Foreign Investors and the Peer Effects to Payout Policies

2.1 Peer Effect and Payout Policies

Many studies show the peer effect of corporate financial policies (Brav et al., 2005; Lieberman & Asaba, 2006). There is a learning channel where smaller and newer firms mimic and learn from veteran firms that are better rated and more renowned. Firms mimic other rivals in dividend policy because they learn strategies from incumbent firms.

Chief executive officers send signals to show their competence and the financial stability of the firm (Deshmukh et al., 2013). Managers are also incentivized to compete with other rivals to force them to suffer from financial restraints (Scharfstein & Stein, 1990). Payouts may depend on competitiveness (Hoberg et al., 2014; MacKay & Phillips, 2005), where firms compete in the product and financial markets to attract investors. The peer effect on signaling occurs when firms without the intention to change corporate policies decide to do so to stand out for customers and investors. Large shareholders or institutional investors also monitor firm efficiency and may intervene in dividend payout policies (Shleifer & Vishny, 1986). Grinstein and Michaely (2005) hypothesized that institutional investors who possess superior informational backgrounds and experience exert their authority to achieve the optimal payout policies.

Firms that share a product market indirectly affect each other’s characteristics, such as profit margins and stock returns. Additionally, studies show that financial decisions are made not only by internal factors but also by competitors’ fiscal decisions that share the market. Firms especially mimic others with desirable credit ratings (Grennan, 2019).

There are two asymmetric sides of dividend payout decisions: payout increase and payout decrease. Grennan (2019) shows that the peer effect exists for dividend increases but not decreases. As the decisions to increase or decrease dividends are driven by different reasons and induce asymmetric consequences, peer influence plays an asymmetric role in such decisions. A dividend decrease is a strong negative signal that shows a firm’s financial trouble (Dhillon & Johnson, 1994; Nissim & Ziv, 2001). In contrast, a dividend increase positively affects stock returns, albeit smaller than the negative effect of the decrease. Due to the asymmetry of response for dividend increases and decreases, the decision-making procedure also differs. Our study considers both dividend increases and decreases and finds that peer effect from dividends exists for both of them.

2.2 Ownership and Monitoring in Korea

The ownership structure is known to influence corporate governance. While Choe et al. (2005) point out that foreign investors have informational disadvantages over domestic investors, other researches show that foreign investors are more experienced and sophisticated when coming into developing financial markets. Jeon et al. (2011) show that foreign investors effectively monitor firms using a broad perspective and play a role in determining payout policies. As shareholders exercise their voting rights and actively monitor firm management, agency cost, and information asymmetry are reduced. Many studies find that institutional investors use their vast international experience to monitor more efficiently and objectively and thus improve corporate governance. The channel through which ownership structure enhances governance is dividend payouts. Dividend payout deters earnings management and enhances the accounting data quality (Dechow & Ge, 2006; Deng et al., 2017; Dichev & Tang, 2009). Consequently, institutional investors prefer firms with dividend payouts and also exert their right to demand dividends. Short et al. (2002) find a positive association between institutional ownership and dividends using a U.K. data set.

More specially, foreign institutional investors play a significant role in monitoring with objectivity. This is important for corporations in Korea because the ownership structure is heavily concentrated (more so for Chaebol firms). Many studies find that family-controlled Chaebol firms show poor performance and lack corporate governance (Campbell & Keys, 2002). The entry of foreign investors alleviates this. Kang and Kim (2013) find that R2, which takes on lower values for lower informational asymmetry and better corporate governance, is negatively associated with payout policies. Furthermore, this relation is more substantial for non-Chaebol firms. Previous findings suggest that the influence of dividends in improving corporate governance depends on the ownership structure. In Korea, foreign institutional investors are expected to be objective and unbounded by local irregularities (e.g., ownership structures such as Chaebol). While many studies examine the relation between Chaebol and corporate governance (Kim & Lee, 2003; Kim et al., 2007), we focus on foreign investors and their role in increasing governance through payouts.

It has been shown that foreign investors prefer firms with specific characteristics beyond conventional profitability, tangibility, and solvency. Kim et al. (2010) show using Korean firms that foreign equity participation is positively related to firm’s efforts for better governance. Jeon et al. (2011) show that foreign investors prefer firms with dividend payouts and that foreign owners play a role in influencing payout decisions once they have acquired voting rights. Kang et al. (2010) show that foreign investors impact corporate dividend policy while local institutional investors “play inadequate roles as stakeholders.” Studies suggest that foreign investor and payout policies are linked through theories such as dividend clientele, adverse selection, and agency problems.

Overall, our main hypothesis is as follows:

H1: The degree of peer influence from other firms’ dividend payouts will increase as monitoring by foreign investors increases.

3.1 Data and Key Variables

This study uses two data sources: KOCOInfo’s TS2000 and FnGuide’s DataGuide. TS2000 provides annual financial statements from the Data Analysis, Retrieval, and Transfer System, monitored by the Financial Supervisory Service. Our sample period is from 2000 to 2018. This sample period was selected to disregard any influence of the dot-com boom. Firm characteristics, payout decisions, and the three-digit industry code of all public firms in Korea are acquired from firm financial statements. Quarterly financial information is limited, and since most Korean firms payout dividends annually, this study conducts yearly analysis.

DataGuide provides information regarding daily stock transactions, which we use to calculate idiosyncratic risk. Using firm idiosyncratic risk, we construct industry risk and our key IV, peer idiosyncratic risk. Explanatory variables used in our study are profitability, life cycle, book-to-market ratio, leverage, tangibility, investment intensity growth rate, cash flow to capital, and size (Grennan, 2019). Variable definitions are provided in the Appendix. Two key variables of interest are foreign ownership share and peer influence. The summary statistics are reported in Table 1. Table 1 shows the differences in firm characteristics for dividend payers and non-payers. Notably, foreign ownership appears to be much higher for dividend payers.

When studying peer effects, difficulties arise when trying to separate market patterns from heterogeneous firm factors. Firms that share the same industry could have similar dividend policy changes due to industry characteristics, not because of peer effects. Additionally, firms may have similar characteristics that attract specific types of investors who demand similar payout policies. To ensure peer effects are correctly measured despite the reasons above, we implement an instrumental variable (IV) approach. Many studies in this field implement this method (Campbell et al., 2001; Grennan, 2019; Hoberg et al., 2014; Leary & Roberts, 2014). We calculate the idiosyncratic risk of each firm and use the idiosyncratic risk of peer firms as an IV. When one firm’s risk increases, it increases its cash holding to withstand any adverse shocks. As a consequence, this procedure affects the dividend payout policy. However, the idiosyncratic risk is unique to its firm by construction and has no direct impact on other firms’ payout policies. Thus, peer firms’ idiosyncratic risks are suitable as an instrumental variable.

This study explores the factors that affect peer influence on dividend payout policies. We hypothesize that foreign ownership alters the level of influence from peers on dividend decisions. One of the key variables in our study observes the changes in dividend policies. DivInc _ijt and DivDec _ijt represent dividend payout changes that firm i from industry j made in year t. DivInc _ijt (DivDec _ijt ) is an indicator variable that shows whether a firm increased (decreased) its dividend payouts. The dividend change over the previous year must be greater than 1% to take on a value of 1. Although it would be nice also to have visibility on repurchasing behavior, the process through which stock repurchases are made is different from that of U.S. firms. Thus, we focus on dividend payouts.

Peer influence is measured by Peer_(−i)jt−1. This variable is the percentage of peer firms in industry j that increased or decreased payouts in year t – 1, excluding firm i. Peer firms are competing firms that share the same three-digit industry code. This variable identifies whether rival firms’ payout changes affect a firm’s dividend decisions. To investigate further, we include the change in foreign ownership share. ΔForeignshare _it represents the net purchase of outstanding stocks by foreign investors. Foreigners are essential for this study because they prefer firms with dividend policies and also exert influence in payout decision-making (Jeon et al., 2011). Their involvement indicates active monitoring of the firm and also the industry.

Risk is an important variable that determines payouts. In this study, we find peer idiosyncratic risk using the method proposed by Grennan (2019). We first compute the value-weighted industry excess return using the 90-day Certificate of Deposit (CD) rate, R ijt . Value is weighted using market capitalization. Then, the daily firm-specific residual return ( ϵ ijdt ) is calculated as below:

where R ijdt represents the return of firm i in industry j on day d of year t. Next, the daily firm-specific residual returns are used to construct the annual idiosyncratic risk of each stock.

where IR _ijt represents the idiosyncratic risk of firm i in industry j at time t. Using firm risk constructed using equation (2), two variables are calculated, industry risk and peer idiosyncratic risk. Industry risk is calculated as follows:

IR _jt represents a risk that industry j bears in year t. ω ijt is the share of the market capitalization of firm i in industry j. In addition, peer idiosyncratic risk (PeerIR) is the average of the idiosyncratic risks of peer firms. We exclude a firm’s own risk when calculating peer risk to eliminate possible correlations. Industry risk is used as an explanatory variable in our regressions, while peer idiosyncratic risk is used as an IV.

3.2 Regression Models

Using the variables defined previously, we test our hypothesis that foreign ownership influences the sensitivity of peer effects. First, we set up empirical specifications to show that peer effects exist among Korean firms as predicted by various prior studies. Following the identification strategy of Grennan (2019), we use peer idiosyncratic risk as our instrument for peer influence. Then, we further study the changes in peer influence by foreign ownership levels. The empirical specification is:

Div ijt is a dummy variable that indicates whether firm i increased or decreased dividend payouts by more than 1% from the previous year. Pee r ( − i ) j ( t − 1 ) represents the fraction of firms that increased or decreased dividends in industry j where firm i operates. X ijt is a vector of control variables that determine the likelihood of changing payout policies. f i is the firm-fixed effect, δ t is the time-fixed effect, and ε ijt is the unobservable error term. Peer influence may embed issues regarding endogeneity. Therefore, the IV approach is used. The IV is the average of peer firms’ idiosyncratic risks, which is correlated with the payout decisions of peer firms but not with firm i’s dividend payouts. We evaluate how β changes as foreign ownership changes.

In addition to the earlier specification, we explore the heterogeneity of the peer effect using foreign ownership and interaction terms as shown below:

where Δ ForeignShar e it represents the change in foreign ownership of firm i in year t. This study considers the asymmetric behavior of payout increases and decreases noted in prior studies and examines the peer effects in dividend increases and decreases separately. We hypothesize that firms with higher foreign ownership will respond more aggressively to dividend changes of peer firms indicated by a positive coefficient for β 2 . In addition, we find the industry averages of profitability, life cycle, market-to-book ratio, tangibility, investment-to-capital ratio, and cash flow-to-capital, and size. These industry averages are included as industry control variables. These controls are analogous to the peer firm averages used in the study of Grennan (2019). Industry averages control for industry characteristics that peer firms have, which attract domestic and foreign investors.

4.1 Univariate Tests

Our univariate analysis separates our sample into four subgroups by peer influence. The likelihood of a dividend increase is found for each group to represent the level of peer influence. The difference between groups with the highest and lowest levels of peer influence is approximately 10%. In contrast, the difference in dividend decreases is only 5%. Peer firms asymmetrically influence the decision to increase or decrease the dividend, as shown in Grennan (2019). We divide our sample into quartiles by the change in foreign share and form four-by-four subgroups. The 16 subgroups show that the difference in the odds of a dividend increase between the groups with the highest and lowest levels of peer influence is greater in subsamples with higher foreign transactions. In groups with high foreign net purchases (increases in foreign share), the difference was 13%. The difference was only 7% in groups with the lowest change in foreign share. Foreign transactions had no impact on the peer effects of dividend decreases. Empirical evidence suggests that the influence of foreign investors on peer effects is more pronounced for dividend increases than for dividend decreases. Results were consistent using IV regression models. Changes in foreign ownership only impact the influence of peers for dividend increases, not decreases. Firms with high foreign ownership changes are shown to be more sensitive to peer effects.

For the univariate analysis, Table 2 presents the odds of dividend increases and decreases across quartiles of peer influence. The values show that a firm’s dividend choices are related to its peers’ dividend decisions. Peer influence is the fraction of rival firms that increased or decreased dividends in the industry to which firm i belongs. Column 1 (3) shows the likelihood of a firm increasing (decreasing) its dividend for each quartile. The probabilities increase as peer influence gets stronger. While Columns 1 and 2 show how the likelihood of increasing dividends changes as peer influence changes, Columns 3 and 4 show the changes in the likelihood of dividends decreasing. The difference between the highest and lowest peer influence groups for a dividend increase and decrease is quite distinct in size. The difference for dividend increase is 0.1016, and for decrease, it is 0.0505. This is consistent with previous findings that dividend increase is affected by peer effects. Dividend decreases induce asymmetric impact and, thus, are decided to know the severe consequences. Due to these reasons, dividend decreases are not driven by peer influences as much as dividend increases.

In addition, the sample is divided into quartiles based on changes in foreign ownership share. Table 3 reports the likelihood of regular dividend payers increasing or decreasing payouts by each quartile of changes in foreign ownership share and strength of peer influence. The groups are divided into a 4 × 4 matrix according to changes in foreign ownership and peer influence. Across foreign ownership changes, the likelihood of changing payout policies is significantly higher when the peer influences are higher. In other words, the difference in the likelihood of increasing dividends between groups with high and low peer pressure increases as foreign ownership increases. The lowest quartile shows a difference of 0.0772, while that of the highest quartile is 0.1305. This gap increases for dividend-increasing decisions (Panel A) but not for dividend decreases (Panel B). The differences between high and low peer influences are all statistically significant. While we observe apparent increases in the likelihood of the announcement of a dividend increase from low to high peer influence as foreigner activity increases, the difference for dividend decreases remains stable from 0.0507 to 0.0478. The univariate results show that peers’ payout policies may influence other firms’ dividend policies. However, whether foreign ownership affects dividend decreases through peer influence is questionable, while dividend-increasing decisions are more clearly affected.

4.2 Empirical Results

To better understand the relationship between foreign ownership and peer effects, we present empirical evidence using IV. In this regression, firm characteristics are controlled. The IV specification includes firm-fixed effect, industry-fixed effect, firm risk, and time-fixed effects. Additionally, the firm-specific covariates described in Table 1 are included. Firms are ranked by change in foreign ownership (net purchase). The upper half of firms are classified as high and the lower half as low. The foreign share dummy is an indicator variable that takes a value of 1 if the foreign ownership change from the previous year is greater than the median and 0 otherwise.

4.3 Interactive Results

In addition to observing the differences in significance and magnitude for high and low changes in foreign ownership, Table 6 reports the results with the interaction terms in the empirical specification. The results are consistent in that foreign ownership change only affects dividend increases, not decreases. Panel A reports the findings using the indicator variable for change in foreign share. Panel B employs the value of change on foreign shares. Column 1 shows coefficients from the IV regression on dividend increases, while Column 2 covers decreases. Column 1 suggests that peer effects of payout changes only exist in the presence of foreigners. Column 2 shows evidence of peer influence on dividend decreases. This effect is independent of foreign ownership. Peer effects exist for dividend decreases but not through foreign ownership. The evidence suggests that foreign investors are only particularly concerned with peers increasing dividends, not decreasing.

4.4 Robustness Check with Net Purchase

In Table 4, we separated our sample into two groups by ranking the change in foreign share. Firms with a change in foreign ownership share greater than the median were put into the high group, and firms with a change in foreign ownership lower than the median were put into the low group. For a robustness check, we divide the sample according to whether a firm’s foreign ownership increased. If it increased from the previous year, that firm was put into the “Net Buy” group. Firms in this group had positive net purchases by foreign investors. If foreigners exited or the ownership share did not change, those firms were put into “Others.” Net purchases for firms put into Others are less than or equal to zero. The results are presented in Table 7 for these two groups separately. Peer influence remains influential for both groups. Consistent with previous results, peer influence is stronger for firms with increased foreign ownership. Peer influence for Others is 0.514, while that for the Net Buy group is 0.621. This difference provides supporting evidence that increases in foreign ownership share induce peer effects to be more pronounced in increasing dividends.

4.5 Potential Channels of the Effect of Foreign Investors on the Dividend Peer Effect

We suspect there could be two channels through which foreign investors and dividend peer effect are related. One is the signaling effect, and the other is the reputation effect. To study the channels, we look at credit ratings and the age of the Chief Executive Officer (CEO) to account for the signaling and reputation effects, respectively. Managers use dividends to signal that the firm is in a stable condition. Signaling is important for firms with high uncertainty or information asymmetry. To account for information asymmetry, we look at firms with and without credit ratings separately. Columns 1 and 2 of Table 8 show how foreign investor participation influence firms without credit ratings more. Evidence suggests that foreign investors play an important role in demanding payouts in industries, with other firms increasing dividends for firms with high information asymmetry. We show that foreign investor activities on peer effect are greater when there is higher information asymmetry.

Ages of CEOs proxy for the reputation effect because younger CEOs have more incentives to build reputations for their own future career paths. While it may be reasonable to assume that younger CEOs mimic other firms’ dividend policies, Columns 3 and 4 of Table 8 show otherwise. The results show that peers do not influence the dividend policies of firms with younger CEOs, nor do the degree change due to foreign investor activities. Table 8 suggests that signaling is a more significant channel when addressing foreign investors and the peer effect on dividends.