Accruals quality, information risk, and institutional investors’ trading behavior: Evidence from the Korean stock market

Accruals quality, information risk, and institutional investors’ trading behavior: Evidence from the Korean stock market

Journal Pre-proofs Accruals quality, information risk, and institutional investors’ trading behavior: Evidence from the Korean stock market Kyung Soon...

774KB Sizes 0 Downloads 11 Views

Journal Pre-proofs Accruals quality, information risk, and institutional investors’ trading behavior: Evidence from the Korean stock market Kyung Soon Kim, Chune Young Chung, Jin Hwon Lee, Sangjun Cho PII: DOI: Reference:

S1062-9408(19)30056-7 https://doi.org/10.1016/j.najef.2019.101081 ECOFIN 101081

To appear in:

North American Journal of Economics & Finance

Received Date: Revised Date: Accepted Date:

28 January 2019 2 September 2019 30 September 2019

Please cite this article as: K. Soon Kim, C. Young Chung, J. Hwon Lee, S. Cho, Accruals quality, information risk, and institutional investors’ trading behavior: Evidence from the Korean stock market, North American Journal of Economics & Finance (2019), doi: https://doi.org/10.1016/j.najef.2019.101081

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2019 Published by Elsevier Inc.

Accruals quality, information risk, and institutional investors’ trading behavior: Evidence from the Korean stock market Kyung Soon Kima, Chune Young Chungb, *, Jin Hwon Leec, and Sangjun Chod a Division

of Business Administration, College of Business, Chosun University, 309 Pilmundaero, Dong-gu, Gwangju, Korea 501-759; E-mail: [email protected] b *(corresponding author) School of Business Administration, College of Business and Economics, Chung-Ang University, 84 Heukseok-ro, Dongjak-gu, Seoul, Korea 06974; E-mail: [email protected] c Department of Business Management, Osan University, 45 Cheonghak-ro, Osan-si, Gyeonggido, Korea 18119; E-mail: [email protected] d Price College of Business, University of Oklahoma, 307 West Brooks, Norman, OK 73019; Email: [email protected] JEL codes: G12; G14; G34; M41 Keywords: accruals quality; institutional investors’ trading behavior; institution type; macroeconomic condition; business diversification

Accruals quality, information risk, and institutional investors’ trading behavior: Evidence from the Korean stock market

Abstract: This study analyzes the effect of firm-specific information risk, measured by accruals quality, on the cost of capital using institutional investors’ trading behavior. Institutional investors in firms with lower accruals quality increase their net selling in later years. Furthermore, these investors’ net selling is relevant to the innate and discretionary factors of accruals quality. This relationship is stronger for foreign institutions than for domestic institutions, and it is mostly observed under favorable macroeconomic conditions. We do not observe this relationship for large business groups connected by shares.

JEL codes: G12; G14; G34; M41

Keywords: accruals quality; institutional investors’ trading behavior; institution type; macroeconomic condition; business diversification

1. Introduction Recent studies investigate whether firm-specific information risk influences pricing decisions in capital markets.1 Specifically, empirical accounting studies proxy firm-specific information risk with earnings quality and analyze the relation between earnings quality and the cost of capital (Aboody et al., 2005; Botosan, 1997; Botosan & Plumlee, 2005; Francis et al., 2004, 2005). Most studies find a negative correlation between the cost of capital and earnings quality. Easley and O'Hara (2004) and Lambert et al. (2007) establish theoretical systems to explain this negative correlation. Easley and O'Hara (2004) focus on information asymmetry among investors. Less informed investors demand return premia for firms with high information asymmetry (that is, firms with high information risk) because they recognize their disadvantages relative to more informed investors. In contrast, Lambert et al. (2007) focus on information precision rather than information asymmetry; they define information precision as the average accuracy of an investor's future cash flow prediction. Francis et al. (2004, 2005) empirically demonstrate these theoretical concepts by analyzing the effect of accruals quality (AQ), as a proxy for firm-level information risk (precision or asymmetry), on the cost of capital and stock returns. They find that the costs of debt and capital are greater for firms with low AQ than for firms with high AQ and that stock returns change significantly depending on the degree of AQ. Many subsequent studies consider whether AQ is a risk factor. Some researchers report that AQ is a risk factor that significantly influences the cost of capital and the pricing decision (Aboody et al., 2005; Barth et al., 2013; Bhattacharya et al., 2012; Kim and Qi, 2010; Ogneva, 2008). However, other studies

Abbreviations: accruals quality (AQ), buy-and-hold abnormal returns (BHAR), International Financial Reporting Standards (IFRS), Korea Exchange (KRX), Korea Fair Trade Commission (KFTC), price-earnings to growth (PEG), private information (PIN) 1

Traditional finance theory explains that firm-intrinsic information does not affect pricing decisions because diversified investments can eliminate intrinsic risk. However, the notion that firm-intrinsic information influences the expected rate of return is supported by several studies, including Merton’s (1987) incomplete information models, Diamond and Verrecchia’s (1991) liquidity effect models, and the asymmetric information models of Brennan et al. (1998) and Admati (1985).

find that AQ does not significantly influence the cost of capital or the pricing decision (Cohen, 2008; Core et al., 2008; Khan, 2008; McInnis, 2010; Mohanram & Rajgopal, 2009). Thus, influence of AQ on these two outcomes remains controversial. This study therefore investigates whether firm-specific information risk in emerging capital markets, such as South Korea, affects investment risk. Unlike previous studies analyzing the effect of AQ on the cost of capital and stock price returns, our study examines whether institutional investors’ trading behavior changes according to the degree of AQ. This approach is primarily motivated by the lack of consensus in the literature, as studies use either the ex-ante cost of capital (e.g., that implied by analyst forecasts) or the ex-post cost of capital (e.g., stock returns) to establish AQ’s pricing effect. Hence, we attempt to bridge this gap from a behavioral perspective using an alternative measure of the cost of capital (i.e., institutional investors’ trading behavior). Furthermore, the qualitative difference between emerging and developed markets precludes us from using the ex-post cost of capital. The proportion of individual investors is relatively higher in emerging capital markets, leading to greater information asymmetry between the two types of market participants. Stock price returns are therefore more vulnerable to irregular noise. Consequently, using stock returns as a proxy for the cost of capital in emerging markets may cause measurement error to obscure our findings. Relatedly, An et al. (2010) examine whether accounting information quality, proxied by AQ, is related to stock returns in Korea and whether this relationship affects the cost of capital. They find no evidence that this risk is systematically priced in stock returns, and it does affect the cost of capital in Korea, unlike in the US. They clearly note that measurement errors due to the Korean market’s unique characteristics may cause different results from those in other settings. They cite as possible explanations for the discrepancy: (ⅰ) different industry classifications for calculating accruals; (ⅱ) Korea’s greater amount of earnings management, which could lead to greater AQ measurement error; (ⅲ) different mechanisms by which information is reflected in stock prices owing to the higher

proportion of individual investors; and (ⅳ) different legal systems (civil laws) and financial regulations and, thus, different trading behaviors of institutional investors.

Considering these unique market conditions, we propose using institutional investors’ net trading behavior as an alternative measure of the ex-post cost of capital. Korea’s capital market, unlike those of other countries, classifies investors as individuals, domestic institutions, and foreigners. In addition, the Korea Exchange (KRX), Korea’s only available securities exchange, discloses the daily trading volumes of individual companies by investor type.2 These unique Korean trading volume data provide new measures to analyze whether AQ affects rational investors’ risk perceptions without needing to proxy for the cost of capital. We predict that if a sophisticated institutional investor identifies AQ as idiosyncratic risk, this investor’s net selling should increase only for firms with low AQ. In other words, we indirectly investigate the pricing effect of AQ by examining whether institutional investors’ trading behavior differs according to the level of firm-specific information risk, as measured by AQ. Thus, we analyze the relationship between the degree of AQ at the end of the fiscal year and institutional investors’ net trading behavior in subsequent years, using a sample of companies listed on the Korean stock market from 2003 to 2014. The objectives and results of this study are as follows. First, we examine whether institutional investors’ net selling volume increases more following accounting information announcements for lower AQ firms than following those for higher AQ firms. If institutional investors can eliminate unsystematic risk (in this case, firm-specific information risk) through diversification, then their trading behavior should not differ across firms with different AQ. However, if they cannot completely eliminate

Most foreign investors in Korea are institutional investors. Thus, we define institutional investor transactions as the sum of domestic institution and foreigner transactions. 2

unsystematic risk through diversification, their selling volumes may be greater for lower AQ firms because they see such firms as riskier. Moreover, we decompose AQ into discretionary and innate factors and analyze the impact of each factor on institutional investors’ trading behavior.3 We find that institutional investors’ net selling increases for lower AQ firms. Furthermore, we find that both innate and discretionary AQ are associated with net selling by institutional investors. This result confirms that of a previous study supporting the pricing effect of firm-specific information risk. Second, we investigate whether different types of institutional investors are differently aware of the risk related to AQ. In Korea, domestic and foreign institutions may differ in terms of the independence of their investments and possible alternative investments. Because most domestic financial institutions (e.g., fund, insurance, and asset management companies) are affiliated with large business groups (e.g., chaebols), their trades may not always be determined independently and economically.4 In addition, changes in the asset compositions of domestic institutions can be relatively inelastic to changes in risk because their asset allocations are mostly concentrated in domestic firms. In contrast, foreign institutions’ interests are less conflicting. Furthermore, their ability to diversify their investments internationally means that their asset allocations are relatively more elastic to changes in risk. Thus, we predict that domestic and foreign institutions have different trading behaviors with respect to firm-specific information risk. We find that domestic and foreign institutional investors both display net selling behavior for firms with lower innate AQ. Foreign institutions react more negatively to low innate AQ than their domestic counterparts do. However, we observe significant net selling behavior for lower

Francis et al. (2004, 2005) measure an individual firm’s AQ as the standard deviation of abnormal accruals during a specific period. They argue that two factors affect the volatility of abnormal accruals: the fundamental risk from business operations, which can change AQ’s size, and the manager’s discretionary choice, which can also alter AQ. Thus, they decompose AQ into fundamental innate and discretionary AQ to assess each component’s effect on the cost of capital and the pricing decision and report that both affect the cost of capital. 3

For instance, Samsung’s subsidiaries include financial firms, such as Samsung Securities, Samsung Fire Insurance, Samsung Life Insurance, Samsung Asset Management, and Samsung Card. Most large domestic conglomerates (i.e., chaebols) also own financial companies.

4

discretionary AQ firms only among foreign investors. These results suggest that awareness of information risk caused by discretionary management factors can differ depending on an institutional investor’s independence and alternative investment options. Third, we investigate whether the relation between AQ and institutional investors’ trading behavior changes with macroeconomic conditions. Kim and Qi (2010) observe a risk premium associated with AQ in periods of economic growth but not during economic depressions. Thus, they illustrate that the link between the risk premium and the variance of AQ can systematically change with macroeconomic conditions. We perform a similar analysis and demonstrate that institutional investors mostly recognize innate AQ under good economic conditions. Thus, our results are consistent with those of Kim and Qi (2010). Lastly, we investigate whether the link between AQ and institutional investors’ trading behavior differs for firms that belong to large business groups. In Korea, firms within large-scale corporate groups (e.g., chaebols) are connected by shares owned by affiliated firms, and owner-managers tend to exercise more control than is assigned. We can consider such firms diversified. A diversified firm can reduce the fundamental risk in its operating activities by constructing a business portfolio or internal capital market. Additionally, because companies within enterprise groups must prepare consolidated financial statements, accounting information may be mutually monitored among companies. Thus, we predict that institutional investors are less aware of the inherent information risks of companies within enterprise groups relative to independent firms. We find that institutional investors’ net selling behavior is usually significant for individual firms but not for corporate group firms. Thus, information risk may be lower for firms within enterprise groups. Our results suggest that AQ’s pricing effect depends on many factors, such as investor type, market prospects, and corporate governance structure. In the following section, we introduce the theoretical background and develop our hypotheses. In section 3, we describe the main variables and the

empirical models used. Section 4 reports the analysis results. Lastly, section 5 concludes and provides suggestions related to current affairs. 2. Theoretical background and hypotheses 2.1 AQ and the cost of capital Easley and O'Hara (2004) and Lambert et al. (2007) theoretically explain the effect of the quality of accounting information on the cost of capital. Easley and O'Hara (2004) suggest that differences in the composition of public and private information (i.e., firm-level information asymmetry) influence the cost of capital and that investors demand higher returns for stockholdings containing more private information. Lambert et al. (2007) explain the effect of accounting information quality on the cost of capital in terms of an investor’s expected cash flow. They claim that announcing a higher information quality affects that firm’s assessed covariance and other firms’ cash flows and that the quality of accounting information is directly related to the cost of capital because it is non-diversifiable. Easley and O'Hara (2004) and Lambert et al. (2007) state that even if an investor can make diversified investments, risk can be recognized differently depending on the firm’s information asymmetry and precision. Francis et al. (2004, 2005) measure a firm’s information asymmetry and precision using accounting earnings quality and empirically demonstrate that the cost of capital changes with these measurements. They define accounting earnings quality (i.e., AQ) as the standard deviation of an individual firm’s past abnormal accruals and show that lower AQ firms (i.e., those with greater abnormal accruals volatility) incur higher costs of capital and lower excess rates. That is, by verifying the link between AQ and the cost of capital, they provide experimental evidence that firm-specific information risk is not diversified and can alter investors’ decisions. However, later studies report conflicting results. Several studies express strong doubts regarding the preliminary results of Francis et al. (2005). Core et al. (2008) conduct adequate verification of asset pricing to investigate whether AQ determines the expected return as a latent risk factor and find no proof that AQ is priced. Similarly, McInnis (2010) explains that previous findings that lower volatility is associated with a lower implied cost of capital are

driven by analysts’ optimism in their long-term earnings forecasts. Mohanram and Rajgopal (2009) fail to provide robust evidence that the information risk measured by private information (PIN) affects future returns and instead find that PIN exhibits no association with the cost of capital implied by analysts’ earnings forecasts. However, most subsequent theoretical studies support the conclusions of Francis et al. (2005) (Aboody et al., 2005; Barth et al., 2013; Bhattacharya et al., 2012; Kim & Qi, 2010; Ogneva, 2012). Aboody et al. (2005) find evidence that earnings quality is priced and that insider trading is more profitable for firms with more exposure to that factor. Barth et al. (2013) show that an earnings transparency measure is negatively related to subsequent excess returns, portfolio mean returns, and the expected cost of capital even after controlling for previously documented determinants of the cost of capital. Unlike Core et al. (2008), who state that the AQ risk factor does not determine pricing, Kim and Qi (2010) report that it is significantly reflected in the price after controlling for low-priced stocks. Furthermore, they find that the risk premium related to AQ is usually observed during periods of economic growth. Bhattacharya et al. (2012) check whether earnings quality and the cost of capital are directly or indirectly linked using path analysis. They observe statistically reliable evidence for both a direct path and an indirect path mediated by information asymmetry (measured by the bid–ask spread and PIN). Although these empirical studies do not always reach the same conclusions, they roughly agree that AQ and the cost of capital are linked. Francis et al. (2004, 2005) define AQ as a proxy for firm-level information asymmetry. They hypothesize that fluctuations in AQ can be attributed to the business’s innate risk factor and the manager’s discretionary risk factor and examine the relation between AQ and the cost of capital. They show that the cost of capital is associated with both types of risk factors. However, further studies disagree on whether the discretionary risk factor affects the cost of capital. Kim and Qi (2010) report that the innate portion of AQ influences the cost of capital, whereas the discretionary portion does not. They explain that this difference is due to managers’ motivation to smooth earnings and signals regarding

future economic conditions, which are incorporated in discretionary AQ (Guay et al., 1996; Subramanyam, 1996). Moreover, Gray et al. (2009) use Australian data to find a link between innate AQ and debt and the cost of capital, but they find no such link for discretionary AQ. Overall, prior studies conclude that investors react relatively more sensitively to innate than to discretionary AQ.

2.2 Overview of the Korean stock market The Korean stock market was fully disclosed in the process of overcoming the Asian financial crisis in the late 1990s. Regulatory authorities classified stock traders as individuals, domestic institutions, and foreigners after the 2000s to efficiently manage foreign capital, and they established a system to announce firms’ daily trading volumes for each investor type. The Korean stock market is more advantageous than other markets for research on trading behavior because institutional investors’ trading behavior (net selling and net buying) can be observed more directly. Because this market clearly divides the trading behaviors of domestic and foreign institutions, it provides reliable data for comparing institutional investors’ risk recognition of accounting information across different types using their trading behavior. The Korean capital market has experienced growth, depression, and stagnation since the 2000s. Since the capital market was opened to overcome the Asian financial crisis in the late 1990s, Korean companies have shifted away from debt-oriented financing toward capital market financing. Korean regulatory institutions also revised their accounting standards based on International Financial Reporting Standards (IFRS) to enhance international comparability, and IFRS have been mandatory since 2011. Furthermore, authorities developed a rigorous public market announcement system and enacted laws to

vitalize capital markets.5 As a result, the Korean capital market grew until the mid-2000s. Korea’s market index skyrocketed owing to the 2008 global financial crisis but it recovered to its previous state by 2010. The end of our sample period, 2011 to 2015, represents a period of stagnation.6 Considering Korea’s economic conditions after the 2000s, we define 2008 to 2010 as a depression and all other periods as favorable, and we compare institutional investors’ risk recognition of AQ (net selling or net purchasing) across different macroeconomic conditions. One distinct trait of Korean companies is that several firms may form a corporate group. These Korean corporate groups (i.e., chaebols) are connected by shares through mutual investments or crossshareholding and controlled by an owner-manager. Because many Korean corporate groups are connected to other industries by shares, the accounting information of a parent company, which owns many subsidiary companies, combines the performances of unrelated diversified subsidiaries. Thus, owning shares in a corporate group is the same as possessing a diversified business portfolio.7 Diversified and affiliated firms in a corporate group can better manage the business risk arising from firm operations owing to diversification and their own capital markets within the group. Because companies in a corporate group must disclose consolidated financial statements, partial monitoring of accounting information may occur among the companies. Thus, firms in corporate groups may have relatively low opportunistic earnings management, and investors in a corporate group’s affiliated firms are less likely to

5

Korea followed the US in adopting fair disclosure regulations in 2002 and revised and improved its Securities and Exchange Act to vitalize its capital market. Moreover, starting in 2009, Korea advanced related regulations by combining capital market and finance laws to develop the financial investment industry.

6

Korea’s market indexes, the KOSPI and the KOSDAQ, reached 628 and 444, respectively, at the end of 2002. However, by the end of 2007, they had grown to 1,897 and 704, respectively. Owing to the global financial crisis, they dropped to 1,124 and 332, respectively, at the end of 2008, but they recovered to 2,051 and 511, respectively, by the end of 2010. In 2015, the last year of this study’s sample period, they were at 1,961 and 682, respectively. The Korean capital market has stagnated since 2011.

7

In Korea, the regulatory institution (i.e., the Fair Trade Commission) publicly announces large corporate group firms and monitors and restricts their governance structures and unfair trades. Generally, most of the listed companies included in large corporate groups are large and own several listed or non-listed subsidiaries.

be concerned about individual investee firms’ intrinsic information risk. We predict that corporate governance causes differences in the intensity of accounting information that reflects firm-specific information risk. Thus, we examine whether the relationship between AQ and institutional investor transaction behavior changes with corporate governance.

2.3 Hypotheses Considering previous studies analyzing the effects of AQ and the unique characteristics of the Korean stock market, we develop several hypotheses about the relationship between AQ and institutional investor transactions. The main purpose of this study is examining whether the firm-specific information risk measured by AQ affects price using institutional investor transactions. Typically, we assume that institutional investors are rational and can make diversified investments. Accordingly, if an institutional investor can eliminate firm-intrinsic risk through diversified investments, AQ should not change the investor’s net trading behavior. In contrast, if the investor cannot eliminate firm-intrinsic risk through diversified investments, he or she will perceive lower AQ firms as riskier and demand a premium on the rate of return. This increase in the cost of capital in turn increases institutional investors’ selling volumes of lower AQ firms’ stocks. Thus, we predict that if AQ is a risk factor that reflects information asymmetry and precision, institutional investors’ net selling tendency should increase for lower AQ firms. Francis et al. (2005) argue that AQ can be influenced by economic fundamentals (i.e., innate AQ) and managers’ selective reporting of accounting policies and estimates (i.e., discretionary AQ). Empirical studies that support the pricing effect of AQ indicate that innate AQ reflects a company’s business model and fundamental risk attributed to the operating environment and that the innate rather than the discretionary component mainly affects price (Chen et al., 2008; Dechow & Dichev, 2002; Gray et al., 2009; Kim & Qi, 2010; Yee, 2006). However, these previous studies are based on developed markets in which internal and external monitoring substantially affect managers’ opportunistic earnings reporting

behavior. Investors’ risk recognition of discretionary AQ may differ between developed markets and countries in which market monitoring is not as strong. Thus, we separate AQ in the Korean stock market, which has relatively low market monitoring, into innate and discretionary AQ and further analyze its effect on institutional investors’ trading behavior in the later periods. Along these lines, we first test the following hypothesis to examine our predictions. H1: After accounting information announcements, institutional investors’ net selling volume should increase more for lower AQ firms than for higher AQ firms. The second focus of our study is investigating whether institutional investors’ trading behavior differs for different institution types. In Korea, institutional investors include domestic and foreign institutions. We presume that investment independence and the ability to diversify investments vary across the two types of institutions and that this difference can drive differences in the intensity of AQ risk recognition. Among domestic institutional investors, pension funds mostly distribute their assets to domestic firms for political reasons, and most financial investment institutions (e.g., securities, insurance, and asset management companies) form special relationships with some large conglomerates, both of which restrict their independence and ability to make diversified investments. Accordingly, domestic institutional investors make inelastic asset allocations relative to the information risk of individual firms. Thus, we expect the link between AQ and institutional investors’ trading behavior to be relatively weak. On the contrary, foreign institutional investors own a higher proportion of mutual funds in Korea, are mutually independent from the Korean government and domestic companies, and have many alternative international investment opportunities. Thus, they may make more economical trading decisions regarding changes in individual firms’ information risk.8 If AQ is a risk factor, we predict that the relation

8

The phenomenon of investors allocating lesser portions of their wealth to foreign capital markets is called the “home bias puzzle” (Tesar & Wener, 1995). Kang and Stulz (1997) argue that information asymmetry is one of the main causes of this puzzle. Hence, foreign investors may react more negatively to firm-specific information risk than domestic investors do.

between AQ and foreign institutional investors’ trading behavior is relatively stronger than that for domestic institutional investors. We set up the following hypothesis to verify this prediction. H2: The relation between institutional investors’ trading behavior and AQ changes according to the institutional investor type (domestic or foreign). The third focus of this study is examining whether Kim and Qi’s (2010) conclusion that AQ is differently reflected in pricing depending on macroeconomic conditions holds for the Korean market. Kim and Qi (2010) empirically show that the pricing effect of AQ is more prominent during economic expansions than during economic depressions. Furthermore, they show that the price effect of AQ according to the economic situation is related to innate rather than discretionary AQ. They therefore claim that AQ contributes to the cost of equity and that the pricing effect of AQ is associated with fundamental risk. Acknowledging prior studies, we analyze whether the relationship between AQ and institutional investors’ trading behavior changes depending on macroeconomic conditions (i.e., favorable economic periods vs. economic depressions) and which components of AQ (i.e., innate and discretionary factors) are linked to macroeconomic conditions. We examine the following hypothesis. H3: The relation between AQ and institutional investors’ trading behavior differs depending on macroeconomic conditions. The last focus of this study is examining whether the relation between AQ and institutional investors’ trading behavior is associated with corporate governance. Korea has a unique corporate governance structure called a chaebol, which is a nonrelated diversified large-scale corporate group that is connected by shares through mutual investments (or cross-shareholding) among firms and controlled by an owner-manager. In Korea, chaebols comprise a great proportion of the market, and institutional investors’ asset allocations are mostly concentrated in them. Until the early 2000s, Korean corporate groups exploited highly leveraged business portfolios. Indisputably, such a business diversification strategy can have both favorable (e.g., lower business risk)

and unfavorable (e.g., higher financial risk) consequences. Relatedly, prior empirical studies report that, in general, the downsides of highly leveraged business portfolios often outweigh their benefits (Ferris et al., 2003). Indeed, independent individual companies that lack the support of a conglomerate should be more prudent with their diversification strategies, whereas members of large corporate groups can take advantage of risky business plans. In fact, earlier studies based on data from India find that affiliates of large business groups can achieve better firm performance from more diversified portfolios (George & Kabir, 2012; Khanna & Palepu, 2000). Recently, Korean corporate groups have substantially increased their equity ratios, creating a new business environment in which they can choose from numerous diversification options that are not limited to strategies that heavily rely on debt. In other words, equity-driven corporate diversification at the group level can effectively lower business risk without substantially increasing financial risk. Hence, institutional investors may associate conglomerate members with less firm-specific information risk relative to independent enterprises. Thus, we examine the degree to which institutional investors’ risk perceptions of AQ change depending on a firm’s affiliation or non-affiliation with a large corporate group. We presume that the unique structure of Korean corporate groups can affect an individual firm’s information risk. Companies connected by shares to those in other industries are no different from a diversified business portfolio. Thus, a diversified firm can reduce the innate or fundamental risk of its operating activities because nonrelated diversified firms tend to dissipate their inherent operating risk. In addition, affiliated firms in a corporate group are likely to have their own capital market. As long as this capital market is efficient enough, the affiliated firms can reduce financial risk and gain more timely investment opportunities.9 Thus, institutional investors’ risk recognition of innate AQ may be lower for

9

Firms can be strongly affiliated owing to interrelated ownership structures, and the internal capital

market provides a soft budget constraint within the firms. This soft budget constraint may have positive or negative effects. For instance, Scharfstein and Stein (2000) show that a soft budget constraint may incur

large corporate groups than for individual firms, and thus, the relation between innate AQ and institutional investors’ net selling may be relatively weaker for corporate group firms. In contrast, we predict that large corporate group firms may provide a counterbalancing force for discretionary risk. First, discretionary risk may be higher for large corporate group firms.10 Demirkan et al. (2012) investigate the effect of discretionary AQ on the cost of capital for single-segment and multiple-segment firms from an agency-problem perspective. They demonstrate that multiple-segment firms have lower discretionary AQ, which in turn raises the cost of capital, reflecting the price. That is, Demirkan et al. (2012) argue that many more agency problems can arise for multiple-segment firms. Applying their perspective to Korea, many more agency problems can arise because large corporate group firms are multiple-segment firms connected by shares, and these agency problems can expand the discretionary factor of AQ. Alternatively, large corporate group firms may have lower discretionary risk. In an individual firm with multiple segments, opportunistic earnings management may prosper because the manager can wield power in each segment. On the contrary, large conglomerates connected by shares, such as chaebols, may yield different results and interpretations from those of Demirkan et al. (2012). As chaebols are required to disclose consolidated financial statements, chaebol affiliates must collaborate

managerial rent-seeking behavior and inefficiency from socialism in internal capital allocations. However, affiliated firms with an internal capital market face more flexible financial constraints and are less subject to investment-cash flow sensitivity (Fee et al., 2009; Lamont, 1997; Lee et al., 2009; Shin & Park, 1999; Shin & Stulz, 1998). In addition, Duchin (2010) shows that affiliated and diversified firms have less operational volatility and, thus, can maintain stable cash flows. He argues that such characteristics can reduce the cost of equity. Similarly, we expect that diversified and affiliated firms in the Korean market reduce their business risk owing to the role of internal capital markets. 10

Some existing studies find that conglomerate-affiliated firms’ accounting quality is worse than that of other firms (Bae & Jeong, 2007; Kim & Yi, 2006). These results indicate that members of conglomerates may have greater agency problems than non-members do.

within a single, unified accounting system, considerably undermining managerial accounting discretion and possibly leading to less opportunistic earnings management. Moreover, Korean business conglomerates are intensively monitored by numerous parties, including the regulatory authority (i.e., the Korea Fair Trade Commission), financial analysts, and institutional investors who own substantial portions of these conglomerates’ shares. Owing to the rigor of both internal and external monitoring mechanisms, conglomerate-affiliates with diversified business portfolios may have lower discretionary risk than individual enterprises do (Jiraporn et al., 2008). Subsequently, institutional investors may not evaluate conglomerate-affiliates’ discretionary risk as fastidiously as they would that of non-affiliated firms. Thus, we predict that the relationship between discretionary AQ and institutional investors’ net trading behavior is more pronounced for individual firms than for large corporate group firms. Because our study adopts the premise that large corporate groups can reduce aggregate business risk by adopting diversified business portfolios can and simultaneously lower managerial discretionary risk through strict internal and external monitoring, we predict that institutional investors’ risk recognition of both innate and discretionary AQ is less sensitive for conglomerate-affiliated firms. We set up the following hypothesis. H4: The relation between AQ and institutional investors’ trading behavior differs across firms that are part of large corporate groups and firms that are not.

3. Measurements of variables and methodology 3.1 Measurements of variables 3.1.1 Measuring firm-specific information risk Accruals quality (Francis et al. 2004, 2005). We adopt the approach of Francis et al. (2004, 2005), who measure individual firm-level information risk using accounting earnings quality. They measure the past

volatility of abnormal accruals for individual firms and define a firm with greater AQ volatility as having higher information risk. They estimate abnormal accruals (νit) by implementing a regression analysis from year t-4 to year t using the annual industry-specific model given by equation (1), and they calculate the standard deviation of abnormal accruals. They define AQ as the standard deviation of abnormal accruals over the past five years. In equation (1), TCAit indicates the total current accruals of firm i in year t.11 Ait is the average total assets of firm i in years t and t-1. CFOit is the cash flow from the operations of firm i in year t. △REVit is firm i’s sales in year t-1 deducted from its sales in year t. PPEit is the gross property, plant, and equipment value of firm i at the end of year t. νit is the residual of equation (4).

𝑇𝐶𝐴𝑖𝑡 𝐴𝑖𝑡

= ∅0 +

∅1𝐶𝐹𝑂𝑖,𝑡 ― 1 𝐴𝑖𝑡

+

∅2𝐶𝐹𝑂𝑖𝑡 𝐴𝑖𝑡

+

∅3𝐶𝐹𝑂𝑖𝑡 + 1 𝐴𝑖𝑡

+

∅4𝑅𝐸𝑉𝑖𝑡 𝐴𝑖𝑡

+

∅5𝑃𝑃𝐸𝑖𝑡 𝐴𝑖𝑡

+ 𝑣𝑖𝑡

(1)

∴ 𝐴𝑄𝑖,𝑡 = 𝜎(𝑣i,t)

Innate and discretionary AQ. Francis et al. (2004, 2005) contend that the volatility of abnormal accruals can be determined by the fundamental or innate risk in an individual firm’s operating activities and the discretionary risk incurred by a manager’s opportunistic earnings management. Thus, they decompose AQ into two components and analyze each component’s effect on the cost of capital. They first use the relationship between individual firm-level operating risk factors (e.g., firm scale, cash flow volatility, sales volatility, the business cycle, and the frequency of loss) and AQ to estimate the innate component, and they allocate the remainder to the discretionary component. They demonstrate that both components of AQ affect the cost of capital and that the innate component has a greater effect than the discretionary

11

We measure total current assets as follows: 𝑇𝐶𝐴𝑖,𝑡 = ∆𝐶𝐴𝑖,𝑡 ―∆𝐶𝐿𝑖,𝑡 ―∆𝐶𝑎𝑠ℎ𝑖,𝑡 + ∆𝑆𝑇𝐷𝐸𝐵𝑇𝑖,𝑡.

In this equation, △CAit is the change in firm i's current assets from year t-1 to year t. △CLit is the change in firm i's current liabilities from year t-1 to year t. △Cashit is the change firm i's cash and cash-convertible assets from year t-1 to year t. △STDEBTit is the change in debt among firm i's current liabilities from t-1 to year t.

component has. We apply their methods to divide AQ into innate and discretionary components and investigate whether each component affects institutional investors’ trading behavior. Innate AQ is the predicted value of equation (2), and discretionary AQ is the residual. In equation (2), AQit represents the AQ of firm i in year t. SIZEit is the size of firm i at the end of year t, defined as the log of market capitalization (one million won). σ(CFO)it indicates the cash flow volatility, which is the standard deviation of adjusted operating cash flows with respect to total assets from t-4 to t. σ(Sales)jt is sales volatility, which is the standard deviation of adjusted sales with respect to total assets, from t-4 to t. OperCycleit is the operating cycle of firm i in year t, which is the log of the sum of days of accounts receivable and days of inventory. NegEarnRatioit is the percentage loss, reflecting the number of losses in individual firms that occurred over the past five years as a percentage.

𝐴𝑄𝑖𝑡 = 𝛽0 + 𝛽1𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽2𝜎(𝐶𝐹𝑂)𝑖𝑡 + 𝛽3𝜎(𝑆𝑎𝑙𝑒𝑠)𝑖𝑡 + 𝛽4𝑂𝑝𝑒𝑟𝐶𝑦𝑐𝑙𝑒𝑖𝑡 (2)

+ 𝛽5𝑁𝑒𝑔𝐸𝑎𝑟𝑛𝑅𝑎𝑡𝑖𝑜𝑗𝑡 + 𝜀𝑖𝑡 𝐼𝑛𝑛𝑎𝑡𝑒𝐴𝑄𝑖𝑡 = 𝐴𝑄𝑖𝑡 𝐷𝑖𝑠𝑐𝐴𝑄𝑖𝑡 = 𝐴𝑄𝑖𝑡 ― 𝐴𝑄𝑖𝑡

3.1.2 Institutional investors’ trading behavior Conventional studies on the relationship between AQ and the cost of capital use either ex-ante or ex-post measures to proxy for the cost of capital. Francis et al. (2004) utilize the cost of capital implied by analyst forecasts. This ex-ante measure has a definitive advantage in that it accurately captures sophisticated analysts’ risk perception levels. Korean analysts, however, may not be able to provide such informative estimates because they face a severe agency problem and offer coverage for only a few firms, which are mainly large companies. Alternatively, stock price returns are an ex-post measure that conveys all market participants’ expectations following accounting disclosures (Francis et al., 2005; Kim & Qi, 2010). The usefulness of this measure as a proxy for the cost of capital, however, largely depends on

market efficiency. Compared with the US, Korea has a higher proportion of individual investors, higher investor protection regulations, and weaker external monitoring influences, leading to greater information asymmetry between individuals and institutions. These differences may undermine the validity of stock returns as a proxy. Instead, institutional investors’ net trading volume has the same advantages as ex-ante and ex-post measures because it offers clear insights into the rationale behind sophisticated institutional investors’ trading decisions in reaction to accounting disclosures. Moreover, because the KRX releases data on companies’ daily trading volumes by investor type, we can capture institutional investors’ net trading behavior more precisely and can include all KRX-listed firms in our sample. Hence, we use institutional investors’ net trading behavior as an alternative proxy for the cost of capital. We calculate institutional investor trading behavior for firm i in year t+1 according to equation (3). INST_NBRi,t+1 is institutional investors’ net purchase volume for firm i’s common stock divided by the weighted average of outstanding shares. This study assumes that the intensity of firm-specific information risk recognition varies owing to several differences between domestic and foreign institutions. Thus, we also measure domestic (D_INST_NBRi,t+1) and foreign institutional investors’ net trading behavior (F_INST_NBRi,t+1) using equations (4) and (5). Additionally, we observe individual investors’ net trading behavior (INDI_NBRi,t+1) using equation (6). In equations (3) to (6), INST_BUYi,t+1,d (INST_SELLi,t+1,d) is the purchase volume (selling volume) of firm i’s common stock on trading date d in year t+1 for both domestic and foreign institutional investors. D_INST_BUYi,t+1,d (D_INST_SELLi,t+1,d) is domestic institutional investors’ purchase (selling) volume of firm i’s common stock on trading date d in year t+1. Moreover, F_INST_BUYi,t+1,d (F_INST_SELLi,t+1,d) is foreign institutional investors’ purchase (selling) volume of firm i’s common stock on trading date d in year t+1. INDI_BUYi,t+1,d (INDI_SELLi,t+1,d) is the purchase (selling) volume of firm i’s common stock on trading date d in year t+1 for individual investors. NSi,t+1 is the weighted average of outstanding shares of firm i in year t+1. n indicates the number of trading dates in year t+1. 𝑛

𝐼𝑁𝑆𝑇_𝑁𝐵𝑅𝑖,𝑡 + 1 = ∑𝑑 = 1(𝐼𝑁𝑆𝑇_𝐵𝑈𝑌𝑖,𝑡 + 1,𝑑 ― 𝐼𝑁𝑆𝑇_𝑆𝐸𝐿𝐿𝑖,𝑡 + 1,𝑑)/𝑁𝑆𝑖,𝑡 + 1

(3)

𝑛

(4)

𝑛

(5)

𝐷_𝐼𝑁𝑆𝑇_𝑁𝐵𝑅𝑖,𝑡 + 1 = ∑𝑑 = 1(𝐷_𝐼𝑁𝑆𝑇_𝐵𝑈𝑌𝑖,𝑡 + 1,𝑑 ― 𝐷_𝐼𝑁𝑆𝑇_𝑆𝐸𝐿𝐿𝑖,𝑡 + 1,𝑑)/𝑁𝑆𝑖,𝑡 + 1 𝐹_𝐼𝑁𝑆𝑇_𝑁𝐵𝑅𝑖,𝑡 + 1 = ∑𝑑 = 1(𝐹_𝐼𝑁𝑆𝑇_𝐵𝑈𝑌𝑖,𝑡 + 1,𝑑 ― 𝐹_𝐼𝑁𝑆𝑇_𝑆𝐸𝐿𝐿𝑖,𝑡 + 1,𝑑)/𝑁𝑆𝑖,𝑡 + 1 𝑛

𝐼𝑁𝐷𝐼__𝑁𝐵𝑅𝑖,𝑡 + 1 = ∑𝑑 = 1(𝐼𝑁𝐷𝐼__𝐵𝑈𝑌𝑖,𝑡 + 1,𝑑 ― 𝐹_𝐼𝑁𝑆𝑇_𝑆𝐸𝐿𝐿𝑖,𝑡 + 1,𝑑)/𝑁𝑆𝑖,𝑡 + 1 (6)

3.2 Research models This study aims to examine whether individual firms’ AQ the components of AQ (innate and discretionary AQ) affect institutional investors’ information risk recognition using their net trading behavior. We verify this study’s hypotheses by setting up models (1) and (2). The dependent variables in models (1) and (2) represent institutional investors’ net trading behavior in year t+1 (i.e., the ratio of net purchase volume less net selling volume to the number of issued shares). We analyze trading behavior by institutional investor type. INST_NBR(t+1) indicates the net trading behavior exhibited by all institutional investors (i.e., the sum of domestic and foreign institutions). D_INST_NBR(t+1) and F_INST_NBR(t+1) indicate the net trading behavior displayed by domestic and foreign institutional investors. The explanatory variable in model (1) is the AQ (AQit) of firm i in year t, which is calculated according to equation (1). In this study, because AQit is measured using the volatility of abnormal accruals, we believe that a greater measured value of AQit for an individual firm indicates a lower AQ (i.e., higher firm-intrinsic information risk). Model (2) separates the AQ (AQit) of firm i in year t into the firm’s fundamental risk factor and the manager’s discretionary risk factor. DiscAQit describes the volatility of accruals due to the manager’s discretionary risk, and InnateAQit reflects that due to the firm’s fundamental risk. Thus, for both measures, a higher value again implies lower information quality. We predict that if institutional investors see AQ (or the components of AQ) as a risk factor, β1 in model (1) (or β1 and β2 in model (2)) should be statistically significantly negative.

𝐷𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 [(1)𝐼𝑁𝑆𝑇_𝑁𝐵𝑅𝑖,𝑡 + 1 (2)𝐷_𝐼𝑁𝑆𝑇_𝑁𝐵𝑅𝑖,𝑡 + 1 (3)𝐹_𝐼𝑁𝑆𝑇_𝑁𝐵𝑅𝑖,𝑡 + 1] 4

= 𝛽0 + 𝛽1𝐴𝑄𝑖𝑡 + 𝛾𝑗

∑𝐼𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑎𝑙 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠

𝑖𝑡

𝑗=1

9

+ 𝛿𝑘



𝑘=1

𝐹𝑖𝑟𝑚 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠𝑖𝑡 + 𝐹𝑖𝑟𝑚_𝐷 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦_𝐷 + 𝑌𝐸𝐴𝑅_𝐷 + 𝜀𝑖𝑡 Model (1) 𝐷𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 [(1)𝐼𝑁𝑆𝑇_𝑁𝐵𝑅𝑖,𝑡 + 1 (2)𝐷_𝐼𝑁𝑆𝑇_𝑁𝐵𝑅𝑖,𝑡 + 1 (3))𝐹_𝐼𝑁𝑆𝑇_𝑁𝐵𝑅𝑖,𝑡 + 1] 4

= 𝛽0 + 𝛽1𝐷𝑖𝑠𝑐𝐴𝑄𝑖𝑡 + 𝛽2𝐼𝑛𝑛𝑎𝑡𝑒𝐴𝑄𝑖𝑡 + 𝛾𝑗

∑𝐼𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑎𝑙 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠

𝑖𝑡

𝑗=1 9

+ 𝛿𝑘∑𝑘 = 1𝐹𝑖𝑟𝑚 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠𝑖𝑡 + 𝐹𝑖𝑟𝑚_𝐷 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦_𝐷 + 𝑌𝐸𝐴𝑅_𝐷 + 𝜀𝑖𝑡 Model (2) In this study, we analyze the relationship between AQ in year t and institutional investors’ trading behavior in year t+1 by including a variety of control variables in the regression equation. First, we include institutional control variables. Institutional investors may react differently to individual firm-level information depending on such factors as the number of shares owned, the trading volume, and the investment period. Accordingly, we control for the characteristics of institutional investors in individual firms. INST_TRit is the ratio of institutional investors' trading volume to the total trading volume of firm i during year t. We use this ratio as a proxy for the percentage of institutional investors’ total shares held in individual firms.12 INST_Qit is the log value of institutional investors’ (the sum of domestic and foreign institutions) total trading volume for firm i in year t. INST_Turn is the turnover rate of institutional investors’ total trading volume for firm i in year t and is measured as the ratio of institutional investors’ total trading volume of common stock to the weighted average of common stock shares outstanding.

12

We measure institutional investors’ (domestic and foreign institutions) total trading volume (the sum of the selling and purchasing volumes) for firm i in year t as the ratio to the total trading volume for all investors (individual investors, domestic institutions, and foreign institutions). We cannot use institutions’ total share ownership rate because this information is not announced in Korea. Thus, Chung et al. (2019) calculate institutional investors’ relative trading volume rate and use that as a proxy for the total share ownership rate. This measurement is appropriate because it exhibits similar relations to the determinants of the total share ownership rate. Thus, we follow their approach.

INST_BLOCK is the percentage of institutional investors that own 5% or more of firm i’s shares in year t. Additionally, if the dependent variables of models (1) and (2) reflect just domestic or foreign institutional investors, we include separately measures for that type as control variables for the traits of institutional investor types. D_INST_TRit (F_INST_TRit) is the relative proportion of domestic (foreign) institutional investors’ trading volume for firm i in year t. We estimate the sum of domestic (foreign) institutional investors’ selling and purchase volumes using the ratio to the selling and purchase volumes for all institutional investors. DINST_Qit (FINST_Qit) is the log value of the total trading volume of domestic (foreign) institutional investors in firm i in year t. DINST_Turnit (FINST_Turnit) is the turnover rate of domestic (foreign) institutional investors’ trading volume for firm i in year t, which is measured as described above. D_INST_BLOCKit (F_INST_BLOCKit) is the percentage of domestic (foreign) institutional investors that own 5% or more of firm i's shares in year t. Second, we include firm control variables. We include variables that reflect the traits of individual firms because institutional investors’ trading reactions can differ according to these characteristics. In this case, SIZEit indicates the size of the firm and is the log of the market value of the equity (one million won) of firm i in year t. BMit is firm i’s book-to-market ratio in year t. DEBTit is firm i’s total debt at the end of year t divided by total assets. ROAit is firm i’s net income in year t divided by total assets. ANALYSTit is the number of security companies that uploaded analyst reports on firm i in year t. BETAit is the slope of the regression equation that estimates the daily and market index returns according to the market model in year t. We also add the idiosyncratic return volatility and transaction costs as additional control variables in the primary regression model to identify their potential influences on institutional investors’ net trading behavior.13 STD_AbReturnit is the standard deviation of the daily

13 Mashruwala et al. (2006) show that the accruals anomaly documented by Sloan (1996) is concentrated in firms with high idiosyncratic stock return volatility. Moreover, the accruals anomaly is found for lowprice and low-volume stocks. Rajgopal and Venkatachalam (2011) find that deteriorating earnings quality is associated with higher idiosyncratic return volatility. These studies indicate that the pricing effect of

abnormal returns of firm i in year t. Daily abnormal returns are found by subtracting SIZE- and BMadjusted (i.e., firm size and book-to-market ratio) portfolio returns from firm i’s daily stock return on day d of year t. Liquidityit is the total trading volume of the common stock of firm i in year t divided by the weighted average of the common stock. PRICEit is the natural log of firm i’s stock price at the end of year t. Firm_D is a firm dummy to control for firm cluster effects, Industry_D is an industry dummy to control for industry effects, and Year_D is a year dummy to control for year effects. To verify hypothesis 1, which considers the relation between AQ and institutional investors’ net trading behavior, we perform the regression analyses in models (1) and (2) by setting institutional investors’ trading behavior in year t+1 as the dependent variable and including the characteristics of institutional investors as control variables. Hypothesis 2 examines whether the risk recognition of AQ (or its components) differs for different institutional investor types. Thus, the dependent variables in models (1) and (2) are the net trading behavior of domestic and foreign institutions in year t+1. We include specific variables for each type of institution as controls. Hypothesis 3 considers whether the link between AQ and institutional investors’ trading behavior creates differences in institutional investors’ recognition of firm-specific information risk under different macroeconomic conditions. Thus, we test models (1) and (2) for periods of favorable and unfavorable macroeconomic conditions. Considering the effect of the 2008 global financial crisis, we define 2008 to 2010 as a period of economic depression and the rest of our sample period as having favorable macroeconomic conditions. Finally, hypothesis 4 examines whether institutional investors’ risk recognition differs for firms that are part of large corporate groups. To verify this hypothesis, we refer to data provided by Korea’s regulatory institution (i.e., the Fair Trade

AQ may be influenced by idiosyncratic stock return volatility and transaction costs. Thus, we include these variables as controls in our regression model to analyze the incremental impact of AQ on institutional investors’ net trading behavior.

Commission) to split the sample into firms that are part of large corporate groups and those that are not, and we conduct a regression analysis following models (1) and (2).

3.3 Sample and descriptive statistics The sample period for this study is from 2003 to 2014. The measurement of AQ uses data from the previous five years, and institutional investors’ trading behaviors are calculated for year t+1, so the measurement period is from 1999 to 2015. We extract accounting, financial, stock, and trading volume data from 1999 to 2015 from FnGuide Data Guide Pro.14 We select sample firms from listed companies (KOSPI and KOSDAQ) during the sample period. To unify the measuring time, we only include firms with fiscal years ending in December, and we exclude firms in the financial industry that can disrupt the accounting figures. We also exclude firms with no collectible financial, stock, or trading volume data. Thus, our final sample includes 13,564 firm-year observations. Table 1 illustrates the descriptive statistics of the variables used for the research models described in Section 3.2. Explanations of the variables are provided in Appendix 1. [Insert Table 1 about here]

4. Results 4.1 Effects of AQ on trading behavior by investor type Table 2 shows whether institutional and individual investors’ net trading behavior in later years differs according to the degree of AQ. In Table 2, we categorize yearly AQ into quartiles based on its

14

FnGuide is a representative Korean company that provides accounting, financial, stock price, and trading volume data; analyst consensuses; and economic references. Data Guide Pro is the firm’s data extraction system. In Korea, many institutions, companies, and researchers use Data Guide Pro to analyze Korean companies. http://www.fnguide.com/

size, record the mean values of individual and institutional investors’ net trading behaviors in year t+1 for each group, and test their statistical significance using Newey-West t-statistics (Newey & West, 1987). Panel A shows institutional and individual investors’ average net trading behavior for each AQ quartile. Positive values indicate net purchases, and negative values indicate net sales. AQ1 includes firms with low accruals volatility (low information risk firms), and AQ4 includes firms with high accruals volatility (high information risk firms). In model (1), which describes institutional investors’ net trading behavior, we observe statistically significant net selling behavior for AQ4 (-0.012, t = -4.753) but not for AQ1 (0.001, t = 0.022). Moreover, the two groups exhibit significantly different trading behavior (-0.073, t = -4.627). This result shows that institutional investors are more likely to net sell stocks with lower AQ. In model (2), which describes individual investors’ net trading behaviors, we observe statistically significant net buying behavior for AQ4 (0.022, t = 7.925) but not for AQ1 (0.001, t = 0.022). Again, the two groups exhibit significantly different trading behavior (0.022, t = 11.450). Our findings in Panel A suggest that individual investors, who have less knowledge of firm-specific information risk, react systematically and favorably to firms’ reported earnings, whereas alert institutional investors associate low AQ with high firm-specific information risk. Panels B and C compare institutional and individual investors’ trading behavior by quartile group by decomposing AQ into discretionary and innate AQ. In Panels B and C, we define the quartiles and conduct the analysis in the same manner as in Panel A, but we focus on discretionary and innate AQ, respectively. In both models in Panel B, we find no statistical difference between the group with the highest accruals volatility (Disc AQ4) and that with the lowest accruals volatility (Disc AQ1) owing to discretionary risk. In contrast, in model (1) of Panel C, we observe a significant difference (-0.098, t = 6.227) at the 1% level between the group with the highest accruals volatility (Innate AQ4) and that with the lowest accruals volatility (Innate AQ1) owing to fundamental risk factor. This result demonstrates that institutional investors are more likely to sell stocks with high fundamental information risk on average. [Insert Table 2 about here]

Table 3 provides the results of the regression analysis of the relationship between AQ (AQ components) and institutional investors’ trading behavior. Models (1) and (3) include firm-specific variables as controls, and models (2) and (4) include institutional investor-specific variables as controls. In models (1) and (2), we find a significant negative link between AQ and institutional investors’ trading behavior at the 1% level. Thus, institutional investors have a greater tendency to net sell lower AQ firms. Furthermore, the coefficients of discretionary and innate AQ in models (3) and (4) are both statistically significant, and the regression coefficient of innate AQ is larger than that of discretionary AQ. Thus, although institutional investors recognize both the discretionary and innate components of AQ as risk factors, they react more sensitively to the innate component. Overall, the results in Tables 2 and 3 suggest that institutional investors tend to display more selling behavior for lower AQ firms, recognizing their greater risk. Our finding supports hypothesis 1 and prior studies that claim that AQ affects the cost of capital and the expected rate of return. [Insert Table 3 about here]

4.2 AQ and trading behavior by institutional investor type Tables 4 and 5 provide the results of examining hypothesis 2, which considers whether risk awareness of AQ varies across institutional investor types, by separating institutional investors’ trading behavior into that of domestic (D_INST_NBR(t+1)) and foreign institutions (F_INST_NBR(t+1)). Table 4 divides the sample into groups based on the degree of AQ (AQ components) and considers whether trading behavior differs on average across institutional investor types in each group. Panel A presents the mean trading behavior by institution type for AQ quartiles. We observe statistically significant net selling behavior for lower AQ firms (AQ4) among both domestic and foreign institutions (domestic institutions: 0.003, t = -3.919, foreign institutions: -0.009, t = -3.452) but not for higher AQ firms (AQ1). In addition,

we find that net selling behavior is significantly greater in the AQ4 group than in the AQ1 group for both domestic and foreign institutions. Panels B and C show the results of measuring the mean trading behavior of each type of institutional investor in each group by decomposing AQ into discretionary and innate AQ and comparing the differences. Panel B shows that the trading behavior of domestic and foreign institutions does not differ across the lowest (Disc AQ4) and highest discretionary AQ firms (Disc AQ1). In contrast, Panel C shows that net selling behavior is greater for the lowest innate AQ firms (innate AQ4) than for the highest such firms (innate AQ1) for both domestic and foreign institutions. Furthermore, the trading behavior gap across the groups is larger for foreign (-0.008, t= -2.892) than for domestic institutional investors (-0.005, t= -1.932). [Insert Table 4 about here] Table 5 gives the results of the regression analysis of the relationship between AQ (or AQ components) and the trading behavior of institutional investor types after controlling for institutional investor characteristics and firm-specific factors. Models (1) and (2) give the results of the regression analysis for domestic institutional investors’ trading behavior, and models (3) and (4) give the results for foreign institutional investors’ trading behavior. In model (1), the coefficient of AQ is statistically significant negative (-0.015, t = -2.351). Moreover, in model (2), the coefficient of innate AQ is significantly negative (-0.039, t = -3.842), but the coefficient of discretionary AQ is not different from zero. This finding reflects domestic institutional investors’ tendency to net sell lower AQ firms, and this relationship is driven by innate AQ. In model (3), the regression coefficient of AQ is statistically significantly negative (-0.036, t = -5.677). In addition, in model (4), innate (-0.063, t = -6.265) and discretionary (-0.021, t= -2.953) AQ have significant negative effects. This result suggests that foreign institutional investors recognize both innate and discretionary AQ as risk factors.

Overall, the findings in Table 5 demonstrate that foreign institutional investors are more sensitive than domestic institutional investors are. Specifically, foreign institutional investors’ trading behaviors vary depending on the manager’s discretionary risk factors, whereas those of domestic institutional investors do not. We presume that these differences are derived from differences in the independence and alternative investment ranges of domestic and foreign institutional investors. Thus, the results in Table 5 partially support hypothesis 2. [Insert Table 5 about here]

4.3 Macroeconomic conditions and AQ’s effect Table 6 provides the results of analyzing whether AQ (or its components) is related to institutional investors’ trading behavior even after controlling for institutional investors traits’ and firmspecific factors under different macroeconomic conditions. Panels A and B show the results for each macroeconomic condition. First, in Panel A-(1), which examines the effect of AQ under favorable economic conditions, the regression coefficient of AQ is statistically significantly negative for both institutional investor types. In Panel A-(2), which separates AQ into discretionary and innate AQ, the coefficients of both components are statistically significantly negative, and this negative link is stronger for foreign institutional investors. In Panel B-(3), which analyzes the effect of AQ during an economic depression, the relation between AQ and institutional investors’ trading behavior is not statistically significant (-0.023, t = -1.511) and is less negatively correlated (-0.023, t = -2.033) with foreign institutional investors’ trading behavior. In Panel B-(4), which tests the effect of AQ components, discretionary AQ has no relation to trading behavior for either institutional investor type, whereas we observe a significant negative link (-0.063, t = 3.579) between innate AQ and foreign institutional investors’ trading behavior.

Overall, institutional investors’ net selling behavior for lower AQ firms increases under both favorable and unfavorable macroeconomic conditions, but the net selling behavior of institutional investors is relatively greater in the former than in the latter period. This finding supports that of Kim and Qi (2010), who state that the pricing effect of AQ is greater under favorable macroeconomic conditions and contend that the pricing effect of AQ is attributable to the fundamental risk factor, which is associated with macroeconomic conditions. [Insert Table 6 about here]

4.4 Corporate governance structure and AQ’s effect The owner-managers of consolidated conglomerates in Korea essentially control all business activities of their member firms through cross shareholdings. The Korea Fair Trade Commission (KFTC) sets forth its own ownership and governance standards through which it determines a comprehensive monitoring list and strictly regulates the abusive business practices of firms on the list.15 Based on the KFTC’s provisions, we divide the sample into corporate group and non-corporate group subsamples and examine the effect of AQ and its components for each group. Table 7 presents the results of comparing institutional investors’ information risk recognition of AQ across corporate group firms that are connected by shares and firms that are not connected. Panel A of Table 7 describes institutional investors’ trading reactions pertaining to AQ and its components for corporate group firms. Panel A-(1) indicates no significant relation between AQ and the trading behavior of all institutional investors. Furthermore, we find no significant relationship between discretionary and innate AQ and all institutional investors’ trading behaviors in Panel A-(2).

The KFTC’s ownership standard targets a firm’s largest shareholder that owns, in combination with its affiliates, at least 30% of the firm’s shares. The KFTC’s governance standard applies to shareholders who are deemed to have absolute dominance over an individual company’s management. 15

Panel B in Table 7 shows institutional investors’ trading reactions pertaining to AQ and its components for non-corporate group firms. In Panel B-(3), the regression coefficient of AQ is statistically significantly negative for all institutional investors’ trading behaviors. In Panel B-(4), the regression coefficients for discretionary and innate AQ indicate significant negative relationships with institutional investors’ trading behavior (INST_NBR). This negative link is stronger for innate AQ and is more sensitive for foreign institutions. Our results indicate that multiple-segment large corporate group firms can adopt diversified business portfolios or establish internal capital markets to effectively lower firm-specific business risk and can also reduce managerial discretionary risk through rigorous internal and external control. Overall, the results in Table 7 support hypothesis 4, which argues that institutional investors perceive less information risk for diversified firms that are connected by shares. [Insert Table 7 about here]

4.5 Alternative measure of the cost of equity and its effect on AQ Our study relies on the assumption that institutional net trading behavior can capture firms’ cost of equity of firms and explain AQ. Thus, we employ alternative measures of the cost of equity based on previous studies and conduct the empirical analysis as a robustness check. We utilize Easton’s (2004) price-earnings to growth (PEG) model for the ex-ante cost of equity and Barber and Lyon’s (1997) size/growth-adjusted holding period abnormal return approach for the ex-post cost of equity. Specifically, Equation (7) describes Easton’s (2004) PEG model: 𝑅_𝑃𝐸𝐺𝑖,𝑡 + 1 = (𝐹𝐸𝑃𝑆𝑡 + 2 ― 𝐹𝐸𝑃𝑆𝑡 + 1)/𝑃𝑡

(7)

Here, R_PEGi,t is the ex-ante measure of the cost of equity. FEPt+1 and FEPt+2 are analyst forecasts of net income per share in years t+1 and t+2, respectively. The analyst forecast data are extracted from FnGuide Pro. Pt is the share price at the end of year t.

Equation (8) measures the proxy variable for the ex-ante cost of equity, that is, buy-andhold abnormal returns (BHAR) (BHARi,t+1) for firm i in year t+1, following Barber and Lyon (1997). 12

12

𝐵𝐻𝐴𝑅𝑖,𝑡 + 1 = ∏𝑘 = 1(1 + 𝑅𝑖,𝑡 + 1,𝑘) ― ∏𝑘 = 1(1 + 𝑅𝐵,𝑡 + 1,𝑘)

(8)

Similar to Barber and Lyon (1997), we utilize the size/growth adjusted BHAR. Ri,t+1,k is the stock return for month k in year t+1 for firm i, and RB,t+1,k is the corresponding 25 monthly portfolio returns sorted by size and growth. Table 8 provides the results based on alternative cost of equity measures. Models (1) and (2) analyze the effects of AQ and its components on the ex-ante cost of equity reflected by analyst forecasted earnings per share. We find that AQ and innate AQ are significantly positively related to the ex-ante cost of equity, whereas discretionary AQ is not significantly associated with this measure. Models (3) and (4) analyze the effect of AQ and its components on the expost cost of equity measured by BHAR. Similar to the findings based on the ex-ante cost of equity, we find that AQ and innate AQ have significantly negative relationships with BHAR but that discretionary AQ has an insignificant relationship with BHAR. Hence, firms with poor AQ have high costs of equity, leading to negative stock performances. Overall, the findings are consistent with those of previous related studies pertaining to the relationship between AQ and cost of equity and our results based on institutional trading behavior. [Insert Table 8 about here]

5. Summary and conclusions This study investigates whether AQ is useful as a substitute for firm-specific information risk and whether AQ affects investor pricing using institutional investors’ trading behavior. First, we find that

institutional investors’ net selling behavior increases for lower AQ firms. This result coincides with those of previous studies that argue that if a firm’s intrinsic information risk is not diversified, it can increase investors’ risk and, subsequently, the cost of capital. We also analyze the effects of the innate and discretionary factors of AQ and find that institutional investors’ net selling behavior increases for firms with higher innate risk. This result is consistent with prior studies that show that innate risk is more relevant to cost of capital than discretionary risk is. Thus, our result provides further evidence for AQ’s influence on price. In addition, we explore individual investors’ trade reactions to AQ and find that individual investors’ net buying volume increases for firms with lower AQ. This result suggests that individual investors, who are less aware of firm-specific information risk, react systematically and favorably to firms’ reported earnings. As the proportions of individual investors and information asymmetry are higher in emerging than in developed markets, stock price returns are more vulnerable to irregular noise in emerging markets. The results confirm the validity of institutional investors’ net trading behavior as an alternative proxy for stock price returns in emerging markets. Second, we find that foreign institutional investors are more sensitive to AQ and its components than domestic institutional investors are. Domestic institutional investors have relatively high conflicts of interest and more assets allocated to domestic firms, and foreign institutional investors are more independent and can make diversified international investments. Our result suggests that investors’ trading responses to information risk can differ depending on the independence of their investments and their ability to make alternative investments. Third, we show that the effects of AQ and its components differ across macroeconomic conditions in Korea. Institutional investors’ net selling behavior for lower innate AQ firms increased more under favorable macroeconomic conditions. This result is similar to that of Kim and Qi (2010), who state that AQ’s pricing effect is based on fundamental risk and that fundamental risk interacts with macroeconomic conditions. Furthermore, these results show that under favorable macroeconomic

conditions, managers’ tendency to make optimistic investments increases, leading to increased fundamental risk that investors cannot eliminate through diversification. Thus, institutional investors recognize this risk factor more strongly and reflect it in the cost of capital. Lastly, we do not observe institutional investors’ net selling behavior for lower AQ firms for corporate groups connected by shares. In Korea, non-related diversified large-scale corporate groups (e.g., chaebols) are connected directly or indirectly by shares. This share connection within corporate groups is assumed to be similar to non-related business diversification, which can reduce operating risk. Our results indicate that fundamental risk that cannot be eliminated can still be partially reduced by constructing a business portfolio or internal capital market and that institutional investors recognize a lower cost of capital for corporate groups. This study extends the existing literature by demonstrating the price effect of AQ using unique institutional investor volume data for the Korean stock market. Our results also provide new evidence that the price effect of AQ may vary depending on the investor type, economic situation, and corporate governance.

Our results have some meaningful economic and policy implications. First, we find that the different risk assessments of accounting information across the two investor types is likely to proliferate the transfer of wealth. To prevent the further exacerbation of such information asymmetry-driven transfers between these parties, market policies that emphasize the positive role of sell-side analysts are urgently needed. Second, until the early 2000s, Korean corporate groups continued to rely heavily on debt when adopting business portfolios, a speculative business strategy that brought both intended and unintended outcomes. In recent years, however, these corporate groups have been developing more equity-based business plans, effectively lowering business risk without raising financial risk. Our result indicates that although firmspecific information risk cannot entirely dissipate even in the most conservative, well-diversified

investment portfolio, residual risk can be at least partially eliminated if the investee firm adopts a wide-ranging corporate diversification strategy.

References Aboody, D., Hughes, J., & Liu, J. (2005). Earnings quality, insider trading, and cost of capital. Journal of Accounting Research, 43 (5), 651–673. An, O., Kim, J., & Kim, D. (2010). Is accounting information quality priced in Korea? Korean Journal of Financial Studies, 39 (1), 133–159. Bae, K. H., & Jeong, S. W. (2007). The value-relevance of earnings and book value, ownership structure, and business group affiliation: evidence from Korean business groups. Journal of Business Finance & Accounting, 34, 740–766. Barber, B., & Lyon, J. (1997). Detecting long-run abnormal stock returns: the empirical power and specification of test statistics. Journal of Financial Economics, 43, 341–372.

Barth, M., Konchitchki, Y., & Landsman, W. R. (2013). Cost of capital and earnings transparency. Journal of Accounting and Economics, 55, 206–224. Bhattacharya, N., Ecker, F., Olsson, P., & Schipper, K. (2012). Direct and mediated associations among earnings quality, information asymmetry, and the cost of equity. The Accounting Review, 87, 449–482. Botosan, C. A. (1997). Disclosure level and the cost of equity capital. The Accounting Review, 72 (3), 323–349. Botosan, C. A., & Plumlee, M. (2005). Assessing alternative proxies for the expected risk premium. The Accounting Review, 80 (January), 21–53. Chen, L. H., Dhaliwal, D. S., & Trombley, M. A. (2008). The effect of fundamental risk on the market pricing of accruals quality. Journal of Accounting, Auditing & Finance, 23 (4), 1– 29. Chung, C. Y., Kim, D., Kim, K. S., Lee, J. H., & Lee, K. J. (2019). Do institutional investors enhance accounting earnings attributes in the Korean market? Emerging Markets Finance & Trade, 55, Forthcoming. Cohen, D. (2008). Does information risk really matter? An analysis of the determinants and economic consequences of financial reporting quality. Asia-Pacific Journal of Accounting and Economics, 15 (2), 69–90. Core, J., Guay, W., & Verdi, R. (2008). Is accruals quality a priced risk factor? Journal of Accounting and Economics, 46 (1), 2–22.

Dechow, P., & Dichev, I. (2002). The quality of accruals and earnings: the role of accrual estimation errors. The Accounting Review, 77 (1), 35–59. Demirkan, S., Radhakrishnan, S., & Urcan, O. (2012). Discretionary accruals quality, cost of capital, and diversification. Journal of Accounting, Auditing and Finance, 27, 496–526. Duchin, R. (2010). Cash holdings and corporate diversification. Journal of Finance 65 (3), 955– 992. Easley, D., & O'Hara, M. (2004). Information and the cost of capital. Journal of Finance, 59 (August), 1553–1583. Easton, P. (2004). PE ratio, and estimating the implied expected rate of return on equity capital. Accounting Review, 79 (1), 73–95. Fee, E., Hadlock, E., & Pierce, J. (2009). Investment, financing constraints, and internal capital markets: evidence from the advertising expenditures of multinational firms. Review of Financial Studies, 22, 2361–2392. Ferris, S. P., Kim, K. A., & Kitsabunnarat, P. (2003). The costs (and benefits?) of diversified business groups: the case of Korean chaebols. Journal of Banking & Finance, 27, 251– 273. Francis, J., LaFond, R., Olsson, P., & Schipper, K. (2004). Cost of equity and earnings attributes. The Accounting Review, 79 (4), 976–1010. Francis, J., LaFond, R., Olsson, P., & Schipper, K. (2005). The market pricing of accruals quality. Journal of Accounting and Economics, 39 (2), 295–327.

George, R., & Kabir, R. (2012). Heterogeneity in business groups and the corporate diversification–firm performance relationship. Journal of Business Research, 65, 412– 420. Gray, P., Koh, P. S., & Tong, Y. (2009). Accruals quality, information risk and cost of capital: evidence from Australia. Journal of Business Finance & Accounting, 36, 51–72. Guay, W., Kothari, J. P., & Watts, R. L. (1996). A market-based evaluation of discretionary accrual models. Journal of Accounting Research, 34 (Supplement), 83–105. Jiraporn, P., Kim, Y. S., & Mathur, I. (2008). Does corporate diversification exacerbate or mitigate earnings management? An empirical analysis. International Review of Financial Analysis, 17, 1087–1109. Kang, J.-K., & Stulz, R. M. (1997). Why is there a home bias? An analysis of foreign portfolio equity ownership in Japan. Journal of Financial Economics, 46 (1), 3–28. Khan, M. (2008). Are accruals mispriced? Evidence from tests of an intertemporal capital asset pricing model. Journal of Accounting and Economics, 45 (1), 55–77. Khanna, T., & Palepu, K. (2000). Is group affiliation profitable in emerging markets? An analysis of diversified Indian business groups. Journal of Finance, 55, 867–891. Kim, D., & Qi, Y. (2010). Accruals quality, stock returns, and macroeconomic conditions. The Accounting Review, 85, 937–978. Kim, J. B., & Yi, C. H. (2006). Ownership structure, business group affiliation, listing status, and earnings management: Evidence from Korea. Contemporary Accounting Research, 23, 427–464.

Lambert, R., Leuz, C., & Verrecchia, R. (2007). Accounting information, disclosure and the cost of capital. Journal of Accounting Research, 45 (2), 385–420. Lamont, O. (1997). Cash flow and investment: evidence from internal capital markets. Journal of Finance, 52, 83–110. Lee, S., Park, K., & Shin, H. (2009). Disappearing internal capital markets: Evidence from diversified business groups in Korea. Journal of Banking and Finance, 33, 326–334. Mashruwala, C., Rajgopal, S., & Shevlin, T. (2006). Why is the accrual anomaly not arbitraged away? The role of idiosyncratic risk and transaction costs. Journal of Accounting and Economics, 42(1-2), 3-33. McInnis, J. (2010). Earnings smoothness, average returns, and implied cost of equity capital. The Accounting Review, 85 (1), 315–341. Mohanram, P., & Rajgopal, S. (2009). Is PIN priced risk? Journal of Accounting and Economics, 47 (3), 226–243. Newey, W. K., & West, K. D. (1987). Hypothesis testing with efficient method of moments estimation. International Economic Review, 777-787. Ogneva, M. (2012). Accrual quality, realized returns, and expected returns: the importance of controlling for cash flow shocks. The Accounting Review, 87 (4), 1415–1444. Rajgopal, S., & Venkatachalam, M. (2011). Financial reporting quality and idiosyncratic return volatility. Journal of Accounting and Economics, 51(1-2), 1-20. Scharfstein, D. & Stein, J. (2000). The dark side of internal capital markets: divisional rent seeking and inefficient investment. Journal of Finance, 55, 2537–2564.

Shin, H. & Park, Y. (1999). Financing constraints and internal capital markets: evidence from Korean “chaebols.” Journal of Corporate Finance, 5, 169–191. Shin, H. & Stulz, R. (1998). Are internal capital markets efficient? Quarterly Journal of Economics, 113, 531–552. Sloan, R. G. (1996). Do stock prices fully reflect information in accruals and cash flows about future earnings?. Accounting review, 289-315. Subramanyam, K. R. (1996). The pricing of discretionary accruals. Journal of Accounting Research, 22 (1–3), 249–281. Tesar, L. L., & Werner, I. M. (1995). Home bias and high turnover. Journal of International Money and Finance, 14 (4), 467–492. Yee, K. K. (2006). Earnings quality and the equity risk premium: a benchmark model. Contemporary Accounting Research, 23 (3), 833–877.

Table 1. Descriptive statistics Mean

Std.

Min.

25%

50%

75%

Max.

Dependent and explanatory variables INST_NBR(t+1)

-0.005

0.060

-0.832

-0.019

0.000

0.013

0.464

D_INST_NBR(t+1)

-0.001

0.045

-0.667

-0.010

0.000

0.007

0.375

F_INST_NBR(t+1)

-0.004

0.045

-0.832

-0.007

0.000

0.006

0.436

INDI_NBR(t+1)

0.010

0.072

-0.571

-0.017

0.001

0.029

1.005

AQ

0.093

0.068

0.000

0.046

0.073

0.119

0.360

-0.005

0.053

-0.206

-0.035

-0.011

0.016

0.198

0.098

0.049

-0.074

0.062

0.087

0.121

0.274

DiscAQ InnateAQ

Institutional investor-specific control variables INST_TR

0.118

0.161

0.000

0.017

0.039

0.152

0.619

D_INST_TR

0.066

0.097

0.000

0.001

0.013

0.095

0.366

F_INST_TR

0.050

0.071

0.000

0.009

0.023

0.053

0.306

INST_TQ

6.031

1.152

0.000

5.531

6.206

6.741

8.966

DINST_TQ

5.078

1.821

0.000

4.468

5.490

6.249

8.759

FINST_TQ

5.639

1.411

0.000

5.133

5.947

6.508

8.604

INST_Turn

0.207

0.253

0.000

0.039

0.114

0.272

2.831

DINST_Turn

0.098

0.160

0.000

0.003

0.027

0.120

1.663

FINST_Turn

0.109

0.132

0.000

0.017

0.063

0.153

1.560

INST_BLOCK

0.045

0.098

0.000

0.000

0.000

0.061

1.063

D_INST_BLOCK

0.032

0.084

0.000

0.000

0.000

0.000

1.000

F_INST_BLOCK

0.013

0.051

0.000

0.000

0.000

0.000

0.811

SIZE

7.882

0.663

6.149

7.434

7.766

8.210

9.921

BM

1.354

1.105

-4.785

0.618

1.086

1.778

7.426

DEBT

0.433

0.215

0.001

0.266

0.432

0.583

1.364

ROA

-0.003

0.176

-1.025

-0.016

0.029

0.070

0.998

ANALYST

0.232

0.382

0.000

0.000

0.000

0.301

1.519

BETA

4.880

0.662

3.149

4.433

4.764

5.207

6.916

Std_AbRETURN

0.031

0.013

0.008

0.022

0.029

0.038

0.094

Firm-specific control variables

Liquidity

0.339

0.644

-1.878

-0.088

0.313

0.728

4.065

PRICE

3.740

0.665

1.813

3.280

3.672

4.158

6.346

Notes: The definitions of variables are presented in Appendix 1.

Table 2. AQ (AQ component) levels and trading behavior by investor type (institutions and individuals) Panel A. AQ and institutional investors’ trading behavior INST_NBR(t+1)  INDI_NBR(t+1)  AQ1 (Low Information Risk) (n=3,391) 0.001 0.001* (0.022) (0.531) AQ 2 (n=3,391) -0.003 0.006*** (-1.516) (2.838) AQ 3 (n=3,391) -0.003 0.007*** (-1.408) (2.595) AQ4 (High Information Risk) (n=3,391) -0.012*** 0.022*** (-4.753) (7.925) *** Difference (4 - 1) -0.073 0.022*** (-4.627) (11.450) Panel B. Discretionary AQ and institutional investors’ trading behavior INST_NBR(t+1)  INDI_NBR(t+1)  Disc AQ1 (Low Discretionary Information Risk) (n=3,391)

-0.007***

(-2.737) Disc AQ 2 (n=3,391) -0.002 (-0.075) Disc AQ 3 (n=3,391) -0.002 (-1.044) Disc AQ4 (High Discretionary Information Risk) -0.008*** (n=3,391) (-2.938) Difference (4 - 1) 0.003 (0.395) Panel C. Innate AQ and institutional investors’ trading behavior INST_NBR(t+1) 

0.013*** (6.017) 0.005* (1.829) 0.005* (1.869) 0.014*** (5.362) 0.000 (0.335) INDI_NBR(t+1) 

Innate AQ1 (Low Fundamental Information Risk) Innate AQ 2

(n=3,391) (n=3,391)

Innate AQ 3

(n=3,391)

Innate AQ4 (High Fundamental Information Risk) (n=3,391) Difference (4 - 1)

-0.000

0.000

(-0.048) -0.000 (-0.018) -0.005*** (-3.145) -0.013*** (-4.648) -0.098*** (-6.227)

(0.299) 0.001 (0.654) 0.010*** (4.389) 0.025*** (8.806) 0.025*** (12.999)

Notes: Values in parentheses are Newey-West t-statistics. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Table 3. Effect of AQ and its components on institutional trading behaviors Dependent Variable: INST_NBR(t+1) (1)   (2)   (3) (4) Coeff. (t-value)  Coeff. (t-value)  Coeff. (t-value)  Coeff. (t-value) Intercept -0.140 -0.218* -0.128 -0.206* (-1.217) (-1.914) (-1.114) (-1.805) AQ -0.051*** -0.052*** (-5.942) (-6.130) DiscAQ -0.030*** -0.027*** (-3.092) (-2.862) InnateAQ -0.094*** -0.104*** (-6.863) (-7.666) INST_R -0.061*** -0.062*** (-9.828) (-10.071) INST_Q 0.003*** 0.003*** (3.277) (3.094) *** INST_Turn -0.034 -0.034*** (-10.075) (-10.043) *** INST_BLOCK -0.037 -0.037*** (-7.282) (-7.351) SIZE 0.050 0.056 0.049 0.055 (1.304) (1.463) (1.272) (1.435) BM -0.001 -0.001 -0.001** -0.001* (-1.547) (-1.182) (-2.072) (-1.838) DEBT -0.023*** -0.021*** -0.023*** -0.022*** (-8.803) (-8.368) (-8.858) (-8.461) ROA 0.012*** 0.010*** 0.010*** 0.007** (3.554) (3.053) (2.840) (2.171) ANALYST -0.007*** 0.010*** -0.007*** 0.011*** (-3.329) (4.194) (-3.242) (4.370) BETA -0.051 -0.051 -0.051 -0.050 (-1.325) (-1.342) (-1.308) (-1.322) Std_AbRETURN -0.342*** -0.336*** -0.322*** -0.315*** (-5.205) (-5.113) (-4.887) (-4.780) Liquidity 0.003*** 0.002* 0.003*** 0.003* (2.459) (1.654) (2.638) (1.862) *** *** *** PRICE 0.004 0.012 0.004 0.011*** (3.162) (7.193) (2.937) (6.935) Firm fixed effects Included Included Included Included Industry fixed effects Included Included Included Included Year fixed effects Included Included Included Included N 13,564 13,564 13,564 13,564 Adj. R2   0.045 0.046 0.069 Notes: *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. The definitions of variables are presented in Appendix 1.

Table 4. AQ level and trading behavior by institutional investor type Panel A. AQ and institutional investors’ trading behavior D_INST_NBR(t+1) AQ1 (Low Information Risk)

AQ2

AQ3

(n=3,391)

0.001

-0.000

(0.290)

(-0.123)

-0.001

-0.002

(-0.708)

(-0.991)

-0.001

-0.002

(-1.449)

(-0.836)

(n=3,391)

(n=3,391)

AQ4 (High Information Risk)

F_INST_NBR(t+1)

(n=3,391)

-0.003*** (-3.919)

Difference (4 - 1)

-0.004* (-1.957)

-0.009*** (-3.452) -0.009*** (-4.183)

Panel B. Discretionary AQ and institutional investors’ trading behavior D_INST_NBR(t+1) Disc AQ1 (Low Discretionary Information Risk) (n=3,391) Disc AQ2

Disc AQ3

-0.002**

-0.005

(-2.019)

(-1.560)

-0.001

-0.001

(-0.879)

(-0.370)

0.001

-0.003

(0.190)

(-1.596)

(n=3,391)

(n=3,391)

Disc AQ4 (High Discretionary Information Risk) (n=3,391)

F_INST_NBR(t+1)

-0.002**

-0.005**

(-2.361)

(-2.045)

-0.000

-0.001

(-0.322)

(-0.972)

Difference (4 - 1)

Panel C. Innate AQ and institutional investors’ trading behavior D_INST_NBR(t+1) Innate AQ1 (Low Fundamental Information Risk) (n=3,391)

F_INST_NBR(t+1)

0.002

-0.002

(0.577)

(-0.752)

Innate AQ2

Innate AQ3

(n=3,391)

(n=3,391)

-0.000

0.000

(-0.222)

(0.304)

-0.004*** (-3.143)

Innate AQ4 (High Fundamental Information Risk) (n=3,391) Difference (4 - 1)

-0.003*** (-3.363) -0.005* (-1.932)

-0.002 (-1.025) -0.010*** (-3.283) -0.008*** (-2.892)

Notes: Values in parentheses are Newey-West t-statistics. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Table 5. Effect of AQ and its components on institutional trading behaviors by national origin Net Selling Volume by Institutional Investor Type D_INST_NBR(t+1) F_INST_NBR(t+1) (1)  (2) (3) 4) Coeff. (t-value)  Coeff. (t-value)  Coeff. (t-value)  Coeff. (t-value) Intercept -0.100 -0.094 -0.123 -0.116 (-1.158) (-1.090) (-1.444) (-1.362) AQ -0.015** -0.036*** (-2.351) (-5.677) DiscAQ -0.005 -0.021*** (-0.747) (-2.953) *** InnateAQ -0.039 -0.063*** (-3.842) (-6.265) *** *** D_INST_TR -0.039 -0.040 (-5.698) (-5.867) D_INST_TQ 0.000 0.000 (0.704) (0.670) D_INST_Turnover -0.057*** -0.056*** (-14.826) (-14.733) D_INST_BLOCK -0.035*** -0.036*** (-7.728) (-7.804) F_INST_TR -0.113*** -0.113*** (-11.983) (-12.075) F_INST_TQ 0.001** 0.001** (2.450) (2.273) F_INST_Turnover -0.038*** -0.038*** (-8.550) (-8.586) *** F_INST_BLOCK -0.090 -0.089*** (-13.296) (-13.210) SIZE 0.023 0.022 0.032 0.031 (0.795) (0.776) (1.111) (1.089) BM 0.000 -0.001 0.000 -0.001 (-0.705) (-1.123) (-0.873) (-1.315) DEBT -0.010*** -0.011*** -0.011*** -0.011*** (-5.375) (-5.440) (-5.776) (-5.818) * *** ROA -0.003 -0.005 0.014 0.012*** (-1.298) (-1.804) (5.776) (4.823) ANALYST 0.009*** 0.009*** 0.008*** 0.008*** (5.022) (5.084) (4.621) (4.745) BETA -0.017 -0.017 -0.028 -0.011 (-0.591) (-0.581) (-0.992) (-0.976) Std_AbRETURN -0.289*** -0.278*** -0.071 -0.059 (-5.808) (-5.569) (-1.424) (-1.181) Liquidity 0.003*** 0.003*** 0.001 0.002 (3.618) (3.682) (1.137) (1.360) PRICE 0.005*** 0.005*** 0.006*** 0.005***

(5.386) (5.247) (5.098) (4.858) Firm fixed effects Included Included Included Included Industry fixed effects Included Included Included Included Year fixed effects Included Included Included Included N 13,564 13,564 13,564 13,564 Adj. R2   0.057 0.057 0.080 0.081 Notes: *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. The definitions of variables are presented in Appendix 1.

Table 6. Relationship between institutional investors’ trading behavior and AQ by macroeconomic conditions Panel A-(1). Favorable economic period: Effect of AQ Net purchasing behavior by institutional investor type in year t+1

Intercept

AQ

INST_NBR

D_INST_NBR

F_INST_NBR

Coeff. (t-value)

Coeff. (t-value)

Coeff. (t-value)

-0.173

-0.065

-0.105

(-1.322)

(-0.650)

(-1.069)

-0.064*** (-6.334)

INST_TRK

-0.060*** (-8.253)

INST_TQK

0.005*** (4.737)

INST_TurnoverK

-0.038*** (-9.425)

INST_BLOCKK

-0.028***

-0.022*** (-2.881) -0.035*** (-4.285) 0.001 (1.398) -0.064*** (-13.840) -0.028***

-0.041*** (-5.407) -0.126*** (-11.419) 0.002*** (3.898) -0.035*** (-6.649) -0.098***

(-4.732)

(-5.315)

(-11.631)

Firm control variables

Included

Included

Included

Firm fixed effects

Included

Included

Included

Industry fixed effects

Included

Included

Included

Year fixed effects

Included

Included

Included

N

9,910

9,910

9,910

Adj. R2

0.076

0.063

0.088

Panel A-(2). Favorable economic period: Effect of AQ components (discretionary and innate AQ) Net purchasing behavior by institutional investor type in year t+1 INST_NBR

D_INST_NBR

F_INST_NBR

Coeff. (t-value)

Coeff. (t-value)

Coeff. (t-value)

Intercept

DiscAQ

-0.160

-0.057

-0.100

(-1.221)

(-0.569)

(-1.017)

-0.036*** (-3.102)

InnateAQ

-0.125*** (-7.591)

INST_TRK

-0.062*** (-8.475)

INST_TQK

0.005*** (4.542)

INST_TurnoverK

-0.038*** (-9.355)

INST_BLOCKK

-0.029***

-0.006 (-0.719) -0.059*** (-4.725) -0.037*** (-4.497) 0.000 (1.325) -0.064*** (-13.692) -0.029***

-0.029*** (-3.335) -0.064*** (-5.225) -0.127*** (-11.478) 0.002*** (3.761) -0.035*** (-6.661) -0.097***

(-4.800)

(-5.405)

(-11.583)

Firm control variables

Included

Included

Included

Firm fixed effects

Included

Included

Included

Industry fixed effects

Included

Included

Included

Year fixed effects

Included

Included

Included

N

9,910

9,910

9,910

Adj. R2

0.078

0.065

0.089

Continuous Panel B-(3). Economic depression: Effect of AQ Net purchasing behavior by institutional investor type in year t+1

Intercept

INST_NBR

D_INST_NBR

F_INST_NBR

Coeff. (t-value)

Coeff. (t-value)

Coeff. (t-value)

-0.177

-0.172

-0.313

AQ

INST_TRK

(-1.334)

(-1.006)

-0.023

0.000

(-1.511)

(0.042)

-0.053*** (-4.431)

INST_TQK

0.005** (-2.379)

INST_TurnoverK

0.092*** (-4.655)

INST_BLOCKK

-0.001***

(-0.982) -0.023** (-2.033)

-0.050*** (-3.939)

-0.043** (-2.320)

0.000 (-0.624)

-0.003** (-2.348)

-0.040*** (-5.943)

-0.055*** (-6.448)

-0.058***

-0.074***

(-5.900)

(-6.422)

(-6.665)

Firm control variables

Included

Included

Included

Firm fixed effects

Included

Included

Included

Industry fixed effects

Included

Included

Included

Year fixed effects

Included

Included

Included

N

3,654

3,654

3,654

Adj. R2

0.056

0.045

0.073

Panel B-(4). Economic depression: Effect of AQ components (discretionary and innate AQ) Net purchasing behavior by institutional investor type in year t+1

Intercept

DiscAQ

InnateAQ

INST_NBR

D_INST_NBR

F_INST_NBR

Coeff. (t-value)

Coeff. (t-value)

Coeff. (t-value)

-0.299

-0.181

-0.154

(-1.274)

(-1.023)

(0.380)

-0.006

-0.004

-0.001

(-0.365)

(-0.356)

(-0.107)

-0.056** (-2.396)

0.007 (0.397)

-0.063*** (-3.579)

INST_TRK

-0.054*** (-4.545)

INST_TQK

-0.005** (-2.421)

INST_TurnoverK

-0.029*** (-4.669)

INST_BLOCKK

-0.055***

-0.049*** (-3.894) 0.000 (-0.625) -0.040 (-5.948)*** -0.058***

-0.045** (-2.429) -0.003** (-2.449) -0.055*** (-6.499) -0.073***

(-5.921)

(-6.402)

(-6.567)

Firm control variables

Included

Included

Included

Firm fixed effects

Included

Included

Included

Industry fixed effects

Included

Included

Included

Year fixed effects

Included

Included

Included

N

3,654

3,654

3,654

Adj. R2

0.057

0.045

0.075

Notes: *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. The definitions of variables are presented in Appendix 1.

Table 7. Relationship between and AQ and institutional investors’ trading behavior by corporate governance Panel A-(1). Corporate group firms: Effect of AQ Net purchasing behavior by institutional investor type in year t+1

Intercept

AQ

INST_TRK

INST_TQK

INST_TurnoverK

INST_NBR

D_INST_NBR

F_INST_NBR

Coeff. (t-value)

Coeff. (t-value)

Coeff. (t-value)

-0.548

-0.229

-0.294

(-1.171)

(-0.599)

(-0.832)

0.008

0.011

-0.005

(0.271)

(0.445)

(-0.214)

-0.074***

-0.062***

-0.106***

(-4.356)

(-3.202)

(-4.448)

0.000

0.000

0.000

(-0.023)

(-0.021)

(0.129)

-0.016**

-0.030***

-0.021*

(-2.252)

(-3.424)

0.001

-0.017

(0.098)

(-1.582)

(-4.956)

Firm control variables

Included

Included

Included

Firm fixed effects

Included

Included

Included

Industry fixed effects

Included

Included

Included

Year fixed effects

Included

Included

Included

N

1,951

1,951

1,951

Adj. R2

0.066

0.088

0.121

INST_BLOCKK

(-1.863) -0.123***

Panel A-(2). Corporate group firms: Effect of AQ components (discretionary and innate AQ) Net purchasing behavior by institutional investor type in year t+1

Intercept

INST_NBR

D_INST_NBR

F_INST_NBR

Coeff. (t-value)

Coeff. (t-value)

Coeff. (t-value)

-0.545

-0.231

-0.290

DiscAQ

InnateAQ

INST_TRK

INST_TQK

INST_TurnoverK

(-1.162)

(-0.604)

(-0.818)

0.025

0.012

0.012

(0.699)

(0.413)

(0.437)

-0.006

0.014

-0.021

(-0.143)

(0.393)

(-0.643)

-0.074***

-0.062***

-0.106***

(-4.366)

(-3.192)

(-4.437)

0.000

0.000

0.000

(-0.050)

(-0.018)

(0.022)

-0.016**

-0.030***

-0.021*

(-2.244)

(-3.429)

0.000

-0.017

(0.035)

(-1.567)

(-4.892)

Firm control variables

Included

Included

Included

Firm fixed effects

Included

Included

Included

Industry fixed effects

Included

Included

Included

Year fixed effects

Included

Included

Included

N

1,951

1,951

1,951

Adj. R2

0.065

0.087

0.121

INST_BLOCKK

(-1.845) -0.121***

Panel B-(3). Non-corporate group firms: Effect of AQ Net purchasing behavior by institutional investor type in year t+1

Intercept

AQ

INST_NBR

D_INST_NBR

F_INST_NBR

Coeff. (t-value)

Coeff. (t-value)

Coeff. (t-value)

-0.191

-0.091

-0.107

(-1.631)

(-1.053)

(-1.229)

-0.056*** (-6.397)

INST_TRK

-0.049***

-0.016** (-2.530) -0.033***

-0.039*** (-5.904) -0.106***

(-6.893) INST_TQK

0.003*** (2.692)

INST_TurnoverK

-0.044*** (-10.643)

INST_BLOCKK

-0.044***

(-4.329) 0.000 (0.658) -0.071*** (-15.487) -0.039***

(-9.544) 0.001*** (2.071) -0.043*** (-8.743) -0.087***

(-7.938)

(-7.728)

(-12.375)

Firm control variables

Included

Included

Included

Firm fixed effects

Included

Included

Included

Industry fixed effects

Included

Included

Included

Year fixed effects

Included

Included

Included

11,613

11,613

11,613

0.073

0.059

0.079

N Adj. R2

Panel B-(4). Non-corporate group firms: Effect of AQ components (discretionary and innate AQ) Net purchasing behavior by institutional investor type in year t+1

Intercept

DiscAQ

INST_NBR

D_INST_NBR

F_INST_NBR

Coeff. (t-value)

Coeff. (t-value)

Coeff. (t-value)

-0.179

-0.086

-0.101

(-1.531)

(-0.987)

(-1.154)

-0.032*** (-3.210)

InnateAQ

-0.110*** (-7.764)

INST_TRK

-0.051*** (-7.111)

INST_TQK

0.002** (2.502)

INST_TurnoverK

-0.043***

-0.006 (-0.877) -0.043*** (-4.114) -0.035*** (-4.504) 0.000 (0.595) -0.070***

-0.024*** (-3.268) -0.067*** (-6.282) -0.107*** (-9.623) 0.001* (1.924) -0.043***

(-10.609) INST_BLOCKK

-0.045***

(-15.396)

(-8.777)

-0.040***

-0.087***

(-7.967)

(-7.768)

(-12.325)

Firm control variables

Included

Included

Included

Firm fixed effects

Included

Included

Included

Industry fixed effects

Included

Included

Included

Year fixed effects

Included

Included

Included

11,613

11,613

11,613

0.075

0.060

0.080

N Adj. R2

Notes: *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. The definitions of variables are presented in Appendix 1. Table 8. Relationship between and AQ and the cost of equity Cost of Equity Ex-ante COE (R_PEG) Ex-post COE (BHAR) (1) (2) (3) (4) Coeff. (t-value)  Coeff. (t-value)  Coeff. (t-value)  Coeff. (t-value) Intercept 0.125 0.092 -1.246 -1.220 (0.390) (0.287) (-1.071) (-1.048) AQ 0.049 -0.180 (1.992)** (-2.086)** Innate AQ 0.136 -0.289 (3.653)*** (-2.096)** Disc AQ 0.010 -0.133 (0.366) (-1.367) SIZE 0.037 0.043 0.543 0.541 (0.342) (0.397) (1.399) (1.392) BM 0.005 0.006 -0.023 -0.023 (2.791)*** (3.257)*** (-3.691)*** (-3.795)*** DEBT 0.082 0.083 0.012 0.012 (11.681)*** (11.810)*** (0.479) (0.450) ROA -0.048 -0.046 0.079 0.073 *** *** ** (-3.245) (-3.113) (2.277) (2.070)** INST_TR -0.029 -0.025 -0.174 -0.178 *** * *** (-2.267) (-1.931) (-2.990) (-3.054)*** ANALYST 0.000 -0.001 0.104 0.105 *** (0.024) (-0.142) (4.530) (4.567)*** BETA -0.061 -0.065 -0.563 -0.561 (-0.565) (-0.609) (-1.445) (-1.441) Std_AbRETURN 0.166 0.115 -4.049 -4.000

(0.671) (0.465) (-6.094)*** (-6.005)*** Liquidity 0.013 0.012 -0.054 -0.053 (3.235)*** (3.033)*** (-4.614)*** (-4.582)*** PRICE -0.001 -0.001 -0.039 -0.040 (-0.320) (-0.294) (-3.030)*** (-3.068)*** Firm fixed effects Included Included Included Included Industry fixed effects Included Included Included Included Year fixed effects Included Included Included Included N 3,217 3,217 13,564 13,564 Adj. R2   0.240 0.202 0.034 0.034 Notes: *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. The definitions of variables are presented in Appendix 1.

Appendix 1. Description of Variables Variable Definition Dependent Variables INST_NBRi,t+1 The institutional (both domestic and foreign) net stock purchase ratio for firm i in year t+1. The net stock purchase ratio is calculated as the yearly sum of daily institutional net common stock purchases (purchasing volume minus selling volume) for firm i in year t+1 divided by the number of total outstanding shares. D_INST_NBRi,t+1 The net stock purchase ratio of firm i’s stock in year t+1 for domestic institutions. The net stock purchase ratio is calculated as the yearly sum of daily domestic institutional net common stock purchases (purchasing volume minus selling volume) for firm i in year t+1 divided by the number of total outstanding shares. F_INST_NBRi,t+1 The net stock purchase ratio of firm i’s stock in year t+1 for foreign institutions. The net stock purchase ratio is calculated as the yearly sum of daily foreign institutional net common stock purchases (purchasing volume minus selling volume) for firm i in year t+1 divided by the number of total outstanding shares. INDI_NBRi,t+1 The individual net stock purchase ratio of firm i’s stock in year t+1. The net stock purchase ratio is calculated as the yearly sum of daily individual net common stock purchases (purchasing volume minus selling volume) for firm i in year t+1 divided by the number of total outstanding shares. Explanatory variables AQit Estimated overall AQ for firm i in year t. Based on Francis et al. (2004), we measure the standard deviation of yearly abnormal accruals during the last five years. DiscAQit Estimated discretionary AQ for firm i in year t. Based on Francis et al. (2004), we subtract the portion affected by management discretion (i.e., information risk) from overall AQ. InnateAQit Estimated innate AQ for firm i in year t. Based on Francis et al. (2004), we subtract the portion affected by innate risk from business operations (i.e., fundamental risk) from overall AQ. Institutional Characteristic Control Variables INST_TRit The proportion of institutional (both domestic and foreign) trading volume of firm i in year t. We estimate this proportion as the yearly sum of the institutional trading volume (sum of purchasing volume and selling volume) divided by the total trading volume of firm i in year t. D_INST_TRit The proportion of domestic institutional trading volume of firm i in year t. We estimate the proportion as the yearly sum of domestic institutions’ trading volume (sum of purchasing volume and selling volume) divided by the total trading volume for firm i in year t.

F_INST_TRit

The proportion of foreign institutional trading volume of firm i in year t. We estimate the proportion as the yearly sum of foreign institutions’ trading volume (sum of purchasing volume and selling volume) divided by the total trading volume for firm i in year t. INST_Qit The log of institutional (both domestic and foreign) investors’ total trading volume of firm i’s common stock in year t. D_INST_Qit The log of domestic institutional investors’ total trading volume of firm i’s common stock in year t. F_INST_Qit The log of foreign institutional investors’ total trading volume of firm i’s common stock in year t. INST_Turnit The institutional (both domestic and foreign) turnover ratio of firm i’s common stock in year t. It is the ratio of institutional investors’ trading volume of firm i’s common stock to total outstanding shares. D_INST_Turnit Domestic institutions’ turnover ratio of firm i’s common stock in year t. It is the ratio of domestic institutional investors’ trading volume of firm i’s common stock to total outstanding shares. F_INST_Turnit Foreign institutions’ turnover ratio of firm i’s common stock in year t. It is the ratio of foreign institutional investors’ trading volume of firm i’s common stock to total outstanding shares. INST_BLOCKit The ownership level for institutional (both domestic and foreign) investors who hold at least 5% of firm i’s outstanding shares at the end of year t. D_INST_BLOCKit The ownership level for domestic institutional investors who hold at least 5% of firm i’s outstanding shares at the end of year t. F_INST_BLOCKit The ownership level for foreign institutional investors who hold at least 5% of firm i’s outstanding shares at the end of year t. Firm Characteristic Control Variables SIZEit The log of the market value (in millions) at the end of year t, indicating the size of firm i. The ratio of the book value of equity to its market value for firm i in year t. BMit DEBTit

The ratio of total liabilities to total assets for firm i in year t.

ROAit

The ratio of net income to total assets for firm i in year t.

ANALYSTit

The log of the sum of one and the estimated number of analysts covering firm i in year t.

BETAit

The estimated beta coefficient for firm i in year t. It is based on the market regression model that uses firm i’s daily stock returns and market index returns in year t. The standard deviation of daily abnormal returns for firm i in year t. Daily abnormal returns are found by subtracting SIZE-and BM-adjusted (i.e., firm size and book-to-market ratio) portfolio returns from firm i’s daily stock return on day d of year t.

Std_AbRETURNit

Liquidityit PRICEit Firm fixed effect Industry_fixed effect YEAR fixed effect

The log of the yearly sum of the daily trading volume divided by total outstanding shares for firm i in year t. The log of firm i’s stock price at the end of year t. Firm dummy to control for firm cluster effects. Industry dummy to control for industry effects. Year dummy to control for year effects.

Alternative Measures for cost of equity

R_PEG i,t

Implied cost of equity based on Easton’s (2004) PEG model in year t+1 for firm i.

BHAR i,t+1

Size/growth-adjusted BHAR in year t+1 for firm i.