Chapter II. The Cascade Effect in the Syndicated Loan Market
6. Conclusions
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
37
estimated coefficients of the relational distance measures is slightly smaller than that obtained when I do not control for these effects. The relative degrees of statistical significance are similar to those reported in Table 2.3, suggesting my empirical results are robust to unobservable firm-level or lender-level heteroskedasticity.
[Insert Table 2.8]
6. Conclusions
The interaction of economic agents is complicated and no single rule can capture all situations. The complicated interaction also forms complex networks. This phenomenon has inspired me to construct my theoretical and empirical work in this paper. I use two cases to model lenders’ interaction in the syndicated loan market. My purpose is to explore how informational cascade could happen in this market and how it would affect loan pricing. I also empirically test the model’s predictions.
The first case is my benchmark where I assume all potential lenders can freely share information held by them. The second one is the cascade case in which I assume each potential lender can only observe the decisions of its predecessors. If potential lenders only have imperfect signals, the actions of their predecessors are important information for evaluating a loan. My model shows that if the lead bank is rational and risk-neutral, the probability of syndication failure is always positive in the benchmark but is zero in the case of cascade. This results in lower ex ante financing cost under the cascade but the ex post interest rate will be higher. The intuition is that the lead bank will increase the interest rate to elicit a positive cascade and ensure that financing is obtained.
To empirically test the models’ predictions, physical distance and relational distance are used to proxy for segmentation of communication amongst lenders. I use average path length, clustering coefficient, and density, which are both taken from the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
38
analysis of syndication networks, to gauge the relational distance. The longer distance signifies that communication or information is more segmented amongst lenders. The results show that the physical distance does not support the predictions, but the relational distance does. I argue that the physical distance is not a good proxy due to innovations in technology, transportation, and communication. The relational distance is a good proxy because the influence of network structures on information dissemination and transmission is well-established in the literature.
As the relational distance is used as a proxy for segmented communication, the results show that the relational distance is positively correlated with loan spread and is negatively correlated with the probability of syndication failure. In addition, the higher relational distance also results in more non-price contract terms, especially the requirements for collateral and guarantees. These findings confirm my model’s predictions. The cascade effect does matter for lending conditions.
My study contributes to the literature in three aspects. First, to my best knowledge, this is the first study to explore cascade effect in the syndicated loan market. As noted in section 1, although some studies have examined the herd behavior in banks’ investment decisions, they did not focus on the syndicated loan market and did not explore how herd behavior affects loan pricing.
Second, several distance measures have been proposed and associated with economic decisions, for example, physical distance (Mian, 2006; Giannetti and Yafeh, 2012), distance of specialization (Cai et al., 2010), and cultural distance (Giannetti and Yafeh, 2012). Also, many researches have proven the existence of relationship lending (e.g., Elyasiani and Goldberg, 2004; Champagne and Kryzanowski, 2007). I extend these ideas to propose a novel proxy, relational distance. It is difficult to capture relationships amongst economic agents and to quantify the relational distance.
I overcome these obstacles by conducting a social network analysis.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
39
Finally, there have been plenty of papers that investigate the determinants of loan prices. In this paper, I analyze not only the factors that affect loan price but also those that cause syndication failures. To my best knowledge, this is also the first study to empirically test the determinants of syndication failures. It is an important issue because syndication failures are costly to borrowers and lenders and may impair investment activities. Understanding the causes of syndication failures helps to improve the success of the syndicated loan market.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
40
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
41
Figure 2.1: An example of the decision rule of the cascade case
This figure illustrates an example of the decision rule of the cascade case. The lowest interest rate for which the potential lender is willing to accept is reported in parentheses.
Figure 2.2: An example of a syndication network.
This figure uses a facility in 1995 as an example to show what the syndication network look like. Each node represents a lender and any line linking two nodes indicates that the two banks had a lead-participant relationship in the period 1990 to 1994.
First lender Second lender Third lender
Reject (3/5) ….
Accept (2/5) : positive cascade g
b
g g
b b
Accept (1/3) r = 1/2
Reject (2/3)
g b
Reject (3/4) : negative cascade
Accept (1/2)
Accept (1/2) : positive cascade
Accept (1/4) : positive cascade
‧
Table 2.1: Descriptive statistics of regression variables
Data are collected form DealScan database and for the period from January 1990 to August 2010. Only facilities which have lender number that is larger than 1 are included. Sample I is the sample used in the analysis for physical distance. Sample II is the sample used in the analysis for relational distance. Definitions and units of the variables are provided in Appendix B, Table B1.
Variable
Sample I Sample II
Obs. Mean Std. Median Obs. Mean Std. Median Panel A: Loan characteristics
Allindrawn 26,083 223.11 144.33 215 56,973 227.24 151.82 225 Amt 31,938 171.50 400.77 66 65,390 265.10 650.60 100 Maturity 27,750 46.71 147.07 46 59,528 47.92 102.60 50 Secured 11,346 0.82 0.38 1 29,080 0.78 0.42 1 Guaranteed 31,938 0.02 0.14 0 65,390 0.05 0.22 0 Covenant 31,938 0.20 0.40 0 65,390 0.33 0.47 0 Lender_num 31,938 4.25 3.76 3 65,390 6.90 7.52 4 Failed 31,938 0.006 0.08 0 65,390 0.005 0.07 0 Panel B: Borrower characteristics
Pub_unrated 24,507 0.23 0.42 0 50,592 0.20 0.40 0 Pub_rated 24,507 0.27 0.44 0 50,592 0.32 0.47 0 Sales 13,794 3,235.24 11,391.40 596.75 30,518 4,881.90 16,288.59 932.06 MB 11,359 3.01 25.94 1.87 24,742 7.38 409.59 1.97 ROA 13,794 1.16 47.54 2.91 30,511 1.59 50.64 3.09 Leverage 13,750 36.70 26.92 33.67 30,513 37.20 27.85 34.08 Panel C: Proxies for the segmentation of communication
Distance 31,938 2,570.99 2,432.64 1,580.41
Path 65,326 1.68 2.15 1.07
Cluster 46,508 0.84 0.26 0.94
Density 65,326 0.79 0.29 0.93
‧
Table 2.2: Effects of physical distance on loan contract terms and syndication failures
The dependent variable in column (1) is all-in-drawn spread. The dependent variable in column (2) is an indicator variable which takes the value 1 if a loan contract includes at least one financial covenant and 0 otherwise. The dependent variable in column (3) is an indicator variable which takes the value 1 if the loan has been secured and zero otherwise. The dependent variable in column (4) is an indicator variable which takes the value 1 if the contract requires guarantees and zero otherwise. In column (5), the dependent variable is an indicator variable which takes the value 1 if the loan failed and 0 otherwise.
Secured/Guaranteed takes the value 1 if a loan is secured or guaranteed. Missing is a missing value indicator which equals 1 if Secured/Guaranteed is missing and 0 otherwise. Definitions of the other variables are provided in Appendix B, Table B1. Distance is standardized throughout the sample. Estimates for the intercepts are not reported in this table. All regressions include loan type, loan purpose, and year fixed effects. Robust standard errors allowing for clustering by year are presented in parentheses. ***, **, and * denote statistical significance at the 0.1%, 1%, and 5%, respectively.
Loan spread Credit_Spread 37.4990*** (6.6034) -0.2989 (0.3437) 0.1474 (0.0973) -0.2305 (0.1650) 0.2779 (0.2454) Term_Spread 13.4258** (4.6859) -0.2341 (0.1536) 0.0613 (0.0330) 0.2233*** (0.0504) -0.1435 (0.1453)
‧
Table 2.3: Effect of relational distance on loan spreads
The dependent variable is all-in-drawn spread. Secured/Guaranteed takes the value 1 if a loan is secured or guaranteed. Missing is a missing value indicator which equals 1 if Secured/Guaranteed is missing and 0 otherwise. Definitions of the other variables are provided in Appendix B, Table B1. Estimates for the intercepts are not reported in this table.
All regressions include loan type, loan purpose, and 2-digit SIC fixed effects. Robust standard errors allowing for clustering by year are presented in parentheses. ***, **, and * denote statistical significance at the 0.1%, 1%, and 5%, respectively.
(1) (2) (3)
Path 3.5792*** (0.6606)
Clustering -44.3250*** (7.4740)
Density -51.6100*** (8.9697)
Log (Amt) -21.2732*** (1.4399) -19.1894*** (1.4915) -21.0105*** (1.4879) Maturity -0.0057 (0.0039) -0.0049 (0.0031) -0.0056 (0.0039) Secured/Guaranteed 59.3465*** (3.9918) 60.0370*** (4.2133) 57.9215*** (3.9096) Missing 8.1895* (3.3282) 8.3715* (3.5560) 8.6066* (3.3020) Covenant -11.8981** (3.4480) -8.4844* (3.6945) -13.0771** (3.6638) Lender_num -0.7946* (0.2765) -0.2291 (0.1819) -0.5435* (0.2093) Pub_unrated -1.9282 (2.8232) -1.4963 (2.2395) -2.1985 (2.8676) Pub_rated -3.1289 (3.7527) -4.3647 (3.0567) -3.0795 (3.7890) Log (Sales) -6.4798*** (1.4194) -5.0598** (1.4370) -6.0614*** (1.3628) MB -0.0016*** (0.0003) -0.0013** (0.0004) -0.0016*** (0.0003) ROA -2.2445*** (0.2952) -2.0234*** (0.2799) -2.2100*** (0.3103) Leverage 0.6966*** (0.0860) 0.6394*** (0.0796) 0.6857*** (0.0833) GDP_PC 15.8162* (6.8117) 13.1458 (7.4952) 18.1406* (6.8772) Credit_Spread 57.0536*** (10.5777) 58.5774*** (12.9945) 56.3230*** (10.9258) Term_Spread 21.0268** (6.9659) 19.8088* (7.4205) 20.3506* (7.0597) Location -2.6788 (2.5588) -4.9733 (2.4562) -2.5183 (2.5595) Fix effect:
‧
Table 2.4: Effects of relational distance on non-price contract terms
The dependent variable in Panel A is an indicator variable which takes the value 1 if a loan contract includes at least one financial covenant and 0 otherwise. The dependent variable in Panel B is an indicator variable which takes the value 1 if a loan is secured and 0 otherwise.
The dependent variable in Panel C is an indicator variable which takes the value 1 if a loan is guaranteed and 0 otherwise. Secured/Guaranteed takes the value 1 if a loan is secured or guaranteed. Missing is a missing value indicator which equals 1 if Secured/Guaranteed is missing and 0 otherwise. Definitions of the other variables are provided in Appendix B, Table B1. Estimates for the intercepts are not reported in this table. All regressions include loan type, loan purpose, and 2-digit SIC fixed effects. Robust standard errors allowing for clustering by year are presented in parentheses. ***, **, and * denote statistical significance at the 0.1%, 1%, and 5%, respectively.
(1) (2) (3)
Panel A:Results for financial covenants
Path 0.0024 (0.0210)
Clustering -0.1619 (0.0937)
Density -0.1143 (0.0928)
Log (Amt) -0.0630* (0.0256) -0.0934*** (0.0278) -0.0617* (0.0263) Maturity -0.0020* (0.0008) -0.0034*** (0.0010) -0.0020* (0.0008) Secured/Guaranteed -0.0621 (0.0618) -0.0510 (0.0673) -0.0712 (0.0621) Missing -1.6189*** (0.1106) -1.6203*** (0.1093) -1.6184*** (0.1125) GDP_PC 0.1406 (0.1659) 0.1626 (0.1578) 0.1465 (0.1638) Credit_Spread 0.1528 (0.0941) 0.1423 (0.0941) 0.1500 (0.0928) Term_Spread 0.0388 (0.0521) 0.0218 (0.0537) 0.0381 (0.0518) Location -0.0082 (0.0926) -0.0096 (0.0868) -0.0032 (0.0929) Fix effect: Maturity 0.0001 (0.0001) 0.0001 (0.0001) 0.0001 (0.0001) Guaranteed 0.1325* (0.0534) 0.1543** (0.0519) 0.1270* (0.0559) Covenant -0.2302* (0.0900) -0.2540** (0.0920) -0.2430** (0.0907)
‧
Lender_num -0.0055 (0.0042) 0.0033 (0.0048) -0.0010 (0.0036) Pub_unrated -0.3078** (0.1052) -0.3356** (0.1195) -0.3088** (0.1093) Credit_Spread -0.0634* (0.0256) -0.0872*** (0.0259) -0.0779** (0.0271) Term_Spread -0.0132 (0.0241) -0.0136 (0.0293) -0.0227 (0.0252) Location -0.1013 (0.0624) -0.1108* (0.0494) -0.0933 (0.0593)
Panel C: Results for guarantees
Path 0.0317*** (0.0088) Leverage 0.0007 (0.0007) 0.0002 (0.0009) 0.0006 (0.0007) GDP_PC 0.6774*** (0.1208) 0.7058*** (0.1170) 0.7001 *** (0.1185) Credit_Spread -0.0944 (0.1252) -0.0766 (0.1237) -0.0971 (0.1227) Term_Spread 0.2102** (0.0641) 0.2034** (0.0648) 0.2055** (0.0632) Location 0.0409 (0.0355) 0.0326 (0.0476) 0.0412 (0.0350) Fix effect:
‧
Table 2.5: Effect of relational distance on syndication failures
The dependent variable is a binary variable which takes the value 1 if the loan failed and 0 otherwise. Secured/Guaranteed takes the value 1 if a loan is secured or guaranteed. Missing is a missing value indicator which equals 1 if Secured/Guaranteed is missing and 0 otherwise.
Definitions of the other variables are provided in Appendix B, Table B1. Estimates for the intercepts are not reported in this table. Only samples with lenders number between two and five are used. Robust standard errors allowing for clustering by year are presented in parentheses. ***, **, and * denote statistical significance at the 0.1%, 1%, and 5%, respectively.
(1) (2) (3) Secured/Guaranteed 0.2997 (0.1537) 0.0172 (0.1343) 0.3122* (0.1527) Missing -0.0759 (0.2062) -0.2438 (0.2051) -0.0723 (0.2071) Covenant -0.8569*** (0.1434) -0.8504*** (0.2078) -0.8470*** (0.1399) Pub_unrated 0.0887 (0.1230) 0.1691 (0.1864) 0.1038 (0.1244) Pub_rated 0.2775* (0.1364) 0.2979 (0.1987) 0.2874* (0.1407) Log (Sales) 0.0034 (0.0361) 0.0428 (0.0455) -0.0019 (0.0356) MB -0.0000 (0.0001) -0.0001 (0.0001) -0.0001 (0.0002) ROA -0.0059 (0.0035) -0.0068* (0.0030) -0.0056 (0.0033) Leverage 0.0029 (0.0019) 0.0032 (0.0034) 0.0028 (0.0019) GDP_PC -0.4569* (0.1811) -0.5941** (0.2414) -0.4609* (0.1779) Credit_Spread 0.3051 (0.2258) 0.5890** (0.2202) 0.3137 (0.2217) Term_Spread -0.0395 (0.1046) -0.0569 (0.1309) -0.0430 (0.1041) Location -0.7130* (0.2885) -0.4634* (0.2349) -0.7180* (0.2901)
‧
Table 2.6: Robustness check: alternative explanations and model specifications
This table presents robustness tests to explore alternative explanations for the relationship between relational distance and loan price. The dependent variable is all-in-drawn spread in all panels. Previous_Relation, Investment_Grade, Covenant Violation, and Foreign_Lender, are indicator variables whose definitions are provided in Appendix B, Table B1.P_Previous_Relation, C_Previous_Relation, and D_Previous_Relation are interactions between Previous_Relation and Path, Clustering, and Density, respectively. The similar notations and definitions can be applied to the other panels. Estimates for the intercepts and control variables are not reported in this table. All regressions include loan type, loan purpose, and 2-digit SIC fixed effects. Robust standard errors allowing for clustering by year are presented in parentheses. ***, **, and * denote statistical significance at the 0.1%, 1%, and 5%, respectively.
(1) (2) (3)
Panel A: Previous relationship between lead bank and borrower Path 4.3131** (1.2442)
Clustering -41.3556*** (8.7480)
Density -54.7697*** (10.7821)
Previous_Relation -5.1585 (4.1042) -0.9542 (9.1590) -16.2604 (7.8482) P_Previous_Relation -1.3678 (1.3015)
C_Previous_Relation -6.0565 (11.4687)
D_Previous_Relation 11.9728 (10.0375)
Panel B: Borrower’s rating
Path 3.1386*** (0.7530)
Clustering -46.9178*** (8.8689)
Density -49.1490*** (9.0554)
Investment_Grade -41.2863*** (4.4285) -43.6701** (14.6073) -49.7238*** (10.1636) P_Investment_Grade 1.2634 (0.8460)
C_Investment_Grade 4.8189 (13.5304)
D_Investment_Grade 12.1345 (8.3259)
Panel C: Covenant violation
Path 4.4761** (1.4665)
Clustering -35.0436* (12.3289)
Density -39.0086* (13.2394)
Covenant_Violation 23.4100*** (2.7215) 26.9637** (8.0773) 46.0926*** (9.6531) P_Covenant_Violation 1.2044 (1.2861)
C_Covenant_Violation -8.0228 (9.6935)
D_Covenant_Violation -26.9496* (12.0530) Panel D: Foreign participants
Path 7.8838** (2.6224)
Clustering -41.8129*** (6.4784)
Density -54.7658*** (10.9528)
Foreign_Lender -1.0971 (4.0324) 0.5974 (9.7382) -16.0555 (8.3753) P_Foreign_Lender -5.1485 (2.6087)
C_Foreign_Lender -4.1282 (10.6922)
D_Foreign_Lender 8.0208 (7.9902)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
49
Table 2.7: Robustness check: endogeneity
This table reports the results of two-stage instrumental variable regressions. Panel A is the first-stage results and Panel B is the second-stage results. All control variables and fix effects in Table 3 are included in both stages but not reported. The dependent variables in column (1), (2), and (3) of Panel A are Path, Clustering, and Density, respectively. The dependent variable in Panel B is all-in-drawn spread. Inactive_num is the instrumental variable and defined as the number of nonlead-bank members who had not joined any loan syndicate in the previous three years. Definitions of the other variables are provided in Appendix B, Table B1. Robust standard errors allowing for clustering by year are presented in parentheses. ***, **, and * denote statistical significance at the 0.1%, 1%, and 5%, respectively.
(1) (2) (3)
Panel A: First stage
Inactive _num 1.7375*** (0.0994) -0.0297*** (0.0034) -0.0547*** (0.0073)
N 20,170 17,342 20170
Adj. R-sq 0.8205 0.1648 0.2023
Test of instrument strength (Ho: the instrument is weak)
F-statistic 305.3780*** 76.4786*** 56.1779***
Test of exogeneity (Ho: the instrument is exogenous)
F-statistic 0. 0376 7.1267* 5.1933* Panel B: Second stage
Path 3.6828*** (0.9323)
Clustering -167.2799** (55.9041)
Density -116.8941** (39.5805)
N 20,170 17,342 20,170
Adj. R-sq 0.5380 0.5249 0.5311
‧
Table 2.8: Robustness check: omitted variables
The dependent variable is all-in-drawn spread. Borrower_Experience is the number of previous loans that the borrower borrows from the syndicated loan markets.
Lender_Experience is the average number of previous loans that the lenders of a given facility participate in the syndicated loan market. Top5 is a dummy variable, which indicates whether the lead bank belongs to the top five lenders in the league table. Z-score is Altman’s (1968) Z-score. Secured/Guaranteed takes the value 1 if a loan is secured or guaranteed. Missing is a missing value indicator which equals 1 if Secured/Guaranteed is missing and 0 otherwise.
Definitions of the other variables are provided in Appendix B, Table B1. Estimates for the intercepts are not reported in this table. All regressions include loan type, loan purpose, and 2-digit SIC fixed effects. Robust standard errors allowing for clustering by year are presented in parentheses. ***, **, and * denote statistical significance at the 0.1%, 1%, and 5%,
Previous_Relation 1.3912 (0.8791) 2.3584* (1.0526) 1.3491 (0.8990) Investment_Grade -45.5459*** (9.0027) -43.0530*** (7.3032) -45.1702*** (9.0547) Covenant_Violation 4.7711*** (0.7752) 4.2724*** (0.8658) 4.8635*** (0.7511) Foreign_Lender -12.4417** (3.9321) -5.9076* (2.6491) -13.6973** (4.0234) Borrower_Experience 0.0346 (0.3418) 0.3504 (0.2975) 0.0602 (0.3441) Lender_Experience -0.0085*** (0.0014) -0.0063** (0.0019) -0.0067*** (0.0014)
Top5 -0.8537 (4.0508) 1.9458 (3.7815) 0.1499 (3.8500) Z_Score -0.6531* (0.2472) -2.4235* (0.8742) -0.6376* (0.2469)
Log (Amt) -20.4164*** (2.0093) -19.6463*** (1.7956) -20.6941*** (1.9692) Maturity -0.9595*** (0.1749) -0.9168*** (0.1809) -0.9451*** (0.1820) Secured/Guaranteed 63.3064*** (4.5449) 64.6131*** (5.4718) 62.8998*** (4.5512) Missing 23.8258*** (4.7392) 25.7356*** (3.6585) 23.5859*** (4.5572) Covenant -15.5452** (4.7237) -8.6916* (3.2024) -15.7647** (4.6035) Lender_num -1.3320* (0.4565) -0.5351 (0.2675) -0.8750* (0.3198)
Pub_unrated 4.0972 (6.6157) 7.3356 (5.3142) 3.9213 (6.5292) Pub_rated 16.8857* (6.9468) 15.8912* (6.6389) 16.4534* (6.9543)
Log (Sales) -1.3783 (2.7821) -1.1280 (2.3908) -1.0359 (2.7370)
MB 0.0445 (0.0497) 0.0721 (0.0444) 0.0493 (0.0535) ROA -1.7446*** (0.3679) -1.7158*** (0.3382) -1.7329*** (0.3804)
Leverage 0.9090*** (0.0989) 0.6273*** (0.1312) 0.8865*** (0.0975) GDP_PC 34.0044*** (5.2237) 22.2191*** (5.1405) 32.7333*** (5.3619) Credit_Spread 54.2489** (13.0833) 55.6013** (16.9992) 53.3515** (13.3535) Term_Spread 12.9894* (4.9710) 10.7183 (5.4424) 12.6624* (4.9237)
Location 6.0115 (5.0293) -0.4282 (4.8717) 5.9940 (5.0546)
‧
Consider the denominator. By performing integration by parts:
1 1 1 1 1 1
Continuing the process of integration by parts:
1 1 0
The same procedure can be applied to the numerator in (A.1):
1 1 1 2 1 Dividing expression (A.2) by (A.3) yields
‧
Proof of equations (2) and (3):
Using the result of equation (A.2):
1
The first order condition is
*
‧
Amt Facility amount Million US$
Maturity Loan maturity Months
Secured 1 if the loan is secured, 0 otherwise 0 or 1 Guaranteed 1 if the loan is guaranteed, 0 otherwise 0 or 1 Covenant 1 if the loan has financial covenants, 0 otherwise 0 or 1 Lender_num Lender number of a given facility
Failed 1 if the deal status is “Cancelled” or “Suspended”, 0 otherwise
0 or 1
Panel B: Borrower characteristics
Pub_unrated 1 if the borrower is public and not rated, 0 otherwise Pub_rated 1 if the borrower is public and rated, 0 otherwise
Sales Borrower’s sales at the end of the year prior to the loan active
MB Borrower’s market to book ratio at the end of the year prior to the loan active
ROA Borrower’s return on assets at the end of the year prior to the loan active
Leverage Borrower’s total debt to total assets ratio at the end of the year prior to the loan active
Panel C: Proxies for the segmentation of communication
Distance Average physical distance among lenders for a given facility
kilometer
Path Average shortest path among lenders for a given facility Cluster The member’s clustering coefficient for a given facility Density The proportion of links in a given network relative to the
total possible links Panel D: other variables
GDP_PC GDP per capita (annual) US$10,000
Credit_Spread The difference between the yields of BAA and AAA corporate bonds
%
Term_Spread The difference between the yields of 10-year and 2-year Treasury bonds
%
Location 1 if the borrower’s principal executive office is the same as the lead bank’s
0 or 1
Previous_Relationship 1 if there exists a previous relationship between the lead bank and borrower
0 or 1
Investment_Grade 1 if the borrower’s credit rating is investment grade 0 or 1 Covenant_Violation 1 if the borrower violate a financial covenant 0 or 1 Foreign_Lender 1 if there is at least one foreign lender in a given facility 0 or 1
‧
Table B2: Effects of physical distance on loan contract terms and syndication failures: Facilities without foreign lender
The dependent variable in column (1) is all-in-drawn spread. The dependent variable in column (2) is an indicator variable which takes the value 1 if a loan contract includes at least one financial covenant and 0 otherwise. The dependent variable in column (3) is an indicator variable which takes the value 1 if the loan has been secured and zero otherwise. The dependent variable in column (4) is an indicator variable which takes the value 1 if the contract requires guarantees and zero otherwise. In column (5), the dependent variable is an indicator variable which takes the value 1 if the loan failed and 0 otherwise. Only facilities without foreign number are included in this analysis. Secured/Guaranteed takes the value 1 if a loan is secured or guaranteed. Missing is a missing value indicator which equals 1 if Secured/Guaranteed is missing and 0 otherwise. Definitions of the other variables are provided in Appendix B, Table B1.
Distance is standardized throughout the sample. Estimates for the intercepts are not reported in this table. All regressions include loan type, loan purpose, and 2-digit SIC fixed effects. Robust standard errors allowing for clustering by year are presented in parentheses. ***, **, and * denote statistical significance at the 0.1%, 1%, and 5%, respectively.
Loan spread
Pub_unrated -2.3848 (6.8450) 0.0332 (0.1010) -0.1407 (0.1034) 0.0301 (0.1256) -0.0402 (0.3014) Pub_rated -8.2764 (6.6067) 0.0405 (0.1373) 0.1577 (0.1327) -0.2277 (0.1940) 0.3179 (0.2054) Log (Sales) -7.3977* (2.6924) -0.1937*** (0.0414) -0.2703*** (0.0422) 0.1105** (0.0357) 0.1078 (0.1237)
‧
Table B3: Effects of physical distance on loan contract terms and syndication failures: Facilities with foreign lenders
The dependent variable in column (1) is all-in-drawn spread. The dependent variable in column (2) is an indicator variable which takes the value 1 if a loan contract includes at least one financial covenant and 0 otherwise. The dependent variable in column (3) is an indicator variable which takes the value 1 if the loan has been secured and zero otherwise. The dependent variable in column (4) is an indicator variable which takes the value 1 if the contract requires guarantees and zero otherwise. In column (5), the dependent variable is an indicator variable which takes the value 1 if the loan failed and 0 otherwise. Only facilities with at least one foreign number are included in this analysis. Secured/Guaranteed takes the value 1 if a loan is secured or guaranteed. Missing is a missing value indicator which equals 1 if Secured/Guaranteed is missing and 0 otherwise. Definitions of the other variables are provided in Appendix B, Table B1. Distance is standardized throughout the sample. Estimates for the intercepts are not reported in this table. All regressions include loan type, loan purpose, and 2-digit SIC fixed effects. Robust standard errors allowing for clustering by year are presented in parentheses. ***, **, and * denote statistical significance at the 0.1%, 1%, and 5%, respectively.
Loan spread Secured/Guaranteed 62.8617*** (3.9781) -0.1181 (0.0870) 0.1717 (0.1756)
Guaranteed -0.0455 (0.1190)
Secured -0.0009 (0.1165)
Missing 11.9394** (4.0893) -1.6441*** (0.1419) -0.5153* (0.2090) -0.6331** (0.1963) Covenant -0.7128 (4.9334) 0.0452** (0.0145) -0.1872** (0.0715) 1.3665*** (0.2794) -0.7964*** (0.2184) Lender_num -0.6799* (0.3237) -0.0189 (0.1078) -0.0272** (0.0102) -0.0146 (0.0130)
Pub_unrated 6.3894 (5.7656) -0.0284 (0.1203) -0.0979 (0.1058) -0.2594 (0.2089) 0.2757 (0.2658) Pub_rated 2.3779 (3.7648) -0.1300*** (0.0318) -0.1575 (0.0925) -0.2626 (0.1789) 0.3675 (0.3035) Credit_Spread 52.4664** (14.1253) -0.2982 (0.1585) 0.2649 (0.2144) -0.0223 (0.2536) 0.6078** (0.2072) Term_Spread 13.6303* (5.8556) -0.0994 (0.0906) 0.0986 (0.0573) 0.2677** (0.0849) -0.1343 (0.1762)
‧
Table B4: Effects of physical distance on loan contract terms and syndication failures: Big loans
The dependent variable in column (1) is all-in-drawn spread. The dependent variable in column (2) is an indicator variable which takes the value 1 if a loan contract includes at least one financial covenant and 0 otherwise. The dependent variable in column (3) is an indicator variable which takes the value 1 if the loan has been secured and zero otherwise. The dependent variable in column (4) is an indicator variable which takes the value 1 if the contract requires guarantees and zero otherwise. In column (5), the dependent variable is an indicator variable which takes the value 1 if the loan failed and 0 otherwise. Only facilities with facility amount equal to or larger than sample median are included in this analysis. Secured/Guaranteed takes the value 1 if a loan is secured or guaranteed. Missing is a missing value indicator which equals 1 if Secured/Guaranteed is missing and 0 otherwise. Definitions of the other variables are provided in Appendix B, Table B1. Distance is standardized throughout the sample. Estimates for the intercepts are not reported in this table. All regressions include loan type, loan purpose, and 2-digit SIC fixed effects. Robust standard errors allowing for clustering by year are presented in parentheses. ***, **, and * denote statistical significance at the 0.1%, 1%, and 5%, respectively.
Loan spread Lender_num -0.5198 (0.5259) 0.0532*** (0.0118) -0.0179 (0.0106) -0.0203 (0.0200)
Pub_unrated 3.7690 (3.2682) 0.0104 (0.0834) -0.2232* (0.0993) -0.1284 (0.1824) -0.0321 (0.3303) Pub_rated 6.4561 (5.4597) 0.0787 (0.1272) 0.0552 (0.0973) -0.1282 (0.1274) 0.0461 (0.3206) Log (Sales) -11.9162*** (2.2766) 0.0124 (0.0491) -0.3235*** (0.0701) 0.1403 (0.0717) 0.0939 (0.0744) MB -0.1224*** (0.0293) 0.0052 (0.0035) -0.0004 (0.0010) -0.0040 (0.0053) 0.0011 (0.0006) ROA -1.3588*** (0.2555) 0.0021 (0.0018) -0.0209** (0.0076) -0.0007 (0.0038) -0.0083 (0.0057) Leverage 0.8273*** (0.1222) -0.0013 (0.0010) 0.0129*** (0.0029) 0.0008 (0.0020) -0.0001 (0.0038) GDP_PC 32.3359*** (54.0221) 1.0560*** (2.3876) 0.4847*** (0.6783) 0.9274*** (1.0385) -0.1013 (2.2041) Credit_Spread 23.4693* (8.5410) -0.3862 (0.4290) 0.0984 (0.0956) -0.2416 (0.2457) -1.5341 (1.0255) Term_Spread 13.0015** (4.1991) -0.3746* (0.1876) -0.0190 (0.0381) 0.2147*** (0.0560) -0.5663* (0.2510)
‧
Table B5: Effects of physical distance on loan contract terms and syndication failures: Small loans
The dependent variable in column (1) is all-in-drawn spread. The dependent variable in column (2) is an indicator variable which takes the value 1 if a loan contract includes at least one financial covenant and 0 otherwise. The dependent variable in column (3) is an indicator variable which takes the value 1 if the loan has been secured and zero otherwise. The dependent variable in column (4) is an indicator variable which takes the value 1 if the contract requires guarantees and zero otherwise. In column (5), the dependent variable is an indicator variable which takes the value 1 if the loan failed and 0 otherwise. Only facilities with facility amount smaller than sample median are included in this analysis. Secured/Guaranteed takes the value 1 if a loan is secured or guaranteed. Missing is a missing value indicator which equals 1 if Secured/Guaranteed is missing and 0 otherwise. Definitions of the other variables are
The dependent variable in column (1) is all-in-drawn spread. The dependent variable in column (2) is an indicator variable which takes the value 1 if a loan contract includes at least one financial covenant and 0 otherwise. The dependent variable in column (3) is an indicator variable which takes the value 1 if the loan has been secured and zero otherwise. The dependent variable in column (4) is an indicator variable which takes the value 1 if the contract requires guarantees and zero otherwise. In column (5), the dependent variable is an indicator variable which takes the value 1 if the loan failed and 0 otherwise. Only facilities with facility amount smaller than sample median are included in this analysis. Secured/Guaranteed takes the value 1 if a loan is secured or guaranteed. Missing is a missing value indicator which equals 1 if Secured/Guaranteed is missing and 0 otherwise. Definitions of the other variables are