GMM estimation of affine term structure models

Specialeforsvar ved Mikkel Lundstrøm Weywadt

 Titel: GMM estimation of affine term structure models

Abstract:  Short rate modelling is a classical tool from interest rate theory that has been thoroughly researched over the years. This has led to the extension that is known as the jump-diffusion where we model the noise of the process through a Gaussian diffusion term and a compound Poisson jump term. The existence of jumps in the interest rate is a prevalent theory as it serves as an explanation for the sudden and violent market shocks. Furthermore in the pensions industry short rate modelling is starting to receive increased attention because of the focus on stochastic solvency models. In this thesis we review the theory on affine term structure models for jump-diffusions and relate the results to the classic stochastic differential equation. One of the most challenging aspects is applying the theory to short rate model calibration. Model calibration is important if we are to project the market into the future, which is an important aspect of assessing the risks of a pensions company. As such we dedicate most of our efforts towards developing an estimation framework that uses the generalised method of moments to calibrate the Vasicek model, with and without jumps. Furthermore we describe some useful numerical techniques that are used for approximating otherwise infeasible analytical solutions

Vejledere: Jesper Lund Pedersen / Morten Bakkedal
Censor:     Cathrine Jacobsen, ATP