GDP & emissions - degrowthing or decoupling?

Author

Ben Anderson & Tom Rushby

Published

November 11, 2022

Friday fagpackets

One of a series…

What’s all this about de-growth?

<sigh>

A follow-up to @tom_rushby’s tweet

Updated after another by @JKSteinberger

Data

Downloaded from the awesome OurWorldInData:

  • absolute and per capita GDP

  • absolute and per capita Greenhouse Gas emissions in two flavours:

    • production (‘territorial’) and

    • consumption

Code 
# fix
dt_abs <- data.table::fread(here::here("data", "co2-emissions-and-gdp.csv"))
dt_pc <- data.table::fread(here::here("data", "co2-emissions-and-gdp-per-capita.csv"))

Where do we care about?

Change these at will and for your purposes…

This would make a really nice shiny app

Code 
entities_of_interest <- c("United Kingdom", "United States", "New Zealand", "Australia",
                          "China", "India", "Norway", "Kenya", "Jamaica")

# check valid
t <- dt_pc[Entity %in% entities_of_interest, .(nObs = .N, 
                                           meanAnnualPerCapTerr = mean(`Annual CO2 emissions (per capita)`, na.rm = TRUE),
                                           meanAnnualPerCapCons = mean(`Annual consumption-based CO2 emissions (per capita)`,na.rm = TRUE),
                                           meanAnnualGDPperCap = mean(`GDP per capita, PPP (constant 2017 international $)`,na.rm = TRUE)), keyby = .(Entity)]

knitr::kable(t, digits = 2) %>%
  kable_styling()
Entity nObs meanAnnualPerCapTerr meanAnnualPerCapCons meanAnnualGDPperCap
Australia 31 17.71 16.11 40887.80
China 31 4.64 3.94 6980.80
India 31 1.19 1.09 3639.23
Jamaica 31 3.32 3.55 9564.31
Kenya 31 0.28 0.39 3594.88
New Zealand 31 8.09 8.69 35562.84
Norway 31 8.93 9.85 57633.00
United Kingdom 31 8.52 10.80 40041.50
United States 31 19.09 20.00 51695.66

Absolute GDP & emissions

Absolute production emissions

First we’ll try production (‘territorial’) emissions. We are not controlling for population size so countries with large populations will visually dominate.

Code 
plotDT <- dt_abs[Entity %in% entities_of_interest]

ggplot2::ggplot(plotDT, aes(x = as.numeric(`GDP, PPP (constant 2017 international $)`)/1000000000, 
                            y = as.numeric(`Annual CO2 emissions`)/1000000,
                            colour = Entity,
                            alpha = Year)) +
  geom_point() +
  labs(y = "Annual CO2 emissions (production-based, gT)",
       x = "GDP $bn (constant 2017 $)",
       caption = figCaption)

Figure 1: GDP vs production emissions over time (selected entities).

Absolute consumption emissions

Note that 2020 consumption-based emissions data is missing so you don’t see the 2020 COVID-19 downtick. Absolute emissions are falling in some places even as GDP is still increasing…

Code 
plotDT <- dt_abs[Entity %in% entities_of_interest]

ggplot2::ggplot(plotDT, aes(x = as.numeric(`GDP, PPP (constant 2017 international $)`)/1000000000, 
                            y = as.numeric(`Annual consumption-based CO2 emissions`)/1000000,
                            colour = Entity,
                            alpha = Year)) +
  geom_point() +
  labs(y = "Annual CO2 emissions (consumption-based, gT)",
       x = "GDP $bn (constant 2017 $)",
       caption = figCaption)

Figure 2: GDP vs consumotion emissions over time (selected entities).

Per capita GDP & emissions

Since we’ll be dividing everything pairwise by the same denominator, nothing much about the shapes should change… but the plots should be much clearer as we’ve removed the affect of population size. Countries with large populations will no longer visually dominate…

Production emissions

First we’ll try production emissions. Figure 3 shows change in this metric over time. Some interesting changes in direction (and otherwise)… basically they all need to be inverse U shaped, i.e. heading for y = 0.

Code 
entities_of_interest <- c(entities_of_interest, "World")

plotDT <- dt_pc[Entity %in% entities_of_interest]

p <-ggplot2::ggplot(plotDT, aes(x = as.numeric(`GDP per capita, PPP (constant 2017 international $)`), 
                            y = as.numeric(`Annual CO2 emissions (per capita)`),
                            colour = Entity,
                            alpha = Year)) +
  geom_point() +
  geom_line() +
  labs(y = "Annual CO2 emissions per capita (production-based, T)",
       x = "GDP per capita (constant 2017 $)",
       caption = figCaption)

p

Figure 3: GDP vs production emissions over time (selected entities)

Let’s repeat that. Hover over the dots to see which is which. This is plotly in action #YMMV.

Code 
p <-ggplot2::ggplot(plotDT, aes(x = as.numeric(`GDP per capita, PPP (constant 2017 international $)`), 
                            y = as.numeric(`Annual CO2 emissions (per capita)`),
                            colour = Entity)) +
  geom_point() +
  geom_line() +
  labs(y = "Annual CO2 emissions per capita (production-based, T)",
       x = "GDP per capita (constant 2017 $)",
       caption = figCaption)

plotly::ggplotly(p)

Figure 4: GDP vs production emissions over time (selected entities)

Now let’s have a look at all the countries but without the lines.

Quite interesting, Qatar for example. Perfect place for a winter sport’s World Cup. Trinidad and Tobago also worth a look. Macao looks like the place to be?

Code 
#plotDT <- dt_pc[Entity %in% entities_of_interest]
plotDT <- dt_pc # for fun

p <- ggplot2::ggplot(plotDT, aes(x = as.numeric(`GDP per capita, PPP (constant 2017 international $)`), 
                            y = as.numeric(`Annual CO2 emissions (per capita)`),
                            colour = Entity,
                            alpha = Year)) +
  geom_point() +
  theme(legend.position = "none") +
  labs(y = "Annual CO2 emissions per capita (production-based, T)",
       x = "GDP per capita (constant 2017 $)",
       caption = figCaption)

plotly::ggplotly(p)

Figure 5: GDP vs production emissions over time (all entities)

Consumption emissions

Next we’ll try consumption emissions. This is shown in Figure 6.

Note that 2020 consumption-based emissions data is missing so you don’t see the down-tick.

Code 
plotDT <- dt_pc[Entity %in% entities_of_interest]

ggplot2::ggplot(plotDT, aes(x = as.numeric(`GDP per capita, PPP (constant 2017 international $)`), 
                            y = as.numeric(`Annual consumption-based CO2 emissions (per capita)`),
                            colour = Entity,
                            alpha = Year)) +
  geom_point() +
  geom_line() +
  labs(y = "Annual CO2 emissions per capita (consumption-based, T)",
       x = "GDP per capita (constant 2017 $)",
       caption = figCaption)

Figure 6: GDP vs consumption emissions over time (selected entities).

Let’s repeat that for all countries. Hover over the dots to see which is which #YMMV.

Well this time Luxembourg is out in the lead…

Code 
#plotDT <- dt_pc[Entity %in% entities_of_interest]
plotDT <- dt_pc # for fun

p <- ggplot2::ggplot(plotDT, aes(x = as.numeric(`GDP per capita, PPP (constant 2017 international $)`), 
                            y = as.numeric(`Annual consumption-based CO2 emissions (per capita)`),
                            colour = Entity,
                            alpha = Year)) +
  geom_point() +
  theme(legend.position = "none") +
  labs(y = "Annual CO2 emissions per capita (consumption-based, T)",
       x = "GDP per capita (constant 2017 $)",
       caption = figCaption)

plotly::ggplotly(p)

Figure 7: GDP vs consumption emissions over time (all entities)

Efficiency

How about looking at efficiency: gC02e per $ GDP?

Following https://timjackson.org.uk/earth-vs-growth/ …

Code 
dt_abs[, `Emissions intensity (production-based)` := `Annual CO2 emissions`/`GDP, PPP (constant 2017 international $)`*1000*1000]
Code 
plotDT <- dt_abs[Entity %in% entities_of_interest]

ggplot2::ggplot(plotDT, aes(Year, `Emissions intensity (production-based)`, colour = Entity)) +
  geom_line() +
  labs(y = "gC02 per $ GDP",
       colour = "Country",
       caption = figCaption)

For the UK, let’s say c. 300g/$ in 1995 reducing to 200g/$ in 2010. So 15 years. To go from 200g/$ to zero at the same rate would take approx. 30 years. So net-zero by 2050?

Regardless of whether that aligns with science-based targets (i.e. absolute zero by 2044) … what happens if the economy grows … by say 2-3% per year? … would that force total emissions up (fixed intensity) or result in efficiency gains (lower intensity)?

Code 
dt_abs[, `Emissions intensity (consumption-based)` := `Annual consumption-based CO2 emissions`/`GDP, PPP (constant 2017 international $)`*1000*1000]
Code 
plotDT <- dt_abs[Entity %in% entities_of_interest]

ggplot2::ggplot(plotDT, aes(Year, `Emissions intensity (consumption-based)`, colour = Entity)) +
  geom_line() +
  labs(y = "gC02 per $ GDP",
       caption = figCaption)
Warning: Removed 22 rows containing missing values (`geom_line()`).

Whom do we love?

quarto

data.table

ggplot2

here

skimr

Data descriptions

Check

Skim the absolute data:

Code 
skimr::skim(dt_abs)
Warning: Couldn't find skimmers for class: integer64; No user-defined `sfl`
provided. Falling back to `character`.

Warning: Couldn't find skimmers for class: integer64; No user-defined `sfl`
provided. Falling back to `character`.
Data summary
Name dt_abs
Number of rows 7795
Number of columns 8
Key NULL
_______________________
Column type frequency:
character 4
numeric 4
________________________
Group variables None

Variable type: character

skim_variable n_missing complete_rate min max empty n_unique whitespace
Entity 0 1.00 4 32 0 255 0
Code 0 1.00 0 8 931 225 0
Annual CO2 emissions 479 0.94 1 21 0 6137 0
GDP, PPP (constant 2017 international $) 1626 0.79 1 21 0 6162 0

Variable type: numeric

skim_variable n_missing complete_rate mean sd p0 p25 p50 p75 p100 hist
Year 0 1.00 2005.06 8.930000e+00 1990.00 1997.00 2005.00 2013.00 2.02000e+03 ▇▇▇▇▇
Annual consumption-based CO2 emissions 3730 0.52 976326173.39 3.366486e+09 197201.67 10759532.52 59975058.69 290854945.27 3.67025e+10 ▇▁▁▁▁
Emissions intensity (production-based) 2105 0.73 263.56 2.409400e+02 9.15 122.67 196.73 309.26 2.30112e+03 ▇▁▁▁▁
Emissions intensity (consumption-based) 4295 0.45 291.92 1.686800e+02 4.19 179.06 251.82 365.36 1.82195e+03 ▇▂▁▁▁

Skim the per capita data

Code 
skimr::skim(dt_pc)
Data summary
Name dt_pc
Number of rows 7745
Number of columns 6
Key NULL
_______________________
Column type frequency:
character 2
numeric 4
________________________
Group variables None

Variable type: character

skim_variable n_missing complete_rate min max empty n_unique whitespace
Entity 0 1 4 32 0 252 0
Code 0 1 0 8 899 224 0

Variable type: numeric

skim_variable n_missing complete_rate mean sd p0 p25 p50 p75 p100 hist
Year 0 1.00 2005.07 8.93 1990.00 1997.00 2005.00 2013.00 2020.00 ▇▇▇▇▇
GDP per capita, PPP (constant 2017 international $) 1579 0.80 17879.29 20132.68 436.72 3678.83 10365.78 25832.03 161971.47 ▇▂▁▁▁
Annual consumption-based CO2 emissions (per capita) 3680 0.52 6.53 6.92 0.05 1.19 4.30 9.89 57.79 ▇▂▁▁▁
Annual CO2 emissions (per capita) 492 0.94 5.11 6.44 0.02 0.77 2.97 7.42 68.72 ▇▁▁▁▁
Code 
# hidden chunk