Re: [Quantlib-dev] Bond redemption, face value and amortising bonds

[Simon I. <Sim...@st...>]

For the asset swap, this would mean having an amortizing floating leg:
which would usually mean that the floating leg has a compatible schedule
(with matching notional paydown dates). That could be very tricky to
implement: I think the best resolution would be to verify that the
floating leg schedule has a coupon payment date within 2 days of the
notional payment date (and thus can be said to have a constant notional
for the coupon).

For the AssetSwap, we would need a Bond accessor function for the
redemption dates and an accessor function for the notional changes.

Simon


-----Original Message-----
From: fer...@gm...
[mailto:fer...@gm...] On Behalf Of Ferdinando Ametrano
Sent: 26 February 2008 09:27
To: Simon Ibbotson
Cc: qua...@li...
Subject: Re: [Quantlib-dev] Bond redemption, face value and amortising
bonds

On Mon, Feb 25, 2008 at 7:44 PM, Simon Ibbotson
<Sim...@st...> wrote:
>  I'd guess that an asset swap on an amortising
>  bond swaps only the interest payments, not the notional repayments.
I agree for par asset swap, in the sense that exchanging the notional
repayments would result in zero net flow, since the repayments should
be identical on the floating leg. This is in accord with what happen
with the final traditional redemption, when only the difference to par
is exchanged. The guideline here is that the bond price as evaluated
discounting all its cashflows on the yield curve should be identical
to the asset swap price when the floating leg spread is set to zero.

When it comes to market asset swap, things changes slightly and might
require further attention.

On a related issue: I haven't taught about it carefully but it might
be worth to have different accessors for interest payments and
redemptions, besides the old all-inclusive cashflows accessor

ciao -- Nando