How is Ho-Lee model calibrated in practice?

umvue · December 7th, 2008, 3:25 am

Ho-Lee model is likedr = theta(t)*dt + sigma*dzSuppose I have zero-coupon yield curve that has these pointsmaturity(yr) zero rate (%)0.25 10.1270.5 10.4691.0 10.5361.5 10.6812.0 10.808Do I then fit a smooth bond price function P with these prices and then use the relationship f(0,t) = - d(log P(0,t)/dt to deduce the theta(t)? If that's how it's done in practice, usually what is the functional form people use for the bond price function?Thanks a lot!

umvue · December 9th, 2008, 1:40 am

ah. I find the answer to my question....http://en.wikipedia.org/wiki/Yield_curv ... arket_data

amit7ul · December 11th, 2008, 10:27 am

you would have to estimate slope of the curve at t=0 and use that to estimate theta.. bootstrapping+interpolation would be required.

umvue · December 11th, 2008, 11:03 pm

Thanks for your reply. I think I know how to bootstrap and then obtain a piece-wisely connected linear yield curve. But I think I only have the line to connect overnight rate and 1mos rate. Do I just use this slope as the slope for T=0? Also, what should be the slope when T hits the undifferentiable points, e.g. 1mos?