CSR, Market Value, and Profitability: International Evidence

B. Burcin Yurtoglu is Professor and Chair of Corporate Finance, and Stevan Bajic is a doctoral candidate at WHU Otto Beisheim School of Management. This post is based on a recent paper authored by Professor Yurtoglu and Mr. Bajic. Related research from the Program on Corporate Governance includes Socially Responsible Firms by Alan Ferrell, Hao Liang, and Luc Renneboog (discussed on the Forum here).

Over the last two decades corporate social responsibility (CSR) has become a mainstream business activity. Companies have increasingly integrated environmental, social, and governance (ESG) aspects into their market and non-market strategies (The Economist, 2008). Businesses remained committed to CSR principles even during the financial crisis: a UN Global Compact–Accenture CEO (2010) study shows that 96% of CEOs believe that sustainability issues should be fully integrated into the strategy and operations of a company (up from 72% in 2007) and that 93% of CEOs consider sustainability issues to be critical to their business. Although CSR is receiving an ever growing attention, there is no clear-cut answer to the fundamental question of whether CSR predicts higher firm value and performance (Griffin and Mahon, 1997; Orlitzky et al., 2003; Margolis and Walsh, 2003; Margolis et al., 2009) which impedes sound policy formulation at the corporate level.

CSR comprises a broad range of activities. For some companies CSR involves treating their employees well by providing high-quality employment benefits and job conditions; for others actively requiring their suppliers to meet certain standards, considering the views of external stakeholders in their decision-making process, cutting their emissions, and for some others to create a philanthropic foundation. Most of these CSR activities involve that companies go beyond statutory requirements in providing goods and services with a public character and in internalizing their externalities. The benefits of CSR are expected to accrue through various channels. This starts with an improved brand and reputation (Baron 2001), being more attractive to potential and existing employees, smoother relationships with regulators, and through meeting the standards required by customers. Involvement in some of these issues carries costs which can be direct, as when expenditures for charitable donations or environmental protection increase, and so lower profits. Costs can also be indirect, as when the firm becomes less flexible and operates at lower efficiency (Claessens and Yurtoglu 2013).

Whether the benefits of CSR activities exceed the costs and drive superior performance and shareholder value is an empirical question. To answer it, a large number of studies have operationalized the latent, unobservable concept of CSR with an observable measure of corporate social performance (CSP) and analyzed whether CSP has a material impact on corporate financial performance (CFP). In doing so, a wide variety of CSP constructs have been employed to predict both market and accounting based corporate performance measures. While these studies suggest a positive relationship between CSP and CFP, the size of the effect seems to be small (Margolis et al. 2009) and some studies find either a negative or insignificant relationship, especially in studies which use data from earlier decades (van Beurden and Gössling 2008). Another systematic meta-analysis (Orlitzky et al., 2003) conclude that the relationship between CSP and CFP is more likely to be positive, but point to the large unexplained variance across studies and call for further analysis. These meta-studies also document substantial variety in empirical strategies used to obtain the reported outcomes.

We argue in our paper that observational CSR studies of whether CSP predicts higher performance and shareholder value have to deal with three fundamental problems which pose severe challenges to identification. These are construct validity, limited data, and endogeneity.

Construct validity. CSP is a construct that imperfectly measures the latent unobservable environmental, social and governance aspects of CSR and it is not possible to quantify the gap between the construct and the underlying concept.

Data. Many studies focus on a single country and use a single and often a specific CSP construct which is either not available for other countries (e.g., environmental disclosure or employee treatment) or has a widely varying meaning and importance for different firms in the same country (e.g., hazardous waste reduction). Since different aspects of CSP are often correlated, a firm with a high performance in one of these aspects, e.g., in reducing emissions, is likely to perform well also on other dimensions, e.g., reducing resources. A study which studies the impact of only one of these aspects may document a relationship between the included aspect and a measure of performance, however, the true driver of superior performance can be an omitted aspect of CSR. Unless one uses a broad CSP measure, the omitted aspects of CSP become an omitted variable, and thus a source of omitted variable bias (OVB) for the included aspects.

Endogeneity. As with many other studies in related areas, a third problem stems from the various forms of endogeneity, especially omitted variable bias (OVB). At the firm level, does superior corporate performance foster better corporate social performance (CSP), as the firm can afford it? Or does better CSP cause better performance and shareholder value? CSR activities arise endogenously as firms choose them in response to internal and external constraints they face. The endogeneity of the CSP-CFP relationship creates estimation problems if CSR choices are made on the basis of unobservable characteristics correlated with the error term in the regression models being estimated.

We respond to these three problems in the following way. First, we employ a broad measure of CSR, provided by ASSET4, reflecting firm level choices and activities in dealing with environmental, social and governance issues. Using an exploratory principal component analysis (PCA), we analyze whether the components of this broad measure capture the unobserved component of CSP. This analysis shows that ESG scores contain two sets of elements which are measuring similar factors, one which highlights the environmental and social factors and a second consisting of governance aspects, highlighting the need to control for governance in CSR studies.

Second, our firm-level panel spans over 2003-2012 and is representative of the population of listed companies from 35 countries differing in legal traditions, institutions, regulations, language, culture, and geographic location. To the best of our knowledge, no prior study employs such a comprehensive dataset.

We then assess whether ESG factors affect firm market value (proxied by Tobin’s q) and operating performance (proxied by EBIT/Assets) and how firm fixed effects (FE) or random effects (RE) results differ from OLS results. We first focus on a broad ESG measure. Next we analyze whether the environmental (E), social (S) and governance (G) components which constitute the broad ESG score predict firm value and profitability in isolation from other components. We also conduct extensive sensitivity tests and discuss the extent to which our results depend on the choice of control variables, the functional form of the dependent variable, and also to the omission of covariates.

We report that studies of particular aspects of CSP, such as employee relations, environmental incidents or toxic emissions, face OVB concerns, because different aspects of CSP are often correlated. In our sample, we find an economically significant relationship between the overall CSP measure and firm market value: with firm fixed effects and extensive covariates, a one-standard-deviation increase in overall ESG predicts 9.6% higher Tobin’s q. This result is driven by the social subscore of the composite CSP measure. On the other hand, we observe a much weaker relationship between CSP and profitability. However; an average effect across many countries does not tell us for which countries, and which firms, the aspect matters. We leave this and related questions to future research.

The full paper is available for download here.

Both comments and trackbacks are currently closed.