Does mandatory IFRS adoption facilitate debt financing?

Transcription

1 Rev Account Stud (2015) 20: DOI /s z Does mandatory IFRS adoption facilitate debt financing? Annita Florou 1 Urska Kosi 2 Published online: 10 June 2015 Springer Science+Business Media New York 2015 Abstract We examine whether the mandated introduction of International Financial Reporting Standards (IFRS) is associated with the propensity to access the public rather than private debt market and the cost of debt. We use a global sample of public bonds and private loans and find that mandatory IFRS adopters are more likely, post-ifrs, to issue bonds than to borrow privately. We also find that mandatory IFRS adopters pay lower bond yield spreads, but not lower loan spreads, after the mandate. These findings are consistent with debt providers responding positively to financial reporting of higher quality and comparability, but only when there is a greater reliance on publicly available financial statements than private communication. Lastly, we document that the observed debt market benefits are concentrated in countries with larger differences between domestic GAAP and IFRS and are present even for EU countries that did not experience concurrent financial reporting enforcement or other institutional reforms. Overall, our study documents positive economic consequences around the mandated IFRS adoption for corporate debt financing and, in particular, for bond financing. Keywords Accounting regulation IFRS Accounting quality Public and private debt markets Cost of debt JEL Classifications G15 K22 M41 M48 & Annita Florou annita.florou@kcl.ac.uk 1 2 Department of Management, King s College London, Franklin-Wilkins Building, 150 Stamford Street, London SE1 8WA, UK Institute of Accounting and Auditing, International Accounting Group, Vienna University of Economics and Business, Welthandelsplatz 1, 1020 Vienna, Austria

2 1408 A. Florou, U. Kosi 1 Introduction Regulators and standards-setters have suggested that a uniform set of high quality accounting standards such as International Financial Reporting Standards (IFRS) can facilitate financing and reduce the cost of capital because investors will face lower levels of information asymmetry and estimation risk (IASB and FASB 2008). In line with these expectations, prior IFRS research focuses on equity markets and reports reduced cost of equity and increased institutional holdings for mandatory IFRS adopters after adoption (Daske et al. 2008; Li 2010; Florou and Pope 2012). We focus on debt markets and explore the debt financing consequences around mandatory IFRS adoption. In particular, we focus on new debt issuance and examine how the mandatory adoption of IFRS relates to the propensity of firms to issue bonds rather than obtain loans and to the cost of debt. Understanding the implications of mandatory IFRS financial reporting for debt markets is important for at least two primary reasons. First, firms access debt markets far more frequently than equity markets. For example, in 2000 and 2011, the average European country s corporate debt market was three times the size of its equity market; the balance between private sector debt and equity in other non-eu countries over the same period was comparable. 1 Given that debt is the most significant source of capital for many firms, the debt market consequences around mandatory IFRS adoption are likely to be economically significant. Second, creditors have different information needs than equity investors, so the findings of IFRS studies on equity markets may not necessarily be generalizable to debt markets. This raises the question of whether financial information reported under IFRS will meet creditor needs (Ball et al. 2014). The adoption of IFRS in many jurisdictions introduced significant changes in accounting rules and in the properties of financial statements. Consequently, IFRS may affect the financing decisions of debt providers, who rely on financial statement information in their decision-making. However, the a priori effect of IFRS on the qualitative attributes of financial information required by creditors is unclear. For example, in contrast to equity, the value of debt is more asymmetrically sensitive to decreases in firm value than to increases. Holders of debt therefore have a greater demand for negative information (Ball et al. 2008a, b). Moreover, lenders use the book value of borrowers assets to determine their debt-service capability. Unlike equity holders, debt holders have a higher priority in bankruptcy cases, and, due to security requirements, they collateralize a firm s assets. Debt holders arguably require current but reliable estimates of asset values (Watts 2003). On the one hand, some IFRS requirements may accelerate the recognition of bad news, such as recognition of impairments under IAS 36, or enhance the reliability of financial statement numbers, such as historical cost asset measurement under IAS 16. In this case, IFRS would facilitate debt financing. But on the other hand, some 1 Over the total amount of private credit and outstanding corporate bonds in the EU was 193 % (as a percentage of GDP), whereas the total value of all shares listed on European stock markets was 59 %. Similarly, the total amount of U.S. corporate debt (both loans and bonds) over the same period was 323 % as opposed to the total market capitalization of stocks which was 126 % (for further details see World Bank, Global Financial Development Database).

3 Does mandatory IFRS adoption facilitate debt financing? 1409 IFRS features may reduce accounting conservatism and cast doubt on the reliability of financial information, such as recognition and measurement of financial instruments under IAS 39 and IFRS 9. In this case, IFRS could hinder the decision-making of debt providers. In summary, the debt financing consequences around mandatory IFRS reporting is an empirical issue. We provide direct evidence of the debt financing consequences around mandatory IFRS adoption, differentiating between public and private debt market deals. Unlike equity finance, corporate debt usually features a choice of multiple contract terms (e.g., interest, maturity, collateral). If capital providers have alternatives for mitigating agency costs, prior literature suggests accounting quality may ultimately be less important in financing decisions (Biddle and Hilary 2006; Beatty et al. 2010). Moreover, private lenders can access nonpublic information, and they may be better able (compared to public bond holders) to ex post monitor borrowers and renegotiate debt contracts, suggesting that initial contract setting may be less critical for the former than the latter (Gordon and Winton 2003). The consequences for new debt issuance at the time of IFRS adoption may thus be felt primarily in the public bond market. Accordingly, we first examine whether the introduction of IFRS is associated with access to the public debt market, that is, the likelihood of raising debt from the bond rather than the loan market. Next, we examine whether the cost of debt decreases around mandatory IFRS reporting and whether any change differs between the public and private debt markets. We use a difference-in-differences research design on a large global sample of public bond issues and private loan agreements from a maximum of 35 countries over the period. We report a number of results that are new to the literature. First, we find that the likelihood a firm will access the bond market increases by 8.4 % for mandatory IFRS adopters after the mandate; the enhanced access to the public debt market increases public debt, as a proportion of total debt, by 9.7 percentage points. Second, we find that the cost of bonds for mandatory IFRS issuers decreases by 36.6 basis points per annum after IFRS, relative to a benchmark sample of non-adopters; this reduction corresponds to almost $1.5 million annual cost savings for an average IFRS bond issuer. In contrast, we find no relationship between the mandatory switch to IFRS and loan rates. These results suggest positive externalities for debt financing at the time of the IFRS mandate but only for bond markets, where reliance on public financial reporting is more prevalent than private communication. Our findings are robust to alternative research designs, most notably the employment of a constant sample of firms, a matched sample approach and endogenous switching regression models. Next, we extend our analysis to consider the role of differences in country-level institutions. The expected effects of mandatory IFRS adoption likely depend on the characteristics of the adopting country s institutions (Hail et al. 2010). Therefore we first consider whether the observed IFRS-related debt market changes depend on the extent of the accounting changes necessary to move from local GAAP standards to IFRS reporting. We expect the debt financing effects to be stronger in countries with more substantial accounting differences (e.g., Florou and Pope 2012; Ball et al. 2014). Consistent with our predictions, we find evidence that the propensity to issue

4 1410 A. Florou, U. Kosi a bond is higher and the cost of bonds is lower primarily for first-time IFRS issuers in countries with a higher divergence between domestic standards and IFRS. Second, we attempt to separate the debt financing effects of mandatory IFRS adoption from the potential confounding effects of concurrent changes in EU market regulations and enforcement suggested by current research (Christensen et al. 2013). Our tests show the presence of debt market changes even for first-time IFRS adopters in EU countries that experienced no contemporaneous enforcement or other institutional reforms. However, consistent with prior studies (e.g., Florou and Pope 2012), we also show that mandatory IFRS adopters from non-eu countries do not experience any debt market changes; therefore, we cannot rule out the potential influences of other concurrent institutional changes. Lastly in line with Ball et al. (2014) and contrary to IFRS studies focusing on equity markets, we provide evidence that the documented debt financing effects do not vary across EU countries as a function of the strength of their enforcement regime. This observation reinforces the argument that inferences of IFRS research on equity markets may not necessarily apply to debt markets. We believe this is the first paper to explicitly show debt financing benefits at the time of mandatory IFRS adoption. We demonstrate that mandatory IFRS reporting is associated with increased access to the bond market and reduced bond yield spreads. Compared to a sample of non-adopters, mandatory IFRS adopters raise more public debt at a lower cost after the mandate. Our findings contribute primarily to the limited literature on the debt market consequences around mandatory IFRS adoption. In particular, our study most closely relates to research by Florou et al. (2013), Wu and Zhang (2014), and Ball et al. (2014). Florou et al. (2013) use an international sample of mandatory adoption firms and find a higher credit relevance of IFRS financial statement information, defined as the ability of accounting numbers to explain credit analysts ratings. Similarly, Wu and Zhang (2014) find that credit ratings are significantly more sensitive to accounting factors following mandatory IFRS adoption but only in strong rule of law countries. In contrast, Ball et al. (2014) adopt a debt contracting approach and find that IFRS financial statements are less useful for covenant design, primarily in private loan contracts. While these studies focus on the IFRS-related financial reporting and contracting implications for debt markets, we concentrate on the IFRS-related valuation implications for debt securities. 2 Our study also complements the work of Daske et al. (2008) and Li (2010), who focus on the equity markets and report a reduction in the cost of equity around mandatory IFRS adoption, though the former yields mixed results. Unlike studies that examine secondary market effects, we analyze the IFRS-related consequences for new debt issuance. The advantage is that the cost of debt is directly observable, while the implied cost of equity is subject to estimation error (Easton and Monahan 2 Similarly to Ball et al. (2014), Kim et al. (2011) focus on bank loan contracts and examine the impact of voluntary adoption of International Accounting Standards (IAS) on various aspects of the terms of private debt agreements. Chen et al. (2013) examine the debt market effects of mandatory IFRS adoption, again in the context of bank loan contracts. Finally, Beneish et al. (2012) examine the macroeconomic debt effects of mandatory IFRS adoption by focusing on country-level foreign debt investment.

5 Does mandatory IFRS adoption facilitate debt financing? ), which can be especially problematic at the time of significant changes in firms accounting standards (Daske et al. 2008). We also contribute to the literature on the differential effects of accounting quality and disclosure across public and private debt markets. In particular, our paper relates to research by Bharath et al. (2008) and Dhaliwal et al. (2011). Based on a sample of firms reporting under US GAAP, Bharath et al. (2008) show that borrowers with higher accounting quality, proxied by abnormal accruals, are more likely to issue public bonds than to borrow from banks. They also report a negative relationship between accounting quality and the cost of debt, although this effect is greater for bonds than for loans. Using a US sample again, Dhaliwal et al. (2011) focus on the role of disclosure. They document that firms with higher quality disclosures are less likely to access the private debt market. Both studies are consistent with higher information asymmetries and adverse selection costs in the public than private debt markets; private lenders can set financing arrangements that compensate for less informative and transparent financial statements. 3 We complement this research by reporting results consistent with IFRS-related benefits for public but not private debt markets. The reminder of this paper is structured as follows. Section 2 discusses prior theory and related literature. Section 3 describes our research design and data, while Section 4 discusses our empirical results, including those of additional analysis and sensitivity tests. Section 5 concludes. 2 Theory and related literature Debt providers face asymmetric information problems when assessing the ability of firms to repay debt, which in turn is dependent on the value of assets and future cash flow prospects. Theory posits that higher quality and more comparable accounting information can reduce information asymmetry and estimation risk (for detailed discussions, see Dye 1990; Verrecchia 2001; Easley and O Hara 2004; and Lambert et al. 2007). Consistent with theory, prior empirical studies have confirmed the role of financial statement information in mitigating agency costs in a debt-financing context. For example, Francis et al. (2005) show that borrowers with higher accrual quality pay less for debt; Zhang (2008) documents lower interest rates for firms that report more conservatively; and Dhaliwal et al. (2011) find that firms with lower disclosure quality are more likely to access the private than public debt market. In summary, prior work suggests that financial reporting attributes are associated with the source and cost of debt (see Armstrong et al. 2010, for a recent literature review). IFRS are predicted to affect financial reporting in many jurisdictions because of more extensive and informative disclosures, better measurement and recognition 3 Ball et al. (2013) also study the implications of the contracting differences between public and private debt lenders, but they focus solely on U.S. cross-listings. Consistent with the bonding hypothesis, which predicts debt market benefits because of stronger creditor protection and legal enforcement in the United States, they demonstrate that cross-listed firms are more likely to raise public debt and also to incur lower yield spreads but only for bonds.

6 1412 A. Florou, U. Kosi rules, and enhanced comparability (Hail et al. 2010). Because the mandatory switch from local GAAP to IFRS is likely to affect financial reporting, it may also affect the propensity of firms to access the debt markets as well as the cost of debt. However, it is difficult to accurately predict all the consequences for debt financing, especially in regard to the recognition and measurement rules. First, the nonlinear structure of debt payoffs gives rise to asymmetric demand by creditors for financial information. Because of its limited upside, the value of debt is less affected by upward prospects, but it is more sensitive to downside risk. As a result, there is a higher demand by debt holders for negative information (Ball et al. 2008a, b). Moreover, debt holders also require reliable information, because it limits the extent of managerial discretion in accounting measurement (Watts 2003). However, the effects of IFRS financial reporting on accounting conservatism and reliability are not clear a priori, primarily due to the fair value orientation. On the one hand, fair value accounting and other related IFRS rules, such as recognition of impairments under IAS 36 and recognition of pension liabilities under IAS 19, should lead to more timely recognition of economic losses. In this case, the information in IFRS financial statements will meet the qualitative requirements of debt holders, thus enabling debt financing. On the other hand, fair value accounting may reduce accounting conservatism and cast doubt on reliability because (a) it can result in the recognition of unrealized economic gains, such as those on trading securities and other financial instruments, according to IAS 39 and IFRS 9 (Schipper 2005; Ball et al. 2014); (b) it requires the recognition of transitory losses and gains in income statement numbers, which may impair the usefulness of those numbers for predicting a borrower s future debtservicing capacity (Ball et al. 2014); and (c) it requires the extensive use of unverifiable estimates and judgments in the valuation of assets and liabilities, which can reduce perceived reliability and increase opportunistic balance sheet and earnings manipulation (Ball et al. 2014). In this case, the financial information provided by IFRS may be less useful to debt holders, thus impeding debt financing. 4 Second, lenders use the book value of assets as a way to determine a firm s ability to service future debt. Asset values are also used by debt holders for collateral purposes (Armstrong et al. 2010). Therefore they require current but reliable estimates of asset values (Watts 2003). Again, the implications of IFRS financial reporting for asset valuation and therefore debt financing are not clear a priori. According to the recognition and measurement rules of property, plant, and equipment under IAS 16 and investment property under IAS 40, companies may use either historical cost or fair value; this choice must be stated in the annual reports following IFRS adoption and be applied consistently. 5 Companies that choose the historical cost method provide more reliable but potentially less relevant future estimates of asset values. Companies that choose the fair value method provide more timely and up-to-date asset values, but they may be less reliable. 6 In summary, 4 In this respect, Ahmed et al. (2013) investigate mandatory IFRS adoption and document less timely loss recognition. 5 IAS 40 requires that the fair value of an investment property asset is either recognized on the balance sheet or disclosed in financial statement footnotes.

7 Does mandatory IFRS adoption facilitate debt financing? 1413 the question of the debt financing consequences around mandatory IFRS reporting is essentially an empirical issue. Furthermore, the debt financing consequences at the time of mandatory IFRS adoption are expected to differ between bond and loan financing. Theory emphasizes significant institutional differences between public and private debt markets in response to information asymmetry concerning credit quality and the quality of investment projects (see Gordon and Winton 2003, for a comprehensive review). First, dispersed public lenders have weaker incentives (compared to private lenders) to engage in costly information collection and borrower monitoring, due to free-rider problems and duplication of monitoring costs (Gordon and Winton 2003). Moreover, firms are less inclined to provide proprietary information to dispersed bond holders than to small groups of loan providers (Bhattacharya and Chisea 1995). Finally, contracts may be much harder to renegotiate based on new information if credit problems arise due to information and coordination problems among diffused bond holders (Diamond 1991; Rajan 1992). In sum, public lenders are less likely to have access to private information and are less likely to engage in producing information, monitoring borrowers, and renegotiating debt following initial issuance. Consequently, bond-financing arrangements tend to be more boilerplate, to rely heavily on published financial statement information, and potentially are more sensitive to changes in financial reporting. In contrast, private lenders (e.g., banks) have both the ability and incentives to ex ante acquire and examine valuable nonpublic information (e.g., management accounts, budgets, and forecasts), which they then use in credit approval decisions; loan providers invest a significant stake of their wealth in borrowers about which the former have become privately informed (Boyd and Prescott 1986). Second, private lenders may also have an information advantage over diffused arm s length lenders (e.g., bond holders) after the capital is provided; loan providers can be efficient ex post monitors by privately interacting with borrowers over time and across different products (Diamond 1991). Third, because of access to firm s private information and superior monitoring, private lenders enjoy generally greater flexibility and have lower costs for renegotiating debt (Rajan 1992); this in turn suggests that initial contract setting may be less critical in the case of private loans, compared to public bonds. In sum, the literature describes loan suppliers as being particularly good at privately investigating and monitoring informationally opaque firms as well as renegotiating debt. As a result, loan-financing arrangements are generally more customized, are often based on non-gaap as well as GAAP information, and may be less sensitive to changes in public financial information quality. 6 Muller et al. (2011) focus on European real estate firms. They find larger declines in information asymmetry in the post-ifrs period, as reflected in lower bid-ask spreads, for mandatory IFRS adoption firms that did not provide fair values before adoption. However, they also show that these firms continue to have higher information asymmetries than voluntary adoption firms, which appears partly attributable to the generally lower reliability of fair values. Similarly, Christensen and Nikolaev (2013) study the accounting treatment of long-term nonfinancial assets by mandatory IFRS adopters in the United Kingdom and Germany. They document that fair value accounting is used exclusively for property, most likely because of the relative liquidity of property markets, which therefore provide more reliable measures of fair value. They also show that firms that report reliable fair value estimates of investment property tend to access the debt markets more frequently.

8 1414 A. Florou, U. Kosi If the mandated introduction of IFRS enhances the comparability and quality of financial statement information required by debt providers, and therefore reduces information asymmetries as well as monitoring and renegotiation costs (which are higher for public bond than private loan agreements), we predict that mandatory IFRS adopters will be more likely to access the public debt market than to borrow privately. We also expect them to incur lower debt costs. However, in light of the information, monitoring and renegotiation advantages of private lenders (compared to public bond holders), we expect any impact on the cost of debt to be smaller for private debt. 3 Research design and data 3.1 Empirical specifications Access to public versus private debt To test the relationship between mandatory IFRS adoption and access to public versus private debt, we use an empirical model based on specifications in Bharath et al. (2008) and Daske et al. (2008): Public market tþ1 ¼ b 0 þ b 1 First Time mandatory t þ b 2 Voluntary t þ b 3 Voluntary t Mandatory þ Rb j Firm Specific controls j;t þ Rb k Country ð1þ Specific controls k;tþ1 þ e tþ1 where two proxies are used for the dependent variable. First, similarly to prior studies, we perform an issue-level analysis after deleting debt issues of the same type by the same firm within 1 year (e.g., Denis and Mihov 2003; Bharath et al. 2008; Dhaliwal et al. 2011). In particular, for each issue, we generate a dummy variable, Public Market Dummy, that equals 1 if a firm borrows in the public market and 0 if it borrows from private lenders. In this case, we use a probit regression on the total sample of bonds and loans with 13,546 observations. 7 Next, we use a firm-level analysis by estimating Eq. (1) in two ways: (a) by probit regression on a sample of 12, unique firm-years with the above Public Market Dummy dependent variable; and (b) by OLS on a sample of 12,574 unique firm-years with a continuous dependent variable, Public Market Continuous, calculated for each firm-year as the ratio between the total amount of public debt and the total amount of public and private debt issued in a particular year. The value of this variable ranges from 0 (i.e., the firm obtains only new loans in a particular year) to 1 (i.e., the firm issues only new bonds in a particular year). 8 7 The number of observations is slightly lower than the total number of issues of 14,003 reported in Table 1. This is because of lost observations when the independent dummy variables (i.e., the industryyear or country indicators) perfectly predict the dependent dummy variable.

9 Does mandatory IFRS adoption facilitate debt financing? 1415 For the IFRS indicator variables, we adopt Daske et al. s (2008) research design and define two binary variables as follows: (a) First-Time Mandatory, which refers to observations of firms that did not report under IFRS before the mandatory adoption. It is equal to 1 for all firm-years, with IFRS reporting ending on or after the local mandated IFRS adoption date and 0 otherwise (note that this indicator captures the effects for first-time mandatory adopters, relative to a sample of nonadopters after IFRS adoption), and (b) Voluntary, which refers to observations of firms that voluntarily switched to IFRS reporting before it was mandated. It is equal to 1 for all firm-years with IFRS reporting by voluntary adopters and 0 otherwise. To capture any incremental effects of mandatory IFRS introduction for the voluntary IFRS adopters, we construct a separate binary variable, Mandatory, that equals 1 for firm-years ending on or after the local mandatory IFRS adoption date and 0 otherwise. We then interact it with the Voluntary indicator. We identify IFRS adopters based on the reporting standards used in each firm-year (Worldscope item #07536) and the coding established by Daske et al. (2012). If first-time mandatory IFRS adopters are more likely to issue public debt following the adoption, we predict a positive sign for b 1 in Eq. (1). Similarly, b 3 is expected to be positive if voluntary adopters also benefit from the mandating of IFRS, possibly because of comparability externalities. 9 In line with prior literature, we include the following determinants expected to be associated with debt access: Size, Tangibility, and Market to Book to proxy for information asymmetries, flotation costs, and growth opportunities, respectively (Denis and Mihov 2003; Bharath et al. 2008); Leverage, Ohlson s O-Score, Investment Grade, and Rated to control for borrowers credit quality and the probability of default (Denis and Mihov 2003; Bharath et al. 2008; Ball et al. 2013); ROA, Returns, and Returns Variability to control for firm performance and firm risk (Kim et al. 2011; Ball et al. 2013); Capital Market Access to proxy for financing needs (Bharath et al. 2008; Dhaliwal et al. 2011); and US GAAP to control for implementation of an alternative set of high quality accounting standards (Daske et al. 2008). We also include GDP Growth and Country Rating to control for potentially time-varying country-level factors. Firm-specific variables come from Worldscope; we obtain GDP Growth and Country Rating data from World Bank and Standard & Poor s Sovereign Ratings, respectively; all variables are described in detail in the Appendix. 8 We re-estimate Eq. (1) using a double-censored (i.e., two-limit) tobit, given that the dependent variable Public Market Continuous has a lower value of 0 and an upper value of 1. Alternatively, we also apply a logit transformation of the dependent variable. In both cases, results remain qualitatively identical. 9 Although we focus on mandatory IFRS adopters, note that prior literature documents different dynamics for voluntary adopters than for first-time adopters of capital market changes at the time of the IFRS mandate. This implies how complex these dynamics are for voluntary adopters. For example, Florou and Pope (2012) report different quarterly incremental institutional holding changes around mandatory IFRS adoption between voluntary and first-time adopters. Similarly, Daske et al. (2008) find a larger decline in the cost of equity for voluntary than first-time adopters when IFRS financial reporting becomes mandatory. Li (2010), in contrast, does not document any change in the cost of equity for voluntary adopters around mandatory IFRS adoption.

10 1416 A. Florou, U. Kosi Cost of debt Based on prior research (Bharath et al. 2008; Daske et al. 2008; Kim et al. 2011), we estimate the following empirical specification to test the relationship between mandatory IFRS adoption and the cost of debt: Cost of debt tþ1 ¼ b 0 þ b 1 First Time mandatory t þ b 2 Voluntary t þ b 3 Voluntary t Mandatory þ Rb j Firm Specific controls j;t þ Rb k Issue Specific Controls k;tþ1 þ Rb 1 Country Specific Controls l;tþ1 þ e tþ1 ; ð2þ Since debt issues by the same firm may have different contractual terms, we perform an issue-level analysis that includes multiple debt issues by the same firm within 1 year (e.g., Kim et al. 2011; Ball et al. 2013). Using single equation models, we first estimate Eq. (2) for 5793 bonds and 8627 loans. Next, we follow Bharath et al. (2008) and estimate an endogenous switching model in order to (a) control for selection bias and (b) enable a direct comparison of coefficient estimates to examine whether mandatory IFRS adoption relates to the cost of public and private debt differently. Because firms do not select debt markets randomly, we can observe the cost of either public or private debt conditional on market choice. As in Bharath et al. (2008), we simultaneously estimate the selection equation of the probit issuebased model specified in Eq. (1), where Capital Market Access, GDP Growth, and Country Rating are the exogenous variables unique to the selection stage. We then run two regression models, one for each debt market as specified in Eq. (2), where the characteristics of the debt contract are the exogenous variables unique to the second-stage regression. Again in line with prior studies (Bharath et al. 2008; Kim et al. 2011), we define Cost of Debt in the public market as the basis point spread over a government bond issued by the same country with comparable maturity and in the same currency. We can thus control for interest rate shifts and the time preference for money. In the private market, we define Cost of Debt as the amount the borrower pays in basis points over LIBOR or the LIBOR equivalent for each loan dollar drawn (including annual fees). If mandatory IFRS adoption is associated with reduced bond costs, we expect b 1 and b 3 to be negative. Moreover, this negative effect is predicted to be smaller or absent for loans. Hence we expect the estimates on these coefficients to be less negative or insignificant, respectively. We use several controls that are predicted to be associated with the cost of debt. These include firm-specific variables that capture observable characteristics of firms default risk: Size, Tangibility, Leverage, Market to Book, O-Score, Investment Grade, Rated, Current Ratio, ROA, Returns, and Return Variability (Bharath et al. 2008; Ball et al. 2013; Baghai et al. 2014). We also control for issue-specific characteristics that are systematically related to debt pricing. These include Amount, Maturity, and Secured (Bharath et al. 2008; Kim et al. 2011). Similar to Eq. (1), we use the US GAAP indicator. To remove the effects of other non-ifrs-related factors from the IFRS coefficients in Eq. (2), we include a

11 Does mandatory IFRS adoption facilitate debt financing? 1417 contemporaneous variable Market Benchmark, defined as the yearly mean of the dependent variable from observations in nonmandatory IFRS reporting countries (Daske et al. 2008). Finally, we include two country factors that may be associated with the cost of debt, namely, Term Spread as a proxy for economic uncertainty (Campbell and Taksler 2003; Graham et al. 2008) and Median Distance to Default (DTD) as an overall country-level measure of corporate default risk (Correia et al. 2012; Baghai et al. 2014). 10 In all models under Eqs. (1) and (2), we include country- and industry-year fixed effects, and we winsorize continuous variables at the 1st and 99th percentiles. 11 All empirical specifications are estimated with White standard errors, adjusted to account for correlations within country-year clusters (Petersen 2009). 12 The appendix provides a detailed description of all variables. 3.2 Sample selection We obtain our sample from public bond issues and private loan arrangements completed by listed nonfinancial firms for the period; we exclude the post-2007 period because of the distortionary effect of the global credit crisis on debt markets. We exclude borrowers who do not have data on all the required firmspecific variables, as well as countries with fewer than 20 observations. Our public debt data come from Securities Data Company s (SDC) Platinum database, which includes 18,684 global bond offerings by public nonfinancial institutions during Using several company identifiers (i.e., ISIN, CUSIP, and ticker symbol) and company name, we successfully match 11,678 bond issues to Worldscope (62.5 % of total bonds by listed firms included in SDC). In line with prior studies (e.g., Bharath et al. 2008; Ball et al. 2013), we exclude convertible bonds and bonds with callable features. For the access to debt analysis, we obtain a final sample comprising of 2346 unique Worldscope firms issuing 4676 bonds (excluding multiple bond issues by the same firm within 1 year). For the cost of debt analysis, we obtain a final sample of 1560 unique borrowers issuing 5793 bonds (with all required issue-specific variables). 10 Term Spread is calculated as the difference between the 10- and two-year government bond rate at the country-month level. We obtain data on monthly government bond rates from Datastream. Median DTD is the country-month median value of the firm-level distance to default (DTD) measure that combines leverage and asset volatility. The firm DTD measures come from the Risk Management Institute (RMI) at the National University of Singapore (NUS). For more details, see and Duan and Wang (2012). The firm-level DTD measure is not available for all our sample firms. Nevertheless, in our primary analysis, we also estimate the cost of debt model after including the firm-level DTD. 11 Our main findings remain robust to the inclusion of separate fixed effects by year and by industry as well as to the inclusion of firm-level fixed effects. 12 We employ one-way clustering at country-year level (i.e., our clustering unit is every year in a given country). We assess the sensitivity of our primary analysis to two-way clustering by country and by year and find that our results persist. However, we note that given our relatively short period (eight year clusters) and the number of our country clusters that can be as low as 13 (depending on the estimated model), two-way clustering by country and by year can be subject to the small cluster problem and lead to biased standard errors (Petersen 2009, p. 460).

12 1418 A. Florou, U. Kosi We obtain private debt data from Dealscan, which provides detailed information on pricing and other loan terms worldwide. The basic units are loans (also called facilities). When a borrower enters into multiple loans at the same contract date, they are grouped together into a package (also called a deal). We consider each facility as a separate observation, since loan characteristics (e.g., the cost of debt, maturity) can vary across facilities. The Dealscan population is comprised of 88,204 private debt issues by nonfinancial firms during We first use ticker symbol, which is the only company identifier provided by Dealscan, to match its population to Worldscope. 22,633 of the above loans have a ticker symbol, of which we successfully match 10,491 (46.4 %); this proportion resembles the 54 % reported by Dichev and Skinner (2002). 13 Following Kim et al. (2011), we also use company name to increase the matched sample, allowing us to match an additional 10,079 loans. In line with Bharath et al. (2008), we select all senior loans, and we restrict the sample to term loans, revolvers, and 364-day facilities. 14 The final sample consists of 9327 loans obtained by 3938 unique Worldscope borrowers for the access to debt analysis and of 8627 private issues by 2454 unique firms for the cost of debt analysis (with all required issue-specific variables). Table 1 summarizes our sample selection process. The matched bond and loan samples reflect (a) the large number of nonpublic borrowers included in the debt databases (e.g., government entities and private firms) and (b) the differences in firm identifiers between the debt databases and Worldscope. The sample selection process of Dealscan and SDC Platinum may be of potential concern. Both provide comprehensive coverage of corporate global debt, but their data collection procedures may be skewed toward US, larger firms and, in the case of Dealscan, perhaps toward larger loans (Strahan 1999; Roberts and Sufi 2009; Ball et al. 2013). Clearly, SDC Platinum will include larger firms, which are more likely to issue bonds. But these firms are also most likely to confront the choice between issuing public and private debt, examined in the current study. The Dealscan loan data for US firms is collected primarily from SEC filings. Data on the private credit of non-us firms, and especially of smaller loans, may be less extensive because the reporting requirements are less stringent. However, information constraints may be more prominent for younger and smaller firms (with possibly smaller loans). Thus excluding such cases may understate our predicted relationships (Strahan 1999; Denis and Mihov 2003) Note that 26 % of total Dealscan observations have a ticker symbol; this percentage is lower than the 43 % reported by Dichev and Skinner (2002). However, their study is based only on U.S. loans. If we focus only on U.S. private issues, we also find that 42 % of the total Dealscan population has a ticker symbol. 14 To the extent that it is less appropriate to compare long-term public debt with private shorter-maturity debt, we assess the sensitivity of our findings to the exclusion of 364-day facilities, which account for 15.3 % of total loans in our sample. Our results remain qualitatively unchanged. 15 Nevertheless, we assess the robustness of our analysis to potential selection biases and find qualitatively identical results (see Sect. 4.4). Also, prior research suggests that some loans covered by Dealscan are renegotiations of existing deals, rather than new deals. Identifying renegotiations is not possible based on the information provided by Dealscan. It would also require manual data collection, which is not feasible because of weak national reporting requirements (Roberts and Sufi 2009). Arguably, a lender who renegotiates with a firm already has private information and is less reliant on the firm s

13 Does mandatory IFRS adoption facilitate debt financing? 1419 To demonstrate the generalizability of our results, we compare our sample of borrowers from SDC Platinum or Dealscan to the Worldscope population during We can confirm that our sampling procedure includes primarily larger firms. Moreover, in line with prior studies, we document that firms from Worldscope that overlap with either SDC Platinum or Dealscan are also more leveraged and more profitable than the general Worldscope universe (Strahan 1999; Dichev and Skinner 2002). Following prior IFRS literature (Daske et al. 2008; Florou and Pope 2012), we identify two subsamples: (a) a treatment sample consisting of issuers from mandatory IFRS adopter countries and (b) a benchmark sample consisting of issuers from countries that do not allow or require IFRS reporting and after excluding voluntary IFRS adopters in these countries. Table 2 shows our sample composition by country and year for the access to debt and the cost of debt analysis. For the sake of brevity, we constrain our discussion to the former; our main observations between the two analyses are similar. Panel A shows results for the bond sample. There are 905 bond issues from 16 countries in the treatment sample, with the U.K., France, and Germany having the highest proportions of observations. Bond issues by voluntary and first-time mandatory IFRS adopters are 19.6 and 20.2 % of the sample, respectively. The benchmark sample consists of 3771 bond issues from 14 countries, with the United States and Japan accounting for the highest number of observations. Panel B shows results for the loan sample. There are 1406 loans from 17 countries in the treatment sample; the United Kingdom, Australia, and Germany dominate. Loans relating to voluntary and first-time mandatory adopters are 14.7 and 24.4 % of the sample, respectively. The benchmark sample has 7921 loans from eight countries, with borrowers from the United States and Japan again the most prevalent. Panel C breaks down the sample composition by year of debt issue. 3.3 Data description Table 3 reports descriptive statistics for the variables used in the regression tests for both the treatment and benchmark samples. Panel A describes the access to debt variables (Public Market Dummy and Public Market Continuous). Panels B and C show the bond- and loan-specific characteristics, respectively, while panel D describes firm- and country-level variables for the full sample of bonds and loans. The descriptive evidence suggests that firms from treatment countries are more likely to issue bonds and to pay lower loan spreads but higher bond spreads than firms from benchmark countries. However, these statistics do not control for the determining factors of the source of debt and borrowing costs, so we investigate these issues further in our regression analysis. In terms of control variables, we find Footnote 15 continued financial statements than a first-time lender would be. However, while the level of dependence on financial statements may vary across private lender types, prior theory predicts that all banks have access to private information and therefore rely less on public financial statements than bond holders (see Sect. 2). We believe our inferences regarding the debt-market effects around mandatory IFRS adoption are unlikely to be biased because of renegotiations.

14 1420 A. Florou, U. Kosi Table 1 Sample selection process Observations Firms Panel A: SDC Platinum-bond sample Bonds of public nonfinancial institutions ( ) 18, Worldscope population of nonfinancial institutions ( ) 217,907 36,514 Matched sample 11, Bonds with convertible/callable features Nonfinancial institutions with selected bonds 11, Missing data on financial variables Countries with less than 20 observations Total sample Multiple debt issues of the same type by the same firm within one year Final sample for access to debt Total sample Missing data on bond variables Final sample for cost of debt Panel B: Dealscan-loan sample Loans of nonfinancial institutions ( ) 88,204 27,797 With ticker symbol 22, Without ticker symbol 65,571 23,064 Worldscope population of nonfinancial institutions ( ) 217,907 36,514 Matched sample 20, Matched on ticker symbol 10, Matched on name 10, Non-senior/short-term loans Nonfinancial institutions with selected loans 19, Missing data on financial variables Countries with less than 20 observations Total sample 14, Multiple debt issues of the same type by the same firm within 1 year Final sample for access to debt Total sample 14,448 3,938 Missing data on loan variables Final sample for cost of debt The sample period is Panels A and B present sample selection for bonds and loans, respectively. We match firms from SDC Platinum to Worldscope based on company identifiers (i.e., ISIN, CUSIP, and ticker symbol) and company name. In Dealscan, we distinguish between borrowers with and without a ticker symbol. We match firms from Dealscan to Worldscope based on ticker symbol and company name significant differences in the mean values of almost all the variables between the treatment and benchmark subsets. These findings are consistent with prior literature (e.g., Kim et al. 2011), which also reports differences in several characteristics

15 Does mandatory IFRS adoption facilitate debt financing? 1421 Table 2 Sample composition by country and year Treatment sample Benchmark sample Access to debt Cost of debt Access to debt Cost of debt Country Issues Firms Issues Voluntary Issues First-time mandatory Issues Firms Issues Voluntary Issues First-time mandatory Country Issues Firms Issues Firms Panel A: bond sample by country Australia Brazil Austria Canada Finland Chile France China Germany India Hong Kong Indonesia Ireland Japan Italy South Korea Netherlands Malaysia Philippines Mexico Singapore Peru Spain Taiwan Sweden Thailand Switzerland US UK Venezuela Total Total

16 1422 A. Florou, U. Kosi Table 2 continued Treatment sample Benchmark sample Access to debt Cost of debt Access to debt Cost of debt Country Issues Firms Issues Voluntary Issues First-time mandatory Issues Firms Issues Voluntary Issues First-time mandatory Country Issues Firms Issues Firms Panel B: loan sample by country Australia Bermuda Finland Canada France Chile Germany Japan Greece South Korea Hong Kong New Zealand Ireland Taiwan Italy US Netherlands Norway Philippines Singapore South Africa Spain Sweden Switzerland UK Total Total

17 Does mandatory IFRS adoption facilitate debt financing? 1423 Table 2 continued Bonds Loans Access to debt Cost of debt Access to debt Cost of debt Issue year Issues Firms Issues Voluntary Issues First-time mandatory Issues Firms Issues Voluntary Issues First-time mandatory Issues Firms Issues Voluntary Issues First-time mandatory Issues Firms Issues Voluntary Issues First-time mandatory Panel C: sample composition by year Total The bond sample is comprised of 4676 issues by 2346 unique firms for the access to debt analysis and 5793 issues by 1560 unique firms for the cost of debt analysis; the loan sample is comprised of 9327 issues by 3938 unique firms for the access to debt analysis and 8627 issues by 2454 unique firms for the cost of debt analysis. The sample period is Countries in the bond and loan samples are split into two groups: countries that mandated IFRS reporting (treatment sample) and countries that do not allow or require IFRS reporting (benchmark sample). Panels A and B report the number of bond and loan issues and the number of firms by country. For the treatment sample, we also report the number of issues relating to voluntary and first-time mandatory IFRS adopters. Panel C reports the number of bonds and loans, the number of firms, as well as the number of issues relating to voluntary and first-time mandatory IFRS adopters by issue year. First-Time Mandatory refers to firms that never reported under IFRS before mandatory adoption and equals 1 for all firm-years with IFRS reporting ending on or after the local mandated IFRS adoption date and 0 otherwise. Voluntary relates to firms that voluntarily switched to IFRS reporting before it was mandated and equals 1 for all firm-years with IFRS reporting by voluntary adopters and 0 otherwise