These links are not active. So please contact me!
Optimisation methods
- the modified Downhill simplex method
- The Simulated Annealing
- Resolution methods for non-linear systems
- Minpack
- TENSOLVE
- FIXvmn (by Mausner)
- Linear state-space models
- Estimation by quasi-maximum kelihood.
- Kalman filter and smoother
- Standard case
- Diffuse case
- Univariate treatment of time series
- Reference: Durbin & Koopman, 2001, Time Series Analysis by State-Space Methods, Oxford.
- Non-linear/ non-gaussian state-space models
- Estimation by quasi-maximum likelihood.
- Sequential importance sampling
- without systematic resampling
- with adaptative resampling
- with systematic resampling
- with smooth resampling
- Auxiliary Particle filter
- Sequential importance sampling with sub-optimal gaussian proposal à la Kalman
- Monte-Carlo
- Gaussian approximation
- with Smolyak gaussian quadrature
- with gaussian cubature
- with the unscented transform
- Gaussian Mixture approximation
- with Smolyak gaussian quadrature
- with gaussian cubature
- with the unscented transform
- References
- Structural VAR
- Estimation by OLS
- Orthogonalization of shocks (à la Cholesky, Short/long-term identification restrictions)
- Impulse-Response functions
- Decomposition of the error-prediction variance
- Structural Markov-Switching VAR
- Estimation by maximum likelihood
- Choice of the specification: all parameters regime-dependent, only the intercept, ...
- Orthogonalization of shocks (à la Cholesky, Short/long-term identification restrictions)
- State-Dependent Impulse-Response functions
- Generalized Impulse-Response functions
- Markov-switching and unobservable components
- Estimation by quasi maximum likelihood
- Kim filter and smoother (1993)
- Generalized Impulse-Response functions
- Simulation-based methods
- Simulated Method of Moments
- Indirect Inference (Gouriéroux, Monfort & Renault, 1993)
- Simultated Quasi Maximum Likelihood (Smith, 1993)
- Efficient Method of Moments (Gallant & Tauchen, 1996)
- Multiple corrections for the variance matrix of instrumental parameters (including Den Haan & )
- Individual adjustment of instrumental parameters and global over-identification test
- Resolution and Bayesian estimation of DSGE models
- Resolution by the projection method (Judd, 1993) using the Smolyak operator
- Estimation of the posterior mode using one of the 7 particle filters available
- Three variants of the Metropolis algorithm to build the posterior distribution of deep parameters
- RW algorithm with one chain
- RW algorithm with multiple independent chains
- Genetic variant on multiple chains