| In response to user requests, we have added several new features to EViews 4.0. Some of these features, like added function support, an expanded Wald test output, and extensions to the model program language, are minor improvements to existing routines. Others, like the new unit root testing features, and extensions to both the state space error, and system instrumental variables specification languages, represent significant improvements in the set of tools for working with, and analyzing your data.
Unit Root Testing
EViews 4.1 includes support for the newest generation of unit root tests. In addition to the existing Augmented Dickey-Fuller and Phillips-Perron tests, EViews now allows you to compute the GLS-detrended Dickey-Fuller (Elliot, Rothenberg, and Stock, 1996), Kwiatkowski, Phillips, Schmidt, and Shin (KPSS, 1992), Elliott, Rothenberg, and Stock Point Optimal (ERS, 1996), and Ng and Perron (NP, 2001) unit root tests.
EViews will also perform Newey-West (1992) and Andrews (1991) automatic bandwidth selection for kernel based estimators, or automatic information criteria based selection of lag length for Dickey-Fuller tests and AR spectral density estimators.
System Extensions
We have expanded the flexibility of instrumental variables specifications estimated by 2SLS and 3SLS.
Previously, all instrumental variables projections in systems were performed on an equation-by-equation basis. In EViews 4.1, the new @stackinst statement provides a new way of specifying instruments for systems of equations that allows for cross-equations on the projections of variables on instruments. You can now stack your equations and instruments prior to performing the projection. The easy-to-use syntax provides you with full control over the instrument stacking so that you may combine the earlier ordinary instrument and new stacked instrument specifications.
Sspace Improvements
EViews 4.1 extends the features of the sspace object in two distinct ways.
First, extensions to the state space syntax in EViews 4.1 allow you to write the error term for any equation as a linear combination of named errors. This syntax allows users to specify a wide range of important models in a more natural fashion.
Second, new sspace object data members give you access to output matrices containing intermediate calculations from the Kalman filter.
Enhanced Model Features
EViews 4.1 provides new modeling tools to aid you in model building and solution. An enhanced command syntax provides you with greater control over the model solution procedure, enhanced tools for working with scenarios, and additional commands to maintain the links in your model.
Miscellaneous Statistical Features
The output of the Wald test has been expanded to provide additional information about the restrictions and the restriction variances. As a result, you can use the expanded output to find the standard errors of functions of your coefficients.
The "Statistics by Classification" view of a series now allows you to compute arbitrary quantiles of your data by group.
New Functions
EViews 4.1 now supports a family of percentage change functions that complement the existing functions. These functions eliminate the need to rescale function values when working in percentage terms.
In addition, we have added functions for computing base-10 and arbitrary base logarithms.
|