This page contains information about Michael Creel's graduate
level econometrics notes. There are a couple of unusual thing about
You may be wondering why the notes are available in this form. It's
simply because I use a lot of free software, and this is a means of
contributing back to the community. Also, I'm hoping to receive
error corrections and contributions from users of the notes.
- they are available in editable form, so that you can modify
them to suit your needs. They are however copyrighted, so you
should learn about the GPL before
modifying and distributing them.
- they contain links that point to example
programs using the Octave
matrix programming language. Octave sometimes gets little
respect as a language for econometrics, with allegations that it
is slow compared to Matlab, etc. A couple of points:
- for plain Octave, these allegations are true, in some
respects at least
- Octave is free, and it runs on all popular operating systems
- plain Octave is perfectly good for ordinary econometrics of
the sort that most students and many researchers need to do
- Octave can easily be linked to parallel matrix algebra
libraries so that it uses all cores of a computer. Plain
Octave doesn't do that by default.
- Octave can be compiled using proprietary compilers, if you
- it is easy to use C++ code for bottlenecks. For serious
research work on computationally demanding problems, this step
- it is easy to use OpenMP or MPI to use Octave for parallel
computing. This can have a great performance effect, and your
code is free and portable.
- if you use these methods, you can get very good performance
- There are examples for stuff like OLS, ML, etc., but
there is also working (GPL'd) code for nonparametric
methods, simulation-based estimation methods, estimation of DSGE
models, and parallel programming applications in econometrics.
The examples make use of some other code for GNU Octave that I
have written - minimizers and other stuff, available
- the example programs together with data and all needed
software are on the accompanying econometrics.iso
image. This is an ISO image which can be used to boot a
computer, but which is more easily used to boot a virtual
computer, running under your usual operating system. This works
with all of the popular operating systems. Using this method,
you can be running the examples in very little time, without the
hassle of installing a lot of software.
A bit more detail: The notes are written using the LyX word processor, which allows
export to LaTeX. The document is about 500 pages. Coverage is
2 semesters of econometric theory, emphasizing models for stationary
data. Pure time series methods are very sparsely treated, and models
for nonstationary are completely absent. I would love to receive a
couple of contributed chapters in these areas.
- the notes,
in pdf form.
(about 1.3GB) contains the notes and all the examples,
ready to run. You don't need to install anything to use it
(except a virtualization platform, if you don't have one
installed). I test the image using Virtualbox,
which is a free download for Windows, MacOS, and Linux. There is
an appliance econometrics.ova.
To use the ISO image in a virtual machine, import the .ova
appliance file into Virtualbox, then in Settings, make the
Storage->IDE controller point to the location where you have
save the econometrics.iso file. If you would like the scripts
used to build this image, they are here
- all of the sources for the notes and examples are directly
To run the examples, you need to have Octave installed. To edit
the notes, you need to install LyX.