The variables used in the regression equations (presented in Chapter 3) are described in Section 4.2. This section describes their descriptive statistics (Section 4.3.1) and discusses the results of their related correlation analysis (Section 4.3.2).
4.3.1. Descriptive Statistics
As stated in Chapter 3, Section 3.2.3, prior studies indicate that the characteristics of revolving loans differ from those of term loans. The descriptive statistics for the revolving and term loan samples and the tests of their differences are therefore shown separately in Table 4.3.69 The latter involve both a t-test and Wilcoxon’s sum test, testing the mean and the median differences, respectively (Chen and Martin, 2011), to determine whether the two loan type characteristics are statistically different.
69 To exclude extreme values, these variables are winsorized at the first and 99th percentiles, following Sufi (2009b) and Nikolaev (2010).
Table 4.3: Descriptive statistics and difference test results for revolving and term loans
This table presents the distribution of variables by showing mean, median, standard deviation (Std. Dev.), minimum (Min), maximum (Max) and t-test and Wilcoxon’s sum test results for revolving and term loans. See Table 4.2 for these variable definitions. Unless other be stated, all variables are stated on ratios.
Variables
Revolving Loans Term Loans
Difference
Mean Median N Mean Median Std.
Dev. Min Max N Mean Median Std.
Dev. Min Max t-Test Wilcoxon’s Sum Test Panel A: Loan Terms
AISD (%) 15,866 1.651 1.500 1.096 0.165 5.000 6,000 2.585 2.500 1.290 0.350 7.500 -0.934*** -1.000***
COLLATERAL 11,383 0.700 1.000 0.458 0.000 1.000 5,095 0.909 1.000 0.287 0.000 1.000 -0.209*** 0.000***
MATURITY (years) 16,166 3.061 3.000 1.762 0.333 7.250 6,254 4.835 5.000 2.082 0.417 10.000 -1.774*** -2.000***
COVINDEX 17,636 1.746 1.000 1.928 0.000 7.000 6,625 2.840 2.000 2.575 0.000 7.000 -1.094*** -1.000***
FACSIZE ($ millions) 17,636 224 100 360 1 2,160 6,625 115 51 173 1 1,034 109*** 49***
LN(FACSIZE) 17,636 4.395 4.605 1.598 0.278 7.678 6,625 3.659 3.929 1.730 -0.660 6.941 0.736*** 0.676***
Panel B: Information Asymmetry
IAINDEX 17,636 2.962 3.000 0.820 1.250 5.000 6,625 2.952 3.000 0.847 1.250 5.000 0.010 0.000 IAINDEX’s components
FIRMAGE (years) 17,636 18.152 11.000 15.814 1.000 54.000 6,625 13.784 9.000 12.810 1.000 54.000 4.368*** 2.000***
LN(FIRMAGE) 17,636 2.446 2.398 1.031 0.000 3.989 6,625 2.202 2.197 0.963 0.000 3.989 0.244*** 0.201***
R&D 17,636 0.020 0.000 0.052 0.000 0.311 6,625 0.021 0.000 0.058 0.000 0.353 -0.001 0.000 FCSTERROR 12,698 0.383 0.080 1.037 0.000 7.877 4,062 0.547 0.115 1.392 0.000 9.680 -0.163*** -0.035***
DISPERSION 11,863 0.113 0.033 0.271 0.000 2.000 3,721 0.162 0.042 0.402 0.000 3.000 -0.048*** -0.008***
(Table 4.3 continued)
Variables
Revolving Loans Term Loans Different Tests
Mean Median N Mean Median Std. Dev. Min Max N Mean Median Std. Dev. Min Max t-Test Wilcoxon’s
Sum Test Panel C: Lending Relationship
REL(D) 17,636 0.485 0.000 0.500 0.000 1.000 6,625 0.426 0.000 0.426 0.000 1.000 0.058*** 0.000 Alternative measures
REL(A) 17,636 0.381 0.000 0.439 0.000 1.000 6,625 0.328 0.000 0.425 0.000 1.000 0.053*** 0.000***
REL(N) 17,636 0.365 0.000 0.427 0.000 1.000 6,625 0.315 0.000 0.413 0.000 1.000 0.050*** 0.000***
Panel D: Other Borrower Characteristics
TOTALASSETS ($ millions) 17,636 2,629 497 5,808 6 34,194 6,625 1411 302 3,488 6 24,308 1218*** 195***
LN(TOTALASSETS) 17,634 6.242 6.209 1.921 1.813 10.440 6,625 5.740 5.712 1.765 1.823 10.099 0.502*** 0.497***
PROFITABILITY 16,777 0.137 0.126 0.210 -1.101 0.711 6,299 0.101 0.122 0.318 -2.166 0.612 0.036*** 0.004**
LEVERAGE 17,602 0.255 0.233 0.217 0.000 1.046 6,618 0.332 0.293 0.276 0.000 1.348 -0.077*** -0.06***
M/B 15,198 1.802 1.445 1.087 0.696 7.024 5,449 1.694 1.404 0.942 0.682 6.282 0.108*** 0.041***
TANGIBILITY 16,865 0.342 0.277 0.250 0.003 0.914 6,384 0.324 0.270 0.233 0.003 0.904 0.018*** 0.007 COVERAGE 16,910 18.432 5.167 51.967 0.000 391.725 6,410 12.752 3.537 39.259 0.000 321.191 5.679*** 1.630***
LN(1+COVERAGE) 16,910 1.922 1.819 1.243 0.000 5.973 6,410 1.630 1.512 1.160 0.000 5.775 0.292*** 0.307***
CONCENTRATION 16,099 0.315 0.279 0.212 0.014 0.899 6,149 0.269 0.216 0.211 0.006 0.897 0.046*** 0.063***
ASSETMATURITY 15,904 5.333 3.015 5.872 0.398 30.610 6,040 4.812 2.913 5.420 0.418 31.200 0.522*** 0.102*
LN(ASSETMATURITY) 15,904 1.188 1.104 0.986 -0.920 3.421 6,040 1.126 1.069 0.924 -0.873 3.440 0.061*** 0.035*
EARNINGSVARIANCE 14,657 0.054 0.034 0.060 0.002 0.351 5,263 0.059 0.039 0.063 0.003 0.381 -0.005*** -0.004***
TAX_ASSETS 17,572 0.022 0.018 0.031 -0.068 0.131 6,589 0.016 0.011 0.031 -0.081 0.124 0.006*** 0.007***
REGULATED 17,636 0.067 0.000 0.250 0.000 1.000 6,625 0.020 0.000 0.139 0.000 1.000 0.047*** 0.000***
CFVOLATILITY 15,936 0.052 0.031 0.066 0.003 0.443 5,858 0.059 0.036 0.075 0.003 0.529 -0.007*** -0.005***
CURRENTRATIO 16,029 1.922 1.592 1.328 0.284 8.192 6,124 1.972 1.603 1.531 0.221 10.314 -0.050** -0.011
(Table 4.3 continued)
Variables
Revolving Loans Term Loans Different Tests
Mean Median N Mean Median Std. Dev. Min Max N Mean Median Std. Dev. Min Max t-Test Wilcoxon’s
Sum Test Panel E: Lender Characteristic
TOPLENDERS 17,636 0.408 0.000 0.491 0.000 1.000 6,625 0.327 0.000 0.469 0.000 1.000 0.081*** 0.000***
Panel F: Other Loan Characteristics
SYNDICATION 17,636 0.877 1.000 0.329 0.000 1.000 6,625 0.856 1.000 0.351 0.000 1.000 0.021*** 0.000***
FIRSTLOAN 17,636 0.180 0.000 0.385 0.000 1.000 6,625 0.196 0.000 0.397 0.000 1.000 -0.015*** 0.000***
TRANCHE 17,636 0.361 0.000 0.480 0.000 1.000 6,625 0.780 1.000 0.414 0.000 1.000 -0.419*** -1.000***
LENDERNUMBER 17,636 6.515 4.000 7.246 1.000 37.000 6,625 6.147 3.000 7.833 1.000 44.000 0.368*** 1.000***
GENPURPOSES 17,636 0.506 1.000 0.500 0.000 1.000 6,625 0.357 0.000 0.479 0.000 1.000 0.149*** 1.000***
RECAPITALIZATION 17,636 0.246 0.000 0.431 0.000 1.000 6,625 0.282 0.000 0.450 0.000 1.000 -0.036*** 0.000***
ACQUISITION 17636 0.141 0.000 0.348 0.000 1.000 6,625 0.317 0.000 0.465 0.000 1.000 -0.177*** 0.000***
OTHERPURPOSE 17,636 0.107 0.000 0.309 0.000 1.000 6,625 0.044 0.000 0.204 0.000 1.000 0.063*** 0.000***
Panel G: Macroeconomic Factors
LIBOR (%) 17,636 4.953 5.426 1.715 1.449 7.750 6,625 4.893 5.426 1.695 1.449 7.750 0.060*** 0.000***
IRVOLATILITY 17,636 0.099 0.094 0.017 0.069 0.145 6,625 0.099 0.094 0.017 0.069 0.145 0.000* 0.000***
TERMPREMIUM (%) 17,636 1.011 0.750 0.880 -0.130 2.740 6,625 0.987 0.750 0.887 -0.130 2.740 0.024* 0.000***
CREDITSPREAD (%) 17,636 0.829 0.700 0.343 0.520 3.330 6,625 0.818 0.750 0.229 0.520 1.310 0.012*** -0.050***
Note: The superscripts ***, **, and * indicate statistical significances at the 1%, 5% and 10% levels respectively.
As shown in Table 4.3, the revolving and term loan samples account for 73 per cent (17,636 loans) and 27 per cent (6,625 loans), respectively, of the final sample. The revolving loan price (AISD), on average, is 1.651 per cent, while that of term loans is higher, about 2.585 per cent.
Non-price terms also differ across loan types (Panel A of Table 3). Revolving loans are generally less likely to pledge collateral compared to term loans: 70 per cent of revolving loans are secured, compared to nearly 91 per cent for term loans.
Term loan maturities (4.835 years) are on average longer those of revolving loans (3.061 years). This is probably because term loans often finance long-term investments. The mean of COVINDEX for term loans (2.840) is also higher than for revolving loans (1.746). Loan sizes on revolvers average USD 224 million versus USD 115 million for term loans. The fact that revolving loans are larger may reflect the borrower size differences shown in Panel D of Table 4.3. These loan term differences may reflect differing borrower characteristics. Since revolving loan borrowers tend to be better borrowers (as seen from the variables in Panel D of Table 4.3), they receive better loan terms.
Overall, as shown in Panel B of Table 4.3, term loan borrowers with an IAINDEX value of 2.952 appear to have fewer information asymmetry problems than revolving loan borrowers with an IAINDEX value of 2.962, but the means of all of its components suggest that term loans face more information asymmetry problems than revolving loans. This may be due to the construction of IAINDEX (as discussed in Section 4.2.3.1), based on yearly ranking (to avoid the secular trends of its components). The IAINDEX mean may therefore reflect borrower information
asymmetries better than the means of these components. Regarding lending relationships, revolving loans are based more on lending relationships (48.5 per cent) than term loans (42.6 per cent) (in Panel B of Table 4.3). The intensities of lending relationships for revolving loans measured by REL(A) and REL(N) (0.381 and 0.365, respectively) are also higher than those of term loans (0.328 and 0.315, respectively).
As mentioned previously, most other borrower characteristic variables (as shown in Panel D of Table 4.3) show that borrowers of revolving loans have better variables than borrowers of term loans. On average, borrowers of revolving loans are nearly twice the size of term loan borrowers (USD 2,63 billion versus USD 1.41 billion). Borrowers with more profitable, lower-leveraged, and higher-growth prospects tend to borrow revolving loans. With respect to lender characteristic (Panel E of Table 4.3), more than 40 per cent of revolving loans come from the top lenders, compared to about 32 per cent of term loans. This suggests that top lenders may prefer revolving to term loans. In relation to other loan characteristics (Panel F of Table 4.3), only 36 per cent of revolving loans are involved in multi-tranches, whereas more than twice the percentage, 78 per cent, of term loans are multi- tranches. Revolving loans are usually for general corporate purposes (about 50 per cent), while term loans are also often used for recapitalization and acquisition (about 28 per cent and 32 per cent, respectively, for a total of 60 per cent).
The t-test and Wilcoxon’s sum test show that most of the two samples’
variables differ significantly. This statistical evidence supports the suggestion, made in Chapter 3, Section 3.2.3, that the two types of loan should be examined separately.
4.3.2. Correlation Analysis
High correlation between independent variables can result in multicollinearity problem and is therefore checked for accordingly. The t-test and Wilcoxon’s sum test in Section 4.3.1 (Table 4.3) find these two loan types to be statistically different, as suggest by prior studies discussed in Section 3.2.3 so Table C1 of Appendix C reports the Pearson product–moment correlation coefficients between the variables used in the regressions for revolving loans (Panel A) and term loans (Panel B) separately.
As shown in Appendix C, most dependent variables in the five equations (Equations 3.1–3.5) have low correlations. They may therefore not cause a multicollinearity problem when included as independent variables. However, LN(FACSIZE) is highly and significantly correlated with LN(TOTALASSETS) in both samples (i.e., 0.80 for revolving loans and 0.74 for term loans). These two variables are therefore not included together in the same equation. In addition, following Dennis, et al. (2000), LIBOR and TERMPREMIUM are the two independent variables in the loan price equation (Equation 3.1). They are highly correlated in both loan samples (i.e., -0.79 for both loan types), but is lower than for Dennis et al. (2000), who find the correlation to be as high as -0.883. This may create a multicollinearity problem, but the multicollinearity tests reported in Appendix E (Section E.1), show that this is not a serious issue.