This file describes the files attached to the paper Tough Love for Lazy Kids: Dynamic Insurance and Equal Bequests by Ctirad Slavik and Kevin Wiseman, Review of Economic Dynamics Ms. No.: RED-14-279R2, editor in charge: Matthias Doepke.

There are 3 sets of files, which all run under standard Windows operating system (Windows 7, for example). In addition, Matlab, Stata 13 and Microsoft Excel are needed to execute/view these files.

1. The file HRS data.xlsx contains the details of out datawork from the Health and Reirement Survey.

2. The folder PSID Data and Codes contains datafiles and codes used for our work with the PSID. The codes included are to be run in this order:
(i) The file data_construction.do, uses the data called Data_reshaped.dta (its construction is explained below) and creates 2 datasets: Data_panel_family.data (a household level dataset) and Data_panel_ID.dta (an individual level dataset). The file also uses the file called cpi_minwage.xlsx, which contains minimum wages and cpi indexes. Execution time approximately 10 seconds.
(ii) The file transfers.do uses Data_panel_family.dta and calculates some basic transfer statistics, as reported in section 2 in the paper. It saves the results in an excel sheet where additional calculations are being performed, as in the attached excel file called SRC_4table_results.xlsx. Execution time 5 seconds.
(iii) The file wealth.do uses the individual dataset Data_panel_ID.dta and calculates wealth statistics for the parents and wealth-wage correlation for the parents (lines 1-5 in Table 3 in the paper). The file also estimates the savings function for the kids, see the discussion in section 5.1 of the paper. Expected execution time 5 seconds.
(iv) The file intergenerational.do estimates the intergenerational wage persistence equation 5.2 (see alsolines 6-7 in Table 3) and uses the file use Data_panel_ID.dta. Execution time approximately 10 minutes.
(v) The file intragenerational.do estimates the life-time (intragenerational) wage persistence equation 5.3 (see alsolines 8-11 in table 3) and uses the file use Data_panel_family.dta. Execution time approximately 5 seconds.


Construction of the PSID data file Data_reshaped.dta
Cross year index provides an interface for users to select the variables they need, instead of downloading all the data sets. Link: http://simba.isr.umich.edu/VS/i.aspx
In order to construct a panel, one has to construct an identifier from ?Individual Data Index? and ?Family Public Data Index?, which identifies the individual dimension for panel data.
(a) Enter Individual Data Index, Survey Information/Interview Information, select id (interview) number, person number, relationship to head, and sequence number. Notes: The unique identifier is constructed according to PSID FAQ: (1968 ID multiplied by 1000) plus Person Number 68. Using the unique identifier, we can reshape the ?wide data? into ?long data?. Sequence number is used to determine whether the individual is in the family unit or not. To identify the head, one has to restrain the head to be in the FU (sequence number) and the relationship to head is head (=1 or 10, PSID changed its classification after some years).
(b) Select the variables of interest, then download the data, save it as Data_raw.dta
(c) Update KeyCode.xlsx file, put all variables according to the format provided by KeyCode.xlsx, one should go to Cross Year Index webpage to obtain the year and corresponding variable names in that year.
(d) Run reshape_data.m in Matlab, it tells you it will generate PSIDrename.do
(e) Run the file PSIDrename.do which reshapes the data from wide to long, using the unique identifier (1968 ID multiplied by 1000 plus Person Number 68). The output is Data_reshaped.dta


3. The folder Matlab Codes and Results contains Matlab codes used to numerically solve the model.
(i) The main code is called code_calib.m. This code solves for the optimal transfers in our model. On line 156 one can choose the public information version (first best) of the model, on line 157 the extended version of the model in which kids also leave transfers (see section 5.3 of the paper). The results are saved in the vector finout (and in the excel file called results.xls). The costs of equal division are calculated in the excel file kappa.xls.  Execution time of code_calib.m (using correct policies from the folder Policies2017) is 10 ? 15 seconds.
(ii) For a version of the model with the constraint that bequests/gifts have to be equal or non-negative, one needs to manually introduce a penalty on lines 14-24 in the code systemnew.m (lines 14- 19 in systemfb.m).
(iii) File code.m in the folder FiguresBench is used to generate Figure 1 in section 4. Execution time 1-2 seconds.
(iv) File codetheta.m in the folder FiguresSignals is used to generate Figure 3 and 4 in section 4 for the extended model with signals. Execution time approximately 30 seconds.