The Equity Risk Premium Just Went Negative. Here's Every Other Time That Happened

"It has never been negative before" — the table below says it has

The "equity risk premium" (ERP) is simply the extra return stocks are expected to offer over safe bonds, measured here as the S&P 500's earnings yield minus the 10-year Treasury yield. As of October 23, 2026, the S&P 500's trailing earnings yield was 3.875% (GuruFocus), while the 10-year Treasury yield stood at 5.00% (Federal Reserve Bank of St. Louis, FRED, October 23, 2026). That makes the simple nominal ERP about -1.1 percentage points -- negative, meaning the stock market's earnings yield is currently lower than the "risk-free" return on a 10-year government bond. Depending on which earnings figure and which bond yield a given analyst uses, estimates right now range from roughly zero to notably negative; this article uses the nominal, nothing-fancy version so every number traces to a dated primary source, and treats the other published estimates as corroborating context rather than a contradiction.

Era	Earnings yield (approx.)	Equity risk premium	Avg. real return, following decade
Early 1980s	Above 12% at times	Strongly positive	~10.3%/yr (about 1-in-20 decades negative)
1997-2002 (dot-com peak to bust)	Below 5%	Near zero to negative	~3.7%/yr (more than 1-in-4 decades negative)
2010-2022	~5% (P/E ~20)	~+4.4 points (real, vs. TIPS)	Not separately broken out in the cited sources
Now (Oct 23, 2026)	3.875%	~-1.1 points (nominal, vs. 10-yr Treasury)	Unknown -- this is the open question

A negative or near-zero ERP sounds alarming by itself, but the useful question for an investor isn't whether the number is negative -- it's how often this has happened before, and what the market actually did afterward. Since 1950, the average ERP has run about +1.0 percentage point using Shiller's long-run dataset, according to an October 2026 analysis from Real Investment Advice. Today's reading sits well below that average. Lining up the handful of past episodes where the premium also collapsed gives a rough, far-from-certain sense of the range of outcomes that followed -- not a forecast.

"The 2010-2022 premium is the baseline we will revert to" — that decade is the outlier in this table

For more than a decade after the financial crisis, the premium swung back decisively in stocks' favor. Fortune's September 29, 2026 analysis (citing Dartmouth finance professor Kenneth French, who calls the ERP "the holy grail of stock market investing") puts the 2010-2022 average 10-year TIPS yield at just 0.6%, while the S&P 500 traded around a P/E of 20 -- a 5% earnings yield. On that real-yield basis, the equity risk premium averaged about 4.4 percentage points across those years, roughly four times today's level by the same methodology. Fortune frames the move since then in stark terms: the ERP has fallen about 77% from that 4.4-point average to roughly 1% as of late September 2026, driven almost entirely by the rise in bond yields rather than a stock-market crash.

The honest caveats matter as much as the pattern. The sample size is small -- a handful of multi-year episodes since 1950, not dozens of independent trials -- so two or three historical analogs are a thin basis for confidence, and each of those episodes had its own distinct cause (inflation shock, dot-com mania, a banking crisis) that may or may not resemble today's deficit-and-term-premium story. The forward earnings yield (based on analysts' expected future earnings rather than trailing results) runs closer to 5.3%, per Real Investment Advice's October 2026 figures, which would put the ERP much closer to zero than deeply negative; which number is "right" depends on whether this earnings growth materializes. There's also a structural argument that today's elevated bond yields reflect federal deficits and term premium rather than a stronger economy, which would make the bond side of the comparison less of a like-for-like "safe alternative" than it was in prior decades. None of this is a reason to ignore the signal, but it is a reason not to treat a single negative reading as a verdict -- the same data point can reasonably support "stocks are unusually expensive relative to bonds right now" and "the bond side of this ratio is unusually distorted right now," and separating those two stories matters more than memorizing the headline number. This article is for general investing education, not a personalized recommendation to buy, sell, or hold any security; anyone acting on valuation measures like this should weigh their own time horizon and risk tolerance, and a licensed financial professional can help with that assessment.

"A negative premium means the market is pricing a crash" — the dot-com version took years to resolve

The dot-com run-up pushed the opposite scenario: as stock prices outran earnings through the late 1990s, earnings yields fell and the ERP shrank toward zero and below. Real Investment Advice's October 2026 piece names 2002 specifically as the last time -- aside from the unusual 2009 earnings collapse -- that stocks yielded this much less than bonds before now, using the 1997-2002 window as a direct analog to today's setup. For starting points where the earnings yield had fallen below 5% (grouping together the 1990s run-up and the dot-com peak), the same analysis found a materially worse forward outcome: an average real return of about 3.7% a year over the following decade, with more than one in four ten-year periods finishing in negative territory. The 2007-2009 financial crisis produced a second, shorter negative-ERP stretch, according to Fortune's September 2026 coverage of the same measure, though neither source gives a specific trough value for that period.

Putting the current nominal reading (about -1.1) next to the historical regimes above, today looks closer to the 1997-2002 and 2007-2009 episodes than to the 2010-2022 stretch or the early-1980s extreme -- which is exactly why commentators keep reaching for 2002 as the comparison. As a purely illustrative exercise, applying the two historical regime averages from Real Investment Advice's study to a hypothetical $10,000, held for ten years in each scenario: at the high-ERP regime's average 10.3% real annual return, it would grow to roughly $26,650; at the low-ERP regime's average 3.7% real annual return, roughly $14,380. That is a historical average applied to a hypothetical sum -- not a projection of what the next ten years will do, and today's reading is more extreme (more negative) than the "below 5%" grouping threshold used to generate the lower figure, meaning even that more modest outcome could prove optimistic.

"The reading tells you what is wrong" — it does not say which side of the ratio moved

That last point is worth sitting with, because it changes what kind of risk this reading describes. A collapsing ERP driven by a falling stock market (as in 2000-2002 or 2008) is a very different event from one driven by rising bond yields while stocks simply hold their ground, which is closer to what's happened recently. Fortune's own figures show the 10-year TIPS yield rising 110 basis points since early March 2026 alone -- a rapid move in the "safe" side of the comparison, not the "risky" side. If elevated yields mostly reflect federal deficits, term premium, and bond-market supply rather than a hotter economy, then a falling ERP in this particular episode says more about what's happening in the Treasury market than about stocks being priced for disaster. That's a meaningfully different story than the 2000 or 2008 versions of the same number falling for the opposite reason.

At the opposite extreme from today, the early 1980s saw S&P 500 earnings yields climb above 12% at times, according to Real Investment Advice's October 2026 historical breakdown -- a function of beaten-down stock prices (after the 1970s' inflation-driven bear market) rather than booming profits. Treasury yields were also unusually high in that period, but stock earnings yields ran well above them for an extended stretch, producing one of the largest positive equity risk premiums on record. In the decade that followed high-ERP starting points like this one, Real Investment Advice's grouping found an average real (inflation-adjusted) return of about 10.3% a year, with roughly one in twenty ten-year periods ending with a loss. That's the "cheap stocks, expensive patience pays off" case -- the reading itself didn't guarantee the result, but it coincided with one of the more favorable setups in the dataset.

This article is for general investing education only, using historical index-level data. It does not constitute personalized investment advice, and past performance does not guarantee future results. All figures are as of the dates noted and sourced to the publications above.

Sources

GuruFocus, "S&P 500 Earnings Yield," data as of October 23, 2026 -- gurufocus.com
Federal Reserve Bank of St. Louis (FRED), 10-Year Treasury Constant Maturity Rate (DGS10), October 23, 2026 -- fred.stlouisfed.org
Real Investment Advice, "Equity Risk Premium Has Vanished, And Stocks Aren't Cheap," Lance Roberts, October 10-12, 2026 -- realinvestmentadvice.com
Fortune, "Historically, stocks have offered a big premium over bonds. Suddenly, the difference has almost vanished," Shawn Tully, September 29, 2026 -- fortune.com

Search This Blog

USStockInsights