Regimes: Nonparametric Identification and Forecasting

Abstract

The practice of categorizing individual time periods with economic “regimes,” such as recession, depression, and expansion, is commonplace and has a strong influence on the return assumptions employed as inputs in asset allocation models. Unfortunately, methodologies used to determine whether a period belongs to a given regime vary in their effectiveness and can be inexact. In this article, the authors offer an alternative nonparametric approach that emphasizes current conditions rather than preset regime characteristics and draws on probabilities to reflect the reality that no two economic periods are identical.The authors also include case studies through which they illustrate and apply their recommendations.