R for economists

Documents, books

• Jeff Racine and Rob Hyndman have an article Using R to Teach Econometrics, Journal of Applied Econometrics, Vol. 17, No. 2, 2002, pp. 149-174. [Associated files]
• Mahmood Arai has written a useful document A brief guide to R for beginners in Econometrics.
• Econ 472, at UIUC, has a nice website which has many examples in R.
• An introduction to modern bayesian econometrics by Tony Lancaster, Blackwell, May 2004. The book uses R.
• I have heard good things about a book which focuses on MLE: Yudi Pawitan (2001), In all Likelihood: Statistical Modelling and Inference using Likelihood, Clarendon Press, Oxford. The associated website has a good set of R codes.
• Edward W. Frees has a book project Longitudinal and Panel Data: Analysis and Applications for the Social Sciences with an associated web page with R codes.
• See my set of files R by example.
• You might find my web page Guidelines for projects based on R to be useful.

CRAN Packages

There are a large number of CRAN packages that are relevant for economists. The range of finance and econometrics which is now directly feasible in R using existing CRAN libraries is amazing. There is now a concept of CRAN views where R gurus keep track of all the packages out there and tell you which are pertinent for you. Two specific views are of interest, Finance and Econometrics.

In order to get started, you need to do the following steps once:

$ sudo R
> install.packages("ctv")
> library(ctv)
> install.views("Finance")
> install.views("Econometrics")

Once you've done this, from here on, in order to stay updated, you must do:

$ sudo R
> library(ctv)
> update.ctv <- function(view, force.bundles = FALSE) {
    f <- function(x) if (x[1] == view) x$packagelist$name
    ctv.pkgs <- unlist( sapply( CRAN.views(), f) )
    idx <- if (force.bundles) 1 else c(1,5)
    installed.pkgs <- c(installed.packages()[,idx])
    dnld.packages <- setdiff(ctv.pkgs, installed.pkgs)
    install.packages(dnld.packages, dependencies = TRUE)
    update.packages()
  }
> update.ctv("Finance")
> update.ctv("Econometrics")

If you are on Mac OS X, you need a slightly different version --

> update.ctv <- function(view, force.bundles = FALSE) {
    f <- function(x) if (x[1] == view) x$packagelist$name
    ctv.pkgs <- unlist( sapply( CRAN.views(), f) )
    idx <- if (force.bundles) 1 else c(1,5)
    installed.pkgs <- c(installed.packages()[,idx])
    dnld.packages <- setdiff(ctv.pkgs, installed.pkgs)
    install.packages(dnld.packages, type="source", dependencies = TRUE)
    update.packages()
  }

Using update.ctv(), any packages/bundles in the view that are not installed are installed and then all installed packages, not just ones in the view, are updated to their most recent version. When force.bundles is TRUE, this forces reinstallation of all bundles in the view. If any packages in a bundle from a view were manually removed without removing all packages in that bundle then this will be necessary in order to restore the entire bundle. The function update.ctv() is by Gabor Grothendieck and Achim Zeileis.

Time series econometrics

• Rob Hyndman has written R functions for doing forecasting.
• A package for multivariate GARCH. At present, it only does simple bivariate gaussian problems (which is also cool!) and says it is at version 0.05.

Finance

• Rmetrics is an ambitious attempt at creating a full finance software system, targeting both research users and practitioners, in R.
• Burns Statistics, does a good deal of interesting finance using R, has source code, datasets, etc. for download. In particular, look at their working papers page.

Other econometrics software

• Tools for bayesian analysis, computation, and communication.

Mailing lists

If you are a serious R user, then subscribing to the mailing lists is strongly recommended. They are freely accessed using your web browser. I am on two of the lists: R-help (Main R Mailing List: Primary help) and R-sig-finance (Special Interest Group for 'R in Finance').

Searching mailing lists is a great tool for finding out stuff, even if you aren't reading them regularly.

Return to R by example

Ajay Shah
ajayshah at mayin dot org