Fair value accounting and corporate debt structure

ADIAC-00338; No of Pages 12 Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Advances in Accounting, incorporating Advances in International Accounting journal homepage: www.elsevier.com/locate/adiac

Fair value accounting and corporate debt structure☆ Haiping Wang a,⁎, Jing Zhang b a b

School of Administrative Studies, York University, 4700 Keele Street, Toronto, Ontario M3J1P3, Canada School of Business, University of Alabama in Huntsville, 301 Sparkman Drive, Huntsville, AL, United States

a r t i c l e

i n f o

Article history: Received 19 October 2016 Received in revised form 23 February 2017 Accepted 23 February 2017 Available online xxxx Keywords: Fair value accounting Debt structure Convertible debt Maturity of debt Information asymmetry Agency cost

a b s t r a c t In this study, we examine the impact of fair value accounting on corporate debt structures, i.e., debt conversion privilege and maturity term. We argue that fair value accounting affects agency conflicts between debtholders and shareholders via its impact on financial reporting quality. Consequently, it should affect corporate decisions on the debt structure. Our empirical results show that ceteris paribus, more use of fair value measures in financial statements are associated with a greater demand for convertible debt and debt with short maturity, and the results are mainly driven by Level 2 and Level 3 fair value measures. These findings suggest that it is the lack of reliability of fair value measures that gives rise to more demand for debt structure tools that mitigate debtholdershareholder agency conflicts. In addition, we find that the negative association between the use of Level 3 fair value measures and the debt conversion privilege or debt maturity term is more pronounced for high-performance firms, suggesting that high-performance firms benefit more by issuing convertible debt or shortening debt maturity. This study provides novel insights regarding the impact of fair value accounting on corporate debt structure. It also provides regulatory implications, calling for better measurement guidance on fair value inputs. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction In this study, we investigate the role that fair value accounting plays in the design of corporate debt contracts. The finance and economics literature has long recognized that the design of debt contracts can be used as a tool to reduce shareholder-debtholder agency conflicts. Several prior studies show that shortening the debt maturity term or including provisions, such as conversion privilege, helps mitigate the agency cost of debt (Barnea, Haugen, & Senbet, 1980; Myers, 1977; Bodie & Taggart, 1978). Agency conflicts between shareholders and debtholders are directly affected by the quality of financial reporting, as more transparent financial reporting lessens debtholders' information disadvantage and facilitates efficient monitoring. Consequently, it is expected that the quality of financial reporting plays a role in corporate decisions regarding debt contract design. Our study specifically focuses on two important debt contract terms: debt conversion privilege and debt maturity, and examine whether the use of fair value accounting in financial statements has an impact on firms' choice on these two terms. Companies with lower (higher) financial reporting quality usually face greater (less) shareholder-debtholder agency conflicts, resulting in higher

☆ We thank seminar participants at the American Accounting Association Annual Meeting 2016, New York. Haiping Wang acknowledges financial support from the 20152016 Liberal Arts & Professional Studies Minor Research Grant. ⁎ Corresponding author. E-mail addresses: [email protected] (H. Wang), [email protected] (J. Zhang).

(lower) demand to use certain debt contract tools as a means of reducing such conflicts. Hence, the way that fair value accounting affects the design of debt contracts depends on how fair value measures affect the quality of financial reporting. Lately, the two major standard setters (i.e., Financial Accounting Standards Board, FASB, and International Accounting Standards Board, IASB) have been making joint efforts toward a more fair value-oriented reporting regime, which may impact the decision-making of various stakeholders, including shareholders, debtholders and corporate managers. Fair value accounting is a double-edged sword. Proponents of fair value accounting claim that it improves the relevance and timeliness of accounting information compared to historical cost accounting, and therefore, improves financial reporting quality. Opponents of fair value accounting express concern over its conceptual caveats and lack of reliability.1 For example, some fair value measures are subject to estimation errors and/or managerial manipulation, as they are based on either the market value of similar items (i.e., Level 2 fair value as defined in SFAS 157) or management's best estimates (i.e., Level 3 fair value as defined in SFAS 157). These estimated fair values are thus less reliable and may lead to lower financial reporting quality. The pros and cons of fair value accounting on financial reporting quality make it an open question as to whether using fair value accounting in financial statements exacerbates or alleviates agency conflicts 1 Penman (2007) provides a theoretical explanation why the fair value concept may be inferior to historical cost under certain circumstances.

http://dx.doi.org/10.1016/j.adiac.2017.02.002 0882-6110/© 2017 Elsevier Ltd. All rights reserved.

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002

2

H. Wang, J. Zhang Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx

between shareholders and debtholders. To answer this question, in this study, we empirically examine the impact of fair value accounting on a firm's decision regarding two debt contract terms: conversion privilege and debt maturity term, which have been documented in the prior literature as two frequently used tools to address shareholder-debtholder agency conflicts. We examine a sample of newly issued public debts between 2008 and 2013 and find a significantly positive association between the proportion of fair value, especially Level 2 and Level 3 fair value measures, and the likelihood of using the conversion feature in debt contracts. This finding indicates that the lack of reliability of fair value measures exacerbates agency conflicts between debtholders and shareholders, leading to more use of the conversion feature in the debt contracts. However, we do not find that Level 1 fair value measures have significant influence on the likelihood of issuing convertible debt, which confirms that the main issue of fair value measure is its reliability. In addition, we find that the debt maturity term is negatively associated with the use of Level 2 and Level 3 fair value measures. This finding supports the argument that the use of Level 2 and Level 3 fair value measures increases agency conflicts between shareholders and debtholders, leading to a higher demand for debts with a short maturity. Again, we do not find the same effect for Level 1 fair value measures, implying that it is the unreliable fair value measures (i.e., Level 2 and Level 3 fair values) that lower financial reporting quality. Furthermore, we conjecture that only high-performance firms are willing to issue convertible debt or short-term debt to overcome agency conflicts, because they are more likely to force debt conversion and less likely to incur debt rollover risk. Indeed, our empirical tests show consistent evidence that the negative association between Level 3 fair value measures and debt conversion privilege or debt maturity term is more pronounced in highperformance firms. We focus on the structure of public debt rather than private debt because unlike private lenders (e.g., banks), which possess significant inside information through private channels, public debtholders mainly rely on public accounting information for their decision-making. Therefore, the impact of fair value application on corporate debt decisions should be most substantial for public debts. In addition, our sample includes companies with fair value disclosures from all industries, which extends the narrow scope of prior fair value studies that focus mainly on financial institutions. In this sense, our evidence regarding the impact of fair value accounting on debt structure is more generalizable and provides implications to a broader audience. Our study makes important contributions to the accounting and finance literature and regulators. First, to the best of our knowledge, our study is one of the very first to document the impact of fair value accounting on corporate decisions related to public debt contract design. Prior fair value literature has mainly focused on the equity market and value relevance of fair value accounting (e.g., Barth, 1994; Petroni & Wahlen, 1995; Barth, Beaver, & Landsman, 1996, 2001; Eccher, Ramesh, & Thiagarajan, 1996; Nelson, 1996; Khurana & Kim, 2003; Song, Thomas, & Yi, 2010; Lee & Park, 2013), while neglecting the role that fair value accounting plays in debt contracting.2 Two recent exceptions are Demerjian, Donovan, and Larson (2016) and Aytekin and Karolyi (2015). Those studies investigate the impact of fair value accounting on debt convents of private loans and syndicated loans respectively. Our study extends this line of literature by providing empirical evidence on the association between fair value accounting and public debt contract design, highlighting the impact of fair value on debt contracting efficiency. Second, our findings document a new consequence of fair value application. That is, Level 2 and Level 3 fair value increases agency cost of 2 In fact, Kothari, Ramanna, and Skinner (2010) criticize this narrow interpretation of accounting's role as mere valuation. According to Holthausen and Leftwich (1983), an important objective of accounting is to facilitate firms' contractual arrangements, including executive compensation agreements and debt contracts.

debt and consequently affects debt contracting efficiency. This evidence also provides policy implications to standard setters such as FASB and IASB, as it suggests that some fair value measures (i.e., Level 2 and Level 3 fair values) suffer from low reliability. Therefore, there is need for more detailed measurement guidance from the standard setters that helps to improve the reliability of certain fair value measures. The rest of the paper is organized as follows. Section 2 presents an overview of fair value accounting and reviews related prior literature. Section 3 describes the theoretical framework and develops the hypotheses. Section 4 describes our research design. Section 5 reports the summary statistics and empirical results. Section 6 conducts additional analyses, and Section 7 concludes. 2. Institutional background and related research 2.1. Background of fair value accounting Despite different wording, the definitions of the term “fair value” are basically equivalent in the FASB and IASB pronouncements.3 The concept of fair value can be interpreted as the exit market price that would result, under close-to-ideal market conditions, in a transaction between knowledgeable, independent and economically rational parties in a complete information set (Hitz, 2007). Following the enactment of SFAS No. 157 (FASB, 2006), firms must disclose the three-tier measurement basis for assets and liabilities reported at fair value. Such a disclosure was not available prior to SFAS No. 157. Specifically, assets and liabilities defined as Level 1 are measured and reported at observable quoted prices in active markets. When an active market is absent, fair value is based on observable valuation inputs that reflect a) quoted prices for similar items in active markets, b) quoted prices for identical or similar items in inactive markets, c) inputs other than quoted prices that are observable, or d) correlated prices. Such a measurement basis is designated as Level 2. When neither of the above two types of inputs are available, fair value relies on models that reflect management's assumptions about economic, market, and firm-specific conditions, which is defined as Level 3 inputs, or “mark-to-model” accounting (FASB, 2006). 2.2. The impact of fair value accounting on the equity market Empirical evidence regarding fair value accounting mainly focuses on the value relevance of accounting numbers (e.g., Barth, 1994; Petroni & Wahlen, 1995; Barth et al., 1996, 2001; Eccher et al., 1996; Nelson, 1996; Khurana & Kim, 2003; Song et al., 2010). Current evidence in this line of research is mixed. In particular, Barth (1994), Petroni and Wahlen (1995) and Eccher et al. (1996) consistently find that the fair values of investment securities are value relevant after controlling for the fair values of other financial instruments. In addition, Barth et al. (1996) show that the fair value estimates of loans and long-term debt are incrementally value relevant beyond the related book values. However, Eccher et al. (1996) and Nelson (1996) find that the value relevance concerning the fair value of loans is weaker, and the fair values of deposits and off-balance sheet items are not value relevant. In a setting of financial institutions, Song et al. (2010) investigate the value relevance of the three-level fair value inputs and find that Level 3 fair values are less value relevant than Level 1 or Level 2 fair values. However, using the closed-end fund setting, Lawrence, Siriviriyakul, and Sloan (2016) find that Level 3 fair values are of similar value relevance to Level 1 and Level 2 fair values. From a slightly different 3 The recent new Standard, Statement of Financial Accounting Standard (SFAS) 157, Fair Value Measurements, defines fair value as “the price that would be received to sell an asset or paid to transfer a liability in an orderly transaction between market participants at the measurement date” (FASB SFAS No. 157 Fair Value Measurements 2006). In a recent convergence project, IASB developed an International Financial Reporting Standard (IFRS) on fair value measurement on the basis on SFAS 157.

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002

H. Wang, J. Zhang Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx

perspective, DeFond, Hung, Li, and Li (2014) investigate the impact of IFRS (International Financial Reporting Standards) on crash risk for both financial and non-financial firms. The authors argue that IFRS affects non-financial firms via increased transparency, while impacting financial firms through three channels: increased transparency, fair value accounting, as well as managers' appetite for risky investment. Another line of research study the impact of fair value on the information environment of analysts, or sophisticated investors on the equity market. Magnan, Menini, and Parbonetti (2015) find that fair value measurement of the financial institutions relates to more dispersed forecasts. In particular, Level 2 fair values are associated with enhanced forecast accuracy, while Level 3 relates to increased forecast dispersion. On the contrary, using a sample that include all industries, Barron, Chung, and Yong (2016) find that Level 3 measurements are able to reduce uncertainty in analysts' information environment. 2.3. The impact of fair value accounting on the debt market A number of studies investigate the impact of fair value accounting on the debt market. Demerjian et al. (2016) and Aytekin and Karolyi (2015) investigate the impact of fair value on corporate debts. Both studies focus on private loans as their research setting and find that lenders tend to exclude the effect of fair value from debt covenants. In addition, Cantrell, McInnis, and Yust (2014) examine the ability of reported loan fair values to predict credit losses and find that net historical loan costs are a better predictor of credit losses than reported loan fair values. Blankespoor, Linsmeier, Petroni, and Shakespeare (2013) provide empirical evidence that leverage measured at fair value is more associated with credit risk, suggesting that the fair value accounting reporting system makes accounting information more informative of firms' credit risk. In a similar vein, Hodder, Hopkins, and Wahlen (2006) test the risk relevance of fair value accounting. They find that the volatility of full fair value income is more strongly associated with certain firm risk, especially interest rate risk. Using financial institutions as their sample, Magnan, Shi, and Wang (2017) find that more extensive use of fair value in the financial statements is associated with higher cost of debt. Our study extends this line of literature by examining how fair value accounting affects the design of public debt contracts in an all-industry setting. 2.4. Accounting policies and corporate decision-making Another related line of research documents that accounting standards play a role in corporate decision-making. Amir, Guan, and Oswald (2010) find that pension accounting affects corporate decisions regarding pension asset allocation. Goncharov and Triest (2011) show that the adoption of fair value accounting affects firms' decisions on dividend payouts. The study by Bharath, Sunder, and Sunder (2008) is similar to our research, in that it investigates the association between accounting quality and debt contract design. However, they use accruals quality as a proxy for accounting quality and find that firms with lower accounting quality are more likely to issue private debt and use a more stringent contract design. 3. Theoretical framework and hypothesis development

3

and across firms (Hitz, 2007). In addition, fair value provides timely updates of a firm's financial position. Such timely updates promptly deliver information regarding the changes of the firm's risk profile, allowing investors to exercise early corrective actions on their investments before further deterioration of their investment value occurs (Plantin, Sapra, & Shin, 2008). On the other hand, fair value is also subject to a number of issues that may lower the quality of financial reporting. First, the fair value concept may not benefit all types of assets and liabilities. For example, in the case of assets that are held for a longer period, their current fair values may not be predictive of the value at maturity, therefore, fair value accounting may not be a good candidate as a measurement basis under such circumstances. Second, fair values may incorporate price bubbles into the financial statements when market prices are inefficient (Penman, 2007). This may result in a good-looking financial statement in disguise, leaving future losses to show up only at a later time when the bubble bursts. Furthermore, when the market prices of certain assets do not exist, fair value is either based on the market prices of similar items (Level 2 fair values), or managerial estimation (Level 3 fair values). Thus, fair value estimates could be subject to estimation errors and/or managerial manipulation, resulting in more information asymmetry between debtholders and the firm. Overall, it is an open question as to how fair value accounting affects the quality of financial reporting. If fair value accounting improves debtholders' information environment and facilitates efficient monitoring, it is expected to mitigate agency conflicts between shareholders and debtholders. In contrast, if fair value accounting exacerbates the information environment of debtholders, then shareholder-debtholder agency conflicts will increase in the use of fair value measures.

3.2. Fair value accounting and convertible debt Convertible debt is a hybrid product of straight debt and equity. Creditors can later convert the debt they hold into shares of common stock in the issuing company at an agreed-upon price. Due to debtholders' information disadvantage, when a firm has risky debt outstanding, managers acting on behalf of the shareholders are likely to substitute high-risk projects for low-risk ones in order to maximize the value of the equity claim, i.e., the “risk-shifting” problem (Lewis, Rogalski, & Seward, 1999). Green (1984) argues that debt issuers can use the conversion feature to alleviate the risk-shifting problem, because the conversion feature changes the payoff structure of the existing shareholders, making them less likely to overinvest in risky projects. If the use of fair value accounting improves debtholders' information environment by delivering more timely, relevant and unbiased information regarding the firm's financial positions, the shareholderdebtholder agency conflicts should be alleviated, resulting in less demand for convertible debts. On the contrary, if debtholders have concerns regarding the reliabilities of fair value inputs, and perceive that information asymmetry is higher in firms whose financial statements contain more fair values, then the demand to use the conversion feature should be greater in such firms. Because of the opposing arguments, we make a non-directional prediction about the association between the use of fair value measures and the issuance of convertible debt.

3.1. Fair value accounting, financial reporting quality and agency cost of debt

H1a. There is no association between the extent of a firm's assets and liabilities measured and reported in fair values and the issuance of convertible debt.

Fair value accounting has both pros and cons. On the one hand, fair value has a number of theoretical merits that should improve the information environment of debtholders. First, fair value is a market-based concept that is not affected by factors specific to a particular firm (Penman, 2007). The fair value construct has the power of incorporating market consensus expectations about future cash flows in an efficient and virtually unbiased manner that is consistent from period to period

Following the enactment of SFAS No. 157, firms must disclose in the financial statements the measurement basis of fair values, that is, Level 1 (valuation based upon market prices), Level 2 (valuation based upon market inputs), and Level 3 (valuation based upon model estimates). All three levels of fair value measurements should have the merit of providing timely and relevant information to debtholders. However, the degree of reliability of the three-level measurements differs, with

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002

4

H. Wang, J. Zhang Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx

Level 1 fair value measures being the most reliable, and Level 3 fair value measures being the least. In other words, compared to Level 1, Level 2 and Level 3 fair value measures are more likely to reduce the overall financial reporting quality, leading to greater demand for convertible debt. Therefore, we propose our Hypothesis 1b as follows: H1b. There is a positive association between the extent of a firm's assets and liabilities measured and reported in Level 2 and Level 3 fair values and the issuance of convertible debt. 3.3. Fair value and debt maturity Myers (1977) argues that, with the existence of information asymmetry, shareholders are likely to forgo low-risk, but positive NPV projects (the “underinvestment” problem), because the payoffs of those projects benefit debtholders more. One effective approach for controlling this problem is to shorten the maturity of the debt4 (Barnea et al., 1980; Myers, 1977; Flannery, 1986). If fair value measures improve debtholders' information environment, which in turn lowers agency conflicts, firms do not need to shorten the debt maturity. Otherwise, if fair value measures have a negative effect on financial reporting quality, we should observe that firms with more use of fair values in financial statements issue debts with shorter maturity. Since the association between fair value accounting and financial reporting quality is unclear, we do not make a directional prediction regarding the impact of fair value on a firm's choice of maturity term. H2a. There is no association between the extent of a firm's assets and liabilities measured and reported in fair values and the maturity of debt. In addition, as Level 2 and especially Level 3 fair values suffer greater degree of reliability problems, which lowers the overall quality of financial reporting, we expect that firms with more use of Level 2 and Level 3 fair value measures have a greater tendency to issue short-term debt. Therefore, we propose the following hypothesis: H2b. There is a negative association between the extent of a firm's assets and liabilities measured and reported in Level 2 and Level 3 fair values and the maturity of debt. 4. Research design We apply a logistic model to examine our first set of hypotheses regarding the association between debt conversion privilege and the use of fair value measurements. The dependent variable, CONVERTIBLE, is an indicator variable that equals 1 if a firm includes conversion feature in the debt contract and 0 otherwise. The independent variable of interest in the model, FAIR (LEVEL1, LEVEL2, LEVEL3) is the proportion of fair value assets and liabilities (their breakdowns) to total assets, which captures the extent to which firms' balance sheets are fair value-oriented (Magnan et al., 2015). If the use of fair value accounting improves (worsens) debtholders' information environment and results in lower (higher) agency conflicts between shareholders and debtholders, we expect to find a negative (positive) coefficient on FAIR in the following regression model: CONVERTIBLE ¼ a þ β0  FAIR ðLEVEL1; LEVEL2; LEVEL3Þ þ β1  BSIZE þ β2  COUPON þ β3  MAT þ β4  SIZE þ β5  LEV þ β6  ROA þ β7  BM þ β8  TANGIBILITY þ β9  TAX þ β10  BIG þ β11  BANK þ β12  YEARDUMMIES þ e ð1Þ

4 For example, Easterwood and Kadapakkam (1994), and Barclay and Smith (1995) both find evidence that firms with more pronounced agency costs of debt tend to issue more short-term debt.

In Model (1), we control for the security-level and firm-level factors that are documented to affect the decision on debt structure. At the security level, we include the amount of debt issue, BSIZE; coupon rate, COUPON, and debt maturity, MAT as control variables because various debt contract features could be simultaneously determined by the firms. At firm level, we first control firm size,5 SIZE, measured as the logarithm of total assets, leverage,6 LEV, measured as total liabilities over total assets, and firms' effective tax rate,7 TAX, measured by the ratio of income tax expense to pretax income. In addition, to control for firms' current performance and growth potential, we include return on assets, ROA, ratio of property, plant and equipment over total assets, TANGIBILITY, and book to market ratio, BM, in the regression model. Last, the quality of auditors plays an important role in the financial reporting quality. We therefore include in the model an indicator variable, BIG, that equals to one if the firms are audited by one of the Big-4 auditors, and zero otherwise. Next, we apply an OLS model to test our second set of hypotheses that fair value accounting influences the maturity term of the debt contract. We predict that fair value usage is negatively (positively) associated with debt maturity, if the use of fair value measurements increase (mitigate) agency conflicts between shareholders and debtholders. Our second regression model is as follows: MAT ¼ a þ β0  FAIR ðLEVEL1; LEVEL2; LEVEL3Þ þ β1  BSIZE þ β2  COUPON þ β3  CONVERTIBLE þ β4  SIZE þ β5  LEV þ β6  ROA þ β7  BM þ β8  TANGIBILITY þ β9  TAX þ β10  BIG þ β11  BANK þ β12  YEARDUMMIES þ e ð2Þ In Model (2), the dependent variable is maturity of debt, MAT, measured as the length of debt maturity terms in years. The main independent variables of interest are the same as in Model (1). The securitylevel control variables are debt size, BSIZE, coupon rate, COUPON, and the inclusion of conversion privilege, CONVERTIBLE, as defined in Model (1). We control for the same firm-level factors that affect a firm's decision on debt structure as previously defined. In all regression models, we include an indicator variable, BANK, based on two-digit SIC (60–69) codes, because financial industry is more likely to be affected by fair value application as it has more items on the financial statements that are measured at fair value (e.g. financial instruments). In addition, we include the year-fixed effect to control for the macroeconomic conditions such as financial crisis period. Moreover, we handle potential cross-sectional correlation within firms in each regression model by relying on robust standard errors that clustered by firm. Detailed variables definitions are provided in Appendix A. 5. Descriptive statistics and empirical results 5.1. Sample firms We rely on a sample of new public debt issued from 2008 to 2013 for our empirical tests. The new debt issue setting is well accepted as preferable for testing the impact of accounting transparency on corporate debt structures (see Yu, 2005; Easton, Monahan, & Florin, 2009; Nikolaev, 2010). In addition, we focus on public debts rather than private borrowings (e.g., bank loans) because unlike private lenders that have better access to firms' inside information (Diamond, 1991), public debtholders rely more on information from financial statements. To 5 According to Barclay and Smith (1995) and Smith and Warner (1979), larger firms, due to more rigorous scrutiny, are motivated to provide higher quality financial statements and thus face less agency conflicts between debtholders and shareholders. 6 Diamond (1991) points out that firms with higher leverage are more likely to be financially distressed or have higher liquidity risk and thus such firms have a greater incentive to involve in risk-shifting activities. 7 Hennessy and Tserlukevich (2008) argue that debt conversion has tax costs. Therefore, the benefits of issuing convertible debt must at least compensate for the tax costs when designing debt structure.

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002

H. Wang, J. Zhang Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx

5

data on fair values. We further remove 480 observations that have missing data on the control variables, and end up with a sample of 2493 unique public corporate debt issues.

compile our sample, we extract all newly issued US public debts during 2008 to 2013 from the Mergent FISD database. We choose the period after 2007 because the breakdowns of fair values (Levels 1, 2 and 3) became available only after SFAS No. 157 was effective in November 2007. Our initial sample includes 3577 public corporate debt issues with available information on our testable debt features. Next we merge this issue-specific dataset with firm-specific financial data from COMPUSTAT. A total of 604 observations are eliminated due to missing

5.2. Descriptive statistics Table 1 Panel A presents the summary statistics of our test variables. On average, 10.3% of our sample firms issued convertible debts, and the

Table 1 Summary statistics. This table provides descriptive statistics for the sample firms. Panel A shows summary statistics for the whole sample. Panel B provides comparisons between firms that issued convertible debts and firms that issued regular debts. Panel provides comparisons between firms that issued long-term debts and firms that issued short-term debts. ***, **, and * indicate significance at the 1%, 5% and 10% levels. Panel A: Summary statistics Variables

N

Mean

Median

SD

P25

P75

CONVERT MAT FAIR LEVEL1 LEVEL2 LEVEL3 HQ BSIZE COUPON SIZE LEV ROA BM TAN TAX BIG BANK

2493 2493 2493 2493 2493 2493 2493 2493 2493 2493 2493 2493 2493 2493 2493 2493 2493

0.103 10.597 0.152 0.057 0.097 0.015 0.663 13.054 4.721 9.398 0.630 0.041 0.563 0.306 0.217 0.957 0.146

0.000 9.992 0.039 0.006 0.013 0.000 1.000 13.122 4.500 9.345 0.615 0.047 0.474 0.201 0.281 1.000 0.000

0.304 8.222 0.240 0.142 0.226 0.078 0.473 0.747 2.355 1.746 0.208 0.088 0.445 0.273 0.441 0.204 0.353

0.000 5.052 0.010 0.000 0.002 0.000 0.000 12.612 3.000 8.221 0.490 0.011 0.288 0.074 0.153 1.000 0.000

0.000 10.055 0.199 0.050 0.073 0.004 1.000 13.528 6.000 10.561 0.760 0.085 0.760 0.542 0.360 1.000 0.000

Panel B: Descriptive statistics – convertible debts vs. non-convertible debts Convertible debts

Non-convertible debts

Variables

N

Mean

Median

N

Mean

Median

Difference (convert-non-convert)

FAIR LEVEL1 LEVEL2 LEVEL3 HQ BSIZE COUPON MAT SIZE LEV ROA BM TAN TAX BIG BANK

256 256 256 256 256 256 256 256 256 256 256 256 256 256 256 256

0.295 0.130 0.111 0.055 0.543 12.279 3.497 8.047 7.313 0.567 -0.027 0.602 0.204 0.061 0.871 0.188

0.176 0.034 0.012 0.000 1.000 12.301 3.250 5.086 7.258 0.536 0.006 0.543 0.098 0.072 1.000 0.000

2237 2237 2237 2237 2237 2237 2237 2237 2237 2237 2237 2237 2237 2237 2237 2237

0.136 0.049 0.096 0.010 0.677 13.143 4.861 10.889 9.637 0.637 0.049 0.558 0.317 0.235 0.966 0.141

0.035 0.005 0.013 0.000 1.000 13.122 4.625 10.014 9.609 0.615 0.050 0.468 0.220 0.290 1.000 0.000

0.159*** 0.081*** 0.015 0.045*** −0.134*** −0.863*** −1.364*** −2.841*** −2.324*** −0.07*** −0.076*** 0.044 −0.113*** −0.173*** −0.095*** 0.046**

Panel C: Descriptive statistics – long-term debts vs. short-term debts Long-term debts

Short-term debts

Variables

N

Mean

Median

N

Mean

Median

Difference (long-term-short-term)

FAIR LEVEL1 LEVEL2 LEVEL3 HQ CONVERT BSIZE COUPON SIZE LEV ROA BM TAN TAX BIG BANK

1275 1275 1275 1275 1275 1275 1275 1275 1275 1275 1275 1275 1275 1275 1275 1275

0.1416 0.0530 0.0964 0.0091 0.7043 0.0353 13.1537 4.9261 9.7026 0.6217 0.0556 0.5411 0.3195 0.2544 0.9757 0.1412

0.0356 0.0063 0.0126 0.0000 1.0000 0.0000 13.1224 4.7500 9.6327 0.6070 0.0540 0.4680 0.2252 0.2930 1.0000 0.0000

1218 1218 1218 1218 1218 1218 1218 1218 1218 1218 1218 1218 1218 1218 1218 1218

0.164 0.062 0.098 0.020 0.620 0.173 12.950 4.506 9.080 0.638 0.026 0.585 0.291 0.178 0.937 0.151

0.046 0.005 0.013 0.000 1.000 0.000 12.899 3.750 9.037 0.622 0.038 0.482 0.178 0.266 1.000 0.000

−0.022*** −0.009** −0.002 −0.011*** 0.084*** −0.138*** 0.204*** 0.420*** 0.623*** −0.017 0.029*** −0.044** 0.028*** 0.077*** 0.039*** −0.010

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002

6

H. Wang, J. Zhang Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx

average length of debt maturity is 10.597 years. Fair value assets and liabilities account for an average of 15.2% of total assets. At the debt level, the average debt size (natural logarithm of the debt offering amount) is 13.054, and the average coupon rate is 4.721%. At the firm level, the sample firms have an average size (natural logarithm of total assets) of 9.398, a leverage ratio of 0.630, a return-on-asset ratio of 0.041, a book-to-market ratio of 0.563, a tangible asset ratio of 0.306, and an effective tax rate of 0.217. Last, about 95.7% of our sample firms use Big Four accounting firms as their auditors, and 14.6% of them are in the financial industry. Overall, our results indicate that there is ample variation across the continuous regression variables. Panel B of Table 1 compares the characteristics of convertible debt issuers with regular straight debt issuers. We find that the convertible debt issuers have a significantly higher proportion of fair value assets and liabilities, suggesting that firms using more fair values are more likely to use the conversion option in their debt contracts. In addition, convertible debts are relatively smaller and have a lower coupon rate and shorter maturity term. Firms that issue convertible debts also appear to be smaller in size, leverage and profitability, have fewer tangible assets, a lower effective tax rate and are less likely to hire a Big Four accounting firm. Panel C compares the characteristics of firms that issued long- and short-term debts. The proportion of fair value assets and liabilities is significantly higher for short-term debt issuers, which provides preliminary evidence that the use of fair value measures leads to the issuance of more short-term debts. Moreover, consistent with prior literature (Stohs & Mauer, 1996; Barclay & Smith, 1995), we find that larger firms, more profitable firms and firms with more tangible assets are more likely to issue long-term debts. Table 2 reports the industry and calendar year distribution of our sample firms. The number of observations in our sample is fairly evenly distributed across years and industries. The three categories of industries with the most observations are: basic industries, the financial industry and the utility industry. This industry distribution shows that the financial industry is not the only main user of fair value measures, indicating the wide application of fair value accounting across industries and supporting the need for studies on the impact of fair value in all industries. Table 3 reports the Pearson correlation statistics among the test and control variables used in the main analyses. The proportion of aggregate fair values and the three-tier breakdowns are positively and significantly correlated with the likelihood of issuing convertible debts, while the proportion of Level 3 fair values is negatively and significantly correlated with the length of the maturity term. This finding indicates that firms with more fair value measures, especially the less reliable ones, are more likely to issue convertible debts or short-term debts. Some control variables are significantly correlated with each other, however, with a low correlation coefficient. Overall, there are low cross-correlations between our main test variables and control variables.

5.3. Empirical results Table 4 reports the results for the first set of hypotheses. In Column (1), we find a positive relation between the use of fair value measures and the likelihood of including the conversion feature in the debt contract, significant at the 1% level, suggesting that firms using more fair value measures in their financial statements are more likely to issue convertible debts. This finding supports the argument that fair value measures exacerbate agency conflicts between debtholders and shareholders, and as a response, firms are more likely to use the conversion feature in their debt contract in order to reduce the agency cost. In Column (2), we use the three-level fair value breakdowns as the test variables and examine how they influence the decision on the use of conversion feature. Results show that the positive relation between the use of fair value measures and the issuance of convertible debts are significant only for Level 2 and Level 3 fair value measures. This result indicates that increased agency conflict is mainly driven by the less reliable fair value measures that are subject to more managerial opportunism and/or estimation errors. In addition, we find that firm size, return on assets and tangibility are negatively and significantly associated with the likelihood of issuing convertible debts, while firm leverage is positively and significantly associated with the likelihood of issuing convertible debts. This finding suggests that firms with a larger size, higher profitability or lower risks have less severe agency problems of debt, and therefore are less likely to issue convertible debts. In addition, we find that convertible debts tend to be smaller in debt size, lower in coupon rate and longer in the maturity term. Table 5 reports the multivariate regression results for our second set of hypotheses. Column (1) shows that the association between aggregate fair value percentage and length of the debt maturity term is not significant. Furthermore, we use the three-level fair value breakdowns as test variables and re-run the regression. Results from Column (2) clearly show that the length of the maturity term are negative and significantly associated with proportions of Level 2 and Level 3 fair values, but insignificantly associated with the proportion of Level 1 fair values. Once again, this finding supports the argument that the use of the less reliable fair value measurements increases debtholder-shareholder agency conflicts, resulting in a higher likelihood of issuing shorterterm debt. Overall, the findings from Tables 4 and 5 suggest that both the debt conversion option and maturity term are used by firms to mitigate agency conflicts between debtholders and shareholders arising from the use of less reliable fair value measurements. 6. Additional tests 6.1. The effect of firm performance So far, we have shown that firms with more use of the less reliable fair value measurements are more likely to issue convertible debt and

Table 2 Industry distribution. This table shows the industry and year distribution of the sample firms. Industry classification is based on the twelve industry group affiliations identified in Campbell (1996). Two-digit SIC codes

Industry

2008

2009

2010

2011

2012

2013

Total

10,12,14,24,26,28,33,34,35,38,39 25,30,36,37,50,55,57 15,16,17,32,52 60–69 1,9,20,21,54 27,58,70,78,79 13,29 72,73,75,80,82,87,89 22,23,31,51,53,56,59 40,41,42,44,45,47 46,48,49 99 Total

Basic industries Consumer durables Construction Finance/real estate Food/tobacco Leisure Petroleum Services Textiles/trade Transportation Utilities Not classable

101 36 7 52 12 13 47 28 10 12 66 0 384

130 39 24 68 26 8 35 35 20 9 53 0 447

134 46 11 58 30 11 34 33 13 8 74 0 452

127 54 19 92 56 11 67 51 18 16 78 7 596

136 43 26 93 38 12 50 56 28 14 61 0 557

11 13 3 1 14 2 0 6 1 0 6 0 57

639 231 90 364 176 57 233 209 90 59 338 7 2493

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002

1 1 −0.048⁎⁎⁎ 1 0.032⁎ −0.022

BANK BIG

1 0.086⁎⁎⁎ 0.077⁎⁎⁎ −0.381⁎⁎⁎

TAX TAN

Variables

(1) CONVERT

FAIR

2.363⁎⁎⁎

1 0.009 −0.016 −0.019 0.355⁎⁎⁎

BM

1 0.145⁎⁎⁎ 0.254⁎⁎⁎ 0.178⁎⁎⁎ −0.079⁎⁎⁎ 0.086⁎⁎⁎ 0.257⁎⁎⁎ 0.201⁎⁎⁎

−0.006 −0.020 0.047⁎⁎ 0.212⁎⁎⁎

1 −0.200⁎⁎⁎ −0.073⁎⁎⁎ 0.132⁎⁎⁎ 0.114⁎⁎⁎ −0.088⁎⁎⁎ 1 −0.218⁎⁎⁎ −0.095⁎⁎⁎

ROA LEV SIZE

−0.018 0.049⁎⁎⁎ 0.445⁎⁎⁎

0.007 −0.149⁎⁎⁎ −0.120⁎⁎⁎ −0.132⁎⁎⁎ 0.028

−0.015 0.399⁎⁎⁎

−0.073⁎⁎⁎ −0.207⁎⁎⁎ −0.033⁎ 0.024 0.091⁎⁎⁎

−0.002 0.323⁎⁎⁎ 0.142⁎⁎⁎ −0.048⁎⁎⁎ 0.179⁎⁎⁎ −0.281⁎⁎⁎

1 −0.077⁎⁎⁎ −0.144⁎⁎⁎ 0.076⁎⁎⁎ −0.045⁎⁎ −0.110⁎⁎⁎

−0.003 −0.018 −0.115⁎⁎⁎ −0.021 0.069⁎⁎⁎ 0.078⁎⁎⁎ −0.075⁎⁎⁎ 0.069⁎⁎⁎ −0.351⁎⁎⁎ −0.045⁎⁎

CONVERT FAIR LEVEL1 LEVEL2 LEVEL3 HQ BSIZE COUPON MAT SIZE LEV ROA BM TAN TAX BIG BANK

1 0.231⁎⁎⁎ 0.187⁎⁎⁎ 0.041⁎⁎ 0.188⁎⁎⁎ −0.086⁎⁎⁎ −0.370⁎⁎⁎ −0.196⁎⁎⁎ −0.119⁎⁎⁎ −0.434⁎⁎⁎ −0.121⁎⁎⁎ −0.274⁎⁎⁎

1 0.693⁎⁎⁎ 0.612⁎⁎⁎ 0.388⁎⁎⁎

1 0.190⁎⁎⁎ 0.017 0.011 −0.022 −0.131⁎⁎⁎ −0.015 −0.032⁎ 0.052⁎⁎⁎ −0.032⁎

1 0.073⁎⁎⁎ 0.017 0.157⁎⁎⁎ −0.093⁎⁎⁎

0.011 −0.091⁎⁎⁎ 0.079⁎⁎⁎ −0.111⁎⁎⁎ −0.331⁎ −0.166⁎⁎⁎ 0.244⁎⁎⁎

1 0.069⁎⁎⁎ −0.135⁎⁎⁎ 0.005 0.175⁎⁎⁎ 0.059⁎⁎⁎ 0.102⁎⁎⁎ 0.056⁎⁎⁎ −0.057⁎⁎⁎ 0.039⁎⁎ 0.078⁎⁎⁎ 0.035⁎

0.014 0.073⁎⁎⁎ 0.211⁎⁎⁎ −0.045⁎⁎

1 −0.222⁎⁎⁎ 0.073⁎⁎⁎ 0.659⁎⁎⁎ −0.016 0.254⁎⁎⁎ −0.047⁎⁎

0.001

1 0.167⁎⁎⁎ −0.048⁎⁎⁎ 0.127⁎⁎⁎ −0.039⁎⁎ 0.069⁎⁎⁎ 0.058⁎⁎⁎ 0.060⁎⁎⁎ −0.049⁎⁎⁎ 1 0.141⁎⁎⁎ −0.322⁎⁎⁎ 0.216⁎⁎⁎ −0.304⁎⁎⁎ 0.212⁎⁎⁎ 0.279⁎⁎⁎ −0.050⁎⁎⁎ −0.198⁎⁎⁎

MAT COUPON BSIZE HQ LEVEL3 LEVEL2

1.335 (0.154) 1.862⁎⁎

LEVEL2 LEVEL3

COUPON

−0.442 (0.144) −1.345⁎⁎⁎ (0.000) 0.058⁎⁎

ROA

(0.012) −1.864⁎⁎⁎ (0.000) 1.777⁎⁎⁎ (0.010) −7.369⁎⁎⁎

BM

(0.000) 2.309⁎⁎⁎

SIZE LEV

TAN TAX BIG

LEVEL1

(2) CONVERT

(0.000) LEVEL1

MAT

FAIR

7

Table 4 Fair value proportion and convertible debt issuance. This table presents the estimation results of hypotheses H1a and H1b. All variables are defined in the Appendix A. The standard errors are clustered at the firm level. ⁎⁎⁎, ⁎⁎, and ⁎ indicate significance at the 1%, 5% and 10% levels. P-values are provided in brackets.

BSIZE

CONVERT

Table 3 Pearson correlation. This table presents a correlation matrix of aggregated fair value percentage, three levels of fair value breakdowns, convertible feature, maturity term and other control variables used in the tests, which are described in the Appendix A. ⁎⁎⁎, ⁎⁎, and ⁎ indicate significance at the 1%, 5% and 10% levels.

H. Wang, J. Zhang Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx

BANK Constant Year fixed effect Pseudo R2 Observations

(0.000) −0.380 (0.504) −0.456⁎⁎ (0.026) −1.026⁎ (0.074) 0.353 (0.354) 21.145⁎⁎⁎ (0.000) Yes 0.664 2493

(0.013) 3.813⁎⁎⁎ (0.000) −0.400 (0.184) −1.372⁎⁎⁎ (0.000) 0.060⁎⁎ (0.011) −1.887⁎⁎⁎ (0.000) 1.912⁎⁎⁎ (0.006) −7.497⁎⁎⁎ (0.000) 2.323⁎⁎⁎ (0.000) −0.459 (0.423) −0.433⁎⁎ (0.033) −0.967⁎ (0.089) 0.244 (0.531) 20.619⁎⁎⁎ (0.000) Yes 0.667 2493

debt with short maturity. Nonetheless, the preference for convertible debt or short-term debt may vary with firm performance. First, after the issuance of convertible debt, only the high-performance firms (i.e. firms with optimistic future prospects) are able to force conversion and reduce their debt burden (Stein, 1992). Therefore, compared to low performance firms, high performance firms have greater incentives to use debt conversion feature to mitigate agency cost of debt. As a result, the association between the use of fair value measurement and convertible debt issuance should be more pronounced in high performance firms. Second, due to the existence of debt rollover risk, shortterm debt may be a less efficient mechanism to mitigate shareholderdebtholder agency conflicts for firms facing greater debt rollover risk. Flannery (1986) argues that high-performance firms are more willing to issue short-term debt so as to mitigate shareholder-debtholder agency conflicts, because those firms are less concerned about debt rollover risk and are likely to re-issue new debts in the future with a better term. In contrast, low-performance firms are more reluctant to issue shortterm debt because they face a greater risk of not being able to rollover the debt in the future. Therefore, we expect the effect of fair value measures on the design of the debt maturity term to be more pronounced in high-performance firms. We use the firms' future performance, measured as the two-year-ahead annual stock return, as a proxy for firm performance.8 We include in the regression an indicator variable, HQ, which is equal to one if the firm's two-year-ahead annual stock return is larger than zero, and zero otherwise. Table 6 presents the empirical results. In Panel A, we find that the interaction between Level 3 fair 8 In an untabulated test, we tried to use one-year-ahead and three-year-ahead annual stock return as a proxy for firm performance. Our results remain qualitatively the same.

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002

8

H. Wang, J. Zhang Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx

Table 5 Fair value proportion and debt maturity. This table presents the estimation results of hypotheses H2a and H2b. All variables are described in the Appendix A. The standard errors are clustered at the firm level. ⁎⁎⁎, ⁎⁎, and ⁎ indicate significance at the 1%, 5% and 10% levels. P-values are provided in brackets. Variables

(1) MAT

FAIR

0.137 (0.889)

LEVEL1

(2) MAT

Panel A: Fair values and convertible debts

COUPON

−2.108⁎⁎⁎ (0.003) 1.902⁎⁎⁎

2.243 (0.179) −1.635⁎⁎⁎ (0.001) −4.727⁎⁎⁎ (0.000) −2.118⁎⁎⁎ (0.002) 1.947⁎⁎⁎

CONVERT

(0.000) 5.158⁎⁎⁎

(0.000) 5.414⁎⁎⁎

ROA

(0.000) 2.610⁎⁎⁎ (0.000) −7.626⁎⁎⁎ (0.000) 11.133⁎⁎⁎

(0.000) 2.695⁎⁎⁎ (0.000) −7.854⁎⁎⁎ (0.000) 11.133⁎⁎⁎

BM

(0.000) −3.908⁎⁎⁎ (0.000) 0.763 (0.353) 0.608 (0.293) 3.167⁎⁎⁎

(0.000) −3.909⁎⁎⁎ (0.000) 0.709 (0.408) 0.597 (0.307) 2.992⁎⁎⁎

(0.000) −1.073⁎⁎ (0.014) 8.233 (0.128) Yes 0.188 2493

(0.000) −0.447 (0.291) 7.776 (0.132) Yes 0.192 2493

LEVEL2 LEVEL3 BSIZE

SIZE LEV

TAN TAX BIG BANK Constant Year fixed effect R2 Observations

Table 6 Fair value proportion, debt contract term and firm performance. This table presents the estimation results of the test on the impact of firm performance on the association between fair value proportion and two debt contract terms: conversion feature and maturity. All variables are described in the Appendix A. The standard errors are clustered at the firm level. ⁎⁎⁎, ⁎⁎, and ⁎ indicate significance at the 1%, 5% and 10% levels. P-values are provided in brackets.

Variables

(1) CONVERT

FAIR

2.079⁎⁎⁎ (0.010) 0.554 (0.564)

FAIR ∗ HQ LEVEL1

LEVEL3

6.2. Bank and nonbank samples The financial industry is one of the most influenced industries by the application of fair value accounting, in that it deals with a large number of financial assets and liabilities which are required to be booked to market. DeFond et al. (2014) argue that IFRS may have different impact on financial firms versus non-financial firms. In particular, they believe the requirement on fair value under IFRS may only affect financial firms on their crash risks, while the impact of IFRS on non-financial firms is restricted to increased transparency in general and may not be associated with fair value. In order to test whether there are systematic differences between financial firms and non-financial firms on the association between fair value and firms' debt structure, we separate the financial firms from nonfinancial firms and re-analyze Models (1) and (2). The descriptive statistics reported in Table 7 Panel A shows that the mean value of the fair value proportion of the financial firms is 0.392, which is approximately four times greater than that of nonfinancial firms,

(0.004) 0.559 (0.725) 0.749 (0.642) 6.815⁎

LEVEL1 ∗ HQ LEVEL2 ∗ HQ LEVEL3 ∗ HQ

COUPON

−0.658⁎⁎ (0.028) −0.421 (0.151) −1.369⁎⁎⁎

MAT

(0.000) 0.058⁎⁎⁎

BSIZE

SIZE LEV ROA BM TAN

values and firm performance is positively and significantly associated with the likelihood of issuing convertible debt. Similarly, results from Panel B show that the interaction between Level 3 fair values and firm performance is negatively and significantly associated with the length of the debt maturity term. This finding is in support of our conjecture that the effect of fair value measurements on the use of debt conversion feature or debt maturity term is more pronounced in high performance firms. Thus, our finding provides empirical support that high performance firms are more likely to use debt conversion feature or shorter maturity as a tool to mitigate debtholders' concern about wealth transfer.

0.997 (0.445) 1.335 (0.359) 3.041⁎⁎⁎

LEVEL2

HQ

TAX BIG BANK Constant Year fixed effect Pseudo R2 Observations

(2) CONVERT

(0.053) −0.769⁎⁎⁎ (0.009) −0.374 (0.195) −1.402⁎⁎⁎

(0.008) −1.847⁎⁎⁎ (0.000) 1.944⁎⁎⁎ (0.005) −7.044⁎⁎⁎

(0.000) 0.061⁎⁎⁎ (0.005) −1.873⁎⁎⁎ (0.000) 2.030⁎⁎⁎ (0.005) −7.089⁎⁎⁎

(0.000) 2.436⁎⁎⁎ (0.000) −0.357 (0.534) −0.397⁎

(0.000) 2.451⁎⁎⁎ (0.000) −0.425 (0.461) −0.407⁎

(0.062) −0.993⁎ (0.079) 0.281 (0.465) 20.910⁎⁎⁎ (0.000) Yes 0.667 2493

(0.053) −0.886 (0.113) 0.131 (0.743) 20.443⁎⁎⁎ (0.000) Yes 0.672 2493

Panel B: Fair values and debt maturity Variables

(1) MAT

FAIR

0.336 (0.650) −0.274 (0.744)

FAIR ∗ HQ LEVEL1

2.761 (0.401) −1.422⁎ (0.061) −2.483⁎⁎⁎

LEVEL2 LEVEL3 LEVEL1 ∗ HQ LEVEL2 ∗ HQ LEVEL3 ∗ HQ HQ BSIZE

(2) MAT

0.386 (0.479) −2.096⁎⁎⁎ (0.003)

(0.003) −0.661 (0.765) −0.093 (0.929) −9.084⁎⁎ (0.015) 0.447 (0.489) −2.129⁎⁎⁎ (0.002)

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002

H. Wang, J. Zhang Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx Table 6 (continued) Panel B: Fair values and debt maturity Variables

(1) MAT

(2) MAT

COUPON

LEV

1.913⁎⁎⁎ (0.000) 5.194⁎⁎⁎ (0.000) 2.603⁎⁎⁎ (0.000) −7.672⁎⁎⁎

1.967⁎⁎⁎ (0.000) 5.533⁎⁎⁎ (0.000) 2.701⁎⁎⁎ (0.000) −7.870⁎⁎⁎

ROA

(0.000) 11.057⁎⁎⁎

(0.000) 11.094⁎⁎⁎

(0.000) −3.936⁎⁎⁎ (0.000) 0.796 (0.342) 0.596 (0.290) 3.165⁎⁎⁎ (0.000) −1.060⁎⁎ (0.018) 7.949 (0.155) Yes 0.188 2493

(0.000) −3.927⁎⁎⁎ (0.000) 0.725 (0.402) 0.608 (0.284) 2.950⁎⁎⁎ (0.000) −0.438 (0.303) 7.564 (0.172) Yes 0.194 2493

CONVERT SIZE

BM TAN TAX BIG BANK Constant Year fixed effect R2 Observations

indicating that the extent to which fair value accounting is used is much greater in the financial industry. Panels B and C report consistent results for financial firms and non-financial firms. In Columns (1) and (2), we find a significant and positive association between the proportion of fair value measures and the decision to issue convertible debts, and this association is mainly driven by the proportion of Level 2 and Level 3 fair value measures. Columns (3) and (4) show that debt maturity is significantly and negatively associated with Level 3 fair value measures, but is insignificantly associated between Level 1 and Level 2 fair value measures. These results suggest that agency conflicts between debtholders and shareholders are mainly driven by the least reliable fair value measures –Level 3 fair values. Overall, our findings remain similar for firms in both the financial and nonfinancial industries, suggesting that the effect of fair value accounting on shareholderdebtholder agency conflict does not only prevail in financial firms.

Table 7 Financial and non-financial firms. This table reports robustness test on our baseline regression. Panel A provides the descriptive statistics for firms in financial and nonfinancial industries. Panel B provides the results on the multivariate regression for firms in financial industries, while Panel C provides the results on the multivariate regression for firms in nonfinancial industries. All variables are described in the Appendix A. The standard errors are clustered at the firm level. ⁎⁎⁎, ⁎⁎, and ⁎ indicate significance at the 1%, 5% and 10% levels. P-values are provided in brackets. Panel A: Summary statistics Financial firms

Non-financial firms

Variables

N

Mean

Median

N

Mean

Median

CONVERT MAT FAIR LEVEL1 LEVEL2 LEVEL3 BSIZE COUPON SIZE LEV ROA BM TAN TAX BIG

364 364 364 364 364 364 364 364 364 364 364 364 364 364 364

0.132 9.596 0.392 0.089 0.341 0.067 12.978 4.809 10.235 0.740 0.021 0.951 0.058 0.188 0.934

0.000 8.037 0.380 0.022 0.196 0.007 12.899 4.675 9.821 0.825 0.011 0.917 0.009 0.290 1.000

2129 2129 2129 2129 2129 2129 2129 2129 2129 2129 2129 2129 2129 2129 2129

0.098 10.768 0.111 0.052 0.056 0.006 13.067 4.706 9.255 0.611 0.045 0.496 0.348 0.222 0.961

0.000 9.995 0.031 0.005 0.010 0.000 13.122 4.500 9.271 0.604 0.055 0.433 0.275 0.281 1.000

Panel B: Financial firms Variables FAIR

(1) CONVERT

LEVEL3

COUPON MAT

LEV

A possible concern for the above findings is that our sample period covers the recent financial crisis during which market-based measurement may be more problematic than under normal economic conditions. For example, during a financial crisis, the liquidity of capital markets is quite poor; as a result, market prices may deviate greatly from the fundamental values of assets and liabilities. Hence, the fair value measures during the financial crisis may be distorted. Moreover, managerial decisions during a financial crisis may not accurately represent regular managerial behavior. Hence, we exclude the financial crisis period (2007–2009) from our sample and re-estimate our regression models as a robustness check. Table 8 reports consistent results. Therefore, our findings do not appear to be sensitive to the inclusion of the financial crisis period.

ROA BM TAN TAX BIG Constant Year fixed effect Pseudo R2 or R2 Observations

(4) MAT

2.048⁎⁎⁎ (0.007)

0.049 (0.935) −1.457⁎⁎⁎ (0.002) −0.061 (0.265)

(0.000) 0.086 (0.887) −1.724⁎⁎⁎ (0.003) −0.066 (0.176)

CONVERT SIZE

(3) MAT

−1.076⁎⁎⁎ (0.005) 1.692⁎⁎⁎

4.813⁎⁎ (0.021) 0.389 (0.525) −3.455⁎⁎⁎ (0.006) −1.103⁎⁎⁎ (0.009) 1.897⁎⁎⁎

(0.000)

(0.000)

1.100 (0.686) 1.378⁎⁎⁎ (0.001) −6.877⁎⁎⁎

2.121 (0.429) 1.445⁎⁎⁎ (0.001) −7.697⁎⁎⁎

(0.000) 3.575 (0.657) −3.094⁎⁎ (0.013) −4.298⁎⁎ (0.021) −0.011 (0.992) 5.064⁎⁎ (0.037) 0.062 (0.991) Yes 0.210 364

(0.000) 4.066 (0.641) −3.057⁎⁎ (0.012) −5.319⁎⁎⁎ (0.003) −0.067 (0.951) 4.953⁎ (0.058) −0.689 (0.912) Yes 0.218 364

−1.426 (0.527) 2.354 (0.320) 4.340⁎⁎⁎

LEVEL2

BSIZE

(2) CONVERT

2.641⁎⁎⁎ (0.005)

LEVEL1

6.3. Post-financial-crisis period

−2.433⁎⁎⁎ (0.006) −1.576 (0.382) −10.840⁎

−2.692⁎⁎⁎

(0.053) 4.257⁎⁎ (0.027) 1.530 (0.278) −1.666⁎⁎⁎

(0.002) −0.427 (0.800) −8.569⁎ (0.073) 4.562⁎⁎ (0.023) 1.561 (0.354) −1.538⁎⁎⁎

(0.003) −3.103⁎⁎ (0.018) 24.976⁎⁎⁎ (0.004) Yes 0.74 364

(0.001) −3.211⁎⁎⁎ (0.007) 26.946⁎⁎⁎ (0.003) Yes 0.755 364

Panel C: Non-financial firms

6.4. Additional controls for firm risk and complexity Variables

One caveat of our research design is that the percentage of assets and liabilities measured at fair values might be associated with the intrinsic risk or complexity of the company. If firm risk and complexity have an effect on debt contract design, then our findings might be subject to omitted correlated variables bias. To alleviate this issue, we include five additional variables: standard deviation of returns (SD_RET), the

9

FAIR LEVEL1 LEVEL2

(1) CONVERT 2.746⁎⁎⁎ (0.000)

(2) CONVERT

2.079⁎ (0.055) 2.652⁎⁎⁎ (0.000)

(3) MAT

(4) MAT

0.205 (0.892) 2.330 (0.162) −2.165 (0.142) (continued on next page)

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002

10

H. Wang, J. Zhang Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx

Table 7 (continued) Panel C: Non-financial firms (1) CONVERT

(2) CONVERT

COUPON

−0.398 (0.206) −1.367⁎⁎⁎

MAT

(0.000) 0.077⁎⁎⁎

7.309⁎ (0.089) −0.384 (0.225) −1.375⁎⁎⁎ (0.000) 0.077⁎⁎⁎

(0.001)

(0.001)

Variables LEVEL3 BSIZE

CONVERT

(3) MAT

(4) MAT

−2.380⁎⁎⁎ (0.009) 1.970⁎⁎⁎

−17.030⁎⁎⁎ (0.001) −2.386⁎⁎⁎ (0.008) 2.007⁎⁎⁎

(0.000)

(0.000)

6.048⁎⁎⁎

6.182⁎⁎⁎ (0.000) 2.921⁎⁎⁎ (0.000) −7.672⁎⁎⁎ (0.000) 11.452⁎⁎⁎ (0.003) −4.075⁎⁎⁎ (0.000) 0.728 (0.387) 0.795 (0.135) 2.333⁎⁎⁎ (0.001) 9.402 (0.221) Yes 0.196 2129

LEV

−1.901⁎⁎⁎ (0.000) 2.552⁎⁎⁎

−1.908⁎⁎⁎ (0.000) 2.440⁎⁎⁎

ROA

(0.002) −7.720⁎⁎⁎

(0.003) −7.802⁎⁎⁎ (0.000) 2.314⁎⁎⁎ (0.001) −0.784 (0.213) −0.370 (0.137) −0.131 (0.873) 20.222⁎⁎⁎ (0.000) Yes 0.672 2129

(0.000) 11.662⁎⁎⁎ (0.002) −4.010⁎⁎⁎ (0.000) 0.808 (0.340) 0.735 (0.165) 2.388⁎⁎⁎ (0.002) 9.364 (0.231) Yes 0.193 2129

BM TAN TAX BIG Constant Year fixed effect Pseudo R2 or R2 Observations

(0.000) 2.328⁎⁎⁎ (0.001) −0.790 (0.213) −0.342 (0.172) −0.164 (0.840) 20.240⁎⁎⁎ (0.000) Yes 0.671 2139

Variables FAIR

(1) CONVERT

LEVEL1

LEVEL3

7. Conclusion In this paper, we empirically examine the impact of fair value accounting on two debt contract terms: debt conversion privilege and debt maturity term. We find a positive association between Level 2 and Level 3 fair value measures and the likelihood of issuing convertible debts, and a negative association between Level 2 and

0.281 (0.855) 2.016 (0.157) 3.226⁎⁎ (0.029) −0.688⁎⁎ (0.034) −1.673⁎⁎⁎ (0.000) 0.078⁎⁎⁎ (0.003)

−2.426⁎⁎⁎ (0.000) 2.310⁎⁎

−2.473⁎⁎⁎ (0.000) 2.500⁎⁎

(0.026) −10.648⁎⁎⁎ (0.000) 2.712⁎⁎ (0.016) −0.354 (0.603) −0.592⁎

(0.014) −10.612⁎⁎⁎ (0.000) 2.703⁎⁎ (0.018) −0.368 (0.589) −0.541⁎

Constant

(0.056) −1.515⁎⁎ (0.017) 0.755 (0.147) 29.804⁎⁎⁎

(0.094) −1.416⁎⁎ (0.022) 0.677 (0.203) 29.489⁎⁎⁎

Year fixed effect Pseudo R2 or R2 Observations

(0.000) Yes 0.753 1662

(0.000) Yes 0.752 1662

COUPON MAT CONVERT SIZE LEV ROA BM TAN TAX

BANK

(3) MAT 0.429 (0.714)

−0.718⁎⁎ (0.032) −1.633⁎⁎⁎ (0.000) 0.074⁎⁎⁎ (0.003)

BSIZE

BIG

standard deviation of trading volumes (SD_VOL), the use of derivatives (DEV), firm beta (BETA), and bid-ask spread (SPREAD) to control for firm risk (Hughes, Liu, & Liu, 2009; Callahan, Rodney, & Spencer, 2012). Additionally, we use the number of geographic segments (GSEGMENT) and the number of business segments (BSEGMENT) as proxies for firm complexity. Table 9 presents the regression results that are qualitatively the same: The positive association between the fair value percentage and the likelihood of issuing convertible debt remains significant, and the effect is mainly driven by the use of Level 2 and Level 3 fair value measures. In addition, we also find a negative and significant association between the use of Level 3 fair value measures and the length of the debt maturity term. Overall, our findings are robust to the inclusion of additional controls for firm risk and complexity. To further ease the concern that our findings are driven by certain firm characteristics that are associated with both the usage of fair value measures and debt structure, we develop four industryadjusted measures of fair value proportion: AB_FAIR, AB_LEVEL1, AB_LEVEL2, AB_LEVEL3, calculated as the proportion of fair value (LEVEL1, LEVEL2, LEVEL3) measures minus the industry average, because the structure of firm assets could be related to industry membership. We reestimate the hypothesis 1 and 2 using the adjusted measures of fair value proportion, and present the findings in Panel B of Table 9. We continue to find a positive and significant association between AB_LEVEL2 or AB_LEVEL3 and debt conversion feature, and a negative and significant association between AB_LEVEL2 or AB_LEVEL3 and debt maturity.

(2) CONVERT

1.942⁎⁎ (0.019)

LEVEL2 (0.000) 2.895⁎⁎⁎ (0.000) −7.426⁎⁎⁎

SIZE

Table 8 Post-financial crisis period. This table reports robustness test on our baseline regressions. All variables are described in the Appendix A. The standard errors are clustered at the firm level. ⁎⁎⁎, ⁎⁎, and ⁎ indicate significance at the 1%, 5% and 10% levels. P-values are provided in brackets. (4) MAT

3.324⁎⁎⁎ (0.007) −2.095⁎⁎⁎

−2.495⁎⁎ (0.016) 2.370⁎⁎⁎

(0.005) −5.933⁎⁎⁎ (0.000) −2.510⁎⁎ (0.011) 2.445⁎⁎⁎

(0.000)

(0.000)

6.441⁎⁎⁎ (0.000) 2.961⁎⁎⁎

6.833⁎⁎⁎ (0.000) 3.072⁎⁎⁎

(0.000) −7.423⁎⁎⁎ (0.000) 15.037⁎⁎⁎ (0.000) −4.631⁎⁎⁎ (0.000) 1.007 (0.402) 1.226⁎⁎⁎

(0.000) −7.914⁎⁎⁎ (0.000) 15.194⁎⁎⁎ (0.000) −4.591⁎⁎⁎ (0.000) 0.919 (0.468) 1.223⁎⁎⁎

(0.000) 3.807⁎⁎⁎ (0.000) −1.083⁎

(0.000) 3.520⁎⁎⁎ (0.000) −0.213 (0.693) 5.107 (0.485) Yes 0.24 1662

(0.066) 5.762 (0.462) Yes 0.232 1662

Level 3 fair value measures and the length of debt maturity. In addition, we also find that the impact of Level 3 fair value measures on debt conversion privilege or debt maturity term is stronger in high performance firms. These findings suggest that the main issues regarding fair value accounting is the lack of reliability, and as a result, firms that use more Level 2 and Level 3 fair value measures in their financial statements are more likely to issue convertible debt and debt with short maturity. Our results are robust to the exclusion of financial/nonfinancial firms, the exclusion of years of the financial crisis, and the inclusion of additional controls for firm risk and complexity. Our findings add to the fair value accounting literature which investigates the usefulness and potential consequences of implementing fair value measures. Our evidence suggests that the less reliable fair value measures (i.e., Level 2 and Level 3 fair values) worsen the quality of financial reporting and increase agency cost of debt. In other words, fair value accounting affects debt contracting efficiency. In addition, our paper is one of the very first to examine the impact of fair value accounting on public debt structures. Prior studies related to fair value accounting have mainly focused on the equity market. Very few has examined the role played by fair value accounting in the debt market. The evidence from our paper shows that fair value measurements have a significant impact on debt contract design. Future research can extend our study by exploring other potential debt market consequences of applying fair value accounting.

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002

H. Wang, J. Zhang Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx Table 9 Additional controls for firm risk and complexity. This table reports robustness test on our baseline regression. Panel A provides the results on the multivariate regression with inclusion of additional controls for firm risk and complexity. Panel B presents the results on the effect of abnormal fair value measures on the use of conversion feature in the debt contract and debt maturity. All variables are described in the Appendix A. The standard errors are clustered at the firm level. ⁎⁎⁎, ⁎⁎, and ⁎ indicate significance at the 1%, 5% and 10% levels. P-values are provided in brackets. Panel A Variables FAIR

(2) CONVERT

3.250⁎⁎⁎ (0.000)

LEVEL1 LEVEL2 LEVEL3

(0.002) 5.982⁎⁎⁎

COUPON MAT

(3) MAT 0.390 (0.739)

1.080 (0.344) 2.381⁎⁎⁎

BSIZE

−0.522 (0.114) −1.550⁎⁎⁎ (0.000) 0.090⁎⁎⁎

(0.000) −0.463 (0.165) −1.640⁎⁎⁎ (0.000) 0.098⁎⁎⁎

(0.000)

(0.000)

−2.173⁎⁎ (0.011) 2.390⁎⁎⁎ (0.000)

(4) MAT

3.019⁎ (0.058) −1.346⁎⁎⁎ (0.006) −10.311⁎⁎⁎ (0.005) −2.178⁎⁎⁎ (0.008) 2.482⁎⁎⁎ (0.000)

LEV

−1.872⁎⁎⁎ (0.000) 2.145⁎⁎⁎

−1.921⁎⁎⁎ (0.000) 2.530⁎⁎⁎

7.311⁎⁎⁎ (0.000) 2.534⁎⁎⁎ (0.000) −7.423⁎⁎⁎

ROA

(0.009) −8.502⁎⁎⁎ (0.000) 1.861⁎⁎⁎ (0.008) −0.329 (0.690) −0.403 (0.177) −1.438⁎⁎

(0.003) −8.610⁎⁎⁎ (0.000) 1.887⁎⁎⁎ (0.009) −0.568 (0.505) −0.353 (0.252) −1.306⁎

(0.000) 9.362⁎⁎ (0.023) −3.546⁎⁎⁎ (0.000) 0.721 (0.335) 0.550 (0.425) 4.399⁎⁎⁎

(0.000) 9.890⁎⁎ (0.020) −3.467⁎⁎⁎ (0.000) 0.652 (0.392) 0.556 (0.426) 4.119⁎⁎⁎

(0.048) 0.759 (0.119) −0.250 (0.644) 0.190⁎⁎⁎

(0.054) 0.532 (0.328) 0.073 (0.902) 0.197⁎⁎⁎

SDBETA

(0.000) −0.067 (0.343) 7.011 (0.520) 5.089⁎⁎

(0.000) −0.077 (0.279) 9.614 (0.378) 5.202⁎

BETA

(0.045) 0.578⁎

(0.000) −0.471 (0.486) −0.598 (0.330) −0.007 (0.911) −0.009 (0.916) −64.801 (0.189) −0.172 (0.927) −0.978⁎⁎⁎

(0.075) 20.772⁎⁎⁎ (0.000) Yes 2148 0.704

(0.052) 0.604⁎ (0.071) 19.527⁎⁎⁎ (0.000) Yes 2148 0.713

(0.000) −1.259⁎⁎ (0.039) −0.359 (0.458) 0.004 (0.947) 0.010 (0.913) −62.821 (0.168) −0.190 (0.924) −0.934⁎⁎⁎ (0.000) 5.155 (0.422) Yes 2148 0.223

(0.000) 4.688 (0.460) Yes 2148 0.232

(1) CONVERT

(2) CONVERT

(1) MAT

(2) MAT

CONVERT SIZE

BM TAN TAX BIG BANK DEV GSEGMENT BSEGMENT SPREAD

Constant Year fixed effect Observations Pseudo R2 or R2

7.904⁎⁎⁎ (0.000) 2.633⁎⁎⁎ (0.000) −7.812⁎⁎⁎

Panel B Variables AB_FAIR

2.909⁎⁎⁎ (0.000)

AB_LEVEL1

AB_LEVEL3

COUPON MAT

−0.192 (0.838) 0.909 (0.371) 2.465⁎⁎⁎ (0.007) 3.813⁎⁎⁎

AB_LEVEL2

BSIZE

Table 9 (continued) Panel B (1) CONVERT

(2) CONVERT

(0.006)

(0.008)

LEV

−1.905⁎⁎⁎ (0.000) 1.929⁎⁎⁎

−1.919⁎⁎⁎ (0.000) 1.915⁎⁎⁎

ROA

(0.005) −7.601⁎⁎⁎ (0.000) 2.407⁎⁎⁎ (0.000) −0.561 (0.321) −0.465⁎⁎

(0.005) −7.601⁎⁎⁎ (0.000) 2.359⁎⁎⁎ (0.000) −0.558 (0.326) −0.448⁎⁎

Constant

(0.021) −1.005⁎ (0.090) 0.894⁎⁎ (0.012) 21.608⁎⁎⁎

(0.029) −1.163⁎⁎ (0.039) 1.029⁎⁎⁎ (0.007) 22.216⁎⁎⁎

Year fixed effect Observations Pseudo R2 or R2

(0.000) Yes 2493 0.667

(0.000) Yes 2493 0.669

Variables CONVERT SIZE

(1) CONVERT

−0.410 (0.173) −1.383⁎⁎⁎ (0.000) 0.063⁎⁎⁎

(0.000) −0.430 (0.126) −1.406⁎⁎⁎ (0.000) 0.063⁎⁎⁎

−2.104⁎⁎⁎ (0.002) 1.903⁎⁎⁎ (0.000)

2.695⁎⁎⁎ (0.010) −1.529⁎⁎ (0.022) −4.261⁎⁎⁎ (0.000) −2.143⁎⁎⁎ (0.002) 1.945⁎⁎⁎ (0.000)

11

BM TAN TAX BIG BANK

(1) MAT

(2) MAT

5.211⁎⁎⁎ (0.000) 2.614⁎⁎⁎ (0.000) −7.628⁎⁎⁎

5.389⁎⁎⁎ (0.000) 2.693⁎⁎⁎ (0.000) −7.846⁎⁎⁎

(0.000) 11.108⁎⁎⁎ (0.000) −3.917⁎⁎⁎ (0.000) 0.728 (0.392) 0.609 (0.293) 3.169⁎⁎⁎ (0.000) −1.047⁎⁎ (0.022) 8.176 (0.124) Yes 2493 0.188

(0.000) 11.344⁎⁎⁎ (0.000) −3.942⁎⁎⁎ (0.000) 0.660 (0.442) 0.618 (0.307) 3.072⁎⁎⁎ (0.000) −1.182⁎⁎⁎ (0.007) 7.916 (0.137) Yes 2493 0.192

Appendix A. Variables definition

CONVERT MAT FAIR LEVEL1

Equals to one if the debt is a convertible debt; zero otherwise The maturity term of the debt in years The proportion of fair value assets and liabilities over total assets at year t The proportion of Level 1 fair value assets and liabilities over total assets at year t LEVEL2 The proportion of Level 2 fair value assets and liabilities over total assets at year t LEVEL3 The proportion of Level 3 fair value assets and liabilities over total assets at year t HQ Equals to 1 if the firm's two-year-ahead annual stock return is larger than zero; 0 otherwise. BSIZE Natural logarithm of the debt issuance amount COUPON The coupon rate of the debt multiplied by 100 SIZE Natural logarithm of total assets at year t LEV Total liabilities scaled by total assets at year t ROA Net income scaled by total assets at year t BM Book value of equity scaled by market value of equity at year t TAN Total PP&E assets scaled by total assets at year t TAX Income tax expense scaled by total pre-tax income at year t BIG Equals to 1 if the firm has one of the Big-4 accounting firms as its auditor at year t; 0 otherwise BANK Equal to one if a firm belongs to financial industry; 0 otherwise DEV Equal to one if the firm has involved in derivatives activities at year t; zero otherwise SD_VOL The standard deviation of stock trading volume at year t SD_RET The standard deviation of stock return at year t GSEGMENT The number of geographic segment at year t BSEGMENT The number of business segment at year t SPREAD The average daily bid-ask spread at year t BETA Firm beta based on the single factor market model using monthly return for firms with at least 30 observations in the previous 60 months AB_FAIR The proportion of fair value assets and liabilities over total assets minus the industry average proportion of fair value assets and liabilities over total assets (based on two-digit SIC) at year t AB_LEVEL1 The proportion of Level 1 fair value assets and liabilities over total assets minus the industry average proportion of Level 1 fair value assets and liabilities over total assets (based on two-digit SIC) at year t AB_LEVEL2 The proportion of Level 2 fair value assets and liabilities over total assets minus the industry average proportion of Level 2 fair value assets and liabilities over total assets (based on two-digit SIC) at year t AB_LEVEL3 The proportion of Level 3 fair value assets and liabilities over total assets minus the industry average proportion of Level 3 fair value assets and liabilities over total assets (based on two-digit SIC) at year t

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002

12

H. Wang, J. Zhang Advances in Accounting, incorporating Advances in International Accounting xxx (2017) xxx–xxx

References Amir, E., Guan, Y., & Oswald, D. (2010). The effect of pension accounting on corporate pension asset allocation. Review of Accounting Studies, 15, 345–366. Aytekin, E., & Karolyi, S. A. (2015). Relevance, reliability, and optimal debt contracting: Evidence from fair value adoption. Working paper. CMU & LBS. Barclay, M., & Smith, C. (1995). The maturity structure of corporate debt. The Journal of Finance, 50(2), 609–631. Barnea, A., Haugen, R., & Senbet, L. (1980). A rational for debt maturity structure and call provisions in the agency theoretic framework. The Journal of Finance, 35(5), 1223–1234. Barron, O., Chung, S. G., & Yong, K. O. (2016). The effect of statement of financial accounting standards no. 157 fair value measurements on analysts' information environment. Journal of Accounting and Public Policy, 35, 395–416. Barth, M. (1994). Fair value accounting: Evidence from investment securities and the market valuation of banks. The Accounting Review, 69(1), 1–25. Barth, M., Beaver, W., & Landsman, W. (1996). Value-relevance of banks' fair value disclosures under FAS no. 107. The Accounting Review, 71(4), 513–537. Barth, M., Beaver, W., & Landsman, W. (2001). The relevance of the value relevance literature for accounting standard setting: Another view. Journal of Accounting and Economics, 31(1–3), 77–104. Bharath, S., Sunder, J., & Sunder, S. (2008). Accounting quality and debt contracting. The Accounting Review, 96(6), 2143–2176. Blankespoor, E., Linsmeier, T., Petroni, K., & Shakespeare, C. (2013). Fair value accounting for financial instruments: Does it improve the association between bank leverage and credit risk? The Accounting Review, 88(4), 1143–1177. Bodie, Z., & Taggart, R. (1978). Future investment opportunities and the value of a call provision on a bond. The Journal of Finance, 33, 1187–1200. Callahan, C., Rodney, S., & Spencer, A. (2012). An examination of the cost of capital implications of FIN 46. The Accounting Review, 87(4), 1105–1134. Campbell, J. (1996). Understanding risk and return. Journal of Political Economy, 104, 298–345. Cantrell, W., McInnis, M., & Yust, G. (2014). Predicting credit losses: Loan fair values versus historical costs. The Accounting Review, 89(1), 147–176. DeFond, M. L., Hung, M., Li, S., & Li, Y. (2014). Does mandatory IFRS adoption affect crash risk? The Accounting Review, 90(1), 265–299. Demerjian, P. R., Donovan, J., & Larson, C. (2016). Fair value accounting and debt contracting: Evidence from adoption of SFAS 159. Journal of Accounting Research, 54(4), 1041–1076. Diamond, D. (1991). Debt maturity structure and liquidity risk. Quarterly Journal of Economics, 106, 709–738. Easterwood, J., & Kadapakkam, P. (1994). Agency conflicts, issue costs, and debt maturity. Quarterly Journal of Business and Economics, 69–80. Easton, D., Monahan, S., & Florin, V. (2009). Initial evidence on the role of accounting earnings in bond market. Journal of Accounting Research, 47(3), 721–766. Eccher, E., Ramesh, K., & Thiagarajan, S. (1996). Fair value disclosures by bank holding companies. Journal of Accounting and Economics, 22(1–3), 79–117. Financial Accounting Standards Board (2006). Statement of financial accounting standards no. 157—Fair value measurements. Norwalk, CT. Flannery, M. (1986). Asymmetric information and risky debt maturity choice. Journal of Finance, 41, 18–38. Goncharov, I., & Triest, S. (2011). Do fair value adjustments influence dividend policy? Accounting and Business Research, 41(1), 51–68.

Green, R. (1984). Investment incentives, debt and warrants. Journal of Financial Economics, 115–136. Hennessy, C., & Tserlukevich, Y. (2008). Taxation, agency conflicts, and the choice between callable and convertible debt. Journal of Economic Theory, 143(1), 374–404. Hitz, J. (2007). The decision usefulness of fair value accounting―A theoretical perspective. European Accounting Review, 16(2), 323–362. Hodder, L., Hopkins, P., & Wahlen, J. (2006). Risk-relevance of fair-value income measures for commercial banks. The Accounting Review, 81(2), 337–375. Holthausen, R., & Leftwich, R. (1983). The economic consequences of accounting choice: Implications of costly contracting and monitoring. Journal of Accounting and Economics, 5, 77–117. Hughes, J., Liu, J., & Liu, J. (2009). On the relations between expected returns and implied cost of capital. Review of Accounting Studies, 14, 246–259. Khurana, I., & Kim, M. (2003). Relative value relevance of historical cost vs. fair value: Evidence from bank holding companies. Journal of Accounting and Public Policy, 22(1), 19–42. Kothari, S. P., Ramanna, K., & Skinner, D. (2010). Implications for GAAP from an analysis of positive research in accounting. Journal of Accounting and Economics, 50(2–3), 246–286. Lawrence, A., Siriviriyakul, S., & Sloan, R. (2016). Who's the fairest of them all? Evidence from closed-end funds. The Accounting Review, 91(1), 207–227. Lee, C., & Park, M. S. (2013). Subjectivity in fair-value estimates, audit quality, and informativeness of other comprehensive income. Advances in Accounting, 29(2), 218–231. Lewis, C., Rogalski, R., & Seward, J. (1999). Is convertible debt a substitute for straight debt or for common equity? Financial Management, 5–27. Magnan, M., Menini, A., & Parbonetti, A. (2015). Fair value accounting: information or confusion for financial markets? Review of Accounting Studies, 20(1), 559–591. Magnan, M., Shi, Y., & Wang, H. (2017). Fair value accounting and the cost of debt. Working paper. Myers, S. (1977). Determinants of corporate borrowing. Journal of Financial Economics, 5, 147–176. Nelson, K. (1996). Fair value accounting for commercial banks: An empirical analysis of FAS no.107. The Accounting Review, 71(2), 161–182. Nikolaev, V. (2010). Debt covenants and accounting conservatism. Journal of Accounting Research, 48(1), 51–89. Penman, S. (2007). Financial reporting quality: Is fair value a plus or a minus? Accounting and Business Research, 37(Special Issue), 33–44. Petroni, K., & Wahlen, J. (1995). Fair values of equity and debt securities and share prices of property liability insurer. The Journal of Risk and Insurance, 62(4), 719–737. Plantin, G., Sapra, H., & Shin, H. (2008). Fair value accounting and financial stability. Chicago GSB research paper no. 08-15. Smith, C., & Warner, J. (1979). On financial contracting: An analysis of bond covenants. Journal of Financial Economics, 7(2), 117–161. Song, C., Thomas, W., & Yi, H. (2010). Value relevance of FAS no. 157 fair value hierarchy information and the impact of corporate governance mechanisms. The Accounting Review, 85(4), 1375–1410. Stein, J. C. (1992). Convertible bonds as backdoor equity financing. Journal of Financial Economics, 32(1), 3–21. Stohs, M., & Mauer, D. (1996). The determinants of corporate debt maturity structure. Journal of Business, 279–312. Yu, F. (2005). Accounting transparency and the term structure of credit spreads. Journal of Financial Economics, 75(1), 53–84.

Please cite this article as: Wang, H., & Zhang, J., Fair value accounting and corporate debt structure, Advances in Accounting, incorporating Advances in International Accounting (2017), http://dx.doi.org/10.1016/j.adiac.2017.02.002