Systemic Risk Driven Portfolio Selection

  • Agostino Capponi, Associate Professor, Dept of Industrial Engineering and Operations Research, Columbia University
  • Alexey Rubtsov, Assistant Professor, Ryerson University, Research Associate, Global Risk Institute
falling domino blocks with hand in the middle stopping them


The balance between risk and return has been at the center of portfolio construction methodologies. Although asset allocation literature has primarily focused on a firm’s individual risk, the Global 2007-2009 financial crisis highlighted the importance of accounting for systemic events, i.e., extreme forms of risk that can have severe consequences on the financial system.

The following is the central thesis studied in this research:

How to construct portfolios that perform well in the face of systemic events.

We developed a portfolio construction methodology based on two well-known risk measures: Value-at-Risk (VaR) and Conditional Value-at-Risk (CoVaR). Furthermore, we derived explicit formulae for the optimal portfolio and explained economic intuition for optimal asset allocation. Our approach is quite flexible and allows the investor to properly balance portfolio risk and portfolio correlation with the financial system when making investment decisions.

The performance of the proposed methodology is assessed on the Canadian and US stock markets. In the US market, we chose the constituents of the S&P500 Financials Index, and for the Canadian market, we chose the constituents of the S&P/TSX Capped Financial Index. The time period was from January 4, 2000 until October 1, 2018, hence covering the period of the Global 2007-2009 financial crisis when noticeable systemic events occurred. Finally, we compared our portfolio criteria with well-known benchmarks. The out-of-sample analysis of the portfolios’ risk-adjusted returns reveals that our optimal portfolio outperforms the benchmarks in times of a market downturn. We also found that the optimal portfolio is less diversified than the benchmark portfolios. These results can be intuitively explained as follows: portfolios that behave well during crisis periods tend to invest in few stocks that have relatively small correlation with the financial system.