Integrating ESG criteria in portfolio optimization: A Moroccan case study using Markowitz’s theory and correlation network analysis

Abstract

This paper evaluates the contribution of integrating environmental, social, and governance (ESG) criteria into portfolio optimization, which is an important gap in the literature on socially responsible investing (SRI) for the context of Morocco. It offers new insights on how ESG constraints affect performance and risk in dynamic portfolios, accounting for time-varying correlations of responsible assets. The paper uses a quantitative approach by considering two models: one using only Markowitz’s theory about mean–variance (MV) optimality without considering the ESG criteria, and another model with a constraint that enforces a minimum acceptable threshold in ESG scores, named here as MV-ESG. The minimum spanning tree (MST) method was adopted for correlation network construction, and it would consider studying dynamic correlation structures on two different periods: crisis and post-crisis. The data used are the closing prices of companies listed on the Casablanca stock exchange (CSE), obtained from their official website, while the ESG scores were obtained from Refinitiv. The results show that responsible portfolios (RPs), optimized by traditional and network approaches, are neither more profitable nor less volatile than their conventional counterparts, with the exception of peripheral portfolios ($P^p$), which show more attractive profitability in the responsible case, but at the cost of higher risk. While the network approach has facilitated better opportunities for higher returns, reduced losses, and better ESG scores than the traditional method, the results show that in a social orientation crisis, performance will be unfavorable; it has not been able to provide positive returns during this period.