Related Party Transactions and Corporate Social Responsibility: Evidence from Korea

Abstract

This study investigates whether related party transactions (RPTs) are associated with corporate social responsibility (CSR). Analyzing Korean public companies, we find that firms with more RPTs show poor CSR performance, especially in terms of soundness and fairness. The results are only maintained in firms with low ownership of controlling shareholders, supporting the conflict-of-interest view related to RPTs. Furthermore, we find that RPTs of firms held low by controlling shareholders do not contribute to firm value. Our results indicate that the expropriation of profits or resources through RPTs negatively affects the CSR decision of affiliations, which acts as a deterrence to long-term growth. In emerging markets where RPTs within a business group are prevalent, a government needs to restrict RPTs which are beneficial only for controlling shareholders and to encourage CSR activities of affiliations with low controlling ownership.

1. Introduction

This study investigates the association between related party transactions (RPTs) and corporate social responsibility (CSR). Specifically, we investigate whether firms with more RPTs consider CSR less important. RPTs are common in business groups in Asian countries such as China and Korea. It can enhance firm value through efficient contract among related parties [1], but prior studies mostly support the negative role of RPT as a tool of tunneling or earnings management [2,3]. Hendratama and Barokah [4] show that RPTs of firms with better CSR have a positive association with firm value despite the general negative relationship between RPTs and firm value [4]. The finding shows the moderating role of CSR, but it does not explain how RPTs affect it. CSR is a deliberate strategy for competitive advantage and an indispensable part of a firm’s sustainable growth; thus, it is important to identify factors that weaken CSR activities. Negative RPTs are likely to prevail when companies have weak corporate governance, which leads to low CSR activities in terms of soundness and fairness. In this study, we explore this possibility.

With the Paris Climate Agreement and the adoption of UN Sustainable Development Goals (UN-SDGs) in 2015, the importance of CSR has been more emphasized, and its definition has evolved from profit generation to shared value creation responding to social expectations [5]. CSR activities make companies more competitive by improving their image and helping them increase their profits [6,7]. CSR initiatives play a role in organizational performance even in small and medium-sized enterprises [8]. Given the CSR’s contribution to competitive advantage and stakeholders’ high interests in CSR, firms need to invest in CSR activities for their sustainable growth [5]. However, CSR investments do not always have priorities because organizations should comply with the needs of actors which provide critical resources for their continuing survival [9].

Controlling shareholders affect affiliations’ decision-making through formal (for example, equity) or informal (for example, family) ties [10]. Tunneling is a typical behavior in which controlling shareholders expropriate the wealth of minority shareholders using RPTs, which have been a concern for standard-setters and regulators [11]. RPTs are transactions between legally independent firms, but “the influenced party may be favored or caused to subordinate its independent interest (FAS No. 57).” This implies that firms with more RPTs are likely to prioritize the interests of their controlling shareholders than their own interests. However, the extent to which related companies try to increase the wealth of controlling shareholders at the expense of their own profits may vary depending on the shareholding of controlling shareholders. Given that business groups tunnel their internal resources out to affiliates in which controlling shareholders hold a larger portion, the negative association between RPTs and firms’ own interests is expected in companies with smaller portion held by controlling shareholders.

This study uses Korean research settings and investigates the association between RPTs and CSR. Business groups whose affiliations are under common control are prevalent in Korea and they make active use of RPTs for wealth transfer to a controlling shareholder [12,13,14]. The lack of proper corporate governance to protect minority shareholders and low cash-flow rights of controlling shareholders exacerbates the negative effect of RPTs in Korea [11,14]. In addition, Korean CSR is at the top of emerging market economies (CSR in emerging market economies is reported in Table 2 of Boubakri et al. [15], and Korea ranks second of 21 countries) and the Korea Economic Justice Institute (KEJI) announces their top 200 CSR companies with their CSR scores every year. Korean audit reports include RPTs in the footnotes. Therefore, high-quality CSR and RPT data are available.

Using 2507 Korean public firm-years with CSR scores from 2011 to 2018, we find the negative association between RPTs and CSR on average, indicating that RPTs signal low CSR activities. Next, we use six subcategories of CSR, soundness, fairness, social contribution, environmental management, consumer protection, and employee satisfaction and find that RPTs are negatively related to soundness and fairness. Similarly to regulators’ concerns, it can be seen that companies engaging in more RPTs are less interested in social responsibility from the perspectives of soundness and fairness. The positive association between RPTs and consumer related CSR can be interpreted as the efforts to restore damaged reputation [16]. We investigate whether the negative association between RPTs and CSR is different depending on the ownership of controlling shareholders and find that the negative relationship remains only in low ownership, consistent with tunneling. In addition, we examine whether the ownership of controlling shareholders affects the association between RPTs and firm value and find an insignificant result in low ownership but a significantly positive relationship in high ownership, supporting the idea that affiliations less owned by controlling shareholders do not seem to make RPT decisions for their own benefits.

Our study makes the following contributions to the literature. First, this study enhances our understanding on the negative effect of RPTs regarding CSR. Extending prior research which has identified various CSR determinants such as company size and industry sector [17], we show RPT can curb CSR activities and suggest an additional need for regulation of RPTs in weak investor protection countries, consistent with the concerns of standard setters and regulators.

Second, by examining the effect of ownership, we present how controlling shareholders can affect affiliations’ CSR activities in emerging markets where business groups are prevalent. This suggests that government driven CSR policies need to focus on business group level CSR in emerging markets. Considering that CSR is often used to cover the agency problem caused by controlling shareholders [16], it is necessary to review whether the CSR of RPT firms with high ownership of controlling shareholders corresponds to true social responsibility.

Finally, we respond to the demand for an under-researched area by adding RPT related CSR evidence for emerging market economies where interests and investments in CSR have been rapidly increasing [15].

The remainder of the study is organized as follows. Section 2 discusses related studies and presents hypothesis development. Section 3 describes sample and research model. Section 4 provides our empirical results and Section 5 concludes this paper.

2. Prior Research and Hypothesis Development

RPTs are self-dealing between the firm and related entities such as affiliated companies and shareholders. The RPT literature has two competing perspectives: the efficient contracting view and the conflict-of-interest view. The efficient contracting view argues that RPTs reduce transaction costs caused by information asymmetry, which can contribute to firm value [1,12]. In contrast, under the conflict-of-interest view [11], RPTs are used for wealth transfer from minority shareholders to controlling shareholders (tunneling) or as a tool of earnings management for managerial private interests [13,18,19].

Prior studies generally support the conflict-of-interest view. Wong and Jian [20] use Chinese firms and find that corporate group controlling firms inflate earnings through RPT sales when they are at the risk of delisting or necessary for stock issuance. Bertrand et al. [21] find non-operating tunneling is prevalent in Indian business groups and Jiang et al. [22] also provide the evidence about severe expropriation from minority shareholders using intercorporate loans in China. In addition, Kohlbeck and Mayhew [3] show that more RPTs are related to financial misstatements in S&P 1500 firms.

Most Korean evidence is consistent with the conflict-of-interest view, which may be due to the well-known characteristic that a controlling shareholder with small cash-flow rights can control its whole business group through a pyramid or circular ownership [10]. Specifically, Korean business groups invest more in poor growth opportunities than non-business groups [23] and the acquisition of Korean business groups increases the wealth of the controlling shareholder while decreases that of minority shareholders [18]. Kim and Yi [13] provide evidence that group-affiliated firms more engage in earnings management and the phenomenon is primarily observed in public firms. Recently, Cho and Lim [14] find that major controlling shareholders enjoy higher earnings rate when they have greater cash-flow rights and that they mainly use RPT sales to achieve their objectives.

Based on the conflict-of-interest view, RPTs hinder the efficient allocation of firm resources, which is likely to reduce investments with long-term effects, just as real earnings management sacrifices long-term growth to increase short-term profits [24].

CSR activities are essential for sustainable growth and their importance has been emphasized, but its reflection in firm performance takes a long time (https://assets.kpmg/content/dam/kpmg/be/pdf/2017/kpmg-survey-of-corporate-responsibility-reporting-2017.pdf, accessed on 5 May 2021). Stakeholder theory provides the ground that CSR is ultimately a long-term investment to increase firm value and many studies support the stakeholder view of CSR by showing that CSR is positively related to firm reputation and performance [25,26,27]. Boubakri et al. [15] also present the evidence that CSR in emerging market economies is less than in developed economies, but it is related to higher firm value and lower cost of equity capital. El Ghoul et al. [28] find that CSR plays an important role as a non-market mechanism when countries have weaker market institutions, which results in improved competitive advantage. Corporate efforts to build a good image and balance economic and socio-environmental elements through CSR extend to cooperation with non-governmental organizations (NGOs) [6,7]. CSR initiatives of small and medium-sized enterprises contribute to firm performance, particularly when management accountants participate in CSR management [8].

Using Indian listed companies, Hendratama and Barokah [4] find a negative association between RPTs and firm value, but this turns positive when firms engage in better CSR. They also provide evidence that some RPTs may enhance firm value in the short term, finally leading to managerial opportunism through tunneling. This increases the possibility that lots of RPTs are related to value-destroying activities. The study shows the moderating role of CSR but does not explain the relationship between RPTs and CSR. Recently, Ryu and Chae [29] find high CSR performance of firms engaging in RPT with Korean listed companies and argue that the result is the opportunistic use of CSR in firms with agency problem. However, RPT needs to be investigated by not only its existence but also the total transaction volume because, if it is not a single large-scale transaction such as M&A, the more RPTs, the greater their influence on various corporate decisions including CSR.

Despite the importance of CSR, its benefits are uncertain and accompanied by a variety of costs such as cost of CSR activities, cost of participating in global value chains, and opportunity costs [30]. The uncertainty of CSR is greater in underdeveloped financial markets and poor corporate governance, which leads to lower CSR [15]. RPTs are prevalent in Asian countries where legal protection for investors is weak and corporate governance is not good. Consistent with the conflict-of-interest view, Korean RPTs are determined for controlling shareholders or managers rather than for the firm itself. Both of them are likely to prefer a definite increase in short-term profits rather than an increase in uncertain future value. Those transactions are less likely to enhance firm value and can reduce the capacity of resources to be invested in CSR, thus RPTs are expected to be negatively associated with CSR. Given that RPTs are more made under weak governance and low economic power, it is likely to decrease CSR performance in terms of soundness and fairness. However, stakeholder-related CSR activities such as social contribution, environmental management, and employee satisfaction can be less due to the decreased resource by RPTs and expenses burden. Therefore, we do not predict specific CSR activity under the influence of RPTs in advance. Our first hypothesis is set as follows:

Hypothesis 1 (H1).

CSR activities are less for firms with more RPTs.

Korean business groups are generally under the influence of an individual controlling over all affiliations and “while the founding families do not necessarily own majority stakes in the companies, the descendants of the founders often retain control by virtue of long association with the businesses”. (https://www.britannica.com/topic/chaebol accessed on 30 April 2021) RPTs are transactions between two companies, but it does not work against all participating companies. Business groups make decisions at a group level or for the founding families and profit transfers are generally conducted to maximize the wealth of controlling shareholders, thus the effect of RPTs on each firm can vary depending on the ownership of controlling shareholders [18,31].

Market-to-book ratio is greater for firms where controlling shareholders own more cash-flow rights [32]. Controlling shareholders enjoy higher profitability by expropriating profits from companies where they hold less shares through RPT sales [14]. This implies that firms whose performance more directly influence the earnings rates of their controlling shareholders have less incentive to reduce CSR because RPTs enable them to utilize more resources and they want to cover their negative image [33]. In contrast, companies that are negatively affected by RPTs are likely to have difficulty in making necessary investments, especially long-term and continuous investments such as CSR activities. We expect the negative association of Hypothesis 1 to remain only when controlling shareholders have less shares. Our second hypothesis is set as follows:

Hypothesis 2 (H2).

CSR activities are less for firms with more RPTs when the ownership of controlling shareholders is low.

3. Research Design and Sample Selection

3.1. Proxy of CSR Measure

This study examines the effect of RPTs on CSR. To measure CSR, we use CSR scores provided by the Korea Economic Justice Institute (KEJI) based on previous studies [34,35,36,37]. Since 1991, in Korea, the KEJI have annually awarded “the Economic Justice Award” by scoring public companies which actively performed CSR. The KEJI has refined the CSR scores (the KEJI index) through several methodological improvements, and the KEJI index has been widely recognized in academia and used in the CSR literature. Since 2011, the KEJI has been using CSR scores (total of 100 points) for soundness (25 points), fairness (20 points), social contribution (15 points), consumer protection (15 points), environmental management (10 points) and employee satisfaction (15 points). See Appendix A for details and scores for each category.

The calculation process of the KEJI index (CSR) consists of three steps as follows. First, it calculates the actual measures of the indicator according to the given formula. In the case of non-financial industry, there are 57 detailed evaluation indicators. Second, the actual value for each indicator is converted into a 100-point score according to the rating guidelines. The KEJI use the linear interpolation, in other words, “value = min score + {(max score − min score) × (actual score − actual min score)}/(actual max score − actual min score)”. Third, it is calculated as the final score according to the rating weighting ratio for each indicator. That is, the value determined in step 2 is graded in 3 to 5 groups in order to provide a final rating. Accordingly, we used the total score of the KEJI index and scores by six categories as CSR proxy.

3.2. Related Party Transaction (RPT) Measure

According to previous literature [38,39], we define transactions with related parties (RPT), as total related party transactions scaled by total sales. We collect related party transactions from the footnote in the audit report and divide them by total sales for firm i and year t:

$$\begin{array}{c}{\mathrm{RPT}}_{\mathrm{it}}=\frac{{\mathrm{The} \mathrm{Sum} \mathrm{of} \mathrm{Related} \mathrm{Party} \mathrm{Transactions}}_{\mathrm{it}}}{{\mathrm{Total} \mathrm{Sales}}_{\mathrm{it}}}\\ =\frac{\left({\mathrm{Sales}}_{\mathrm{it}}+{\mathrm{Other} \mathrm{Revenues}}_{\mathrm{it}}+{\mathrm{Purchases}}_{\mathrm{it}}+{\mathrm{Other} \mathrm{Expenses}}_{\mathrm{it}}\right)}{{\mathrm{Total} \mathrm{Sales}}_{\mathrm{it}}}\end{array}$$

(1)

3.3. Regression Model

To examine our hypotheses, we use the following multiple regression Equation (2). The second hypothesis is investigated by conducting the regression for two subsamples (high and low controlling shareholder’s equity ratio). The subscript i and t denotes firm and year, respectively.

CSR_i,t = β₀ + β₁RPT_i,t + β₂SIZE_i,t + β₃LEV_i,t + β₄ROA_i,t + β₅FOR_i,t + β₆LAR_i,t + β₇TA_i,t + β₈BIG4_i,t + β₉CASH_i,t +
β₁₀AGE_i,t + β₁₁QR_i,t + β₁₂BTM_i,t + Industry Dummy + Year Dummy + ε_i,t

(2)

Our interest variable is RPT and Hypothesis 1 expects firms with more RPTs to engage in less CSR activities. If the first hypothesis is supported, β₁ will be negative. The second hypothesis examines whether the ownership of controlling shareholders affects the result, and we predict that β₁ will be negative for low controlling shareholder’s equity ratio subsample.

In investigating the association between RPT and CSR, we control for various factors related to CSR performance following previous studies [36,40]. SIZE is the natural logarithm of the total assets and LEV is the ratio of total liabilities to total assets. ROA is net income scaled by total assets at the beginning of the fiscal year. FOR is foreign ownership ratio and LAR is the ownership ratio of largest (controlling) shareholders. TA is total accruals (net income less cash flows from operating activities) scaled by total assets at the beginning of the fiscal year. BIG4 is the indicator variable which equals one if the auditor belongs to the big four accounting firms (PwC, KPMG, Deloitte, and E&Y), and zero otherwise. CASH is the sum of cash and cash equivalents to total assets at the beginning and AGE is defined as natural logarithm of the number of years elapsed since establishment date. QR is the ratio of total current assets to total current liabilities and BTM is the book value to the market value of the equity. It is expected that CSR performance will be higher as firms are larger, profitable, and have greater foreign ownership. In contrast, more leveraged and higher ownership of controlling shareholders is predicted to negatively affect CSR performance. We also include industry and year dummies to control for their fixed effects. In all regression analyses, t-statistics are adjusted at the firm level.

3.4. Sample Selection

We used Korean public firms from 2011 to 2018. For the homogeneity of sample, the target period starts from 2011, when all listed companies in Korea mandatorily adopted International Financial Reporting Standards (IFRS), and CSR indicators were also changed from 7 to 6 (Until 2010, the KEJI have included “contribution to economic development” in the index with a proportion of 10%. As it was excluded in the KEJI index since 2011, we used sample from 2011 to maintain the comparability of the sample.) indicators.

Table 1. Sample Selection Procedure and Distribution.

Panel A: Sample Selection
Sample Selection Process		Number of firms
Sample which CSR data are available		2789
Deduction: financial industry companies		(195)
Deduction: firms without RPTs data		(87)
Final Sample Selected		2507
Panel B: Distribution by Year
Year	Number of firms	Percent
2011	274	10.93
2012	356	14.20
2013	355	14.16
2014	343	13.68
2015	290	11.57
2016	356	14.20
2017	352	14.04
2018	181	7.22
Total	2507	100.00
Panel C: Distribution by Industry
Industry	Number of firms	Percent
Construction Industry	61	2.43
Education Services	9	0.36
Wholesale and Retail Business	206	8.22
Facilities and Business Supporting Services	20	0.80
Food and Accommodations Industry	7	0.28
Art and Leisure-related Services	10	0.40
Transportation Industry	86	3.43
Electricity, Water and Gas	45	1.79
Science & Technical Support Services	155	6.18
Manufacturing Industry	1776	70.84
Publication, Information Business and Broadcasting	102	4.07
Others	30	1.20
Total	2507	100.00

shows sample selection procedures in Panel A and distribution in Panel B and C. As provided in Panel A of Table 1; we begun with 2789 firm-years that are available both the KEJI index (CSR) and related financial data from FnGuide database. After that, we excluded 195 firm-years belonging to financial industries whose CSR indicators are different from those of non-financial industries. In addition, 87 observations were additionally deleted because the control variables needed for this study could not be obtained. Thus, final sample of the study consisted of 2507 firm-year observations.

Panel B and C of Table 1 provide our sample distribution by year and industry. As shown in Panel B of Table 1, 274 to 356 companies were included in the sample every year, excluding 2018. Meanwhile, in 2018, the KEJI provided data on only 200 companies, and the sample decreased to 181 companies. As shown in Panel C of Table 1, approximately 82 percent of our samples belongs to Manufacturing (71%) and Wholesale and Retail (8%).

4. Empirical Results

4.1. Descriptive Statistics and Pearson Correlations

Table 2. Descriptive Statistics.

	N	Mean	S.D.	MIN	25%	Median	75%	MAX
CSR	2507	62.246	3.122	55.030	60.087	62.171	64.385	69.919
soundness	2507	16.616	1.727	12.505	15.442	16.647	17.791	20.479
fairness	2507	15.320	1.418	11.500	14.850	15.350	16.850	17.150
contribution	2507	6.235	1.509	3.759	4.823	6.468	7.269	10.371
consumer	2507	5.838	1.805	4.550	4.550	5.150	6.050	10.250
environment	2507	9.167	1.745	4.550	9.650	9.650	10.250	10.700
employee	2507	9.075	1.141	6.545	8.292	9.145	9.935	11.279
RPT	2507	0.286	0.314	0.000	0.046	0.168	0.430	1.375
SIZE	2507	19.809	1.279	17.509	18.961	19.609	20.416	23.853
LEV	2507	0.375	0.186	0.026	0.224	0.367	0.522	0.798
ROA	2507	0.047	0.043	−0.053	0.019	0.039	0.065	0.214
FOR	2507	0.102	0.131	0.000	0.013	0.046	0.144	0.642
LAR	2507	0.455	0.155	0.120	0.341	0.463	0.564	0.820
TA	2507	−0.015	0.058	−0.165	−0.046	−0.016	0.011	0.187
BIG4	2507	0.667	0.471	0.000	0.000	1.000	1.000	1.000
CASH	2507	0.065	0.069	0.000	0.015	0.042	0.090	0.345
AGE	2507	3.498	0.684	1.099	3.332	3.714	3.932	4.443
QR	2507	2.838	5.242	0.197	1.068	1.587	2.710	42.002
BTM	2507	1.218	0.728	0.158	0.682	1.057	1.599	3.623
TQ	2507	1.142	0.627	0.453	0.772	0.969	1.273	4.213

Notes: Appendix B presents variable definitions. All continuous variables are winsorized at 1% and 99% levels.

shows descriptive statistics for the main variables used in the analyses. All of the continuous variables except the indicator variables are winsorized at the top and bottom 1% tails of distribution. The mean and median values of CSR, total CSR scores of the 6 categories (soundness, fairness, social contribution, consumer protection, environment management, and employee satisfaction), are 62.246 and 62.171, respectively. The mean (median) and standard deviation of RPT are 0.286 (0.168) and 0.314, respectively, indicating that approximately 29% of sales are related transactions on average. The size of a company (SIZE) is relatively widely distributed with an average of 19.809 and a standard deviation of 1.279. On average, the debt-to-asset ratio (LEV) is about 37.5%, and about 66.7% of the sample appoint BIG4 as auditor (BIG4). The average values for FOR and LAR presents about 10.2% and 45.5%, respectively. The means of current ratio (QR) and book-to-market ratio (BTM) are 283.8% and 121.8%, respectively. The average (median) Tobin’s Q (TQ) of our sample is 1.142 (0.969).

Table 3. Pearson Correlations.

	CSR	RPT	SIZE	LEV	ROA	FOR	LAR	TA	BIG4	CASH	AGE	QR	BTM
RPT	−0.023
	0.258
SIZE	0.186	0.152
	0.000	0.000
LEV	−0.083	−0.104	0.166
	0.000	0.000	0.000
ROA	0.193	0.004	0.041	−0.240
	0.000	0.852	0.038	0.000
FOR	0.233	0.102	0.463	−0.162	0.283
	0.000	0.000	0.000	0.000	0.000
LAR	−0.274	0.095	−0.051	−0.097	−0.043	−0.189
	0.000	0.000	0.011	0.000	0.031	0.000
TA	−0.008	0.047	−0.068	−0.042	0.176	−0.061	0.009
	0.702	0.019	0.001	0.037	0.000	0.002	0.668
BIG4	0.090	0.119	0.390	0.032	0.053	0.222	0.031	−0.024
	0.000	0.000	0.000	0.109	0.008	0.000	0.125	0.229
CASH	0.009	−0.124	−0.111	−0.163	0.254	0.089	−0.143	−0.065	0.012
	0.648	0.000	0.000	0.000	0.000	0.000	0.000	0.001	0.543
AGE	−0.019	−0.027	−0.042	−0.030	−0.151	−0.064	−0.055	−0.003	−0.094	−0.098
	0.353	0.183	0.037	0.132	0.000	0.001	0.006	0.873	0.000	0.000
QR	0.001	0.179	−0.089	−0.446	0.011	0.054	0.077	0.067	−0.052	0.072	0.003
	0.959	0.000	0.000	0.000	0.585	0.006	0.000	0.001	0.010	0.000	0.875
BTM	−0.150	−0.012	−0.062	−0.057	−0.336	−0.189	0.168	0.015	−0.077	−0.131	0.175	0.109
	0.000	0.565	0.002	0.004	0.000	0.000	0.000	0.451	0.000	0.000	0.000	0.000
TQ	0.182	0.016	0.077	0.034	0.374	0.284	−0.186	−0.019	0.064	0.168	−0.159	−0.079	−0.692
	0.000	0.421	0.000	0.085	0.000	0.000	0.000	0.352	0.001	0.000	0.000	0.000	0.000

Notes: Appendix B presents variable definitions. p-values are below correlation coefficients.

shows the Pearson correlations among our key variables. RPT is negatively correlated with CSR but is not statistically significant. We observe that there are significantly positive correlations between CSR and SIZE, ROA, FOR, BIG4 and TQ at the 1% level. Conversely, CSR showed significantly negative correlations with LEV, LAR, and BTM. RPT, our variable of interest, has a statistically positive relationship with SIZE, FOR, LAR, BIG4, and QR under 1% significance, and negative relationship with LEV and CASH. Since correlation analysis alone cannot grasp the association between RPTs and CSR and the effect of largest (controlling) shareholder’s ownership on it, it is necessary to perform multivariate regression analysis in which related variables are controlled.

4.2. Related Party Transactions (RPTs) and Corporate Social Responsibility (CSR)

Table 4. The Relationship between RPTs and CSR.

Variables	Coefficient		t-Statistics
Intercept	59.288	***	35.45
RPT	−0.625	**	−2.50
SIZE	0.280	***	3.54
LEV	−1.335	***	−2.63
ROA	8.788	***	5.07
FOR	1.185	*	1.76
LAR	−5.026	***	−10.37
TA	−1.782	*	−1.74
BIG4	0.193		1.15
CASH	−3.165	***	−3.04
AGE	−0.032		−0.26
QR	0.011		0.67
BTM	−0.392	***	−3.77
Year Dummy	Included
Industry Dummy	Included
Adj. R2	0.327
F statistics	37.58 ***
MAX VIF	1.68
N	2507

Notes: Appendix B presents variable definitions. Standard errors are adjusted at firm levels. ***, **, * indicate significance at 1%, 5%, and 10% levels (two-tailed), respectively.

presents the results for the relationship between RPTs and CSR. We find that the coefficient on RPT is significantly negative (coefficient = −0.625, t-value = −2.50), supporting the negative effect of RPTs on CSR as expected in Hypothesis 1.

To identify CSR activities that are negatively affected by RPTs, we reexamine the analysis using the sub-CSR scores and the results are shown in

Table 5. The Relationship between RPTs and CSR Using Sub-CSR scores.

	Dependent Variables: Sub-CSR Scores
Variables	(1) Soundness	(2) Fairness	(3) Social Contribution	(4) Consumer Protection	(5) Environment Management	(6) Employee Satisfaction
	Coefficient (t-statistics)	Coefficient (t-statistics)	Coefficient (t-statistics)	Coefficient (t-statistics)	Coefficient (t-statistics)	Coefficient (t-statistics)
Intercept	14.054 ***	22.570 ***	2.436 ***	1.666 ***	10.203 ***	8.221 ***
	(16.17)	(26.70)	(3.56)	(4.81)	(25.88)	(14.29)
RPT	−0.633 ***	−0.237 *	0.006	0.098 *	0.009	0.146
	(−4.79)	(−1.79)	(0.05)	(1.69)	(0.15)	(1.55)
SIZE	0.270 ***	−0.328***	0.228 ***	0.173 ***	−0.031	−0.027
	(7.33)	(−8.25)	(7.35)	(10.17)	(−1.64)	(−0.97)
LEV	−1.571 ***	−0.085	0.079	0.252 **	0.023	−0.015
	(−5.86)	(−0.37)	(0.33)	(2.48)	(0.22)	(−0.08)
ROA	2.448 ***	0.392	4.367 ***	0.518	−0.374	1.364 **
	(2.70)	(0.52)	(4.97)	(1.47)	(−0.92)	(2.24)
FOR	1.031 ***	0.430	−0.159	0.418 ***	−0.197	−0.294
	(3.12)	(1.37)	(−0.57)	(2.65)	(−1.34)	(−1.07)
LAR	−4.005 ***	−0.169	−0.249	−0.236 **	0.023	−0.344 *
	(−15.04)	(−0.71)	(−1.20)	(−2.26)	(0.20)	(−1.93)
TA	−1.744 ***	0.145	0.927 *	−0.439 **	0.171	−0.866 **
	(−3.31)	(0.31)	(1.87)	(−2.27)	(0.68)	(−2.40)
BIG4	0.268 ***	−0.101	0.095	−0.059	−0.020	0.012
	(2.97)	(−1.36)	(1.30)	(−1.63)	(−0.65)	(0.18)
CASH	−0.989 *	−0.896 *	−0.940 **	0.044	0.345	−0.772 *
	(−1.75)	(−1.76)	(−2.01)	(0.20)	(1.52)	(−1.82)
AGE	−0.012	−0.039	−0.077	−0.032	0.046 *	0.076 *
	(−0.18)	(−0.63)	(−1.54)	(−1.29)	(1.70)	(1.87)
QR	0.001	0.005	0.001	−0.002	−0.003	0.008 *
	(0.12)	(0.79)	(0.12)	(−0.96)	(−1.41)	(1.74)
BTM	−0.199 ***	0.049	−0.270 ***	0.017	0.040 *	−0.036
	(−3.30)	(1.00)	(−5.43)	(0.75)	(1.82)	(−0.83)
Year	Included	Included	Included	Included	Included	Included
Industry	Included	Included	Included	Included	Included	Included
Adj. R2	0.431	0.301	0.237	0.918	0.885	0.372
F-statistics	58.46 ***	33.24 ***	24.10 ***	873.41 ***	598.91 ***	45.80 ***
N	2507	2507	2507	2507	2507	2507
Clustering	Firm	Firm	Firm	Firm	Firm	Firm
Max VIF	1.68	1.68	1.68	1.68	1.68	1.68

Notes: Appendix B presents variable definitions. Standard errors are adjusted at firm levels. ***, **, * indicate significance at 1%, 5%, and 10% levels (two-tailed), respectively.

. Model (1), (2), (3), (4), (5), and (6) are the result for soundness, fairness, social contribution, consumer protection, environmental management, and employee satisfaction, respectively.

In model (1) and (2), the coefficients on RPT are significantly negative, suggesting that firms with more RPTs are likely to have poor governance and transparency, consistent with prior studies and regulators’ concerns [10,18,31]. The significantly positive result of model (4) can be interpreted as firms use consumer related CSR activities such as protecting consumer rights and product quality improvement to mitigate anti-corporate mood caused by RPTs [16]. RPT do not show significant relationship with the other three perspectives (social contribution, environmental management, employee satisfaction), implying that RPT firms do not negatively affect all CSR.

To investigate whether RPT transaction type affects our results, we measure RPT separately by RPT_S (Sales + Other Revenues) and RPT_P (Purchases + Other Expenses) and reexamine the analysis. As shown in

Table 6. The Relationship between the Type of RPTs and CSR.

Variables	Dependent Variables: CSR
	(1) Sales and Other Revenues: RPT_S			(2) Purchase and Other Expenses: RPT_R
	Coefficient		t-Statistics	Coefficient		t-Statistics
Intercept	59.404	***	35.38	59.330	***	35.52
RPT_S	−0.643	*	−1.94
RPT_P				−0.895	*	−1.83
SIZE	0.275	***	3.47	0.270	***	3.41
LEV	−1.373	***	−2.71	−1.270	**	−2.51
ROA	8.881	***	5.09	8.691	***	5.03
FOR	1.098	*	1.66	1.224	*	1.82
LAR	−5.079	***	−10.48	−5.046	***	−10.40
TA	−1.846	*	−1.79	−1.754	*	−1.71
BIG4	0.194		1.15	0.174		1.04
CASH	−3.164	***	−3.01	−2.863	***	−2.74
AGE	−0.034		−0.27	−0.022		−0.18
QR	0.011		0.68	0.005		0.33
BTM	−0.383	***	−3.66	−0.398	***	−3.79
Year Dummy	Included			Included
Industry Dummy	Included			Included
Adj. R2	0.327			0.326
F-statistics	37.38 ***			37.32 ***
N	2507			2507
Clustering	Firm			Firm
Max VIF	1.68			1.68

Notes: Appendix B presents variable definitions. Standard errors are adjusted at firm levels. ***, **, * indicate significance at 1%, 5%, and 10% levels (two-tailed), respectively.

, the coefficients of RPT_S and RPT_P are significantly negative, suggesting that RPTs act as a deterrent to CSR regardless of transaction type.

Regarding control variables, SIZE, ROA have positive coefficients while LEV, LARGE, CASH, BTM are negatively associated with CSR, consistent with prior studies [36,40].

4.3. Ownership Effect on the Relationship between RPTs and CSR

We divide the sample into two based on the ownership of controlling shareholders and examine its influence on the association between RPTs and CSR.

Table 7. Ownership Effect on the Relationship between RPTs and CSR.

Variables	Dependent Variables: CSR
	Sample with Low Controlling Ownership			Sample with High Controlling Ownership
	Coefficient		t-Statistics	Coefficient		t-Statistics
Intercept	59.409	***	22.96	57.990	***	23.91
RPT	−1.152	***	−2.87	−0.168		−0.57
SIZE	0.338	***	2.76	0.247	**	2.05
LEV	−1.568	**	−2.15	−1.197	*	−1.72
ROA	8.726	***	3.45	10.195	***	4.33
FOR	0.221		0.18	1.923	**	2.38
LAR	−4.251	***	−3.57	−4.184	***	−3.52
TA	−2.180		−1.58	−1.810		−1.15
BIG4	−0.106		−0.42	0.428	*	1.91
CASH	−2.939	**	−2.07	−3.863	***	−2.71
AGE	−0.165		−0.91	0.057		0.33
QR	0.040		1.49	0.001		0.08
BTM	−0.545	***	−3.55	−0.263	*	−1.88
Year Dummy	Included			Included
Industry Dummy	Included			Included
Adj. R2	0.303			0.303
F-statistics	17.19 ***			17.18 ***
N	1254			1253
Clustering	Firm			Firm
Max VIF	2.17			1.68

Notes: Appendix B presents variable definitions. Standard errors are adjusted at firm levels. ***, **, * indicate significance at 1%, 5%, and 10% levels (two-tailed), respectively.

show the results of low and high ownership of controlling shareholders, respectively.

The coefficient of RPT is significantly negative (coefficient = −1.152, t-value = −2.87) in firm-years with low ownership of controlling shareholders while insignificant (coefficient = −0.168, t-value = −0.57) in high ownership sub-sample; this is consistent with Hypothesis 2. Given that controlling shareholders make decisions of affiliations for their benefits by favoring firms which they have more shares [10,14], the result may indicate that firms with low ownership of controlling shareholders cannot afford to invest in CSR due to the low availability of resources caused by RPTs.

4.4. Additional Analysis

4.4.1. RPTs and Firm Value

Based on prior studies [14,18], we assume that firms with higher ownership of controlling shareholders benefit from RPTs. However, our sample and target period are not same with prior studies, thus we need to reexamine the assumption by using our sample for the robustness of our results. We investigate whether the association between RPT and firm value is different depending on the ownership of controlling shareholders and present the results in

Table 8. The Association between RPTs and Firm Value and the Role of Ownership Effect.

Variables	Dependent Variables: TQ
	Full Sample			Sample with Low Controlling Ownership			Sample with High Controlling Ownership
	Coeff		t-Stat	Coeff		t-Stat	Coeff		t-Stat
Intercept	1.257	***	3.20	1.898	***	3.67	0.505		0.77
RPT	0.043		0.65	−0.079		−0.76	0.163	*	1.94
SIZE	−0.032		−1.39	−0.052	**	−2.24	−0.006		−0.14
LEV	0.588	***	4.50	0.695	***	3.72	0.581	***	3.38
BIG4	0.031		0.77	0.061		1.05	0.011		0.23
ROA	4.848	***	7.82	4.998	***	5.92	4.741	***	4.77
FOR	0.877	***	2.85	0.987	**	2.57	0.749	*	1.66
LAR	−0.466	***	−3.23	−1.364	***	−4.56	0.063		0.22
AGE	−0.079	**	−2.21	−0.096	*	−1.77	−0.062		−1.25
Year Dummy	Included			Included			Included
Industry Dummy	Included			Included			Included
Adj. R2	0.288			0.342			0.239
F-statistics	35.75 ***			23.69 ***			14.28 ***
N	2507			1254			1253
Clustering	Firm			Firm			Firm
Max VIF	1.62			2.13			1.48

Notes: Appendix B presents variable definitions. Standard errors are adjusted at firm levels. ***, **, * indicate significance at 1%, 5%, and 10% levels (two-tailed), respectively.

. Model (1), (2), and (3) show the results of full, low ownership, and high ownership, respectively.

We find a significantly positive coefficient (coefficient = 0.163, t-value = 1.94) on RPT only in the high ownership sub-sample; this is consistent with our assumption. In the low ownership sub-sample, the coefficient on RPT is negative, but not significant (coefficient = −0.079, t-value = −0.76), suggesting that RPTs for firms with low ownership of controlling shareholders have no positive effect on firm value.

4.4.2. Endogeneity Issue

Since the interest variable of the study, RPT, may have an endogenous relationship with CSR, there could exist bias in the OLS estimates by Equation (2). Accordingly, 2SLS model using instrumental variables is used to control endogeneity. First, as in Equation (3), we obtain the instrumented (predicted) value of RPT (IvRPT) using the lagged RPT as instrumental variable. Controls are vectors of variables that influence RPT. Then, we replace IvRPT to the model (4).

1st: RPT_i,t = β₀ + β₁RPT_i,t−1 + β₂Controls_i,t−1 + ε_i,t

(3)

2nd: CSR_i,t = β₀ + β₁IvRPT_i,t + β₂Controls_i,t + ε_i,t

(4)

In

Table 9. Sensitivity Test: Using Instrumental Variable.

Variables	Dependent Variables: CSR
	Full Sample			Sample with Low Controlling Ownership			Sample with High Controlling Ownership
	Coeff		t-Stat	Coeff		t-Stat	Coeff		t-Stat
Intercept	60.106	***	27.96	62.047	***	19.04	55.567	***	17.21
IvRPT	−0.848	**	−2.35	−1.370	**	−2.19	−0.346		−0.82
SIZE	0.278	***	2.82	0.248		1.57	0.331	**	2.24
LEV	−1.083	*	−1.71	−1.002		−1.04	−1.022		−1.18
ROA	9.473	***	4.40	8.254	***	2.77	11.914	***	4.00
FOR	0.829		1.05	0.994		0.67	0.738		0.83
LAR	−5.637	***	−9.88	−4.277	***	−3.47	−4.873	***	−3.55
TA	−2.142		−1.60	−2.487		−1.46	−1.269		−0.59
BIG4	0.233		1.21	0.021		0.07	0.341		1.25
CASH	−3.967	***	−3.11	−3.415	**	−1.99	−5.164	***	−2.86
AGE	−0.025		−0.17	−0.252		−1.23	0.090		0.43
QR	0.014		0.52	0.076		1.43	−0.007		−0.24
BTM	−0.432	***	−3.26	−0.644	***	−3.30	−0.222		−1.20
Year Dummy	Included			Included			Included
Industry Dummy	Included			Included			Included
Adj. R2	0.354			0.306			0.312
F-statistics	27.66 ***			12.69 ***			13.04 ***
N	1595			798			797
Clustering	Firm			Firm			Firm
Max VIF	1.75			2.04			1.75

Notes: Appendix B presents variable definitions. Standard errors are adjusted at firm levels. ***, **, * indicate significance at 1%, 5%, and 10% levels (two-tailed), respectively.

, IvRPT considering endogeneity still shows a significantly negative value (coefficient = −0.848, t-value = −2.35) with CSR and maintains only sample with low controlling ownership. This shows that the results of this study are still maintained even after controlling for endogeneity (Additionally, in this study, referring to Dyck et al. [41], we conduct the Granger causality test. As a result of the analysis, the relationship between RPT and CSR still shows a negative relationship, but the statistical significance disappears. This is interpreted as a result suggesting that the possibility of simultaneity or reverse causality between RPT and CSR may exist.).

4.4.3. Characteristics of the Controlling Shareholder

Results of this study show that the relationship between RPTs and CSR is differentiated by the ownership of the controlling shareholder. However, in addition to the ownership, the purpose of utilizing RPTs for CSR may vary depending on the characteristics of the controlling shareholder itself. For example, if the controlling (largest) shareholder is an institutional investor or a foreigner, the degree of RPT utilization may be different than if it were not.

Table 10. The Relationship between RPTs and CSR by the characteristics of controlling shareholder.

Variables	Dependent Variables: CSR
	Controlling Shareholder = Institutional Investors			Controlling Shareholder = Foreigners			Other Sample
	Coeff		t-Stat	Coeff		t-Stat	Coeff		t-Stat
Intercept	40.126		1.64	45.072	***	5.08	60.280	***	34.17
RPT	−1.946		−0.64	5.001	**	2.57	−0.714	***	−2.86
SIZE	0.932		1.22	0.598		1.08	0.229	***	2.62
LEV	−6.697		−0.66	7.748		1.54	−1.214	**	−2.33
ROA	−8.332		−0.50	18.371		1.51	9.239	***	5.10
FOR	7.773	***	2.80	−1.064		−0.18	1.077		1.20
LAR	0.278		0.05	−1.493		−0.34	−5.078	***	−9.74
TA	24.913	*	1.77	−12.798	**	−1.25	−2.174	**	−2.10
BIG4	1.400		0.63	−1.881		−0.95	0.262		1.55
CASH	−10.746		−0.80	−1.486	***	−0.18	−2.977	***	−2.77
AGE	0.391		0.28	−1.278	*	−1.75	−0.031		−0.24
QR	−1.746		−0.65	0.342		1.41	0.014		0.88
BTM	1.338	***	2.65	1.154	*	1.69	−0.413	***	−3.87
Year Dummy	Included			Included			Included
Industry Dummy	Included			Included			Included
Adj. R2	0.799			0.746			0.328
F-statistics	2.94 ***			4.86 ***			36.16 ***
N	45			54			2408
Clustering	Firm			Firm			Firm
Max VIF	8.07			8.45			1.88

Notes: Appendix B presents variable definitions. Standard errors are adjusted at firm levels. ***, **, * indicate significance at 1%, 5%, and 10% levels (two-tailed), respectively.

presents the results of analyzing the relationship between RPTs and CSR according to the characteristics of controlling shareholders.

As a result of examining the relationship between RPT and CSR by controlling shareholder, there is no significant relationship in the institutional investor sample and a significant positive relationship in the foreigner’s sample. These results imply that CSR patterns using RPT is differentiated according to the largest (controlling) shareholder’s incentives by the characteristics of the largest shareholder.

5. Conclusions and Discussion

This study analyzes the association between RPTs and CSR using Korean public companies. Consistent with conflict-of-interest view about RPTs, we report empirical evidence that firms with more RPTs show poor CSR performance, indicating that RPTs signal low CSR activities. To clarify what CSR activities are affected by RPTs, we use the sub-CSR scores and find that soundness and fairness are lower with more RPTs. This implies that RPTs in an environment where fair transaction is not expected due to weak governance structures which hinder the efficient allocation of resources, making it difficult to invest properly in CSR. The result is only maintained in firm-years with low ownership of controlling shareholders whose RPTs do not contribute to firm value. This suggests that RPTs negatively affect CSR of affiliations with a small number of shares held by controlling shareholders, known to be sacrificed in the tunneling process. The finding also shows the importance of ownership structure in RPTs. As argued in the efficient contract view, RPTs can make it possible for trading companies to efficiently get and use resources as an alternative market if there is no public market, which contributes to firm value. However, the existence of controlling shareholder pursuing private interests through transactions between affiliations overshadows their positive aspects and can be an obstacle to a company’s long-term growth.

5.1. Practical and Theoretical Suggestions

This study presents several practical suggestions for managers and regulators. The results show that excessive RPTs not only do not contribute to firm value, but also hinder CSR activities. To meet the needs of various stakeholders is a key strategy for competitive advantage in recent business. The negative image of RPTs may also deteriorate the company’s reputation, thus managers need to care about RPT-related decisions for long-term sustainability. Further, the findings of this study indicate that firms with more RPTs are vulnerable from the perspective of soundness and fairness. In other words, they have weak governance and less transparency in transactions, which can negatively affect the wealth of investors and counterparts. To lessen its bad effect and emphasize the importance of CSR for sustainable growth, regulatory bodies in countries with weak investor protection should monitor RPTs closely especially for firms which are likely to conduct RPTs at the expense of their own interests.

Our results theoretically reveal that companies perform more RPTs as a tunneling tool, consistent with conflict-of-interest view. It may seem contrary to the prior finding that a firm’s engagement in RPT urges managers with agency costs to opportunistically use CSR activities [29]. Low-level RPT companies may be able to conceal agency issues through CSR, but it is not easy for firms that do RPTs above a certain level. Therefore, our study extends the literature by showing that managers’ CSR decisions depends on the level of RPTs and suggests that the theory needs to consider of the RPT level.

5.2. Limitations

Our research has several limitations. First, we use Korean settings where business groups whose chairmen’s control power over affiliations is very strong are prevalent and regulators continue to monitor RPTs for fear of their negative impact on the economy. Therefore, our results may not be generalizable to other studies conducted in different research environments from Korea in terms of business groups, ownership structure or RPTs. Second, there are various types of RPTs, but this study measure RPT with sales, other revenues, purchases, and other expenses, and it cannot use collateral or payment guarantees between related parties due to the limited data availability. Thus, our findings may not be same if the variable is measured to reflect additional RPTs. Third, CSR incentives of private and non-profit firms can be different from those of public companies which have lots of stakeholders and try to maintain a good corporate image and reputation for the purposes of better profitability. Therefore, the results may vary depending on the CSR willingness in private and non-profit sectors. Lastly, there could be a possibility that the research model of this study failed to consider all of the variables that affect CSR.