Dumbbell Chart of Pay Gaps in Europe

paygap_eustat_eu_dumbbell.Rmd

---
title: "Gender Pay Gap: Dumbbell Plot (EU only)" 
author:
    - name: "Cédric Scherer"
      url: www.cedricscherer.com
output: 
    distill::distill_article:
        highlight: kate
        code_folding: false  
        toc: true            
        toc_depth: 2         
---


```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, message = FALSE, warning = FALSE, 
                      dev = "ragg_png", dpi = 350, retina = 1)
```

# Setup

```{r packages}
library(tidyverse)
library(here)
library(janitor)
library(ggrepel)
library(ggtext)
library(colorspace)
library(pdftools)
```


# Data

```{r data}
paygap_raw <- read_csv(here::here("sdg_05_20_linear.csv"))

paygap <-
  paygap_raw %>% 
  janitor::clean_names() %>% 
  dplyr::select(country = geo, year = time_period, gap = obs_value) %>%
  filter(!country %in% c("CH", "NO", "IS")) %>%  
  mutate(country = if_else(str_detect(country, "EU27_"), "EU27", country))

paygap_comparison <-
  paygap %>% 
  filter(year == 2020 | year <= 2010) %>%
  mutate(
    period = if_else(year == 2020, "today", "reference"),
    gap = -gap
  ) %>% 
  group_by(country, period) %>% 
  summarize(
    gap = mean(gap, na.rm = TRUE),
    n = n()
  ) %>% 
  ungroup() %>% 
  pivot_wider(
    id_cols = c(country),
    names_from = period,
    values_from = c(gap, n)
  ) %>% 
  filter(!is.na(gap_today)) %>% 
  dplyr::select(-n_today) %>% 
  mutate(
    country = if_else(str_detect(country, "EU|EA"), paste0("<b style='font-family:Compagnon-Medium;'>", country, "</b>"), country),
    country_today = fct_reorder(country, gap_today),
    country_ref = fct_reorder(country, gap_reference),
    change = gap_today - gap_reference,
    trend = case_when(
      change < -.5 ~ "Larger pay gap",
      change > .5 ~ "Smaller pay gap",
      TRUE ~ "Stable pay gap"
    ),
    trend = factor(trend, levels = c("Larger pay gap", "Stable pay gap", "Smaller pay gap")),
    v1 = if_else(change > 0, 1.45, -.45),
    v2 = if_else(change > 0, -.25, 1.25),
    percent_lab = paste0(scales::percent(abs(gap_today), scale = 1, accuracy = .1, trim = FALSE),
                         "<br><span style='font-size:8.2pt;font-family:Compagnon-Roman;'>",
                         if_else(change > 0, paste0("–", sprintf("%2.1f", abs(change)), "%"),
                                             paste0("+", sprintf("%2.1f", abs(change)), "%")),
                         "</span>")
  )
```


```{r dumbbell-graph, fig.height=12, fig.width=16}
pos <- as.numeric(paygap_comparison$country_today[which(str_detect(paygap_comparison$country_today, "EU|EA"))])

ggplot(paygap_comparison, 
       aes(x = as.numeric(country_today), y = gap_reference, color = trend)) +
  
  ## reference groups ----------------------------------------------------------
  annotate(
    geom = "rect",
    xmin = pos - .55, xmax = pos + .55,
    ymin = -30, ymax = -1.5,
    fill = "grey30", alpha = c(.05, .075)
  ) +
  annotate(
    geom = "text",
    x = c(pos[1] - .75, pos[2] + .75), y = -28.2,
    label = c("The 19 European Union countries that\nuse the euro as the official currency",
              "The 27 European Union countries\n(excluding the UK that left the EU)"),
    hjust = c(1, 0), color = "grey50", family = "Compagnon-Roman", 
    size = 3.5, lineheight = .85
  ) +
  
  ## y grid --------------------------------------------------------------------
  geom_hline(
    yintercept = -1:-5*5,
    color = "grey80", linetype = "42", size = .3
  ) +
  geom_hline(
    yintercept = c(0, -30),
    color = "grey80", size = .6
  ) +
  annotate(
    geom = "rect",
    xmin = 26.5, xmax = Inf,
    ymin = .5, ymax = -.5,
    fill = "white"
  ) +
  
  ## y-axis labels -------------------------------------------------------------
  annotate(
    geom = "text",
    x = -1, y = -1:-5*5,
    label = scales::percent(1:5*5, scale = 1, accuracy = 1, trim = FALSE),
    hjust = 0, vjust = 1.7, color = "grey30", family = "Compagnon-Light", size = 6.7
  ) +
  annotate(
    geom = "text",
    x = -1, y = c(1.7, 0, -30),
    label = c("Women's gross hourly earnings were...", 
              "equal to those of male paid employees in the EU", 
              "30% below those of male paid employees in the EU"),
    hjust = 0, vjust = 1.7, color = "#212121", family = "Compagnon-Roman", size = 6.7
  ) +
  
  ## legend --------------------------------------------------------------------
  annotate(
    geom = "segment",
    x = 6.5, xend = 6.5, y = -8.5, yend = -6, size = .8,
    arrow = arrow(type = "closed", length = unit(2.3, "mm")),
    color = "grey60"
  ) +
  annotate(
    geom = "point",
    x = 6.5, y = -8.5, size = 2.7,
    color = "grey60"
  ) +
  annotate(
    geom = "text",
    x = 6.75, y = c(-6, -8.1, -9),
    label = c("Pay gap 2020", 
              "Pay gap 2002–10", 
              "Size encodes number\nof reference years"),
    hjust = 0, color = "grey30", family = "Compagnon-Roman", 
    size = c(4, 4, 3.3), lineheight = .7
  ) +
  
  ## dumbbell labels -----------------------------------------------------------
  ggtext::geom_richtext(
    aes(label = country, vjust = v1,
        color = trend, color = after_scale(darken(color, .2))), 
    family = "Compagnon-Roman", size = 5.3, 
    fill = NA, label.color = NA, label.padding = unit(rep(.1, 4), "lines"), show.legend = FALSE
  ) +
  ggtext::geom_richtext(
    aes(label = percent_lab, y = gap_today, vjust = v2,
        color = trend, color = after_scale(darken(color, .2))), 
    family = "Compagnon-Medium", size = 3.3, lineheight = .85,
    fill = NA, label.color = NA, label.padding = unit(rep(.1, 4), "lines"), show.legend = FALSE
  ) +
  
  ## dumbbell ------------------------------------------------------------------
  geom_point(
    aes(alpha = trend == "Stable pay gap", size = n_reference), 
    #size = 2, 
    show.legend = FALSE
  ) +
  geom_segment(
    aes(xend = as.numeric(country_today), yend = gap_today), size = 1.1,
    arrow = arrow(type = "closed", length = unit(2.3, "mm"))
  ) +
  
  ## coord + scales ------------------------------------------------------------
  coord_cartesian(expand = FALSE, clip = "off") +
  scale_x_continuous(limits = c(-1, 27.5), guide = "none") +
  scale_y_continuous(breaks = 0:-6*5, limits = c(-31, 1.7)) +
  scale_color_manual(
    values = c("#B55951", "#F0B48E", "#669d8f"), 
    name = NULL
  ) +
  scale_alpha_manual(values = c(1, 0)) +
  scale_size_area(max_size = 3.5) +
  
  ## styling -------------------------------------------------------------------
  theme_minimal(base_size = 16, base_family = "Compagnon-Roman") +
  theme(
    panel.grid = element_blank(),
    axis.ticks.length.x = unit(.4, "lines"),
    axis.ticks.x.top = element_line(color = "grey80", size = .6),
    axis.text.x.top = element_text(color = "grey30", hjust = c(.5, .7, .3, .5, .5)),
    axis.text.y = element_blank(),
    plot.caption = element_textbox_simple(family = "Compagnon-Light", size = 13.3,
                                          color = "grey30", lineheight = 1.5,
                                           margin = margin(t = 50, b = 0), hjust = 0),
    legend.position = c(.16, .73),
    legend.text = element_text(color = "grey30", size = 11.5),
    legend.key.width = unit(2.2, "lines"),
    plot.margin = margin(40, 40, 30, 40)
  ) +
  labs(x = NULL, y = NULL,
       caption = "<b style='color:#212121;font-family:Compagnon-Roman;font-size:19.2pt;'>In 2020, women's gross hourly earnings were on average 13% below those of men in the EU.</b><br>Many countries were able to <b style='color:#577A71;font-family:Compagnon-Roman;'>decrease the pay gap</b> between male and female employees: most notably <b style='color:#577A71;font-family:Compagnon-Roman;'>Luxembourg</b> with the smallest gap in 2020 of 0.7% and <b style='color:#577A71;font-family:Compagnon-Roman;'>Romania</b> with a decrease of 13.6% within 18 years to 2.4%, ranking second. However, in several countries the pay gap was almost <b style='color:#C38B64;font-family:Compagnon-Roman;'>stable (Hungary and France)</b> or even <b style='color:#854D49;font-family:Compagnon-Roman;'>increased (Latvia, Portugal, Croatia, and Malta)</b> compared to the gap in the reference period from 2002 to 2010 for all years available via Eurostat.<br><br>As an unadjusted indicator, the gender pay gap gives an overall picture of the differences between men and women in terms of earnings and measures — a concept which is broader than discrimination in the sense of: <i style='color:#577A71;font-family:Compagnon-Bold;font-size:15.7pt;'>equal pay for work of equal value.</i><br><br>Graphic: Cédric Scherer • Data: Eurostat (SDG_05_20; no data for Greece and Ireland) • #InternationalWomensDay2022")

ggsave("paygap_eu_dumbbell.pdf", width = 16, height = 12, device = cairo_pdf)
pdf_convert("paygap_eu_dumbbell.pdf", filenames = "./plots/paygap_eu_dumbbell.png", dpi = 700)
```

***

<details><summary>Session Info</summary>

```{r sessionInfo}
Sys.time()
sessionInfo()
```

</details>

paygap_eustat_eu_slope.Rmd

---
title: "Gender Pay Gap: Slope Plot (EU only)" 
author:
    - name: "Cédric Scherer"
      url: www.cedricscherer.com
output: 
    distill::distill_article:
        highlight: kate
        code_folding: false  
        toc: true            
        toc_depth: 2         
---


```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, message = FALSE, warning = FALSE, 
                      dev = "ragg_png", dpi = 350, retina = 1)
```

# Setup

```{r packages}
library(tidyverse)
library(here)
library(janitor)
library(ggrepel)
library(ggtext)
library(colorspace)
library(pdftools)
```


# Data

```{r data}
paygap_raw <- read_csv(here::here("sdg_05_20_linear.csv"))

paygap <-
  paygap_raw %>% 
  janitor::clean_names() %>% 
  dplyr::select(country = geo, year = time_period, gap = obs_value) %>%
  filter(!country %in% c("EU28", "EA19", "CH", "NO", "IS")) %>%  
  mutate(country = if_else(str_detect(country, "EU27_"), "EU27", country))

paygap_comparison <-
  paygap %>% 
  group_by(country) %>% 
  filter(year == 2020 | year == min(year)) %>%
  pivot_wider(
    id_cols = c(country),
    names_from = year,
    values_from = gap,
    names_prefix = "gap_"
  ) %>% 
  pivot_longer(
    cols = -c(country, gap_2020),
    names_to = "year_ref",
    values_to = "gap_ref",
    names_prefix = "gap_"
  ) %>% 
  ungroup() %>% 
  filter(!is.na(gap_ref), !is.na(gap_2020)) %>% 
  mutate(
    year_ref = as.numeric(year_ref),
    change = gap_2020 - gap_ref,
    trend = case_when(
      country == "EU27" ~ "European Union",
      change > .5 ~ "Larger pay gap",
      change < -.5 ~ "Smaller pay gap",
      TRUE ~ "Stable pay gap"
    ),
    trend = factor(trend, levels = c("Larger pay gap", "Stable pay gap", "Smaller pay gap", "European Union")),
    country = if_else(country == "EU27", "EU", country), 
    country_lab = paste(country, scales::percent(gap_2020, scale = 1, accuracy = .1, trim = FALSE)) 
  ) %>% 
  arrange(desc(trend))
```


```{r slope-graph, fig.height=14, fig.width=10}
ggplot(paygap_comparison, aes(x = year_ref, y = -gap_ref, color = trend)) +
  
  ## reference period ----------------------------------------------------------
  annotate(
    geom = "rect",
    xmin = 2001.7, xmax = 2010.3,
    ymin = -30, ymax = -1.2,
    fill = "grey30", alpha = .05
  ) +
  annotate(
    geom = "text",
    x = 2006, y = -27.5, 
    label = "Earliest information\non gender pay gaps", 
    family = "Compagnon-Light",
    color = "grey17", size = 3.9,
    lineheight = .85
  )  +
  annotate(
    geom = "segment",
    x = 2001.7, xend = 2010.3, 
    y = -26.7, yend = -26.7,
    color = "grey70", size = .4,
    arrow = arrow(type = "closed", ends = "both", unit(1.6, "mm"), angle = 50)
  ) +
  
  
  ## y grid --------------------------------------------------------------------
  annotate(
    geom = "segment",
    x = 2000.6, xend = 2020.4, 
    y = -1:-4*5, yend = -1:-4*5,
    color = "grey80", linetype = "42", size = .3
  ) +
  annotate(
    geom = "segment",
    x = 2000.6, xend = 2023.5, 
    y = -5:-6*5, yend = -5:-6*5,
    color = "grey80", linetype = "42", size = .3
  ) +
  annotate(
    geom = "segment",
    x = 2000.6, xend = 2023.5, 
    y = 0, yend = 0,
    color = "grey80", size = .6
  ) +
  annotate(
    geom = "segment",
    x = 2020, xend = 2020, 
    y = 1, yend = 0,
    color = "grey80", size = .6
  ) +
  
  ## annotation bubbles --------------------------------------------------------
  annotate(
    geom = "segment",
    x = 2020.2, xend = 2020.2, 
    y = -22, yend = -23.2,
    color = "grey80", size = .3
  ) +
  annotate(
    geom = "text",
    x = 2020.2, y = -23.9, 
    label = "Area equals\nabsolute change", 
    family = "Compagnon-Light",
    color = "grey30", size = 3.9,
    lineheight = .85
  ) +
  
  ## y-axis labels -------------------------------------------------------------
  annotate(
    geom = "text",
    x = 2000.6, y = -1:-5*5,
    label = scales::percent(1:5*5, scale = 1, accuracy = 1, trim = FALSE),
    hjust = 0, vjust = 1.7, color = "grey30", family = "Compagnon-Light", size = 4.8
  ) +
  annotate(
    geom = "text",
    x = 2000.6, y = c(1, 0, -30),
    label = c("Women's gross hourly earnings were...", 
              "equal to those of male paid employees in the EU", 
              "30% below those of male paid employees in the EU"),
    hjust = 0, vjust = 1.7, color = "#212121", family = "Compagnon-Roman", size = 4.8
  ) +
  
  ## annotation countries ------------------------------------------------------
  geom_text_repel(
    aes(x = 2020.2, y = -gap_2020, 
        color = trend, label = country_lab,
        color = after_scale(darken(color, .2))),
    family = "Compagnon-Roman", size = 3.8, hjust = 0,
    xlim = c(2021.5, NA), direction = "y", max.overlaps = 20,
    segment.size = .3, segment.alpha = .33, #segment.linetype = "13",
    segment.curvature = .05, segment.ncp = 3, segment.angle = 10,
    inflect = TRUE, box.padding = .2, show.legend = FALSE
  ) +
  
  ## slope -------------------------------------------------------------------
  geom_point(
    aes(x = 2020.2, y = -gap_2020, size = abs(change)),
    color = "white", show.legend = FALSE
  ) +
  geom_point(
    aes(x = 2020.2, y = -gap_2020, size = abs(change)),
    shape = 1, stroke = .6, show.legend = FALSE
  ) +
  geom_segment(
    data = filter(paygap_comparison, trend == "Smaller pay gap"),
    aes(xend = 2019.8, yend = -gap_2020), size = .3,
    arrow = arrow(type = "closed", length = unit(1.3, "mm"))
  ) + 
  geom_segment(
    data = filter(paygap_comparison, trend %in% c("Larger pay gap", "Stable pay gap")),
    aes(xend = 2019.8, yend = -gap_2020), size = .6, 
    arrow = arrow(type = "closed", length = unit(2.5, "mm"))
  ) + 
   geom_segment(
    data = filter(paygap_comparison, trend == "European Union"),
    aes(xend = 2019.8, yend = -gap_2020), size = .6, 
    arrow = arrow(type = "closed", length = unit(1.7, "mm"))
  ) + 
  geom_point(show.legend = FALSE) +
  
  ## coord + scales ------------------------------------------------------------
  coord_cartesian(expand = FALSE, clip = "off") +
  scale_x_continuous(
    breaks = c(sort(unique(paygap_comparison$year_ref)), 2020), 
    limits = c(2000.6, 2023.5), position = "top"
  ) +
  scale_y_continuous(breaks = 0:-6*5, limits = c(-31, 1)) +
  scale_color_manual(
    values = c("#B55951", "#F0B48E", "#91B9AE", "#212121"), 
    name = NULL
  ) +
  scale_size_area(max_size = 7.5) +
  
  ## styling -------------------------------------------------------------------
  theme_minimal(base_size = 15, base_family = "Compagnon-Roman") +
  theme(
    panel.grid = element_blank(),
    axis.ticks.length.x = unit(.4, "lines"),
    axis.ticks.x.top = element_line(color = "grey80", size = .6),
    axis.text.x.top = element_text(color = "grey30"),
    axis.text.y = element_blank(),
    plot.caption = element_textbox_simple(family = "Compagnon-Light", size = 11.5,
                                          color = "grey30", lineheight = 1.4,
                                           margin = margin(t = 35, b = 5), hjust = 0),
    legend.position = c(.87, .12),
    legend.text = element_text(color = "grey30", family = "Compagnon-Light", size = 11.5),
    legend.key.width = unit(1.8, "lines"),
    plot.margin = margin(40, 40, 30, 40)
  ) +
  labs(x = NULL, y = NULL,
       caption = "<b style='color:#212121;font-family:Compagnon-Roman;'>In 2020, women's gross hourly earnings were on average 13% below those of men in the EU.</b><br>Many countries were able to <b style='color:#6E9288;font-family:Compagnon-Roman;'>decrease the pay gap</b> between male and female employees: most notably <b style='color:#6E9288;font-family:Compagnon-Roman;'>Luxembourg</b> with the smallest gap in 2020 of 0.7% and <b style='color:#6E9288;font-family:Compagnon-Roman;'>Romania</b> with a decrease of 13.6% within 18 years to 2.4%, ranking second. However, in several countries the pay gap was almost <b style='color:#C38B64;font-family:Compagnon-Roman;'>stable (France, Lithuania, and Italy)</b> or even <b style='color:#854D49;font-family:Compagnon-Roman;'>increased (Latvia, Portugal, Croatia, and Malta)</b> compared to the earliest reference value listed by Eurostat.<br><br>As an unadjusted indicator, the gender pay gap gives an overall picture of the differences between men and women in terms of earnings and measures — a concept which is broader than discrimination in the sense of: <i style='color:#6E9288;font-family:Compagnon-Bold;'>equal pay for work of equal value.</i><br><br><b style='font-size:8.4pt;'>Graphic: Cédric Scherer • Data: Eurostat (SDG_05_20; no data for Greece and Ireland) • #InternationalWomensDay2022</b>")

ggsave("paygap_eu_slope.pdf", width = 10, height = 14, device = cairo_pdf)
pdf_convert("paygap_eu_slope.pdf", filenames = "paygap_eu_slope.png", dpi = 700)
```

***

<details><summary>Session Info</summary>

```{r sessionInfo}
Sys.time()
sessionInfo()
```

</details>