[gretl main page]

gretl databases textbook datasets big datafiles custom data collections data utilities

Terminology

In gretl parlance, a database is a (usually large) collection of series that may be of mixed frequencies and observation ranges. The data are stored in binary format, as single-precision "floats". You access such files via gretl's File/Databases menu. From there you can import selected series into gretl's workspace. Gretl databases have a .bin filename suffix, and are accompanied by an index file (suffix .idx) containing information on the series.

A gretl datafile, on the other hand, is a collection of series that is homogeneous with respect to frequency and observation range (though there may be missing values for some variables at some observations). Such a file may be read directly into gretl's workspace. Gretl datafiles typically have a .gdt suffix; by default they are in XML format, conforming to the Document Type Definition (DTD) given in gretldata.dtd.

For further details, please see the gretl manual.

1. gretl databases

The gretl distribution contains one sample database, a collection of US macroeconomic time series from the Federal Reserve Bank of St Louis. Several more databases are available for free downloading, or online access. Note that with gretl 0.40 and higher you can access these additional databases via the menu item File/Databases/On database server (provided you are connected to the internet).

2. Textbook datasets

Datasets and replication scripts for Ramu Ramanathan's Introductory Econometrics with Applications, and selected data from William Greene's Econometric Analysis, are provided with the base gretl package. Additional textbook datasets are available as follows.

Wooldridge: gretl offers support for the datasets used in Jeffrey Wooldridge's textbook, Introductory Econometrics: A Modern Approach. The datasets are available as a Windows self-installer, wooldridge_data.exe, or as a compressed archive, wooldridge_data.tar.gz.

Gujarati: gretl supports the datasets in Damodar Gujarati's Basic Econometrics. Please look here for details.

Stock and Watson: gretl supports the datasets in James Stock and Mark Watson's Introduction to Econometrics. Replication scripts are also provided. The package is available as a Windows self-installer, stock_watson.exe, or as a compressed archive, stock_watson.tar.gz.

New: The additional datasets for the second edition of Stock and Watson are now available for Windows, stock_watson_2.exe, or as a compressed archive, stock_watson_2.tar.gz.

Verbeek: The datasets for the second edition of Marno Verbeek's Guide to Modern Econometrics are available for Windows, verbeek_data.exe, or as a compressed archive, verbeek.tar.gz.

Davidson and MacKinnon: gretl supports the datasets in their Econometric Theory and Methods. The datasets are available as a Windows self-installer, ETM_data.exe, or as a compressed archive, ETM_data.tar.gz.

Hill, Griffiths and Lim: Lee Adkins provides gretl datasets and scripts for the third edition of Principles of Econometrics. A companion free pdf book is provided too: Using gretl for Principles of Econometrics.

With the Windows self-installers, just download them to the desktop then double-click on the installer icon. (Once the datasets are installed you can delete the installer.)

The gzipped archive versions should be decompressed in the gretl "share" directory. For example, if gretl is installed under /usr you could install the Stock and Watson data and scripts like this (as the root user):

cd /usr/share/gretl
wget http://ricardo.ecn.wfu.edu/pub/gretl/stock_watson.tar.gz
tar xvfz stock_watson.tar.gz
rm stock_watson.tar.gz

3. Big datafiles

The following packages each contain one or more gretl datafiles with accompanying header and label files. They should be unzipped into the user's gretl directory (by default $HOME/gretl on unix-type systems, or c:\userdata\gretl\user on MS Windows).

You should follow up the links in the first column above to get full details on these data sets. I am in the process of assembling my own notes on these packages; for the moment I have only done this for the Penn World Table.

Penn World Table

PWT56 is a rich macroeconomic panel dataset, spanning 152 countries over the years 1950 to 1992. The package for gretl comprises three main data files:

• pwt56.gdt: the entire pwt56. This dataset has many missing observations.

• pwt56_60_89.gdt: A panel data set of 120 countries for the 30 years 1960-89, containing 20 variables. This data set has no missing observations.

• pwt56_1985.gdt: More data are available for 1985 than for any other year, so I also made a pure cross-sectional data file for all 152 countries for this year. This file has some missing observations, but these will not be quite so difficult to deal with in a cross section as in a panel data set.

The package includes full documentation and sample scripts to analyse the data.

I suggest you consult the gretl manual for advice on dealing with panel data.

4. Custom data collections

If you use gretl in your teaching you may wish to set up a custom collection of data files (or scripts) for your students to access. Look here for details.

5. Data utilities

Besides offering pre-packaged databases, gretl comes with some scripts that can be used to create databases using data available via the internet. These can be found in the utils subdirectory of the gretl source package (see the gretl main page). To run the scripts you need to have perl installed on your computer, and you need to be connected to the internet.