Additional Regression Tools

This section describes additional tools that may be used to augment the techniques described in
“Basic Regression Analysis”.

• This first portion describes special EViews expressions that may be used in specifying and estimating models with Polynomial Distributed Lags (PDLs) or dummy variables (
“Special Equation Expressions”.

• In the second section, we describe methods for heteroskedasticity consistent, heteroskedasticity and autocorrelation consistent, and cluster robust covariance estimation (
“Robust Standard Errors”).

• Next we describe tools for weighted least squares (
“Weighted Least Squares”) and nonlinear least squares (
“Nonlinear Least Squares”estimation.

• Lastly, we describe the indicator saturation approach which extends basic regression analysis by testing for outliers and structural breaks in the regression specification (
“Indicator Saturation”).

Note that parts of this discussion refer to estimation of models which have autoregressive (AR) and moving average (MA) error terms. These concepts are discussed in greater depth in
“Time Series Regression”.