What do U.S. equity investors fear the most? Market crashes? Sure. High inflation? Definitely. But what’s worse than both? A lost decade.
Lost decades are periods where U.S. stocks generate little to no total return for roughly a 10-year period (or longer). The three most prominent lost decades in U.S. stock market history occurred in the 1930s, the 1970s, and the 2000s. In each of these periods, U.S. stocks declined by 50% (or more) and then took around a decade to recover.
The good news is that lost decades don’t happen too often. In the 10 decades from 1920-2020, only 3 of them would be classified as such. But, do you know what’s even better? When we take into account how a typical person invests their money, these lost decades become shorter and less prevalent.
How do I know? Let me show you.
Why Buying Over Time Smooths Returns
Most of the time when investors look at historical performance they consider lump sum (or one-time) investments in a single asset class. They look at snapshots between point A and point B. Between one point in time and another.
Unfortunately, this isn’t how most people invest, especially not today. Thanks to the rise of index funds/ETFs and automated 401(k) contributions, many individual investors are purchasing shares of multiple asset classes over time.
However, this diversified, dollar cost averaging-based approach leads to a new issue: how do I calculate historical performance? With a lump sum investment, calculating historical performance is easy. You take the final value (of an asset), divide by the initial value, and subtract 1. That’s your overall return.
But, if you invest the same amount of money on a specific schedule, the math is no longer so simple. Now you have to calculate the return of every individual purchase and average them out. This isn’t easy to do without Excel or a computer program.
While this does make it harder to calculate your return, thankfully, it also smooths out your return as well. More importantly, when buying over time, no single decline ends up mattering all that much. What really matters is where you end up at the end of the period.
Thankfully, we have computers that can run these calculations for us. And when you do, you will see that buying over time makes the lost decades far less painful than they might otherwise seem.
How DCA Shortens Lost Decades
To illustrate this, consider the chart below which shows the rolling average dollar growth of a 10-year dollar cost averaging (DCA) simulation in an 80/20 U.S. stock/bond portfolio over all periods from 1920 to April 2025. What the plot shows is the average inflation-adjusted value of each dollar invested monthly over the 10-year span, assuming annual rebalancing and dividend reinvestment.
So, if the average real dollar growth is equal to $1, then that means that your investments, on average, matched the rate of inflation. If the real dollar growth is less than $1, then your investments underperformed inflation, and, if it is greater than $1, then they outperformed inflation.
Putting this in numeric terms, if you invested $833.33 a month for 120 months (or $100,000 across 10 years) and the average real dollar growth was equal to $1, then your final portfolio value would’ve been $100,000 (i.e. it matched inflation).
The good news? Historically you would’ve outperformed inflation in 89% of 10-year periods when buying over time:
And sometimes it outperformed magnificently so. For example, consider the first (leftmost) point on the plot, which shows an average dollar growth of $2.50 for the period ending December 1929. This means that for every dollar you invested each month from January 1920 through December 1929 into an 80/20 portfolio, you would’ve had $2.50 on average at the end.
So, if you invested $833.33 a month starting in January 1920, you would’ve ended up with around $250,000 by December 1929 (after reinvesting dividends, adjusting for inflation, and doing an annual rebalance). That’s one of the best 10-year DCA periods on record.
Compare this to period beginning in January 1965 and ending in December 1974. During that 10-year window, every dollar invested would’ve shrunk to $0.68 on average. So the same $833.33 a month ($100,000 in total) would’ve declined in value to only $68,000 by December 1974. That’s the lowest point in the data.
But the outliers aren’t what’s important. What matters is how few periods there are where you lose purchasing power over a 10-year period. Even during the DotCom bubble, there was only a 13-month period (from October 1998 to November 1999) where anyone who started buying would’ve underperformed inflation.
In other words, if you invested the same amount of money each month into an 80/20 U.S. stock/bond portfolio from January 2000 to December 2009, you would’ve slightly outperformed CPI.
I know this isn’t a great return, but considering that this is one of the worst 10-year periods in U.S. stock market history, you can see how remarkable this is. And if you extend your simulation (to a 20-year buying window), then the probability of you losing money in real terms during any period is even lower:
The craziest part about this plot is the 20-year period ending around 2000. Every dollar invested into an 80/20 portfolio from the early 1980s until the late 1990s would’ve grown to $4 in real terms (on average)! I doubt we ever see another 20-year period like it again.
Either way, lost decades are far less painful when you buy over time. And, since this is how most people tend to invest, even the worst decades rarely feel lost. This suggests that many of the fears we have about such periods tend to be exaggerated.
Of course, it’s easy to say this in hindsight. Buying over time works with U.S. stocks because U.S. equities and the U.S. economy have always recovered. If either of these truths fail to materialize in the future, then my historical analysis will no longer hold.
Thankfully, I’m not worried about such an event occurring. Betting against America hasn’t worked yet—and I wouldn’t start now.
Happy investing and thank you for reading!
If you liked this post, consider signing up for my newsletter.
This is post 454. Any code I have related to this post can be found here with the same numbering: https://github.com/nmaggiulli/of-dollars-and-data
Leave a Reply