my_indicators <- c(
  int_pov_line = "SI.POV.DDAY",
  nat_pov_line = "SI.POV.NAHC",
  undernourishment = "SN.ITK.DEFC.ZS",
  lower_secondary = "SE.SEC.CMPT.LO.ZS",
  slums = "EN.POP.SLUM.UR.ZS",
  gdp_capita ="NY.GDP.PCAP.CD", 
  pop = "SP.POP.TOTL"
  )

Data Manipulation

In this section, we downloaded the whole data set for all of our variables, with the whole range of time covered by the World Bank. Then, we defined our main indicators: the proportion of the population living under the national poverty line (povna), the proportion of the population living under the international poverty line (povint); and the difference between both percentages from each country (povmix).It is important to note that the variable povmix is calculated subtracting the percentage of the international poverty line to the percentage of the national poverty line. This decision responds to the fact that national poverty lines are usually set at higher thresholds.

Applying filters, we also eliminated all the missing data for our variables in our investigation. Moreover, for our indicators, we used coding to reduce our data set to the latest data available for each country, thus assuring that our indicators have information on the highest number of countries possible. In this sense, we have not decided a start date nor and end date nor something like that. Instead, we worked with the whole time period, only extracting the data from the latest year available for each country and for each variable.

To make the work more visual and the data easier to understand, we created a series of tables, maps and graphs, showing and analyzing different bits of data. We will explain them as they show up along our chunks.

data_set <- wbstats::wb_data(my_indicators,) %>% 
  left_join(wbstats::wb_countries() %>% select(-country), by="iso3c")

data_set %>% 
  group_by(country) %>%
  drop_na(nat_pov_line) %>%
  drop_na(int_pov_line) %>% 
  drop_na(undernourishment) %>% 
  drop_na(lower_secondary) %>% 
  drop_na(slums)

## # A tibble: 87 x 27
## # Groups:   country [53]
##    iso2c.x iso3c country      date slums gdp_capita lower_secondary int_pov_line
##    <chr>   <chr> <chr>       <dbl> <dbl>      <dbl>           <dbl>        <dbl>
##  1 AR      ARG   Argentina    2016  14.7     12790.            89.2          1.1
##  2 AR      ARG   Argentina    2018  14.7     11795.            90.9          1.4
##  3 BF      BFA   Burkina Fa~  2018  57.1       805.            43.0         33.7
##  4 BD      BGD   Bangladesh   2005  70.8       499.            54.2         25.1
##  5 BD      BGD   Bangladesh   2010  61.6       781.            55.5         19.2
##  6 BD      BGD   Bangladesh   2016  49.4      1402.            79.4         14.3
##  7 BY      BLR   Belarus      2016  45.2      5040.           107.           0  
##  8 BY      BLR   Belarus      2018  33.2      6360.            97.8          0  
##  9 CN      CHN   China        2010  29.1      4550.            99.1         11.2
## 10 CI      CIV   Cote d'Ivo~  2018  60.1      2314.            49.4          9.2
## # ... with 77 more rows, and 19 more variables: nat_pov_line <dbl>,
## #   undernourishment <dbl>, pop <dbl>, iso2c.y <chr>, capital_city <chr>,
## #   longitude <dbl>, latitude <dbl>, region_iso3c <chr>, region_iso2c <chr>,
## #   region <chr>, admin_region_iso3c <chr>, admin_region_iso2c <chr>,
## #   admin_region <chr>, income_level_iso3c <chr>, income_level_iso2c <chr>,
## #   income_level <chr>, lending_type_iso3c <chr>, lending_type_iso2c <chr>,
## #   lending_type <chr>

povna = data_set %>%
     group_by(country) %>%
     drop_na(nat_pov_line) %>%
     arrange(-date) %>% 
     slice(1)  

povint = data_set %>%
     group_by(country) %>%
     drop_na(int_pov_line) %>%
     arrange(-date) %>% 
     slice(1) 

povmix = data_set %>%
     group_by(country) %>%
     drop_na(int_pov_line) %>%
     drop_na(nat_pov_line) %>%
     arrange(-date) %>% 
     slice(1) %>% 
     mutate(difference = nat_pov_line - int_pov_line)

#"Countries with the highest proportion of population living below the national poverty line. Arranged by region"
povna_reg = povna %>% 
  left_join(codelist %>% rename(GeoAreaCode=iso3n)) %>% 
  group_by(region) %>% 
  slice_max(nat_pov_line) %>% 
  select(region,iso3c,country, nat_pov_line, date) %>% 
  arrange(-nat_pov_line)

povna_reg %>%
  ungroup()  %>%
  mutate(date=as.integer(date)) %>%
  flextable::flextable() %>%
  flextable::colformat_double(digits = 2)

#"Countries with the highest proportion of population living below the international poverty line. Arranged by region"
povint_reg = povint %>% 
  left_join(codelist %>% rename(GeoAreaCode=iso3n)) %>% 
  group_by(region) %>% 
  slice_max(int_pov_line) %>% 
  select(region,iso3c,country, int_pov_line, date) %>% 
  arrange(-int_pov_line)
  
povint_reg %>% 
  ungroup()  %>%
  mutate(date=as.integer(date)) %>%
  flextable::flextable() %>%
  flextable::colformat_double(digits = 2)

In these two tables we show the highest values of each region. We can see that there is little difference in the percentages from one poverty line to the other. What may be striking is the fact that only one country repeats: Yemen. For the rest of regions, differences appear as to which country has the highest value. Another important fact to note is that the two countries of North America (Canda and the US) have not established any kind of National Poverty Line.

#"Countries with the highest difference between the percentages of national poverty line and international poverty line. Arranged by region"
povmix_reg = povmix %>% 
  left_join(codelist %>% rename(GeoAreaCode=iso3n)) %>% 
  group_by(region) %>% 
  slice_max(difference) %>% 
  select(region,iso3c,country, difference, date) %>% 
  arrange(-difference)

povmix_reg %>% 
    ungroup()  %>%
  mutate(date=as.integer(date)) %>%
  flextable::flextable() %>%
  flextable::colformat_double(digits = 2)

This table shows the countries with the highest difference between the percentages of the National Poverty Line and the International Poverty Line. As we have noted above, national poverty lines are usually set at higher thresholds. This explains the huge differences that appear in this table, highlighting Guatemala, Azerbaijan, Gambia and Yemen.

#"Countries with the lowest proportion of population living below the national poverty line. Arranged by region"
povna_reg2 = povna %>% 
  left_join(codelist %>% rename(GeoAreaCode=iso3n)) %>% 
  group_by(region) %>% 
  slice_min(nat_pov_line) %>% 
  select(region,iso3c,country,nat_pov_line, date) %>% 
  arrange(nat_pov_line)

povna_reg2 %>% 
  ungroup()  %>%
  mutate(date=as.integer(date)) %>%
  flextable::flextable() %>%
  flextable::colformat_double(digits = 2)

#"Countries with the lowest proportion of population living below the international poverty line. Arranged by region"
povint_reg2 = povint %>% 
  left_join(codelist %>% rename(GeoAreaCode=iso3n)) %>% 
  group_by(region) %>% 
  slice_min(int_pov_line) %>% 
  select(region,iso3c,country,int_pov_line, date) %>% 
  arrange(int_pov_line)

#"Countries with the lowest difference between national poverty line and international poverty line. Arranged by region"
povmix_reg2 = povmix %>% 
  left_join(codelist %>% rename(GeoAreaCode=iso3n)) %>% 
  group_by(region) %>% 
  slice_min(difference) %>% 
  select(region,iso3c,country,difference, date) %>% 
  arrange(difference)

povmix_reg2 %>% 
  ungroup()  %>%
  mutate(date=as.integer(date)) %>%
  flextable::flextable() %>%
  flextable::colformat_double(digits = 2)

For this series of data, we have decided to omit the table for the international poverty line. This is because it did not provide relevant data: it only showed countries with 0.0 values. The conclusions that we can extract here is that in effect the national poverty lines are way stricter than the international poverty line. Also, for the table showing the lowest difference in countries, we can see an interesting fact: negative values. This means that for these countries (Malawi, Solomon Islands and India), the national poverty lines are in fact set below the international poverty line, contrary to the common standard.

#"Countries with the highest proportion of population living below the national poverty line"
povna20L = povna %>%
     group_by(region) %>%
     ungroup() %>%
     arrange(-nat_pov_line) %>%
     select(region,country,nat_pov_line,date) %>%
     slice(1:20)

povna20L %>%
  ungroup()  %>%
  mutate(date=as.integer(date)) %>%
  flextable::flextable() %>%
  flextable::colformat_double(digits = 2)

#"Countries with the highest proportion of population living below the international poverty line"
povint20L = povint %>%
     group_by(region) %>%
     ungroup() %>%
     arrange(-int_pov_line) %>%
     select(region,country,int_pov_line,date) %>%
     slice(1:20)

povint20L %>%
  ungroup()  %>%
  mutate(date=as.integer(date)) %>%
  flextable::flextable() %>%
  flextable::colformat_double(digits = 2)

In these two tables we can see the countries with the highest percentages for both variables. Although there are differences between the two tables, it is clear that Sub-Saharan Africa predominates in both of them.

#"Countries with the lowest proportion of population living below the national poverty line"
povna20W = povna %>%
     group_by(region) %>%
     ungroup() %>%
     arrange(nat_pov_line) %>%
     select(region,country,nat_pov_line,date) %>%
     slice(1:15)

povna20W %>%
  ungroup()  %>%
  mutate(date=as.integer(date)) %>%
  flextable::flextable() %>%
  flextable::colformat_double(digits = 2)

#"Countries with the lowest proportion of population living below the international poverty line"
povint20W = povint %>%
     group_by(region) %>%
     ungroup() %>%
     arrange(int_pov_line) %>%
     select(region,country,int_pov_line,date) %>%
     slice(1:10)

povint20W %>%
  ungroup()  %>%
  mutate(date=as.integer(date)) %>%
  flextable::flextable() %>%
  flextable::colformat_double(digits = 2)

We have created these two tables to see once again the huge difference between national and international poverty lines. Both of them show the lowest values. However, while for the international poverty line there are a lot of countries with values of 0.0, the national poverty line shows a very different tendency. China is the only country which presents a value of 0.0, and the top 15 (Czech Republic), goes as high as 9.5%.

library(tmap)
data(World)

map = World %>% 
  filter(iso_a3 != "ATA") %>% 
  left_join(povna %>% rename(iso_a3=iso3c) %>%      select(iso_a3,country,nat_pov_line),
     by="iso_a3")

map %>% ggplot(aes(fill = nat_pov_line)) +
  geom_sf() +
  scale_fill_viridis_c() +
  theme(legend.position="bottom") +
  labs(
    title = "Map of National Poverty Line",
    fill = NULL,
    caption = paste("Source: World Bank") 
  ) +
   coord_sf()

This map is quite illustrative: it shows the distribution of poverty throughout the world according to the proportion of the population living below national poverty lines. We can see that the Global South presents worrysome results.

map = World %>% 
  filter(iso_a3 != "ATA") %>% 
  left_join(povint %>% rename(iso_a3=iso3c) %>%      select(iso_a3,country,int_pov_line),
     by="iso_a3")

map %>% ggplot(aes(fill = int_pov_line)) +
  geom_sf() +
  scale_fill_viridis_c() +
  theme(legend.position="bottom") +
  labs(
    title = "Map of International Poverty Line",
    fill = NULL,
    caption = paste("Source: World Bank") 
  ) +
   coord_sf()

This map illustrates the distribution of poverty according to the proportion of the population living below the international poverty line. We can see that the picture is considerably better if we compare it to the previous map. However, the trend persists in Sub-Saharan Africa, where results remain as worse as with the national poverty lines.

map = World %>%
  filter(iso_a3 != "ATA") %>%
  left_join(povmix %>% rename(iso_a3=iso3c) %>% select(iso_a3,country,difference),
     by="iso_a3")

map %>% ggplot(aes(fill = difference)) +
  geom_sf() +
  scale_fill_viridis_c() +
  theme(legend.position="bottom") +
  labs(
    title = "Map of the Difference between National and International Poverty Line",
    fill = NULL,
    caption = paste("Source: World Bank")
  ) +
   coord_sf()

This map illustrates the difference between both poverty lines along the world. We can see that the difference is higher in the Global South.

povint %>%
  group_by(region) %>% 
  arrange(int_pov_line) %>% 
  slice(1) %>% 
  select(region,country,gdp_capita) %>%  flextable::flextable() %>% 
  flextable::colformat_double(digits = 0)

povint %>%
  group_by(region) %>% 
  arrange(-int_pov_line) %>% 
  slice(1) %>% 
  select(region,country,gdp_capita) %>%  flextable::flextable() %>% 
  flextable::colformat_double(digits = 0)

These tables show the highest and lowest values of poverty (fixed to the poverty international line) for each region, next to its GDP figure. We can see that there is a strong trend relating GDP per capita and proportion of population living in poverty.

povna %>% 
  group_by(region) %>%  
  drop_na(pop) %>%      
  summarize(`Weighted average`=weighted.mean(nat_pov_line,pop,na.rm=TRUE),
 `Simple average`=mean(nat_pov_line,na.rm=TRUE),
 Countries=n(),
 Missing=sum(is.na(nat_pov_line)),
 `Pop (millions)`=sum(pop)/1e6
            ) %>%      
  arrange(`Weighted average`) %>% flextable::flextable() %>% 
  flextable::colformat_double(digits = 2) 

povint %>% 
  group_by(region) %>%  
  drop_na(pop) %>%      
  summarize(`Weighted average`=weighted.mean(int_pov_line,pop,na.rm=TRUE),
 `Simple average`=mean(int_pov_line,na.rm=TRUE),
 Countries=n(),
 Missing=sum(is.na(int_pov_line)),
 `Pop (millions)`=sum(pop)/1e6
            ) %>%      
  arrange(`Weighted average`) %>% flextable::flextable() %>% 
  flextable::colformat_double(digits = 2) 

We have created these tables to show the weighted and simple average of the different rates of poverty for each region in the world. We can see differences between both types of averages, the reason being the existence of bigger and smaller countries that distort the simple average. On the other hand, we can see that there exists a notable difference between the averages of the national poverty line (which are considerably higher) and the averages of the international poverty line.

povmix %>% 
  group_by(region) %>%  
  drop_na(pop) %>%      
  summarize(National=weighted.mean(nat_pov_line,pop,na.rm=TRUE),
 International=weighted.mean(int_pov_line,pop,na.rm=TRUE),
 Countries=n(),
 `Pop (millions)`=sum(pop)/1e6) %>% 
  ggplot(aes(
      x = International,
      y = National
      )) +
  geom_point(aes(size=`Pop (millions)`, color=region)) +
  geom_smooth(se = FALSE,method="lm") +
  geom_abline(slope=1,intercept=0) +
  scale_size_continuous(
    labels = scales::number_format(scale = 1/1e6, suffix = "m"),
    breaks = seq(1e8,1e9, 2e8),
    range = c(1,20)
    ) +
  theme_minimal() +
  labs(
    title = "Comparison between both poverty lines",
    x = "International Povery Line",
    y = "National Poverty Line",
    size = "Population",
    color = NULL,
    caption = "Source: World Bank, through `wbstats`"
   
  ) 

These graph illustrated the difference of averages that we have just mentioned. The more far from the black line, the greater the difference within that region. We can see that every region is placed above the black line, meaning that the averages of national poverty lines are always higher. The blue line represents the deviation of the total calculation.

Moreover, the great differences that we have found throughout this project can be seen very clearly. We see, for example, that the region of East Asia & Pacific is the one with the least proportion of population living in poverty and, at the same time, the one region presenting the least difference between averages. Something similar happens with the Sub-Saharan Africa region, although it is located in the opposite side of the poverty spectrum.

povmix %>%
  ggplot(aes(
      x = gdp_capita, 
      y = nat_pov_line
      )) +
  geom_point(aes(size=pop, color=region)) +
  geom_smooth(se = FALSE) +
  scale_x_log10(
    labels = scales::dollar_format(),
    breaks = scales::log_breaks(n = 4)
  ) + 
  scale_size_continuous(
    labels = scales::number_format(scale = 1/1e6, suffix = "m"),
    breaks = seq(1e8,1e9, 2e8),
    range = c(1,20)
    ) +
  theme_minimal() +
  labs(
    title = "National Poverty Line according to 
    GDP per capita",
    x = "GDP per Capita",
    y = "National Poverty Line",
    size = "Population",
    color = NULL,
    caption = "Source: World Bank, through `wbstats`"
  ) 

povmix %>%
  ggplot(aes(
      x = gdp_capita, 
      y = int_pov_line
      )) +
  geom_point(aes(size=pop, color=region)) +
  geom_smooth(se = FALSE) +
  scale_x_log10(
    labels = scales::dollar_format(),
    breaks = scales::log_breaks(n = 4)
  ) + 
  scale_size_continuous(
    labels = scales::number_format(scale = 1/1e6, suffix = "m"),
    breaks = seq(1e8,1e9, 2e8),
    range = c(1,20)
    ) +
  theme_minimal() +
  labs(
    title = "International Poverty Line according to 
    GDP per capita",
    x = "GDP per Capita",
    y = "Internatopnal Poverty Line",
    size = "Population",
    color = NULL,
    caption = "Source: World Bank, through `wbstats`"
  ) 

These two graphs are pretty illustrative: they show a clear negative correlation between GDP per capita and the proportion of population living in poverty. These graphs are the visual representation of the tables shown above, and they confirm that the higher the GDP per capita, the lower the poverty in the country.