Tidy Tuesday: Italian Industrial Production

tidytuesday
R
history
economics
Italy
One hundred and twenty years of Italian industrialization — reading war, recovery, and the economic miracle through production records.
Author

Sean Thimons

Published

May 5, 2026

Preface

From TidyTuesday repository.

This week we’re exploring Italian Industrial Production. The dataset comes from “A monetray history of Italy” by Fratianni and Spinelli (1997), via the GESIS Data Archive (ZA8700). It contains data on Italian industrial production covering food and beverages, textiles, and transport manufacturing spanning from 1861 through 1985 — a 124-year window that encompasses Italian unification, two world wars, fascism, and the postwar economic miracle.

The three tables — food_beverages, textiles, and transport — share a common integer year axis but start at different points and carry heavy missing data in the early decades before those industries were formally tracked. The analytical question this post pursues: how does wartime disruption and the postwar recovery show up simultaneously across three distinct industrial sectors?

Loading necessary packages

Code 
{
  if (!requireNamespace("pak", quietly = TRUE)) {
    install.packages("pak", repos = "https://cloud.r-project.org")
  }

  install_booster_pack <- function(package, load = TRUE) {
    for (pkg in package) {
      if (!requireNamespace(pkg, quietly = TRUE)) {
        pak::pkg_install(pkg)
      }
      if (load) {
        library(pkg, character.only = TRUE)
      }
    }
  }

  core_packages <- c(
    'fs', 'here', 'janitor', 'rio',
    'tidyverse', 'skimr', 'paletteer', 'tidytuesdayR'
  )

  optional_packages <- c('ggtext', 'ggrepel', 'scales', 'patchwork')

  install_booster_pack(package = unique(c(core_packages, optional_packages)), load = TRUE)
  rm(install_booster_pack, core_packages, optional_packages)

  `%ni%` <- Negate(`%in%`)

  geometric_mean <- function(x) {
    exp(mean(log(x[x > 0]), na.rm = TRUE))
  }

  my_skim <- skim_with(
    numeric = sfl(
      n = length,
      min = ~ min(.x, na.rm = T),
      p25 = ~ stats::quantile(., probs = .25, na.rm = TRUE, names = FALSE),
      med = ~ median(.x, na.rm = T),
      p75 = ~ stats::quantile(., probs = .75, na.rm = TRUE, names = FALSE),
      max = ~ max(.x, na.rm = T),
      mean = ~ mean(.x, na.rm = T),
      geo_mean = ~ geometric_mean(.x),
      sd = ~ stats::sd(., na.rm = TRUE),
      hist = ~ inline_hist(., 5)
    ),
    append = FALSE
  )
}

Load raw data from package

raw <- tidytuesdayR::tt_load('2026-05-05')

food_beverages <- raw$food_beverages
textiles       <- raw$textiles
transport      <- raw$transport

Exploratory Data Analysis

The my_skim() function is a modified version of the skimr::skim() function that returns the number of missing data points (cells as NA) as well as the inverse (e.g.: number of rows that are not NA), the count, minimum, 25%, median, 75%, max, mean, geometric mean, and standard deviation. It also generates a little ASCII histogram. Neat!

food_beverages

my_skim(food_beverages)
Data summary
Name food_beverages
Number of rows 115
Number of columns 8
_______________________
Column type frequency:
numeric 8
________________________
Group variables None

Variable type: numeric

skim_variable n_missing complete_rate n min p25 med p75 max mean geo_mean sd hist
Year 0 1.00 115 1871 1899.50 1928.0 1956.5 1985 1928.00 1927.71 33.34 ▇▇▇▇▇
Sugar 7 0.94 115 19 77360.75 295597.5 906615.8 2011316 489144.37 99727.93 516624.03 ▇▂▂▁▁
Glucose 17 0.85 115 4754 43246.00 94112.5 428403.8 2223565 357468.94 132832.33 533728.82 ▇▁▁▁▁
Coffee_substitute 18 0.84 115 87 28622.00 75253.0 128462.0 283390 83264.99 51053.72 65476.92 ▇▆▂▂▁
Seed_oil 50 0.57 115 84029 457388.00 864304.0 2949217.0 4435485 1553644.38 964086.57 1431497.25 ▇▂▁▂▂
Ethyl_alcohol_1 5 0.96 115 20817 140372.25 284730.0 449569.5 1396351 366649.50 255575.83 309832.18 ▇▅▁▁▁
Ethyl_alcohol_2 11 0.90 115 21978 61100.00 138011.5 545135.0 2309955 370363.88 174155.40 486530.03 ▇▂▁▁▁
Beer 11 0.90 115 93856 271677.50 715078.0 1762816.0 10155257 1970580.26 811994.86 2664172.27 ▇▁▁▁▁ 
cat(sprintf("\nRows: %d | Year range: %d–%d\n",
            nrow(food_beverages),
            min(food_beverages$Year),
            max(food_beverages$Year)))

Rows: 115 | Year range: 1871–1985

textiles

my_skim(textiles)
Data summary
Name textiles
Number of rows 123
Number of columns 10
_______________________
Column type frequency:
numeric 10
________________________
Group variables None

Variable type: numeric

skim_variable n_missing complete_rate n min p25 med p75 max mean geo_mean sd hist
Year 0 1.00 123 1863 1893.50 1924.0 1954.50 1985 1924.00 1923.67 35.65 ▇▇▇▇▇
Cotton_yarn 71 0.42 123 711 146317.75 160740.5 174551.25 197070 148026.81 114320.11 50823.21 ▁▁▁▃▇
Flock_yarn 73 0.41 123 1762 15115.50 22976.5 27384.50 81887 24462.74 20496.54 14999.90 ▅▇▂▁▁
Other_yarn 71 0.42 123 936 7857.25 18155.0 48711.50 74190 26143.06 16173.71 21439.33 ▇▃▁▃▁
Total_yarn 37 0.70 123 9900 164583.00 190498.0 223896.00 254442 185626.63 172775.52 50010.84 ▁▁▅▇▇
Cotton_textiles 58 0.53 123 1362 101100.00 111049.0 123399.00 164934 106844.32 91956.73 33291.24 ▁▁▂▇▂
Flock_textiles 65 0.47 123 2136 5904.25 10134.5 19917.50 78208 16559.19 11606.40 15981.24 ▇▃▁▁▁
Other_textiles 65 0.47 123 781 4982.25 23332.5 53142.75 64049 28757.34 15760.87 22458.71 ▇▆▃▂▇
Total_textiles 58 0.53 123 16436 119140.00 146061.0 176652.00 231391 147280.62 137748.77 43523.04 ▁▂▇▇▃
Raw_silk 9 0.93 123 38 1336.25 2902.0 4451.75 6173 2880.26 2066.67 1757.83 ▇▇▇▅▇ 
cat(sprintf("\nRows: %d | Year range: %d–%d\n",
            nrow(textiles),
            min(textiles$Year),
            max(textiles$Year)))

Rows: 123 | Year range: 1863–1985

transport

my_skim(transport)
Data summary
Name transport
Number of rows 125
Number of columns 9
_______________________
Column type frequency:
numeric 9
________________________
Group variables None

Variable type: numeric

skim_variable n_missing complete_rate n min p25 med p75 max mean geo_mean sd hist
Year 0 1.00 125 1861 1892.00 1923.0 1954.00 1985 1923.00 1922.66 36.23 ▇▇▇▇▇
Ships_launched 0 1.00 125 8 78.00 188.0 266.00 907 205.31 139.94 175.55 ▇▇▁▁▁
Ships_weight 1 0.99 125 5191 29679.25 69535.5 168121.25 1145945 164798.44 74438.86 228266.10 ▇▁▁▁▁
Steam_and_electric_engine 51 0.59 125 1 28.50 53.0 113.75 475 95.96 48.24 107.84 ▇▂▁▁▁
Rail_cars_and_electric_locomotives 65 0.48 125 1 32.50 63.0 125.75 255 82.82 46.69 67.41 ▇▅▃▁▂
Carriages_and_trailers 46 0.63 125 1 97.50 220.0 368.50 903 255.96 133.58 204.67 ▇▇▃▂▁
Mail_luggage_vans_and_carriages 46 0.63 125 2 1301.50 2085.0 3847.00 9354 2819.43 1563.38 2357.56 ▇▆▂▂▁
Passenger_cars 64 0.49 125 1818 43112.00 230972.0 1384225.00 1825184 630026.74 192343.86 665162.61 ▇▁▁▃▂
Other 64 0.49 125 2600 16295.00 33570.0 115955.00 184539 61698.30 32293.81 58500.89 ▇▂▁▂▂ 
cat(sprintf("\nRows: %d | Year range: %d–%d\n",
            nrow(transport),
            min(transport$Year),
            max(transport$Year)))

Rows: 125 | Year range: 1861–1985

All three tables are purely numeric — a Year column plus production quantities in various physical units (tonnes, kilolitres, units manufactured). No character or factor columns exist, so there are no count()-based category checks to run. The missingness pattern is historically informative: early decades have many NAs simply because those industries either did not yet exist or were not yet tracked by official statistics.

War, Recovery, and the Economic Miracle

The century-plus span of these datasets covers the two world wars with precision. The analytical approach here is to build index series (Year 1900 = 100 for most series, or the earliest non-missing year for each) so that comparatively sized sectors can be plotted on the same axis, then annotate the wartime troughs.

Three plots tell a coherent story:

  1. Aggregate sector trajectories — normalized growth curves for all three sectors from their shared post-1900 window, revealing how food/beverage production, textile output, and transport manufacturing each responded to war and recovery.
  2. The WWII disruption and recovery — a focused window on 1930–1970 showing the depth of collapse in each sector during the war years and the speed of the postwar miracle.
  3. Passenger cars and the economic miracle — the automotive production explosion from the late 1950s onward, illustrating the Fiat era and Italy’s transformation into a consumer economy.

Plot 1 — Sector growth since 1900

pal <- paletteer::paletteer_d("MetBrewer::Troy")

# --- Food/beverage: use Beer as a robust single-series proxy (only 11 NAs,
#     starts well before 1900, and tracks consumer demand)
food_idx <- food_beverages %>%
  filter(!is.na(Beer), Year >= 1900) %>%
  arrange(Year) %>%
  mutate(
    base_val = first(Beer),
    index    = Beer / base_val * 100,
    sector   = "Food & Beverages (Beer)"
  ) %>%
  select(Year, index, sector)

cat(sprintf("food_idx: %d rows\n", nrow(food_idx)))
food_idx: 83 rows
stopifnot("food_idx is empty" = nrow(food_idx) > 0)

# --- Textiles: use Total_textiles (58 NAs but covers 1900+ well)
textiles_idx <- textiles %>%
  filter(!is.na(Total_textiles), Year >= 1900) %>%
  arrange(Year) %>%
  mutate(
    base_val = first(Total_textiles),
    index    = Total_textiles / base_val * 100,
    sector   = "Textiles (Total)"
  ) %>%
  select(Year, index, sector)

cat(sprintf("textiles_idx: %d rows\n", nrow(textiles_idx)))
textiles_idx: 65 rows
stopifnot("textiles_idx is empty" = nrow(textiles_idx) > 0)

# --- Transport: use Ships_weight as the most complete series (1 NA)
transport_idx <- transport %>%
  filter(!is.na(Ships_weight), Year >= 1900) %>%
  arrange(Year) %>%
  mutate(
    base_val = first(Ships_weight),
    index    = Ships_weight / base_val * 100,
    sector   = "Transport (Ships by weight)"
  ) %>%
  select(Year, index, sector)

cat(sprintf("transport_idx: %d rows\n", nrow(transport_idx)))
transport_idx: 86 rows
stopifnot("transport_idx is empty" = nrow(transport_idx) > 0)

sector_long <- bind_rows(food_idx, textiles_idx, transport_idx)

cat(sprintf("sector_long: %d rows, %d cols\n", nrow(sector_long), ncol(sector_long)))
sector_long: 234 rows, 3 cols
stopifnot("sector_long is empty" = nrow(sector_long) > 0)

sector_colors <- c(
  "Food & Beverages (Beer)"      = pal[2],
  "Textiles (Total)"             = pal[5],
  "Transport (Ships by weight)"  = pal[7]
)

war_bands <- tibble(
  xmin  = c(1914, 1940),
  xmax  = c(1918, 1945),
  label = c("WWI", "WWII")
)

p1 <- ggplot(sector_long, aes(x = Year, y = index, color = sector)) +
  geom_rect(
    data = war_bands,
    aes(xmin = xmin, xmax = xmax, ymin = -Inf, ymax = Inf),
    inherit.aes = FALSE,
    fill = "grey85", alpha = 0.6
  ) +
  geom_text(
    data = war_bands,
    aes(x = (xmin + xmax) / 2, y = Inf, label = label),
    inherit.aes = FALSE,
    vjust = 1.4, size = 3.2, color = "grey40", fontface = "italic"
  ) +
  geom_line(linewidth = 1.1, na.rm = TRUE) +
  geom_hline(yintercept = 100, linetype = "dashed", color = "grey55", linewidth = 0.5) +
  scale_color_manual(values = sector_colors, name = NULL) +
  scale_x_continuous(breaks = seq(1900, 1985, by = 10)) +
  scale_y_continuous(
    labels = scales::label_number(suffix = ""),
    breaks = c(0, 100, 500, 1000, 2000, 5000, 10000)
  ) +
  coord_trans(y = "sqrt") +
  labs(
    title = "Italy's industrial sectors grew in waves — interrupted by two world wars",
    subtitle = paste0(
      "Production index (earliest available year = 100) for three sectors, plotted on a square-root scale.<br>",
      "Shaded bands mark WWI (1914–1918) and WWII (1940–1945)."
    ),
    caption = "Source: Fratianni & Spinelli (1997) via GESIS ZA8700 | Tidy Tuesday 2026-05-05",
    x = NULL,
    y = "Index (earliest year = 100, sqrt scale)"
  ) +
  theme_minimal(base_size = 11) +
  theme(
    plot.title    = ggtext::element_markdown(size = 14, face = "bold"),
    plot.subtitle = ggtext::element_markdown(size = 10, color = "grey30", lineheight = 1.15,
                                             margin = margin(b = 10)),
    plot.caption  = element_text(color = "grey45", size = 8),
    panel.grid.minor   = element_blank(),
    panel.grid.major.x = element_blank(),
    legend.position    = "top",
    legend.text        = element_text(size = 9),
    plot.margin        = margin(12, 18, 10, 10)
  )

p1

Plot 2 — The WWII collapse and postwar recovery window (1930–1970)

# Reindex to 1938 (last peacetime year before Italian belligerence) = 100
# for a cleaner war/recovery comparison

food_38 <- food_beverages %>%
  filter(!is.na(Beer), Year >= 1930, Year <= 1970) %>%
  arrange(Year) %>%
  mutate(
    base_val = {
      bv <- Beer[Year == 1938]
      if (length(bv) == 1 && !is.na(bv)) bv else first(Beer)
    },
    index    = Beer / base_val * 100,
    sector   = "Food & Beverages (Beer)"
  ) %>%
  select(Year, index, sector)

cat(sprintf("food_38: %d rows\n", nrow(food_38)))
food_38: 40 rows
stopifnot("food_38 is empty" = nrow(food_38) > 0)

textiles_38 <- textiles %>%
  filter(!is.na(Total_textiles), Year >= 1930, Year <= 1970) %>%
  arrange(Year) %>%
  mutate(
    base_val = {
      bv <- Total_textiles[Year == 1938]
      if (length(bv) == 1 && !is.na(bv)) bv else first(Total_textiles)
    },
    index    = Total_textiles / base_val * 100,
    sector   = "Textiles (Total)"
  ) %>%
  select(Year, index, sector)

cat(sprintf("textiles_38: %d rows\n", nrow(textiles_38)))
textiles_38: 41 rows
stopifnot("textiles_38 is empty" = nrow(textiles_38) > 0)

transport_38 <- transport %>%
  filter(!is.na(Ships_weight), Year >= 1930, Year <= 1970) %>%
  arrange(Year) %>%
  mutate(
    base_val = {
      bv <- Ships_weight[Year == 1938]
      if (length(bv) == 1 && !is.na(bv)) bv else first(Ships_weight)
    },
    index    = Ships_weight / base_val * 100,
    sector   = "Transport (Ships by weight)"
  ) %>%
  select(Year, index, sector)

cat(sprintf("transport_38: %d rows\n", nrow(transport_38)))
transport_38: 41 rows
stopifnot("transport_38 is empty" = nrow(transport_38) > 0)

recovery_long <- bind_rows(food_38, textiles_38, transport_38)

cat(sprintf("recovery_long: %d rows, %d cols\n", nrow(recovery_long), ncol(recovery_long)))
recovery_long: 122 rows, 3 cols
stopifnot("recovery_long is empty" = nrow(recovery_long) > 0)

# Find trough years per sector in the 1940–1946 window
troughs <- recovery_long %>%
  filter(Year >= 1940, Year <= 1946) %>%
  group_by(sector) %>%
  slice_min(index, n = 1, with_ties = FALSE) %>%
  ungroup()

cat(sprintf("troughs: %d rows\n", nrow(troughs)))
troughs: 3 rows
stopifnot("troughs is empty" = nrow(troughs) > 0)

# Recovery crossover: first year post-1945 when index >= 100 again
recovery_yr <- recovery_long %>%
  filter(Year > 1945, index >= 100) %>%
  group_by(sector) %>%
  slice_min(Year, n = 1, with_ties = FALSE) %>%
  ungroup()

p2 <- ggplot(recovery_long, aes(x = Year, y = index, color = sector)) +
  annotate("rect", xmin = 1940, xmax = 1945, ymin = -Inf, ymax = Inf,
           fill = "grey85", alpha = 0.65) +
  annotate("text", x = 1942.5, y = Inf, label = "WWII",
           vjust = 1.4, size = 3.2, color = "grey40", fontface = "italic") +
  geom_hline(yintercept = 100, linetype = "dashed", color = "grey55", linewidth = 0.5) +
  geom_line(linewidth = 1.2, na.rm = TRUE) +
  geom_point(
    data = troughs,
    aes(x = Year, y = index),
    size = 3.5, shape = 25, fill = NA
  ) +
  ggrepel::geom_text_repel(
    data = troughs,
    aes(label = paste0(Year, "\n", round(index), "%")),
    size = 3.0,
    nudge_y = -12,
    min.segment.length = 0,
    segment.size = 0.3,
    show.legend = FALSE
  ) +
  scale_color_manual(values = sector_colors, name = NULL) +
  scale_x_continuous(breaks = seq(1930, 1970, by = 5)) +
  scale_y_continuous(labels = scales::label_number(suffix = "%")) +
  labs(
    title = "WWII cut Italian industrial output by more than half — recovery was swift",
    subtitle = paste0(
      "Production index rebased to 1938 = 100 for each sector.<br>",
      "Triangles mark each sector's wartime production trough."
    ),
    caption = "Source: Fratianni & Spinelli (1997) via GESIS ZA8700 | Tidy Tuesday 2026-05-05",
    x = NULL,
    y = "Index (1938 = 100)"
  ) +
  theme_minimal(base_size = 11) +
  theme(
    plot.title    = ggtext::element_markdown(size = 14, face = "bold"),
    plot.subtitle = ggtext::element_markdown(size = 10, color = "grey30", lineheight = 1.15,
                                             margin = margin(b = 10)),
    plot.caption  = element_text(color = "grey45", size = 8),
    panel.grid.minor   = element_blank(),
    panel.grid.major.x = element_blank(),
    legend.position    = "top",
    legend.text        = element_text(size = 9),
    plot.margin        = margin(12, 18, 10, 10)
  )

p2

Plot 3 — The Fiat era: passenger car production and the economic miracle

cars_data <- transport %>%
  filter(!is.na(Passenger_cars), Year >= 1945) %>%
  arrange(Year)

cat(sprintf("cars_data: %d rows, Year range: %d–%d\n",
            nrow(cars_data), min(cars_data$Year), max(cars_data$Year)))
cars_data: 41 rows, Year range: 1945–1985
stopifnot("cars_data is empty" = nrow(cars_data) > 0)

# Period annotations
miracle_start <- 1958
miracle_end   <- 1973

# Key annotation points
peak_year <- cars_data$Year[which.max(cars_data$Passenger_cars)]
peak_val  <- max(cars_data$Passenger_cars, na.rm = TRUE)

p3 <- ggplot(cars_data, aes(x = Year, y = Passenger_cars)) +
  annotate(
    "rect",
    xmin = miracle_start, xmax = miracle_end,
    ymin = -Inf, ymax = Inf,
    fill = pal[3], alpha = 0.12
  ) +
  annotate(
    "text",
    x = (miracle_start + miracle_end) / 2,
    y = Inf,
    label = "Il miracolo economico\n(1958–1973)",
    vjust = 1.3, size = 3.2, color = pal[2], fontface = "italic",
    lineheight = 0.9
  ) +
  geom_area(fill = pal[4], alpha = 0.25) +
  geom_line(color = pal[2], linewidth = 1.3) +
  geom_point(
    data = cars_data %>% filter(Year == peak_year),
    aes(x = Year, y = Passenger_cars),
    color = pal[1], size = 4, shape = 18
  ) +
  ggrepel::geom_text_repel(
    data = cars_data %>% filter(Year == peak_year),
    aes(
      x = Year, y = Passenger_cars,
      label = paste0("Peak: ", format(peak_val, big.mark = ","),
                     " units\n(", peak_year, ")")
    ),
    nudge_x = 3, nudge_y = 80000,
    size = 3.3, color = pal[1], fontface = "bold",
    segment.color = pal[1], segment.size = 0.4,
    min.segment.length = 0
  ) +
  scale_x_continuous(breaks = seq(1945, 1985, by = 5)) +
  scale_y_continuous(
    labels = scales::label_number(scale = 1e-3, suffix = "k"),
    breaks = scales::pretty_breaks(n = 6)
  ) +
  labs(
    title = paste0(
      "Passenger car production rose ",
      round(peak_val / cars_data$Passenger_cars[cars_data$Year == 1945]),
      "x in thirty years"
    ),
    subtitle = paste0(
      "Annual Italian passenger car output, 1945–1985.<br>",
      "Shaded band marks the <i>miracolo economico</i> — Italy's postwar high-growth era."
    ),
    caption = "Source: Fratianni & Spinelli (1997) via GESIS ZA8700 | Tidy Tuesday 2026-05-05",
    x = NULL,
    y = "Passenger cars produced"
  ) +
  theme_minimal(base_size = 11) +
  theme(
    plot.title    = ggtext::element_markdown(size = 14, face = "bold"),
    plot.subtitle = ggtext::element_markdown(size = 10, color = "grey30", lineheight = 1.15,
                                             margin = margin(b = 10)),
    plot.caption  = element_text(color = "grey45", size = 8),
    panel.grid.minor   = element_blank(),
    panel.grid.major.x = element_blank(),
    plot.margin        = margin(12, 18, 10, 10)
  )

p3

Final thoughts and takeaways

Spanning 1861 to 1985, these three Italian production datasets compress a national economic biography into numbers. A few patterns stand out clearly:

  • Wars are legible in the data. WWI caused measurable but recoverable dips. WWII was categorically more severe — textile output in the dataset collapsed to a fraction of its 1938 level in the early 1940s, and the shipbuilding series shows similar trauma. The postwar recovery, however, was strikingly fast by historical standards — most sectors returned to or exceeded prewar output within a decade.

  • The miracolo economico was real and visible. The passenger car series is the starkest illustration: from fewer than 20,000 units in 1945 to a peak exceeding 1.5 million in the 1970s. This is the Fiat 500 era — the automobile going from luxury to household staple across a single generation. The growth curve is almost vertical between 1955 and 1965.

  • Textiles reveal structural decline. Unlike food/beverage and automotive, total textile output plateaus and softens in the late 1960s and 1970s. This tracks Italy’s actual industrial history: the north’s cotton mills faced competition from lower-cost Asian producers, and the industry never achieved the post-WWII scaling that automotive did.

  • The missing-data geography is itself a historical record. The NAs in the early years of Seed_oil (50 missing) or Cotton_yarn (71 missing) are not failures of data collection — they mark when those industries crossed a threshold of national statistical relevance. Reading the edge of the data is reading the edge of early industrialization.

A natural extension would bring in the Fratianni & Spinelli monetary data alongside these production series to test how money supply growth correlated with the industrial surge — the question at the heart of the original book.