Expert estimates of the equity risk premium (ERP) remained remarkably consistent across the first two decades of this century, a newly published CFA Institute research paper reveals.
The CFA report shows the most common ERP 10-year forecasts from a 2021 panel of financial luminaries including Cliff Asness, Rob Arnott and Roger Ibbotson converged at 4 per cent in line with findings from the inaugural event in 2001 and follow-up roundtable in 2011.
“It almost seems as though a 4% value is the financial equivalent of a cosmological constant,” the CFA ‘monograph’ says.
But the paper, which details the wide-ranging expert debate on ERP and associated issues, suggests the seemingly entrenched view on expected returns from shares could provide “false reassurance” to asset allocators.
“In other words, could it be that this 4% ‘cosmological constant’ is not really derived solely from actual forecasts but rather ends up being somewhat of a ‘goldilocks’ number that comfortably fits with a variety of investor hopes and institutional structures?”
While the average ERP estimate was anchored at 4 per cent, the paper titled ‘Revisiting the equity risk premium’ notes forecasts ranged from zero to more than 6 per cent in 2001 and from 1 per cent to 6 per cent plus in the 2021 follow-up event (which featured most of the same participants).
“… for those experts who provided estimates in both 2001 and 2021, those offering lower versus higher estimates in the first instance also did so in the second instance,” the CFA report says.
Given the reasonably wide range of forecasts and inherent biases of experts who make them, the paper says asset allocators and others should probably take a flexible approach to modeling ERP numbers.
“Whichever approach the analyst chooses, the results will be influenced by estimates and methods,” the report says. “In an era in which we observe macro forces acting on asset markets and changing premia regimes, it might behoove the analyst not only to use the consensus equity premium number as an asset allocation anchor but also to test allocation sensitivity to different estimates and methods.”
In an April 2023 update Asness, head of quant manager AQR, and Arnott, the founder of factor investment specialist Research Affiliates (or RAFI), revised down their previous 2021 ERP estimates by 1 per cent and 0.3 per cent, respectively, with both landing at about 3 per cent for the expected annual outperformance of shares above the risk-free rate over the next 10 years.
However, Arnott says that if asset markets revert to historical valuation means, the 10-year ERP will rise to 0.7 per cent from his current estimate of 0.1 per cent.
Several other expert panel members including Yale professor of finance, Roger Ibbotson, and Antti Ilmanen (AQR) also dialed down their ERP forecasts in 2023 compared to two years prior.
Apart from filing the latest ERP numbers, the 126-page roundtable transcript showcases an absorbing debate from some of the most-respected asset allocation experts in the business as they bat around various calculations, theories and thought experiments such as the following one posed to panel member and Morgan Stanley financial researcher, Martin Leibowitz, by a book collaborator, Sidney Homer in the 1970s.
Homer asked Leibowitz to calculate the value of one drachma invested by a Roman centurion compounded at 4 per cent over the intervening centuries – a hard job at the time but co-panelist, Laurence Siegel, supplied the answer in 2021 within minutes as 2.6 × 1034 drachmas.
“The next question was an even better question,” Leibowitz says. “Sidney asked, ‘What happened to it all?’”