Errors and Term Premia

Specialeforsvar: Jonas Sofus Nervil Kistorp

Titel: Errors and Term Premia
Understanding the Effect of Noise Contaminated Yields on the One-Step Approach

Abstract: This thesis analysis two popular methods used to estimate the term structure of interest rates; the Adrian, Crump, and Moench model and the Kalman filter in conjunction with an affine term structure model. We test the ability of the methods to recapture the original “true” simulated term premia, when faced with yields contaminated by measurement errors in a one-step approach. The two main contributions are; the exploration of the ACM model in a Monte Carlo study, and how different distribution of measurement errors affect estimation of affine term structure models. We find, that the ACM model struggles with estimating the term premia when subjugated to noisy non-synthetic yields, whereas the Kalman filter with the Stochastic Market Price of Risk model can to a large extent disentangle Gaussian and approximately Gaussian measurement errors from the underlying term structure.

Vejleder: David Glavind Skovmand
Censor:   Mads Stenbo Nielsen, CBS