Murat K. Munkin, "Count Roy Model with Finite Mixtures", Journal of
Applied Econometrics, Vol. 37, No. 6, 2022, pp. 1160-1181.

The data set is derived from eight annual surveys of the Medical
Expenditure Panel Survey (MEPS) conducted between 1996-2003.  MEPS is
publicly available at the Agency for Healthcare Research and Quality
(AHRQ). The sample is restricted to only unemployed individuals, aged
65 years and older, whose insurance status did not change during the
survey period and who are covered by Medicare, but not by Medicaid.

The data file meps.txt has 9818 observations and 40 variables. This
file is an ASCII file in DOS format, which is zipped in the file
mkm-data.zip. Unix/Linux users should use "unzip -a".

The data are saved in column, space delimited format with variables
appearing in the order listed below. Definitions and summary
statistics of these variables are given in Table 1 of the paper. Below
is a Matlab code to load the data set.

load meps.txt;
ssiratio = meps(:,1);	
year97 = meps(:,2);	
year98 = meps(:,3);	
year99 = meps(:,4);	
year00 = meps(:,5);	
year01 = meps(:,6);	
year02 = meps(:,7);	
year03 = meps(:,8);	
age = meps(:,9);	
famsze = meps(:,10);	
educyr = meps(:,11);
officebased = meps(:,12);
physician = meps(:,13);
nonphysician = meps(:,14);
drug = meps(:,15);
female = meps(:,16);
hisp = meps(:,17);
marry = meps(:,18);	
northe = meps(:,19);	
mwest = meps(:,20);	
south = meps(:,21);	
phylim = meps(:,22);
actlim = meps(:,23);
vegood = meps(:,24);	
good = meps(:,25);	
fair = meps(:,26);	
poor = meps(:,27);
msa = meps(:,28);	
income = meps(:,29);
injury = meps(:,30);
totchr = meps(:,31);
black = meps(:,32);	
supp = meps(:,33);	
chronic = meps(:,34);
chronic1 = meps(:,35);	
chronic2 = meps(:,36);	
chronic3 = meps(:,37);	
chronic4 = meps(:,38);
chronic4plus = meps(:,39);
chronic5plus = meps(:,40);