In the examples used here, the bond portfolio consisted of government bonds. Of course, corporate and municipal bonds are widely held in bond portfolios. Unfortunately, there is no corporate bond futures contract. A municipal bond futures contract exists in the United States, based on an index of municipal bonds, but its volume is relatively light and the contract may not be sufficiently liquid for a large-size transaction. Government bond futures contracts tend to be relatively liquid. In fact, in the United States, different contracts exist for government securities of different maturity ranges, and most of these contracts are relatively liquid.
If one uses a government bond futures to manage the risk of a corporate or municipal bond portfolio, there are some additional risks to deal with. For instance, the relationship between the yield change that drives the futures contract and the yield change that drives the bond portfolio is not as reliable. The yield on a corporate or municipal bond is driven not only by interest rates but also by the perceived default risk of the bond. We might believe that the yield beta is 1.20, meaning that the yield on a corporate bond portfolio is about 20 percent more volatile than the implied yield that drives the futures contract. But this relationship is usually estimated from a regression of corporate bond yield changes on government bond yield changes. This relationship is less stable than if we were running a regression of government bond yield changes on yield changes of a different government bond, the one underlying the futures.
In addition, corporate and municipal bonds often have call features that can greatly distort the relationship between duration and yield change and also make the measurement of duration more complicated. For example, when a bond’s yield decreases, its price should increase. The duration is meant to show approximately how much the bond’s price should increase. But when the bond is callable and the yield enters into the region in which a call becomes more likely, its price will increase by far less than predicted by the duration. Moreover, the call feature complicates the measurement of duration itself. Duration is no longer a weighted-average maturity of the bond.
Finally, we should note that corporate and municipal bonds are subject to default risk that is not present in government bonds. As the risk of default changes, the yield spread on the defaultable bond relative to the default-free government bond increases. This effect further destabilizes the relationship between the bond portfolio value and the futures price so that duration-based formulas for the number of futures contracts tend to be unreliable.
It is tempting to think that if one wants to increase (decrease) duration and buys (sells) futures contracts, that at least the transaction was the right type even if the number of futures contracts is not exactly correct. The problem, however, is that changes in the bond portfolio value that are driven by changes in default risk or the effects of call provisions will not be matched by movements in the futures contract. The outcome will not always be what is expected.
Another problem associated with the modified duration approach to measuring and managing bond portfolio risk is that the relationship between duration and yield change used here is an instantaneous one. It captures approximately how a bond price changes in response to an immediate and very small yield change. As soon as the yield changes or an instant of time passes, the duration changes. Then the number of futures contracts required would change. Most bond portfolio managers do not perform these kinds of frequent adjustments, however, and simply accept that the transaction will not work precisely as planned.
We should also consider the alternative that the fund could adjust the duration by making transactions in the bonds themselves. It could sell relatively low-duration bonds and buy relatively high-duration bonds to raise the duration to the desired level. There is still no guarantee, however, that the actual duration will be exactly as desired. Likewise, to reduce the duration to zero, the fund could sell out the entire bond portfolio and place the proceeds in cash securities that have low duration. Reducing the duration to essentially zero would be easier to do than increasing it, because it would not be hard to buy bonds with essentially zero duration. Liquidating the entire portfolio, however, would be quite a drastic thing to do, especially given that the fund would likely remain in that position for only a temporary period.
Raising the duration by purchasing higher-duration bonds would be a great deal of effort to expend if the position is being altered only temporarily. Moreover, the transaction costs of buying and selling actual securities are much greater than those of buying and selling futures.