Caroline Karlsson presenterar sitt examensarbete

Examining Affine General Equilibrium Models
	
Abstract
	
This thesis project, explores Eraker's (2006) Affine General
Equilibrium Models for bond pricing. In the model, an infinitely lived
representative agent receives utility of consumption. Instead of using
time-separable expected utility functions, recursive Epstein--Zin
preferences are used. For these preferences, the utility today depends
on future utility and they allow the elasticity of intertemporal
substitution to be disentangled from the coefficient of relative risk
aversion. According to Eraker and other authors, the disentanglement
together with appropriate dynamics of state variables allow different
key market phenomena to be explained. Eraker uses affine processes,
since these are flexible and he wants to create a framework that
rivals with no--arbitrage models.

I have extended one of Eraker's example models with an extra
volatility process with the end goal being an even better fit between
simulated bond yields and yields of US Treasury bonds. However, as
opposed to Eraker, I have chosen to model state variables as
first--order vector autoregressions in order to preserve the discrete
time setting. I find this convenient since Epstein--Zin preferences
are given in discrete time.