Let's simplify it. Two portfolios - both with same duration but one is heavily weighted in the 30 year bond. Longer bonds have higher duration so if curve steepens in the long end, the portfolio with the larger exposure to 30 year bonds (a 30 year key rate no?) will under perform vs the other.
Hi @brian.field I agree with your interpretation. I think this is analogous to how a barbell strategy tends to under-peform (over-perform) when there is a yield curve steepening (flattening). In Tuckman's example, both the benchmark index and portfolio share the same duration of 4.339 years, but the portfolio is more concentrated at the 30-year key rate. Tuckman only shows the key rate durations of the benchmark, where the 30-year key rate duration is only 0.349 years; i.e., a 10 basis basis shock to the 30-year key rate implies approximately at 0.035% price change in the bond portfolio. If we imagine the manager's key rate duration profile, it sums to approximately 4.339 years (i.e., same duration as benchmark) but the 30-year key rate is higher than 0.349. So it doesn't need to be a barbell per se, but it would be more concentrated at the 30-year key rate exposure.
Then under a yield curve flattening:
The 30-year rate decreases more than, say, the 2-year rate. The manager's portfolio, which has the higher 30-year key rate duration, outperforms due to have a greater proportion of the portfolio "weighted" toward the 30-year rate exposure; or the other flattening:
The 30-year rate increases less than, say, the 2-year rate. This is an overall loss on the long position, but the manager's portfolio experiences a lesser decrease (ie, outperforms) than the benchmark due to being more weighted toward the 30-year rate and less weighted toward to the shorter rates that decreased more. I hope that's helpful and your weekend is going well!