Friday, August 19, 2011

Should we blend?

A client alerted me about a new Q&A that is on the GIPS(R) (Global Investment Performance Standards) website. I'll save you the trouble of looking it up, and provide the particulars here. First, the question: 

"For one of our composites, we use a portfolio-weighted custom benchmark that is created monthly using the benchmarks of the individual portfolios in the composite. The GIPS standards require that if a firm changes the benchmark, the firm must disclose the date of, description of, and reason for the change. Given the nature of the benchmark, the benchmark is subject to change each month. What must be disclosed to satisfy this requirement?"

The response is interesting: "The GIPS standards require that if a custom benchmark or combination of multiple benchmarks is used, the firm must disclose the benchmark components, weights, and rebalancing process. In this example, the benchmark may change every month as part of the normal procedure. It is required in this instance to disclose that the benchmark is rebalanced monthly using the weighted average returns of the benchmarks of all of the portfolios included in the composite. A firm is not required to disclose how the underlying portfolio benchmarks and weights have changed each month. If the benchmark for the composite were to change from a portfolio-weighted custom benchmark created monthly using the benchmarks of the individual portfolios in the composite to a market index, this would be a benchmark change that must be disclosed

"In the spirit of full disclosure and fair representation, firms must disclose the components that comprise the portfolio-weighted custom benchmark, including the weights that each component represents, as of the most recent annual period end. Firms should also offer to provide this information for prior periods upon request.

"Sample disclosure: The Long US Government/Credit Custom Benchmark is calculated using the benchmarks of portfolios in the Composite. The benchmark is rebalanced monthly based on the beginning values of portfolios included in the composite. As of December 31, 2009, the breakdown of the benchmark is 88.2% Barclays Capital US Long Government/Credit Index and 11.8% Barclays Capital US Long Government/Credit A+ Index. The breakdown of the custom benchmark for different time periods is available upon request."

Our client, and we as well, found this approach to benchmarking quite unusual. Usually when we think of a custom benchmark it is constructed to align with a strategy for which a single benchmark doesn't work (e.g., a balanced approach, whereby we'd find an equity and fixed income component). Here, the manager has a composite where the underlying portfolios all align with different benchmarks. I would respond to the question with a series of questions.

Does the manager actually manage differently for each of the client benchmarks, or does the manager manage against a common benchmark? If it's a common benchmark, then it should be the one that's used. If the manager manages differently for each client, in accordance with their respective benchmark, then why aren't they using different composites, since benchmark is an acceptable criteria for composite construction?

If the manager feels that the differences between the portfolios is not material, then I would advise them to pick one of the benchmarks and use that. What value is a benchmark that reflects a dynamic mix of underlying benchmarks, each representing the strategy for of a different client? And asset-weighting them means that the larger account's benchmark rules, yes? But why? And don't forget, the firm can provide additional benchmarks for reference purposes, too!

I think this question calls into the issue of the role of the composite benchmark: is it not to provide a basis to compare the manager's successful implementation of their strategy? If the manager is managing across multiple benchmarks, they may have success with one and failure with another: combining them means what, then?

This is an interesting topic, I think, and one worthy of some reflection, too. Your thoughts are welcome.

2 comments:

  1. Stephen Campisi, CFAAugust 19, 2011 at 5:59 PM

    I want to make sure that I understand your client's proposed scenario: Is the client saying that the blended benchmark is adjusted each month by changing either the weightings of the individual benchmarks that make up the blend, or the individual benchmarks themselves or both? As a simple example, one month the blended benchmark may be 55% Russell 3000 + 35% EAFE + 10% Barclays Aggregate while the next month the blended benchmark may be 60% Russell 3000 + 40% Barclays Aggregate. Occasionally, one hears of clients wanting to change their blended benchmark to match their tactical asset allocation view and so they align the benchmark to these market exposures. In doing so, they eliminate the structural differences between the benchmark and the portfolio, thereby eliminating their allocation effect and are left with only their selection impact. This seems to be a rather useless exercise.

    IF this is what the question addresses, then there may need for guidance on the definition of a benchmark, noting that it is a strategic, long-term measure of the performance from a passive market exposure and therefore the "opportunity cost" of investing with an active manager.

    Let's also encourage performance analysts to see blended benchmarks as a standard yardstick rather than an exception. While GIPS is organized around helping a firm's marketing function by presenting the performance for the manager of a single investment product (usually a single fund) this is not the sum of all investment activity. In fact, this is only a small slice of what's really going on. After all, investors hold diversified portfolios of publicly available stocks and bonds, along with real estate, commodities, hedge funds, private equity and other "alternative" investments. A "plan sponsor" type of overall portfolio manager is responsible for the management of the client's total portfolio of assets and this "manager of managers" is making the tactical asset allocation calls as well as selecting the fund-level product managers who execute the strategy within each particular mandate. The value of the tactical calls as well as the value of the active implementation are components of the return on the overall portfolio. The proper measurement and evaluation in this context is a constant exposure blended benchmark. That said, this idea of a "tactical benchmark" is simply misguided and to be avoided.

    ReplyDelete
  2. Steve, thanks for your question and opportunity to elucidate further, as it's clear that it hasn't been {clear, that is!).

    First, it's not our client who posed the question; our client pointed the Q&A out to us.

    Second, "if life could only be so simple" (as you describe). I agree with what you suggest, but that's not what the questioner asked: it's a case where, for example, where we have a bond strategy, where three different bond indexes are used by three different clients (one per). The manager BLENDS these three client-individualized indexes to form one (i.e., the blending of three bond indexes).

    ReplyDelete

Note: Only a member of this blog may post a comment.