Practical Empirical Research Using gretl and hansl

• Author: Artur Tarassow
• DOI: http://doi.org/10.1111/1467-8462.12324
• Published date: 01 June 2019
• Date: 01 June 2019

The Australian Economic Review, vol. 52, no. 2, pp. 255–271 DOI: 10.1111/1467-8462.12324

Abstract

This article provides an introduction to the free

open-source statistics and econometrics soft-

ware gretl. The software is an attempt to bridge

the different demands for teaching at different

levels. Furthermore, it also provides function-

alities required for professional work. We also

introduce gretl’s powerful and intuitive script-

ing language ‘hansl’. This article can be seen

as an illustrated step-by-step guide that uses a

real-life data analysis example. Finally, I

estimate a vector autoregressive time-series

model using gretl for structural analyses as

well as for forecasting purposes.

1. Introduction

Data analysis is a vital part of the economic

training and daily life of a professional

economist. Economists are trained to be

sophisticated users of data, and econometrics

—the interplay of economic theory, statistics

and mathematics—is a subject which econom-

ics students usually have to study. Under-

graduates studying economics are typically

exposed to courses in statistics and economet-

rics, covering the practice and interpretation of

least-squares regression, data management and

data visualisation. Graduate students learn

advanced treatments of the subject, likely

involving Maximum Likelihood estimation,

simulation and so on. Professional economists

—whether in public or private institutions—

are likely to find themselves using, or possibly

developing, econometric methods.

Gretl is an open-source and freely available

statistics and econometrics software that

attempts to bridge the various demands for

teaching at different levels as well as for

professional work. It comprises a full-featured

graphical interface but also a powerful script-

ing language called ‘hansl’. This scripting

language can be seen as a domain-specific

language for statistics and econometrics. Gretl

handles datasets—a matrix but with a richer

structure including some meta-information—

similar to Eviews but can also deal with pure

matrices and arrays such as Matlab, Gauss and

Julia for advanced programming purposes.

In the following, I will introduce gretl—and

to some extend hansl—alongside a practical

empirical example using time-series data. The

article is structured as follows. The next section

discusses the econometrics software market

and presents gretl and hansl. Section 3

introduces gretl’s graphical user interface and

* Tarassow: Department of Socioeconomics, Welckerstr.

8, University of Hamburg, 20354 Hamburg, Germany,

email <atecon@cryptolab.net>. I am grateful to Allin

Cottrell, Riccardo (Jack) Lucchetti, Sven Schreiber and

Jenny Lye for helpful comments. Any remaining errors or

omissions are strictly my own. I gratefully acknowledge

financial support from the Graduate School at the

University of Hamburg and the hospitality of the

University of Melbourne during a visit in March 2017.

°

C2019 The University of Melbourne, Melbourne Institute: Applied Economic & Social Research,

Faculty of Business and Economics

Published by John Wiley & Sons Australia, Ltd

For the Student

Practical Empirical Research Using gretl and hansl

Artur Tarassow

256 The Australian Economic Review June 2019

explains how to conduct basic data manage-

ment as well as some initial analyses by

providing an overview about essential com-

mands. In Section 4 a vector autoregressive

time-series model is estimated and I will show

how the model’s dynamics can be analysed for

structural or forecasting purposes. Section 5

briefly introduces the work with matrices

before Section 6 summarises the article.

2. Econometric Software and gretl

2.1 The Econometrics Software ‘Market’

The demands placed on an econometric

software are a delicate issue. For teaching

purposes it makes sense to have undergraduate

students work with reasonably user-friendly

software—in the easiest case controlling the

software by a graphical user interface through

point-and-click using the mouse. However, it

makes little sense to introduce a software to

undergraduates that does not support more

advanced methods and does not offer a fully-

fledged scripting and programming environ-

ment—aspects that are required for an

advanced treatment in graduate classes and

for professionals. And of course there is a

premium on teaching ‘marketable skills’rather

than dead-end expertise.

Gretl is a program that attempts to bridge

these different demands. It comprises a full-

featured graphical user interface (GUI): its

underlying functionality can be driven either

by gretl’s scripting language or by the

apparatus of menus, dialogue boxes and so

on. The lead developers—Allin Cottrell from

Wake Forest University and Riccardo (Jack)

Lucchetti from Universit

a Politecnica delle

Marche—attempt to ensure that almost every-

thing that can be done via scripting can also be

done via the GUI, and vice versa, with only a

few exceptions (Cottrell 2017).

Gretl’s main competitors are the major

proprietary econometrics packages, Stata and

Eviews, and also the major open-source

statistical software project, R. As in gretl,

datasets and series are also basic in Stata and

Eviews, and econometric functionality is

supported by a wide range of built-in

commands. Even though both Stata and

Eviews support scripting, their respective

languages are quite odd from the point of

view of a programmer used to general-purpose

scripting languages or Matlab-like interfaces

for matrix manipulation. Gretl offers the

common apparatus of fully-fledged program-

ming languages (function-calling, function-

definition, declaration of and assignment to

named variables of various types) similar to

what is available in Matlab, R and Python.

2.2 gretl and hansl

Gretl is an acronym for Gnu Regression,

Econometrics and Time-series Library. The

software is available for Windows, Mac OS X

as well as Linux through the official gretl

homepage.

1

It is free, open-source software

that may be redistributed and/or modified

under the terms of the GNU General Public

License as published by the Free Software

Foundation.

Gretl comprises a common command line

(CLI) program and a GUI client, and makes use

of reliable free and open-source software

packages in the ‘background’for computa-

tions, for example, (multi-threaded) LAPACK/

BLAS, fftw, GTK, gnuplot, etc. The first gretl

version was released in January 2000 and has

been under active development since then. Its

user interface is available in 17 languages

thanks to an active community. Gretl is shipped

with a User Guide of 380þpages, a Command

Reference of 210þpages and a ‘A hansl

Primer’tutorial to gretl’s scripting language.

The latest stable release is version 2019a which

was published on 24 January 2019.

Gretl’s unique selling point is that it offers a

high-level matrix-oriented language with a

syntax similar to Matlab or Gauss, and an

intuitive high-level scripting language called

hansl that is attuned to econometrics. This

makes gretl especially interesting for lecturing

statistics and econometrics but also for serious

research.

Hansl—Hansl’sANeatScripting Language

—was developed over time and is now a very

advanced high-level scripting language with

more than 140 commands specifically tailored

°

C2019 The University of Melbourne, Melbourne Institute: Applied Economic & Social Research, Faculty of Business and

Economics