The Long Run Is Lying to You

In his latest Cliff´s Perspective, The Long Run Is Lying to You, Cliff Asness is writing about the underperformance of value equity investing. I think this is a very interesting piece that deserves attention and the approach help explain a few other things important to asset allocation, in general.

Warm up: The US equity premium

Cliff starts by running a regression, explaining the realized US equity premium in 1950 – 2020 of 6.5% p.a. and finds that 1.30% p.a. (of the 6.5%) is purely due to valuation changes of stocks over the sample. He runs the following regression:

S&P 500 Return = 5.2% + 0.95 * change in valuation

The change in valuation is captured by a change in the Shiller CAPE. So, the realized equity premium is 6.5%, and the intercept is recorded as 5.2% and the rest (1.3%) of the realized equity premium is captured by changes in the CAPE over the sample. So, if you do not believe that valuation (or CAPE) is going to change from 2020 levels - going forward – a reasonably expectation of S&P returns will be 5.2%, and it would be wrong to use the higher historically realized equity premium of 6.5%.  This is also Cliff´s preferred expectation of the equity premium going forward.

US equity market versus “the rest of the world” – what to expect going forward ?

From 1990 to 2020 the US equity market outperformed the rest of the world, by around 4.4% p.a. Now, if you adjust for purely valuation changes in US stocks versus stocks in “the rest of the world”, the outperformance is reduced to a low 1.1% p.a.. The 1.1% of outperformance is explained by real earnings outperformance in the US. The rest, or 75% of the realized outperformance is explained by favorable valuation changes in the US - US stocks have simply turned more expensive compared to “the rest of the world”, driven by investor demand and the expectation that US stocks will be a better bet. In other words, if you do not believe this will extend into the future, you have a good argument for chasing stocks outside of the US. 

Underperformance of Value - explained ?

The value factor (HML) has only produced a return of 1.9% p.a. in 1950 to 2020, which is very low indeed. This gets worse if you look at the data from 1990 to the present with a 0% return. So, value as a factor seems to be worthless in recent times. Cliff argues that if you control for valuation changes of value stocks versus other stocks in the sample 1950-2020, the value factor actually give rise to a much more healthy 3.0% p.a. return. In other words, the low returns from value stocks is explained by the fact that value stocks have cheapened (are out of fashion with investors) relative to other stocks.  This effect is even more pronounced in the sample 1990 – 2020, where the return is 4.1% for value stocks (if you keep valuation changes over the sample constant). The question is: if you do NOT believe that value stocks will keep trading more cheap going forward (or if you believe some sort of mean reversion in relative valuation), you need to increase your expected returns on value stocks considerably going forward.