Sustainable Finance Assignment 2

Key Findings

• Although it varies by region, there is a positive relationship between sovereign debt and consumption emissions (scope 1, 2, & 3 emissions).

• The United States, China, and Japan have the highest CO2 emissions financed by sovereign debt.

• There is a positive relationship between PPP-adjusted GDP, sovereign debt, and CO2 emissions financed by sovereign debt.

Introduction

PCAF has recently released a new framework for calculating different types of emissions. This paper specifically looks at how to what this new framework means for sovereign debt. Using data from various organizations it calculates overall consumption emissions for different sovereigns and uses this information to calculate emissions financed from sovereign debt. This information is then used to calculate the relationships between emissions from sovereign debt, PPP adjusted GDP, regions of the world, and sovereign debt. I find that there is a slightly postive relationship between sovereign debt and total consumption emissions, that the United States and China have the most emissions financed by sovereign debt, and that PPP adjusted GDP, sovereign debt, and emissions from sovereign debt have all shown a positive correlation over time.

Body

There is a slightly positive relationship between sovereign debt and consumption emissions, although it varies by region. Figure 1 shows the relationship between consumption emissions and sovereign debt by country. A slight correlation can by seen, especially in South Asia, North America, and Europe and Central Asia. However, East Asia and the Pacific and Latin America show more unclear relationships. When looking at Figure 2, the regional differences in sovereign debt and consumption emissions is even more profound. North America and East Asia and the Pacific stand out as having far more sovereign debt and consumption emissions that the rest of the world. However, while North American consumption emissions are relatively constant no matter their debt, East Asia and the Pacific seems to see more emissions when it has more debt. There is definitely at least some kind of correlation between debt and consumption emissions, even if it is not a very strong one. This makes sense because debt can typically imply more production in many cases. This could impact the way that sovereign debt is seen in the future, especially as emissions become more important.

Code

#| error: false
#| output: false
#| warning: false
#| include: false
#| message: false

library(tidyverse) 
library(janitor)
library(here)
library(readxl)
library(tidyr)
library(dplyr)

library(rnaturalearth)
library(countrycode)
library(wbstats)
library(ggplot2)
library(gganimate)
#install.packages('gganimate')
library(gapminder)
#install.packages('gifski')
library(gifski)
#Here, I read in data for domestic emissions from 1990 to 2018.
scope_1_domestic <- read_csv("~/Raw Data Finace Project 2/Scope 1 Emissions.csv")


#I omit empty data entries because they have no value at this point.

scope_1_domestic <- scope_1_domestic %>% 
  na.omit()

#I change data type so that emissions in the year 1990 are seen as numerical and not categorical. This will allow me to better sort the data later.

scope_1_domestic$"1990" <- as.numeric(as.character(scope_1_domestic$"1990"))  # Convert one variable to numeric

#I clean the data so that year is one column and domestic emissions are one column.

scope_1_domestic <- scope_1_domestic %>% 
  pivot_longer("1990":"2018",
    names_to = "year",
    values_to = "domestic_emissions"
  )

#I rename the data so that it is standardized.

scope_1_domestic <- scope_1_domestic %>% 
  rename("country_name" = "Country/Region")

#Create country code index function.

country_name_regex_to_iso3c <- function(country_name) {
  country_name |>
    countrycode(origin = "country.name", 
                                     destination = "iso3c",
                                     origin_regex = TRUE)
}

#Create iso3c index function.

iso3c_to_country_name <- function(iso3c) {
  iso3c |>
  countrycode(origin = "iso3c", destination = "country.name")
}


#I add a standardized country name and ISO code for each country.

scope_1_domestic <- scope_1_domestic %>% 
  mutate(iso3c = country_name_regex_to_iso3c(country_name)) |>
  mutate(country_name = iso3c_to_country_name(iso3c)) 

#Download data for export emissions for each country from 1995-2018.

scope_1_export <- read_excel("~/Raw Data Finace Project 2/Export Emissions!.xlsx", skip = 6)

#I clean the data so that year is one column and export emissions are one column.

 scope_1_export <- scope_1_export %>% 
  pivot_longer("1995":"2018",
    names_to = "year",
    values_to = "export_emissions"
  )
 
 #Standardize column names,
 
 scope_1_export <- scope_1_export %>% 
  rename("country_name" = "Country")
 
 #Standardize country names and add country code.
 
 scope_1_export <- scope_1_export %>% 
  mutate(iso3c = country_name_regex_to_iso3c(country_name)) |>
  mutate(country_name = iso3c_to_country_name(iso3c)) 
 
 #Delete empty data entries.
 
 scope_1_export <- scope_1_export %>% 
  na.omit()
 
#Join domestic emissions and export emissions so that I now have all scope 1 emissions data.
 
 scope_1_and_export <- 
  full_join(scope_1_domestic, scope_1_export, by = c("year" = "year", "country_name" = "country_name", "iso3c" = "iso3c"))
 
 #Read in scope 2 emissions info from data base.
 
 scope_2 <- read_excel("~/Raw Data Finace Project 2/Scope 2 Emissions.xlsx", skip = 6)
 
 #I clean the data so that year is one column and scope 2 emissions are one column.
 
 scope_2 <- scope_2 %>% 
  pivot_longer("1995":"2018",
    names_to = "year",
    values_to = "scope_2_emissions"
  )
 
 #Standardize column and country names and add ISO index.
 
 scope_2 <- scope_2 %>% 
  rename("country_name" = "Country")

scope_2 <- scope_2 %>% 
  mutate(iso3c = country_name_regex_to_iso3c(country_name)) |>
  mutate(country_name = iso3c_to_country_name(iso3c)) 

#Omit missing data

scope_2 <- scope_2 %>% 
  na.omit()

#Download scope 3 emissions data.

scope_3 <- read_excel("~/Raw Data Finace Project 2/Scope 3 Emissions.xlsx", skip = 6)

#rearrange data so that year is in a column and scope 3 emissions are in a column.

scope_3 <- scope_3 %>% 
  pivot_longer("1995":"2018",
    names_to = "year",
    values_to = "scope_3_emissions"
  )

#Standardize country names and column names and add ISO index.

scope_3 <- scope_3 %>% 
  rename("country_name" = "Country")

scope_3 <- scope_3 %>% 
  mutate(iso3c = country_name_regex_to_iso3c(country_name)) |>
  mutate(country_name = iso3c_to_country_name(iso3c)) 

#Remove missing data.

scope_3 <- scope_3 %>% 
  na.omit()

#combine scope 1 and scope 2 emissions data.

scopes_1_and_2 <- 
  full_join(scope_1_and_export, scope_2, , by = c("year" = "year", "country_name" = "country_name", "iso3c" = "iso3c"))

#Combine scope 1, 2 and 3 emissions data into one data base.


all_scopes_all_data <-
    full_join(scopes_1_and_2, scope_3, , by = c("year" = "year", "country_name" = "country_name", "iso3c" = "iso3c"))


#Create column with total emissions where I add scope 1, 2, and 3, emissions.

all_scopes_all_data <- all_scopes_all_data %>% 
  mutate(consumption_emissions = domestic_emissions + scope_2_emissions + scope_3_emissions - export_emissions)

#Remove unnecessary columns.

all_scopes_all_data <- 
  select(all_scopes_all_data, -unit)

#Download GDP PPP data.

GDP_PPP <- read_excel("~/Raw Data Finace Project 2/GDP_by_PPP.xlsx", skip = 3)

#Rearrange data so that year is a column and GDP PPP is a column.

GDP_PPP <- GDP_PPP %>% 
  pivot_longer("1960":"2021",
    names_to = "year",
    values_to = "GDP_PPP_dollars"
  )

#Remove missing data.

GDP_PPP <- GDP_PPP %>% 
  na.omit()

#Standardize column names and remove unnecessary columns.

GDP_PPP <- GDP_PPP %>% 
  rename("country_name" = "Country Name")

GDP_PPP <- GDP_PPP %>% 
  rename("iso3c" = "Country Code")

GDP_PPP <- 
  select(GDP_PPP, -"Indicator Name")

GDP_PPP <- 
  select(GDP_PPP, -"Indicator Code")

#Add GDP PPP to data base with scope 1, 2, and 3 emissions.

all_data_scopes_GDP_PPP <-
   full_join(all_scopes_all_data, GDP_PPP, , by = c("year" = "year", "country_name" = "country_name", "iso3c" = "iso3c"))

#Download nominal GDP data.

nom_GDP <- read_excel("~/Raw Data Finace Project 2/Nominal GDP.xlsx", skip = 3)

#Rearrange it so that year is a column and nominal GDP is a column

nom_GDP <- nom_GDP %>% 
  pivot_longer("1960":"2021",
    names_to = "year",
    values_to = "nom_GDP_dollars"
  )

#Omit missing data

nom_GDP <- nom_GDP %>% 
  na.omit()

#Standardize column names and remove unnecessary columns.

nom_GDP <- nom_GDP %>% 
  rename("country_name" = "Country Name")

nom_GDP <- nom_GDP %>% 
  rename("iso3c" = "Country Code")

nom_GDP <- 
  select(nom_GDP, -"Indicator Name")

nom_GDP <- 
  select(nom_GDP, -"Indicator Code")

#Add nominal GDP tp data set with scopes 1-3 emissions and GDP PPP.

all_data_scopes_GDP_PPP_and_nom_na <-
   full_join(all_data_scopes_GDP_PPP, nom_GDP, , by = c("year" = "year", "country_name" = "country_name", "iso3c" = "iso3c"))

debt_pct_gdp <- read_excel("~/Raw Data Finace Project 2/Government Debt.xlsx")


debt_pct_gdp <- debt_pct_gdp %>%
  pivot_longer("1950":"2021",
    names_to = "year",
    values_to = "pct_debt_gdp"
  )

debt_pct_gdp <- debt_pct_gdp %>%
  na.omit()

debt_pct_gdp <- debt_pct_gdp %>%
  rename("country_name" = "Country")

debt_pct_gdp <- debt_pct_gdp %>%
  mutate(iso3c = country_name_regex_to_iso3c(country_name)) |>
  mutate(country_name = iso3c_to_country_name(iso3c)) 

all_data_debt_no_na <-
  full_join(all_data_scopes_GDP_PPP_and_nom_na, debt_pct_gdp, , by = c("year" = "year", "country_name" = "country_name", "iso3c" = "iso3c"))

all_data_debt_no_na <- all_data_debt_no_na %>%
  mutate(debt = (pct_debt_gdp/100) * nom_GDP_dollars)

iso3c_to_x <- purrr::partial(countrycode, origin = "iso3c")

all_data_debt_no_na <- all_data_debt_no_na %>%
  mutate(region = iso3c_to_x(iso3c, destination = "region"))

all_data_debt_no_na <- all_data_debt_no_na %>%
  mutate(scope_1_emissions_per_dollar_gdp_ppp = domestic_emissions/GDP_PPP_dollars)

all_data_debt_no_na <- all_data_debt_no_na %>%
  mutate(consumption_emissions_per_dollar_gdp_ppp = consumption_emissions /GDP_PPP_dollars)

#millions on tons of CO2/dollar

all_data_debt_no_na <- all_data_debt_no_na %>%
  drop_na(region)

all_data_debt_no_na <- all_data_debt_no_na %>%
  mutate(financed_emission = (debt/GDP_PPP_dollars) * consumption_emissions)


financed_emissions_no_na <- all_data_debt_no_na %>%
  drop_na(financed_emission)

Code

all_data_debt_no_na %>%
  drop_na(consumption_emissions, debt) %>%
  ggplot(aes(x = consumption_emissions, y = debt/1000000)) + 
    geom_point(na.rm = TRUE) +
    geom_smooth(na.rm = TRUE) +
    facet_wrap(~region, scales = "free_x", "free_y") +
    labs(title = "Figure 1: Sovereign Debt v. Consumption Emissions by Region 1995-2018",
       subtitle = str_wrap("Regions have very different relationships between debt and emissions, but there is a somewhat positive relationshsip between the two.", width = 70),
       x = "Conusmption Emissions (millions of tons of CO2)",
       y = "Sovereign Debt (USD)", 
       caption = "Sources: Climate Watch, OECD, IMF, World Bank | Latest Data: 2018 | Calculations by Ryan Showman"
       ) + 
      scale_y_continuous(labels = scales::label_dollar(suffix = "M"), expand = c(0,0)) +
      theme_minimal()

Code

all_data_debt_no_na %>%
  drop_na(consumption_emissions, debt) %>%
  ggplot(aes(x = consumption_emissions, y = debt/1000000, color = region)) +
    geom_point(na.rm = TRUE, alpha = 1/5) +
    labs(title = "Figure 2: Sovereign Debt v. Consumption Emissions by Region 1995-2018",
       subtitle = str_wrap("Some regions have somewhat constant debt no matter their emissions, while others have a more linear relationship between debt and emissions. ", width = 70),
       x = "Conusmption Emissions (millions of tons of CO2)",
       y = "Sovereign Debt (USD)", 
       caption = "Sources: Climate Watch, OECD, IMF, World Bank | Latest Data: 2018 | Calculations by Ryan Showman"
       ) +
  scale_y_continuous(labels = scales::label_dollar(suffix = "M"), expand = c(0,0)) +
      theme_minimal()

The United States, China, and Japan have the highest emissions financed from sovereign debt globally. Figures 3 and 4 show that the United States has, by far, the most financed emissions globally. Compared to the United States, China has a little over half as many emissions financed by sovereign debt and Japan has a third. The rest of the world’s emissions financed by sovereign debt is almost minimal compared to the United States. This is interesting because while the United States, China and Japan also have the world’s highest GDP, Germany, who has the world’s 4th largest GDP, has the 10th most emissions financed by sovereign debt. This reveals that emissions financed by sovereign debt is not directly related to GDP, even if it looks like it at first viewing. This is hopeful, because it means that with the right policies, countries can have high GDP and lower emissions financed from sovereign debt. This does make one wonder if the type of economy (service or manufacturing or commodity based) has an impact on emissions financed by sovereign debt.

Code

financed_emissions_2018 <- financed_emissions_no_na %>%
  filter(year == "2018")

# get the map data in sf format (google it)
world_map_data <- rnaturalearth::ne_countries(returnclass = "sf") 

world_map_data

world_map_financed_emissions <- world_map_data |> 
  left_join(financed_emissions_2018, by = c("iso_a3" = "iso3c")) 

Code

world_map_financed_emissions %>%
  filter(iso_a3 != "ATA") %>% 
  ggplot(aes(fill = financed_emission/1000)) +
  # this makes a map using the sf data.  So easy, right?
  geom_sf() +
    theme(legend.position="bottom") +
  labs(
    title = "Figure 3: Emissions Financed by Sovereign Debt 2018",
    subtitle = str_wrap("Sovereign Debt finances the most emissions in China and the United States", width = 70),
    fill = "Financed Emissions (billions of tons of CO2)",
    caption = "Sources: Climate Watch, OECD, IMF, World Bank | Latest Data: 2018 | Calculations by Ryan Showman"
  )

Code

financed_emissions_2018 <- financed_emissions_2018 %>%
  arrange(financed_emission)

financed_emissions_2018$positive_or_negative <- ifelse(financed_emissions_2018$financed_emission < 0, "negative", "positive")  

financed_emissions_2018 %>%
  filter(financed_emission > 200) %>%
  ggplot(aes(x = fct_reorder(.f = country_name, .x = financed_emission), y = financed_emission, fill = region)) +
  geom_bar(stat="identity", width= .9) +
  labs(title = "Figure 4: Emissions Financed by Sovereign Debt by Country 2018",
       subtitle = str_wrap("The United States, China, and Japan finance the most emissions from sovereign debt.", width = 70),
       x = "Country",
       y = "Emissions Financed from Sovereign Debt (millions of tons of CO2)", 
       fill = "Region",
       caption = "Sources: Climate Watch, OECD, IMF, World Bank | Latest Data: 2018 | Calculations by Ryan Showman"
       ) +
      theme_minimal() +
  coord_flip()

There is a positive relationship between sovereign debt, PPP adjusted GDP, and emissions financed by GDP. Figure 5 shows the relationship between a country’s average PPP adjusted GDP and their average financed emissions. There is a generally positive correlation although in countries with lower PPP adjusted GDP there seems to be a “U” shaped curve as GDP grows. This could be because countries with extremely low GDP could have more sovereign debt financed emissions due to sovereign debt first going to finance industry which is not economically efficient, while countries with slightly higher GDP use the debt to invest in industries where they have a comparative advantage, which may be more manufacturing based. Figure 6 shows a similar relationship between average sovereign debt and average sovereign debt financed emissions. It shows countries with lower average sovereign debt having a less consistent relationship with emissions financed by sovereign debt, while countries with greater overall average sovereign debt having more consistently higher average emissions financed by sovereign debt. This makes sense because if countries do not have much sovereign debt, then their emissions financed by sovereign debt will most likely depend more on a few projects rather than overall economic activity. Figure 7 ties everything together, demonstrating that sovereign debt, PPP adjusted GDP, and emissions financed by sovereign debt have always had a positive correlation over time. Overall, this means that investors need to be more aware that buying sovereign debt generally contributes to more emissions.

Code

financed_emissions_no_na %>%
  group_by(year) %>% 
  summarize(mean_financed_emission = mean(financed_emission), mean_gdp_ppp = mean(GDP_PPP_dollars), mean_debt = mean(debt)) %>%
   ggplot(aes(x = mean_gdp_ppp/1000000, y = mean_financed_emission)) +
  geom_line() +
  labs(title = "Figure 5: Average PPP Adjusted GDP v. Average Emissions Financed by Sovereign Debt",
       subtitle = str_wrap("There is a positive relationship between emissions financed by sovereign debt and PPP adjusted GDP. ", width = 70),
       x = "Average PPP Adjusted GDP (USD)",
       y = "Average Emissions Financed by Sovereign Debt", 
       caption = "Sources: Climate Watch, OECD, IMF, World Bank | Latest Data: 2018 | Calculations by Ryan Showman"
       ) +
  scale_x_continuous(labels = scales::label_dollar(suffix = "M"), expand = c(0,0)) +
      theme_minimal()

Code

financed_emissions_no_na %>%
  group_by(year) %>% 
  summarize(mean_financed_emission = mean(financed_emission), mean_gdp_ppp = mean(GDP_PPP_dollars), mean_debt = mean(debt)) %>% 
ggplot(aes(x = mean_debt/1000000, y = mean_financed_emission)) +
  geom_line() +
  labs(title = "Figure 6: Average Sovereign Debt v. Average Emissions Financed by Sovereign Debt",
       subtitle = str_wrap("There is a positive relationship between emissions financed by sovereign debt and sovereign debt.", width = 70),
       x = "Average Sovereign Debt (USD)",
       y = "Average Emissions Financed by Sovereign Debt", 
       caption = "Sources: Climate Watch, OECD, IMF, World Bank | Latest Data: 2018 | Calculations by Ryan Showman"
       ) +
  scale_x_continuous(labels = scales::label_dollar(suffix = "M"), expand = c(0,0)) +
      theme_minimal()

Code

theme_set(theme_bw())

graph_financed_emissions <- financed_emissions_no_na

  graph_financed_emissions$"year" <- as.numeric(as.character(graph_financed_emissions$"year"))

graphic <- ggplot(graph_financed_emissions, 
  aes(x = GDP_PPP_dollars/1000000, y=financed_emission, size = debt/1000000, colour = region)
  ) +
  geom_point(show.legend = TRUE, alpha = 0.7) +
  scale_color_viridis_d() +
  scale_size(range = c(2, 12)) +
  scale_x_continuous(labels = scales::label_dollar(suffix = "M"), expand = c(0,0))

graphic + transition_time(year) +
  labs(title = "Year: {frame_time}",
      subtitle = str_wrap("Figure 7: Emissions Financed by Sovereign Debt v. PPP Adjusted GDP v. Debt over Time ", width = 70), 
       x = "PPP Adjusted GDP (USD)",
      y = "Emissions Financed by Sovereign Debt (millions of tons of CO2)",
      size = "Sovereign Debt (millions of USD)",
      colour = "Region",
      caption = "Sources: Climate Watch, OECD, IMF, World Bank | Latest Data: 2018 | Calculations by Ryan Showman"
       ) 

Conclusion

Through the evidence highlighted in this essay, there seems to be a generally clear relationship between sovereign debt, GDP, and increased CO2 emissions. However, there are certainly a number of outliers that defy the trend. To deal with this, investors should be more aware of CO2 emissions financed by sovereign debt and try to buy sovereign debt from countries with lower CO2 emissions. To ensure this, countries with higher CO2 emissions financed by sovereign debt should be priced to be seen as more risky in the market.