With the displacement of the single, balanced pension fund manager by multiple specialty managers, control of total portfolio diversification has passed largely into the hands of the plan sponsor. The sponsor’s allocation of plan assets to managers having different investment “styles” can have significant impact on portfolio risks and returns. For example, the ability to rotate among different managers in order to take advantage of the market’s cyclical tendency to reward first one investment style, then another, can significantly improve total portfolio performance; conversely, employing several managers whose styles are subject to similar extramarket risks can amplify the plan’s risk of investment losses.
To differentiate between manager styles, sponsors have historically relied on subjective judgment and, more recently, complicated quantitative techniques such as fundamental factor models. A simple, yet objective, method for categorizing managers may be found in cluster analysis—a technique that has proved effective in grouping common stocks according to shared exposures to different types of extramarket risk. Tests indicate that cluster analysis, using correlation coefficients between performance histories to group managers, can accurately identify and differentiate between managers having different investment styles.