Activity Number:
|
159
|
Type:
|
Invited
|
Date/Time:
|
Tuesday, August 13, 2002 : 8:30 AM to 10:20 AM
|
Sponsor:
|
JBES
|
Abstract - #300357 |
Title:
|
Iterative and Recursive Estimation in Structural Non-adaptive Models
|
Author(s):
|
Eric Renault*+ and Valentin Patilea
|
Affiliation(s):
|
Université de Montréal and Université de Orleans
|
Address:
|
C.P. 6128, succursale Centre-ville, Montreal, Quebec, H3C 3J7, Canada
|
Keywords:
|
nonlinear state space modelling ; backfitting ; option pricing
|
Abstract:
|
The paper proposes an inference method, called latent backfitting, well-suited for a wide class of econometric models where the structural relationships of interest define the observed endogenous variables as a known function of unobserved state variables and unknown parameters. This nonlinear state space specification opens the door for an iterative EM-like strategy: In the E-steps the states variables are forecasted given the observations and a value of the parameters; in the M-steps these forecasts are used to define estimations in the latent world. The iterative estimation we propose is particularly useful for latent regression models and for state variables models of common use in finance. In the first case, it can be seen as an extension of standard backfitting based on generalized residuals. In the latter case, implied state variables are exactly recovered for a given value of the parameters. Recursive procedures are also proposed for less CPU-demanding estimation. The asymptotic properties of our estimators are based on a contraction mapping condition. An empirical illustration in the context of continuous time models of the term structure of interest rates is provided.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |