By now every serious investor understands the consequences rising interest rates will have on bond portfolios. For more than four years we’ve been reminded that when rates go up, bond prices fall—and the longer a bond fund’s duration, the greater the losses will be.

The conventional wisdom is to keep your bond duration short if you expect rates to rise. The problem is, the iShares DEX Short Term Bond (XSB) has a yield to maturity of just 1.38% these days—once you deduct fees, that’s less than a savings account at an online bank. And unlike a savings account (which effectively has a duration of zero), short-term bonds will still lose value if rates move higher.

It should come as no surprise that the financial industry has come up with a product that tries to address this issue: it’s called the floating-rate note. A “floater” has a maturity date like a conventional bond, but its coupon is tied to a benchmark such as the Canadian Dealer Offered Rate (or CDOR, which is this country’s version of LIBOR). The coupon is adjusted every month or every quarter.

Because of this reset, floaters have a duration close to zero and won’t lose value if rates rise—indeed, they will benefit by paying higher coupons almost immediately. The iShares DEX Floating Rate Note (XFR) boasts a duration of just 0.13, yet its yield to maturity is 1.26%, just 12 basis points lower than that of XSB. Last year, as short-term rates ticked up slightly, XFR returned 1.90%, significantly more than both short-term bonds and cash.

Whatever floats your note

So what’s the catch? Why bother with short-term bonds or savings accounts when you can get higher yields and virtually no interest rate risk with floaters? As with any investment, there are always trade-offs to consider.

The first is that floaters will underperform if short-term rates decline or stay flat. While rates have been poised to rise for a long time, the fact is the CDOR has barely budged since October 2010. More important, during a period of turmoil in the equity markets, rates are likely to fall as investors rush to safety, so high-quality conventional bonds are a better diversifier in a balanced portfolio. In 2008, XSB returned over 8%, providing a soft cushion as stock markets plummeted. Floaters could not have offered that kind of protection.

The second potential pitfall with floaters is credit risk. The good news is this doesn’t seem to be an issue with XFR, which is about 85% government bonds: more than half the holdings are rated AAA, and nothing is rated lower than A. When we next experience a panic in the stock market, these high-quality bonds are not likely to see a significant decline in value.

But that’s not true of actively managed funds, which may hold floaters that are nothing more than junk bonds. The BMO Floating Rate Income Fund, for example, lost more than 48% in 2008 as high-yield bonds cratered along with stocks and real estate. The Trimark Floating Rate Income Fund also took a 28% haircut that year. Clearly these funds are not a substitute for short-term investment-grade bonds or cash.

The only other ETF in this asset class is the Horizons Active Floating Rate Bond (HFR), which falls between the two extremes. It’s exposure (through an interest rate swap) is to corporate bonds, but they’re all investment-grade, with about 70% rated A or higher. A fund with this exposure can certainly lose money during a 2008-type crisis, but it won’t suffer anything like the carnage we saw with junk bonds. I would consider this ETF an alternative to something like the Vanguard Canadian Short-Term Corporate Bond (VSC).

In my view, floating-rate notes are little more than a tactical bet on where interest rates are headed, which is inconsistent with a passive investing strategy. For long-term investors, a traditional bond allocation (whether it’s a ladder or a broad-based ETF) will provide more protection when equity markets take a tumble, and that’s the most important role of fixed income in a portfolio. And for those who need to park cash for short-term needs, a high-interest savings account is a far safer alternative.