Last Updated Feb 17, 2010 9:47 AM EST
To review, GNMAs carry higher yields because their maturity is uncertain, and the uncertainty always works against you. If rates fall, the expected maturity shortens and reinvestment risk increases. If rates rise, the expected maturity lengthens and price (and inflation) risk increases.
However, looking at asset classes or investments in isolation isn't sufficient; you should look at how their addition impacts the risk and return of your entire portfolio. Looking at things in the whole is the only right way to view things. We can do just that using Morningstar's database, which goes back 20 years.
During this period (1990-2009), the Vanguard GNMA Fund returned 7.0 percent with a standard deviation of 4.6 percent, and five-year Treasuries returned 6.7 percent with a standard deviation of 6.2 percent. We will compare two portfolios that are rebalanced annually:
- Portfolio A has an allocation of 60 percent to the S&P 500 Index and 40 percent to Vanguard's GNMA fund.
- Portfolio B has an allocation of 60 percent S&P 500 and 40 percent five-year Treasuries.
- Portfolio A returned 8.2 percent per year with a standard deviation of 12.2 percent.
- Portfolio B also returned 8.2 percent but did so with a standard deviation of 11.6 percent.
What is important to note is that this 20-year period was a relatively favorable one for GNMAs. Interest rates were relatively stable, and there was never a spike in inflation that would have caused rates to rise (and maturities to extend) as there was in the 1970s and early 1980s. If data were available going back that far, it seems likely that comparison would have been even worse.
Consider that the standard deviation of the five-year Treasury note was 6.2 percent per year from 1990-2009. However, from 1970-1989, the volatility of the five-year Treasury note was 6.9 percent per year. For the full period 1970-2009, the volatility was 6.6 percent per year. Volatility, in either direction, is bad for investors in GNMAs, as they only earn the risk premium when rates are relatively stable. (Of course, it's possible that the greater volatility was accompanied by higher before the fact risk premiums.)
To summarize, investors in GNMAs have not been rewarded with greater returns, let alone greater risk-adjusted returns. The bottom line is that there are far worse "mistakes" than investing in GNMAs. (It's better to take maturity risk than credit risk, at least you are assured of return of principal.) However, in my opinion there are sufficient reasons to avoid them. In light of the evidence, the risks don't seem worth taking.