Thursday, July 17, 2014

Risk & Valuations, Part 11: Allocations with Stocks & Bonds

I previously published a series of posts where I concluded that, in practice, nominal fixed-rate bonds have not provided any diversification benefits in the classic bonds & stocks portfolio.  How does the model of risk and valuations in this current series of posts mesh with that idea?

I apologize for a little sloppiness here.  I am going to separate risk premiums out into the unlevered Equity Risk Premium (UERP) (the additional premium earned for holding equities over debt), real risk free interest rates (RRFR) (the premium earned for holding debt), and the inflation premium (the premium earned for the expected decay in the value of cash).  UERP comes out of my annual Black-Scholes valuation estimates for non-financial corporations from the Federal Reserve's Z.1 report.  RRFR is the rate on 10 year treasury bonds, minus inflation.  Ideally, for the inflation premium, I would have some indicator derived from inflation protected bonds, but since I don't have that data for a long time frame, I am just using the GDP price deflator.  In practice, this seems to be a decent approximation over the long term, but it probably adds a bit of point-to-point noise.

Here is a graph of these three premiums, stacked.  Even with my noisy estimates, the total required returns to unlevered equities (RRFR + UERP) appear to have a steady trend around 8%.  (The RRFR + UERP data series doesn't appear to reject the null hypothesis of an Augmented Dickey Fuller stationarity test.  But, a regression of UERP against inflation and RRFR returns a negative coefficient on RRFR of about -0.4 with a p-value less than .001.  In other words, there has been a strong negative relationship between UERP and RRFR, which should tend to stabilize the combined value as RRFR fluctuates over time, but with some possible fluctuations over time.)  A trend like this is what we would expect if there is a relatively stable return to productive assets, and the relative returns to RRFR and UERP reflect risk trading between debt and equity holders, as I described in the early sections.

There isn't a statistically significant relationship between inflation and UERP.  This makes sense, because the expected growth rate in net earnings should roughly adjust along with expected inflation.  In theory, unleveraged returns on assets should be roughly neutral to inflation.  In fact, this data presents weak evidence that UERP is lowest at inflation expectations between about 2% and 4%, and rises at inflation levels above and below that.  This is the relationship I previously found between inflation and RGDP growth.

So, this suggests that bonds and equities might provide beneficial diversification if real, fixed rate bonds are used instead of nominal bonds.  This makes sense if considered in the paradigm of the asset allocation process I described in Part 5, where a portfolio of stocks and bonds is rebalanced to match corporate leverage.  To the extent that changes in RRFR and UERP reflect their negative relationship, that portfolio should experience more stable returns as a result of its stock & bond allocation.

As discussed in the earlier series of posts regarding stock and bond allocations, this would explain the apparent benefit of home ownership.  Homes provide a sort of inflation protected return, as pre-paid rent, so they mimic the return on a RRFR portfolio allocation.  On the negative side, homes require a highly leveraged, non-diversified position in a particular real estate asset that is highly correlated with many risk factors a given household would be exposed to.  And the asymmetric risks of a mortgage also work against the stability of the household portfolio.  On the positive side, the inflation protection homes provide is probably more highly correlated to a household's local cost of living than a TIPS bond would be.  Also, limited access to homeownership tends to create above-market returns to home ownership.  This was probably not the case in the 2000's, but has generally been the case, including in the present market.  If home prices do manage to climb another 20 or 30% in the current recovery (as they should), many households might be better served with TIPS bonds in place of home ownership, as the advantage of above-market returns would again be reduced.

I think the problem with nominal bonds in a diversified portfolio can be explained with the risk trading model.  The inverse relationship between RRFR and UERP can be explained with risk trading.  But, inflation has tended to follow a path unrelated to RRFR and UERP.  Rather, it is a product of Federal Reserve policy, which seems to follow its own long-term biases.  And trends in inflation have tended to cover very long time periods in the mature Federal Reserve Era.  Inflation has generally been fairly low, except for a 20 year period from about 1965 to 1985.  So, mean reversion from the influence of inflation has not happened except in very, very long holding periods.  This has meant that the effect of inflation has been to push relative portfolio performance to the extremes for portfolios in a typical investor life-cycle.

This data only goes back to about 1960.  In the period from after the stock market crash of the late 1920's until the stagflation period of the 1970's, the experienced equity premium was extremely high.  This probably resulted from a combination of (1) very low bond rates throughout the period, (2) very high GDP growth rates, and (3) required equity returns that started out very high and declined over time, providing additional capital gains.  Gains over time come, in part, from the required returns that were in place at the time of the initial investment, but they also come from changes in required returns over the holding period, which provide trading gains.  Since much of the trading gains in the 1990's were related to falling risk free interest rates, both bonds and equities did well in the 1990's.  There was a small increase in the premium experienced from equities, but not as much as one would think simply by looking at the returns of the S&P 500.  Rates continued to fall in the 2000's, and ERP was high, but relatively stable.  This, combined with the fact that equities took the brunt of the demand shocks experienced in the two recessions, meant that experienced equity premiums in the 2000's were negative compared to bonds.  All portfolios, especially those initiated near the bottoms of the recent equity collapses, should experience very high equity premiums again over the next couple of decades as the pendulum swings back on relative RRFR and UERP levels, but continued demand shocks might make this a bumpy ride.  Every expert at the end of the bar seems to think every sign of progress is a bubble, and respected economic pundits refer to "new highs" in the stock market as if that is a problem to be solved, even as equity holders over the past 10 to 20 years have had Great Depression level returns relative to bonds.  There's a strange, self-destructive mood in this country, and so I'm afraid we might be in for some more demand shocks.  Except for that risk, we are probably in the midst of another decent period of high relative equity returns.

No comments:

Post a Comment