Fixed-Income Traders: What do you think about the HJM model?

Dave baby's picture
Rank: Monkey | 64

-read an interesting article about the future of FI sell-side trading as it relates to risk management. What do you guys think about the classical approach to risk management, i.e., duration, convexity..., being replaced by the Heath, Jarrow, Morton model?

Comments (10)

Sep 22, 2007

The HJM model (along with the LIBOR market and swap market models) are both relatively old inventions that are mostly used nowadays to serve as a theoretical framework to build off, much like Black-Scholes. There are more recent models in the literature today (for example, the Flesaker-Hughston model), and I'm sure the investment banks have developed for proprietary use some even more complicated ones.

Sep 24, 2007

uhhh....using HJM or any model is a completely different question from the duration, convexity or other sensitivities of your book.

One guy might be driving an audi, one guy a lexus, but hopefully everyone is wearing a seatbelt. understand?

Sep 24, 2007
Jimbo:

uhhh....using HJM or any model is a completely different question from the duration, convexity or other sensitivities of your book.

One guy might be driving an audi, one guy a lexus, but hopefully everyone is wearing a seatbelt. understand?

...not really. Jimbo, do you duration and convexity hedge? Or do you use a new modeling approach, similar to HJM?

Sep 24, 2007
BigDaveLax:
Jimbo:

uhhh....using HJM or any model is a completely different question from the duration, convexity or other sensitivities of your book.

One guy might be driving an audi, one guy a lexus, but hopefully everyone is wearing a seatbelt. understand?

...not really. Jimbo, do you duration and convexity hedge? Or do you use a new modeling approach, similar to HJM?

Of course we hedge duration and convexity. If not I would be out of a job, and fast.

HJM is a modelling framework. There are multiple models that describe the evolution of rates in the future. Which model you pick, how you calibrate it, etc is one story.

A model itself is a pricing tool, and a risk outputting tool. Using a model is not the same as hedging a position. It's like saying do you hedge your delta, or do you use black scholes?...the answer is both, you use black scholes (or an HJM model or Cheyette, or whatever) to price the structure and generate the risk profile (Duration, convexity, vega,etc) and then hedge that. Different models may show the risk differently, but whichever one you pick, you need to manage the risk.

Sep 24, 2007

instruments oftentimes move differently than how models predict...Thus the usefulness of having an experience person reading the models and executing the hedges as they know how to use the model for their purposes...

Learn More

Boost your resume and land a finance job by passing the FINRA SIE. 264 pages & 1981 smart flashcards written by a former 8X top Fidelity instructor. Try it for 0 bananas here.

Sep 24, 2007

BigDave, it's pretty clear you have a limited understanding of what HJM is and how models relate to what traders do. Here's a good starting point for the model http://en.wikipedia.org/wiki/Heath-Jarrow-Morton_f...

Sep 24, 2007
PoppingMyCollar:

BigDave, it's pretty clear you have a limited understanding of what HJM is and how models relate to what traders do. Here's a good starting point for the model http://en.wikipedia.org/wiki/Heath-Jarrow-Morton_f...

well, there's no way to know much about it unless you're in the market or an academic. but choice of model is very important...you have to have a handle about their appropriateness for what you're pricing, strengths and weaknesses, and balance against computational constraints....BGM for example is fairly widely used but some banks will not use it b/c it is so computationally intensive. so that's the sort of constraint you might not have thought of...

if you are a saleman using the sort of sales tools many banks have and you send them a pricing request on a european swaption built in a bermudan swaption template(with all but the first exercise zeroed out) they will be very curious as to why you are trying to chose the model on their behalf...

Am happy to answer more general modelling questions btw, this sort of thing interests me.

Jimbo

Sep 28, 2007
Comment