What GMO's 7-year forecast says about valuations in 2026

GMO's 7-year asset class forecast as of 31 July 2026 projects a real return of minus 7.2% a year for US large cap equities over the next seven years, minus 5.9% for US small caps and minus 0.7% for emerging market equities, against plus 4.4% for Japanese small value stocks and plus 3.0% for international deep value. US bonds are forecast at 2.5% real and US cash at 1.4%. The forecast is a valuation exercise, and the valuation it reacts to is visible in Robert Shiller's data, which put the cyclically adjusted price earnings ratio of the S&P 500 at 40.7 in October 2026 against a series average of about 17.8 since 1881. This article sets out the numbers and the method. It is a description of one firm's model and of public valuation data, and it is not a recommendation.

The July 2026 numbers

GMO publishes two sets of forecasts, one for a "normal" interest rate environment consistent with a 1.3% equilibrium real cash rate and one for a "low" environment at 0.0%. All figures are annualised real returns over seven years.

Source: GMO, 7-Year Asset Class Real Return Forecasts, as of 31 July 2026. GMO states the chart represents forecasts for asset classes and not for any GMO fund, and that forecasts are forward looking statements based on its reasonable beliefs and not a guarantee.

Three features stand out. The spread between the most and least attractive equity classes is more than 11 points a year. Value and small cap classes outside the US are the only equity categories forecast to beat US bonds in the normal scenario. And the low rate scenario lifts every forecast except cash by 1 to 3 points without turning US large caps positive.

How the method works

GMO's research library page describes the series as seven year real return forecasts; the published chart is the firm's output rather than a full statement of its model. In outline, the approach assumes that valuations and profitability revert towards long run norms over a seven year horizon, and decomposes the expected return into the yield received, the growth in underlying fundamentals, and the gain or loss from the change in valuation as it moves back towards fair value. An asset priced far above its historical norm therefore carries a large negative valuation term, which is what produces a minus 7.2% figure for US large caps when the index trades at a CAPE near 40. The two rate scenarios exist because the fair value that prices revert towards depends on the real cash rate assumed. A lower equilibrium rate justifies higher sustainable valuations, which is why the low rate column is kinder to every asset except cash itself.

The method has known limitations. Reversion can take much longer than seven years or arrive through earnings growth rather than price falls, and the forecast says nothing about the path.

What the valuation data show

Shiller's dataset, updated through the start of October 2026, gives the CAPE at 40.7, using the 2 October close of the S&P 500. It was 37.7 in December 2024, 39.6 in December 2025 and 40.0 at the end of July 2026 when GMO ran its forecast. The series peaked at 44.2 in December 1999 and averaged about 17.8 across its full history. The same file records the US 10 year Treasury yield at 5.24% in October 2026, up from 4.32% in April, so the valuation of equities has risen while the competing risk free yield has also risen, a combination that mean reversion models treat as the least favourable.

Other institutions arrive at a similar direction with gentler numbers. J.P. Morgan's 2026 family office report quotes its own 10 to 15 year capital market assumptions as forecasting no asset class above 10% nominal, while about 55% of its family office respondents target 7% to 10% and almost a third target more than 11%.

Relevance to a family office

For an office whose strategic allocation is roughly a third listed equities, most of it North American, the GMO figures and the CAPE describe a concentration in the one asset class that valuation models like least. That is information about risk rather than an instruction. The practical questions it raises concern the benchmark against which the portfolio is judged, covered in benchmarking family office returns, and whether the rebalancing policy would respond to a prolonged drawdown in listed equities while private holdings are marked slowly, covered in rebalancing a portfolio that is 40% illiquid. The allocation that most offices actually hold is set out in family office asset allocation in 2026.

GMO's forecasts have been bearish on US equities for most of the past decade and US equities have risen for most of it, which is the strongest argument against treating the model as a timing tool and the strongest argument for treating a CAPE above 40 as a fact worth knowing.