September 24th, 2010, 2:01 pm
What I wrote is just an instantaneous relation. Your price model is Vasicek, sodP = (blah1) dt + sig1 dW , which means E[dP(t)^2]/dt = sig1^2Your return model is GBM, so dP/P = (blah2) dt + sig2 dW, which means E[(dP(t)/P(t))^2]/dt = sig2^2 or E[dP(t)^2] /dt= P(t)^2 sig2^2 So, the noise terms (and ONLY the noise terms) are instantaneously consistent at time t if sig1 = P(t) sig2, which is what I wrote before.But since sig1 and sig2 are constants and P(t) is not, you can really only match things up at a single time,say time-0. To answer your question directly, in the GBM model, the vol of return is fixed, but *not* the vol of price. Apart from the noise terms, your models are not consistent no matter how you line up the vols.You can't even argue that one is objective and the other is risk-neutral. But, that is a separate issue -- I just throw in the thought.
Last edited by
Alan on September 23rd, 2010, 10:00 pm, edited 1 time in total.