The CFA has produced an excellent paper; you can download it here.
Long-run stock returns look extraordinary, yet few investors became rich. This brief explains why dividend reinvestment assumptions overstate historical equity returns, reframing the equity risk premium as “the return nobody got.”
Equity markets have delivered remarkable long-run returns on paper, yet those gains rarely translated into lasting, generational wealth for most investors. “Stocks for the Long Run Revisited: Dividends and ‘The Return Nobody Got’” examines this disconnect by rethinking how long-term equity performance has been measured and, more importantly, how dividends are treated in historical return series.
Dividends have accounted for a substantial share of equity returns over at least the past century. Standard total-return calculations implicitly assume that dividends are continuously reinvested, whether back into the issuing stock, into a perfectly tracked index, or into other securities. These assumptions are analytically convenient, but they describe an investment behavior that was largely unavailable, impractical, or uncommon for most investors before the rise of low-cost index funds.
Drawing on insights from Rob Arnott, Hendrik Bessembinder, Edward McQuarrie, Roger Ibbotson, Jeremy Siegel, and Elroy Dimson, the brief shows that dividends were often consumed, taxed, or reinvested imperfectly. Transaction costs, sales loads, bid–ask spreads, delayed reinvestment, portfolio drift, and management fees materially reduced realized outcomes. When dividends are treated as cash flows that fund consumption rather than as capital that compounds indefinitely, long-run equity returns fall sharply, and the implied equity risk premium narrows accordingly.
The brief also explores how declining transaction costs and the advent of indexing reshaped the economics of investing, contributing to higher equilibrium equity valuations in recent decades. These structural changes help explain why historical averages may offer limited guidance for future expectations.
Ultimately, this brief reframes long-run equity returns as a measure of asset-class potential rather than a record of typical investor experience. By distinguishing theoretical returns from realizable outcomes, it offers a more grounded perspective on equity performance, valuation, and the true nature of the long-term equity premium
These are personal reflections on the idea of a equity risk premium.
I am not a CFA. I am only entitled to put two of those letters after my name, so this (a non-investment professional opining on investment) feels as some accountants may feel if non-accountants critique current accounting, which seems to me to be long on disclosure but short on clarity and relevance.
I was a pensions trustee for 35 years.
For the first fourteen years I listened to actuarial investment consultants and some asset gatherers talk about risk premia. In my head I (and most fellow trustees) felt: what is this? We can’t pay pensions with “premia”. We pay pensions with cash from investment.
You recalled from John Hamilton’s Melton Mowbray talk of a year ago his anecdote of the fund manager who believed a return of minus 26% when markets had fallen 28% was a good outcome, deserving of a performance fee.
I won’t name that fund manager, who’s still active today, but his view was fairly typical at that time.
We didn’t pay pensions with relative returns, however, but from absolute returns.
Mr McCaffrey has had the benefit of being around the CFA’s Equity Risk Forum which has been meeting and discussing since 2001 among a group of leading investors, finance experts, and academics.
My problem with ideas of an equity risk premium is that it is defined as the expected excess return on equities relative to a risk‑free asset. But this expectation is never directly observable and must be inferred from models, surveys, or history.
This makes it closer to a modelling construct than a directly measurable economic quantity, and different reasonable methods can yield different numbers at the same point in time or even over time.
I spent my last twenty-one years as a pension trustee believing in a different form of attribution analysis from that which many actuarial investment consultants used.
Roger Ibbotson is one of the academic members of Mr McCaffrey’s Forum.
My fellow trustees and I stumbled across what for us was a far more useful method of decomposing investment returns. This was very similar to Ibbotson’s decomposition of returns, primarily developed by Ibbotson and Peng Chen (2003, updated by Straehl & Ibbotson, 2017), which breaks down long-term equity returns into foundational supply-side components, rather than relying solely on demand-side market price movements.
The Ibbotson framework links stock returns to the cash flows they generate for investors, separating returns into inflation, income (dividends/buybacks), and real growth in valuation fundamentals (earnings, P/E ratios, book value).
We applied the analysis not just to listed equity investment mandates but also to mandates in other asset classes, including bonds/debt/gilts, property and private markets.
Instead of mandates being set relative to market indices, we tendered our investment mandates in terms of initial yield, expected yield and expected growth. We also monitored subsequent performance that way too, using money-weighted rates of return rather than the time-weighted rates of return preferred by most actuarial investment consultants.
Ideas of there being a “risk-free” asset against which relative returns from other assets can be measured was, for us anyway, unhelpful. And in the UK,hopefully, the very idea has become discredited through the LDI crisis of 2022/23?
You pay pensions with cash realised from absolute returns, not with relative return “premia” or discounts.
I couldn’t get the Slideshare link to McCaffrey’s report to work, but it’s available here too:
rpc.cfainstitute.org/sites/default/files/docs/research-reports/rf_mccaffrey_stocksforthelongrun_revisited_online.v2.pdf
Thanks to Derek introducing us to this a year ago, our DB shared ambition scheme is now viewing our investment decisions against a Decomposition Analysis separating out:
* Current yield (dividend and interest effectively received into the trustee’s bank account to pay benefits or in our case reinvest elsewhere);
* Growth and inflation to the Current Yield
* Revaluation Yield (basically everything else reflecting the expected future realisation proceeds over the initial investment or current value)
We have set a Scheme wide cash target for Current Yield and assess Growth and Revaluation Yields against our valuation assumptions.
Picking up on Paul McCaffrey’s point about dividends being the return nobody got, we are suggesting that the Revaluation Yield forecasts should be further broken down into 5 accumulating components:
1 Revenue growth based on growth forecasts of the underlying investments
2 Profit margin representing the underlying investments’ ability to turn revenue per share into earnings through profit margins.
3 (Share Count) The effect of expected changes in the Fund’s holdings in underlying investments arising from realisations or additional investments, or the dilution of holding through external capital issue.
4 Valuation impact based on expected changes in price-to-earnings ratios.
5 Dividend yield estimates based on current levels, adjusted for any expected changes to the amount of cash companies return to shareholders.
Perhaps this is a case of one (Scottish) chartered accountant talking to another non CFA (although in this case one with the same letters in a different order)!
The difficulty we have found is getting our investment consultants and fund managers to present proposals on this basis (we find extremely difficult to overcome liability discount rate matching bias). So yes, a Pensions Playpen on the topic would be extremely valuable.