The future is hard to backtest

People have always asked me for financial advice because they knew I worked in the stock market. At first, I’d try to explain that I was a trader, not an asset manager, and that my trading horizon ranged from minutes to days. That usually got me a blank stare. The difference didn’t register. People kept asking and I eventually decided it was easier to just tell them something.

So here’s the advice I’ve been offering for three decades now: Open a Vanguard account and buy its S&P 500 index fund.

I based that profound advice partly on my personal experience: Trading is near impossible, so there’s no point trying to beat the market.

And partly on history: Over the long run, stocks have always been the best investment.

The data for the latter point is provided by Jeremy Siegel, author of the foundational text of investing advice: Stocks for the Long Run.

You don’t have to read the whole thing. “The most important chart of this book,” Siegel says, is the first one:

Cobbling together two centuries of data, Siegel found that over 220 years, US equities have averaged a compounding real return of 6.9% per year.

6.9% is a lot, especially as it’s adjusted for inflation. It means that US equities have doubled investors’ purchasing power every 10 years — for two centuries.

To put that in perspective, Siegel notes that $1 million invested in the US stock market in 1802, with dividends reinvested, would have grown to $54 trillion by 2021.

That is only a thought experiment: It would have required buying one-fifth of the entire 1802 stock market and never consuming any of the returns. But Siegel believes it's directionally instructive: “In the long run, history has shown that stocks are safer than bonds for long-term investors whose goal is to preserve the purchasing power of their wealth.”

On a long enough time horizon, his data suggests they might even be perfectly safe:

This makes investing pretty easy: If you hold stocks for at least 30 years, you have a 100% chance of outperforming risk-free T-bills.

Is that really how it works, though? Is past performance a guide to future performance?

The advisors at Vanguard seem to think so.

“Stocks have had a 10% average annual return over the long run with a lot of ups and downs along the way,” their website explains. “That means, all things considered, a stock-heavy mix is likely to return more than a bond-heavy mix over the long term.”

Elsewhere on the site they add that “investors shouldn't expect future long-term returns to differ significantly from the markets' long-term historical averages.”

That is reassuring.

But why, exactly? Is there any reason to believe the next century of investing will be anything like the last two?

Siegel’s dataset begins in the very different world of 1802 when New York investors had just six stocks to choose from and no exchange to trade them on. The city’s stock and bond traders met informally in a coffee house.

Incredibly, two of the six stocks are still trading: the Bank of New York (now Bank of New York Mellon) and the Manhattan Company (now JPMorgan).

Most of the rest of Siegel’s dataset would be almost unrecognizable to modern investors.

In the 19th century, stocks were so distrusted that a company’s dividend yield was typically higher than its bond yield.

In 1920, the S&P 500 traded at a multiple of just 5x earnings and paid a dividend yield of 6%.

In 1976, Ben Graham advised investors to buy stocks that traded below the value of their cash and other readily sellable assets — assigning zero value to long-term assets like property and factories. There were more than 300 of these rock-bottom bargains at the time — about 10% of all US-listed companies.

Graham additionally recommended buying stocks at below 7x earnings or above a 7% dividend yield. 

In 1982, they would not have been hard to find. The S&P 500 traded at 7x earnings that year and a 6% dividend yield, depressed by a fed funds rate as high as 19%.

For most of the next four decades, stock prices went up — in large part because bond yields went down. But also because stock market investing became a mainstream pursuit. 

401(k)s, ETFs, CNBC, Jim Cramer, E*Trade, Robinhood, zero-commission trading, and WallStreetBets brought more and more people into the market.

Now, US households have a record 32% of their assets in equities…

…and the S&P 500 trades on a Shiller P/E ratio of 42x.

None of these investing tailwinds are likely to repeat. Bond yields can’t go from 16% to 4% again (because they’re already 4%). Trading fees can’t go to zero again. ETFs can’t be reinvented. Investing can’t go mainstream for the first time again.

These one-time events have structurally raised the valuation of equities: It’s unlikely we’ll ever get to buy the S&P 500 at 7x earnings again.

Structurally higher valuations should — all else being equal — lower the future return from investing in equities.

Things will not be equal, of course. 

They might be better!

Artificial intelligence might eliminate scarcity, for example, which would probably be good for the stock market. Or corporate America could simply continue to get bigger, more efficient, and more profitable.

But the last century will be difficult to beat for US investors because a lot went right for the US: the post-war manufacturing boom, winning the Cold War, exporting free-market capitalism and the US dollar to the world, importing the best human capital from the world, inventing the internet, and incubating almost all of the tech companies. Among other good things.

With that in mind, Vanguard’s observation that stocks have historically returned 10% might more accurately be stated as stocks in the country that became history’s most dominant economic and technological superpower returned 10%.

That is good to know, but perhaps not very predictive. 

To his credit, Siegel acknowledges this. “One must be aware of the political, institutional, and legal framework in which these returns were generated,” he cautions. “The superior performance of stocks over the past two centuries might be explained by the growing dominance of nations committed to free-market economics.”

Perhaps more surprisingly, Vanguard seems to acknowledge it, too.

Roger Aliaga-Díaz, Vanguard’s chief economist, currently recommends a 40/60 allocation between stocks and bonds, an inversion of the default 60/40 portfolio.

The 60/40 rule of thumb, which so many of us follow, is based on a 1994 paper on retirement planning with an instructive title: “Determining Withdrawal Rates Using Historical Data.”

In other words, most of us are investing based on the assumption that historic returns are predictive of future ones.

Aliaga-Díaz thinks that’s a mistake: “Our analysis of fundamental drivers points to greater odds that longer term returns will be subdued below historical averages.”

That is bad news for investors, of course, if he’s right.

But either way, his methodology is even worse news: It implies that we should be thinking less about historical data and more about the future.

Which is hard, because there’s no data on the future. 

I’m done giving investment advice.

(For real this time.)

— Byron Gilliam