Some Basic Instructions for Using EViews

To run EViews on the CSB/SJU computing network, you will find the EViews program under Start / All Programs / Academic Software / Economics (or in Network Programs / Academic Applications / Economics / EViews). 

To access EViews from off-campus via the internet, you have to log-on to the Citrix server at:  https://citrix.csbsju.edu/Citrix/MetaFrame/auth/login.aspx; after entering your CSB/SJU computing network username and password, click on the Academic Software folder, then Economics, and then EViews 9 to start the program.

The current version is 9, although you may find earlier versions also on the network -- they'll work fine, too.

EViews has on-line HELP which can provide you with assistance in working with the data and program. This is a professional-quality econometrics software package with some pretty sophisticated features – yet it is not hard to use for even simple, basic econometric work.

After starting EViews, you will need to open/load a work-file by using the File menu to Open one of the Workfiles – the program will prompt you where it should look for the *.wf1 files, if not in the current directory.  (If you want to create a new workfile, refer to HELP for instructions.)

Once you have loaded a workfile, to select a data series, simply click on the variable name. You can select multiple series by holding the Ctrl key and clicking on the variables you wish to select and put into a Group; holding the Shift key allows you to select a continuous block of listed variables.  You can show the Group variables in list format, a spreadsheet, or graph(s).

To run a regression, click on the Quick menu and Estimate Equation. It will prompt you to enter either the variables (dependent variable first, then independent variables; and don’t forget to include c for the constant term) or an explicit equation.

EViews allows you to enter log (and other) transforms directly into the regression equation variables – so you can enter Log(var1) c Log(var2) without having to separately create the log of var1 as a separate data series (using the GENR function).  Similarly, you can specify new variables created by arithmetic from existing variables – so you can enter (var1*100/var2) or (var1*var2).  This easily beats what you’d have to do in Excel.

Once the regression equation has been estimated, you can observe a plot of the residuals, actual and fitted values by clicking on Residuals.  Statistics returns the estimated equation, Estimate returns to the prompt for entering variables or an equation.

If your estimated equation indicates the presence of serial or auto-correlation (Durbin-Watson statistic below 1.7 or 1.8) as mentioned in the text, you can attempt to correct for it by successively adding AR(n) terms (with "n" = 1, 2, 3, ...) to your regressor list until the D-W statistic increases to or above an acceptable level (2.0) or does not improve.

You can also adjust the sample periods (SMPL) within the data set range in order to estimate the equation for different periods to see and test if the estimated coefficients or relationships change.