Generally, history provides useful lessons for the future. For fixed-income markets, however, the recent history is downright misleading. To wit, let us compare the total performance of equities, as proxied by the S&P 500 Total Return index, and fixed-income markets, as measured by the Barclays Treasury Bond index.
Bonds have performed extraordinarily well since December 1999, returning an annual average of 6.1%, vs. just 1.7% for stocks. As a result, simple mixes of stocks and bonds have done remarkably well over this period. The same can be said for risk-parity approaches that leverage up bonds. This is only in retrospect, obviously. Very few investors expected this outcome; in fact, raise your hand if you were fully invested in Treasuries during this period.
We are not condemned to ignore history, however. Instead, we can use it to gain useful insights for prospects for fixed-income markets. These extraordinary gains were a direct result of the sustained decrease in yields over this period: Yields dropped to 0.9% from 6.5%. This drop of 5.6 percentage points represents an annual average drop of 43 basis points.
We can now use this performance information to decompose returns into their sources. From the drop in yields, we can compute an approximation of the price effect using duration. The Barclays index had an average duration of about 5.5 years over this period. Multiplying by the yield drop of 0.43 percentage points gives an annual price appreciation of 2.4%. Next, we turn to the income effect, which can be measured by the average yield of 3.4% over this period. Hence, we have decomposed the annual return on the Treasury index over this period into sources that add up to a number close to the actual return.
The same decomposition can be used to evaluate prospects for these markets. We start with the income effect, which we take as the current yield of 0.9% on the index. To simplify, round this up to 1%. This is the only thing we can be sure of, however. More importantly, we have to account for the price effect. It is fair to say that yields will not continue dropping at the same rate for an extended period, lest yields become negative. Eventually, nominal yields should compensate investors for expected inflation. Long-term implied inflation is currently around 2.5%. Therefore, adding 50 basis points for the real rate, it would be reasonable for yields to slowly revert from 1% to a 3% long-run value.
For a more detailed analysis, we can describe the distribution of future interest rates via Monte Carlo simulations of a single-factor model of rates with mean reversion. The starting point at time zero is the current rate of 1%. Next, we estimate the speed of mean reversion from historical data. This can be expressed in numbers of years required to close half the gap from 1% to 3%. In this case, it would take five years, on average, to move from 1% to 2%. The graph describes the evolution of average future rates, along with upper 95% and lower 5% confidence bands. The lower band seems realistic, as it includes the extreme situation of Japanese bond yields, which have been languishing for a while around 0.5%.
We can then compute the total return (income and price effects) on the Treasury index along each simulation path, maintaining the same duration at each point in time. This can be used to create a distribution of annualized returns for Treasuries over a fixed horizon, say the next five years. This analysis reveals unappealing results. The distribution has low expected returns and substantial downside risk. In addition, this analysis assumes that changes in rates will be incremental; in reality, we are likely to see sharp upward jumps in rates.
Overall, long-term prospects for fixed-income instruments seem bleak. Investors should be actively looking into other asset classes, away from directional interest rate risk.