Benjamin Graham was Warren Buffett’s teacher and mentor. He also ran an investment fund which specialized in uncovering undervalued stocks.

One day in 1926, Graham was at his desk, reading through a government report on railroads, when he noticed a potentially valuable footnote. It referenced assets held by a number of oil pipeline companies, but there wasn’t a lot of detail. So Graham boarded a train to Washington and found his way to the Interstate Commerce Commission (ICC), where he thought he might find more information.

What Graham uncovered in the dusty offices of the ICC confirmed what the footnote had suggested: that many pipeline companies, due to a quirk of history, held assets that were worth far more than the companies themselves. Northern Pipe Line’s shares, for example, were trading at $65 per share, but the company held bonds worth $95 per share. It was about as obvious an investment opportunity as ever existed. Graham began purchasing Northern shares and after discussions with the company’s management eventually unlocked $110 per share for investors. This investment was an almost no-lose situation, but it was one that other investors hadn’t noticed because the information was so inaccessible.

Today, the world is different. The sort of data that Graham had to look for in a government filing cabinet is now readily available online. Rarely, if ever, is there public information that isn’t in digital form. It’s for that reason that the market is regarded as being more efficient today. Stock prices are seen as being more accurate because all of the relevant data has been factored in.

This development is generally seen as positive. Greater access to information allows investors to make more fully informed decisions. You may recall Enron, for example, where an accounting fraud brought down the company. What caused the fraud to unravel? A reporter named Bethany McLean took a close look at the company’s annual report online and concluded that the numbers didn’t add up. She published her initial findings in the March 2001 issue of Fortune, and by December the company filed for bankruptcy. It was a stunning fall, all prompted by the research of a single reporter working from her desk. Looking at examples like this, today’s market seems far more efficient than it was in Graham’s time, when Enron might have been able to continue its fraud undetected for far longer.

In a recent paper, though, Clifford Asness, an investment researcher and fund manager, suggests that conventional wisdom might be wrong—that the market might be less efficient today than it was in the past, despite the improved access to data. To support his claim, Asness looks at market valuation as a proxy for market efficiency. Specifically, he looked at the valuation gap between the most expensive and least expensive stocks and examined how that gap has changed over time. What he found is that this ratio has been rising steadily for years, and through a number of lenses, it has become more extreme. You can see an illustration of this in his paper.

This valuation trend is evidence of inefficiency, Asness argues, for the simple reason that it’s irrational for investors to overpay for stocks. It’s a pillar of market efficiency, in fact, that prices should reflect all available data. Asness reasons that if investors were reading the data rationally, they would be making different decisions, and that would cause the valuation gap to close. But that’s not what’s happening.

Why would this be the case? If investors have more information today, why would they be making decisions that seem contrary to the data? Asness explores a variety of possible explanations. In the end, he concludes, ironically, that it’s the Internet itself that’s made the market less efficient. On the one hand, the web can be a source of reliable information—as it was for Bethany McLean. But on balance, Asness calls it a “fever swamp” of misinformation. In the past, there was the notion of the wisdom of crowds, which posited that groups of people together make better decisions. But Asness feels that today social media has produced the opposite result, turning large groups of people into “a coordinated clueless and even dangerous mob.”

We saw this sort of un-wise mob dynamic with the meme stock craze in 2021, when groups of investors bid up the shares of bankrupt and nearly-bankrupt companies. Their leader: a YouTube personality who called himself Roaring Kitty. That year also saw a boom in special purpose acquisition companies (SPACs) and other questionable investments.

While that period was perhaps extreme, the valuation data Asness presents suggests that the market is still less efficient than it used to be. The gap, in other words, between the most highly valued growth stocks and the most depressed value stocks doesn’t have a rational basis. If investors were looking more carefully at valuations, this argument goes, some of the most expensive growth stocks would see their valuations moderate, and some of the more depressed value stocks would see their price rise.

What antidote does Asness suggest? He believes that eventually reason will prevail. “Assuming a valuation change will continue to go on forever is obvious folly.” But he also knows that the current situation may not correct any time soon. “…depressingly, I’m saying you can do the right thing and still be wrong for about 30 years.” In other words, the market looks like it’s being irrational, but it might stay that way, perhaps for a while.

Where does this leave investors? In building a portfolio, I’d avoid making any big bets. If you start with a fund that tracks an index like the S&P 500—which includes both growth and value stocks—you’ll benefit regardless of which way things turn out. But recognizing that the market today is more heavily weighted toward growth stocks, you might incorporate a modest position in a fund that tilts toward value. That way, the bulk of your portfolio will be diversified, but you’ll have a thumb on the scale toward the stocks which, according to the data, are undervalued. If they eventually come back to life, you’ll benefit. But if they don’t, you won’t have given up too much.

Most importantly, I’d avoid going to any one extreme. On the one hand, I’d stay away from riskier, growth-oriented funds like QQQ. To be sure, their performance has been head and shoulders above the rest, but past performance does not guarantee future results. And according to the data, these are the stocks that are most overpriced. At the same time, as Asness says, the market isn’t always rational. So I wouldn’t interpret the data he presents as reason to go in the other direction, to become overly conservative. When it comes to the stock market, as I’ve suggested before, try to take a balanced, center lane approach.