README.txt


This is the replication archive for:

John M. Abowd, Francis Kramarz, Sebastien Perez-Duarte, and Ian M. Schmutte "Sorting Between and Within Industries: A Testable Model of Assortative Matching" (July 2014 version).


Last modified on 10 August 2014.


The raw data (in the "assets/" subdirectory) are moments from the 2004 LEHD Snapshot, the BEA National Income Accounts, JOLTS vacancy rates, and CPS unemployment. We convert raw data to analysis files using the code in the "build/" subdirectory. The resulting analysis files appear in the "data/" subdirectory. The program "progs/est_AKPS_main.m" runs all analysis programs (under the "progs/" subdirectory) on the analysis files and generates raw output. Publication tables and figures are extracted from the spreadsheet progs/orig_output/output_run_01.xls

"progs/est_AKPS_main.m" describes all code and data dependencies. The analysis requires the MATLAB Statistics and Optimization Toolboxes. We also use version 28.13 of the Adaptive Simulated Annealing (ASA) code (C++) provided by Lester Ingber (http://www.ingber.com/#ASA-CODE) along with version 1.39 of ASAMIN.m, Shinichi Sakata's MATLAB gateway function to ASA. For completeness, we include a pre-compiled version of the MEX file as part of this archive. Depending on your platform it may be necessary to rebuild asamin.mexZZZ. We have therefore also included the relevant versions of Ingber's and Sakata's source code with this archive (under the "assets/" subdirectory).


Contact Ian M. Schmutte (schmutte@uga.edu) with any questions.