Corporate social performance and firm debt levels: Impacts of the covid-19 pandemic and institutional environments.

This paper examines the relation between corporate social performance (CSP) and firm debt levels and explores the channels between them by focusing on the ongoing health crisis, the COVID-19 pandemic. We use a large sample of public firms from 31 countries between 2002 and 2020. Employing pooled ordinary least squared and firm fixed effects models, after controlling for endogeneity and sample selection bias, we find that during the pre-COVID economic condition, CSP has a significantly positive impact on firm debt levels by reducing financial constraints and enhancing stakeholder engagement. However, during the outbreak, CSP becomes costlier and reveals more managerial agency problems for firms that make such associations attenuated. Furthermore, our evidence suggests that in countries with better institutional environments, the CSP-firm debt levels relation is less pronounced. These results have several implications in terms of investment and capital structure decisions.

Introduction

The debate over how corporate social performance (CSP) affects a firm's financing decisions has received growing attention in recent years, notably in the wake of the COVID-19 pandemic. A review study on CSP by Gillan et al. (2021) highlights the crucial role of CSP on firm financial situations and argues that CSP can bring various advantages to firms, e.g., reducing a firm's risk (see Hong and Kacperczyk, 2009; El Ghoul et al., 2011; Oikonomou et al., 2012; Albuquerqueet al., 2019), ultimately diminishing a firm's costs of capital (see Heinkel et al., 2001; Chava, 2014; Ng and Rezaee, 2015; Pástor et al., 2021), and improving the trust between a firm and its stakeholders (Hong et al., 2019; Lins et al., 2017). On the other hand, the COVID-19 pandemic and the subsequent lockdown brought about an exogenous and unparalleled shock to firms that dramatically affect corporate financing (e.g., Bannigidadmath et al., 2022; Liu et al., 2022; Del Lo et al., 2022). For instance, Li et al. (2020) document an unprecedented increase in commercial loans on banks’ balance sheets during the COVID-19. Halling et al. (2020) show that bond issuance with longer maturities increases significantly in the health crisis, especially for highly-rated bonds. Even though CSP investment is costly, these practices keep the firms immune during the period of crisis (e.g., Albuquerque et al., 2020; W. Ding et al., 2021). Consequently, the COVID-19 shock may have significant impacts on the CSP - firm's financial situations link.

Despite the nexus between CSP and firm-level financial performance has been debated for many years, there is limited proof of the impact of sustainable activities on firm debt financing as well as the mechanisms underneath this link (Goss and Roberts, 2011; Lee and Faff, 2009). In addition, whereas CSP-firm debt levels association can be endogenous to economic situations, we still know little about the impact of the COVID-19 pandemic on this relation. Specifically, recent studies highlight the insurance role of CSP during the COVID-19 crisis that may help firms increase customer loyalty and investor belief (Albuquerque et al., 2020; W. Ding et al., 2021). Prior studies also suggest that CSP investments are costlier and reveal more managerial agency problems during a crisis (Bénabou and Tirole, 2010; Almeida et al., 2009; Buchanan et al., 2018). Ultimately, this paper helps to bridge the knowledge gaps by exploring the mechanisms of the association between CSP and firm debt levels. We further investigate the debated question that is how the CSP-firm debt levels association is affected by the ongoing health crisis, the COVID-19 pandemic. In addition, given that COVID-19 is a global crisis, we conduct an international study that allows us to evaluate the sensitivity of CSP-firm debt levels across countries with different institutional settings.

The CSP-firm debt levels association can be explained by several theories. The agency theory indicates that the board is responsible for monitoring the managers on the firm's CSP policy. Firms’ managers may be reluctant to invest in sustainability practices that do not immediately enhance the financial performance of firms but lead to a firm's financial distress. However, superior sustainability performance helps firms to improve their reputation which consequently makes it easier for firms to access external finance. In contrast, legitimacy theory suggests that unsustainable firms can improve their legitimacy by applying sustainable technologies (Bansal and Clelland, 2004). Because these technologies are costly to begin with and unsure about the long-run benefits, they may result in uncertainty and financial stress for firms. Stakeholder theory holds that good relationships with stakeholders play an essential role in the better performance of firms (Freeman, 2010). If a firm's stakeholders care about sustainability, such firms are more likely to invest in sustainable activities. This helps the firm to improve customer loyalty, investor attitude, and creditor risk assessment. In turn, CSP firms may have fewer financial constraints. On the other hand, trade-off theory suggests that the economic costs of investing in sustainability are greater than the benefits of such investments. To enhance CSP, firms may need to shift their resources from core business to sustainable activities (Trumpp and Guenther, 2017). These practices, in turn, may have negative impacts on a firm's financial performance (Preston and O'Bannon, 1997).

To test our hypotheses, we use a cross-country panel dataset from the Thomson Reuters ESG Database of publicly listed firms in 31 countries for the period from 2002 to 2020. The results confirm that there is a significant and positive relationship between CSP and firm debt levels. We further investigate the mechanisms of this link and find that financial constraints and stakeholder engagements are significantly associated with this nexus in the predicted direction. In addition, the COVID-19 pandemic makes the investment in CSP more costly which reduces the positive role of these initiatives on a firm's capital structure decision. We also find that high-quality institutions moderate the CSP-firm debt levels association.

Our findings make further contributions to the literature. First, although CSP and financial policy are considered among the most important strategic decisions that the firm should even take, to the best of our knowledge, this is the first paper to investigate the association between CSP and firm debt levels by empirically identifying the mechanisms underneath such link. Previous studies argue that CSR could be considered a helpful tool for high-levered firms to avoid bankruptcy (Bae et al., 2019). Firms also employ CSR as a self-defense scheme against managerial discretion costs, resulting in higher demand for using debt to solve agency conflicts and monitor the management (Villarón-Peramato et al., 2018). However, these studies do not empirically examine in which ways CSR affects firm leverage. In this paper, we show that superior CSP enhances firm debt levels by reducing the firm's financial constraints and improving the firm's stakeholder engagements.

Second, our paper fills the gap that investigates the impact of the COVID-19 pandemic on the association between CSP and firm debt levels. Due to the cash-flow shortage arising from COVID-19, firms seek more external funds to manage their liquidity needs (Halling et al., 2020; Li et al., 2020). Firms with high debt levels are even more vulnerable to the pandemic and exposed to higher risk as they are more financially inflexible and face higher bankruptcy risk. This paper shows that during the normal economic condition (i.e., pre-COVID-19), CSP reduces firm risks and helps firms easily access external funds. However, under the negative impact of the COVID-19 outbreak, social sustainability investment becomes costly leading to overinvestment concerns and revealing more severe managerial agency problems. By uncovering the attenuated effect of such a pandemic on CSP-firm debt levels relation, our findings bring further intuitions on the impact of the COVID-19 pandemic on a firm's financing decisions.

Third, this paper highlights the important role of institutional arrangements in capital structure literature. For instance, Fan et al. (2012) find that firms in countries with poor institutions are potentially highly leveraged. Ho et al. (2021) show that strong institutional settings could substitute the role of CSP in reducing leverage adjustment costs. In this paper, we indicate that high-quality institutions could be considered a less costly channel for firms to reduce agency problems and moderate the positive impact of CSP on firm debt levels.

The remainder of the paper is organised as follows. Section 2 summaries the literature and develops the hypotheses. Section 3 explains the data and empirical method. Section 4 discusses the results whereas Section 5 presents the robustness checks. Section 6 concludes the paper.

Literature review and hypothesis development

CSP and firm debt levels

CSP is a firm that simultaneously incorporates environmental protection, economic growth, social efficiency, and corporate governance into its operations to benefit the firm and society (Artiach et al., 2010). Lately, the research on CSP has concentrated on how CSR/CSP may influence the risk and the access to financial resources. For instance, Lee and Faff (2009) and Stellner et al. (2015) show that firms investing in CSP face lower idiosyncratic and credit risks. Firms that care about employees’ welfare such as health, safety and well-being are viewed positively by investors, which enhances the firm value and reduces firm risk (Jo and Na, 2012; Perez-Gladish et al., 2012; Lee and Kim, 2016; Benlemlih et al., 2018). Albuquerque et al. (2020) suggest a causal link between ESG to reduced systematic risk and increased valuations. Hoepner et al. (2019) show that activism on ESG/CSR issues, particularly environmental issues, can lower a firm's downside risk. Better CSP firms are also awarded relatively high ratings by credit rating agencies (Attig et al., 2013; Oikonomou et al., 2014). Seltzer et al. (2020) provide evidence that firms with better environmental performance tend to have higher credit ratings and lower yield spreads, especially when they are located in states with more stringent environmental regulations. These firms, in turn, have better access to financing with a lower cost of capital (Gao et al., 2021; La Rosa et al., 2018, Bolton and Kacperczyk, 2020, Pastor et al., 2020). Particularly, Cheng et al. (2014) find that sustainable firms face lower financial constraints and access easier to external resources. Goss and Roberts (2011) document that banks would discount about 7 to 18 basis points on debts for such firms compared to others. Ultimately, their underlying asset values have lower volatility and lower expected losses from lower default risk.

Furthermore, the literature shows that social and sustainability activities enhance the mutual belief and cooperation between firms and their stakeholders, thus, improve the engagement between them. Godfrey (2005) and Bae et al. (2019) claim that CSR acts as insurance that protects stakeholders, therefore, investments in CSR are regarded as an insurance premium. This, in turn, can reduce negative reactions from customers and diminish competitors’ incentive to exploit the weak points of highly levered firms (Hong et al., 2019; Lins et al., 2017). Moreover, customers are more likely to appreciate high-CSP firms. They believe that such firms will not break implicit contracts even during periods of negative shocks. Lins et al. (2017) suggest that firms investing in social capital have higher trust from their stakeholders and will be paid off when the general level of trust in the market suffers the breakout. Thereby, high-CSP firms have higher profitability, growth, and sales compared to their counterparties, thus, can raise more debt.

Based on the previous discussion, we suggest hypotheses as follows:

Firms with superior CSP have higher debt levels than firms with low CSP.

Superior CSP enhances firm debt levels by reducing the firm's financial constraints.

Superior CSP enhances firm debt levels by improving the firm's stakeholder engagement.

The impact of the COVID-19 pandemic

The impact of CSP on firm debt levels can be endogenous to economic situations. Lins et al. (2017) show that CSR practices keep the firms immune during the period of crisis. As a result, during the global financial crisis, intensity CSR firms had higher profitability, growth, sales, and raised even more debt. Recent studies also confirm the insurance role of CSP during the COVID-19 pandemic. Albuquerque et al. (2020) and W. Ding et al. (2021) claim that CSR engagements can reduce the firm-value drop and lower the risk caused by the breakout. Huang and Ye (2021) show that CSR significantly reduces the risk of over-levered firms during the COVID-19 crisis and protects them from market depression. In general, the recent studies highlight the significance of CSP in maintaining firm value and lowering risk during negative shocks. It implies that in the COVID-19 health crisis, the role of CSP in determining firm debt levels can be even more pronounced. Therefore, we extend H1 with the impact of the COVID-19 pandemic as follows:

The positive relation between CSP and firm debt levels is strengthened during the COVID-19 pandemic.

Some prior studies suggest that ESG investments are costly that generate overinvestment concerns, particularly over the crisis duration. Almeida et al. (2009) show that firms must deal with limited financial resources and are likely to lower their investment during the uncertainty period. Furthermore, CSP investments could reveal managerial agency problems, i.e., firm managers engage in CSP to increase their utility rather than the welfare of shareholders (e.g., B´enabou and Tirole, 2010; Cheng et al., 2014; Di Giuli and Kostovetsky, 2014; Masulis and Reza, 2015). Johnson et al. (2000) and Masulis and Reza (2015) indicate that the crisis enhances further agency conflicts and potentially magnifies the costs of CSP activities. Hong et al. (2012) argue that ESG/CSR performance increases along with idiosyncratic stock returns during the Internet bubble shock. Buchanan et al. (2018) confirm that the financial crisis leads to more severe agency problems and the subsequent costs of ESG activities cause superior CSP firms to suffer larger decreases in firm values. In addition, recent studies show that financial flexibility is specifically valuable during the pandemic, resulting in the effects of COVID-19 being more severe among firms with high debt (R. Ding et al., 2021; R. Fahlenbrach et al., 2021; Ramelli and Wagner, 2020, Liu et al., 2020). From the above, we propose an alternative hypothesis on the impact of the health crisis as COVID-19 on the relation between CSP and firm debt levels as follows:

The positive relation between CSP and firm debt levels is moderated during the COVID-19 pandemic.

The role of institutional environments

Institutional settings are generally considered as an external mechanism to eliminate agency problems (e.g., Ho et al., 2021; Hunjra et al., 2020). La Porta et al. (2002; 2006) find that countries with better institutions have more effective financial markets. Such countries have intense investor protection and legal enforcement that guarantee the implementation of stakeholder rights. Giannetti (2003) further confirms that better institutions reduce information asymmetry and mitigate agency problems. Alves and Ferreira (2011) and Fan et al. (2012) show that firms from countries with higher levels of corruption and less legal enforcement deal with more asymmetric information and have higher debt ratios. These settings are less costly for firms to eliminate agency problems and enhance stakeholder engagement. Consistently, Breuer et al., (2018) confirm that the higher ESG/CSR performance reduces (increases) the cost of capital in countries with strong (weak) investor protection. Hunjra et al. (2020) also find that institutional quality moderates the negative impact of financial development on environmental sustainability. Consequently, we argue that better institutional settings can diminish the positive impact of CSP on firm debt levels.

The positive relation between CSP and firm debt levels is mitigated in countries with a strong institutional environment, ceteris paribus.

Data and methodology

Data

The data set used in this study is retrieved from different sources during the 2002–20201 period. Specifically, we obtain data on ESG factors from the Thomson Reuters ESG database to estimate CSP. Firm financial data is attained from the Thomson Reuters Worldscope via the Datastream databases. The institutional settings information is collected from La Porta et al. (1998) and D. Kaufmann et al. (2009). Only common-securities firms are retained whereas firms that did not have two consecutive years’ data are eliminated to reduce short panel bias. We winsorize both the dependent and independent variables at the 1st and 99th percentiles to eliminate the potential impact of extreme values. Finally, our sample comprised 22,056 firm-year observations from 31 countries.

Methodology

To investigate the relation between CSP and firm debt levels, we first estimate the following baseline regression:where a firm is indexed by i, country by j, and time by t. We use market debt levels (MLEV) as the dependent variable. CSP is the corporate social sustainability performance variable, and X is a set of the firm- and country-level control variables, including market-to-book ratio (MTB), profitability (PROF), tangibility (TANG), depreciation (DEP), liquidity (LIQ), GDP growth rate (GGDP), and inflation rate (INF).2 Following the recent literature (W. Ding et al., 2021; La Rosa et al., 2018, Wellalage and Kumar, 2021), we estimate this model using ordinary least squares. We run regression specifications with and without industry-fixed effects. We also add year and country fixed effects to control for time-series correlation across countries. Moreover, given that both CSP and debt ratios are firm-level choices, we employ firm fixed effects estimators to control for time-invariant unobserved firm-specific factors that may be correlated with the CSP variables and the debt levels. Standard errors are robust to heteroscedasticity and autocorrelation.

Next, to test the channels that link CSP and firm debt levels, we involve the interaction terms of CSP and the channels including financial constraints and stakeholder engagement into equation Eq. (1):where Channels is the channels that link CSP and firm debt levels.

Finally, we include the interaction between CSP and the dummy variable of COVID-19 and the interaction between CSP and institutional variables in Eq. (1) to examine the effect of the COVID-19 crisis and institutional environments, respectively, on the link between CSP and firm debt levels: where COVID is the dummy variable that equals to one if the year is 2020 and zero otherwise. INSENV is the institutional environment variables including the rule of laws (RL), corruption (CO), accounting standards (AS), government effectiveness (GE), regulation quality (RQ), and control of corruption (CC).

Empirical results

Descriptive statistics

Panel A and B of Table 1 report the summary statistics of the main variables for each country and the whole sample, respectively. Overall, descriptive statistics of dependent and independent variables are comparable to those in previous literature (An et al., 2016; Ho et al., 2021). For instance, the mean of market debt ratio in our study is 0.228. The mean scores of ESG and ESCG that are used to proxy for CSP in our sample are 52.595 and 46.342, respectively. Developed countries have higher ESG and ESGC scores relative to developing countries indicating that the formers care more about sustainability. Panel C reports the Pearson correlation coefficients among explanation variables of our main analysis. Except for the high correlation between ESG score and ESGC score, there is no evidence that such variables are highly correlated. Therefore, multicollinearity is unlikely to be a major concern in our regression analysis.

Table 1

Descriptive statistics and correlation coefficients.

This table presents the descriptive statistics including number of observations and means of firm characteristics for each country and full sample in Panel A and entire-sample summary statistics in Panel B. Panel C presents the Pearson correlation coefficients among explanation variables of the main analysis. Stars indicate significant at the 5% level (p < 0.05). The variable definitions are in Appendix A.

Panel A. Descriptive statistics
COUNTRY	Obs.	ESG score	ESGC score	LEV	MTB	PROF	TANG	DEP	LIQ	GGDP	INFL
AUSTRALIA	950	47.391	42.716	0.197	3.330	0.088	0.326	0.040	3.161	2.745	2.414
AUSTRIA	113	58.050	52.607	0.261	1.726	0.118	0.343	0.053	1.443	1.491	1.903
BELGIUM	149	54.866	49.583	0.236	1.856	0.100	0.241	0.045	2.022	1.521	1.997
BRAZIL	106	57.400	46.267	0.318	3.544	0.161	0.316	0.035	1.918	1.749	6.360
CANADA	724	46.896	44.264	0.196	2.809	0.094	0.508	0.051	2.993	1.819	1.760
CHINA	543	39.285	36.141	0.259	2.908	0.097	0.302	0.027	1.444	7.612	2.151
DENMARK	240	53.231	48.336	0.178	5.489	0.165	0.240	0.048	1.880	1.157	1.619
FINLAND	309	58.713	53.086	0.234	2.645	0.139	0.236	0.043	1.619	1.156	1.444
FRANCE	671	65.467	54.865	0.234	2.174	0.117	0.203	0.046	1.390	1.132	1.293
GERMANY	806	60.976	50.401	0.217	2.264	0.122	0.253	0.045	1.697	1.417	1.376
GREECE	51	47.123	45.049	0.302	1.530	0.099	0.344	0.031	1.666	0.597	2.693
HONG KONG	454	42.689	40.176	0.193	2.876	0.112	0.270	0.030	2.308	3.346	3.021
INDONESIA	66	48.365	48.365	0.095	3.954	0.250	0.524	0.041	3.239	5.400	5.204
ISRAEL	69	47.338	40.602	0.290	4.129	0.127	0.218	0.039	1.475	3.506	1.552
ITALY	171	62.406	49.136	0.326	2.066	0.121	0.235	0.043	1.256	0.093	1.709
JAPAN	3950	53.064	49.617	0.203	1.674	0.104	0.290	0.042	2.086	0.850	0.235
MALAYSIA	131	47.543	45.710	0.263	4.006	0.146	0.421	0.042	2.137	5.094	2.264
MEXICO	22	38.065	35.765	0.326	3.509	0.154	0.468	0.045	1.568	2.014	4.095
NETHERLANDS	251	66.480	52.581	0.225	2.467	0.117	0.251	0.045	1.603	1.255	1.611
NEW ZEALAND	76	47.158	44.996	0.255	4.268	0.176	0.349	0.051	1.772	2.645	2.079
NORWAY	157	59.797	52.604	0.227	2.190	0.148	0.387	0.064	1.652	1.489	1.989
PHILIPPINES	25	37.688	36.588	0.346	2.595	0.135	0.354	0.035	1.527	6.173	2.669
SINGAPORE	74	44.264	41.766	0.210	2.180	0.101	0.196	0.028	2.329	6.004	2.194
SOUTH KOREA	703	52.739	46.428	0.259	1.789	0.109	0.350	0.042	1.586	3.226	1.870
SPAIN	137	55.174	49.203	0.339	5.060	0.119	0.237	0.044	1.528	1.508	1.717
SWEDEN	409	59.580	52.829	0.269	2.907	0.139	0.203	0.039	1.567	2.289	1.155
SWITZERLAND	563	54.654	46.785	0.181	3.724	0.118	0.217	0.040	2.353	1.792	0.275
THAILAND	21	53.108	53.108	0.233	2.463	0.152	0.388	0.037	2.142	4.605	1.281
TURKEY	103	52.392	49.652	0.314	3.093	0.149	0.350	0.040	1.661	6.144	8.239
UNITED KINGDOM	1347	55.959	49.492	0.213	3.304	0.133	0.214	0.041	1.619	1.590	2.135
UNITED STATES	8665	50.693	43.259	0.242	4.030	0.110	0.201	0.039	2.568	1.826	1.935
All countries	22,056	52.595	46.342	0.228	3.093	0.113	0.253	0.041	2.214	1.829	1.598
Panel B. Entire-sample summary statistics
	Mean	Median	SD	Min.	P25	P75	Max.
ESG score	52.595	52.170	17.815	0	38.73	66.72	97.89
ESGC score	46.342	44.090	15.968	0	34.73	57.22	95.6
LEV	0.228	0.217	0.168	0	0.099	0.327	1.155
MTB	3.093	2.16	4.042	−8.470	1.31	3.58	29.21
PROF	0.113	0.119	0.135	−1.469	0.076	0.168	0.455
TANG	0.253	0.212	0.185	0	0.110	0.355	0.922
DEP	0.041	0.037	0.025	0	0.025	0.051	0.199
LIQ	2.214	1.648	2.180	0.128	1.210	2.404	28.525
GGDP	1.829	2.000	2.114	−9.132	1.420	2.762	15.240
INFL	1.598	1.593	1.414	−1.736	0.467	2.450	14.714
Panel C. Pearson correlation coefficients
	ESG score	ESGC score	MTB	PROF	TANG	DEP	LIQ	GGDP	INFL
ESG score	1
ESGC score	0.751*	1
MTB	−0.022*	−0.037*	1
PROF	0.112*	0.078*	0.066*	1
TANG	0.059*	0.055*	−0.125*	0.109*	1
DEP	0.073*	0.054*	−0.029*	0.153*	0.398*	1
LIQ	−0.201*	−0.128*	0.019*	−0.180*	−0.176*	−0.183*	1
GGDP	−0.081*	−0.070*	0.077*	0.040*	0.022*	−0.078*	0.009	1
INFL	−0.056*	−0.098*	0.100*	0.086*	0.017	−0.042*	−0.012	0.302*	1

We firstly regress the market debt ratio on CSP measures (ESG and ESGC score) to study how firms’ CSP activities affect their debt levels. Table 2 shows the results from estimating Eq. (1) using pooled OLS (models 1 and 3), pooled OLS with industry, year, and country dummies (models 2 and 4), and firm-fixed effects model (models 5–6). Our variables of interest are two CSP proxies: ESG and ESGC score. All coefficient estimates of our CSP proxies are positive and significantly associated with the market debt ratio at a 1% level. It implies that firms with superior CSP are more likely to use debt.

Table 2

CSP and firm debt levels – baseline regression results.

This table reports the baseline regression results for the impact of CSP on firm debt levels. The dependent variable is market debt ratio (LEV). CSP is proxied by ESG and ESG scores. Control variables include firm and country characteristics. ***, **, * reveal significance at the 1%, 5%, and 10% levels, respectively. T-statistics are reported in parenthesis. The variable definitions are in Appendix A.

	POOLED OLS				FE
VARIABLES	ESG		ESGC		ESG	ESGC
	(1)	(2)	(3)	(4)	(5)	(6)
CSP	0.0489***	0.0498***	0.0316***	0.0471***	0.0299***	0.0171***
	(7.4231)	(7.5981)	(4.3362)	(6.6279)	(3.6878)	(2.9316)
MTB	0.0025***	0.0018***	0.0025***	0.0018***	0.0002	0.0002
	(8.7748)	(6.3135)	(8.7734)	(6.3354)	(0.7551)	(0.7753)
PROF	−0.1840***	−0.1646***	−0.1810***	−0.1629***	−0.1807***	−0.1808***
	(−21.0449)	(−19.1609)	(−20.7157)	(−18.9778)	(−21.8810)	(−21.8801)
TANG	0.1158***	0.0743***	0.1157***	0.0744***	0.0159	0.0155
	(16.9362)	(7.9084)	(16.9033)	(7.9154)	(1.1313)	(1.1030)
DEP	0.1878***	0.3784***	0.1915***	0.3754***	0.1892***	0.1907***
	(3.7672)	(7.0153)	(3.8395)	(6.9577)	(3.0979)	(3.1219)
LIQ	−0.0220***	−0.0206***	−0.0225***	−0.0209***	−0.0087***	−0.0087***
	(−39.9412)	(−36.9342)	(−41.1561)	(−37.9667)	(−14.6243)	(−14.6530)
GGDP	−0.0002	−0.0027**	−0.0004	−0.0027**	−0.0025***	−0.0025***
	(−0.3987)	(−2.3734)	(−0.7132)	(−2.3640)	(−3.7549)	(−3.8043)
INF	0.0045***	−0.0054***	0.0045***	−0.0054***	−0.0045***	−0.0046***
	(5.2563)	(−3.6105)	(5.2996)	(−3.5490)	(−5.2045)	(−5.3163)
Constant	0.2198***	0.2446***	0.2318***	0.2483***	0.2452***	0.2506***
	(43.6060)	(8.3816)	(46.7576)	(8.5125)	(32.2409)	(34.6831)
Observations	18,974	18,974	18,974	18,974	18,974	18,974
R-squared	0.1269	0.2316	0.1252	0.2310	0.0751	0.0749
Industry FE		YES		YES
Year FE		YES		YES	YES	YES
Country FE		YES		YES
Firm FE					YES	YES

These results are not only statistically but economically significant. Specifically, a standard deviation increases in CSP estimates, on average, boosts a firm's debt ratio by 0.273% - 0.871%.3 Thus, provided that the average market debt ratio in the sample is 0.228, a shift of market debt levels for an average firm is about 1.197% (= 0.273%/0.228) to 3.820% (= 0.871%/0.228). The results support our hypothesis H1 that firms with superior CSP use more debt than their counterparties. The results are aligned with the findings by Bae et al. (2019) and Villarón-Peramato et al. (2018), whereas this study investigates the direct impact of CSP on firm debt levels with the mechanisms underneath this link.

We next explore two possible mechanisms that relate to a firm's CSP and debt levels, which are financial constraint and stakeholder engagements. Table 3 presents the results for financial constraint measures including KZ index (models 1–2) and dividend dummy (models 3–4) and stakeholder engagements measure (models 5–6). In all models, the coefficients of CSP proxies are positive and statistically significant at the 1% level indicating that the positive impact of CSP on firm debt levels persists after controlling for the firm's financial constraints and stakeholder engagements. Furthermore, the coefficients on the interactions CSP*KZ, CSP*DIVIDEND, and CSP*STAENG are negatively and statistically significantly at the 1% level. It indicates that the positive relation between CSP and firm debt levels is moderated for firms with low financial constraints and/or high stakeholder engagement. These results are related to Cheng et al. (2014) who find that CSP firms face lower financial constraints and access easier to external resources. They are also consistent with Hong et al. (2019) and Lins et al. (2017) who suggest that CSR can reduce negative reactions from customers and diminish competitors’ incentive to exploit the weak points of highly levered firms. Ultimately, these findings support our hypotheses H2a and H2b.

Table 3

Channels that link CSP and firm debt levels.

This table reports the regression results of Eq. (2) to test the channels that link CSP and firm debt levels. The dependent variable is market debt ratio (LEV). CSP is proxied by ESG and ESG scores. Financial constraint is proxied by KZ Index (KZ) and dividend payout (DIVIDEND). STAENG is the stakeholder engagement. Control variables include firm and country characteristics. ***, **, * reveal significance at the 1%, 5%, and 10% levels, respectively. T-statistics are reported in parenthesis. The variable definitions are in Appendix A.

VARIABLES	Financial Constraint				Stakeholder Engagement
	KZ Index		Dividend Dummy
	ESG (1)	ESGC (2)	ESG (3)	ESGC (4)	ESG (5)	ESGC (6)
CSP	0.0970***	0.0717***	0.0516***	0.0487***	0.0926***	0.0913***
	(12.4035)	(8.3229)	(7.8336)	(6.8269)	(6.5327)	(6.2841)
CSP*KZ	−0.0844***	−0.0567***
	(−7.2941)	(−4.4549)
KZ	0.1672***	0.1481***
	(25.6680)	(23.4451)
CSP*DIVIDEND			−0.0000*	−0.0000*
			(−1.9595)	(−1.8713)
DIVIDEND			0.0000	0.0000
			(0.9800)	(1.0397)
CSP*STAENG					−0.0008***	−0.0009***
					(−2.7820)	(−3.5567)
STAENG					0.0004**	0.0005***
					(2.2090)	(3.4721)
MTB	0.0022***	0.0023***	0.0018***	0.0018***	0.0020***	0.0020***
	(8.5662)	(8.7631)	(6.2324)	(6.2697)	(6.6560)	(6.6789)
PROF	−0.0547***	−0.0536***	−0.1660***	−0.1644***	−0.1682***	−0.1672***
	(−6.6540)	(−6.5179)	(−19.2120)	(−19.0423)	(−17.8432)	(−17.7321)
TANG	0.0626***	0.0647***	0.0728***	0.0730***	0.0733***	0.0726***
	(7.1893)	(7.4291)	(7.7398)	(7.7571)	(7.2783)	(7.2147)
DEP	0.2535***	0.2448***	0.3919***	0.3875***	0.3281***	0.3268***
	(5.0775)	(4.8939)	(7.2586)	(7.1754)	(5.5272)	(5.5023)
LIQ	−0.0149***	−0.0157***	−0.0205***	−0.0209***	−0.0211***	−0.0213***
	(−28.3861)	(−30.2156)	(−36.8616)	(−37.9129)	(−35.5564)	(−36.1460)
GGDP	−0.0018*	−0.0018*	−0.0028**	−0.0028**	−0.0021*	−0.0022*
	(−1.6872)	(−1.6864)	(−2.4158)	(−2.4010)	(−1.7627)	(−1.7770)
INF	−0.0037***	−0.0037***	−0.0052***	−0.0051***	−0.0044***	−0.0043***
	(−2.6633)	(−2.6453)	(−3.4540)	(−3.3832)	(−2.8085)	(−2.7478)
Constant	0.1951***	0.2100***	0.2439***	0.2478***	0.2235***	0.2273***
	(7.2238)	(7.7638)	(8.3659)	(8.4997)	(7.4453)	(7.6107)
Observations	18,916	18,916	18,929	18,929	16,343	16,343
R-squared	0.3464	0.3435	0.2329	0.2323	0.2421	0.2415
Industry FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES
Country FE	YES	YES	YES	YES	YES	YES

To investigate how the health crisis, COVID-19 pandemic, reshapes firm debt levels sensitivity to social and sustainability performance, we extend our data sample until the year 2020 and estimate model Eq. (3). The results are shown in Table 4 . Specifically, whereas the positive relation between CSP and firm debt levels remains unchanged, the coefficients on the interaction between CSP and COVID-19 dummy variable are negatively and statistically significant at the 1% level. It implies that such a health crisis tends to increase the agency problems and enlarge the costs of social and sustainability activities, thus, mitigating the positive effects of CSP on firm debt levels. The results are consistent with hypothesis H3b that the COVID-19 health crisis mitigates the positive effects of CSP on firm debt levels. Buchanan et al. (2018) suggest that the crisis tends to increase the agency problems that cause superior CSP firms to suffer larger decreases in firm values. Recent studies (e.g., Ding et al., 2021; R. Fahlenbrach et al., 2021; Ramelli and Wagner, 2020, Liu et al., 2020) show that the effects of COVID-19 are more severe among firms with high debt. Therefore, such a health crisis tends to enlarge the costs of social and sustainability activities and moderate the positive link between CSP and firm debt levels.

Table 4

Impact of COVID-19 on the association between CSP and firm debt levels.

This table reports the regression results of Eq. (3) that investigates the impact of COVID-19 pandemic on the CSP-firm debt levels association. The dependent variable is market debt ratio (LEV). CSP is proxied by ESG and ESG scores. COVID19 is a dummy variable that equals to 1 if the year is 2020 and zero otherwise. Control variables include firm and country characteristics. ***, **, * reveal significance at the 1%, 5%, and 10% levels, respectively. T-statistics are reported in parenthesis. The variable definitions are in Appendix A.

	POOLED OLS				FE
VARIABLES	ESG		ESGC		ESG	ESGC
	(1)	(2)	(3)	(4)	(5)	(6)
CSP	0.0006***	0.0011***	0.0006***	0.0011***	0.0009***	0.0009***
	(11.0512)	(8.4267)	(10.8714)	(8.2090)	(7.1931)	(6.6492)
CSP*COVID19	−0.0003*	−0.0007***	−0.0004***	−0.0008***	−0.0006***	−0.0007***
	(−1.8617)	(−3.7408)	(−2.5956)	(−4.2796)	(−3.2478)	(−3.7354)
COVID19	0.0683	−0.0244	0.1054**	0.0157	0.0311	0.0850
	(1.4397)	(−0.4512)	(2.1677)	(0.2840)	(0.5550)	(1.4983)
MTB	−0.0000**	−0.0000	−0.0000**	−0.0000	−0.0000	−0.0000
	(−2.2471)	(−1.3802)	(−2.1807)	(−1.4304)	(−0.0837)	(−0.0670)
PROF	0.2359***	0.2375***	0.2364***	0.2382***	0.2192***	0.2199***
	(36.5764)	(36.6419)	(36.6433)	(36.7315)	(31.9859)	(32.0676)
TANG	0.2439***	0.2503***	0.2442***	0.2506***	0.2621***	0.2626***
	(38.7959)	(38.9023)	(38.8315)	(38.9291)	(38.0918)	(38.1397)
DEP	0.3329***	0.3316***	0.3328***	0.3316***	0.3227***	0.3224***
	(55.8903)	(54.8914)	(55.8922)	(54.8730)	(50.5648)	(50.5218)
LIQ	−0.0164***	−0.0162***	−0.0164***	−0.0162***	−0.0065***	−0.0060***
	(−11.4410)	(−10.7824)	(−11.4547)	(−10.7786)	(−3.4863)	(−3.2589)
GGDP	−0.0013	−0.0100***	−0.0015	−0.0102***	−0.0084***	−0.0086***
	(−0.8697)	(−3.5768)	(−1.0288)	(−3.6458)	(−3.1395)	(−3.2392)
INF	0.0027	−0.0078*	0.0028	−0.0078*	−0.0062	−0.0063
	(1.2009)	(−1.8992)	(1.2457)	(−1.8983)	(−1.5814)	(−1.6101)
Constant	0.1586***	0.2393***	0.1588***	0.2386***	0.1476***	0.1451***
	(26.0595)	(3.1152)	(26.0901)	(3.1033)	(6.6856)	(6.5706)
Observations	22,056	22,056	22,056	22,056	22,056	22,056
R-squared	0.6532	0.6597	0.6530	0.6595	0.6742	0.6740
Industry FE		YES		YES
Year FE		YES		YES	YES	YES
Country FE		YES		YES
Firm FE					YES	YES

We next investigate how institutional environments alter the positive impacts of CSP on firm debt levels. Previous studies indicate that such settings are broadly considered as an external control mechanism to eliminate agency problems, moderate asymmetric information, and improve stakeholder engagement (Alves and Ferreira, 2011; Breuer et al., 2018, Fan et al., 2012; Hunjra et al., 2020, Öztekin, 2015). As a result, institutional settings can mitigate the positive relation between CSP and firm debt levels as proposed in H4. We examine Eq. (4) to test this hypothesis. Specifically, we include six different measures of institutional environments and their interactions with CSP measures. Table 5 presents the results. Supporting the baseline findings, the coefficients of CSP measures are positively significant at the 1% level across models. Moreover, the interactions between CSP and institutional settings are negatively significant in all models, indicating that better institutions are an effective way for firms to mitigate agency conflicts and enhance stakeholder engagement. Subsequently, in countries with better institutions, the positive impact of CSP on firm debt levels is relatively attenuated (H4).

Table 5

Impact of institutional environments on the association between CSP and firm debt levels.

This table reports the regression results of Eq. (4) that investigates the impact of institutional environments on CSP-firm debt levels association. The dependent variable is market debt ratio (LEV). CSP is proxied by ESG and ESG scores. INSENV is the institutional environment variable that is proxied by the rule of law (RL), corruption (CO), accounting standards (AS), government efficiency (GE), regulation quality (RQ), and control of corruption (CC). Control variables include firm and country characteristics. ***, **, * reveal significance at the 1%, 5%, and 10% levels, respectively. T-statistics are reported in parenthesis. The variable definitions are in Appendix A.

VARIABLES	RL		CO		AS		GE		RQ		CC
	ESG (1)	ESGC (2)	ESG (3)	ESGC (4)	ESG (5)	ESGC (6)	ESG (7)	ESGC (8)	ESG (9)	ESGC (10)	ESG (11)	ESGC (12)
CSP	0.1136***	0.1289***	0.1678***	0.1945***	0.3743***	0.2494**	0.1642***	0.1980***	0.1177***	0.1175***	0.0725***	0.0992***
	(4.0324)	(4.2442)	(3.2569)	(3.5416)	(3.4600)	(2.1804)	(4.9084)	(5.5156)	(4.2367)	(3.9452)	(3.4663)	(4.3673)
CSP*INSENV	−0.0453**	−0.0549***	−0.0140**	−0.0174***	−0.0044***	−0.0027*	−0.0768***	−0.0984***	−0.0520***	−0.0514**	−0.0186	−0.0357**
	(−2.5011)	(−2.7935)	(−2.3372)	(−2.7263)	(−3.0124)	(−1.7752)	(−3.6056)	(−4.3021)	(−2.6703)	(−2.4476)	(−1.3649)	(−2.4159)
INSENV	0.0020	0.0040	0.4111***	0.4251***	−0.0090***	−0.0103***	−0.0028	0.0060	−0.0694***	−0.0730***	−0.0292**	−0.0230**
	(0.0869)	(0.1717)	(7.2035)	(7.4549)	(−5.2987)	(−6.0427)	(−0.1701)	(0.3708)	(−4.3363)	(−4.6866)	(−2.4702)	(−1.9643)
MTB	0.0021***	0.0022***	0.0017***	0.0017***	0.0017***	0.0017***	0.0021***	0.0021***	0.0021***	0.0021***	0.0021***	0.0022***
	(7.1592)	(7.1899)	(5.7830)	(5.8068)	(5.8514)	(5.8269)	(7.0573)	(7.0789)	(6.9632)	(6.9885)	(7.1595)	(7.1958)
PROF	−0.1870***	−0.1860***	−0.1577***	−0.1559***	−0.1566***	−0.1555***	−0.1862***	−0.1849***	−0.1870***	−0.1863***	−0.1883***	−0.1871***
	(−19.5305)	(−19.4444)	(−18.1848)	(−17.9942)	(−18.0199)	(−17.9224)	(−19.4678)	(−19.3461)	(−19.5689)	(−19.5170)	(−19.6812)	(−19.5725)
TANG	0.0813***	0.0812***	0.0749***	0.0749***	0.0756***	0.0746***	0.0819***	0.0819***	0.0819***	0.0815***	0.0817***	0.0818***
	(8.3219)	(8.3079)	(7.7954)	(7.7874)	(7.8328)	(7.7278)	(8.3809)	(8.3842)	(8.3879)	(8.3496)	(8.3635)	(8.3670)
DEP	0.3665***	0.3685***	0.3697***	0.3701***	0.3675***	0.3713***	0.3617***	0.3647***	0.3586***	0.3615***	0.3673***	0.3663***
	(6.5489)	(6.5931)	(6.7607)	(6.7752)	(6.7069)	(6.7820)	(6.4727)	(6.5287)	(6.4183)	(6.4786)	(6.5693)	(6.5586)
LIQ	−0.0216***	−0.0219***	−0.0204***	−0.0207***	−0.0203***	−0.0206***	−0.0216***	−0.0219***	−0.0216***	−0.0219***	−0.0215***	−0.0219***
	(−35.0422)	(−35.9012)	(−36.2443)	(−37.2206)	(−36.0949)	(−37.0495)	(−35.1018)	(−35.9648)	(−35.1850)	(−36.0383)	(−34.9936)	(−35.8943)
GGDP	−0.0017	−0.0017	−0.0015	−0.0015	−0.0017	−0.0016	−0.0015	−0.0015	−0.0024**	−0.0024**	−0.0004	−0.0004
	(−1.4057)	(−1.4239)	(−1.2444)	(−1.2220)	(−1.3502)	(−1.2868)	(−1.2263)	(−1.2667)	(−2.0164)	(−2.0426)	(−0.3100)	(−0.3253)
INF	−0.0066***	−0.0065***	−0.0066***	−0.0065***	−0.0069***	−0.0068***	−0.0043**	−0.0044**	−0.0038**	−0.0037**	−0.0053***	−0.0052***
	(−3.8673)	(−3.8111)	(−4.1778)	(−4.1506)	(−4.3517)	(−4.2720)	(−2.4768)	(−2.5351)	(−2.3231)	(−2.2769)	(−3.2561)	(−3.2017)
Constant	0.2684***	0.2669***	−3.2564***	−3.3723***	0.9796***	1.0777***	0.2754***	0.2629***	0.3998***	0.4070***	0.3312***	0.3235***
	(5.4991)	(5.4907)	(−6.5968)	(−6.8354)	(7.4177)	(8.1906)	(6.9317)	(6.6304)	(10.0023)	(10.2908)	(8.9343)	(8.7284)
Observations	16,804	16,804	18,391	18,391	18,337	18,337	16,804	16,804	16,804	16,804	16,804	16,804
R-squared	0.2422	0.2421	0.2303	0.2298	0.2291	0.2283	0.2431	0.2431	0.2453	0.2450	0.2427	0.2428
Industry FE	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
Country FE	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES

Endogeneity issues and robustness checks

We first test the robustness of baseline results to potential endogeneity issues by following Cheng et al. (2014) in identifying a shock to the firm's social and sustainability performance. The study indicates that when a firm is rated for the first time by an agency, it will improve its CSP more than other firms. Therefore, we expect that the initiation of the rating process will enhance the positive association between CSP and firm debt levels. The results are in Table 6 . Consistent with our expectation, the interaction term between CSP and initiation of the rating process (CSP*INITIAL_RATING) is positively significant at the 5% level across models. It implies that, on average, superior CSP firms increase their debt levels more than their counterparties if their social and sustainability activities are initially rated.

Table 6

Exogenous shock: Impact of initial rating to the association between CSP and firm debt levels.

This table reports regression results of examining the initial rating as the exogenous shock to CSP performance. The dependent variable is market debt ratio (LEV). CSP is proxied by ESG and ESG scores. INNITIAL_RATING is a dummy variable that equals to 1 if the firm is initiated coverage by Thomson Reuter ESG database and zero otherwise. Control variables include firm and country characteristics. ***, **, * reveal significance at the 1%, 5%, and 10% levels, respectively. T-statistics are reported in parenthesis. The variable definitions are in Appendix A.

VARIABLES	ESG (1)	ESGC (2)
CSP	0.0428***	0.0402***
	(6.1855)	(5.3964)
CSP*INITIAL_RATING	0.0545**	0.0561**
	(2.3251)	(2.2930)
INITIAL_RATING	−0.0316***	−0.0340***
	(−3.0177)	(−3.2164)
MTB	0.0018***	0.0018***
	(6.2974)	(6.3191)
PROF	−0.1656***	−0.1644***
	(−19.2608)	(−19.1281)
TANG	0.0742***	0.0742***
	(7.8941)	(7.9013)
DEP	0.3787***	0.3767***
	(7.0227)	(6.9842)
LIQ	−0.0205***	−0.0208***
	(−36.7559)	(−37.6142)
GGDP	−0.0026**	−0.0026**
	(−2.2942)	(−2.2683)
INFL	−0.0058***	−0.0058***
	(−3.8198)	(−3.8009)
Constant	0.2521***	0.2607***
	(8.5196)	(8.8376)
Observations	18,974	18,974
R-squared	0.2320	0.2316
Industry FE	YES	YES
Year FE	YES	YES
Country FE	YES	YES

We also address the endogeneity issues by employing the instrumental variable approach. We employ the average CSP for each industry in a specific country, excluding the impact of the focal firm as an instrument for CSP. The intuition is that the CSP of a focal firm is systematically affected by the CSP of other firms within the same industry-country (Ioannou and Seafeim, 2012). It is argued that firms from an industry-country pair with higher investment in CSP will have superior social and sustainability performance than companies from a low CSP investment-industry-country pair. The results from the first stage reported in Table 7 show that the coefficients of instrumental variables are positively significant at the 1% level (models 1 and 3). It indicates that firms operating within a high CSP industry-country pair have superior social and sustainability performance, which is consistent with our argument. Furthermore, the significant positive relation between CSP and firm debt levels is maintained at the 1% level (model 2 and 4), confirming our baseline finding.

Table 7

Instrumental variable approach.

This table describes the regression from our instrument variable approach. In the first-stage, we regress CSP variables (ESG and ESGC score) on the instrument and the controls. In the second stage, we regress dependent variable (market debt ratio) on the predicted values of CSP variables and the control variables. Control variables include firm and country characteristics. ***, **, * reveal significance at the 1%, 5%, and 10% levels, respectively. T-statistics are reported in parenthesis. The variable definitions are in Appendix A.

VARIABLES	ESG		ESGC
	First (1)	Second (2)	First (3)	Second (4)
IV	0.8523***		0.7702***
	(0.0090)		(0.0115)
CSP		0.0493***		0.0896***
		(0.0115)		(0.0162)
MTB	−0.0001	0.0018***	−0.0002	0.0018***
	(0.0003)	(0.0003)	(0.0003)	(0.0003)
PROF	0.1289***	−0.1646***	0.0907***	−0.1665***
	(0.0078)	(0.0086)	(0.0079)	(0.0087)
TANG	−0.0066	0.0743***	−0.0032	0.0736***
	(0.0086)	(0.0094)	(0.0087)	(0.0094)
DEP	−0.0946*	0.3784***	−0.0598	0.3760***
	(0.0494)	(0.0538)	(0.0497)	(0.0538)
LIQ	−0.0115***	−0.0206***	−0.0072***	−0.0206***
	(0.0005)	(0.0006)	(0.0005)	(0.0006)
GGDP	−0.0016	−0.0027**	−0.0017	−0.0027**
	(0.0011)	(0.0011)	(0.0011)	(0.0011)
INFL	−0.0023*	−0.0054***	−0.0032**	−0.0053***
	(0.0014)	(0.0015)	(0.0014)	(0.0015)
Constant	0.0864***	0.2448***	0.1201***	0.2313***
	(0.0268)	(0.0294)	(0.0271)	(0.0297)
Observations	18,974	18,974	18,974	18,974
R-squared	0.4262	0.2316	0.2758	0.2296
Industry FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
Country FE	YES	YES	YES	YES

We next test the robustness of our baseline findings by re-estimating our model by employing (1) alternative measures of debt ratio including book debt ratio and active debt ratio,4 (2) alternative econometric method that is dynamic system GMM, (3) alternative dataset that extent the data period to 2002–2020, and (4) alternative subsamples including developed, developing, common laws, and civil law countries. Results are in Table 8, Table 9, Table 10, Table 11 . We find that all robustness checks across models and tables, the coefficients of CSP proxies are positively significant indicating that our baseline results do not alter the approach of estimating the debt ratio, econometric method, the data period, and the sample of countries.

Table 8

Robustness check: alternative measure of leverage including book leverage and active leverage as dependent variable.

This table reports the robustness checks using book leverage and active leverage as dependent variable. CSP is proxied by ESG and ESG scores. Control variables include firm and country characteristics. ***, **, * reveal significance at the 1%, 5%, and 10% levels, respectively. T-statistics are reported in parenthesis. The variable definitions are in Appendix A.

VARIABLES	BLEV						ALEV
	ESG			ESGC			ESG			ESGC
	OLS (1)	OLS (2)	FE (3)	OLS (4)	OLS (5)	FE (6)	OLS (7)	OLS (8)	FE (9)	OLS (10)	OLS (11)	FE (12)
CSP	0.0661***	0.0679***	0.0266***	0.0349***	0.0375***	0.0031	0.0425***	0.0419***	0.0693***	0.0278***	0.0410***	0.0335***
	(9.2890)	(9.7392)	(3.4439)	(4.4366)	(4.9568)	(0.5056)	(6.3356)	(6.2398)	(9.3696)	(3.7503)	(5.6342)	(5.6164)
MTB	−0.0088***	−0.0062***	−0.0036***	−0.0088***	−0.0062***	−0.0036***	0.0030***	0.0023***	0.0006***	0.0031***	0.0023***	0.0006***
	(−28.3521)	(−20.5214)	(−15.8677)	(−28.3091)	(−20.4721)	(−15.8565)	(10.3925)	(8.0532)	(2.7615)	(10.3962)	(8.0744)	(2.7411)
PROF	−0.3185***	−0.2942***	−0.2486***	−0.3139***	−0.2896***	−0.2489***	−0.2866***	−0.2635***	−0.2362***	−0.2841***	−0.2622***	−0.2381***
	(−33.7364)	(−32.2076)	(−27.5555)	(−33.2382)	(−31.6822)	(−27.5762)	(−31.9553)	(−29.7002)	(−27.1890)	(−31.6952)	(−29.5876)	(−27.3492)
TANG	0.2175***	0.1265***	0.0697***	0.2175***	0.1272***	0.0647***	0.1184***	0.0825***	0.0455***	0.1183***	0.0826***	0.0357**
	(29.4700)	(12.6632)	(4.5464)	(29.4194)	(12.7047)	(4.2332)	(16.9998)	(8.5612)	(3.0978)	(16.9734)	(8.5687)	(2.4358)
DEP	−0.2405***	0.1415**	0.2389***	−0.2351***	0.1371**	0.2320***	0.2495***	0.4045***	0.0856	0.2528***	0.4016***	0.0717
	(−4.4685)	(2.4676)	(3.5824)	(−4.3616)	(2.3864)	(3.4794)	(4.9103)	(7.2944)	(1.3362)	(4.9719)	(7.2414)	(1.1181)
LIQ	−0.0280***	−0.0226***	−0.0062***	−0.0287***	−0.0233***	−0.0063***	−0.0207***	−0.0195***	−0.0073***	−0.0212***	−0.0198***	−0.0074***
	(−47.1234)	(−38.0635)	(−9.5354)	(−48.6859)	(−39.6128)	(−9.6486)	(−36.9023)	(−34.0045)	(−11.6151)	(−37.9748)	(−34.8676)	(−11.8125)
GGDP	−0.0043***	−0.0069***	−0.0073***	−0.0045***	−0.0069***	−0.0073***	−0.0004	−0.0028**	−0.0013***	−0.0006	−0.0028**	−0.0013***
	(−6.9230)	(−5.6427)	(−19.2395)	(−7.3542)	(−5.6353)	(−19.2586)	(−0.7648)	(−2.3711)	(−3.5003)	(−1.0127)	(−2.3645)	(−3.6011)
INFL	−0.0055***	−0.0045***	−0.0111***	−0.0055***	−0.0044***	−0.0114***	0.0047***	−0.0047***	−0.0022***	0.0047***	−0.0046***	−0.0026***
	(−5.9790)	(−2.8003)	(−16.4953)	(−5.9574)	(−2.7472)	(−16.9165)	(5.4001)	(−3.0243)	(−3.4611)	(5.4376)	(−2.9685)	(−4.0907)
Constant	0.2820***	0.1968***	0.2628***	0.3019***	0.2127***	0.2775***	0.2200***	0.2202***	0.2158***	0.2302***	0.2228***	0.2410***
	(51.7968)	(6.3417)	(38.6521)	(56.3587)	(6.8449)	(47.4570)	(42.9080)	(7.4159)	(33.1007)	(45.6590)	(7.5054)	(42.9561)
Observations	18,974	18,974	18,974	18,974	18,974	18,974	18,690	18,690	18,690	18,690	18,690	18,690
R-squared	0.2416	0.3526	0.1324	0.2389	0.3502	0.1317	0.1333	0.2264	0.0691	0.1321	0.2261	0.0658
Industry FE		YES			YES			YES			YES
Year FE		YES	YES		YES	YES		YES	YES		YES	YES
Country FE		YES			YES			YES			YES
Firm FE			YES			YES			YES			YES

Table 9

Robustness check: alternative econometric method as dynamic system GMM.

This table represents the robustness test using dynamic system GMM as an alternative econometric method. CSP is proxied by ESG and ESG scores. Control variables include firm and country characteristics. ***, **, * reveal significance at the 1%, 5%, and 10% levels, respectively. T-statistics are reported in parenthesis. The variable definitions are in Appendix A.

VARIABLES	ESG (1)	ESGC (2)
CSP	0.0126***	0.0081**
	(3.0260)	(2.1872)
MTB	0.0022***	0.0023***
	(5.7133)	(6.5108)
PROF	−0.1533***	−0.1495***
	(−9.2025)	(−9.5029)
TANG	0.0269***	0.0223***
	(4.5780)	(4.0596)
DEP	0.0132	0.0117
	(0.3328)	(0.2958)
LIQ	−0.0055***	−0.0050***
	(−7.4657)	(−6.7631)
GGDP	−0.0010*	−0.0006
	(−1.8197)	(−1.1858)
INFL	−0.0006	−0.0002
	(−0.8841)	(−0.3714)
Constant	0.0537***	0.0422***
	(7.4716)	(6.6501)
Observations	15,893	15,893
Number of id	2398	2398
Industry FE	YES	YES
Year FE	YES	YES
Country FE	YES	YES
AR(1)	0.0000	0.0000
AR(2)	0.3289	0.3646
P-value Hansen test	0.0000	0.0000

Table 10

Robustness check: alternative data period.

This table reports the robustness checks using the data period of 2002–2020. CSP is proxied by ESG and ESG scores. Control variables include firm and country characteristics. ***, **, * reveal significance at the 1%, 5%, and 10% levels, respectively. T-statistics are reported in parenthesis. The variable definitions are in Appendix A.

	POOLED OLS				FE
VARIABLES	ESG		ESGC		ESG	ESGC
	(1)	(2)	(3)	(4)	(5)	(6)
CSP	0.0006***	0.0008***	0.0006***	0.0007***	0.0007***	0.0006***
	(12.0250)	(7.9650)	(11.5217)	(7.1778)	(6.5281)	(5.5025)
MTB	−0.0000**	−0.0000	−0.0000*	−0.0000	−0.0000	−0.0000
	(−2.1198)	(−1.4922)	(−1.9455)	(−1.5487)	(−0.1866)	(−0.1793)
PROF	0.2366***	0.2380***	0.2371***	0.2384***	0.2198***	0.2203***
	(36.7342)	(36.7174)	(36.7951)	(36.7613)	(32.0857)	(32.1212)
TANG	0.2443***	0.2505***	0.2447***	0.2507***	0.2622***	0.2626***
	(38.8905)	(38.9191)	(38.9434)	(38.9334)	(38.0894)	(38.1291)
DEP	0.3339***	0.3314***	0.3343***	0.3314***	0.3224***	0.3222***
	(56.2975)	(54.8456)	(56.3643)	(54.8153)	(50.5224)	(50.4708)
LIQ	−0.0164***	−0.0162***	−0.0165***	−0.0162***	−0.0065***	−0.0060***
	(−11.4829)	(−10.7765)	(−11.5046)	(−10.7575)	(−3.4960)	(−3.2510)
GGDP	−0.0007	−0.0093***	−0.0010	−0.0095***	−0.0078***	−0.0080***
	(−0.5406)	(−3.3388)	(−0.7220)	(−3.3992)	(−2.9262)	(−3.0176)
INF	0.0026	−0.0073*	0.0027	−0.0073*	−0.0058	−0.0058
	(1.1634)	(−1.7862)	(1.2191)	(−1.7643)	(−1.4789)	(−1.4916)
Constant	0.1581***	0.2368***	0.1585***	0.2358***	0.1460***	0.1432***
	(26.1809)	(3.0814)	(26.2314)	(3.0656)	(6.6118)	(6.4858)
Observations	22,061	22,061	22,058	22,058	22,061	22,058
R-squared	0.6532	0.6595	0.6529	0.6592	0.6740	0.6737
Industry FE		YES		YES
Year FE		YES		YES	YES	YES
Country FE		YES		YES
Firm FE					YES	YES

Table 11

Robustness check: subsample tests.

This table reports the robustness checks using different subsamples including developed, developing, common law, and civil law countries. CSP is proxied by ESG and ESG scores. Control variables include firm and country characteristics. ***, **, * reveal significance at the 1%, 5%, and 10% levels, respectively. T-statistics are reported in parenthesis. The variable definitions are in Appendix A.

VARIABLES	Developed countries		Developing countries		Common law countries		Civil law countries
	ESG (1)	ESGC (2)	ESG (3)	ESGC (4)	ESG (5)	ESGC (6)	ESG (7)	ESGC (8)
CSP	0.0496***	0.0460***	0.0487***	0.0628***	0.0298***	0.0339***	0.0689***	0.0555***
	(7.1689)	(6.1061)	(2.6107)	(3.3113)	(3.1233)	(3.1323)	(8.1540)	(6.4657)
MTB	0.0016***	0.0016***	0.0099***	0.0099***	0.0010***	0.0010***	0.0043***	0.0043***
	(5.5058)	(5.5453)	(8.4706)	(8.4428)	(3.0201)	(3.0495)	(7.2285)	(7.0956)
PROF	−0.1492***	−0.1474***	−0.6699***	−0.6665***	−0.0919***	−0.0911***	−0.3787***	−0.3781***
	(−17.0199)	(−16.8312)	(−13.4955)	(−13.4408)	(−8.7871)	(−8.7292)	(−22.1943)	(−22.1228)
TANG	0.0726***	0.0727***	0.1030***	0.1044***	0.0458***	0.0458***	0.1728***	0.1708***
	(7.3256)	(7.3305)	(3.7099)	(3.7661)	(3.5386)	(3.5406)	(12.8572)	(12.6868)
DEP	0.4406***	0.4352***	0.2860	0.2919	0.3471***	0.3461***	0.2896***	0.2945***
	(7.7721)	(7.6747)	(1.5592)	(1.5979)	(4.6204)	(4.6077)	(3.9115)	(3.9728)
LIQ	−0.0197***	−0.0201***	−0.0252***	−0.0252***	−0.0173***	−0.0174***	−0.0302***	−0.0311***
	(−34.3970)	(−35.4429)	(−10.6422)	(−10.6360)	(−25.1466)	(−25.5987)	(−29.8546)	(−30.9670)
GGDP	−0.0020	−0.0019	−0.0072***	−0.0071***	0.0025	0.0024	−0.0039***	−0.0037***
	(−1.4886)	(−1.4481)	(−3.1331)	(−3.1122)	(1.0109)	(0.9872)	(−3.1039)	(−2.9921)
INFL	−0.0077***	−0.0076***	0.0018	0.0021	−0.0037	−0.0039	−0.0007	−0.0005
	(−4.6274)	(−4.5662)	(0.5455)	(0.6316)	(−1.1417)	(−1.1941)	(−0.4041)	(−0.3087)
Constant	0.1070**	0.1117**	0.2706***	0.2620***	0.2293***	0.2297***	0.3125**	0.3203**
	(2.4506)	(2.5565)	(5.0821)	(4.9161)	(6.7662)	(6.7826)	(2.5011)	(2.5600)
Observations	17,519	17,519	1455	1455	10,509	10,509	8465	8465
R-squared	0.2291	0.2285	0.5020	0.5035	0.2203	0.2203	0.3644	0.3625
Industry FE	YES	YES	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES	YES	YES
Country FE	YES	YES	YES	YES	YES	YES	YES	YES

Conclusion

This letter explores the relation between CSP and firm debt levels conditional on the economic conditions and institutional environments. We find that CSP and firm debt levels have a positive association through two mechanisms: easing a firm's financial constraints and enhancing stakeholder engagement. More importantly, this relation is attenuated during the health crisis, which is the COVID-19 pandemic, and in countries with strong institutional settings. Our findings remain unchanged from a battery of robustness and endogeneity checks.

Our study has important implications at both organizational and national levels. Firms’ executives could consider investing more in sustainable activities to have more debt financing. However, during the COVID-19 pandemic, such investment becomes costly and does not help firms much with their debt levels. Governments and authorities should also consider using regulations and institutional quality to control firms’ debt levels.

Recommendations for further research, there may be other mechanisms, apart from financial constraint and stakeholder engagements, such as corporate governance or the concentration of ownership that explain the positive relation between CSP and firm debt levels. Therefore, in future research, for example by considering other mechanisms under CSP-firm debt levels relation or extending the scope to a broader number of countries, especially developing ones.

CRediT authorship contribution statement

Min Bai: Conceptualization, Methodology, Supervision, Validation, Writing – review & editing. Ly Ho: Data curation, Investigation, Visualization, Software, Writing – original draft.

Declaration of Competing Interest

None.

Variables	Acronym	Description	Data source
A. Firm-level variable
Book leverage	BLEV	Book value of total debt divided by book value of total assets	Worldscope
Active leverage	ALEV	Book value of total debt divided by the sum book value of total assets and the total net income	Worldscope
Market leverage	MLEV	Book value of total debt to sum of market value of equity and book value of total debt	Worldscope
ESG score	ESG	Measure firm's ESG performance including three components: environment, social, and governance	Thomson Reuters ESG database
ESG combined score	ESGC	Incorporates ESG controversies captured from global media sources that materially and substantially impact the firms.	Thomson Reuters ESG database
Tangibility	TANG	Net property, plant and equipment to book value of assets	Worldscope
Growth opportunity	MTB	Ratio of book value of assets less book value of equity plus market value of equity to book value of assets	Worldscope
Profitability	PROF	Earning before interests, taxes, depreciation and amortization to book value of assets	Worldscope
Depreciation	DEP	Depreciation and amortization to book value of assets	Worldscope
Liquidity	LIQ	Total current assets to total assets	Worldscope
Stakeholder engagement	STAENG	Measures the degree to which a focal company explains the formal processes in place for engagement with its stakeholders. The higher the score is, the stronger the firm's stakeholder engagement is.	Thomson Reuters ESG database
Dividend pay-out	DIVIDEND	A dummy that takes value of one if firm pay the dividend and zero otherwise	Worldscope
KZ index	KZ	The index that consists of a linear combination of five accounting ratios: cash flow to total assets, the market to book ratio, debt to total assets, dividends to total assets, and cash holding to total assets.	Self-calculated following Baker, Stein, and Wurgler (2003)
B. Country-level variable
GDP growth rate	GGDP	Annual GDP growth rate	World Development Indicator
Inflation rate	INFL	Annual growth in consumer price index	World Development Indicator
Rule of law	RL	Measures the law and order tradition in the country. The index is scaled from zero to 10, with lower scores for less tradition for law and order.	La Porta et al. (1998)
Accounting standard	AS	An indicator equal to one if the accounting standard index, which was created by examining and rating companies’ 1990 annual reports for their inclusion or omission of 90 specific accounting items, covering general information, income statements, balance sheets, funds flow statement, accounting standards, stock data, and special items, is greater than its annual median, and zero otherwise	La Porta et al. (1998)
Level of corruption	CO	Measures the corruption level of the government in the country. The index is scale from zero to 10, with lower scores for lower level of corruption.	La Porta et al. (1998)
Government effectiveness	GE	Measures perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies.	D. Kaufmann et al. (2009)
Regulation quality	RQ	Measures perceptions of the ability of the government to formulate and implement sound policies and regulations that permit and promote private sector development.	D. Kaufmann et al. (2009)
Control of corruption	CC	Measures perceptions of the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as "capture" of the state by elites and private interests.	D. Kaufmann et al. (2009)