estimation
----------
The estimation stage uses two programs, a "master" program (called "meister_de" or "meister_us"), which controls the process of minimizing the criterion function, and a "slave" program (called "knecht_de" or "knecht_us") that solves the household's optimization problem and calculates simulated moments for a given parameter constellation.
You need exactly one instance of the master and at least one slave. As the slaves do all the work, most of which can be parallelized, it is a good idea to run several slave processes at once. I typically used 30 to 40 slaves; the maximum number that makes sense is 160. Slave processes shut down automatically once the master terminates.
Please note that you have to change the "knecht_**.java" file to include the address of the server on which the master program is run (in the line in which the SyncClient class is instantiated). If you are running the master an all slaves on the same machine, "localhost" will work. You may also have to change the port that is used (if you do so, change it in the "meister_**.java" program, too).
The whole optimization procedure is recorded in the log file "meisterlog.txt", which also contains the final estimation results.

derivatives
-----------
The architecture of the program calculating the derivatives is the same as in the estimation step. The master process will create files "de_deriv(N).csv" or "us_deriv(N).csv", where (N) is a number corresponding to the distance used in calculating the approximate derivative. When constructing the standard deviations reported in the paper, "**_deriv0.csv" was used.

analysis
--------
The simple program in "Analyse.java" calculates some statistics for a given scenario. "Stats.csv" contains aggregates by age of various variables, including assets, human capital, wages, and unemployment. The other files generated contain the tables summarizing the state and behaviour (claims to UI, reservation wage, hazard rate, human capital, etc.) of "typical" agents losing their job at a certain age and remaining unemployed. These tables were used to construct the reservation wage and hazard rate graphs in the paper.