Why I Don't Like the Callan Periodic Table of Investment Returns

You might have seen the Callan Periodic Table of Investment Returns before. It displays the relative performance of 9 asset classes in each year, typically over a 20 year period of time.

Some believe this table to be useful in that they believe that it demonstrates the value of diversification and that "past performance does not predict future performance" (e.g., Bogleheads' Wiki).

However, I don't like the Callan table at all for the reasons outlined below.

1. The Callan table eliminates valuable information.

The ordering of the table condenses ratio data, which carry the most information of any type of data, to ordinal data, which conveys far less information.

There are four levels of measurement with all data: nominal, ordinal, interval, and ratio.

Nominal data only uses numbers as labels, like telephone numbers and players' numbers in sports. Letters could just as easily be used.

Ordinal data are ranked in some way. The order in which racers finish is an example.

Interval data are measured using a scale where the points on the scale are placed at an equal distance from one another (e.g., the distance between '3' and '4' is the same as the distance between '4' and '5'). Temperature scales like Fahrenheit and Celsius use an interval scale.

Ratio data contain all of the properties of interval data but also have a zero point that indicates the complete absence of what's being measured. Measures of weight and time use ratio data.

Now let's move to the problem of converting ratio data to ordinal data using the example of college football teams, where the top 25 are ranked. If these rankings are accurate, it means that the team ranked 19th is better than the 20th ranked player, for instance. But it doesn't tell us how much better the 19th ranked team is than the 20th ranked team (or the 21st, 22nd, etc.). It could be that there is a big gap in the quality of the 19th ranked team and the 20th ranked team but only a tiny gap between the 18th and 19th ranked teams. This is the sort of thing that sports fans argue about ad nauseum.

Using a different analogy, if two horses in a race finish within .1 second of each other, but the third horse finishes 3 seconds later, using only the order in which the horses finished the race (i.e., ordinal data) obscures this very meaningful difference in relative performance that ratio data (i.e., the time it took each horse to finish the race) make very clear.

The Callan table takes ratio data (i.e., returns from various asset classes) and ranks them in order of lowest to highest returns by calendar year, thereby converting them to ordinal data. If we see that one asset class outperformed another in a given year, the color coding scheme doesn't tell us how much better its outperformance was.

Why would an investor want to lose valuable

2. The Callan table can easily lead investors to the wrong conclusions.

An investor looking at the Callan table could easily conclude that the seemingly random year-to-year ranked returns of the asset classes means that they have all had similar cumulative returns over the long-term. But this is actually 100% false.

For instance, the Callan table above shows that ex-U.S. stocks in developed nations were in the top 4 ranked asset classes in 11 of the 20 years displayed, which looks good. By comparison, large-cap (i.e., large company) U.S. stocks were only in the top 4 ranked asset classes in 10 of the 20 years. This could easily lead someone to believe that ex-U.S. stocks outperformed U.S. large-cap stocks, but that is incorrect. Average annualized real returns of ex-U.S. stocks were 4.25%, while U.S. large-cap stocks returned 6.93%. Such important distinctions are completely lost when we're only looking at ordinal data for individual years.

3. The Callan table can encourage a short-term mindset.

The Callan table displays the ranked returns of asset classes over individual years. But no one should be investing in volatile asset classes like stocks, real estate, or high yield bonds for only a year! Among buy-and-hold investors, one likely needs to have at least a 5 year and possibly a 10 year investment horizon for owning such asset classes to be prudent because they can easily lose significant value over shorter periods.

As long-time financial educator Paul Merriman has said, "a year in the markets is just noise."

If you aren't concerned with short-term performance, and you shouldn't be if you are investing for the long-term, then don't pay attention to it.

4. The Callan table may lead investors to believe that they need more complexity in their portfolio than they truly do.

Seeing 9 different asset classes that seem to have fluctuated randomly from year to year may lead investors to conclude that one would have needed all 9 asset classes in their portfolio over that time frame in order to be properly diversified. But again, this is false.

Investing equally in the 9 asset classes shown in the Callan table from 2002-2021 would have returned 5.27% annualized after inflation for U.S. investors. But if U.S. bonds, ex-U.S. bonds, high yield bonds, and cash were all substituted for only U.S. bonds, reducing the 9 asset classes to 6, the returns would have been identical, and the portfolio would have had less volatility and a smaller maximum drawdown.

It's obviously easier to manage a portfolio with 6 asset classes than one with 9. And most investors probably don't need more than 4-5 asset classes for their portfolios to be well diversified.

But more importantly, investors who believe that they need 9 (or more) asset classes may believe that they cannot manage their portfolio on their own and seek out help from an investment advisor. Such advisors routinely charge fees of 1% of the assets they managing, and this seemingly small fee can actually cost a single investor millions of dollars. Yes, millions.

Final Thoughts

While it's undeniable that tables, graphs, and other means of visually displaying data can be highly beneficial to helping us understand those data, reducing the information in those data, as is done in the Callan table, can do more harm than good.

If you want to look at the past performance of various asset classes, using a free tool like Portfolio Visualizer that displays cumulative returns, volatility, drawdowns, and other metrics will be far more useful than the Callan table.