Human behavior analysis under financial information science: Evidence from corporate social responsibility

Abstract

Purpose

The purpose of this paper is to investigate the relationships among corporate social responsibility (CSR), analyst forecast accuracy and firms’ earnings management behavior using US-based firms.

Design/methodology/approach

The authors use the Kinder, Lydenberg, Domini (KLD) database to construct CSR performance scores and divide all firms into ten groups from high to low as a proxy for CSR performance. The authors obtained an initial sample of 33,364 firm-year observations from 1991 to 2012. Filtering for records which exist in the KLD, Compustat, and Center for Research in Security Prices databases lefts a total of 16,807 firm-year observations and CSR evaluation reports for 5,896 firms.

Findings

The authors find that high CSR-score firms have lower rates of analyst forecast error than their low CSR-score counterparts, suggesting that CSR performance is a useful means of forecasting earnings. Furthermore, firms with better CSR performance have significantly lower accrual-based earnings management behavior. However, the level of the manipulation behavior of real earnings management (REM) activities increased significantly in better CSR firms, suggesting that high CSR-score firms substituted REM methods for accrual-based methods. REM methods are consistent with the stipulations of the Sarbanes-Oxley Act and allow high CSR-score firms to better manipulate earnings behavior. These results hold after the authors control for various factors related to firm financial characteristics.

Originality/value

Overall, the findings have important implications for investors and regulators to more easily assess firms’ earnings manipulation behavior and earnings stability under CSR performance and financial information in financial markets.

Citation

Hsu, F.J. and Chen, Y.-C. (2018), "Human behavior analysis under financial information science: Evidence from corporate social responsibility", Library Hi Tech, Vol. 36 No. 4, pp. 685-704. https://doi.org/10.1108/LHT-11-2016-0130