gretl is a free, open-source econometrics package that pairs a point-and-click interface with its own scripting language for regression, time-series, and panel-data work, aimed at students, teachers, and researchers who want transparent, reproducible estimation.
gretl, short for GNU Regression, Econometrics and Time-series Library, is built for estimation rather than general statistics. You load a dataset, open the model dialog, and specify a dependent variable, regressors, instruments, or a time-series structure; the model window returns coefficients, standard errors, diagnostics, and test results you can extend with a single click. Behind that window sits hansl, gretl's own scripting language, which records what the menus did and lets you build loops, functions, and multi-model studies. The same program runs graphical and command-line sessions, so a classroom demonstration and a batch estimation job share one engine. Data arrives from CSV, Excel, Stata, and SPSS files, from SQLite databases, or from the sample datasets bundled with the package.
gretl is unusually transparent for a statistics program. Every estimation is written into a session log in plain script, so the full path from raw data to final table can be re-run from a single file, and results appear in a text window you can edit, save, or typeset. Graphs are generated as editable plot specifications rather than static pictures, which makes redrawing them with different labels or ranges quick. Packages from the project's own repository add extra estimators, tests, and import filters, and the program can hand data to R or Julia for work outside its scope. Users who prefer a menu-driven front end for psychology-style analyses often keep Jamovi Desktop nearby, exporting to it from gretl's CSV output.
Estimation speed matters less than knowing what was estimated, and that is where gretl pulls ahead of drag-and-drop statistics tools. A regression on panel data with fixed effects, a GARCH model on returns, or a cointegration test on macro series takes a few dialog choices, and each produces diagnostics alongside the coefficients instead of leaving you to hunt for them. The script that records those choices can be saved, versioned, and pasted into a paper's appendix, which turns reproducibility from an intention into a byproduct. Teaching benefits as well: students see the menu action and the generated command at the same time, so the syntax stops feeling like a separate subject. Because gretl is free and open source, an entire cohort can install it without licensing paperwork, and its modest footprint runs comfortably on older lab machines.
Key capabilities and highlights.
Dialogs cover ordinary least squares, instrumental variables and two-stage least squares, GMM, logit and probit, tobit, and survival-style limited dependent variable models. Each window collects the variables, sample restrictions, and robust standard error options in one place before the model is estimated.
gretl's built-in language handles loops, conditionals, user-defined functions, and string processing, so a repeated estimation across many series can be automated instead of retyped. Scripts run from the GUI or in batch mode, and the same syntax drives the program's own menu commands.
Every action in gretl is appended to a session script in readable commands. Save that script and the entire analysis can be replayed on a fresh copy of the data, which is exactly what a referee or a co-author needs to verify a reported coefficient.
ARIMA models, VAR and VECM systems, unit root tests, cointegration tests, and GARCH variants for volatility are all reachable from the model menu. Forecasting, impulse responses, and residual diagnostics appear in the same window as the estimates, keeping the workflow on one screen.
Pooled, fixed-effects, random-effects, and between estimators sit alongside specification tests such as Hausman, plus options for clustered and robust standard errors. Panel datasets can be structured by unit and period inside the program, so lagged and differenced variables are generated without manual editing.
A matrix engine supports linear algebra directly in hansl, letting you implement an estimator that no menu offers. Scripts can also build their own dialog boxes, so a custom routine behaves like a native gretl command once it is packaged.
CSV, Excel, Stata, SPSS, and gretl's own data files load without conversion, and SQLite databases open natively with other database engines reachable through ODBC. Import filters handle delimiter and date quirks, which saves the usual round of manual cleanup before an estimation.
Model tables export to LaTeX, RTF, CSV, or plain text, and graphs are stored as editable specifications that redraw at a different size or label set. Combined with the session log, that makes the jump from finished estimation to a formatted results table fairly short.
Browse historical releases and previous builds.
Common questions about gretl.
gretl is used for econometric estimation and testing: regression, instrumental variables, limited dependent variables, panel models, and time-series work such as ARIMA, VAR, and GARCH, along with the diagnostic tests that accompany them.
Yes. It is free and open-source software, so there is no licence fee for students, teachers, or research groups, and the source code can be inspected or modified under its licence terms.
No. Most analyses can be completed entirely through menus and dialog boxes. Scripting only becomes necessary when you want to automate a repeated task, write a custom estimator, or reproduce a study from a saved session file.
hansl is the scripting language built into gretl. It supports loops, functions, matrices, and string handling, and the graphical interface generates hansl commands as you click, so the language doubles as a learning aid and an automation route.
It can. Common formats including CSV, Excel, Stata, and SPSS load directly, its own data files open natively, and SQLite databases can be queried from inside the program, with ODBC available for other database sources.
Yes. Packages distributed through the project's own repository add extra estimators, tests, and import filters, and hansl's matrix engine lets you code a procedure yourself and expose it as a menu command.
Model output can be exported to LaTeX, RTF, CSV, or plain text, and graphs are saved as specifications you can redraw with different labels or dimensions, which suits journal submissions and teaching handouts alike.
It works well in a classroom because the generated script appears alongside each result, so students connect a menu action to the underlying command. Being free and open source also removes the licensing barrier for a whole cohort.
gretl is a free, open-source econometrics package that pairs a point-and-click interface with its own scripting language for regression, time-series, and panel-data work, aimed at students, teachers, and researchers who want transparent, reproducible estimation.
JASP is a free, open-source statistics application that runs classical and Bayesian analyses side by side in one live interface, aimed at students, researchers, and teachers who want readable output without writing R code.
Minitab Statistical Software 22 is a commercial statistics package built around quality improvement, combining guided analyses, control charts, process capability, and design of experiments so engineers, analysts, and students can turn process data into decisions without programming.
JASP is a free, open-source statistics application that runs classical and Bayesian analyses side by side in one live interface, built for students, researchers, and teachers who want readable output without writing R code.
Jamovi Desktop is a free, open-source statistics application with a spreadsheet-style workspace where analyses update live as you change variables, built for students, researchers, and analysts who need reliable results without writing code.
Comments