1
IFRS 7 Risk Disclosures and Syndicated
Loan Pricing:
The role of the Board of Directors
Master Thesis, MSc Accountancy University of Groningen Faculty of Economics and Business
24 June 2019 G.G. van Koten Student number: S3540383 Tel. +31 6 31246987 Email: g.g.van.koten@student.rug.nl Supervisor: dr. Y. Karaibrahimoglu
ABSTRACT
Since financial year 2007 firms are required to apply IFRS 7 in their financial reporting. This standard requires firms to disclose in a more structure manner their risk exposure arising from their use of financial instruments. With this study we aim to answer the question whether IFRS 7 provides useful information to lenders. We theorize and find that IFRS 7 disclosures provide good information to lenders, with unfavourable consequences for firms. We find that, controlling for firm specific characteristics, that a better IFRS 7 disclosure increases the perceived riskiness leading to higher loan prices. Studying the effects of board characteristics, we find that the concentration of independent directors on the board to have a negative effect on the way lenders use the financial instrument disclosure. In general, we find that firms in the post-adoption period of IFRS 7 pay significantly higher loan prices.
3 CONTENTS
1 INTRODUCTION ... 4
2 INSTITUTIONAL BACKGROUND, LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT ... 7
2.1 IFRS 7 – FINANCIAL INSTRUMENT RISK DISCLOSURE ... 7
2.2 SYNDICATE LOAN MARKET ... 7
2.3 HYPOTHESES DEVELOPMENT ... 8 2.4 CONCEPTUAL MODEL ... 12 3 METHODOLOGY ... 13 3.1 SAMPLE ... 13 3.2 RESEARCH DESIGN ... 14 3.3 VARIABLES... 15 4 RESULTS ... 20 4.1 DESCRIPTIVE STATISTICS ... 20 4.2 CORRELATION MATRIX ... 20 4.3 MULTIVARIATE ANALYSIS ... 22 4.4 ADDITIONAL TESTS ... 24
5 DISCUSSION AND CONCLUSION... 30
5.1 DISCUSSION AND CONCLUSION ... 30
5.2 CONCLUSSION ... 31
5.3 LIMITATIONS ... 32
5.4 FURTHER RESEARCH ... 32
1
INTRODUCTION
Starting year-end 2007, firms were required by the International Accounting Standards Board (IASB) through IFRS 7 to publish a Financial Instrument Risk Disclosure in their annual reports. The aim of this reporting standard is to aid investors in their decision- making process by providing them with more information of higher quality and transparency (IASB, 2005). Due to their asymmetrical payoff and different risk preference, the information provided by the financial instrument risk disclosure should be of importance to the creditors of the firms (Karaibrahimoglu and Porumb, 2019). But more than a decade has passed since IFRS 7 became mandatory and it is still not clear whether financial instrument risk disclosures are indeed useful to creditors, both in practice and in the literature. Even though the IASB thinks financial instrument risk disclosures provide valuable information to investors, they are not sure whether the information in those disclosures is useful for creditors (IASB, 2017). Also, academic literature cannot give an answer to the question whether and/or how the IFRS 7 disclosure is being used by creditors in their decision making (e.g. Elshandidy, Shrives, Bamber and Abraham, 2018; Kravet and Muslu, 2013; Miihkinen, 2012).
The reason why the disclosures are so important to investors is due to the information asymmetry that exists between them and the firm’s managers. One way of explaining why this asymmetry exists is through the agency theory. Agency theory tries to explain the problems that arise due to the separation of ownership and control in a firm. Investors do not have access to the same information as the managers of the firms, while at the same time they are also unable to monitor the action of the managers. This provides with ample opportunity for managers to act opportunistically, to prevent this from happening investors make costs to align the interests of the managers with their own (Ang, Cole and Lin, 2000). Those costs can be reduced by reducing the information advantage of the managers (Huang and Zhang, 2011). One way to decrease the information asymmetry for creditors and thus reduce the agency costs, is through disclosures like the one provided by IFRS 7. However, risk disclosures like IFRS 7 are generally assumed to contain unfavourable information, which have the possibility to increase the perceived riskiness for creditors. A recent study shows that a firm’s risk disclosure contains risk that are firm specific (Campbell et al., 2014), risk disclosures that are firm specific have been found to be useful to investors (Campbell et al., 2014; Hope, Hu and Lu, 2016). Kravet and Muslu (2013) found that firms with a more extensive risk disclosure are perceived as riskier by investors.
5 Our aim is to examine the effect of IFRS 7 disclosures on the subsequent loan prices, we expect
that firms who have a more extensive risk disclosure to be perceived by lenders as riskier and thus should receive a higher loan price. To research whether this is indeed the case the following research question has been formulated:
Research Question 1: To what extent does the financial instrument risk disclosure
influence the subsequent loan price provided by lenders?
Elshandidy et al. (2018) explains however that it is not only that we do not understand how risk disclosures influence the decision-making process of lenders, we also do not know what might influence this relationship. A factor that might moderate this relationship are certain corporate governance characteristics, especially those related to the board of directors. Research suggest that different characteristics of a firm’s board of directors influences the way a firm discloses risk in her disclosure (e.g. Agyei-Mensah, 2017; Elshandidy and Neri, 2015; Abraham and Cox 2007).
We are interested in understanding to what extent the board of directors moderates the relationship between the risk disclosure and the subsequent loan price provided by lenders. To guide the research the following research question has been formulated:
Research Question 2: To what extent does the board of directors influence the
relationship between the disclosed financial instrument risks and the loan price? By researching this we contribute to the growing body of literature in the field of risk reporting. Currently there has not been done any research on how debt markets process risk disclosures and/or if certain factors influence this process (e.g. Elshandidy et al., 2018; Kravet and Muslu, 2013; Miihkinen, 2012). Our results contribute to growing body of academic literature by showing that risk information disclosed in IFRS 7 provides relevant information to lenders, affecting their decision-making process when setting loan prices (Karaibrahimoglu and Porumb, 2019). However, we found that board independence to moderate this process, leading firms to paying lower loan prices. In general, we find that firms pay significantly higher loan prices in the post-adoption period of IFRS 7.
The results also aid the IASB and other practitioners in assessing the disclosed information. In the beginning of this chapter we showed that the IASB is not sure whether IFRS 7 provides useful information to investors. Our results contribute to practice by showing that IFRS 7 risk disclosures provide relevant information for lenders in assessing the risk of lending to a
borrower. It also contributes to practice by showing that lenders are mostly interested in financial instrument risks associated with systematic risks rather than unsystematic risk.
The remainder of the thesis is structured as followed. In the following chapter we start by discussing the theoretical background of this research, here the underlying theory will be described followed by the formulation of multiple hypotheses. In the subsequent chapter we will discuss the methodology of this research, describing the data and discussing the method for researching the data. After this we will present our findings in the results section. And to conclude this thesis we will present the discussion and conclusion in the last chapter.
7
2
INSTITUTIONAL BACKGROUND, LITERATURE REVIEW AND
HYPOTHESIS DEVELOPMENT
2.1 IFRS 7 – FINANCIAL INSTRUMENT RISK DISCLOSURE
In 2005 the IASB issued IFRS 7, this accounting standard replaces the IAS 30 standard and continues the disclosure requirements under IAS 32 (IFRS 7). IFRS 7 deals with the disclosure of the risks raising from the usage of financial instruments by the firm (IASB, 2005). Although early adoption was encouraged by the IASB, the effective adoption date for firms was for annual reports for year-end 2007 (IASB, 2005). In IFRS 7 no clear definition is given for what exactly constitutes as a financial instrument. In the previous standard, IAS 32, a financial instrument is defined as “a contract that gives rise to a financial asset of one entity and a financial liability or equity instrument of another entity.” (IAS 32, p.2). The goal of IFRS 7 is to enable the user of the disclosure to assess “(1) the significant of financial instruments for the entity’s financial position and performance; and (2) the nature and extent of risks arising from financial instrument to which the entity is exposed during the period and at the reporting data, and how the entity manages those risks.” (IFRS 7, para.1).
The IFRS 7 disclosure focuses on the risk that arise from the usage of financial instrument by the firm as well as on how the firm tends to manage those particular risks, the risks that are typically discussed in the disclosure are credit risk, liquidity risk and market risk (IFRS 7, para.32). In the standard it is stated that both the qualitative- and the quantitative dimension of the risk has to be discussed in the disclosure. The qualitative section of the disclosure has to discuss “(a) the exposures to risk and how they arise; (b) its objectives, policies and processes for managing the risk and the methods used to measure the risk; and (c) any changes in (a) and (b) from the previous period.” (IFRS 7, para.33). With regard to the quantitative disclosure a firm has to provide “summary quantitative data about its exposure to that risk at the end of the reporting period.” (IFRS 7, para.34).
2.2 SYNDICATE LOAN MARKET
For most firms the primary source of outside financing are loans provided by banks (Ferreira and Matos, 2012). Even though banks are a special kind of creditor, they have still a different pay-off structure and risk preference compared to equity investors. Since lenders do not share in the profits of increased risk-taking behaviour of firms, and in general prefer to take not too much risks, they make use of their loan contract to protect themselves. Loan pricing is one of
the contract terms used to mitigate risk, by charging a higher spread on the loan the lender can reduce the adverse effects of risks. The spread is the difference between that what is charged and the underlying risk-free interest rate. The higher the perceived risk of the lending to a borrower the more spread the lender will charge to the borrower (e.g. Porumb et al., 2019; Kim et al., 2011a). Other loan terms are also used by lenders to mitigate risk, researcher commonly look at the maturity, loan size and number of lenders (e.g. Porumb et al., 2019; Kim, Song and Zhang, 2011; Kim, Tsui and Yi, 2011). The maturity of the loan is another good indicator for the underlying risk of the loan, firms with a higher perceived riskiness will on average receive loans with a lower maturity (e.g. Porumb et al., 2019; Graham and Qiu, 2008). The number of lenders brings us to a special kind of loan, the syndicated loan.
A syndicate loan is a loan in which a group of lenders, led by a lead arranger, lend money to a borrower (e.g. Bharath, Dahiya, Saunders and Srinivasan, 2011). Lenders can have different reasons for why they use this type of lending, a loan can be for example too risky for a lender and chooses to syndicate the loan to mitigate risks. Because the risk is shared by a number of lenders, one might expect that for a riskier loan more lenders will be involved (Porumb et al., 2019). A problem however with this type of lending is that an additional moral hazard problem is created, a syndicate moral hazard problem. This so-called syndicate moral hazard problem is that the lead arranger bears the full costs for monitoring the borrower without receiving the full benefit for this monitoring activity, giving the lead arranger an incentive to shirk (Bharat et al., 2011; Sufi 2007). The reason why this is possible in the first place is due to information asymmetry, the other members of the syndicate cannot observe whether the lead arranger is really monitoring the borrower (Bharat et al., 2011; Sufi 2007). The other members of the syndicate are fully aware of this problem and will therefore demand that the syndicated loan has more strict lending terms. This information asymmetry and the associated costs can however be reduced through the firm’s disclosure activity (Huang and Zhang, 2011), explaining why IFRS 7 is also relevant to banks who normally are assumed to have access to inside information.
2.3 HYPOTHESES DEVELOPMENT
2.2.1. The effect of the financial instrument risk disclosure
The financial instrument risk disclosure is a mandatory disclosure under IFRS 7, in this disclosure a firm has to disclose which risks the company faces due to the financial instruments it holds. Information provided in the financial instrument risk disclosure is relevant information
9 source for lenders, it however contains information about unfavourable scenarios for the firm.
The disclosure explains the firm’s exposure to firm- and market specific conditions, the uncertainties that financial instruments create and how it might impact the firm. Managers have the tendency to disclose positive information or at the very least delay the disclosure of negative news (Kothari, Shu and Wysocki, 2009). Therefore, requiring firms to disclose financial instrument risk could be considered desirable. Even though it might be desirable, does it also deliver useful information?
Bean and Irvine (2015) asked this question to financial analyst, they found that although in contained some usefulness it was still mostly boilerplate and generic. But other research findings seem to suggest that in fact the contrary might be the case. Campbell et al. (2014) show that in a US-setting most risk disclosures contain risks that are firm-specific, and thus not boiler-plate. Those risk disclosures contain exactly the kind of information that is useful to investors (Hope, Hu and Lu, 2016; Campbell et al., 2014). If this is indeed the case, firms that have a more extensive risk disclosure could therefore be expected to disclose more risk information and should therefore be perceived to be riskier. For the US-setting this reasoning gets supported by the findings of Kravet and Muslu (2013), who find that US firms with more extensive risk disclosures are perceived as riskier by the market1.
Due to their asymmetric payoff structure and risk preference lenders tend to dislike borrowers who are perceived as riskier, they do not share in the rewards of additional risk but will share in the possible losses of the additional risk. When facing risk lenders will only consider lending money under stricter contract terms. We expect that firms with a more extensive and qualitative financial instrument risk disclosure will be perceived as riskier by lenders, therefore they will be offered more strict lending terms when borrowing money. To test this argument, the following hypothesis has been formulated:
Hypothesis 1: The financial instrument risk disclosure has a positive effect on the subsequent
loan price of the firm.
2.2.2. The effect of the board of directors
Agency theory assumes that due to the separation of ownership and control, the agent has an information advantage over the principal (Jensen and Meckling, 1976). Furthermore, it states
1 We would mention that there is also an alternate view among scholars on how IFRS 7 should influence the decision-making
process of lenders. Those scholars argue that even though the information disclosed is unfavorable for investors, it still reduces the information asymmetry between managers and investors. They therefore expect a negative relationship between IFRS 7 risk disclosures and the subsequent loan price.
that to maximize their own benefit, agents behave opportunistically at the expense of the principal (Jensen and Meckling, 1976). An example of opportunistic behaviour the creditors have to deal with is the risk that managers might be inclined to make decisions that benefit the shareholders at the expense of creditors (Christensen et al., 2016; Jensen and Meckling, 1976). A reason why this is possible in the first place is because creditors are at an information disadvantage compared to the manages of the firm. Managers tend to behave opportunistically towards investors by immediately disclosing the favourable information, while delaying the disclosure of less favourable information (Kothari, Shu and Wysocki, 2009). To prevent managers from behaving sub-optimally the lenders have to take preventive actions, the costs associated with those preventive actions however are costly. One way to lower those costs is by reducing the information advantage the agent has over the creditors (Huang and Zhang, 2011). An important corporate governance mechanism that can help with this is the board of directors.
The board of directors fulfils multiple important functions within firms, literature often describes the monitoring, discipling and advising management as important functions (e.g. De Andres and Vallelado, 2008; Adams & Ferreira, 2007; Hillman and Dalziel, 2003). Monitoring by the board of directors can help reduce the agency costs by limiting the extent in which management can behave opportunisticly, limiting the damage from opportunistic behaviour for both lenders and other investors (Hillman and Dalziel, 2003). Which in this case would mean that management discloses unfavourable information in a timely manner. Besides, Elshandidy and Neri (2015) found that higher levels of disclosure can be achieved by firms when the board of directors is able to monitor the management properly.
An important element of the board of directors are the non-executive directors. Non-executive directors fulfil and important role within firms, no wonder that it is a commonly looked at characteristics when researching the effect of corporate governance (e.g. Agyei-Mensah, 2017; Abraham and Cox, 2007). Non-executive directors influence the way the firms disclose its information, something which is clearly visible in the disclosure of financial information (Chen and Jaggi, 2000). But more important for this research they also seem to influence the risk disclosures of a firm. Researchers have linked non-executive directors to an increase in the quality of the disclosure (Abraham and Cox, 2007; Beasley, 1996) as well as the extent of compliance (Agyei-Mensah, 2017). This however only seems the case when non-executives are independent, for non-executives with ties to the firms there has not been found anything significant (Abraham and Cox, 2007). This is in line with argument made by agency theorists
11 that this can only be achieved when incentives of the board are properly aligned with investors
(Fama, 1980; Jensen and Meckling, 1976), which might not be the case with dependent directors who might have a conflict of interest.
Among researchers it is commonly accepted that board of directors that have a higher concentration of independent directors are better in monitoring and questioning management decisions as well as evaluating the firm’s performance (e.g. Villiers, Naiker and Van Staden, 2011; Kesner & Johnson, 1990). However, Villiers et al. (2011) argues that the presence of independent directors per se is not helpful since when they are in a minority they might be influenced by the CEO or other executives. They therefore propose to instead measure board independence in the form of concentration of independent directors on the board. Effective monitoring by non-executive directors have been found to reduce the agency costs for investors (Daily et al., 2003; Hillman & Dalziel, 2003). Likely because, in case of banks, they do not have to do as much expensive monitoring themselves (Diamond, 1984).
By ensuring higher quality risk disclosure, as well as the monitoring function needed to ensure that management discloses unfavourable information in a timely manner the independent directors provide an important governance function for lenders. However, this is only possible when the presence of independent directors on the board is large enough to resists any undue influence of the firm’s executives. We expect that the percentage of independent directors on the board to have a negative effect on the way lenders use the financial instrument risk disclosure. This brings us to the following hypothesis:
Hypothesis 2: The percentage of independent directors on the board have a negative effect on
the relation between financial instrument risk disclosure and the subsequent loan price.
Another important board characteristic that influences the effectiveness of the board ability to monitor the firms is the expertise of individual directors, and then especially financial expertise of the individual directors. Not all financial experts are valued equal, however. According to Defond, Hann and Hu (2005), market participants value accounting financial experts more than non-accounting experts. Their reasoning is that financial accounting experts on the board are better in monitoring the firms reporting process, leading to a higher quality financial reporting. Higher quality financial reporting has been linked to lower cost of capital (Lambert, Leuz and Verrecchia, 2007), possibly due to the fact that in case of lenders they have to do less expensive monitoring themselves (Diamond, 1984). This however only seems to be the case for firms who
have a strong corporate governance in place (e.g. Felo and Solieri, 2009; Defond, Hann and Hu, 2005). This suggests that in firms with less effective corporate governance mechanisms the knowledge of the accounting expert can be less effectively used.
Financial experts, especially in the field of accounting, are better in monitoring the firms reporting process leading to higher quality financial reporting. We expect that banks reward firms for their more qualitative disclosure in the form of more favourable lending terms. We expect that the number of financial experts negatively influences the relation between financial instrument risk disclosure and subsequent lending terms. Based on this we formulate the third hypothesis as:
Hypothesis 3: The number of financial experts on the board of directors has a negative effect
on the relationship between the financial instrument risk disclosure and the subsequent loan price.
2.4 CONCEPTUAL MODEL
Figure 1 - Conceptual model
H2:
H3:
-Financial Instrument Risk
Disclosure Loan Pricing
The Percentage of Independent Directors on the Board
The Number of Financial Experts on the Board
13
3
METHODOLOGY
3.1 SAMPLE
For this research we look at all UK premium listed firms, for the period 2007 – 2016. Choosing UK as a research setting provides a wide variety of benefits. One of them is the amount of research that is done in a UK setting, which enabled us to come with more accurate predictions. Secondly the UK also provides a high degree of generalizability to other western markets, increasing the usefulness of our findings to a greater number of practitioners and researchers. This study chose to use data for a ten-year period, from 2007 to 2016. The year 2007 is the obvious start date since it marks the adoption year of IFRS 7.
DealScan provides us with the data for the various syndicate loan contracts of UK premium listed firms, it however does not provide a convenient way to identify firms. Chava and Roberts (2008) provided us with the link between DealScan and Compustat, allowing us to identify firm’s GVKEY in DealScan which were subsequently used to identify firm’s ISIN codes. Using the ISIN codes we were able to match the various loan contracts with IFRS 7 data, Board of Directors data and Control variables.
Using DealScan we obtained data on 2719 syndicate loans for UK listed firms in the period 2007 – 2016. Of those loans 1368 was missing some data on spread or maturity and thus were dropped from the sample, for 1003 other loans there was no corresponding IFRS 7 data available and thus were also excluded from the sample. And lastly, for 25 other loans there were missing data on control variables and thus were also dropped from the sample. The total sample used to answer hypothesis 1 consists of 323 observations. We also have to adjust the sample some more to be able to use them to answer hypothesis 2 and hypothesis 3, 23 observations had to be dropped due to missing data on the Board of Directors. See Table 1 below for the complete breakdown of the sample.
TABLE 1
COMPOSITION OF THE SAMPLES
Total Number of Loan Contracts 2719
Missing Data on Loan Contracts -1368
Missing Data on IFRS 7 Disclosures -1003
Missing Data on Control Variables -25
Sample Size Research Question 1 323
Missing Data on Board of Directors -21
3.2 RESEARCH DESIGN
For this research we make use of an Ordinary Least Square (OLS) regression to test our various hypotheses. We make use of an OLS regression because we make use of a data sample consisting out of continuous data and dummy variables. Each observation refers to a loan, we cannot use firm as identifier because a firm can have multiple loans in the same year. To test our first hypothesis, we make use of the following model:
𝑆𝑝𝑟𝑒𝑎𝑑𝑖 = 𝛽0 + 𝛽1 𝐼𝐹𝑅𝑆7𝑖 + 𝛽2 𝑆𝑖𝑧𝑒𝑖 + 𝛽3 𝑅𝑂𝐴𝑖 + 𝛽4 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖 + 𝛽5 𝑀𝑎𝑡𝑢𝑟𝑖𝑡𝑦𝑖 + 𝛽6 𝐴𝑚𝑜𝑢𝑛𝑡𝑖 + 𝛽7 𝑁𝑟𝐿𝑒𝑛𝑑𝑒𝑟𝑠𝑖 + 𝛽8 𝐿𝑜𝑎𝑛𝑇𝑦𝑝𝑒𝑖 + 𝛽9 𝑌𝑒𝑎𝑟_𝐷𝑖 + 𝛽10 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦_𝐷𝑖 + 𝛽11𝐶𝑜𝑢𝑛𝑡𝑟𝑦𝑜𝑓𝑆𝑦𝑛𝑑𝑖𝑐𝑎𝑡𝑖𝑜𝑛_𝐷𝑖 + 𝜀𝑖
Model 1
In the above model IFRS7 can stand for either the principle component of IFRS 7 or coverage. The formula is applied three different times. The first time using the principle component, the second time using coverage and the last time using both variables in the regression.
To test our second hypothesis, we make use of a slightly different model, in this model we have added the interaction variable 𝐼𝐹𝑅𝑆7𝑖 𝑥 𝐵𝑜𝑎𝑟𝑑_𝐼𝑛𝑑𝑖 . This variable is made by calculating the interaction terms between one of the three possible forms of IFRS7, as mentioned earlier, and the percentage of independent directors on the board. To test our second hypothesis, we make use of model 2 presented below:
𝑆𝑝𝑟𝑒𝑎𝑑𝑖= 𝛽0 + 𝛽1 𝐼𝐹𝑅𝑆7𝑖 + 𝛽2 𝐵𝑜𝑎𝑟𝑑_𝐼𝑛𝑑𝑖 + 𝛽3 𝐼𝐹𝑅𝑆7𝑖 𝑥 𝐵𝑜𝑎𝑟𝑑_𝐼𝑛𝑑𝑖 + 𝛽4 𝑆𝑖𝑧𝑒𝑖 + 𝛽5 𝑅𝑂𝐴𝑖
+ 𝛽6 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖 + 𝛽7 𝑀𝑎𝑡𝑢𝑟𝑖𝑡𝑦𝑖+ 𝛽8 𝐴𝑚𝑜𝑢𝑛𝑡𝑖 + 𝛽9 𝑁𝑟𝐿𝑒𝑛𝑑𝑒𝑟𝑠𝑖 + 𝛽10 𝐿𝑜𝑎𝑛𝑇𝑦𝑝𝑒𝑖
+ 𝛽11 𝑌𝑒𝑎𝑟_𝐷𝑖 + 𝛽12 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦_𝐷𝑖 + 𝛽13 𝐶𝑜𝑢𝑛𝑡𝑟𝑦𝑜𝑓𝑆𝑦𝑛𝑑𝑖𝑐𝑎𝑡𝑖𝑜𝑛_𝐷𝑖 + 𝜀𝑖
Model 2
For our last hypothesis we make use of an interaction term between IFRS7 and the number of financial experts on the board, this is represented by 𝐼𝐹𝑅𝑆7𝑖 𝑥 𝐹𝑖𝑛𝐸𝑥𝑖 . To test of third and final hypothesis we make use of model 3:
𝑆𝑝𝑟𝑒𝑎𝑑𝑖= 𝛽0 + 𝛽1 𝐼𝐹𝑅𝑆7𝑖 + 𝛽2 𝐹𝑖𝑛𝐸𝑥𝑖 + 𝛽3 𝐼𝐹𝑅𝑆7𝑖 𝑥 𝐹𝑖𝑛𝐸𝑥𝑖 + 𝛽4 𝑆𝑖𝑧𝑒𝑖 + 𝛽5 𝑅𝑂𝐴𝑖 + 𝛽6 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖
+ 𝛽7 𝑀𝑎𝑡𝑢𝑟𝑖𝑡𝑦𝑖+ 𝛽8 𝐴𝑚𝑜𝑢𝑛𝑡𝑖 + 𝛽9 𝑁𝑟𝐿𝑒𝑛𝑑𝑒𝑟𝑠𝑖 + 𝛽10 𝐿𝑜𝑎𝑛𝑇𝑦𝑝𝑒𝑖 + 𝛽11 𝑌𝑒𝑎𝑟_𝐷𝑖
+ 𝛽12 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦_𝐷𝑖 + 𝛽13 𝐶𝑜𝑢𝑛𝑡𝑟𝑦𝑜𝑓𝑆𝑦𝑛𝑑𝑖𝑐𝑎𝑡𝑖𝑜𝑛_𝐷𝑖 + 𝜀𝑖
15 3.3 VARIABLES
3.3.1. Dependent variable
The primary goal of this study is to learn how the financial instrument risk disclosure influences the loan pricing of the syndicated loan. In the previous chapter we briefly discussed the spread as a metric for loan pricing.
▪ Spread: The spread is the interest rate the borrower has to pay over the LIBOR rate. However, DealScan only provides the “all-in-spread” (AIS). The AIS contains beside the spread also the annual fees the borrower has to pay to the syndicate, so called service costs. For the remainder of this research we will be referring to this AIS as the spread of the loan. We made use of Thomson Reuters DealScan to obtain the data with regard to the Spread charged on syndicate loans.
3.3.2. Independent variables
For this research we are interested in understanding how the financial instrument risk disclosure influences the loan pricing of syndicated loans. Miihkinen (2012) proposes a couple of metrics to measure the various quality aspects of the financial instrument risk disclosure. The metrics that are relevant for this research are respectively the quantity of IFRS 7, number of risk categories, the usage of a hard disclosure and the coverage. The first three variables will be combined using a Principle Component Analysis, we do this because it is likely that the combination of the various aspects of the IFRS 7 influences the perceived riskiness rather than only one metric of the disclosure. All data was hand collected from financial instrument risk disclosures found in annual reports.
▪ Principle Component IFRS7: We computed a principle component using the following three IFRS 7 metrics:
❖ Quantity IFRS 7: In our literature review we discussed how we expect firms to disclose meaningful and useful information in their financial instrument risk disclosure. If this is indeed the case, firms with lengthier disclosure are therefore expected to be perceived as riskier. We measure this aspect of IFRS 7 by taking the natural logarithm of the total number of words used in the IFRS 7 disclosure (Miihkinen, 2012).
❖ Number of Risk Categories: This is the total number of risk categories used in the disclosure of the financial instrument risks. We have identified a total of six different
risk categories, which are respectively interest rate risk, foreign currency risk, other price risk, liquidity risk, credit risk and remaining risk. This is measured by the total number of risk categories that are identified in the IFRS 7 risk disclosure.
❖ Hard Disclosure: And lastly, we also take into consideration if the firm made use of tables when explaining their risks various risk. This will be measured by calculating the percentage of the risk categories that is supported by a hard disclosure.
▪ Coverage: For an investor to properly understand the risks a company face, a firm has to provide a balanced description of the various risk categories (Miihkinen, 2012). To measure whether firms provide a balanced description of the risk categories we compute scores for a firm’s IFRS 7 coverage using the method described by Miihkinen, 2012:
𝑪𝒐𝒗𝒆𝒓𝒂𝒈𝒆 = (𝟏 𝑯⁄ )
𝑵𝒖𝒎𝒃𝒆𝒓 𝒐𝒇 𝑹𝒊𝒔𝒌 𝑪𝒂𝒕𝒆𝒈𝒐𝒓𝒊𝒆𝒔
The H is the concentration of the risk disclosure measured using the Herfindahl Index. We make use of an inversed Herfindahl score to provide us with a more accurate score.
3.3.3. Moderating variables
Our second research question concerns itself with the question of how specific board of director’s characteristics moderate the way the IFRS 7 disclosures are being used by the syndicated loan market. In the previous chapter we formulated two hypotheses with regard to the board of directors, the first one focusing on the percentage of the board that qualifies as independent and the second one focused on the number of financial experts on the board. We made use of BoardEx to obtain the data regarding those characteristics.
▪ Board Independence (%): Is the percentage of directors on a board that are independent. As we discussed previously, when a larger percentage of the board is independent it is assumed that they are less likely being influence by the CEO and thus better at monitoring management (Villiers et al., 2011). To determine who is independent we make use of the BoardEx database, in here they disclose what role the director fulfils on the board (e.g. Independent Director). Board independence (%) is measured by calculated the percentage of the board that qualifies as independent.
▪ Number of Financial Experts: Is the total number of financial experts that sit on the board of directors. We choose to qualify someone as a financial expert when he or she holds an accounting qualification (e.g. Certified Public Accountants and Chartered Public
17 Accountants). As was outlined in our theory section, it seems that investors mostly value
the presence of accountants compared to other types of financial experts (Defond, Hann and Hu, 2005). This variable is measured by counting the number of directors on the board holding an accounting qualification.
3.3.4. Control variables
During this study we also control for certain effects that have been found to infer or possibly infer with the way the financial instrument risk disclosure affects the loan terms provided by the debt market.
▪ Firms Size: Firm size has been found to be an important driver of disclosure quality (e.g. Elshandidy et al., 2013; Miihkinen, 2012; Abraham and Cox, 2007). There are different methods used to measure the size of the firm, some refer to the sales of the firm (Miihkinen, 2012; Abraham and Cox, 2007) while others refer to the market capitalisation (Elshandidy et al., 2013). We think that using net sales is the most appropriate way to measure firm size since most of the categories of financial instrument risk we identified earlier are related to a firm’s normal business operations (e.g. foreign currency and credit risk). Firm size is measured as the natural logarithm of net sales (Miihkinen, 2012; Abraham and Cox, 2007). The data is obtained from WRDS Compustat Global.
▪ Profitability: Secondly, we also control for firm profitability. Profitability has been commonly controlled for by many researchers in studies done on firm disclosure practices (e.g. Elshandidy and Neri, 2015; Miihkinen, 2012; Abraham and Cox, 2007). Two commonly used metrics for firm profitability are the return on equity (e.g. Elshandidy and Neri, 2015) and return on assets (e.g. Miihkinen, 2012). Since lenders do not share in the returns shareholders make and are therefore not interested in the return on equity, because of this we choose to measure profitability in terms of return on assets. The data is obtained from WRDS Compustat Global.
▪ Leverage: How Leverage exactly influences the disclosure practices of a firm is currently undecided. On one hand you have studies that found a positive relation between leverage and disclosure quality (e.g. Elshandidy et al., 2013), other studies found a negative relation between those two (e.g. Miihkinen, 2012) and some found no significant relation at all (e.g. Abraham and Cox, 2007). We measure leverage as the ratio total liabilities to total assets (Abraham and Cox, 2007). The data is obtained from WRDS Compustat Global.
▪ Maturity: The maturity of the loan is another way for lenders to mitigate risk that is associated with lending money to third parties. It is therefore no surprise that it a commonly
looked at loan characteristic when researching contracting terms (e.g. Kim, Song and Zhang, 2011; Kim, Tsui and Yi, 2011; Porumb et al., 2019). The maturity of a loan is the duration of the loan until the end date. Maturity will be measured by taking the natural logarithm of the maturity in months, the data with regards to the maturity of the loan is obtained from Thomson Reuters DealScan.
▪ Amount (Loan Size): The size of the loan is also an important characteristic of a loan; therefore, we will also control for this effect. The natural logarithm of the loan amount will be used, loan amount is measured in millions. Data about loan size is obtained from Thomson Reuters DealScan.
▪ Number of Lenders: Another loan characteristic we control for the number of lenders that are involved in the syndication of the loan. When working with syndicate loans one has to control for the numbers of lenders involved because it functions as a way for lenders to mitigate risks (e.g. Simons, 1993). Number of lenders is measured as the total number of lenders that are involved in the syndication of the loan. The data with regards to the numbers of lenders involved in the loan is obtained from Thomson Reuters DealScan.
▪ Loan Type: The last loan characteristic we control for is the loan type. We control for whether the loan is a term loan or another type of loan, a term loan has a fixed repayment schedule and is therefore often considered as being less risky. We create a dummy variable which takes a value of 1 for terms loans and the value of 0 if another type of loan. The data with regards to the loan type of the loan is obtained from Thomson Reuters DealScan.
3.3.5. Dummy Variables
▪ Year: We will create a dummy variable for the various years the loans are syndicated. Controlling for year is a common practice when researching loan contracts (e.g. Porumb et al., 2019; Kim, Song and Zhang, 2011).
▪ Industry: The industry in which the firm operates is an often controlled for variable when researching loan contracts of firms (e.g. Porumb et al., 2019; Miihkinen, 2012; Kim, Song and Zhang, 2011). We will control for the different industries by creating a dummy variable based on the industry classifications of Fama and French. Using the Fama and French (1988) method, firms will be assigned to one of seventeen groups based on their Standard Industrial Classification (SIC) code.
▪ Country of Syndication: Also, the country of syndication might have some effect on the loan terms. Countries across the world can have rather large differences in legislation for investor protection and the enforcement of those legislation (La Porta, Lopez-de-Silanes,
19 Shleifer and Vishny, 1998). The implication of this is that also the risk for lenders varies
from country to country and will affect how strict lending terms might be. We control for this effect by creating a dummy variable based on the country of syndication, for our sample this means that our loan can be syndicated in one of eleven different countries.
TABLE 2
LIST OF VARIABLES
Variable Description
Dependent variables
Spread The basis points that have to be paid over LIBOR.
Independent variables
Quantity IFRS 7 Is measured as the natural logarithm of (1+ total number of words used in the IFRS7 disclosure).
Number of Risk Categories The total number of risk categories that have been discussed in the IFRS 7 risk disclosure.
Hard Disclosure Measured as the natural logarithm of (1 + percentage of risk categories that is supported with a hard disclosure).
Principle Component IFRS7 The outcome of the principle component analysis using the quantitative aspect of IFRS 7, the total number of risk categories and the percentage of risk categories that is supported by a table.
Coverage Coverage is computed as (1 / H) / the number of main risk categories. The H is the Herfindahl index score, for the computation please see 3.3.2 in the text.
Moderating variables
Board Independence (%) The percentage of the board that qualifies as independent. Number of Financial Experts The number of financial experts on the board.
Control variables
Firm Size Calculated as the natural logarithm of (1 + Net Sales). Profitability The ratio Net Income to Total Assets.
Leverage The ratio Total Liabilities to Total Assets.
Maturity The natural logarithm of the duration of the loan measured in months.
Amount The loan amount measured as the natural logarithm of the loan size measured in millions.
Number of Lenders Number of lenders involved in the syndication process of the loan.
Loan Type Taking the value of 1 for term loans and the value of 0 for all other kinds of loans.
Year Dummy Dummy variable for the year in which the loan has been syndicated.
Industry Dummy A dummy variable used to control for the industry in which the firm operates.
Country of Syndication Dummy A dummy variable for the different countries in which the loan contract can be syndicated.
4
RESULTS
4.1 DESCRIPTIVE STATISTICS
In table 3 we present the descriptive statistics of the variables used in the three models. The first sample is used to test hypothesis 1 (model 1), while the second sample is used to test both hypothesis 2 (model 2) and hypothesis 3 (model 3).
TABLE 3
Sample used for Model 1 Sample used for Model 2 & Model 3
Variables Obs. Mean Std. Dev. Min. Max Obs. Mean Std. Dev. Min. Max
Dependent Variables
Spread 323 227,263 141,557 20 725 302 232,635 141,725 20 725
Independent Variables
Quantity IFRS 7 323 7,397 0,490 5,799 8,488 302 7,406 0,501 5,799 8,448
Number of Risk Categories 323 4,390 0,8430 2 6 302 4,427 0,811 2 6
Hard Disclosure 323 1,210 0,478 0 1,946 302 1,213 0,482 0 1,946
Principle Component 323 0,011 0,947 -1,95 1,621 302 0,011 0,959 -1,95 1,621
Coverage 323 0,888 0,174 0,401 1,201 302 0,882 0,176 0,401 1,201
Moderating Variables
Board Independence (%) - - - 302 0,563 0,133 0,167 0,846
Number of Financial Experts - - - 302 1,513 1,117 0 4
Control Variables Maturity 317 3,788 0,487 2,303 4,970 296 3,803 0,473 2,303 4,970 Number of Lenders 297 8,721 6,471 1 27 276 8,681 6,608 1 27 Amount 323 5,772 1,329 2,235 8,599 302 5,733 1,330 2,235 8,599 Loan Type 323 0,272 0,446 0 1 302 0,265 0,442 0 1 Size 323 7,364 1,482 3,663 12,846 302 7,331 1,504 3,663 12,846 ROA 323 0,024 0,076 -0,27 0,239 302 0,024 0,069 -0,27 0,239 Leverage 323 0,628 0,209 0,065 1,298 302 0,636 0,207 0,065 1,298
All the variables have been winsorized on a 1% and 99% level, except the Number of Risk Categories and the Loan Type.
All the continuous data variables have been winsorized at the 1% and 99% level. The remaining variables, the discrete variables, have not been winsorized. The average spread is around 227 basis points in the sample used for model 1, the spread is around 232 basis points in the sample used for model 2 and model 3.
4.2 CORRELATION MATRIX
Next, we will discuss our correlation matrix for our sample. In table 4 on the next page we present the results of our correlation analysis, here the results of sample 2 are presented. We choose to only show the results for the second sample since the only difference is in a few decimals, the overall picture is the same.
21 The only correlations that are worth mentioning are rather interesting, which are the correlations between the loan size, the number of lenders and
the firm size. That the size of the loan and firm size are correlated (0,64) with each other is rather logical, larger firms tend to have a larger financing need and are therefore expected to take on larger loans compared to smaller firms. Also, the correlation between number of lenders and loan size is explainable. Preferably, lenders do not want that one loan has a too large share of its loan portfolio due to the risks for the bank. So, when dealing with large loans more lenders will be involved to mitigate this risk. And lastly, we have the correlation between number of lenders and firm size (0,52). Larger firms tend to borrow larger amount, the loan therefore is either too large for one lender to handle or gives too much exposure for the lender to a single firm in both cases raising the need for loan syndication.
TABLE 4
CORRELATION MATRIX
VARIABLES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12]
Spread [1] -
Principle Component IFRS7 [2] 0,204 -
Coverage [3] 0,002 0,297 - Board Independence [4] -0,330 -0,112 -0,133 - Financial Expertise [5] -0,281 -0,060 0,086 -0,001 - Maturity [6] 0,040 0,135 0,106 -0,028 -0,066 - Number of Lenders [7] -0,241 -0,102 -0,094 0,284 -0,128 -0,008 - Amount [8] -0,346 -0,083 -0,025 0,359 -0,026 0,009 0,610 - Loan Type [9] 0,143 0,018 -0,057 -0,049 -0,081 0,119 -0,133 -0,068 - Size [10] -0,228 -0,051 -0,104 0,379 -0,123 -0,081 0,525 0,643 0,040 - ROA [11] -0,155 -0,020 0,077 -0,055 0,032 0,095 -0,047 0,047 -0,073 -0,028 - Leverage [12] 0,135 -0,054 -0,147 -0,066 -0,114 -0,047 0,053 0,085 0,075 0,300 0,007 - All the variables presented above are displayed and explained in Table 2 on page 18. For this correlation matrix above we make use of sample 2, sample 2 is almost the same as the sample except for 20 observation which have been dropped due to missing observation on board of directors. The differences between sample 1 and sample 2 is neglectable since the difference is only a few decimal points.
4.3 MULTIVARIATE ANALYSIS
On the next page we present the results of our multivariate analyses of model 1, model 2 and model 3 in table 5, we will discuss our observations with regard to each of the three models and the corresponding hypothesis. In our first hypothesis we expected that the usage of IFRS 7 disclosures would have a positive effect on the subsequent loan price, this due to an increase in the perceived riskiness by lenders. For model 1 we observe that the disclosure quality of the IFRS 7 disclosure to have a significant positive effect on loan price (P<0.05), seemingly confirming our expectations. But also, in model 2 and model 3 we observe that the principle component has a positive significant effect on loan price (P<0.05; P<0.10). All in all, we find enough evidence to support our hypothesis that IFRS 7risk disclosures have a positive effect on loan price. Therefore, we can accept our fist hypothesis:
Hypothesis 1: The financial instrument risk disclosure has a positive effect on the subsequent
loan price of the firm.
Our second model, presented in table 5, focuses on the effect that board independence has on the way lenders use the financial instrument risk disclosure in determining loan price. We theorized and found that the interaction between IFRS 7 and board independence to have a negative effect on the subsequent loan price. We expected this because it limits the extent in which managers can disclose opportunistically, resulting in a complete and timely disclosure of unfavourable risk information. We find enough evidence to support our claim that board independence has a negative effect on the way lenders use the financial instrument risk disclosure. Therefore, we can accept our second hypothesis:
Hypothesis 2: The percentage of independent directors on the board have a negative effect on
the relation between financial instrument risk disclosure and the subsequent loan price.
In our third and last model, also presented in table 5, we focus on the effect that the number of financial experts have on the way financial instrument risk disclosure is being used in determining loan price. We argued that financial expertise should lead to a better monitoring of the financial reporting process of a firm, giving additional assurance to lenders that the disclosure is complete and timely. Based on this we expected that a larger number of financial experts on the board to have negative effect on the way lenders use IFRS 7 disclosures, leading to a lower subsequent loan price. Our findings however do not support this, we do not find
23 evidence that the interaction between IFRS 7 and financial expertise to have a significant effect
on loan price. Therefore, we must reject our third hypothesis:
Hypothesis 3: The number of financial experts on the board of directors has a negative effect
on the relationship between the financial instrument risk disclosure and the subsequent loan price.
With regard to our control variables in all three the models we do not observe anything out of the ordinary. Almost all significant effects are in line with how one might expect the respective control variables to influence the firms loan price. The only control variable that is somewhat ambiguous is the negative effect that loan size has on loan price, we think that this effect is because of the large correlation that exists between firm size and loan size.
TABLE 5
RESULTS OF MULTIVARIATE REGRESSION
Model 1 Model 2 Model 3
1a 1b 1c 2a 2b 2c 3a 3b 3c
VARIABLES Spread Spread Spread Spread Spread Spread Spread Spread Spread
CPA_IFRS7 0.0984** 0.106** 0.0847* 0.0882* 0.0869* 0.101** (0.0442) (0.0453) (0.0471) (0.0485) (0.0455) (0.0471) coverage -0.0122 -0.0345 -0.0486 -0.0612 -0.0334 -0.0538 (0.0432) (0.0439) (0.0480) (0.0482) (0.0474) (0.0488) PCA_IFRS7 x Board_Ind -0.103* -0.112* (0.0567) (0.0581) coverage x Board_Ind 0.0655 0.0789 (0.0570) (0.0573) Board_Ind -0.0864 -0.144** -0.131** (0.0547) (0.0615) (0.0607) PCA_IFRS7 x FinEx 0.0508 0.0364 (0.0427) (0.0447) coverage x FinEx 0.0436 0.0437 (0.0383) (0.0400) FinEx -0.152*** -0.178*** -0.160*** (0.0442) (0.0434) (0.0448) Size -0.259*** -0.250*** -0.261*** -0.176* -0.151 -0.167* -0.209** -0.203** -0.227** (0.0914) (0.0921) (0.0914) (0.100) (0.101) (0.100) (0.0951) (0.0964) (0.0962) ROA -0.0418 -0.0452 -0.0409 -0.0567 -0.0505 -0.0586 -0.0585 -0.0470 -0.0529 (0.0390) (0.0394) (0.0391) (0.0477) (0.0483) (0.0476) (0.0467) (0.0471) (0.0470) Leverage 0.241*** 0.238*** 0.243*** 0.202*** 0.189*** 0.182*** 0.213*** 0.190*** 0.203*** (0.0520) (0.0526) (0.0521) (0.0577) (0.0596) (0.0597) (0.0555) (0.0567) (0.0568) Maturity 0.00109 0.0129 0.00262 0.0135 0.0113 0.0134 0.000173 0.00304 0.00178 (0.0474) (0.0477) (0.0475) (0.0522) (0.0528) (0.0522) (0.0511) (0.0513) (0.0512) Amount -0.145** -0.156** -0.139** -0.136** -0.138** -0.122* -0.134** -0.130* -0.111*
(0.0616) (0.0623) (0.0622) (0.0656) (0.0673) (0.0666) (0.0641) (0.0665) (0.0665) NLenders -0.0385 -0.0434 -0.0410 -0.0270 -0.0459 -0.0341 -0.0748 -0.0857* -0.0853* (0.0478) (0.0484) (0.0479) (0.0500) (0.0506) (0.0501) (0.0496) (0.0506) (0.0503) LoanType 0.168* 0.166* 0.168* 0.179* 0.194** 0.190** 0.166* 0.189** 0.175* (0.0888) (0.0896) (0.0888) (0.0957) (0.0973) (0.0960) (0.0944) (0.0948) (0.0949) Constant -0.963 -0.933 -0.961 0.332* 0.344* 0.376* 0.260 0.262 0.249 (0.846) (0.854) (0.847) (0.194) (0.196) (0.196) (0.187) (0.187) (0.187)
Year Dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes
Industry Dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes
Country of Syndication
Dummy Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 291 291 291 270 270 270 270 270 270
R-squared 0.533 0.524 0.534 0.523 0.510 0.528 0.541 0.533 0.544 *** p<0.01, ** p<0.05, * p<0.1. Standard errors are presented in parentheses. All regression variables have been standardized, this means they all have a mean of zero and a standard deviation of one. The results of model 1 are calculated using sample 1, for model 2 and model 3 we made use of sample 2. About our control variables we do not observe any irregularities, all the control variables that are significant have an effect that is in line with what one might expect.
4.4 ADDITIONAL TESTS
4.4.1. Pre-Post Analysis
The first additional test we are going to do is a Pre-Post analysis. In 2007 the disclosure of financial instrument risks became mandatory for all stock listed firms in the UK. Using a Pre-Post analysis, we are going to test whether there are observable differences in loan price between the Pre-adoption period compared to the Post-adoption period. Testing whether there is a difference between the two periods limits the possibility that our findings are due to “coincidence” rather than the implementation of IFRS 7 in 2007.
To test this, we created a dummy variable for the period 2005 – 2008 which will replace the IFRS 7 variable in our three models. This dummy variable takes the value of 0 for loan in the period 2005 – 2006 and the value of 1 for loans in the period 2007 – 2008. This dummy variable will be used in the three models replacing the IFRS 7 variable.
We present our findings below in table 6. Our findings clearly show that firm have to pay a significantly (P<0.01) higher loan price in the after the adoption of IFRS 7. This shows that it is likely that IFRS 7 indeed increased the perceived riskiness and that our prior observations are thus correct.
25
TABLE 6
Additional Test - Pre-Post Analysis
Model 1 Model 2 Model 3
VARIABLES Spread Spread Spread
PrePost (Pre=0; Post=1) 0.692*** 0.515*** 0.506***
(0.143) (0.133) (0.136) Board_Ind -0.140** (0.0559) FinEx -0.0763 (0.0519) Size -0.110 -0.0519 -0.101 (0.101) (0.101) (0.100) ROA -0.248*** -0.235*** -0.267*** (0.0620) (0.0609) (0.0631) Leverage 0.0898** 0.0919** 0.109** (0.0450) (0.0441) (0.0468) Maturity 0.150*** 0.134*** 0.152*** (0.0380) (0.0378) (0.0379) Amount -0.101 -0.115* -0.0921 (0.0665) (0.0654) (0.0664) NLenders -0.0795* -0.0625 -0.103** (0.0460) (0.0455) (0.0484) LoanType 0.0697 0.0654 0.0899 (0.0931) (0.0913) (0.0937) Constant 0.346 0.299 0.309 (0.540) (0.530) (0.538)
Year Dummy Yes Yes Yes
Country of Syndication Dummy Yes Yes Yes
Industry Dummy Yes Yes Yes
Observations 154 154 154
R-squared 0.551 0.583 0.572
*** p<0.01, ** p<0.05, * p<0.1. Standard errors are presented in parentheses. All regression variables have been standardized, this means they all have a mean of zero and a standard deviation of one. The results of model 1 are calculated using sample 1, for model 2 and model 3 we made use of sample 2. About our control variables we do not observe any irregularities, all the control variables that are significant have an effect that is in line with what one might expect.
4.4.2. Differences between Systematic Risk and Unsystematic Risk
Our second additional test focuses whether there is a significant difference in effect on loan price between the systematic risk components of IFRS 7 and the disclosed unsystematic risks. The idea behind having a diversified loan portfolio is that one can eliminate the exposure to firm and industry specific risks (unsystematic risks) and thus only has to deal with the exposure to market risks (unsystematic risk). With the assumption that banks have a diversified loan portfolio, thus eliminating the exposure to unsystematic risks, they should therefore be more interested in the disclosure of systematic risks in the IFRS 7 disclosure. We therefore predict that lenders should be more interested in the systematic risks disclosed in the financial instrument risk disclosure than the unsystematic risks.
We are going to test this by applying our three models again, this time making a distinction between the systematic risk and unsystematic risk related to the quantitative section and the coverage of the disclosure.
We predict and find that the disclosure of systematic risks in the IFRS 7 disclosure are in general more important than the disclosure of unsystematic risks. For model 1, table 7, we observe a slight positive effect (P<0.05) of the systematic risk’s principle component of IFRS 7 on loan price. However, this is only observed when looking at the principle component in isolation of the coverage variable. In table 8 we present the results for model 2. Here we also observe the effect of the principle component on the systematic risks side (P<0.10). For the interaction effect of board independence with the IFRS 7 variables we observe three significant effects on the systematic risks side (P<0.01; P<0.05; P<0.10) and two significant effects on the unsystematic risks side (P<0.05; P<0.10). Overall, the lenders seem to favour disclosure of systematic risks, but the evidence is not strong enough to draw a definitive conclusion. And lastly, we have model 3 of which the results are presented in table 9. For this model we observe significant effects of the systematic risk disclosed in IFRS 7 on loan price, but no significant effects on the unsystematic risks side. In general, the effects are in line with what we expected, except the positive interaction effect between the systematic risk’s principle component and financial expertise of the board (P<0.10). A possible explanation might be that boards with more financial expertise discloses information that is different in content, this study however does not look at the content of the disclosure and can therefore not confirm this. This however does not affect our general observation that all the observed effects of IFRS 7 are mostly due to the disclosure of systematic risks.
27
TABLE 7
Additional Test - Systematic Risk versus Unsystematic Risk - Model 1
Systematic risks Unsystematic risks
1 2 3 1 2 3
VARIABLES Spread Spread Spread Spread Spread Spread PCA_IFRS7_Sys 0.0984** 0.148 (0.0442) (0.227) coverage_sys 0.0629 0.0465 (0.0462) (0.0465) PCA_IFRS7_Unsys 0.0394 0.113 (0.0671) (0.114) coverage_unsys 0.0512 0.0723 (0.0424) (0.0486) Size -0.259*** -0.239** -0.247*** -0.262*** -0.239** -0.281*** (0.0914) (0.0922) (0.0928) (0.0945) (0.0928) (0.0943) ROA -0.0418 -0.0542 -0.0655 -0.0431 -0.0479 -0.0618 (0.0390) (0.0398) (0.0412) (0.0395) (0.0393) (0.0405) Leverage 0.241*** 0.249*** 0.254*** 0.242*** 0.239*** 0.263*** (0.0520) (0.0530) (0.0528) (0.0529) (0.0525) (0.0530) Maturity 0.00109 0.0122 0.0125 0.0105 0.0158 0.0207 (0.0474) (0.0474) (0.0479) (0.0476) (0.0476) (0.0479) Amount -0.145** -0.159** -0.149** -0.165*** -0.146** -0.149** (0.0616) (0.0617) (0.0616) (0.0628) (0.0627) (0.0640) NLenders -0.0385 -0.0358 -0.0366 -0.0344 -0.0504 -0.0311 (0.0478) (0.0483) (0.0482) (0.0501) (0.0492) (0.0500) LoanType 0.168* 0.156* 0.144 0.169* 0.175* 0.158* (0.0888) (0.0896) (0.0895) (0.0897) (0.0900) (0.0893) Constant -0.963 -0.904 -0.512 -0.887 -0.0285 -0.113 (0.846) (0.852) (0.909) (0.858) (0.863) (0.954)
Year Dummy Yes Yes Yes Yes Yes Yes
Country of Syndication Yes Yes Yes Yes Yes Yes
Industry Dummy Yes Yes Yes Yes Yes Yes
Observations 291 291 291 291 290 290
R-squared 0.533 0.527 0.540 0.524 0.527 0.544
*** p<0.01, ** p<0.05, * p<0.1. Standard errors are presented in parentheses. All regression variables have been standardized, this means they all have a mean of zero and a standard deviation of one. About our control variables we do not observe any irregularities, all the control variables that are significant have an effect that is in line with what one might expect.
TABLE 8
Systematic Risk versus Unsystematic Risk - Model 2
Systematic risks Unsystematic risks
1 2 3 1 2 3
VARIABLES Spread Spread Spread Spread Spread Spread PCA_IFRS7_Sys 0.0847* 0.0662 (0.0471) (0.0474) PCA_IFRS7_Sys x Board_Ind -0.103* -0.0729 (0.0567) (0.0575) coverage_sys 0.0711 0.0586 (0.0489) (0.0493) coverage_sys x Board_Ind -0.114*** -0.0928** (0.0388) (0.0400) PCA_IFRS7_Unsys -0.0111 -0.00272 (0.0734) (0.108) PCA_IFRS7_Unsys x Board_Ind 0.113 0.0404 (0.0946) (0.104) coverage_unsys 0.0652 0.0665 (0.0439) (0.0492) coverage_unsys x Board_Ind -0.153** -0.141* (0.0681) (0.0759) Board_Ind -0.0864 -0.0761 -0.0671 -0.114** -0.127** -0.129** (0.0547) (0.0552) (0.0552) (0.0559) (0.0550) (0.0559) Size -0.176* -0.154 -0.168* -0.167 -0.162 -0.163 (0.100) (0.0995) (0.0993) (0.105) (0.102) (0.105) ROA -0.0567 -0.0631 -0.0660 -0.0425 -0.0387 -0.0371 (0.0477) (0.0490) (0.0488) (0.0486) (0.0480) (0.0484) Leverage 0.202*** 0.183*** 0.186*** 0.218*** 0.203*** 0.207*** (0.0577) (0.0598) (0.0602) (0.0596) (0.0575) (0.0596) Maturity 0.0135 0.00979 0.0109 0.00951 0.0368 0.0345 (0.0522) (0.0517) (0.0518) (0.0527) (0.0528) (0.0534) Amount -0.136** -0.138** -0.130** -0.153** -0.136** -0.137** (0.0656) (0.0653) (0.0651) (0.0677) (0.0668) (0.0681) NLenders -0.0270 -0.0521 -0.0408 -0.0440 -0.0532 -0.0545 (0.0500) (0.0499) (0.0501) (0.0526) (0.0508) (0.0526) LoanType 0.179* 0.168* 0.167* 0.178* 0.193** 0.191** (0.0957) (0.0964) (0.0960) (0.0973) (0.0965) (0.0970) Constant 0.332* 0.330* 0.339* 0.276 0.301 0.286 (0.194) (0.191) (0.193) (0.197) (0.193) (0.198)
Year Dummy Yes Yes Yes Yes Yes Yes
Country of Syndication Dummy Yes Yes Yes Yes Yes Yes
Industry Dummy Yes Yes Yes Yes Yes Yes
Observations 270 270 270 270 269 269
R-squared 0.523 0.527 0.536 0.509 0.521 0.521
*** p<0.01, ** p<0.05, * p<0.1. Standard errors are presented in parentheses. All regression variables have been standardized, this means they all have a mean of zero and a standard deviation of one. About our control variables we do not observe any irregularities, all the control variables that are significant have an effect that is in line with what one might expect.
29
TABLE 9
Systematic Risk versus Unsystematic Risk - Model 3
Systematic risks Unsystematic risks
1 2 3 1 2 3
VARIABLES Spread Spread Spread Spread Spread Spread PCA_IFRS7_Sys 0.0869* 0.0639 (0.0455) (0.0457) PCA_IFRS7_Sys x FinEx 0.0508 0.0808* (0.0427) (0.0439) coverage_sys 0.108** 0.128** (0.0500) (0.0518) coverage_sys x FinEx -0.0991** -0.0941** (0.0410) (0.0415) PCA_IFRS7_Unsys -0.00645 0.0207 (0.0703) (0.103) PCA_IFRS7_Unsys x FinEx -0.00141 -0.00406 (0.0585) (0.0810) coverage_unsys 0.0553 0.0602 (0.0431) (0.0508) coverage_unsys x FinEx 0.0221 0.0201 (0.0372) (0.0491) FinEx -0.152*** -0.157*** -0.133*** -0.173*** -0.173*** -0.172*** (0.0442) (0.0434) (0.0443) (0.0440) (0.0435) (0.0442) Size -0.209** -0.148 -0.175* -0.182* -0.165* -0.170* (0.0951) (0.0947) (0.0946) (0.0987) (0.0970) (0.101) ROA -0.0585 -0.0797* -0.0941* -0.0514 -0.0519 -0.0512 (0.0467) (0.0480) (0.0482) (0.0471) (0.0471) (0.0475) Leverage 0.213*** 0.170*** 0.192*** 0.200*** 0.199*** 0.201*** (0.0555) (0.0600) (0.0602) (0.0567) (0.0560) (0.0573) Maturity 0.000173 -0.0100 -0.00414 0.000591 0.00215 0.00163 (0.0511) (0.0509) (0.0508) (0.0520) (0.0517) (0.0522) Amount -0.134** -0.143** -0.126** -0.148** -0.141** -0.143** (0.0641) (0.0634) (0.0632) (0.0657) (0.0657) (0.0669) NLenders -0.0748 -0.0806 -0.0810 -0.0769 -0.0812 -0.0786 (0.0496) (0.0494) (0.0491) (0.0528) (0.0512) (0.0530) LoanType (Term=1; Other=0) 0.166* 0.148 0.119 0.182* 0.191** 0.192** (0.0944) (0.0946) (0.0949) (0.0952) (0.0952) (0.0960)
Constant 0.260 0.245 0.259 0.267 0.238 0.232
(0.187) (0.185) (0.184) (0.190) (0.188) (0.194)
Year Dummy Yes Yes Yes Yes Yes Yes
Country of Syndication Dummy Yes Yes Yes Yes Yes Yes
Industry Dummy Yes Yes Yes Yes Yes Yes
Observations 270 270 270 270 269 269
R-squared 0.541 0.546 0.557 0.530 0.534 0.534
*** p<0.01, ** p<0.05, * p<0.1. Standard errors are presented in parentheses. All regression variables have been standardized, this means they all have a mean of zero and a standard deviation of one. Regarding our control variables we do not observe any irregularities, all the control variables that are significant have an effect that is in line with what one might expect.
5
DISCUSSION AND CONCLUSION
5.1 DISCUSSION AND CONCLUSION
5.1.1. Effect of financial instrument risk disclosure on subsequent loan pricing.
In our literature review we argued that financial instrument risks disclosures contain meaningful and relevant information for investors. Information in IFRS 7 disclosures is often regarded as unfavourable information to investors, telling them what could go wrong and should therefore be expected to lead to a higher perceived riskiness. This argumentation is supported by what we observe in our findings, across all our test and additional tests we observe a significant positive effect of IFRS 7 on loan price.
An important finding for this hypothesis, and the remaining two, is that using a Pre-Post analysis we find that in the post adoption period of IFRS 7 firms pay a significantly higher loan price. This support the notion that our findings are not mere coincidence but really are due to IFRS 7 and not due to some unobserved confounding effect.
However, our additional test show that it seems that not all risk information has the same value to investors. Our additional test show that that the disclosure of systematic risk is mostly affecting the loan price. This is in line with the general assumption that investors with a diversified portfolio are only exposed to systematic risks and should only be interested in information that helps them assess this risk. The evidence is however weak and would require additional testing to confirm whether it is really the systematic risks that is affecting loan price.
5.1.2. The effect of the board of directors on the decision-making process of lenders
Researchers found that managers have the tendency to behave opportunistically by delaying, or in some cases indefinitely, the disclosure of unfavourable information (Kothari, Shu and Wysocki, 2009). Since IFRS 7 disclosures contain unfavourable information we argued that one might expect that managers also with this type of disclosure behaves opportunistically. It is generally accepted that independent directors can limit the extent in which management can behave opportunistically. In our literature review we therefore argued that a larger concentration of independent directors should lead to a better and more timely disclosure of financial instrument risks, which lenders will reward in the form of lower loan prices.
The results of our test show that the interaction between the IFRS 7 disclosure and the percentage of independent directors have a significant negative effect on the subsequent loan price. This support our claim that an independent board limits the degree of opportunistic
