William Bernstein Discusses Tilting Hey, rjb.
I get where you're coming from, but Bernstein was attempting to write a popular statistical model of a portfolio that would approach the Efficient Frontier, while claiming that, since you can't know future asset class performance, you can't know where the Efficient Frontier will lie. In order to do that, he looked at historical asset class performance and correlation. For foreign, since he's writing in the late 90s, the only thing he has long term data for is the EAFE. For small he uses the CRSP 9-10, which gets much farther into microcap territory than VTSMX.
When he starts talking about asset allocation in Chapter 5, he admits up front he is "an asset class junkie," and is willing to own "20 or 30" different asset classes. But the rub is that he wants everything to be a separate asset class. We have 15 years of hindsight and recency bias showing an across the board increase in correlations. For instance, Bernstein approaches bonds as a risk control tool, and assumes correlations of .777 between SC and the S&P 500, and .483 between the EAFE and the S&P 500. It's not so much that Bernstein doesn't want to use total market indices, but that doing so doesn't allow him to really make his broader point re: diversification because there is a lack of data. He ends up using the European stocks as a proxy for the EAFE because there wasn't a Vanguard DM fund yet.
The portfolio Ted points to, he calls the "Level-One Asset Palette," and he designs it for those who find "reading this book ... the equivalent of root canal work." Quickly after he presented it as his Lazy Portfolio. In the book he presents second level and third level "palettes," which include EM, Foreign smallcap, REITs, Natural Resource stocks, short term bonds, TIPS, foreign bonds, and valuation factors.
Not sure if that helps or not, but that's what I gather his reasoning is. Personally, if you're going for as much growth and diversification across 4 asset classes as you could easily get, I would think something like CRSP 9-10 (VB or VBR), foreign small (VSS), Real Assets (VNQ, RWO, VDE, or ALPS), and either an intermediate or hedged foreign bond fund (BND or BNDX) would be better. But some of those funds didn't exist 15 years ago.