Bob Shiller is concerned about the rise of “everything is different” chatter. Seeing parallels in the dot-com bust in 2000-2002 with current conditions, the Yale Professor tells Bloomberg that the equity market looks frothy, based on his calculation of the cyclically-adjusted price-earnings ratio (CAPE).

Shiller’s valuation metric is now at 29.77, based on the current calculation — the highest level since 2002. There’s no rule that says the stock market can’t go higher from here, perhaps much higher and for a lengthy stretch of time. But Shiller’s research, summarized in his book Irrational Exuberance, advises that relatively high CAPE readings tend to be associated with low/negative expected stock-market returns. By that standard, CAPE’s elevated level looks worrisome.

The caveat is that the historical record is only a rough estimate at best for deciding what comes next. There’s also a debate about what defines a genuinely high CAPE reading. Although the current 29.77 level is high by recent standards, it’s been much higher – over 40 in 2000.

Meanwhile, the economic climate is different in 2017 vs. 2000. As noted yesterday, US recession risk is low at the moment. Also, the economy is in no danger of overheating for the foreseeable future. Do those factors mean that CAPE’s rise isn’t as troubling as it appears? Maybe, but Shiller’s still anxious. “The market is way over-priced,’’ he warns.

But that was true last month as well, or so it appeared, when I profiled bubble risk on Feb. 14. Fast forward a month and the stock market is higher and so is CAPE. Identifying bubbles in real time and predicting when or if they’ll pop is difficult bordering on impossible. That’s a reminder that going all in (or out) with regards to equity allocations is almost always a bad idea.

Still, the question is whether there’s any support for Shiller’s CAPE-based hypothesis that suggest expected return looks challenged at the moment? As one answer let’s consider a version of an econometric estimate of bubble risk that makes periodic appearances on these pages. The statistical engine is what’s known as the Augmented Dickey-Fuller Test, which is applied to the S&P 500 on a rolling 36-month basis. (For details about the methodology, see this post from 2014.) The output, p-values, can range from 0 to 100, with 100 indicating extreme bubble risk. The bad news: the current reading is roughly 0.95, which is to say about as high as it gets.