Asset redeployability, competition, and cost of debt

(1)

1

Master Thesis

Asset redeployability, competition, and cost of debt

University of Amsterdam

Amsterdam Business School

MSc Finance, Quantitative Finance track

Sijing Li (11252014)

Date: 01-07-2018

Supervisor: Jeroen Ligterink

(2)

2

Statement of originality

This document is written by Student Sijing Li, who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

This paper is aiming to study the effect of competition on firms’ cost of debt. It is also the first to study the interactive effect of competition and asset redeployability on the cost of debt. I found positive and significant effects of both competition and asset redeployability on the cost of debt. Besides that, firms being in the most competitive industries are on average pay higher rates than firms being in the less competitive industries. Given that firms being in the group with the most redeployable assets, the positive effect of competitive environment on the cost of debt through the channel of rival firm threats is weakened, and the total effect of competition turns out to be negative. In other words, if firms’ assets are the most redeployable assets in the group, the cost of debt will be lower even the industry competitiveness increases. These findings suggest that the after considering the interactive effect of asset redeployability and competition, the adverse effect of competition on the cost of debt through the channel of potential assets buyers in liquidation outweighs the positive effect through the channel of rival firm threats. Thus, more redeployable assets strengthen the adverse effect of competition on the cost of debt.

(4)

4

1. Introduction

The intensity of industry competition is a crucial factor for the operations of firms. Some firms face less competition, while other firms belong to industries with intense competition. Firm performance not only depends on their characteristics but also the competition from the industry counts. When firms borrowing from external financing sources, firm performance will be a vital factor for creditors pricing the debt. Since competitive environment impacts the firm performance, there could exist a linkage between competition and the cost of debt. The determinants of loan pricing are diverse, competition faced by firms is one of the potential determinants which have not received too much attention in the literature.

The basic idea of my research question is to investigate the role of competition in determining the firm’s cost of debt. Two factors are of utmost importance when banks provide capital to firms and price the loans, which are the likelihood of firms defaulting on loans and the liquidation value can be recovered in the case of default. Since market competition can affect both the liquidation value of assets and corporate default risk, competition could be a determinant for the cost of debt. In general, competition potentially determines firms’ cost of debt in two ways. On the one hand, more intense market competition will provide a more competitive secondary market for firm’s assets in liquidation. Therefore, firms operating in the more competitive industry are expected to have higher liquidation value as a result of abundant potential buyers for firm’s assets in liquidation with more competition. Banks are more likely to allow those firms to issue more debts or lower the required return by charging lower interests on debt given a higher firm’s liquidation value. On the other hand, firms in the more competitive industry face more threats from rival firms, which results in more business risk for incumbent firms (Bolton et al., 1990). Banks take this factor into account and will tighten the terms of loans accordingly. In turn, taking leverage will be more expensive for those firms. In sum, due to the fact of these two opposite effects of competition on firm’s cost of debt, it is interesting to study which effect is more significant by performing an empirical study. Previous literature mainly argues how the total effect of competition on the cost of debt, seldom of them argue the effect separately from these two channels.

(6)

6

Also, competition is not the only critical variable of interests in the research question. Asset redeployability is the other element I want to include in the study. Adding asset redeployability is not solely for making the empirical analysis more productive, most importantly, it is because of its link with the competitive environment. Competition can determine the liquidation value of the firm by affecting the availability of potential asset buyers, while asset redeployability is another crucial dimension of determining liquidation value of assets. Firms with more redeployable assets are more likely to receive financing contracts with better terms, e.g., longer maturity and lower interest rates (Benmelech et al., 2005). Because of the deterministic effect of the liquidation value of assets in financial contracting, the indirect effect of asset redeployability on firm’s borrowing cost is not negligible. Since both competition and assets redeployability can determine the liquidation value of assets in some way, it is worth investigating the interactive effect of asset redeployability and competition on the cost of borrowing.

In sum, this paper aims to investigate how asset redeployability and competition, individually and interactively, determine firms’ cost of debt. Both asset redeployability and the availability of potential purchasers for firm’s assets are two primary dimensions for evaluating the liquidation value of firm’s assets, while the assets’ liquidation value is of great importance for the pricing of debt. Especially when contracts are incomplete, and there are transaction costs (Benmelech et al., 2005), higher liquidation value of firms gives creditors more securities to recover the debt. In turn, creditors are willing to lend at a lower rate given a relative higher firm’s liquidation value. Moreover, not only the effect of competitive environment and asset redeployability on the cost of debt will be studied, but also some analysis on how these variables determine the lender structure, the upfront fee, and the annual fee will be investigated.

This research question contributes to the big question that how competitive environment influences the firm’s financing cost. This empirical study differs from previous studies in the following aspects. Firstly, as mentioned above, competition determines the cost of borrowing from two channels. It worth investigating whether the adverse effect of threats from rival firms will outweigh the positive effect of more

(7)

7

potential buyers in liquidation or the other way around. The effect of competitive environment on the firm’s cost of debt have received little attention. By analyzing the effect of competition from different channels, a new angle of analysis on the role of competition in determining the corporate cost of borrowing will be presented. In Valta’s (2012) paper, he also studies the role of competition in determining firms’ cost of debt. However, the focus in his paper is how different measures used for competition affect the results. The part explaining the effect of competition through two channels are missing. Secondly, although there exist some literature analyzing the asset redeployability or competition separately, seldom of previous literature study the interactive effect of asset redeployability and competition on firm’s cost of borrowing. However, the link between these two elements makes it worth investigating. Although in Valta’s (2012) paper he includes a proxy for asset specificity, measured by the proportion of machinery and equipment to total assets, it cannot fully capture the redeployability of firms’ assets. In my empirical study, not only the effect of competition will be analyzed from two aspects, but also a more accurate proxy for asset redeployability will be used.

The primary empirical study is based on a large sample of loan contracts, which are borrowed by U.S. firms both existed in COMPUSTAT and LPC DealScan database from the year 1997 to 2015. I find substantial evidence that banks charge higher loan price as the competitiveness of industry increases. Banks also charge a significantly higher price on average for firms being in the most competitive industries relative to the others. Besides that, the effect of asset redeployability is consistent with the previous study. The cost of debt is significantly decreasing as the firm-level asset redeployability score increases. Using COMPUSTAT Herfindahl-Hirschman Index (HHI) adjusted for firm size as a proxy for the industry-level competition, loans borrowed by firms being in the most competitive industries tend to pay 9.4% (16 basis points) more on average than the comparable loans borrowed by firms in less competitive industries, considering other factors affecting loan spread. This spread difference can also be translated into an amount of 705,000 USD additional borrowing cost per year, on average. As for the interactive effects of competition and asset redeployability, the effect of competition on the cost of debt, given firms’ assets being in the most

(8)

8

redeployable group, turns out to be negative. Possessing more redeployable assets can help firms being in the competitive environment to lower the cost of debt. The significant effects of competition and asset redeployability are robust to alternative samples, such as the package-level sample and the sample within a shorter period. The results also hold after controlling for loan type and loan purpose fixed effects. Across all the specifications, there exists a positive relationship between the intensity of competition and the firm’s cost of debt. Moreover, the findings also suggest that competition captures some additional risks originating from the competitive environment on top of the risks captured by the proxy for corporate default risk. There exists no literature explicitly studying the interactive effect of competition and asset redeployability on the cost of debt. However, some previous researchers find the individual effects of competition and asset redeployability on the cost of debt. Valta (2012) found a significant positive effect of competition on the cost of debt by using multiple proxies for product market competition, where the similar effect has also been found in my empirical study. Previous literature finds that firms with more redeployable assets will receive financial contracts with longer maturities and lower interest rates. I also find a similar effect for asset redeployability, where more redeployable assets associate with a lower loan spread.

The paper proceeds as follows. Section 2 gives a comprehensive literature review related to competition, asset redeployability and cost of debt. Section 3 outlines the research hypotheses explicitly based on the previous literature and theoretical background. A brief overview of the sample and some summary statistics and a correlation matrix are presented in section 4. After that, section 5 explains the methodology used and discuss the empirical results and findings. Some results regarding robustness checks are done in section 6. Section 7 concludes the paper.

2. Literature review

2.1 Why choosing debt financing?

Debt financing is vital for firms because it is one of the primary sources of external financing and the cost of debt will impact the firms’ operational flexibility and decision making. According to the trade-off theory of leverage, there exists advantages and

(9)

9

potential risks of taking leverage. Because of deductibility of interest payment, the tax advantage of taking leverage is one of the primary reasons for choosing debt financing (Kraus and Litzenberger, 1973). Besides that, the obligated interest payment of debt reduces the free cash flow available at managers’ discretion, which mitigates the agency costs arising from the conflict between managers and shareholders (Jensen, 1986). Scott (1976) argue that bankruptcy cost is relevant for the optimal capital structure. Moreover, cost of debt also covers the agency costs between managers and investors. Williamson (1988) stresses that the choice of debt financing primarily lies in the characteristics of the underlying assets. He investigates the determinants of the liquidation value of assets by focusing on the potential buyers of assets. He points out that a higher asset redeployability, meaning that assets can be used alternatively for other purposes, results in higher liquidation value. Based on these redeployable assets, firms’ debt capacity is enhanced. Because even if managers fail to manage the firm appropriately, creditors can still recover the debt principal by redeploying those assets. Therefore, the important role of asset redeployability in firm’s liquidation value and the firm’s lending capacity is identified in this paper.

Above all, what other factors determining the pricing of corporate debt? Merton (1974) propose that the value of the issues of corporate debt fundamentally depend on three items, namely the required rate of return on riskiness, the various provisions and restrictions in the indenture and the default risk. The research done by Fries et al. (1997) investigates the valuation of securities and optimal capital structure from the perspective of entry and exit of firms in a competitive industry. They argue that the role of the price elasticity of demand for industry output is crucial to the pricing for corporate debt. All these empirical findings suggest that the market condition or the competitive environment indeed impact the financing cost paid by firms.

2.2 Competition and cost of debt

What role does competition play in debt pricing? To investigate the relationship between competition and cost of borrowing, the model used by Valta (2012) indicates that the cost of borrowing can be written as a function of firm’s default risk and the firm’s liquidation value, which are the two top essential factors banks will consider

(10)

10

when pricing the financial contract. Both the firm’s default risk and liquidation value are affected by market competition to some extent. Thus, there should exist a link between competition and the cost of borrowing. Furthermore, firms’ operational decisions and business risk can be directly and indirectly affected by the competitive environment.

The effect of competition on the cost of debt can be generalized into two channels. Firstly, competition raises firm’s cost of debt through the channel of default risk. According to Irvine and Pontiff (2008), increased idiosyncratic volatility of cash flow results from increased competition. Competition reduces the free cash flow and increases cash flow risk. Therefore, the firm’s default risk will also be higher. It is expected that firms with more default risk must pay higher rates for their debt. Bolton et al. (1990) argue that firm’s performance affects its financing costs and its access to capital. They state that firms with more cash reduce the cash flow of those financially constrained rival firms such that they can be driven out of the market. Therefore, the threats from rival firms increase the business risk of incumbent firms. Another finding of their paper is that firms relying heavily on external financing are also exposed to more fierce competition. The adverse effects of more intensive competition, such as reduced profitability, lower market share, will lead firms to turn to depend more on internal funds. However, this may give rise to financial slack and reduce the feasibility for investors to monitor the firm. Thus, external financing will be even more expensive for those firms. Furthermore, financially strong rival firms could adopt aggressive pricing strategies to significantly increase the business risks of incumbent firms, who have limited access to external financing (Bolton et al., 1990). Hou and Robinson (2006) predict that lower stock returns are more likely to exist in industries with high barriers to entry. Based on the test-series tests, they found that industry concentration can be used as a proxy for a risk factor sensitivity. In their study, competitive industries indeed have more pronounced distress risk, and therefore investors ask for a higher rate of return which includes the premium for such risk. In other words, investors tend to price the debts higher for those firms in industries with the more intense competition. Valta (2012) further investigates how product market competition impacts the firm’s cost of debt. By using various proxies for product market

(11)

11

competition and instrumenting competition with import tariffs to bypass endogeneity issues, he finds that the cost of bank debt is significantly higher for firms with more intensive product market competition. Therefore, when banks are pricing the debt contract, the risk arising from the firms’ competition environment will be considered. Secondly, the other channel which competition impacts the cost of debt is the asset liquidity. Ortiz-Molina and Phillips (2010) point out that asset liquidity is an essential determinant for the firm’s operational flexibility, which should be one of the determinants for the cost of borrowing as well. Moreover, they also found that firms with higher asset liquidity and during periods of high asset liquidity in the industry tend to have a lower cost of capital. Competition significantly impacts the availability of potential buyers for firm’s assets, which indirectly determines the liquidation value of firm’s assets. Therefore the market structure plays a vital role in determining the firm’s lending capacity.

The two channels mentioned above can explain how competition will impact the firms’ cost of borrowing. Whether the adverse effect of competition through the channel of default risk outweighs the positive effect of competition through the channel of the higher liquidation value of assets needs to be examined. Therefore, this is the primary question the empirical study aims to answer.

2.3 Asset redeployability, competition, and cost of debt

The aim of this research paper does not solely focus on the role of competition, but also on its interaction with asset redeployability. Benmelech et al. (2005) found evidence that firms with more redeployable assets will receive financial contracts with longer maturities and lower interest rates. Especially when contracts are incomplete, and transaction costs exist, liquidation value is utmost important, and it is imperatively vital for debt to be secured against firm’s assets and allow lenders to recover the firm’s liquidation value. Besides that, this paper also sheds light on the importance of liquidation value in financial contracting. As mentioned earlier, both asset redeployability and the availability of potential purchasers for firm’s assets in liquidation are two primary dimensions for the evaluation of firm’s liquidation value. Therefore, controlling asset redeployability in the regression not only increases the

(12)

12

explanatory power of the regression. The connection between competition and asset redeployability is essential. Competition is expected to lower firms’ cost of borrowing because more competition implies more potential buyers for firms’ assets. It is expected that firms with more redeployable assets have a higher liquidation value of assets and external financing can be less costly given a higher liquidation value of assets. However, the effect of competition on the cost of borrowing is complex and ambiguous. Therefore, it would be interesting to study how the magnitude and significance of the effect of competition change after considering the asset redeployability.

2.4 Hypotheses development

According to the literature reviewed above, it is not apparent that which effect embedded in the competition will be dominant. Firm’s competitive environment will determine their cost of borrowing oppositely from two perspectives. If the positive effect of competition through the threats from rival firms dominates, it is expected that the net effect of competition will be adverse. If the adverse effect of competition through the channel of liquidation value dominates, then competition will in general decrease the firm’s cost of borrowing.

H1: the positive effect of competition through the channel of rival firm threats on firm’s cost of borrowing is dominant, implying that more intense competition, in general, will raise firm’s borrowing cost.

As mentioned earlier, asset redeployability plays a crucial role in determining the liquidation value of firm’s assets. Therefore, it is necessary to include this effect in the research. The more redeployable the firm’s assets, the higher the assets’ liquidation value. Thus, it is expected that firms with a higher assets redeployability index will need to pay less for their debt.

H2: Firms employing more redeployable assets pay lower rates for their debt.

Assets redeployability and the availability of potential purchases for assets in liquidation are two crucial dimensions of liquidation value. It is interesting to study how these two dimensions interacting with each other and what total effect will be

(13)

13 on the cost of debt.

H3: For firms with higher asset redeployability, the adverse effect of competition on the cost of debt through the channel of more availability of potential assets buyers will be enhanced.

In other words, the adverse effect of competition on the cost of debt is expected to be dominant given that firms have the most redeployable assets.

3. Methodology

In this section, I explain the empirical strategy to study the effect of competition and asset redeployability on the cost of debt. I follow the methodology from Graham (2008) and Valta (2012) and with some variations:

𝐿𝑜𝑎𝑛 𝑠𝑝𝑟𝑒𝑎𝑑_{𝑖,𝑡,𝑗}

= 𝐶𝑂𝑀𝑃𝑈𝑆𝑇𝐴𝑇 𝐻𝐻𝐼_𝑘,𝑡 + 𝐴𝑠𝑠𝑒𝑡 𝑟𝑒𝑑𝑒𝑝𝑙𝑜𝑦𝑎𝑏𝑖𝑙𝑖𝑡𝑦_𝑖,𝑡 + 𝑙𝑜𝑎𝑛 𝑐ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐𝑠_{𝑖,𝑡,𝑗} + 𝐹𝑖𝑟𝑚 𝑐ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐𝑠_{𝑖,𝑡−1} + 𝑀𝑎𝑐𝑟𝑜𝑒𝑐𝑜𝑛𝑜𝑚𝑖𝑐 𝑓𝑎𝑐𝑡𝑜𝑟𝑠_𝑡−1+ 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑓𝑖𝑥𝑒𝑑 𝑒𝑓𝑓𝑒𝑐𝑡𝑠 + 𝑦𝑒𝑎𝑟 𝑓𝑖𝑥𝑒𝑑 𝑒𝑓𝑓𝑒𝑐𝑡𝑠 + 𝜖_{𝑖,𝑡,𝑗}

As mentioned earlier, the natural logarithm of all-in-spread drawn will be used to proxy firm’s cost of debt for the consideration of skewed distribution. The dependent variable is the loan spread paid by firm i for facility j at time t. As for the proxy for industry competition, I employ the industry-level COMPUSTAT Herfindahl index adjusted for average firm size, which is from the dataset constructed by Hoberg and Philips (2010). Index k represents different industries. For firm i at time t, there is also a corresponding asset redeployability score (Kim and Kung, 2016).

Apart from the competition and asset redeployability, there are other determinants for a firm’s cost of debt. The empirical model controls for firm’s profitability, corporate default risk, leverage ratio, asset tangibility, cash flow volatility, firm size and investment opportunities. Besides that, some loan characteristics, such as loan maturity and loan size, are included as well. All the firm-characteristic variables and macroeconomic factors controlled are lagged for one quarter, which is before the starting quarter of the facility. Moreover, the industry fixed effects at three-digit

(14)

14

Standard Industrial Classification (SIC) code level are included in all the regression specifications to capture the differential patterns of loan pricing across industries. For some of the regressions, year fixed effects are included to control for unobserved time-variant factors. Error terms include all the other unobserved determinants for the cost of debt, which are assumed to be independent of the variables already controlled in the model.

Among the firm characteristics, I employ the modified Z-score from Graham et al. (1998) to proxy corporate default risk. In their research, Z-score is used to capture the probability of corporate bankruptcy, which is a modified version of Z-score from Altman’s (1968). Based on the definition of Z-score, it may capture some direct and indirect bankruptcy cost, such as the tense relationships with suppliers and creditors, the legal fees, etc. (Graham, 1996). Moreover, cash flow volatility is calculated to capture firm’s business risk. It is expected that firm with a higher cash flow volatility will pay a higher price for the loan contracts because a higher cash flow volatility implies more business risk. Two macroeconomic factors are included in the empirical model as well. Macroeconomic factors contain the information about investors’ expectation and market economic conditions, which would impact the banks’ loan pricing. Collin-Dufresne et al. (2001) suggest that the dispersion of the yields between AAA corporate bond and BAA corporate bond becomes more significant in an economic downturn. During the economic downturn, investors require higher compensation because of the higher likelihood of corporate default. Most of the key variables of interest and control variables will be further discussed in the data section. A table of variable descriptions can also be found in Appendix A.

4. Data and sample construction

4.1 Data sources

To investigate the effect of competition on firm’s actual cost of debt, the sample is constructed by using various databases. Starting with collecting the historical information on loan pricing and loan characteristics from 1997 to 2017 from the LPC DealScan database, I gather all the spreads and fees for each facility and other detailed information such as maturity, facility amount, facility starting date. The linking table

(15)

15

for DealScan and COMPUSTAT is constructed by combining the linking table created by Chava et al. (2008) and the extension of the linking table after the year 2012 (Keil, 2018). By merging the facility dataset with the linking table, each facility ID has a corresponding GVKEY. Facility dataset with variable GVKEY can be further merged with COMPUSTAT data. After dropping the observations with loan maturity, loan size, loan pricing or GVKEY missing, the final facility dataset contains 75,061 observations. Each observation corresponds to a different loan.

From COMPUSTAT quarterly database, all the needed quarterly accounting data can be obtained. The COMPUSTAT dataset covers the period of 1997 to 2017. All the financial firms and utility firms are excluded from the sample, which corresponds to SIC code ranging from 6000 to 6999 and 4900 to 4999. Financial firms are excluded because their leverage levels are relatively higher than firms in other industries, where more leverage indicates a higher likelihood of financial distress. As for utility firms, their cash holdings are subject to regulatory supervision. Therefore, the cost of debt paid by these two types of firms might not be comparable to firms in other industry. In sum, the final merged dataset is a sample of 14794 unique facilities issued by 3002 U.S. firms. For each facility ID, there are corresponding GVKEY, the quarter in which the facility starts, relevant firm characteristics, loan information and macroeconomic factors. The final dataset is not a standard panel data because there might exist multiple facilities for a firm in the same quarter. Also, because most of the loans covered by LPC DealScan database are borrowed by larger firms, which might lead to potential selection bias for the final sample. However, since the focus of the study is on the effect of industry-level competition and firm size will also be controlled in the study. The absolute firm size would not bias the results that much. However, it is still questionable whether the conclusions and findings in my empirical study can be generalized to small- and medium-sized firms. Firm characteristics and operational patterns of small- and medium-sized firms differ significantly from large firms. There are more instabilities and volatilities for small- and medium-sized firms, and the access to external financing is limited to some extent. A more competitive environment could post more threats to the profitability of small- and medium-sized firms, and the potential volatilities could also affect firms’ borrowing capacity. Also, banks tend to

(16)

16

provide better terms for firms with sound performance. Therefore, it is expected that for small- and medium-sized firms, the positive relationship between competition and the cost of debt remains, and the size of the effect would be more significant.

4.2 Descriptive statistics

4.2.1 Loan spread, COMPUSTAT HHI, and Asset redeployability

The primary proxy for firm’s borrowing cost is loan spread, which measured as all-in-spread drawn in the DealScan database. This measure is defined as the all-in-spread the borrower pays in basis points over LIBOR or LIBOR equivalent for each dollar drawn down. Moreover, this measure also adds to the borrowing spread of the loan over LIBOR with any annual fee paid to the bank group. The higher the spread paid for the loan, the more expensive the loan is. Loan spread better captures the borrowing cost of the firm than merely using interest rate charged on loan. The logarithm of loan spreads is used for the empirical study to address the problem of skewness of the loan spread data (Graham et al., 2008). The other loan characteristics, such as loan maturity and facility amount, can be obtained from the facility table in DealScan database. To capture the effect of competition, the primary Herfindahl index used is the COMPUSTAT HHI adjusted for average firm size. A higher Herfindahl index implies a more concentrated industry. Therefore, more competition corresponds to a lower Herfindahl index. HHI is a widely-used proxy for market competition in the literature. As suggested by Hoberg and Philips (2010), COMPUSTAT HHI adjusted for firm size is a better proxy than a simple Herfindahl index calculated with COMPUSTAT data. Smaller firms are more likely to generate lower sales than larger firms due to the limitations of operation and capacity. A more comprehensive explanation for the construction of adjusted COMPUSTAT HHI can be found in Hoberg and Philips (2010). By merging the Herfindahl index dataset with the main COMPUSTAT dataset, each firm will have the corresponding industry-level Herfindahl index given the industry the firm belongs to. After that, merging the DealScan dataset with COMPUSTAT dataset, industry Herfindahl index can be matched with each facility through the link of GVKEY. Another variable of interest is asset redeployability. In previous empirical studies, asset redeployability has been widely used to measure a firm’s liquidation value of

(17)

17

assets in studying corporate strategies. To better capture the redeployability of corporate assets, the measure constructed by Kim et al. (2016) will be employed. In their empirical study, they decompose the notion of asset redeployability into asset specificity and asset liquidity, which results in a more comprehensive measure for capturing the redeployability of assets. Their measure indicates the relative ranking of redeployability for the asset composition of each industry. A higher score implies that industries have relatively more redeployable assets across and within industries. Their measure not only considers the correlation of assets’ usage across different industries but also breaks down the capital expenditure into different types of assets, which virtually covers all the industry group in the United States. The detailed explanation for how to construct this measure can be found in Kim et al. (2016). To ensure the robustness of empirical results, they use different weights to calculate the score. Based on the dataset constructed by them, both the firm-year level and the industry-year level asset redeployability score are available to use. The firm-level redeployability score is derived from the value-weighted average of the industry-level score. For my research question, I will merge the firm-year asset redeployability score with the COMPUSTAT dataset by NAICS code such that it can be further matched with facility data. Besides that, the firm-level redeployability score can capture more variations across firms than industry-level data.

4.2.2 Control variables

Some firm characteristics, which commonly mentioned in previous studies, are used in this analysis to better capture the effects of the variable of interests on firm’s cost of debt. Total assets measured in billion USD is used to proxy firm size. Larger firms have more financing sources and more operating flexibilities. Most of the larger firms have a sound reputation and more financial stabilities. Therefore, banks will provide debt contracts with better terms. Charging a lower price for the debt would be one of the possibilities of better terms. The tangibility of firm assets, defined as the ratio of Property, Plant and Equipment (PPE) to total assets, is also included. Firms with more tangible assets might pay less for their debt because of the higher potential liquidation value of assets. Besides that, profitability is another important firm characteristic controlled for loan pricing. It is expected that more profitable firms can borrow at a

(18)

18 lower rate, other things equal.

A firm’s leverage ratio, defined as the sum of current liabilities and long-term liabilities divided by the total assets (Graham et al., 2008), is also relevant for the study because firms already taking high leverage is more likely to default. Banks will tighten the terms for firms with higher corporate default probabilities. Therefore, it is expected that firms taking a higher level of leverage pay more for debt. Although this measure of the leverage ratio is not the best proxy, which is the ratio of long-term liabilities to total assets. However, after trying both measures in the relevant regressions, the significance and the magnitudes of the effects for the key variables are similar under both cases. Therefore, the inclusion of the current liabilities in the measure will continue being used for measuring leverage ratio in my empirical studies.

Corporate default probability is one of the critical determinants for banks pricing the loan contracts. Therefore, the accounting measure for default probability, Z-score, is also included in the primary regression model. The expected effect of Z-score on corporate borrowing cost is adverse since a smaller value of Z-score implies a higher level of financial distress (Graham et al., 1998). The measure for Z-score in my research is the modified version of Altman’s z-score used in the Graham’ paper (1998), which does exclude the ratio of market value of equity divided by book value of total debt in the calculation. Because a similar item, the market-to-book ratio, will be explicitly included in the empirical model, the modified score will be used. The formula of Z-score explained in Altman’s paper (1968) explicitly include the part of the market-to-book ratio.

Moreover, the market-to-book ratio is used as a proxy for the growth opportunities of the firm. On the one hand, it is expected that firms with more growth opportunities can borrow at a lower rate because there is a potential for better performance and higher profits. On the other hand, growth firms might have more instabilities and subject to more asymmetric information. Besides that, growth firms are likely to have more financial needs and might be more dependent on external financing. From this perspective, a higher market-to-book ratio might lead to a higher cost of debt. However, if market-to-book ratio represents the extra value over the book value of

(19)

19

assets that the debtor can recover in the case of liquidation, then the higher the ratio, the lower the cost of debt will be. Therefore, the expected sign of market-to-book ratio is ambiguous.

Furthermore, cash flow volatility, calculated as the moving standard deviation of the quarterly cash flow to total assets ratio in every eight quarters, is used to proxy the profitability risk of the firm. Firms with more cash flow volatility are expected to pay more for the debt. Besides that, two loan characteristics are included in the study as well. Loan maturity measured in months is the active loan period from the starting date. For more extended loan period, banks have the exposures to corporate business risk for a more extended period. Therefore, loans with longer maturity tend to have a higher price. Also, loan size measured by the ratio of facility amount to total assets is controlled because this variable also influences the loan pricing. Macroeconomics factors such as credit spread and term spread can also indirectly impact the cost of borrowing. Especially during the economic downturn, investors are exposed to higher risk of corporate defaults. Therefore, they require a higher compensation which can be reflected from the larger size of credit spread.

4.2.3 Descriptive statistics

The final sample is an unbalanced dataset, and all the ratios are winsorized at the 1st

and 99th_{percentiles to alleviate the impacts from outliers. Panel A in Table 1 is the}

summary statistics for the primary variables, reporting the number of observations, mean, median, standard deviation for each variable. The statistics show that the average loan spread is around 169 basis points over LIBOR, which are comparable to the related studies, for instance, Valta (2012) and Graham et al. (2008). Besides that, the facility amount varies considerably among the full sample. The mean and median of maturity show that most of the loans are medium-term loans. For COMPUSTAT Herfindahl index and asset redeployability score, their mean is nearly equal to the median. COMPUSTAT HHI are measured at the industry level, and a higher index implies a lower industry competition.

As for the firm characteristics of the borrowing firms in the sample, the average leverage ratio is 0.451, and the average firm size is USD 7.3 billion. The average Z-score

(20)

20

is relatively low in the sample. According to the definition of Altman’s Z-score, a score lower than 1.8 means that the company is probably headed for bankruptcy, while a score higher than 3.0 implies nearly no bankruptcy risk (Altman, 1968). The Z-score formula used in my empirical study exclude the portion of market-to-book ratio. If adding the 60% of the average market-to-book ratio back to the average Z-score, the mean Altman’s Z-score is around 2.2, which is a reasonable average corporate default probability for the whole sample. Therefore, the criteria for Altman’s Z-score will not be applied to the modified version. Besides that, on average, 29.5% of firms’ assets are tangible assets for the whole sample. The credit spread is the difference between AAA and BAA corporate bond. Therefore, the sample average of credit spreads is negative. Overall, all the statistics of variables are within reasonable ranges after dropping out some outliers.

Panel B presents an overview of the loan pricing and the other two loan characteristics, namely the loan maturity and facility amount, which measured in months and millions USD respectively. This table presents the compositions of the whole sample, which includes 14794 different facilities issued by 3002 different firms. Based on the deciles of COMPUSTAT Herfindahl index, Q1 is the group of firms with the index in the lowest three deciles, which implies the most competitive industries. Q3 includes the firms with the highest three deciles of the index, which implies the group of firms belong to the least competitive industries. Q2 is the group of firms whose corresponding index is in the middle four deciles of Herfindahl index.

For the group of firms belonging to more competitive industries, which is Q1 in panel B, the all-in-spread drawn is on average 180.98 basis points, meaning that those firms on average pay the loans 180.98 basis points over the LIBOR rate for each dollar amount drawn down. Compared with the other two groups, firms in Q1 on average pay more for the debt and receive loans with shorter maturity and smaller loan size. For firms in group Q3, the loan prices are on average 20 basis points less than the price paid by Q1. Besides that, the loan maturity and loan size are taken by the firms in this group are longer and more substantial.

(21)

21 Table 1 Descriptive statistics

Panel A: Summary statistics

This table presents the sample descriptive statistics for the period of 1997 to 2015. The final dataset is measured as a quarterly basis. All-in-spread drawn measured in basis points is the proxy for the cost of debt. Asset redeployability is measured by the score. The higher the score, the more redeployable the assets are. COMPUSTAT HHI adjusted for average firm size is the primary proxy for competition. The higher the Herfindahl index, the lower the competition level. Firm size is proxy by total asset in billion USD. Z-score is the proxy for corporate default probability.

Variables Obs. Mean Median Std. Dev.

All-in-spread drawn (bps) 14,794 168.994 150.000 113.998

Loan maturity (months) 14,304 48.466 60.000 22.605

Facility amount (millions USD) 14,794 440.634 200.000 898.737

COMPUSTAT HHI 14,794 0.069 0.060 0.036 Asset redeployability 14,794 0.402 0.480 0.108 Tangibility 13,920 0.295 0.230 0.220 Leverage 13,087 0.451 0.450 0.175 Profitability 13,083 0.042 0.040 0.023

Total assets (millions USD) 13,943 7328.958 1429.670 26040.840

Market-to-book 12,668 1.874 1.530 1.949

Z-score 12,128 1.090 1.040 0.477

Cash flow volatility 11,403 0.037 0.030 0.045

Credit spread (bps) 14,516 -0.921 -0.880 0.309

Term spread (bps) 14,516 1.710 1.740 1.116

Panel B: Competition and loan characteristics

This table presents the sample descriptive statistics for the period of 1997 to 2015. The final dataset is measured as a quarterly basis. All-in-spread drawn measured in basis points is the proxy for the cost of debt. Asset redeployability is measured by the score. The higher the score, the more redeployable the assets are. COMPUSTAT HHI adjusted for average firm size is the primary proxy for competition. The higher the Herfindahl index, the lower the competition level. Firm size is proxy by total asset in billion USD. Z-score is the proxy for corporate default probability.

Loan characteristics Q1

Most competitive

Q2 Q3

Least competitive

All-in-spread drawn (bps) 180.98 165.04 162.28

Loan maturity (months) 47.08 48.81 49.39

(22)

22 Panel C: Correlation matrix

This table reports the partial correlation matrix for the variables included in the study. aisd represents all-in-spread drawn, facamount is the facility amount. HHI is the Compustat HHI, and Redep is the asset redeployability score. For the other variables, more explanations can be found in Appendix A.

aisd maturity facamount HHI Redep

All-in-spread drawn 1 Maturity 0.1375 1 Facility amount -0.1658 -0.0585 1 COMPUSTAT HHI -0.0342 0.0044 0.1435 1 Redeployability -0.0739 -0.0573 -0.0529 -0.1094 1 Tangibility -0.0026 -0.0031 0.0081 0.0892 -0.4387 Leverage 0.1148 0.0177 0.0433 0.1212 0.0771 Profitability -0.1277 -0.0331 -0.0206 -0.0333 0.0115 Total assets -0.0890 -0.0567 0.3192 0.1423 -0.0945 Market-to-book -0.2048 -0.0416 0.0264 -0.0487 0.0378 Cash flow volatility 0.0902 -0.0457 -0.094 -0.0491 0.1225 Credit spread -0.2162 0.0978 -0.0275 -0.005 0.0467

Term spread 0.2001 -0.0287 0.0479 0.0037 -0.0296

Panel B roughly reports the relationship between competition and loan pricing across different groups. It seems that more competition will increase the firm’s cost of borrowing, which will be further examined in the following empirical study.

Panel C in Table 1 is the partial correlation matrix for the cost of debt and other firm and loan characteristics. COMPUSTAT Herfindahl index is negatively related to all-in-spread drawn. It is shown that larger firms with more profits, less cash flow volatility, more redeployable assets, more growth opportunities and more tangible assets are likely to borrow at a lower rate. In addition, there exists a positive relationship between firm’s leverage-taking and cost of debt.

5. Empirical results

The empirical analysis will be primarily based on three regression tables. Table 2 is the regression results for the underlying regression model, which examines the individual effect of competition and asset redeployability on the corporate cost of borrowing. Table 3 presents the regressions with some group dummy variables and study the

(23)

23

interactive effect of competition and firm asset redeployability on the corporate cost of borrowing. All the dependent variables in Table 2 and 3 are log loan spread. As for Table 4, regressions on the lender structure, the upfront fee, and the annual fee are performed.

5.1 Competition, asset redeployability, and cost of debt

This section mainly discusses the results reported in Table 2. The regressions in Table 2 examine the segregate effect of competition and asset redeployability on the cost of debt. Column 1 presents the regression results with only COMPUSTAT HHI and Redeployability score as explanatory variables. The relationship is strongly significant even without controlling any other relevant explanatory variables for the cost of debt. The coefficient of industry-level COMPUSTAT HHI suggests that a higher COMPUSTAT Herfindahl index is associated with lower cost of debt. Since a higher Herfindahl index suggests more competition in the industry, the sign of the coefficient implies that firms in more competitive industries on average pay more for debt. In addition, the coefficient of redeployability suggests that a higher redeployability score is associated with lower loan spread, which means firms with more redeployable assets pay lower rates for their debt.

In column 2, two loan characteristics are controlled, which both have significant effects on loan spread. Besides that, the signs of these two control variables are as expected, because loan pricing is increasing with the maturity and amount on average. After controlling these two loan characteristics, the size of the effect of COMPUSTAT HHI and Redeployability shrink, while the effect remained significant. The explanatory power of the regressions in the first two column are limited, which are less than 1 percent. Therefore, in column 3, some common-used firm characteristics for the cost of debt are included. In column 4, both loan characteristics and firm characteristics are controlled and the explanatory power increase to 17 percent. In column 5, two additional macroeconomics factors are included to capture the movements in the economic environment. According to the results from column 1 to 5, the significance of the critical variables is nearly not affected, although the magnitudes of the effects are smaller, on average.

(24)

24

All dependent variables are log (loan spread)

Variables (1) (2) (3) (4) (5) (6)

Key variable of interests

Compustat HHI -2.243*** -1.889*** -1.397** -1.168** -1.110** -2.743*** (-4.06) (-3.80) (-2.37) (-2.20) (-2.15) (-3.59) Redeployability -0.561*** -0.482*** -0.923*** -0.811*** -0.727*** -1.045*** (-3.99) (-3.72) (-4.99) (-4.74) (-4.33) (-3.21) Loan characteristics Log(maturity) 0.213*** 0.207*** 0.217*** 0.184*** (-12.01) (-10.98) (-11.41) (-10.72) Loan size 0.504*** 0.527*** 0.620*** 0.547*** (-7.82) (-6.88) (-8.13) (-8.37) Firm characteristics Total Assets -0.0038 -0.00286 -0.00322 -0.0055*** (-1.42) (-1.15) (-1.25) (-2.89) Tangibility -0.348*** -0.288*** -0.261*** -0.308*** (-3.99) (-3.59) (-3.33) (-2.99) Leverage 0.167 0.174* 0.256*** 0.156* (-1.6) (-1.78) (-2.73) (-1.69) Profitability -0.495 -0.965 -0.556 -0.594 (-0.60) (-1.18) (-0.69) (-0.83) Z-score -0.152*** -0.149*** -0.160*** -0.206*** (-3.76) (-3.98) (-4.36) (-5.95) Market-to-Book -0.152*** -0.143*** -0.124*** -0.102*** (-6.22) (-5.87) (-5.23) (-5.19)

Cash Flow volatility 1.913*** 1.823*** 1.929*** 1.776***

(-3.59) (-3.37) (-3.6) (-3.34) Macroeconomics factors Credit spread -0.330*** -0.144** (-12.96) (-2.20) Term spread 0.113*** 0.0553*** (-14.62) (-2.8) Constant 5.250*** 4.317*** 5.814*** 4.867*** 4.174*** (-73.92) (-45.46) (-53.16) (-38.65) (-31.35) N 14794 14296 9997 9733 9733 9729 R-squared 0.014 0.061 0.126 0.17 0.23 0.373

Year effects No No No No No Yes

Industry effects No No No No No Yes

Table 2 Effects of competition and asset redeployability on the cost of debt

This table presents the regression results for the effect of competition and asset redeployability on the corporate cost of borrowing. All the dependent variables are log (loan spread). Regression (1) to (4) are the variations of the underlying regression model. Regression (5) is the basic OLS model with all the explanatory variables. Regression (6) is the model with industry and year fixed effects controlled. Robust t-statistics are reported in parentheses. All the standard errors are clustered at the firm level. *, **, *** represents significance at 10%,5% and 1% level, respectively.

(25)

25

The last column in Table 2 covers loan characteristics, firm characteristics, and macroeconomics factor altogether. Also, year and industry fixed effects are included, which significantly improves the explanatory power from 1.4% (column 1) to 37.3% (column 6). The coefficient of the COMPUSTAT HHI is -2.743 and is significant at 1% significance level, which implies that a 0.01 increase in the Herfindahl index is associated with a 0.027 log points decrease in loan spread. Because the sample means of COMPUSTAT HHI adjusted for average firm size is 0.069, therefore a 0.01 increase in the index will be used for the analysis instead of assuming a one-unit increase in the index. An amount of 0.027 log point changes in loan spread corresponds to a roughly 1.027 basis points change in all-in-spread drawn, which is a sizable effect on the cost of borrowing because the facilities in the sample are large, on average. The sign of the coefficient also indicates that the positive effect through the channel of rival firm threats is dominant over the adverse effect through the liquidation value of assets. The coefficient of the redeployability score is -1.045 and is highly significant as well. The mean redeployability score in the sample is 0.402. Therefore, assuming the score increase by 0.1 would be reasonable for explaining the coefficient. Thus, the log spread would decrease by 0.1042 log points. The sign of the coefficient proves the second hypothesis that firms with more redeployable assets will pay a lower rate on average for debt.

The coefficients of control variables in column 6 are most significant and with expected signs. Keeping other things equal, if total assets increase by one billion, the cost of debt will decrease by 0.0055 log point. Therefore, cost of debt is decreasing in firm size, which possibly because of more stability and more sound reputation in larger firms. In addition, firms with more tangible assets borrow debt at lower spread because creditors can be better recovered in liquidation. For firms already taking a high level of leverage, their cost of debt tends to be higher on average. The loan spread will increase by 0.0156 log point if leverage ratio increases by 0.1. The reason might be that more leveraged firms have higher default risk on average than the other peers. Moreover, more profitable firms might pay less for debt although this effect is insignificant according to the result. Z-score is a necessary control variable in the regression model because it is the proxy for the corporate default risk. A higher

(26)

Z-26

score suggests a lower default risk. The coefficient implies that a one-unit increase in Z-score would entail the cost of spread decreasing by 0.206 log spread. Therefore, lower corporate default risk is associated with a lower cost of debt, which is consistent with the expectation. The Market-to-book ratio proxy for the growth opportunities of the firm, while growth opportunities can affect the cost of debt from two perspectives. On the one hand, more growth opportunities imply a possibility of better performance and more profits, which is a good signal for creditors. On the other hand, growth firms face more uncertainties and asymmetric information, which make it harder for creditors to monitor. Also, growth firms have more financing needs and might be more dependent on external financing. Therefore, the expected sign of Market-to-book ratio is uncertain. However, if Market-to-Book ratio represents the extra firm value over the book value of assets that the creditors can recover in the case of liquidation, then the higher the ratio, the lower the borrowing cost would be. The coefficient of Market-to-Book ratio is -0.102, implying that more growth opportunities will lower the cost of debt or a higher value of debt recovered in case of liquidation will lower the cost of debt.

Furthermore, more cash flow volatilities will increase the cost of debt as the sign of the coefficient is positive. As for two macroeconomic factors, both coefficients are significant and with expected signs. The effect of a credit spread is negative, meaning that when the spread gets larger, the cost of debt will be lower. Since the definition of credit spread here is the yield difference between AAA and BAA corporate bond. Therefore, the spread should be a negative amount. During the economic downturn, the spread will become larger in absolute value because investors would ask for even higher compensation for BAA bond. Therefore, cost of debt on average is lower during sound economic conditions.

In sum, the empirical results presented in Table 2 provide evidence for the first two hypotheses. Individually, the more competition within the industry, the higher the cost of debt on average for the firms within this industry. For firms with more redeployable assets, the borrowing cost on average is lower as expected.

(27)

27

5.2 Interactive effect of competition and asset redeployability

In this section, the analysis is mainly based on the results reported in Table 3. After studying the segregate effect of competition and asset redeployability on the cost of debt, the research question can be moved further into the interactive effect of both factors. The third hypothesis conveys the idea that the redeployability of firm assets will change the effect of competition. As mentioned earlier, competition environment determines the potential availability of assets buyers in liquidation, which further determines the liquidation value of assets. Asset redeployable is linked with competition because both can determine liquidation value of assets in specific ways. Therefore, Table 3 presents three regressions, which aim to study how these two factors interact.

The first regression in Table 3 is based on the full sample. Compared with the last regression in Table 2, the only difference is the inclusion of ‘competitive’ dummy variable. Based on COMPUSTAT HHI, the whole sample is divided into ten deciles. The competitive dummy equals to one when the corresponding firm for the observation belongs to the most competitive industries. The most competitive industries are those whose COMPUSTAT HHI is in the lowest three deciles. The coefficient of competitive dummy variable implies the effect of firms being in the most competitive industries on the corporate cost of debt. According to the results, the magnitudes and signs of the effects of COMPUSTAT HHI and asset redeployability remain nearly the same as the findings of the last regression in Table 2. The coefficient of the competitive dummy, 0.088, is significant at 5% significance level and it means that firms being in the most competitive industries will on average pay 8.8% more borrowing cost than the firms in less competitive industries. The effect of the competitive dummy is consistent with the coefficient of COMPUSTAT HHI, which implies more competition is associated with a higher borrowing cost.

The second regression in Table 3 is a modified version of the first regression. By solely including a redeployable dummy variable, this regression aims to test for the effect of whether being in the group with the most redeployable assets would pay less for debt on average than the rest of firms. Redeployability score is used as the criteria for

(28)

28

All dependent variables are log (loan spread)

Variables (1) (2) (3)

Key variables of interest

Compustat HHI -2.516*** -2.760*** -3.106*** (-3.26) (-3.60) (-3.73) Competitive Dummy 0.088** 0.094*** (2.43) (2.60) Redeployability -1.064*** -1.164*** -1.229*** (-3.28) (-3.57) (-3.76) Redeployable Dummy 0.0517 -0.172 (0.61) (-1.38)

Compustat HHI *Redeployable 3.737**

(2.17) Loan characteristics Log(maturity) 0.184*** 0.185*** 0.183*** (10.74) (10.78) (10.74) Loan size 0.538*** 0.547*** 0.536*** (8.24) (8.37) (8.21) Firm characteristics Total assets -0.00551*** -0.00551*** -0.00546*** (-2.89) (-2.92) (-2.90) Tangibility -0.313*** -0.309*** -0.303*** (-3.05) (-2.99) (-2.96) Leverage 0.170* 0.157* 0.178* (1.85) (1.70) (1.96) Profitability -0.612 -0.592 -0.652 (-0.86) (-0.83) (-0.91) Z-score -0.203*** -0.206*** -0.202*** (-5.88) (-5.94) (-5.88) Market-to-Book -0.103*** -0.102*** -0.103*** (-5.20) (-5.20) (-5.20)

Cash Flow volatility 1.761*** 1.769*** 1.773***

(3.38) (3.34) (3.34) Macroeconomics factors Credit spread -0.141** -0.144** -0.152** (-2.16) (-2.21) (-2.22) Term spread 0.0551*** 0.0550*** 0.0537*** (2.78) (2.78) (2.70) N 9,729 9,729 9,729 R-squared 0.374 0.373 0.375

Year effects Yes Yes Yes

Industry effects Yes Yes Yes

Table 3 Interactive effect of competition and asset redeployability

This table reports the results regarding the interactive effect of both factors. Regression (1) tests for the effect of being in the most competitive industries on the cost of debt. Regression (2) studies whether firms with the most redeployable asset on average pay less for debt than the rest of firms. Regression (3) study the interactive effect of competition and asset redeployability by interacting COMPUSTAT HHI with the redeployable dummy variable. Both year and industry fixed effects are added in each regression. Robust t-statistics are reported in parentheses. All the standard errors are clustered at the firm level. *, **, *** represents significance at 10%,5% and 1% level, respectively.

(29)

29

constructing the redeployable dummy. Redeployable dummy equals to one when the asset redeployability score of the firm is in the highest three deciles. However, the reason for this finding could be due to the diversity of determinants for the cost of debt. Although asset redeployability is a crucial dimension for the liquidation value of assets, the indirect effect on the cost of borrowing is not strong enough compared to a more direct determinant, in this case, competition.

The third column of Table 3 is the primary regression aiming for testing the third hypothesis because it studies the interactive effect of competition and asset redeployability. The inclusion of both competitive dummy and the redeployable dummy is intended to study the effect of firms being in the most competitive industries or/and with the most redeployable assets on the cost of borrowing. A similar effect of competitive dummy has been found in the last regression, meaning that firms belonging to the more competitive industries on average pay 9.4% more than the spread paid by the firms in less competitive industries. Also, the effect of redeployable dummy remains insignificant. As for the COMPUSTAT HHI and redeployability score, their effects remain significant, and the size of effects become more substantial compared to the previous regressions. In other words, cost of debt becomes more sensitive to the changes in competitiveness and asset redeployability in this regression model. Most importantly, the effect of the interaction is significant at 5% significance level. The coefficient of the interaction implies that given firms belong to the group with the most redeployable assets, if the intensity of industry competition increased, which corresponds to a decrease in COMPUSTAT HHI, the total effect of COMPUSTAT HHI on the cost of debt would be 0.631 (3.737-3.106). If COMPUSTAT HHI increases by 0.1 unit, the cost of debt will increase by 0.631 log point. The main finding of this regression states that more intense industry competition would lower the average spread paid by firms, given that firms are being in the group with the most redeployable assets. Considering the linkage between competition and asset redeployability, the relationship between competition and cost of debt changed from positive to negative. As mentioned in the previous section, the effect of competition on the cost of debt is expected to be negative through the channel of availability of potential asset buyers in case of liquidation. In other words, the adverse

(30)

30

effect of competition on the cost of debt will be strengthened given that firms have the most redeployable assets, which is consistent with the third hypothesis.

5.3 Lender structure, upfront fee, and annual fee

In this section, I investigate the effect of competition and asset redeployability on some other loan contract items, namely the lender structure in a loan contract and the transaction fees charged for the loan by the lenders. The relevant regression results are presented in Table 4. Again, competition environment could affect the risk characteristics of firms. It is mentioned earlier that the rival firm threats in a competitive industry could be too severe to increase the corporate business risk considerably. From this perspective, the number of lenders may increase together with the informational issues existed in loan contracting, However, for more competitive industries, the availability of potential buyers for assets in liquidation is more abundant as well. Therefore, it is likely that creditors can be fully recovered from the sale of assets. Therefore, the need to diversify their risk through sharing the loan with other lenders is less.

Column 1 of Table 4 reports the regression results for the number of lenders in a loan contract. The key finding is that a higher COMPUSTAT HHI is associated with more number of lenders in a loan contract. Alternatively speaking, the number of lenders on average are less for more competitive industries. The coefficient, 27.84, suggests that a 0.1 increase in COMPUSTAT HHI would decrease the number of lenders by 2.784 on average. This finding implies that the availability of potential buyers for assets in liquidation would make lenders feel more secure that they can recover the debt. There is less demand for lenders to share the credit risk with others. Besides that, more redeployable assets cannot significantly determine the lender structure. Credit risk and informational issues are the more relevant determinants for lender structure. In addition, longer maturity and larger loan size would raise the number of lenders, because there are more risks embedded in the loan contract such that creditors have a stronger motive for syndication, which is consistent with the diversification motive of lenders. By sharing a small portion of the syndicated size, lenders limit their exposure to the risky loans (Graham et al., 2008).

(31)

31

(1) (2) (3)

Variables Number of lender Log (Upfront fee) Log (Annual fee) Key variables of interest

Compustat HHI _27.84*** _-1.982*** _-2.128*** (-8.36) (-2.77) (-3.30) Redeployability 1.566 -0.736* -1.221*** (-1.11) (-1.82) (-3.15) Loan characteristics Log(maturity) 0.723*** 0.309*** 0.187*** (-7.75) (-9.57) (-9.34) Loan size 0.863** 0.171 1.194*** (-2.09) (-1.5) (-8.1) Firm characteristics Total assets 0.0142*** -0.00315*** -0.00812*** (-5.35) (-4.29) (-9.11) Tangibility -2.496*** -0.736*** -0.839*** (-5.05) (-5.30) (-5.20) Leverage 4.395*** -0.0536 0.264* (-9.76) (-0.45) (-1.82) Profitability 4.016 -0.866 -0.141 (-1.12) (-0.81) (-0.14) Z-score -1.076*** -0.113** -0.0537 (-6.03) (-1.97) (-1.06) Market-to-Book 0.170** -0.0709*** -0.196*** (-2.46) (-3.63) (-9.07)

Cash Flow volatility -9.252*** 1.821*** 2.408***

(-4.70) (-3.76) (-3.64) Macroeconomics factors Credit spread 0.575*** -0.319*** -0.386*** (-3.03) (-6.38) (-4.67) Term spread 0.402*** 0.117*** 0.139*** (-7.05) (6.75) (10.63) N 13,069 1,780 2,170 R-squared 0.127 0.388 0.493

Industry effects Yes Yes Yes

Table 4 Regressions on lender structure, upfront fee, and annual fee

This table presents the results of three regressions on the number of the lender, upfront fee, and annual fee, respectively. The key variables of interest in all three regressions are COMPUSTAT HHI and Redeployability score. From the table, all the effect of competition is significant at 1% significance level. Industry fixed effects are added in each regression. Robust t-statistics are reported in parentheses. All the standard errors are robust to heteroscedasticity. *, **, *** represents significance at 10%,5% and 1% level, respectively.

(32)

32

Moreover, the relationship between firm size and the number of lenders is positive. The information asymmetry issue is less severe for larger firms such that they can borrow from syndication of more lenders. It is as expected that lower corporate default probability (Z-score) is associated with fewer lenders. Similarly, for lower corporate cash flow volatility, the syndicated loan tends to consist of fewer lenders because of less business risk. Furthermore, it is worth mentioning that increase in credit and term spread are related to more number of lenders. During the economic downturn, both credit and terms spread will become more dispersed. Since credit spread is the yield difference between AAA and BAA bond, the spread will become even more negative during the economic downturn. In other words, economic downturn leads to a decrease in credit spread, which further leads to a decrease in the number of lenders. Because of an increase in market-wide default risk, there is a stronger monitoring motive for syndication. A more concentrated lender structure would lead to a more efficient group monitoring, and the restructuring activities in case of default are more likely to succeed.

I further examine whether the transaction cost charged by lenders are higher for more competitive industries and lower for more redeployable assets. In Table 2 and 3, I both find that more intense competition will raise the loan spread paid by firms. For transaction cost, this positive effect of competition remains. Column 2 and 3 are regressions on the loan transaction costs charged by the lenders. In a syndicated loan contract, the lead bank always acts as an agent bank for the rest of the banks to administer and arrange the loan. The fee paid by the borrower is tied to the riskiness and complexity of the loan. Specifically, I examine the upfront fees and annual fees charged by banks. According to the definition, an upfront fee is a one-time fee, which is often paid at the closing of the deal. The upfront payment is paid to the lead bank and can be shared with other syndicate banks. An annual fee is merely the annual fee charged against the entire commitment amount. Because of the limitation of fee information in the sample, the two regressions are based on the sample with 1780 and 2170 non-missing observations, respectively. The results in column 2 and 3 imply that both upfront fee and annual fee decrease if COMPUSTAT HHI increases or competitiveness enhances. Compared with the effects of competition found in

Asset redeployability, competition, and cost of debt

Master Thesis