Grice is a portfolio manager at Switzerland-based Calibrium AG. He previously served as an investment strategist at Societe Generale, and he spoke at that firm’s annual investment conference in London on July 9.
1. Value investing is an intellectual fraud
“What is investing if not for value?” Grice asked, rhetorically. He defined traditional value investing as betting with the odds in your favor or buying dollars for $.75.
The “value” adjective is not necessary, Grice said. “It is like fast sprinting or wet swimming.”
Grice then drew a distinction between fundamental, Graham and Dodd-style value investing, and factor, or quantitatively-based, strategies. He acknowledged that the 1992 Fama-French research showed that you could exploit statistical patterns of cheapness.
He then presented data comparing the small-cap Russell 2000 index relative to its value counterpart, going back 15 years. The annual value premium has been -44 basis points, Grice said. It is not just the Russell indices where value has failed. He said that using the MSCI world index data, the value premium has been -38 basis points over the same period; with emerging markets it was -12 basis points.
It is widely known that growth stocks have outperformed value in the U.S. over most of the last decade, and that value has been the winning strategy over longer time frames. Grice’s analysis was important because he compared value to a broader index and over multiple markets, using only the last 15 years.
But what he said next should be carefully considered by devotees of quantitative investing.
The value premium is not like the liquidity premium, Grice said, which inherently justifies a higher return for its risk. The same is true of the premia associated with credit risk or duration.
“Why should I get persistent risk premium for a cheap multiple?” Grice asked.
“Value investing is far too easy,” he said. “Anyone with a Bloomberg for Factset terminal can do this. It was a historical anomaly because it was cheap.”
“It’s now gone,” Grice said. “The value anomaly has been arbed [arbitraged] out. Value does not equal cheap.”
Quantitatively-based value investing is not the same as fundamental research, he said. Fundamentally-based investing is “hard,” he said, “and it’s got to be harder than quantitative analysis.”
The challenge for advisors is that there is no way to conclusively prove Grice’s claim. We know that quantitatively-based value strategies, unlike the broad capitalization-weighted index, cannot be pursued by all investors. Eventually capital flows to value strategies must erode returns, but we have no way to know for sure when this will happen – or if it already has.
Grice’s assertion should not be dismissed. One cannot be certain that value will “revert to the mean” and resume its long-term outperformance relative to growth the broader market. We’ll have to await further research to see the extent to which asset flows to value strategies, which have been substantial over the 15-year period he studied, have eroded returns.
This is really an interesting claim, and one that I (and I’m sure many others) have wondered for years now. There’s another important piece of this claim that is not discussed in this piece – idk if Grice addressed it or not. That is, what if the value premium is gone? What does that mean for investments based on that premise? What should an investor expect from that portfolio relative to a growth portfolio or a balanced portfolio? My question really concerns forward looking returns. If the value premium is gone, does that mean value stocks and all the rest of the stocks now have the same forward looking return? If this is the case, then value investing may not be an advantage, but it may also not be harmful.
2. Absolute return is a myth (everything is relative)
It is generally understood that valuations, using metrics such as the Shiller CAPE ratio, are predictive of returns over long time horizons. The U.S. market is in the top quintile of historical valuations, Grice said, implying a sub-2% real return over the next decade.
“That seems clear cut and obvious,” he said, “but look closer and you see cracks.”
Grice provided data on the range of those return forecasts. Top decile valuations imply returns ranging from -2.4% to 8.7%. Thus, for the next 10 years one could “quite reasonably” make 8.7%, he said. “So how practically useful is this information? It should be a warning sign.”
Using U.S. equities, Grice then compared a buy-and-hold strategy to one based on market timing. His timing model was based on the CAPE ratio and it adjusted allocations depending on the quintile of historical valuations (he did not provide more specific details). His market timing model beat buy-and-hold, he said, but the bulk of the returns came in 1920s.
“There was no value added in the last 70 years,” he said. In addition to the S&P 500, Grice said that finding held for the FTSE 100, DAX 30 and Nikkei 225.
“Valuations matter,” Grice said, “but they’re difficult to measure and get right.”
Forward-looking return estimates are better than naïve extrapolations, he said. But the problem with focusing on U.S. equities is that there have been only seven signals in 100 years, based on valuations being in the top or bottom quintile. “There is not enough data to have statistically significant results,” he said. “The problem is that we don’t get enough movement.”
But if you look at signals across multiple asset classes, you get more useful data. Grice has analyzed valuation-based strategies using bonds and equities. He constructed a model that determines the equity-bond allocation based on the relative valuations of the two asset classes, and said that it outperforms a naïve buy-and-hold strategy. (He said his model used mean-variance optimization based on Sharpe ratios, but did not provide additional details.) Indeed, he said, his relative performance-driven portfolio “did very well.”
“The importance of valuation reveals itself when you add more asset classes to the portfolio,” he said. He has tested his findings for markets in Japan, across Europe and in different European countries. “Valuation is phenomenally powerful but only on a relative basis,” Grice said.
Right now, he said, equities are attractive relative to bonds. “Equities are not that bad,” he said, “given the unattractiveness of the opportunity set.”
OMG I could not love this more. I had the same issue trying to make an investment model out of CAPE. My conclusion was that 7 signals in 100 years can only lead to a model that is data-mining, in addition to being basically useless. I wonder how this compares to Meb Faber’s dynamic allocation model?
His third point is “there is no bond bubble.” You can click on the link to read that part, bonds are boring.