A Methodology for the Identification and Measurement of Corporate Social Responsibility Constructs with Existing Analyst Data

Primary Author: Michael Craven

Faculty Sponsor: Bernard Wong-On-Wing

 

Primary College/Unit: Carson College of Business

Category: Business, Communication, and Politial Sciences

Campus: Pullman

 

Abstract:

PRINCIPAL TOPIC

Research on Corporate Social Responsibility (CSR) and business outcomes has produced inconsistent results, which may be associated with difficulties in measuring CSR. Prevailing measurement methodologies use existing databases with indicators of strengths and weaknesses on environmental, social, and governance (ESG) issues including MSCI ESG KLD STATS. However, current research assumes that the strengths and weaknesses of each category accurately reflect differentiable constructs, that strengths are equivalently reduced by weaknesses, and that each indicator is equal when measuring those constructs.

 

METHOD

A scale development methodology is created to overcome the limitations of the existing data and to empirically identify and measure CSR constructs from the analyst data. Because indicators are added and removed, the indicators are screened with a co-measurement based variation of Cronbach’s Alpha to ensure sufficient covariance for further analysis. These indicators are then analyzed with logistic regression based exploratory factor analysis to identify the latent CSR constructs represented by the data and confirmatory factor analysis to provide continuous measures from the dichotomous indicators.

 

RESULTS AND IMPLICATIONS

The results indicate that prior measurement models for CSR poorly reflect the underlying aspects of CSR. In most categories, the assumption of differentiable strengths and weaknesses misidentifies the extant constructs. The assumption of equal weighting of indicators does not hold. Work is needed to validate these new measures against know CSR outcomes, such as fines for pollution. These empirically identified CSR constructs and measurements will resolve a long-standing issue and contribute to resolving the inconsistent results of prior CSR research.