On Reproducible Econometric Research

Manuscript

This Web page provides accompanying material for the article

Koenker R, Zeileis A (2009). "On Reproducible Econometric Research," Journal of Applied Econometrics, 24(5), 833-847. doi:10.1002/jae.1083

An older technical report version (including some additional material) is available as

Koenker R, Zeileis A (2007). "Reproducible Econometric Research (A Critical Review of the State of the Art)," Report 60, Department of Statistics and Mathematics, Wirtschaftsuniversität Wien, Research Report Series. URL http://epub.wu-wien.ac.at/.

Resources

• Technology: The paper has been written in LaTeX, all empirical analyses and simulations have been carried out in R, employing a literate econometric practice approach and combining LaTeX and R based on Sweave.
• Document sources: The master source file (with combined LaTeX and R code) is JAE-RER.Rnw. From this file, the LaTeX file JAE-RER.tex is dynamically produced. This in turn can be compiled to JAE-RER.pdf. For compilation with LaTeX, our BibTeX data base repro.bib and the BibTeX style file jae.bst are required. (Additionally, the sources for the older technical report version are available as RER.Rnw.)
• Data: The data sets for the two empirical analyses from the paper are available as flat text files. The growth regression data from Durlauf & Johnson (1995) is available in data.dj (as obtained from the Journal of Applied Econometrics Data Archive). The wage-equation meta-model data from Koenker (1988) is available in its raw form as rk.raw and processed to CSV format (comma-separated values) in rk.csv. R code for the pre-processing is linked below.
• R code: Extended R code that not only re-produces the replication exercises from the paper but also provides more complete results for each case study is available in R session transcripts (.Rt files). These are flat text files containing R source code and comments:
  • cross-country growth regressions,
  • wage-equation meta-models,
  • power of fluctuation tests simulation.
• Version control: This Web page and all the resources provided above are under version control in a Subversion (SVN) repository. Current (and previous) versions are available for anonymous checkout from svn://svn-statmath.wu-wien.ac.at/repro/.
• R packages: R itself and all the packages employed for this project are available from the Comprehensive R Archive Network (CRAN). In particular, the following packages have been used: lmtest, sandwich, MASS. (Additionally, strucchange and lattice have been used in the technical report version.)

Related material

• An earlier technical report "Reproducible Econometric Research" by Roger Koenker is available in the following formats: [HTML], [PDF], [PostScript].
• Further material on reproducibility in econometrics research compiled by Roger Koenker is available.
• Another replication case study in forensic econometrics "Validating Multiple Structural Change Models" by Achim Zeileis and Christian Kleiber has been published in the Journal of Applied Econometrics as doi:10.1002/jae.856. An extended research report version is available from ePub WU.
• The R Project for Statistical Computing
• Journal of Applied Econometrics Data Archive
• CRAN Task View: Econometrics in R
• Journal of Statistical Software Special Volume: Econometrics in R