In addition to all Common Features, the Specification tests dialog includes a "Tests" frame. In this frame you can select the various types of tests you wish to bootstrap. The choices are:
| • | Serial correlation: The LM tests for no serial correlation. |
| • | Ljung-Box: The Ljung-Box Portmanteau test for no serial correlation. |
| • | Univariate ARCH: The LM tests for no ARCH on an equation-by-equation basis. |
| • | Multivariate ARCH: The LM test for no multivariate ARCH versus first order ARCH. |
| • | Normality: The Wald-type tests for no skewness, no excess kurtosis (relative to the Gaussian distribution), and a joint hypothesis. |
| • | Omnibus: The Shenton-Bowman/Doornik-Hansen omnibus test for no skewness and no excess kurtosis (relative to the Gaussian distribution). |
| • | Lag order: The LM and Wald tests of a greater (fewer) number of lags. |
| • | Weak exogeneity: Wald tests of a variable being weakly exogenous with respect to α and β. |
| • | Granger causality: Wald tests of no Granger causality. |
|
Figure: The dialog for running the bootstrap of specification tests. |
