Revisiting Beta: How Well Has Beta Predicted Returns?
After the capital asset pricing model (CAPM) was developed in the 1960s and 1970s, financial researchers started to test how well this theoretical model actually worked in the real world.
Amid expanding computing power and greater data access, the 1980s became a critical era for gauging CAPM’s validity as analysts explored beta’s effectiveness in anticipating future returns.
Surprisingly, the general consensus that emerged was that beta’s return forecasting power was pretty weak.
In the 60 years or so since CAPM emerged, how well has the model and beta anticipated returns over the decades? To find out, we analyzed every firm that has traded on the NYSE and NASDAQ and built portfolios of companies based on their systematic risk (beta) using monthly returns and a 12-month rolling calculation.
If a firm had a beta under 0.5, it was allocated to the low beta portfolio. Firms with beta greater than 1.5 were allocated to its high beta counterpart.
Using these groupings, we examined how the portfolios performed over the subsequent year — both on a median and market-cap-weighted basis. Portfolios were then reconstructed according to new beta calculations each year.
|High-Beta Portfolio Median Return||Low-Beta Portfolio Median Return||High-Beta Portfolio Market-Weighted Return||Low-Beta Portfolio Market-Weighted Return||Percent of Years in Accordance with CAPM|
It turns out, the 1980s were a terrible time for beta. On an annualized basis, a low beta portfolio performed 6 percentage points better on average than its high beta counterpart over the decade, generating an 18.14% vs. a 12.12% return
We then examined the percentage of years that reflected CAPM predictions on an ordinal basis throughout the decade. In only four out of the 10 years did CAPM accurately forecast returns. That is, positive market return years should correspond to high beta beating low beta portfolios and negative market return years to low beta beating high beta portfolios. This means that CAPM did worse than a random walk over this time period and helps explain why researchers of the era were so skeptical of the model.
But the 1980s were something of an outlier. As the decades progressed, beta and CAPM become a better predictor. From 2010 to 2020, CAPM was right in 10 of the 11 years.
Indeed, in every decade since the 1980s, a high beta portfolio generated slightly more than 5 percentage point premium over its low beta peer on an annualized basis. That is, the high beta portfolio averaged a 15.53% return compared to the low-beta’s 10.34% return.
All told, the results highlight that beta is not as bad a predictor of future returns as is often thought. The 1980s were a terrible time for beta and CAPM, but since that decade, beta has been a decent forecaster of future returns.
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: ©Getty Images / Eskay Lim / EyeEm
Professional Learning for CFA Institute Members
CFA Institute members are empowered to self-determine and self-report professional learning (PL) credits earned, including content on Enterprising Investor. Members can record credits easily using their online PL tracker.
7 thoughts on “Revisiting Beta: How Well Has Beta Predicted Returns?”
I cannot reproduce your results. They seem completely wrong.
Backtesting using Portfolio123, my results are very different from yours. I’m seeing this: 2000 to 2010, low beta 14.65%, high beta 9.12%; 2010 to 2020, low beta 10.29%, high beta 8.05%. I’m using the exact same parameters as you are: all NYSE and NASDAQ stocks, rebalancing once a year, measuring beta with 12 one-month returns.
I have done a lot of other tests as well, and the results are consistent with the above, or even more extreme in favor of low beta. I tried using rolling backtests, varying the limits from 0.5 and 1.5 to 0.8 and 1.2, varying the beta measurements to weekly beta, weekly three-year beta, weekly five-year beta. Every single test I’ve run gives low beta stocks the advantage.
In fact, ever since 1973, researchers have found that low beta outperforms high beta, not only in equities but across all asset classes. The reason for this is mathematical. See https://backland.typepad.com/investigations/2018/06/why-low-beta-outperforms.html
There is s reason that a certain stock is a high beta stock. Stock is more sensitive than the market. This article justifies the notion.
It would be interesting to see what caused the breakdown of CAPM in the 1980s.
Just back-of-the-envelope using your table, but the average level of interest rates (10yr) and their volatility seems to have a strong relationship with the hit rate (Percent of Years in Accordance with CAPM). Clearly not a lot of data points, however.
Your conclusions are very surprising as there has been a lot of other academic research which reach very different conclusions.
Surely you know about Eugene Fama and Kenneth French’s 1992 study in which they put virtually all U.S. stocks into beta deciles for the period 1963-1990. Per Burton Malkiel’s book, A Random Walk Down Wall Street, “The remarkable result,…, is that there was essentially no relationship between the return of these decile portfolios and their beta measures.” Other studies have reached similar conclusions as regards other non-US markets.
Hence, the (relatively) recent introduction of minimum volatility equity portfolios – which are also low beta. (“Recent” relative performance has been subpar, though, in line with your past decade results.)
Any particular reasons why beta predictions differ in the 1980s as compared to 2010-2020?
Thank you for this article !
Did you run the analysis by industry, by High/mid/small cap, leverage, ROCE, liquidity level on the market or any other relevant metric ? And did you find any relevant correlation ?
Also, do you know why the 80s were different from any other decade ?