The influence of financial constraint on peer influence of cash dividend decision – Evidence from China


 We analyse the influence of financial constraints on the peer effect of dividend decision in China by employing the Carhart four-factor model to construct instrument variables of peer influence. We find that (1) the decision of whether to pay cash dividends (DIV) is significantly influenced by peers, and the estimated marginal effect is 0.53%, but the question of whether to pay catering dividends and the extent of such dividends to be paid are not significantly affected by peers. (2) Under the semi-mandatory dividend policy in China, financial constraints will significantly reduce peer influence on the dividend level. (3) Peer influence on DIV is more pronounced among companies that face high financial constraints.


Introduction
Cash dividend decision is one of the major financial management decisions of listed companies, but it is also influenced by China's regulatory policy. In order to protect the interests of investors and raise the dividend level of listed companies, the Securities Regulatory Commission has issued the policy that clearly states that the cumulative cash dividends (DIV) of the past 3 years should not be <20% of the average annual profits realised in the last 3 years if the companies want to issue new securities. This requirement is raised to 30% in 2008. Scholars name the policy as 'semi-mandatory dividend policy" (Li et al., 2010) [1], which has a certain impact on the dividend behaviour of Chinese companies. On the one hand, it makes the companies who need to issue new securities to refinance, which will increase their willingness to pay dividends and readily comply with the policy. However, companies that do not have such needs will lack incentives (Wang and Zhang, 2012. Wei et al., 2014) [2,3] and thus their willingness to pay dividends may even become weakened (Gao et al., 2018) [4]. According to the dividend data of listed companies in China from 2008 to 2019, the proportion of listed companies distributing DIV rose from 52.64% to 69.7%, but the average cash dividend payment rate raised only from 22.69% to 26.49%, and cash dividend payment rate of listed companies which paid DIV decreased from 43.10% to 38%. The studies comprising the existing literature do not answer the question of why the dividend willingness is improved but the dividend level has not significantly improved in China.
Some scholars have found that the company's dividend decision is influenced by the peer companies, resulting in the convergence of dividend behaviour. Adhikari and Agrawal (2018) [5] take the American-listed company as the sample and find that the company dividend policy is affected by peer companies in the same industry. The main reason is the competition imitation mechanism. Grennan (2019) [6] also empirically analyses the peer influence of dividend decision change. Ding and Li (2020), Li et al. (2020) and Wang et al. (2021) [7][8][9] studied the regional and industrial peer influence on the dividend payouts of listed companies in China, but none of the above studies take into account the reality of financial constraints faced by most enterprises in China and the influence of the semi-compulsory dividend policy.
Therefore, considering the semi-mandatory dividend policy, this paper researches the influence of financial constraints on peer influence on cash dividend decision, and the conduction path of peer influence between enterprises with different financial constraints. The main contribution of this paper is (1) it enriches the research on the peer influence of dividend decision under the semi-compulsory dividend policy in China, (2) it researches the influence of financial constraints on dividend decision peer influence; and (3) it studies the conduction path of peer influence on dividend decision between enterprises with varying financial constraints.

Peer influence and dividend decisions
Peer influence arises from the imitation behaviour between enterprises. Lieberman and Asaba (2006) [10] put forward two theories on why enterprises imitate each other. One is the information-based theory, where firms follow other companies that are perceived as having superior information. The other one is rivalry-based theories, where firms imitate others to maintain competitive parity or limit rivalry. The information-based theory is related to the uncertainty of the external environment. If the company manager cannot accurately predict the results of a decision or an action, the company will very likely accept the information contained in the actions of other companies and imitate the decision-making actions of other peer companies. In fact, enterprises need to refer to a large number of internal and external information when making decisions. Moreover, because of the information asymmetry in the market, the cost of information acquisition is high, and even some external information is difficult to obtain. Companies in the same industry will imitate each other for the reasons that the macro-environment, industry policy and supply chain environment are very similar, and the basic elements and external information needed for decision-making are the same. Therefore, imitation and learning from peers may be a low-cost and efficient information channel, which can reduce the cost of information acquisition and the risk of decision-making.
Another kind of imitative motivation based on information is herd behaviour. As any behaviour of the enterprise conveys certain information to the outside world, this information may be used by stakeholders, such as investors, as useful information to evaluate the enterprise. Business managers may ignore private information in order to improve or maintain a good brand reputation and imitate others' behaviours to avoid negative evaluation. The dividend signal theory shows that the company pays dividends in order to transmit information to shareholders. The dividend principal-agent theory holds that dividend payment is beneficial to reduce the agent cost and reflects a good level of corporate governance. Therefore, companies may imitate peers in order to obtain a positive evaluation.
Another reason why peer individuals imitate each other is that competition produces imitation behaviour (Adhikari and Agrawal, 2018) [5]. When companies with comparable resource endowments and market posi-tions compete with each other, companies can adopt differentiation strategies to distinguish them from competitors, and this reduces the possibility of imitation and may generate higher profits. However, any differentiation strategy often needs higher investment and is associated with higher risk, which could even lead to business failure. In contrast, homogenisation strategy can also limit the exertion of competitors' competitive advantage by imitating them, so imitation can also reduce competition. There are generally 'leaders' and 'followers' with different competitive advantages in an industry. 'Followers" often imitate the behaviours of 'leaders' in order to narrow the gap and reduce the intensity of competition. Dividend is an important factor of company information and company valuation. Regulators, third-party supervision and auditors may compare the target companies with the industry average level or the level of peer companies, so the target companies have competitive imitation motivation. Therefore, the company is likely to imitate the dividend decision behaviour of the peer companies. Therefore, we build up the hypothesis H1: H1: Companies' dividend decisions are positively influenced by peer companies.

Financial constraints and dividend decisions
Financial constraints refer to the difficulty in obtaining financing when there is a need for capital to invest, which often arises due to the frictions in the supply of capital. The chief source of friction is information asymmetries between investors and the firm (Tirole, 2006) [11], and because of that, the investment cannot reach the optimal level he condition (Fazzari et al., 1988) [12]. The difference of internal and external capital cost leads to the preference of internal capital, so companies prefer endogenous capital and adopt residual dividend policy. A lot of research suggests that the smaller the external financial constraints on companies, the higher the dividend level. China's capital market is still in rapid development and most companies are facing financial constraints, and so they do not pay dividend (Zhang and Wang, 2015) [13]. To protect the interests of investors, the Chinese government introduced a semi-mandatory dividend policy, which requires the companies that want to issue new securities to pay a certain percent of profits in the form of DIV. Hence, companies with refinancial needs are deliberately deployed to meet regulatory conditions (Wei et al, 2017) [14], leading to a policy pandering motive. Therefore, we build up the hypotheses H2 and H3: H2: The greater the financial constraints, the more likely the company is to pay dividends. H3: The greater the financial constraints, the lower the cash dividend level of the company.

Peer influence, financial constraints and dividend decisions
When the enterprise's own profit cannot meet the investment demand, there is an external financial demand. At this time, it is difficult to obtain external financial support, unless the obligation is accepted to pay higher financial costs and put up with additional financial constraints. When enterprises face higher financial constraints, the payment of DIV will reduce the inner funds of enterprises; however, the external financial demand will be greater, which will lead to the reduction of financial flexibility and the loss of company value. When enterprises face higher financial constraints, they will put their own factors in a more important position, and at the same time, they will not be able to imitate their peers even if they want, which leads to weakening of the impact of peer companies. Therefore, we build up the hypothesis H4: H4: The greater the financial constraints, the less the company's dividend decision is affected by the peer companies.

Data
We use the data of China's stock-listed companies as the primary data from the China CSMAR database. The full sample runs from 2010 to 2019. Following previous studies (e.g. Leary and Roberts, 2014) [15], we exclude financial and ST companies, companies that have only one peer in 1 year or listed for <3 years. The final screening of 2589 listed companies results in a total of 18,337 samples. We winsorise all variables at the 0.5 and 99.5 percentile levels to minimise the influence of outliers.

Dependent variables
In analysing the cash dividend decision of company i, we use three measures as decision of cash dividend. The first one is whether to pay a cash dividend (Div_dum). The second one is whether to pay the catering dividend (Div_cat). Following previous studies of Chen et al. (2015) [16], the ratio of total accumulated DIV in the last 3 years to the arithmetic average of distributable profits in the last 3 years is between 30% and 40%, i.e., Div_cat equals 1, otherwise equals 0. The third one is Div_lev, which is measured by the ratio of cash dividend per share to earnings per share.

Explanatory variables
The explanatory variable is peer influence dividend decisions. Peer firms are defined as all firms in the same first-level industry classification of China Securities Regulatory Commission (CSRC) as peer companies, except firm i, in a given year corresponding to dependent variables, measured by Peer_dum, Peer_cat and Peer_lev, respectively. Peer_dum refers to the proportion of companies that pay cash dividends among peers, Peer_cat refers to the proportion of companies that pay catering dividends among peers and Peer_lev refers to the average cash dividend level of the peers.
Because the interaction between companies in the same industry will produce endogenous problems, there are also identification challenges in empirical tests of peer influence. Similar to those used by Adhikari and Agrawal (2018) [5], we use two instrumental variables, which are peer average idiosyncratic equity shocks (Peer_ir) and peer idiosyncratic volatilities (Peer_risk). The calculation processes are as follows: First, we use the Carhart four-factor model to calculate the monthly idiosyncratic stock return of the company i. The models are as follows: R ijt refers to the stock return of company i in industry j in month t. The MKTt is the excess monthly market return, SMBt is the size factor, HMLt is the book-to-market factor and MOMt is the momentum factor. R ijt is the calculated expected return of company i in industry j in month t. The difference between the R ijt and R ijt is the monthly Idiosyncratic Return. The descriptive statistics of Carhart four-factor model coefficient, real return, expected return and monthly Idiosyncratic Return are shown in Table 1.
Then, we sum 12 months Idiosyncratic Return of company i in year T as the annual Idiosyncratic Return. The average annual Idiosyncratic Return of peers in industry j in year T is the Peer_ir ijT . The standard deviation of the Idiosyncratic Return of peers in industry j in year T is Peer_risk ijT .

Financial constraints
Financial constraint is measured by the KZ index, which is the most popular measure of financial constraints (Farre-Mensa, 2016) [17]. KZ index is also widely used as a measure of financial constraints in China (Li and Huang, 2020) [18]. In order to fit the actual condition of China, we combine currency funds and transactional financial assets as cash holdings (CASH); net operating cash flows (OCF) and DIV are standardised according to total assets at the end of last year. The other two factors are the assets-liabilities ratio (LEV) and Tobin's Q. By using ordinal Logit regression, the regression coefficients were significant at 1% and pseudo R 2 is 0.253. The KZ index regression model is:

Control variables
Following the existing literature, the control variables include enterprise size (Size), cash ratio (Cash), total return on assets (ROA), growth rate (Growth), Tobin's Q, asset-liability ratio (Lev), established life (Life), equity nature (Soe) and equity concentration (Shr1). The definition of variables is shown in Table 2.

Models
We use model (4) to test whether the company i's dividend decision is influenced by peers in China. To test the impact of financial constraints on company i's dividend decision and the peer influence of dividend decision, we add KZ index and interaction item of peer influence and KZ index to model (4) to form model (5). If the interaction item's coefficient is significantly negative, it means financial constraints will weaken the peer influence of dividend decision.
4 Empirical analysis

Peer influence on dividend decision
In order to test whether the peers influence listed companies' dividend decision in China. The regression results show that controlling other related variables, the regression coefficient of Peer_dum is 2.191 and is significantly positive at the level of 5%, indicating that there exists peer influence. The result means that the higher the proportion of peer companies that pay DIV, the greater the possibility is for the company that will pay DIV. The estimated marginal effect of Peer_dum is 0.0053 when other variables are kept at their means, which suggests that compared to a company with no-dividend-paying peers, a company with all-dividend-paying peers is 0.53% more likely to pay dividends. The coefficients of Peer_cat and Peer_lev are positive but not significant, indicating that there is no significant peer influence on the level of DIV of listed companies in China. This conclusion is different from that of Adhikari and Agrawal (2018) and Grennan (2019), who take the western  developed markets as the research background. The reason for this result in China may be mainly explained by the financial constraints. The capital market of China is an emerging market, and most enterprises face higher financial constraints, which limits their ability to imitate peers to issue higher dividend levels. Besides, the semi-mandatory dividend policy of China makes the enterprises adopt a strong policy catering motive and pay less attention to the dividend decision behaviour of the peers, which leads to a situation wherein the influence on a company from its peers is not obvious.

Financial constraints, peer influence and dividend decision
To test the impact of financial constraints on the peer influence on dividend decision of listed companies in China, we use model (5). The results are shown in Table 5. Table 5 show that the peer influence is still only significantly positive as respects the decision of whether or not to pay dividends. The coefficients of financial constraints are 0.120, 0.182 and -0.008; all are significant at the level of 1%, which shows that financial constraint has a positive effect on the decision of whether or not listed companies pay cash and whether to pay dividends.

Columns (1), (3) and (5) in
Because peer influence is an endogenous variable, interaction item of peer influence and KZ index is also an endogenous variable. We construct the interaction variables of Peer_ir with KZ index and Peer_risk with KZ index as the instrumental variables of the interaction item. Columns (2), (4) and (6) in Table 5 add interaction items of peer influence and KZ index, which mainly focus on the relations of financial constraint and the peer influence of dividend decision. The coefficient of interaction item in Column (2) is -0.002, but not significant. It means that the higher the proportion of peers who pay DIV, the more likely a company is to pay DIV, no matter what level of financial constraints it is facing. The coefficient of interaction item in Column (4) is - 1.1977 and is significant at 1%, which indicates that the higher the proportion of peer companies that pay catering dividends, the less likely the enterprise is to pay catering dividends if the financial constraints are higher. This is because the higher the proportion of companies that pay dividends among their peers, the greater the competition is in the capital market and the higher the refinancial costs. So the company that needs to refinance will tend to raise funds from banks; as a result, the peer influence is weakened. The coefficient of interaction item in Column (4) is -0.029, which shows that the financial constraints in China not only limit the ability to improve the level of DIV but also limit the ability to imitate peer companies to pay higher DIV. It is an important reason why peer influence is not significant.

Robustness check
In this paper, the robustness test is carried out in the following ways and the conclusions are robust.

Sample selection bias
In order to reduce the impact of bonus share on cash dividend issuance decisions, we exclude the sample that gives bonus share, and the regression results show that the decision of whether to pay DIV is positively affected by peers. However, there is still no significant peer influence on whether to pay catering dividends and dividend level, as shown in Columns (1), (2) and (3) of Table 6.

Add control variables
Firms may use similar dividend policies simultaneously in response to common industry shocks. To solve the endogenous problem of possible missing variables, we add industry risk (Ind_risk) as a control variable. We use the standard deviation of the difference between monthly industry return and monthly market return as industry risk. The regression results are shown in Columns (4), (5) and (6) of Table 6.

Alternative financial constraint measure
SA index is also widely used by scholars to measure financial constraints. We replace the KZ index in model (5) with the SA index. The regression results show that the peer influence is only significantly positive as regards the decision of whether to pay DIV, and is not significant as regards the decision of whether to pay catering dividends and the extent of payment that can be made; and the coefficient of interaction item is consistent with the previous part (Table 7).

Further research: Peer influence conduction path from the perspective of financial constraints
The studies by Adhikari and Agrawal (2018) [5] have found that cash dividend decisions of small-scale companies are influenced by both small-scale peers and large-scale peers, while large-scale companies are only affected by large-scale peers, which means peers influence is asymmetric. From the perspective of financial constraints, we study the transmission path between different financial constraint companies.
We define a company whose KZ index is higher than the average KZ index in industry j in year T as a high financial constraint company. Otherwise, it is a low financial constraint company. There are 8186 samples in the low financial constraint group, as against 10,151 samples in the high financial constraint group, reflecting that most companies in China are facing high financial constraints.
The peer influence of low financial constraint and that of high financial constraint are calculated according to different financial constraint levels. Model (5) is used to analyse whether the company is influenced by peers with the same financial constraints level or with different financial constraints levels. Then the path of peer influence conduction is analysed.
The results in Table 8 show that, for low financial constrained enterprises, the coefficient of low financial constrained peer influence is 1.824 and that of high financial constrained peer influence is 2.142, but they are not significant. For high financial constrained companies, the coefficient of low financial constrained peer influence is 2.478 and that of high financial peer influence is 2.868; all are significant at the level of 10%, indicating that enterprises facing higher financial constraints are more likely to imitate peers, regardless of the financial constraints of peer companies. In other words, there are two paths to conduct the peer influence of the decision on whether to pay cash dividends: one is that the low financial constraint enterprise transmits to the high financial constraint enterprise and the other is that the high financial constraint enterprise transmits to the high financial constraint enterprise.

Conclusions
Based on the sample of stock-listed companies from 2010 to China, this paper analyses the peer influence on cash dividend policy and the influence of financial constraints on dividend decision peer influence. It is found that (1) the decision of whether to pay DIV is significantly influenced by peers, but the decision of whether to pay catering dividends and the extent of dividend that can be paid are not significantly affected by peers.
(2) Under the semi-mandatory dividend policy in China, financial constraints will significantly increase the willingness of companies to pay DIV, but will significantly reduce the dividend level and make the peer influence on the dividend level not significant. (3) Peer influence on DIV decision is more pronounced among companies that face high financial constraints, meaning that the peer influence is mainly conducted from low financial constrained companies and high financial constrained companies to high financial constrained companies. T h i s p a g e i s i n t e n t i o n a l l y l e f t b l a n k