A historical look at the debate between stocks and bonds

Digging deep into numbers prior to modern bookkeeping produces surprising results that challenge our beliefs

A fundamental belief of mine — which I’ve repeated often in private conversation, public forums, classrooms, this column and my book — is that stocks beat bonds over the long run. And the longer the run, the larger the beat-down.

That belief marks me as an acolyte of Jeremy Siegel’s classic book, “Stocks For the Long Run.”

As an aside, if you haven’t read his book, you should go out and buy it right now. Siegel first published it in 1994 but updated editions have been reissued as recently as 2014.

But back to the topic at hand.

It’s always hard to hear or read something that contradicts what we deeply believe. And yet a new academic paper from professor Edward McQuarrie shows important exceptions to my deeply held belief.

In his paper, “The First Eighty Years of the US Bond Market: Investor Total Return from 1793, Combining Federal, Municipal, and Corporate Bonds,” McQuarrie provides evidence for long time periods when bonds actually beat stocks. What? No! On the one hand, this academic paper appeals most to finance nerds (guilty as charged!), dealing as it does with price and investor-return data more than 200 years old, plus the challenges of integrating and interpreting this information with imperfect records. (Did you know that people didn’t even have the Yahoo Finance app on their phones back then?)

It’s hard to compile accurate financial data from the olden days. As McQuarries writes, the works of historical researchers like him and others means “the stock component of nineteenth century investor returns has become steadily less fictional.”

On the other hand, it’s not just an academic discussion because we look to history as an important guide to how we allocate our accounts to stocks and bonds.

If my deeply held belief about the past is wrong, maybe it will lead me to be more cautious about what will happen in the future.

McQuarrie describes something I did not know or appreciate when I became an acolyte of Siegel’s “stocks for the long run” cult. Specifically, the observable data on stocks and bonds before about 1850 is very scarce. There was no continuous stream of U.S. Treasury bonds to form what Siegel estimated as the “risk-free” rate of bond yields.

Studies like Siegel’s about what a bond investor could earn over the long run consisted of quite a bit of educated guessing and interpolation between data points. McQuarrie has built a more thorough data set of bond prices and yields that includes municipal and corporate bonds, in additional to available federal bond information.

A central point of McQuarrie’s study is that bonds in the 19th century came in all sorts of risky forms beyond the “safe” U.S. Treasury bond, so a bond investor of those days would have experienced a much more volatile ride than would a bond investor of today. At the same time, the average bond investor — because a portfolio would have included these riskier bonds — would have earned a higher yield. That would have produced a higher return than stocks over long periods of time.

Using McQuarrie’s time series, he estimates that in the 50 years following 1793, for example, bond investors beat stock investors in terms of total return.

Fifty years is far longer than I would have previously believed. It’s so long, in fact, that it poses a fundamental challenge to one of my deepest investment beliefs.

McQuarrie further argues that there isn’t a fundamental baseline advantage of stocks over bonds. Instead, returns from both asset classes are far more random, within a set of boundaries, than Siegel has argued and more random than what I have previously thought.

What does an investor today do with this information? Philosophically, we should be less dogmatic about what the past tells us. And maybe more skeptical about past estimated data, especially as we look into distant history when records were not as good as they are today.

Practically speaking, however, the last century or so still supports Siegel’s (and therefore my) approach. In addition, we have much better price and yield data for the last 100 years, so we can make interpretations with more confidence.

I’m going to keep advocating my case, although with more humility. Since 1926 — for which we have great data — we can say the following:

Stocks outperform bonds in most years, although not every year. The more years you hold stocks or bonds, the greater the difference in performance.

For example, looking back to performance since 1926, if you had a one-, five-, 10-, 15- or 20-year holding period, we know how frequently in percentage terms a basket of stocks beat a basket of bonds.

Over a one-year period, a basket of stocks beat a basket of bonds 64 percent of the time; over five years, 70 percent of the time; 10 years, 80 percent; 15 years, 87 percent; and over 20 years, stocks beat bonds more than 99 percent of the time. (Hat tip to financial adviser David Hultstrom of Financial Architects LLC in Woodstock, Ga., for compiling those stats.)

That last stat means that in the last 92 years, there has not been a 20-year holding period in which bonds have beaten stocks. The clear implication is that if your holding period for an investment is 20 years or more (like it is in your retirement account), then you are making a low-probability bet by holding bonds. And that low-probability bet is probably going to lower your returns.

McQuarrie’s paper reminds us that although that’s been true for roughly the last 100 years, it wasn’t true if you look back 200 years or more. I’m not leaving the cult, but I’m modestly pointing out the value of questioning deeply held beliefs.