Mandatory adoption of IFRS and timely loss recognition across Europe: The effect of corporate finance incentives

International Review of Financial Analysis 38 (2015) 70–82

Contents lists available at ScienceDirect

International Review of Financial Analysis

Mandatory adoption of IFRS and timely loss recognition across Europe: The effect of corporate finance incentives☆ Ann L.-C. Chan a, Audrey W.-H. Hsu b,⁎, Edward Lee c a b c

Department of Accounting, National Chengchi University, Taipei 11605, Taiwan Department of Accounting, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei 106, Taiwan Manchester Accounting and Finance Group, Manchester Business School, University of Manchester, Crawford House, Oxford Road, Manchester M13 9PL, UK

a r t i c l e

i n f o

Article history: Received 19 June 2014 Received in revised form 25 December 2014 Accepted 2 February 2015 Available online 7 February 2015 JEL classification: F30 G15 M41 Keywords: IFRS Costs of debt capital Timely loss recognition Corporate finance

a b s t r a c t We examine whether firms have increased their timely loss recognition with the mandatory adoption of International Financial Reporting Standards (IFRS) across Europe since 2005. We estimate firm-specific asymmetric timeliness using the Khan and Watts (2009) C-score, which accounts for size, market-to-book, and leverage. We use firms that voluntarily adopted IFRS before the mandatory adoption date as a control sample to address the effect of unidentified confounding events. We find increased timely loss recognition relative to this control sample only among mandatory IFRS adopters with a higher cost of debt and in countries less dependent on private debt or bank financing. Our results are robust to controls for firm characteristics such as interest coverage, return on assets, earnings volatility, loss, accrual quality, beta, and growth, as well as both industry and country effects. We confirm that corporate finance incentives play a decisive role in determining firms' timeliness of loss recognition after mandatory IFRS adoption. © 2015 Published by Elsevier Inc.

1. Introduction Recent literature on the mandatory adoption of International Financial Reporting Standards (IFRS) across the European Union focuses extensively on the economic consequences of this new accounting regime for the capital markets.1 These studies generally reveal that mandatory IFRS adoption has a beneficial impact on the capital market, although mainly among countries with better legal enforcement. However, capital market economic consequences are only indirect effects of the change in accounting standards and therefore merely provide ☆ We would like to thank Hans Bonde Christensen, Judy Day, Steven Lin, Norman Strong, Peter Taylor, and Martin Walker for their useful comments in the development of this paper. ⁎ Corresponding author. Tel.: +886 233661131. E-mail addresses: [email protected] (A.L.-C. Chan), [email protected] (A.W.-H. Hsu), [email protected] (E. Lee). 1 For example, existing studies have evaluated cost of equity capital (e.g., Daske, Hail, Leuz, & Verdi, 2008; Li, 2010), stock return volatility (e.g., Beuselinck, Joos, Khurana, & Van der Meulen, 2009; Landsman, Maydew, & Thornock, 2012), costs of debt capital (e.g., Florou & Kosi, 2013; Wu & Zhang, 2009), sell-side analyst forecast (e.g., Byard, Li, & Yu, 2011; Tan, Wang, & Welker, 2011), institutional ownership (e.g., DeFond, Hu, Hung, & Li, 2011; Florou & Pope, 2012), and value relevance (Agostino, Drago, & Silipo, 2011). See Leuz and Wysocki (2008) and Bruggemann, Hitz, and Sellhorn (2013) for extensive literature reviews.

http://dx.doi.org/10.1016/j.irfa.2015.02.002 1057-5219/© 2015 Published by Elsevier Inc.

indirect evidence of the impact of IFRS. The direct effect of IFRS should be changes in the quality of accounting disclosure, which in turn would provide more direct evidence of the influence of the new accounting standards. Despite this, studies of the mandatory IFRS adoption have so far paid relatively less attention to its effect on accounting disclosure quality and those that do yield mixed findings regarding its benefits.2 We examine the impact of mandatory IFRS adoption across Europe on accounting disclosure quality by focusing on timely loss recognition. The inverse relationship between timely loss recognition and costs of debt is well established in the literature (e.g., Ahmed, Billings, Morton, & Stanford-Harris, 2002; Ball, Robin, & Sadka, 2008; Ball & Shivakumar, 2005; Watts, 2003a,b; Zhang, 2008). Timely recognition of loss benefits lenders since it enhances debt contracting efficiency by causing poorly performing borrowers to breach debt covenants in a timely fashion (e.g., Zhang, 2008). However, an inconsistency exists in the existing literature, between the expected and the observed impact of mandatory IFRS adoption on timely loss recognition. Ball (2006) predicts that IFRS will increase timely loss recognition, which will in turn enhance debt

2 For instance, Christensen, Lee, and Walker (2008) show improved accounting quality among voluntary but not mandatory adopters in Germany. Jeanjean and Stolowy (2008) show no reduction in earnings management among mandatory adopters in Australia, France, and the UK.

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82

contracting efficiency and could therefore reduce the costs of debt capital. Although Florou and Kosi (2013) show that the costs of public debt is indeed lower after the mandatory adoption of IFRS, Ahmed, Neel, and Wang (2013) and Chen, Tang, Jiang, and Lin (2010) both find that timely loss recognition instead declines among mandatory IFRS adopters. In other words, it appears that the direction of the new accounting standards' direct effect (i.e., the decrease in the timeliness of loss recognition) cannot substantiate the direction of its indirect effect (i.e., the decrease in the costs of debt). Therefore, empirical evidence of the decline in both loss recognition timeliness and costs of debt capital, following mandatory IFRS adoption, is difficult to reconcile.3 We argue that firms with higher costs of debt are more likely to increase their timely loss recognition following mandatory adoption of IFRS. Our argument is based on three grounds. First, the mandatory adoption of IFRS could render the capital market more sensitive to accounting disclosures than under previous domestic standards. By applying a more consistent set of accounting standards across a large set of countries, IFRS facilitates cross-border financial statement comparability (Ball, 2006). Second, the increased use of financial reports by investors after mandatory adoption of IFRS may lead firms to recognize losses more timely. Prior studies indicate that timely loss recognition is more pronounced when public financial disclosure is a more likely solution for the information asymmetry problem (e.g., Ball, Kothari, & Robin, 2000; Ball, Robin, & Wu, 2003). Thus, we would expect companies to recognize economic losses in a more timely fashion, especially when they face greater agency conflicts with debtholders. Third, we expect companies incurring higher costs of debts may exhibit greater increase of timely loss recognition than companies with lower costs of debt in the post-IFRS period due to greater contracting pressure. The mandatory adoption of IFRS provides firms with an opportunity to reduce agency costs of debt by reflecting economic losses in a more timely fashion, particularly for those with greater desire to reduce their costs of capital. The perceived economic benefit is expected to be greater under IFRS than previous domestic accounting standards as the capital market would pay more attention to accounting disclosures under IFRS and the improved cross-country accounting comparability under IFRS facilitates firms to acquire capital from foreign investors. Meanwhile, we also expect that firms with high costs of debt are more likely to increase the timeliness of their loss recognition after mandatory IFRS adoption if they are domiciled in countries with more prevalent public debt markets. One reason is that, between public and private lenders, the former group of investors have higher information costs and are more dependent on financial reporting information (e.g., Bharath, Sunder, & Sunder, 2008; Diamond, 1991; Fama, 1985). To test our assertion, we apply a sample that comprises 11,860 firmyear observations, from sixteen European countries, over the period from 2002 to 2007. Following existing studies of mandatory IFRS adoption effects (e.g., Byard et al., 2011; Daske et al., 2008; Li, 2009), our treatment sample consists of firms that have adopted IFRS since 2005 on a mandatory basis and our control sample consists of firms that voluntarily adopted IFRS before 2005. If the effect we predict is indeed associated with mandatory IFRS adoption, then it should occur only in the treatment sample. If this effect exists in the control sample as well, then it is likely to be caused by unidentified confounding events or time trends. We measure firm-specific timely loss recognition using the C-score developed by Khan and Watts (2009), which incorporates the effects of size, market-to-book ratio and leverage. Instead of the Basu (1997) regression approach, we use the C-score because it is more likely to capture firm- and time-specific changes in timely loss recognition, which is more appropriate for our research setting. We estimate the firm-specific costs of debt using interest expense divided by 3

Florou and Kosi (2013) sampled 21 countries and Ahmed et al. (2013) sampled 19 studies. The samples of these two studies overlapped by 18 countries. Therefore, it is unlikely that the differences in their results are driven by differences in the countries sampled.

71

total interest-bearing debt, following existing studies such as Pittman and Fortin (2004) and Francis, La Fond, Olsson, and Schipper (2005).4 Our analyses also control for firm characteristics that may determine firms' incentives to recognize losses on a timely basis as well as both industry and country effects. Finally, we determine the pervasiveness of private debt and bank-based financing among our sampled countries, following the approach of Bushman and Piotroski (2006). Our findings are as follows. First, using a difference-in-differences research design, we observe that mandatory adopters (treatment sample) with higher costs of debt are associated with incrementally higher C-scores relative to voluntary adopters (control sample) after 2005, when IFRS was enacted. Second, the aforementioned observations exist only among mandatory IFRS adopters domiciled in countries with a lower pervasiveness of private debt or bank-based financing. The results imply that the increase in timely loss recognition that can be attributed to mandatory IFRS adoption depends on firms' corporate finance incentives to reduce the cost of borrowing and on whether the firm is domiciled in countries where public debt is more prevalent. We contribute to the growing literature on IFRS in three ways. First, we are one of the first studies that argue that accounting conservatism is the means by which the benefit of mandatory IFRS adoption is realized. While many studies have shown that IFRS adoption is associated with lower costs of debt capital (e.g., Florou & Kosi, 2013; Wu & Zhang, 2009), there is little evidence on whether firms change their accounting choices around the adoption of IFRS to alleviate the agency conflicts with debtholders. We show that loss recognition timeliness is one mechanism that firms can employ to address debtholder–shareholder agency conflicts in the post-IFRS period, which is consistent with the prediction by Ball (2006, p. 12). Second, we show that corporate finance incentives play an influential role in determining whether firms commit to higher accounting disclosure quality after their mandatory IFRS adoption, which is consistent with the theory of Leuz (2010, pp. 248–250). While the existing literature provides mixed findings of the impact of mandatory IFRS adoption on timely loss recognition (e.g., Ahmed et al., 2013; Dimitropoulos, Asteriou, Kousenidis, & Leventis, 2013), our study highlights the need to consider the conditioning effect of corporate finance incentives. Finally, we also reconcile the inconsistency in the existing literature, between the expected and the observed impact of mandatory IFRS adoption on timely loss recognition (Ahmed et al., 2013; Ball, 2006; Chen et al., 2010). We show that the decrease in the timeliness of loss recognition after IFRS, documented by concurrent studies (e.g., Ahmed et al., 2013; Chen et al., 2010), is caused by a background time trend and not mandatory IFRS adoption. Our paper is organized as follows. Section 2 reviews the relevant literature and develops the hypotheses. Section 3 describes the methodology, sample, and data. Section 4 presents the empirical findings. Section 5 concludes. 2. Literature review and hypothesis development 2.1. Timely loss recognition and the costs of debt Higher accounting disclosure quality decreases information asymmetry, which in turn lowers lenders' perceived risk and reduces adverse selection problems (Verrecchia, 2001). When lending money to firms, investors need to assess the default risk of borrowers, based on all available information. Capital suppliers are likely to perceive firms that withhold information and have greater information uncertainty as riskier and therefore charge a higher premium to compensate (Diamond & Verrecchia, 1991). Accounting information is used in debt covenants 4 We use this measure because it enables us to cover a much larger sample size than other measures such as yield spread from bond issues or syndicated loans would allow. These alternative measures are often limited by data availability and could also incur sample selection bias. For instance, Dealscan has a greater coverage of syndicated loans for the US than for European firms.

72

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82

and thus affects lenders' contractual rights and their monitoring of borrowers (Holthausen & Leftwich, 1983; Watts & Zimmerman, 1986). Leftwich (1983) shows, for the US, that even where debt contracts are written with non-GAAP metrics, the numbers are usually backed out from reported GAAP numbers. Day and Taylor (1996) and Moir and Sudarsanam (2007) show, for the UK, that debt covenants rely on accounting-based metrics. Iatridis (2011) also find that timely loss recognition enhance contracting efficiency. Sengupta (1998) uses yield-to-maturity and the total interest costs of new debt issues to proxy the costs of debt and shows that it is lower among US firms with higher disclosure quality ratings from financial analysts. Beatty, Ramesh, and Weber (2002) show that US firms with more reporting flexibility, which they measure as the ability to include voluntary and mandatory accounting changes to compute covenant compliance, are charged higher interest. Francis et al. (2005) use overall interest cost to proxy the costs of debt and document that it is lower among US firms with lower information risk, as measured by accrual quality. Bharath et al. (2008) show that US firms with lower accounting quality face more stringent price terms, in both public and private debt contracts. They also show that firms with lower accounting quality tend to prefer private debt and argue that this is because banks have superior information access and processing abilities that reduce their adverse selection costs (as borrowers). The use of accounting-based metrics in debt covenants also causes lenders to demand timely loss recognition (e.g., Watts, 2003a,b). Debt contracting efficiency is enhanced by timely loss recognition because it ensures the early transfer of decision rights to lenders when firms are in distress (e.g., Ball & Shivakumar, 2005; Ball et al., 2003). This facilitates early intervention by lenders to prevent managerial opportunism (e.g. sub-optimal investments, acquisitions, and asset sales) and the agency costs of debt (e.g. dividend payouts, stock repurchases, and further issuance of debt). Ball et al. (2008) show, through tests on country-level observations, that timely loss recognition is significantly associated with debt markets. Nikolaev (2010) shows that the reliance on covenants in public debt contracts is greater when the accounting regime has a higher degree of timely loss recognition. Ahmed et al. (2002) observe that US firms with greater amounts of conflict between lenders and shareholders adopt more conservative accounting practices and that firms with such practices are associated with lower costs of debt, proxied by senior debt ratings. Zhang (2008) shows that conservative borrowers are more likely to violate debt covenants and that lenders offer lower interest rates to such borrowers. Li (2009) also analyzes country-level observations and confirms lower costs of debt among countries with more conservative financial reporting systems. Therefore, the literature clearly establishes an inverse relationship between firms' timeliness of loss recognition and their costs of debt capital. 2.2. Impact of IFRS adoption Prior to the mandatory adoption of IFRS across a large group of countries (e.g. in Europe from 2005 onward), empirical evidence on the impact of IFRS adoption are acquired largely from voluntary adoption setting. Leuz and Verrecchia (2000) and Leuz (2003) find evidence of a reduction in information asymmetry, as proxied by bid-ask spreads, trading volume, and share price volatility, among German voluntary adopters. Barth, Landsman, and Lang (2008) document less earnings management, more timely loss recognition, and a greater value relevance of accounting numbers, among voluntary IFRS adopters, in twenty-one countries. However, the main drawback of examining a voluntary adoption setting is the difficulty of differentiating the effect of the accounting standards per se from the firms' individual financial reporting incentives. Institutional environments can affect earnings quality (Cahan, Emanuel, & Sun, 2009). For instance, Ball et al. (2003) show that firms in East Asian countries, where accounting standards of common law origin are adopted, have disclosure quality that is not necessarily better than that of firms in code law countries. This suggests

that the East Asian countries have institutional environments that give preparers less incentives to issue high quality financial reports. This implies that standards per se do not necessarily determine accounting disclosure quality and that financial reporting incentives also play an essential role. Christensen et al. (2008) confirm this argument by showing that improvements in accounting disclosure quality after IFRS only occur among voluntary adopters and not mandatory adopters, in Germany. Daske, Hail, Leuz, and Verdi (2013) classify IFRS adopters into “serious” and “label” groups, based on changes in the number of pages of their annual reports before and after the adoption year, to account for the degree of compliance incentive among an international sample of voluntary adopters. They find a reduction in the cost of equity capital for the “serious” but not the “label” adopters and interpret this as evidence that financial reporting incentives determine economic benefits. Due to the drawbacks involved in analyzing voluntary adoption, more recent studies on the impact of IFRS have tended to focus on the mandatory adoption setting, made widely available by the switch to IFRS in Europe, since 2005. The growing empirical literature on the mandatory adoption of IFRS has focused substantially on its economic consequences for the capital market. Zeff (1978) defines economic consequences as “the impact of accounting reports on the decision-making behavior of business, government, unions, investors and creditors”. Daske et al. (2008) examine the effect of IFRS on equity market liquidity, which is proxied by the effect on the stock price impact and the bid-ask spread. Li (2009) focuses on the impact of IFRS on the implied cost of equity capital, derived from analyst-forecast earnings and stock prices. Beuselinck et al. (2009) evaluate whether IFRS influences share price synchronicity, which is the portion of stock return variability attributed to market-wide systematic factors. Using the switch from UK GAAP to IFRS, Iatridis (2010) finds that the implementation of IFRS reduces the scope for earnings management and leads to more value relevant accounting measures. Jiao, Koning, Mertens, and Roosenboom (2012) find an increased forecast accuracy and agreement in the European Union after the switch to IFRS in the first year. Landsman et al. (2012) assess the impact of IFRS on stock return volatility around earnings announcements. Florou and Kosi (2013) examine whether IFRS influences the costs of debt, which they measure using bond issues and yield spreads. Wu and Zhang (2009) test whether credit rating correlates with accounting information to a greater extent after IFRS. Byard et al. (2011) focus on the IFRS impact on analyst forecasts, while Tan et al. (2011) evaluate whether IFRS increases following by cross-border analysts. DeFond et al. (2011) and Florou and Pope (2012) each investigate the effect of IFRS on institutional ownership. What all these empirical studies based on mandatory IFRS adoption settings have in common is the general finding of positive economic consequences, although this is mainly concentrated in countries with better legal enforcement and/or outsider economies. However, economic consequences in the capital markets are an indirect effect of the change in accounting standards. Therefore, these empirical studies are not drawing their inferences about the impact of mandatory IFRS adoption directly from its direct effects on accounting disclosure quality. Unlike studies of economic consequences, there is relatively less attention in the mandatory IFRS adoption literature paid to the evaluation of changes in accounting disclosure quality. For those examining the impact on disclosure quality, the studies have so far revealed mixed results from mandatory IFRS adoption. For example, using the accounting quality measures of Barth et al. (2008), Christensen et al. (2008) show that mandatory IFRS adopters in Germany did not improve the quality of their financial reporting as their voluntary adopter counterparts had done. In a study of firms in Australia, France, and the UK, Jeanjean and Stolowy (2008) document that there is no decline in earnings management following mandatory IFRS. In fact, they find increased earnings management among French firms. However, using the data in Greece, Dimitropoulos et al. (2013) find that accounting information during the post-IFRS period

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82

demonstrates less earnings management and higher value relevance, when compared to the relative amounts reported during the pre-IFRS period. Clarkson, Hanna, Richardson, and Thompson (2011) examined the impact of the adoption of IFRS in Europe and Australia on the value relevance of earnings and book value. They introduced a cross product term, equal to the product of earnings per share and book value per share, into the traditional linear pricing models. With a sample of 3488 firms adopting IFRS in 2005, they find that the product term is more significant when using IFRS accounting numbers compared to using local GAAP accounting. As for the evidence of timely loss recognition, both Ahmed et al. (2013) and Chen et al. (2010) use international samples and reveal a lower timeliness of loss recognition among mandatory IFRS adopters. In contrast, Dimitropoulos et al. (2013) argue that the adoption of IFRS in Greece is associated with more timely loss recognition. Their study is based on 101 companies listed in Athens Stock Exchange, where 25 were early voluntary adopters before 2005. In their Table 5, they first show that voluntary IFRS adopters exhibited significant timely loss recognition than the other firms following the Greek GAAP in the pre-IFRS period (2001–2004). They then find that the coefficient on timely loss recognition in the post-IFRS period (2005–2008) is much higher than the coefficient in the pre-IFRS period. Despite this, Karampinis and Heves (2011) also use Greece as the setting, but they find weak evidence that accounting income incorporates bad news in a more asymmetric fashion in the post-IFRS period (2005–2007) than in the pre-IFRS period (2002–2004). The authors attribute this to weak shareholders' protection, low litigation risk and poor monitoring mechanism, and interpret their results to suggest that simultaneous infrastructure changes are required in order to make any material improvements in financial reporting. To sum up, the evidence from the studies of accounting disclosure quality still provides a mixed picture of the impact of mandatory IFRS adoption. 2.3. Hypothesis development We argue that one way in which managers can mitigate the agency conflicts with debtholders in the post-IFRS period is to bind themselves to a higher level of loss recognition timeliness as a signal for not expropriating debtholders' wealth. Our argument is based on three premises. First, mandatory IFRS adoption leads to greater use of financial reports from investors because information of financial reporting becomes more relevant to investors in the post-IFRS period (Ball, 2006). IFRS is a set of principles-based accounting standards, which can better reflect the underlying economic substance than rules-based accounting standards because the latter emphasize on legal forms (Ball, 2006; Nelson, 2003; Nobes, 2005) and were applied by many countries prior to IFRS. For instance, existing studies show that more firm-specific information are available in the market following both mandatory (Beuselinck et al., 2009) and voluntary (Loureiro & Taboada, 2014) adoption of IFRS. Furthermore, IFRS enables greater accounting comparability (Ball, 2006), which allows investors to benchmark firms' performance not only against domestic peers, but also against foreign competitors. Existing studies provide empirical evidence consistent with capital market benefits attributable to improved accounting comparability (Brochet, Jagolinzer, & Riedl, 2013; Wu & Zhang, 2010). Thus, IFRS provides firms with more opportunities to entice investors in the capital market than under previous domestic accounting standards. Second, the increased use of financial reports by debtholders after mandatory adoption of IFRS can lead firms to recognize losses more timely. Prior studies (e.g., Ball & Shivakumar, 2005; Ball et al., 2000; Ball et al., 2003) argue that only when investors have demands for financial reporting can we observe an increase in loss recognition timeliness. For example Ball et al. (2003) find that loss recognition timeliness is much more pronounced in countries where public financial disclosure is a more likely solution for the information asymmetry between

73

investors and companies. Timely loss recognition is especially important to lenders if their contractual rights are hinged entirely on financial reporting information (Ball et al., 2008). Thus, if adopting IFRS increases the demand for higher quality in financial statements from the investors, companies are more likely to recognize economic losses in a more timely fashion, especially when the agency cost between managers and investors is high. Third, we argue that managers in firms with higher costs of debt perceive more benefits associated with loss recognition timeliness than firms with lower costs of debt, and are more likely to make this commitment by reflecting economic losses in a more timely fashion. One main reason is that the increased timely loss recognition can reduce cost of debts (e.g., Ahmed et al., 2002; Zhang, 2008). Timely recognition of loss benefits lenders by increasing the timeliness of poorly-performed firms breaching debt covenants. Between higher and lower costs of debt firms, we expect the former group of firms to have greater incentives to take actions to reduce agency costs of debt and attract more debt investors. IFRS offers higher costs of debt firms with a better chance to attract more debt investors not only domestically but also from abroad. This accounting regime change may be perceived as a useful “game changer” for higher costs of debt firms since it provides increased relevance of accounting disclosures to the capital markets and increased cross-border comparability, which are both relatively less available under previous domestic accounting standards. To the extent that timely loss recognition can reduce cost of debts (Nikolaev, 2010; Zhang, 2008), we expect firms with higher costs of debt to have more incentives to apply such accounting practices, which would in turn give them greater potential to attract debt investors and to eventually reduce their costs of debt. Taking together the three reasons above,5 we formulate the first testable hypothesis of our study as follows: H1. Firms with higher costs of debt are more likely to increase their timely loss recognition following mandatory adoption of IFRS. Prior studies argue that the country's institutional environments can affect the outcome of adopting IFRS (Ball et al., 2003; Li, 2010). Countries where companies tend to obtain external capital from public market, on average, have higher investor protection and more credible accounting information than from private market (Lopez-de-Silanes, La Porta, Shleifer, & Vishny, 1998). Since public lenders are more dependent on financial reporting than private lenders (e.g., Bharath et al., 2008), firms in countries where public debt is more prevalent are more likely to cater to the demands of their investors for timely loss recognition. The demand for timely loss recognition is lower in countries with more prevalent private debt market as information asymmetry between companies and creditors is more likely to be resolved through insider communications with management (Ball et al., 2000). Moreover, Ball et al. (2008) show that timely loss recognition is more pronounced in countries with a higher proportion of public debt markets. As public lenders have higher information costs and are more dependent on financial reporting information than private lenders (e.g., Bharath et al., 2008; Diamond, 1991; Fama, 1985), we expect that evidence of firms with high costs of debt recognizing economic losses in a more timely manner after mandatory IFRS adoption exists mainly in

5 Our prediction is based on the three reasons altogether. However, our conjecture may not be valid if each reason is considered separately. For example, if timely loss recognition can help reduce costs of debts, one may argue that high quality firms may easily separate from low quality firms via voluntary IFRS adoption before 2005. However, the possibility of this might be low. First, in the pre-IFRS period, not all countries permit early adoption. Second, for companies domiciled in countries which allow early adoption may not recognize economic losses more timely because their debt markets do not rely much on financial reporting in assessing debtholders' credit quality in the pre-IFRS period (our first reason). Prior research argues that when the debt market does not pay attention to accounting information, firms have no incentives to recognize losses timely (our second reason) even if they have contracting demands. Thus, we form our hypothesis based on three rationales.

74

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82

countries with more prevalent public debt markets. We form our second testable hypothesis as follows: H2. Evidence in support of hypothesis H1 will be more pronounced among countries with more prevalent public debt markets. There are two possible critiques of our prediction. First, firms seeking to reduce their costs of debt could increase their timely loss recognition prior to mandatory IFRS adoption. However, we argue that increasing loss recognition timeliness could be costly to firms and they may choose not to do so without the additional benefit of attracting foreign lenders that occurs under IFRS. Domestic standards in many European countries are driven by tax-book conformity and would be unable to offer the benefits of attracting foreign investors to compensate for this cost.6 In other words, they wait for mandatory IFRS adoption to increase timely loss recognition so that some of this cost can be offset by the perceived benefits of international accounting comparability. Second, firms seeking the benefits of increased international accounting comparability could have adopted IFRS on a voluntary basis prior to 2005. However, there are many reasons why firms may not seek voluntary adoption. To begin with, not all countries permit early adoption. Even if early adoption is permitted, the switch to IFRS may affect existing debt contracts based on domestic accounting standards (e.g., Ormrod & Taylor, 2004). Firms that are already paying higher costs of debt may be especially sensitive to this since they are likely to be in distress and close to breaching covenants (e.g., Christensen, Lee, & Walker, 2009). Firms may also perceive that there will be a greater benefit from international accounting comparability when more countries simultaneously replace their domestic standards with IFRS, such as was the case in Europe in 2005.

where NI is net income scaled by the share price at the beginning of the fiscal year. R is the annual return of firm i over the twelve-month period from the fifth month of the fiscal year t to the fourth month of the fiscal year t + 1 and DR is a dummy variable taking the value one when R is negative, and zero otherwise; SIZE is the natural logarithm of the market value of total assets; MB is the market-to-book ratio of shareholders' equity; and LEV is leverage measured as total liabilities to total assets. For each country, we derive the yearly parameters for λ0, λ1, λ2 and λ3 from Eq. (1), and calculate the average weights. The average weights are subsequently used to estimate annual firm-specific CSCORE using the following Eq. (2). CSCOREit ¼ λ0 þ λ1 SIZEit þ λ2 MBit þ λ3 LEV it

ð2Þ

When estimating the model, we delete firm years with missing data and a negative book value of equity. We also exclude observations falling in the top or bottom 1% of earnings, returns, size, market-to-book, and leverage in each year. Khan and Watts (2009) argue that firms with high C-scores reflect losses more timely than gains in their financial reporting.8 3.2. Hypotheses tests To test our hypothesis H1, we conduct a difference-in-differences analysis based on the following regression: CScoreit ¼ δ0 þ δ1 Post it−1 þ δ2 CODit−1 þ δ3 MAN i þ δ4 ðPost it−1  CODit−1 Þ þ δ5 ðPost it−1  MANi Þ þ δ6 ðCODit−1  MANi Þ þ δ7 ðPost it−1  CODit−1  MAN i Þ þ δ8 IntCovit−1 þ δ9 ROAit−1 þ δ10 σ ðEarnÞit−1 þ δ11 Lossit−1 þ δ12 AQ it−1 þ δ13 Betait−1

3. Methodology and sample

þ δ14 Growthit−1 þ δ15 Interest it−1 þ τit þ λit þ εit

3.1. Measure of timely loss recognition: C SCORE Following Khan and Watts (2009), we use C-score as a firm-year measure of conservatism in our main analyses. Khan and Watts (2009) employ a parsimonious method to estimate a firm-yearspecific conservatism measure for the Basu (1997)-typed earnings conservatism, whereby earnings are regressed on contemporaneous stock returns, with a dummy variable and an associated interaction term included in the regression to capture the timely recognition of economic losses.7 As timely loss recognition based on Basu (1997) cannot measure the level of firm-year conservatism and can limit the nature of the hypothesis testing that can be conducted, Khan and Watts (2009) draw on Basu's (1997) measure of asymmetric timeliness of earnings and construct the C-score — a firm-year-specific measure of conservatism (Lai & Taylor, 2008). Specifically, we first estimate Eq. (1) using yearly cross-sectional regressions across available firms for each country.

NIit ¼ α 0 þ α 1 DRit þ α 2 Rit ðμ 0 þ μ 1 SIZEit þ μ 2 MBit þ μ 3 LEV it Þ þ α 3 DRit  Rit ðλ0 þ λ1 SIZEit þ λ2 MBit þ λ3 LEV it Þ þ δ0 SIZEit þ δ1 MBit þ δ2 LEV iv þ δ3 DRit  SIZEiv þ δ4 DRit  MBit þ δ5 DRit  LEV it Þ þ ε i

ð1Þ

6 Prior studies (e.g., Naranjo, Saavedra, & Verdi, 2014) argue that IFRS adoption is less likely to systematically affect tax benefits and rates. Thus, our results are not affected by the impact of adopting IFRS on these issues. 7 The earnings–return regression of Basu (1997): NIit = α0 + α1DRit + α2Rit + α3RitDRit + εit, where NI is net income scaled by the share price at the beginning of the fiscal year. R is the annual return of firm i over the twelve-month period from the fifth month of the fiscal year t to the fourth month of the fiscal year t + 1 and DR is a dummy variable taking the value one when R is negative, and zero otherwise. According to Basu (1997), a positive α3 demonstrates timely loss recognition.

ð3Þ

where (for each company i in year t): CScore is the firm- and yearspecific measure of timely loss recognition, estimated following Khan and Watts (2009), which incorporates the effects of firm size, marketto-book, and leverage; Post indicates mandatory IFRS adoption period, which equals 1 for the years 2005 to 2007, and 0 otherwise; COD proxies costs of debt, which is measured as interest expense divided by interestbearing debt; MAN is an indicator equal to 1 if a firm does not adopt IFRS until 2005, and 0 otherwise; IntCov proxies interest coverage and is measured as the scaled decile rank of operating income divided by interest expense; ROA captures profitability and is measured as the scaled decile rank of the return on assets; and σ(Earn) proxies earnings volatility and is measured as the scaled decile rank of the standard deviation of net income before extraordinary items over the past five years. Loss equals 1 if a firm incurs losses for two consecutive years and 0 otherwise; AQ captures accrual quality and is the scaled decile rank of the absolute value of performance-matched abnormal accruals, following Kothari et al. (2005) and McNichols (2000); Beta is the firm's CAPM beta, estimated from the past 60 months (minimum 18 months) of rolling regression; Growth captures growth opportunity and is calculated as the percentage change in sales from last year; Interest is a country's interest rate; τ is a country fixed effect; and λ is the industry fixed effect. Control variables such as interest coverage, profitability, earnings volatility, accrual quality and beta follow Francis et al. (2005). In Eq. (3), the main effect we test is indicated by the coefficient δ7. Coefficient δ7 estimates the incremental timely loss recognition for the mandatory adopters relative to early IFRS adopters as firms increase one unit of cost of debt in the post-IFRS period. To support our hypothesis H1, we need to observe a significantly positive value of δ7 for firms that adopted IFRS on a mandatory basis for the first time in 2005. 8 In additional tests, we also assess the construct validity of C-score as a metric of timely loss recognition for Basu (1997) and use alternative measures of conservatism to check the robustness of our results in Section 4.3.

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82

To test our hypothesis H2, we examine the mandatory adopter sample and employ a difference-in-differences research design by including two country-level indicators of the prevalence of public debt markets, and their interaction with COD and Post. Following Beck, DemirgucKunt, and Levine (2000) and Bushman and Piotroski (2006), the first indicator is the ratio of the country's private bond market capitalization to its market capitalization of equity (PRIVDEBT).9 A higher value of PRIVDEBT indicates a greater use of private debt than public debt. The second indicator is calculated as the ratio of the country's deposit money bank assets to its market capitalization of equity (BANKMKT). A higher value of BANKMKT indicates a greater degree of bank-based financing, and therefore less prevalent public debt. Both indicators are obtained from the Bank Structure and Economic Development database (World Bank).10 Specifically our tests of hypothesis H2 are based on the following regressions: CScoreit ¼ δ0 þ δ1 Post it−1 þ δ2 CODit−1 þ δ3 HLDEBT i þ δ4 ðPost it−1  CODit−1 Þ þ δ5 ðPost it−1  HLDEBT i Þ þ δ6 ðCODit−1  HLDEBT i Þ þ δ7 ðPost it−1  CODit−1  HLDEBT i Þ þ δ8 IntCovit−1 þ δ9 ROAit−1 ð4aÞ þ δ10 σ ðEarnÞit−1 þ δ11 Lossit−1 þ δ12 AQ it−1 þ δ13 Betait−1 þ δ14 Growthit−1 þ δ15 Interest it−1 þ τ it þ λit þ εit CScoreit ¼ δ0 þ δ1 Post it−1 þ δ2 CODit−1 þ δ3 HLBANK i þ δ4 ðPost it−1  CODit−1 Þ þ δ5 ðPost it−1  HLBANK i Þ þ δ6 ðCODit−1  HLBANK i Þ þ δ7 ðPost it−1  CODit−1  HLBANK i Þ þ δ8 IntCovit−1 þ δ9 ROAit−1 þ δ10 σ ðEarnÞit−1 þ δ11 Lossit−1 þ δ12 AQ it−1 þ δ13 Betait−1 þ δ14 Growthit−1 þ δ15 Interest it−1 þ τ it þ λit þ εit

ð4bÞ

where HLDEBT is equal to 1 (0) when PRIVDEBT is above (below) the yearly cross-sectional median; HLBANK is equal to 1 (0) when BANKMKT is above (below) the yearly cross-sectional median. Coefficient δ4 estimates the incremental timely loss recognition of firms with high costs of debt in countries with low PRIVDEBT from 2005 onward; Coefficient δ7 estimates the incremental timely loss recognition for firms with high costs of debt in countries with high PRIVDEBT (BANKMKT) relative to countries with low PRIVDEBT (BANKMKT) from 2005 onward. To support our hypothesis H2, we must observe a significantly negative value of δ7. As the results can be confounded by other factors such as changes in macroeconomic conditions or changes in accounting standards over time, we use early voluntary adopters as the control group, which comprises firms that adopted IFRS voluntarily prior to 2005. If we find this effect in both the mandatory adopters and voluntary adopters, then we cannot infer that it is attributable to mandatory IFRS adoption, since it is more likely to be driven by unidentified confounding events or a background time trend. To support H1, we need to observe a significantly positive value of δ3 in Eq. (3) for mandatory IFRS adopters, but not for early adopters. Similarly, to support H2, we need to observe a significantly negative value of δ7 in Eqs. (4a) and (4b) for mandatory IFRS adopters, but not for early adopters. 3.3. Sample Our sample is obtained from the Compustat Global Vantage database. We exclude financial industries, cross-listers/ADRs and UK-AIM 9 Beck et al. (2000) introduce these indicators as part of a new dataset providing statistics on the size, activity, and efficiency of financial intermediaries across countries and through time. They argue that this new dataset improves researchers' assessments of the development, structure, and performance of the financial structure across countries, since it draws on a wider source of information than the IMF's International Financial Statistics and the IFC's Emerging Market Database. 10 An alternative to country-level variables, such as PRIVDEBT and BANKMKT, would be firm-level proxies of dependence on private vs public debt and bank loans. However, the data to construct such proxies for European firms are not widely available and would therefore reduce the sample size of our analyses substantially, which would in turn reduce the ability of our findings to be generalized across a larger set of firms.

75

Table 1 Sample size. Country

Total

2002

2003

2004

2005

2006

2007

UK France Germany Sweden Switzerland Netherlands Italy Norway Finland Spain Greece Denmark Belgium Austria Ireland Total

3222 1962 1871 814 749 466 721 383 478 27 208 387 273 215 84 11,860

606 307 312 125 131 88 79 62 80 1 4 74 41 37 14 1961

583 355 335 140 134 90 116 65 80 2 19 72 46 39 15 2091

562 356 313 151 128 82 128 69 82 3 41 67 47 34 15 2078

515 331 310 133 119 78 126 62 78 3 41 62 45 33 12 1948

512 326 310 132 126 73 137 66 81 9 52 66 53 36 16 1995

444 287 291 133 111 55 135 59 77 9 51 46 41 36 12 1787

Notes: This table presents the number of firms in each year in each of the sixteen European countries sampled, over the six-year sample period (2002–2007). It also shows the total number of firm-year observations for each individual country and the overall sample. The sample covers both the first time mandatory adopters (treatment group) and the early voluntary adopters (control group).

firms to reduce sample noise. In all the countries we sampled, the IFRS reporting is required for fiscal years ending on or after December 31, 2005. Our initial sample consists of 23,225 firm-year observations from European countries including Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, the Netherlands, Norway, Sweden, Switzerland, and the UK over the eight-year period from 2002 to 2007. We do not extend our sample across 2008–2009 to avoid the global financial crisis during the period. The financial statement variables, including accounting standards are collected from Compustat Global. By reference to the variable of accounting standards, we categorize companies as early IFRS adopters if their switch from local standards to IFRS before 2005 and as mandatory adopters if they switch from local standards to IFRS in 2005. To estimate beta, we obtain market data from Compustat Global Issue as beta is measured relative to the country market index. We require at least 18 monthly observations to calculate market beta. As for country institutional factors, we collect the data from the Bank Structure and Economic Development database (World Bank). We require all the necessary data to compute the variables used in the firm-period regressions described above, and delete the outliers of all variables at the top and bottom 1% of their distributions. The final sample comprises 11,860 firm-year observations, with 1668 firm-year observations for early adopters. Table 1 presents our sample. The UK accounts for over a quarter of the total sample, while France and Germany together account for one third. Since the mandatory adoption of IFRS commenced in the fiscal year 2005, our sample covers a three-year pre-adoption period (2002 to 2004) and a three-year post-adoption period (2005 to 2007). The sample includes both first time mandatory adopters (the treatment group) and early voluntary adopters (the control group).11 Table 2 reports country-specific PRIVDEBT, BANKMKT, and CScore. The five countries with the lowest prevalence of private bond markets are Finland, Greece, Ireland, Switzerland, and the UK. Finland, France, Sweden, Switzerland, and the UK are the five countries with the lowest dominance by bank-based financing. The correlation between PRIVDEBT and BANKMKT is 0.75. This high correlation between these two proxies of the pervasiveness of public debt implies that their test results can serve as mutual robustness checks. The five countries with the highest average C-scores are Austria, Germany, Greece, Sweden, and the UK. France, Ireland, Italy, Norway, and Spain are the five countries with 11 We exclude cross-listers/ADRs and UK-AIM firms, in order to reduce sample noise. We obtain results similar to our main analyses in untabulated sensitivity tests where we include these firms.

76

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82

Table 2 Country-level variables and C-scores. Country

PRIVDEBT

BANKMKT

Average CScore

UK France Germany Sweden Switzerland Netherlands Italy Norway Finland Spain Greece Denmark Belgium Austria Ireland

0.12 0.49 0.87 0.39 0.14 0.57 0.93 0.47 0.19 0.38 0.04 2.05 0.55 1.69 0.34

1.03 1.27 2.86 0.94 0.68 1.55 2.1 1.71 0.56 1.63 1.59 2.62 1.6 5.72 2.19

0.60 0.37 0.54 0.46 0.44 0.41 0.4 0.34 0.42 0.06 0.47 0.43 0.41 0.48 0.35

Table 3 presents descriptive statistics for the firm-specific variables used in our analysis. On average, European firms in our sample have a C-score of 0.482 (median of 0.355). The C-score is positively correlated with σ(Earn) and AQ, which indicates a positive relationship between timely loss recognition and information asymmetry and is therefore broadly consistent with LaFond and Watts (2008). The C-score is positively correlated with Loss and negatively correlated with IntCov and ROA, which is consistent with lenders demanding less timely loss recognition from well performing firms. Finally, the C-score is negatively related with both Beta and Growth, which implies that riskier firms are less timely in their loss recognition. 4. Empirical findings 4.1. Test of hypothesis H1

Notes: This table presents the mean C-score and institutional characteristics of our sixteen sampled European countries. CScore is the firm-year measure of conditional conservatism, based on Khan and Watts (2009); based on the Bank Structure and Economic Development database (World Bank), PRIVDEBT is the ratio of the country's private bond market capitalization to the country's market capitalization of equity securities and BANKMKT is the ratio of the country's deposit money bank assets to the country's market capitalization of equity securities. As PRIVDEBT and BANKMKT have country-year values, we report the time-series mean values for each country across 2002–2007.

the lowest average C-scores. At the country level, the C-score has correlations of 0.09 and 0.11 with PRIVDEBT and BANKMKT, respectively. These low correlations ensure that our sub-samples, partitioned by PRIVDEBT and BANKMKT to test hypothesis H2, will consist of firms with broadly similar average C-scores.

Table 4 presents the results of the regression analysis using a difference-in-differences design, which are based on robust standard errors clustered by firm and year (Petersen, 2009). Column (1) reports the results for the basic model that does not control for company characteristics and global interest rates, Column (2) reports the results controlling for company characteristics and global interest rates but do not control for the country and industry effects, and Column (3) reports the results controlling for company characteristics, global interest rates, and both the country and industry effects. Focusing on Column (3), we first find that the coefficient of POST is consistently significantly negative (−0.128, t-statistic = −6.05) and the coefficient POSTxMAN is insignificant. These findings suggest a decline in timely loss recognition for both mandatory and voluntary IFRS adopters after 2005 among firms that do

Table 3 Firm-level variables. Panel A: Summary statistics Variables

Mean

Standard deviation

25%

50%

75%

CScore COD IntCov ROA σ(Earn) LOSS AQ BETA GROWTH

0.482 0.094 0.025 11.223 1.408 0.242 0.061 0.750 1.398

4.226 0.127 0.111 51.311 17.398 0.428 0.060 0.603 19.534

0.169 0.047 0.001 1.315 0.019 0.000 0.019 0.307 0.965

0.335 0.062 0.037 4.294 0.040 0.000 0.042 0.658 1.053

0.495 0.089 0.073 9.906 0.100 0.000 0.081 1.090 1.158

Panel B: Correlation analyses Variable

CScore

COD

IntCov

ROA

σ(Earn)

LOSS

AQ

BETA

GROWTH

CScore

1.000

−0.011 (0.248) 1.000

−0.321 (0.000) −0.087 (0.000) 1.000

−0.286 (0.000) −0.138 (0.000) 0.791 (0.000) 1.000

0.102 (0.000) 0.182 (0.000) −0.214 (0.000) −0.211 (0.000) 1.000

0.175 (0.000) 0.114 (0.000) −0.737 (0.000) −0.642 (0.000) 0.356 (0.000) 1.000

0.059 (0.000) 0.096 (0.000) −0.153 (0.000) −0.166 (0.000) 0.239 (0.000) 0.247 (0.000) 1.000

−0.167 (0.000) 0.033 (0.000) −0.038 (0.000) −0.026 (0.004) 0.200 (0.000) 0.086 (0.000) 0.087 (0.000) 1.000

−0.230 (0.000) −0.105 (0.000) 0.362 (0.000) 0.292 (0.000) −0.165 (0.000) −0.264 (0.000) −0.056 (0.000) 0.055 (0.000) 1.000

COD IntCov ROA σ(Earn) LOSS AQ BETA GROWTH

Notes: This table reports summary statistics (Panel A) and correlations (Panel B). The sample covers both first time mandatory adopters and voluntary adopters (the control group) in sixteen European countries over the period from 2002 to 2007. CScore is the firm-year measure of conditional conservatism, based on Khan and Watts (2009); COD is interest expense divided by interest bearing debt; IntCov is the ratio of operating income to interest expense; ROA is the return on assets; σ(Earn) is the standard deviation of net income before extraordinary items over the past five years; LOSS is an indicator equal to 1 if a firm incurs losses for two consecutive years, and 0 otherwise; AQ is the absolute value of performance-matched abnormal accruals, following Kothari et al. (2005) and McNichols (2000); BETA is a firm's beta; and GROWTH is the growth rate in sales. Panel B includes Pearson correlations, with p-values reported in parentheses.

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82 Table 4 Timely loss recognition after mandatory IFRS adoption, conditional on costs of debt. Regression analyses

Intercept POST COD MAN POST × COD POST × MAN COD × MAN POST × COD × MAN IntCov ROA σ(Earn) LOSS AQ BETA GROWTH INTEREST Country effect Industry effect Obs Adj R2

(1) Regression 1

(2) Regression 2

(3) Regression 3

Coeff

Coeff

Coeff

t-Stat b

0.340 −0.153c −0.036 0.001 −0.157 0.047 0.094 0.261c

No No 11,860 0.112

(2.53) (−1.81) (−0.18) (0.00) (−1.62) (0.23) (0.43) (1.82)

t-Stat a

t-Stat a

0.648 −0.141c −0.033 −0.018 −0.177c 0.049 0.098 0.269c

(3.49) (−1.75) (−0.17) (−0.12) (−1.67) (0.24) (0.44) (1.85)

0.534 −0.128a 0.013 0.015 −0.147 0.040 0.048 0.241c

(12.69) (−6.05) (0.55) (0.55) (−1.06) (1.45) (1.38) (1.89)

−0.187 0.164 −0.132 0.064 −0.280 −0.376a 0.000 −0.008 No No 11,860 0.137

(−1.18) (0.99) (−1.20) (0.74) (−0.52) (−3.57) (0.03) (−0.62)

−0.118 0.144 −0.077 0.090b −0.229 −0.357a −0.000 −0.005 Yes Yes 11,860 0.225

(−1.16) (1.01) (−1.48) (2.22) (−0.95) (−2.63) (−0.24) (−0.64)

Notes: This table presents the pooled regression analyses of the full sample. POST is a dummy variable equal to 1 for years after mandatory IFRS adoption (i.e. 2005, 2006, and 2007) and 0 otherwise; COD is interest expense divided by interest-bearing debt; MANt is an indicator equal to 1 if a firm does not adopt IFRS until 2005, and 0 otherwise; IntCov is the scaled decile rank of the ratio of operating income to interest expense; ROA is the scaled decile rank of the return on assets; σ(Earn) is the scaled decile rank of the standard deviation net income before extraordinary items over the past five years; LOSS is an indicator equal to 1 if a firm incurs losses for two consecutive years and 0 otherwise; AQ is the scaled decile rank of the absolute value of performance-matched abnormal accruals, following Kothari et al. (2005) and McNichols (2000); BETA is the firm's beta; GROWTH is the growth rate in sales; and INTEREST is a country's interest rate. t-Statistics are based on standard errors clustered by firm and year. Superscripts a, b, and c indicate significance at the 0.01, 0.05, and 0.10 levels respectively for two-tailed t-tests.

not have debt. The fact that this is observed in both the treatment and the control sample suggests that it cannot be attributed to the mandatory adoption of IFRS from 2005 onward. Instead, it is attributable to unidentified confounding events or a background time trend that influences all firms irrespective of whether they adopt IFRS on a mandatory or a voluntary basis. This suggests that the reduction in timely loss recognition observed in existing studies (e.g., Ahmed et al., 2013; Chen et al., 2010) may not necessarily be an IFRS effect. We also find that the coefficient of POST × COD is insignificant (−0.147, t-statistic = −1.06). The results indicate that voluntary IFRS adopters who switched to IFRS before year 2005 are not associated with incrementally higher loss recognition timeliness in the post-IFRS period, even if they have a higher cost of debt. However, the coefficient on the three interaction term POST × COD × MAN is significantly positive (0.241, t-statistic = 1.89). Across the three columns, the coefficients of POST × COD × MAN are consistently significantly positive. This suggests that, relative to voluntary IFRS adopters, mandatory IFRS adopters exhibit relatively larger increases in loss recognition timeliness from year 2005 onwards when their costs of debt are high. The results are in support of our prediction in H1. In terms of economic significance, for a one-standard-deviation increase in COD, the change in C-score in the post-IFRS period is 3.06% higher for the mandatory adopters than the early adopters.12 These results are robust to controls for firm characteristics (i.e. solvency, profitability, earnings volatility, loss-making tendency, earnings quality, risk, and growth prospects) as well as both country and industry effects.

12 This is calculated as 0.127 (a one standard deviation of COD) ∗ 0.241 (the coefficient on POST × COD × MAN) = 0.0306.

77

In summary, the findings in Table 4 suggest that firms paying a higher cost of debt increase the timeliness of their loss recognition after mandatory IFRS adoption.13 Among firms we assumed would benefit more from IFRS, our results confirm the prediction of Ball (2006) that IFRS will increase timely loss recognition. 4.2. Testing hypothesis H2 Table 5 presents the results of hypothesis H2. To statistically compare the timely loss recognition for firms with high costs of debts in countries with high PRIVDEBT than in countries with low PRIVDEBT, we use a difference-in-differences research design. Column (1) reports the results for mandatory adopters and Column (2) reports the results for early adopters. The results show that the coefficient on the interactive term POST × COD × HLDEBT is significantly negative (− 0.560, t-statistic = − 2.07) for mandatory adopters, but not for voluntary adopters (0.724, t-statistic = 0.78). The Wald test (p-value b 0.01) also indicates that the coefficients of POST × COD × HLDEBT are significantly different between mandatory and voluntary adopters. These results suggest that, compared to countries with high PRIVDEBT, mandatory adopters whose cost of debt are higher recognize losses more timely during 2005–2009 in countries with low PRIVDEBT. In terms of economic significance, for a one-standard-deviation increase in COD, the incremental C-score for firms in countries with high PRIVDEBT is 7.11% lower than those in countries with low PRIVDEBT during the post-IFRS period. This suggests that our evidence in support of hypothesis H1 exists only among firms in countries with a greater pervasiveness of public debt market, which in turn confirms our prediction in hypothesis H2. We essentially observe that mandatory IFRS adopters with higher costs of debt increase their timely loss recognition after 2005 only in countries with more active public lenders, who are more dependent on financial reporting than private lenders are. This is consistent with such firms increasing their timely loss recognition after mandatory IFRS adoption only when their lenders are sensitive to changes in accounting disclosure quality. Turning to the voluntary IFRS adopters (i.e. our control sample), the coefficients of POST × COD × HLDEBT are statistically insignificant. This strengthens the inference that our findings in favor of hypotheses H1 and H2 are indeed attributable to mandatory IFRS adoption rather than unidentified confounding effects. Table 6 replicates the tests in Table 5, substituting PRIVDEBT with BANKMKT to identify firms from countries with a greater pervasiveness of public lending. Our variable of interest is on POST × COD × HLBANK. Column (1) shows that the coefficients of POST × COD × HLBANK are significantly negative (− 0.522, t-statistic = − 1.98) for the mandatory IFRS adopters (i.e. our treatment sample). This suggests that, for a one-standard-deviation increase in COD, the incremental C-score in countries with high BANKMKT is 6.63% lower than the C-score in countries with low BANKMKT during the post-IFRS period. Column (2) shows that the coefficients of POST × COD × HLBANK are insignificant among the voluntary adopters (i.e. our control sample). The Wald test also indicates that the coefficients of POST × COD × HLBANK are significantly different between mandatory and voluntary adopters. Thus, both columns in Table 6 yield findings broadly similar to those shown in the corresponding panels in Table 5. In other words, our evidence in support of the prediction made in hypothesis H2 is robust to alternative proxies for the prevalence of public debt markets. 4.3. Additional analyses and robustness tests 4.3.1. UK sample alone To address the concern that cross-country differences may contaminate our main results for H1, we re-examine the tests based on the UK

13 Untabulated analyses performing separate regressions for mandatory adopters and early adopters lead to similar conclusions.

78

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82

Table 5 Timely loss recognition after mandatory IFRS adoption, conditional on costs of debt and the development of private bond markets. (1) Mandatory adoption

Intercept POST COD HLDEBT POST × COD POST × HLDEBT COD × HLDEBT POST × COD × HLDEBT IntCov ROA σ(Earn) LOSS AQ BETA GROWTH INTEREST Country effect Industry effect Obs Adj R2

(2) Voluntary adoption

Coeff

t-Stat

Coeff

t-Stat

Coeff

t-Stat

Coeff

t-Stat

0.606a −0.103 0.014 0.385b −0.023 −0.005 −0.005 −0.351b −0.004 −0.044 −0.132 0.081 −0.185 −0.334a −0.000 −0.002 No No 10,192 0.13

(4.49) (−1.08) (0.15) (2.51) (−0.22) (−0.03) (−0.03) (−2.39) (−0.03) (−0.26) (−1.11) (0.96) (−0.35) (−2.97) (−0.02) (−0.11)

0.626a −0.177c −0.110 0.588a −0.024 0.071 0.141 −0.560b 0.101 −0.153 −0.179 0.040 −0.150 −0.255b −0.000 −0.002 Yes Yes 10,192 0.18

(3.30) (−1.93) (−0.94) (3.61) (−0.16) (0.42) (0.72) (−2.07) (0.43) (−0.76) (−1.13) (0.41) (−0.24) (−2.08) (−0.04) (−0.14)

0.470 −0.161 −0.007 0.042 0.020 0.039 −0.009 0.497 −0.553 0.556 −0.015b 0.023 −0.676 −0.257b −0.008b −0.001 No No 1668 0.33

(1.23) (−0.46) (−0.01) (0.06) (0.08) (0.09) (−0.01) (0.64) (−1.16) (1.10) (−2.07) (0.09) (−0.44) (−1.98) (2.25) (−0.13)

−0.290 −0.140 −0.011 −0.011 0.021 0.058 −0.326 0.728 −0.269 0.313 0.838b 0.171 0.440 0.267c −0.015 −0.001 Yes Yes 1668 0.60

(−0.42) (−0.35) (−0.02) (−0.01) (0.04) (0.11) (−0.45) (0.78) (−0.38) (0.46) (−2.80) (0.54) (0.27) (1.90) (0.44) (−0.10)

Notes: Column (1) reports the regression results for mandatory adopters who first adopted IFRS in 2005. Column (2) reports the regression results for voluntary adopters who first adopted IFRS before 2005. HLDEBT is equal to 1 (0) when PRIVDEBT is above (below) the yearly cross-sectional median; POST is a dummy variable equal to 1 for years after mandatory IFRS adoption (i.e. 2005, 2006, and 2007) and 0 otherwise; COD is interest expense divided by interest bearing debt; IntCov is the scaled decile rank of the ratio of operating income to interest expense; ROA is the scaled decile rank of the return on assets; σ(Earn) is the scaled decile rank of the standard deviation of net income before extraordinary items over the past five years; LOSS is an indicator equal to 1 if a firm incurs losses for two consecutive years, and 0 otherwise; AQ is the scaled decile rank of the absolute value of performance-matched abnormal accruals, following Kothari, Leone, and Wasley (2005) and McNichols (2000); BETA is the firm's beta; GROWTH is the growth rate in sales; and INTEREST is a country's interest rate. t-Statistics are based on standard errors clustered by firm and year. Superscripts a, b, and c indicate significance at the 0.01, 0.05, and 0.10 levels respectively for two-tailed t-tests.

sample since it accounts for nearly a quarter of all our firm-year observations. In addition, Brochet et al. (2013) argue that UK provides a unique setting to examine the effects of IFRS on the comparability of accounting information. As the domestic accounting standards of the UK are quite similar to IFRS, the information quality benefits from IFRS adoption are less likely attributable to changes in standard quality, but more attributable to changes in financial statement comparability.

Moreover, early adoption is not allowed in the UK. Table 7 reports the results for UK listed companies. The results show that the coefficients of POST × COD are significantly positively in all columns, indicating that firms with higher costs of debt have incrementally higher timely loss recognition from 2005 onward under IFRS than they did previously under the UK standards. These findings strengthen our prediction that mandatory IFRS adoption improves the information

Table 6 Timely loss recognition after mandatory IFRS adoption, conditional on costs of debt and the ratio of deposit money bank assets. (1) Mandatory adoption Coeff Intercept POST COD HLBANK POST × COD POST × HLBANK COD × HLBANK POST × COD × HLBANK IntCov ROA σ(Earn) LOSS AQ BETA GROWTH INTEREST Country effect Industry effect Obs Adj R2

t-Stat a

0.602 −0.100 0.018 0.360b −0.016 −0.013 −0.013 −0.328b −0.012 −0.033 −0.139 0.083 −0.174 −0.330a −0.000 −0.001 No No 10,192 0.13

(4.55) (−1.13) (0.16) (2.54) (−0.12) (−0.07) (−0.07) (−2.50) (−0.07) (−0.20) (−1.19) (0.98) (−0.33) (−2.94) (−0.03) (−0.07)

(2) Voluntary adoption Coeff a

0.645 −0.168c −0.095 0.539a −0.073 0.062 0.141 −0.522b 0.099 −0.151 −0.179b 0.041 −0.150 −0.257b −0.000 −0.002 Yes Yes 10,192 0.15

t-Stat

Coeff

t-Stat

Coeff

t-Stat

(3.33) (−1.77) (−0.86) (3.54) (−0.34) (0.34) (0.63) (−1.98) (0.46) (−0.75) (−2.11) (0.42) (−0.24) (−2.02) (−0.04) (−0.09)

0.471 −0.164 −0.012 0.051 0.032 0.047 −0.004 0.486 −0.532 0.545 −0.003b 0.023 −0.693 −0.257 0.002 0.001 No No 1668 0.30

(1.23) (−0.51) (−0.03) (0.02) (0.10) (0.10) (−0.01) (0.60) (−1.12) (1.07) (−2.02) (0.09) (−0.45) (−0.82) (0.02) (0.11)

−0.323 −0.133 −0.019 −0.004 0.071 0.043 −0.347 0.742 −0.252 0.298 −0.826b 0.169 0.474 0.271 0.005 0.001 Yes Yes 1668 0.33

(−0.44) (−0.36) (−0.03) (−0.01) (0.09) (0.09) (−0.50) (0.82) (−0.37) (0.44) (−1.98) (0.54) (0.23) (0.36) (0.04) (0.12)

Notes: Column (1) reports the regression results for mandatory adopters who first adopted IFRS in 2005. Column (2) reports the regression results for voluntary adopters who first adopted IFRS before 2005. HLBANK is equal to 1 (0) when BANKMKT is above (below) the yearly cross-sectional median; POST is a dummy variable equal to 1 for years after mandatory IFRS adoption (i.e. 2005, 2006, and 2007) and 0 otherwise; COD is interest expense divided by interest bearing debt; IntCov is the scaled decile rank of the ratio of operating income to interest expense; ROA is the scaled decile rank of the return on assets; σ(Earn) is the scaled decile rank of the standard deviation of net income before extraordinary items over the past five years; LOSS is an indicator equal to 1 if a firm incurs losses for two consecutive years, and 0 otherwise; AQ is the scaled decile rank of the absolute value of performance-matched abnormal accruals, following Kothari et al. (2005) and McNichols (2000); BETA is the firm's beta; GROWTH is the growth rate in sales; and INTEREST is a country's interest rate. t-Statistics are based on standard errors clustered by firm and year. Superscripts a, b, and c indicate significance at the 0.01, 0.05, and 0.10 levels respectively for two-tailed t-tests.

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82 Table 7 Results based on the UK sample. (1)

Intercept POST COD POST × COD IntCov ROA σ(Earn) LOSS AQ BETA GROWTH INTEREST Industry effect Obs Adj R2

(2)

(3)

Coeff

t-Stat

Coeff

t-Stat

Coeff

t-Stat

0.477a −0.312 −0.227 0.987a

(2.93) (−1.22) (−0.79) (2.81)

0.847b −0.281 −0.186 0.977b 0.511 −0.425 −0.448 0.084 −0.145 −0.456 0.003 −0.001 No 3205 0.14

(2.47) (−1.07) (−0.64) (2.57) (0.79) (−0.73) (−0.71) (0.28) (−0.08) (−0.75) (0.05) (−0.09)

1.043 −0.374 −0.132 0.860b 0.497 −0.407 −0.162 0.135 −0.804 −0.652 −0.002 −0.002 Yes 3205 0.32

(1.34) (−1.06) (−0.34) (2.08) (0.57) (−0.52) (−0.19) (0.33) (−0.33) (−0.79) (−0.03) (−0.10)

No 3205 0.05

Notes: This table presents the pooled regression analyses of UK mandatory adopters. POST is a dummy variable equal to 1 for years after mandatory IFRS adoption (i.e. 2005, 2006, and 2007), and 0 otherwise; COD is interest expense divided by interest-bearing debt; IntCov is the scaled decile rank of the ratio of operating income to interest expense; ROA is the scaled decile rank of the return on assets; σ(Earn) is the scaled decile rank of the standard deviation net income before extraordinary items over the past five years; LOSS is an indicator equal to one if a firm incurs losses for two consecutive years and zero otherwise; AQ is the scaled decile rank of the absolute value of performance-matched abnormal accruals, following Kothari et al. (2005) and McNichols (2000); BETA is the firm's beta; GROWTH is the growth rate in sales; and INTEREST is a country's interest rate. t-Statistics are based on standard errors clustered by firm and year. Superscripts a, b, and c indicate significance at the 0.01, 0.05, and 0.10 levels respectively for two-tailed t-tests.

environment through enhanced comparability of financial statements across borders.

4.3.2. Sample excluding UK, France, and Germany We also replicate all the tests of hypotheses H1 and H2 by removing UK, France and German firms, separately. Recall from Table 1 that the UK accounts for nearly a quarter of all our firm-year observations and from Table 2 that the UK has low values of both PRIVDEBT and BANKMKT. If our findings are concentrated only among UK firms, then our inferences cannot be generalized to the non-UK countries in our sample. Similarly, France and Germany are the second and the third largest countries in our sample. Untabulated results indicate that the coefficients of interest become weaker after excluding UK, Germany and France firms, separately; however, they remain significant at the 10% significance level. Thus, both UK and continental European firms seeking to reduce their costs of borrowing increase the timeliness of their loss recognition after mandatory IFRS adoption. This outcome is expected, since the increased financial statement comparability offered by IFRS should not only benefit UK firms.

79

4.3.3. Endogeneity issue of early adopters To address the endogeneity issue with the voluntary IFRS adopters of our control sample, we conduct additional tests using the Heckman (1979) two-stage model. The first stage is a Probit model that accounts for the choice to early adopt IFRS. We consider economic characteristics, including firm size (SIZEit, measured by the log of total assets), financial leverage (LEVit, measured by the ratio of total debt to common equity), growth opportunities (MBit, measured by market value of equity divided by book value of equity). The inverse Mills ratio generated from the first stage is added to our Eqs. (3), (4a) and (4b). Untabulated results show that the coefficient of POST × COD in Eq. (3), the coefficient of POST × COD × HLDEBT in Eq. (4a) and the coefficient of POST × COD × HLBANK in Eq. (4b) remain insignificant for early adopters.

4.3.4. The validity of C score Following Khan and Watts (2009), we assess the validity of CSCORE as a measure of conservatism in the European Union setting. To do so, we examine whether CSCORE can effectively distinguish firms with different levels of earnings conservatism in a consistent way as Basu's (1997) return-based asymmetric timeliness coefficients. We sort firms into quintiles based on their CSCORE and estimate the Basu regression within each group. Table 8 shows that the Basu asymmetric timeliness measure (DR × R) is increasing nearly monotonically from the lowest to the highest group of CSCORE. Specifically, for the coefficient on DR × R, the difference between the highest and the lowest CSCORE group is significantly positive (0.509, t-statistic = 5.77). Overall, these results suggest that CSCORE is effective in distinguishing various degrees of earnings conservatism.

4.3.5. Alternative measures of earnings conservatism We assess the robustness of our results using two alternative measures of earnings conservatism — the extent to which earnings include negative total accruals before depreciation (Givoly & Hayn, 2000), and the extent to which earnings include negative non-operating accruals (Givoly & Hayn, 2000). To test H1, Table 9 Panel A shows the results using the conservatism indicator of negative total accruals before depreciation, and Panel B shows the results using the indicator of nonoperating accruals. Both indicators are scaled by net assets at the beginning period. In both panels, the results show that the coefficients on POST × COD are significantly negative for mandatory adopters, which suggest that firms with high costs of debt tend to reflect bad news more timely through the recognition of “negative” accruals in accounting following IFRS. However, the coefficient on POST × COD is insignificant for early adopters. The results support our H1. Similarly, untabulated results using these two alternative conservatism measures are also consistent with H2.

Table 8 Coefficients from basic Basu regressions sorted by CSCORE. CSCORE quintile

1

2

3

4

5

Intercept

0.052 (13.27)a −0.003 (−0.25) 0.029 (2.72)a 0.055 (2.32)b 0.057

0.064 (14.56)a −0.003 (−0.31) 0.032 (3.51)a 0.237 (6.94)a 0.103

0.066 (10.96)a 0.017 (1.23) 0.046 (4.18)a 0.370 (8.83)a 0.191

0.057 (6.99)a 0.088 (4.58)a 0.054 (0.84) 0.578 (12.27)a 0.182

0.057 (6.85)a 0.059 (3.24)a −0.100 (−1.55) 0.654 (13.91)a 0.241

DR R DR × R Adj. R

2

Difference (1 vs. 5)

0.509 (5.77)a

Notes: Firms are sorted annually into quintiles by CSCORE. The regression of Basu (1997) is estimated for each group. The earnings–return regression of Basu (1997): NIit = α0 + α1DRit + α2Rit + α3RitDRit + εit, where NI is net income scaled by the share price at the beginning of the fiscal year. R is the annual return of firm i over the twelve-month period from the fifth month of the fiscal year t to the fourth month of the fiscal year t + 1. DR is a dummy variable taking the value 1 when R is negative, and 0 otherwise. A positive coefficient of DR × R indicates the asymmetric timeliness of earnings in reflecting economic losses more quickly than economic gains. Superscripts a, b, c indicate statistical significance at the 0.01, 0.05, and 0.10 levels respectively for two-tailed t-tests.

80

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82

Table 9 Alternative measures of earnings conservatism. (1) Mandatory adoption

(2) Voluntary adoption

Coeff

Coeff

t-Stat

t-Stat

Panel A: Total accruals before depreciation (7.49) Intercept 0.024a POST 0.004 (1.79) c COD −0.004 (−1.85) a (−2.55) POST × COD −0.008 (−10.23) IntCov −0.040a (6.60) ROA 0.027a (4.65) σ(Earn) 0.013a (−24.86) LOSS −0.054a (−22.28) AQ −0.304a BETA −0.013a (−5.11) GROWTH 0.000 (0.91) INTEREST −0.001 (−0.56) Industry effect Yes Obs 10,192 0.48 Adj R2

0.009a −0.001 −0.002 −0.005 −0.004 −0.000 0.001 −0.005a 0.816a −0.001 0.001 −0.000 Yes 1668 0.19

(3.40) (−0.38) (−1.04) (−1.01) (−1.28) (−0.10) (0.59) (−3.24) (76.21) (−0.68) (1.41) (−0.99)

Panel B: Non-operating accruals Intercept −0.022a POST 0.003 COD −0.001 POST × COD −0.006b IntCov −0.041a ROA 0.020a σ(Earn) 0.013a LOSS −0.054a AQ −0.302a BETA −0.012a GROWTH 0.001 INTEREST −0.000 Industry effect Yes Obs 10,192 0.48 Adj R2

−0.023a −0.007c −0.002 0.004 −0.040a 0.015c 0.016b −0.044a −0.332a −0.006 0.001 −0.002 Yes 1668 0.17

(−5.25) (−1.73) (−0.44) (0.75) (−4.76) (1.82) (2.34) (−10.68) (−12.52) (−1.25) (0.35) (−0.07)

(−5.17) (1.01) (−1.47) (2.06) (−10.10) (3.30) (4.71) (−24.36) (−22.08) (−5.11) (0.88) (−0.15)

Notes: POST is a dummy variable equal to 1 for years after mandatory IFRS adoption (i.e. 2005, 2006, and 2007), and 0 otherwise; COD is interest expense divided by interest-bearing debt; IntCov is the scaled decile rank of the ratio of operating income to interest expense; ROA is the scaled decile rank of the return on assets; σ(Earn) is the scaled decile rank of the standard deviation net income before extraordinary items over the past five years; LOSS is an indicator equal to one if a firm incurs losses for two consecutive years and zero otherwise; AQ is the scaled decile rank of the absolute value of performance-matched abnormal accruals, following Kothari et al. (2005) and McNichols (2000); BETA is the firm's beta; GROWTH is the growth rate in sales; and INTEREST is a country's interest rate. t-Statistics are based on standard errors clustered by firm and year. Superscripts a, b, and c indicate significance at the 0.01, 0.05, and 0.10 levels respectively for two-tailed t-tests.

4.3.6. Is more loss timeliness associated with lower costs of debt? We examine whether the increased timely loss recognition after the mandatory IFRS adoption leads to a reduction in costs of debt. In particular, we expect those with higher costs of debt in the pre-IFRS period to experience larger increases in timely loss recognition in the post-IFRS period, thereby reducing costs of debts. To test this prediction, we split the sample based on the median level of costs of debt in the preIFRS period and create a dummy indicator, HSAMPLE, equal to 1 (0) for those with costs of debt above (below) the median. We then employ the following model to test whether firms with high costs of debt in the pre-IFRS period experience a reduction in costs of debt capital when they recognize economic losses in a more timely manner: CCODi ¼ δ0 þ δ1 CCSCOREi þ δ2 HSAMPLEi þ δ4 CCSCOREi  HSAMPLEi þ δ5 CIntCovi þ δ6 CAQ i þ δ7 Cσ ðEarnÞi þ δ8 CROAi þ δ9 CLEV i ð5Þ þ τ i þ λi þ ε i

where CCOD is the change in the mean of COD from the pre-IFRS period to the post-IFRS period; CCScore is the change in the mean of CScore from the pre-IFRS period to the post-IFRS period; CIntCov is the change in the mean of IntCov from the pre-IFRS period to the post-IFRS period; CAQ is the change in the mean of AQ from the pre-IFRS period to the

post-IFRS period; Cσ(Earn) is the change in the mean of σ (Earn) from the pre-IFRS period to the post-IFRS period; CROA is the change in the mean of ROA from the pre-IFRS period to the post-IFRS period; CLEV is the change in the mean of LEV from the pre-IFRS period to the postIFRS period; τ is a country fixed effect; and λ is the industry fixed effect. We predict a negative δ4 as, for firms with high costs of debts in the pre-IFRS period, higher conservatism in the post-IFRS period is associated with a greater reduction in costs of debt. Untabulated results show a significantly negative value of δ4 (−0.109, t-statistic = −1.95), indicating that the mandatory adoption of IFRS results in more timely loss recognition among firms with higher costs of debt in the pre-IFRS period, and therefore a reduced costs of debt in the post-IFRS period.14

4.3.7. Other tests We further examine whether the increase in timely loss recognition after mandatory IFRS occurs only in firms with higher costs of debt but not in firms with higher costs of equity. Although existing studies (e.g., Ball et al., 2008) suggest that timely loss recognition mainly caters for the needs of lenders, studies such as LaFond and Watts (2008) suggest that lower information asymmetry due to greater timely loss recognition can benefit outside equity investors. Firms with a higher cost of debt can also have a higher cost of equity, e.g. those in financial distress. To this extent, the increase in timely loss recognition after mandatory IFRS adoption that we observe among such firms may be due to managerial incentives to reduce the cost of equity rather than the costs of debt, as we argue.15 We replicate all our analyses by including an additional interactive term between POST and Beta to determine whether the relationship between the C-score and POST is conditional on the cost of equity and actually subsumes the coefficient of POST × COD.16 Untabulated results show that the coefficients of POST × Beta are consistently insignificant and, despite the presence of the new term, the coefficients of POST × COD remain significantly positive in all regressions. In other words, we confirm that the increase in loss recognition timeliness among mandatory IFRS adopters indeed exists only among those with a higher cost of debt but not necessarily among those with a higher cost of equity. This finding is in line with the argument of Ball et al. (2008) that timeliness of loss recognition is more important to lenders than shareholders. Finally, we also perform several other tests as a robustness check. We exclude the transition period, and control for institutional environments. We measure the quality of legal enforcement using the average score of the efficiency of the judicial system, rule of law and corruption as indicated in Lopez-de-Silanes et al. (1998). We further control for other determinants of conservatism, namely, size, leverage, and the book-to-market ratio. To mitigate the concern that accounting choice 14 To ensure that our main results are not driven the possibility that early adopters may have lower cost of debts before the mandatory IFRS adoption, we employ the following model in the pre-IFRS period, we employ the following model in the pre-IFRS period:

CODi ¼ δ0 þ δ1 MAN i þ δ2 CSCOREi þ δ3 IntCovi þ δ4 AQ i þ δ5 σ ðEarnÞi þ δ6 ROAi þ δ7 LEV i þ τ i þ λi þ ε i Our untabulated results suggest that before the mandatory IFRS adoption, voluntary IFRS adopters do not exhibit lower cost of debts compared to firms following the local standard. 15 Of course, even if this is the case, our evidence will still show that the incentives to reduce the cost of capital in general induce firms to increase their timely loss recognition after mandatory IFRS adoption in order to benefit from the opportunity to increase crossborder investment due to improvements in the international comparability of their accounting information. However, to ensure our interpretation of the evidence is not only partially but completely correct, we need to determine whether firms that increase their loss recognition timeliness after mandatory IFRS adoption do so to cater to lenders or equity investors. 16 Our use of beta to proxy the cost of equity follows the well established Capital Asset Pricing Model (CAPM). Other proxies of the cost of equity capital, such as those derived from equity analyst earnings forecasts and market price (e.g. Li, 2010), would reduce the cross-section of our sample substantially because equity analysts mainly cover larger firms.

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82

may be different for old and new debt, we use the median value of costs of debt in the pre-IFRS period of our sample (2002–2004) as an indicator of the costs for old debt and rerun all the tests. Untabulated results of these additional analyses lead to similar conclusions.

5. Summary and conclusion We find evidence among some firms that confirms the prediction of Ball (2006) that IFRS adoption increases timely loss recognition. Our finding contrasts with other concurrent studies (e.g., Ahmed et al., 2013; Chen et al., 2010) which instead suggest that mandatory IFRS adoption reduces the timeliness of loss recognition. Specifically, we first show that mandatory IFRS adopters (treatment sample) with a higher cost of debt are associated with higher Khan and Watts (2009) C-scores from 2005 onward, and that the same effect does not exist among their voluntary IFRS adopter (control sample) counterparts. The finding from the treatment sample confirms our prediction that firms seeking to reduce their costs of borrowing are associated with a higher degree of timely loss recognition in the post-IFRS period. The contrast in the finding from the control sample mitigates the possibility that the effect we observe among the mandatory IFRS adopters could be attributed to some unidentified confounding event. Second, we also find the evidence of firms with high costs of debt recognizing economic losses in a more timely manner after IFRS adoption exists mainly in countries with a lower pervasiveness of private debt or bank-based financing. This is consistent with high cost of debt firms catering to the demands of public lenders for higher accounting disclosure quality. It is in the public debt market that the effort to increase timely loss recognition is more likely to induce a decline in the cost of borrowing. The fact that our findings occur in countries where public debt is prevalent, further strengthens our inference of a mandatory IFRS adoption effect on the timeliness of loss recognition among firms with higher costs of debt. Finally, we show a significant reduction in C-scores among firms with lower costs of debt, after 2005. This result occurs among both mandatory and voluntary IFRS adopters and in countries with high and low pervasiveness of public debt. In other words, we reveal a reduction in the timeliness of loss recognition that is likely to be caused by background time trend effects and unlikely to be due to mandatory IFRS adoption. Therefore, it is possible that the reduction in timely loss recognition, which previous studies (e.g., Ahmed et al., 2013; Chen et al., 2010) attributed to IFRS, could be caused by this time trend. The growing IFRS literature focuses more on evaluating its indirect effects, i.e. economic consequences, than its direct effect, i.e. accounting disclosure quality. We examine the direct effect of IFRS and reveal that it induces a positive impact but mainly among firms seeking to reduce their costs of debt and only among countries where public debt is more prevalent. Our approach of identifying the IFRS effect on accounting disclosure quality through firms' corporate finance incentives could be applied to other studies beyond the subject of timely loss recognition. Our results provide important policy implications to regulators and standard setters. In the future, when the standard setters (e.g., IASB or FASB) want to improve financial reporting, they must recognize the importance of timely loss recognition. FASB attempts to ban conservatism to achieve neutrality of information (Watts, 2003a). Our studies show that timely loss recognition enhances contracting efficiency, and that European firms exhibit more timely loss recognition when they have incentives to improve debt contracting in the post-IFRS periods. If standard setters remove conservatism principle, it is likely that this will impair debt contracting efficiency and impose significant costs in debt markets. Finally, our results are subject to one limitation. While our results are supportive of the conditional-conservatism interpretation of the Basu coefficient, they do not rule out the possibility that the Basu coefficient may be due to factors other than conservatism (e.g. Dietrich, Muller, & Reidl, 2007).

81

References Agostino, M., Drago, D., & Silipo, D. (2011). The value relevance of IFRS in the European banking industry. Review of Quantitative Finance and Accounting, 36(3), 437–457. Ahmed, A., Billings, B., Morton, R., & Stanford-Harris, M. (2002). The role of accounting conservatism in mitigating bondholder–shareholder conflict over dividend policy and in reducing debt costs. The Accounting Review, 77(4), 867–890. Ahmed, A., Neel, S., & Wang, D. (2013). Does mandatory adoption of IFRS improve accounting quality? Preliminary evidence. Contemporary Accounting Research, 30(4), 1344–1372. Ball, R. (2006). International Financial Reporting Standards (IFRS): Pros and cons for investors. Accounting and Business Research, International Accounting Policy Forum, 5–27. Ball, R., Kothari, S., & Robin, A. (2000). The effect of international institutional factors in properties of accounting earnings. Journal of Accounting and Economics, 29(1), 1–51. Ball, R., Robin, A., & Sadka, G. (2008). Is financial reporting shaped by equity markets or by debt markets? An international study of timeliness and conservatism. Review of Accounting Studies, 13(2–3), 168–205. Ball, R., Robin, A., & Wu, J. (2003). Incentives versus standards: Properties of accounting income in four East Asian countries. Journal of Accounting and Economics, 36(1–3), 235–270. Ball, R., & Shivakumar, L. (2005). Earnings quality in UK private firms: Comparative loss recognition timeliness. Journal of Accounting and Economics, 39(1), 83–128. Barth, M., Landsman, W., & Lang, M. (2008). International accounting standards and accounting quality. Journal of Accounting Research, 46(3), 467–498. Basu, S. (1997). The conservatism principle and the asymmetric timeliness of earnings. Journal of Accounting and Economics, 24(1), 3–37. Beatty, A., Ramesh, K., & Weber, J. (2002). The importance of accounting changes in debt contracts: The cost of flexibility in covenant calculations. Journal of Accounting and Economics, 33(2), 205–227. Beck, T., Demirguc-Kunt, A., & Levine, R. (2000). A new database on the structure and development of the financial sector. World Bank Economic Review, 14(3), 597–605. Beuselinck, C., Joos, P., Khurana, I., & Van der Meulen, S. (2009). Mandatory IFRS reporting and stock price informativeness. Available at SSRN http://ssrn.com/abstract= 1381242 Bharath, S., Sunder, J., & Sunder, S. (2008). Accounting quality and debt contracting. The Accounting Review, 83(1), 1–28. Brochet, F., Jagolinzer, A., & Riedl, E. (2013). Mandatory IFRS adoption and financial statement comparability. Contemporary Accounting Research, 30(4), 1373–1400. Bruggemann, U., Hitz, J., & Sellhorn, T. (2013). Intended and unintended consequences of mandatory IFRS adoption: Review of extant evidence and suggestions for future research. European Accounting Review, 22(1), 1–37. Bushman, R., & Piotroski, J. (2006). Financial reporting incentives for conservative accounting: The influence of legal and political institutions. Journal of Accounting and Economics, 42(1–2), 107–148. Byard, D., Li, Y., & Yu, Y. (2011). The effect of mandatory IFRS adoption on financial analysts' information environment. Journal of Accounting Research, 49(1), 69–96. Cahan, S., Emanuel, D., & Sun, J. (2009). The effect of earnings quality and country level institutions on the value relevance of earnings. Review of Quantitative Finance and Accounting, 33(4), 371–391. Chen, H., Tang, Q., Jiang, Y., & Lin, Z. (2010). The role of International Financial Reporting Standards in accounting quality: Evidence from the European Union. Journal of International Financial Management and Accounting, 21(3), 220–278. Christensen, H., Lee, E., & Walker, M. (2008). Incentives or standards: What determine accounting quality change around IFRS adoption? Available at SSRN http://ssrn. com/abstract=1013054 Christensen, H., Lee, E., & Walker, M. (2009). Do IFRS reconciliations convey information? The effect of debt contracting. Journal of Accounting Research, 47(5), 1167–1199. Clarkson, P., Hanna, J.D., Richardson, G.D., & Thompson, R. (2011). The impact of IFRS adoption on the value relevance of book value and earnings. Journal of Contemporary Accounting and Economics, 7, 1–17. Daske, H., Hail, L., Leuz, C., & Verdi, R. (2008). Mandatory IFRS reporting around the world: Early evidence on the economic consequences. Journal of Accounting Research, 46(5), 1085–1142. Daske, H., Hail, L., Leuz, C., & Verdi, R. (2013). Adopting a label: Heterogeneity in the economic consequences around IAS/IFRS adoptions. Journal of Accounting Research, 51(3), 495–547. Day, I., & Taylor, P. (1996). Loan contracting by UK corporate borrowers. Journal of International Banking Law, 8, 318–325. DeFond, M., Hu, X., Hung, M., & Li, S. (2011). The impact of mandatory IFRS adoption on mutual fund ownership: The role of comparability. Journal of Accounting and Economics, 51(3), 240–258. Diamond, D. (1991). Monitoring and reputation: The choice between bank loans and directly placed debt. Journal of Political Economy, 99(4), 689–721. Diamond, D., & Verrecchia, R. (1991). Disclosure, liquidity and the cost of capital. The Journal of Finance, 66(4), 1325–1359. Dietrich, J.R., Muller, K.A., & Reidl, E.J. (2007). Asymmetric timeliness tests of accounting conservatism. Review of Accounting Studies, 12(1), 95–124. Dimitropoulos, P.E., Asteriou, D., Kousenidis, D., & Leventis, S. (2013). The impact of IFRS on accounting quality: Evidence from Greece. Advances in Accounting, 29(1), 108–123. Fama, E. (1985). What's different about banks? Journal of Monetary Economics, 15(1), 29–39. Florou, A., & Kosi, U. (2013). Does mandatory IFRS adoption facilitate debt financing? Available at SSRN http://ssrn.com/abstract=1508324 Florou, A., & Pope, P. (2012). Mandatory IFRS adoption and institutional investment decisions. The Accounting Review, 87(6), 1993–2025.

82

A.L.-C. Chan et al. / International Review of Financial Analysis 38 (2015) 70–82

Francis, J., La Fond, R., Olsson, P., & Schipper, K. (2005). The market pricing of accruals quality. Journal of Accounting and Economics, 39(2), 295–327. Givoly, D., & Hayn, C. (2000). The changing time-series properties of earnings, cash flows and accruals: Has financial reporting become more conservative? Journal of Accounting and Economics, 29(3), 287–320. Heckman, J. (1979). Sample selection bias as a specification error. Econometrica, 47, 153–161. 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. Iatridis, G. (2010). International Financial Reporting Standards and the quality of financial statement information. International Review of Financial Analysis, 19, 193–204. Iatridis, G.E. (2011). Accounting disclosures, accounting quality and conditional and unconditional conservatism. International Review of Financial Analysis, 20, 88–102. Jeanjean, T., & Stolowy, H. (2008). Do accounting standards matter? An exploratory analysis of earnings management before and after IFRS adoption. Journal of Accounting and Public Policy, 27(6), 480–494. Jiao, T., Koning, M., Mertens, G., & Roosenboom, P. (2012). Mandatory IFRS adoption and its impact on analysts' forecasts. International Review of Financial Analysis, 21, 56–63. Karampinis, N.I., & Heves, D.L. (2011). Mandating IFRS in an unfavourable environment: The Greek experience. The International Journal of Accounting, 46(3), 304–332. Khan, M., & Watts, R. (2009). Estimation and empirical properties of a firm-year measure of accounting conservatism. Journal of Accounting and Economics, 48(2–3), 132–150. Kothari, S., Leone, A., & Wasley, C. (2005). Performance matched discretionary accrual measures. Journal of Accounting and Economics, 39(1), 163–197. LaFond, R., & Watts, R. (2008). The information role of conservatism. The Accounting Review, 83(2), 447–478. Lai, C., & Taylor, S. (2008). Estimating and validating a firm-year-specific measure of conservatism: Australian evidence. Accounting and Finance, 48(4), 673–695. Landsman, W., Maydew, E., & Thornock, J. (2012). The information content of annual earnings announcement and mandatory adoption of IFRS. Journal of Accounting and Economics, 53(1–2), 34–54. Leftwich, R. (1983). Accounting information in private markets: Evidence from private lending agreements. The Accounting Review, 58(1), 23–42. Leuz, C. (2003). IAS versus US GAAP: Information asymmetry-based evidence from Germany's new market. Journal of Accounting Research, 41(3), 445–472. Leuz, C. (2010). Different approaches to corporate reporting regulation: How jurisdictions differ and why. Accounting and Business Research, 40(3), 229–256. Leuz, C., & Verrecchia, R. (2000). The economic consequences of increased disclosure. Journal of Accounting Research, 38(3), 91–124. Leuz, C., & Wysocki, P. (2008). Economic consequences of financial reporting and disclosure regulation: A review and suggestions for future research. Available at SSRN http://ssrn.com/abstract=1105398 Li, X. (2009). Accounting conservatism and cost of capital: International analysis. Available at SSRN http://ssrn.com/abstract=1261971 Li, S. (2010). Does mandatory adoption of International Financial Reporting Standards in the European Union reduce the cost of equity capital? The Accounting Review, 85(2), 607–636.

Lopez-de-Silanes, F., La Porta, R., Shleifer, A., & Vishny, R. (1998). Law and finance. Journal of Political Economy, 106(6), 1113–1155. Loureiro, G., & Taboada, A. (2014). Do improvement in the information environment affect real investment decisions? Working paper (Available at SSRN: http://ssrn.com/ abstract=1952593). McNichols, M. (2000). Research design issues in earnings management studies. Journal of Accounting and Public Policy, 19(4–5), 313–345. Moir, L., & Sudarsanam, S. (2007). Determinants of financial covenants and pricing of debt in private debt contracts: The UK evidence. Accounting and Business Research, 37(2), 151–166. Naranjo, P., Saavedra, D., & Verdi, R. (2014). Financial reporting regulation and financing decisions. Working paper (Available at SSRN: http://papers.ssrn.com/sol3/papers. cfm?abstract_id=2147838). Nelson, M. (2003). Behavioral evidence on the effects of principles- and rules-based standards. Accounting Horizons, 17(1), 91–104. Nikolaev, V. (2010). Debt covenants and accounting conservatism. Journal of Accounting Research, 48(1), 51–89. Nobes, C. (2005). Rules-based standards and the lack of principles in accounting. Accounting Horizons, 19(1), 25–34. Ormrod, P., & Taylor, P. (2004). The impact of the change to international accounting standards on debt covenants: A UK perspective. Accounting in Europe, 1(1), 71–94. Petersen, M. (2009). Estimating standard errors in finance panel data sets: Comparing approaches. Review of Financial Studies, 22(1), 435–480. Pittman, J., & Fortin, S. (2004). Auditor choice and the cost of debt for newly public firms. Journal of Accounting and Economics, 37(1), 113–136. Sengupta, P. (1998). Corporate disclosure quality and the cost of debt. The Accounting Review, 73(4), 459–474. Tan, H., Wang, S., & Welker, M. (2011). Analyst following and forecast accuracy after mandated IFRS adoptions. Journal of Accounting Research, 49(5), 1307–1357. Verrecchia, R. (2001). Essays on disclosure. Journal of Accounting and Economics, 32(1–3), 97–180. Watts, R. (2003a). Conservatism in accounting part I: Explanations and implications. Accounting Horizons, 17(3), 207–221. Watts, R. (2003b). Conservatism in accounting part II: Explanations and research opportunities. Accounting Horizons, 17(4), 287–301. Watts, R., & Zimmerman, J. (1986). Positive accounting theory. Engelwood Cliffs, NJ: Prentice-Hall. Wu, J., & Zhang, I. (2009). The adoption of internationally recognized accounting standards: Implications for the credit markets. Available at SSRN http://ssrn.com/ abstract=1425209 Wu, J., & Zhang, I. (2010). Accounting integration and comparability: Evidence from relative performance evaluation around IFRS adoption. Working paper (Available at SSRN: http://ssrn.com/abstract=1650782 or http://dx.doi.org/10.2139/ssrn.1650782). Zeff, S. (1978). The rise of economic consequences. The Journal of Accountancy, 146, 56–63. Zhang, J. (2008). The contracting benefits of accounting conservatism to lenders and borrowers. Journal of Accounting and Economics, 45(1), 27–54.