This book provides an overview of quantitative portfolio risk analysis, concentrating mainly on primary asset classes such as stocks, bonds, real estate, and foreign exchange. Our approach relies on statistical modeling of asset returns, framed by economic theory and cognizant of the institutional settings of contemporary capital markets.
One key message that emerges from empirical research over the last forty years is that risk regimes change, often suddenly and in unexpected ways. This emphasizes the importance of a solid understanding of the statistical modeling used for risk analysis and risk management. To prepare adequately for sudden shifts, and to adjust quickly to them, a risk analyst needs a deep understanding of the assumptions and vulnerabilities of his modeling approach.
We stress the need for a multidisciplinary perspective. There is no single framework for portfolio risk analysis that works in all situations. A risk analyst must be able to take a purely statistical modeling approach, a microeconomic or macroeconomic perspective, or an institutional– behavioral perspective, carefully balancing the contributions of each of these points of view, in order to properly analyze portfolio risk in the ever-changing environment of global capital markets.
Two competing yardsticks in portfolio risk analysis are return variance and worst-case losses; the latter is usually measured by value-at-risk or expected shortfall. The advantage of variance-based modeling is its statistical reliability and analytical elegance. The drawback is that return variance completely characterizes the portfolio return distribution only in the unrealistic case in which returns follow a normal distribution. More generally, two portfolios with very different loss profiles can have the same return variance. It is imperative to note that although variance– covariance analysis can be very informative about return dispersion and