System of Equations 

Basic

• Linear and nonlinear estimation.

• Least squares, 2SLS, equation weighted estimation, Seemingly Unrelated Regression, Three-Stage Least Squares.

• GMM with White and HAC weighting matrices.

• AR estimation using nonlinear least squares on a transformed specification.

• Full Information Maximum Likelihood (FIML).

VAR/VEC

• Estimate structural factorizations in VARs by imposing short- or long-run restrictions.

• Impulse response functions in various tabular and graphical formats with standard errors calculated analytically or by Monte Carlo methods.

• Impulse response shocks computed from Cholesky factorization, one-unit or one-standard deviation residuals (ignoring correlations), generalized impulses, structural factorization, or a user-specified vector/matrix form.

• Impose and test linear restrictions on the cointegrating relations and/or adjustment coefficients in VEC models.

• View or generator cointegrating relations from estimated VEC models.

• Extensive diagnostics including: Granger causality tests, joint lag exclusion tests, lag length criteria evaluation, correlograms, autocorrelation, normality and heteroskedasticity testing, cointegration testing, other multivariate diagnostics.

Multivariate ARCH

• Conditional Constant Correlation (p,q), Diagonal VECH (p,q), Diagonal BEKK (p,q), with asymmetric terms.

• Extensive parameterization choice for the Diagonal VECH's coefficient matrix.

• Exogenous variables allowed in the mean and variance equations; nonlinear and AR terms allowed in the mean equations.

• Bollerslev-Wooldridge robust standard errors.

• Normal or Student's t multivariate error distribution.

• A choice of analytic or (fast or slow) numeric derivatives. (Analytics derivatives not available for some complex models.)

• Generate covariance, variance, or correlation in various tabular and graphical formats from estimated ARCH models.

State Space

• Kalman filter algorithm for estimating user-specified single- and multiequation structural models.

• Exogenous variables in the state equation and fully parameterized variance specifications.

• Generate one-step ahead, filtered, or smoothed signals, states, and errors.

• In- and out-of-sample forecasting, using n-step ahead or smoothed values.

• Examples include time-varying parameter, multivariate ARMA, and quasilikelihood stochastic volatility models.  

 Previous: Estimation  |  Next: Testing and Evaluation