Articles |
Monte Carlo Estimation for Nonlinear Non-Gaussian State Space Models
Department of Econometrics, VU University Amsterdam, De Boelelaan 1105, 1081 HV Amsterdam, The Netherlands bjungbacker{at}feweb.vu.nl s.j.koopman{at}feweb.vu.nl
Received for publication 1 December 2005. Revision received 1 May 2007.
We develop a proposal or importance density for state space models with a nonlinear non-Gaussian observation vector y 
p(y¦
) and an unobserved linear Gaussian signal vector p(). The proposal density is obtained from the Laplace approximation of the smoothing density p(¦y). We present efficient algorithms to calculate the mode of p(¦y) and to sample from the proposal density. The samples can be used for importance sampling and Markov chain Monte Carlo methods. The new results allow the application of these methods to state space models where the observation density p(y¦) is not log-concave. Additional results are presented that lead to computationally efficient implementations. We illustrate the methods for the stochastic volatility model with leverage.
Key Words: Importance sampling • Kalman filtering • Markov chain Monte Carlo • Newton–Raphson • Posterior mode • Simulation smoothing • Stochastic volatility model
References
-
Anderson B. D. O., Moore J. B. Optimal Filtering (1979) Englewood Cliffs: Prentice-Hall.
Black F. Studies of stock price volatility changes. Proc. Bus. Econ. Statist. Sec. (1976) Alexandria: Am. Statist. Assoc. 177–81.
Carter C. K., Kohn R. On Gibbs sampling for state space models. Biometrika (1994) 81:541–53.
Casella G., Robert C. P. Rao-Blackwellisation of sampling schemes. Biometrika (1996) 83:81–94.
de Jong P. Smoothing and interpolation with the state space model. J. Am. Statist. Assoc. (1989) 84:1085–8.[CrossRef][ISI]
de Jong P., Shephard N. The simulation smoother for time series models. Biometrika (1995) 82:339–50.
Durbin J., Koopman S. J. Monte Carlo maximum likelihood estimation of non-Gaussian state space models. Biometrika (1997) 84:669–84.
Durbin J., Koopman S. J. Time Series Analysis by State Space Methods (2001) Oxford: Oxford University Press.
Durbin J., Koopman S. J. A simple and efficient simulation smoother for state space time series analysis. Biometrika (2002) 89:603–16.
Fahrmeir L., Kaufmann H. On Kalman filtering, posterior mode estimation and Fisher scoring in dynamic exponential family regression. Metrika (1991) 38:37–60.[CrossRef]
Fruhwirth-Schnatter S. Data augmentation and dynamic linear models. J. Time Ser. Anal. (1994) 15:183–202.
Fruhwirth-Schnatter S. Efficient Bayesian parameter estimation. State Space and Unobserved Components Models—Harvey A. C., Koopman S. J., Shephard N., eds. (2004) Cambridge: Cambridge University Press. 123–51.
Geweke J. Bayesian inference in econometric models using Monte Carlo integration. Econometrica (1989) 57:1317–39.[CrossRef][ISI]
Golub G. H., Van Loan C. F. Matrix Computations (1997) 2nd ed. Baltimore: The Johns Hopkins University Press.
Journel A. Geostatistics for conditional simulation of ore bodies. Econ. Geol. (1974) 69:673–87.[Abstract]
Kohn R., Ansley C. F. A fast algorithm for signal extraction, influence and cross-validation. Biometrika (1989) 76:65–79.
Koopman S. J. Disturbance smoother for state space models. Biometrika (1993) 80:117–26.
Nelson D. B. Conditional heteroskedasticity in asset pricing: A new approach. Econometrica (1991) 59:347–70.[CrossRef][ISI]
Nocedal J., Wright S. J. Numerical Optimization (1999) New York: pringer Verlag.
Ripley B. D. Stochastic Simulation (1987) New York: Wiley.
Shephard N. Stochastic Volatility: Selected Readings (2005) Oxford: Oxford University Press.
Shephard N., Pitt M. K. Likelihood analysis of non-Gaussian measurement time series. Biometrika (1997) 84:653–67.
So M. K. P. Posterior mode estimation for nonlinear and non-Gaussian state space models. Statist. Sinica (2003) 13:255–74.
Yu J. On leverage in a stochastic volatility model. J. Economet. (2005) 127:165–78.[CrossRef]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||