3. RESEARCH DESIGN
3.3 Empirical Model
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
20
(2) The frequency of interest payments per year (INTERESTPAY)
INTERESTPAY is a dummy variable in this study. I find some firms pay their interests only at maturity and period is more than one year, so I control for the number that a firm pays interest in a year.
(3) Ways to pay off (METHODPAY)
There are many ways to pay off, including lump sum, installment, and other ways. I expect that different way will lead to different loan terms.
Therefore, I control for this variable.
Detailed definitions of the below variables are summarized in Appendix A.
3.3 Empirical Model
This study extends the literature on bond by investigating the relationship between AC experts, the price, and non-price terms in bonds. I consider four important loan terms commonly discussed in prior studies, including bond rate, bond maturity, convertible bond issuance, and collateral requirement. I test my hypotheses based on the following regression model:
𝑪𝒐𝑫 = 𝛼0+ 𝛼1𝐹𝐼𝑁𝐸𝑋𝑃𝑖𝑡 + 𝛼2 𝐴𝐶𝐶𝐸𝑋𝑃𝑖𝑡 + 𝛴𝛼𝑖 𝐺𝑜𝑣𝑒𝑟𝑛𝑎𝑛𝑐𝑒 𝐶ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐𝑠𝑖𝑡 +𝛴𝛼𝑖 𝐹𝑖𝑟𝑚 𝐶ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐𝑠𝑖𝑡 + 𝛴𝛼𝑖 𝐵𝑜𝑛𝑑 𝐶ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐𝑠𝑖𝑡+ 𝜀
The main interest in this study is four AC expertise indicator variables
(FINEXP-D, ACCEXP-(FINEXP-D, FINEXP-O, ACCEXP-O). I also control for a set of control variables
related to loan terms. To be consistent with my all hypotheses, I expect 𝛼1 and 𝛼2 to‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
21
be negative when the dependent variables are RATE, CONVERTIBLE, and
COLLATERAL; but to be positive when the dependent variable is MATURITY.
3.3.1 The Impact of AC Experts on Bond Rate
In this study, I use three different regression models to examine the association between experts and loan terms. For each model using different dependent variables, I include a set of control variables documented to be associated with loan terms such as government regulations, characteristics of AC and board, and financial information of firms. To evaluate the impact of experts on loan terms, I specify the three regression models as follows.
1. Model 1
𝑅𝐴𝑇𝐸 = 𝛼0+ 𝛼1FINEXP + 𝛼2𝐴𝐶𝑆𝐼𝑍𝐸 + 𝛼3 𝐴𝐶𝑀𝐸𝐸𝑇 + 𝛼4 𝐴𝑇𝑇𝐸𝑁𝐷𝐴𝑁𝐶𝐸 +𝛼5𝐵𝐷𝑆𝐼𝑍𝐸 + 𝛼6 𝐵𝐷𝑀𝐸𝐸𝑇 + 𝛼7𝐴𝐶𝐿𝐴𝑊 + 𝛼8𝑅𝑂𝐴 + 𝛼9𝐸𝐵𝐼𝑇𝐷𝐴 +𝛼10𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛼11𝐵𝐼𝐺4 + 𝛼12𝐼𝑁𝑇𝐸𝑅𝐸𝑆𝑇𝑃𝐴𝑌
+𝛼13𝑀𝐸𝑇𝐻𝑂𝐷𝑃𝐴𝑌 + 𝛼14𝐶𝑂𝐿𝐿𝐴𝑇𝐸𝑅𝐴𝐿 + 𝜀
(1) 2. Model 2
𝑅𝐴𝑇𝐸 = 𝛼0+ 𝛼1ACCEXP + 𝛼2𝐴𝐶𝑆𝐼𝑍𝐸 + 𝛼3 𝐴𝐶𝑀𝐸𝐸𝑇 + 𝛼4 𝐴𝑇𝑇𝐸𝑁𝐷𝐴𝑁𝐶𝐸 +𝛼5𝐵𝐷𝑆𝐼𝑍𝐸 + 𝛼6 𝐵𝐷𝑀𝐸𝐸𝑇 + 𝛼7𝐴𝐶𝐿𝐴𝑊 + 𝛼8𝑅𝑂𝐴 + 𝛼9𝐸𝐵𝐼𝑇𝐷𝐴 + 𝛼10𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛼11𝐵𝐼𝐺4 + 𝛼12𝐼𝑁𝑇𝐸𝑅𝐸𝑆𝑇𝑃𝐴𝑌
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
22
+𝛼13𝑀𝐸𝑇𝐻𝑂𝐷𝑃𝐴𝑌 + 𝛼14𝐶𝑂𝐿𝐿𝐴𝑇𝐸𝑅𝐴𝐿 + 𝜀
(2) 3. Model 3
𝑅𝐴𝑇𝐸 = 𝛼0+ 𝛼1FINEXPONLY+ 𝛼2ACCEXPONLY+ 𝛼3𝐴𝐶𝑆𝐼𝑍𝐸 + 𝛼4 𝐴𝐶𝑀𝐸𝐸𝑇 +𝛼5 𝐴𝑇𝑇𝐸𝑁𝐷𝐴𝑁𝐶𝐸 + 𝛼6𝐵𝐷𝑆𝐼𝑍𝐸 + 𝛼7 𝐵𝐷𝑀𝐸𝐸𝑇 + 𝛼8𝐴𝐶𝐿𝐴𝑊 + 𝛼9𝑅𝑂𝐴
+𝛼10𝐸𝐵𝐼𝑇𝐷𝐴 + 𝛼11𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛼12𝐵𝐼𝐺4 + 𝛼13𝐼𝑁𝑇𝐸𝑅𝐸𝑆𝑇𝑃𝐴𝑌 +𝛼14𝑀𝐸𝑇𝐻𝑂𝐷𝑃𝐴𝑌 + 𝛼15𝐶𝑂𝐿𝐿𝐴𝑇𝐸𝑅𝐴𝐿 + 𝜀
(3)
RATE is the stated rate on bond when bond is issued. This measure is equal to the
interest the borrower pays to the lenders. The other variables are defined in the prior section and summarize in appendix A. Over recent decades, numerous studies document the relationship between financial reporting quality and loan terms (Anderson et al., 2004; Bharath, J. Sunder, and S. V. Sunder, 2008; Paige-Fields, Fraser, and Subrahmanyam, 2012). More recently, Chan et al. (2013) extend those arguments, and document that clawback provisions increase financial reporting quality and reduce the degree of information asymmetry between borrowers and lenders, where leads to a decrease in the interest rates charged by lenders. To be consistent with third hypothesis, H3, I expect the signs of 𝛼1 and 𝛼2 in Model 3 to be negative, implying that bondholders are more likely to respond favorably to firms with financial or accounting experts on the AC by charging lower bond rate. Moreover, in Model 3, to be consistent with third hypothesis, H3, I also expect 𝛼1 to be less than 𝛼2 . This implication is that the impact of accounting experts on the rate would be larger than the impact of financial experts.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
23
3.3.2 The Impact of AC Experts on Bond Maturity
To evaluate the impact of AC experts on loan terms, I specify the following equations:
1. Model 1
𝑀𝐴𝑇𝑈𝑅𝐼𝑇𝑌 = 𝛼0+ 𝛼1FINEXP + 𝛼2𝐴𝐶𝑆𝐼𝑍𝐸 + 𝛼3 𝐴𝐶𝑀𝐸𝐸𝑇 + 𝛼4 𝐴𝑇𝑇𝐸𝑁𝐷𝐴𝑁𝐶 +𝛼5𝐵𝐷𝑆𝐼𝑍𝐸 + 𝛼6 𝐵𝐷𝑀𝐸𝐸𝑇 + 𝛼7𝐴𝐶𝐿𝐴𝑊 + 𝛼8𝑅𝑂𝐴 + 𝛼9𝐵𝐼𝐺4 +𝛼10𝑅𝐹 + 𝜀
(4) 2. Model 2
𝑀𝐴𝑇𝑈𝑅𝐼𝑇𝑌 = 𝛼0+ 𝛼1𝐴𝐶𝐶𝐸𝑋𝑃 + 𝛼2𝐴𝐶𝑆𝐼𝑍𝐸 + 𝛼3 𝐴𝐶𝑀𝐸𝐸𝑇 + 𝛼4 𝐴𝑇𝑇𝐸𝑁𝐷𝐴𝑁𝐶𝐸 +𝛼5𝐵𝐷𝑆𝐼𝑍𝐸 + 𝛼6 𝐵𝐷𝑀𝐸𝐸𝑇 + 𝛼7𝐴𝐶𝐿𝐴𝑊 + 𝛼8𝑅𝑂𝐴 + 𝛼9𝐵𝐼𝐺4
+𝛼10𝑅𝐹 + 𝜀
(5) 3. Model 3
𝑀𝐴𝑇𝑈𝑅𝐼𝑇𝑌 = 𝛼0 + 𝛼1𝐹𝐼𝑁𝐸𝑋𝑃𝑂𝑁𝐿𝑌+ α2𝐴𝐶𝐶𝐸𝑋𝑃𝑂𝑁𝐿𝑌+ 𝛼3𝐴𝐶𝑆𝐼𝑍𝐸 +𝛼4 𝐴𝐶𝑀𝐸𝐸𝑇 + 𝛼5 𝐴𝑇𝑇𝐸𝑁𝐷𝐴𝑁𝐶𝐸 + 𝛼6𝐵𝐷𝑆𝐼𝑍𝐸 +𝛼7 𝐵𝐷𝑀𝐸𝐸𝑇 + 𝛼8𝐴𝐶𝐿𝐴𝑊 + 𝛼9𝑅𝑂𝐴
+𝛼10𝐵𝐼𝐺4 + 𝛼11𝑅𝐹 + 𝜀
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
24
(6)
MATURITY is the length of time between bond’s issue date and maturity date. The
other variables are defined in prior section and summarized in appendix A. The extent literature documents some degree of association between maturity and borrower’s information quality (Chan et al., 2013). In addition, Bharath and Dittmar (2006) also indicate that the lower quality borrowers face a higher economic cost in securing the funds for shorter maturity. These prior arguments imply that there is a negative relationship between maturity and debt cost. To be consistent with third hypothesis, H3, I expect the signs of 𝛼1 and 𝛼2 to be positive, implying that bondholders are more likely to offer a longer maturity to those borrowers with financial or accounting experts on the AC. Moreover, to be consistent with third hypothesis, H3, I also expect 𝛼1 to be less than 𝛼2 . This implication is that the impact of accounting experts on maturity would be larger than the impact of financial experts.
3.3.3 The Impact of AC Experts on Convertible Bonds Issuance
To evaluate the impact of AC experts on loan terms, I specify the following equations:
1. Model 1
𝐶𝐻𝐴𝑁𝐺𝐸 = 𝛼0+ 𝛼1𝐹𝐼𝑁𝐸𝑋𝑃 + 𝛼2𝐴𝐶𝑆𝐼𝑍𝐸 + 𝛼3 𝐴𝐶𝑀𝐸𝐸𝑇 + 𝛼4 𝐵𝐷𝑆IZE +𝛼5𝐵𝐷𝑀𝐸𝐸𝑇 + 𝛼6 𝐴𝐶𝐿𝐴𝑊 + 𝛼7𝑅𝑂𝐴 + 𝛼8𝐸𝐵𝐼𝑇𝐷𝐴 + 𝜀
(7)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
25
2. Model 2
𝐶𝐻𝐴𝑁𝐺𝐸 = 𝛼0+ 𝛼1𝐴𝐶𝐶𝐸𝑋𝑃 + 𝛼2𝐴𝐶𝑆𝐼𝑍𝐸 + 𝛼3 𝐴𝐶𝑀𝐸𝐸𝑇 + 𝛼4 𝐵𝐷𝑆𝐼𝑍𝐸 +𝛼5𝐵𝐷𝑀𝐸𝐸𝑇 + 𝛼6 𝐴𝐶𝐿𝐴𝑊 + 𝛼7𝑅𝑂𝐴 + 𝛼8𝐸𝐵𝐼𝑇𝐷𝐴 + 𝜀
(8) 3. Model 3
𝐶𝐻𝐴𝑁𝐺𝐸 = 𝛼0 + 𝛼1𝐹𝐼𝑁𝐸𝑋𝑃𝑂𝑁𝐿𝑌+ 𝛼2𝐴𝐶𝐶𝐸𝑋𝑃𝑂𝑁𝐿𝑌+ 𝛼3𝐴𝐶𝑆𝐼𝑍𝐸 + 𝛼4 𝐴𝐶𝑀𝐸𝐸𝑇 +𝛼5 𝐵𝐷𝑆𝐼𝑍𝐸 + 𝛼6𝐵𝐷𝑀𝐸𝐸𝑇 + 𝛼7 𝐴𝐶𝐿𝐴𝑊 + 𝛼8𝑅𝑂𝐴 + 𝛼9𝐸𝐵𝐼𝑇𝐷𝐴 + 𝜀
(9)
CNANGE is a dummy variable equals to 1 if the bond is issued with conversion
right that is a value-added component of convertible bond as the bondholder’s right to convert the bond into common stock in the issuing company, and 0 otherwise. The other variables are defined in prior section and summarize in appendix A. Prior studies suggest that riskier firms have higher propensity to issue the convertible bonds to attract creditors and investors (Brigham, 1966; Brennan and Schwartz, 1977). These prior arguments imply that there is a relationship between convertible bond issuance and debt cost. To be consistent with third hypothesis, H3, I expect the signs of 𝛼1 and 𝛼2 to be negative, implying that bondholders are less likely to require conversion right of bond to those borrowers with financial or accounting experts on the AC. Moreover, to be consistent with third hypothesis, H3, I also expect 𝛼1 to be less than 𝛼2 . This implication is that the impact of accounting experts on non-convertible bond issuance would be larger than the impact of financial experts.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
26
3.3.4 The Impact of AC Experts on Collateral Requirement
To evaluate the impact of AC experts on loan terms, I specify the following equations:
1. Model 1
𝐶𝑂𝐿𝐿𝐴𝑇𝐸𝑅𝐴𝐿 = 𝛼0+ 𝛼1𝐹𝐼𝑁𝐸𝑋𝑃 + 𝛼2𝐴𝐶𝑀𝐸𝐸𝑇 + 𝛼3 𝐴𝑇𝑇𝐸𝐷𝐴𝑁𝐶𝐸 +𝛼4 𝐵𝐷𝑆𝐼𝑍𝐸 + 𝛼5𝐵𝐷𝑀𝐸𝐸𝑇 + 𝛼6 𝐴𝐶𝐿𝐴𝑊 + 𝛼7𝑆𝐼𝑍𝐸 +𝛼8𝑅𝑂𝐴 + 𝛼9𝑅𝐹 + 𝜀
(10)
2. Model 2
𝐶𝑂𝐿𝐿𝐴𝑇𝐸𝑅𝐴𝐿 = 𝛼0+ 𝛼1ACCEXP + 𝛼2𝐴𝐶𝑀𝐸𝐸𝑇 + 𝛼3 𝐴𝑇𝑇𝐸𝐷𝐴𝑁𝐶𝐸 +𝛼4 𝐵𝐷𝑆𝐼𝑍𝐸 + 𝛼5𝐵𝐷𝑀𝐸𝐸𝑇 + 𝛼6 𝐴𝐶𝐿𝐴 + 𝛼7𝑆𝐼𝑍𝐸 +𝛼8𝑅𝑂𝐴 + 𝛼9𝑅𝐹 + 𝜀
(11) 3. Model 3
𝐶𝑂𝐿𝐿𝐴𝑇𝐸𝑅𝐴𝐿 = 𝛼0+ 𝛼1FINEXPONLY+ 𝛼2ACCEXPONLY+ 𝛼3𝐴𝐶𝑀𝐸𝐸𝑇 +𝛼4 𝐴𝑇𝑇𝐸𝐷𝐴𝑁𝐶𝐸 + 𝛼5 𝐵𝐷𝑆𝐼𝑍𝐸 + 𝛼6𝐵𝐷𝑀𝐸𝐸𝑇
+𝛼7 𝐴𝐶𝐿𝐴𝑊 + 𝛼8𝑆𝐼𝑍𝐸 + 𝛼9𝑅𝑂𝐴 + 𝛼10𝑅𝐹 + 𝜀
(12)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
27
COLLATERAL is a dummy variable equal to 1 if the bond is collateralized, and 0
otherwise. The other variables are defined in prior section and summarize in appendix A. Paige-Fields et al. (2012) show that borrowers with more experience and more diverse boards are less likely to be required to provide collateral in loan contract. In addition, Chan et al. (2013) document a negative relation between financial quality and collateral requirement in loan contract. Also, Bharath et al. (2006) indicate that poorer accounting quality has a significant economic effect on borrowers in terms of higher collateral. Accordingly, these prior arguments imply that there is a positive relationship between collateral requirement and debt cost. To be consistent with third hypothesis, H3, I expect the signs of 𝛼1 and 𝛼2 to be negative, implying that bondholders are less likely to require collateral to those borrowers with financial or accounting experts on the AC. Moreover, to be consistent with third hypothesis, H3, I also expect 𝛼1 to be less than 𝛼2 . This implication is that the impact of accounting experts on non-collateral requirement would be larger than the impact of financial experts.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
28