gfic

This repository contains all .tex files and source code for the following paper:

• Minsu Chang & Francis J. DiTraglia, "A Generalized Focused Information Criterion for GMM", Journal of Applied Econometrics, forthcoming.

The file "main.pdf" contains a copy of the paper itself, including the online-only appendix at the end of the document.

Required Software and Packages

Calculations for the simulations and empirical examples were carried out in a mixture of R and C++ using the package Rcpp as an interface to C++ from R. A full list of the packages needed to replicate our results is as follows:

• MASS
• Matrix
• Rcpp
• RcppArmadillo
• parallel
• ggplot
• RColorBrewer
• tikzDevice
• plm

Note that plm is used only to load the data for our empirical example (see below for details) and ggplot, RcolorBrewer and tikzDevice are used only to generate the Figures (B1-B10) that appear in the online-only appendix. The package parallel is used only the generate the Random versus Fixed Effects results from the online-only appendix (Figures B1 and B2). For more information, see the comments beginning in line 15 of simulations/REvsFE/RUN_ME.R.

Simulation Studies

Code for our simulation studies appears in the directory simulations which contains three subdirectories:

• DynamicPanel contains code to replicate Table 2 from the body of the paper and figures B3 through B10 from the online Appendix.
• SRvsLR contains code to replicate Table 1 from the body of the paper.
• REvsFE contains code to replicate Figures B1 and B2 from the online Appendix.

To replicate any of these examples, navigate to the corresponding directory and create a subdirectory called results if it does not already exist. Then run the script RUN_ME.R.

Empirical Example

Our empirical example uses the dataset of Baltagi and Levin (1992) and Griffin and Xiong (2000), as provided by the plm package in R. The dataset is a panel of 1309 observations for 46 U.S. states over the period 1963-1992 containing the following variables:

• (1) state = State abbreviation.
• (2) year = Year.
• (3) price = Price per pack of cigarettes.
• (4) pop = Population.
• (5) pop16 = Population above the age of 16.
• (6) cpi = Consumer price index with (1983=100).
• (7) ndi = Per capita disposable income.
• (8) sales = Cigarette sales in packs per capita.
• (9) pimin = Minimum price in adjoining states per pack of cigarettes.

All code for our empirical example is contained in the directory empirical_example. Our code loads the dataset directly from the "plm" package. In case this dataset is excluded from future versions of the package, we include a .csv file Cigar.csv in the empirical_example directory. To replicate the empirical example, navigate to the appropriate directory and run the script RUN_ME.R