One vote for Matlab
Which software programs (or programing skills) are most common/useful for economics phd students?
My first thoughts include statistical analysis packages & math programs: SPSS, SAS, MatLab, Mathematica, etc. Maybe excel too for basic stuff...
Thought I'd check the thoughts of people here too. If you think it doesn't really matter, say why.
Some obvious follow up questions might be:
Does anyone use any open source versions of this type of software?
Whats the best way to learn these programs? [to get comfortable with them before using them for a research project]
My Profile and Results Stanford GSB Finance '16
R is open source and is quite a powerful statistics package.
And Matlab is quite useful. At my university, one can buy a student license for a pretty low price, so perhaps you should look into that. As for the best way to learn, I would just play around with it, using the help menus when you get stuck. Alternatively, I expect there are some notes for a course in Matlab (or whatever you decide to learn) publicly available somewhere.
MATLAB is good for manual econometrics and useful for PhD metrics assignments if they want you to do a lot of bootstrap and montecarlo simulations by hand.
For general purpose econometrics software my vote is with Stata. It has a strong community that makes the base software more useful and exporting output into latex and other publication programs is quite efficient.
Attending UC Berkeley
MATLAB is all we've needed so far to do whatever econometrics and dynamic programming we've been asked to do for problem sets. I can't imagine anything requiring more than Matlab at this stage of the second semester (5 weeks of classes to go... and then it's prelim study time)
Also, what I would have found useful would have been to get familiar with LaTex. However, most of my classmates and I now use LyX which is just a great compromise to begin learning Tex.
I now hand in all weekly problem sets fully typed even though this is strictly required only for our Macro class... it makes it a lot easier to review them and have them checked for errors by my classmates before handing them in. It's also pretty sweet when you get to copy and paste large chunks from prior homeworks into current homeworks which require similar processes and you can just edit and submit.
I highly recommend it, and regret not typing all my first semester stuff, which is going to be much more difficult to review for prelims.
Dropbox is a necessity... Use my referral code and we'll both get a bonus... click Dropbox. Also, if you are going to join Amazon Student (free) or Student Prime I'll get a small credit if you use my referral link.
What is most useful depends on your interests. For anyone but a theorist, either SAS or Stata skills are very important. SPSS is not so useful because it seems to focus much more on ANOVA methods used by other social scientists rather than multiple regression methods. It maybe okay for experimentalists (lots of experimental psychologists use it) but better to learn SAS or Stata.
Matlab is certainly useful for macro and for doing econometrics that you can't find in canned SAS or Stata procedures. However, I personally have hardly used Matlab at all in my three years in a PhD program.
Certainly everyone should learn how to use LaTeX for writing papers and making slides. Since I'm an experimentalist I'll give a shout out to z-tree, which is great for me but totally useless for anyone not doing experiments.
Here's a big shout out to R. I use it predominately in my work as an RA and find it very useful. It has effective tutorials that get you started with the basics. Because it's open-source, there are a lot of user-written packages and help documentation that you can locate just by googling. The fact that it's free doesn't hurt either.
The one downside to R is that it's not very fast for large problems. Programs that operate at a more basic level like matlab or most notably Fortran are much faster. It just depends on what you want to be doing. If you are engaging in things like high dimension non-linear stochastic optimization, you'll probably want a faster program. The downside to these types of programs is that you will be writing many more lines of code and (Fortran especially) they are less intuitive.
A symbolic interpreter like mathematica is beneficial in that it can take derivatives and solve difficult algebra. One open source manifestation of this type of software is Yacas; though I believe that no one is working on debugging it any longer.
One effective method I've found for learning--beyond a very basic level--is to read a paper in which some computational step has been taken and try to duplicate the results. This has the added benefit of familiarizing yourself with some economic literature at the same time.
One vote for STATA and one for MATLAB. The latter is such a great program. Used for everything from econ to engineering. If you want another one, GAUSS.
For optimization, GAMS is the standard.
Regarding R. I tried for a long time to learn it, and I just can't. My roommate is an EE PhD student and can't even figure it out (he is a programming wiz).
Profile: here || Attending: UC Berkeley ARE
There are currently 1 users browsing this thread. (0 members and 1 guests)
Bookmarks