Stats › Valuation › Equity premium debate
The equity risk premium debate
The equity risk premium is meant to be a simple idea. Stocks are riskier than government bonds, so over time they ought to pay you more, and the premium is how much more. It's one of the load bearing numbers in finance. It sits inside pension assumptions, company valuations, retirement plans and just about every long range forecast you'll ever be shown.
The trouble is there isn't one of it. There are at least four numbers in circulation, all called the equity risk premium, and they aren't rival estimates of the same thing. One measures what actually got delivered over the past century. One is reverse engineered out of today's prices. One is what practitioners say they expect when you ask them. And one is a forecast for the future, made by the same researchers who assembled the historical record. They currently sit at roughly 5.0, 4.18, 5.5 and 3.0 to 3.5. Averaging those would be meaningless, and people do it anyway.
This page lays out all four, explains what each one is answering, and covers the thing underneath them all: a 1985 paper that showed standard economic theory could only justify a tiny fraction of the premium that actually turned up, and forty years later still nobody agrees why. Five sources, dated and linked at the foot of the page, and the notes section flags several places where figures get mislabelled in circulation. Worth reading those before you quote anything from here.
There is no single equity risk premium and the four figures in circulation are not competing estimates of the same quantity. The realised US premium over bonds from 1900 to 2025 was about 5.0 percentage points. The implied forward looking premium was 4.18% on 1 July 2026. Surveys of practitioners give 5.5% and of chief financial officers 4.42%. And the researchers behind the long run dataset put the prospective world premium at 3.0% to 3.5%. The underlying problem has not gone away: in 1985 Mehra and Prescott showed standard models could justify at most 0.35% against a realised 6.18%, and the gap is still called a puzzle.
Members only
The rest of this page is for members
Below this point there are 9 sections, 1 chart and 5 named sources, roughly 2100 words of it. Every figure carries the source it came from and the date the data is from.
You can keep browsing every statistic in the library for free. The intro and the summary are always open.