What we worry about, and what actually matters
A common observation among long-horizon investors is that they spend roughly 95% of their time worrying about things that explain roughly 5% of their long-run returns — and 5% of their time on things that explain 95%.
The bulk of investor attention goes to monthly macro releases, central-bank speech parsing, quarterly earnings noise, and political headlines. These move markets in the short term and have very low predictive power for 20-year outcomes.
Key takeaway — Across 60 years of MSCI World data, starting valuation, starting dividend yield, and earnings-growth realisation explain over 90% of the variance in 20-year equity returns. The other factors most investors track contribute almost nothing.
The three factors
Factor one: starting cyclically-adjusted earnings yield (CAPE earnings yield).
The inverse of Shiller's CAPE. When equity earnings yield is high relative to history, subsequent 20-year returns are high. When it is low, they are low. This is the most well-documented relationship in equity finance and survives most attempts to break it. The R-squared on its own is around 0.55 for global equity.
Factor two: starting dividend yield plus expected real growth.
Sometimes called the "Gordon growth" decomposition. Long-run real total return decomposes almost cleanly into: starting dividend yield + earnings growth + valuation re-rating. Over 20-year windows the re-rating term contributes almost zero on average; the first two dominate.
Factor three: realised earnings growth versus expected.
This is where forecasting is hardest but the impact is largest at the country and sector level. Markets that underperform expected earnings growth over a 20-year window lose. Markets that exceed it (Asia in 1990-2010, US tech in 2010-2020) win disproportionately. The variance here is driven by structural factors: demographics, productivity, capital efficiency, regulatory regime — not by macro flow.
What this implies for portfolio decisions
The implications are uncomfortable for the industry:
- Country selection at the 20-year horizon is dominated by the three factors, not by anything in next quarter's GDP print.
- Active managers in highly efficient markets cannot beat the index by being right about the three factors any more than the price already reflects them.
- The cost of being wrong on the three factors is enormous, and the cost of being wrong on most other factors is trivially recoverable.
Where most investors get the three factors wrong
Three common errors:
- Anchoring on trailing earnings yield rather than CAPE. Trailing earnings are too noisy and reflect cycle position; CAPE smooths the cycle and is the relationship that holds.
- Treating recent earnings growth as the expected forward growth. Recent growth is mean-reverting; expected forward growth should be modelled from structural factors. The US 2010-2020 cycle is a cautionary tale on both sides — under-extrapolating early, over-extrapolating late.
- Ignoring currency. For non-base-currency equity exposure, FX moves can dominate equity returns over 5–10-year windows. Over 20 years FX tends to mean-revert toward PPP, but the variance around the mean is high.
A practical reading list for the long horizon
If you have an hour a week to spend on long-horizon equity allocation:
- Spend 15 minutes on the current CAPE / earnings-yield map across major markets;
- Spend 15 minutes on the dividend-yield map;
- Spend 15 minutes on the structural-growth assessment — demographics, productivity, capital-formation rates — of the markets you have exposure to;
- Spend 15 minutes on everything else: macro headlines, central-bank communications, sector rotations.
This is roughly the inverse of the time-allocation most investors actually use.
Bottom line
The factors that drive 20-year equity returns are well-documented, stable across regimes, and almost entirely ignored by investors who spend their time on noisier data. Re-allocating attention to the three factors that actually matter is the highest-leverage change any long-horizon investor can make — and it costs nothing but the discipline to do it.