--- title: "Inequality and Climate Change Impacts" subtitle: "Does Economic Inequality Increase Country Emissions?" author: "Dr. Kuehl" institute: "School of Public Policy - Chaing Mai University" date: today format: beamer: theme: simple slide-number: true progress: true transition: fade footer: "Kuehl | Mock Talk" execute: echo: false warning: false message: false editor_options: chunk_output_type: console --- ```{r} #| label: setup library(tidyverse) library(ggplot2) library(knitr) library(kableExtra)#for tables library(modelsummary) library(wbstats) setwd("~/Dropbox/POLS 641/CMU Su26/finalproject/Mock") ## Load dataset built in other file gwd <- read.csv("gwd.csv") ``` ## Research Puzzle ::::: columns ::: {.column width="55%"} **Motivating Question** - Climate Change is an existential global challenge - Countries and people need to reduce their emissions *yesterday* - Economic inequality leads to elites having more political power - National climate policies have a large impact on emissions ::: ::: {.column width="45%"} ![](images/taylor.webp) ::: ::::: ------------------------------------------------------------------------ ## Research Question ::::: columns ::: {.column width="55%"} **Motivating Question** - Emissions are unequal within countries - Oxfam: The world's richest 1% are responsible for more carbon emissions than the poorest 66% - **Do more unequal countries do less to mitigate climate change?** ::: ::: {.column width="45%"} ![](images/oxfam.png) ::: ::::: ## Background and Literature **What we know** - There is wide varition in global emissions - Much is driven by wealth - Wealth -\> Consumption -\> Emissions - Generally, people are supportive of climate policies, but pushback from rich elites can hinder climate action ::: notes Speaker notes: These don't show up on the screen, but can be useful when building ::: ------------------------------------------------------------------------ ## Hypotheses I derive the following expectations: ::: incremental - **H1 (Main):** Ceterus paribus, as *Income inequality* increases, *Per Capita carbon emissions* will also increase - **Null Hypothesis** There is no relationship between income inequality and carbon emissions. ::: ## Income Inequality - Various Measures ```{r} #| label: fig-distribution_all #| fig-height: 4.5 vars_to_plot <- c("gini_posttax", "gini_pretax", "top10_share", "top1_share", "wealth_top10_share", "wealth_top1_share") gwd |> select(iso3c, all_of(vars_to_plot)) |> pivot_longer(-iso3c, names_to = "measure", values_to = "value") |> mutate(measure = case_match(measure, "gini_pretax" ~ "Gini (pre-tax income)", "gini_posttax" ~ "Gini (post-tax income)", "top10_share" ~ "Top 10% income share", "top1_share" ~ "Top 1% income share", "wealth_top10_share" ~ "Top 10% wealth share", "wealth_top1_share" ~ "Top 1% wealth share"), measure = factor(measure, levels = c( "Gini (pre-tax income)", "Gini (post-tax income)", "Top 10% income share", "Top 1% income share", "Top 10% wealth share", "Top 1% wealth share"))) |> ggplot(aes(x = value, fill = measure)) + geom_density(alpha = 0.55, color = NA) + scale_fill_brewer(palette = "Set2") + labs(title = "Distribution of inequality measures across countries, 2022", subtitle = "All values on a 0–100 scale", x = "Value", y = "Density", fill = NULL, caption = "Source: World Inequality Database") + theme_minimal(base_size = 12) + theme(plot.title = element_text(face = "bold"), plot.subtitle = element_text(color = "grey40"), legend.position = "bottom", panel.grid.minor = element_blank()) ``` ## Primary Independent Variable ::::: columns ::: {.column width="55%"} - Primary IV: Income Inequality - The distribution of wealth within a country - Operationalized using the share of income to the top 10% within each country - Date from the World Income Database ::: ::: {.column width="45%"} ```{r} #| label: fig-distribution #| fig-height: 6 # Pull the two countries' values for the annotation ggplot(gwd, aes(x = wealth_top10_share)) + geom_histogram(binwidth = 2, fill = "#2c7bb6", color = "white", na.rm = TRUE) + geom_vline(aes(xintercept = mean(wealth_top10_share, na.rm = TRUE)), color = "red", linetype = "dashed", linewidth = 0.8) + geom_vline(aes(xintercept = median(wealth_top10_share, na.rm = TRUE)), color = "orange", linetype = "dashed", linewidth = 0.8) + geom_text(data = subset(gwd, country.x %in% c("United States", "Thailand")), aes(x = wealth_top10_share, y = 0, label = country.x), vjust = -1, angle = 90, size = 3, color = "black") + geom_vline(data = subset(gwd, country.x %in% c("United States", "Thailand")), aes(xintercept = wealth_top10_share), color = "black", linetype = "dotted") + annotate("text", x = mean(gwd$wealth_top10_share, na.rm = TRUE), y = Inf, label = "Mean", color = "red", vjust = 2, hjust = -0.1, size = 3) + annotate("text", x = median(gwd$wealth_top10_share, na.rm = TRUE), y = Inf, label = "Median", color = "orange", vjust = 4, hjust = -0.1, size = 3) + labs( title = "Distribution of Top 10% Wealth Share", x = "Top 10% Wealth Share (%)", y = "Count" ) + theme_minimal() ``` ::: ::::: ## Dependent Variable ::::: columns ::: {.column width="55%"} Carbon Emissions - The per capita contribution to climate change - Operationalized as CO2 Emissions in metric tons per capita - Data from the world bank *All data are from 2023 or most recent* ::: ::: {.column width="45%"} ```{r} #| fig-height: 6 ggplot(gwd, aes(x = co2_per_cap)) + geom_histogram(binwidth = 1, fill = "#2c7bb6", color = "white", na.rm = TRUE) + geom_vline(aes(xintercept = mean(co2_per_cap, na.rm = TRUE)), color = "red", linetype = "dashed", linewidth = 0.8) + geom_vline(aes(xintercept = median(co2_per_cap, na.rm = TRUE)), color = "orange", linetype = "dashed", linewidth = 0.8) + geom_vline(data = subset(gwd, country.x %in% c("United States", "Thailand")), aes(xintercept = co2_per_cap), color = "black", linetype = "dotted") + geom_text(data = subset(gwd, country.x %in% c("United States", "Thailand")), aes(x = co2_per_cap, y = 0, label = country.x), vjust = -1, angle = 90, size = 3, color = "black") + annotate("text", x = mean(gwd$co2_per_cap, na.rm = TRUE), y = Inf, label = "Mean", color = "red", vjust = 2, hjust = -0.1, size = 3) + annotate("text", x = median(gwd$co2_per_cap, na.rm = TRUE), y = Inf, label = "Median", color = "orange", vjust = 4, hjust = -0.1, size = 3) + labs( title = "Distribution of CO2 Per Capita", x = "CO2 Per Capita (tonnes)", y = "Count" ) + theme_minimal() gwd <- gwd[!gwd$co2_per_cap %in% tail(sort(gwd$co2_per_cap, na.last = NA), 2), ] ``` ::: ::::: ## Descriptive Statistics and Controls ```{r} #| label: tbl-descriptives #| tbl-cap: "Summary Statistics" gwd$gdp_pc <- gwd$gdp_pc/1000 #convert gdp per capita var_labels <- c( "income_share_top10" = "Income Share: Top 10%", "co2_per_cap" = "CO2 per Capita (tonnes)", "gdp_pc" = "GDP per Capita (thousands USD)", "ag_land" = "Agricultural Land (%)", "gdp_growth" = "GDP Growth (%)", "electricity_access" = "Electricity Access (%)" ) gwd %>% select(income_share_top10, co2_per_cap, gdp_pc, ag_land, gdp_growth, electricity_access) %>% summarise(across(everything(), list( Mean = ~mean(.x, na.rm = TRUE), Median = ~median(.x, na.rm = TRUE), SD = ~sd(.x, na.rm = TRUE), Min = ~min(.x, na.rm = TRUE), Max = ~max(.x, na.rm = TRUE), N = ~sum(!is.na(.x)) ))) %>% pivot_longer(everything(), names_to = c("Variable", "Stat"), names_sep = "_(?=[^_]+$)") %>% pivot_wider(names_from = Stat, values_from = value) %>% mutate( Variable = var_labels[Variable], across(c(Mean, Median, SD, Min, Max), ~round(.x, 2)), N = as.integer(N) ) %>% kbl( col.names = c("Variable", "Mean", "Median", "SD", "Min", "Max", "N"), align = c("l", rep("c", 6)), caption = "Table 1: Descriptive Statistics" ) %>% kable_styling( bootstrap_options = c("striped", "hover"), full_width = FALSE, font_size = 25 ) %>% row_spec(0, bold = TRUE) %>% column_spec(1, bold = TRUE, width = "12em") ``` ------------------------------------------------------------------------ ## Correlation Matrix ```{r} #| label: corr #| tbl-cap: "Correlation Matrix" library(ggcorrplot) cor_data <- gwd %>% select("income_share_top10", "co2_per_cap", "gdp_pc", "ag_land", "gdp_growth", "electricity_access") %>% cor(use = "pairwise.complete.obs") ggcorrplot(cor_data, method = "square", type = "lower", lab = TRUE, lab_size = 4, colors = c("#d73027", "white", "#2c7bb6"), title = "Correlation Matrix", ggtheme = theme_minimal()) ``` ## Initial Scatter Plot ```{r} ggplot(gwd, aes(x = wealth_top10_share, y = co2_per_cap)) + geom_point(alpha = 0.6, color = "#2c7bb6", na.rm = TRUE) + geom_smooth(method = "lm", se = TRUE, color = "red", na.rm = TRUE) + geom_point(data = subset(gwd, country.x %in% c("United States", "Thailand")), aes(x = wealth_top10_share, y = co2_per_cap), color = "black", size = 3) + geom_text(data = subset(gwd, country.x %in% c("United States", "Thailand")), aes(x = wealth_top10_share, y = co2_per_cap, label = country.x), vjust = -1, size = 3, color = "black") + labs( title = "CO2 Per Capita vs. Top 10% Wealth Share", x = "Top 10% Wealth Share (%)", y = "CO2 Per Capita (tonnes)" ) + theme_minimal() ``` ## Primary Regression Results ```{r} #| label: tbl-main-results #| tbl-cap: "OLS Regression — Main Results" m1 <- lm(co2_per_cap ~ income_share_top10, data = gwd) m2 <- lm(co2_per_cap ~ income_share_top10 + gdp_pc, data = gwd) m3 <- lm(co2_per_cap ~ income_share_top10 + gdp_pc + ag_land + gdp_growth, data = gwd) m4 <- lm(co2_per_cap ~ income_share_top10 + gdp_pc + ag_land + gdp_growth + electricity_access, data = gwd) modelsummary( list("Bivariate" = m1, "+ GDP Per Capta" = m2, "Full Model" = m3, "Add Controls"=m4), stars = c("*" = .1, "**" = .05, "***" = .01), gof_map = c("nobs", "r.squared", "adj.r.squared"), coef_rename = c( "democracy" = "Democracy", "log(gdp_pc)" = "ln GDP per Capita", "trade_open" = "Trade Openness" ), output = "kableExtra", fmt = 3 ) |> kable_styling(font_size = 18, full_width = FALSE) |> row_spec(0, bold = TRUE) ``` ------------------------------------------------------------------------ ## Logged Regression Results Logging GDP per Capita and CO2 per Capita ```{r} gwd$log_co2_per_cap <- log(gwd$co2_per_cap) gwd$log_gdp_pc <- log(gwd$gdp_pc) m1_log <- lm(log_co2_per_cap ~ income_share_top10, data = gwd) m2_log <- lm(log_co2_per_cap ~ income_share_top10 + log_gdp_pc, data = gwd) m3_log <- lm(log_co2_per_cap ~ income_share_top10 + log_gdp_pc + ag_land + gdp_growth, data = gwd) m4_log <- lm(log_co2_per_cap ~ income_share_top10 + log_gdp_pc + ag_land + gdp_growth + electricity_access, data = gwd) modelsummary( list("Bivariate" = m1_log, "+ GDP Per Capita" = m2_log, "+ Land & Growth" = m3_log, "+ Electricity" = m4_log), stars = c("*" = .1, "**" = .05, "***" = .01), gof_map = c("nobs", "r.squared", "adj.r.squared"), coef_rename = c( "income_share_top10" = "Income Share Top 10%", "log_gdp_pc" = "ln GDP per Capita", "ag_land" = "Agricultural Land", "gdp_growth" = "GDP Growth", "electricity_access" = "Electricity Access" ), output = "kableExtra", fmt = 3 ) |> kable_styling(font_size = 18, full_width = FALSE) |> row_spec(0, bold = TRUE) ``` ## Plotting with CO2 Logged ```{r} ggplot(gwd, aes(x = wealth_top10_share, y = log_co2_per_cap)) + geom_point(alpha = 0.6, color = "#2c7bb6", na.rm = TRUE) + geom_smooth(method = "lm", se = TRUE, color = "red", na.rm = TRUE) + geom_point(data = subset(gwd, country.x %in% c("United States", "Thailand")), aes(x = wealth_top10_share, y = log_co2_per_cap), color = "black", size = 3) + geom_text(data = subset(gwd, country.x %in% c("United States", "Thailand")), aes(x = wealth_top10_share, y = log_co2_per_cap, label = country.x), vjust = -1, size = 3, color = "black") + labs( title = "CO2 Per Capita vs. Top 10% Wealth Share", x = "Top 10% Wealth Share (%)", y = "Logged CO2 Per Capita (tonnes)" ) + theme_minimal() ``` ## Standardized Regression ```{r} #| label: fig-coefplot #| fig-cap: "Coefficient Plot " #| fig-height: 4 # Z-score independent variables gwd$z_income_share_top10 <- scale(gwd$income_share_top10) gwd$z_log_gdp_pc <- scale(gwd$log_gdp_pc) gwd$z_ag_land <- scale(gwd$ag_land) gwd$z_gdp_growth <- scale(gwd$gdp_growth) gwd$z_electricity_access <- scale(gwd$electricity_access) # Rerun models with standardized IVs m1_z <- lm(log_co2_per_cap ~ z_income_share_top10, data = gwd) m2_z <- lm(log_co2_per_cap ~ z_income_share_top10 + z_log_gdp_pc, data = gwd) m3_z <- lm(log_co2_per_cap ~ z_income_share_top10 + z_log_gdp_pc + z_ag_land + z_gdp_growth, data = gwd) m4_z <- lm(log_co2_per_cap ~ z_income_share_top10 + z_log_gdp_pc + z_ag_land + z_gdp_growth + z_electricity_access, data = gwd) # Coefficient plot modelplot( list("Bivariate" = m1_z, "+ GDP Per Capita" = m2_z, "+ Land & Growth" = m3_z, "+ Electricity" = m4_z), coef_omit = "Intercept", coef_rename = c( "z_income_share_top10" = "Income Share Top 10%", "z_log_gdp_pc" = "ln GDP per Capita", "z_ag_land" = "Agricultural Land", "z_gdp_growth" = "GDP Growth", "z_electricity_access" = "Electricity Access" ) ) + geom_vline(xintercept = 0, linetype = "dashed", color = "red") + labs( title = "Coefficient Plot — DV: ln CO2 per Capita (Standardized IVs)", x = "Coefficient (SD units)", color = "Model" ) + theme_minimal() ``` ## Alternative Measures of Wealth Inequality Includes all control variables ```{r} gwd$log_co2_per_cap <- ifelse(gwd$co2_per_cap > 0, log(gwd$co2_per_cap), NA) m4_log <- lm(log_co2_per_cap ~ income_share_top10 + log_gdp_pc + ag_land + gdp_growth + electricity_access, data = gwd) m4_gini_post <- lm(log_co2_per_cap ~ gini_posttax + log_gdp_pc + ag_land + gdp_growth + electricity_access, data = gwd) m4_gini_pre <- lm(log_co2_per_cap ~ gini_pretax + log_gdp_pc + ag_land + gdp_growth + electricity_access, data = gwd) m4_top1 <- lm(log_co2_per_cap ~ top1_share + log_gdp_pc + ag_land + gdp_growth + electricity_access, data = gwd) m4_wealth10 <- lm(log_co2_per_cap ~ wealth_top10_share + log_gdp_pc + ag_land + gdp_growth + electricity_access, data = gwd) m4_wealth1 <- lm(log_co2_per_cap ~ wealth_top1_share + log_gdp_pc + ag_land + gdp_growth + electricity_access, data = gwd) modelsummary( list( "Income Share Top 10" = m4_log, "Gini (post-tax)" = m4_gini_post, "Gini (pre-tax)" = m4_gini_pre, "Income Share Top 1" = m4_top1, "Wealth Top 10" = m4_wealth10, "Wealth Top 1" = m4_wealth1 ), stars = c("*" = .1, "**" = .05, "***" = .01), gof_map = c("nobs", "r.squared", "adj.r.squared"), coef_omit = "log_gdp_pc|ag_land|gdp_growth|electricity_access|Intercept", coef_rename = c( "income_share_top10" = "Income Share Top 10%", "gini_posttax" = "Gini (post-tax)", "gini_pretax" = "Gini (pre-tax)", "top1_share" = "Income Share Top 1%", "wealth_top10_share" = "Wealth Share Top 10%", "wealth_top1_share" = "Wealth Share Top 1%" ), output = "kableExtra", fmt = 3 ) |> kable_styling(font_size = 18, full_width = FALSE) |> row_spec(0, bold = TRUE) ``` ------------------------------------------------------------------------ ## Contributions & Further Research **There does not appear to be a strong relationship between wealth inquality and CO2 Emissions** - Potential Omitted Variable Bias - Political Insitutions - Democracies and Interest group systems - ie how elites preferences are heard - Oil Dependent Economies - Regional Differences - Could look at climate policy as DV, rather than emissions ## Contributions & Further Research **Qualitative Follow-Up** To probe the mechanism, future work should conduct process tracing to better understand if or how economic elites shape climate policy. - United States as a most crucial case - Thailand as "typical" case Case studies of wealthy people/companies in each country ## Discussion **Policy Implications** These results suggest wealth inequality is not a meaningful impediment to effective climate policy **Limitations** - One year snapshot - Country Level of analysis - Aggrgate Numbers ## {.center} ### Thank You **I look forward to your questions** ::: {.callout-tip appearance="minimal"} *Slides code available on course website* :::