Code

# Libraries
library(geobr)
library(sf)
library(ggplot2)
library(dplyr)
library(tidyr)
library(stringr)
library(MetBrewer)
library(showtext)

# Color palette
cores = met.brewer("Hiroshige", 20)[c(1, 8, 15, 20)]
# Import Gill Sans locally
font_add("Gill Sans", "GillSans.ttc")
showtext_auto()

# Data ------------------------------------------------------------------

# Shapes

# Import city shapefile
munis <- read_municipality(year = 2022, showProgress = FALSE)
# Import state shapefile
geostate <- read_state(showProgress = FALSE)

# Import city data from Brazil (see data/raw/R/painel_municipios.R)
panel <- readr::read_csv(here::here("static/data", "cities_brazil.csv"))

## Data cleaning --------------------------------------------------------

# Vector to select GDP columns
pib_cols = c("pib_agriculture", "pib_industrial", "pib_services",
             "pib_govmt_services")

# Compute GDP share by city
pib_share = panel |> 
  select(all_of(c("code_muni", "pib", pib_cols))) |> 
  pivot_longer(cols = 3:last_col(), names_to = "sector", values_to = "total") |> 
  mutate(
    share = total / pib * 100,
    # Improve text labels
    sector_label = str_remove(sector, "pib_"),
    sector_label = str_to_title(sector_label),
    sector_label = str_replace(sector_label, "Govmt_services", "Government"),
    sector_label = str_replace(sector_label, "Industrial", "Industry")
    )

# Convert to wider
tab_pib_share = pib_share |> 
  pivot_wider(
    id_cols = "code_muni",
    names_from = "sector",
    values_from = "share"
    )

# Find the most relevant economic sector by city
top_pib_share = pib_share |> 
  filter(share == max(share), .by = "code_muni") |> 
  select(code_muni, top_sector = sector_label)

# Join tabular data with city shapefile
map_share = munis |> 
  left_join(tab_pib_share, by = "code_muni") |> 
  left_join(top_pib_share, by = "code_muni")

# Map -------------------------------------------------------------------

m1 = ggplot() +
  # State borders
  geom_sf(data = geostate, lwd = 0.5, color = "gray30", fill = "white") +
  # City colors
  geom_sf(
    data = na.omit(map_share),
    aes(fill = top_sector, color = top_sector),
    lwd = 0.01
    ) +
  # Remove junk from the east coast
  coord_sf(xlim = c(NA, -35)) +
  # Use Hiroshige colors
  scale_fill_manual(name = "", values = cores) +
  scale_color_manual(name = "", values = cores) +
  labs(
    title = "Top GDP contribution by City (2021)",
    subtitle = "Colors represent the most relevant economic contributing group in each city",
    caption = "Source: National Accounts, IBGE (2023)") +
  ggthemes::theme_map(base_family = "Gill Sans") +
  # Thematic elements
  theme(
    # Background
    panel.background = element_rect(fill = "#ffffff", color = NA),
    plot.background = element_rect(fill = "#ffffff", color = NA),
    # Legend
    legend.position = "left",
    legend.justification = 0.5,
    legend.key.size = unit(1, "cm"),
    legend.key.width = unit(1.5, "cm"),
    legend.text = element_text(size = 12),
    legend.margin = margin(),
    # Labels (title, subtitle)
    plot.title = element_text(size = 38, hjust = 0.5),
    plot.subtitle = element_text(size = 14, hjust = 0.5),
    # Plot margins
    plot.margin = margin(10, 5, 5, 10)
  )

# Histogram -------------------------------------------------------------

pib_bars = top_pib_share |> 
  count(top_sector) |> 
  mutate(top_sector = forcats::fct_rev(top_sector))

p_hist = ggplot(pib_bars, aes(top_sector, n)) +
  geom_col(aes(fill = top_sector)) +
  # Number labels
  geom_text(
    aes(y = n, label = n, color = factor(c(0, 0, 0, 1))),
    size = 4,
    family = "Gill Sans",
    nudge_y = c(-200, -200, 200, -200)
    ) +
  # Text labels
  geom_text(
    aes(y = 50, label = top_sector, color = factor(c(0, 0, 0, 1))),
    family = "Gill Sans",
    size = 4,
    hjust = 0
  ) +
  # Flip axis
  coord_flip() +
  scale_fill_manual(values = rev(cores)) +
  scale_color_manual(values = c("black", "white")) +
  labs(x = NULL, y = NULL, title = "Number of cities") +
  guides(fill = "none", color = "none") +
  theme_minimal(base_family = "Gill Sans") +
  theme(
    plot.title = element_text(size = 14, hjust = 0.5),
    panel.grid.major = element_blank(),
    panel.grid.minor = element_blank(),
    axis.text = element_blank()
    )

GDP Contribution by city

The number of cities where Agriculture is the main driver of the local economy rose from 1049 in 2020 to 1272 in 2021. Agriculture is most prevalent in the Midwest cities.
Public Administration is the most prevalent economic activity among cities in the North and Northeast of the country. General services are most prevalent in the Southeast.
Industrial activities constitute the primary economic driver for fewer than 8% of cities. The few industrial cities are spatially disperse across the country, exhibiting no clear pattern.

Dados: IBGE, Contas Nacionais, Produto Interno Bruto dos Municípios (2023)
Tipografia: Gill Sans
Paleta: Hiroshige (MetBrewer)