Practical analysis for investment professionals
20 October 2015

The “Relatively” Easy Way to Forecast Long-Term Returns

Posted In: Drivers of Value

Long-term returns are relatively easy to forecast.

Short-term returns are dominated by randomness, but long-term forecasts for most asset classes can, in part, be derived mathematically (give or take some arguing about the assumptions).

But why bother with long-term return expectations — for example, 10-year forecasts? For most multi-asset managers or tactical asset allocators, 10 years is an eternity. Investment managers are judged on much shorter time frames.

For asset owners or asset managers compiling a strategic asset allocation, however, long-term forecasts are relevant and necessary. When combined with estimates for risk and correlation, these forecasts allow investors to fine-tune their long-term benchmarks and consider trade-offs between asset classes to enhance the implied risk and return profile of the fund.

In the following table, I have aggregated the results from three major asset managers — JP Morgan, Northern Trust, and BNY Mellon — that publish their long-term return forecasts for major asset classes. Here are the average expected returns:

Average Long-Term Return Forecasts

Asset Class Average Forecast (per annum)
US inflation 2%–2.5%
US cash 2%–2.5%
US 10-year bonds 2%–2.5%
Commodities 2%–3%
Hedge funds 4%–5%
US equities 6%–7%
Global equities 6%–8%
Private equity 8%–9%


Let’s think about how these estimates are derived and whether they are realistic.

Fixed-income securities are the obvious starting point. If we buy a 10-year Treasury today with a redemption yield of 2.5% and hold it to redemption, we know that the return will be 2.5% per annum (assuming that the US government doesn’t default).

The Return from US Equities

Now, let’s consider US equities. The simplest expression of the truly long-term return from US equities follows a classical formula, as described by Richard Grinold and Kenneth Kroner:

Long-term return from equities = Dividend yield + Inflation + Real earnings growth

Long-term return from equities = 2.0% + 2.25% + 2.25% = 6.5%

So, at first glance, if you believe the assumptions — that inflation will be around 2.25% and that dividends will grow pretty much in line with long-run GDP expectations — then the forecast above is reasonable. What’s not to like? Let’s unwrap this in more detail.

First, should we adjust for buybacks? In reality, the payback to long-term (buy and hold) investors will be both in dividends and in capital return from share buybacks. It’s reasonable to assume that substituting buybacks for dividends makes no substantive difference to total long-term returns, although some of the publications linked in this post explore the building blocks behind this in impressive detail.

Second, is it reasonable to assume that dividend growth (or earnings growth) will keep pace with the real economy? Can the profit share of GDP hold at its current level? A recent report from McKinsey & Company is forecasting that more competitive world markets will trigger a 20% fall in global profit share by 2025.

Also, even if profit share holds near to recent highs, can the companies that currently make up the index maintain their own profit share as new players and technologies emerge? My personal expectation is that earnings growth will not match real GDP growth in the long run. You may have your own view.

Third is the question of equity market valuation. If we are considering a finite time horizon (let’s say 10 years), then our formula above only holds if the dividend yield remains constant. If it is likely to change, we need to make a valuation adjustment.

It is for this reason that the estimate of long-term US equity returns from our fourth research publication is starkly different from those above. Rob Arnott’s team at Research Affiliates forecasts that, over the next 10 years, the valuation of the US equity market (as measured by the Shiller CAPE ratio) will revert halfway back to its long-term average. This implies a valuation adjustment of 2.4% per annum. When added to a dividend yield of 2% and their estimate of dividend growth of 1.4% per annum, this gives a prospective 10-year total return from US equities of just 1.0% per annum.

Whom do you believe?

Visit Savvy Investor for other articles on long-term return forecasts or asset allocation.

If you liked this post, don’t forget to subscribe to the Enterprising Investor.

All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.

Image credit: ©

About the Author(s)
Andrew Perrins

Andrew Perrins is CEO and co-founder of Savvy Investor, a professional network for institutional investors that curates pension and investment research and white papers from around the web. After qualifying as an actuary in 1987, Perrins worked as director of global asset allocation for Abbey Life and Chase Manhattan Bank.

7 thoughts on “The “Relatively” Easy Way to Forecast Long-Term Returns”

  1. Aleksei K says:

    If you invest your whole portfolio in a single 10Y Treasury bond today (let’s call it intermediate gov bond asset class), in 8 years you will only hold 2Y treasury bond, so your portfolio will drift from intermediate gov bonds to short term government bonds, and hence your annualized return of 2.5% percent will show return of a portfolio which drifts from intermediate treasury to essentially cash (before maturing). It will be by no means a return of a 10Y treasury bond asset class. Now, a fund investing in intermediate gov bonds will roll low duration securities into new 10Y securities to maintain asset class structure. The resulting return will be a mix of effects from yield curves and paths, and most definitely will not be 2.5%.

    Pretty much the same thing, i.e. asset class drift, will happen to most small cap, or growth stocks or any other portfolio without rebalancing.

    And while 1% return for equities over the next 10 years is quire believable (given current valuations), there is a good deal of chance that returns exactly 10 years from now will be anything BUT 1% (nor 6-8%). So anyone really wishing to have a believable forecast of what will be long term return of various asset classes would require much more sophisticated tools.

    But then, “I’ve see rich businesspersons; I’ve see rich speculators; but I’ve never seen a rich forecaster.” @nntaleb

    1. Agreed – the return from maintaining a portfolio of 10Y Treasuries will certainly be different from that of a 10Y bond held to redemption. I wasn’t seeking to describe how to forecast the former – as you say that it more complicated.

      The publications referred to above include varying levels of “sophistication” in their forecasts. The core building blocks are similar, but it’s impossible to write a short article like this without leaving out a whole load of interesting detail.

      In practice of course, forecasts will be too high or too low. It may have been instructive to add confidence intervals to these forecasts; e.g. the 10Y return from equities is 6.5% pa (plus or minus 5% pa with 90% confidence) but then we would get into a discussion of the return distribution of 10 year returns, which is a fascinating discussion, but a different topic altogether.

      1. Aleksei K says:

        I appreciate your reply.

        Investment profession is structured in a way that discourages practitioners to discuss practical matters (how they really analyse, forecast and invest). On the other hand, you can see terrible forecasts from “gurus” and prominent bank analysts all around you all the time (which obviously doesn’t add credibility to profession).

        I wonder why CFA (which is probably the highest standard in investment profession today) generally stays away from any practical discussion. The exam itself is mainly theoretical and any scent of applicability is usually from years ago.

        I presume exam material will stay theoretical forever and it’s fine. But beyond the exam, why not write about practical issues instead of rewriting once and again theoretical textbook material?

  2. Jeremy S says:

    This is such a critical subject – can I suggest you cover it again in more details. There are few things more important in long term investment management than understanding the difference between the views of, say, JP Morgan, BNY and Northern Trust on the one hand and ,say, Robert Arnott and GMO on the other.

    Issues which are critical to me are:
    1) If you adjust dividend flow for share buybacks then surely you should adjust for stock issuance?

    2) Growth historically has disproportionately come from smaller unquoted companies. Given this and a likelihood of profit share of GDP returning to mean, dividend growth may considerably undershoot economic growth ?

    A cynic might argue that many houses simply use expected numbers sufficiently high to justify the equity exposures and – as stated above- no one can really know. Nevertheless, the debate is critical. Perhaps you can invite a proponent on each side to set out their view in detail.

    1. Hi Jeremy, thank you, I agree it’s difficult to do justice to the topic in a short article. The two issues you highlight are critical, as you say.

      In terms of the first issue, we have some good articles on the Savvy Investor site at – I’d suggest starting with the Research Affiliates article which directly addresses your point on issuance versus buybacks.

Leave a Reply

Your email address will not be published. Required fields are marked *

By continuing to use the site, you agree to the use of cookies. more information

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.