Introduction Eviews for Orientation course Econometrics

Marius Ooms January 2002, update November 2004

Afdeling Econometrie vrije Universiteit amsterdam

Acknowledgement This introduction is partly extracted from Vogelvang (2005).

Contents
1 Introduction 1.1 Intranet and Internet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 The 2.1 2.2 2.3 1 1

2.4

software EViews 2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 EViews Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 An Eviews Basics Demonstration . . . . . . . . . . . . . . . . . . . . . . . . 4 2.3.1 Eviews Workles and Data Import . . . . . . . . . . . . . . . . . . . 4 2.3.2 Creating a Workle . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.3.3 Importing Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.3.4 Checking Imported Data and Saving the Workle . . . . . . . . . . . 6 2.3.5 Eviews menus and Creation of Simple Objects: Groups and Equations 6 2.3.6 Command Window, Data Transformations and expressions . . . . . 7 2.3.7 AutoSeries, Changing Data points . . . . . . . . . . . . . . . . . . . 8 2.3.8 Examples of EViews output: Graphs and export Graphs to Word . . 9 2.3.9 Scatter diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.3.10 Eviews Tables, editing and Exporting Tables to Word . . . . . . . . 11 Advanced use of Eviews Help Examples: Maximum Likelihood . . . . . . . 13 15

3 Conclusion

Introduction

This Eviews-introduction is part of the material for a basic course in Econometrics. It should be used together with a copy of the Eviews software, a textbook and the corresponding datasets. The datasets will be made available to all students via the computer network. Since Eviews, Econometric Views is widely used in the Econometrics curriculum and numerical examples in econometric textbooks are often based on Eviews results, we have chosen the package for this course. During the course the student has to make a number of computer exercises in order to understand the methods in practice. EViews 4 is used to solve the exercises. The faculty of Economics has a student licence for EViews 2.0, so EViews can be used at home too. This lower level of EViews does not present major problems for this course. Eviews 2.0 is an older program and does not always work well with long directory names and le names. So you may have to change the names of some folders and les at home. A more recent version of Eviews (3.1) for students is available for use at home for 30 Euro. Ask the teacher. Eviews 5 has become available in 2004, but Eviews 4 commands should continue to work also in Eviews 5. A student version of Eviews 5 is not available. Eviews and its documentation are only available in English. The Faculty of Economics also employs English versions of the Windows operating system. That is why we have not attempted to write this introduction in Dutch. Please look up English words in the documentation that you really dont understand. After following the instructions in this document hands-on, you should be able to learn the operation of other procedures in Eviews more easily.

1.1

Intranet and Internet

This text and corresponding data and procedures will be available in dierent formats in on the internet site: http://www.feweb.vu.nl/econometriclinks/orientatie An extensive list of econometric links is permanently available on http://econometriclinks.com or http://www.feweb.vu.nl/econometriclinks It provides links to (collections) of online datasets, code, journals, textbooks and more. In particular http://www.feweb.vu.nl/econometriclinks/software.html presents an up-to-date list of links to econometric and statistical software packages, with links to dierent freely available Eviews tutorials. The Department of Econometrics of the Free University has its site on http://www.feweb.vu.nl/ectrie/ 1

The Eviews homepage is at http://www.eviews.com This is a very useful site, as software upgrades can be downloaded freely from its download section.

2
2.1

The software EViews

EViews, Econometric Views, is described in this text. Eviews is a widely used program in education, government and industry. It originated from the program TSP, Time Series Processor, http://www.tspintl.com. Nowadays, Eviews and TSP coexist, with Eviews specializing in user-friendliness and TSP specializing in speed, scientic documentation and numerical accuracy. Originally intended for time series data, Eviews is now also t for the analysis of crosssection data and panel data. Eviews provides most standard econometric techniques. The Eviews-terminology for its procedures follows the standard of most Econometrics textbooks. This terminology may dier slightly from the standard statistics-terminology. Moreover, some standard statistics procedures like the BoxPlots, are not (yet) implemented in Eviews 4. EViews produces pictures and tables that can easily be imported in word processing systems like Word, Scientic WorkPlace or MikTex.

2.2

EViews Basics

Eviews is started and closed like other Windows-programs. In a standard installation you will nd it via the Window Start-Menu on the bottom of your screen, for example via Start -Programs -Eviews -Eviews 4. The Main Eviews Window will pop up. Figure 1 shows this menu. The Main Menu options File, Edit, Window and Help follow the standard Windows-conventions. Objects, View, Procs, Quick and Options incorporate the special Eviews-features. The bottom line, or status line, of the main window provides additional info on the status of various settings of the program. The only relevant part at this point is the Path: information. This tells you in which folder Eviews looks for input les rst.

Figure 1: Eviews Main Window Eviews Help The Eviews Help-system is comprehensive, it encompasses nearly the entire users guide. Figure 2 shows the main help topics for Eviews novices. 2

Figure 2: Eviews 4 Help: Table of Contents The Chapters of the Eviews Windows Help books can be long. It is not possible to do a keyword search within a chapter in the Windows Help system. However the Help system also contains .pdf les which can be browsed by Chapter and which can be searched using Acrobat Reader, using Ctrl-F or the binocular icon within Acrobat Reader. Use the Help-function later to explore other options of EViews that are not discussed in this document. Standard Windows functionality The Introduction to Eviews of the Eviews-Help also explains Windows Basics for people who are not familiar with the Windows-system. Eviews works as a standard Windows-program, like Windows-Explorer. It has the advantage that all menus are text based, so you dont have to learn the meaning of new icons. Pointing and clicking with the mouse is sucient to browse through all the options and to perform many tasks. In standard circumstances left clicking on an item selects an Eviews object, right clicking a selected object opens a menu. Double clicking an object opens the specic Econometric View of an object. Many of the standard Windows editing-keys Ctrl-C (Copy), Ctrl-X (Cut), Ctrl-V (Paste), ctrl-Z (Undo last edit action) work as expected. A general Undo button to correct your own faults is not available and the program is not entirely robust on the network. Therefore: save your results regularly, as it is not impossible that the program crashes unexpectedly and all what has been done is lost!

2.3

An Eviews Basics Demonstration

By repeating the actions of this Demonstration you will get to know most of the basic concepts of Eviews: Workles, Objects, Series, Sample, Expressions, Series, Auto-Series, Groups, Graphs, compare the table of Contents in Figure 2. Knowing these basics is necessary to nd your own way in Eviews later on. We discuss how to 1. create Workles, setting a Sample and Import Data 2. save Workles 3. navigate most of the Eviews-windows 4. create simple Eviews Objects like Groups 5. use Expressions in the Command window 6. perform standard data tranformations 7. create, adjust and export Graphs At this stage of the course we cannot use econometric procedures, but some preliminary runs can be done with the data that are provided with the book. We can perform a straightforward data analysis: graphs can be made, correlations can be computed, data can be seasonally adjusted, etc. 2.3.1 Eviews Workles and Data Import

The Chapter titled Workle basics of the Eviews Help-system states: At the heart of the EViews design is the concept of an object. In brief, objects are collections of related information and operations that are bundled together into an easy-to-use unit. Virtually all of your work in EViews will involve using and manipulating various objects. EViews holds all of its objects in object containers. You can think of object containers as ling cabinets or organizers for the various objects with which you are working. The most important object container in EViews is the workle and your rst step in any project will be to create a new workle or to load an existing workle into memory. Therefore, the workle is the starting point in EViews. We create a new workle for every new project. Within workles one works with objects, like series, equations, groups of variables, graphs, tables, etc.. All these objects are part of the workle. So the rst thing to do in EViews is to create the workle and to import data. This will create the rst objects: a number of series. In the rst session one has to import data into the workle. An existing workle can be opened in any EViews-session later on. EViews can import data in a variety of formats, like ASCII-text and Excel. To import the data, one has to know how the data has been organized in the le. The data les for this course have the standard spreadsheet organization. The observations for the dierent variables are arranged in columns, with variables names on top.

2.3.2

Creating a Workle

Here is how you start working with Eviews on the ASCII dataset in the text le into Macro95.dat, which is available in the folder DataCarlsonThorne. This le contains quarterly data on 22 US macroeconomic variables running from 1959.1 to 1995.2. (Re)Start EViews in Windows. Click on File and New ; Then click Workle. A window appears with questions about the data: the frequency and the sample period. Click on the correct frequency (Quarterly) and ll in the correct Start date and End date. Click OK and change the name Untitled in Macro95, by using File SaveAs. Choose a directory where you have the permission to save your results! The workle gets the Eviews-extension .wf1. The new workle is now saved as Macro95. Do not forget to tick(mark) Update Default Directory. This changes the Path -Setting in the status line at the bottom of the Eviews Window. EViews will start the next session in the same directory you have just chosen, which is very convenient.

2.3.3

Importing Data

Now data can be imported in the workle. Click on Procs and Import and Read Text-LotusExcel (in our situation the data are in a text le) and browse for the le, Macro95.dat that has to be imported. Click OK, and Eviews asks you to select from a range of inputoptions, see Figure 3. EViews asks whether the data le is organized in Columns or in Rows (formerly called: by observation or by variable). If the data le contains the names of the variables, only the number of variables in the le has to be lled in. In our case there are 22 variables, the rst one being a date-variable, so we always know the dates of the corresponding observations in the same row of the data set.

Figure 3: Importing ASCII data from a textle There is also a preview window to check that the correct data will be imported and to see that the data really are organized in columns. If the data are not neatly organized in a matrix, do not mark File laid out as rectangle.

If your ASCII-data are not in the right format, you should edit them in a text-editor. Eviews contains a primitive text editor. Select File -Open -TextFile from the main menu and open Macro95.da0 (a copy of Macro95.dat, since the basic version of Eviews 4 does not allow opening .dat les in this way). to see how it looks in a text editor. Much better freely availble text-editors exist, see e.g. OxEdit, www.oxedit.com, an editor that is also installed in the computer lab. 2.3.4 Checking Imported Data and Saving the Workle

If all has been done correctly, the names of the variables appear in the workle window, otherwise wrong names like ser01, ser02, etc. are seen in the workle. The dierent variables are called Series in Eviews, even if the observations are not ordered according to time, in which case Eviews labels them as undated or irregular series. Series are Eviews Objects that we use most often. Statistical and econometric methods can now be used to analyze the variables. Extra data can be imported at a later stage. Double click a series name (with the left mouse button). A new window with its own buttons appears. See the options under View, check a spreadsheet View and a graphical view of the Series and see whether the data have been imported correctly. If your are satised with the imported data you should Save the Current Workle using File -Save from Eviews Main Window. To reuse your workle in a next session, start again by clicking on File. Then the names of the last used workles are visible in a history list and one can select directly a workle from this list. It is also possible to double click on the name of the workle in the Microsoft explorer or Norton Windows Commander. Then EViews starts with that workle if the extension (.wf1) has been associated with the EViews program in your le browser. 2.3.5 Eviews menus and Creation of Simple Objects: Groups and Equations

Once you have successfully loaded data into a workle it is time to browse through the special options of the main Eviews menu of Figure 1 again. The Object -drop-down menu allows you to create all types of new objects and to rename and delete existing objects. You can create a new Series if you want to type the data in yourself. The Quick -menu provides quick shortcuts for experienced Eviews users. A useful application of the Quick -menu is provided in 2.3.7 The drop down menus under View, Procs in the main Eviews menu are context sensitive: The range of options depends on the object(s) selected at the moment of the analysis. Similar View and Procs menus are also available in the windows of the objects themselves. Check this out for the a Series-object. Double click on a variable name in the Worklewindow: an Eviews-Series -window pops up. Under View youll nd the graphs and descriptive statistics that we need in a rst stage data analysis, under Procs youll nd more advanced options like Exponential Smoothing. Note that the same options are now available under View and Procs in the main Eviews menu.

Within a workle it is possible to create other Objects like Groups of variables. EviewsGroups are extremely useful objects. The Eview-Help says: A group is simply a list of series identiers. It is not a copy of the data in the series. Thus, if you change the data for one of the series in the group, you will see the changes reected in the group. A group is created very easily. With the left-handed mouse button you select one series rst. To create a group more series can be added to the selection by keeping Ctrl pushed down and clicking with the left-hand mouse button on the relevant variable names, after which it becomes a group by selecting Open selected (under View ), One window and Open group. See also Figure 8 below, which shows an example of a group selection in the (background) Workle-window. If you want to specify an equation -object you should click on the name on the left-handside variable of the equation rst. When you specify a regression equation later on it has to be the rst variable in the list. It is also possible to click on the selected variable names with the right-hand mouse button. Under View and Procs many possibilities arise to analyse the group (group statistics, various graph options, etc.); also regression equations are simply formed. After a procedure has been applied (e.g. least squares) one can proceed with the analysis with new possibilities given under View again (e.g. residual tests, coecient tests ). The best way to learn the various possibilities of a procedure is to click on the various buttons and to see what the possibilities are for the object in question. The structure is logical and the help function is often adequate, provided you know the basic terms an techniques, of course. Remember to save the workle at the end of the session, so that everything what has been done (transformed variables, equations, maybe groups etc.) is available in the next session. Each result (object) can be sent to the (laser) printer by clicking on Print in the toolbar of that particular window, but it is better to export them to another program like Word and make a simple report, before sending the results to the printer. 2.3.6 Command Window, Data Transformations and expressions

Transformations can most easily be specied in the white top-window, the so-called Command window. Type series lgcq = log(gcq) which generates a new series lcgq (after you press the Enter-key) and computes the logs of the consumption series gcq or use series dgcq = gcq - gcq(-1) which takes the rst dierences: it subtracts consumption in quarter t 1 from consumption in quarter t. Typing show dgcq in the Command-window brings up the Series-Window of dgcq. This is a quick way to start checking whether your transformation was successful. You can also transform series by clicking on the button Genr in the workle-window 7

(then e.g. only lgcq = log(gcq) has to be specied). Many other Eviews Expressions : transformations, functions and operators are available, see the Function Reference of the main Help(-Reference) menu. The Command Window keeps record of the commands by maintaining a history list. This can be an advantage if similar operations have to done repeatedly. Old command lines can be selected, adjusted and rerun: simply select the line in the command window and press the Enter -key. Most procedures can be applied in the command window. The Command reference of the main Help -menu provides all the commands. Once you know some commands this can save you time. For example: a quick way to re-import the data is to type read macro95.dat 22 in the command window, provided the le macro95.dat is in the current Eviews-Path (see status line at bottom Main Window). After generating many Eviews-subwindows it is sometime hard to nd the command window on your screen. You can always retrace the command window by clicking Window on the main menu of the main Eviews window. The command-window is always no. 1 in that list. 2.3.7 AutoSeries, Changing Data points

It is not always necessary to transform variables and save the results in a separate series. Transformed data enlarge the contents of the workle. It is also possible to use the AutoSeries (Automatically generated Series) feature of Eviews. For example, the specication of d(gcq) in the specication of a group or in the specication of an equation refers to the rst differences (i.e. the changes over 1 period) of gcq, without generating observations in a new variable permanently. We shall see how this works by looking at the eect of a change in one of the observations of the series gcq on the Auto-observations in the AutoSeries d(gcq): Generate a new group consisting of gcq and d(gcq): Type group transgcq gcq d(gcq) show transgcq in the command window. A new group, titled transgcq has been formed. Open the Spreadsheet View of transgcq: Click on View -Spreadsheet in the Group window of transgcq. Edit the values of one of the observations of gcq: Click on Edit+ in the Group window of transgcq, click on the cell of gcq that you want to change. The corresponding value appears near the top of the window. Change the value and push Enter on your keyboard. The corresponding values of d(gcq) change as well! Note that you cannot edit the values of d(gcq). Now you know why. Note that you also could have used Quick -button of the main menu to bring up the previous spreadsheet view. Use Quick -Show and type gcq d(gcq) in the resulting Show window and press OK. You can then change the name of the group by pushing the Name 8

button in the group-window and entering the name transgcq AutoSeries can also involve more complicated expressions like d(log(gcq)), which computes the rst dierences of the logs of gcq. 2.3.8 Examples of EViews output: Graphs and export Graphs to Word

First an example is given of Multiple Graphs, Line in an EViews Group window. The series gcq and gidpq have been opened as a group rst. These graphs are then created in the group window under View. See Figure 4. These graphs should correspond to Figures 17.1

Figure 4: Multiple Line Graphs from Group Window and 17.2 on pages 826 and 827 in the book of Carlson and Thorne. Check the sample if they do not correspond. In order to use the resulting picture in another program you should save it as a Windowsmetale with the extension .wmf : Use the Copy command under Edit in the main menu of the main Eviews-window, and choose Save Metale to disk or Copy to disk le, choose the .wmf format, then select a directory and choose a name, for example eviewsgraph1.wmf. The option Copy to clipboard is not advised. The picture can subsequently be imported in a Word document using Insert -Picture From File. See Figure 5 You can also Edit the graph with Eviews. Labels can be dragged to another place to a new place in the graph: Click on the label, hold the mouse button pressed and move the label. The label text can been changed as well: Double click on the label or use the right-hand mouse button. Clicking on the picture with the right-hand mouse button (or double clicking with the left-hand button: gives more options) produces an Options window to change the general outlook for dierent types of graphs, like a regression line in a scatter diagram or a zero

Figure 5: Inserting Eviews picture in .wmf format into Word line in a line diagram. See Figure 6 for Eviews 3 and Figure 7 4 for Eviews 4. . .

Figure 6: Graph Options Eviews 3

2.3.9

Scatter diagram

To create a scatter diagram select in the Group window: Graph, Scatter and the scatter type. Optional is the regression line. In Figure 8 there is no regression line. ;

10

Figure 7: Graph Options Eviews 4

Figure 8: A scatter plot from the Groups Window 2.3.10 Eviews Tables, editing and Exporting Tables to Word

At the end of this demonstration we show how to generate, edit, save and export numerical output of Eviews. (Re)Show the group transgcq that we created in the section on Autoseries. Look for transgcq in the Workle-window, or create it as a new group using Quick -Show and entering gcq d(gcq).

11

We create descriptive statistics from the pull-down menu under View in the Group window: View -Descriptive Stats -Common Sample. The standard descriptive statistics of Eviews appear. See Figure 9.

Figure 9: Descriptive Statistics in a Group Window Suppose now we want to edit and export this output. Then we should push the Freeze button of the Group-window, or we could use Object -Freeze Output from Eviews main menu. A Table -window pops up in a spreadsheet format. You can edit the output of the table after pushing the Edit+/- button in the Table-Window. You can also change the appearance by changing the font and cell-width in the table. Once you are satisied with the results you may want to export them to Word. First select the part of the table that you want to export with your mouse. Then export this part of the table using Edit -Copy from the main menu of Eviews, see Figure 10. Eviews will ask whether you want to export the numbers formatted or unformatted. Choose formatted if you do not want to do further computations with the numbers. Finally, import the Eviews-output in Word using Edit -Paste. See Figure 11. In Windowslanguage the last operations are described as: Copy the contents of the Eviews-table to the Windows-clipboard. Paste the contents from the Windows-clipboard into your Word document. The Paste-icon on Words main toolbar, shows a picture of the clipboard and a sheet of paper being removed, see Figure 11.

12

Figure 10: Exporting contents of an Eviews Table

Figure 11: Importing contents of an Eviews Table in Word

2.4

Advanced use of Eviews Help Examples: Maximum Likelihood

The help les for advanced objects often describe two ways to estimate models. The rst (user manual) examples show how to create an object using menus and then tell what to enter in the Object Window. Many other (reference manual) examples just tell you the Eviews-commands, i.e. the commands that should be entered in the command window. Here we describe how to use the Eviews Help examples for the LogL (loglikelihood) object and the corresponding ml (maximum likelihood) function, that is the most important member of this object. For user manual help, see Help -Contents -The Log likelihood (LogL) 13

Object. For reference manual help, see Help -Help Reference -Object Reference -Logl.

In order to use the examples of the manuals we have to create a workle and example series. We do this preparatory work in the command window: create series series series a x z y 1900 2000 = 3*@rnorm = 2*@rnorm^2+0.2 = x + z + @rnorm

Now we can start to apply the examples of the basic Eviews user Help system. The rst example creates the loglikelihood using the main menu. This is an example of the type that you nd near the beginning of the Help Chapters by user tasks and estimation methods in the Contents Index of the Help -Eviews Help topics System. To create a LogL Object use the menus Object -New Object -LogL. Then you should enter the following lines into your log likelihood object, to be more specic: in the Likelihood Specication View of the LogL object. @logl res = var = logl1 param logl1 y - c(1) - c(2)*x - c(3)*z c(4) * z^c(5) = log(@dnorm(res/@sqrt(var))) - log(var)/2 c(1) 1 c(2) 1 c(3) 1 c(4) 1 c(5) 1

Then simply click the Estimate button in the likelihood window toolbar to open the Estimation Options dialog. By default, EViews uses the values stored in the coecient vector, (c()), prior to estimation. The default values are zero, which results in zero values for the variance series var, which makes that the series of loglikelihood values logl1 cannot be computed and initial estimates cannot be obtained. If a param statement is included in the specication, the values specied in the statement will be used instead. This is a necessary addition to the Eviews example.

The second way is to create the loglikelihood object (i.c. with name hsmodel), add its members (i.c. the series logl1, res and var) and do the estimation only by commands in the command window. These commands can also be used in Eviews programs. logl hsmodel hsmodel.append hsmodel.append hsmodel.append hsmodel.append hsmodel.append hsmodel.ml

logl1 res = var = logl1 param

y - c(1) - c(2)*x - c(3)*z c(4) * z^c(5) = log(@dnorm(res/@sqrt(var))) - log(var)/2 c(1) 1 c(2) 1 c(3) 1 c(4) 1 c(5) 1

14

This is an example of the type that you nd in the Reference Manual: Help -Help Reference and that you nd near the end of the Help Chapters by Estimation Method in the Contents Index of the Help -Eviews Help topics System. Remember that these (object oriented) commands (with dots) should be typed in the Command Window! Window -Command.

Conclusion

This introduction is not complete. In the coming weeks we shall provide exercises to help you get the necessary skills and to help you understand the econometric concepts discussed in the relevant chapters of the textbook.

References
Anonymous (1997a). Eviews 3, Command and Programming Reference. Quantitative Micro Software. Anonymous (1997b). Eviews 3, Users Guide. Quantitative Micro Software. Vogelvang, B. (2005). Econometrics, Theory and Applications with Eviews. Harlow, England: Pearson Education.

15