Every year the large asset managers publish their capital market assumptions: tables of expected annual returns for equities, government bonds, credit and emerging market debt over the next five or ten years. They circulate widely, get quoted in the press, and end up shaping how a lot of people allocate their savings. They are also routinely misread.
This piece is about how to read those forecasts: what they actually claim, how they are built, why they are almost always wrong in the short run and still worth attention over the long one, and what a trader or investor should sensibly do with them.
What an expected return forecast actually says
The first misreading is the most common: a five-year expected return of, say, seven per cent a year is not a prediction that the asset will return seven per cent next year, or in any given year. It is the centre of a distribution. The same models that produce the central figure usually produce a range around it, and that range is wide, often wide enough to include years of double-digit losses.
The second misreading is treating the number as a forecast of events. It is not. It is mostly arithmetic applied to today's starting conditions. For a bond, the expected return is dominated by the yield you can lock in today, plus or minus the effect of the curve as the bond ages. For equities, it is typically the sum of the dividend yield, expected earnings growth, and an assumption about whether valuation multiples drift towards some long-run average. Change the starting valuation and the whole number changes, which is precisely why these forecasts move so much from one year to the next without anyone predicting anything new.
Why the starting point does most of the work
This is the part worth internalising, because it explains the swing in sentiment over the last few years better than any narrative about the economy.
Through the decade of near-zero and negative rates, expected returns on government bonds were close to nothing, for the simple reason that you cannot earn much from a bond bought at a yield close to zero. That was not pessimism; it was arithmetic. When policy rates rose sharply, the same arithmetic delivered the opposite result: bonds bought at materially higher yields carry materially higher expected returns, and forecasts published from 2023 onwards duly showed government debt back in positive territory, with US Treasuries typically ahead of euro area government bonds and investment grade credit ahead of both.
Equity forecasts work the same way in reverse. A market that has already re-rated upwards has a lower expected return from here, not a higher one, because part of the future return has been pulled forward into the price. This is the single most counter-intuitive feature of these tables: strong recent performance usually lowers the expected return, and a bad year usually raises it. Investors who chase the asset class that did best last year are, in the language of the forecasts, buying the one with the worst outlook.
Where the numbers come from, and what they hide
Three assumptions do most of the work in any of these models, and they are worth looking for whenever you see a headline figure.
The first is the valuation assumption: whether multiples are assumed to revert towards a historical average, and over what period. This single choice can swing an equity forecast by several percentage points a year, and reasonable people disagree about it strongly.
The second is the earnings growth assumption, which is usually modest, in the low single digits in real terms. When a forecast leans on an unusually high growth number, often justified by a structural story such as technology adoption, that is where the optimism is hiding.
The third is currency. An expected return is always quoted in a base currency, and for an investor whose money and expenses are in a different one, the exchange rate over five years can matter more than the asset return itself. A seven per cent annual return in dollars is not seven per cent for a euro-based investor unless the pair happens to cooperate, which over five years it may not. This is not a footnote; for foreign assets it is frequently the largest single source of uncertainty. Anyone holding foreign assets is running a currency position whether they intended to or not, which is why the mechanics and cost of hedging that exposure deserve more attention than they usually get.
What they are useful for, and what they are not
Expected return forecasts are useful for one thing above all: setting realistic expectations about what a balanced portfolio can deliver over a long horizon, and therefore how much you need to save and for how long. Used that way they are a planning tool, and a good one, because the arithmetic behind them is sound even when the specific numbers turn out wrong.
They are also useful as a relative signal. When the gap between the expected return on bonds and on equities narrows, the case for taking equity risk weakens, and vice versa. That comparison survives even if both absolute numbers are off, because the errors tend to move together.
They are close to useless for timing. Nothing in a five-year forecast tells you what happens in the next six months, and using one to justify a trade is a category error. The horizon of the forecast and the horizon of a trade have nothing to do with each other, and anyone building short-term positions from long-term capital market assumptions has confused two different activities. If your decisions live on a horizon of days or weeks, the relevant inputs are positioning, volatility and levels, not a five-year mean.
The practical conclusion: diversification is the part that survives
Whatever the numbers say in a given year, the conclusion drawn from them is nearly always the same, and it is the right one: combine assets whose returns do not move together. That is not a platitude. It is the only recommendation in this whole field that does not depend on the forecast being accurate, because the benefit of diversification comes from the correlation structure rather than from the level of returns.
The practical caution is that correlations are not fixed. Assets that diversified each other for a decade can start moving together precisely when it matters, which is what happened when bonds and equities fell in the same year during the inflation shock. Checking how the pieces of a portfolio are actually behaving, rather than how they behaved historically, is a live exercise: our correlation page tracks the current relationships between the major currencies, and our article on intermarket correlations covers how gold, oil and bonds interact with them.
How to read the next set of forecasts
When the next round of capital market assumptions lands, three questions will tell you most of what you need to know. What starting yield or valuation is the number built on, since that is where most of it comes from. What is being assumed about multiples reverting, since that is where the disagreement lives. And in which currency is the figure quoted, since for anyone outside that currency it may be the largest variable of all.
Answer those three and the table stops being a prophecy and becomes what it always was: an organised statement of what today's prices imply, if nothing surprising happens. Something surprising usually happens. That is not a reason to ignore the arithmetic, only a reason not to mistake it for knowledge of the future.
