In the first white paper, we showed that the standard CAPM fails in practice, we outlined the broad reasons why this is the case, and we described a high-level approach to solve the problem. To do that, we must first develop a full understanding of the drivers of certain economic and financial variables. We focus here on these drivers, explain in more detail why the standard CAPM fails, and lay the groundwork for the re-engineering required for it to succeed.
As a prerequisite to this work, we establish solid, consistent data on Treasuries going back to 1871 (similar to our work on earnings described in the second white paper). To this effect, we reconstruct 20-year nominal Treasuries and we back-cast 20-year Treasury inflationprotected securities, or TIPS. The approach differs for the two series because nominal Treasuries existed in 1871, whereas actual TIPS only came into existence in 1998. Please see Appendix C for the methodology and results.
This section contains ten main insights:
First, while the modern US economy is much less volatile than its forebear, it is subject to periods of acceleration and deceleration whose frequency has been largely invariant over a century and half.
Second, there is no difference between the real rate and the nominal rate until the mid1950s because long-term inflation expectations were zero until that time and the so-called Fisher effect was not yet operative.
Third, market-based breakeven inflation expectations, as measured by the difference between nominal rates and TIPS, are a poor measure of long-term inflation expectations because they also carry a host of unrelated risks.
Fourth, equity and Treasury volatilities have not responded in kind to the reduction in volatility of the economy. This indicates that investors now react to smaller signals than they would have in the past because their frame of reference is scaled to a world of much lower volatility.
Fifth, the instability of the equity-Treasury correlations over time and across the frequency spectrum is the core reason why the traditional CAPM fails in practice.
Sixth, TIPS have a complex, largely unrecognized, relationship with risk that requires disentangling. The results of this disentanglement are helpful as we develop a deeper understanding of both equity and Treasury markets, particularly so with respect to the latter.
Seventh, the market becomes more sophisticated as time passes: Investors discover how to price previously ignored risks and do so specifically by asset class, and learned behaviors, such as portfolio rebalancing, become more common. As a result, the primordial soup of a single “non-cash financial asset” class evolves into multiple asset classes with very distinct characteristics.
Eighth, the broadening of the motivations for holding Treasuries has subtly changed the meaning embedded in their price signal. This broadening leads to a negative risk premium, which economists call the “convenience yield.” The emergence of a convenience yield from the mid-1960s onward needs to be addressed and reversed in order to normalize the meaning of the price signal over the 150-year observation period.
Ninth, the combination of high inflation and high investor personal tax rates turns high real pre-tax interest rates into low post-tax ones.
And, finally, tenth, arbitrage is not 100% efficient, particularly when the market does not have a full understanding of what is being arbitraged.
In Section 2.2, we show the work that led to each of these findings to a level of detail that is consistent with their importance. In Section 2.3, we draw the implications of these findings from a technical modeling point of view. Readers who are most interested in the results of this work rather than its inner workings can skip ahead to Section 3.