Goldman, Dana and Nicole Maestas, "Medical Expenditure Risk and Household Portfolio Choice", Journal of Applied Econometrics, Vol. 28, No. 4, 2013, pp. 527-550. Overview: This article primarily uses the Health and Retirement Study (HRS) produced by the University of Michigan. The HRS Conditions of Use prohibit transfer of HRS data files to third parties. This article uses both publicly released data and restricted use data. Public release data can be accessed at: http://hrsonline.isr.umich.edu/index.php?p=reg. From this website, users can register for and download the RAND HRS versions of the data, listed under Data Products, "RAND Contributions." Information about the process for accessing restricted use data can be found at: http://hrsonline.isr.umich.edu/index.php?p=reslis Data Sources for Key Variables: HRS Public Release 1998 and 2000 waves--demographics, health status, income and wealth, pension balances, other variables HRS Restricted Release--county/state geographic identifiers CMS Medicare Managed Care Market Penetration Quarterly Data, st/cty files for 1998--number of Medicare beneficiaries in county and number of beneficiaries enrolled in a Medicare HMO Area Resource File 2003--county HMO (Medicare + Non-Medicare) market penetration in 1998, county population in 2000, average Medicare expenditure (A and B) in county in 2000 Processing Steps/Order of Programs: The file gm-progs.zip contains a number of program files, either SAS files or Stata do files. These are all ASCII files in DOS format. Unix/Linux users should use "unzip -a". 1. Make extract of HRS public files with age 65 + only To generate 1998 and prior data, run the following files in sequence: hinsur1.sas hinsur2.sas h00wave4.sas hinsur4.sas To generate 2000 data, run the following files in sequence: hinsur100.sas hinsur200.sas h00new.sas hinsur400.sas Convert output SAS data files to STATA files 2. Read in CMS data merge98.do 3. Merge HRS geocodes, CMS data, and HRS Public Release Data from Steps 1 and 2 hrs_cms_merge.do 4. Create pension wealth variables from HRS Public Release Data penwealth.sas, draws on initialize.sas and epdv.sas for macros 5. Read in 2000 Area Resource File and merge to HRS by county ARF0203.do HRS_Weiss_fsize_arf_qm.do 6. Read in additional HRS variables pull_nursinghome.do pull_more_hrs_vars.do 7. Merge files, clean and construct key analysis variables makevars.do 8. Run descriptives for tables descriptives 5-10-11.do 9. Run Main Models: Probit and DFML model ThreePart62_2POS_deciles.do, which calls LikeFunction2.do 10. Run alternative versions for DFML model for wealth tertiles and reclassifying bonds as safe ThreePart62_2POS_deciles_lowtert.do ThreePart62_2POS_deciles_midtert.do ThreePart62_2POS_deciles_hitert.do ThreePart62_2POS_deciles_bondssafe.do