, S.C., John, K., Lang, L.H.P., 1990. Troubled debt restructurings: an empirical study of private reorganization of firms in default. J. Financ. Econ. 27, 315–354. Houston, J.F., James, C.M., 1996. Bank information monopolies and the choice between public and private debt claims. J. Financ. 51, 1863–1889. Inoue, K., Kato, H.K., Bremer, M., 2008. Corporate restructuring in Japan: who monitors the monitor? J. Bank. Financ. 32, 2628–2635. Jensen, M.C., 1989. Active investors, LOBs, and the privatization of bankruptcy. J. Appl. Corp. Financ. 2, 35–44. Ongena, S, Smith, C., 2001. The duration of bank relationships. J. Financ. Econ. 61, 449–475. Petersen, M., Rajan, R., 1994. The benefits of lending relationships: evidence from small business data. J. Financ. 49, 3–37. Rajan, R., Winton, A., 1995. Covenants and collateral as incentives to monitor. J. Financ. 50, 1113–1146. Ramakrishnan, R., Thakor, A., 1984. Information reliability and a theory of financial intermediation. Rev. Econ. Stud. 51, 415–432. Schenone, C., 2004. The effect of banking relationships on the firm’s IPO underpricing. J. Financ. 59, 2903–2958. Shen, C.H., Wang, C.A., 2005. Does bank relationship matter for a firm’s investment and financial constraints? The case of Taiwan. Pacific-Basin Financ. J. 13, 163–184.