The following article is reprinted from the July, 1997 issue of On the Edge,
the Interactive Data Fixed Income Analytics bimonthly newsletter.
Using the BondEdge Optimizer to Help Formulate
Portfolio Strategy:
MBS portfolios, yield curve views and duration constraints
Wesley Phoa, Ph.D.
President of Research
An important application of the BondEdge
Optimizer is to help investors formulate portfolio strategy. For example, what is the
optimal structure for an MBS portfolio - i.e. what is the optimal mix between current,
discount and premium coupons? The answer will depend on the investor's yield curve view
and portfolio duration target. Traditionally, investors devise and analyze various
combinations by hand. The Optimizer does the same job more thoroughly - because it can
look at all possible combinations - and more efficiently - because the search is fully
automated and highly efficient.
To illustrate how to use the Optimizer in this way, the following nine yield curve
scenarios were analyzed as of 5/30/97, corresponding to all possible combinations of
parallel and steepening/flattening scenarios:
- 6-month yield changing by -50 bp, 0 bp, +50 bp;
- 10-year to 30-year yields changing by -50 bp, 0 bp, +50 bp.
In each case, intermediate yields were generated by linearly interpolating the assumed
yield shifts at the 6-month and 10-year points. The problem was to maximize total return
over a 3-month holding period, while ensuring that initial portfolio duration was equal to
a specified target level. The universe of available securities included 6.0%-10.0% coupon
FNMA pass-throughs (most recent production year in each case).
In the absence of a duration target, the optimizer would select a single security - the
one which had the highest total return under the given scenario. In order to achieve a
duration target, the optimizer must instead find an appropriately weighted portfolio of
two or more securities. It is easy to see that, for this particular problem, an optimal
portfolio will contain at most two securities.
For example, suppose an investor believes that the entire yield curve will rally by 50
bp. If the portfolio's duration target is 5.0, it should be structured around 7.5% coupons
(58%) and 6.0% coupons (42%). If the duration target is 3.0, the portfolio should be
structured around 8.0% coupons (77%) and 9.5% coupons (23%).
Thus, from a huge number of feasible alternatives, the BondEdge Optimizer efficiently
selects the most promising potential trade for each yield curve view; in the presence of
duration constraints, the "best" combinations are far from obvious. Of course,
the next step is to carry out further analysis on the proposed trades; it rarely makes
sense to blindly follow an optimizer's prescriptions.
Note that in some scenarios, it is actually optimal to hold cash rather than being
fully invested in mortgage pass-throughs. In practice, of course, one would allow more
flexibility - for example, an investor would be able to hold Treasury bonds as well as
cash. This could be accommodated by adding more securities to the universe.
