Benchmarking: Simple Concept, Difficult Execution

Most people who have been around investing for even a brief period understand the concept of a benchmark- comparing your portfolio results to the results of a “control” group of assets. The idea is to measure performance against a neutral, broad based basket of relevant assets. Naturally, many gravitate towards making this comparison using some of the biggest and most commonly observed indices such as the S&P 500, the Bloomberg Barclays Aggregate Bond Index or the MSCI All Country World Index. For a variety of reasons, the task of creating appropriate benchmarks and even calculating their returns is more difficult than it appears and we wanted to spend some time explaining why those complexities exist and how we view benchmarks.

Why are benchmarks difficult to create?

Most simplistically, the task of benchmarking is asking the question: “If my assets were completely diversified across the investable universe, what would my results have been?” This is an impossible question to answer. Consider all the different assets you would have to value to answer this question. Sure, it’s easy to value stocks on a daily basis. But what about art, classic cars, or raw land? Even valuing financial assets can be difficult. When volatility spiked in March 2020, the liquidity in the bond market dried up so much that certain bonds weren’t trading hands at all. How do you know the price when nothing is moving?

Logically, we limit our benchmarks to assets that can be valued regularly. Most of the time, this means publicly traded financial assets. However, even in the universe of financial securities, there are some pretty esoteric assets out there. This leads to an even more winnowed down list of securities that are easily valued and easily investable.

The bond market is less liquid than the stock market and contains far more securities. Not all these securities trade each day. Furthermore, unlike stocks, which trade on stock exchanges, most bond trading takes place via over-the-counter (OTC) transactions. OTC transactions occur when two individuals make a trade directly- bypassing exchanges. This means in many cases, the prices at which these trades are occurring are not being reported. This is another reason it can be difficult to track bond prices broadly.


Bond indices also suffer from a few problems not present in stock indices. The first is the so-called “Bums Problem.” Essentially, riskier credit issuers tend to issue more credit. For example, if a company is really struggling to generate revenue, they may need to issue more bonds to stay afloat. Because bond indices (like stock indices) weight returns according to market capitalization, more weight is thereby given to the bonds of riskier companies because they have issued more debt. The second problem is that bond indices will often include bonds of all durations. While companies (and their shares) are expected to exist forever, their bonds are not- they eventually mature. The issue for investors is that the longer the period of time to maturity, the more volatile the price of the bond will be over time. This means you may be comparing your performance to a basket of bonds that you would never consider owning due to their volatility. The catch is that for indices with no duration constraints, the index provider still has to include their performance to stay “neutral.”

In the world of publicly traded equities, there is ample information on the price of shares and the number of shares outstanding. This means it is relatively easy to calculate the market cap of each publicly traded company. The fixed income market, which includes everything from treasury bonds to commercial paper, is more opaque. While indices for different types of fixed income exist, it’s difficult to find good benchmarks that encompass multiple asset classes within the fixed income market. For example, the most well-known bond index, the Bloomberg Barclays US Aggregate index tracks the performance of the entire US investment-grade bond market, but does not include lower quality debt. For this reason, especially in the fixed income world, some benchmarks are created based on the allocation other investors are making to the asset class in similar portfolios, rather than the market cap of the assets.

Less theoretically, there is an additional reason it can be difficult to create the perfect benchmark. Whether you’re looking at stock or fixed income indices, a company owns the rights to the index and its data. This index then needs to be licensed from the index provider for use in benchmarks- and it’s not cheap. This data can cost hundreds of thousands of dollars per year for a portfolio performance software provider to license for use by financial advisors. As such, these software providers limit the number of indices available for use by their advisor clients.

Why is benchmark performance difficult to calculate?

Intuitively, most people would assume that calculating the performance of the benchmark is as simple as looking at the beginning and ending value of each component, then weighting that performance appropriately, but this leads to some problems. Choosing different components for inclusion in a benchmark implicitly means that the different components will be generating different returns. This means that the weight of each component will “drift” over time. Let’s consider a very simple example of a benchmark that consists of 60% S&P 500 (SPX) and 40% Bloomberg Barclays US Aggregate Index (AGG). Let’s assume that in Q1, the SPX return is 20% and the AGG returns -5%. This would cause the weight of the SPX to increase to 65% and the weight of the AGG to fall to 35%, thereby drifting away from our original allocation over time.

To remedy this, the benchmark needs to be “rebalanced” to maintain the intended allocation. The rebalanced performance numbers then need to be strung together (chained) over time to calculate the cumulative return of the benchmark. However, the frequency at which the rebalancing occurs will affect the calculated return over time. If rebalancing occurs more frequently, the benchmark will “capture” more returns when a component of an index is falling relative to another (the benchmark calculation will be more negative than the straight line returns of the indices). Likewise, longer periods between rebalancing will lead to the benchmark capturing more upside when components rising relative to others and less downside when they are falling.

The difference is starkest during times of heighted volatility. Let’s look at two periods with major market moves to illustrate this point. One on the downside (Q1 2020) and one with strong gains (Q1 2019). For this example, we’ll again use our very simple benchmark detailed above (60% SPX, 40% AGG).

Hilton Head Investment Adivor

As you can see, different return periods can dramatically impact the calculated returns if you only look at a single point in time. In many cases, the returns are close and ultimately get you to (almost) the same place over longer periods of time. However, there are some large divergences present- although it should be noted that some of these “divergences” are only observable after the fact.

Bluffton Investment Advisor

The Q1 2020 data shows an even sharper contrast, with daily results “outperforming” weekly results for essentially the entire month of February and then again in the last couple weeks of March. In fact, the difference between daily and weekly calculation would have resulted in over 1 percentage point of difference at the end of the quarter. This may not sound like much, but it represents a 9.4% difference! This effect is entirely due to the fact that one return period allows the allocation to drift more than the other.

There is no right or wrong answer, but the return period of the benchmark should always match the return period for the portfolio. In our firm, we use daily benchmark performance data and daily portfolio return data.

What is the takeaway?

We have established that benchmarks: 1) Are difficult to create, 2) May contain assets that are inappropriate for you, and 3) Will have different results depending on the return period. So how should benchmarks be viewed?

To be clear, it is important to compare your investment results to a benchmark. In fact, we make a commitment to all clients that we will provide the performance data of their portfolios and the performance of the benchmark no less than quarterly. Performance that deviates significantly from the benchmark is a concern. However, we do not recommend anyone become overly fixated on or concerned with performance that exceeds or lags a benchmark by a percent or less in one particular quarter. Pay more attention to longer term trends. It’s important to remember that it is just a snapshot in time and the difference could be entirely due to the return period of the benchmark.

As a planning-based firm, we try to remind clients that the goal for investment performance is the minimum return required to meet their goals. For clients with a large amount of assets and few goals, the focus is typically more on asset preservation than growth. For younger clients who have a long period of time until retirement, the focus is more often on long term appreciation. Regardless, you should know the long-term rate of growth you need and choose the appropriate asset allocation- this should ultimately be your measuring stick. The only way to get this answer is through detailed financial planning.