6.3 An example: Evaluating the impact of Medicaid expansion on uninsured rate

Let’s take a concrete example.

Say, we are interested in evaluating the impacts of ACA-Medicaid expansion on insurance outcomes. Following the supreme court decision in 2012 that deemed Medicaid expansion voluntary, 26 states expanded Medicaid in 2014, while the rest did not. In fact, 19 states did not expand Medicaid until 2018.

Data for this example comes from one of my projects and can be downloaded from github. We are looking at the county-level data, where our outcome variable is the uninsured rate. We want to evaluate the ATT. What is the effect of ACA-Medicaid expansion reform on uninsured rates?

We will draw our attention to the states that expanded Medicaid in 2014 plus the 19 states that did not expand Medicaid until 2018. States that expanded Medicaid between 2014 and 2018 are dropped from the sample. This is to avoid the case of bad comparison, an issue that arises from comparing early treated units with later treated units. We will reflect on this topic later.

For now, let’s take a quick look at the data.

user = 2
if(user == 1){
    source("/home/user1/Dropbox/Medicaid_South/code/filepath.r")
}else{
    source("/Users/vshrestha/Dropbox/Medicaid_South/code/filepath.r")
}


library(pacman)
p_load(fixest, dplyr, ggplot2, tidyverse, patchwork, arrow)


# load in county level uninsured rate data merged with other variables 
mort_allcauses <-  read_feather( file.path(datapath, "NVSS_data_county_2010to2017_merged_allcauses.feather"))  %>% 
                    mutate(treat = ifelse(is.na(treat) == T, "control 3", treat))  %>% 
                    filter(yearexpand == 2014 & age == 0 & race_name == "white")  %>% 
                    dplyr::select("countyfips", "year", "state.abb", "expand", "yearexpand", "sahieunins138", "GovernorisDemocrat1Yes")  %>% 
                    filter(duplicated(.))   %>%  
                    arrange(countyfips, year) # sort by countyfips and year
                     # the select() function is masked 
                    # by other packages, so use dplyr::select() instead 

# only keep the years 2013 and 2014 for the canonical case
dat_canonical  <- mort_allcauses  %>% 
                    filter(year %in% c(2013, 2014))

head(dat_canonical)

## # A tibble: 6 × 7
##   countyfips  year state.abb expand yearexpand sahieunins138 GovernorisDemocrat1Yes
##        <dbl> <dbl> <chr>      <dbl>      <dbl>         <dbl>                  <int>
## 1       1001  2013 AL             0       2014          39.6                      0
## 2       1001  2014 AL             0       2014          31.9                      0
## 3       1003  2013 AL             0       2014          45.1                      0
## 4       1003  2014 AL             0       2014          43.8                      0
## 5       1005  2013 AL             0       2014          37.3                      0
## 6       1005  2014 AL             0       2014          34                        0

dat_canonical  %>% tabyl(state.abb, expand)

##  state.abb   0   1
##         AL 134   0
##         AR   0 150
##         AZ   0  30
##         CA   0 115
##         CO   0 126
##         CT   0  16
##         DE   0   6
##         FL 134   0
##         GA 318   0
##         HI   0   8
##         IA   0 198
##         ID  88   0
##         IL   0 204
##         KS 210   0
##         KY   0 240
##         MA   0  28
##         MD   0  48
##         ME  32   0
##         MI   0 166
##         MN   0 174
##         MO 230   0
##         MS 163   0
##         NC 200   0
##         ND   0 106
##         NE 185   0
##         NH   0  20
##         NJ   0  42
##         NM   0  66
##         NV   0  33
##         NY   0 124
##         OH   0 176
##         OK 154   0
##         OR   0  72
##         RI   0  10
##         SC  92   0
##         SD 131   0
##         TN 190   0
##         TX 507   0
##         UT  58   0
##         VA 264   0
##         VT   0  28
##         WA   0  78
##         WI 144   0
##         WV   0 110
##         WY  46   0

cat("The expansion states are: \n")

## The expansion states are:

table(dat_canonical$state.abb[dat_canonical$expand == 1])

## 
##  AR  AZ  CA  CO  CT  DE  HI  IA  IL  KY  MA  MD  MI  MN  ND  NH  NJ  NM  NV  NY  OH  OR 
## 150  30 115 126  16   6   8 198 204 240  28  48 166 174 106  20  42  66  33 124 176  72 
##  RI  VT  WA  WV 
##  10  28  78 110

cat("The non-expansion states are: \n")

## The non-expansion states are:

table(dat_canonical$state.abb[dat_canonical$expand == 0])

## 
##  AL  FL  GA  ID  KS  ME  MO  MS  NC  NE  OK  SC  SD  TN  TX  UT  VA  WI  WY 
## 134 134 318  88 210  32 230 163 200 185 154  92 131 190 507  58 264 144  46

length(table(dat_canonical$state.abb[dat_canonical$expand == 1]))

## [1] 26

length(table(dat_canonical$state.abb[dat_canonical$expand == 0]))

## [1] 19

The two groups are as follows:

1. Expansion states (treated): AR, AZ, CA, CO, CT, DE, HI, IA, IL, KY, MA, MD, MI, MN, ND, NH, NJ, NM, NV, NY, OH, OR, RI, VT, WA, WV

2. Non-expansion states (control): AL, FL, GA, ID, KS, ME, MO, MS, NC, NE, OK, SC, SD, TN, TX, UT, VA, WI, WY