Econometrics and Statistics

EViews 10 offers a exciting new additions and improvements to its set of econometric and statistical features. The following is a brief outline of the most important new features, followed by additional discussion and pointers to full documentation.

Smooth Threshold Regression Estimation

Smooth Transition Autoregressive (STAR) modeling (Teräsvirta, 1994) is an extremely popular approach for nonlinear time series analysis. STAR models, which are a special case of Smooth Transition Regression (STR) models, embed regime-dependent linear auto-regression specifications in a smooth transition nonlinear regression framework.

EViews tools for estimation of two-regime STR models with unknown parameters for the shape and location of the smooth threshold. EViews estimation supports several different transition functions, provides model selection tools for selecting the best threshold variable from a candidate list, and offers the ability to specify regime varying and non-varying variables and variables that appear in only one regime.

To estimate a smooth transition model, Quick/Estimate Equation... from the main EViews menu, select THRESHOLD - Threshold Regression from the main Method dropdown menu near the bottom of the dialog, and click on the Smooth radio button in the Threshold type setting.

The options page allows you specify the transition function, covariance estimation method (including various robust estimators), and optimization settings.

Following estimation, EViews offers specialized views for the transition function and weights along with support for tests for linearity against STR alternatives and tests of no remaining nonlinearity and parameter constancy, alongside conventional tests for heteroskedasticity and serial correlation.

• See “Smooth Transition Regression” for discussion and additional features.

Finite-Sample Adjusted Heteroskedasticity-Consistent Covariances

EViews 10 offers support for several new heteroskedasticity robust coefficient covariances in least squares regression. These alternatives provide different approaches to accounting for bias in finite samples by adjusting the weights given to residuals on the basis of leverage (Long and Ervin, 2000; Cribari-Neto and da Silva, 2011).

This general class of heteroskedasticity consistent sandwich covariance estimators may be written as:

(0.1) |

where are observation-specific weights that are chosen to improve finite sample performance.

EViews allows you to estimate your covariances using several choices for . In addition to the standard White covariance estimators (HC0, HC1), EViews supports the bias-correcting HC2, pseudo-jackknife HC3 (MacKinnon and White, 1985), and the leverage weighting HC4, HC4m, and HC5 (Cribari-Neto, 2004; Cribaro-Neto and da Silva, 2011; Cribari-Neto, Souza, and Vasconcellos, 2007 and 2008).

The full range of these estimators is available for equations estimated by linear least squares.

Select a method, enter any required parameters, and click on OK to proceed to estimation.

• See “Alternative HC Estimators” for discussion and additional detail.

Cluster-Robust Covariances

In many settings, observations may be grouped into different groups or “clusters” where errors are correlated for observations in the same cluster and uncorrelated for observations in different clusters (Liang and Zeger, 1986; Wooldridge, 2003; Cameron and Miller, 2015). EViews offers support for consistent estimation of coefficient covariances that are robust to either one and two-way clustering.

When estimating an equation using LS, the Covariance method dropdown on the Options may be used to select Cluster robust covariance estimation.

You will then be prompted to enter the name of one or two series in the Cluster series edit field, and to select the (cluster robust) CR method from among approaches ranging from d.f. corrected to methods using finite sample corrections based on the leverage for observations.

• See “Cluster-Robust Covariances” for discussion and additional detail.

VARs with Linear Restrictions

The basic -variable VAR(p) specification has coefficients so that even moderate sized VARs require estimation of a large number of parameters. When VARs are applied to macroeconomic data with limited sample sizes, model over-parameterization is a frequent problem as there are too few observations to estimate precisely the VAR parameters.

EViews now offers support for the linear restriction approach to handling this over-parameterization problem.

Select Quick/Estimate VAR... or type var in the command window to display the estimation dialog. Click on the VAR Restrictions tab to specify your restrictions:

The Restrictions section on the left should be used to select the VAR elements that you wish to restrict. You may click on the entries for the lag matrices (L1, L2, L3, L4) and the vectors of coefficients associated with each exogenous variable (C) to select an element to restrict. The right side of the dialog will change to show the current settings for the selected element.

In settings where restricting a large number individual elements is inconvenient, or where you wish to impose restrictions across lag or exogenous variable coefficients, you may find it more convenient to specify restrictions using text expressions. In this case, click on the Text node of the Restrictions tree to display the Restriction Text edit field.

When restrictions have been applied, the Basics tab of the dialog will change to offer additional options. In the newly displayed Restriction estimation section in the bottom left o the dialog, you will be prompted to specify the method of estimating your restricted coefficients:

In you do not select the Iterate GLS weighting option, EViews will estimate the parameters of the model using one-step GLS estimation.

If you do select Iterate GLS weighting you will be prompted to specify the iteration and convergence properties of the estimation. To estimate with OLS, you may select Iterate GLS weighting and specify a Max Iterations of 0.

Click on OK to estimate restricted VAR specification.

For additional discussion, see

• See “Estimating a VAR with Linear Restrictions in EViews” for discussion and additional detail.

• See Var::ls

SVAR Specification and Estimation

A structural VAR (SVAR) uses additional identifying restrictions and estimation of structural matrices to transform VAR errors into uncorrelated structural shocks. Obtaining structural shocks is central to a wide range of VAR analysis, including impulse response, forecast variance decomposition, historical decomposition, and other forms of causal analysis. See, for example, Amisano and Giannini (1997), Martin, Hurn, and Harris (2013).

Prior versions of EViews supported estimation of SVARs that employed either short-run A-B restrictions or long-run F restrictions for identification, but not models that included both. Further, the interface for specification of the SVAR restrictions was somewhat cumbersome.

EViews 10 introduces an entirely new SVAR estimation engine following Rubio-Ramirez, Waggoner, and Zha (2010) that allows for models that combine short and long-run restrictions. In addition the new specification interface is a much more intuitive method of imposing your restrictions.

As before, estimation of a SVAR may be performed by selecting Proc/Estimate Structural Factorization... from the menu of an estimated VAR. The all new Identifying Restrictions tab allows you to specify your SVAR restrictions.

The Restriction Preset drop-down menu provides a variety of pre-built restriction templates that can be applied to the SVAR model. Several of the presets can be used as is, while others require additional restrictions to meet the order condition.

Alternately, the selection area on the left of the dialog allows you to view the current pattern matrices for any and all of the four canonical matrices as well as the current collection of text expression restrictions. Selecting a matrix element will change the dialog display to show the current restrictions for that matrix.

As before, you may use the Text selection to enter your restrictions in text. Text entry will be required for more complicated that restrictions that involve more than one canonical matrix element. In addition, there are a number of special functions that make it easy to specify common restrictions such as lower triangular or diagonal.

The top portion of your SVAR output will show the restrictions, parameterization, and estimated coefficients,

while the bottom portion shows the structural matrix estimates,

For additional discussion, see

• See “Structural (Identified) VARs” for discussion and additional detail.