Now the finding that matters most to anyone who allocates capital, and which surprises almost everyone the first time. Across countries and over long horizons, real GDP growth and real equity returns correlate weakly and sometimes negatively. Elroy Dimson, Paul Marsh and Mike Staunton have reported this repeatedly from their long-run global returns dataset, and it survives most reasonable ways of cutting the data.
How do you decompose a stock index return?
The reason is visible once you decompose an index return into what actually produces it:
(1 + R) = (1 + dividend yield) × (1 + real EPS growth) × (1 + inflation) × (1 + share-count effect) × (1 + repricing)Five terms, and only one of them is anything like GDP growth. Take two stated index builds over the decade (both figures are a stated vintage; check each component before quoting it):
| Term | US large-cap, 2014 to 2024 | India large-cap, FY2014 to FY2024 |
|---|---|---|
| Dividend yield | 1.80% | 1.30% |
| Real EPS growth | 4.00% | 5.50% |
| Inflation | 3.00% | 5.20% |
| Share-count effect | +0.60% | −0.80% |
| Repricing (P/E) | 18.0 → 24.0, +2.9186% | 18.5 → 22.5, +1.9767% |
| Total nominal return | 12.9042% | 13.7340% |
| Real total return | 9.6158% | 8.1122% |
Advertisement
Why doesn't GDP growth show up in stock returns?
Three things in that table break the intuitive link between an economy and its index.
The share-count effect is the one that catches people out. American companies retired shares over the decade, adding 0.60 points a year to the return on each surviving share. Indian companies issued them, subtracting 0.80 points.
Aggregate corporate earnings can grow with the economy while earnings per share grow more slowly, because the new capital that funded the growth was raised from new shareholders. Fast-growing economies issue more equity, so the wedge is systematically worse in exactly the places where growth is fastest.
Repricing contributed 2.9186 points of the American 12.9042, which is 22.62% of the total return, and 1.9767 of the Indian 13.7340, which is 14.39%. That component is a change in what buyers will pay for a rupee or a dollar of earnings. It is not growth in any sense, it cannot repeat indefinitely, and it is the first thing to strip out before extrapolating.
And the growth-to-earnings wedge runs opposite to the intuition. American real EPS grew 4.00% against real GDP growth of 2.62%, a wedge of plus 1.380 points. Indian real EPS grew 5.50% against real GDP growth of 5.90%, a wedge of minus 0.397 points.
A listed index is a selected, changing subset of an economy. The unlisted, informal and newly formed parts of a fast-growing one capture a large share of the growth without ever appearing in it.
The realised risk premium tells the same story from the other end. Against a start-of-window ten-year sovereign yield of 2.60% in the United States and 8.70% in India, the realised premium was 10.30 points and 5.03 points. The faster economy delivered the smaller realised premium, by 5.27 points.
Do earnings track growth within a single economy?
One clarification, because the earlier macro work states something that sounds opposite. Within a single economy over a century, real corporate earnings do broadly track output per worker, and productivity remains the trend line under the market. The weak correlation here is a cross-country result over decade-length windows, and the three terms above are exactly why the two statements coexist: share count, index coverage and repricing all move faster than productivity and all differ between markets.
None of this says emerging-market equity is a poor asset. It says the causal chain most people carry, fast GDP growth to fast earnings growth to high equity returns, has a leak at each of the two joints, and the size of each leak is measurable from published data. Where the forward-looking premium comes from is a discount-rate question the valuation phase already answered, and it is not a growth question.