Estimation of portfolio efficiency considering social responsibility: evidence from the multi-horizon diversification DEA

¹ School of Business Administration, Hunan University, Changsha 410082, P.R. China
² School of Business, Hunan Normal University, Changsha 410081, P.R. China

^* Corresponding author: xiaohelu1986@163.com

Received: 10 May 2020
Accepted: 18 February 2021

Abstract

With the introduction of the concept of social responsibility investment/green investment, more and more investors have realized the importance of such investment, which has prompted portfolio managers to more comprehensively consider both financial and non-financial performance of portfolios in different time horizons. DEA (Data Envelopment Analysis), as a data-driven evaluation approach, has been widely used in performance evaluation of portfolios. However, the existing studies are mostly limited to single-horizon problems, and the evaluation indicators are mostly financial indicators, while ignoring the impact of non-financial indicators (e.g., social responsibility indicators). More importantly, the input-output process of portfolios in the multi-horizon framework also needs to be clarified. In this paper, we first define the input-output process of portfolios from the multi-horizon perspective, and then propose the corresponding stochastic output possibility sets based on portfolio returns and social responsibility indicators. We use the expectation and variance measures to derive the deterministic estimation of the above stochastic sets, where the expectations and variances of portfolio returns and social responsibility indicators are all regarded as outputs. We construct the multi-horizon diversification DEA models both with and without social responsibility constraints. Finally, we select the 20 component stocks of China ESG100 index to illustrate the difference between the multi-horizon models and the single-horizon models, and further discuss the impact of social responsibility on the portfolio efficiency and its ranking. The empirical results show that compared with the single-horizon models, the proposed models can provide portfolio managers with an improvement strategy to balance the performance of portfolio returns and social responsibility indicators in different time horizons. Further, we also find that the social responsibility has a greater impact on the portfolio efficiency and its ranking, especially when the portfolio managers pay more attention to the social responsibility performance.