Table 6.9 presents the results of the pooled OLS for Model (1) based on the book debt ratio for the sample of rated firms only. Column 1 shows the OLS results of the book debt ratio regressed upon credit ratings and their squares. Column 2 shows the results of the full model, containing all the control variables, and Column 3 displays the results for Model (1) based on the rated firms’ sample without the inclusion of the utility firms.
145 Column 1 shows that the F-value is significant at the 1% level, suggesting that the model without control variables is significant in explaining variations in capital structures. Therefore, the null hypothesis, that the slopes of the explanatory variables are
simultaneously zero, is rejected. An adjusted-R2 of 16.6% suggests that CR and CR2 can
jointly explain 16.6% of the variation in the debt ratio of rated firms. Consistent with the hypotheses, the coefficient of CR is positive and significant at the 1% level while the
coefficient of CR2 is negative and significant at the 1% level, indicating an inverted U-
shaped relationship between the credit ratings and capital structures of UK firms.
Table 6.9
Pooled Time-series Cross-sectional Regression of Book Debt Ratio on Credit Ratings and Control Variables (Rated Firms only)
Variables 1 2 3 (Constant) CR CR2 LOS PROF FAR MBR LIQD TECHdum INDdum CSdum CGdum HCdum UTLdum BMdum OGdum 0.029 (0.84) 0.167 (6.62)*** -0.017 (-3.85)*** 0.533 (5.11)*** 0.187 (6.99)*** -0.021 (-4.45)*** -0.041 (-7.57)*** 0.099 (2.07)** 0.107 (4.19)*** 0.025 (5.22)*** -0.046 (-6.68)*** 0.003 (0.07) -0.007 (-0.27) 0.002 (0.11) 0.105 (4.25)*** 0.024 (0.68) 0.037 (1.43) 0.022 (0.81) 0.074 (2.29)** 0.565 (4.88)*** 0.169 (5.84)*** -0.017 (-3.40)*** -0.041 (-6.95)*** 0.128 (2.48)** 0.078 (2.95)*** 0.023 (4.65)*** -0.043 (-5.98)*** -0.013 (-0.31) -0.006 (-0.22) 0.005 (0.24) 0.104 (4.17)*** 0.023 (0.64) 0.027 (0.98) 0.077 (2.36)** Adj R2 F Sig N .166 87.516 .000 874 .338 21.963 .000 874 .429 30.741 .000 760
Notes: This table displays the OLS regression results for the rated firms only with (Columns 1 and 2) and without utility firms (Column 3) for Model (1). Coefficients are reported outside parenthesis while t- values are in the parenthesis. ***, ** and * denotes p-values significant at the 1%, 5% and 10% levels, respectively. Variables are defined as total debt to total assets (TDTA) as dependent variable, numerical code 1-5 for credit rating (CR), credit rating square (CR2), log of sales (LOS) refers to natural logarithm of sales, profitability (PROF) is the ratio of earnings before interest, taxes and depreciation to total assets, fixed assets ratio (FAR) is the ratio of fixed assets to total assets, market to book ratio (MBR) is the book value of the assets minus the book value of the equity minus market value of equity divided by book value of assets and liquidity ratio (LIQD)is the ratio of current assets to total assets, technology dummy (TECHdum), industrial dummy (INDdum), consumer services dummy (CSdum), consumer goods dummy (CGdum), heath care dummy (HCdum), utility dummy (UTLdum), basic material dummy (BMdum) and oil and gas dummy (OGdum).
After adding the restrictions of firm-level factors into the model, the adjusted-R2 increases
from 16.6% to 33.8% with an F-value of 21.96 at p<0.01.This adjusted-R2 is higher than
the prior UK studies reporting 18% (Rajan and Zingales, 1995), 8.22% (Bevan and
Danbolt, 2002) and 31% (Jong et al., 2008) adjusted-R2. This is possibly because the
146 the overall model. Such differences however, can also be attributable to a different sample period and sample size along with differences in the specification of the model. For example, these studies do not control for industry effects and/or liquidity in their models.
The coefficients of CR and CR2 remain qualitatively similar after other firm-level factors
are controlled for in the model. The coefficients are highly significant and are of expected
signs. The positive coefficient on CR and the negative coefficient on CR2 indicate that the
leverage increases with 18.7 percentage points but the rate of increase simultaneously decreases by 2.1 percentage points with each consecutive squared rating. After it has attained its peak, leverage then diminishes with the increase in CR (i.e., with a decrease in the credit quality of the firm) which would imply a non-linear relationship. This suggests that firms with high and low credit ratings have lower leverage in their capital structures
compared to their counterpart mid rated firms. This provides strong evidence for the credit
rating – capital structure hypothesis (CR-CS) in predicting that concerns for the costs and benefits of credit ratings drive firms to follow conservative debt policies despite having
better access to debt markets, as is suggested by the credit rating – market access
hypothesis (CR-MA).
The non-linear relationship between the credit ratings and capital structures of firms suggests that previous studies such as that by Mittoo and Zhang (2010) have been unable to fully capture the complex relationship between the credit ratings and capital structures of firms. For example, they empirically find a negative relationship between credit ratings and leverage. Mittoo and Zhang argue that before acquiring credit ratings, speculative grade firms were constrained by debt capacity and their rating status facilitated them in accessing public debt markets, resulting in high levels of gearing. However, inconsistently with Mittoo and Zhang (2010), the results of the present study (Table 6.9) indicate that, similar to high rated firms, lowest rated firms within the speculative grade have also relatively low levels of leverage. Low rated firms are likely to face supply-side constraints due to their credit ratings relative to medium rated and high rated firms. Moreover, they can be expected to have higher concerns for the costs imposed by their credit ratings, as downgrades would have relatively more serious implications than their counterpart high rated and medium rated firms. Consistent with the CR-CS hypothesis, the results indicate that they will prefer to have low gearing ratios. For such firms, as predicted by the CR-CS hypothesis, the costs of low ratings and any subsequent downgrades are higher than the benefits of employing more leverage. As discussed in Chapter 2, the institutional settings
147 of the UK market would also increase the concerns over low ratings. For example, the creditor friendly bankruptcy code and lower proportion of low rated firms would make low rated firms particularly concerned about their credit ratings leading to low gearing ratios. This finding, however, is inconsistent with Lemmon and Zender (2004), who argue that rated firms, irrespective of whether they are investment grade or speculative grade, will have better access to debt markets and have high levels of leverage.
Consistent with the hypothesis and prior study by Mittoo and Zhang (2010), high rated firms seem to have relatively low gearing ratios. This suggests that despite having better access to debt markets, as is suggested by the CR-MA hypothesis, high rated firms have a preference for low gearing ratios, which appears to be due to the higher incentive to maintain their credit ratings. The CR-CS hypothesis implies that for high rated firms, the benefits of high ratings outweigh the benefits of high leverage. High rated firms arguably have low cost of capital, easier access to the commercial paper market, favourable terms and conditions in debt contracts, access to alternative sources of financing and they can also benefit from greater financial flexibility due to their high credit ratings. Apart from these financial benefits, high rated firms can also enjoy the non-financial benefits of high ratings, such as a good managerial reputation in the labour market, employee loyalty and favorable suppliers’ terms and conditions. As high rated firms, over a period of time, have gained a market reputation for being successful and highly creditworthy firms, they should therefore have more incentive to maintain their credit ratings than other rated firms. As predicted by the CR-CR hypothesis, these benefits of high credit ratings induce high rated firms to choose low gearing ratios.
It should be noted that the implications of the CR-CS hypothesis differ from traditional trade-off theory. The trade-off theory, which predicts a negative relationship between risk and leverage, implies that high rated firms, which arguably have low chances of bankruptcy, have high leverage. However, the implications of the CR-CS hypothesis are distinct from the trade-off theory as it suggests that the benefits of high ratings are material for high rated firms, which would lead high rated firms to choose low levels of gearing.
Mid rated firms seem to have a preference for high gearing ratios. Given that these firms have better credit ratings than low rated firms, they have less constrained access than low rated firms. Despite the fact that the CR-CS hypothesis predicts that considerations for ratings in capital structure decisions should be somewhat similar across different rating
148 levels, mid-rated firms arguably have less concern for their credit ratings. Mid-rated firms would be likely to require large changes in their capital structures to get into a category where they would benefit from being top rated. Moreover, the high gearing ratios of mid rated firms also suggest that they are stable firms with a limited risk of falling towards low ratings. As these firms are far from low and high ratings categories, their good credit ratings help them in accessing more debt. This imply that the results of prior empirical studies (Faulkender and Petersen, 2006; Judge and Mateus, 2009; Mittoo and Zhang, 2010) are dominated by mid-rated firms. Overall, the results of the present study suggest that credit ratings have a non-linear relationship with the capital structures of firms.
An interesting observation about the capital structures of rated firms is that the characteristics, which appear to affect the capital structures of rated firms, are different from those suggested by prior studies. For example, for the rated firms’ sample, large firms have less debt and the coefficient is significantly different from zero. This is inconsistent with prior literature, which suggests that larger firms have higher leverage due to their better access to debt markets, more remote chance of failure and economies of scale. One possible reason could be due to the positive relationship between size and leverage; large firms are expected to have high credit ratings and therefore have low leverage in their capital structure. Nevertheless, it is noted that even after controlling for the size of the firm, which has previously been used as a proxy for firms’ access to capital markets and chances of bankruptcy (see for example, Titman and Wessels, 1988; Rajan and Zingales, 1995; Bevan and Danbolt, 2002, amongst others), the credit ratings variables maintain their statistical significance. The size variable does not have any serious multicollinearity or abnormal correlation with the credit ratings as suggested by the bi-variate analysis, VIF and Eigenvalues presented in Section 6.1.2, Appendices 6A and 6B. Consolidating the findings of correlation, multicollinearity statistics and the regression output of Table 6.9, the results suggest that firms’ credit ratings offer some unique benefits and costs to the firms at each rating level, which can play a role in determining the capital structures of the firms.
Similarly, significant at the 5% level, profitability has a positive sign, which means that when rated firms are profitable they have higher leverage. Although this is inconsistent
with the pecking-order theory and prior empirical evidence (Jong et al., 2008; Rajan and
Zingales, 1995; Bevan and Danbolt, 2002), the results are more supportive of the trade-off theory. Profitable firms are less likely to fail but they may have to pay high corporate taxes
149 because of their high profits. Such firms have therefore a higher incentive to safely employ more leverage, in order to reduce their tax burdens. As hypothesised, the coefficient of tangibility has a positive relationship with the leverage of the firms. The significance and sign of the coefficient signifies the role of collateralisable assets towards the leverage.
Inconsistently with Myers (1984), rated firms with high growth opportunities are shown to be likely to have high debt as well. It seems that, when firms have an advantageous position in the market by possessing credit ratings, they are likely to behave differently when faced with a higher growth opportunities set. It indicates that rated firms possibly have less underinvestment problems that lead these firms to choose high gearing in the presence of growth opportunities. The relationship of liquidity and leverage is also found to be negative and significant at the 1% level which is in line with past studies such as
Ozkan (2001) and Deesomsak et al. (2004), who report a negative relationship. Most of the
industry dummies on the other hand are insignificant, which is either due to the small sample size or implies that the rated firms generally have similar capital structures across the industries.
The inclusion of utility firms in the sample may receive some criticism. The utility firms, being largely regulated, are governed under a different set of regulations than the non- regulated firms. Therefore, they are likely to have different capital structures, which cannot be directly compared with other rated firms. Previous studies investigating firms’ capital structures tend to exclude utility firms from their sample (Stohs and Mauer, 1996; Ozkan, 2000, 2001 and 2002; Shyam-Sunder and Myers, 1999). To be consistent with the prior literature, Column 3 presents the regression results for Model (1), estimated only for the sample without the utility firms. It can be noted that the results remain qualitatively similar those reported in to Column 2. However, the fit of the model has considerably improved.
The CR and CR2 have a similar sign to Column 3 and remain statistically significant at the
1% level. Moreover, the control variables also possess similar signs and significance levels, suggesting that the inclusion of the utility firms has not distorted the previous analysis. Also worth mentioning here is that the utility firms also show that the credit ratings have a non-linear relationship with leverage.
Overall, the results of the present section provide strong support to accept hypotheses H1a,
H1b andH1c that the implications of the CR-CS hypothesis induce a non-linear relationship
150 to the factors proposed by traditional theories of capital structure, credit ratings seem to have a higher contribution in explaining the capital structure decisions of rated firms. The results of the control variables seem to indicate that rated firms have a different capital structure and are affected by the same firm characteristics in different ways as well. Caution has to be exercised when attempting to understand the capital structure of such firms, as this small group has unique characteristics which may not be observed collectively with other firms and may require a separate analysis.