Corporate Social Responsibility and Firm Debt Maturity

Since the study of Modigliani and Miller (1958), theories on debt maturity have seen considerable evolution. Under the assumption of a perfect market, Modigliani and Miller (1958) show that the structure of debt maturity has no impact on the value of a company; however, in a market in which we admit the existence of some imperfections, such as agency conflicts (Myers 1977), information asymmetry (Flannery 1986), credit risk (Diamond 1991), and taxation (Brick and Ravid 1985), the determinants of debt maturity are relevant to the company.

In line with these studies, Maksimovic and Titman (1991) show that firms that want to commit themselves credibly to providing better employee benefits need to have lower debt ratios and to limit their use of debt.

A reviewer has suggested including the overall CSR score squared as an additional interest variable to investigate whether a non-linear relationship between CSR and debt maturity exists. Our findings from this analysis show that the CSR score squared is negative and statistically significant at the 5 % level, confirming our argument that high CSR firms are likely to use more short-term debt as a monitoring mechanism when CSR performance exceeds an optimal level.

Wood (1995) argues that MSCI ESG (previously known as KLD) provides researchers with the best CSR information available in the US. Sharfman (1996) confirms this, stating that the MSCI ESG database is a valid measure and provides a good evaluation of companies’ social performance. In his view, it can be used by researchers as a standard for measuring CSR. Jo and Harjoto (2011) support this, arguing that MSCI ESG is the most comprehensive and widely used source of data for CSR research.

MSCI ESG has assessed firms in the field of human rights since 1995. These dimensions are not available for the years before 1995 and thus we follow Kim et al. (2012) and Galema et al. (2008) in excluding this dimension to construct our overall CSR score. However, in robustness checks, we find that the results remain unchanged if we include human rights in our overall CSR measure. Our results also remain unchanged when we use Servaes and Tamayo’s (2013) definition of CSR by excluding corporate governance from our overall CSR score.

Several studies (Johnson 2003; Custódio et al. 2013) include the size squared as an additional control variable, to capture the non-linear relation between debt maturity and firm size as predicted by Diamond (1991). However, in our sample, we note that there is a strong correlation between the two measures (the correlation between size and size squared is 0.97). In addition, by integrating the size squared into our basic model (Model 2, Table 6), we find a strong multicollinearity between variables: the variance inflation factors (VIF) of size squared and size are higher than 40. The VIF for the control variables (including size) is around two after excluding size squared. We decided not to include size squared in our control variables. However, in unreported results, we find that even when we do include it; our results remain unchanged, showing a negative and significant relationship between CSR and long-term debt. Furthermore, the size squared coefficient is negative and significant, confirming the non-linear relationship between credit risk and debt maturity.

Johnson (2003) finds an average percentage of 45 % of debt maturing in more than three years (between 1986 and 1995). Chen et al. (2012) report an average of the percentage of debt maturing in more than three years equal to 52.5 % (for a sample between 1974 and 2010), while Custódio et al. (2013) report an average percentage of debt maturing in more than three years equal to 44 %.

In analysis that is not reported because of a lack of space, we run our regression model after bisecting the sample at the median of the scores for each of the control variables. In each of these regressions, the dependent variable is debt maturity, which is contingent on the CSR score, size, abnormal earnings, leverage, term spread, and return on asset volatility. Our findings from this analysis are that the negative impact of CSR on debt maturity is stronger for large firms with a high market-to-book ratio, high return on asset volatility, and high abnormal results, and for firms with low asset maturity. The effect of CSR on debt maturity is independent of the level of firm leverage. The CSR coefficients are significant at the 1 % level in all these models.

We use four matching techniques: one-to-one matching without replacement, nearest neighbor matching with n = 20, nearest neighbor matching with n = 40, and finally a Gaussian Kernel matching.

For example, Demirgüç-Kunt and Maksimovic (1999), Scherr and Hulbert (2001), and Fan et al. (2012) report that a relevant proxy of debt maturity is the ratio of debt maturing in more than one year to total debt. Ozkan (2000) and Datta et al. (2005) use the percentage of debt maturing in more than five years.

In the regression framework provided by Gatchev et al. (2009, 2010), the system includes six equations and takes into account five different sources of funds, namely, the changes excess cash, short-term debt issues, long-term debt issues, equity issues, and share repurchases. In our context, we consider it more relevant to focus only on short-term debt issues, long-term debt issues, and equity issues, as these are the three sources of funds that matter most in our context.

The restrictions we use are as follows: (1) β₁₁ + β₁₂ + β₁₃ = 0, (2) β₂₁ + β₂₂ + β₂₃ = 1, (3) β₄₁ + β₄₂ + β₄₃ = 1, (4) β₃₁ + β₃₂ + β₃₃ = 1.