This paper measures the trade-off between financial efficiency and SRI in the traditional mean-variance optimization. We compare the optimal portfolios of an SR-insensitive investor and her SR-sensitive counterpart in order to assess the cost associated with SRI. Our contribution is twofold. First, we extend the Markowitz (1952) model4 by imposing an SR threshold. This leads to four possible SR-efficient frontiers: a) the SR-frontier is the same as
the non-SR frontier (i.e. no cost), b) only the left portion is penalized (i.e. a cost for high-risk-aversion investors only), c) only the right portion is penalized (i.e. a cost for low-risk aversion investors only), and d) the full frontier is penalized (i.e. a cost for all investors). Despite its crucial importance, practitioners tend to leave the investor’s risk aversion out of the SRI story. Our paper on the other hand offers a fully operational mean-variance framework for SR portfolio management, a framework that can be used for all asset classes (stocks, bonds, commodities, mutual funds, etc.). It makes explicit the consequences of any given SR threshold on the determination of the optimal portfolio. To illustrate this, we complement our theoretical approach by an empirical application to emerging bond portfolios. The rest of the paper is organized as follows. Section 2 proposes the theoretical framework for
the SR mean-variance optimization in the presence of risky assets only. Section 3 adds a riskfree asset. Section 4 applies the SRI methodology to emerging sovereign bond portfolios. Section 5 concludes.
Download this article in PDF format