I would caution views that the equity premium is more immune to underperformance than the value premium. According to the earlier linked Fama-French paper Volatility and Premiums
– even over a 50-year investment lifetime, an investor has a one-in-twenty chance of a negative realized equity premium, which is even higher than the one-in-thirty five chance of a negative realized value premium (if you believe the Fama-French research). And over the last 16 years (2000-2015) Vanguard Total Bond Market has outperformed Vanguard Total Stock market by 0.65% annualized ( 5.11% vs. 4.46%) obviously highly period dependent – but that’s the point – it can happen.
And if I look at the table Taylor posted for the 15 years to 6-30-2000, Small cap value returned 11.3% while large cap blend returned 15.2%. And Larry posted the returns 15-years to date with SV = 9.9%, S&P500 = 4.8%. So across the two periods SV had returns of 11.3% and 9.9%, while large cap = 15.2% vs. 4.8%. To me, SV provided more consistent returns over the two periods, even if it underperformed LB in the 15 yrs to mid-2000 (I will take an 11.3% return any day). Over these periods the value premium had low correlation with the equity premium providing a diversification benefit (i.e. lower deviations in returns over the two periods). However, the challenge is that correlations across factors premiums varies over time, and in some cases the correlations become increasingly positive just at the wrong time – as in 1929-1932.
Here are some returns over three bear markets: 1929-1932, 1973-74, 2000-2002, together with returns for the 10 years since the start of the bear market.
: LV and SV declined by more than the overall market (deflation increased the real cost of debt which is higher in LV companies, and even more so in SV companies). Over the 10 year period from 1929-1938, nominal annualized returns of LV was still negative but better than the overall market (-0.9%, and -1.6% annualized), and both had positive real returns given the -2.0% inflation rate over the period. Small Value lagged both, with a -6.7% nominal annualized return. In sum: large value had better performance than small value over this period, declining less in 1929-1932 and having higher 10 yr annualized returns from 1929-1932 than both SV and the overall market.
: LV declined by less than the overall market (-12.4% vs. -22.7% annualized), while SV decline by slightly more than the overall market (-24.3% vs. -22.7% annualized). Given the +10% inflation rate over this period, real returns were much worse. Over the 10 year period 1973-1982, SV and LV both outperformed the market (20.2% and 13.5% vs. 7.5% annualized return). And with the +8.7% annualized inflation rate over this period, the real annualized market return was still negative after 10 years. In sum: in this period, LV provided greater downside diversification in 1973-74, while SV had higher annualized returns over 1973-1982.
: Both LV and SV had positive annualized returns over this period (+0.2% vs. +12.1%) while the market declined (-14.7% annualized). Over the 10yr period 2000-2009, the annualized nominal market return was -0.9%, which was lower in real terms given the +2.5% annualized inflation rate, while the return from LV and SV were +4.0% and +12.7% respectively. In sum: in this this period SV provided greater downside diversification in 2000-02, and provide higher annualized returns over the 2000-09.
Annualized return (%)
TSM = CRSP1-10
LV = Dimensional US Large Value Index
SV = Dimensional US Small Value Index
What do I conclude from the above?
: The diversification benefits of LV and SV have varied across market downturns and over different timeframes (i.e. during actual downturn vs. 10yrs from start of downturn). LV provided better 10yr diversification from 1929-1938 than SV, but both amplified 1929-1932 declines. LV provided better diversification in 1973-74, while SV had best 1973-1982 returns. SV provided best downside diversification in 2000-02 and higher 10 year returns from 2000-09. Given these differing effects, personally, I take a middle (value) path - with 50:50 LV:SV type combination – rather than all LV or all SV (or all TSM). For what its worth a 50:50 LV:SV combination had higher annualized returns than each individually in the 17 years since 1927 when inflation was above 5 percent. This broadly equates to a value load about double the size load. And adding longer term bonds (personally would not extend beyond intermediate) can help in 1929-1932 type periods.
We each need make our own choices. If TSM is what you are comfortable with and can stick with it - then go with that - but would caution thinking that the equity premium is more immune to underperformance than the value premium, as this could lead to disappointment.
Obviously no guarantees.