Abacus
Early View
Original Article
Open Access

Does Financial Statement Comparability Reduce Differences in Sentiment-induced Investor Trading Behaviour?

Eun Hye Jo, 

Eun Hye Jo

George Mason University, Korea

Search for more papers by this author
Jung Wha (Jenny) Lee, 

Corresponding Author

Jung Wha (Jenny) Lee

George Mason University, Korea

[email protected]

Search for more papers by this author
First published: 17 June 2025
https://doi.org/10.1111/abac.12371

Abstract

This study examines whether comparable financial information can mitigate differences between individual and institutional investors’ trading behaviour, particularly behaviour that is shaped by investor sentiment. The results indicate that the higher the comparability, the smaller the gap in trading behaviour driven by investor sentiment between institutional and individual investors. In particular, the tendency of individual investors to purchase stocks associated with high investor sentiment is mitigated by comparable financial information. However, the study also reveals that comparability does not significantly affect institutional investors’ investment decisions. Furthermore, the effect of comparability in minimizing differences in trading behaviours resulting from investor sentiment is more pronounced for firms that are difficult to evaluate (such as young, volatile, non-dividend-paying firms with extreme market-to-book ratios, and those with limited analyst coverage) as well as firms operating in less competitive industries. Overall, these findings collectively indicate that comparability benefits investors and highlights which types of investors stand to gain from it.

Traditional finance theories have long debated the efficiency of capital markets, relying on strict assumptions about the rationality of investors. However, recent advances in behavioural finance theory have highlighted the limitations of conventional economic models, which fail to account for the observed psychological biases and irrational behaviours of investors in financial markets. In other words, changes in stock prices cannot be entirely explained by systematic risk (Lee et al., 1991; Siegel, 1992; Shleifer, 2000; Bakshi and Chen, 2005). Rather, the sentiments of investors participating in the capital market play a key role in investor information processing, which can affect stock prices (Brown and Cliff, 2004). Investor sentiment refers to the collective beliefs held by investors regarding future cash flows and investment risks, which are not solely based on fundamental information (Baker and Wurgler, 2006, 2007). An example of investor sentiment is the surge in stock prices in the Korean stock market driven by political tendencies ahead of the US presidential election.

Experts pointed out that “politically-themed stocks” in which prices soar regardless of a company's intrinsic value occur only in Korea. It is not easy to find an international case where the stock price soars only through social, kin, and academic ties rather than expectations of policy benefits. If Wall Street had betted on companies that benefited from Biden's policies during the last US presidential election, the stock price of the CEOs who graduated from the same university as Biden would have surged.

The surge in politically themed stocks ahead of the presidential election suggests psychological factors that cannot be explained by fundamentals inducing mispricing. Previous behavioural finance studies have empirically measured investor sentiment and then demonstrated the relationship between sentiment and stock prices (Baker and Stein, 2004; Baker and Wurgler, 2006; Kumar and Lee, 2006; Yang and Zhou, 2015; Ryu et al., 2018). Recent studies have shown the effect of investor sentiment on corporate decision-making, such as earnings management (Simpson, 2013), disclosures (Bergman and Roychowdhury, 2008), stock issuance (Baker and Wurgler, 2002), dividends (Baker and Wurgler, 2004; Li and Lie, 2006), and capital investment (Gilchrist et al., 2005). These previous studies have shown that when investor sentiment exists in the market, managers make strategic decisions consistent with investor sentiment-driven expectations.

Kahneman and Tversky (1973) posit that human cognitive abilities are limited, and decision-making relies on available information. In addition, individuals tend to seek out information that aligns with their beliefs: they tend to overreact to information that is consistent with their beliefs and underreact to information that contradicts their beliefs (Lord et al., 1979; Fiske and Taylor, 1991). Therefore, investors participating in the capital market may evaluate the value of a company based on specific information that aligns with their sentiments.

Stock market investors can be categorized into two groups: sophisticated and unsophisticated investors (Bartov et al., 2000). Sophisticated investors possess extensive knowledge and a deep understanding of accounting information, and make informed investment decisions. Unsophisticated investors have limited ability to analyse information, and rely more on psychological factors and intuition rather than data (Kyle, 1985; Black, 1986). Several studies have classified institutional investors as sophisticated investors and individual investors as unsophisticated investors (Kyle, 1985; Black, 1986; Bartov et al., 2000; Bae et al., 2006; Huang and Shiu, 2009; Kim et al., 2015). The following quotation clearly shows the characteristics of institutional and individual investors in the Korean stock market:

According to Korea Exchange (KRX), as a result of analyzing 224 stocks in the 19th presidential election, 96.6% of investors were individual investors; individual investors continue to net buy whenever the politically-themed stock rises. Institutional investors became the sellers of individual investors’ purchases. Individual investors suffered losses in 186 stocks, 83% of the total, and the average loss was 617,000 won per account. (Korea Exchange, 2017)

Institutional investors pay a relatively high cost to acquire high-quality information, whereas individual investors make meagre investments in information acquisition (Bhattacharya, 2001; Battalio and Mendenhall, 2005). In addition, institutional investors are more likely to notice company managers’ opportunistic earnings reporting behaviour based on their accounting knowledge and financial statement analysis ability (Bhattacharya et al., 2007). Therefore, institutional investors can determine whether a company's stock price is correctly valued in the market. In contrast, individual investors have relatively low accounting knowledge and analytical skills (Bartov et al., 2000; Collins et al., 2003; Bhattacharya et al., 2007; Miller, 2010) and are greatly impacted by psychological factors that affect the market (Barber and Odean, 2000). Consequently, individual investors buy overvalued stocks owing to investor sentiment. In contrast, institutional investors follow the opposite strategy, indicating that the trading behaviours of individual and institutional investors are quite different. In other words, the higher the investor sentiment, the more significant the difference between the trading behaviours of individual and institutional investors.

When financial statements are more comparable to those of industry peers, these peers serve as benchmarks. Comparable financial statements thus help investors to process information at a lower cost (De Franco et al., 2011; Kim et al., 2013) and to better assess a company's relative financial performance (Choi et al., 2019). Benchmarking firms with highly comparable firms can help individual investors with low sophistication better evaluate whether a company's stock is overvalued or undervalued at a low cost. Therefore, we expect that comparability would reduce the difference in trading behaviours between institutional and individual investors induced by investor sentiment.

The Korean stock market presents an opportunity to explore whether comparability can mitigate differences in trading behaviours between individual and institutional investors stemming from investor sentiment. First, the Korean stock market is dominated by individual investors. Most individual investors in the Korean market are considered to engage in noisy and uniform trading besides being influenced by behavioural biases (Ahn et al., 2008; Ryu, 2015; Yang et al., 2017; Ryu and Yang, 2018). Since individual investors are more likely to be influenced by market sentiment (Kumar and Lee, 2016), the Korean stock market is generally considered sensitive to market sentiment. Second, Korea has a strong agency problem between controlling shareholders and minority shareholders. The Korean economy has been dominated by large business groups called chaebol (Larker and Tayan, 2011). Chaebol owners control all member firms through a pyramid structure and cross holdings despite having small cash flow rights (Claessens et al., 2010). Controlling power is heavily concentrated in an individual or a single family, which may increase the risk of expropriation of minority investors by controlling shareholders (Ang et al., 2000; Bae et al., 2002). Investor protection is important to investors in Korea as strict laws can mitigate the risk of expropriation for minority investors. However, Korea has a weak investor protection mechanism (Wang et al., 2021; The World Bank Annual Doing Business Report, 2020). Therefore, a mechanism is needed to protect individual investors, who account for a significant portion of the Korean stock market. Third, understanding investors’ trading behaviour requires data on buying and selling activity. Previous studies using US data have examined individual investors’ accounts to analyse their trading records (Odean, 1999; Barber and Odean, 2000; Kumar and Lee, 2006). While this approach can identify the effect of individual characteristics on individual trading patterns and investment returns, it is a limited sample from a small number of brokerage houses and is not representative of the entire individual investor population. In contrast, the Korea Exchange (KRX) provides daily buying and selling trading volumes (amounts) of all companies listed on the KRX by investor, classified as domestic individuals, domestic institutions, and foreign institutions, which allows us to directly observe the trading behaviours of each company.

This paper exploits unique data on the trading volume of each investor of Korean listed firms (KSE). We measure trading volume by dividing the net buying volume, which is the result of subtracting the selling volume from the buying volume of each individual and institutional investor, by the sum of the buying volume and the selling volume (i.e., trading volume = net buying volume divided by the sum of the buying and selling volume). We find that differences between individual and institutional investors’ trading volume driven by investor sentiment are lower for firms that are more comparable with their peers. Breaking investors down into individual and institutional investors, we find that the tendency of individual investors to net buy stocks with high investor sentiment is lower for more comparable firms. In contrast, institutional investors’ tendency to net sell stocks with high investor sentiment is independent of comparability. Our results suggest that individual investors benefit more from comparable information than institutional investors. Also, our cross-sectional analyses present that the effect of comparability in mitigating the differences in trading behaviours between investors is strong for firms that are difficult to value (e.g., young, volatile, non-dividend-paying, extreme market-to-book firms, and firms with low analyst coverage) and firms in a less competitive industry. Our findings are robust to alternative measures of comparability (Francis et al., 2014), alternative measures of investor sentiment, and redefinitions of sophisticated investors.

This study contributes to the literature on the benefits of comparability. Previous studies shed light on some benefits, for example, facilitating analysts’ forecasting ability (De Franco et al., 2011), improving acquisition decisions (Chen et al., 2018), reducing the cost of debt (Kim et al., 2013; Fang et al., 2016), lowering stock price crash risk (Kim et al., 2016), and improving investors’ information processing ability (Choi et al., 2019). Although Kim et al. (2016) and Choi et al. (2019) focus on the benefits for equity investors, our distinction between individual and institutional investors is important in identifying which types of investors benefit more from comparability. The Financial Accounting Standards Board (FASB) (1980, p. 40) notes the importance of comparability: ‘investing and lending decisions essentially involve evaluations of alternative opportunities, and they cannot be made rationally if comparable information is unavailable’. Our finding that comparability improves individual investors’ investment decisions suggests that comparability is useful to investors, especially to unsophisticated individual investors when they evaluate alternative opportunities. At the same time, it highlights the role of comparability in improving the efficiency of capital allocation, as comparability leads unsophisticated individual investors to better investment decisions (Durnev et al., 2003; Durnev et al., 2004).

REVIEW OF PRIOR RESEARCH AND HYPOTHESES DEVELOPMENT

Investor Sentiment

The first area of research examines investor reactions according to investor sentiment by analysing the relationship between investor sentiment and stock returns (Brown and Cliff, 2004; Lemmon and Portniaguina, 2006; Schmeling, 2009; Stambaugh et al., 2012). Baker and Wurgler (2006, 2007) suggest that investor sentiment significantly affects stocks with difficult valuations. Specifically, start-ups, unprofitable stocks, non-dividend-paying firms, and companies with high return volatility show high subsequent stock returns because their stock prices are undervalued when investor sentiment is low. When investor sentiment is high, on the other hand, their stocks are overvalued, resulting in low future returns. Cornell et al. (2017) report that high-quality accounting information alleviates the problem of mispricing caused by investor sentiment. Kim and Ryu (2021) examine trading behaviours in the Korean market and show that as investor sentiment increases, net buying by individual investors increases, whereas net selling by institutional and foreign investors increases.

The second area of research examines corporate actions in response to investor sentiment. Baker and Wurgler (2000) show that firms issue more equity when the stock market is overvalued, arguing that managers intentionally select the timing of stock issuance. Bergman and Roychowdhury (2008) examine corporate disclosures according to investor sentiment and find that when investor sentiment is optimistic, managers do not frequently disclose earnings forecasts to maintain the optimistic expectations of investors and financial analysts. However, when investor sentiment is pessimistic, managers frequently disclose positive earnings forecasts to increase the current estimates of future earnings, indicating that managers respond to investor sentiment by strategically making corporate decisions.

The third area of research has examined the behaviour of external stakeholders, such as financial analysts and auditors, in response to investor sentiment. Leone et al. (2013) find that the Big 5 auditors are less likely to express a going concern opinion on dot-com companies that attempted to go public on the Nasdaq during the dot-com bubble in the US in early 2000 when investor sentiment was high. However, Cornell et al. (2017) show that financial analysts are more optimistic when investor sentiment is high.

Our research aligns with the first research stream, which explores investor reactions to investor sentiment through stock returns. As stock prices reflect changes in the expectations of the market as a whole while trading volume reflects changes in the expectations of individual investors (Beaver, 1968), we focus on trading volume. More importantly, we break down investors into individual and institutional investors and then explore the effect of financial statement comparability, one of the qualitative characteristics of accounting information, on each investor's response to investor sentiment.

Financial Statement Comparability

When a firm's financial statements are more comparable with its peers within the industry, information users can acquire and process more information at a low cost (De Franco et al., 2011; Kim et al., 2013). Consistent with this view, many previous studies have examined the benefits of comparability in various aspects. De Franco et al. (2011) find that companies with more comparable financial statements have more analyst coverage and their earnings forecasts are more accurate. Zhang (2018) finds that comparability allows auditors to conveniently identify exceptions or abnormalities, reducing the time required to collect audit evidence and consequently improve audit work efficiency. Additionally, Kim et al. (2013) and Fang et al. (2016) report that comparability lowers bid–ask spreads and loan spreads, respectively, suggesting that comparability plays a role in mitigating information asymmetry in debt markets. Furthermore, comparability improves mergers and acquisition decisions. Chen et al. (2018) find that acquiring a company's acquisition decision-making becomes more efficient when companies to be acquired are more comparable, indicating that comparability enables information users to better evaluate the corporate value and make more efficient decisions.

Some previous studies have examined the benefits that equity investors derive from comparable information. Baik et al. (2013) examine the benefits of comparability through the speed of information reflected in stock prices. When new information enters the market, such as an economic shock, the market assesses the impact of this information on the stock price of each company. Baik et al. (2013) report that comparability accelerates the rate at which new information is reflected in stock prices by enabling investors to quickly analyse the impact of new economic events on each company. In a similar vein, Choi et al. (2019) report that FERC (the ability of current period stock returns to reflect future earnings) is higher for more comparable firms as comparability helps investors to better predict future earnings. In addition, Kim et al. (2016) demonstrate that comparability allows investors to make more efficient decisions. Stock price crash risk occurs when bad news accumulates within a company and is suddenly revealed when it can no longer be withheld. Kim et al. (2016) report that comparability lowers the stock price crash risk due to the accumulation of bad news, providing evidence that comparability is useful for investors’ investment decision-making. While these previous studies have shown the benefits of comparability to investors from various viewpoints, they fall short of showing which investors benefit more from comparability. The Korean stock market is dominated by individual investors, thus, if comparability fails to provide benefits to individual investors its usefulness in the Korean market may be weakened. In this respect, examining which group of investors most enjoys the utility of comparability is important.

Hypothesis Development

In behavioural finance, financial information, as well as irrational factors in politics, the social environment, and individual sentiment, influence asset prices. According to Kahneman and Tversky (1973), humans rely on limited information to make decisions as they cannot use all the available information. They react more to information that is consistent with their beliefs but ignore information that contradicts them (Lord et al., 1979). Furthermore, behavioural finance argues that current stock prices are a function of rationality and irrationality. Therefore, if investors are more optimistic about a company's prospects than its fundamental value suggests, its stock price will be overestimated (Baker and Wurgler, 2006, 2007).

Individual investors make little investment in information acquisition, whereas institutional investors have high-quality information through significant investment in information acquisition (Bhattacharya, 2001; Mikhail et al., 2007; Ayers et al., 2011). Additionally, institutional investors have a higher analytical ability to interpret financial statements than individual investors (Bartov et al., 2000; Collins et al., 2003; Miller, 2010). Moreover, individual investors mistakenly believe that a company's intrinsic value is high even though it maintains a high stock price through upward earnings management (Bhattacharya et al., 2007). Individual investors tend to make investment decisions based on psychological bias rather than factual information (Barber and Odean, 2000); therefore, they buy stocks with high investor sentiment (Kim and Ryu, 2021). On the other hand, institutional investors net short stocks with high investor sentiment, as they have superior information, financial statement understanding, and analytical skills (Kim and Ryu, 2021). Therefore, the higher the investor sentiment, the more significant the difference in trading behaviour between individual and institutional investors.

When a firm's financial statements are more comparable with its peers in the industry, investors can obtain and process information at a lower cost (De Franco et al., 2011). By benchmarking highly comparable firms, investors can better evaluate a firm's relative performance and future earnings (Choi et al., 2019; Ozkan et al., 2012), allowing investors to better assess corporate value at a low cost. Additionally, since highly comparable information can serve as a reference for mapping economic events into financial statements (De Franco et al., 2011; Francis et al., 2014), investors can better understand how economic events are converted into financial statements. By using highly comparable companies that have experienced the same economic events, investors, including unsophisticated individual investors, can more easily evaluate whether a company's value is overvalued or undervalued and whether irrational factors, such as investor sentiment, are reflected in corporate value. Therefore, we expect differences in trading behaviours between institutional and individual investors to be lower for more comparable firms.

Previous studies have suggested that comparability can benefit institutional investors. Previous studies (De Franco et al., 2011; Kim et al., 2013; Zhang, 2018) have shown that analyst forecasts are more accurate, audit work is more efficient, and the split rating is reduced for more comparable firms, and this suggest that comparability benefits more informed users. However, the benefits institutional investors gain from comparable accounting information may be smaller than those of individual investors. As institutional investors possess private information and have expertise in information processing and financial statement interpretation, the additional informational advantage gained through comparable accounting information may be low. These findings are in line with Choi et al. (2019), who argue that comparability helps unsophisticated investors to a greater extent than other investors. Also, institutional investors’ trading is more driven by portfolio rebalancing criteria (Porter et al., 1996), which may open less room for comparability to operate. Correspondingly, we expect comparability to benefit individual investors more than institutional investors to a greater extent. Therefore, the difference between the trading behaviours of individual and institutional investors due to investor sentiment is expected to be low for more comparable firms. Accordingly, the first hypothesis is established as follows:

H1.The higher the comparability, the smaller the difference between the trading behaviours of individual and institutional investors due to investor sentiment.

The impact of comparability in reducing disparities in trading behaviour among investors is anticipated to be stronger for companies that are difficult to value. Chen et al. (2018) provide evidence that the relationship between acquired firms’ comparability and acquiring firms’ acquisition profitability is significant for inter-industry acquisitions but not for intra-industry acquisitions. This implies that comparability is more useful when the acquirer has limited information about the acquired firms; thus, it is challenging to evaluate corporate value. Additionally, companies with complex valuations (young, volatile, non-dividend-paying, high growth) are greatly influenced by investor sentiment (Baker and Wurgler, 2006; Mian and Sankaraguruswamy, 2012; Seybert and Yang, 2012). Therefore, we expect that the effect of comparability in mitigating the trading differences between investors would be large for companies that are difficult to evaluate and heavily influenced by investor sentiment. Accordingly, we establish the following hypothesis:

H2-1.The effect of comparability in reducing the difference between the trading behaviours of individual and institutional investors due to investor sentiment is significant for companies whose corporate value is challenging to evaluate.

The effect of comparability in lowering differences in trading behaviour among investors due to investor sentiment is expected to be stronger for firms in a less competitive industry. First, product market competition affects managerial disclosure decisions, preventing optimistic investor sentiment. Previous studies suggest that product market competition affects managerial disclosure decisions (Ali et al., 2014; Huang et al., 2017). Since the entry of new companies weakens the competitive position of existing firms, firms are quick to recognize bad news and hide good news to deter new entrants to their industry (Darrough and Stoughton, 1990; Clinch and Verrechia, 1997; Gelb and Greenstein, 2004; Li, 2010; Dhaliwal et al., 2014). The disclosure of bad news makes it difficult to create an optimistic investor sentiment about the company itself (Bergman and Roychowdhury, 2008), and the effect of comparability in mitigating trading behaviour between individual and institutional investors due to investor sentiment may be small. Second, product market competition plays a disciplinary role, hindering optimistic investor sentiment. Previous studies have shown that managers respond to investor sentiment expectations by portraying their firms in a way that maximizes their attractiveness to investors (Baker et al., 2007). For example, managers conduct earnings management (Simpson, 2013), report more optimistic management earnings forecasts (Hurwitz, 2018), disclose pro forma earnings in excess of GAAP earnings (Brown et al., 2012), and engage in opinion shopping (Amin et al., 2021). However, product market competition acts as a governance mechanism and curtails the opportunistic behaviour of managers (Alchian, 1950; Stigler, 1958; Hart, 1983; Scharfstein, 1988; Giroud and Mueller, 2010, 2011). Several studies provide evidence of the disciplinary role of competition (Chen et al., 2014; Majeed et al., 2018). Therefore, competition, as an external disciplinary mechanism, would curtail the opportunistic activities of managers to maintain or prolong investor sentiment.

Taken together, since product market competition makes it difficult to create an optimistic investor sentiment, the role of comparability in reducing differences in trading behaviour among investors due to investor sentiment is expected to be small in a fiercely competitive industry. Accordingly, we establish the following hypothesis:

H2-2.The effect of comparability in reducing the difference between the trading behaviours of individual and institutional investors due to investor sentiment is significant for companies in a less competitive industry.

RESEARCH DESIGN

Research Model

We use the following regression equation (1) to examine whether comparability mitigates differences in trading behaviours between different investors due to investor sentiment:
ΔDIFF t + 1 = β 0 + β 1 SENT t + β 2 COMP t + β 3 SENT t × COMP t + β 4 SIZE t + β 5 ROA t + β 6 RET t + β 7 LEV t + β 8 LNPRICE t + β 9 ΔVKOSPI t + β 10 ΔTERM t + β 11 ΔCREDIT t + β 12 STDBUS t + β 13 DISCL t + β 14 BOD t + β 15 BIND t + β 16 LARGE t + Year FE + Industry FE + ε $$ {\varDelta DIFF}_{t+ 1}={\beta}_0+{\beta}_1{SENT}_t+{\beta}_2{COMP}_t+{\beta}_3{SENT}_t\times {COMP}_t+{\beta}_4{SIZE}_t+{\beta}_5{ROA}_t+{\beta}_6{RET}_t+{\beta}_7{LEV}_t+{\beta}_8{LNPRICE}_t+{\beta}_9{\varDelta VKOSPI}_t+{\beta}_{1 0}{\varDelta TERM}_t+{\beta}_{1 1}{\varDelta CREDIT}_t+{\beta}_{1 2}{STDBUS}_t+{\beta}_{1 3}{DISCL}_t+{\beta}_{1 4}{BOD}_t+{\beta}_{1 5}{BIND}_t+{\beta}_{1 6}{LARGE}_t+ Year\ FE+ Industry\ FE+\varepsilon $$ ()
The dependent variable, DIFF, in equation (1) is the difference in net buying volume between individual and institutional investors (IBS_INDI - IBS_INST); ΔDIFFt+1 is the change in DIFF between the periods t and t+1. Barber and Odean (2000) use a method to calculate the asymmetry of buy and sell trades by investors. IBS_INDI is calculated by subtracting the selling volume of an individual investor (∑INDI_SELL) from the buying volume of an individual investor (∑INDI_BUY) during an accounting period and dividing it by the sum of the buying and selling volume. Similarly, IBS_INST is calculated by subtracting the selling volume of an institutional investor (∑INS_SELL) from the buying volume of an institutional investor (∑INST_BUY) during an accounting period and dividing it by the sum of buying and selling volume.
IBS _ INDI INST i , t = ∑ INDI INST _ BUY i , t − ∑ INDI INST _ SELL i , t ∑ INDI INST _ BUY i , t + ∑ INDI INST _ SELL i , t $$ IBS\_ INDI{(INST)}_{i,t}=\frac{\sum INDI(INST)\_{BUY}_{i,t}-\sum INDI(INST)\_{SELL}_{i,t}}{\sum INDI(INST)\_{BUY}_{i,t}+\sum INDI(INST)\_{SELL}_{i,t}} $$ ()

SENTt is the average investor sentiment over the year t based on Ryu et al.'s (2018) measure. The specific measurement method is described in the next section. COMP is the financial statement comparability based on De Franco et al.'s (2011) measure, the most widely used measure in previous studies. We hypothesize that comparability mitigates the difference in investors’ trading behaviour due to investor sentiment, so we expect β3 to be negative.

To examine the significance of comparability in reducing the disparity in trading behaviours between individual and institutional investors driven by investor sentiment, we divide the entire sample based on the complexity of corporate valuations. We classify companies as either difficult or easy to evaluate. Specifically, when a company's analyst coverage, age, and stock price volatility are below the sample's median, the companies are defined as low coverage, young, and stable. Those above the median are high-coverage, mature, and volatile. Baker and Wurgler (2006) indicate that the effect of investor sentiment on the valuation of market-to-book ratio (MTB)-sorted portfolios is U-shaped. Therefore, after dividing MTB into quartiles, the companies corresponding to the first and fourth quartiles are defined as extreme MTB firms, and those corresponding to the second and third quartiles are defined as non-extreme MTB firms. Low coverage, young, volatile, extreme MTB, and non-dividend-paying firms are defined as difficult to value; high-coverage, mature, stable, non-extreme-MTB, and dividend-paying firms are defined as firms that are easy to value. If Hypothesis 2-1 is supported, it is expected that β3 will be more negative in the hard-to-value group than in the easy-to-value group.

Previous studies have used industry concentration to measure the degree of competition within an industry (DeFond and Park, 1999; Li, 2010; Ali et al., 2014; Dhaliwal et al., 2014). We use the Herfindahl-Hirschman Index (HHI) as a proxy for industry concentration.
HHI j , t = ∑ i = 1 N S i , j , t 2 $$ {HHI}_{j,t}=\sum \limits_{i=1}^N{S}_{i,j,t}^2 $$ ()

where Sijt is the percentage market share of the sale of the individual company i in industry j in year t. A high (low) value of HHI means that few (many) companies are competing in the industry. In other words, a high (low) HHI value means a low (high) market competition. To test Hypothesis 2-2, we divide the entire sample by the median value of HHI and see if there is a significant difference in β3 between the high and low HHI groups.

According to Kumar and Lee (2006), individual investors prefer low-priced stocks that can be traded in a small amount; therefore, we control the natural logarithm of the closing price (LNPRICE). We control the stock return (RET) to manage investors’ momentum strategies (Szakmary et al., 2010). As previous studies have reported that institutional investors prefer companies with high profitability and low leverage (Eakins et al., 1998; Dahlquist and Robertsson, 2001), we control return on assets (ROA) and leverage (LEV). The variable related to market uncertainty (VKOSPI) is controlled for because the more significant the market's fate, the more investors’ trading could change; for example, they may move toward low-risk assets like bonds and spots from relatively high-risk assets such as stocks. The stock market volatility index (VIX) is an index that measures the volatility of future stock market returns expected by investors (Bloom, 2009; Barth and So, 2014; Kim et al., 2016). In this study, VKOSPI, the volatility index of the Korean stock market, designed by the KRX to suit the Korean stock market, is used. In addition, we control macroeconomic variables such as term spread and credit spread. Term spread (TERM) is measured by subtracting the risk-free rate of return (risk-free rate is the 91-day certificate of deposit rate) from the yield on the five-year government bond. Credit spread (CREDIT) is the return on a BBB credit-rated bond subtracted from the return on a AA credit-rated bond. To control the fluctuation in business sentiment during the accounting period, we add the standard deviation of business sentiment (STDBUS) as a control variable. Also, we control the frequency of managerial voluntary disclosure (DISCL) because disclosures provided by companies are likely to influence investors’ trading behaviour. Additionally, to control the effect of corporate governance and ownership structure (Chung and Zhang, 2011), we add board size (BOD), board independence (BIND), and largest shareholders’ ownership (LARGE) to the control variables. Lastly, we include year and fixed effects in the model and estimate the model with robust standard errors clustered by firm.

Investor Sentiment

Many studies have estimated market-wide investor sentiment and industry-specific investor sentiment (Baker and Wurgler, 2006; Ali and Gurun, 2009; Simpson, 2013; Wang, 2018). Baker and Wurgler (2006) developed a market-wide composite investor sentiment index based on the six underlying proxies for sentiment: the closed-end fund discount, share turnover, the number and first-day returns on IPOs, the equity shares in new issues, and the dividend premium. However, Ryu et al. (2018) argue that investor sentiment might differ across company, and the Korean stock market does not have a large deviation in dividend yield, unlike the US market; thus, it is not easy to measure data such as closed-end fund discounts. Thus, applying the variables used by Baker and Wurgler (2006) to the Korean market is challenging. In addition, unlike previous studies that have developed an investor sentiment index using annual or monthly data, developing investor sentiment based on daily data can better reflect changes in investor sentiment. Ryu et al. (2018) and Seok et al. (2019) devise a firm-specific investor sentiment index tailored to the Korean market based on daily data.

To construct a sentiment index for a particular firm i on trading day t, they calculate the following four stock market components: the relative strength index (RSI), the psychological line index (PLI), the adjusted turnover rate (ATR), and the logarithm of trading volume (LTV). RSI measures the magnitude of recent price changes to evaluate the overbought or oversold part of a stock's price. The following equation calculates RSI:
RSI it = RS i , t 1 + RS it × 100 $$ { RS I}_{it}=\frac{RS_{i,t}}{\left(1+{RS}_{it}\right)}\times 100 $$ ()
RS t = ∑ k = 0 13 max 0 p i , t − k − p i , t − 1 − k / ∑ k = 0 13 max 0 p i , t − 1 − k − p i , t − k $$ {RS}_t=\sum \limits_{k=0}^{13}\max \left(0,{p}_{i,t-k}-{p}_{i,t-1-k}\right)/\sum \limits_{k=0}^{13}\max \left(0,{p}_{i,t-1-k}-{p}_{i,t-k}\right) $$ ()
where Pit is the closing price of stock i at time t. RSI is generally calculated based on a 14-day period (Chen et al., 2010). After obtaining the value of RS, we adjust it so that RSI has a value between zero and 100, gauging over-buy if the RSI value is 80 and above and over-sell if it is 20 or below. RSI is used as a proxy for investor sentiment in studies such as Chen et al. (2010). PLI is denoted in equation (6). It identifies changes in investor sentiment for a certain period to indicate whether the sentiment is positive or negative. This is the ratio of the number of days a stock's price increases in the preceding 12 days to the overall number of days (12 in this case). Here, the 12 days is based on the primitive rhythm of 12 days in human psychological change. If the number of days a stock price increases is nine in the preceding 12 days, PLI stands at 75, and if the stock price declines on all the days, it stands at zero. PLI measures the extent to which a stock is overbought (75 and above) or oversold (25 or below). Both RSI and PLI measure the extent to which a stock is overbought or oversold and show whether investor sentiment is positive or negative regarding that particular stock. PLI is used as a proxy for investor sentiment in studies such as Yang and Gao (2014).
PLI it = 100 × ∑ K = 0 11 max 0 , P i , t − k − p i , t − 1 , k P i , t − k − p i , t − 1 − k / 12 $$ {PLI}_{it}=100\times \sum \limits_{K=0}^{11}\left(\frac{\max \left(0,{P}_{i,t-k}-{p}_{i,t-1,k}\right)}{P_{i,t-k}-{p}_{i,t-1-k}}\right)/12 $$ ()
Trading turnover is calculated by dividing the volume by the total number of shares outstanding on the same date. To determine whether investors are optimistic or pessimistic, ATR, measured by equation (7), is used as a proxy for investor sentiment (Yang and Zhang, 2014; Yang and Zhou, 2015). The positive (negative) ATR value indicates a bullish (bearish) stock price.
ATR i , t = Volume i , t share i , t × RET i , t RET i , t $$ {ATR}_{i,t}=\frac{Volume_{i,t}}{share_{i,t}}\times \frac{RET_{i,t}}{\left|{RET}_{i,t}\right|} $$ ()
Previous studies have used trading volume as a proxy for investor sentiment (Baker and Stein, 2004; Liao et al., 2011). LTV is the natural logarithmic value of the daily trading volume of individual shares.
LTV i , t = ln Volume i , t $$ {LTV}_{i,t}=\ln \left({Volume}_{i,t}\right) $$ ()
Next, using principal component analysis, we estimate Fi. RSI, Fi. PLI, Fi. ATR, and Fi. LTV is the first principal factor of the five components. Then, we construct common sentiment by linearly combining these factors and elements, as shown in equation (9). Finally, controlling for the market risk factor (MARKET), we estimate the residual for stock i and define the residual as firm-specific investor sentiment on day t. MARKET is defined as the difference between the Korea Stock Exchange (KOSPI) return and the risk-free rate.
S i , t = F i , RSI × RSI i , t + F i , PLI × PLI i , t + F i , ATR × ATR i , t + F i , LTV × LTV i , t $$ {S}_{i,t}={F}_{i, RSI}\times {RSI}_{i,t}+{F}_{i, PLI}\times {PLI}_{i,t}+{F}_{i, ATR}\times {ATR}_{i,t}+{F}_{i, LTV}\times {LTV}_{i,t} $$ ()
SENT i , t = a 0 + a 1 MARKET t + ε i , t $$ {SENT}_{i,t}={a}_0+{a}_1\ {MARKET}_t+\varepsilon i,t $$ ()

The daily investor sentiment for individual companies is calculated and then averaged by year. However, bias may occur in this case due to the investor sentiment index's asymmetric distribution or extreme value. Therefore, we rank the sample based on the annual average value of daily investor sentiment within the industry and year, and then divide by the number of observations and refer to it as SENT. Thus, it has a value of 0<SENT<=1, and the higher the value, the more optimistic investor sentiment is in the same industry.

Financial Statement Comparability

We measure comparability according to De Franco et al. (2011). If the accounting functions of two firms mapping economic events to financial information are similar, the accounting information between the two firms is considered comparable. We estimate the accounting function of firm i through equation (11), using the preceding 16 quarters.
EARNINGS i , t = a i + β i RETURN + ε i , t $$ {EARNINGS}_{i,t}={a}_i+{\beta}_i\ RETURN+{\varepsilon}_{i,t} $$ ()
In equation (11), EARNINGS (a proxy for financial information) is the quarterly net income scaled by the market value at the beginning of the quarter, and RETURN (a proxy for economic events) is the quarterly stock return. The estimated coefficients α ̂ i , β ̂ i $$ {\hat{\alpha}}_i,{\hat{\beta}}_i $$ represent the accounting function of firm i. Similarly, we estimate the accounting function of firm j α ̂ j β ̂ j $$ \left({\hat{\alpha}}_j,{\hat{\beta}}_j\right) $$ through its earnings and returns from the previous 16 quarters. To measure the closeness of the accounting function between firms i and j under the same economic event, we use each company's accounting function with the economic events of a single firm.
E Earnings iit = a ̂ i + β ̂ i RETURN it $$ E{(Earnings)}_{iit}={\hat{a}}_i+{\hat{\beta}}_{\mathrm{i}}{RETURN}_{it} $$ ()
E Earnings ijt = a ̂ j + β ̂ j RETURN it $$ E{(Earnings)}_{ijt}={\hat{a}}_j+{\hat{\beta}}_{\mathrm{j}}{RETURN}_{it} $$ ()
(Earnings)iit is the predicted earnings of firm i from firm i's function, given its stock returns, and E(Earnings)ijt is the predicted earnings of firm j from firm j's function, given firm i's stock returns. If the accounting function of firms i and j are comparable, they produce similar accounting information when experiencing the same economic event. Therefore, the predicted earnings calculated from equations (12) and (13) will be similar. The similarity of expected earnings given the same economic event, that is, comparability, is measured by the following equation:
CompAcct ijt = − 1 / 16 × ∑ t − 15 t E Earnings iit − E Earnings ijt $$ {CompAcct}_{ijt}=-1/16\times \sum \limits_{t-15}^t\left|E\left({Earnings}_{iit}\right)-E\left({Earnings}_{ijt}\right)\right| $$ ()

Comparability between firms i and j (CompAcctijt) is calculated by putting an absolute value on the difference between (Earnings)iit and E(Earnings)ijt and then multiplying by –1. We create combinations for all firms in the same industry and then estimate CompAcctijt for each combination. Next, we form a firm-year measure of comparability in two ways. The first comparability measure is CompINDit, the median CompAccctijt for firm i's comparability. The second measure of comparability is Comp4it, which is the average of the four highest CompAccctijt. These two measures of comparability are the most commonly used in previous studies.

Sample Selection

Our initial sample includes firm-year observations listed on the KRX from 2011 through 2019. The following firms are excluded to ensure comparability: firms with insufficient stock trading data to measure investor sentiment, and companies with capital erosion due to the non-reliability of their financial statements. We only include companies with December as the year-end for consistency. This produces a final sample of 3,436 firm-year observations. Financial data and stock data are obtained from the KIS-VALIUE database and the TS-2000 database, respectively. To minimize the influence of outliers, we winsorize the samples corresponding to the upper and lower 1% of all continuous variables. Panel A of Table 1 provides the industry distribution of our sample, with the manufacturing industry accounting for the largest portion at 69.53%, followed by professional, scientific, and technical activity industries at 11.76%, and wholesale and retail trade industries at 9.98%. Panel B of Table 1 presents the sample distribution across years. Table B reveals that sample is not concentrated in any particular year and is generally equal across years.

Table 1. SAMPLE SELECTION
Panel A: Industry distribution of the sample
Industry N %
Manufacturing 2,389 69.53
Professional, scientific, and technical activities 404 11.76
Wholesale and retail trade 343 9.98
Construction 110 3.20
Information and communication 78 2.27
Transportation and storage 48 1.40
Electricity, gas, steam, and air conditioning supply 40 1.16
Business facilities management and business support services 16 0.47
Agriculture, forestry, and fishing 8 0.23
Total 3,436 100
Industrial classification follows the Korean Standard Industrial Classification (KSIC), which is consistent with the International Standard Industry Classification recommended by the United Nations.
Panel B: Early distribution of the sample
Year N %
2011 374 10.88
2012 376 10.94
2013 380 11.06
2014 383 11.15
2015 384 11.19
2016 382 11.12
2017 385 11.20
2018 386 11.23
2019 386 11.23
Total 3,436 100 (%)

EMPIRICAL RESULTS

Descriptive Statistics

Table 2 shows the descriptive statistics of four variables measured based on Seok et al. (2019) and Ryu et al. (2018); these are used to calculate investor sentiment, this study's main variable of interest. All variables show positive values and are generally similar to previous studies (Ryu et al., 2018; Seok et al., 2019).

Table 2. DESCRIPTIVE STATISTICS OF THE VARIABLES THAT CONSTITUTE THE INVESTMENT SENTIMENT INDEX
Variable Mean Std. dev Q1 Median Q3
RSI 59.763 17.619 48.000 60.714 72.549
PLI 44.625 13.937 33.333 41.667 50.000
LTV 11.040 2.235 9.732 11.247 12.528
ATR 0.151 4.358 –0.282 0.000 0.258
  • RSI = relative strength index; PLI = psychological line index; LTV = logarithm of the trading volume; ATR = adjusted turnover rate.

Panels A and B of Table 3 present the yearly buying and selling volume of individual, institutional, foreign, and other investors during the sample period. The individual investors’ purchase volume is more than 84% of the total buying volume. Meanwhile, the buying and selling of institutional, foreign, and other investors stand at about 6.25%, 9.03%, and less than 1%, respectively. The same goes for sell trades. As shown in Panels A and B, individual investors account for more than 80% of the total trading volume in the Korean stock market. Although the trading volume of individual investors is high, if the trading amount per unit is small, the impact on the stock market may be low. Panels C and D show the annual buying and selling amounts of individual, institutional, and other investors during the sample period. In Panel C, the buying amount of individual investors accounts for approximately 49% of the total buying amount, and the buying amount of institutional investors accounts for about 22% of the total buying amount. The same is valid for sell trades. In terms of trading volume and amount, the proportion of individual investors is remarkably high in the Korean market. In other words, individual investors are the leading players in the Korean stock market. Individual investors, known as noise investors who lack information, are easily affected by psychological bias, and are sensitive to market sentiment (Barber and Odean, 2009; Chung et al., 2016), suggesting that investor sentiment is prevalent in the Korean stock market.

Table 3. TRADING VOLUME AND TRADING AMOUNT BY INVESTOR TYPE
Panel A: Buying volume
Investor 2011 2012 2013 2014 2015 2016 2017 2018 2019 Average
Individual 72,990 107,668 69,347 55,954 98,061 78,619 67,107 76,696 95,009 80,161
(84.01%)
Institutional 7,627 5,867 5,057 5,193 5,540 4,966 5,369 6,842 7,241 5,967
(6.25%)
Foreigner 6,166 6,401 6,151 6,435 8,526 8,459 9,635 12,897 12,890 8,618
(9.03%)
Other 949 710 541 549 776 642 620 671 659 680
(0.71%)
Total 87,732 120,647 81,096 68,130 112,903 92,686 82,731 97,105 115,798 95,425
Unit: 1,000,000, (%) = trading volume by each investor/total volume. Source: The Bank of Korea (https://ecos.bok.or.kr).
Panel B: Selling volume
Investor 2011 2012 2013 2014 2015 2016 2017 2018 2019 Average
Individual 72,792 107,822 69,195 55,731 97,323 78,432 66,988 76,356 94,504 79,905
(83.74%)
Institutional 7,659 5,960 5,091 5,268 5,918 5,174 5,652 7,150 7,360 6,137
(6.43%)
Foreigner 6,307 6,094 6,068 6,358 8,702 8,245 9,337 12,771 12,983 8,541
(8.95%)
Other 975 771 742 773 961 834 754 828 951 843
(0.88%)
Total 87,732 120,647 81,096 68,130 112,903 92,686 82,731 97,105 115,798 95,425
Unit: 1,000,000, (%) = trading volume by each investor/total volume. Source: The Bank of Korea (https://ecos.bok.or.kr).
Panel C: Buying amount
Investor 2011 2012 2013 2014 2015 2016 2017 2018 2019 Average
Individual 942,973 600,477 455,857 435,938 715,568 548,533 599,416 818,028 577,440 593,907
(48.89%)
Institutional 421,317 295,919 234,764 230,252 252,334 218,308 270,680 329,238 288,123 264,952
(21.81%)
Foreigner 311,019 285,283 284,842 295,237 338,583 328,073 404,131 431,853 349,954 339,745
(27.97%)
Other 26,751 14,584 10,913 14,550 20,745 17,754 19,933 18,744 11,975 16,150
(1.33%)
Total 1,702,060 1,196,263 986,375 975,977 1,327,230 1,112,669 1,294,160 1,597,864 1,227,493 1,214,754
Unit: 1,000,000,000 (WON), (%) = trading amount by each investor/total amount. Source: The Bank of Korea (https://ecos.bok.or.kr).
Panel D: Selling amount
Investor 2011 2012 2013 2014 2015 2016 2017 2018 2019 Average
Individual 944,793 616,027 461,496 438,774 715,972 557,138 608,745 810,983 589,242 644,241
(50.57%)
Institutional 409,366 292,285 229,705 230,946 252,799 223,517 273,106 332,128 279,268 280,482
(22.01%)
Foreigner 319,161 268,595 281,647 290,560 343,015 317,239 397,803 437,615 349,118 331,954
(26.05%)
Other 28,741 19,357 13,526 15,697 15,444 14,775 14,505 17,138 9,864 17,398
(1.37%)
Total 1,702,060 1,196,263 986,375 975,977 1,327,230 1,112,669 1,294,160 1,597,864 1,227,493 1,274,075
Unit: 1,000,000,000 (WON), (%) = trading amount by each investor/total amount. Source: The Bank of Korea (https://ecos.bok.or.kr).

Table 4 shows the descriptive statistics. The mean and median of ΔDIFFt+1 corresponding to the dependent variable in this study are 0.004 and 0.016, respectively, showing positive values. This means that the increase in individual investors’ net buying volume is greater than institutional investors’ net buying volume during the same period. According to the variable described above, the mean of SENT has a value between zero and one. Another variable of interest, comparability, has a mean (median) of CompIND of –0.028 (–0.016) and a mean (median) of Comp4 of –0.017 (–0.007). The mean and median of HHI are 0.143 and 0.109, respectively. This figure is higher than that of the US, indicating that Korea has a relatively uncompetitive market. This may be due to a difference in the speed of technological development in Korea and the US.

Table 4. DESCRIPTIVE STATISTICS
Variables N Mean Std. dev. Median Min. Max.
SENTt 3,436 0.542 0.295 0.541 0.019 1.000
ΔDIFFt+1 3,436 0.004 0.451 0.016 –2.020 1.992
CompINDt 3,436 –0.028 0.050 –0.016 –1.022 –0.003
Comp4t 3,436 –0.017 0.044 –0.007 –0.917 –0.001
HHIt 3,436 0.143 0.106 0.109 0.025 0.867
SIZEt 3,436 26.871 1.464 26.648 24.162 31.262
AGEt 3,436 44.250 17.696 45.000 9.000 94.000
STDRETt 3,436 0.025 0.011 0.022 0.010 0.060
DIVt 3,436 0.727 0.446 1.000 0.000 1.000
MTBt 3,436 1.305 1.213 0.930 0.262 7.883
ROAt 3,436 0.016 0.071 0.023 –0.314 0.211
RETt 3,436 0.090 0.435 –0.005 –0.563 2.099
LEVt 3,436 0.397 0.210 0.404 0.011 0.875
LNPRICEt 3,436 9.410 1.625 9.262 6.284 13.763
ΔVKOSPIt 3,436 –1.240 2.589 –0.470 –5.458 2.860
ΔTERMt 3,436 –0.068 0.386 0.066 –0.729 0.518
ΔCREDITt 3,436 –0.016 0.182 –0.078 –0.187 0.400
STDBUSt 3,436 2.584 1.457 1.938 1.133 6.064
DISCLt 3,436 1.799 3.193 0.000 0.000 26.000
BODt 3,436 1.677 0.342 1.792 1.099 2.639
BINDt 3,436 0.380 0.138 0.333 0.000 1.000
LARGEt 3,436 0.440 0.164 0.446 0.100 0.841
  • All continuous variables are winsorized at the top and bottom one percentile. See the Appendix for variable definitions.

Table 5 shows Pearson correlations among the main variables. SENT, the variable of interest, has a significantly positive correlation with ΔDIFFt+1 at the 1% level, indicating that in firms with high investor sentiment there is likely to be a large difference in trading behaviour between individual and institutional investors. SENT has a significantly positive correlation with HHI, STDRET, MTB, ROA, RET, and LNPRICE at 1%. This means that investor sentiment is high in companies with industry concentration (i.e., the degree of competition within the industry is low), companies with high volatility in returns, and companies with high growth potential and profitability. In particular, there is a positive relationship between SENT and LNPRICE, indicating that the higher the investor sentiment, the higher the stock price. This means that investor sentiment has a significant impact on stock price determination, which is consistent with previous studies that investors’ irrationality can be another factor in determining stock prices in the capital market (e.g., Baker and Wurgler, 2006; Schmeling, 2009).

Table 5. CORRELATION COEFFICIENTS
(2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22)
1)SENTt 0.11 0.09 0.08 0.12 0.02 –0.08 0.15 0.03 0.10 0.15 0.56 –0.06 0.09 –0.01 –0.01 0.01 0.01 0.08 0.04 0.02 –0.05
2)ΔDIFFt+1 1.00 0.03 0.03 0.01 0.02 –0.01 –0.04 0.04 0.03 0.02 0.08 –0.01 0.06 0.03 –0.04 –0.04 –0.04 0.01 0.02 0.02 –0.01
3)CompINDt 1.00 0.97 –0.01 0.03 –0.02 –0.25 0.31 –0.06 0.20 0.03 –0.29 0.16 0.03 –0.01 –0.01 –0.03 0.07 0.04 0.01 0.11
4)Comp4t 1.00 –0.02 0.04 –0.02 –0.23 0.29 –0.06 0.18 0.04 –0.26 0.15 0.01 –0.01 0.01 –0.01 0.09 0.05 0.01 0.08
5)HHIt 1.00 0.09 –0.05 –0.01 –0.07 –0.07 –0.06 –0.04 0.06 –0.04 –0.01 0.03 0.01 0.01 0.04 0.03 0.06 –0.06
6)SIZEt 1.00 0.06 –0.31 0.28 –0.09 0.19 –0.12 0.13 0.61 0.02 –0.01 –0.01 –0.01 0.63 0.47 0.59 –0.01
7)AGEt 1.00 –0.05 0.01 –0.11 –0.02 –0.01 –0.03 0.06 0.09 –0.02 –0.03 –0.06 –0.04 0.05 0.01 –0.07
8)STDRETt 1.00 –0.39 0.30 –0.24 0.34 0.20 –0.31 0.14 –0.11 0.06 0.13 –0.14 –0.15 –0.12 –0.22
9)DIVt 1.00 –0.18 0.38 –0.03 –0.29 0.45 –0.02 –0.01 0.02 0.02 0.15 0.16 0.09 0.19
10)MTBt 1.00 –0.05 0.12 0.07 0.11 0.10 0.02 –0.07 –0.04 0.07 0.01 0.01 –0.15
11)ROAt 1.00 0.15 –0.26 0.34 0.01 0.03 –0.02 0.00 0.14 0.07 0.03 0.15
12)RETt 1.00 –0.02 0.09 –0.04 0.01 –0.05 0.06 –0.05 –0.05 –0.10 –0.02
13)LEVt 1.00 –0.25 –0.06 0.01 0.03 0.04 0.07 0.05 0.12 –0.12
14)LNPRICEt 1.00 0.02 0.03 –0.05 –0.02 0.38 0.32 0.29 0.09
15)ΔVKOSPIt 1.00 –0.01 –0.42 –0.30 0.01 –0.03 0.04 0.01
16)ΔTERMt 1.00 –0.61 –0.48 0.01 –0.01 –0.01 0.01
17)ΔCREDITt 1.00 0.75 –0.01 0.02 –0.01 0.01
18)STDBUSt 1.00 –0.01 0.02 –0.02 –0.01
19)DISCLt 1.00 0.34 0.40 –0.14
20)BODt 1.00 0.43 –0.11
21)BINDt 1.00 –0.10
22)LARGEt 1.00
  • This table presents the Pearson correlation coefficients. See the Appendix for variable definitions. Bold indicates significance at the 0.01 level (based on two-tailed tests).

Main Results

Table 6 shows whether comparability mitigates differences in investors’ trading behaviour due to investor sentiment. Column (1) shows the results of CompIND, and column (2) shows the results of Comp4. In columns (1) and (2), the regression coefficients of SENT are both 0.010, showing a significantly positive value at 1%. This means that the greater the investor sentiment in the Korean stock market, the greater the difference in trading behaviours between individual and institutional investors. That is, institutional investors, so-called sophisticated investors, and individual investors, called unsophisticated, react differently to stocks with high investor sentiment. The regression coefficients of SENT×CompIND and SENT×Comp4 are –0.224 and –0.243, respectively, showing significant negative values. This finding shows that the difference in trading behaviours between individual and institutional investors due to investor sentiment is mitigated by comparability, which supports Hypothesis 1.

Table 6. THE EFFECT OF COMPARABILITY ON THE RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND TRADING BEHAVIOUR
Variable (1) ΔDIFFt+1 (2) ΔDIFFt+1
Intercept –0.121 –0.099
(–0.57) (–0.47)
SENTt 0.010*** 0.010***
(3.42) (3.36)
CompINDt –0.969
(–1.53)
Comp4t –1.238*
(–1.72)
SENTt×CompINDt –0.224*
(–1.80)
SENTt×Comp4t –0.243*
(–1.95)
SIZEt –0.003 –0.004
(–0.27) (–0.39)
ROAt –0.026 –0.018
(–0.18) (–0.42)
RETt 0.017 0.018
(0.68) (0.72)
LEVt 0.035 0.032
(0.73) (0.69)
LNPRICEt 0.017** 0.016**
(2.29) (2.26)
ΔVKOSPIt –0.007 –0.007
(–1.17) (–1.18)
ΔTERMt –0.105*** –0.106***
(–3.71) (–3.76)
ΔCREDITt –0.029 –0.035
(–0.20) (–0.24)
STDBUSt 0.004 0.005
(0.14) (0.19)
DISCLt –0.003 –0.003
(–1.39) (–1.27)
BODt 0.003 0.005
(0.09) (0.16)
BINDt 0.036 0.044
(0.48) (0.60)
LARGEt 0.006 0.008
(0.12) (0.15)
Year and industry fixed effects Yes Yes
Cluster Firm Firm
Adj. R2 0.0283 0.0279
Obs. 3,436 3,436
  • All continuous variables are winsorized at the top and bottom one percentile. t-statistics are presented in parentheses. t-values are based on standard errors clustered by firm. ***, **, and * indicate significance at the 0.01, 0.05, and 0.10 levels, respectively (based on two-tailed tests). See the Appendix for variable definitions.

Next, to examine in detail how comparability affects the trading behaviours of individual and institutional investors, we examine the trading behaviours of individual and institutional investors separately. Results are presented in Table 7. Columns (1) and (2) represent individual investors’ trading behaviour (ΔIBS_INDIt+1), and columns (3) and (4) show the trading behaviour of institutional investors (ΔIBS_INSTt+1), respectively. The coefficient of SENT is 0.004 in columns (1)–(2), whereas it is –0.008 in columns (3)–(4). Given that the higher investor sentiment is within a bullish market (Baker and Wurgler, 2007), institutional investors appear to net sell stocks with high investor sentiment to hedge or rebalance their portfolios (Kang et al., 2010). This finding that SENT remains significant after controlling for stock return suggests that a company's performance is not the only factor determining investors’ buying and selling decisions. In other words, investor sentiment is important in investors’ trading decisions. Additionally, this finding represents that institutional investors tend to net sell stocks with high investor sentiment, while individual investors tend to net buy stocks that institutional investors have net sold, which is consistent with Kim and Ryu's (2021) finding. The coefficient of SENT×CompIND in column (1) is negative and significant (–0.019, t-stat= –1.86). Similarly, in column (2), the coefficient of SENT ×Comp4 is negative and significant (–0.021, t-stat= –2.22). This means that the tendency of individual investors to net buy stock with high sentiment is mitigated by comparability. This finding suggests that comparability helps individual investors identify overvalued stocks driven by optimistic investor sentiment and thus make better trading decisions.

Table 7. THE EFFECT OF COMPARABILITY ON THE RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND TRADING BEHAVIOUR: DIVIDING INVESTORS INTO INDIVIDUAL AND INSTITUTIONAL INVESTORS
Variable (1) ΔIBS_INDIt+1 (2) ΔIBS_INDIt+1 (3)ΔIBS_INSTt+1 (4)ΔIBS_INSTt+1
Intercept –0.011 –0.011 0.252 0.254
(–0.29) (–0.29) (1.16) (1.17)
SENTt 0.004*** 0.004*** –0.008*** –0.008***
(8.71) (8.70) (–2.94) (–2.93)
CompINDt –0.050 0.613
(–0.89) (0.94)
Comp4t –0.077 0.790
(–1.30) (1.05)
SENTt×CompINDt –0.019* 0.098 0.107
(–1.86) (0.87) (0.94)
SENTt×Comp4t –0.021**
(–2.22)
SIZEt –0.001 –0.001 0.001 0.002
(–0.43) (–0.44) (0.15) (0.16)
ROAt –0.019 –0.019 0.006 0.005
(–1.08) (–1.06) (0.04) (0.03)
RETt 0.003 0.003 –0.001 –0.001
(0.78) (0.81) (–0.02) (–0.05)
LEVt 0.007 0.007 –0.048 –0.049
(0.92) (0.93) (–1.01) (–1.02)
LNPRICEt 0.002 0.002 –0.015* –0.015*
(1.40) (1.40) (–1.95) (–1.95)
ΔVKOSPIt –0.009*** –0.009*** 0.050*** 0.050***
(–4.26) (–4.27) (5.39) (5.41)
ΔTERMt 0.029*** 0.029*** 0.053** 0.053**
(4.98) (4.99) (2.00) (2.01)
ΔCREDITt 0.135*** 0.135*** –0.287** –0.287**
(5.58) (5.59) (–2.51) (–2.51)
STDBUSt 0.012*** 0.013*** –0.084*** –0.084***
(2.75) (2.77) (–3.70) (–3.73)
DISCLt –0.001 –0.001 0.003 0.003
(–0.72) (–0.71) (1.48) (1.47)
BODt 0.003 0.003 –0.001 –0.001
(0.62) (0.62) (–0.03) (–0.04)
BINDt 0.001 0.001 –0.025 –0.025
(0.06) (0.07) (–0.35) (–0.36)
LARGEt 0.003 0.003 –0.009 –0.010
(0.29) (0.30) (–0.17) (–0.19)
Year and industry fixed effects Yes Yes Yes Yes
Cluster Firm Firm Firm Firm
Adj. R2 0.0703 0.0704 0.0286 0.0287
Obs. 3,436 3,436 3,436 3,436
  • All continuous variables are winsorized at the top and bottom one percentile. t-statistics are presented in parentheses. t-values are based on standard errors clustered by firm. ***, **, and * indicate significance at the 0.01, 0.05, and 0.10 levels, respectively (based on two-tailed tests). See the Appendix for variable definitions.

In constrast, in columns (3) and (4), the coefficients of SENT×CompIND and SENT×Comp4 are insignificant. This means that comparability does not affect the relationship between investor sentiment and institutional investors’ trading. In other words, it suggests that the tendency of institutional investors to net sell stocks with high investor sentiment is not further improved by comparability. Institutional investors have high-quality information and expertise in analysing financial statements (Bartov et al., 2000; Collins et al., 2003; Miller, 2010). Hence, the additional information obtained through comparable accounting information could be low. Table 7 shows that the effect of comparability on improving investors’ decisions (e.g., selling or holding investments) (FASB, 2010) is not equal across investors and demonstrates that comparability provides more benefits to individual investors than institutional investors. In addition, given that investors experience low subsequent stock returns after periods of high investor sentiment (Miwa, 2016), we can anticipate that comparability can prevent individual investors from suffering economic losses in advance.

Table 8 shows the results of examining whether the effect of comparability in mitigating differences in investors’ trading behaviours due to investment sentiment differs between companies that are difficult to value and those that are easy to value. Firms whose corporate value is difficult to assess typically are covered by fewer financial analysts, are young, volatile, non-dividend-paying, with extreme MTB; these are presented in columns (1), (3), (5), (7), and (9), respectively. Alternatively, high analyst coverage, mature, stable, dividend-paying, and non-extreme MTB companies are presented in columns (2), (4), (6), (8), and (10), respectively. In Panel A, the results of CompIND are reported, and in Panel B, the results of Comp4 are reported. Panels A and B consistently show that the coefficient of SENT×CompIND (Comp4) is significantly negative in the case of low coverage, young, volatile, non-dividend-paying, and extreme MTB companies. In contrast, non-significant values are shown for high coverage, mature, stable, dividend-paying, and non-extreme MTB companies. This result shows that the effect of comparability in mitigating the differences in trading behaviour due to investor sentiment is evident for hard-to-value companies, which supports our Hypothesis 2-1. The more difficult it is to evaluate a corporate value, the greater the amount and quality of information investors can obtain from comparable information to its industry peers, thereby maximizing the comparability effect.

Table 8. THE EFFECT OF COMPARABILITY ON THE RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND TRADING BEHAVIOUR: DEPENDING ON COMPLEX VALUATION
Panel A: Use CompIND
Variable (1) Low coverage (2) High coverage (3) Young (4) Mature (5) Volatile (6) Stable (7) Non-dividend (8) Dividend (9) Extreme MTB (10) Non-extreme MTB
Intercept –0.496 –0.017 –0.365 –0.161 –0.153 –0.175 –0.001 –0.331 –0.398 –0.202
(–1.01) (–0.07) (–1.05) (–0.46) (–0.45) (–0.47) (0.00) (–1.19) (–1.16) (–0.60)
SENTt 0.011*** 0.015*** 0.016*** 0.010*** 0.008** 0.014*** 0.004 0.014*** 0.007** 0.019***
(2.58) (5.84) (4.21) (2.80) (2.30) (3.29) (0.77) (4.70) (1.98) (5.31)
CompINDt –1.398** 0.156 –0.653 –0.567 –0.862* 1.403 –1.189* 1.082 –1.265** 0.108
(–2.13) (0.21) (–1.04) (–0.92) (–1.81) (1.21) (–1.96) (1.28) (–2.30) (0.13)
SENTt×CompINDt –0.297** 0.043 –0.200** 0.012 –0.143* 0.081 –0.165* –0.011 –0.272** 0.031
(–2.31) (0.40) (–2.09) (0.11) (–1.84) (0.45) (–1.75) (–0.06) (–2.54) (0.23)
SIZEt 0.001 –0.003 –0.006 0.001 –0.009 –0.003 –0.008 –0.001 0.003 –0.007
(0.01) (–0.29) (–0.38) (0.09) (–0.64) (–0.17) (–0.36) (–0.04) (0.21) (–0.45)
ROAt –0.093 0.091 –0.047 0.104 –0.009 –0.092 0.054 –0.131 –0.280 0.371*
(–0.44) (0.60) (–0.25) (0.53) (–0.06) (–0.38) (0.25) (–0.68) (–1.54) (1.82)
RETt –0.026 0.030 –0.037 0.029 0.006 0.172** –0.018 0.034 0.039 –0.045
(–0.62) (1.12) (–0.93) (0.85) (0.19) (2.19) (–0.38) (1.05) (1.09) (–1.18)
LEVt 0.101 0.011 0.058 0.048 0.154** –0.098 0.089 0.062 0.054 0.061
(1.27) (0.23) (0.86) (0.71) (2.45) (–1.29) (0.81) (1.11) (0.81) (0.90)
LNPRICEt 0.031** 0.001 0.021* 0.012 0.022** 0.011 0.024 0.014 0.019* 0.020*
(2.34) (0.02) (1.74) (1.08) (1.99) (0.98) (1.13) (1.57) (1.72) (1.71)
ΔVKOSPIt –0.067*** –0.050*** –0.068*** –0.051*** –0.005 –0.099*** 0.011 –0.084*** –0.045*** –0.076***
(–3.16) (–4.77) (–3.88) (–3.20) (–0.27) (–5.65) (0.44) (–6.50) (–2.59) –4.83)
ΔTERMt –0.016 –0.023 –0.006 –0.040 –0.174*** 0.121** –0.221*** 0.048 –0.063 0.023
(–0.27) (–0.82) (–0.12) (–0.91) (–3.32) (2.56) (–3.05) (1.34) (–1.31) (0.52)
ΔCREDITt 0.466* 0.418*** 0.603*** 0.268 –0.296 1.308*** –0.477 0.780*** 0.247 0.650***
(1.81) (3.35) (2.91) (1.39) (–1.53) (5.66) (–1.51) (5.06) (1.18) (3.41)
STDBUSt 0.093* 0.096*** 0.141*** 0.053 0.016 0.171*** –0.028 0.144*** 0.076* 0.119***
(1.89) (4.14) (3.70) (1.44) (0.48) (3.70) (–0.47) (4.96) (1.90) (3.39)
DISCLt –0.009 0.001 –0.003 –0.004 –0.002 –0.004 0.008 –0.005 –0.006 –0.001
(–0.80) (0.01) (–0.68) (–0.77) (–0.33) (–0.96) (0.79) (–1.38) (–1.26) (–0.19)
BODt 0.028 –0.022 0.047 –0.036 –0.005 0.032 –0.010 0.002 0.020 –0.018
(0.62) (–0.79) (1.14) (–0.95) (–0.13) (0.78) (–0.16) (0.07) (0.51) (–0.46)
BINDt 0.085 –0.010 –0.043 0.094 0.064 –0.041 0.020 0.030 0.013 –0.010
(0.64) (–0.14) (–0.38) (0.90) (0.64) (–0.37) (0.12) (0.36) (0.13) (–0.09)
LARGEt 0.010 0.005 0.081 –0.027 0.018 0.022 –0.035 0.005 –0.064 0.097
(0.11) (0.09) (1.00) (–0.35) (0.24) (0.26) (–0.30) (0.08) (–0.79) (1.25)
Year and industry fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Cluster Firm Firm Firm Firm Firm Firm Firm Firm Firm Firm
Adj. R2 0.0039 0.0709 0.0276 0.0096 0.0043 0.0736 0.0210 0.0420 0.0181 0.0259
Obs 1,780 1,656 1,749 1,687 1,942 1,494 941 2,495 1,720 1,716
Panel B: Use Comp4
Variable (1) Low coverage (2) High coverage (3) Young (4) Mature (5) Volatile (6) Stable (7) Non-dividend (8) Dividend (9) Extreme MTB (10) Non-extreme MTB
Intercept –0.501 –0.032 –0.366 –0.146 –0.121 –0.179 0.031 –0.339 –0.391 –0.210
(–1.03) (–0.13) (–1.05) (–0.42) (–0.36) (–0.49) (0.06) (–1.22) (–1.11) (–0.62)
SENTt 0.011*** 0.015*** 0.016*** 0.010*** 0.008** 0.014*** 0.004 0.014*** 0.007** 0.019***
(2.61) (5.76) (4.16) (2.85) (2.37) (3.17) (0.84) (4.77) (2.00) (5.31)
Comp4t –1.988** 0.711 –0.777 –1.832** –1.315** –0.158 –1.711** 1.133 –1.885*** 0.778
(–2.52) (0.72) (–1.01) (–2.31) (–2.23) (–0.10) (–2.40) (0.90) (–2.82) (0.73)
SENTt×Comp4t –0.350*** 0.000 –0.287** –0.195 –0.258*** –0.159 –0.268** –0.221 –0.327*** 0.049
(–2.63) (0.00) (–2.34) (–1.41) (–2.60) (–0.63) (–2.32) (–0.87) (–2.94) (0.27)
SIZEt 0.001 –0.002 –0.006 –0.001 –0.011 –0.003 –0.010 0.001 0.002 –0.006
(0.00) (–0.21) (–0.38) (–0.01) (–0.75) (–0.18) (–0.44) (0.01) (0.13) (–0.42)
ROAt –0.086 0.085 –0.055 0.113 –0.011 –0.055 0.061 –0.136 –0.272 0.347*
(–0.41) (0.56) (–0.30) (0.57) (–0.07) (–0.22) (0.27) (–0.71) (–1.49) (1.70)
RETt –0.025 0.030 –0.033 0.033 0.006 0.182** –0.018 0.034 0.040 –0.046
(–0.60) (1.13) (–0.84) (0.96) (0.22) (2.32) (–0.39) (1.05) (1.14) (–1.20)
LEVt 0.105 0.010 0.058 0.055 0.159** –0.093 0.096 0.062 0.058 0.060
(1.31) (0.20) (0.86) (0.80) (2.52) (–1.22) (0.87) (1.11) (0.86) (0.89)
LNPRICEt 0.032** –0.001 0.021* 0.013 0.023** 0.011 0.025 0.014 0.020* 0.020*
(2.37) (–0.01) (1.74) (1.16) (2.04) (0.98) (1.16) (1.55) (1.75) (1.72)
ΔVKOSPIt –0.067*** –0.050*** –0.068*** –0.051*** –0.005 –0.098*** 0.010 –0.083*** –0.046*** –0.076***
(–3.16) (–4.75) (–3.90) (–3.21) (–0.29) (–5.61) (0.40) (–6.49) (–2.62) (–4.82)
ΔTERMt –0.015 –0.023 –0.004 –0.039 –0.169*** 0.120** –0.215*** 0.048 –0.062 0.022
(–0.26) (–0.81) (–0.08) (–0.88) (–3.24) (2.53) (–2.97) (1.35) (–1.29) (0.51)
ΔCREDITt 0.469* 0.418*** 0.604*** 0.281 –0.292 1.302*** –0.468 0.782*** 0.253 0.645***
(1.82) (3.35) (2.91) (1.46) (–1.52) (5.63) (–1.48) (5.07) (1.21) (3.39)
STDBUSt 0.094* 0.095*** 0.142*** 0.055 0.018 0.169*** –0.024 0.143*** 0.078* 0.119***
(1.91) (4.11) (3.73) (1.50) (0.53) (3.67) (–0.41) (4.95) (1.94) (3.38)
DISCLt –0.008 –0.001 –0.003 –0.003 –0.001 –0.004 0.008 –0.005 –0.006 –0.001
(−0.78) (–0.02) (–0.67) (–0.71) (–0.26) (–0.97) (0.81) (–1.41) (–1.16) (–0.19)
BODt 0.029 –0.022 0.048 –0.038 –0.005 0.032 –0.010 0.002 0.020 –0.019
(0.63) (–0.81) (1.17) (–1.00) (–0.13) (0.78) (–0.16) (0.07) (0.51) (–0.50)
BINDt 0.086 –0.013 –0.045 0.095 0.068 –0.040 0.023 0.026 0.018 –0.012
(0.65) (–0.18) (–0.41) (0.92) (0.68) (–0.36) (0.14) (0.31) (0.17) (–0.11)
LARGEt 0.014 0.004 0.079 –0.021 0.021 0.024 –0.031 0.004 –0.062 0.095
(0.14) (0.08) (0.97) (–0.27) (0.29) (0.28) (–0.27) (0.06) (–0.77) (1.23)
Year and industry fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Cluster Firm Firm Firm Firm Firm Firm Firm Firm Firm Firm
Adj. R2 0.0050 0.0710 0.0286 0.0120 0.0060 0.0730 0.0181 0.0616 0.0185 0.0262
Obs 1,780 1,656 1,749 1,687 1,942 1,494 941 2,495 1,720 1,716
  • Low (high) coverage firms are firms with the number of analysts below (above) the sample median. Young (mature) firms have ages below (above) the sample median. Volatile (stable) firms are firms with volatility below (above) the sample median. Extreme MTB (non-extreme MTB) firms are firms with MTB in the top and bottom (middle) quartiles. All continuous variables are winsorized at the top and bottom one percentile. t-statistics are shown in parentheses. t-values are based on standard errors clustered by firm and year. ***, **, and * indicate significance at the 0.01, 0.05, and 0.10 levels, respectively (based on two-tailed tests). See the Appendix for variable definitions.

Table 9 shows the results of analysing whether the effect of comparability on investors’ trading behaviours differs depending on industry concentration. Whereas columns (1) and (2) show a high degree of industry concentration (the degree of competition within the industry is low), columns (3) and (4) indicate low industry concentration (the degree of competition within the industry is high). SENT×CompIND (Comp4) produces significant negative coefficient values in columns (1) and (2) with a low degree of competition within the industry, while insignificant values in columns (3) and (4), indicating that the benefits of comparable information are negligible in areas with increased competition within the industry. The intense competition encourages companies to recognize bad news quickly and adopt conservative accounting measures to hinder the entry of new competitors (Dhaliwal et al., 2014), lowering investor sentiment. The positive correlation coefficient between HHI and SENT presented in Table 5 supports this argument. Fierce competition makes it harder to create optimistic investor sentiment itself. Thus, the effect of comparability in mitigating differences in trading behaviour between individual and institutional investors induced by investor sentiment is minimal.

Table 9. THE EFFECT OF COMPARABILITY ON THE RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND TRADING BEHAVIOUR
Variable High HHI Low HHI
(1) ΔDIFFt+1 (2) ΔDIFFt+1 (3) ΔDIFFt+1 (4) ΔDIFFt+1
Intercept 0.083 0.039 –0.576 –0.566
(0.27) (0.12) (–1.64) (–1.61)
SENTt 0.012*** 0.012*** 0.012*** 0.012***
(3.38) (3.34) (3.47) (3.50)
CompINDt –1.362** 0.719
(–2.45) (0.87)
Comp4t –1.835*** 0.099
(–2.61) (0.11)
SENTt×CompINDt –0.296*** 0.099
–2.74) (0.69)
SENTt×Comp4t –0.327*** 0.002
(–2.88) (0.01)
SIZEt –0.017 –0.016 0.009 0.009
(–1.21) (–1.06) (0.60) (0.58)
ROAt –0.094 –0.089 0.100 0.107
(–0.53) (–0.47) (0.54) (0.57)
RETt –0.007 –0.006 0.010 0.011
(–0.18) (–0.15) (0.30) (0.33)
LEVt 0.020 0.033 0.092 0.092
(0.35) (0.50) (1.35) (1.35)
LNPRICEt 0.022** 0.025** 0.011 0.012
(2.00) (2.14) (1.09) (1.10)
ΔVKOSPIt –0.055*** –0.054*** –0.068*** –0.068***
(–3.27) (–3.15) (–4.13) (−4.11)
ΔTERMt –0.036 –0.029 –0.008 –0.007
(–0.78) (–0.62) (–0.17) (–0.16)
ΔCREDITt 0.372* 0.384* 0.539*** 0.539***
(1.84) (1.87) (2.74) (2.74)
STDBUSt 0.066* 0.061 0.140*** 0.139***
(1.73) (1.55) (3.73) (3.71)
DISCLt –0.002 –0.003 –0.005 –0.005
(–0.38) (–0.62) (–0.96) (–0.97)
BODt 0.022 0.021 –0.008 –0.009
(0.54) (0.50) (–0.22) (–0.25)
BINDt 0.111 0.100 –0.052 –0.050
(1.03) (0.90) (–0.52) (–0.50)
LARGEt 0.034 0.015 0.005 0.006
(0.46) (0.18) (0.07) (0.08)
Year and industry fixed effects Yes Yes Yes Yes
Cluster Firm Firm Firm Firm
Adj. R2 0.0273 0.0191 0.0248 0.0243
Obs. 1,720 1,720 1,716 1,716
  • All continuous variables are winsorized at the top and bottom one percentile. t-statistics are presented in parentheses. t-values are based on standard errors clustered by firm. ***, **, and * indicate significance at the 0.01, 0.05, and 0.10 levels, respectively (based on two-tailed tests). See the Appendix for variable definitions.

ADDITIONAL ANALYSES AND SENSITIVITY CHECKS

Additional Tests

According to previous studies, investors tend to process information more systematically and meticulously when they have pessimistic beliefs (Taylor, 1991; Bless et al., 1996; Simpson, 2013). If a company's stock price is low due to low sentiment, they will likely try to boost their stock price (Lambert, 2006). For example, managers can signal to the market that stocks are undervalued through corporate actions such as stock acquisitions (Lambert, 2006). In a similar vein, Bergman and Roychowdhury (2008) provide evidence that managers disclose earnings forecasts to stimulate stock prices during a low investor sentiment period. Therefore, when investor sentiment is pessimistic, the effect of comparability may be small because managers make various efforts to raise their stock prices. To test this possibility, similar to Wang (2018), we partition the sample into three segments based on SENT and use two dummy variables, HIGH and LOW, as proxies for the optimistic and pessimistic sentiments. We use the following regression equation (15):
ΔDIFF t + 1 = β 0 + β 1 HIGH LOW t + β 2 COMP t + β 3 HIGH LOW t × COMP t + β 4 SIZE t + β 5 ROA t + β 6 RET t + β 7 LEV t + β 8 LNPRICE t + β 9 ΔVKOSPI t + β 10 ΔTERM t + β 11 ΔCREDIT t + β 12 STDBUS t + β 13 DISCL t + β 14 BOD t + β 15 BIND t + β 16 LARGE t + Year FE + Industry FE + ε $$ {\varDelta DIFF}_{t+ 1}={\beta}_0+{\beta}_1 HIGH{(LOW)}_t+{\beta}_2{COMP}_t+{\beta}_3 HIGH{(LOW)}_t\times {COMP}_t+{\beta}_4{SIZE}_t+{\beta}_5{ROA}_t+{\beta}_6{RET}_t+{\beta}_7{LEV}_t+{\beta}_8{LNPRICE}_t+{\beta}_9{\varDelta VKOSPI}_t+{\beta}_{1 0}{\varDelta TERM}_t+{\beta}_{1 1}{\varDelta CREDIT}_t+{\beta}_{1 2}{STDBUS}_t+{\beta}_{1 3}{DISCL}_t+{\beta}_{1 4}{BOD}_t+{\beta}_{1 5}{BIND}_t+{\beta}_{1 6}{LARGE}_t+ Year\ FE+ Industry\ FE+\varepsilon $$ ()

HIGH (LOW) is a dummy variable with a value of one if SENT is 0.8 or more (0.3 or less) and zero otherwise. If the coefficients of both HIGH × COMP and LOW × COMP have significant negative values, the indication is that comparability mitigates the differences in investors’ trading behaviour due to investor sentiment, regardless of optimistic or pessimistic sentiment. The results of equation (15) are presented in Table 10. The regression coefficients of HIGH×CompIND and HIGH×Comp4 in columns (1) and (2) are –0.415 and –0.438, respectively. They show a significant negative value, whereas the regression coefficients of LOW×CompIND and LOW×Comp4 in columns (3) and (4) are insignificant. This means that the impact of comparability in mitigating differences in investors’ trading behaviours due to investor sentiment is more evident when sentiment is optimistic than pessimistic. These findings highlight the need for investors to make investment decisions based on comparable accounting information, especially when investor sentiment is high.

Table 10. THE EFFECT OF COMPARABILITY ON THE RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND TRADING BEHAVIOUR: SEPARATING HIGH AND LOW INVESTOR SENTIMENT
Variable (1) ΔDIFFt+1 (2) ΔDIFFt+1 (3) ΔDIFFt+1 (4) ΔDIFFt+1
Intercept –0.152 –0.163 –0.171 –0.172
(–0.67) (–0.66) (–0.62) (–0.62)
HIGHt 0.040* 0.041*
(1.93) (1.94)
CompINDt –0.214 –0.046
(–0.34) (–0.06)
Comp4t –0.698 –0.074
(–0.92) (–0.08)
HIGHt×CompINDt –0.415**
(–2.15)
HIGHt×Comp4t –0.438**
(–2.28)
LOWt –0.060*** –0.061***
(–2.73) (–2.75)
LOWt×CompINDt –0.328
(–0.40)
LOWt×Comp4t –0.108
(–0.11)
SIZEt –0.011 –0.007 –0.002 –0.002
(–1.01) (–0.66) (–0.17) (–0.17)
ROAt –0.095 –0.083 0.107 0.098
(–0.60) (–0.49) (0.65) (0.59)
RETt 0.012 0.013 0.092*** 0.092***
(0.58) (0.60) (2.61) (2.60)
LEVt 0.001 0.025 0.073 0.074
(0.01) (0.43) (1.26) (1.27)
LNPRICEt 0.019** 0.021** 0.011 0.012
(2.36) (2.33) (1.20) (1.21)
ΔVKOSPIt –0.054*** –0.001 –0.054*** –0.054***
(–4.82) (–0.32) (–4.49) (–4.47)
ΔTERMt –0.034 –0.217*** –0.042 –0.042
(–1.08) (–5.21) (–1.23) (–1.23)
ΔCREDITt 0.310** –0.765*** 0.444*** 0.441***
(2.17) (–3.24) (3.04) (3.03)
STDBUSt 0.104*** 0.029 0.083*** 0.084***
(3.68) (1.24) (2.94) (2.94)
DISCLt –0.003 –0.004 –0.001 –0.001
(–1.19) (–1.62) (–0.40) (–0.40)
BODt 0.022 0.016 –0.008 –0.008
(0.68) (0.50) (–0.23) (–0.23)
BINDt 0.143* 0.137 0.001 0.002
(1.70) (1.60) (0.00) (0.02)
LARGEt 0.024 0.036 –0.023 –0.023
(0.40) (0.56) (–0.34) (–0.34)
Year and industry fixed effects Yes Yes Yes Yes
Cluster Firm Firm Firm Firm
Adj. R2 0.0271 0.0320 0.0332 0.0323
Obs. 2,506 2,506 2,559 2,559
  • All continuous variables are winsorized at the top and bottom one percentile. t-statistics are presented in parentheses. t-values are based on standard errors clustered by firm. ***, **, and * indicate significance at the 0.01, 0.05, and 0.10 levels, respectively (based on two-tailed tests). See the Appendix for variable definitions.

Sensitivity Test

In this section, we examine whether the study results are robust when using trading amounts, alternative measures of comparability, alternative measures of investor sentiment, and when extending the definition of sophisticated investors to foreign investors.

Using Trading Amount

We examine whether the results are robust when using trading amount instead of trading volume in measuring each investor's trading behaviour. The variable (ΔADIFF) is the difference between individual and institutional investors’ net buying amounts. Table 11 shows the results using ΔADIFFt+1 as a dependent variable. The analysis shows that the coefficients of SENT in columns (1) and (2) are both 0.009, showing significant positive values. That is, the higher the investor sentiment, the larger the difference in trading amount between investors. The coefficients of SENT×CompIND and SENT×Comp4 are –0.244 and –0.259, which are significantly negative. This finding shows that comparability mitigates the differences in trading amount as well as the trading volume between individual and institutional investors driven by investor sentiment.

Table 11. THE EFFECT OF COMPARABILITY ON THE RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND TRADING BEHAVIOUR: USING THE TRADING AMOUNT
Variable (1) ΔADIFFt+1 (2) ΔADIFFt+1
Intercept –0.123 –0.109
(–0.58) (–0.51)
SENTt 0.009*** 0.009***
(3.22) (3.21)
CompINDt –1.045*
(–1.65)
Comp4t –1.296*
(–1.80)
SENTt×CompINDt –0.244*
(–1.94)
SENTt×Comp4t –0.259**
(–2.05)
SIZEt –0.003 –0.003
(–0.35) (–0.35)
ROAt –0.031 –0.029
(–0.23) (–0.66)
RETt 0.016 0.019
(0.63) (0.72)
LEVt 0.035 0.034
(0.73) (0.73)
LNPRICEt 0.017** 0.016**
(2.39) (2.28)
ΔVKOSPIt –0.007 –0.007
(–1.18) (–1.17)
ΔTERMt –0.106*** –0.107***
(–3.77) (–3.81)
ΔCREDITt –0.038 –0.044
(–0.26) (–0.29)
STDBUSt 0.008 0.009
(0.29) (0.33)
DISCLt –0.003 –0.003
(–1.26) (–1.16)
BODt 0.005 0.006
(0.18) (0.21)
BINDt 0.027 0.027
(0.37) (0.37)
LARGEt 0.015 0.006
(0.28) (0.12)
Year and industry fixed effects Yes Yes
Cluster firm Firm
Adj. R2 0.0273 0.0275
Obs. 3,436 3,436
  • All continuous variables are winsorized at the top and bottom one percentile. t-statistics are presented in parentheses. t-values are based on standard errors clustered by firm. ***, **, and * indicate significance at the 0.01, 0.05, and 0.10 levels, respectively (based on two-tailed tests). See the Appendix for variable definitions.

Use of Alternative Comparability Measures

Earning consists of cash flows and accruals, which are determined according to accrual basis accounting. If two entities apply different accounting principles to accruals, they will have different accrual structures, and the comparability of accounting information will be low. From this perspective, Francis et al. (2014) measure comparability through the similarity of accrual structures. We use this measure to examine whether our results are robust even when using alternative comparability measures. Specifically, suppose that the accounting information of two companies, i and j, are comparable. In that case, the difference in accruals between the companies is expected to be small if the structure of the accruals is similar. We replace earnings in equations (11)–(13) with accruals and then perform the same procedure described in the description of comparability measurement. CompIND_AC and Comp4_AC are accrual-based comparabilities. Table 12 shows the results of using alternative measures of comparability using accruals. The coefficients of SENT×CompIND_AC and SENT×Comp4_AC4 are –0.165 and –0.166, respectively. This shows the robustness of our results.

Table 12. THE EFFECT OF COMPARABILITY ON THE RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND TRADING BEHAVIOUR: USING ALTERNATIVE COMPARABILITY
Variable (1) ΔDIFFt+1 (2) ΔDIFFt+1
Intercept –0.187 –0.297
(–0.88) (–1.41)
SENTt 0.010*** 0.140***
(3.54) (4.02)
CompINDt –0.714
(–1.42)
Comp4t –0.710
(–1.43)
SENTt×CompIND_ACt –0.165*
(–1.68)
SENTt×Comp4_ACt –0.166*
–1.73)
SIZEt –0.003 –0.003
(–0.35) (–0.34)
ROAt –0.052 –0.039
(–0.38) (–0.28)
RETt 0.011 0.024
(0.44) (1.14)
LEVt 0.037 0.034
(0.80) (0.72)
LNPRICEt 0.017** 0.018**
(2.31) (2.49)
ΔVKOSPIt –0.021*** –0.025***
(–2.73) (–3.25)
ΔTERMt –0.095*** –0.104***
(–4.06) (–4.43)
ΔCREDITt 0.070 0.068
(0.66) (0.63)
STDBUSt 0.049 0.059**
(2.13) (2.56)
DISCLt –0.003 –0.003
(–1.35) (–1.24)
BODt 0.006 0.005
(0.21) (0.16)
BINDt 0.034 0.031
(0.47) (0.41)
LARGEt 0.008 0.003
(0.15) (0.05)
Year and industry fixed effects Yes Yes
Cluster Firm Firm
Adj. R2 0.0297 0.0298
Obs. 3,436 3,436
  • All continuous variables are winsorized at the top and bottom one percentile. t-statistics are presented in parentheses. t-values are based on standard errors clustered by firm. ***, **, and * indicate significance at the 0.01, 0.05, and 0.10 levels, respectively (based on two-tailed tests). See the Appendix for variable definitions.

Use of Alternative Investor Sentiment Measures

We examine whether our results remain consistent using the market-wide investor sentiment index. Several previous studies use the consumer sentiment index (Lemmon and Portniaguina, 2006; Schmeling, 2009) as a proxy for investor sentiment. In Korea, the Bank of Korea publishes the Composite Consumer Sentiment Index (CCSI) every month. A CCSI that is higher (lower) than 100 indicates that consumers are more optimistic (pessimistic) about the current economy than they were in the past. MSENT is a dummy variable that takes the value one if the annual average of the CCSI is greater than 100, and zero otherwise. Table 13 shows the results of using alternative measures of investor sentiment. The coefficients of MSENT×CompIND and MSENT×Comp4 are –1.409, and –1.472, respectively. This shows the robustness of our results.

Table 13. THE EFFECT OF COMPARABILITY ON THE RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND TRADING BEHAVIOUR: USING ALTERNATIVE INVESTOR SENTIMENT
Variable (1) ΔDIFFt+1 (2) ΔDIFFt+1
Intercept 55.966*** 55.789***
(4.42) (4.41)
MSENTt 0.574*** 0.572***
(4.43) (4.42)
CompINDt 0.364
(0.77)
Comp4t 0.486
(0.83)
MSENTt×CompINDt –1.409*
(–1.82)
MSENTt×Comp4t –1.472*
(–1.77)
SIZEt 0.001 0.001
(0.06) (0.08)
ROAt 0.033 0.001
(0.23) (0.00)
RETt 0.086*** 0.087***
(4.99) (5.10)
LEVt 0.034 0.031
(0.67) (0.63)
LNPRICEt 0.016** 0.017**
(2.03) (2.11)
ΔVKOSPIt –0.486*** –0.485***
(–4.60) (–4.60)
ΔTERMt 4.327*** 4.314***
(4.32) (4.32)
ΔCREDITt 6.801*** 6.781***
(4.46) (4.46)
STDBUSt 1.633*** 1.627***
(4.47) (4.46)
DISCLt –0.002 –0.002
(–0.91) (–0.89)
BODt 0.006 0.006
(0.20) (0.19)
BINDt 0.027 0.026
(0.36) (0.34)
LARGEt –0.001 –0.001
(–0.02) (–0.01)
Year and industry fixed effects Yes Yes
Cluster Firm Firm
Adj. R2 0.0295 0.0295
Obs. 3,436 3,436
  • All continuous variables are winsorized at the top and bottom one percentile. t-statistics are presented in parentheses. t-values are based on standard errors clustered by firm. ***, **, and * indicate significance at the 0.01, 0.05, and 0.10 levels, respectively (based on two-tailed tests). See the Appendix for variable definitions.

Expanding the Scope of Sophisticated Investors

Based on previous studies, we define institutional investors as sophisticated investors and individual investors as unsophisticated investors. Some studies suggest that foreign investors can be classified as sophisticated investors (Chan et al., 2008). According to Grinblatt and Keloharju (2000), institutional investors are more professional than individual investors, and foreign investors are more experienced than institutional investors. Additionally, previous studies find that foreign investors have higher returns than domestic investors in the Korean stock market (Ko and Kim, 2004), supporting the view that foreign investors can be classified as sophisticated investors. Based on this view, we expand the range of sophisticated investors to include foreign investors. The difference in the net buying volume between individual and sophisticated investors (i.e., institutional and foreign investors) is defined as TDIFFt (IBS_INDI – IBS_INST-IBS_FOR). Table 14 shows the results of ΔTDIFFt +1, the difference between the following and current TDIFF, as a dependent variable.

Table 14. THE EFFECT OF COMPARABILITY ON THE RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND TRADING BEHAVIOUR: EXTENDING THE DEFINITIONS OF PROFESSIONAL INVESTORS
Variable (1) ΔTDIFFt+1 (2) ΔTDIFFt+1
Intercept –0.119 –0.264
(–0.51) (–1.01)
SENTt 0.017*** 0.019***
(5.23) (5.70)
CompINDt –0.979
(–1.51)
Comp4t –1.481*
(–1.92)
SENTt×CompINDt –0.227*
–1.81)
SENTt×Comp4t –0.270**
(–2.01)
SIZEt 0.001 0.001
(0.03) (0.03)
ROAt –0.050 –0.021
(–0.34) (–0.14)
RETt 0.016 0.004
(0.59) (0.15)
LEVt 0.028 0.051
(0.54) (0.93)
LNPRICEt 0.0177** 0.018**
(2.06) (2.01)
ΔVKOSPIt 0.007 –0.062***
(1.60) (–5.43)
ΔTERMt –0.0717** 0.049
(–2.20) (1.50)
ΔCREDITt –0.074 0.654***
(–0.43) (4.70)
STDBUSt –0.023 0.094***
(–1.22) (3.44)
DISCLt –0.003 –0.004
(–1.24) (–1.41)
BODt –0.013 –0.012
(–0.41) (–0.35)
BINDt 0.047 0.030
(0.60) (0.37)
LARGEt 0.011 0.004
(0.18) (0.06)
Year FE, Industry FE Yes Yes
Cluster Firm Firm
Adj. R2 0.0346 0.0461
Obs. 3,436 3,436
  • All continuous variables are winsorized at the top and bottom one percentile. t-statistics are presented in parentheses. t-values are based on standard errors clustered by firm. ***, **, and * indicate significance at the 0.01, 0.05, and 0.10 levels, respectively (based on two-tailed tests). See the Appendix for variable definitions.

In Table 14, the coefficients of SENT×CompIND and SENT×Comp4 are –0.227 and –0.270, respectively, indicating significant negative values. This means that the difference between the trading behaviours of unsophisticated and sophisticated investors due to investor sentiment is mitigated by comparability. Although not tabulated, when we use ΔIBS_FORt+1 as a dependent variable instead of ΔTADIFFt+1, SENT×CompIND and SENT×Comp4 show insignificant values. Considering that most foreign investors in the Korean market are institutional investors (Bank of Korea), they benefit little from comparable information, as they already have expertise in valuation.

CONCLUSION

Individual investors who lack expertise in financial analysis dominate the Korean stock market, and thus investor sentiment can be easily observed. This study examines whether comparability mitigates the difference between the trading behaviours of individual and institutional investors due to investor sentiment. We find that differences between individual and institutional investors’ trading behaviours driven by investor sentiment are lower for firms that are more comparable with their industry peers. When we divided investors into individual and institutional investors, the tendency of individual investors to net buy stocks with high investor sentiment was lower for more comparable firms. In contrast, institutional investors’ tendency to net sell stocks with high investor sentiment was unrelated to comparability. In cross-sectional tests, we find that the effect of comparability in mitigating the differences between the trading behaviours of investors due to investor sentiment is evident for firms that are hard to value (e.g., young, volatile, non-dividend-paying, extreme MTB firms, and firms with low analyst coverage) and firms in a less competitive industry.

Our study is the first to reveal that comparability mitigates the divergence in trading behaviours between individual and institutional investors caused by investor sentiment. Furthermore, our research extends prior studies that have explored the advantages of comparability for equity investors (Baik et al., 2013; Kim et al., 2016; Choi et al., 2019). This study sheds light on which types of investors can benefit more from comparability by exploiting the unique data on trading volume and amounts for different types of investors.

This study has the following limitations. First, we use representative and alternative measures of comparability, but there are doubts as to whether these proxies fully measure the comparability concept. Second, there may be a problem with omitted variables because we do not control for all variables that could affect each investor's trading behaviour.

APPENDIX A: VARIABLE DEFINITIONS

RSIt Relative strength index
PLIt Psychological line index
ATRt Adjusted turnover rate
LTVt Logarithm of trading volume
SENTt Investor sentiment index for each company
MSENTt Indicator variable which equals one if the annual average of CCSI is greater than 100, and zero otherwise
HIGHt Indicator variable which equals one if SENT is greater than or equal to 0.8, and zero for the middle sentiment group (0.3<SENT<0.8)
LOWt Indicator variable which equals one if SENT is less than or equal to 0.3, and zero for the middle sentiment group (0.3<SENT<0.8)
IBS_INDIt

∑ INDI _ BUY i , t − ∑ INDI _ SELL i , t ∑ INDI _ BUY i , t + ∑ INDI _ SELL i , t $$ \frac{\sum INDI\_{BUY}_{i,t}-\sum INDI\_{SELL}_{i,t}}{\sum INDI\_{BUY}_{i,t}+\sum INDI\_{SELL}_{i,t}} $$

where INDI_BUY(SELL) = volume of buying (selling) by individual investors for the year
IBS_INSTt

∑ INST _ BUY i , t − ∑ INST _ SELL i , t ∑ INST _ BUY i , t + ∑ INST _ SELL i , t $$ \frac{\sum INST\_{BUY}_{i,t}-\sum INST\_{SELL}_{i,t}}{\sum INST\_{BUY}_{i,t}+\sum INST\_{SELL}_{i,t}} $$

where INST_BUY(SELL) = volume of buying (selling) by institutional investors for the year
IBS_FORt

∑ FOR _ BUY i , t − ∑ FOR _ SELL i , t ∑ FOR _ BUY i , t + ∑ FOR _ SELL i , t $$ \frac{\sum FOR\_{BUY}_{i,t}-\sum FOR\_{SELL}_{i,t}}{\sum FOR\_{BUY}_{i,t}+\sum FOR\_{SELL}_{i,t}} $$

where FOR_BUY(SELL) = volume of buying (selling) by foreign investors for the year
ΔDIFFt+1 DIFF for the year t+1 minus DIFF for the year t, where DIFF=IBS_INDI-IBS_INST
ΔADIFFt+1

ADIFF for the year t+1 minus ADIFF for the year t, where ADIFF=IBS_AINDI-INS_AINST

IBS_AINDI= ∑ INDI _ ABUY i , t − ∑ INDI _ ASELL i , t ∑ INDI _ ABUY i , t + ∑ INDI _ ASELL i , t $$ \frac{\sum INDI\_{ABUY}_{i,t}-\sum INDI\_{ASELL}_{i,t}}{\sum INDI\_{ABUY}_{i,t}+\sum INDI\_{ASELL}_{i,t}} $$

IBS_AINST= ∑ INST _ ABUY i , t − ∑ INST _ ASELL i , t ∑ INST _ ABUY i , t + ∑ INST _ ASELL i , t $$ \frac{\sum INST\_{ABUY}_{i,t}-\sum INST\_{ASELL}_{i,t}}{\sum INST\_{ABUY}_{i,t}+\sum INST\_{ASELL}_{i,t}} $$

INDI_ABUY(INST_ABUY) = individual (institutional) investor's buying amount

INDI_ASELL (INST_ASELL) = individual(institutional) investor's selling amount
ΔTDIFFt+1 TDIFF for the year t+1 minus TDIFF for the year t, where TDIFF=IBS_INDI-IBS_INST- IBS_FOR
CompAcctijt

Comparability between firms i and j (CompAcctijt), calculated as the negative value of the average absolute difference between the predicted earnings using firm i and j's functions

CompAcct ijt = − 1 / 16 × ∑ t − 15 t E Earnings iit − E Earnings ijt $$ {CompAcct}_{ijt}=-1/16\times \sum \limits_{t-15}^t\left|E\left({Earnings}_{iit}\right)-E\left({Earnings}_{ijt}\right)\right| $$

Comp4t Average of the four highest CompAcctijt
CompINDt Median value of CompAcctijt
CompAcct_ACijt

Comparability between firms i and j (CompAcct_ACijt), calculated as the negative value of the average absolute difference between the predicted accrual using firm i and j's functions:

CompAcct _ AC ijt = − 1 / 16 × ∑ t − 15 t E Accrual iit − E Accrual ijt $$ CompAcct\_{AC}_{ijt}=-1/16\times \sum \limits_{t-15}^t\left|E\left({Accrual}_{iit}\right)-E\left({Accrual}_{ijt}\right)\right| $$

Comp4_ACt Average of the four highest CompAcct_ACijt
CompIND_ACt Median value of CompAcct_ACijt
SIZEt Natural logarithm of total assets
ROAt Net income divided by total assets
RETt Stock return for the year
LEVt Total liabilities divided by total assets
LNPRICEt Natural logarithm of closing price
VKOSPIt KOSPI Volatility Index
TERMt Term spread; return on a five-year government bond after subtracting the risk-free rate, where the risk-free rate is the 91-day certificate of deposit rate
CREDITt Credit spread; return on BBB credit-rated bonds subtracted from the return on AA credit-rated bonds
STDBUSt Standard deviation of monthly business sentiment
DISCLt Frequency of firms' voluntary disclosure
BODt Natural logarithm of board size
BINDt Number of independent directors divided by the number of total board members
LARGEt Ratio of shares that largest shareholders owned in the firm divided by the total number of shares outstanding

Biography

  • Eun Hye Jo and Jung Wha (Jenny) Lee ([email protected]) are at George Mason University, Korea.

  • 1 For example, Freems Corporation, where presidential candidate Lee and the president of the university alums serve, and Dongshin Construction, located in the hometown of candidate Lee, recorded high stock prices. In addition, Daewon Cable, where presidential candidate Yoon and university alums serve as outside directors, also recorded values in the upper limit.
  • 2 The ownership of individual investors in the Korean stock market is relatively high. According to the Wall Street Journal, in the US, the average ownership of institutional investors is 80%, and that of individual investors is 15%. In Korea, on the other hand, the ownership of institutional and individual investors is 13% and 50%, respectively (Wall Street Journal, 2020). This implies that individual investors dominate the Korean stock market. Regarding total trading volume during our sample period, individual investors account for about 84%, foreign investors 9.03%, and institutional investors 6.25%. Regarding trading amount, individual investors account for 48.89%, foreign investors 27.97%, and institutional investors 21.81% (see Table 3). The number of individual investors is also rapidly increasing in the Korean market. Domestic individual investors have nearly tripled within five years, from 5.56 million in 2018 to 14.24 million in 2022 (Korea Securities Depository, 2022). These statistics indicate that the power of individual investors is growing as a leading force in the Korean stock market.
  • 3 This index has six dimensions of investor protection: 1) extent of disclosure index; 2) extent of director liability index; 3) ease of shareholder suits index; 4) extent of shareholder rights index; 5) extent of ownership and control index; and 6) extent of corporate transparency index. They calculate an overall minority investor protection score from zero to 10, with the highest values indicating strong investor protection. According to these reports, Korea's score in 2020 was 74, but considering the score of other East Asian countries, such as Taiwan's 76, Singapore's 86, and Malaysia's 88, it is difficult to say that Korea has a high level of investor protection. Wang et al. (2021), for example, classify Korea as a country with weak investor protection.
  • 4 According to Ayers et al. (2011), individual investors use the seasonal random walk model to predict profits, but institutional investors base their earnings expectations on analysts’ forecasts.
  • 5 Miller (2010) finds that the more complex the financial reporting, the sharper the decline in the trading volume of individual investors. This supports the notion that individual investors have a low ability to understand and analyse financial statements.
  • 6 We are grateful to an anonymous reviewer who has provided us with this valuable perspective.
  • 7 In contrast, competition can make companies reluctant to disclose their private information (Daniel, 1983). For consistency with the rest of the hypothesis, we set Hypothesis 2-2 as an alternative form.
  • 8 We measure HHI based on the Korean Standard Industrial Classification.
  • 9 VKOSPI is a Korean-style volatility index similar to the volatility index (VIX) calculated and announced by the Chicago Board Options Exchange (CBOE). The KRX announced VKOSPI in April 2009.
  • 10 We obtained the credit spread from the Bank of Korea Economic Statistical System (https://ecos.bok.or.kr/ ).
  • 11 We collected the frequency of managers’ voluntary disclosure from the Korea Investor's Network for Disclosure System (KIND) operated by the KRX. https://kind.krx.co.kr/.
  • 12 Most sentiment studies targeting the Korean capital market use the measurement devised by Ryu et al. (2018) and Seok et al. (2019) (Kim et al., 2019; Ryu et al., 2020; Yun and Kim, 2022).
  • 13 The risk-free rate is the 91-day certificate of deposit rate. The risk-free rate is obtained from the Bank of Korea Economic Statistical System (https://ecos.bok.or.kr/).
  • 14 Previous studies show that market-level investor sentiment has different effects on stock returns in each industry (Chen et al., 2013; Sayim et al., 2013; Molchanov and Stangl, 2018), which implies that investor sentiment can be derived for each industry.
  • 15 We confirm that the results of this study are qualitatively similar when the residual derived from equation (10) is used as the investor sentiment.
  • 16 The Korean Composite Stock Price Index (KOSPI) is the South Korea's representative stock market index with a market capitalization of approximately 2,176 billion in U.S. dollars in 2020. As of 2020, South Korea's market capitalization ranks 9th worldwide (Please refer to https://www.theglobaleconomy.com/rankings/stock_market_capitalization_dollars/).
  • 17 Given that adoption of IFRS affects comparability (Lang et al., 2010; Yip and Young, 2012), our sample begins from 2011 as Korea adopts IFRS from 2011.
  • 18 As the investor sentiment index is estimated based on daily data, it is excluded from the sample if it is designated as a management item during the sample period or a discontinuity in stock trading due to trading suspension.
  • 19 The KIS-VALUE and TS-2000 are equivalent to COMPUSTAT and CRSP in the US, respectively.
  • 20 Before adjusting the investor sentiment index to a value between zero and one, the average is –0.366, whereas the standard deviation is 4.322, showing an immense value. This value is similar to that in a previous study (Baek and Kwak, 2020). It shows significant differences in the level of investor sentiment across companies.
  • 21 According to Grullon and Michaely (2007), the average HHI in the US market is 0.0872. Compared to the US market, the Korean market is relatively less competitive.
  • 22 Greater analyst coverage is associated with an improvement in the dissemination of information, reducing information asymmetry among investors (Merton, 1987).
  • 23 When we use the interaction SENT*COMP*HIGH HHI, we find a significantly negative coefficient of this interaction term.
  • 24 Francis et al. (2014) show that auditor style affects accounting comparability through accruals reporting. Decomposing accounting comparability into accrual comparability and cash flow comparability, Kim et al. (2014) find an inverse relationship between comparability and financial analysts’ forecast errors arising from accrual comparability rather than cash flow comparability. Park (2013) provides empirical evidence which shows that analysts’ forecast accuracy is higher for firms with higher accrual comparability.
  • 25 The CCSI is a standardized and synthesized index of six major individual indices: current living conditions, living conditions prospects, household income prospects, consumption spending forecasts, current economic judgment, and future economic prospects. The Bank of Korea calculates this index through a direct questionnaire.
  • 26 According to the Bank of Korea (https://ecos.bok.or.kr), foreign institutional investors’ trading volume accounted for 97% of foreign investors’ trading volume over the period 2011–2019.

The full text of this article hosted at iucr.org is unavailable due to technical difficulties.