Defined contribution plan participants are increasingly flocking to target-date funds as a single turnkey solution for diversification in self-directed 401(k) plans. It's easy to see the allure. Through a single investment, the plan participant gains exposure to varied asset classes, multiple mutual funds and an allocation appropriate for their target retirement date. However, while simple for the investor to understand, those fiduciaries who must analyze the performance of target-date funds often face a daunting task of evaluating an expanding universe of funds against others with the same retirement date, but with widely different asset allocations.
The overall performance of target-date funds is driven by two factors — the fund's asset allocation and the performance of the underlying funds that make up the target-date fund. As noted in Brinson, Hood and Beebower's 1986 study, more than 90% of the variance in balanced plan results are due to differences in asset allocation. Setting performance benchmarks for target-date funds becomes a challenge because even slight variations in asset allocations can translate to significant differences in performance. Rather than attempt to understand a target-date fund's performance by analyzing it against other funds with the same retirement date, William Sharpe's returns-based style analysis offers a useful methodology to determine a fund's asset allocation and provide greater clarity of the factors influencing its performance.
One common approach is to establish performance benchmarks by categorizing target-date funds by their stated retirement date. In theory, when looking at funds with the same time horizon, an analyst could use an appropriately balanced benchmark or category average return as a reference point for investment performance. Where this approach becomes flawed, however, is that funds are classified only by their stated time horizon objective and not by the underlying holdings and asset allocations. Comparing two funds that have the same objective but widely differing asset allocations and divergent performance is not a true “apples-to-apples” comparison.
The idea behind returns-based style analysis is that the returns of any investment product can be replicated by using a set of passively managed, investible products, such as index funds or exchange-traded funds. When using RBSA to develop a clear picture of a fund's asset allocation and benchmark its performance, three rules need to be applied. First, the set of indexes or independent variables must be representative and cover the opportunity set of all possible asset classes. Second, the indexes must be exclusive, or non-overlapping, to avoid double counting. Finally, the indexes must have correlations low enough so they are distinguishable from one another.
When these rules are applied correctly, and used with a set of very broad asset class indexes, the vast majority of target fund performance and asset allocation can be explained through RBSA. For target-date funds, an appropriate RBSA would use U.S. stocks, foreign stocks, bonds and cash as the independent variables.
RBSA can be used to infer a fund's asset allocation, as it produces what is known as a “style benchmark” to identify the optimal combination of indexes that tracks the performance of the fund's assets. Unlike a crude system of comparing groups of funds by their retirement dates, the style benchmark defines each target-date fund's asset allocation.
The style benchmark's validity is cross-checked via the R-squared, a statistical measure that represents the percentage of a fund or security's movements that can be explained by movements in a benchmark index. When the R-squared to style benchmark percentage is 90 or higher, we know that the vast majority of performance variability can be explained by the RBSA. In a recent study of 277 target-date funds, 258 funds (or 93%) had an R-squared of 90% or higher. The average R-squared was 96.8%. This analysis concludes that RBSA does a very good job of inferring target-date fund asset allocations.
With the style benchmark established, we can conduct in-depth analysis on the sources of performance. As an example, when looking at two hypothetical fund families (Family A and Family B) across multiple target dates, Family A would appear to significantly outperform Family B in each of the target dates. However, the reason Family A is universally outperforming Family B is unclear. Whether it is due to better asset allocations and weighting of equities and fixed income, or the individual funds used to build the target-date fund adding value within their respective asset classes, the factors behind Family A's higher performance are unknown.
However, after applying RBSA and creating style benchmarks, it becomes clear that Family A has an across-the-board overweight to equities, while Family B is weighted more toward fixed income. In our hypothetical scenario, knowing the stronger performance of equity markets during the period we are looking at directly correlates to the systematic stronger performance of Family A.
The RBSA approach offers a more accurate comparison of target-date funds by categorizing funds by their asset allocation. Knowing that asset allocation is the primary driver of a target-date fund's performance, comparing the performance of a group of funds categorized by their retirement date does not provide a true benchmark against differently allocated funds within the category. Rather than comparing apples to apples, this approach is akin to comparing apples to oranges.
Marc Odo serves as director of applied research with Zephyr Associates Inc.
