Yield Curve Strategies - Reading 23 Q17

The problem states that the analyst expects a stable yield curve and proposes 3 strategies:

  1. Sell 3 year bond, use proceeds to buy 10 year bonds

  2. Sell 5 year bonds, use proceeds to buy 30 year MBS

  3. Sell 10 year bond, buy call option on 10 year bond future

The answer is Strategy #2. I understand that during a stable yield curve, we can Buy & Hold, Ride the Yield Curve, Carry Trade and Sell Convexity.

Strategy 1 is a carry trade and Strategy 2 is selling convexity, and Strategy 3 is buying convexity.

Is there a preference/ highest to lowest expected return from the 4 strategies during a stable yield curve? I’m guessing Buy & Hold would have the lowest expected return, but what about the other 3? How come we didn’t pick Strategy 1?

I would’ve answered 1 as being most profitable. Strategy 2 where you short 5 year bonds with equal duration to buy 30 year bonds will incur negative carry actually. This is because for one the term spread of 30 over 5 is not sufficient if you consider the fact that you have to short more 5 year bonds than you an go long 30 year bonds (duration adjusted yield is lower) and two the roll-down return on the 5 year bond is significantly higher if the curve is upward sloping (and usually flat beyond 20-25+). The MBS spread over a regular bond will most likely not be sufficient to cover this opportunity cost.

I actually figured it out. I left the most important piece of information out. I woke up at 5am to do this and obviously my brain wasn’t awake yet. At the beginning of this question, it states that there is a mandate that allows portfolio duration to fluctuate +/- 0.30 per year from benchmark. And before they list the strategies, they have a table showing the duration of each bond. Additionally, when they state Strategy 2, they specifically say MBS has effective duration of 4.75.

3 year bond has duration 2.92; 5 year bond has duration of 4.74, and 10 year bond has duration of 8.82.

So Strategy 1 would violate the mandate as as selling 3 year bonds and buying 10 year bonds clearly deviates the duration by more than +/- 0.30 whereas Strategy 2 you sell 5 year bond (4.74 duration) and buy 30 year MBS (4.75) leaving duration virtually unchanged.

i actually missed the whole ±0.30 duration deviation per year.

I just figured that Abram predicted a stable environment and that strategy 2 bought MBSs, which one would do in a stable environment, i.e. sell convexity.

I guess you can only eliminate strategy 1 by applying the constraint of a deviation tolerance for the portfolio’s duration?