Readme file for the replication files of the paper 
„Should unemployment insurance be asset tested?“
by Sebastian Koehne, IIES Stockholm, sebastian.koehne@iies.su.se, and Moritz Kuhn, University of Bonn, mokuhn@uni-bonn.de
The replication files are organized in two sections. In the first section (Data), we explain how to reproduce the results on liquid asset holdings based on 2004 Survey of Consumer Finances data. In the second section (Code), we explain how to reproduce the quantitative results on the optimal UI design using a set of MATLAB files.

Data
Replicate empirical results
1. credit constraints and borrowing interest rate from section 5
2. results on liquid asset holdings from section 5.1 (Table 1)
3. Gini coefficients of liquid asset holdings from section 7.3 
All files are in folder data/dofiles
The first step to extract data from the raw data can be skipped and the provided data files can be used. In this case please start with step 3 that uses the provided data files, otherwise start at step 1 to extract the data from the raw SCF data files. The raw SCF data files have been downloaded from the Survey of Consumer Finances webpage (http://www.federalreserve.gov/econresdata/scf/scfindex.htm)
1. Run extractdata.do. This creates file scf2004wealth.dta. It calls scflabel.do during execution. It uses raw data of the SCF 2004 survey from folder rawdata to generate a data file with the consolidated household balance sheet positions.
2. Run creditcard.do. This creates creditcardinfo2004.dta. It uses raw data of the SCF 2004 from folder rawdata to generate a data file with credit card information.
3. Run liquidwealth.do. It merges data files scf2004wealth.dta and creditcardinfo2004.dta and generates the empirical results from the paper. Step 3 can be done without running steps 1 and 2 based on the provided data files.
The file liquidwealth.do displays as output the Gini coefficients of liquid wealth, the calibration targets for the credit limit, the interest rate on borrowing, and a table that corresponds to the first four columns of Table 1. 


Code
Replicate numerical results
1. Model calibration
2. Policy experiments to determine
a. Optimal replacement rate
b. Optimal degree of asset dependence
The policy experiments include computing the transitional dynamics.
3. Decomposition of the welfare effects. 
All files are in folder code. There are different subfolders for the benchmark model (folder main) and the model with Epstein-Zin preferences (folder EpsteinZin). To reproduce the results from the paper, open the respective folder and the MATLAB main.m file. The instructions apply to both cases.
Lines 23- 30 determine the code execution and are set to reproduce the results from the paper. All parameters and calibration targets are set starting from line 40. The initial setting of all parameters is such that results for the benchmark model (tables 2 - 4) are reproduced. Necessary parameter changes to generate robustness results from table 5 are reported in the main.m file. To reproduce the case with Epstein-Zin preferences the files in folder EpsteinZin must be used. The case of endogenous taxes in table 5 requires extending the code for a simple loop over a fine set of tax rates. The optimal tax rate is reported in the paper.
All results are exported to an Excel file temp.xls that is written to the current folder, e.g. the main folder in case of the benchmark model. The temp.xls file contains several work sheets that contain the results for different tables of the paper.
Results of Table 2 are taken from sheet replacement_rate. The columns on sheet replacement_rate correspond to the columns of Table 2
A:	benefit level
B:	median assets
C:	unemployment rate
D:	tax rate
E:	steady state welfare (the derivation of the welfare change in terms of equivalent variation has been done manually)
D:	welfare change as consumption equivalent variation including the transition

Results of Table 3 are on sheet asset_test. The columns on sheet asset_test correspond to the columns of Table 3
A:	slope of benefit function
B:	replacement rate at the borrowing constraint
C:	replacement rate at median asset holdings
D:	median asset holdings
E:	unemployment rate
D:	welfare change as consumption equivalent variation including the transition


Results of Table 4 are taken from sheet decomposition. The columns on sheet asset_test correspond to the columns of Table 4
A:	slope of benefit function
B:	inequality effect (percentage points)
C:	uncertainty effect (percentage points)
D:	level effect from consumption (percentage points)
E:	transition effect (percentage points)
F:	level effect from effort (percentage points)
G:	equivalent variation


Please contact any of the authors if you experience problems running these files.