4. EMPIRICAL RESULTS
4.3 Regression Results
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
30
4.3 Regression Results
4.3.1 The Impact of AC Experts on Bond Rate
In this section, I examine the relative effectiveness of the accounting experts in reducing the bond rate. Table 4 shows the results from the regression analysis of financial and accounting experts, as well as the control variables, on bond rate. In all regressions, we use a two-tailed test for the coefficients. Most importantly, most results of variance inflation factor (VIF) are less than 10, suggesting that there is no strong linear relation in each regression.
In Table 4, three models are reported for comparative benchmarking purposes.
Model 1 of Table 4 reports that there is a negative but insignificant association between the financial experts and bond rate, implying that the presence of at least one financial expert on the AC is not significantly considered by bondholders. Model 2 explores the effect of the AC with at least one accounting expert on bond rate. Consistent with my second hypothesis H2, the result reveals that the coefficient of ACCEXP-D is negative and significant at 10% level, suggesting that accounting experts are associated with the lower bond rate. The results present that, on average, the rate of bond issued by firm with at least one accounting expert on AC is about 0.153percent lower than firm without any accounting expert. The results, coupled with the findings of Model 2, provide preliminary evidence that the differential bond rate is driven primarily by the accounting experts rather than financial experts.
Model 3 is the main interest in my study. As predicted in H2, Model 2 reveals that the coefficient of ACCEXP-O is negative and significant, indicating that accounting experts are negatively associated with rate and significantly at 5% level. It also presents that the coefficient of FINEXP-O is negative but insignificant, suggesting
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
31
that only financial experts in the AC are not associated with the bond rate. These results provide supporting evidence of my third hypothesis H3 in this study.
For these three regressions in Table 4, the coefficients of ACSIZE are insignificant but negative, consistent with a prior study (Felo et al., 2003). The results here indicate that after controlling for accounting and financial experts, there is no association between AC size and loan term. The coefficients of ACMEET are insignificant but positive, suggesting that AC meetings are insignificantly associated with rate. In addition, the coefficients of ATTENDANCE and BDSIZE are both insignificantly associated with rate. However, the negative and significant coefficient of BDSIZE indicates that the board size is significantly negative associated with rate since, consistent with prior research (Anderson et al., 2004). The coefficients of ACLAW, ROA, EBITDA, LEVERAGE and BIG4 are insignificant, suggesting that whether firms are required to establish the AC and the firm specific characteristics are insignificantly associated with rate after controlling for accounting and financial experts.
To summarize, the results in Table 4 show the evidence that compared to firms with at least one financial expert or only financial experts on the AC, firms with at least one accounting expert or only accounting experts on AC have better loan term by lower rate. In addition, the results also provide the evidence that the number of board meeting is significantly associated with lower rate.
4.3.2 The Impact of AC Experts on Bond Maturity
In this section, I examine the relative effectiveness of experts in increasing the bond maturity. Table 5 presents the results from the regression analysis of financial and accounting experts, as well as the control variables, on bond maturity. In all regressions,
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
32
we use a two-tailed test for the coefficients. Most importantly, most results of VIF are less than 10, suggesting that there is no strong linear relation in each regression.
In Table 5, three models are reported for comparative benchmarking purposes.
Model 1 of Table 5 reports that there is a positive but insignificant association between the financial experts and bond maturity, implying that the presence of at least one financial expert on the AC is insignificantly related to shorter maturity. Model 2 explores the impact of the AC with at least one accounting expert on bond maturity.
Consistent with my second hypothesis H2, the result reveals that the coefficient of ACCEXP-D is positive and significant at 5% level, suggesting that accounting experts are associated with the longer maturity of bond. The results present that, on average, the maturity of bond published by firm with at least one accounting expert on AC is about 1.1percent lower than maturity by firm without any accounting expert. The results, coupled with the findings of Model 2, provide preliminary evidence that the differential maturity is driven primarily by the accounting experts rather than financial experts.
Model 3 is the main interest in my study. As predicted in H2, Model 2 reveals that the coefficient of ACCEXP-O is positive and significant at 5% level, indicating that accounting experts are significantly associated with bond maturity. It also presents that the coefficient of FINEXP-O is positive but insignificant, suggesting that only financial experts on AC are not associated with bond maturity. These results provide supporting evidence of my third hypothesis H3.
For these three regressions in Table 5, in governance characteristics variables, except the coefficients of BDSIZE, the coefficients of ACSIZE, ACMEET, ATTENDENACE, and BDMEET are all insignificant. This result indicates that after controlling for accounting and financial experts, there is no association between AC
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
33
size, number of board and AC meetings, or attendance of members and loan term. The coefficients of BDSIZE are significantly positive, suggesting that board size is significantly associated with maturity. This result is consistent with prior study (Bharath et al. 2006). In addition, the coefficients of ACLAW show that whether firms are required to establish an AC is significant and positive, suggesting that required establishment is associated with maturity. And the coefficients of ROA and BIG4 are both insignificant. Finally, the coefficients of RF are negative and significant, indicating that there is significantly negative relationship between the risk-free rate of interest and maturity of bond.
To summarize, the results in Table 5 show the evidence that compared to firms with at least one financial expert or only financial experts on the AC, firms with at least one accounting expert or only accounting experts on AC have better loan term by longer maturity. In addition, the results also provide the evidence that the bigger board size and the lower free-risk of rate contribute to better loan terms.
4.3.3 The Impact of AC Experts on Convertible Bonds Issuance
In this section, I examine the relative effectiveness of experts in reducing the propensity of convertible bond issuance. Table 5 presents the results from the regression analysis of financial and accounting experts, as well as the control variables, on convertible bond. In all regressions, we use a two-tailed test for the coefficients. Most importantly, most results of VIF are less than 10, suggesting that there is no strong linear relation in each regression.
In Table 6, three models are reported for comparative benchmarking purposes.
Model 1 of Table 6 reports that there is a positive but insignificant association between the financial experts and convertible bond, implying that the presence of at least one
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
34
financial expert on the AC is insignificantly related to convertible bonds issuance from bondholders’ point of view. Model 2 explores the impact of the AC with at least one accounting expert on convertible bond issuance. Consistent with my second hypothesis H2, the result reveals that the coefficient of ACCEXP-D is negative and significant at 5% level, suggesting that accounting experts are associated with the nonconvertible bond. The results present that, on average, the possibility of convertible bond published by firm with at least one accounting expert on AC is about 5.9percent lower than without any accounting expert. The results, coupled with the findings of Model 2, provide preliminary evidence that whether bond is convertible is driven primarily by accounting experts rather than financial experts.
Model 3 is the main interest in my study. As predicted in H2, Model 2 reveals that the coefficient of ACCEXP-O is negative and significant at 10% level, indicating that accounting experts are significantly associated with convertible bond issuance. It also presents that the coefficient of FINEXP-O is positive but insignificant, suggesting that only financial experts in the AC are not associated with the convertible bond. These results provide supporting evidence of my third hypothesis H3.
For these three regressions in Table 6, in governance characteristics variables, except the coefficients of BDSIZE, the coefficients of ACSIZE, ACMEET, and BDMEET are all insignificant. This result indicates that after controlling for accounting and financial experts, there is no significant association between AC size or number of board and AC meetings and loan terms. The coefficients of BDSIZE are significantly negative, suggesting that board size is significantly associated with whether convertible bonds are issued. In addition, the coefficients of ACLAW are significant and negative, suggesting that after controlling for accounting and financial experts, firms required to
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
35
establish an AC are less likely to issue the convertible bonds. And the coefficients of ROA are insignificant, indicating that there is insignificant relationship between return on assets and whether firms issue the convertible bonds. Finally, the coefficients of EBITDA are negative and significant, indicating that there is significantly negative relationship between the earnings before interest, tax, depreciation, and amortization and whether firms issue the convertible bonds.
To summarize, the results in Table 6 show the evidence that compared to firms with at least one financial expert or only financial experts on AC, firms with at least one accounting expert or only accounting experts on AC have better loan term by lower propensity to issue the convertible bonds. In addition, the results also provide the evidence that the bigger board size, required establishment of an AC, and the higher earnings before interest, tax, depreciation, and amortization contribute to better loan terms.
4.3.4 The Impact of AC Experts on Collateral Requirement
In this section, I examine the relative effectiveness of experts in reducing the requirement of collateral. Table 7 presents the results from the regression analysis of financial and accounting experts, as well as the control variables, on collateral requirement. In all regressions, we use a two-tailed test for the coefficients. Most importantly, most results of VIF are less than 10, suggesting that there is no strong linear relation in each regression.
In Table 7, three models are reported for comparative benchmarking purposes.
Model 1 of Table 7 reports that, consistent with my first hypothesis H3, there is a positive but insignificant association between the financial experts and collateral requirement, implying that the presence of at least one financial expert on the AC is not
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
36
significantly related to collateral requirement. Model 2 explores the impact of the AC with at least one accounting expert on collateral requirement. Consistent with my second hypothesis H2, the result reveals that the coefficient of ACCEXP-D is negative and significant at 10% level, suggesting that accounting experts are associated with the nonconvertible bond. The results present that, on average, the propensity of collateral requirement on firms with at least one accounting expert on AC is about 4.4percent lower than without any accounting expert. The results, coupled with the findings of Model 2, provide preliminary evidence that collateral requirement is driven primarily by accounting experts rather than financial experts.
Model 3 is the main interest in my study. As predicted in H2, Model 2 reveals that the coefficient of ACCEXP-O is negative and significant, indicating that accounting experts are negatively associated with collateral and significantly at 10% level. It also presents that the coefficient of FINEXP-O is positive but insignificant, suggesting that only financial experts in the AC are not associated with collateral. F-test shows that there is significant difference in the coefficient for FINEXP-O and ACCEXP-O. More importantly, collateral is attributable to AC with only accounting experts rather than with only financial experts. These results provide supporting evidence of my third hypothesis H3.
For these three regressions in Table 7, in governance characteristics variables, the coefficients of ACMEET, ATTENDANCE, BDSIZE, and BDMEET are all insignificant. This result indicates that after controlling for accounting and financial experts, there is no significant association between board size, number of board and AC meetings, or average attendance of AC members and the loan term. In addition, the coefficients of ACLAW are insignificant and negative, suggesting that after controlling
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
37
for accounting and financial experts, firms required to establish an AC are less likely to be required the collaterals. And the coefficients of SIZE and ROA are insignificant, indicating that there is insignificant relationship between firm size or return on assets and required collateral. Finally, the coefficients of RF are insignificant, indicating that the free-risk of rate is insignificant associated with whether collaterals are required.
In sum, the results in Table 7 show the evidence that compared to firms with at least one financial expert or only financial experts on AC, firms with at least one accounting expert or only accounting experts on AC have better loan terms by lower propensity to be required of the collaterals.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
38