Does Corporate Social Responsibility (CSR) Improve Credit Ratings? Evidence from Geographic Identification

• Published date: 01 September 2014
• Author: Pornsit Jiraporn,Adisak Boeprasert,Kiyoung Chang,Napatsorn Jiraporn
• DOI: http://doi.org/10.1111/fima.12044
• Date: 01 September 2014

Does Corporate Social Responsibility

(CSR) Improve Credit Ratings?

Evidence from Geographic Identification

Pornsit Jiraporn, Napatsorn Jiraporn, Adisak Boeprasert,

and Kiyoung Chang∗

We show that a firm’s CSR policy is significantly influenced by the CSR policies of firms in the

same three-digit zip code, an effect possibly due to investor clienteles, local competition, and/or

social interactions. We then exploit the variation in CSR across the zip codes to estimate the effect

of CSR on credit ratings under the assumption that zip code assignments are exogenous. Wef ind

that more socially responsiblefirms enjoy more favorable credit ratings.In particular, an increase

in CSR by one standard deviation improvesthe f irm’s creditrating by as much as 4.5%.

We find that a firm’s corporate social responsibility (CSR) policy is significantly influenced

by the CSR policies of firms in the same three-digit zip code, an effect possibly due to investor

clientele, local competition, and/or social interactions. Wethen exploit the variation in CSR across

the zip codes to estimate the effect of CSR on credit ratings under the assumption that zip code

assignments are exogenous. We find that more socially responsible firms enjoy more favorable

credit ratings. In particular, an increase in CSR by one standard deviation improves the firm’s

credit rating by as much as 4.5%.

Although a tremendous volume of research has been conducted regarding the issue of corporate

social responsibility (CSR), its effects on corporations are still not completely understood. One

difficulty is identifying exogenous shocks to CSR policies. Due to market segmentation, investor

clientele, social interactions, and/or local competition, we hypothesize that firms located in close

proximity to one another will have similar CSR policies. Exploiting the variation in CSR policies

across geographic locations, we then estimate the effect of CSR on credit ratings since ratings

significantly influence the cost of debt for the firm. Firms with f avorable credit ratings have better

access to capital markets and can borrow at a much lower cost. Additionally, strong credit ratings

also inspire investor confidence.

Our proximity measure is based on zip codes. The US Postal Service (USPS) allocates zip

codes based on efficiency in mail delivery. Zip code changes are also rare and usually reflect

changes in macroeconomic factors such as demographics and urban development. Thus, zip

code assignments are unlikely related to corporate policies or outcomes and can be considered

exogenous.

The idea of corporate isomorphism and peer pressure is hardly new. Corporations that invest

in CSR have strong incentives to publicize their CSR activities and make them as visible as they

∗Pornsit Jiraporn is an Associate Professor of Finance in the School of Graduate Professional Studies at Pennsylvania

State University in Malvern, PAand a Visiting Associate Professor of Finance at The National Institute of Development

Administration (NIDA) and SASIN Graduate Institute of Business Administration (GIBA), Chulalongkorn University in

Bangkok, Thailand. Napatsorn Jiraporn is an Assistant Professorof Marketing at State University of New York(SUNY) at

Oswego in Oswego,NY. Ardisak Boeprasert is a PhD candidate at the National Institute of Development Administration

(NIDA), Bangkok, Thailand and The Petroleum Authority of Thailand in Bangkok, Thailand. Kiyoung Chang is an

Associate Professor of Financeat the University of South Florida-Sarasota-Manatee in Sarasota, FL 34243.

Financial Management •Fall 2014 •pages 505 - 531

506 Financial Management rFall 2014

can asocial spending is “akin to advertising” (Dorfman and Steiner, 1954; Milgrom and Roberts,

1986; Navarro, 1988; Webb and Farmer, 1996; Sen and Bhattacharya, 2001). Therefore, CSR

activities are highly observable. We argue that the observable nature of CSR makes it likely

that firms are influenced by their geographic peers when formulating their own CSR policy.

Our results strongly confirm this argument. In particular, the empirical evidence indicates that

the degree of CSR of a given firm is significantly influenced by the average degree of CSR

of the geographically proximate firms. We find that firms situated in the same three-digit zip

code exhibit similar CSR policies. The influence of the surrounding firms on CSR remains

significant even after controlling for a number of firm characteristics, such as f irm size, leverage,

profitability, research and development (R&D) spending, capital expenditures, and advertising,

as well as possible variations over time and across industries.

We explore the impact of CSR on credit ratings using an identification strategy based on the

geographic similarity in CSR policies. Part of the firm’s CSR can be attributed to its neighbors.

Therefore, it comes from outside the firm and can be considered exogenous. It is this variation

in CSR across geographical locations that we exploit in our empirical strategy. Using the average

CSR level of the surrounding firms as an instrumental variable, we estimate a two-stage least

squares (2SLS) analysis and demonstrate that a higher degree of CSR leads to better credit ratings.

Therefore, CSR is recognized and viewed positively by credit rating agencies. To ascertain the

impact of the endogeneity bias, we compare the ordinary least square (OLS) and the 2SLS

estimates and find that the difference between the two estimates is quite large. In particular,

without properly accounting for endogeneity, the effect of CSR on credit ratings would be

underestimated by approximately 45%.

We are also aware of a possible endogeneity bias that can be attributed to unobservable firm

characteristics. Certain firm attributes not included in the model may influence both CSR and

credit ratings, possibly leading to a spurious correlation between CSR and credit ratings. To ensure

that our results are not confounded by the omitted variable bias, we execute a two-stage fixed

effects analysis, which controls for both possible reverse causality and for the omitted variable

bias. We obtain consistent results. The impact of CSR on credit ratings is not only statistically

significant, but is also economically meaningful. In particular, an increase in CSR by one standard

deviation improves the firm’s credit rating by as much as 4.5%.1

The literature on the effect of CSR on corporate outcomes is divided. While the more recent

evidence appears to demonstrate the beneficial effects of CSR (Orlitzky, Schmidt, and Rynes,

2003; Saiia, Carroll, and Buchholtz, 2003; Godfrey,2005; Brammer and Millington, 2008; Benson

and Davidson, 2010), a large number of other studies find either negative or insignificant results

(Aupperle, Carroll, and Hatfield, 1985; McGuire, Sundgren, and Schneeweis, 1988; Wright

and Ferris, 1997; Teoh, Welch, and Wazzan, 1999). The literature concerning credit ratings has

documented a large number of firm characteristics that influence credit ratings. For example,

default risk is found to be inversely related to credit ratings (Lamy and Thompson, 1988; Ziebart

and Reiter, 1992), whereas firm size is positively related to credit ratings (Horrigan, 1966, Kaplan

and Urwitz, 1979; Boardman and McEnally, 1981; Bhoraj and Sengupta, 2003). Closely related

are recent studies that demonstrate that better corporate governance is associated with better

1The instrumental variable (IV) approach is especially appropriate in this context as the IV estimation addresses the

attenuation bias resulting from mismeasured explanatory variables, which, unless properly addressed,would bias coeffi-

cient estimates toward zero. Since it is difficult to precisely measure CSR, there could be measurement errors. The IV

approach helps mitigates the bias attributable to these possible measurement errors (Miguel, Satyanath, and Sergenti,

2004).