Choosing your computational toolkit

I work with large datasets, messy data handling, models, etc. I often get asked about the computational tools that are useful for dealing with these kinds of problems. There are `menu driven systems' where you click some buttons and get some work done - but these are useless for anything nontrivial. To do serious economics these days, you have to be writing computer programs. And this is true of any field - e.g. empirical macroeconomics - and not just of "computational finance" (which is a hot buzzword these days).

Attributes

You should pay attention to four elements: price, freedom, elegant and powerful computer science, and network effects.

Low price
Low price is better than high price. Price = 0 is obviously best of all.
Freedom
Freedom here is in many aspects. A good software system is one that doesn't tie me down in terms of hardware / OS, so that I'm able to keep moving. E.g. if I move from Linux to Apple, and I have to buy a fresh Stata binary, that drives up the cost. E.g. if I move from Linux to Apple, and a certain package isn't even available for Apple, that's not nice - it intrudes in my freedom of choice of platform.
Another aspect of freedom is in working with colleagues, collaborators and students. With commercial software, this becomes a problem, because your colleagues may not have the same software that you are using. Here free software really wins spectacularly.
Good practice in research involves a great accent on reproducibility. Reproducibility is important both so as to avoid mistakes, and because the next person working in your field should be standing on your shoulders. This requires an ability to release code. This is only possible with free software.
Computer science
Systems like SAS and Gauss use archaic computer science. The code is inelegant. The language is not powerful. In this day and age, writing C or fortran by hand is "too low level". Hell, with gauss, even a minimal thing like online help is tawdry. I like a system to be built by people who know their computer science - it should be an elegant, powerful language. All standard CS knowledge should be nicely in play to give me a gorgeous system. Good computer science gives you more productive humans.
Network effects
Lots of economists use gauss, and give out gauss source code, so there is a network effect in favour of gauss. A similar thing is right now happening with statisticians and R.

The alternatives

System Applicability Price Freedom CS Network effects
SAS SAS seeks to be a single monolithic system that will comprehensively do all your economics and statistics. Huge! Zero Terrible Some economists do write SAS code so I guess some useful SAS codes are floating around.
Stata Stata is a `younger, cleaner statistics package' as compared with SAS. High Zero More elegant than SAS, but really not a place you want to write programs. Quite a few economists use Stata, and there is a wealth of code coming out through `stata technical bulletins' (STB).
Gauss Gauss is the workhorse of econometricians worldwide. High None Gauss is really poor CS - worse than Stata but better than SAS. Strong
Ox Ox seeks to be a low price but unfree system to compete with Gauss. Zero But it's not GPL, and it is effectively only for Microsoft Windows. It looks like a clean language, better than gauss and stata. Small (there are some good stochastic volatility codes out there).
Octave (Matlab) Octave is a freeware matlab. So it's good for writing formulas in matrix notation. I felt it isn't good at dealing with files / string manipulation / regexes, and that it is not strong on probability and statistics. I also felt it's graphics are weak. Free Free Excellent A few people in macroeconomics are writing matlab codes, and I suppose these should work with Octave, but beyond that, No. (And if these codes use matlab toolboxes like `optimisation', then you're out of luck).
R R is a complete statistics package, integrating great computer science, matrix notation, good graphics, probability and statistics. Free Free Truly superb Weak - as yet, only a few economists are using R.
Grocer Grocer is an econometrics system written using scilab (a matrix scientific computation system). Free Free Fair Small

From 2003 onwards, I have been doing all my work in R. I am also looking at grocer with interest - some of the hard work that has gone into it is yielding invaluable functions which would be hard work to reimplement.

In addition, there is "background work" of massaging data, manipulating files, and so on. I do this using Unix tools: shell, shell prompt utilities, awk and perl.

Ajay Shah, December 2006