|
|
|
|
|
|
From: Vladimir
Affiliation:
Address:
Date: 26 Nov 2003
Time: 18:34:43
I feel that I have to clarify couple things. I was WRONG saying that convexity (further referred to as C) had nothing to do with parallel shifts. Indeed, that term (dy)^2 in the C equation implies that we are talking about parallel shifts. I regret I did this simple mistake. But I actually had a different thing in mind. The original message assertained that "in the world of parallel YC shifts, positive convexity is good, negative is bad". And this is just not true. Independent of curve reshapings, higher convexity is more valuable due to its hedging property. And this fact indeed has nothing to do with how YC evolves over time. I guess I just had to make myself more clear. In general, D is about small parallel shifts, while C is about cash flow dispersion in a portfolio as well as expected volatility.
I would disagree with Lowery that "positive convexity is always good". In certain cases negative convexity might be a desireable property.
It might be problematic and confusing to monitor separate convexity limits in addition to duration limits. Though I agree with Lowery that they provide a very useful information, monitoring them and setting limits might just add to traders' confusion. On the other hand, DV01 (for the whole accrual book)incorporates both D and C. I can give an example of Citigroup, which monitors accrual book interest rate exposure through DV01.
And I absolutely agree with Lowery that in the A/L context, scenario simulation (exposure to non-parallel shifts) is more important than D,C and DV01 put together.
|
|
