Good. I understand there to be three oversimplifications in my crude simulation. In order to benefit from the roll return, we have to take into account the entire yield curve, rather than assuming a single "prevailing interest rate." And we have to assume that bonds sold prior to maturity. And we have to assume that the timing of the choice is based on optimization that is a function of the yield curve.skierincolorado wrote: ↑Fri Oct 22, 2021 12:38 amSo most bond funds would not hold to maturity, which actually substantially boosts returns via roll yield. If a bond fund holds a ladder from 8 years to 4 years and then sells them, it's not holding the bonds to maturity. The yield will therefore be more than the YTM....
.Since you are doing a ladder holding to maturity - this doesn't apply. But for most real bond funds roll yield is a significant part of returns. Rolling down the yield curve to shorter durations = price increases (all else equal). This is especially true where the curve is steep - less than 10 years of duration...
So yes I *strongly* agree with your overall point. Bond returns can be quite good even when rates rise, especially once you factor in the roll return on most bond funds, and if the rate increase is modest. The 5 year rate is 1.16% today. Even if it is 1.66% two years from now, holding and rolling 5 year bonds will still have significantly positive reutrns.
I'm not clear whether "roll return" plays a part in a bond index fund. To put it another way, what does the aggregate index assume about the timing of purchase and sale? (Maybe the managers of bond index funds are able to exploit roll return to overcome transaction costs in tracking the index?)
I also understand that these all work together to create an extra source of return--that is, bond funds should do better than the simulation.
Even in my oversimplified simulations, I find that the effect of a slow, gradual interest rate increase is much less than duration x interest rate change. And I can reproduce your observation that there isn't necessarily any loss at all.