This document describes the SPSS scripts that combine all relevant observations for each country into the files that are used in the next stage ("de_equiv_p.sav", "de_equiv2_p.sav", "us_equiv_p.sav") and generate variables that are not available. Please note that you will have to adjust the paths used in read/write operations in these scripts. In file names, "**" stands for a 2-digit year, such as "93" or "01".

GSOEP:
The "status.SPS" and "status2.SPS" scripts generate a self-employment variable based on households' employment history. Then, the "create**.SPS" scripts combine all employment-related variables from the GSOEP files into files called "soep**.sav".
The "de_equiv**.SPS" scripts combine this information with variables contained in the CNEF into data files called "de_eqiv**.sav", which are then merged by"de_equiv.SPS" and "de_equiv2.SPS", and then finally amended with macro variables by "de_equiv_p.SPS". The final output files are "de_equiv_p.sav", which contains households from the 1991 to 1997 waves, and "de_equiv2_p.sav" containing the 2000 and 2001 waves. These latter two waves are the only ones with data on consumption; they are only used to estimate the preference parameter phi.

PSID:
Employment data are organized in the family files of the PSID. Create a mapping between household interview numbers and individual id's ("indi.sav") by running "indi.sps". Load the raw family data into SPSS (using the syntax files supplied with the data for 91 and 92 and "raw**syntax.SPS" for the other waves), construct the required variables using "raw2final**.SPS" and save the result as "psid**.sav". Then run "rawtofinalplus**.SPS" to add individual id's.
"us_equiv**.SPS" merges this information with the CNEF, while "us_equiv.SPS" combines the seven waves and "us_equiv_p.SPS" adds some macro variables to create the final output, "us_equiv_p.sav"

Please note that in the following stage, Stata will be used, so that the SPSS ".sav" files need to be converted into Stata's ".dta" format.
