The urban-rual gap in expenditure for health care by household, VHLSS 2018

Daily Graph Series

Data pre-processing and Visualization

# Clear R environment: 
rm(list = ls())

# Load some R packages (download from https://www.mediafire.com/file/ojayuymg68eraei/HO1_remove_some_columns.dta/file): 
library(dplyr)
library(stringr)
library(haven)

#==========================================
#  Data set: HO1_remove_some_columns.dta 
#==========================================

# Load data: 

read_dta("F:/HO1_remove_some_columns.dta") -> ho1

#------------------------
#  Create Household ID
#------------------------

# Function adds zero: 

add_zero <- function(x) {
  
  y <- case_when(x < 10 ~ str_c("0", x), TRUE ~ as.character(x))
  
  return(y)
}

# Some columns for creating household ID: 

code_components <- c("tinh", "huyen", "xa", "diaban", "hoso")

# Use the function: 

ho1 %>% 
  mutate(tinh_n = add_zero(tinh), 
         huyen_n = add_zero(huyen), 
         xa_n = add_zero(xa), 
         diaban_n = add_zero(diaban), 
         hoso_n = add_zero(hoso)) %>% 
  mutate(h_code = str_c(tinh_n, huyen_n, xa_n, diaban_n, hoso_n)) -> ho1

# Remove duplications: 

ho1 %>% filter(!duplicated(h_code)) -> ho1_nonDup

#--------------------------
#  Extract province names
#--------------------------

# Extract province info:
library(stringi)

ho1_nonDup$tinh %>% 
  attributes() %>% 
  .$labels %>% data.frame() -> df_province

# Rename for DF: 
names(df_province) <- "province_code"

# Create some columns and relabel for provinces: 

df_province %>% 
  mutate(province_vie = row.names(df_province)) %>% 
  mutate(province_eng = stri_trans_general(province_vie, "Latin-ASCII")) %>% 
  mutate(province_eng = str_replace_all(province_eng, "Tinh |Thanh pho ", "")) %>% 
  mutate(province_code = add_zero(province_code)) -> df_province

# Join the two data sets: 

full_join(ho1_nonDup, 
          df_province %>% select(province_code, province_eng), 
          by = c("tinh_n" = "province_code")) -> ho1_nonDup

# Relabel for region: 

ho1_nonDup %>% mutate(ttnt = case_when(ttnt == 1 ~ "Urban", TRUE ~ "Rural")) -> ho1_nonDup

# Calculate mean of financial aid and expenditure for health care (rural vs urban): 

ho1_nonDup %>% 
  group_by(ttnt, province_eng) %>% 
  summarise(avg_medical_aid = mean(m3c15, na.rm = TRUE)) %>% 
  ungroup() %>% 
  arrange(province_eng) -> medical_aid

ho1_nonDup %>% 
  group_by(ttnt, province_eng) %>% 
  summarise(avg_medical_exp = mean(m3ct, na.rm = TRUE)) %>% 
  ungroup() %>% 
  arrange(province_eng) -> medical_exp

# Calculate averall means: 

ho1_nonDup %>% 
  group_by(province_eng) %>% 
  summarise(Overall = mean(m3c15, na.rm = TRUE)) %>% 
  ungroup() %>% 
  arrange(province_eng) -> medical_overall_aid

ho1_nonDup %>% 
  group_by(province_eng) %>% 
  summarise(Overall = mean(m3ct, na.rm = TRUE)) %>% 
  ungroup() %>% 
  arrange(province_eng) -> medical_overall_exp

# Convert to wide form and add overall health expenditure: 

library(tidyr)

medical_aid %>% 
  spread(key = "ttnt", value = "avg_medical_aid") %>% 
  full_join(medical_overall_aid) -> medical_aid_wide

medical_exp %>% 
  spread(key = "ttnt", value = "avg_medical_exp") %>% 
  full_join(medical_overall_exp) -> medical_exp_wide

# Rearrange by overall expenditure: 

medical_aid_wide %>% 
  arrange(Overall) %>% 
  mutate(province_eng = factor(province_eng, province_eng)) -> medical_aid_wide

medical_exp_wide %>% 
  arrange(Overall) %>% 
  mutate(province_eng = factor(province_eng, province_eng)) -> medical_exp_wide

#=======================
#  Data Visualization
#=======================

# Load some R packages: 

library(ggeconodist) # install.packages("ggeconodist", repos = "https://cinc.rud.is")
library(ggplot2)
library(showtext)

# Select Ubuntu Condensed font: 
showtext.auto()
font_add_google(name = "Ubuntu Condensed", family = "ubu")
my_font <- "ubu"

# Figure 1: 

medical_aid_wide %>% 
  ggplot(aes(x = province_eng)) + 
  geom_econodist(aes(ymin = Rural, median = Overall, ymax = Urban), 
                 median_col = "firebrick", 
                 stat = "identity", 
                 median_point_size = 1.5, 
                 show.legend = TRUE) + 
  coord_flip() +
  theme_econodist() + 
  scale_y_continuous(expand = c(0, 0), limits = range(0, 6000), position = "right") + 
  labs(title = "Figure 1: The urban-rual gap in financial aid for health care by household, 2018", 
       caption = "Data Source: VHLSS 2018, GSO") +  
  theme(axis.title.y = element_blank()) + 
  theme(axis.text.y = element_text(family = my_font, size = 10)) + 
  theme(axis.text.x = element_text(family = my_font)) + 
  theme(plot.caption = element_text(family = my_font, size = 8, face = "italic")) + 
  theme(plot.title = element_text(family = my_font, size = 16)) -> g1 
  
grid.newpage()

g1 %>% 
  left_align(c("title", "caption")) %>% 
  add_econodist_legend(
    econodist_legend_grob(
      tenth_lab = "Rural", 
      ninetieth_lab = "Urban", 
      med_lab = "Overall", 
      med_col = "firebrick", 
      family = my_font, 
      label_size = 12,
    ), 
    below = "title"
  ) %>% 
  grid.draw()

# Figure 2: 

medical_exp_wide %>% 
  ggplot(aes(x = province_eng)) + 
  geom_econodist(aes(ymin = Rural, median = Overall, ymax = Urban), 
                 median_col = "firebrick", stat = "identity", median_point_size = 1.5, show.legend = TRUE) + 
  coord_flip() +
  theme_econodist() + 
  scale_y_continuous(expand = c(0, 0), limits = range(0, 12000), position = "right") + 
  labs(title = "Figure 2: The urban-rual gap in expenditure for health care by household, 2018", 
       caption = "Data Source: VHLSS 2018, GSO") +  
  theme(axis.title.y = element_blank()) + 
  theme(axis.text.y = element_text(family = my_font, size = 10)) + 
  theme(axis.text.x = element_text(family = my_font)) + 
  theme(plot.caption = element_text(family = my_font, size = 8, face = "italic")) + 
  theme(plot.title = element_text(family = my_font, size = 16)) -> g2 

grid.newpage()

g2 %>% 
  left_align(c("title", "caption")) %>% 
  add_econodist_legend(
    econodist_legend_grob(
      tenth_lab = "Rural", 
      ninetieth_lab = "Urban", 
      med_lab = "Overall", 
      med_col = "firebrick", 
      family = my_font, 
      label_size = 12,
    ), 
    below = "title"
  ) %>% 
  grid.draw()