Why Were Changing Our Performance Benchmarks

A s a longtime observer of portfolio manager performance, I have noticed a few common warning signs that there might be trouble brewing with a money manager. They are, in no real order of importance:

1. When there is a major change in the management team of a fund

2. When a specific time frame of historical portfolio performance is no longer reported by the firm

3. When a firm changes the benchmark for its performance reporting

I will discuss numbers one and two at another time. Today I want to discuss changing portfolio benchmarks because for the first time since we began actively managing money, we are making a major benchmark change at Pinnacle Advisory Group.

As I have written in this space more than once, choosing a benchmark for a ‘go anywhere’ globally invested portfolio is an exercise in subjectivity; there is no agreed upon benchmark for such a portfolio. An advisor or consumer could make the case that almost any institutionally available asset class should be included in the benchmark, and their case would be as good as any other advisor choosing a different mix of institutionally available asset classes. Therefore we see investment companies with simple two or three asset-class benchmarks and we see companies with a dozen asset classes in the benchmark. (You could make a good case for more or fewer asset-classes, as well.) All of which brings us to the interesting story of how we made our benchmark choices at Pinnacle Advisory Group.

More than a decade ago I met with Michael Kitces, our Chief of Financial Planning Research, and asked him how we could construct an investment methodology that would allow us to own any asset class for our clients (as long as it had an excellent value proposition) while at the same time systematically managing risk in a way they would understand. Historically investment companies used benchmarks to provide risk management by “buying and holding” the asset classes in the same ratio found in the relevant benchmark. Advisors would rationally point to the historical volatility of the benchmark as a guide for explaining the likely future volatility of the portfolio. We decided to decouple what we were allowed to own in client portfolios from the asset classes in the benchmark, therefore allowing us to tactically and actively manage the accounts. However, we would still use the historical benchmark volatility, and current benchmark volatility, as a guide for managing risk. We decided on a very simple two asset class benchmark for our purposes, which turned out to be the S&P 500 Index and the Barclay’s Aggregate Bond Index (then called the Lehman Aggregate Bond Index.) At that time the firm did little relative return reporting for clients, so we made the easy decision to use the same two asset class benchmark for performance reporting and for volatility comparisons. And for eight years the system worked. No harm, no foul.

However, in the past two years our simple two asset class benchmark has become something of a concern. First, as a matter of sound portfolio policy, we insist on managing a diversified portfolio consisting of many asset classes. Diversification – both among asset classes, and within asset classes remains one of our two major rules for investing (the second is to avoid owning overvalued asset classes). Unfortunately, when the S&P 500 Index is the winning asset class, there is little chance to beat our two asset class benchmark without abandoning diversification. For most of our history as active managers, there were a variety of risk asset classes, including international stocks, emerging market stocks, and a variety of commodity or commodity-like (gold) asset classes that outperformed. But recently the U.S. stock market has rocketed ahead of the other asset classes because U.S. Central Bankers are even crazier than international Central Bankers… and investors have rewarded them with inflated U.S. stock prices. The result is that at times we have delivered poor relative performance due to our benchmark choice. This doesn’t mean that we would have necessarily outperformed a different benchmark over the same period of time as we have often acknowledged, we have been cautious during this bull market. Nevertheless, delivering negative relative performance compared to a benchmark that wasn’t chosen for performance purposes has been vexing, to say the least.

The challenge has been compounded by our growing institutional investment business; our institutional clients have been asking why we use a two asset class benchmark when we run portfolios that are more diversified. The issue came to a head during a recent GIPS (Global Investment Performance Standards) audit, when our reviewer pointed out that according to GIPS guidelines our benchmark “should best reflect how we actually invest client portfolios.” Our two asset class benchmark fails this test.

Moving the Goal Posts?

While we agreed that it was time to change the benchmarks, we faced the issue of what such a change would do to our historical relative performance. Broadly speaking, we had two choices:

• We could keep our current benchmark for historical performance reporting and change to a new benchmark beginning at the current date
• We could change the benchmark retroactively for all reporting from inception

Somewhat counter intuitively, we were hoping the change would not have a significant impact on our relative performance historically if it did, the conspiracy-minded might wonder if we were simply changing benchmarks to make ourselves look better. As it turned out, switching to our new benchmark did affect some of our relative performance, though most changes were insignificant. We examined our performance from inception, during the bull market of 2002 to 2007, during the bear market of 2007 to 2009, and the current bull market from 2009 to present; we also checked the trailing one-year, three-year, five-year, and ten-year performance. Our study revealed that the only time period with a dramatic change was the trailing five-year period, which showed a substantial improvement. Generally speaking, the from inception performance showed no change relative to the new benchmark, the 2002 to 2007 performance was slightly worse, while the 2007 to 2009 and 2009 to current performance was slightly better.

The fact is, our new benchmark which now includes international stocks and commodities, as well as a small percentage of cash better reflects how we actually invest our client portfolios. We hope this change makes it easier for clients to evaluate our investment performance, and that it incentivizes our analysts to continue to create diversified portfolios to best manage risk for our clients. Our new benchmark weightings are shown in the chart below.