|
Page 1 of 6
 All active managers must exhibit some level of tracking error against their target benchmark; if they didn't, they would be managing index funds (and doing so perfectly). While the level of tracking error differs among active managers, the implications of tracking error on an aggregate portfolio level are considerable.
Mean variance optimization (MVO) is a tool often utilized by investment professionals to determine optimal allocations for client portfolios. Traditionally, indexes have been used to represent different asset classes during the MVO process. Active managers, however, are frequently used to replace the tested index when the portfolio is actually implemented. This creates a potential mismatch problem, since active managers exhibit tracking error, something that is not considered in an MVO process based entirely on passive indexes.
Active investment management introduces tracking error not only at the individual investment level but also at the aggregate portfolio level. Previous research on tracking error has focused primarily on the impacts of style drift (i.e., tracking error) for an individual investment and not the aggregate impact on a portfolio. The purpose of this paper is to provide insight on the impact of introducing active managers during the portfolio construction and optimization process.
Literature Review
There is anything but common consensus when it comes to determining the net value of active management. A number of researchers contend that active management adds little value (see Carhart (1997), Malkiel (1995) or the updated Brinson, Hood and Beebower study from 1995), while opposing research has been published noting its potential benefits (Pastor and Stambaugh (2002), Wermers (2000)). The purpose of this paper is not to weigh in on this great debate, but rather to explore the implications of tracking error on an aggregate portfolio basis.
Tokat, Wicas and Kinniry (2006) provide a follow-up to the original Brinson, Hood and Beebower (1986) study and discuss the impact on variability for actively managed funds versus an appropriate benchmark portfolio. The authors conclude that actively managed balanced mutual funds had lower performance and increased volatility relative to their indexed static policy portfolios (64 percent of actively managed funds underperformed their policy portfolios over three and five years). The authors do note, though, that there were considerable differences in performance among actively managed funds and that active management does create an opportunity for a portfolio to outperform appropriate market benchmarks (assuming the appropriate manager is selected).
Loeper (2003) has also addressed the probability implications of using active management in a white paper titled "Active vs. Passive Management: The Debate Continues." Loeper notes that given the potential uncertainty surrounding active management, which in the aggregate tends to lead to underperformance, investors should accept the return of an index and only incur risk where it cannot be controlled. The issue of uncertainty is perhaps the fundamental issue that must be addressed when deciding active versus passive: whether the potential benefits from active management are worth the increased uncertainty.
That increased uncertainty is, by definition, the tracking error. To generate alpha, an active manager must deviate from his or her target benchmark. While tracking error can differ considerably among portfolio managers, some tracking error is inevitable.
A Word On Mean Variance Optimization
MVO has come a long way since its introduction. Although it was originally conceptualized for use in individual stock portfolios (or assets with dissimilar correlations), it has become an increasingly popular tool among investment professionals with broadly diversified asset categories. Introducing diversified investments (many of which tend to be highly correlated), though, substantially reduces the benefits of MVO, both on a theoretical and practical level (Nawrocki 2000). Also, a variety of constraints typically must be employed during the optimization process to create a portfolio that can actually be implemented for a client. Although Frost and Savarino (1988) have noted potential benefits from including constraints during the optimization process, imposing too many constraints can reduce (or entirely destroy) the benefits of the MVO process.
|
Previous Page
