IFRS adoption and investor perceptions of earnings quality : evidence from Korea ”

This study examines the consequences of International Financial Reporting Standards (IFRS) adoption in terms of the investor perception of earnings quality in the Korean stock market. Building on evidence from Ecker et al. (2006) suggesting that returnbased earnings quality (E-loading), as captured by the sensitivity of stock returns to accounting information risk, accurately represents investor perceptions of earnings information risk, the authors examine whether E-loading is different between the preand post-IFRS adoption periods. Using KSE-listed firms from 2006 to 2014, the authors find the evidence that the extent of stock return sensitivity to information risk embedded in financial statements is greater in the period of post-IFRS adoption than in the period of pre-IFRS adoption. This finding indicates that even though accounting-based earnings quality improves after the adoption of IFRS, investors perceive earnings information after the adoption of IFRS as riskier than before. In addition, the difference in investor perception is more pronounced for firms with low accruals quality as captured by discretionary accruals, indicating that the effect of IFRS adoption on return-based earnings quality is distinctive from that on accounting-based earnings quality. The paper contributes to the literature on IFRS by exploring the effect of IFRS adoption through a new perspective on earnings quality in capital market. Heejeong Shin (Korea), Su-In Kim (Korea), Sangmi Kim (Korea) BUSINESS PERSPECTIVES LLC “СPС “Business Perspectives” Hryhorii Skovoroda lane, 10, Sumy, 40022, Ukraine www.businessperspectives.org IFRS Adoption and Investor Perceptions of Earnings Quality: Evidence from Korea Received on: 14th of November, 2017 Accepted on: 3rd of January, 2018


INTRODUCTION
The purpose of this study is to examine the consequences of IFRS adoption in terms of earnings quality that investors perceive from financial statements. A large number of studies argue that financial statements under IFRS provide earnings information with high predictability for future earnings, and thus, the accounting numbers on such financial statements are relatively value-relevant for firm valuation (Barth et al., 2008;Ding et al., 2007;Bellas et al., 2007;Morais & Curto, 2007). However, regarding IFRS adoption, researchers and practitioners consider low reliability in accounting information to be problematic, because the principle-based regulation in IFRS leaves accounting practice choices to the manager's discretion. Concurrently, this regulation leads implicitly to earnings management, which is motivated by managers who have incentives to exploit IFRS regulation (Christensen et al., 2008). Furthermore, the disclosure of consolidated financial statements and the fair value of assets on financial statements necessarily cause accounting numbers to fluctuate between pre-and post-IFRS adoption periods, providing earnings information that is barely connected to the historical financial statement information under non-IFRS (Choi et al., 2011). For these reasons, whether the earnings quality with respect to information reliability improves or worsens after IFRS adoption has been a critical issue in Korea (Lee et al., 2016;Moon, 2016).
While there are several studies reporting the consequences of IFRS adoption for accounting-based earnings quality in Korea, their findings are mostly limited to early adoption (i.e., [2009][2010] or the early period of mandatory adoption (i.e., 2011). Additionally, most of these studies report the positive impacts of the adoption of IFRS with respect to earnings quality as captured by the discretionary accruals (Moon, 2011;Choi & Shon, 2012;Park et al., 2012). Of course, there exists some research that documents the change in earnings quality in the post-IFRS period in terms of value relevance or earnings persistence, but these results are mixed (Choi, 2013;Moon, 2016). Moreover, the economic consequences of IFRS adoption differ by accounting intermediaries, such as accounting practitioners, auditors, and analysts. According to survey results by Lee et al. (2016), while accounting practitioners recognize that accounting numbers in financial statements under IFRS reflect firm performance and financial conditions with more accuracy and faithfulness, neither auditors nor analysts approve them. This implies that accounting information users perceive less improvement in earnings quality.
In this context, we raise the question of whether the adoption of IFRS has improved earnings quality in a way that investors experience meaningfully. It is possible that even though accounting-based earnings quality has been enhanced by the adoption of IFRS, investors may perceive earnings from financial statements under IFRS regulation as relatively risky information because of low earnings persistence or poor cross-sectional consistency in accounting practices among firms after the adoption of IFRS. However, there is little research documenting the current consequences of IFRS adoption in terms of market-based earnings quality in Korea. In capturing investor perceptions of earnings quality, we build on Ecker et al.'s (2006) evidence that return-based earnings quality ( ) as captured by the sensitivity of stock returns to accounting information risk, accurately represents investor perceptions of earnings information risk. By examining the difference in the priced factor to earnings information risk between the pre-and post-IFRS adoption periods, we provide some insight into the consequences of IFRS adoption in terms of investor perceptions of earnings quality.
Using Korean Stock Exchange (KSE) listed firms from 2006 to 2014, we find that while the average of accruals quality measured as discretionary accruals decreases over the post-IFRS adoption period, market-based earnings quality ( ) -loading E is higher after IFRS adoption than before. This result indicates that despite improvements in accounting-based earnings quality, investors assess earnings quality from financial statements under IFRS to be relatively low compared to those under non-IFRS. Additionally, we show that for the post-IFRS adoption period, the market-based earnings quality is moderated by accruals-based earnings quality. This finding indicates that regarding the effect of IFRS adoption on earnings quality, return-based earnings quality is distinctive from accounting-based earnings quality. Collectively, our findings suggest that in spite of the improvement in accountingbased earnings quality since IFRS adoption, investors believe earnings information from financial statements is less reliable.
Our findings contribute to the academic and practical fields in several ways. First, given that many papers suggest the consequences of IFRS adoption on the accounting-based perspective, we show consequences on the market-based perspective by examining the difference between pre-and post-IFRS adoption in investor perceptions of information risk. Second, we argue that market-based earnings quality worsens after IFRS adoption, and thus, we provide authorities in charge of accounting regulations and enforcement with the opportunity to reassess regulations and policies regarding the effectiveness of IFRS regulation. Overall, our paper extends prior literature on IFRS regulation by showing the effect of IFRS adoption through a new perspective on earnings quality. This paper is divided into the following sections: section 1 discusses research background and our research question and then develops hypotheses, section 2 describes methodology, including the sample, measurement of variables, and model specification, section 3 presents the empirical results. Finally, last section concludes.

IFRS (International Financial
Reporting Standards) adoption is considered to be a historic accounting regime change that drastically affects the information environment. Since firms are required to disclose earnings information based on economic earnings regarding the firm's intrinsic value under IFRS, financial statement information is expected to be more relevant (Barth et al., 2008). Moreover, IFRS, as a principle-based accounting standard, provides practitioners with accounting principles and general rules, thus making many of the accounting expert's decisions. Because the adoption of IFRS as a new regulation in accounting standards has changed financial statements, researchers and practitioners, including investors, have been concerned about the change in earnings quality between the pre-and post-IFRS adoption periods. In this regard, a large body of studies reports mostly the positive effects of IFRS adoption in terms of earnings quality (Barth et al., 2008;Ding et al., 2007). Specifically, Barth et al. (2008) examine the consequences of IFRS adoption using 1,896 observations of 21 countries that adopted IFRS voluntarily from 1994 to 2003. The researchers show that earnings management for firms that adopted IFRS is low relative to non-IFRS IFRS firms, and improvement in earnings quality for IFRS firms is higher than for non-IFRS firms. Bellas et al. (2007) report that the adoption of IFRS enhances the value relevance of accounting information. Moreover, Morais, and Curto (2008) document that earnings management has been reduced since the adoption of IFRS. There exist previous studies that document the change in earnings quality in the post-IFRS period in terms of value relevance and earnings persistence; however, the results of these studies are mixed. Ahmed et al. (2013) investigate whether the mandatory adoption of IFRS influences earnings quality, following Barth et al. (2008). The researchers show that firms in post-IFRS adoption reveal higher smooth earnings, higher earnings management based on accruals, and lower sensitivity regarding early recognition of loss.
There are several studies reporting the consequences of IFRS adoption regarding accounting-based earnings quality in Korea. The findings from these studies are mostly limited to the early adoption period (i.e., 2009-2010) or early in the period of mandatory adoption (i.e., 2011). Moon (2011) argues that earnings quality improves after IFRS adoption by showing lower accruals and earnings smoothing of 57 firms with early adoption due to changes in accounting standards from K-GAAP to K-IFRS. Choi and Shon (2012) and Park et al. (2012) also show that the early adoption of IFRS reduces both total accruals and discretionary accruals.
For the mandatory adoption of IFRS, Choi (2013) and Choi (2014) investigate whether accounting information is more value relevant in the post-IFRS adoption period and finds that total equity and net income, in particular, are more value relevant in consolidated financial statements. Moon (2016) shows that accruals quality and both persistence and predictability in cash flows are higher after IFRS adoption than before.
However, Lee et al. (2016) report that the economic consequences of IFRS adoption differ by accounting intermediaries, such as accounting practitioners, auditors, and analysts. According to their survey results, while accounting practitioners recognize that accounting numbers in financial statements under IFRS reflect firm performance and financial condition with more accuracy and faithfulness, neither auditors nor analysts approve them. This implies that accounting information users perceive less improvement in earnings quality.
In this regard, we investigate whether investors perceive earnings from financial statements as more/less risky information after IFRS adoption. In capturing investor perceptions of earnings quality, we build on Ecker et al.'s (2006) evidence that return-based earnings quality ( ) -load , ing E as captured by the sensitivity of stock returns to accounting information risk, accurately represents investor perceptions of earnings information risk. Ecker et al. (2006) proposed a model to capture market-based earnings quality measured as a priced factor of information risk by regression of returns minus risk-free returns on unknown factors of the Fama-French 3 factor model (1993), with discretionary accruals from Dechow and Dichev (2002) as a proxy for earnings quality. In that model, the sensitivity of stock returns to earnings quality is a priced factor of earnings quality or information risk and named as -loading. E We use -loading E to explore the difference in how investors perceive earnings quality from financial statements between the pre-and post-IFRS adoption periods. Our research is motivated by the intuition that accounting-based earnings quality is not meaningful to some investors.
To test our research question, we form the null hypothesis as follows.
H0: The extent of investor perceptions of information risk from financial statements is not different between pre-and post-IFRS adoption periods.

Market-based earnings quality (E-loading)
To the extent that the investors price the information risk through the estimation of earnings quality, we assume that a priced factor of information risk incorporates investor perceptions of earnings quality. Ecker et al. (2006) suggested -loading E capturing the sensitivity of stock returns to earnings as the proxy for the investor perception of earnings quality. To test our hypothesis, we measure -loading E following Ecker et al. (2006) and examine the difference in the level of -loading E between the pre-and post-IFRS adoption periods.
where i R is daily stock returns of firm . i f R is daily free-risk rate with 5-year maturity provided from the Bank of Korea (http://www.bok.or.kr). m R is daily market return based on equally weighted index. SMB indicates daily SMB factor, measured as the difference of stock returns between small sized firms and larger sized firms.
HML indicates daily HML factor, the difference of stock returns between high-and low-book to market firms. Lastly, AQ indicates earnings quality factor estimated based on Dechow and Dechev (2002). The estimation of regression of market returns less free risk returns on , AQ that is the coefficient 4 , it α represents the investor perception of earnings quality.

Model specification
Assuming that -loading E

. BIG
Finally, industry-fixed effect is also controlled in this model. Note that the level of -loading E signifies the extent of investors' perceptions of accounting information risk, not the low accounting quality perse. We expect 1 β of IFRS to be positive (negative) when investors perceive earnings from financial statements for the periods of post-IFRS adoption to be more (less) risky information than those in the periods of pre-IFRS adoption. In other words, if there is no difference of investor perception of earnings quality over the IFRS adoption periods, 1 β of IFRS is expected be zero or statistically insignificant, supporting our null hypothesis.

Sample selection
Our sample consists of common stocks listed on the Korea Stock Exchange from 2006 to 2014. We retrieve annual earnings data, daily stock prices, and other financial variables from the K i value II database of NICE Investors Service Co., Ltd. (https://www.kisvalue.com). Free-risk rates with 5-year maturity we obtain information from the Bank of Korea (http://www.bok.or.kr/main/ko-rMain.action). We impose the following restrictions: we delete (1) firms with a fiscal year end in non-December, (2) firms that belong to the financial and insurance industries, (3) firms with impairment of capital, and (4) Table 1

The relation between IFRS adoption periods and market-based earnings quality
We conduct cross-sectional regression to test our hypothesis and find that -loading E

( )
3 f E is significantly positively related to IFRS. As presented in Table 3 for the sample excluding firms from 2009 to 2010, which is the early adoption period. Assuming that higher -loading E signifies the extent of investors' perceptions of the information risk from financial statements, this result indicates that investors perceive accounting information after the adoption of IFRS to be riskier than before. Our finding is robust to controlling for other variables. Notes: The definitions of variables are defined in Model 1. The notation ***, **, and * denotes significance at the 1%, 5%, and 10% levels, respectively.

The moderate effect of accruals quality on the relation between IFRS adoption and market-based earnings quality
Although our finding indicates that the market prices information risk more highly for IFRS-based financial statement information, it is still doubtful that -loading E is a relevant proxy for representing overall earnings quality that investors perceive.  Moreover, it is possible that the deterioration in accounting based-earnings quality after IFRS adoption within our sample criteria may induce relatively higher -loading. E Thus, to validate our hypothesis more clearly, we examine whether the effect of IFRS adoption on investor perceptions ( ) -loading E of earnings quality is moderated by accounting-based earnings quality (i.e., accruals quality). If the estimation of -loading E in Eq. (1) subsumes the effect of accruals quality, then the relation between investors' perceptions ( ) -loading E and IFRS will not be affected by accruals quality. However, if -loading E represents more general earnings quality independent of accruals quality, the relation between investors' perceptions ( ) -loading E and IFRS will be incrementally positive for firms with low accruals quality.
Before we run the regression model regarding our conjecture, we examine the difference in accruals quality between the pre-and post-IFRS periods.
Accruals quality variables ( ) DA are measured as the residuals (i.e., discretionary accruals) from regression for the model in Dechow et al. (1995) and Kothari et al. (2005), in which accruals quality is negatively related to the level of discretionary accruals in total accruals. In Table 4 and then regress -loading E on interaction terms. Even within the periods of post-IFRS adoption, if investors perceptions of the information risk from financial statements is higher when accruals quality is low, the interaction term ( ), β is expected to be significantly positive. For definitions of test variables, please refer to Model 1 and Table 4.  are not significant. This result implies that investors impose larger risk premiums for firms with low accruals quality after the adoption of IFRS than before, but they seem to be indifferent to the effect of accruals quality per se on information risk during the pre-IFRS period.  Notes: The definitions of variables are defined in Model 1 and Model 2. The notation ***, **, and * denotes significance at the 1%, 5%, and 10% levels, respectively.

Asymmetric effect of accruals quality on the investor perception of information risk
Although the investor perception of information risk varies with the extent of accruals quality for the post-IFRS period, we cast a further question of the asymmetric effect of accruals quality on the investor's perception of information risk. We decompose the accruals quality variable ( ) Empirical results will show more evidence that investors are able to distinguish firms with high earnings quality from those with low earnings quality regarding financial statement information.
If investors recognize the extent of accruals quality after IFRS adoption, the incremental effect of accruals quality on the investor's perception of information risk is more pronounced for firms with low earnings quality. Notes: The definitions of variables are defined in Model 1 and Model 2. The notation ***, **, and * denotes significance at the 1%, 5%, and 10% levels, respectively.

CONCLUSION
Assuming that the market-based information risk measure ( ) -loading E suggested by Ecker et al. (2006) accurately represents investor perceptions of earnings quality, we investigate whether -loading E is different pre-and post-IFRS adoption. Using KSE-listed firms from 2006 to 2014, we find evidence that -loading E is higher in post-IFRS adoption period compared to pre-IFRS adoption period. This result indicates that investors consider the risk embedded in accountingbased information in the post-IFRS adoption period to be riskier than before. Although accruals quality has improved since IFRS adoption, the results show that the positive relation between -loading E and the post-IFRS period is pronounced for firms with low accruals quality. This means that investors give more weight to information risk in pricing firms with low accruals quality after the adoption of IFRS than before and that higher -loading E in the post-IFRS period does not subsume the effect of accruals quality on information risk.
Our paper extends the literature on the consequences of IFRS adoption by showing the change of investor perceptions of information risk over the IFRS adoption periods. Moreover, our findings contribute to the practice field by providing evidence that market-based earnings quality from financial statements worsens after IFRS adoption, thus suggesting that authorities in charge of accounting regulations and enforcement have the opportunity to inspect the effectiveness of IFRS regulations.