************************************************************************** This file explains how to generate the results in "Bequest Motives and the Annuity Puzzle", by Lee Lockwood. ************************************************************************** GENERATING FIGURES 1-4: (Figures 1, 2, and 3 show the willingness to pay for annuities as a function of bequest motives. Figure 4 shows the relationship between different measures of the strength of bequest motives.) Run the Matlab m-file, BAP_wtp.m. This produces Figures 1-4 as well as the data underlying these figures. To run the file, open Matlab, ensure that BAP_wtp.m is in the current working directory (which can be ascertained by typing the command "pwd"), and then type "BAP_wtp" to execute the code. GENERATING TABLES 3 AND 5: (Table 3 shows summary statistics for the sample of single retirees in the HRS used to simulate the demand for annuities with different bequest motives from the saving literature. Table 5 shows the relationship between proxies for bequest motives and annuity ownership rates in the HRS.) 1. Visit the Health and Retirement Study (HRS) website, http://hrsonline.isr.umich.edu/index.php?p=reg, and, if necessary, register as a user. It is free to register. 2. Download the RAND HRS Data File (v.J), randJstataSE.zip, and follow the HRS's instructions by saving it to "C:\randhrs\". (Version J is the most recent version as of this writing.) 3. Download the 1992 HRS (Final) (v.2.0), h92core.zip. Unzip h92sta.zip and save HOUSEHLD.DO and HOUSEHLD.DCT to "C:\hrs1992\stata\". Unzip h92da.zip and save HOUSEHLD.DA to "C:\hrs1992\data\". Double-click HOUSEHLD.DO to create HOUSEHLD.DTA. 4. Download the 2006 HRS Core (Final) (v.2.0), h06core.zip. Unzip h06sta.zip and save H06Q_H.DO and H06Q_H.DCT to "C:\hrs2006\stata\". Unzip h06da.zip and H06Q_H.DA to "C:\hrs2006\data\". Double-click H06Q_H.DO to create H06Q_H.DTA. 5. Open the Stata do-files, Create_bmots_data.do Create_ann_data.do, and BAP_sim.do in a text editor. In each, modify the line that begins "cd " to point to the same working directory, e.g., "C:\Documents and Settings\Administrator\My Documents\Annuities\". Then double click Create_bmots_data.do, Create_ann_data.do, and, last, BAP_sim.do. BAP_sim.do produces a Stata log file, BAP_sim.smcl, that contains Tables 3 and 5. GENERATING TABLE 4: (Table 4 contains the simulated and empirical annuity ownership rates for the population of single retirees from the HRS.) To generate the simulated annuity ownership rates in Table 4, run the Matlab m-file, BAP_sim2.m. The two other Matlab m-files, Data_moments_BAP.m and Health_uncertainty.m, must be in the same directory as BAP_sim2.m. Set the three options at the top of the file to simulated annuity ownership under alternative models: with and without medical spending, with restrictions on the minimum size of annuities, and with fixed costs of buying annuities. The file contains directions for how to do this. The models used in the different columns of Table 4 in the paper are: column / medical spending / min annuity size / fixed cost 1 no 0 0 2 yes 0 0 3 yes 20k 0 4 yes 20k 2k The simulation (BAP_sim2.m) can take several days to run. To see the calibration of the "Hurd and Smith (2002)" bequest motive used in the simulation, run the Matlab m-file, Est_HS02.m. This shows that a linear bequest motive with theta = 25.5^(-2) matches Hurd and Smith's (2002) estimates of average anticipated bequests before and after the stock market boom in the 1990s (see their Table 10) almost perfectly: Expected bequests in wave 1: Model = 69.3, Data = 69.5 Expected bequests in wave 2: Model = 122.5, Data = 122.7 The do-file BAP_sim.do, discussed in the section on generating Tables 3 and 5, produces the empirical annuity ownership rate and selects the sample of single retirees from the HRS whose annuity demand is simulated.