Aurora Borealis
1 November 2016 CFA Institute Journal Review

Multi-Level Geometric Attribution, Revamped (Digest Summary)

  1. Biharilal Deora, CFA, CIPM

In an effort to make reporting more meaningful, flexible, and intuitive, the author develops a framework for the practical implementation of true geometric attribution for complex multi-manager funds with multiple currencies through a smoothing mechanism.

What’s Inside?

The standard Brinson and Fachler (BF) geometric attribution does not show intuitive results for situations with multi-level fund structures (i.e., multi-manager funds and multiple currencies). The author develops a methodology by which each attribution effect can be allocated directly to the management decision and various effects can be grouped against different levels, effect types, regions, and time periods.

How Is This Research Useful to Practitioners?

Performing standard BF attribution for a multi-level portfolio should ideally face two issues. First, because there would be a difference between the portfolio’s strategic asset mix and the index mix, the portfolio manager would be incorrectly penalized—even if he had made a good allocation decision (e.g., to underweight a sector that has underperformed)—because the allocation effect would be amplified with the first-level asset mix.

Second, it would become difficult to isolate currency impact on a fixed-income portfolio with the overall portfolio’s currency return being non-zero. Given the number of variables involved, including the different levels of the decision (i.e., top asset mix and individual portfolio mix coupled with currency attribution methodology and smoothing mechanisms), this approach can serve as a useful tool for multi-period attribution or for attribution for a more complex grouping structure based on time series.

How Did the Author Conduct This Research?

The author has modeled a situation in which a balanced fund represents a two-level structure with all independent investment decisions. At the first level, an investment committee decides the strategic asset mix for cash, fixed income, and equity; at the second level, investment managers are set to beat their own benchmarks with defined weights to regions or types of instruments.

Performing standard BF attribution will not give ideal results, with currency impact as an add-on to the portfolio. In order to measure the same, the author uses naive currency attribution and breaks down the local currency effect by each asset class and region and also measures the impact of asset class attribution. Although the total level allocation and selection effect are geometrically linked into the value added, the individual sector/asset class requires an arithmetic roll. To achieve this, the author uses a smoothing factor from the parent level down (i.e., strategic asset allocation) and prorates it with the current level smoothing (i.e., portfolio level) to distribute it among allocation and selection effects and thus showcase the true performance picture to the end user.

Abstractor’s Viewpoint

The investment industry is undergoing a big debate on active versus passive management. With global funds increasingly available to investors, it is important to identify if the fund performance is based on the manager’s style or skill or on luck or market factors. Investors often cannot understand the difference based on comparison with benchmarks alone. The author’s methodology helps investors obtain a realistic picture of performance and helps provide a transparent framework for investment manager selection and remuneration.

We’re using cookies, but you can turn them off in Privacy Settings.  Otherwise, you are agreeing to our use of cookies.  Accepting cookies does not mean that we are collecting personal data. Learn more in our Privacy Policy.