Mean-Variance Portfolio Analysis of the Locational Value of Generation Assets
West Virginia University Department of Economics
I will address the problem of optimal investment in generation based on mean-variance portfolio analysis. It is assumed the investor can freely create a portfolio of shares in buses on the electrical network. Investors are risk averse, and seek to minimize the variance of the weighted average Locational Marginal Price (LMP) in their portfolio, and to maximize its expected value. I conduct simulations using a commonly used IEEE 68-bus network that approximately represents the New York - New England system and calculate LMPs in accordance with the PJM methodology for a full AC optimal power flow solution. Results indicate that the network topology is a crucial determinant of the investment decision as line congestion makes it difficult to deliver power to certain nodes at system peak load. Determining those nodes is an important task for an investor in generation as well as the transmission system operator.