Testing and Evaluation

• Actual, fitted, residual plots.

• Wald tests for linear and nonlinear coefficient restrictions; confidence ellipses showing the joint confidence region of any two functions of estimated parameters.

• Omitted and redundant variables LR tests, residual and squared residual correlograms and Q-statistics, residual serial correlation and ARCH LM tests.

• White, Breusch-Pagan, Godfrey, Harvey and Glejser heteroskedasticity tests.

• Chow breakpoint and forecast tests, Quandt-Andrews unknown breakpoint test, Ramsey RESET tests, OLS recursive estimation.

• ARMA equation diagnostics: graphs or tables of the inverse roots of the AR and/or MA characteristic polynomial, compare the theoretical (estimated) autocorrelation pattern with the actual correlation pattern for the structural residuals, display the ARMA impulse response to an innovation shock.

• Easily save results (coefficients, coefficient covariance matrices, residuals, gradients, etc.) to EViews objects for further analysis.

• <See also Estimation and Systems of Equations for specialized testing procedures>.

Previous: System of Equations  |  Next: Forecasting and Simulation