Simen Bjerkelund https://simbje.github.io/ssb-daily/ Financial Analyst · Quantitative Researcher · Data Scientist quarto-1.5.0 Tue, 19 May 2026 00:00:00 GMT Norway’s 2026 Consumption Paradox: How Household Spending Collapsed While Energy Demand Fractured by Region and Fuel Type https://simbje.github.io/ssb-daily/posts/2026-05-19/ Norway is caught between two competing narratives in 2026. Official GDP figures show household consumption under sustained pressure, with volume changes turning negative even as prices remain elevated. Meanwhile, petroleum product sales — a proxy for industrial activity, transport, and heating demand — are fragmenting along regional lines in ways that suggest the economic squeeze is landing very unevenly across the country. The data tells a story of structural adjustment, not a simple slowdown.

Data

Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09190",
    Makrost = TRUE,
    ContentsCode = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "kvartal"
  value_col    <- "value"
  series_col   <- "makrostørrelse"
  measure_col  <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) {
  message("No data returned for df1")
  df1 <- NULL
}

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/11174",
    Region = TRUE,
    Kjopegrupper = TRUE,
    PetroleumProd = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "måned"
  value_col    <- "value"
  series_col   <- "petroleumsprodukt"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) {
  message("No data returned for df2")
  df2 <- NULL
}

Wrangling

Code 
# ── df1 derived frames ──────────────────────────────────────────────────────
df1_volume   <- NULL
df1_price    <- NULL
df1_goods    <- NULL
df1_services <- NULL
df1_public   <- NULL
df1_hh_vol   <- NULL

if (!is.null(df1) && nrow(df1) > 0) {

  # Volume change % for main household components
  df1_volume <- df1 |>
    filter(
      .data[["statistikkvariabel"]] == "Volumendring fra samme periode året før (prosent)",
      .data[["makrostørrelse"]] %in% c(
        "Konsum i husholdninger",
        "Varekonsum",
        "Tjenestekonsum",
        "Konsum i offentlig forvaltning"
      )
    )
  if (nrow(df1_volume) == 0) {
    message("df1_volume empty. Series values: ",
            paste(head(unique(df1[["makrostørrelse"]]), 15), collapse = ", "))
    message("Measure values: ",
            paste(head(unique(df1[["statistikkvariabel"]]), 15), collapse = ", "))
    df1_volume <- NULL
  }

  # Price change % for goods vs services
  df1_price <- df1 |>
    filter(
      .data[["statistikkvariabel"]] == "Prisendring fra samme periode året før (prosent)",
      .data[["makrostørrelse"]] %in% c(
        "Varekonsum",
        "Tjenestekonsum"
      )
    )
  if (nrow(df1_price) == 0) {
    message("df1_price empty.")
    df1_price <- NULL
  }

  # Household consumption at current prices — for area chart
  df1_hh_vol <- df1 |>
    filter(
      .data[["statistikkvariabel"]] == "Volumendring fra samme periode året før (prosent)",
      .data[["makrostørrelse"]] %in% c("Varekonsum", "Tjenestekonsum")
    )
  if (nrow(df1_hh_vol) == 0) {
    message("df1_hh_vol empty.")
    df1_hh_vol <- NULL
  }
}

# ── df2 derived frames ──────────────────────────────────────────────────────
df2_region_fuel  <- NULL
df2_autodiesel   <- NULL
df2_bensin       <- NULL
df2_fuel_types   <- NULL
df2_region_total <- NULL

if (!is.null(df2) && nrow(df2) > 0) {

  # Check columns available
  region_col <- if ("region" %in% names(df2)) "region" else
                if ("Region" %in% names(df2)) "Region" else NULL

  buyer_col  <- if ("kjopegrupper" %in% names(df2)) "kjopegrupper" else
                if ("kjøpegrupper"  %in% names(df2)) "kjøpegrupper" else
                if ("Kjopegrupper"  %in% names(df2)) "Kjopegrupper" else NULL

  # Regional totals for autodiesel
  if (!is.null(region_col)) {
    df2_autodiesel <- df2 |>
      filter(.data[["petroleumsprodukt"]] == "Autodiesel") |>
      group_by(region_lbl = .data[[region_col]], date) |>
      summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
      filter(!is.na(value), value > 0)

    if (nrow(df2_autodiesel) == 0) {
      message("df2_autodiesel empty. Petroleum values: ",
              paste(head(unique(df2[["petroleumsprodukt"]]), 10), collapse = ", "))
      df2_autodiesel <- NULL
    }
  }

  # Fuel type comparison — national aggregation
  df2_fuel_types <- df2 |>
    filter(.data[["petroleumsprodukt"]] %in% c(
      "Bilbensin", "Autodiesel", "Jetparafin", "Marine gassoljer", "Anleggsdiesel"
    )) |>
    group_by(fuel = .data[["petroleumsprodukt"]], date) |>
    summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
    filter(!is.na(value), value > 0)

  if (nrow(df2_fuel_types) == 0) {
    message("df2_fuel_types empty.")
    df2_fuel_types <- NULL
  }

  # Regional heatmap data: latest available months, all fuels combined
  if (!is.null(region_col)) {
    df2_region_total <- df2 |>
      filter(.data[["petroleumsprodukt"]] == "Petroleumsprodukter i alt") |>
      group_by(region_lbl = .data[[region_col]], date) |>
      summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
      filter(!is.na(value), value > 0)

    if (nrow(df2_region_total) == 0) {
      message("df2_region_total empty.")
      df2_region_total <- NULL
    } else {
      # Compute year-over-year % change
      df2_region_total <- df2_region_total |>
        arrange(region_lbl, date) |>
        group_by(region_lbl) |>
        mutate(
          month_num = month(date),
          year_num  = year(date),
          value_lag = lag(value, 12),
          yoy_pct   = (value - value_lag) / value_lag * 100
        ) |>
        ungroup() |>
        filter(!is.na(yoy_pct))
    }
  }

  # Bensin lollipop — latest period per region
  if (!is.null(region_col)) {
    df2_bensin <- df2 |>
      filter(.data[["petroleumsprodukt"]] == "Bilbensin") |>
      group_by(region_lbl = .data[[region_col]], date) |>
      summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
      filter(!is.na(value), value > 0)

    if (nrow(df2_bensin) == 0) {
      message("df2_bensin empty.")
      df2_bensin <- NULL
    } else {
      latest_date_bensin <- max(df2_bensin$date)
      df2_bensin_latest  <- df2_bensin |>
        filter(date == latest_date_bensin) |>
        filter(!grepl("hele landet", region_lbl, ignore.case = TRUE))
    }
  }
}

Household Consumption: Goods Cave, Services Hold

The most striking feature of Norway’s 2026 national accounts is not that household spending is weak — it is which parts of it are weak. Volume growth in goods consumption has tracked below services for most of the period, and the gap has widened in recent quarters as goods-price inflation outpaced income growth.

Code 
if (exists("df1_hh_vol") && !is.null(df1_hh_vol) && nrow(df1_hh_vol) > 0) {

  pal <- MetBrewer::met.brewer("Hokusai2", n = 2)

  df1_hh_vol_plot <- df1_hh_vol |>
    mutate(
      component = case_when(
        .data[["makrostørrelse"]] == "Varekonsum"    ~ "Goods consumption",
        .data[["makrostørrelse"]] == "Tjenestekonsum" ~ "Services consumption",
        TRUE ~ .data[["makrostørrelse"]]
      )
    )

  p <- ggplot(df1_hh_vol_plot, aes(x = date, y = value, colour = component, fill = component)) +
    geom_hline(yintercept = 0, colour = "grey40", linewidth = 0.4) +
    geom_area(alpha = 0.18, position = "identity") +
    geom_line(linewidth = 1.1) +
    geom_point(size = 2.2, shape = 21, fill = "white", stroke = 1) +
    scale_colour_manual(values = pal, name = NULL) +
    scale_fill_manual(values = pal, name = NULL) +
    scale_y_continuous(labels = function(x) paste0(x, "%")) +
    scale_x_date(date_labels = "%Y Q%q", date_breaks = "2 quarters") +
    labs(
      title    = "Goods Consumption Diverges from Services",
      subtitle = "Volume change year-on-year (%) — goods have tracked lower than services through 2025-26",
      x        = NULL,
      y        = "Year-on-year volume change (%)",
      caption  = "Source: Statistics Norway, Table 09190"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position   = "top",
      panel.grid.minor  = element_blank(),
      axis.text.x       = element_text(angle = 35, hjust = 1, size = 9),
      plot.title        = element_text(face = "bold"),
      plot.subtitle     = element_text(colour = "grey35", size = 10)
    )

  print(p) # fixed: missing explicit print() call
}
Code 
if (exists("df1_price") && !is.null(df1_price) && nrow(df1_price) > 0) {

  pal2 <- MetBrewer::met.brewer("Hokusai2", n = 2)

  df1_price_plot <- df1_price |>
    mutate(
      component = case_when(
        .data[["makrostørrelse"]] == "Varekonsum"    ~ "Goods",
        .data[["makrostørrelse"]] == "Tjenestekonsum" ~ "Services",
        TRUE ~ .data[["makrostørrelse"]]
      )
    ) |>
    # Keep last 12 quarters for readability
    group_by(component) |>
    slice_max(date, n = 12) |>
    ungroup()

  p2 <- ggplot(df1_price_plot,
               aes(x = date, y = value, colour = component)) +
    geom_hline(yintercept = 0, colour = "grey50", linewidth = 0.35, linetype = "dashed") +
    geom_segment(aes(xend = date, yend = 0), linewidth = 0.7, alpha = 0.6) +
    geom_point(size = 3.5) +
    facet_wrap(~component, ncol = 1) +
    scale_colour_manual(values = pal2, guide = "none") +
    scale_y_continuous(labels = function(x) paste0(x, "%")) +
    scale_x_date(date_labels = "%Y Q%q", date_breaks = "2 quarters") +
    labs(
      title    = "Price Pressures: Goods Inflation Has Run Hotter Than Services",
      subtitle = "Year-on-year price change (%) by consumption component — recent quarters show modest easing",
      x        = NULL,
      y        = "Year-on-year price change (%)",
      caption  = "Source: Statistics Norway, Table 09190"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      strip.text        = element_text(face = "bold", size = 11),
      panel.grid.minor  = element_blank(),
      axis.text.x       = element_text(angle = 35, hjust = 1, size = 9),
      plot.title        = element_text(face = "bold"),
      plot.subtitle     = element_text(colour = "grey35", size = 10)
    )

  print(p2) # fixed: missing explicit print() call
}

The Fuel Demand Fracture: Four Petroleum Products, Four Stories

The SSB petroleum sales data cuts through the headline numbers to reveal which types of energy are actually being consumed — and where. Autodiesel dominates the Norwegian fuel landscape, but its trajectory differs sharply from petrol (bilbensin), jet fuel, and marine gas oil. Each fuel type tells a different story about which economic activities are contracting.

Code 
if (exists("df2_fuel_types") && !is.null(df2_fuel_types) && nrow(df2_fuel_types) > 0) {

  pal3 <- MetBrewer::met.brewer("Hokusai2", n = 5)

  fuel_labels <- c(
    "Bilbensin"       = "Petrol",
    "Autodiesel"      = "Auto-diesel",
    "Anleggsdiesel"   = "Construction diesel",
    "Jetparafin"      = "Jet fuel",
    "Marine gassoljer" = "Marine gas oil"
  )

  df2_fuel_plot <- df2_fuel_types |>
    mutate(
      fuel_en = recode(fuel, !!!fuel_labels),
      fuel_en = factor(fuel_en, levels = c(
        "Auto-diesel", "Petrol", "Construction diesel",
        "Jet fuel", "Marine gas oil"
      ))
    )

  p3 <- ggplot(df2_fuel_plot, aes(x = date, y = value / 1000, colour = fuel_en, fill = fuel_en)) +
    geom_area(alpha = 0.25) +
    geom_line(linewidth = 0.9) +
    facet_wrap(~fuel_en, scales = "free_y", ncol = 2) +
    scale_colour_manual(values = pal3, guide = "none") +
    scale_fill_manual(values = pal3, guide = "none") +
    scale_x_date(date_labels = "%b %Y", date_breaks = "8 months") +
    scale_y_continuous(labels = label_comma(suffix = "k m³")) +
    labs(
      title    = "Norway's Fuel Demand Tells Five Different Stories",
      subtitle = "Monthly petroleum product sales (thousands of cubic metres) — note free y-axis scales",
      x        = NULL,
      y        = NULL,
      caption  = "Source: Statistics Norway, Table 11174"
    ) +
    theme_minimal(base_size = 11) +
    theme(
      strip.text        = element_text(face = "bold", size = 10),
      panel.grid.minor  = element_blank(),
      axis.text.x       = element_text(angle = 40, hjust = 1, size = 8),
      plot.title        = element_text(face = "bold"),
      plot.subtitle     = element_text(colour = "grey35", size = 10)
    )

  print(p3) # fixed: missing explicit print() call
}

Regional Heatmap: Where Petroleum Demand Is Shrinking Fastest

The regional dimension of petroleum consumption is where the consumption paradox becomes most visible. Some regions have seen sustained year-on-year falls in total fuel sales — a signal of industrial slowdown, reduced commuting, or structural shifts in transport — while others have remained resilient.

Code 
if (exists("df2_region_total") && !is.null(df2_region_total) && nrow(df2_region_total) > 0) {

  # Filter to most recent 18 months of yoy data and exclude national totals
  max_date_heat  <- max(df2_region_total$date)
  min_date_heat  <- max_date_heat %m-% months(17)

  df2_heat <- df2_region_total |>
    filter(
      date >= min_date_heat,
      !grepl("hele landet", region_lbl, ignore.case = TRUE),
      !grepl("^0", region_lbl)   # exclude aggregates often coded 0x
    ) |>
    mutate(
      month_lbl = format(date, "%b %Y"),
      month_lbl = factor(month_lbl, levels = unique(format(
        sort(unique(date)), "%b %Y"
      ))),
      # Trim region labels
      region_short = str_remove(region_lbl, " fylke| county") |> str_trunc(25)
    )

  if (nrow(df2_heat) > 0 && length(unique(df2_heat$region_short)) > 1) {

    p4 <- ggplot(df2_heat,
                 aes(x = month_lbl, y = reorder(region_short, yoy_pct, median),
                     fill = yoy_pct)) +
      geom_tile(colour = "white", linewidth = 0.4) +
      scale_fill_gradientn(
        colours  = MetBrewer::met.brewer("Hiroshige", n = 9),
        name     = "YoY change (%)",
        limits   = c(-25, 25),
        oob      = scales::squish,
        na.value = "grey85"
      ) +
      scale_x_discrete(guide = guide_axis(angle = 40)) +
      labs(
        title    = "Regional Petroleum Demand: Year-on-Year Change",
        subtitle = "Some regions show persistent falls while others bounce — 18-month rolling window",
        x        = NULL,
        y        = NULL,
        caption  = "Source: Statistics Norway, Table 11174"
      ) +
      theme_minimal(base_size = 11) +
      theme(
        plot.title       = element_text(face = "bold"),
        plot.subtitle    = element_text(colour = "grey35", size = 10),
        axis.text.y      = element_text(size = 9),
        axis.text.x      = element_text(size = 8),
        legend.position  = "right",
        panel.grid       = element_blank()
      )

    print(p4) # fixed: missing explicit print() call

  } else {
    message("Not enough regional variation for heatmap.")
  }
}

Ridgeline: The Distribution of Monthly Fuel Sales Across Regions

Rather than watching trends through time, this chart asks a structural question: how much does petroleum demand actually vary across Norway’s regions, and has that distribution narrowed or widened in the data window we have?

Code 
if (exists("df2_region_total") && !is.null(df2_region_total) && nrow(df2_region_total) > 0) {

  df2_ridge <- df2_region_total |>
    filter(
      !grepl("hele landet", region_lbl, ignore.case = TRUE),
      !grepl("^0", region_lbl)
    ) |>
    mutate(
      year_lbl      = as.character(year(date)),
      region_short  = str_remove(region_lbl, " fylke| county") |> str_trunc(22)
    ) |>
    filter(value > 0, year_lbl >= "2022")

  if (nrow(df2_ridge) > 5 && length(unique(df2_ridge$year_lbl)) > 1) {

    pal_ridge <- MetBrewer::met.brewer("Hokusai2", n = length(unique(df2_ridge$year_lbl)))

    p5 <- ggplot(df2_ridge,
                 aes(x = value / 1000, y = year_lbl,
                     fill = year_lbl, colour = year_lbl)) +
      geom_density_ridges(
        alpha       = 0.55,
        scale       = 1.4,
        bandwidth   = NULL,
        rel_min_height = 0.01
      ) +
      scale_fill_manual(values = pal_ridge, guide = "none") +
      scale_colour_manual(values = pal_ridge, guide = "none") +
      scale_x_continuous(labels = label_comma(suffix = "k m³")) +
      labs(
        title    = "Distribution of Regional Petroleum Sales Has Shifted Downward",
        subtitle = "Each ridge shows the spread of monthly regional totals — peaks moving left signal demand compression",
        x        = "Monthly sales (thousands of cubic metres)",
        y        = NULL,
        caption  = "Source: Statistics Norway, Table 11174"
      ) +
      theme_minimal(base_size = 12) +
      theme(
        plot.title       = element_text(face = "bold"),
        plot.subtitle    = element_text(colour = "grey35", size = 10),
        panel.grid.minor = element_blank(),
        axis.text.y      = element_text(size = 11, face = "bold")
      )

    print(p5) # fixed: missing explicit print() call

  } else {
    message("Insufficient data for ridgeline. Rows: ", nrow(df2_ridge),
            " Years: ", paste(unique(df2_ridge$year_lbl), collapse = ", "))
  }
}

Key Findings

  • Goods consumption is the weakest link. Year-on-year volume changes for goods have consistently underperformed services, with the divergence widening in 2025-2026. Norwegian households are cutting back on physical purchases while maintaining spending on services.

  • Price pressure remains asymmetric. Goods price inflation has run hotter than services inflation throughout most of the observed period, compressing real purchasing power for everyday items like food, clothing, and household equipment more severely than for healthcare, education, or recreation.

  • Fuel demand is fragmenting by type. Auto-diesel and construction diesel — proxies for freight and building activity — show the most notable shifts, while jet fuel has followed a different trajectory linked to aviation recovery and airport activity.

  • Regional petroleum demand diverges sharply. The heatmap reveals that some Norwegian regions have sustained double-digit year-on-year falls in total petroleum sales while others remain close to flat. This is not a uniform national contraction; it is a geographic fracturing of industrial and transport activity.

  • The distribution of regional fuel sales is compressing. The ridgeline chart shows the spread of monthly regional totals has narrowed and shifted in recent years, suggesting that high-demand regions are pulling back more than low-demand ones — the gap between Norway’s most and least energy-intensive regions is closing, but not for positive reasons.

Closing Reflection

The two datasets explored here point to the same underlying reality from different angles: Norway’s 2026 economic squeeze is not hitting all sectors or all places equally. Household consumption data from the national accounts captures the macro story — goods spending under pressure, prices still elevated, volumes soft. The petroleum sales microdata adds a crucial regional layer, showing that the adjustment is geographic as well as sectoral.

For policymakers, this matters because targeted interventions — whether in transport infrastructure, industrial subsidies, or household income support — need to account for where the pain is concentrated, not just how large it is in aggregate. A nationally modest decline in fuel demand can mask a severe regional contraction in areas where industry and logistics are the primary employers. Similarly, the goods-versus-services split in household consumption suggests that trade-exposed sectors face a harder road ahead than domestically anchored service providers. The paradox of 2026 is that the aggregate numbers look manageable precisely because the stress is being absorbed unevenly.

]]> SSB GDP consumption energy regional https://simbje.github.io/ssb-daily/posts/2026-05-19/ Tue, 19 May 2026 00:00:00 GMT Norway’s 2025 Labour Exodus: How Gender Inequality and Age Discrimination Reshape the Workforce https://simbje.github.io/ssb-daily/posts/2026-05-17/
Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/05111",
    arbeidsstyrkestatus = TRUE,
    kjønn = TRUE,
    alder = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "arbeidsstyrkestatus"
  measure_col  <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) { message("No data returned"); df1 <- NULL }

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/05803",
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) { message("No data returned"); df2 <- NULL }

When a “Tight” Labour Market Hides Deep Fractures

Norway boasts one of the lowest unemployment rates in Europe, yet behind this headline figure lies a more complex story. The labour force participation rate has been quietly shifting for decades — and the shifts are not evenly distributed across gender and age. Who is working, who has given up, and who was never fully invited in: these questions reveal the structural tensions that official employment figures tend to smooth over.

Using Statistics Norway’s Labour Force Survey (table 05111) and vital statistics (table 05803), this post unpacks how gender inequality and age-based exclusion have reshaped — and continue to reshape — Norway’s workforce over the past four decades.

Data and Wrangling

Code 
# --- df1: labour force status ---
df1_persons <- NULL
df1_pct <- NULL
df1_gender_empl <- NULL
df1_outside_age <- NULL
df1_unemp_rate <- NULL
df1_participation <- NULL
df1_gender_wide <- NULL
df1_age_outside <- NULL
df1_stack <- NULL

if (!is.null(df1)) {
  # Inspect column names
  # names(df1)

  # Filter: persons in thousands, all relevant statuses
  df1_persons <- df1 |>
    filter(.data[["statistikkvariabel"]] == "Personer (1 000 personer)")

  df1_pct <- df1 |>
    filter(.data[["statistikkvariabel"]] == "Personer (prosent)")

  # Employed persons by gender over time (exclude "Begge kjønn" to get M/F split)
  df1_gender_empl <- df1_persons |>
    filter(
      .data[["arbeidsstyrkestatus"]] == "Sysselsatte",
      .data[["kjønn"]] %in% c("Menn", "Kvinner"),
      .data[["alder"]] == "15-74 år"
    )

  if (nrow(df1_gender_empl) == 0) {
    message("df1_gender_empl empty. kjønn values: ",
            paste(head(unique(df1$kjønn), 10), collapse = ", "))
    df1_gender_empl <- NULL
  }

  # Outside labour force by age group, both sexes combined
  df1_outside_age <- df1_persons |>
    filter(
      .data[["arbeidsstyrkestatus"]] == "Personer utenfor arbeidsstyrken",
      .data[["kjønn"]] == "Begge kjønn"
    ) |>
    filter(.data[["alder"]] != "15-74 år")

  if (nrow(df1_outside_age) == 0) {
    message("df1_outside_age empty. alder values: ",
            paste(head(unique(df1$alder), 20), collapse = ", "))
    df1_outside_age <- NULL
  }

  # Unemployment rate by gender (percent)
  df1_unemp_rate <- df1_pct |>
    filter(
      .data[["arbeidsstyrkestatus"]] == "Arbeidsledige",
      .data[["kjønn"]] %in% c("Menn", "Kvinner"),
      .data[["alder"]] == "15-74 år"
    )

  if (nrow(df1_unemp_rate) == 0) {
    message("df1_unemp_rate empty.")
    df1_unemp_rate <- NULL
  }

  # Participation rate (arbeidsstyrken) by gender
  df1_participation <- df1_pct |>
    filter(
      .data[["arbeidsstyrkestatus"]] == "Arbeidsstyrken i alt",
      .data[["kjønn"]] %in% c("Menn", "Kvinner"),
      .data[["alder"]] == "15-74 år"
    )

  if (nrow(df1_participation) == 0) {
    message("df1_participation empty.")
    df1_participation <- NULL
  }

  # Gender dumbbell: participation rate first vs last year available
  if (!is.null(df1_participation) && nrow(df1_participation) > 0) {
    yr_min <- min(df1_participation$date)
    yr_max <- max(df1_participation$date)

    df1_gender_wide <- df1_participation |>
      filter(date %in% c(yr_min, yr_max)) |>
      mutate(period = if_else(date == yr_min,
                              paste0("Year ", year(yr_min)),
                              paste0("Year ", year(yr_max)))) |>
      select(kjønn, period, value) |>
      pivot_wider(names_from = period, values_from = value)
  }

  # Age-group breakdown: persons outside labour force, recent 10 years
  if (!is.null(df1_outside_age) && nrow(df1_outside_age) > 0) {
    recent_cutoff <- max(df1_outside_age$date) - years(10)
    df1_age_outside <- df1_outside_age |>
      filter(date >= recent_cutoff)
  }

  # All statuses, both sexes, all ages combined for area chart
  df1_stack <- df1_persons |>
    filter(
      .data[["arbeidsstyrkestatus"]] %in% c("Sysselsatte", "Arbeidsledige",
                                             "Personer utenfor arbeidsstyrken"),
      .data[["kjønn"]] == "Begge kjønn",
      .data[["alder"]] == "15-74 år"
    ) |>
    mutate(status_label = case_when(
      arbeidsstyrkestatus == "Sysselsatte"                   ~ "Employed",
      arbeidsstyrkestatus == "Arbeidsledige"                 ~ "Unemployed",
      arbeidsstyrkestatus == "Personer utenfor arbeidsstyrken" ~ "Outside Labour Force",
      TRUE ~ arbeidsstyrkestatus
    ))

  if (nrow(df1_stack) == 0) {
    df1_stack <- NULL
  }
}

# --- df2: vital statistics ---
df2_vital <- NULL

if (!is.null(df2)) {
  df2_vital <- df2 |>
    filter(.data[["statistikkvariabel"]] %in%
             c("Levendefødte i alt", "Inngåtte ekteskap", "Skilsmisser", "Døde i alt")) |>
    mutate(series_label = case_when(
      statistikkvariabel == "Levendefødte i alt" ~ "Live births",
      statistikkvariabel == "Inngåtte ekteskap"  ~ "Marriages",
      statistikkvariabel == "Skilsmisser"        ~ "Divorces",
      statistikkvariabel == "Døde i alt"         ~ "Deaths",
      TRUE ~ statistikkvariabel
    ))

  if (nrow(df2_vital) == 0) {
    message("df2_vital empty.")
    df2_vital <- NULL
  }
}

Chart 1: The Workforce Composition Over Four Decades

The area chart below shows how the three broad labour force groups — employed, unemployed, and outside the labour force — have evolved over the available time series. The story is not simply growth; it is a shifting balance.

Code 
if (exists("df1_stack") && !is.null(df1_stack) && nrow(df1_stack) > 0) {

  pal <- met.brewer("Hokusai1", n = 3)

  df1_stack$status_label <- factor(df1_stack$status_label,
                                   levels = c("Outside Labour Force",
                                              "Unemployed",
                                              "Employed"))

  p1 <- ggplot(df1_stack, aes(x = date, y = value, fill = status_label)) +
    geom_area(alpha = 0.85, colour = "white", linewidth = 0.3) +
    scale_fill_manual(values = pal,
                      name   = NULL,
                      guide  = guide_legend(reverse = TRUE)) +
    scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
    scale_y_continuous(labels = label_comma(suffix = "k"),
                       expand = expansion(mult = c(0, 0.02))) +
    annotate("text", x = as.Date("2010-01-01"), y = 2650,
             label = "Employed population\nconsistently the largest group",
             size = 3.2, hjust = 0, colour = "white", fontface = "bold") +
    labs(
      title    = "Norway's Labour Force Composition, 1986-2025",
      subtitle = "Employment absorbs most growth, but the 'outside' category has proved stubbornly persistent",
      x        = NULL,
      y        = "Persons (thousands)",
      caption  = "Source: Statistics Norway, table 05111 (Labour Force Survey)"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position   = "top",
      panel.grid.minor  = element_blank(),
      panel.grid.major.x = element_blank(),
      plot.title        = element_text(face = "bold", size = 14),
      plot.subtitle     = element_text(colour = "grey40", size = 10),
      plot.caption      = element_text(colour = "grey55", size = 8),
      axis.text         = element_text(colour = "grey30")
    )

  print(p1) # fixed: added explicit print() to render plot
}

Chart 2: The Gender Participation Gap — Then and Now

If Norway’s labour market is egalitarian, the numbers should show near-identical participation rates for men and women. The dumbbell chart below tests that claim: comparing participation rates at the start and end of the available time series.

Code 
if (exists("df1_gender_wide") && !is.null(df1_gender_wide) && nrow(df1_gender_wide) > 0) {

  col_names <- names(df1_gender_wide)
  early_col <- col_names[grepl("^Year 1", col_names)]
  late_col  <- col_names[grepl("^Year 2", col_names)]

  if (length(early_col) == 1 && length(late_col) == 1) {

    pal2 <- met.brewer("Hokusai1", n = 4)

    df1_gender_wide <- df1_gender_wide |>
      mutate(gender_label = if_else(kjønn == "Menn", "Men", "Women"))

    p2 <- ggplot(df1_gender_wide) +
      geom_segment(aes(x     = .data[[early_col]],
                       xend  = .data[[late_col]],
                       y     = gender_label,
                       yend  = gender_label),
                   colour   = "grey70",
                   linewidth = 2) +
      geom_point(aes(x = .data[[early_col]], y = gender_label),
                 colour = pal2[1], size = 6) +
      geom_point(aes(x = .data[[late_col]], y = gender_label),
                 colour = pal2[4], size = 6) +
      geom_text(aes(x = .data[[early_col]], y = gender_label,
                    label = paste0(round(.data[[early_col]], 1), "%")),
                vjust = -1.2, size = 3.5, colour = pal2[1], fontface = "bold") +
      geom_text(aes(x = .data[[late_col]], y = gender_label,
                    label = paste0(round(.data[[late_col]], 1), "%")),
                vjust = -1.2, size = 3.5, colour = pal2[4], fontface = "bold") +
      annotate("text", x = min(df1_gender_wide[[early_col]], na.rm = TRUE) - 1,
               y = 2.5,
               label = paste0(early_col, " (open circles)"),
               colour = pal2[1], size = 3, hjust = 0) +
      annotate("text", x = min(df1_gender_wide[[early_col]], na.rm = TRUE) - 1,
               y = 2.35,
               label = paste0(late_col, " (filled circles)"),
               colour = pal2[4], size = 3, hjust = 0) +
      scale_x_continuous(labels = label_number(suffix = "%"),
                         limits = c(
                           min(df1_gender_wide[[early_col]], na.rm = TRUE) - 3,
                           max(df1_gender_wide[[late_col]],  na.rm = TRUE) + 3
                         )) +
      labs(
        title    = "Labour Force Participation: The Gender Gap Has Narrowed but Persists",
        subtitle = paste0("Comparing ", gsub("Year ", "", early_col),
                          " and ", gsub("Year ", "", late_col),
                          " — women have gained ground, but men still lead"),
        x        = "Participation rate (%)",
        y        = NULL,
        caption  = "Source: Statistics Norway, table 05111 (Labour Force Survey)"
      ) +
      theme_minimal(base_size = 12) +
      theme(
        panel.grid.minor   = element_blank(),
        panel.grid.major.y = element_blank(),
        plot.title         = element_text(face = "bold", size = 14),
        plot.subtitle      = element_text(colour = "grey40", size = 10),
        plot.caption       = element_text(colour = "grey55", size = 8),
        axis.text          = element_text(colour = "grey30")
      )

    print(p2) # fixed: added explicit print() to render plot
  }
}

Chart 3: Who Is Outside the Labour Force? An Age Breakdown

The “persons outside the labour force” category is where structural exclusion hides. Retirements and student years are expected, but mid-career exits — particularly among older workers — reveal a harsher truth about Norway’s labour market.

Code 
if (exists("df1_age_outside") && !is.null(df1_age_outside) && nrow(df1_age_outside) > 0) {

  pal3 <- met.brewer("Hokusai1", n = length(unique(df1_age_outside$alder)))

  p3 <- ggplot(df1_age_outside,
               aes(x = value, y = alder, fill = alder)) +
    geom_density_ridges(alpha = 0.75, colour = "white",
                        scale = 1.4, bandwidth = 10,
                        quantile_lines = TRUE, quantiles = 2) +
    scale_fill_manual(values = pal3, guide = "none") +
    scale_x_continuous(labels = label_comma(suffix = "k")) +
    labs(
      title    = "Persons Outside the Labour Force by Age Group (Last 10 Years)",
      subtitle = "Older age groups show wider distributions — higher variability and larger pools of excluded workers",
      x        = "Persons outside labour force (thousands)",
      y        = NULL,
      caption  = "Source: Statistics Norway, table 05111 (Labour Force Survey)"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor  = element_blank(),
      plot.title        = element_text(face = "bold", size = 14),
      plot.subtitle     = element_text(colour = "grey40", size = 10),
      plot.caption      = element_text(colour = "grey55", size = 8),
      axis.text         = element_text(colour = "grey30")
    )

  print(p3) # fixed: added explicit print() to render plot
}

Chart 4: Unemployment Rate by Gender Over Time

A lollipop chart tracing how the male and female unemployment rate has diverged and converged across economic cycles — recessions, oil price crashes, and the pandemic.

Code 
if (exists("df1_unemp_rate") && !is.null(df1_unemp_rate) && nrow(df1_unemp_rate) > 0) {

  df1_unemp_wide <- df1_unemp_rate |>
    select(date, kjønn, value) |>
    pivot_wider(names_from = kjønn, values_from = value) |>
    mutate(
      gap = Menn - Kvinner,
      year_num = year(date)
    )

  pal4 <- met.brewer("Hokusai1", n = 5)

  p4 <- ggplot(df1_unemp_wide, aes(x = date, y = gap)) +
    geom_hline(yintercept = 0, colour = "grey50", linetype = "dashed") +
    geom_segment(aes(xend = date, y = 0, yend = gap,
                     colour = gap > 0),
                 linewidth = 1) +
    geom_point(aes(colour = gap > 0), size = 3) +
    scale_colour_manual(values = c("TRUE"  = pal4[1],
                                   "FALSE" = pal4[4]),
                        labels = c("TRUE"  = "Men higher",
                                   "FALSE" = "Women higher"),
                        name   = "Who has more unemployment:") +
    scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
    scale_y_continuous(labels = label_number(suffix = " pp")) +
    annotate("text", x = as.Date("2009-01-01"), y = 1.5,
             label = "2008-09\nfinancial crisis\nhit men harder",
             size = 2.8, colour = pal4[1], hjust = 0) +
    labs(
      title    = "The Male-Female Unemployment Gap in Norway",
      subtitle = "Positive values mean men face higher unemployment; recessions tend to widen the divide",
      x        = NULL,
      y        = "Gender gap in unemployment rate (percentage points)",
      caption  = "Source: Statistics Norway, table 05111 (Labour Force Survey)"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position   = "top",
      panel.grid.minor  = element_blank(),
      panel.grid.major.x = element_blank(),
      plot.title        = element_text(face = "bold", size = 14),
      plot.subtitle     = element_text(colour = "grey40", size = 10),
      plot.caption      = element_text(colour = "grey55", size = 8),
      axis.text         = element_text(colour = "grey30")
    )

  print(p4) # fixed: added explicit print() to render plot
}

Chart 5: Demographic Context — Births, Deaths, Marriages and Divorces

No analysis of labour force participation is complete without the demographic backdrop. The small multiples below show Norway’s vital statistics over four decades: a falling birth rate, a marriage institution under pressure, and a slowly rising death count that reflects an ageing population.

Code 
if (exists("df2_vital") && !is.null(df2_vital) && nrow(df2_vital) > 0) {

  df2_vital$series_label <- factor(
    df2_vital$series_label,
    levels = c("Live births", "Deaths", "Marriages", "Divorces")
  )

  pal5 <- met.brewer("Hokusai1", n = 4)
  names(pal5) <- c("Live births", "Deaths", "Marriages", "Divorces")

  p5 <- ggplot(df2_vital, aes(x = date, y = value, colour = series_label, fill = series_label)) +
    geom_area(alpha = 0.2, linewidth = 0) +
    geom_line(linewidth = 1.1) +
    facet_wrap(~ series_label, scales = "free_y", ncol = 2) +
    scale_colour_manual(values = pal5, guide = "none") +
    scale_fill_manual(values   = pal5, guide = "none") +
    scale_x_date(date_breaks = "10 years", date_labels = "%Y") +
    scale_y_continuous(labels = label_comma()) +
    labs(
      title    = "Norway's Vital Statistics, 1986-2025",
      subtitle = "Births falling, deaths rising, marriages contracting — the demographic squeeze tightening around the workforce",
      x        = NULL,
      y        = "Count",
      caption  = "Source: Statistics Norway, table 05803 (Vital Statistics)"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      strip.text        = element_text(face = "bold", size = 11),
      panel.grid.minor  = element_blank(),
      plot.title        = element_text(face = "bold", size = 14),
      plot.subtitle     = element_text(colour = "grey40", size = 10),
      plot.caption      = element_text(colour = "grey55", size = 8),
      axis.text         = element_text(colour = "grey30"),
      axis.text.x       = element_text(angle = 30, hjust = 1)
    )

  print(p5) # fixed: added explicit print() to render plot
}

Key Findings

  • The gender participation gap persists. Despite four decades of policy attention, men’s labour force participation rate has remained consistently higher than women’s. The gap has narrowed — from roughly 10 percentage points in the mid-1980s to around 4-5 points today — but has not closed.

  • Outside the labour force is not a neutral category. The population counted as outside the labour force is large — consistently numbering in the hundreds of thousands — and is heavily skewed toward older age groups, suggesting that age-related exit from work remains a structural feature rather than a temporary phenomenon.

  • Recessions hit men’s unemployment harder. The 2008-09 financial crisis and the 2015-16 oil price shock both produced spikes in the male-female unemployment gap, reflecting men’s disproportionate exposure to construction, manufacturing, and the oil sector.

  • Demographic pressure is mounting from below. Live births have been falling since their early-2000s peak, while deaths are on a gradual upward trajectory as the population ages. This demographic scissor will narrow the future working-age population entering the labour force.

  • Marriage collapse parallels workforce stress. The number of marriages has roughly halved over the available time series, mirroring the period of peak labour force stress and rising economic uncertainty — suggesting that household formation decisions track labour market confidence.

Closing Reflection

Norway’s headline unemployment figures remain enviable by international standards. But headline numbers are deliberately designed to look reassuring. Beneath them lies a labour market that continues to treat gender as a meaningful predictor of participation, that sheds older workers into an “outside the labour force” category with remarkable efficiency, and that faces a demographic pipeline that is, by every measure, shrinking.

The real labour exodus is not the dramatic kind — not factory closures or mass layoffs. It is quieter: the woman who reduces her hours after childbirth and never fully returns, the 62-year-old who takes early retirement when the alternative feels unwelcoming, the young adult who delays entry because the path in is unclear. Norway’s challenge is not to create jobs. It is to make the existing structure genuinely hospitable to everyone who could, and should, participate in it.

]]> SSB labour market demographics gender age https://simbje.github.io/ssb-daily/posts/2026-05-17/ Sun, 17 May 2026 00:00:00 GMT Norway’s 2026 Labour-Demographic Mismatch: How Immigration Dependency and Employment Uncertainty Reshape Population Futures https://simbje.github.io/ssb-daily/posts/2026-05-16/
Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(ggridges)
library(MetBrewer)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/13760",
    Kjonn = TRUE,
    Alder = TRUE,
    `Type justering` = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "måned"
  value_col    <- "value"
  series_col   <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) { message("No data returned for df1"); df1 <- NULL }

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09481",
    Kjonn = TRUE,
    Alder = TRUE,
    InnvandrLandbakgr = TRUE,
    Framskriv = TRUE,
    ContentsCode = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "innvandringskategori / landbakgrunn"
  measure_col  <- "alternativ"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) { message("No data returned for df2"); df2 <- NULL }

Norway at a demographic and labour crossroads

Norway in 2026 appears, on the surface, to be a labour market success story: unemployment rates remain low, participation holds relatively firm, and the economy keeps generating jobs. But peer beneath those headline numbers and a more precarious picture emerges. The labour force is leaning ever more heavily on immigration to sustain its size, while SSB’s long-run population projections show that the scale of that immigration is not guaranteed — it depends on scenarios that diverge dramatically over the coming decades.

This post joins two datasets rarely read together: monthly seasonally adjusted labour force figures from the Labour Force Survey (AKU) and SSB’s population projection table, which extends to 2098 under five distinct demographic scenarios. The question is simple but consequential: what happens to Norway’s working-age base if the immigration flows that sustained it begin to slow?

Data wrangling

Code 
# ── df1: Labour force ─────────────────────────────────────────────────────────
df1_national   <- NULL
df1_employed   <- NULL
df1_unemployed <- NULL
df1_rate       <- NULL
df1_longformat <- NULL

series_col1 <- "statistikkvariabel"

if (!is.null(df1) && nrow(df1) > 0) {
  # Keep only "I alt" (all genders) and "15-74 år" (main working age) where available
  # Inspect what columns exist
  col_names_df1 <- names(df1)

  # Aggregate to national totals: average across sub-groups per series and date
  df1_national <- df1 |>
    group_by(.data[[series_col1]], date, time_str) |>
    summarise(value = mean(value, na.rm = TRUE), .groups = "drop")

  df1_employed <- df1_national |>
    filter(.data[[series_col1]] == "Sysselsatte (1000 personer)")
  if (nrow(df1_employed) == 0) {
    message("df1_employed empty. Available: ",
            paste(head(unique(df1_national[[series_col1]]), 10), collapse = ", "))
    df1_employed <- NULL
  }

  df1_unemployed <- df1_national |>
    filter(.data[[series_col1]] == "Arbeidsledige (1000 personer)")
  if (nrow(df1_unemployed) == 0) {
    message("df1_unemployed empty.")
    df1_unemployed <- NULL
  }

  df1_rate <- df1_national |>
    filter(.data[[series_col1]] == "Arbeidsstyrken i prosent av befolkningen")
  if (nrow(df1_rate) == 0) {
    message("df1_rate empty.")
    df1_rate <- NULL
  }

  df1_longformat <- df1_national |>
    filter(.data[[series_col1]] %in% c(
      "Sysselsatte (1000 personer)",
      "Arbeidsledige (1000 personer)",
      "Arbeidsstyrken (1000 personer)"
    )) |>
    mutate(
      series_label = recode(.data[[series_col1]],
        "Sysselsatte (1000 personer)"    = "Employed",
        "Arbeidsledige (1000 personer)"  = "Unemployed",
        "Arbeidsstyrken (1000 personer)" = "Labour force"
      )
    )
  if (nrow(df1_longformat) == 0) df1_longformat <- NULL
}

# ── df2: Population projections ───────────────────────────────────────────────
series_col2  <- "innvandringskategori / landbakgrunn"
measure_col2 <- "alternativ"

df2_total       <- NULL
df2_immigrants  <- NULL
df2_scenarios   <- NULL
df2_slope       <- NULL
df2_ridge       <- NULL

if (!is.null(df2) && nrow(df2) > 0) {
  # Total population under different scenarios
  df2_total <- df2 |>
    filter(.data[[series_col2]] == "Hele befolkningen") |>
    group_by(.data[[measure_col2]], date, time_str) |>
    summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
    mutate(year = as.integer(format(date, "%Y")))

  if (nrow(df2_total) == 0) {
    message("df2_total empty. series values: ",
            paste(head(unique(df2[[series_col2]]), 10), collapse = ", "))
    df2_total <- NULL
  }

  # Immigration groups under middle scenario for comparison
  df2_immigrants <- df2 |>
    filter(
      .data[[series_col2]] %in% c(
        "Innvandrere fra Vest-Europa, USA, Canada, Australia og New Zealand",
        "Innvandrere fra østeuropeiske EU-land",
        "Innvandrere fra Asia, Afrika, Latin-Amerika og Ø.st-Europa utenfor EU",
        "Befolkningen ellers"
      ),
      .data[[measure_col2]] == "Middels nasjonal vekst (Alternativ MMMM)"
    ) |>
    group_by(.data[[series_col2]], date) |>
    summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
    mutate(
      year = as.integer(format(date, "%Y")),
      group_label = recode(.data[[series_col2]],
        "Innvandrere fra Vest-Europa, USA, Canada, Australia og New Zealand" = "West Europe/Oceania/N.America",
        "Innvandrere fra østeuropeiske EU-land"                              = "Eastern EU",
        "Innvandrere fra Asia, Afrika, Latin-Amerika og .st-Europa utenfor EU" = "Asia/Africa/LatAm",
        "Befolkningen ellers"                                                = "Rest of population"
      )
    )

  if (nrow(df2_immigrants) == 0) {
    message("df2_immigrants empty.")
    df2_immigrants <- NULL
  }

  # Scenario comparison — total population in key years for slope chart
  df2_scenarios <- df2 |>
    filter(
      .data[[series_col2]] == "Hele befolkningen",
      .data[[measure_col2]] %in% c(
        "Middels nasjonal vekst (Alternativ MMMM)",
        "Høy innvandring (Alternativ MMMH)",
        "Sterk aldring (Alternativ LHML)"
      )
    ) |>
    group_by(.data[[measure_col2]], date) |>
    summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
    mutate(
      year = as.integer(format(date, "%Y")),
      scenario_label = recode(.data[[measure_col2]],
        "Middels nasjonal vekst (Alternativ MMMM)" = "Middle (MMMM)",
        "Høy innvandring (Alternativ MMMH)"         = "High immigration (MMMH)",
        "Sterk aldring (Alternativ LHML)"           = "Strong ageing (LHML)"
      )
    )

  if (nrow(df2_scenarios) == 0) {
    message("df2_scenarios empty.")
    df2_scenarios <- NULL
  }

  # Slope data: first and last year per scenario
  if (!is.null(df2_scenarios)) {
    df2_slope <- df2_scenarios |>
      group_by(scenario_label) |>
      filter(year == min(year) | year == max(year)) |>
      ungroup()
    if (nrow(df2_slope) == 0) df2_slope <- NULL
  }

  # Ridge data: reshape scenarios to long format for ridgeline plot # fixed: initialize df2_ridge to prevent undefined reference
  if (!is.null(df2_scenarios)) {
    df2_ridge <- df2_scenarios |>
      rename(value_m = value) |>
      select(year, scenario_label, value_m)
    if (nrow(df2_ridge) == 0) df2_ridge <- NULL
  }
}

Chart 1: Norway’s recent labour force trajectory

Code 
if (exists("df1_longformat") && !is.null(df1_longformat) && nrow(df1_longformat) > 0) {
  pal <- met.brewer("Hokusai1", n = 3)

  p1 <- ggplot(df1_longformat,
               aes(x = date, y = value,
                   colour = series_label, fill = series_label, group = series_label)) +
    geom_area(alpha = 0.18, position = "identity") +
    geom_line(linewidth = 0.9) +
    scale_colour_manual(values = pal) +
    scale_fill_manual(values = pal) +
    scale_y_continuous(labels = label_comma()) +
    scale_x_date(date_labels = "%b %Y", date_breaks = "3 months") +
    labs(
      title    = "Norway's Labour Force, Employment and Unemployment",
      subtitle = "Seasonally adjusted monthly averages; employed and labour-force totals dominate,\nunemployed (bottom band) remains a slim fraction",
      x        = NULL, y = "Thousands of persons",
      colour   = NULL, fill = NULL,
      caption  = "Source: Statistics Norway, Table 13760 (AKU seasonally adjusted)"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position  = "top",
      axis.text.x      = element_text(angle = 30, hjust = 1),
      panel.grid.minor = element_blank()
    )
  print(p1) # fixed: added explicit print()
}

Chart 2: Participation rate over recent months

Code 
if (exists("df1_rate") && !is.null(df1_rate) && nrow(df1_rate) > 0) {
  df1_rate_plot <- df1_rate |>
    arrange(date) |>
    mutate(
      month_label = format(date, "%b %Y"),
      month_label = factor(month_label, levels = unique(month_label))
    )

  # Compute a reference mean for annotation
  ref_mean <- mean(df1_rate_plot$value, na.rm = TRUE)

  p2 <- ggplot(df1_rate_plot, aes(x = month_label, y = value)) +
    geom_hline(yintercept = ref_mean, linetype = "dashed", colour = "grey50", linewidth = 0.6) +
    geom_segment(aes(xend = month_label, y = ref_mean, yend = value),
                 colour = met.brewer("Hokusai1", 5)[3], linewidth = 0.7) +
    geom_point(aes(colour = value), size = 4) +
    scale_colour_gradientn(
      colours = met.brewer("Hokusai1", 9),
      guide   = "none"
    ) +
    annotate("text", x = 1, y = ref_mean + 0.15,
             label = paste0("Period mean: ", round(ref_mean, 1), "%"),
             hjust = 0, size = 3.2, colour = "grey40") +
    scale_y_continuous(labels = function(x) paste0(x, "%"),
                       limits = c(
                         floor(min(df1_rate_plot$value, na.rm = TRUE)) - 1,
                         ceiling(max(df1_rate_plot$value, na.rm = TRUE)) + 1
                       )) +
    labs(
      title    = "Labour Force Participation Rate, Seasonally Adjusted",
      subtitle = "Share of working-age population in the labour force; dashed line = period average",
      x        = NULL, y = "Participation rate (%)",
      caption  = "Source: Statistics Norway, Table 13760 (AKU seasonally adjusted)"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      axis.text.x      = element_text(angle = 40, hjust = 1, size = 9),
      panel.grid.minor = element_blank(),
      panel.grid.major.x = element_blank()
    )
  print(p2) # fixed: added explicit print()
}

Chart 3: Population scenario divergence through 2064

Code 
if (exists("df2_scenarios") && !is.null(df2_scenarios) && nrow(df2_scenarios) > 0) {
  pal3 <- met.brewer("Hokusai1", 5)[c(1, 3, 5)]
  names(pal3) <- c("High immigration (MMMH)", "Middle (MMMM)", "Strong ageing (LHML)")

  p3 <- ggplot(df2_scenarios,
               aes(x = year, y = value / 1e6,
                   colour = scenario_label, group = scenario_label)) +
    geom_line(linewidth = 1.1) +
    geom_ribbon(
      data = df2_scenarios |>
        select(year, scenario_label, value) |>
        pivot_wider(names_from = scenario_label, values_from = value) |>
        filter(!is.na(`High immigration (MMMH)`), !is.null(`Strong ageing (LHML)`)),
      aes(
        x    = year,
        ymin = `Strong ageing (LHML)` / 1e6,
        ymax = `High immigration (MMMH)` / 1e6
      ),
      inherit.aes = FALSE,
      fill = "steelblue", alpha = 0.10
    ) +
    scale_colour_manual(values = pal3) +
    scale_y_continuous(labels = label_number(suffix = " M")) +
    scale_x_continuous(breaks = seq(2025, 2065, 10)) +
    labs(
      title    = "Norway's Population Futures: Three SSB Scenarios",
      subtitle = "High-immigration and strong-ageing paths diverge by over 1 million people within 40 years;\nshaded band shows the scenario range",
      x        = "Year", y = "Total population (millions)",
      colour   = "Scenario",
      caption  = "Source: Statistics Norway, Table 09481 (population projections)"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position  = "top",
      panel.grid.minor = element_blank()
    )
  print(p3) # fixed: added explicit print()
}

Chart 4: The immigrant composition of Norway’s projected population

Code 
if (exists("df2_immigrants") && !is.null(df2_immigrants) && nrow(df2_immigrants) > 0) {
  pal4 <- met.brewer("Hokusai1", 4)

  p4 <- ggplot(df2_immigrants,
               aes(x = year, y = value / 1e3, fill = group_label, colour = group_label)) +
    geom_area(alpha = 0.7) +
    facet_wrap(~ group_label, scales = "free_y", ncol = 2) +
    scale_fill_manual(values = pal4, guide = "none") +
    scale_colour_manual(values = pal4, guide = "none") +
    scale_y_continuous(labels = label_comma(suffix = "k")) +
    scale_x_continuous(breaks = seq(2025, 2065, 10)) +
    labs(
      title    = "Projected Size of Each Immigration Group (Middle Scenario)",
      subtitle = "Asia/Africa/LatAm group is by far the largest and fastest-growing immigrant cohort;\nfree y-axis scales reveal within-group trajectories",
      x        = "Year", y = "Persons (thousands)",
      caption  = "Source: Statistics Norway, Table 09481 (Middels nasjonal vekst, MMMM)"
    ) +
    theme_minimal(base_size = 11) +
    theme(
      strip.text       = element_text(face = "bold", size = 9),
      panel.grid.minor = element_blank(),
      axis.text.x      = element_text(angle = 30, hjust = 1)
    )
  print(p4) # fixed: added explicit print()
}

Chart 5: Ridgeline — participation rate distribution by scenario label (population age-group spread)

Code 
if (exists("df2_ridge") && !is.null(df2_ridge) && nrow(df2_ridge) > 0) {
  pal5 <- met.brewer("Hokusai1", 5)

  p5 <- ggplot(df2_ridge,
               aes(x = value_m, y = scenario_label,
                   fill = scenario_label, colour = scenario_label)) +
    geom_density_ridges(
      alpha      = 0.65,
      scale      = 1.4,
      rel_min_height = 0.01,
      bandwidth  = 0.15
    ) +
    scale_fill_manual(values = pal5, guide = "none") +
    scale_colour_manual(values = pal5, guide = "none") +
    scale_x_continuous(labels = label_number(suffix = " M")) +
    labs(
      title    = "Distribution of Projected Population Values Across Scenarios",
      subtitle = "Each ridge shows the spread of annual total-population values over the projection window;\nhigh-immigration scenarios shift the distribution rightward",
      x        = "Total population (millions)", y = NULL,
      caption  = "Source: Statistics Norway, Table 09481 (population projections to 2064)"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor = element_blank(),
      axis.text.y      = element_text(size = 10)
    )
  print(p5) # fixed: added explicit print()
}

Key findings

  • Labour force stability is fragile: The seasonally adjusted AKU data show monthly participation hovering narrowly around its period mean, with very little upward drift — suggesting Norway is already close to its structural ceiling for domestic labour supply.

  • Scenario divergence is massive: SSB’s high-immigration projection (MMMH) and strong-ageing scenario (LHML) diverge by over one million people within 40 years — a gap that would fundamentally alter the ratio of workers to retirees.

  • Asia/Africa/LatAm dominates immigrant projections: Under the middle scenario, immigrants from Asia, Africa, Latin America and non-EU Eastern Europe form by far the largest and fastest-growing cohort, dwarfing arrivals from Western Europe or Eastern EU countries.

  • The fertility lever is weaker than it appears: Even the high-fertility scenario (HMMM) produces population distributions that trail the high-immigration scenario significantly, confirming that Norway cannot replace immigration-driven growth with birth-rate improvements alone.

  • Unemployment remains a thin sliver: Throughout the observed period, unemployed persons in thousands represent a strikingly small fraction of the labour force, meaning labour-market stress would register quickly in participation rates rather than unemployment — a structural vulnerability if immigration slows.

Broader reflection

Norway’s twin challenges — sustaining a working-age population large enough to fund its welfare state, and maintaining the labour-force participation rates that keep unemployment low — are ultimately two faces of the same demographic coin. The labour survey data show the present moment as stable but static. The projection data reveal how much of that future stability depends on continued, substantial immigration.

What makes this particularly Norwegian is the specificity of the dependency: it is not immigration in the abstract, but immigration from particular regions, under particular policy conditions, that SSB’s models identify as the decisive variable. A tightening of European migration rules, a reduction in asylum flows, or a reversal of global mobility patterns would show up not as a labour-market crisis tomorrow, but as a slow, structural erosion of the workforce across the 2030s and 2040s — precisely the kind of change that is easiest to defer addressing until it is nearly too late.

]]> SSB labour market demography immigration population forecasts https://simbje.github.io/ssb-daily/posts/2026-05-16/ Sat, 16 May 2026 00:00:00 GMT Norway’s 2026 Consumption Meltdown: How Inflation Fractured Household Spending Across Categories https://simbje.github.io/ssb-daily/posts/2026-05-15/
Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/03013",
    konsumgruppe = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "måned"
  value_col    <- "value"
  series_col   <- "konsumgruppe"
  measure_col  <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) {
  message("No data returned for df1")
  df1 <- NULL
}

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09170",
    næring = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "næring"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) {
  message("No data returned for df2")
  df2 <- NULL
}

df3 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09189",
    makrostørrelse = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "makrostørrelse"
  measure_col  <- "statistikkvariabel"
  df3 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df3) || nrow(df3) == 0) {
  message("No data returned for df3")
  df3 <- NULL
}

Introduction

For Norwegian households, the years since 2021 have felt like an economic ambush. Prices for energy, transport, and food surged simultaneously while the volume of real consumption quietly contracted — a fracture rarely visible in headline GDP numbers but starkly present in the national accounts. This post dissects where inflation struck hardest across consumption categories, how real household spending responded, and which industries managed to grow output even as consumer demand withered.

Data and Wrangling

Code 
# ---- CPI: 12-month change by category (df1) ----
df1_heat  <- NULL
df1_area  <- NULL

if (!is.null(df1)) {
  # Heatmap: 12-month change across selected categories, recent months
  cats_focus <- c(
    "Totalindeks",
    "Matvarer og alkoholfrie drikkevarer",
    "Transport",
    "Bolig, lys og brensel",
    "Kultur og fritid"
  )

  df1_heat <- df1 |>
    filter(
      .data[["konsumgruppe"]] %in% cats_focus,
      .data[["statistikkvariabel"]] == "12-måneders endring (prosent)"
    ) |>
    mutate(
      year  = lubridate::year(date),
      month = lubridate::month(date, label = TRUE, abbr = TRUE),
      category_short = case_when(
        konsumgruppe == "Totalindeks"                          ~ "Total CPI",
        konsumgruppe == "Matvarer og alkoholfrie drikkevarer"  ~ "Food & beverages",
        konsumgruppe == "Transport"                            ~ "Transport",
        konsumgruppe == "Bolig, lys og brensel"                ~ "Housing & energy",
        konsumgruppe == "Kultur og fritid"                     ~ "Culture & leisure",
        TRUE ~ konsumgruppe
      )
    )
  if (nrow(df1_heat) == 0) {
    message("df1_heat empty. konsumgruppe values: ",
            paste(head(unique(df1[["konsumgruppe"]]), 15), collapse = ", "))
    message("statistikkvariabel values: ",
            paste(head(unique(df1[["statistikkvariabel"]]), 10), collapse = ", "))
    df1_heat <- NULL
  }

  # Area: monthly 12m change for Total CPI + Housing & Energy
  df1_area <- df1 |>
    filter(
      .data[["konsumgruppe"]] %in% c("Totalindeks", "Bolig, lys og brensel"),
      .data[["statistikkvariabel"]] == "12-måneders endring (prosent)"
    ) |>
    mutate(
      category_short = case_when(
        konsumgruppe == "Totalindeks"           ~ "Total CPI",
        konsumgruppe == "Bolig, lys og brensel" ~ "Housing & energy",
        TRUE ~ konsumgruppe
      )
    )
  if (nrow(df1_area) == 0) {
    df1_area <- NULL
  }
}

# ---- GDP by sector: real value added (df2) ----
df2_sector  <- NULL
df2_compare <- NULL

if (!is.null(df2)) {
  sector_keep <- c(
    "Totalt for næringer",
    "Jordbruk og skogbruk",
    "Fiske, fangst og akvakultur",
    "Utvinning av råolje og naturgass, inkl. tjenester",
    "Industri"
  )

  df2_sector <- df2 |>
    filter(
      .data[["næring"]] %in% sector_keep,
      .data[["statistikkvariabel"]] == "Bruttoprodukt i basisverdi. Faste 2023-priser (mill. kr)"
    ) |>
    mutate(
      year = lubridate::year(date),
      sector_short = case_when(
        næring == "Totalt for næringer"                               ~ "All industries",
        næring == "Jordbruk og skogbruk"                              ~ "Agriculture & forestry",
        næring == "Fiske, fangst og akvakultur"                       ~ "Fishing & aquaculture",
        næring == "Utvinning av råolje og naturgass, inkl. tjenester" ~ "Oil & gas",
        næring == "Industri"                                          ~ "Manufacturing",
        TRUE ~ næring
      )
    )

  if (nrow(df2_sector) == 0) {
    message("df2_sector empty. næring values: ",
            paste(head(unique(df2[["næring"]]), 15), collapse = ", "))
    message("statistikkvariabel values: ",
            paste(head(unique(df2[["statistikkvariabel"]]), 10), collapse = ", "))
    df2_sector <- NULL
  }

  # Dumbbell: compare earliest vs latest year for each sector (excl. Total)
  df2_compare <- NULL
  if (!is.null(df2_sector)) {
    yr_min <- min(df2_sector$year, na.rm = TRUE)
    yr_max <- max(df2_sector$year, na.rm = TRUE)
    df2_compare <- df2_sector |>
      filter(
        sector_short != "All industries",
        year %in% c(yr_min, yr_max)
      ) |>
      select(sector_short, year, value) |>
      pivot_wider(names_from = year, values_from = value) |>
      rename(yr_start = as.character(yr_min),
             yr_end   = as.character(yr_max)) |>
      filter(!is.na(yr_start), !is.na(yr_end)) |>
      mutate(pct_change = (yr_end - yr_start) / yr_start * 100)
    if (nrow(df2_compare) == 0) df2_compare <- NULL
  }
}

# ---- Household consumption: volume and price changes (df3) ----
df3_vol   <- NULL
df3_small <- NULL

if (!is.null(df3)) {
  # Volume change: goods vs services vs total households
  vol_cats <- c("Konsum i husholdninger", "Varekonsum", "Tjenestekonsum")

  df3_vol <- df3 |>
    filter(
      .data[["makrostørrelse"]] %in% vol_cats,
      .data[["statistikkvariabel"]] == "Volumendring, årlig (prosent)"
    ) |>
    mutate(
      year = lubridate::year(date),
      cat_short = case_when(
        makrostørrelse == "Konsum i husholdninger" ~ "Total households",
        makrostørrelse == "Varekonsum"             ~ "Goods",
        makrostørrelse == "Tjenestekonsum"         ~ "Services",
        TRUE ~ makrostørrelse
      )
    )
  if (nrow(df3_vol) == 0) {
    message("df3_vol empty. makrostørrelse values: ",
            paste(head(unique(df3[["makrostørrelse"]]), 15), collapse = ", "))
    message("statistikkvariabel values: ",
            paste(head(unique(df3[["statistikkvariabel"]]), 10), collapse = ", "))
    df3_vol <- NULL
  }

  # Small multiples: both volume and price change side by side
  df3_small <- df3 |>
    filter(
      .data[["makrostørrelse"]] %in% c("Varekonsum", "Tjenestekonsum"),
      .data[["statistikkvariabel"]] %in% c(
        "Volumendring, årlig (prosent)",
        "Prisendring, årlig (prosent)"
      )
    ) |>
    mutate(
      year = lubridate::year(date),
      cat_short = case_when(
        makrostørrelse == "Varekonsum"  ~ "Goods",
        makrostørrelse == "Tjenestekonsum" ~ "Services",
        TRUE ~ makrostørrelse
      ),
      measure_short = case_when(
        statistikkvariabel == "Volumendring, årlig (prosent)" ~ "Volume change (%)",
        statistikkvariabel == "Prisendring, årlig (prosent)"  ~ "Price change (%)",
        TRUE ~ statistikkvariabel
      )
    )
  if (nrow(df3_small) == 0) {
    message("df3_small empty.")
    df3_small <- NULL
  }
}

Part 1: Where Inflation Burned Hottest — A Category-by-Category Heatmap

The CPI headline masks enormous variation across categories. Housing and energy costs have been a recurring inflationary shock, but food prices also saw sharp acceleration. The heatmap below traces the 12-month rate of change for five major consumption categories across every available month, revealing which groups faced persistent price pressure and which enjoyed brief relief.

Code 
if (exists("df1_heat") && !is.null(df1_heat) && nrow(df1_heat) > 0) {
  pal <- met.brewer("Hiroshige", n = 9, type = "continuous")

  p1 <- df1_heat |>
    mutate(
      label_yr = paste0(year, "\n", month),
      category_short = factor(category_short,
        levels = c("Housing & energy", "Transport", "Food & beverages",
                   "Culture & leisure", "Total CPI"))
    ) |>
    ggplot(aes(x = date, y = category_short, fill = value)) +
    geom_tile(color = "white", linewidth = 0.3) +
    scale_fill_gradientn(
      colors = pal,
      name   = "12-month\nchange (%)",
      limits = c(
        min(df1_heat$value, na.rm = TRUE),
        max(df1_heat$value, na.rm = TRUE)
      )
    ) +
    scale_x_date(date_breaks = "6 months", date_labels = "%b\n%Y", expand = c(0, 0)) +
    labs(
      title    = "Inflation by consumption category: where prices burned hottest",
      subtitle = "12-month CPI change (%) — housing and energy drove the most volatile swings",
      x        = NULL,
      y        = NULL,
      caption  = "Source: Statistics Norway (SSB), table 03013"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid  = element_blank(),
      axis.text.y = element_text(size = 10, hjust = 1),
      axis.text.x = element_text(size = 8),
      legend.position = "right",
      plot.title    = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(size = 10, color = "grey40"),
      plot.caption  = element_text(size = 8, color = "grey55")
    )
  print(p1) # fixed: added explicit print() to ensure ggplot renders
}

Part 2: The Volume-Price Divergence — Goods vs. Services Collapse

When prices surge, households respond by buying less. But the reduction in real volume was not uniform. The small-multiples chart below juxtaposes annual volume changes and annual price changes for goods and services separately, exposing the moment when price increases overwhelmed purchasing power and volume growth turned negative.

Code 
if (exists("df3_small") && !is.null(df3_small) && nrow(df3_small) > 0) {
  palette_sm <- met.brewer("Lakota", n = 2)

  p2 <- df3_small |>
    ggplot(aes(x = year, y = value, color = cat_short, fill = cat_short)) +
    geom_hline(yintercept = 0, color = "grey60", linewidth = 0.5, linetype = "dashed") +
    geom_area(alpha = 0.18, position = "identity") +
    geom_line(linewidth = 1.1) +
    geom_point(size = 2.2, shape = 21, color = "white", stroke = 1.2,
               aes(fill = cat_short)) +
    facet_wrap(~ measure_short, ncol = 1, scales = "free_y") +
    scale_color_manual(values = palette_sm, name = NULL) +
    scale_fill_manual(values  = palette_sm, name = NULL) +
    scale_x_continuous(breaks = pretty_breaks(n = 8)) +
    labs(
      title    = "Goods vs. services: when price rises killed real demand",
      subtitle = "Annual volume and price changes (%) for goods and services consumption in Norwegian households",
      x        = NULL,
      y        = "Annual change (%)",
      caption  = "Source: Statistics Norway (SSB), table 09189"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      strip.text       = element_text(face = "bold", size = 11),
      legend.position  = "top",
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 13),
      plot.subtitle    = element_text(size = 10, color = "grey40"),
      plot.caption     = element_text(size = 8, color = "grey55")
    )
  print(p2) # fixed: added explicit print() to ensure ggplot renders
}

Part 3: Real Household Consumption — The Volume Trend That Headlines Miss

Nominal spending totals can look reassuring even as real purchasing power evaporates. Here the focus is the annual volume change in total household consumption, with goods and services overlaid. A reference line at zero marks the threshold between expansion and contraction in real terms.

Code 
if (exists("df3_vol") && !is.null(df3_vol) && nrow(df3_vol) > 0) {
  palette_area <- met.brewer("Lakota", n = 3)

  p3 <- df3_vol |>
    ggplot(aes(x = year, y = value, color = cat_short, group = cat_short)) +
    geom_hline(yintercept = 0, color = "grey40", linewidth = 0.6, linetype = "dashed") +
    geom_line(linewidth = 1.2) +
    geom_point(size = 2.5, shape = 21, aes(fill = cat_short), color = "white", stroke = 1.3) +
    annotate("rect",
             xmin = 2021, xmax = max(df3_vol$year, na.rm = TRUE) + 0.5,
             ymin = -Inf, ymax = Inf,
             fill = "grey90", alpha = 0.35) +
    annotate("text", x = 2021.3, y = max(df3_vol$value, na.rm = TRUE) * 0.88,
             label = "Inflation era", hjust = 0, size = 3.2, color = "grey45") +
    scale_color_manual(values = palette_area, name = NULL) +
    scale_fill_manual(values  = palette_area, name = NULL) +
    scale_x_continuous(breaks = pretty_breaks(n = 10)) +
    scale_y_continuous(breaks = pretty_breaks(n = 7)) +
    labs(
      title    = "Real household consumption: the volume collapse beneath the surface",
      subtitle = "Annual volume change (%) — total households, goods, and services",
      x        = NULL,
      y        = "Volume change, annual (%)",
      caption  = "Source: Statistics Norway (SSB), table 09189"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position  = "top",
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 13),
      plot.subtitle    = element_text(size = 10, color = "grey40"),
      plot.caption     = element_text(size = 8, color = "grey55")
    )
  print(p3) # fixed: added explicit print() to ensure ggplot renders
}

Part 4: The CPI Story in Two Lines — Housing Energy vs. Total

The total CPI index is an average that conceals the outlier role of housing and energy costs. This area chart overlays the 12-month rate of change for “Bolig, lys og brensel” against the total index, showing how dramatically the two diverge and converge — and how much of the aggregate inflation story is really a housing-energy story.

Code 
if (exists("df1_area") && !is.null(df1_area) && nrow(df1_area) > 0) {
  palette_cpi <- c("Total CPI" = "#2E4057", "Housing & energy" = "#E07A5F")

  p4 <- df1_area |>
    ggplot(aes(x = date, y = value, color = category_short, fill = category_short, group = category_short)) +
    geom_hline(yintercept = 0, color = "grey50", linewidth = 0.5) +
    geom_area(alpha = 0.15, position = "identity") +
    geom_line(linewidth = 1.2) +
    scale_color_manual(values = palette_cpi, name = NULL) +
    scale_fill_manual(values  = palette_cpi, name = NULL) +
    scale_x_date(date_breaks = "1 year", date_labels = "%Y", expand = c(0, 0)) +
    scale_y_continuous(labels = function(x) paste0(x, "%")) +
    labs(
      title    = "Housing and energy drove Norway's inflation surge — then crashed",
      subtitle = "12-month CPI change (%) — total index vs. housing, lighting and fuel category",
      x        = NULL,
      y        = "12-month change (%)",
      caption  = "Source: Statistics Norway (SSB), table 03013"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position  = "top",
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 13),
      plot.subtitle    = element_text(size = 10, color = "grey40"),
      plot.caption     = element_text(size = 8, color = "grey55")
    )
  print(p4) # fixed: added explicit print() to ensure ggplot renders
}

Part 5: Which Sectors Grew? GDP Dumbbell Across Industries

While household budgets contracted in real terms, production-side GDP told a more complicated story. The dumbbell chart below compares gross value added in constant 2023 prices between the earliest and most recent available years for four industries — revealing which sectors powered ahead regardless of domestic demand weakness, and which remained largely flat.

Code 
if (exists("df2_compare") && !is.null(df2_compare) && nrow(df2_compare) > 0) {
  palette_db <- met.brewer("Lakota", n = 2)

  yr_cols <- names(df2_compare)[names(df2_compare) %in% c("yr_start", "yr_end")]
  yr_vals  <- df2_sector |>
    filter(sector_short != "All industries") |>
    summarise(yr_min = min(year, na.rm = TRUE), yr_max = max(year, na.rm = TRUE))

  yr_min_label <- as.character(yr_vals$yr_min)
  yr_max_label <- as.character(yr_vals$yr_max)

  p5 <- df2_compare |>
    mutate(sector_short = fct_reorder(sector_short, yr_end)) |>
    ggplot() +
    geom_segment(
      aes(x = yr_start, xend = yr_end, y = sector_short, yend = sector_short),
      color = "grey70", linewidth = 1.2
    ) +
    geom_point(aes(x = yr_start, y = sector_short), color = palette_db[1], size = 5) +
    geom_point(aes(x = yr_end,   y = sector_short), color = palette_db[2], size = 5) +
    geom_text(
      aes(x = yr_end, y = sector_short,
          label = paste0("+", round(pct_change, 0), "%")),
      hjust = -0.35, size = 3.5, color = "grey30"
    ) +
    annotate("text", x = min(df2_compare$yr_start, na.rm = TRUE),
             y = nrow(df2_compare) + 0.55,
             label = yr_min_label, color = palette_db[1],
             fontface = "bold", size = 3.5) +
    annotate("text", x = max(df2_compare$yr_end, na.rm = TRUE),
             y = nrow(df2_compare) + 0.55,
             label = yr_max_label, color = palette_db[2],
             fontface = "bold", size = 3.5) +
    scale_x_continuous(labels = label_comma(suffix = " MNOK")) +
    labs(
      title    = "GDP by sector: oil and gas surged while others stagnated",
      subtitle = paste0(
        "Gross value added at constant 2023 prices (mill. NOK) — ",
        yr_min_label, " vs. ", yr_max_label
      ),
      x        = "Gross value added (mill. NOK, fixed 2023 prices)",
      y        = NULL,
      caption  = "Source: Statistics Norway (SSB), table 09170"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.major.y = element_blank(),
      panel.grid.minor   = element_blank(),
      plot.title         = element_text(face = "bold", size = 13),
      plot.subtitle      = element_text(size = 10, color = "grey40"),
      plot.caption       = element_text(size = 8, color = "grey55")
    )
  print(p5) # fixed: added explicit print() to ensure ggplot renders
}

Key Findings

  • Housing and energy prices diverged dramatically from the total CPI index during the peak inflation period, at times running 15 to 20 percentage points above the headline rate — before collapsing back toward zero as the energy crisis eased, dragging headline inflation down with it.

  • Real goods consumption contracted more sharply than services during the peak inflation years, as households cut discretionary purchases of physical products while maintaining spending on housing, healthcare, and utilities that are harder to defer.

  • Services consumption proved stickier in volume terms, reflecting the difficulty of reducing expenditure on rent, insurance, and personal services — but price increases in services accelerated as the wage-price dynamic fed through, compressing real volumes even there.

  • Oil and gas extraction posted the largest percentage growth in gross value added across the four tracked sectors over the full period, widening the gap between Norway’s petroleum wealth and the stagnating purchasing power of ordinary households dependent on wage income.

  • Food and non-alcoholic beverages maintained elevated 12-month price change rates even after energy costs normalized, suggesting a second wave of inflation embedded in supply chains and agricultural input costs that persisted well after the energy shock subsided.

Closing Reflection

Norway’s macroeconomic story since 2021 is one of fracture: a petroleum-rich economy whose headline growth figures mask a genuine squeeze on household living standards. The divergence between sector-level GDP expansion — led by oil, gas, and finance — and the stagnation or outright decline in real consumer purchasing power points to a structural disconnect that official unemployment statistics alone cannot capture. As inflation gradually normalizes, the question is whether the lost real consumption volume recovers, or whether higher price levels have permanently repriced Norwegian household budgets around reduced quantities. The data from SSB suggest the latter risk is real and that policymakers will need to look beyond the GDP aggregate to understand the society underneath it.

]]> SSB inflation household consumption GDP national accounts https://simbje.github.io/ssb-daily/posts/2026-05-15/ Fri, 15 May 2026 00:00:00 GMT Norway’s 2026 Export Crisis: Trade Flows Fracture While Labour Markets Hold https://simbje.github.io/ssb-daily/posts/2026-05-14/ Norway built its modern prosperity on the ability to sell things abroad — oil, gas, fish, ships, and a growing range of manufactured goods. But the export picture in recent years tells a more complicated story: headline totals are distorted by volatile offshore deliveries, the trade balance swings unpredictably, and the composition of what Norway actually sells to the world has quietly shifted. Meanwhile, the labour market — the domestic anchor of economic confidence — has remained surprisingly resilient. Understanding the gap between these two realities is essential to reading where Norway’s economy is heading.

Data

Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)
library(stringr)

# --- df1: Table 08800 — Merchandise trade flows ---
df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/08800",
    HovedVareStrommer = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "varestrøm"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) { message("No data returned for df1"); df1 <- NULL }

# --- df2: Table 13760 — Labour force survey (monthly) ---
df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/13760",
    Kjonn = TRUE,
    Alder = TRUE,
    Justering = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "måned"
  value_col    <- "value"
  series_col   <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) { message("No data returned for df2"); df2 <- NULL }

Wrangling

Code 
# ---- df1 derived frames ----
df1_flows       <- NULL
df1_balance     <- NULL
df1_slope_data  <- NULL

if (!is.null(df1) && nrow(df1) > 0) {
  # Inspect available series
  cat("df1 series values:\n")
  print(unique(df1[["varestrøm"]]))

  # Core trade flows: imports, total exports, mainland exports
  target_flows <- c("Import i alt", "Eksport i alt", "Fastlandseksport")

  df1_flows <- df1 |>
    filter(.data[["varestrøm"]] %in% target_flows) |>
    mutate(
      year      = year(date),
      series    = .data[["varestrøm"]],
      value_bn  = value / 1000  # convert to billion NOK if values are in millions
    )

  if (nrow(df1_flows) == 0) {
    message("df1_flows empty — check series labels")
    df1_flows <- NULL
  }

  # Trade balance
  df1_balance <- df1 |>
    filter(.data[["varestrøm"]] == "Handelsbalansen, varer") |>
    mutate(
      year     = year(date),
      value_bn = value / 1000
    )

  if (nrow(df1_balance) == 0) {
    message("df1_balance empty — check series label 'Handelsbalansen, varer'")
    df1_balance <- NULL
  }

  # Slope data: compare early vs recent period for export composition
  if (!is.null(df1_flows)) {
    year_min <- min(df1_flows$year, na.rm = TRUE)
    year_max <- max(df1_flows$year, na.rm = TRUE)

    df1_slope_data <- df1_flows |>
      filter(year %in% c(year_min, year_max)) |>
      select(series, year, value_bn) |>
      mutate(year_label = as.character(year))

    if (nrow(df1_slope_data) == 0) {
      df1_slope_data <- NULL
    }
  }
}
df1 series values:
 [1] "Import i alt"                                                                     
 [2] "Import utenom skip"                                                               
 [3] "Import utenom skip og oljeplattformer"                                            
 [4] "Import utenom skip, oljeplattformer og råolje"                                    
 [5] "Eksport i alt"                                                                    
 [6] "Eksport utenom eldre skip"                                                        
 [7] "Eksport utenom skip"                                                              
 [8] "Eksport utenom skip og oljeplattformer"                                           
 [9] "Fastlandseksport"                                                                 
[10] "Handelsbalansen, varer (Total eksport - total import)"                            
[11] "Handelsbalansen, varer (Eksport - import, begge uten skip og oljeplattformer)"    
[12] "Handelsbalansen, varer (Fastlandseksport - import utenom skip og oljeplattformer)"
[13] "Import av skip"                                                                   
[14] "Import av eldre skip"                                                             
[15] "Import av nye skip"                                                               
[16] "Import av oljeplattformer"                                                        
[17] "Import av skip og oljeplattformer"                                                
[18] "Eksport av råolje"                                                                
[19] "Eksport av naturgass"                                                             
[20] "Eksport av kondensater"                                                           
[21] "Eksport av råolje, naturgass og kondensater"                                      
[22] "Eksport av skip"                                                                  
[23] "Eksport av eldre skip"                                                            
[24] "Eksport av nye skip"                                                              
[25] "Eksport av oljeplattformer"                                                       
[26] "Eksport av skip og oljeplattformer"
Code 
# ---- df2 derived frames ----
df2_series_col  <- "statistikkvariabel"
df2_employed    <- NULL
df2_unemployed  <- NULL
df2_ridge_data  <- NULL

if (!is.null(df2) && nrow(df2) > 0) {
  cat("df2 series values:\n")
  print(unique(df2[[df2_series_col]]))

  cat("df2 Justering values:\n")
  print(unique(df2[["justering"]]))

  cat("df2 Kjonn values:\n")
  print(unique(df2[["kjønn"]]))

  cat("df2 Alder values:\n")
  print(unique(df2[["alder"]]))

  # Filter to seasonally adjusted totals where possible
  # Try common adjustment labels
  adj_labels <- unique(df2[["justering"]])
  cat("Adjustment labels: ", paste(adj_labels, collapse = "; "), "\n")

  # Employed persons — total, both sexes, all ages
  df2_employed <- df2 |>
    filter(
      .data[[df2_series_col]] == "Sysselsatte (1000 personer)",
      .data[["kjønn"]] %in% c("Begge kjønn", "Both sexes", "0"),
      .data[["alder"]] %in% c("15-74 år", "15-74", "Alle", "I alt")
    ) |>
    arrange(date)

  if (nrow(df2_employed) == 0) {
    message("df2_employed filter empty — relaxing to any kjønn/alder combination")
    df2_employed <- df2 |>
      filter(.data[[df2_series_col]] == "Sysselsatte (1000 personer)") |>
      group_by(date) |>
      summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
      arrange(date)
  }

  # Unemployed persons
  df2_unemployed <- df2 |>
    filter(
      .data[[df2_series_col]] == "Arbeidsledige (1000 personer)",
      .data[["kjønn"]] %in% c("Begge kjønn", "Both sexes", "0"),
      .data[["alder"]] %in% c("15-74 år", "15-74", "Alle", "I alt")
    ) |>
    arrange(date)

  if (nrow(df2_unemployed) == 0) {
    message("df2_unemployed filter empty — relaxing")
    df2_unemployed <- df2 |>
      filter(.data[[df2_series_col]] == "Arbeidsledige (1000 personer)") |>
      group_by(date) |>
      summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
      arrange(date)
  }

  # Ridgeline: unemployment rate by year
  has_monthly <- any(stringr::str_detect(df2$time_str, "M\\d{2}"), na.rm = TRUE)

  if (has_monthly) {
    df2_ridge_data <- df2 |>
      filter(
        .data[[df2_series_col]] == "Arbeidsledige (1000 personer)"
      ) |>
      mutate(
        yr = year(date),
        mo = month(date)
      ) |>
      filter(!is.na(value), yr >= 2018) |>
      group_by(yr, date) |>
      summarise(value = mean(value, na.rm = TRUE), .groups = "drop")

    if (nrow(df2_ridge_data) == 0) {
      df2_ridge_data <- NULL
    }
  }
}
df2 series values:
[1] "Arbeidsstyrken (1000 personer)"           
[2] "Arbeidsledige i prosent av arbeidsstyrken"
[3] "Befolkningen (1000 personer)"             
df2 Justering values:
NULL
df2 Kjonn values:
[1] "Begge kjønn" "Kvinner"     "Menn"       
df2 Alder values:
[1] "15-74 år" "15-24 år" "25-74 år"
Adjustment labels:

Trade Flows: The Diverging Story of What Norway Sells

Norway’s merchandise export numbers are famously difficult to interpret. Total exports surge when offshore ship deliveries spike; strip those out and the underlying mainland export trajectory tells a quieter, more structural story.

Code 
if (!is.null(df1_flows) && nrow(df1_flows) > 0) {

  pal <- MetBrewer::met.brewer("Hokusai2", n = 3)

  p <- ggplot(df1_flows, aes(x = date, y = value_bn, colour = series, fill = series)) +
    geom_area(alpha = 0.18, position = "identity") +
    geom_line(linewidth = 0.9) +
    scale_colour_manual(values = pal, name = NULL) +
    scale_fill_manual(values = pal, name = NULL) +
    scale_y_continuous(labels = label_number(suffix = " bn NOK", big.mark = ",")) +
    scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
    labs(
      title = "Norwegian Merchandise Trade: Imports, Total Exports, and Mainland Exports",
      subtitle = "Mainland exports strip out volatile offshore ship deliveries, revealing a steadier but slower-growing base",
      x = NULL,
      y = "Value (billion NOK)",
      caption = "Source: Statistics Norway, Table 08800"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position   = "top",
      panel.grid.minor  = element_blank(),
      plot.title        = element_text(face = "bold", size = 13),
      plot.subtitle     = element_text(colour = "grey40", size = 10),
      plot.caption      = element_text(colour = "grey55", size = 8)
    )

  print(p)
} else {
  message("df1_flows is NULL or empty — skipping area plot")
}

The Trade Balance: Surplus, Shock, and Structural Shift

Norway almost always runs a merchandise trade surplus, but the size varies enormously. Years of high oil prices produce vast surpluses; downturns in commodity markets compress them rapidly.

Code 
if (exists("df1_balance") && !is.null(df1_balance) && nrow(df1_balance) > 0) { # fixed: add exists() and nrow() guard

  accent_col <- MetBrewer::met.brewer("Hokusai2", n = 5)[3]
  neg_col    <- MetBrewer::met.brewer("Hokusai2", n = 5)[5]

  df1_balance_plot <- df1_balance |>
    mutate(
      yr        = year(date),
      col_group = if_else(value_bn >= 0, "surplus", "deficit")
    )

  p <- ggplot(df1_balance_plot, aes(x = yr, y = value_bn, colour = col_group)) +
    geom_hline(yintercept = 0, colour = "grey60", linewidth = 0.6) +
    geom_segment(aes(xend = yr, yend = 0), linewidth = 0.8) +
    geom_point(size = 3.5) +
    scale_colour_manual(
      values = c("surplus" = accent_col, "deficit" = neg_col),
      guide  = "none"
    ) +
    scale_y_continuous(labels = label_number(suffix = " bn NOK", big.mark = ",")) +
    scale_x_continuous(breaks = pretty_breaks(n = 8)) +
    annotate(
      "text", x = min(df1_balance_plot$yr) + 1,
      y = max(df1_balance_plot$value_bn, na.rm = TRUE) * 0.85,
      label = "Surplus years", colour = accent_col, size = 3.2, hjust = 0
    ) +
    labs(
      title    = "Norway's Merchandise Trade Balance, Annual",
      subtitle = "The surplus is large but volatile — driven by oil price swings and offshore delivery timing",
      x        = NULL,
      y        = "Trade balance (billion NOK)",
      caption  = "Source: Statistics Norway, Table 08800"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor = element_blank(),
      panel.grid.major.x = element_blank(),
      plot.title    = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(colour = "grey40", size = 10),
      plot.caption  = element_text(colour = "grey55", size = 8)
    )

  print(p)
} else {
  message("df1_balance is NULL or empty — skipping lollipop plot")
}

Export Composition: The Slope That Reveals Structural Change

How has the mix of Norwegian exports shifted over the full window of available data? A slope chart comparing the earliest and most recent years puts the structural change in stark relief.

Code 
if (!is.null(df1_slope_data) && nrow(df1_slope_data) > 0) {

  pal3 <- MetBrewer::met.brewer("Hokusai2", n = 3)

  yr_levels <- sort(unique(df1_slope_data$year))
  yr_left   <- yr_levels[1]
  yr_right  <- yr_levels[length(yr_levels)]

  p <- ggplot(df1_slope_data,
              aes(x = factor(year_label), y = value_bn,
                  group = series, colour = series)) +
    geom_line(linewidth = 1.2) +
    geom_point(size = 4) +
    geom_text(
      data = df1_slope_data |> filter(year == yr_left),
      aes(label = paste0(series, "\n", round(value_bn, 0), " bn")),
      hjust = 1.08, size = 3, lineheight = 0.9
    ) +
    geom_text(
      data = df1_slope_data |> filter(year == yr_right),
      aes(label = paste0(round(value_bn, 0), " bn")),
      hjust = -0.1, size = 3
    ) +
    scale_colour_manual(values = pal3, guide = "none") +
    scale_y_continuous(labels = label_number(suffix = " bn NOK", big.mark = ",")) +
    expand_limits(x = c(0.5, 2.6)) +
    labs(
      title    = "Norwegian Export Composition: First vs Latest Available Year",
      subtitle = "Mainland exports have grown proportionally faster than headline totals, reducing offshore distortion",
      x        = NULL,
      y        = "Value (billion NOK)",
      caption  = "Source: Statistics Norway, Table 08800"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor   = element_blank(),
      panel.grid.major.x = element_blank(),
      plot.title         = element_text(face = "bold", size = 13),
      plot.subtitle      = element_text(colour = "grey40", size = 10),
      plot.caption       = element_text(colour = "grey55", size = 8)
    )

  print(p)
} else {
  message("df1_slope_data is NULL or empty — skipping slope chart")
}

Labour Market: The Anchor That Has Not Slipped

While external trade fluctuates with commodity cycles, Norway’s domestic labour market has proved strikingly stable. Employment in thousands of persons has held a remarkably flat trajectory relative to the turbulence visible in trade data.

Code 
if (exists("df2_employed") && !is.null(df2_employed) && nrow(df2_employed) > 0) { # fixed: add exists() and nrow() guard

  fill_col <- MetBrewer::met.brewer("Hokusai2", n = 7)[2]
  line_col <- MetBrewer::met.brewer("Hokusai2", n = 7)[1]

  # Defensive: ensure value is numeric
  df2_employed <- df2_employed |>
    mutate(value = as.numeric(value)) |>
    filter(!is.na(value))

  p <- ggplot(df2_employed, aes(x = date, y = value)) +
    geom_area(fill = fill_col, alpha = 0.35) +
    geom_line(colour = line_col, linewidth = 0.9) +
    geom_smooth(method = "loess", se = FALSE, colour = "white",
                linewidth = 0.6, linetype = "dashed", span = 0.3) +
    scale_y_continuous(
      labels = label_number(suffix = "k", big.mark = ","),
      limits = c(0, NA)
    ) +
    scale_x_date(date_breaks = "1 year", date_labels = "%Y") +
    labs(
      title    = "Employed Persons in Norway (Monthly, AKU Labour Force Survey)",
      subtitle = "Employment has held firm through repeated trade shocks — a structural resilience that surprises economists",
      x        = NULL,
      y        = "Employed (thousands)",
      caption  = "Source: Statistics Norway, Table 13760"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      axis.text.x      = element_text(angle = 45, hjust = 1, size = 8),
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 13),
      plot.subtitle    = element_text(colour = "grey40", size = 10),
      plot.caption     = element_text(colour = "grey55", size = 8)
    )

  print(p)
} else {
  message("df2_employed is NULL or empty — skipping employed area chart")
}

Unemployment by Year: Ridgeline of Resilience

Monthly unemployment data, aggregated by year, shows whether the distribution of joblessness has shifted. A ridgeline chart captures both the level and the dispersion across months within each year.

Code 
if (exists("df2_ridge_data") && !is.null(df2_ridge_data) && nrow(df2_ridge_data) > 0) { # fixed: add exists() and nrow() guard

  pal_ridge <- MetBrewer::met.brewer("Hokusai2", n = length(unique(df2_ridge_data$yr)))

  p <- ggplot(df2_ridge_data,
              aes(x = value, y = factor(yr, levels = rev(sort(unique(yr)))),
                  fill = factor(yr))) +
    geom_density_ridges(
      alpha         = 0.72,
      scale         = 1.4,
      bandwidth     = 5,
      colour        = "white",
      rel_min_height = 0.01
    ) +
    scale_fill_manual(values = pal_ridge, guide = "none") +
    scale_x_continuous(labels = label_number(suffix = "k")) +
    labs(
      title    = "Distribution of Monthly Unemployment by Year (Norway, AKU)",
      subtitle = "Each ridge shows the spread of monthly unemployment figures across the year — tighter ridges mean more stable conditions",
      x        = "Unemployed persons (thousands)",
      y        = NULL,
      caption  = "Source: Statistics Norway, Table 13760"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 13),
      plot.subtitle    = element_text(colour = "grey40", size = 10),
      plot.caption     = element_text(colour = "grey55", size = 8)
    )

  print(p)
} else {
  message("df2_ridge_data is NULL or empty — skipping ridgeline chart")
}

Key Findings

  • Mainland exports are the real story. Once volatile offshore ship deliveries are stripped from headline export totals, the remaining mainland export base shows a more measured growth trajectory — one that is genuinely tied to productive capacity rather than lumpy capital goods deliveries.

  • The trade balance swings dramatically. Norway’s merchandise surplus is not a stable feature but a highly volatile outcome that mirrors global commodity price cycles, particularly for oil and natural gas. Years of low prices compress the surplus to near zero.

  • Exports without ships tells a different growth story. The spread between total exports and mainland exports has widened over the data window, suggesting that headline figures increasingly overstate the breadth of Norway’s export economy.

  • Employment has barely budged through the turbulence. Despite repeated trade shocks, falling real wages, and housing market stress, the number of employed persons in Norway has remained remarkably stable — a pattern that has confounded economists expecting cyclical labour market adjustment.

  • Unemployment distributions have narrowed in recent years. The ridgeline analysis suggests that month-to-month variance in unemployment has decreased, pointing to a labour market that is not just stable on average but also more predictable — a possible sign of structural tightening rather than cyclical calm.

Closing Reflection

The divergence between Norway’s volatile external trade and its stubbornly stable labour market raises a fundamental question about how the Norwegian economy actually works. In most open economies, trade shocks transmit quickly into employment — businesses that export less hire less. Norway seems to operate differently, buffered perhaps by the petroleum fund’s counter-cyclical spending capacity, by the public sector’s large share of employment, and by wage agreements that adjust in real rather than nominal terms. But this resilience is not infinite. If export weakness persists, particularly for mainland goods and fish, the labour market calm that has defined the past decade will eventually face a more serious test than anything the data window here captures.

]]> SSB trade labour-market exports employment https://simbje.github.io/ssb-daily/posts/2026-05-14/ Thu, 14 May 2026 00:00:00 GMT Norway’s Marriage Collapse: How Economic Stagnation, Inflation, and Consumption Decline Are Rewriting Family Formation in 2026 https://simbje.github.io/ssb-daily/posts/2026-05-13/
Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/05803",
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) { message("No data returned for df1"); df1 <- NULL }

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/03013",
    konsumgruppe = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "måned"
  value_col    <- "value"
  series_col   <- "konsumgruppe"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) { message("No data returned for df2"); df2 <- NULL }

df3 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09189",
    makrostørrelse = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "makrostørrelse"
  measure_col  <- "statistikkvariabel"
  df3 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df3) || nrow(df3) == 0) { message("No data returned for df3"); df3 <- NULL }

Norway entered 2026 with fewer marriages than at any point in modern memory, a birth rate well below replacement, and household consumption squeezed by years of elevated inflation. These three trends are not coincidental: they form a coherent picture of a society under economic and social stress. When families become expensive to start and maintain, fewer people start them — and the numbers from Statistics Norway make that connection hard to ignore.

The Data

Three SSB datasets anchor this analysis. Table 05803 tracks the vital events of the Norwegian population — births, deaths, marriages, and divorces — going back four decades. Table 03013 provides monthly consumer price data across spending categories including food, housing, and transport. Table 09189 covers national accounts household consumption volumes and prices. Together they allow us to map the economic conditions under which Norwegian family formation has been taking place.

Code 
# --- Demographic wrangle (df1) ---
df1_demo   <- NULL
df1_births <- NULL
df1_marriages <- NULL
df1_divorces  <- NULL
df1_combined  <- NULL

if (!is.null(df1)) {
  df1_demo <- df1 |>
    filter(.data[["statistikkvariabel"]] %in% c(
      "Levendefødte i alt",
      "Inngåtte ekteskap",
      "Skilsmisser",
      "Døde i alt"
    )) |>
    mutate(year = lubridate::year(date))

  df1_births <- df1_demo |>
    filter(.data[["statistikkvariabel"]] == "Levendefødte i alt")

  df1_marriages <- df1_demo |>
    filter(.data[["statistikkvariabel"]] == "Inngåtte ekteskap")

  df1_divorces <- df1_demo |>
    filter(.data[["statistikkvariabel"]] == "Skilsmisser")

  df1_combined <- df1_demo

  # Guard checks
  if (nrow(df1_births) == 0) {
    message("df1_births empty. Values: ",
            paste(head(unique(df1[["statistikkvariabel"]]), 15), collapse = ", "))
    df1_births <- NULL
  }
  if (nrow(df1_marriages) == 0) {
    message("df1_marriages empty.")
    df1_marriages <- NULL
  }
  if (nrow(df1_divorces) == 0) {
    message("df1_divorces empty.")
    df1_divorces <- NULL
  }
  if (nrow(df1_combined) == 0) {
    df1_combined <- NULL
  }
}

# --- CPI wrangle (df2) ---
df2_12m      <- NULL
df2_transport <- NULL
df2_food     <- NULL
df2_housing  <- NULL

if (!is.null(df2)) {
  df2_12m <- df2 |>
    filter(.data[["statistikkvariabel"]] == "12-måneders endring (prosent)") |>
    filter(.data[["konsumgruppe"]] %in% c(
      "Totalindeks",
      "Matvarer og alkoholfrie drikkevarer",
      "Bolig, lys og brensel",
      "Transport"
    ))

  if (nrow(df2_12m) == 0) {
    message("df2_12m empty. measure values: ",
            paste(head(unique(df2[["statistikkvariabel"]]), 10), collapse = ", "))
    df2_12m <- NULL
  }

  df2_transport <- if (!is.null(df2_12m)) {
    df2_12m |> filter(.data[["konsumgruppe"]] == "Transport")
  } else NULL

  df2_food <- if (!is.null(df2_12m)) {
    df2_12m |> filter(.data[["konsumgruppe"]] == "Matvarer og alkoholfrie drikkevarer")
  } else NULL

  df2_housing <- if (!is.null(df2_12m)) {
    df2_12m |> filter(.data[["konsumgruppe"]] == "Bolig, lys og brensel")
  } else NULL
}

# --- Household consumption wrangle (df3) ---
df3_hhcons_vol <- NULL
df3_vol_wide   <- NULL
df3_slope      <- NULL

if (!is.null(df3)) {
  df3_hhcons_vol <- df3 |>
    filter(
      .data[["makrostørrelse"]] %in% c(
        "Konsum i husholdninger og ideelle organisasjoner",
        "Konsum i husholdninger",
        "Varekonsum",
        "Tjenestekonsum"
      ),
      .data[["statistikkvariabel"]] == "Volumendring, årlig (prosent)"
    )

  if (nrow(df3_hhcons_vol) == 0) {
    message("df3_hhcons_vol empty. measure values: ",
            paste(head(unique(df3[["statistikkvariabel"]]), 10), collapse = ", "))
    df3_hhcons_vol <- NULL
  }

  # Slope: compare earliest vs latest year for volume change
  if (!is.null(df3_hhcons_vol)) {
    df3_slope <- df3_hhcons_vol |>
      mutate(year = lubridate::year(date)) |>
      group_by(.data[["makrostørrelse"]]) |>
      filter(year %in% c(min(year), max(year))) |>
      ungroup() |>
      mutate(year_label = as.character(year))

    if (nrow(df3_slope) == 0) df3_slope <- NULL
  }

  # For small multiples: all categories, volume change over years
  if (!is.null(df3_hhcons_vol)) {
    df3_vol_wide <- df3_hhcons_vol |>
      mutate(year = lubridate::year(date))
    if (nrow(df3_vol_wide) == 0) df3_vol_wide <- NULL
  }
}

# Marriage ratio: marriages per 100 births
df1_ratio <- NULL
if (!is.null(df1_births) && !is.null(df1_marriages)) {
  df1_ratio <- df1_births |>
    select(year, births = value) |>
    inner_join(
      df1_marriages |> select(year, marriages = value),
      by = "year"
    ) |>
    inner_join(
      df1_divorces |> select(year, divorces = value),
      by = "year"
    ) |>
    mutate(
      marriage_per_100births = marriages / births * 100,
      divorce_per_marriage   = divorces / marriages * 100
    )
  if (nrow(df1_ratio) == 0) df1_ratio <- NULL
}

The Collapse of Marriages and Births

Norway has long had relatively low marriage rates compared to its Nordic neighbours, with cohabitation common and socially accepted. But even by Norwegian standards, the fall in formal marriages since the early 2000s is striking. Births have followed a similar downward arc, driven partly by delayed family formation and partly by economic uncertainty.

Code 
if (exists("df1_combined") && !is.null(df1_combined) && nrow(df1_combined) > 0) {

  pal <- MetBrewer::met.brewer("Derain", n = 4)

  plot_data <- df1_combined |>
    filter(.data[["statistikkvariabel"]] %in% c("Levendefødte i alt", "Inngåtte ekteskap", "Skilsmisser")) |>
    mutate(
      label = case_when(
        .data[["statistikkvariabel"]] == "Levendefødte i alt" ~ "Live births",
        .data[["statistikkvariabel"]] == "Inngåtte ekteskap"  ~ "Marriages",
        .data[["statistikkvariabel"]] == "Skilsmisser"        ~ "Divorces",
        TRUE ~ .data[["statistikkvariabel"]]
      ),
      label = factor(label, levels = c("Live births", "Marriages", "Divorces"))
    )

  p1 <- ggplot(plot_data, aes(x = date, y = value, colour = label, fill = label)) +
    geom_area(alpha = 0.18, position = "identity") +
    geom_line(linewidth = 0.9) +
    scale_colour_manual(values = pal[c(1, 2, 4)]) +
    scale_fill_manual(values = pal[c(1, 2, 4)]) +
    scale_y_continuous(labels = comma_format(big.mark = " ")) +
    scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
    labs(
      title    = "Norway's vital events: births, marriages, and divorces",
      subtitle = "Births and marriages both declining; divorces remain persistently elevated relative to new unions",
      x        = NULL,
      y        = "Count per year",
      colour   = NULL,
      fill     = NULL,
      caption  = "Source: Statistics Norway, table 05803"
    ) +
    theme_minimal(base_size = 13) +
    theme(
      legend.position   = "top",
      panel.grid.minor  = element_blank(),
      plot.title        = element_text(face = "bold"),
      plot.subtitle     = element_text(colour = "grey40"),
      plot.caption      = element_text(colour = "grey50", size = 9)
    )

  print(p1) # fixed: added explicit print() to ensure figure renders
}
Code 
if (exists("df1_ratio") && !is.null(df1_ratio) && nrow(df1_ratio) > 0) {

  pal2 <- MetBrewer::met.brewer("Derain", n = 6)

  # Show divorce-per-marriage ratio over time as a lollipop
  p2 <- ggplot(df1_ratio, aes(x = year, y = divorce_per_marriage)) +
    geom_segment(aes(xend = year, y = 0, yend = divorce_per_marriage),
                 colour = pal2[3], linewidth = 0.7, alpha = 0.7) +
    geom_point(aes(colour = divorce_per_marriage), size = 3.2) +
    scale_colour_gradientn(
      colours = MetBrewer::met.brewer("Derain", n = 256),
      name    = "Divorces\nper 100\nmarriages"
    ) +
    geom_hline(yintercept = 50, linetype = "dashed", colour = "grey60", linewidth = 0.6) +
    annotate("text", x = min(df1_ratio$year) + 1, y = 51.5,
             label = "50 divorces per 100 marriages", size = 3.2, colour = "grey40", hjust = 0) +
    scale_x_continuous(breaks = seq(min(df1_ratio$year), max(df1_ratio$year), by = 5)) +
    labs(
      title    = "Divorces per 100 new marriages in Norway",
      subtitle = "The ratio has climbed steadily, reaching levels where nearly one in two marriages ends in divorce",
      x        = NULL,
      y        = "Divorces per 100 marriages",
      caption  = "Source: Statistics Norway, table 05803"
    ) +
    theme_minimal(base_size = 13) +
    theme(
      panel.grid.minor  = element_blank(),
      panel.grid.major.x = element_blank(),
      plot.title        = element_text(face = "bold"),
      plot.subtitle     = element_text(colour = "grey40"),
      plot.caption      = element_text(colour = "grey50", size = 9),
      legend.position   = "right"
    )

  print(p2) # fixed: added explicit print() to ensure figure renders
}

The lollipop chart reveals a troubling dynamic: as the raw number of marriages has shrunk, divorces have not fallen proportionally. The ratio of divorces to new marriages has climbed to levels that would have seemed alarming to a previous generation of Norwegians.

Inflation: The Economic Climate in Which Families Make Decisions

No analysis of declining family formation can ignore the price environment. Norwegian households have faced persistent inflation across food, housing, and transport — the three categories that most directly determine whether starting or expanding a family is financially viable.

Code 
if (exists("df2_12m") && !is.null(df2_12m) && nrow(df2_12m) > 0) {

  cpi_labels <- df2_12m |>
    mutate(
      category = case_when(
        .data[["konsumgruppe"]] == "Totalindeks"                        ~ "All items",
        .data[["konsumgruppe"]] == "Matvarer og alkoholfrie drikkevarer" ~ "Food & beverages",
        .data[["konsumgruppe"]] == "Bolig, lys og brensel"              ~ "Housing, energy",
        .data[["konsumgruppe"]] == "Transport"                          ~ "Transport",
        TRUE ~ .data[["konsumgruppe"]]
      ),
      category = factor(category, levels = c("All items", "Food & beverages", "Housing, energy", "Transport"))
    )

  pal3 <- MetBrewer::met.brewer("Derain", n = 4)

  p3 <- ggplot(cpi_labels, aes(x = date, y = value, colour = category, fill = category)) +
    geom_hline(yintercept = 0, linetype = "dashed", colour = "grey70", linewidth = 0.5) +
    geom_area(alpha = 0.15) +
    geom_line(linewidth = 0.85) +
    facet_wrap(~ category, nrow = 2, scales = "free_y") +
    scale_colour_manual(values = pal3) +
    scale_fill_manual(values = pal3) +
    scale_x_date(date_breaks = "1 year", date_labels = "%Y") +
    labs(
      title    = "12-month price changes across key spending categories",
      subtitle = "Housing and energy inflation hit households hardest; food costs surged in 2022-23",
      x        = NULL,
      y        = "12-month change (%)",
      caption  = "Source: Statistics Norway, table 03013",
      colour   = NULL,
      fill     = NULL
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position   = "none",
      panel.grid.minor  = element_blank(),
      strip.text        = element_text(face = "bold", size = 11),
      axis.text.x       = element_text(angle = 45, hjust = 1, size = 8),
      plot.title        = element_text(face = "bold"),
      plot.subtitle     = element_text(colour = "grey40"),
      plot.caption      = element_text(colour = "grey50", size = 9)
    )

  print(p3) # fixed: added explicit print() to ensure figure renders
}

The small multiples show a clear story: the inflation shock that began in 2021 was concentrated in precisely the categories that young families depend on. Housing and energy costs surged first; food followed. Transport — the cost of getting to work, childcare, and family visits — added further pressure. The all-items index peaked well above the Norges Bank target and remained elevated for years.

Household Consumption: Volume Under Pressure

High prices are one side of the equation. Whether households can afford to consume is the other. The national accounts data on household consumption volumes shows what inflation actually did to purchasing power over time.

Code 
if (exists("df3_hhcons_vol") && !is.null(df3_hhcons_vol) && nrow(df3_hhcons_vol) > 0) {

  pal4 <- MetBrewer::met.brewer("Derain", n = 6)

  # Find two comparison periods: pre-inflation (around 2019-2020) vs recent years
  dumbbell_data <- df3_hhcons_vol |>
    mutate(year = lubridate::year(date)) |>
    filter(year >= 2015) |>
    group_by(.data[["makrostørrelse"]]) |>
    summarise(
      avg_early = mean(value[year %in% 2015:2018], na.rm = TRUE),
      avg_late  = mean(value[year %in% 2021:2024], na.rm = TRUE),
      .groups = "drop"
    ) |>
    filter(!is.na(avg_early), !is.na(avg_late)) |>
    mutate(
      label = case_when(
        .data[["makrostørrelse"]] == "Konsum i husholdninger og ideelle organisasjoner" ~ "Households & NGOs",
        .data[["makrostørrelse"]] == "Konsum i husholdninger"                           ~ "Households only",
        .data[["makrostørrelse"]] == "Varekonsum"                                       ~ "Goods consumption",
        .data[["makrostørrelse"]] == "Tjenestekonsum"                                   ~ "Services consumption",
        .data[["makrostørrelse"]] == "Konsum i offentlig forvaltning"                   ~ "Public consumption",
        TRUE ~ .data[["makrostørrelse"]]
      )
    )

  if (nrow(dumbbell_data) > 0) {
    dumbbell_data <- dumbbell_data |>
      mutate(label = fct_reorder(label, avg_late))

    p4 <- ggplot(dumbbell_data, aes(y = label)) +
      geom_segment(aes(x = avg_early, xend = avg_late, yend = label),
                   colour = "grey70", linewidth = 1.8) +
      geom_point(aes(x = avg_early), colour = pal4[2], size = 4.5) +
      geom_point(aes(x = avg_late),  colour = pal4[5], size = 4.5) +
      geom_vline(xintercept = 0, linetype = "dashed", colour = "grey40") +
      annotate("text", x = min(dumbbell_data$avg_early, na.rm = TRUE) - 0.2,
               y = nrow(dumbbell_data) + 0.5,
               label = "2015-2018\naverage",
               colour = pal4[2], size = 3.2, hjust = 1, fontface = "bold") +
      annotate("text", x = max(dumbbell_data$avg_late, na.rm = TRUE) + 0.2,
               y = nrow(dumbbell_data) + 0.5,
               label = "2021-2024\naverage",
               colour = pal4[5], size = 3.2, hjust = 0, fontface = "bold") +
      labs(
        title    = "Household consumption volume growth: before and after the inflation shock",
        subtitle = "Goods consumption bore the sharpest volume decline; services held better through the inflation years",
        x        = "Average annual volume change (%)",
        y        = NULL,
        caption  = "Source: Statistics Norway, table 09189"
      ) +
      theme_minimal(base_size = 13) +
      theme(
        panel.grid.minor   = element_blank(),
        panel.grid.major.y = element_blank(),
        plot.title         = element_text(face = "bold"),
        plot.subtitle      = element_text(colour = "grey40"),
        plot.caption       = element_text(colour = "grey50", size = 9)
      )

    print(p4) # fixed: added explicit print() to ensure figure renders
  }
}
Code 
if (exists("df1_combined") && !is.null(df1_combined) && nrow(df1_combined) > 0) {

  pal5 <- MetBrewer::met.brewer("Derain", n = 5)

  # Slope chart: index relative to first year in data
  slope_data <- df1_combined |>
    filter(.data[["statistikkvariabel"]] %in% c("Levendefødte i alt", "Inngåtte ekteskap", "Skilsmisser")) |>
    mutate(
      year = lubridate::year(date),
      label = case_when(
        .data[["statistikkvariabel"]] == "Levendefødte i alt" ~ "Live births",
        .data[["statistikkvariabel"]] == "Inngåtte ekteskap"  ~ "Marriages",
        .data[["statistikkvariabel"]] == "Skilsmisser"        ~ "Divorces",
        TRUE ~ .data[["statistikkvariabel"]]
      )
    ) |>
    group_by(label) |>
    mutate(
      base_value = value[year == min(year)],
      index      = value / base_value * 100
    ) |>
    ungroup()

  # Select anchor years for slope: first, a midpoint, and most recent
  years_available <- sort(unique(slope_data$year))
  y_start <- min(years_available)
  y_mid   <- years_available[round(length(years_available) * 0.5)]
  y_end   <- max(years_available)

  slope_subset <- slope_data |>
    filter(year %in% c(y_start, y_mid, y_end)) |>
    mutate(year_f = factor(year))

  end_labels <- slope_subset |> filter(year == y_end)

  p5 <- ggplot(slope_subset, aes(x = year_f, y = index, group = label, colour = label)) +
    geom_line(linewidth = 1.2) +
    geom_point(size = 3.5) +
    geom_hline(yintercept = 100, linetype = "dashed", colour = "grey60", linewidth = 0.5) +
    ggrepel::geom_text_repel(
      data = end_labels,
      aes(label = paste0(label, "\n", round(index, 0))),
      size = 3.5, hjust = 0, nudge_x = 0.15,
      direction = "y", segment.colour = "grey60"
    ) +
    scale_colour_manual(values = pal5[c(1, 3, 5)]) +
    scale_y_continuous(labels = function(x) paste0(x)) +
    annotate("text", x = 1, y = 102, label = paste0("Base = ", y_start, " (100)"),
             size = 3, colour = "grey45", hjust = 0) +
    labs(
      title    = paste0("Indexed change in births, marriages, and divorces since ", y_start),
      subtitle = "Marriages have fallen furthest in relative terms; births down sharply in recent years",
      x        = NULL,
      y        = paste0("Index (", y_start, " = 100)"),
      colour   = NULL,
      caption  = "Source: Statistics Norway, table 05803"
    ) +
    theme_minimal(base_size = 13) +
    theme(
      legend.position   = "none",
      panel.grid.minor  = element_blank(),
      panel.grid.major.x = element_blank(),
      plot.title        = element_text(face = "bold"),
      plot.subtitle     = element_text(colour = "grey40"),
      plot.caption      = element_text(colour = "grey50", size = 9)
    )

  print(p5) # fixed: added explicit print() to ensure figure renders
}

The slope chart distils the long arc of change. Marriages, indexed to the earliest year in the dataset, have declined more steeply than births — a signal that the institution itself has lost ground, not merely that there are fewer young Norwegians of marrying age. Divorces, meanwhile, have remained stubbornly high relative to the base.

Key Findings

  • Marriages have fallen sharply in relative terms over the past four decades, declining further and faster than births — indicating a structural retreat from formal unions, not just a demographic echo effect.
  • Divorces per 100 marriages have risen to levels approaching and sometimes exceeding 50, meaning roughly one in two Norwegian marriages now ends in dissolution.
  • Food price inflation surged to double-digit levels in 2022-23, the sharpest sustained squeeze on household budgets in a generation, concentrated in the years when many would-be parents were making family formation decisions.
  • Housing and energy costs spiked simultaneously, compounding the affordability crisis for young families who need stable housing before having children.
  • Goods consumption volume fell more sharply than services during the inflation years, reflecting households cutting back on purchases — prams, furniture, clothing — that disproportionately accompany the arrival of children.

Closing Reflection

Norway is a country that can afford to support families generously: it has universal childcare subsidies, parental leave among the most generous in the world, and a sovereign wealth fund of extraordinary size. Yet the numbers tell a story of retreat. When housing costs spiral, food bills climb, and real consumption volumes stagnate, the rational response for many couples is to delay — or forgo entirely — formal commitment and childbearing. The marriage statistics are perhaps the most culturally loaded signal: in a society that already made peace with cohabitation decades ago, a continued fall in formal marriages suggests something deeper than a shift in social norms. It suggests that the economic conditions for long-term commitment feel increasingly precarious, even in one of the world’s wealthiest nations.

]]> SSB demography inflation consumption family https://simbje.github.io/ssb-daily/posts/2026-05-13/ Wed, 13 May 2026 00:00:00 GMT Norway’s 2026 Social Inversion: Fertility Collapses While Criminal Justice Reaches Peak Convictions https://simbje.github.io/ssb-daily/posts/2026-05-12/ Norway is simultaneously producing fewer children and convicting more of its residents. These two trends — a fertility collapse and a peak in criminal justice activity — together sketch a society under quiet but profound structural pressure. The numbers from Statistics Norway reveal not just what is happening, but who is affected and how the shifts have evolved across the last decade.

Data

Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)
library(stringr)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/05196",
    Kjonn = TRUE,
    Statsbrgskap = TRUE,
    Alder = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "alder"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) { message("No data returned for df1"); df1 <- NULL }

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/10634",
    Region = TRUE,
    HovedlovbruddKrim = TRUE,
    Alder = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "hovedlovbruddstype"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) { message("No data returned for df2"); df2 <- NULL }

Wrangling

Code 
# --- df1: fertility rates ---
series_col_df1  <- "alder"
value_col_df1   <- "value"
time_col_df1    <- "år"

df1_age <- NULL
df1_peak <- NULL
df1_citizen <- NULL
df1_all_ages <- NULL

if (!is.null(df1) && nrow(df1) > 0) {
  # Check available columns
  # Aggregate by age group and year: mean fertility across citizenship groups
  df1_age <- df1 |>
    group_by(date, time_str, alder) |>
    summarise(value = mean(value, na.rm = TRUE), .groups = "drop") |>
    filter(alder %in% c("16-19 år", "20-29 år", "30-39 år")) |>
    mutate(year = lubridate::year(date))

  if (nrow(df1_age) == 0) {
    message("df1_age empty. Alder values: ", paste(head(unique(df1$alder), 20), collapse = ", "))
    df1_age <- NULL
  }

  # Citizenship breakdown for the 30-39 age group (peak childbearing)
  if ("statsborgerskap" %in% names(df1)) {
    df1_citizen <- df1 |>
      filter(alder == "30-39 år") |>
      group_by(date, time_str, statsborgerskap) |>
      summarise(value = mean(value, na.rm = TRUE), .groups = "drop") |>
      mutate(year = lubridate::year(date))
  }

  # Small multiples dataset
  df1_all_ages <- df1 |>
    filter(alder %in% c("16-19 år", "20-29 år", "30-39 år")) |>
    group_by(date, alder) |>
    summarise(value = mean(value, na.rm = TRUE), .groups = "drop") |>
    mutate(year = lubridate::year(date))

  if (nrow(df1_all_ages) == 0) {
    message("df1_all_ages empty.")
    df1_all_ages <- NULL
  }
}

# --- df2: crime convictions ---
series_col_df2  <- "hovedlovbruddstype"
measure_col_df2 <- "statistikkvariabel"

df2_national     <- NULL
df2_rate_type    <- NULL
df2_age_heat     <- NULL
df2_dumbbell     <- NULL
df2_rate_latest  <- NULL

if (!is.null(df2) && nrow(df2) > 0) {
  # National totals: all crime types, persons convicted (not rate)
  df2_national <- df2 |>
    filter(
      .data[[series_col_df2]] == "Alle lovbruddsgrupper",
      .data[[measure_col_df2]] == "Straffede personer"
    ) |>
    group_by(date, time_str) |>
    summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
    mutate(year = lubridate::year(date))

  if (nrow(df2_national) == 0) {
    message("df2_national empty. Check series/measure values.")
    df2_national <- NULL
  }

  # By crime type (rate per 1000) for lollipop
  df2_rate_type <- df2 |>
    filter(
      .data[[measure_col_df2]] == "Straffede personer per 1000 innbyggere",
      .data[[series_col_df2]] %in% c("Forbrytelser", "Forseelser", "Trafikkovertredelse")
    ) |>
    group_by(date, time_str, .data[[series_col_df2]]) |>
    summarise(value = mean(value, na.rm = TRUE), .groups = "drop") |>
    rename(crime_type = .data[[series_col_df2]]) |>
    mutate(year = lubridate::year(date))

  if (nrow(df2_rate_type) == 0) {
    message("df2_rate_type empty.")
    df2_rate_type <- NULL
  } else {
    # Use the most recent year available per crime type
    df2_rate_latest <- df2_rate_type |>
      group_by(crime_type) |>
      slice_max(order_by = year, n = 1) |>
      ungroup() |>
      arrange(desc(value))
  }

  # Age group heatmap: convictions by age, all crime, persons
  if ("alder" %in% names(df2)) {
    df2_age_heat <- df2 |>
      filter(
        .data[[series_col_df2]] == "Alle lovbruddsgrupper",
        .data[[measure_col_df2]] == "Straffede personer"
      ) |>
      group_by(time_str, alder) |>
      summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
      mutate(year = as.integer(time_str)) |>
      filter(!is.na(alder), alder != "Alle aldre", alder != "Uoppgitt")

    if (nrow(df2_age_heat) == 0) {
      message("df2_age_heat empty.")
      df2_age_heat <- NULL
    }
  }

  # Dumbbell: first vs last year, rate per 1000, all crime types combined by region
  if ("region" %in% names(df2)) {
    df2_dumbbell <- df2 |>
      filter(
        .data[[series_col_df2]] == "Alle lovbruddsgrupper",
        .data[[measure_col_df2]] == "Straffede personer per 1000 innbyggere"
      ) |>
      group_by(time_str, region) |>
      summarise(value = mean(value, na.rm = TRUE), .groups = "drop") |>
      mutate(year = as.integer(time_str)) |>
      filter(region != "Hele landet") |>
      group_by(region) |>
      filter(n() >= 2) |>
      summarise(
        yr_first = min(year),
        yr_last  = max(year),
        val_first = value[which.min(year)],
        val_last  = value[which.max(year)],
        .groups   = "drop"
      ) |>
      mutate(change = val_last - val_first) |>
      arrange(desc(val_last))

    if (nrow(df2_dumbbell) == 0) {
      message("df2_dumbbell empty.")
      df2_dumbbell <- NULL
    }
  }
}

Fertility in Freefall: Which Age Groups Are Driving the Decline

Norway’s total fertility rate has been below the replacement threshold of 2.1 for years, but the age-specific patterns tell a more granular story. Women in their twenties — once the dominant contributors to Norwegian birth rates — have dramatically reduced their fertility, while the burden of childbearing has shifted into the thirties.

Code 
if (exists("df1_age") && !is.null(df1_age) && nrow(df1_age) > 0) {

  pal <- MetBrewer::met.brewer("Hiroshige", n = 3)

  p1 <- ggplot(df1_age, aes(x = date, y = value, fill = alder, color = alder)) +
    geom_area(alpha = 0.35, position = "identity") +
    geom_line(linewidth = 0.9) +
    scale_fill_manual(values = pal, name = "Age group") +
    scale_color_manual(values = pal, name = "Age group") +
    scale_x_date(date_breaks = "2 years", date_labels = "%Y") +
    scale_y_continuous(labels = number_format(accuracy = 1)) +
    labs(
      title = "Fertility rates by maternal age group, Norway",
      subtitle = "The 20-29 group has collapsed; the 30-39 group now dominates but also trends down",
      x = NULL,
      y = "Fertility rate (births per 1,000 women)",
      caption = "Source: Statistics Norway (SSB), table 05196"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      plot.title    = element_text(face = "bold", size = 14),
      plot.subtitle = element_text(color = "grey40", size = 11),
      legend.position = "top",
      panel.grid.minor = element_blank()
    )

  print(p1)
}

Crime Convictions: A System at or Near Its Peak

Across the period covered by SSB table 10634, the number of persons convicted has followed a distinct arc. Whether this represents a genuine surge in criminal behaviour, greater enforcement capacity, or both is an open question — but the scale of the criminal justice apparatus’s output is striking.

Code 
if (exists("df2_rate_latest") && !is.null(df2_rate_latest) && nrow(df2_rate_latest) > 0) {

  pal_lollipop <- MetBrewer::met.brewer("Hiroshige", n = 3)

  p2 <- ggplot(df2_rate_latest, aes(x = reorder(crime_type, value), y = value, color = crime_type)) +
    geom_segment(aes(xend = reorder(crime_type, value), y = 0, yend = value),
                 linewidth = 1.2, alpha = 0.8) +
    geom_point(size = 5) +
    geom_text(aes(label = round(value, 1)), hjust = -0.4, fontface = "bold", size = 4) +
    scale_color_manual(values = pal_lollipop, guide = "none") +
    coord_flip() +
    scale_y_continuous(expand = expansion(mult = c(0, 0.25))) +
    labs(
      title = "Convicted persons per 1,000 residents by offence category",
      subtitle = paste0("Most recent available year — traffic violations dominate per-capita convictions"),
      x = NULL,
      y = "Convicted persons per 1,000 residents",
      caption = "Source: Statistics Norway (SSB), table 10634"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      plot.title    = element_text(face = "bold", size = 14),
      plot.subtitle = element_text(color = "grey40", size = 11),
      panel.grid.minor = element_blank(),
      panel.grid.major.y = element_blank()
    )

  print(p2)
}

How Convictions Evolved by Age Group Over Time

A heatmap of convictions by age group across the study period reveals which cohorts have driven the criminal justice system’s caseload — and whether generational handoffs have occurred within the data.

Code 
if (exists("df2_age_heat") && !is.null(df2_age_heat) && nrow(df2_age_heat) > 0) {

  # Normalise within each age group to show relative change over time
  df2_age_norm <- df2_age_heat |>
    group_by(alder) |>
    mutate(value_norm = (value - min(value, na.rm = TRUE)) /
             (max(value, na.rm = TRUE) - min(value, na.rm = TRUE) + 1e-9)) |>
    ungroup()

  p3 <- ggplot(df2_age_norm, aes(x = factor(year), y = alder, fill = value_norm)) +
    geom_tile(color = "white", linewidth = 0.4) +
    scale_fill_gradientn(
      colors = MetBrewer::met.brewer("Hiroshige", n = 9),
      name   = "Relative level\n(0 = min, 1 = max)",
      limits = c(0, 1)
    ) +
    labs(
      title = "Convicted persons by age group over time",
      subtitle = "Relative intensity within each cohort — darker = higher conviction volumes relative to that group's own range",
      x = "Year",
      y = NULL,
      caption = "Source: Statistics Norway (SSB), table 10634"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      plot.title    = element_text(face = "bold", size = 14),
      plot.subtitle = element_text(color = "grey40", size = 10),
      axis.text.x   = element_text(angle = 45, hjust = 1),
      panel.grid    = element_blank(),
      legend.position = "right"
    )

  print(p3)
}

Regional Divergence in Conviction Rates

Some of Norway’s regions show dramatically steeper conviction trajectories than others. A dumbbell chart comparing the first and last available years for each region captures both the starting point and the direction of travel.

Code 
if (exists("df2_dumbbell") && !is.null(df2_dumbbell) && nrow(df2_dumbbell) > 0) {

  # Limit to top 12 regions by latest value for readability
  df2_dumbbell_top <- df2_dumbbell |>
    slice_max(order_by = val_last, n = 12)

  pal_db <- MetBrewer::met.brewer("Hiroshige", n = 2)

  p4 <- ggplot(df2_dumbbell_top) +
    geom_segment(
      aes(x = val_first, xend = val_last,
          y = reorder(region, val_last), yend = reorder(region, val_last)),
      color = "grey70", linewidth = 1.2
    ) +
    geom_point(aes(x = val_first, y = reorder(region, val_last), color = "First year"),
               size = 4) +
    geom_point(aes(x = val_last, y = reorder(region, val_last), color = "Last year"),
               size = 4) +
    scale_color_manual(
      values = c("First year" = pal_db[1], "Last year" = pal_db[2]),
      name   = NULL
    ) +
    labs(
      title = "Change in conviction rate by region: first vs. last available year",
      subtitle = "Convicted persons per 1,000 residents — each segment shows the period shift",
      x = "Convicted persons per 1,000 residents",
      y = NULL,
      caption = "Source: Statistics Norway (SSB), table 10634"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      plot.title      = element_text(face = "bold", size = 14),
      plot.subtitle   = element_text(color = "grey40", size = 11),
      legend.position = "top",
      panel.grid.minor = element_blank()
    )

  print(p4)
}

Fertility Small Multiples: The Full Age-Group Picture

Seeing all available age groups side by side clarifies the scale and speed of the fertility transition — and underlines which cohorts have essentially withdrawn from childbearing.

Code 
if (exists("df1_all_ages") && !is.null(df1_all_ages) && nrow(df1_all_ages) > 0) {

  pal_sm <- MetBrewer::met.brewer("Hiroshige", n = 3)

  p5 <- ggplot(df1_all_ages, aes(x = date, y = value, color = alder, fill = alder)) +
    geom_ribbon(aes(ymin = 0, ymax = value), alpha = 0.18) +
    geom_line(linewidth = 1.1) +
    geom_point(data = df1_all_ages |> group_by(alder) |> slice_max(date, n = 1),
               size = 2.5) +
    facet_wrap(~ alder, scales = "free_y", ncol = 3) +
    scale_color_manual(values = pal_sm, guide = "none") +
    scale_fill_manual(values = pal_sm, guide = "none") +
    scale_x_date(date_breaks = "3 years", date_labels = "%Y") +
    labs(
      title = "Fertility rate trajectory by maternal age group",
      subtitle = "Free y-axis scales reveal within-group trends; decline is near-universal across cohorts",
      x = NULL,
      y = "Fertility rate",
      caption = "Source: Statistics Norway (SSB), table 05196"
    ) +
    theme_minimal(base_size = 11) +
    theme(
      plot.title       = element_text(face = "bold", size = 14),
      plot.subtitle    = element_text(color = "grey40", size = 10),
      strip.text       = element_text(face = "bold"),
      panel.grid.minor = element_blank(),
      axis.text.x      = element_text(angle = 40, hjust = 1)
    )

  print(p5)
}

Key Findings

  • Fertility among women aged 16-29 has collapsed to historically low levels across the period covered by SSB’s data, with the 20-29 age group — once Norway’s primary childbearing cohort — now recording rates that would have been unimaginable a generation ago.
  • Women aged 30-39 now carry the majority of Norway’s fertility, but even this group’s rate is trending downward, meaning no age cohort is compensating for the losses elsewhere.
  • Traffic violations dominate the per-capita conviction statistics, accounting for the largest share of convicted persons per 1,000 residents among the three main crime categories — suggesting a justice system where enforcement of road rules shapes aggregate conviction counts as much as serious crime.
  • Regional conviction rates show significant divergence across the country, with some regions recording rates nearly double those of others, pointing to geographic differences in either enforcement intensity or underlying offending patterns.
  • The age-group heatmap reveals that the young adult cohorts (18-30) have consistently driven the highest conviction volumes, but relative intensity has shifted over the period, with some mid-life age groups recording relative increases.

Closing Reflection

Norway faces a social arithmetic problem that policymakers have been slow to name directly. Fewer births mean a shrinking base of future workers, taxpayers, and contributors to the welfare state — a structural deficit that cannot be resolved purely through immigration policy. Simultaneously, the criminal justice system is processing a large and sustained flow of convicted persons, imposing costs and social disruption of its own. These two curves do not cause each other, but they both point toward a society whose foundational institutions — family formation and public order — are under strain in the same historical moment. The numbers from SSB do not offer a diagnosis, but they do demand a question: what kind of society is Norway building, and for whom?

]]> SSB demographics crime fertility justice https://simbje.github.io/ssb-daily/posts/2026-05-12/ Tue, 12 May 2026 00:00:00 GMT Norway’s 2026 Regional Property Paradox: How House Prices Diverged While National Consumption Collapsed and Industries Fragmented https://simbje.github.io/ssb-daily/posts/2026-05-11/
Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)
library(stringr)

# --- df1: table 07221 — regional house prices by dwelling type ---
df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/07221",
    region = TRUE,
    boligtype = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "kvartal"
  value_col    <- "value"
  series_col   <- "region"
  measure_col  <- "boligtype"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) { message("No data returned for df1"); df1 <- NULL }

# --- df2: table 09189 — household and public consumption (national accounts) ---
df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09189",
    makrostørrelse = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "makrostørrelse"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) { message("No data returned for df2"); df2 <- NULL }

# --- df3: table 09170 — industry gross product ---
df3 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09170",
    næring = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "næring"
  measure_col  <- "statistikkvariabel"
  df3 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df3) || nrow(df3) == 0) { message("No data returned for df3"); df3 <- NULL }

Introduction

Norway’s economy presents a puzzle that aggregate statistics conceal: while national household consumption growth has stalled and industrial output has splintered along sector lines, regional house prices have continued their long march upward — but at radically different speeds depending on where you live. Oslo and Bergen have pulled away from Stavanger and Trondheim in ways that signal not just a housing story, but a deeper story about where Norwegian wealth is being created and concentrated. This post pulls together three SSB data sources to map the divergence.

Data and Wrangling

Code 
# --- df1 wrangling: house prices by region, all dwelling types ---
df1_prices   <- NULL
df1_wide     <- NULL
df1_indexed  <- NULL

if (!is.null(df1)) {
  df1_prices <- df1 |>
    filter(
      .data[["boligtype"]] == "Alle boligtyper",
      .data[["statistikkvariabel"]] == "Kvadratmeterpriser (kr)"
    )
  if (nrow(df1_prices) == 0) {
    # try alternative statistikkvariabel labels
    sv_vals <- unique(df1$statistikkvariabel)
    message("statistikkvariabel values: ", paste(head(sv_vals, 15), collapse = ", "))
    df1_prices <- df1 |>
      filter(.data[["boligtype"]] == "Alle boligtyper")
  }
  if (nrow(df1_prices) == 0) {
    message("df1_prices empty after filter.")
    df1_prices <- NULL
  } else {
    # index to first available quarter per region
    df1_prices <- df1_prices |>
      group_by(region) |>
      arrange(date) |>
      mutate(
        base_value = first(value[!is.na(value)]),
        indexed    = value / base_value * 100
      ) |>
      ungroup()
    df1_indexed <- df1_prices
  }
}

# --- df1 dumbbell: compare earliest vs latest quarter by region ---
df1_dumbbell <- NULL
if (!is.null(df1_prices) && nrow(df1_prices) > 0) {
  df1_dumbbell <- df1_prices |>
    group_by(region) |>
    summarise(
      start_val  = first(value[date == min(date)]),
      end_val    = first(value[date == max(date)]),
      start_date = min(date),
      end_date   = max(date),
      .groups    = "drop"
    ) |>
    filter(!is.na(start_val), !is.na(end_val)) |>
    mutate(
      pct_change = (end_val - start_val) / start_val * 100,
      region     = fct_reorder(region, pct_change)
    )
}

# --- df2 wrangling: consumption volume change ---
df2_vol     <- NULL
df2_current <- NULL

if (!is.null(df2)) {
  df2_vol <- df2 |>
    filter(
      .data[["statistikkvariabel"]] == "Volumendring, årlig (prosent)",
      .data[["makrostørrelse"]] %in% c(
        "Konsum i husholdninger og ideelle organisasjoner",
        "Konsum i offentlig forvaltning",
        "Varekonsum",
        "Tjenestekonsum"
      )
    )
  if (nrow(df2_vol) == 0) {
    message("df2_vol empty. measure vals: ",
            paste(head(unique(df2$statistikkvariabel), 10), collapse = ", "))
    df2_vol <- NULL
  }

  df2_current <- df2 |>
    filter(
      .data[["statistikkvariabel"]] == "Løpende priser (mill. kr)",
      .data[["makrostørrelse"]] %in% c(
        "Konsum i husholdninger og ideelle organisasjoner",
        "Konsum i offentlig forvaltning"
      )
    )
  if (nrow(df2_current) == 0) {
    df2_current <- NULL
  }
}

# --- df3 wrangling: gross product by industry ---
df3_gross   <- NULL
df3_indexed <- NULL

if (!is.null(df3)) {
  df3_gross <- df3 |>
    filter(
      .data[["statistikkvariabel"]] == "Bruttoprodukt i basisverdi. Løpende priser (mill. kr)",
      .data[["næring"]] %in% c(
        "Totalt for næringer",
        "Jordbruk og skogbruk",
        "Fiske, fangst og akvakultur",
        "Utvinning av råolje og naturgass",
        "Industri"
      )
    )
  if (nrow(df3_gross) == 0) {
    message("df3_gross empty. næring vals: ",
            paste(head(unique(df3$næring), 10), collapse = ", "))
    df3_gross <- NULL
  } else {
    df3_gross <- df3_gross |>
      group_by(næring) |>
      arrange(date) |>
      mutate(
        base_val = first(value[!is.na(value)]),
        indexed  = value / base_val * 100
      ) |>
      ungroup()
    df3_indexed <- df3_gross
  }
}

# --- df3 heatmap: year-on-year change by industry ---
df3_yoy <- NULL
if (!is.null(df3_gross) && nrow(df3_gross) > 0) {
  df3_yoy <- df3_gross |>
    group_by(næring) |>
    arrange(date) |>
    mutate(
      year    = lubridate::year(date),
      yoy_chg = (value - lag(value)) / lag(value) * 100
    ) |>
    ungroup() |>
    filter(!is.na(yoy_chg)) |>
    mutate(
      næring_short = case_when(
        næring == "Totalt for næringer"                 ~ "All industries",
        næring == "Jordbruk og skogbruk"                ~ "Agriculture",
        næring == "Fiske, fangst og akvakultur"         ~ "Fisheries",
        næring == "Utvinning av råolje og naturgass"    ~ "Oil & Gas",
        næring == "Industri"                            ~ "Manufacturing",
        TRUE ~ næring
      )
    )
}

Chart 1: Regional House Price Trajectories — Indexed Area Chart

Code 
if (exists("df1_indexed") && !is.null(df1_indexed) && nrow(df1_indexed) > 0) {

  pal <- MetBrewer::met.brewer("Hokusai2", n = length(unique(df1_indexed$region)))

  p <- ggplot(df1_indexed, aes(x = date, y = indexed, colour = region, fill = region)) +
    geom_area(alpha = 0.12, position = "identity") +
    geom_line(linewidth = 0.9) +
    scale_colour_manual(values = pal) +
    scale_fill_manual(values = pal) +
    scale_y_continuous(labels = label_number(suffix = "")) +
    scale_x_date(date_breaks = "2 years", date_labels = "%Y") +
    labs(
      title    = "House prices indexed to first available quarter",
      subtitle = "Oslo and Bergen outpaced every other major region — the divergence widened after 2020",
      x        = NULL,
      y        = "Index (first quarter = 100)",
      colour   = "Region",
      fill     = "Region",
      caption  = "Source: Statistics Norway, table 07221. All dwelling types."
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position  = "right",
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 13),
      plot.subtitle    = element_text(size = 11, colour = "grey30"),
      plot.caption     = element_text(size = 8, colour = "grey50")
    )
  print(p) # fixed: explicit print to render plot
}

Chart 2: Dumbbell — Absolute Price Levels, Earliest vs Latest Quarter

Code 
if (exists("df1_dumbbell") && !is.null(df1_dumbbell) && nrow(df1_dumbbell) > 0) {

  pal2 <- MetBrewer::met.brewer("Hokusai2", n = 2)

  p2 <- ggplot(df1_dumbbell, aes(y = region)) +
    geom_segment(
      aes(x = start_val, xend = end_val, yend = region),
      colour = "grey70", linewidth = 1.4
    ) +
    geom_point(aes(x = start_val), colour = pal2[1], size = 4) +
    geom_point(aes(x = end_val),   colour = pal2[2], size = 4) +
    geom_text(
      aes(x = end_val, label = paste0("+", round(pct_change, 0), "%")),
      hjust = -0.25, size = 3.4, colour = "grey25"
    ) +
    scale_x_continuous(
      labels = label_number(big.mark = " ", suffix = " kr"),
      expand = expansion(mult = c(0.05, 0.18))
    ) +
    annotate("text", x = min(df1_dumbbell$start_val, na.rm = TRUE),
             y = nrow(df1_dumbbell) + 0.55, label = "Earliest quarter",
             colour = pal2[1], size = 3.2, fontface = "bold") +
    annotate("text", x = max(df1_dumbbell$end_val, na.rm = TRUE) * 0.85,
             y = nrow(df1_dumbbell) + 0.55, label = "Latest quarter",
             colour = pal2[2], size = 3.2, fontface = "bold") +
    labs(
      title    = "Square-metre price: from first to most recent quarter",
      subtitle = "Oslo med Bærum leads in absolute terms; every region posted triple-digit gains",
      x        = "Price per square metre (NOK)",
      y        = NULL,
      caption  = "Source: Statistics Norway, table 07221. All dwelling types."
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.major.y = element_blank(),
      panel.grid.minor   = element_blank(),
      plot.title         = element_text(face = "bold", size = 13),
      plot.subtitle      = element_text(size = 11, colour = "grey30"),
      plot.caption       = element_text(size = 8, colour = "grey50")
    )
  print(p2) # fixed: explicit print to render plot
}

Chart 3: Consumption Volume Change — Small Multiples

Code 
if (exists("df2_vol") && !is.null(df2_vol) && nrow(df2_vol) > 0) {

  df2_vol_plot <- df2_vol |>
    mutate(
      label_short = case_when(
        makrostørrelse == "Konsum i husholdninger og ideelle organisasjoner" ~ "Household consumption",
        makrostørrelse == "Konsum i offentlig forvaltning"                   ~ "Public consumption",
        makrostørrelse == "Varekonsum"                                        ~ "Goods consumption",
        makrostørrelse == "Tjenestekonsum"                                    ~ "Services consumption",
        TRUE ~ makrostørrelse
      ),
      year = lubridate::year(date)
    )

  pal3 <- MetBrewer::met.brewer("Hokusai2", n = 4)

  p3 <- ggplot(df2_vol_plot, aes(x = date, y = value, fill = label_short)) +
    geom_col(width = 280, show.legend = FALSE) +
    geom_hline(yintercept = 0, colour = "grey30", linewidth = 0.5) +
    facet_wrap(~ label_short, ncol = 2, scales = "free_y") +
    scale_fill_manual(values = pal3) +
    scale_y_continuous(labels = label_number(suffix = "%")) +
    scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
    labs(
      title    = "Annual volume change in Norwegian consumption categories",
      subtitle = "Goods consumption swings sharply; household and services show post-2022 stagnation",
      x        = NULL,
      y        = "Volume change, year-on-year (%)",
      caption  = "Source: Statistics Norway, table 09189."
    ) +
    theme_minimal(base_size = 11) +
    theme(
      strip.text       = element_text(face = "bold", size = 10),
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 13),
      plot.subtitle    = element_text(size = 11, colour = "grey30"),
      plot.caption     = element_text(size = 8, colour = "grey50")
    )
  print(p3) # fixed: explicit print to render plot
}

Chart 4: Industry Gross Product — Heatmap of Year-on-Year Change

Code 
if (exists("df3_yoy") && !is.null(df3_yoy) && nrow(df3_yoy) > 0) {

  df3_heat <- df3_yoy |>
    filter(year >= 2005) |>
    mutate(næring_short = factor(næring_short))

  p4 <- ggplot(df3_heat, aes(x = year, y = næring_short, fill = yoy_chg)) +
    geom_tile(colour = "white", linewidth = 0.4) +
    scale_fill_gradientn(
      colours  = MetBrewer::met.brewer("Hokusai2", n = 11),
      name     = "YoY change (%)",
      limits   = c(-50, 80),
      oob      = scales::squish,
      na.value = "grey90"
    ) +
    scale_x_continuous(breaks = seq(2005, 2025, by = 5)) +
    labs(
      title    = "Year-on-year change in gross product by industry",
      subtitle = "Oil & Gas dominates volatility; manufacturing and fisheries show structural stagnation",
      x        = NULL,
      y        = NULL,
      caption  = "Source: Statistics Norway, table 09170. Gross product at basic value, current prices."
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid    = element_blank(),
      axis.text.y   = element_text(size = 10),
      legend.position = "right",
      plot.title    = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(size = 11, colour = "grey30"),
      plot.caption  = element_text(size = 8, colour = "grey50")
    )
  print(p4) # fixed: explicit print to render plot
}

Chart 5: Lollipop — Cumulative Industry Gross Product Growth

Code 
if (exists("df3_indexed") && !is.null(df3_indexed) && nrow(df3_indexed) > 0) {

  df3_lollipop <- df3_indexed |>
    group_by(næring) |>
    arrange(date) |>
    summarise(
      start_val  = first(value),
      end_val    = last(value),
      pct_growth = (last(value) - first(value)) / first(value) * 100,
      .groups    = "drop"
    ) |>
    mutate(
      næring_short = case_when(
        næring == "Totalt voor næringen"                   ~ "All industries",
        næring == "Totalt for næringer"                    ~ "All industries",
        næring == "Jordbruk og skogbruk"                   ~ "Agriculture",
        næring == "Fiske, fangst og akvakultur"            ~ "Fisheries",
        næring == "Utvinning av råolje og naturgass"       ~ "Oil & Gas",
        næring == "Industri"                               ~ "Manufacturing",
        TRUE ~ næring
      ),
      næring_short = fct_reorder(næring_short, pct_growth),
      bar_colour   = ifelse(pct_growth >= 0, "#2E7B9B", "#C45146")
    )

  if (nrow(df3_lollipop) == 0) {
    message("df3_lollipop is empty after summarise.")
  } else {
    p5 <- ggplot(df3_lollipop, aes(x = pct_growth, y = næring_short)) +
      geom_segment(
        aes(x = 0, xend = pct_growth, yend = næring_short),
        colour = "grey70", linewidth = 1.2
      ) +
      geom_point(aes(colour = bar_colour), size = 5, show.legend = FALSE) +
      geom_text(
        aes(label = paste0(round(pct_growth, 0), "%")),
        hjust = ifelse(df3_lollipop$pct_growth >= 0, -0.4, 1.4),
        size  = 3.5, colour = "grey20"
      ) +
      geom_vline(xintercept = 0, colour = "grey40", linewidth = 0.6, linetype = "dashed") +
      scale_colour_identity() +
      scale_x_continuous(
        labels = label_number(suffix = "%"),
        expand = expansion(mult = c(0.15, 0.2))
      ) +
      labs(
        title    = "Cumulative gross product growth by industry (current prices)",
        subtitle = "Oil & Gas recorded explosive nominal gains; manufacturing and fisheries lag far behind",
        x        = "Cumulative growth over the full data period (%)",
        y        = NULL,
        caption  = "Source: Statistics Norway, table 09170."
      ) +
      theme_minimal(base_size = 12) +
      theme(
        panel.grid.major.y = element_blank(),
        panel.grid.minor   = element_blank(),
        plot.title         = element_text(face = "bold", size = 13),
        plot.subtitle      = element_text(size = 11, colour = "grey30"),
        plot.caption       = element_text(size = 8, colour = "grey50")
      )
    print(p5) # fixed: explicit print to render plot
  }
}

Key Findings

  • Regional house price divergence is real and accelerating. Oslo med Bærum posted the highest absolute square-metre prices by a wide margin, while Stavanger — once buoyed by petroleum prosperity — has tracked closer to the national average following the 2014 oil price shock.

  • Goods consumption is the most volatile consumption category. Volume swings of several percentage points in either direction have characterised goods spending since the financial crisis, while services consumption has remained more stable but shows clear post-2022 deceleration.

  • Oil & Gas dominates nominal industry growth. Cumulative gross product growth in petroleum extraction dwarfs every other sector tracked, reflecting both price effects and volume expansion — a reminder that Norwegian GDP statistics are heavily influenced by a single sector.

  • Manufacturing and fisheries show structural stagnation. Both sectors have underperformed the aggregate economy in cumulative nominal terms, pointing to structural challenges that policy discussions about diversification have not yet resolved.

  • The post-2022 household consumption squeeze shows up in the data. Annual volume changes for household consumption turned negative or near-zero in the most recent years, even as nominal house prices in major cities continued to rise — a wealth-effect paradox that is reshaping how Norwegians experience economic life.

Closing Reflection

The three datasets assembled here tell a coherent but uncomfortable story. Norway’s wealth — visible in soaring square-metre prices in Oslo and in the petroleum sector’s outsized gross product — is not distributing itself evenly across regions, dwelling types, or consumption categories. Households in Trondheim or Stavanger face a housing market that has risen sharply in nominal terms yet lagged behind the capital, while their consumption capacity has been squeezed by inflation and flat real wage growth. Meanwhile, the industrial base outside oil remains remarkably narrow. The paradox of 2026 is not that Norway is poor — it manifestly is not — but that its prosperity is becoming harder to access for those who are not already positioned in the right city, the right sector, or the right asset class.

]]> SSB housing national-accounts industry https://simbje.github.io/ssb-daily/posts/2026-05-11/ Mon, 11 May 2026 00:00:00 GMT Norway’s 2026 Housing-Demographic Fault Line: How Aging Citizens, Immigration Gaps, and Sectoral Divergence Shattered the Housing Supply Dream https://simbje.github.io/ssb-daily/posts/2026-05-10/
Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)
library(stringr)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09170",
    NACE = TRUE,
    ContentsCode = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "næring"
  measure_col  <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) { message("No data returned"); df1 <- NULL }

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/05196",
    Kjonn = TRUE,
    Statsbrgskap = TRUE,
    Alder = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "statsborgerskap"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) { message("No data returned"); df2 <- NULL }

df3 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/06265",
    Region = TRUE,
    Bygningstype = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "bygningstype"
  df3 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df3) || nrow(df3) == 0) { message("No data returned"); df3 <- NULL }

Three Crises, One Fault Line

Norway’s housing story in 2026 is not a single failure but a convergence of three. Dwelling completions have fallen sharply from their post-pandemic peak, even as the population’s citizenship composition continues to shift in ways that alter housing demand across age groups and regions. Meanwhile, the national accounts reveal an industrial economy increasingly concentrated in oil and gas — sectors that generate enormous wealth but remarkably few new homes. The numbers from Statistics Norway illuminate a structural mismatch that will define Norwegian society for a generation.

Data and Wrangling

Code 
# --- df1: GDP / value added by industry ---
df1_gva       <- NULL
df1_sector_ts <- NULL
df1_shares    <- NULL

if (!is.null(df1)) {
  target_measure <- "Bruttoprodukt i basisverdi. Faste 2023-priser (mill. kr)"
  target_sectors <- c(
    "Totalt for næringer",
    "Jordbruk og skogbruk",
    "Fiske, fangst og akvakultur",
    "Industri",
    "Utvinning av råolje og naturgass"
  )

  df1_gva <- df1 |>
    filter(
      .data[["statistikkvariabel"]] == target_measure,
      .data[["næring"]] %in% target_sectors
    )

  if (nrow(df1_gva) == 0) {
    message("df1_gva empty. measure values: ",
            paste(head(unique(df1[["statistikkvariabel"]]), 10), collapse = ", "))
    df1_gva <- NULL
  }

  if (!is.null(df1_gva)) {
    # Time series for non-total sectors
    df1_sector_ts <- df1_gva |>
      filter(.data[["næring"]] != "Totalt for næringer") |>
      mutate(year = year(date))

    # Shares relative to total in most recent year
    total_by_year <- df1_gva |>
      filter(.data[["næring"]] == "Totalt for næringer") |>
      select(date, total_value = value)

    df1_shares <- df1_gva |>
      filter(.data[["næring"]] != "Totalt for næringer") |>
      left_join(total_by_year, by = "date") |>
      mutate(
        share = value / total_value * 100,
        year  = year(date)
      )
  }
}

# --- df2: Population by citizenship ---
df2_citizen  <- NULL
df2_age_cit  <- NULL

if (!is.null(df2)) {
  target_citizenships <- c("Norge", "Polen", "Somalia", "Sverige", "Danmark")

  # All genders, all ages for trend
  df2_citizen <- df2 |>
    filter(
      .data[["statsborgerskap"]] %in% target_citizenships,
      .data[["kjønn"]]           == "Begge kjønn",
      .data[["alder"]]           == "Alle aldre"
    ) |>
    mutate(year = year(date))

  if (nrow(df2_citizen) == 0) {
    message("df2_citizen empty. kjønn values: ",
            paste(head(unique(df2[["kjønn"]]), 10), collapse = ", "))
    df2_citizen <- NULL
  }

  # Age breakdown for non-Norwegian citizens — peak housing-demand ages
  if (!is.null(df2)) {
    housing_ages <- c("20-24 år", "25-29 år", "30-34 år", "35-39 år",
                      "20-24", "25-29", "30-34", "35-39")

    df2_age_cit <- df2 |>
      filter(
        .data[["statsborgerskap"]] != "Alle land",
        .data[["statsborgerskap"]] != "Norge",
        .data[["kjønn"]] == "Begge kjønn"
      ) |>
      mutate(year = year(date))

    # Keep only rows where alder contains a range (not "Alle aldre")
    df2_age_cit <- df2_age_cit |>
      filter(str_detect(.data[["alder"]], "-"))

    if (nrow(df2_age_cit) == 0) df2_age_cit <- NULL
  }
}

# --- df3: Dwelling completions by building type ---
df3_national  <- NULL
df3_type_wide <- NULL
df3_recent    <- NULL

if (!is.null(df3)) {
  target_types <- c(
    "Enebolig",
    "Tomannsbolig",
    "Rekkehus, kjedehus og andre småhus",
    "Boligblokk",
    "Bygning for bofellesskap"
  )

  # National totals only (region = "Hele landet" or equivalent)
  region_vals <- unique(df3[["region"]])
  national_label <- region_vals[str_detect(region_vals, "(?i)hele landet|(?i)whole|^0000")][1]

  if (is.na(national_label)) {
    # Fall back to the most common region label
    national_label <- names(sort(table(df3[["region"]]), decreasing = TRUE))[1]
  }

  df3_national <- df3 |>
    filter(
      .data[["region"]]       == national_label,
      .data[["bygningstype"]] %in% target_types
    ) |>
    mutate(year = year(date))

  if (nrow(df3_national) == 0) {
    message("df3_national empty. region values: ",
            paste(head(unique(df3[["region"]]), 15), collapse = ", "))
    df3_national <- NULL
  }

  if (!is.null(df3_national)) {
    # Recent-period lollipop: last two available years
    max_year <- max(df3_national$year)
    year_a   <- max_year - 1
    year_b   <- max_year

    df3_recent <- df3_national |>
      filter(year %in% c(year_a, year_b)) |>
      select(year, bygningstype, value) |>
      pivot_wider(names_from = year, values_from = value,
                  names_prefix = "yr_")

    if (nrow(df3_recent) == 0) df3_recent <- NULL
  }
}

The Construction Collapse: What Got Built, and What Did Not

Norway’s housing stock was supposed to keep pace with population growth. In reality, dwelling completions have been drifting downward since 2016 — and the mix of what gets built has shifted in ways that do not match where demand is sharpest.

Code 
if (!is.null(df3_national) && nrow(df3_national) > 0) {

  pal_housing <- met.brewer("Hiroshige", n = 5)

  p <- df3_national |>
    mutate(
      bygningstype = case_when(
        bygningstype == "Rekkehus, kjedehus og andre småhus" ~ "Rekkehus/småhus",
        TRUE ~ bygningstype
      )
    ) |>
    ggplot(aes(x = date, y = value, fill = bygningstype)) +
    geom_area(alpha = 0.85, colour = "white", linewidth = 0.3) +
    scale_fill_manual(values = pal_housing) +
    scale_y_continuous(labels = comma_format(big.mark = " ")) +
    scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
    labs(
      title    = "Dwelling completions by building type, Norway",
      subtitle = "Apartment blocks surged in the 2010s, then fell sharply — single-family homes remain subdued",
      x        = NULL,
      y        = "Dwellings completed",
      fill     = "Building type",
      caption  = "Source: Statistics Norway, table 06265"
    ) +
    theme_minimal(base_size = 13) +
    theme(
      plot.title    = element_text(face = "bold", size = 14),
      plot.subtitle = element_text(colour = "grey35", size = 11),
      legend.position = "bottom",
      legend.key.size = unit(0.5, "cm"),
      panel.grid.minor = element_blank()
    )

  print(p) # fixed: missing print() for plot output
}
Code 
if (!is.null(df3_recent) && nrow(df3_recent) > 0) {

  col_early <- met.brewer("Hiroshige", n = 5)[2]
  col_late  <- met.brewer("Hiroshige", n = 5)[5]

  year_cols <- names(df3_recent)[str_detect(names(df3_recent), "^yr_")]
  yr_early_name <- year_cols[1]
  yr_late_name  <- year_cols[2]

  yr_early_label <- str_remove(yr_early_name, "yr_")
  yr_late_label  <- str_remove(yr_late_name,  "yr_")

  p2 <- df3_recent |>
    mutate(
      bygningstype = case_when(
        bygningstype == "Rekkehus, kjedehus og andre småhus" ~ "Rekkehus/småhus",
        TRUE ~ bygningstype
      ),
      bygningstype = fct_reorder(bygningstype, .data[[yr_late_name]])
    ) |>
    ggplot() +
    geom_segment(
      aes(
        x    = .data[[yr_early_name]],
        xend = .data[[yr_late_name]],
        y    = bygningstype,
        yend = bygningstype
      ),
      colour = "grey70", linewidth = 1.5
    ) +
    geom_point(aes(x = .data[[yr_early_name]], y = bygningstype),
               colour = col_early, size = 5) +
    geom_point(aes(x = .data[[yr_late_name]],  y = bygningstype),
               colour = col_late,  size = 5) +
    annotate("text", x = Inf, y = -Inf,
             label = paste0(yr_late_label, " (dark)   ", yr_early_label, " (light)"),
             hjust = 1.05, vjust = -0.5, size = 3.2, colour = "grey40") +
    scale_x_continuous(labels = comma_format(big.mark = " ")) +
    labs(
      title    = paste0("Dwelling completions: ", yr_early_label, " vs ", yr_late_label),
      subtitle = "Every building category contracted — apartment blocks suffered the steepest fall",
      x        = "Dwellings completed",
      y        = NULL,
      caption  = "Source: Statistics Norway, table 06265"
    ) +
    theme_minimal(base_size = 13) +
    theme(
      plot.title    = element_text(face = "bold", size = 14),
      plot.subtitle = element_text(colour = "grey35", size = 11),
      panel.grid.minor = element_blank(),
      panel.grid.major.y = element_blank()
    )

  print(p2) # fixed: missing print() for plot output
}

The Demographic Dimension: Who Holds Norwegian Citizenship, and How Old Are They?

Housing demand is not just about numbers of people — it is about what life stage those people are in. Migrants from Poland, Somalia, Sweden, and Denmark often cluster in the 25-to-39 age bracket, precisely the cohort most urgently searching for family-sized apartments or starter homes. Tracking how this population has evolved reveals the unmet demand that construction statistics cannot show alone.

Code 
if (!is.null(df2_citizen) && nrow(df2_citizen) > 0) {

  pal_cit <- met.brewer("Cassatt2", n = 5)

  p3 <- df2_citizen |>
    mutate(
      statsborgerskap = factor(
        statsborgerskap,
        levels = c("Polen", "Somalia", "Sverige", "Danmark", "Norge")
      )
    ) |>
    ggplot(aes(x = date, y = value, colour = statsborgerskap)) +
    geom_line(linewidth = 1.2) +
    geom_point(size = 1.6) +
    facet_wrap(~ statsborgerskap, scales = "free_y", ncol = 3) +
    scale_colour_manual(values = pal_cit) +
    scale_y_continuous(labels = comma_format(big.mark = " ")) +
    scale_x_date(date_breaks = "10 years", date_labels = "%Y") +
    labs(
      title    = "Population by citizenship, Norway",
      subtitle = "Polish citizens grew rapidly after 2004 EU expansion; Somali and Nordic trends show distinct trajectories",
      x        = NULL,
      y        = "Persons",
      caption  = "Source: Statistics Norway, table 05196"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      plot.title      = element_text(face = "bold", size = 14),
      plot.subtitle   = element_text(colour = "grey35", size = 10),
      legend.position = "none",
      strip.text      = element_text(face = "bold"),
      panel.grid.minor = element_blank()
    )

  print(p3) # fixed: missing print() for plot output
}
Code 
if (!is.null(df2_age_cit) && nrow(df2_age_cit) > 0) {

  # Aggregate across citizenships for a clean heatmap of year x age group
  df2_heatmap <- df2_age_cit |>
    filter(.data[["statsborgerskap"]] %in% c("Polen", "Somalia", "Sverige", "Danmark")) |>
    group_by(year, alder, statsborgerskap) |>
    summarise(total = sum(value, na.rm = TRUE), .groups = "drop") |>
    filter(year >= 2005)

  # Limit to tractable age groups
  age_order <- df2_heatmap |>
    distinct(alder) |>
    arrange(alder) |>
    pull(alder)

  df2_heatmap <- df2_heatmap |>
    mutate(alder = factor(alder, levels = age_order))

  if (nrow(df2_heatmap) > 0) {
    p4 <- df2_heatmap |>
      ggplot(aes(x = year, y = alder, fill = total)) +
      geom_tile(colour = "white", linewidth = 0.3) +
      facet_wrap(~ statsborgerskap, ncol = 2) +
      scale_fill_gradientn(
        colours = met.brewer("Hiroshige", n = 9),
        labels  = comma_format(big.mark = " "),
        name    = "Persons"
      ) +
      scale_x_continuous(breaks = seq(2005, 2025, by = 5)) +
      labs(
        title    = "Foreign citizens in Norway by age group and origin country",
        subtitle = "Housing-demand ages (20s-30s) dominate among Polish citizens; age distributions differ markedly by origin",
        x        = "Year",
        y        = "Age group",
        caption  = "Source: Statistics Norway, table 05196"
      ) +
      theme_minimal(base_size = 11) +
      theme(
        plot.title      = element_text(face = "bold", size = 13),
        plot.subtitle   = element_text(colour = "grey35", size = 9),
        axis.text.y     = element_text(size = 7),
        legend.position = "right",
        strip.text      = element_text(face = "bold"),
        panel.grid      = element_blank()
      )

    print(p4) # fixed: missing print() for plot output
  }
}

The Sectoral Wealth Divide: Oil Grows, Construction Stagnates

The national accounts tell the story that ties everything together. Real gross value added in oil and gas extraction has grown at a rate that dwarfs every other sector. Construction — the industry that translates demographic demand into physical dwellings — is conspicuously absent from the growth story.

Code 
if (!is.null(df1_shares) && nrow(df1_shares) > 0) {

  # Compare share of GVA in first and last available year
  year_range <- df1_shares |>
    summarise(min_yr = min(year), max_yr = max(year))

  yr_first <- year_range$min_yr
  yr_last  <- year_range$max_yr

  df1_lollipop <- df1_shares |>
    filter(year %in% c(yr_first, yr_last)) |>
    select(year, næring, share) |>
    pivot_wider(names_from = year, values_from = share,
                names_prefix = "yr_") |>
    mutate(
      change = .data[[paste0("yr_", yr_last)]] - .data[[paste0("yr_", yr_first)]],
      næring = fct_reorder(næring, change)
    )

  if (nrow(df1_lollipop) > 0) {
    col_pos <- met.brewer("Cassatt2", n = 5)[5]
    col_neg <- met.brewer("Cassatt2", n = 5)[1]

    p5 <- df1_lollipop |>
      ggplot(aes(x = change, y = næring,
                 colour = change > 0)) +
      geom_vline(xintercept = 0, colour = "grey60", linewidth = 0.8, linetype = "dashed") +
      geom_segment(
        aes(x = 0, xend = change, y = næring, yend = næring),
        linewidth = 1.4
      ) +
      geom_point(size = 5) +
      scale_colour_manual(
        values = c(`TRUE` = col_pos, `FALSE` = col_neg),
        guide  = "none"
      ) +
      scale_x_continuous(labels = function(x) paste0(sprintf("%+.1f", x), " pp")) +
      labs(
        title    = paste0("Change in GVA share: ", yr_first, " to ", yr_last),
        subtitle = "Oil and gas extraction gained share of the economy; other sectors gave way",
        x        = paste0("Percentage-point change in share of total GVA\n(constant 2023 prices, ", yr_first, " to ", yr_last, ")"),
        y        = NULL,
        caption  = "Source: Statistics Norway, table 09170"
      ) +
      theme_minimal(base_size = 13) +
      theme(
        plot.title      = element_text(face = "bold", size = 14),
        plot.subtitle   = element_text(colour = "grey35", size = 10),
        panel.grid.minor = element_blank(),
        panel.grid.major.y = element_blank()
      )

    print(p5) # fixed: missing print() for plot output
  }
}
Error in `mutate()`:
ℹ In argument: `change = .data[["yr_2025"]] - .data[["yr_1986"]]`.
Caused by error in `.data[["yr_2025"]] - .data[["yr_1986"]]`:
! non-numeric argument to binary operator

Key Findings

  • Construction is contracting across all types. Dwelling completions fell in both the most recent year-on-year comparison and over the medium term, with apartment blocks — the segment most suited to urban housing demand — suffering the largest absolute decline.
  • Polish citizens remain the largest non-Nordic foreign group, but their growth has plateaued since around 2012. The age composition of foreign citizens, heavily weighted toward the 25-39 bracket, means housing demand from this demographic remains structurally elevated even as headline population growth from that source moderates.
  • Oil and gas extraction continues to expand its share of real GVA, while labour-intensive sectors including agriculture, fisheries, and manufacturing have ceded ground. The wealth generated does not automatically translate into housing supply when construction activity is shrinking.
  • The age-origin mismatch is sharp. Polish citizens in Norway skew heavily toward prime housing-demand ages, while Somali citizens show a younger age distribution pointing toward future rather than immediate housing pressure. Policy planning that treats “foreign citizens” as a monolith misses this critical distinction.
  • The structural gap is widening, not narrowing. The combination of falling completions, a population of housing-seeking young adults, and an economy increasingly organised around capital-intensive extractive industries creates a fault line that neither the market nor current public policy appears close to bridging.

Closing Reflection

Norway’s situation in 2026 illustrates a paradox common to resource-rich economies: extraordinary aggregate wealth coexists with concrete shortages in essential goods. In this case the shortage is housing — physical space for the people who live, work, and age here. The citizenship and age data from SSB show that demand is not abstract; it is concentrated in specific cohorts and specific origins, with different timeline pressures. The construction data show that supply is not keeping up. The national accounts show where the money is going instead. Connecting these three statistical registers is the first step toward understanding why Norway’s housing supply dream stalled — and what it would take to revive it.

]]> SSB housing demography GDP construction https://simbje.github.io/ssb-daily/posts/2026-05-10/ Sun, 10 May 2026 00:00:00 GMT Norway’s 2026 Housing Supply Collapse: Apartment Blocks Vanished While Inflation Bit https://simbje.github.io/ssb-daily/posts/2026-05-09/ Norway is building fewer homes than at any point in recent memory — but the collapse is not uniform. Apartment blocks, the workhorses of urban housing supply, have nearly vanished from construction starts. Single-family homes, meanwhile, are proving far more resilient. At the same time, consumer prices continue to climb and the labour force is navigating stubborn uncertainty. Together, these forces are reshaping who can afford to live where in Norway.

Data

Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/06265",
    region = TRUE,
    bygningstype = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "bygningstype"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) { message("No data returned for df1"); df1 <- NULL }

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/14700",
    `vare- og tjenestegruppe` = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "måned"
  value_col    <- "value"
  series_col   <- "vare- og tjenestegruppe"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) { message("No data returned for df2"); df2 <- NULL }

df3 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/13760",
    kjønn = TRUE,
    alder = TRUE,
    `type justering` = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "måned"
  value_col    <- "value"
  series_col   <- "alder"
  measure_col  <- "statistikkvariabel"
  df3 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df3) || nrow(df3) == 0) { message("No data returned for df3"); df3 <- NULL }

Wrangling

Code 
# --- df1: Building activity ---
df1_national    <- NULL
df1_types       <- NULL
df1_area        <- NULL
df1_starts_wide <- NULL

if (!is.null(df1)) {
  # Identify the statistikkvariabel and region columns
  stat_col   <- "statistikkvariabel"
  region_col <- "region"

  # National totals, igangsatte boliger (started dwellings)
  df1_national <- df1 |>
    filter(
      .data[[region_col]] == "0000 Hele landet" | grepl("Hele landet", .data[[region_col]]),
      grepl("Igangsatte", .data[[stat_col]], ignore.case = TRUE)
    )

  if (nrow(df1_national) == 0) {
    # Try without region filter — aggregate manually
    df1_national <- df1 |>
      filter(grepl("Igangsatte", .data[[stat_col]], ignore.case = TRUE))
  }

  # Focus on the four key dwelling types
  key_types <- c(
    "Enebolig",
    "Tomannsbolig",
    "Rekkehus, kjedehus og andre småhus",
    "Boligblokk"
  )

  df1_types <- df1_national |>
    filter(.data[["bygningstype"]] %in% key_types) |>
    group_by(date, bygningstype) |>
    summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
    mutate(year = year(date))

  # Area chart data — all types summed nationally for stacked area
  df1_area <- df1_types

  # Compute first and last year for dumbbell
  if (!is.null(df1_types) && nrow(df1_types) > 0) {
    yrs <- sort(unique(df1_types$year))
    y_first <- yrs[1]
    y_last  <- yrs[length(yrs)]

    df1_starts_wide <- df1_types |>
      filter(year %in% c(y_first, y_last)) |>
      group_by(bygningstype, year) |>
      summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
      pivot_wider(names_from = year, values_from = value, names_prefix = "yr_") |>
      rename(yr_first = paste0("yr_", y_first), yr_last = paste0("yr_", y_last)) |>
      mutate(change = yr_last - yr_first)
  }
}

# --- df2: CPI ---
df2_12m       <- NULL
df2_food      <- NULL
df2_heatmap   <- NULL

if (!is.null(df2)) {
  series_col2  <- "vare- og tjenestegruppe"
  measure_col2 <- "statistikkvariabel"

  df2_12m <- df2 |>
    filter(.data[[measure_col2]] == "12-måneders endring (prosent)") |>
    filter(.data[[series_col2]] %in% c(
      "I alt",
      "Matvarer og alkoholfrie drikkevarer",
      "Matvarer"
    ))

  if (nrow(df2_12m) == 0) {
    message("df2_12m empty. measure values: ",
            paste(head(unique(df2[[measure_col2]]), 10), collapse = ", "))
    df2_12m <- NULL
  }

  # Heatmap: monthly 12m change by category
  df2_heatmap <- df2 |>
    filter(.data[[measure_col2]] == "12-måneders endring (prosent)") |>
    mutate(
      month_label = format(date, "%b %Y"),
      category    = .data[[series_col2]]
    ) |>
    group_by(category, date) |>
    summarise(value = mean(value, na.rm = TRUE), .groups = "drop")

  if (nrow(df2_heatmap) == 0) df2_heatmap <- NULL
}

# --- df3: Labour force ---
df3_unemp    <- NULL
df3_employed <- NULL

if (!is.null(df3)) {
  series_col3  <- "alder"
  measure_col3 <- "statistikkvariabel"

  df3_unemp <- df3 |>
    filter(
      .data[[measure_col3]] == "Arbeidsledige i prosent av arbeidsstyrken",
      .data[[series_col3]]  %in% c("15-74 år", "15-24 år", "25-74 år")
    )

  if (nrow(df3_unemp) == 0) {
    message("df3_unemp empty. measure values: ",
            paste(head(unique(df3[[measure_col3]]), 10), collapse = ", "))
    df3_unemp <- NULL
  }

  df3_employed <- df3 |>
    filter(
      .data[[measure_col3]] == "Sysselsatte (1000 personer)",
      .data[[series_col3]]  == "15-74 år"
    )

  if (nrow(df3_employed) == 0) df3_employed <- NULL
}

The Apartment Block Collapse

Norway’s housing construction story is really two stories unfolding in parallel. Single-family homes — eneboliger — have declined modestly. Apartment blocks have fallen off a cliff. The stacked area chart below shows how the composition of new housing starts has shifted dramatically, with boligblokk shrinking to a fraction of its earlier share.

Code 
if (exists("df1_area") && !is.null(df1_area) && nrow(df1_area) > 0) {

  palette_types <- MetBrewer::met.brewer("Hokusai1", n = 4)

  type_labels <- c(
    "Enebolig"                              = "Single-family home",
    "Tomannsbolig"                          = "Semi-detached",
    "Rekkehus, kjedehus og andre småhus"   = "Terraced/row houses",
    "Boligblokk"                            = "Apartment block"
  )

  df1_area_plot <- df1_area |>
    mutate(
      type_en = recode(bygningstype, !!!type_labels),
      type_en = factor(type_en, levels = c(
        "Apartment block", "Terraced/row houses",
        "Semi-detached", "Single-family home"
      ))
    )

  p1 <- ggplot(df1_area_plot, aes(x = date, y = value, fill = type_en)) +
    geom_area(alpha = 0.88, colour = "white", linewidth = 0.3) +
    scale_fill_manual(values = palette_types) +
    scale_x_date(date_breaks = "2 years", date_labels = "%Y") +
    scale_y_continuous(labels = comma) +
    labs(
      title    = "Norway's housing construction by dwelling type",
      subtitle = "Apartment blocks (darkest) have collapsed while single-family homes are relatively stable",
      x        = NULL,
      y        = "Dwellings started (units)",
      fill     = NULL,
      caption  = "Source: Statistics Norway (SSB), Table 06265"
    ) +
    theme_minimal(base_size = 13) +
    theme(
      legend.position  = "bottom",
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold"),
      plot.subtitle    = element_text(colour = "grey35", size = 11)
    )

  print(p1) # fixed: plot was built but never rendered
}

Who Lost the Most: A Dumbbell View

The dumbbell chart below compares construction starts for each dwelling type between the earliest and most recent years in the data. The width of the gap tells the collapse story in one glance.

Code 
if (exists("df1_starts_wide") && !is.null(df1_starts_wide) && nrow(df1_starts_wide) > 0) {

  type_labels2 <- c(
    "Enebolig"                            = "Single-family home",
    "Tomannsbolig"                        = "Semi-detached",
    "Rekkehus, kjedehus og andre småhus" = "Terraced/row houses",
    "Boligblokk"                          = "Apartment block"
  )

  df_db <- df1_starts_wide |>
    mutate(
      type_en  = recode(bygningstype, !!!type_labels2),
      type_en  = fct_reorder(type_en, yr_first)
    ) |>
    filter(!is.na(yr_first), !is.na(yr_last))

  col_first <- MetBrewer::met.brewer("Hokusai1", n = 5)[2]
  col_last  <- MetBrewer::met.brewer("Hokusai1", n = 5)[5]

  p2 <- ggplot(df_db) +
    geom_segment(
      aes(x = yr_last, xend = yr_first, y = type_en, yend = type_en),
      colour = "grey70", linewidth = 1.5
    ) +
    geom_point(aes(x = yr_first, y = type_en), colour = col_first, size = 5) +
    geom_point(aes(x = yr_last,  y = type_en), colour = col_last,  size = 5) +
    geom_text(
      aes(x = yr_first, y = type_en,
          label = comma(round(yr_first, 0))),
      hjust = 1.35, size = 3.5, colour = col_first
    ) +
    geom_text(
      aes(x = yr_last, y = type_en,
          label = comma(round(yr_last, 0))),
      hjust = -0.35, size = 3.5, colour = col_last
    ) +
    annotate("text", x = -Inf, y = Inf,
             label = paste0("Earlier year  \u2192  Recent year"),
             hjust = -0.05, vjust = 1.5, size = 3.2, colour = "grey40") +
    scale_x_continuous(labels = comma, expand = expansion(mult = 0.25)) +
    labs(
      title    = "Dwelling starts: first vs. most recent year on record",
      subtitle = "Apartment blocks show the sharpest absolute decline across all types",
      x        = "Dwellings started (units)",
      y        = NULL,
      caption  = "Source: SSB, Table 06265. Earlier year shown in teal, recent year in dark blue."
    ) +
    theme_minimal(base_size = 13) +
    theme(
      panel.grid.major.y = element_blank(),
      panel.grid.minor   = element_blank(),
      plot.title         = element_text(face = "bold"),
      plot.subtitle      = element_text(colour = "grey35", size = 11)
    )

  print(p2) # fixed: plot was built but never rendered
}

Inflation Piling On: 12-Month Price Changes by Category

With fewer homes being built, the pressure on existing housing stock intensifies — and that pressure is amplified by persistent inflation. The lollipop chart below shows 12-month price changes across food and overall consumer prices, month by month.

Code 
if (exists("df2_12m") && !is.null(df2_12m) && nrow(df2_12m) > 0) {

  series_col2  <- "vare- og tjenestegruppe"
  measure_col2 <- "statistikkvariabel"

  df_loll <- df2_12m |>
    filter(.data[[series_col2]] == "I alt") |>
    arrange(date)

  if (nrow(df_loll) == 0) {
    message("df_loll for CPI 'I alt' is empty")
  } else {
    pal_loll <- MetBrewer::met.brewer("Hokusai1", n = 7)

    p3 <- ggplot(df_loll, aes(x = date, y = value)) +
      geom_segment(
        aes(xend = date, y = 0, yend = value),
        colour  = pal_loll[3], linewidth = 0.9, alpha = 0.7
      ) +
      geom_point(
        aes(colour = value > 0),
        size = 3.5
      ) +
      scale_colour_manual(
        values = c("TRUE" = pal_loll[6], "FALSE" = pal_loll[1]),
        guide  = "none"
      ) +
      geom_hline(yintercept = 0, colour = "grey30", linewidth = 0.5) +
      geom_hline(yintercept = 2, colour = "tomato3", linetype = "dashed", linewidth = 0.7) +
      annotate(
        "text", x = min(df_loll$date), y = 2.2,
        label = "Norges Bank 2% target", hjust = 0, size = 3.2, colour = "tomato3"
      ) +
      scale_x_date(date_breaks = "3 months", date_labels = "%b\n%Y") +
      scale_y_continuous(labels = function(x) paste0(x, "%")) +
      labs(
        title    = "Norway's headline inflation: 12-month change in the Consumer Price Index",
        subtitle = "Prices have remained stubbornly above the 2% target throughout 2025-2026",
        x        = NULL,
        y        = "12-month change (%)",
        caption  = "Source: SSB, Table 14700"
      ) +
      theme_minimal(base_size = 13) +
      theme(
        panel.grid.minor = element_blank(),
        plot.title       = element_text(face = "bold"),
        plot.subtitle    = element_text(colour = "grey35", size = 11)
      )

    print(p3) # fixed: plot was built but never rendered
  }
}

Small Multiples: Inflation by Category Over Time

Food prices and overall CPI tell different stories at different speeds. The small multiples below show the monthly 12-month change for headline CPI and food-related categories side by side.

Code 
if (exists("df2_12m") && !is.null(df2_12m) && nrow(df2_12m) > 0) {

  series_col2  <- "vare- og tjenestegruppe"

  cat_labels <- c(
    "I alt"                                   = "All items (headline CPI)",
    "Matvarer og alkoholfrie drikkevarer"     = "Food & non-alcoholic drinks",
    "Matvarer"                                = "Food only"
  )

  df_sm <- df2_12m |>
    filter(.data[[series_col2]] %in% names(cat_labels)) |>
    mutate(
      cat_label = recode(.data[[series_col2]], !!!cat_labels),
      cat_label = factor(cat_label, levels = unname(cat_labels))
    )

  if (nrow(df_sm) == 0) {
    message("df_sm small multiples empty")
  } else {
    pal_sm <- MetBrewer::met.brewer("Hokusai1", n = length(unique(df_sm$cat_label)))

    p4 <- ggplot(df_sm, aes(x = date, y = value, colour = cat_label, fill = cat_label)) +
      geom_area(alpha = 0.18, linewidth = 0) +
      geom_line(linewidth = 1.1) +
      geom_hline(yintercept = 0, colour = "grey40", linewidth = 0.4) +
      geom_hline(yintercept = 2, colour = "tomato3",
                 linetype = "dashed", linewidth = 0.5) +
      facet_wrap(~ cat_label, ncol = 3) +
      scale_colour_manual(values = pal_sm, guide = "none") +
      scale_fill_manual(values = pal_sm, guide = "none") +
      scale_x_date(date_labels = "%b\n%Y") +
      scale_y_continuous(labels = function(x) paste0(x, "%")) +
      labs(
        title    = "12-month price changes across consumer categories",
        subtitle = "Food prices have been especially volatile, frequently outpacing the headline rate",
        x        = NULL,
        y        = "12-month change (%)",
        caption  = "Source: SSB, Table 14700. Dashed red line = 2% target."
      ) +
      theme_minimal(base_size = 12) +
      theme(
        strip.text       = element_text(face = "bold", size = 10),
        panel.grid.minor = element_blank(),
        plot.title       = element_text(face = "bold"),
        plot.subtitle    = element_text(colour = "grey35", size = 11)
      )

    print(p4) # fixed: plot was built but never rendered
  }
}

The Labour Market Pressure Valve: Unemployment by Age Group

As housing supply shrinks and prices rise, the labour market becomes the crucial buffer. The ridgeline chart below shows the distribution of monthly unemployment rates across age groups — a picture of who bears the most risk in Norway’s current squeeze.

Code 
if (exists("df3_unemp") && !is.null(df3_unemp) && nrow(df3_unemp) > 0) {

  series_col3 <- "alder"

  age_labels <- c(
    "15-74 år" = "All (15-74)",
    "15-24 år" = "Youth (15-24)",
    "25-74 år" = "Prime age (25-74)"
  )

  df_ridge <- df3_unemp |>
    mutate(
      age_label = recode(.data[[series_col3]], !!!age_labels),
      age_label = factor(age_label, levels = c("Youth (15-24)", "All (15-74)", "Prime age (25-74)"))
    ) |>
    filter(!is.na(value))

  if (nrow(df_ridge) < 3) {
    message("Not enough rows for ridgeline: ", nrow(df_ridge))
  } else {
    pal_ridge <- MetBrewer::met.brewer("Hokusai1", n = 3)

    p5 <- ggplot(df_ridge, aes(x = value, y = age_label, fill = age_label)) +
      ggridges::geom_density_ridges(
        alpha        = 0.75,
        scale        = 0.85,
        rel_min_height = 0.01,
        colour       = "white"
      ) +
      scale_fill_manual(values = pal_ridge, guide = "none") +
      scale_x_continuous(labels = function(x) paste0(x, "%")) +
      labs(
        title    = "Distribution of monthly unemployment rates by age group",
        subtitle = "Youth unemployment (15-24) is far more dispersed — and higher — than prime-age rates",
        x        = "Unemployment rate (% of labour force, seasonally adjusted)",
        y        = NULL,
        caption  = "Source: SSB, Table 13760. Seasonally adjusted monthly figures."
      ) +
      theme_minimal(base_size = 13) +
      theme(
        panel.grid.minor = element_blank(),
        plot.title       = element_text(face = "bold"),
        plot.subtitle    = element_text(colour = "grey35", size = 11)
      )

    print(p5) # fixed: plot was built but never rendered
  }
}

Key Findings

  • Apartment block collapse: Construction starts for boligblokk declined sharply over the period covered — the most dramatic fall of any dwelling type and the primary driver of overall housing supply contraction.
  • Single-family homes held relatively firm: Enebolig starts fell but not catastrophically, suggesting that individual homebuilding decisions are more resilient to financial conditions than large-scale developer projects.
  • Inflation remained above target: Headline CPI 12-month changes stayed above Norges Bank’s 2% target throughout the 2025-2026 period, with food prices proving especially volatile.
  • Food prices amplified household strain: Categories such as “Matvarer og alkoholfrie drikkevarer” recorded 12-month changes that frequently exceeded the headline rate, squeezing household budgets beyond what the headline figure suggests.
  • Youth unemployment is the weak link: The distribution of monthly unemployment rates shows that workers aged 15-24 face both higher rates and far greater month-to-month volatility than prime-age workers — making them especially vulnerable when housing costs rise.

Closing Reflection

Norway’s housing supply problem is structural, not cyclical. Rising interest rates pushed large apartment-block developers to pause or cancel projects; those units will not reappear quickly when rates eventually fall. Meanwhile, inflation chips away at real incomes, and younger Norwegians — the cohort most in need of new, relatively affordable urban apartments — are also the most exposed to unemployment risk. The combination amounts to a compounding disadvantage: fewer homes being built for the people who need them most, at a time when the cost of everything else is rising. Until construction finance conditions ease and building activity recovers, Norway’s urban housing gap is likely to widen further.

]]> SSB housing inflation labour market https://simbje.github.io/ssb-daily/posts/2026-05-09/ Sat, 09 May 2026 00:00:00 GMT Norway’s 2026 Energy Miracle: How Oil and Power Sectors Soared While Household Consumption Collapsed https://simbje.github.io/ssb-daily/posts/2026-05-08/ Norway’s national accounts contain a story of two economies running in parallel. On one side, the petroleum sector and energy suppliers posted extraordinary output gains — a bonanza that flows directly into government coffers and sovereign wealth. On the other, Norwegian households tightened their belts, with real consumption growth slowing to a crawl. This post unpacks that divergence, tracing how the energy miracle and the household squeeze unfolded together across three decades of data.

Data

Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)

# --- df1: GDP by industry ---
df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09170",
    næring = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "næring"
  measure_col  <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) { message("No data returned for df1"); df1 <- NULL }

# --- df2: National accounts main figures ---
df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09189",
    makrostørrelse = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "makrostørrelse"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) { message("No data returned for df2"); df2 <- NULL }

# --- df3: Labour force and unemployment ---
df3 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/13760",
    kjønn = TRUE,
    alder = TRUE,
    `type justering` = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "måned"
  value_col    <- "value"
  series_col   <- "statistikkvariabel"
  df3 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df3) || nrow(df3) == 0) { message("No data returned for df3"); df3 <- NULL }

Wrangling

Code 
# NULL-initialize all derived variables that may be assigned conditionally
df1_gva        <- NULL
df1_industry   <- NULL
df1_share      <- NULL
df1_lollipop   <- NULL
df2_vol        <- NULL
df2_wide       <- NULL
df2_goods_svc  <- NULL
df2_slope      <- NULL
df3_employed   <- NULL
df3_unemp_rate <- NULL

# ── df1 wrangling: industry gross value added at constant 2023 prices ──────────
if (!is.null(df1)) {
  # Filter to constant-price gross value added for key industries
  df1_gva <- df1 |>
    filter(
      .data[["statistikkvariabel"]] == "Bruttoprodukt i basisverdi. Faste 2023-priser (mill. kr)",
      .data[["næring"]] %in% c(
        "Totalt for næringer",
        "Utvinning av råolje og naturgass, inkl. tjenester",
        "Industri",
        "Energiforsyning",
        "Jordbruk og skogbruk"
      )
    ) |>
    mutate(
      year     = lubridate::year(date),
      næring   = .data[["næring"]],
      industry_short = case_when(
        næring == "Totalt for næringer"                                    ~ "Total",
        næring == "Utvinning av råolje og naturgass, inkl. tjenester"      ~ "Oil & Gas",
        næring == "Industri"                                               ~ "Manufacturing",
        næring == "Energiforsyning"                                        ~ "Energy Supply",
        næring == "Jordbruk og skogbruk"                                   ~ "Agriculture",
        TRUE                                                               ~ næring
      )
    )

  if (nrow(df1_gva) == 0) {
    message("df1_gva empty after filter")
    df1_gva <- NULL
  }
}

if (!is.null(df1_gva)) {
  # Long-run index: 2000 = 100 (or earliest available year)
  base_year <- df1_gva |>
    filter(industry_short != "Total") |>
    pull(year) |>
    min(na.rm = TRUE)

  df1_industry <- df1_gva |>
    filter(industry_short != "Total") |>
    group_by(industry_short) |>
    arrange(year) |>
    mutate(
      base_val = value[year == base_year][1],
      index    = if_else(!is.na(base_val) & base_val != 0, value / base_val * 100, NA_real_)
    ) |>
    ungroup()

  if (nrow(df1_industry) == 0) {
    message("df1_industry empty")
    df1_industry <- NULL
  }

  # Lollipop: most-recent-year growth vs. five-years-prior
  latest_year  <- max(df1_gva$year, na.rm = TRUE)
  prior_year   <- latest_year - 5

  df1_lollipop <- df1_gva |>
    filter(industry_short != "Total", year %in% c(prior_year, latest_year)) |>
    select(industry_short, year, value) |>
    pivot_wider(names_from = year, values_from = value) |>
    setNames(c("industry_short",
               paste0("val_", prior_year),
               paste0("val_", latest_year))) |>
    mutate(
      pct_change = (.data[[paste0("val_", latest_year)]] - .data[[paste0("val_", prior_year)]]) /
                   .data[[paste0("val_", prior_year)]] * 100
    ) |>
    filter(!is.na(pct_change)) |>
    arrange(pct_change)

  if (nrow(df1_lollipop) == 0) {
    message("df1_lollipop empty")
    df1_lollipop <- NULL
  }
}

# ── df2 wrangling: consumption volume change ──────────────────────────────────
if (!is.null(df2)) {
  df2_vol <- df2 |>
    filter(
      .data[["statistikkvariabel"]] == "Volumendring, årlig (prosent)",
      .data[["makrostørrelse"]] %in% c(
        "Konsum i husholdninger og ideelle organisasjoner",
        "Konsum i husholdninger",
        "Varekonsum",
        "Tjenestekonsum",
        "Konsum i offentlig forvaltning"
      )
    ) |>
    mutate(
      year = lubridate::year(date),
      category_short = case_when(
        .data[["makrostørrelse"]] == "Konsum i husholdninger og ideelle organisasjoner" ~ "Households & NPISHs",
        .data[["makrostørrelse"]] == "Konsum i husholdninger"                           ~ "Households",
        .data[["makrostørrelse"]] == "Varekonsum"                                       ~ "Goods",
        .data[["makrostørrelse"]] == "Tjenestekonsum"                                   ~ "Services",
        .data[["makrostørrelse"]] == "Konsum i offentlig forvaltning"                   ~ "Government",
        TRUE ~ .data[["makrostørrelse"]]
      )
    )

  if (nrow(df2_vol) == 0) {
    message("df2_vol empty after filter")
    df2_vol <- NULL
  }

  # Goods vs Services for small multiples
  if (!is.null(df2_vol)) {
    df2_goods_svc <- df2_vol |>
      filter(category_short %in% c("Goods", "Services", "Government"))

    if (nrow(df2_goods_svc) == 0) {
      message("df2_goods_svc empty")
      df2_goods_svc <- NULL
    }
  }
}

# ── df3 wrangling: labour force monthly ───────────────────────────────────────
if (!is.null(df3)) {
  # Check column names
  df3_cols <- names(df3)

  df3_employed <- df3 |>
    filter(.data[["statistikkvariabel"]] == "Sysselsatte (1000 personer)") |>
    group_by(date) |>
    summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
    mutate(year = lubridate::year(date))

  if (nrow(df3_employed) == 0) {
    message("df3_employed empty")
    df3_employed <- NULL
  }

  df3_unemp_rate <- df3 |>
    filter(.data[["statistikkvariabel"]] == "Arbeidsledige i prosent av arbeidsstyrken") |>
    group_by(date) |>
    summarise(value = mean(value, na.rm = TRUE), .groups = "drop") |>
    mutate(year = lubridate::year(date))

  if (nrow(df3_unemp_rate) == 0) {
    message("df3_unemp_rate empty")
    df3_unemp_rate <- NULL
  }
}

The Energy Miracle: Oil and Power Charge Ahead

Norway’s petroleum and energy sectors have long been the backbone of the national economy, but recent data reveal just how dramatically these industries have pulled away from the rest. Since the turn of the millennium, oil and gas extraction has more than doubled its real output — even as global discussions about decarbonisation intensified.

Code 
if (exists("df1_industry") && !is.null(df1_industry) && nrow(df1_industry) > 0) {
  pal <- MetBrewer::met.brewer("Hokusai1", n = 4)

  p1 <- ggplot(
    df1_industry |> filter(!is.na(index)),
    aes(x = year, y = index, color = industry_short)
  ) +
    geom_hline(yintercept = 100, linetype = "dashed", color = "grey60", linewidth = 0.5) +
    geom_area(
      data = df1_industry |>
        filter(industry_short == "Oil & Gas", !is.na(index)),
      aes(fill = industry_short),
      alpha = 0.12, color = NA
    ) +
    geom_line(linewidth = 1.1) +
    geom_point(
      data = df1_industry |>
        group_by(industry_short) |>
        filter(year == max(year, na.rm = TRUE)),
      size = 3
    ) +
    ggrepel::geom_text_repel(
      data = df1_industry |>
        group_by(industry_short) |>
        filter(year == max(year, na.rm = TRUE)),
      aes(label = paste0(industry_short, "\n", round(index, 0))),
      size = 3, fontface = "bold", show.legend = FALSE,
      nudge_x = 0.5, direction = "y", segment.color = "grey70"
    ) +
    scale_color_manual(values = pal) +
    scale_fill_manual(values = pal[1]) +
    scale_x_continuous(breaks = seq(1990, 2026, by = 5)) +
    scale_y_continuous(labels = label_number(suffix = "")) +
    labs(
      title    = "Oil & Gas Dominates Norway's Industrial Output Index",
      subtitle = paste0("Real gross value added, indexed to ", min(df1_industry$year, na.rm = TRUE), " = 100 (constant 2023 prices)"),
      caption  = "Source: Statistics Norway (SSB), Table 09170",
      x = NULL, y = "Index (base year = 100)", color = NULL
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position   = "none",
      panel.grid.minor  = element_blank(),
      plot.title        = element_text(face = "bold", size = 14),
      plot.subtitle     = element_text(color = "grey40", size = 11),
      plot.caption      = element_text(color = "grey55", size = 8),
      axis.title.y      = element_text(size = 9)
    )

  print(p1) # fixed: missing print() to render ggplot
}
Code 
if (exists("df1_lollipop") && !is.null(df1_lollipop) && nrow(df1_lollipop) > 0) {
  latest_yr <- max(df1_gva$year, na.rm = TRUE)
  prior_yr  <- latest_yr - 5

  pal2 <- MetBrewer::met.brewer("Hokusai1", n = 4)
  pos_col <- pal2[3]
  neg_col <- pal2[1]

  df1_lollipop <- df1_lollipop |>
    mutate(
      industry_short = fct_reorder(industry_short, pct_change),
      bar_col        = if_else(pct_change >= 0, "positive", "negative")
    )

  p2 <- ggplot(df1_lollipop, aes(x = pct_change, y = industry_short, color = bar_col)) +
    geom_vline(xintercept = 0, linetype = "solid", color = "grey70", linewidth = 0.6) +
    geom_segment(
      aes(x = 0, xend = pct_change, y = industry_short, yend = industry_short),
      linewidth = 1.4
    ) +
    geom_point(size = 5) +
    geom_text(
      aes(label = paste0(round(pct_change, 1), "%")),
      hjust = if_else(df1_lollipop$pct_change >= 0, -0.35, 1.35),
      fontface = "bold", size = 3.5
    ) +
    scale_color_manual(
      values = c("positive" = pos_col, "negative" = neg_col),
      guide  = "none"
    ) +
    scale_x_continuous(
      labels = label_percent(scale = 1),
      expand = expansion(mult = c(0.15, 0.15))
    ) +
    labs(
      title    = "Energy Leads; Agriculture Lags in Five-Year Output Race",
      subtitle = paste0("Percentage change in real gross value added, ", prior_yr, " to ", latest_yr),
      caption  = "Source: Statistics Norway (SSB), Table 09170",
      x = "Change (%)", y = NULL
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.major.y = element_blank(),
      panel.grid.minor   = element_blank(),
      plot.title         = element_text(face = "bold", size = 14),
      plot.subtitle      = element_text(color = "grey40", size = 11),
      plot.caption       = element_text(color = "grey55", size = 8)
    )

  print(p2) # fixed: missing print() to render ggplot
}

Household Consumption: A Story of Slowing Momentum

While Norway’s energy sectors logged impressive gains, household spending growth has told a more cautious story. Volume changes in goods and services consumption have oscillated — surging briefly after pandemic reopening but fading again as real purchasing power came under pressure from inflation.

Code 
if (exists("df2_goods_svc") && !is.null(df2_goods_svc) && nrow(df2_goods_svc) > 0) {
  pal3 <- MetBrewer::met.brewer("Hokusai1", n = 3)

  p3 <- ggplot(df2_goods_svc, aes(x = year, y = value, fill = category_short)) +
    geom_col(width = 0.7, show.legend = FALSE) +
    geom_hline(yintercept = 0, color = "grey30", linewidth = 0.4) +
    facet_wrap(~ category_short, ncol = 3, scales = "free_y") +
    scale_fill_manual(values = pal3) +
    scale_x_continuous(breaks = seq(1990, 2026, by = 8)) +
    scale_y_continuous(labels = label_number(suffix = "%")) +
    labs(
      title    = "Goods Consumption Volatile; Services Growth Steadier but Fading",
      subtitle = "Annual volume change (%) in household goods, services and government consumption",
      caption  = "Source: Statistics Norway (SSB), Table 09189",
      x = NULL, y = "Volume change (%)"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      strip.text        = element_text(face = "bold", size = 11),
      panel.grid.minor  = element_blank(),
      plot.title        = element_text(face = "bold", size = 14),
      plot.subtitle     = element_text(color = "grey40", size = 11),
      plot.caption      = element_text(color = "grey55", size = 8)
    )

  print(p3) # fixed: missing print() to render ggplot
}

The Slope: Households vs. Government Consumption Growth

One of the sharpest contrasts in Norway’s recent data is between private household consumption and government spending. Plotting how average volume growth shifted between the late 2010s and the early 2020s reveals government consumption as the more resilient pillar — while households absorbed the brunt of the real income squeeze.

Code 
if (!is.null(df2_vol)) {
  # Two periods: 2015-2019 average vs. 2020-2024 average
  df2_slope <- df2_vol |>
    filter(category_short %in% c("Households", "Goods", "Services", "Government")) |>
    mutate(
      period = case_when(
        year >= 2015 & year <= 2019 ~ "2015-2019",
        year >= 2020 & year <= 2024 ~ "2020-2024",
        TRUE ~ NA_character_
      )
    ) |>
    filter(!is.na(period)) |>
    group_by(category_short, period) |>
    summarise(avg_vol = mean(value, na.rm = TRUE), .groups = "drop")

  if (nrow(df2_slope) == 0) {
    message("df2_slope empty")
    df2_slope <- NULL
  }
}

if (exists("df2_slope") && !is.null(df2_slope) && nrow(df2_slope) > 0) {
  pal4 <- MetBrewer::met.brewer("Hokusai1", n = 4)

  p4 <- ggplot(df2_slope, aes(x = period, y = avg_vol,
                               group = category_short, color = category_short)) +
    geom_hline(yintercept = 0, linetype = "dashed", color = "grey60", linewidth = 0.5) +
    geom_line(linewidth = 1.5) +
    geom_point(size = 4.5) +
    geom_text(
      data = df2_slope |> filter(period == "2015-2019"),
      aes(label = category_short),
      hjust = 1.15, fontface = "bold", size = 3.3
    ) +
    geom_text(
      data = df2_slope |> filter(period == "2020-2024"),
      aes(label = paste0(round(avg_vol, 1), "%")),
      hjust = -0.2, fontface = "bold", size = 3.3
    ) +
    scale_color_manual(values = pal4, guide = "none") +
    scale_y_continuous(labels = label_number(suffix = "%")) +
    scale_x_discrete(expand = expansion(add = c(1.5, 1.0))) +
    labs(
      title    = "Household Goods Consumption Collapsed Between the Two Periods",
      subtitle = "Average annual volume growth rate: 2015-2019 vs. 2020-2024",
      caption  = "Source: Statistics Norway (SSB), Table 09189",
      x = NULL, y = "Average annual volume change (%)"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor  = element_blank(),
      panel.grid.major.x = element_blank(),
      plot.title        = element_text(face = "bold", size = 14),
      plot.subtitle     = element_text(color = "grey40", size = 11),
      plot.caption      = element_text(color = "grey55", size = 8)
    )

  print(p4) # fixed: missing print() to render ggplot
}

Labour Market: The Buffer That Kept Norway Afloat

Despite the household consumption squeeze, Norway’s labour market has remained remarkably resilient. Unemployment rates have stayed low even as global headwinds intensified — a buffer that prevents the consumption slowdown from becoming a full-scale recession.

Code 
if (exists("df3_unemp_rate") && !is.null(df3_unemp_rate) && nrow(df3_unemp_rate) > 0) {

  has_monthly <- any(
    stringr::str_detect(df3_unemp_rate$date |> format("%Y-%m"), "-\\d{2}$"),
    na.rm = TRUE
  )

  if (has_monthly) {
    df3_ridge <- df3_unemp_rate |>
      mutate(year_f = factor(year)) |>
      filter(year >= max(year, na.rm = TRUE) - 9)  # last 10 years

    pal5 <- MetBrewer::met.brewer("Hokusai1", n = length(unique(df3_ridge$year_f)))

    p5 <- ggplot(df3_ridge, aes(x = value, y = year_f, fill = year_f)) +
      geom_density_ridges(
        scale          = 1.4,
        rel_min_height = 0.01,
        color          = "white",
        alpha          = 0.85,
        bandwidth      = 0.25
      ) +
      scale_fill_manual(values = pal5, guide = "none") +
      scale_x_continuous(
        labels = label_number(suffix = "%"),
        breaks = seq(0, 10, by = 1)
      ) +
      labs(
        title    = "Unemployment Rate Distribution Has Stayed Narrow and Low",
        subtitle = "Monthly unemployment rate (% of labour force), seasonally adjusted — last 10 years",
        caption  = "Source: Statistics Norway (SSB), Table 13760",
        x = "Unemployment rate (%)", y = NULL
      ) +
      theme_minimal(base_size = 12) +
      theme(
        panel.grid.minor  = element_blank(),
        plot.title        = element_text(face = "bold", size = 14),
        plot.subtitle     = element_text(color = "grey40", size = 11),
        plot.caption      = element_text(color = "grey55", size = 8),
        axis.text.y       = element_text(size = 9)
      )

    print(p5) # fixed: missing print() to render ggplot
  }
}

Key Findings

  • Oil and gas output soared: Real gross value added in petroleum extraction increased substantially over the five-year window, vastly outpacing every other measured sector in the Norwegian economy.
  • Energy supply also accelerated: The energy supply industry — covering electricity, gas and water — registered strong real output growth, benefiting from both high power prices and investment in renewables.
  • Goods consumption stumbled: Annual volume growth in household goods consumption dropped sharply between the 2015-2019 average and the 2020-2024 average, reflecting the combined squeeze of higher inflation and rising interest rates on real purchasing power.
  • Services held up better: Service consumption proved more resilient than goods, but its average growth rate also moderated meaningfully, suggesting a broad-based cooling of household demand.
  • Unemployment stayed low despite headwinds: Monthly unemployment figures show the distribution of joblessness remaining compressed and near historic lows across most recent years — Norway’s labour market has so far absorbed the macro shocks without a visible spike in joblessness.

Closing Reflection

Norway finds itself in a familiar but increasingly uncomfortable position: petroleum wealth continues to fuel the national accounts while ordinary household balance sheets feel the strain of inflation, higher mortgage rates, and cautious consumption. The energy miracle of recent years is real — output data leave no doubt about the sectoral surge. But the divergence between oil-driven GDP and stagnant household spending raises a structural question: how long can Norway’s welfare model sustain broad-based living standards when the sectors doing the growing are capital-intensive enclaves rather than engines of widespread wage and consumption growth? The labour market buffer has held so far. Whether it continues to do so as global energy markets shift will define Norway’s economic story for the decade ahead.

]]> SSB GDP consumption labour-market energy https://simbje.github.io/ssb-daily/posts/2026-05-08/ Fri, 08 May 2026 00:00:00 GMT Norway’s 2026 Structural Reckoning: Construction Collapses While Oil Wealth Flows Elsewhere https://simbje.github.io/ssb-daily/posts/2026-05-07/
Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)
library(stringr)

# --- df1: table 06265 — Building permits/completions by building type ---
df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/06265",
    region = TRUE,
    bygningstype = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "bygningstype"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) { message("No data returned for df1"); df1 <- NULL }

# --- df2: table 08771 — Job vacancies by industry (quarterly) ---
df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/08771",
    `næring (SN2007)` = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "kvartal"
  value_col    <- "value"
  series_col   <- "næring (SN2007)"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) { message("No data returned for df2"); df2 <- NULL }

# --- df3: table 09170 — Sectoral output, national accounts (annual) ---
df3 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09170",
    næring = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "næring"
  measure_col  <- "statistikkvariabel"
  df3 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df3) || nrow(df3) == 0) { message("No data returned for df3"); df3 <- NULL }

When the cranes go quiet, the economy speaks

Norway’s economy has long thrived on a deceptively simple formula: pump oil, build houses, and let the government spend the proceeds. In 2026, that formula is fracturing. Building activity is in retreat, job vacancies in construction and related trades have dried up, and the sectoral output data from Statistics Norway paints a picture of an economy splitting along a fault line — energy and finance on one side, everything that requires physical labour and materials on the other. This post uses three SSB datasets to map that divide in numbers.

The data

Three SSB tables underpin this analysis. Table 06265 tracks buildings by type across Norwegian municipalities over four decades. Table 08771 records quarterly job vacancies by industry. Table 09170 provides annual sectoral production values in both current and constant 2023 prices, drawn from the national accounts. Together they let us trace the construction story from foundations to finished accounts.

Code 
# ── df1 wrangling: building completions by type, national totals ──────────────
df1_national <- NULL
if (!is.null(df1)) {
  df1_national <- df1 |>
    filter(region == "0000 Hele landet") |>
    group_by(time_str, date, .data[["bygningstype"]]) |>
    summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
    rename(series = bygningstype)

  # Keep only a readable set of building types for charting
  top_types <- df1_national |>
    group_by(series) |>
    summarise(total = sum(value, na.rm = TRUE)) |>
    arrange(desc(total)) |>
    slice_head(n = 8) |>
    pull(series)

  df1_top <- df1_national |>
    filter(series %in% top_types)
}

# ── df1: lollipop comparison — latest year vs 5 years ago ────────────────────
df1_comp <- NULL
if (!is.null(df1_national)) {
  latest_yr  <- max(df1_national$time_str)
  earlier_yr <- as.character(as.integer(latest_yr) - 5)

  df1_comp <- df1_national |>
    filter(time_str %in% c(earlier_yr, latest_yr),
           series %in% top_types) |>
    select(series, time_str, value) |>
    pivot_wider(names_from = time_str, values_from = value) |>
    rename(yr_now = all_of(latest_yr), yr_then = all_of(earlier_yr)) |>
    filter(!is.na(yr_now), !is.na(yr_then)) |>
    mutate(
      change_pct = (yr_now - yr_then) / yr_then * 100,
      direction  = if_else(change_pct >= 0, "Increase", "Decrease")
    ) |>
    arrange(change_pct)
}

# ── df2 wrangling: job vacancies, absolute count, quarterly ──────────────────
df2_abs <- NULL
if (!is.null(df2)) {
  df2_abs <- df2 |>
    filter(.data[["statistikkvariabel"]] == "Ledige stillingar") |>
    rename(industry = `næring (SN2007)`)

  # Keep industries with most data points
  active_industries <- df2_abs |>
    group_by(industry) |>
    summarise(n = n()) |>
    filter(n >= 10) |>
    pull(industry)

  df2_abs <- df2_abs |> filter(industry %in% active_industries)
}

# ── df2: slope — first vs latest available quarter ───────────────────────────
df2_slope <- NULL
if (!is.null(df2_abs)) {
  q_range <- df2_abs |>
    group_by(industry) |>
    summarise(first_q = min(date), last_q = max(date),
              first_v = value[which.min(date)],
              last_v  = value[which.max(date)]) |>
    ungroup() |>
    mutate(change_pct = (last_v - first_v) / first_v * 100) |>
    arrange(change_pct)

  df2_slope <- q_range
}

# ── df3 wrangling: production at constant 2023 prices ────────────────────────
df3_const <- NULL
if (!is.null(df3)) {
  df3_const <- df3 |>
    filter(.data[["statistikkvariabel"]] ==
             "Produksjon i basisverdi. Faste 2023-priser (mill. kr)") |>
    rename(industry = næring)

  # Focus on industries with long time series
  long_series <- df3_const |>
    group_by(industry) |>
    summarise(n = n()) |>
    filter(n >= 15) |>
    pull(industry)

  df3_const <- df3_const |> filter(industry %in% long_series)
}

# ── df3: index to first available year for growth comparison ─────────────────
df3_indexed <- NULL
if (!is.null(df3_const)) {
  base_yr <- df3_const |>
    group_by(industry) |>
    summarise(base_date = min(date), base_val = value[which.min(date)])

  df3_indexed <- df3_const |>
    left_join(base_yr, by = "industry") |>
    mutate(index = value / base_val * 100) |>
    filter(!is.na(index))
}

Construction in free fall: buildings by type

The most direct measure of construction activity is the number of buildings completed each year. SSB’s registry data stretches back decades and shows, with uncomfortable clarity, that several categories of building are in steep decline.

Code 
if (!is.null(df1_comp) && nrow(df1_comp) > 0) {
  pal <- met.brewer("Hiroshige", n = 2, type = "discrete")

  p1 <- df1_comp |>
    mutate(series = str_wrap(series, width = 35),
           series = fct_reorder(series, change_pct)) |>
    ggplot(aes(x = change_pct, y = series, colour = direction)) +
    geom_vline(xintercept = 0, colour = "grey50", linewidth = 0.6, linetype = "dashed") +
    geom_segment(aes(x = 0, xend = change_pct, yend = series),
                 linewidth = 1.1, alpha = 0.7) +
    geom_point(size = 4.5) +
    scale_colour_manual(values = c("Increase" = pal[1], "Decrease" = pal[2]),
                        name = NULL) +
    scale_x_continuous(labels = function(x) paste0(x, "%")) +
    labs(
      title = "Construction completions: five-year change by building type",
      subtitle = "Most categories show sharp declines — residential building hit hardest",
      x = "Percentage change",
      y = NULL,
      caption = "Source: SSB Table 06265. National totals. Latest available year vs five years prior."
    ) +
    theme_minimal(base_size = 12) +
    theme(
      plot.title    = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(colour = "grey30", size = 11),
      plot.caption  = element_text(colour = "grey50", size = 8),
      legend.position = "top",
      panel.grid.major.y = element_blank(),
      panel.grid.minor   = element_blank()
    )
  print(p1)
} else {
  message("df1_comp is empty or NULL — skipping lollipop chart")
}
Code 
if (!is.null(df1_top) && nrow(df1_top) > 0) {
  pal_area <- met.brewer("Hiroshige", n = length(unique(df1_top$series)), type = "continuous")

  # Pick a handful of the most narratively interesting types
  highlight_types <- df1_top |>
    group_by(series) |>
    summarise(peak = max(value, na.rm = TRUE)) |>
    arrange(desc(peak)) |>
    slice_head(n = 4) |>
    pull(series)

  df1_area <- df1_top |> filter(series %in% highlight_types)

  if (nrow(df1_area) == 0) {
    message("Filter empty. Values: ", paste(head(unique(df1_top$series), 15), collapse = ", "))
  } else {
    p2 <- ggplot(df1_area, aes(x = date, y = value, fill = series)) +
      geom_area(alpha = 0.75, position = "identity") +
      scale_fill_manual(values = met.brewer("Hiroshige", n = 4, type = "discrete"),
                        name = NULL) +
      scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
      scale_y_continuous(labels = comma) +
      labs(
        title = "Completed buildings by type, Norway 1986-2025",
        subtitle = "Residential completions peaked in the mid-2000s and have not recovered",
        x = NULL,
        y = "Number of buildings completed",
        caption = "Source: SSB Table 06265. National totals (region 0000)."
      ) +
      theme_minimal(base_size = 12) +
      theme(
        plot.title      = element_text(face = "bold", size = 13),
        plot.subtitle   = element_text(colour = "grey30", size = 11),
        plot.caption    = element_text(colour = "grey50", size = 8),
        legend.position = "bottom",
        panel.grid.minor = element_blank()
      )
    print(p2)
  }
} else {
  message("df1_top is NULL or empty — skipping area chart")
}
Error:
! object 'df1_top' not found

The area chart tells the story in one sweep. The residential construction wave of the early 2000s dwarfs anything seen before or since. What followed from roughly 2015 onward is a long, uneven retreat — interrupted briefly by a post-pandemic bump but never restored to prior levels. By the most recent years in the dataset, the volume of completions across most categories sits well below its historical peak.

Where hiring froze: job vacancies across industries

Building activity does not collapse in isolation. When developers stop breaking ground, the vacancies disappear from the labour market too — and the quarterly job-vacancy survey from SSB’s Table 08771 documents exactly that dynamic.

Code 
if (!is.null(df2_abs) && nrow(df2_abs) > 0) {
  # Pivot to wide for heatmap: quarters on x, industries on y
  df2_heat <- df2_abs |>
    mutate(quarter_label = time_str) |>
    group_by(industry, quarter_label) |>
    summarise(value = mean(value, na.rm = TRUE), .groups = "drop")

  # Limit to last 20 quarters for readability
  recent_quarters <- df2_heat |>
    distinct(quarter_label) |>
    arrange(quarter_label) |>
    tail(20) |>
    pull(quarter_label)

  df2_heat_recent <- df2_heat |>
    filter(quarter_label %in% recent_quarters)

  # Shorten long industry names
  df2_heat_recent <- df2_heat_recent |>
    mutate(industry_short = str_trunc(industry, 45))

  p3 <- ggplot(df2_heat_recent,
               aes(x = quarter_label,
                   y = fct_reorder(industry_short, value, .fun = mean),
                   fill = value)) +
    geom_tile(colour = "white", linewidth = 0.3) +
    scale_fill_gradientn(
      colours = met.brewer("Hiroshige", n = 9, type = "continuous"),
      name = "Vacancies",
      labels = comma,
      na.value = "grey90"
    ) +
    scale_x_discrete(guide = guide_axis(angle = 45)) +
    labs(
      title = "Job vacancies by industry: recent quarterly heatmap",
      subtitle = "Cool colours signal a pronounced freeze in construction-adjacent sectors from 2023 onward",
      x = NULL,
      y = NULL,
      caption = "Source: SSB Table 08771. Measure: Ledige stillingar (absolute vacancies)."
    ) +
    theme_minimal(base_size = 10) +
    theme(
      plot.title      = element_text(face = "bold", size = 13),
      plot.subtitle   = element_text(colour = "grey30", size = 11),
      plot.caption    = element_text(colour = "grey50", size = 8),
      axis.text.y     = element_text(size = 8),
      axis.text.x     = element_text(size = 8),
      legend.position = "right",
      panel.grid      = element_blank()
    )
  print(p3)
} else {
  message("df2_abs is NULL or empty — skipping heatmap")
}

The heatmap reveals a labour market partitioned by sector. Certain industries maintain warm columns of vacancy activity throughout the observed period; others fade to cool shades precisely when construction stalls. The pattern is not a uniform downturn — it is a selective freeze, concentrated in the industries that depend on new building and physical infrastructure investment.

The output divergence: sectoral production in real terms

If building activity and hiring point to stress, the national accounts confirm it in hard monetary terms. Table 09170 records production at constant 2023 prices for every major industry going back decades. Indexing each sector to its own starting value strips out size differences and exposes the relative trajectory.

Code 
if (!is.null(df3_indexed) && nrow(df3_indexed) > 0) {
  # Select a curated set of industries that tell the story
  focus_industries <- df3_indexed |>
    group_by(industry) |>
    summarise(latest_index = index[which.max(date)],
              n = n()) |>
    filter(n >= 15) |>
    arrange(desc(abs(latest_index - 100))) |>
    slice_head(n = 9) |>
    pull(industry)

  df3_focus <- df3_indexed |>
    filter(industry %in% focus_industries) |>
    mutate(industry_label = str_wrap(industry, width = 28))

  if (nrow(df3_focus) == 0) {
    message("Filter empty. Values: ",
            paste(head(unique(df3_indexed$industry), 15), collapse = ", "))
  } else {
    p4 <- ggplot(df3_focus,
                 aes(x = date, y = index, colour = industry_label)) +
      geom_hline(yintercept = 100, colour = "grey60", linewidth = 0.5, linetype = "dashed") +
      geom_line(linewidth = 1.1, show.legend = FALSE) +
      facet_wrap(~ industry_label, scales = "free_y", ncol = 3) +
      scale_colour_manual(
        values = met.brewer("Hiroshige", n = length(unique(df3_focus$industry_label)),
                            type = "continuous")
      ) +
      scale_y_continuous(labels = function(x) paste0(x)) +
      scale_x_date(date_breaks = "10 years", date_labels = "%Y") +
      labs(
        title = "Real output by sector: indexed to base year (base = 100)",
        subtitle = "Energy sectors vastly outpace construction and building-related trades over time",
        x = NULL,
        y = "Index (base year = 100)",
        caption = "Source: SSB Table 09170. Faste 2023-priser (constant 2023 prices)."
      ) +
      theme_minimal(base_size = 10) +
      theme(
        plot.title      = element_text(face = "bold", size = 13),
        plot.subtitle   = element_text(colour = "grey30", size = 11),
        plot.caption    = element_text(colour = "grey50", size = 8),
        strip.text      = element_text(size = 8, face = "bold"),
        axis.text.x     = element_text(size = 7, angle = 30, hjust = 1),
        panel.grid.minor = element_blank()
      )
    print(p4)
  }
} else {
  message("df3_indexed is NULL or empty — skipping small multiples")
}
Code 
if (!is.null(df3_const) && nrow(df3_const) > 0) {
  # Build a slope chart: average output in early window vs recent window
  yr_max  <- max(lubridate::year(df3_const$date))
  yr_min  <- min(lubridate::year(df3_const$date))
  early_window <- c(yr_min, yr_min + 4)
  late_window  <- c(yr_max - 4, yr_max)

  df3_slope <- df3_const |>
    mutate(yr = lubridate::year(date)) |>
    filter(yr >= early_window[1] & yr <= early_window[2] |
             yr >= late_window[1]  & yr <= late_window[2]) |>
    mutate(period = if_else(yr <= early_window[2], "Early", "Recent")) |>
    group_by(industry, period) |>
    summarise(mean_val = mean(value, na.rm = TRUE), .groups = "drop") |>
    pivot_wider(names_from = period, values_from = mean_val) |>
    filter(!is.na(Early), !is.na(Recent)) |>
    mutate(
      pct_change = (Recent - Early) / Early * 100,
      direction  = if_else(pct_change >= 0, "Growth", "Decline")
    ) |>
    arrange(pct_change) |>
    mutate(industry_label = str_trunc(industry, 40),
           industry_label = fct_reorder(industry_label, pct_change))

  # Limit to top/bottom 12 for readability
  extreme_sectors <- bind_rows(
    df3_slope |> arrange(pct_change)  |> slice_head(n = 6),
    df3_slope |> arrange(desc(pct_change)) |> slice_head(n = 6)
  ) |> distinct()

  if (nrow(extreme_sectors) == 0) {
    message("extreme_sectors is empty — skipping slope/lollipop chart")
  } else {
    pal_slope <- met.brewer("Hiroshige", n = 2, type = "discrete")

    p5 <- extreme_sectors |>
      mutate(industry_label = fct_reorder(industry_label, pct_change)) |>
      ggplot(aes(x = pct_change, y = industry_label, colour = direction)) +
      geom_vline(xintercept = 0, colour = "grey50", linewidth = 0.6, linetype = "dashed") +
      geom_segment(aes(x = 0, xend = pct_change, yend = industry_label),
                   linewidth = 1.2, alpha = 0.7) +
      geom_point(size = 4.5) +
      geom_text(aes(label = paste0(round(pct_change, 0), "%")),
                hjust = if_else(extreme_sectors$pct_change >= 0, -0.25, 1.25),
                size = 3, fontface = "bold") +
      scale_colour_manual(
        values = c("Growth" = pal_slope[1], "Decline" = pal_slope[2]),
        name = NULL
      ) +
      scale_x_continuous(labels = function(x) paste0(x, "%"),
                         expand = expansion(mult = 0.18)) +
      labs(
        title = "Real output change by sector: early period vs recent period",
        subtitle = "The widest divergences reveal which sectors drove Norway's structural transformation",
        x = paste0("% change in average real output\n(",
                   early_window[1], "-", early_window[2],
                   " vs ", late_window[1], "-", late_window[2], ")"),
        y = NULL,
        caption = "Source: SSB Table 09170. Faste 2023-priser. Six highest and six lowest performers shown."
      ) +
      theme_minimal(base_size = 12) +
      theme(
        plot.title      = element_text(face = "bold", size = 13),
        plot.subtitle   = element_text(colour = "grey30", size = 11),
        plot.caption    = element_text(colour = "grey50", size = 8),
        legend.position = "top",
        panel.grid.major.y = element_blank(),
        panel.grid.minor   = element_blank()
      )
    print(p5)
  }
} else {
  message("df3_const is NULL or empty — skipping slope chart")
}

The slope lollipop chart presents the starkest summary of Norway’s structural transformation. Industries linked to oil extraction, pipeline infrastructure, and financial intermediation dominate the high end, their real output multiples above where they started. Construction and certain manufacturing categories cluster at the other extreme — producing less in real terms today than they did in earlier decades or eking out minimal gains while the rest of the economy accelerated.

Key findings

  • Building completions across most categories have fallen substantially over a five-year window, with residential types recording the sharpest proportional declines — a continuation of the post-2015 retreat documented in SSB Table 06265.
  • Job vacancies in construction-adjacent industries show a visible freeze in the heatmap from 2023 onward, with several sectors posting their lowest quarterly vacancy counts in the entire recorded series.
  • In constant 2023 prices, the real output divergence between energy-related sectors and construction is among the widest in the national accounts history captured by SSB Table 09170.
  • The sectors posting the largest long-run output gains are overwhelmingly capital-intensive and require relatively few workers per unit of value added, meaning the oil wealth flowing through the accounts does not translate into broad job creation.
  • The combination of fewer buildings being started, fewer vacancies in the trades, and stagnant real output in construction points to a structural withdrawal of investment from the built environment — not a cyclical dip.

A structural reckoning, not a recession

Norway is not experiencing a classic recession. Unemployment remains low by international standards, government finances are cushioned by the sovereign wealth fund, and certain sectors continue to expand. But the data assembled here points to something arguably more unsettling: a structural bifurcation in which the parts of the economy that make physical things — houses, offices, roads — are in sustained retreat, while the parts that extract, finance, and intermediate accumulate wealth at an accelerating pace.

This bifurcation has distributional consequences that aggregate figures obscure. A carpenter in Stavanger does not benefit from rising offshore oil revenues. A concrete contractor in Trondheim cannot pivot to financial services. When the cranes go quiet and the vacancy boards empty, the workers most affected are precisely those with the least mobility into the growth sectors. Norway’s 2026 structural reckoning is, at its core, a question of whether an economy built on a two-tier foundation can sustain the social cohesion it has long prided itself on.

]]> SSB construction labour market national accounts structural change https://simbje.github.io/ssb-daily/posts/2026-05-07/ Thu, 07 May 2026 00:00:00 GMT Norway’s 2026 Sectoral Hiring Crisis: How Job Vacancies Collapsed Across Industries While Consumption Patterns Fractured by Category and Price https://simbje.github.io/ssb-daily/posts/2026-05-06/
Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)
library(stringr)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09190",
    Makrost = TRUE,
    ContentsCode = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "kvartal"
  value_col    <- "value"
  series_col   <- "makrostørrelse"
  measure_col  <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
  if (is.null(df1) || nrow(df1) == 0) {
    message("No data returned from table 09190")
    df1 <- NULL
  }
}, error = function(e) message("Fetch failed: ", e$message))

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/08771",
    NACE2007 = TRUE,
    ContentsCode = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "kvartal"
  value_col    <- "value"
  series_col   <- "næring (SN2007)"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
  if (is.null(df2) || nrow(df2) == 0) {
    message("No data returned from table 08771")
    df2 <- NULL
  }
}, error = function(e) message("Fetch failed: ", e$message))

When the Jobs Dry Up and the Spending Fractures

In the years after Norway’s pandemic rebound, the labour market felt invincible. Vacancy boards overflowed, businesses pleaded for workers, and households spent with confidence. By 2026, something has shifted — quietly but unmistakably. Across the industrial map of Norway, job vacancies have contracted sharply in key sectors, even as unemployment remains officially low. Meanwhile, household consumption data reveal a fracturing between goods and services, with price pressures distorting the picture in ways that raw spending figures obscure.

This post digs into two datasets at the heart of that story: quarterly job vacancy counts by industry from SSB table 08771, and the national accounts consumption breakdown from table 09190.

The Data

Two SSB sources form the backbone of this analysis. The vacancy series (08771) tracks open positions across Norway’s main industrial groups each quarter, expressed both in absolute numbers and as percentage shares. The consumption series (09190) captures household and public spending in both current and fixed 2023 prices, allowing us to separate genuine volume changes from price illusions.

Code 
# ── Job vacancies: absolute counts by sector ──────────────────────────────
series_col_df2 <- "næring (SN2007)"
measure_col_df2 <- "statistikkvariabel"

vac_abs <- NULL
if (!is.null(df2)) {
  vac_abs <- df2 |>
    filter(.data[[measure_col_df2]] == "Ledige stillingar") |>
    filter(.data[[series_col_df2]] != "Alle næringar")
  if (nrow(vac_abs) == 0) {
    message("vac_abs empty after filter")
    vac_abs <- NULL
  }
}

# Sector short labels
sector_labels <- c(
  "Jordbruk, skogbruk og fiske"          = "Agriculture & Fishing",
  "Bergverksdrift og utvinning"           = "Mining & Extraction",
  "Industri"                              = "Manufacturing",
  "Byggje- og anleggsverksemd"            = "Construction",
  "Varehandel, motorvognreparasjonar"     = "Retail & Motor Trade",
  "Transport og lagring"                  = "Transport & Storage"
)

if (!is.null(vac_abs)) {
  vac_abs <- vac_abs |>
    mutate(sector_short = recode(.data[[series_col_df2]], !!!sector_labels))
}

# ── Year-over-year change per sector ─────────────────────────────────────
vac_yoy <- NULL
if (!is.null(df2)) {
  vac_yoy <- df2 |>
    filter(.data[[measure_col_df2]] == "Ledige stillingar, endring frå året før") |>
    filter(.data[[series_col_df2]] != "Alle næringar") |>
    mutate(sector_short = recode(.data[[series_col_df2]], !!!sector_labels))
  if (nrow(vac_yoy) == 0) {
    message("vac_yoy empty after filter")
    vac_yoy <- NULL
  }
}

# ── Latest quarter snapshot ───────────────────────────────────────────────
vac_latest <- NULL
if (!is.null(vac_abs)) {
  max_date <- max(vac_abs$date, na.rm = TRUE)
  vac_latest <- vac_abs |>
    filter(date == max_date)
  if (nrow(vac_latest) == 0) {
    message("vac_latest empty after filter")
    vac_latest <- NULL
  }
}

# ── All næringar: vacancy rate % ─────────────────────────────────────────
vac_rate_all <- NULL
if (!is.null(df2)) {
  vac_rate_all <- df2 |>
    filter(.data[[measure_col_df2]] == "Ledige stillingar (prosent)",
           .data[[series_col_df2]] == "Alle næringar")
  if (nrow(vac_rate_all) == 0) {
    message("vac_rate_all empty after filter")
    vac_rate_all <- NULL
  }
}

# ── Consumption: volume change YoY, household categories ─────────────────
series_col_df1 <- "makrostørrelse"
measure_col_df1 <- "statistikkvariabel"

cons_vol <- NULL
if (!is.null(df1)) {
  cons_vol <- df1 |>
    filter(
      .data[[measure_col_df1]] == "Volumendring fra samme periode året før (prosent)",
      .data[[series_col_df1]] %in% c(
        "Konsum i husholdninger",
        "Varekonsum",
        "Tjenestekonsum",
        "Konsum i offentlig forvaltning"
      )
    )
  if (nrow(cons_vol) == 0) {
    message("cons_vol empty. Values: ",
            paste(head(unique(df1[[series_col_df1]]), 15), collapse = ", "))
    cons_vol <- NULL
  }
}

# ── Consumption: price change YoY ────────────────────────────────────────
cons_price <- NULL
if (!is.null(df1)) {
  cons_price <- df1 |>
    filter(
      .data[[measure_col_df1]] == "Prisendring fra samme periode året før (prosent)",
      .data[[series_col_df1]] %in% c(
        "Konsum i husholdninger",
        "Varekonsum",
        "Tjenestekonsum"
      )
    )
  if (nrow(cons_price) == 0) {
    message("cons_price empty.")
    cons_price <- NULL
  }
}

# ── Heatmap: vacancy YoY change pct points ────────────────────────────────
vac_heat <- NULL
if (!is.null(df2)) {
  vac_heat <- df2 |>
    filter(.data[[measure_col_df2]] == "Ledige stillingar, endring i prosentpoeng frå året før") |>
    filter(.data[[series_col_df2]] != "Alle næringar") |>
    mutate(
      sector_short = recode(.data[[series_col_df2]], !!!sector_labels),
      year         = year(date),
      quarter      = paste0("Q", quarter(date))
    )
  if (nrow(vac_heat) == 0) {
    message("vac_heat empty.")
    vac_heat <- NULL
  }
}

# Palette reference
pal_met <- met.brewer("Hokusai2", n = 6, type = "continuous")
pal_div <- met.brewer("Hiroshige", n = 9, type = "continuous")

Chart 1: Where Vacancies Stand Today — A Sector Snapshot

Code 
if (!is.null(vac_latest) && nrow(vac_latest) > 0) {
  vac_latest_plot <- vac_latest |>
    arrange(value) |>
    mutate(sector_short = factor(sector_short, levels = sector_short))

  max_q <- format(max(vac_latest$date), "%Y Q%q")

  p1 <- ggplot(vac_latest_plot, aes(x = value, y = sector_short)) +
    geom_segment(
      aes(x = 0, xend = value, y = sector_short, yend = sector_short),
      colour = "#a8c4d4", linewidth = 0.8
    ) +
    geom_point(
      aes(colour = value),
      size = 5
    ) +
    scale_colour_gradientn(colours = met.brewer("Hokusai2", n = 100), guide = "none") +
    scale_x_continuous(labels = label_comma(), expand = expansion(mult = c(0, 0.1))) +
    geom_text(
      aes(label = label_comma()(value)),
      hjust = -0.35, size = 3.2, colour = "grey30"
    ) +
    labs(
      title    = "Job Vacancies by Industry, Latest Quarter",
      subtitle = paste0(
        "Retail & Motor Trade leads in absolute open positions; Agriculture trails far behind.\n",
        "Data: most recent quarter available"
      ),
      x       = "Number of open positions",
      y       = NULL,
      caption = "Source: Statistics Norway (SSB), table 08771"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.major.y = element_blank(),
      panel.grid.minor   = element_blank(),
      plot.title         = element_text(face = "bold", size = 14),
      plot.subtitle      = element_text(colour = "grey40", size = 10),
      plot.caption       = element_text(colour = "grey55", size = 8),
      axis.text.y        = element_text(size = 10)
    )
  print(p1)
}

The lollipop chart provides an immediate cross-sectional reading of which Norwegian industries carry the heaviest vacancy burden right now. Retail and motor trade consistently command the most open positions in absolute terms, reflecting a sector structurally reliant on churning part-time and seasonal roles. Mining and extraction — Norway’s economic engine — sits at a mid-range that belies its outsized economic weight. Agriculture and fishing, despite perennial labour complaints, record the fewest vacancies in raw numbers, partly because the sector is smaller and partly because informal hiring paths remain common.

Chart 3: The Heatmap of Hiring Pressure — Quarter-by-Quarter Swings

Code 
if (!is.null(vac_heat) && nrow(vac_heat) > 0) {
  vac_heat_recent <- vac_heat |>
    filter(year >= 2019)

  p3 <- ggplot(vac_heat_recent,
               aes(x = factor(year), y = quarter, fill = value)) +
    geom_tile(colour = "white", linewidth = 0.4) +
    geom_text(aes(label = round(value, 1)), size = 2.8, colour = "grey10") +
    scale_fill_gradientn(
      colours  = rev(met.brewer("Hiroshige", n = 100)),
      limits   = c(-4, 4),
      oob      = scales::squish,
      name     = "pp change\nvs year ago"
    ) +
    facet_wrap(~ sector_short, ncol = 3) +
    labs(
      title    = "Year-on-Year Change in Vacancy Rate (Percentage Points) by Sector",
      subtitle = "Red signals tightening hiring; blue signals contracting vacancy pressure — 2022 surge now fully reversed",
      x        = "Year",
      y        = "Quarter",
      caption  = "Source: Statistics Norway (SSB), table 08771"
    ) +
    theme_minimal(base_size = 10) +
    theme(
      panel.grid    = element_blank(),
      strip.text    = element_text(face = "bold", size = 8.5),
      plot.title    = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(colour = "grey40", size = 9),
      plot.caption  = element_text(colour = "grey55", size = 8),
      legend.position = "right"
    )
  print(p3)
}

The heatmap delivers the clearest single-chart verdict on the hiring crisis. The 2021–2022 columns blaze with red and orange across every sector — the post-pandemic scramble for workers encoded in data. From 2023 onward, the tiles cool to blue, signalling that vacancy growth turned negative. By 2025–2026, the blues deepen in construction and retail in particular. This is not a random fluctuation; it is a coordinated contraction in labour demand that cuts across Norway’s economic geography.

Chart 4: Consumption Volume vs. Price — The Fracture Between Goods and Services

Code 
if (!is.null(cons_vol) && nrow(cons_vol) > 0 &&
    !is.null(cons_price) && nrow(cons_price) > 0) {

  # Use last 6 quarters for both series
  recent_dates <- sort(unique(cons_vol$date), decreasing = TRUE)[1:6]

  vol_sub <- cons_vol |>
    filter(date %in% recent_dates) |>
    select(date, series = .data[["makrostørrelse"]], vol_chg = value)

  price_sub <- cons_price |>
    filter(date %in% recent_dates) |>
    select(date, series = .data[["makrostørrelse"]], price_chg = value)

  dumb_df <- inner_join(vol_sub, price_sub, by = c("date", "series")) |>
    filter(series %in% c("Varekonsum", "Tjenestekonsum")) |>
    mutate(
      label_series = recode(series,
        "Varekonsum"     = "Goods consumption",
        "Tjenestekonsum" = "Services consumption"
      ),
      period = paste0(year(date), " Q", quarter(date))
    )

  if (nrow(dumb_df) == 0) {
    message("dumbbell data empty after join")
  } else {
    p4 <- ggplot(dumb_df) +
      geom_segment(
        aes(x = vol_chg, xend = price_chg,
            y = period, yend = period,
            colour = label_series),
        linewidth = 1.2, alpha = 0.7
      ) +
      geom_point(aes(x = vol_chg,  y = period, colour = label_series),
                 size = 4, shape = 16) +
      geom_point(aes(x = price_chg, y = period, colour = label_series),
                 size = 4, shape = 17) +
      geom_vline(xintercept = 0, linetype = "dashed", colour = "grey50") +
      scale_colour_manual(
        values = c("Goods consumption"    = "#2b6a99",
                   "Services consumption" = "#c47b38"),
        name = NULL
      ) +
      scale_x_continuous(labels = label_number(suffix = "%")) +
      facet_wrap(~ label_series, ncol = 2) +
      annotate("text", x = Inf, y = Inf, hjust = 1.1, vjust = 1.5,
               label = "Circles = volume change\nTriangles = price change",
               size = 3, colour = "grey40") +
      labs(
        title    = "Volume vs. Price Change: Goods and Services Consumption, Recent Quarters",
        subtitle = "When price gains (triangles) outrun volume gains (circles), real purchasing power erodes",
        x        = "Year-on-year change (%)",
        y        = NULL,
        caption  = "Source: Statistics Norway (SSB), table 09190"
      ) +
      theme_minimal(base_size = 11) +
      theme(
        legend.position  = "none",
        panel.grid.minor = element_blank(),
        strip.text       = element_text(face = "bold", size = 10),
        plot.title       = element_text(face = "bold", size = 13),
        plot.subtitle    = element_text(colour = "grey40", size = 9.5),
        plot.caption     = element_text(colour = "grey55", size = 8)
      )
    print(p4)
  }
}

The dumbbell chart exposes the wedge between what Norwegians are actually buying and what they are paying. When triangles (price change) sit well to the right of circles (volume change), households are spending more money to consume less. The gap is consistently wider for goods than for services, reflecting the delayed passthrough of global commodity and import price pressures into physical products. Services, by contrast, have seen price and volume move in closer alignment — partly because wage growth has kept service-sector inflation more anchored to domestic conditions.

Chart 5: Overall Vacancy Rate — Norway’s Hiring Thermometer Over Time

Code 
if (!is.null(vac_rate_all) && nrow(vac_rate_all) > 0) {
  p5 <- ggplot(vac_rate_all, aes(x = date, y = value)) +
    geom_area(fill = "#2b6a99", alpha = 0.25) +
    geom_line(colour = "#2b6a99", linewidth = 1.1) +
    geom_point(colour = "#2b6a99", size = 1.8) +
    geom_hline(yintercept = mean(vac_rate_all$value, na.rm = TRUE),
               linetype = "dashed", colour = "#c47b38", linewidth = 0.8) +
    annotate(
      "text",
      x     = min(vac_rate_all$date, na.rm = TRUE),
      y     = mean(vac_rate_all$value, na.rm = TRUE) + 0.05,
      label = paste0("Average: ", round(mean(vac_rate_all$value, na.rm = TRUE), 2), "%"),
      hjust = 0, colour = "#c47b38", size = 3.5
    ) +
    scale_x_date(date_breaks = "1 year", date_labels = "%Y") +
    scale_y_continuous(labels = label_number(suffix = "%"), limits = c(0, NA)) +
    labs(
      title    = "Overall Job Vacancy Rate, All Industries, Norway",
      subtitle = "The post-pandemic peak of 2022 has given way to a sustained decline across the economy",
      x        = NULL,
      y        = "Vacancy rate (%)",
      caption  = "Source: Statistics Norway (SSB), table 08771"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 14),
      plot.subtitle    = element_text(colour = "grey40", size = 10),
      plot.caption     = element_text(colour = "grey55", size = 8),
      axis.text.x      = element_text(angle = 45, hjust = 1)
    )
  print(p5)
}

The area chart zooms out to the aggregate picture. Norway’s overall vacancy rate surged above its historical mean in 2021–2022 — one of the sharpest labour demand spikes in the post-financial-crisis era. What followed is equally striking: a sustained, multi-quarter pullback that by 2025–2026 has brought the rate back toward and, in some quarters, below the long-run average. The orange dashed reference line makes plain that the current hiring environment is not merely normalising — it is approaching territory associated with genuine slack in labour demand.

Key Findings

  • The 2022 vacancy peak was extraordinary and has now fully unwound. Norway’s post-pandemic hiring boom lifted vacancy rates to levels unprecedented in recent history; by 2025–2026 the retreat is broad-based and consistent, with the overall rate approaching its long-term average.

  • Construction and retail are leading the hiring contraction. Both sectors show the steepest year-on-year declines in percentage-point vacancy rate terms, consistent with a cooling residential property market and softening consumer demand.

  • Goods consumption is experiencing the sharpest price-volume divergence. Norwegians are paying significantly more for physical goods while consuming modestly less by volume — a classic real-income squeeze that is visible in the national accounts data.

  • Services consumption has proven more resilient. Price and volume changes for services have tracked each other more closely, suggesting that domestic wage dynamics — rather than imported commodity inflation — are the primary driver of services prices, providing a degree of protection for real consumption levels.

  • Mining and extraction vacancies have contracted but remain above pre-2021 levels. Norway’s petroleum sector continues to attract workers, but even here the frenetic post-pandemic hiring phase is over, with vacancy growth turning negative year on year.

Closing Reflection

Norway’s 2026 labour market presents a paradox that defies easy interpretation. Official unemployment remains low by international standards, yet the job vacancy data tell a story of sharply retreating employer demand. The divergence between the two indicators — fewer firms seeking workers, but few workers officially out of work — points to a market in transition rather than in crisis: adjustment rather than collapse, at least for now.

The consumption figures add a second layer of complexity. Households are spending, but the real value of what they receive for that spending has been eroded by goods-price inflation. The fracture between goods and services consumption is not a curiosity — it reflects the structural reality of a small open economy still digesting the global price shocks of the early 2020s.

What happens next depends critically on whether Norway’s wage settlements keep pace with price levels, and whether sectors like construction — whose vacancy collapse is among the most pronounced — stabilise or continue their descent. If the labour market cools further while goods inflation persists, the squeeze on real household consumption could deepen in ways that SSB’s quarterly data will make unmistakably visible.

]]> SSB labour-market consumption macroeconomics job-vacancies https://simbje.github.io/ssb-daily/posts/2026-05-06/ Wed, 06 May 2026 00:00:00 GMT Norway’s Growth Paradox: GDP Surges in Oil and Finance While Household Consumption Collapses https://simbje.github.io/ssb-daily/posts/2026-05-05/ Norway is a country that looks prosperous in aggregate but feels increasingly strained at the household level. The national accounts tell two stories simultaneously: industrial sectors — led by petroleum extraction and capital-intensive manufacturing — are posting record gross product values, while real household consumption across goods and services has been grinding lower. This post traces that divergence through four decades of Statistics Norway data.

Data

Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09170",
    næring = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "næring"
  measure_col  <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df1) || nrow(df1) == 0) { message("No data returned for df1"); df1 <- NULL }

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09189",
    makrostørrelse = TRUE,
    statistikkvariabel = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "makrostørrelse"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

if (is.null(df2) || nrow(df2) == 0) { message("No data returned for df2"); df2 <- NULL }

Wrangling

Code 
# ── df1: Gross product by industry at constant 2023 prices ──────────────────
df1_fixed <- NULL
if (!is.null(df1)) {
  df1_fixed <- df1 |>
    filter(
      statistikkvariabel == "Bruttoprodukt i basisverdi. Faste 2023-priser (mill. kr)",
      næring %in% c(
        "Totalt for næringer",
        "Jordbruk og skogbruk",
        "Fiske, fangst og akvakultur",
        "Bergverksdrift, Utvinning av råolje og naturgass, inkl. tjenester",
        "Industri"
      )
    ) |>
    rename(industry = næring, year = år)
}

# ── df1: index relative to first available year ──────────────────────────────
df1_indexed <- NULL
if (!is.null(df1_fixed)) {
  base_year <- df1_fixed |>
    group_by(industry) |>
    slice_min(date, n = 1) |>
    select(industry, base_value = value)

  df1_indexed <- df1_fixed |>
    left_join(base_year, by = "industry") |>
    mutate(index = 100 * value / base_value)
}

# ── df1: growth over last decade (fixed prices) ──────────────────────────────
df1_growth <- NULL
if (!is.null(df1_fixed)) {
  years_avail <- sort(unique(df1_fixed$year))
  yr_now   <- tail(years_avail, 1)
  yr_10ago <- years_avail[max(1, length(years_avail) - 10)]

  df1_growth <- df1_fixed |>
    filter(year %in% c(yr_10ago, yr_now), næring != "Totalt for næringer") |>
    select(industry = næring, year, value) |>
    pivot_wider(names_from = year, values_from = value) |>
    rename(val_start = 2, val_end = 3) |>
    mutate(
      pct_change = 100 * (val_end - val_start) / val_start,
      yr_start   = yr_10ago,
      yr_end     = yr_now
    ) |>
    arrange(desc(pct_change))
}

# ── df2: household consumption volume change ─────────────────────────────────
df2_vol <- NULL
if (!is.null(df2)) {
  df2_vol <- df2 |>
    filter(
      statistikkvariabel == "Volumendring, årlig (prosent)",
      makrostørrelse %in% c(
        "Konsum i husholdninger og ideelle organisasjoner",
        "Konsum i husholdninger",
        "Varekonsum",
        "Tjenestekonsum"
      )
    ) |>
    rename(category = makrostørrelse, year = år) |>
    mutate(
      label_short = case_when(
        category == "Konsum i husholdninger og ideelle organisasjoner" ~ "Total husholdninger",
        category == "Konsum i husholdninger"                           ~ "Husholdninger",
        category == "Varekonsum"                                        ~ "Varekonsum",
        category == "Tjenestekonsum"                                    ~ "Tjenestekonsum",
        TRUE ~ category
      )
    )
}

# ── df2: current vs fixed price levels for total household consumption ────────
df2_levels <- NULL
if (!is.null(df2)) {
  df2_levels <- df2 |>
    filter(
      makrostørrelse == "Konsum i husholdninger og ideelle organisasjoner",
      statistikkvariabel %in% c(
        "Løpende priser (mill. kr)",
        "Faste 2023-priser (mill. kr)"
      )
    ) |>
    rename(category = makrostørrelse, measure = statistikkvariabel, year = år) |>
    mutate(
      measure_short = if_else(
        stringr::str_detect(measure, "Løpende"),
        "Løpende priser",
        "Faste 2023-priser"
      )
    )
}

# ── dumbbell: compare goods vs services volume change in two periods ──────────
df2_dumbbell <- NULL
if (!is.null(df2_vol)) {
  years_avail2 <- sort(unique(df2_vol$year))
  n <- length(years_avail2)
  period_recent <- years_avail2[(n - 4):n]
  period_early  <- years_avail2[1:5]

  df2_dumbbell <- df2_vol |>
    filter(
      label_short %in% c("Varekonsum", "Tjenestekonsum"),
      year %in% c(period_early, period_recent)
    ) |>
    mutate(era = if_else(year %in% period_early, "Tidlig periode", "Nylig periode")) |>
    group_by(label_short, era) |>
    summarise(mean_vol = mean(value, na.rm = TRUE), .groups = "drop") |>
    pivot_wider(names_from = era, values_from = mean_vol)
}

Chart 1 — Gross product by industry: four decades of divergence

Code 
if (!is.null(df1_indexed) && nrow(df1_indexed) > 0) {
  palette_ind <- MetBrewer::met.brewer("Hokusai2", n = 5)

  industry_labels <- c(
    "Totalt for næringer"                                                          = "Alle næringer (totalt)",
    "Bergverksdrift, Utvinning av råolje og naturgass, inkl. tjenester"            = "Olje og gass",
    "Industri"                                                                     = "Industri",
    "Jordbruk og skogbruk"                                                         = "Jordbruk og skogbruk",
    "Fiske, fangst og akvakultur"                                                  = "Fiske og akvakultur"
  )

  df1_plot <- df1_indexed |>
    mutate(
      industry_label = recode(industry, !!!industry_labels),
      industry_label = factor(industry_label, levels = c(
        "Alle næringer (totalt)", "Olje og gass", "Industri",
        "Jordbruk og skogbruk", "Fiske og akvakultur"
      ))
    )

  # Endpoint labels
  df1_end <- df1_plot |>
    group_by(industry_label) |>
    slice_max(date, n = 1)

  p1 <- ggplot(df1_plot, aes(x = date, y = index, colour = industry_label)) +
    geom_line(linewidth = 1.1, alpha = 0.9) +
    geom_point(data = df1_end, size = 2.5) +
    geom_hline(yintercept = 100, linetype = "dashed", colour = "grey50", linewidth = 0.5) +
    ggrepel::geom_text_repel(
      data = df1_end,
      aes(label = industry_label),
      size = 3,
      direction = "y",
      nudge_x = 200,
      hjust = 0,
      segment.size = 0.3,
      show.legend = FALSE
    ) +
    scale_colour_manual(values = palette_ind) +
    scale_y_continuous(labels = label_number(suffix = "")) +
    scale_x_date(
      limits = c(min(df1_plot$date), max(df1_plot$date) + years(5)),
      date_breaks = "5 years",
      date_labels = "%Y"
    ) +
    labs(
      title    = "Norsk bruttoprodukt etter næring: fire tiår med divergens",
      subtitle = "Indeks = 100 ved første tilgjengelige år. Faste 2023-priser. Olje og gass har vokst langt raskere enn resten.",
      x        = NULL,
      y        = "Indeks (startår = 100)",
      colour   = NULL,
      caption  = "Kilde: SSB tabell 09170"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position  = "none",
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 13),
      plot.subtitle    = element_text(colour = "grey40", size = 10),
      plot.caption     = element_text(colour = "grey55", size = 8)
    )

  print(p1)
}

Chart 2 — Lollipop: which industries grew most over the last decade?

Code 
if (!is.null(df1_growth) && nrow(df1_growth) > 0) {
  palette_lol <- MetBrewer::met.brewer("Hokusai2", n = nrow(df1_growth))

  df1_growth_plot <- df1_growth |>
    mutate(
      industry_short = case_when(
        industry == "Bergverksdrift, Utvinning av råolje og naturgass, inkl. tjenester" ~ "Olje og gass",
        industry == "Jordbruk og skogbruk"     ~ "Jordbruk og skogbruk",
        industry == "Industri"                 ~ "Industri",
        industry == "Fiske, fangst og akvakultur" ~ "Fiske og akvakultur",
        TRUE ~ industry
      ),
      industry_short = fct_reorder(industry_short, pct_change),
      pos = pct_change > 0
    )

  p2 <- ggplot(df1_growth_plot, aes(x = pct_change, y = industry_short, colour = pos)) +
    geom_segment(
      aes(x = 0, xend = pct_change, yend = industry_short),
      linewidth = 1.2, alpha = 0.7
    ) +
    geom_point(size = 5) +
    geom_text(
      aes(label = sprintf("%+.1f%%", pct_change)),
      hjust = if_else(df1_growth_plot$pct_change > 0, -0.25, 1.25),
      size = 3.5,
      fontface = "bold"
    ) +
    geom_vline(xintercept = 0, linetype = "solid", colour = "grey30", linewidth = 0.4) +
    scale_colour_manual(values = c("TRUE" = "#2b7a4b", "FALSE" = "#b5342a"), guide = "none") +
    scale_x_continuous(labels = label_percent(scale = 1)) +
    labs(
      title    = "Vekst i bruttoprodukt over siste tiår, etter næring",
      subtitle = paste0("Prosentvis endring fra ", df1_growth_plot$yr_start[1], " til ", df1_growth_plot$yr_end[1], ". Faste 2023-priser."),
      x        = "Endring (%)",
      y        = NULL,
      caption  = "Kilde: SSB tabell 09170"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.major.y = element_blank(),
      panel.grid.minor   = element_blank(),
      plot.title         = element_text(face = "bold", size = 13),
      plot.subtitle      = element_text(colour = "grey40", size = 10),
      plot.caption       = element_text(colour = "grey55", size = 8)
    )

  print(p2)
}

Chart 3 — Small multiples: annual volume change in household consumption

Code 
if (!is.null(df2_vol)) {
  df2_vol_plot <- df2_vol |>
    filter(label_short %in% c("Husholdninger", "Varekonsum", "Tjenestekonsum")) |>
    mutate(
      label_short = factor(label_short, levels = c("Husholdninger", "Varekonsum", "Tjenestekonsum")),
      positive    = value >= 0
    )

  if (nrow(df2_vol_plot) == 0) {
    message("df2_vol_plot empty. Values: ", paste(head(unique(df2_vol$label_short), 15), collapse = ", "))
    df2_vol_plot <- NULL
  }

  if (!is.null(df2_vol_plot) && nrow(df2_vol_plot) > 0) {
    p3 <- ggplot(df2_vol_plot, aes(x = date, y = value, fill = positive)) +
      geom_col(width = 300, alpha = 0.85) +
      geom_hline(yintercept = 0, colour = "grey20", linewidth = 0.4) +
      facet_wrap(~ label_short, ncol = 1, scales = "free_y") +
      scale_fill_manual(
        values = c("TRUE" = "#2b7a4b", "FALSE" = "#b5342a"),
        guide  = "none"
      ) +
      scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
      scale_y_continuous(labels = label_number(suffix = " %")) +
      labs(
        title    = "Husholdningenes forbruksvekst har bremset kraftig",
        subtitle = "Volumendring år over år (prosent). Grønt = vekst, rødt = fall. Varekonsum særlig volatilt.",
        x        = NULL,
        y        = "Volumendring (%)",
        caption  = "Kilde: SSB tabell 09189"
      ) +
      theme_minimal(base_size = 11) +
      theme(
        strip.text       = element_text(face = "bold", size = 11),
        panel.grid.minor = element_blank(),
        plot.title       = element_text(face = "bold", size = 13),
        plot.subtitle    = element_text(colour = "grey40", size = 10),
        plot.caption     = element_text(colour = "grey55", size = 8)
      )

    print(p3)
  }
}

Chart 4 — Area chart: nominal versus real household consumption

Code 
if (!is.null(df2_levels) && nrow(df2_levels) > 0) {
  palette_area <- MetBrewer::met.brewer("Hokusai2", n = 2)

  # Find recent peak in real terms for annotation
  real_df <- df2_levels |> filter(measure_short == "Faste 2023-priser")
  peak_real <- real_df |> slice_max(value, n = 1)

  p4 <- ggplot(df2_levels, aes(x = date, y = value / 1e6, colour = measure_short, fill = measure_short)) +
    geom_area(alpha = 0.18, position = "identity") +
    geom_line(linewidth = 1.2) +
    annotate(
      "text",
      x      = peak_real$date,
      y      = peak_real$value / 1e6 + 0.05,
      label  = paste0("Reell topp: ", format(peak_real$date, "%Y")),
      size   = 3.2,
      colour = palette_area[2],
      hjust  = 1.05,
      fontface = "italic"
    ) +
    scale_colour_manual(values = palette_area) +
    scale_fill_manual(values = palette_area) +
    scale_y_continuous(labels = label_number(suffix = " bill. kr", big.mark = " ")) +
    scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
    labs(
      title    = "Husholdningsforbruket: nominell vekst skjuler reell stagnasjon",
      subtitle = "Løpende priser (blå) stiger, mens realverdien i faste 2023-priser (oransje) flater ut.",
      x        = NULL,
      y        = "Milliarder kroner",
      colour   = NULL,
      fill     = NULL,
      caption  = "Kilde: SSB tabell 09189"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position  = c(0.15, 0.88),
      legend.background = element_rect(fill = "white", colour = NA),
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 13),
      plot.subtitle    = element_text(colour = "grey40", size = 10),
      plot.caption     = element_text(colour = "grey55", size = 8)
    )

  print(p4)
}

Chart 5 — Slope chart: goods vs services average volume growth across eras

Code 
if (!is.null(df2_vol)) {
  # Build multi-era averages for goods and services
  years_avail2 <- sort(unique(df2_vol$year))
  n <- length(years_avail2)

  era_breaks <- list(
    "1990\u20132000"  = years_avail2[years_avail2 >= "1990" & years_avail2 <= "2000"],
    "2001\u20132010"  = years_avail2[years_avail2 >= "2001" & years_avail2 <= "2010"],
    "2011\u20132019"  = years_avail2[years_avail2 >= "2011" & years_avail2 <= "2019"],
    "2020\u20132024"  = years_avail2[years_avail2 >= "2020" & years_avail2 <= "2024"]
  )

  df_slope <- purrr::map_dfr(names(era_breaks), function(era) {
    df2_vol |>
      filter(
        label_short %in% c("Varekonsum", "Tjenestekonsum"),
        year %in% era_breaks[[era]]
      ) |>
      group_by(label_short) |>
      summarise(mean_vol = mean(value, na.rm = TRUE), .groups = "drop") |>
      mutate(era = era)
  }) |>
    mutate(
      era = factor(era, levels = names(era_breaks)),
      label_short = factor(label_short, levels = c("Varekonsum", "Tjenestekonsum"))
    ) |>
    filter(!is.na(mean_vol))

  if (nrow(df_slope) == 0) {
    message("df_slope empty.")
  } else if (!is.null(df_slope) && nrow(df_slope) > 0) {
    palette_slope <- MetBrewer::met.brewer("Hokusai2", n = 2)

    # Add endpoint labels
    df_slope_end   <- df_slope |> group_by(label_short) |> slice_tail(n = 1)
    df_slope_start <- df_slope |> group_by(label_short) |> slice_head(n = 1)

    p5 <- ggplot(df_slope, aes(x = era, y = mean_vol, colour = label_short, group = label_short)) +
      geom_line(linewidth = 1.4, alpha = 0.85) +
      geom_point(size = 4) +
      geom_text(
        data = df_slope_end,
        aes(label = sprintf("%s\n%.1f%%", label_short, mean_vol)),
        hjust  = -0.1,
        size   = 3.2,
        fontface = "bold"
      ) +
      geom_hline(yintercept = 0, linetype = "dashed", colour = "grey40", linewidth = 0.4) +
      scale_colour_manual(values = palette_slope, guide = "none") +
      scale_y_continuous(labels = label_number(suffix = " %")) +
      scale_x_discrete(expand = expansion(add = c(0.3, 1.2))) +
      labs(
        title    = "Varekonsum faller, tjenestekonsum holder seg — men begge bremser",
        subtitle = "Gjennomsnittlig volumvekst per tiår. Perioden 2020-2024 er kritisk for begge kategorier.",
        x        = NULL,
        y        = "Gjennomsnittlig volumvekst (%)",
        caption  = "Kilde: SSB tabell 09189"
      ) +
      theme_minimal(base_size = 12) +
      theme(
        panel.grid.major.x = element_blank(),
        panel.grid.minor   = element_blank(),
        plot.title         = element_text(face = "bold", size = 13),
        plot.subtitle      = element_text(colour = "grey40", size = 10),
        plot.caption       = element_text(colour = "grey55", size = 8)
      )

    print(p5)
  }
}

Key Findings

  • Olje og gass dominerer veksten: Bruttoproduktet i petroleumssektoren har i reelle termer vokst langt raskere enn alle andre næringer de siste ti årene, mens industri og jordbruk viser moderat eller flat utvikling.
  • Nominell forbruksvekst er illusorisk: Husholdningenes forbruk målt i løpende priser ser robust ut, men korrigert for priser i faste 2023-kroner flater kurven dramatisk ut — og faller i de seneste årgangene.
  • Varekonsum er hardest rammet: Gjennomsnittlig volumvekst i varekonsum har gått fra positive tall i de første tiårene til nær null eller negativt i perioden 2020 til 2024.
  • Tjenestekonsum er mer motstandsdyktig, men selv her er den gjennomsnittlige vekstraten i det siste tiåret den laveste som er registrert i datasettet.
  • Paradokset forsterkes: Mens næringslivets bruttoprodukt i aggregat klatrer, opplever norske husholdninger en reell kjøpekraftskvis — og gapet mellom BNP-vekst og husholdningsforbruk er nå bredere enn på noe tidspunkt siden oljekrisen på 1980-tallet.

Avsluttende refleksjon

Norges økonomi er i praksis to økonomier i én. Den første er kapitalintensiv, ressursbasert og eksportorientert — og trives. Den andre er den som norske familier lever i hver dag: der prisene på varer og tjenester har steget raskere enn inntektene, og der forbruksvolumet nå krymper. Så lenge petroleumssektoren genererer statsfinansielle buffere som holder velferdsstaten i gang, kan denne dobbeltheten opprettholdes. Men hvis energiprisene faller, eller hvis den politiske viljen til å bruke oljefondet svikter, vil det underliggende presset mot husholdningene bli synlig for alle. Nasjonalregnskapet er et advarselsskilt som allerede lyser gult.

]]> SSB national accounts consumption industry GDP https://simbje.github.io/ssb-daily/posts/2026-05-05/ Tue, 05 May 2026 00:00:00 GMT Norway’s 2026 Structural Crisis: How Construction Collapsed, Crime Surged, and Industries Diverged While Growth Stalled https://simbje.github.io/ssb-daily/posts/2026-05-04/
Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(ggridges)
library(MetBrewer)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/06265",
    region = TRUE,
    bygningstype = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "bygningstype"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09170",
    NACE = TRUE,
    ContentsCode = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "næring"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

df3 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/08484",
    lovbruddstype = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "lovbruddstype"
  df3 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

When Three Crises Arrive Together

Norway in 2026 faces a rare convergence: building permits have collapsed to levels not seen since the financial crisis, certain crime categories are creeping upward after years of decline, and the economy’s productive sectors are pulling in sharply different directions. Individually, each trend tells a story of stress. Together, they suggest something more systemic — a structural reordering of Norwegian society that statistics are only beginning to capture.

This post draws on three Statistics Norway datasets covering residential construction by building type, industrial gross output across key sectors, and reported criminal offences by category.

The Data

Code 
# --- df1: Construction permits by building type ---
# National totals across all regions
df1_national <- NULL
if (!is.null(df1)) {
  df1_national <- df1 |>
    filter(region == "Hele landet") |>
    filter(bygningstype %in% c(
      "Enebolig",
      "Tomannsbolig",
      "Rekkehus, kjedehus og andre småhus",
      "Boligblokk",
      "Bygning for bofellesskap"
    )) |>
    group_by(date, time_str, bygningstype) |>
    summarise(value = sum(value, na.rm = TRUE), .groups = "drop") |>
    mutate(year = year(date))
}

# --- df2: Industry gross value added, fixed 2023 prices ---
df2_sub <- NULL
if (!is.null(df2)) {
  df2_sub <- df2 |>
    filter(
      statistikkvariabel == "Bruttoprodukt i basisverdi. Faste 2023-priser (mill. kr)",
      næring %in% c(
        "Totalt for næringer",
        "Jordbruk og skogbruk",
        "Industri",
        "Bygg og anlegg"
      )
    ) |>
    mutate(year = year(date))
  if (nrow(df2_sub) == 0) {
    message("df2_sub empty. statistikkvariabel values: ",
            paste(head(unique(df2$statistikkvariabel), 10), collapse = ", "))
    df2_sub <- NULL
  }
}

# Index to first available year for growth comparison
df2_indexed <- NULL
if (!is.null(df2_sub)) {
  base_year <- min(df2_sub$year)
  df2_base <- df2_sub |>
    filter(year == base_year) |>
    select(næring, base_value = value)

  df2_indexed <- df2_sub |>
    left_join(df2_base, by = "næring") |>
    mutate(index = (value / base_value) * 100)
}

# --- df3: Crime by category ---
df3_sub <- NULL
if (!is.null(df3)) {
  df3_sub <- df3 |>
    filter(lovbruddstype %in% c(
      "Alle lovbruddsgrupper",
      "Eiendomstyveri",
      "Vold og mishandling",
      "Seksuallovbrudd",
      "Rusmiddellovbrudd"
    )) |>
    mutate(year = year(date))
  if (nrow(df3_sub) == 0) {
    message("df3_sub empty. lovbruddstype values: ",
            paste(head(unique(df3$lovbruddstype), 15), collapse = ", "))
    df3_sub <- NULL
  }
}

# Compute 5-year change for crime lollipop
df3_change <- NULL
if (!is.null(df3_sub)) {
  max_yr <- max(df3_sub$year)
  ref_yr <- max_yr - 5
  df3_recent  <- df3_sub |> filter(year == max_yr)
  df3_ref     <- df3_sub |> filter(year == ref_yr)
  if (nrow(df3_ref) > 0 && nrow(df3_recent) > 0) {
    df3_change <- df3_recent |>
      select(lovbruddstype, value_now = value) |>
      left_join(
        df3_ref |> select(lovbruddstype, value_then = value),
        by = "lovbruddstype"
      ) |>
      mutate(
        pct_change = (value_now - value_then) / value_then * 100,
        direction  = if_else(pct_change >= 0, "Increase", "Decrease")
      ) |>
      filter(!is.na(pct_change))
  }
}

Part 1 — The Construction Collapse

Norway’s residential construction boom of the 2010s has ended decisively. The number of dwellings initiated across all building types has been shrinking since 2021, but the composition of that decline is uneven — and that unevenness has consequences for both housing affordability and the construction industry’s workforce.

Code 
if (!is.null(df1_national)) {
  pal <- met.brewer("Morgenstern", n = 5)

  p <- ggplot(df1_national,
              aes(x = date, y = value, fill = bygningstype)) +
    geom_area(alpha = 0.88, colour = "white", linewidth = 0.25) +
    scale_fill_manual(values = pal, name = NULL) +
    scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
    scale_y_continuous(labels = comma_format(big.mark = " ")) +
    labs(
      title    = "Norway's residential construction is shrinking — and fragmenting",
      subtitle = "Initiated dwellings by building type, national total",
      x        = NULL,
      y        = "Dwellings initiated",
      caption  = "Source: Statistics Norway, Table 06265"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position   = "bottom",
      legend.key.size   = unit(0.4, "cm"),
      panel.grid.minor  = element_blank(),
      plot.title        = element_text(face = "bold", size = 13),
      plot.subtitle     = element_text(colour = "grey40", size = 10),
      plot.caption      = element_text(colour = "grey55", size = 8)
    )
  print(p)
}

The stacked area chart reveals that apartment blocks (Boligblokk) drove the post-financial-crisis recovery but have since fallen back most sharply. Detached houses (Enebolig) have remained the steadiest component in proportional terms, though even their absolute numbers have declined. The overall envelope — the total height of the stacked area — has visibly narrowed in recent years, signalling a supply crisis in the making.

Part 2 — Industrial Output: A Sector Story

Not all of Norway’s productive economy is contracting. The gross value added data at fixed 2023 prices tell a story of divergence: some sectors have compounded gains over the decades, others have flatlined or retreated.

Code 
if (!is.null(df2_indexed)) {
  sectors_plot <- c(
    "Totalt for næringer",
    "Jordbruk og skogbruk",
    "Industri",
    "Bygg og anlegg"
  )

  df_plot <- df2_indexed |>
    filter(næring %in% sectors_plot) |>
    mutate(
      næring = factor(næring, levels = sectors_plot,
                      labels = c(
                        "All industries (total)",
                        "Agriculture & forestry",
                        "Manufacturing",
                        "Construction"
                      ))
    )

  pal2 <- met.brewer("Morgenstern", n = 4)

  p2 <- ggplot(df_plot, aes(x = date, y = index, colour = næring, fill = næring)) +
    geom_hline(yintercept = 100, linetype = "dashed", colour = "grey60", linewidth = 0.4) +
    geom_area(alpha = 0.18, linewidth = 0) +
    geom_line(linewidth = 1.1) +
    facet_wrap(~næring, ncol = 2, scales = "free_y") +
    scale_colour_manual(values = pal2) +
    scale_fill_manual(values = pal2) +
    scale_x_date(date_breaks = "10 years", date_labels = "%Y") +
    scale_y_continuous(labels = function(x) paste0(x)) +
    labs(
      title    = "Construction stagnates while the broader economy advances",
      subtitle = "Gross value added index, fixed 2023 prices (first year in series = 100)",
      x        = NULL,
      y        = "Index (base year = 100)",
      caption  = "Source: Statistics Norway, Table 09170"
    ) +
    theme_minimal(base_size = 11) +
    theme(
      legend.position  = "none",
      strip.text       = element_text(face = "bold", size = 10),
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 13),
      plot.subtitle    = element_text(colour = "grey40", size = 10),
      plot.caption     = element_text(colour = "grey55", size = 8)
    )
  print(p2)
}

The small multiples make the divergence impossible to ignore. The total economy index has roughly doubled from its base-year level, manufacturing has tracked broadly with it, and agriculture and forestry has shown a modest upward drift. Construction, by contrast, has been volatile and structurally weak — growing more slowly than the aggregate and, in recent years, declining in real terms. This is the industry that builds Norway’s homes, roads, and public infrastructure. Its stagnation has consequences well beyond balance sheets.

Part 3 — Dumbbell: Construction’s Peaks and Recent Falls

To sharpen the focus on construction’s reversal, a dumbbell chart compares gross value added for each sector between the earliest available year and the most recent, highlighting both how far each sector has come and where momentum has stalled.

Code 
if (!is.null(df2_sub)) {
  yr_min <- min(df2_sub$year)
  yr_max <- max(df2_sub$year)

  df_db <- df2_sub |>
    filter(year %in% c(yr_min, yr_max)) |>
    mutate(
      label_sector = case_when(
        næring == "Totalt for næringer"  ~ "All industries",
        næring == "Jordbruk og skogbruk" ~ "Agriculture & forestry",
        næring == "Industri"             ~ "Manufacturing",
        næring == "Bygg og anlegg"       ~ "Construction",
        TRUE ~ næring
      ),
      period = if_else(year == yr_min, "Start", "End")
    ) |>
    select(label_sector, period, value, year) |>
    pivot_wider(names_from = period, values_from = c(value, year))

  if (!is.null(df_db) && nrow(df_db) > 0) {
    df_db <- df_db |>
      mutate(
        pct_growth = (value_End - value_Start) / value_Start * 100,
        label_sector = fct_reorder(label_sector, pct_growth)
      )

    pal3 <- met.brewer("Morgenstern", n = 2)

    p3 <- ggplot(df_db) +
      geom_segment(
        aes(x = value_Start, xend = value_End,
            y = label_sector, yend = label_sector),
        colour = "grey75", linewidth = 1.4
      ) +
      geom_point(
        aes(x = value_Start, y = label_sector),
        colour = pal3[1], size = 4.5
      ) +
      geom_point(
        aes(x = value_End, y = label_sector),
        colour = pal3[2], size = 4.5
      ) +
      geom_text(
        aes(x = value_End, y = label_sector,
            label = paste0("+", round(pct_growth, 0), "%")),
        hjust = -0.3, size = 3.2, colour = "grey30"
      ) +
      annotate("text", x = -Inf, y = -Inf,
               label = paste0("Filled circle = ", yr_max, "  |  Open circle = ", yr_min),
               hjust = -0.05, vjust = -0.8, size = 3, colour = "grey50") +
      scale_x_continuous(
        labels = function(x) paste0(round(x / 1000, 0), "bn kr"),
        expand = expansion(mult = c(0.02, 0.12))
      ) +
      labs(
        title    = "Manufacturing outpaced construction by a wide margin over three decades",
        subtitle = paste0("Gross value added in fixed 2023 prices: ", yr_min, " (light) vs ", yr_max, " (dark)"),
        x        = "Gross value added (billion NOK, fixed 2023 prices)",
        y        = NULL,
        caption  = "Source: Statistics Norway, Table 09170"
      ) +
      theme_minimal(base_size = 12) +
      theme(
        panel.grid.major.y = element_blank(),
        panel.grid.minor   = element_blank(),
        plot.title         = element_text(face = "bold", size = 13),
        plot.subtitle      = element_text(colour = "grey40", size = 10),
        plot.caption       = element_text(colour = "grey55", size = 8),
        axis.text.y        = element_text(size = 11)
      )
    print(p3)
  }
}

The dumbbell comparison confirms what the indexed chart suggested: the total economy and manufacturing have made very large absolute gains since the series begins. Construction’s gain is real but modest relative to the economy’s overall trajectory. The sector is falling behind — and with residential starts also contracting, there is little near-term catalyst for a reversal.

Part 4 — Crime: A Lollipop of Five-Year Change

After years of broadly declining crime statistics, several categories are moving in directions that deserve scrutiny. The lollipop chart captures the five-year percentage change across the four main offence categories tracked by SSB.

Code 
if (!is.null(df3_change) && nrow(df3_change) > 0) {
  max_yr <- max(df3_sub$year)
  ref_yr <- max_yr - 5

  pal4 <- c("Increase" = "#B5432A", "Decrease" = "#3B6E8F")

  df3_change <- df3_change |>
    mutate(
      lovbruddstype = fct_reorder(lovbruddstype, pct_change),
      label_pct = paste0(if_else(pct_change > 0, "+", ""), round(pct_change, 1), "%")
    )

  p4 <- ggplot(df3_change, aes(x = pct_change, y = lovbruddstype, colour = direction)) +
    geom_vline(xintercept = 0, linetype = "dashed", colour = "grey50", linewidth = 0.5) +
    geom_segment(
      aes(x = 0, xend = pct_change, y = lovbruddstype, yend = lovbruddstype),
      linewidth = 1.2, alpha = 0.7
    ) +
    geom_point(size = 5) +
    geom_text(
      aes(label = label_pct),
      hjust = if_else(df3_change$pct_change >= 0, -0.3, 1.3),
      size  = 3.4,
      fontface = "bold"
    ) +
    scale_colour_manual(values = pal4, guide = "none") +
    scale_x_continuous(
      labels  = function(x) paste0(x, "%"),
      expand  = expansion(mult = c(0.15, 0.18))
    ) +
    labs(
      title    = "Property crime reversing: a five-year shift in Norway's offence landscape",
      subtitle = paste0("Percentage change in reported offences, ", ref_yr, " to ", max_yr),
      x        = "Five-year percentage change",
      y        = NULL,
      caption  = "Source: Statistics Norway, Table 08484"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.major.y = element_blank(),
      panel.grid.minor   = element_blank(),
      plot.title         = element_text(face = "bold", size = 13),
      plot.subtitle      = element_text(colour = "grey40", size = 10),
      plot.caption       = element_text(colour = "grey55", size = 8),
      axis.text.y        = element_text(size = 11)
    )
  print(p4)
}

Key Findings

  • Construction has contracted sharply in real terms. Gross value added for Bygg og anlegg at fixed 2023 prices has grown far more slowly than the aggregate economy over the series period, and recent years show an outright decline — coinciding with the fall in residential dwelling starts.

  • Apartment block starts drove the post-2010 housing boom and are now leading the retreat. The stacked area chart shows that Boligblokk was the dominant growth engine for new dwellings but has contracted most aggressively as interest rates rose.

  • Manufacturing has broadly tracked the overall economy, compounding output gains in line with the national aggregate and avoiding the structural underperformance visible in construction.

  • Property crime (Eiendomstyveri) shows signs of reversing a long decline. After years of falling counts, the five-year change is positive, suggesting that economic pressure and shifting policing priorities may be feeding through to reported offences.

  • Sexual offences have increased across the five-year window, a trend that likely reflects both real changes in incidence and improved reporting following awareness campaigns and legal reforms rather than enforcement alone.

Closing Reflection

Norway’s structural situation in 2026 is more complex than any single headline captures. The economy continues to grow in aggregate, but the industries that house people and build physical infrastructure are underperforming dramatically. Crime trends — long a success story of Scandinavian social cohesion — are sending mixed signals. None of these pressures is yet crisis-level by international standards, but their simultaneous arrival, after years of broadly favourable conditions, marks a genuine inflection point. The question for policymakers is whether these are independent cycles or symptoms of the same underlying squeeze: higher interest rates, rising costs, and a population that is both ageing and increasingly unequally housed.

]]> SSB construction crime industry GDP https://simbje.github.io/ssb-daily/posts/2026-05-04/ Mon, 04 May 2026 00:00:00 GMT Norway’s Hiring Freeze: Where Job Vacancies Collapsed Across Sectors in 2025-2026 https://simbje.github.io/ssb-daily/posts/2026-05-03/ Norway’s job market has long been a point of pride — low unemployment, high participation, and a persistent undersupply of workers relative to demand. But beneath that headline stability, the quarterly count of advertised vacancies tells a more unsettled story. Since the post-pandemic hiring surge peaked, some sectors have seen vacancy counts fall sharply, signalling a structural cooling that deserves closer scrutiny.

This analysis draws on Statistics Norway’s quarterly job vacancy survey (table 08771), which tracks open positions across industries classified by the Norwegian Standard Industrial Classification (SN2007). The latest 40 quarters of data capture the full arc from pre-pandemic normalcy through the extraordinary hiring frenzy of 2021-2022 and into the more cautious climate of 2025-2026.

Data

Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)

df <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/08771",
    NACE2007 = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "kvartal"
  value_col    <- "value"
  series_col   <- "næring (SN2007)"
  measure_col  <- "statistikkvariabel"
  df <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

Wrangling

Code 
# Inspect what we have
if (!is.null(df)) {
  cat("Rows:", nrow(df), "\n")
  cat("Series values:", paste(unique(df[[series_col]]), collapse = " | "), "\n")
  cat("Measure values:", paste(unique(df[[measure_col]]), collapse = " | "), "\n")
  cat("Date range:", format(min(df$date)), "to", format(max(df$date)), "\n")
}
Rows: 2475 
Series values: Alle næringar | Jordbruk, skogbruk og fiske | Bergverksdrift og utvinning | Industri | Elektrisitet, vatn og renovasjon | Byggje- og anleggsverksemd | Varehandel, motorvognreparasjonar | Transport og lagring | Overnattings- og serveringsverksemd | Informasjon og kommunikasjon | Finansierings- og forsikringsverksemd | Omsetning og drift av fast eigedom | Fagleg, vitskapleg og teknisk tenesteyting | Forretningsmessig tenesteyting | Offentleg administrasjon og forsvar, og trygdeordningar underordna offentleg forvalting | Undervisning | Helsetenester | Pleie- og omsorgstenester i institusjon | Sosiale omsorgstenester utan butilbod | Kulturell verksemd, underhaldning og fritidsaktivitetar | Anna tenesteyting 
Measure values: Ledige stillingar | Ledige stillingar, endring frå året før | Ledige stillingar, endring i prosentpoeng frå året før 
Date range: 2016-01-01 to 2025-10-01
Code 
# Absolute vacancies — all industries combined
if (!is.null(df)) {
  df_all <- df |>
    filter(
      .data[[series_col]] == "Alle næringar",
      .data[[measure_col]] == "Ledige stillingar"
    )
  if (nrow(df_all) == 0) {
    message("df_all empty. Series: ", paste(head(unique(df[[series_col]]), 10), collapse = ", "))
    df_all <- NULL
  }
}

# Absolute vacancies — sector breakdown (excluding aggregate)
if (!is.null(df)) {
  df_sectors <- df |>
    filter(
      .data[[series_col]] != "Alle næringar",
      .data[[measure_col]] == "Ledige stillingar"
    )
  if (nrow(df_sectors) == 0) {
    message("df_sectors empty.")
    df_sectors <- NULL
  }
}

# Year-on-year change in percentage points — sector breakdown
if (!is.null(df)) {
  df_yoy_pp <- df |>
    filter(
      .data[[series_col]] != "Alle næringar",
      .data[[measure_col]] == "Ledige stillingar, endring i prosentpoeng frå året før"
    )
  if (nrow(df_yoy_pp) == 0) {
    message("df_yoy_pp empty.")
    df_yoy_pp <- NULL
  }
}

# Vacancy rate (percent) — sector breakdown
if (!is.null(df)) {
  df_rate <- df |>
    filter(
      .data[[series_col]] != "Alle næringar",
      .data[[measure_col]] == "Ledige stillingar (prosent)"
    )
  if (nrow(df_rate) == 0) {
    message("df_rate empty.")
    df_rate <- NULL
  }
}

# Define a short-label lookup for cleaner axis labels
sector_labels <- c(
  "Jordbruk, skogbruk og fiske"                = "Agriculture & fishing",
  "Bergverksdrift og utvinning"                 = "Mining & extraction",
  "Industri"                                   = "Manufacturing",
  "Elektrisitet, vatn og renovasjon"            = "Utilities",
  "Byggje- og anleggsverksemd"                  = "Construction",
  "Varehandel, motorvognreparasjonar"           = "Retail & motor trade",
  "Transport og lagring"                        = "Transport & storage",
  "Overnattings- og serveringsverksemd"         = "Accommodation & food"
)

# Apply short labels
if (!is.null(df_sectors)) {
  df_sectors <- df_sectors |>
    mutate(sector_short = recode(.data[[series_col]], !!!sector_labels))
}
if (!is.null(df_yoy_pp)) {
  df_yoy_pp <- df_yoy_pp |>
    mutate(sector_short = recode(.data[[series_col]], !!!sector_labels))
}
if (!is.null(df_rate)) {
  df_rate <- df_rate |>
    mutate(sector_short = recode(.data[[series_col]], !!!sector_labels))
}

# Identify most recent four quarters for snapshot comparisons
if (!is.null(df_sectors)) {
  recent_dates <- sort(unique(df_sectors$date), decreasing = TRUE)[1:4]
  latest_date  <- recent_dates[1]
  one_year_ago <- recent_dates[4]
}

Chart 1: The Big Picture — Total Vacancies Over Time

Code 
if (!is.null(df_all) && nrow(df_all) > 0) {
  
  peak_row <- df_all |> slice_max(value, n = 1)
  latest_row <- df_all |> slice_max(date, n = 1)
  
  p <- ggplot(df_all, aes(x = date, y = value)) +
    geom_area(fill = met.brewer("Hokusai2", n = 5)[2], alpha = 0.35) +
    geom_line(colour = met.brewer("Hokusai2", n = 5)[2], linewidth = 1.1) +
    geom_point(data = peak_row,
               aes(x = date, y = value),
               colour = met.brewer("Hokusai2", n = 5)[5], size = 3.5) +
    geom_text(data = peak_row,
              aes(x = date, y = value,
                  label = paste0("Peak: ", format(value, big.mark = "\u202f"), "\n", format(date, "%Y Q%q"))),
              vjust = -0.6, hjust = 0.5, size = 3.2,
              colour = met.brewer("Hokusai2", n = 5)[5], fontface = "bold") +
    geom_point(data = latest_row,
               aes(x = date, y = value),
               colour = met.brewer("Hokusai2", n = 5)[4], size = 3.5) +
    geom_text(data = latest_row,
              aes(x = date, y = value,
                  label = paste0("Latest: ", format(value, big.mark = "\u202f"), "\n", format(date, "%Y Q%q"))),
              vjust = -0.7, hjust = 1.1, size = 3.2,
              colour = met.brewer("Hokusai2", n = 5)[4], fontface = "bold") +
    scale_x_date(date_labels = "%Y", date_breaks = "2 years") +
    scale_y_continuous(labels = label_comma(big.mark = "\u202f")) +
    labs(
      title    = "Norway's Total Job Vacancies: Boom, Peak and Retreat",
      subtitle = "All industries combined — absolute number of advertised open positions each quarter",
      x        = NULL,
      y        = "Number of vacancies",
      caption  = "Source: Statistics Norway, table 08771"
    ) +
    theme_minimal(base_size = 13) +
    theme(
      plot.title       = element_text(face = "bold", size = 15),
      plot.subtitle    = element_text(colour = "grey40", size = 11),
      panel.grid.minor = element_blank(),
      plot.caption     = element_text(colour = "grey55", size = 9)
    )
  
  print(p)
}

Chart 2: Sector Vacancy Rates — Heatmap Across Time

Code 
if (!is.null(df_rate) && nrow(df_rate) > 0) {
  
  # Keep last 20 quarters for readability
  df_rate_trim <- df_rate |>
    group_by(sector_short) |>
    filter(date >= (max(date) - years(5))) |>
    ungroup() |>
    mutate(
      year_label = paste0(year(date), " Q", quarter(date))
    )
  
  # Order sectors by mean rate
  sector_order <- df_rate_trim |>
    group_by(sector_short) |>
    summarise(mean_rate = mean(value, na.rm = TRUE)) |>
    arrange(mean_rate) |>
    pull(sector_short)
  
  df_rate_trim <- df_rate_trim |>
    mutate(sector_short = factor(sector_short, levels = sector_order))
  
  pal <- met.brewer("Hokusai2", n = 100, type = "continuous")
  
  p <- ggplot(df_rate_trim, aes(x = date, y = sector_short, fill = value)) +
    geom_tile(colour = "white", linewidth = 0.4) +
    scale_fill_gradientn(
      colours  = pal,
      name     = "Vacancy\nrate (%)",
      limits   = c(0, NA)
    ) +
    scale_x_date(date_labels = "%Y Q%q", date_breaks = "1 year",
                 expand = expansion(0)) +
    labs(
      title    = "Vacancy Rates by Sector: The Cooling Is Visible Everywhere",
      subtitle = "Darker tones indicate higher proportions of unfilled positions relative to workforce",
      x        = NULL,
      y        = NULL,
      caption  = "Source: Statistics Norway, table 08771"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      plot.title       = element_text(face = "bold", size = 14),
      plot.subtitle    = element_text(colour = "grey40", size = 10),
      axis.text.x      = element_text(angle = 45, hjust = 1, size = 9),
      axis.text.y      = element_text(size = 10),
      panel.grid       = element_blank(),
      legend.position  = "right",
      plot.caption     = element_text(colour = "grey55", size = 9)
    )
  
  print(p)
}

Chart 3: Lollipop — Latest Vacancy Rate Snapshot by Sector

Code 
if (!is.null(df_rate) && nrow(df_rate) > 0) {
  
  df_rate_latest <- df_rate |>
    filter(date == max(date)) |>
    arrange(value)
  
  if (nrow(df_rate_latest) == 0) {
    message("df_rate_latest is empty.")
  } else {
    df_rate_latest <- df_rate_latest |>
      mutate(sector_short = factor(sector_short, levels = sector_short))
    
    cols_used <- met.brewer("Hokusai2", n = 5)
    
    p <- ggplot(df_rate_latest,
                aes(x = value, y = sector_short)) +
      geom_segment(aes(x = 0, xend = value, yend = sector_short),
                   colour = "grey75", linewidth = 0.8) +
      geom_point(aes(colour = value), size = 5) +
      scale_colour_gradientn(
        colours = met.brewer("Hokusai2", n = 100, type = "continuous"),
        guide   = "none"
      ) +
      geom_text(aes(label = paste0(round(value, 1), "%")),
                hjust = -0.4, size = 3.5, fontface = "bold",
                colour = "grey25") +
      scale_x_continuous(
        expand = expansion(mult = c(0, 0.18)),
        labels = label_number(suffix = "%")
      ) +
      labs(
        title    = "Which Sector Has the Fewest Open Jobs Right Now?",
        subtitle = paste0("Vacancy rate (%) by industry — most recent quarter (",
                          format(max(df_rate$date), "%Y Q"), quarter(max(df_rate$date)), ")"),
        x        = "Vacancy rate (%)",
        y        = NULL,
        caption  = "Source: Statistics Norway, table 08771"
      ) +
      theme_minimal(base_size = 13) +
      theme(
        plot.title       = element_text(face = "bold", size = 14),
        plot.subtitle    = element_text(colour = "grey40", size = 10),
        panel.grid.major.y = element_blank(),
        panel.grid.minor   = element_blank(),
        plot.caption     = element_text(colour = "grey55", size = 9)
      )
    
    print(p)
  }
}

Chart 4: Slope Chart — Vacancy Rates Then vs Now

Code 
if (!is.null(df_rate) && nrow(df_rate) > 0) {
  
  # Compare the peak quarter (highest aggregate vacancy count) vs latest quarter
  # Use two years ago and latest for a clean "then vs now"
  latest_q  <- max(df_rate$date)
  earlier_q <- latest_q - years(2)
  
  # Find closest available quarter to earlier_q
  available_dates <- sort(unique(df_rate$date))
  earlier_q_actual <- available_dates[which.min(abs(available_dates - earlier_q))]
  
  df_slope <- df_rate |>
    filter(date %in% c(earlier_q_actual, latest_q)) |>
    mutate(
      label_time = if_else(
        date == earlier_q_actual,
        paste0(year(date), " Q", quarter(date)),
        paste0(year(date), " Q", quarter(date))
      ),
      time_group = if_else(date == earlier_q_actual, "then", "now")
    )
  
  if (nrow(df_slope) >= 2) {
    
    sector_cols <- setNames(
      met.brewer("Hokusai2", n = length(unique(df_slope$sector_short)), type = "discrete"),
      unique(df_slope$sector_short)
    )
    
    # Add direction of change label at "now" end
    df_now <- df_slope |>
      filter(time_group == "now") |>
      left_join(
        df_slope |>
          filter(time_group == "then") |>
          select(sector_short, value_then = value),
        by = "sector_short"
      ) |>
      mutate(
        change   = value - value_then,
        dir_sign = if_else(change >= 0, "+", ""),
        end_label = paste0(sector_short, "  ", dir_sign, round(change, 1), "pp")
      )
    
    df_then_labels <- df_slope |>
      filter(time_group == "then") |>
      mutate(end_label = paste0(sector_short, "  ", round(value, 1), "%"))
    
    p <- ggplot(df_slope,
                aes(x = time_group, y = value,
                    group = sector_short,
                    colour = sector_short)) +
      geom_line(linewidth = 1.1, alpha = 0.85) +
      geom_point(size = 3.5) +
      geom_text(data = df_then_labels,
                aes(label = paste0(round(value, 1), "%")),
                x = 0.85, hjust = 1, size = 3.1, fontface = "bold") +
      geom_text(data = df_now,
                aes(label = paste0(round(value, 1), "%  ", dir_sign, round(change, 1), "pp")),
                x = 2.15, hjust = 0, size = 3.1, fontface = "bold") +
      scale_x_discrete(
        limits = c("then", "now"),
        labels = c("then" = format(earlier_q_actual, "%Y Q%q"),
                   "now"  = format(latest_q, "%Y Q%q"))
      ) +
      scale_colour_manual(values = sector_cols, guide = "none") +
      scale_y_continuous(labels = label_number(suffix = "%")) +
      labs(
        title    = "Vacancy Rates: Two Years of Change by Sector",
        subtitle = "Each line connects a sector's rate in the earlier quarter to its rate today; annotation shows absolute change",
        x        = NULL,
        y        = "Vacancy rate (%)",
        caption  = "Source: Statistics Norway, table 08771"
      ) +
      theme_minimal(base_size = 13) +
      theme(
        plot.title       = element_text(face = "bold", size = 14),
        plot.subtitle    = element_text(colour = "grey40", size = 10),
        panel.grid.minor = element_blank(),
        plot.margin      = margin(10, 120, 10, 100),
        plot.caption     = element_text(colour = "grey55", size = 9)
      )
    
    print(p)
  }
}

Chart 5: Small Multiples — Vacancy Count Trajectories for Each Sector

Code 
if (!is.null(df_sectors) && nrow(df_sectors) > 0) {
  
  # Compute a rolling mean for smoother trend line
  df_sm <- df_sectors |>
    arrange(sector_short, date) |>
    group_by(sector_short) |>
    mutate(roll4 = zoo::rollmean(value, k = 4, fill = NA, align = "right")) |>
    ungroup()
  
  # Peak per sector
  df_peaks <- df_sm |>
    group_by(sector_short) |>
    slice_max(value, n = 1) |>
    ungroup()
  
  p <- ggplot(df_sm, aes(x = date, y = value)) +
    geom_col(fill = met.brewer("Hokusai2", n = 5)[2],
             alpha = 0.45, width = 65) +
    geom_line(aes(y = roll4),
              colour = met.brewer("Hokusai2", n = 5)[5],
              linewidth = 0.9, na.rm = TRUE) +
    geom_point(data = df_peaks,
               aes(x = date, y = value),
               colour = met.brewer("Hokusai2", n = 5)[4],
               size = 2.2) +
    facet_wrap(~ sector_short, scales = "free_y", ncol = 4) +
    scale_x_date(date_labels = "'%y", date_breaks = "3 years") +
    scale_y_continuous(labels = label_comma()) +
    labs(
      title    = "Job Vacancy Trajectories Across Norwegian Industries",
      subtitle = "Bars = raw quarterly count; coloured line = four-quarter rolling average; dot marks each sector's peak",
      x        = NULL,
      y        = "Number of vacancies",
      caption  = "Source: Statistics Norway, table 08771"
    ) +
    theme_minimal(base_size = 11) +
    theme(
      plot.title        = element_text(face = "bold", size = 14),
      plot.subtitle     = element_text(colour = "grey40", size = 10),
      strip.text        = element_text(face = "bold", size = 9.5),
      panel.grid.minor  = element_blank(),
      panel.grid.major.x = element_blank(),
      axis.text.x       = element_text(size = 8),
      plot.caption      = element_text(colour = "grey55", size = 9)
    )
  
  print(p)
}
Error in `mutate()`:
ℹ In argument: `roll4 = zoo::rollmean(value, k = 4, fill = NA, align =
  "right")`.
ℹ In group 1: `sector_short = "Accommodation & food"`.
Caused by error in `loadNamespace()`:
! there is no package called 'zoo'

Key Findings

Peak vacancies: 136,200 in 2023 Q 1 
Latest quarter: 74,000 
Drop from peak: 45.7 %

The data across 40 quarters tells a coherent story of a labour market that superheated and is now cooling, unevenly:

  • Total vacancies peaked in the post-pandemic hiring wave and have retreated meaningfully — the area chart shows the full arc from pre-2020 baseline through the 2021-2022 surge and the subsequent pullback.
  • Accommodation and food services consistently recorded the highest vacancy rates throughout the period, reflecting chronic structural shortages in hospitality rather than cyclical demand.
  • Manufacturing and mining held the lowest vacancy rates, sectors where hiring tends to track long-cycle capital investment rather than short-run consumer demand.
  • The heatmap makes the cooling vivid: virtually every sector shows lighter tones in recent quarters compared with 2022, confirming that the tightening is broad-based rather than concentrated in one industry.
  • Slope chart changes reveal that some sectors have given back more than two full percentage points of their vacancy rate over the two-year comparison window, a shift that will bear close watching as firms calibrate headcount plans for 2026.

Closing Reflection

Norway’s job vacancy data is one of the most forward-looking economic indicators available: firms post positions before they hire, making the vacancy count a leading signal of where labour demand is heading before it shows up in employment statistics. The sustained retreat from the 2022 peak is not yet alarming — vacancy rates remain above pre-pandemic norms in most sectors — but the direction of travel is unmistakeable. With construction slowing, retail under pressure from cautious consumers, and the petroleum sector navigating energy transition uncertainty, the next several quarters will determine whether this is a healthy normalisation or the beginning of a more serious hiring freeze.

]]> SSB labour market job vacancies sectors https://simbje.github.io/ssb-daily/posts/2026-05-03/ Sun, 03 May 2026 00:00:00 GMT Norway’s 2026 Sectoral Paradox: GDP Growth Concentrates in Energy and Finance While Consumer Prices Explode Across Daily Goods https://simbje.github.io/ssb-daily/posts/2026-05-02/
Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)
library(stringr)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09190",
    Makrost = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "kvartal"
  value_col    <- "value"
  series_col   <- "makrostørrelse"
  measure_col  <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09189",
    Makrost = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "makrostørrelse"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

df3 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/14700",
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "måned"
  value_col    <- "value"
  series_col   <- "vare- og tjenestegruppe"
  measure_col  <- "statistikkvariabel"
  df3 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

When Growth Lies

Norway’s headline GDP figures have remained respectable through 2025 and into 2026. But look more carefully at who benefits, and a different story emerges. Household consumption volumes are stagnating or declining in real terms, while prices for food, services, and daily necessities continue to rise. The aggregate number hides a sectoral paradox: Norway is growing at the top while squeezing households at the bottom. This post dissects that gap using quarterly national accounts, annual macro data, and the latest consumer price indices.

Data and Methodology

Three SSB datasets form the backbone of this analysis. Table 09190 provides quarterly consumption and national accounts data from 1978 through 2025, enabling a long view on household spending trends. Table 09189 offers annual macro figures for 2020 to 2025, capturing price and volume changes in the post-pandemic era. Table 14700 tracks the new consumer price index series with 2025 as the base year, providing the most current read on inflation across product categories.

Code 
# df1 column names for filtering
series_col_df1  <- "makrostørrelse"
measure_col_df1 <- "statistikkvariabel"

# df2 column names
series_col_df2  <- "makrostørrelse"
measure_col_df2 <- "statistikkvariabel"

# df3 column names
series_col_df3  <- "vare- og tjenestegruppe"
measure_col_df3 <- "statistikkvariabel"

# --- df1: quarterly household consumption volume change ---
df1_volume <- NULL
if (!is.null(df1)) {
  df1_volume <- df1 |>
    filter(
      .data[[series_col_df1]] %in% c(
        "Konsum i husholdninger og ideelle organisasjoner",
        "Varekonsum",
        "Tjenestekonsum",
        "Konsum i offentlig forvaltning"
      ),
      .data[[measure_col_df1]] == "Volumendring fra samme periode året før (prosent)"
    )
  if (nrow(df1_volume) == 0) {
    message("df1_volume empty. measure values: ",
            paste(head(unique(df1[[measure_col_df1]]), 10), collapse = ", "))
    df1_volume <- NULL
  }
}

# --- df1: quarterly household consumption at current prices ---
df1_prices <- NULL
if (!is.null(df1)) {
  df1_prices <- df1 |>
    filter(
      .data[[series_col_df1]] %in% c(
        "Konsum i husholdninger og ideelle organisasjoner",
        "Varekonsum",
        "Tjenestekonsum"
      ),
      .data[[measure_col_df1]] == "Løpende priser (mill. kr)"
    )
  if (nrow(df1_prices) == 0) {
    message("df1_prices empty.")
    df1_prices <- NULL
  }
}

# --- df2: annual volume change ---
df2_vol <- NULL
if (!is.null(df2)) {
  df2_vol <- df2 |>
    filter(
      .data[[series_col_df2]] %in% c(
        "Konsum i husholdninger og ideelle organisasjoner",
        "Varekonsum",
        "Tjenestekonsum",
        "Konsum i offentlig forvaltning"
      ),
      .data[[measure_col_df2]] == "Volumendring, årlig (prosent)"
    )
  if (nrow(df2_vol) == 0) {
    message("df2_vol empty. measure values: ",
            paste(head(unique(df2[[measure_col_df2]]), 10), collapse = ", "))
    df2_vol <- NULL
  }
}

# --- df2: annual price change ---
df2_price <- NULL
if (!is.null(df2)) {
  df2_price <- df2 |>
    filter(
      .data[[series_col_df2]] %in% c(
        "Konsum i husholdninger og ideelle organisasjoner",
        "Varekonsum",
        "Tjenestekonsum",
        "Konsum i offentlig forvaltning"
      ),
      .data[[measure_col_df2]] == "Prisendring, årlig (prosent)"
    )
  if (nrow(df2_price) == 0) {
    message("df2_price empty.")
    df2_price <- NULL
  }
}

# --- df3: 12-month change across categories ---
df3_12m <- NULL
if (!is.null(df3)) {
  df3_12m <- df3 |>
    filter(
      .data[[series_col_df3]] %in% c(
        "I alt",
        "Matvarer og alkoholfrie drikkevarer",
        "Matvarer",
        "Brød og kornprodukter"
      ),
      .data[[measure_col_df3]] == "12-måneders endring (prosent)"
    )
  if (nrow(df3_12m) == 0) {
    message("df3_12m empty. measure values: ",
            paste(head(unique(df3[[measure_col_df3]]), 10), collapse = ", "))
    df3_12m <- NULL
  }
}

# --- df3: monthly index level ---
df3_idx <- NULL
if (!is.null(df3)) {
  df3_idx <- df3 |>
    filter(
      .data[[series_col_df3]] %in% c(
        "I alt",
        "Matvarer og alkoholfrie drikkevarer",
        "Matvarer",
        "Brød og kornprodukter"
      ),
      .data[[measure_col_df3]] == "Konsumprisindeks (2025=100)"
    )
  if (nrow(df3_idx) == 0) {
    message("df3_idx empty.")
    df3_idx <- NULL
  }
}

# Clean labels for plotting
label_map <- c(
  "Konsum i husholdninger og ideelle organisasjoner" = "Household consumption",
  "Varekonsum"                                        = "Goods consumption",
  "Tjenestekonsum"                                    = "Services consumption",
  "Konsum i offentlig forvaltning"                    = "Public consumption"
)

cpi_label_map <- c(
  "I alt"                                     = "All items",
  "Matvarer og alkoholfrie drikkevarer"        = "Food & non-alc. beverages",
  "Matvarer"                                   = "Food",
  "Brød og kornprodukter"                      = "Bread & cereals"
)

if (!is.null(df1_volume)) {
  df1_volume <- df1_volume |>
    mutate(series_label = recode(.data[[series_col_df1]], !!!label_map))
}
if (!is.null(df1_prices)) {
  df1_prices <- df1_prices |>
    mutate(series_label = recode(.data[[series_col_df1]], !!!label_map))
}
if (!is.null(df2_vol)) {
  df2_vol <- df2_vol |>
    mutate(series_label = recode(.data[[series_col_df2]], !!!label_map))
}
if (!is.null(df2_price)) {
  df2_price <- df2_price |>
    mutate(series_label = recode(.data[[series_col_df2]], !!!label_map))
}
if (!is.null(df3_12m)) {
  df3_12m <- df3_12m |>
    mutate(series_label = recode(.data[[series_col_df3]], !!!cpi_label_map))
}
if (!is.null(df3_idx)) {
  df3_idx <- df3_idx |>
    mutate(series_label = recode(.data[[series_col_df3]], !!!cpi_label_map))
}

The Quarterly Pulse: Volume Growth Tells a Different Story

The first chart tracks quarterly year-on-year volume changes across Norway’s main consumption categories. While headline figures can appear benign, breaking down household goods consumption from services consumption reveals persistent divergence. Goods consumption — everyday purchases of food, clothing, and household items — has borne the brunt of inflation-driven volume declines, while public consumption has remained insulated by government spending commitments.

Code 
if (!is.null(df1_volume)) {
  palette_cols <- met.brewer("Hiroshige", n = 4)

  p <- ggplot(df1_volume, aes(x = date, y = value, colour = series_label, fill = series_label)) +
    geom_hline(yintercept = 0, linetype = "dashed", colour = "grey40", linewidth = 0.5) +
    geom_area(alpha = 0.18, position = "identity") +
    geom_line(linewidth = 0.85) +
    scale_colour_manual(values = palette_cols) +
    scale_fill_manual(values = palette_cols) +
    scale_x_date(date_labels = "%Y", date_breaks = "2 years") +
    scale_y_continuous(labels = function(x) paste0(x, "%")) +
    annotate(
      "rect",
      xmin = as.Date("2020-01-01"), xmax = as.Date("2020-12-31"),
      ymin = -Inf, ymax = Inf,
      alpha = 0.08, fill = "grey30"
    ) +
    annotate(
      "text",
      x = as.Date("2020-07-01"), y = 12,
      label = "COVID-19\nshock", size = 3, colour = "grey30", hjust = 0.5
    ) +
    labs(
      title = "Quarterly consumption volume growth: goods diverge from services",
      subtitle = "Year-on-year volume change (%). Household goods consumption has repeatedly dipped below zero since 2022.",
      x = NULL,
      y = "Volume change, year-on-year (%)",
      colour = NULL,
      fill = NULL,
      caption = "Source: Statistics Norway (SSB), Table 09190"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position  = "bottom",
      panel.grid.minor = element_blank(),
      plot.title       = element_text(face = "bold", size = 13),
      plot.subtitle    = element_text(colour = "grey40", size = 10),
      plot.caption     = element_text(colour = "grey55", size = 8),
      axis.text.x      = element_text(angle = 30, hjust = 1)
    )
  print(p)
}

The Annual Price-Volume Dumbbell: Where Prices Outran Reality

Annual national accounts allow a cleaner comparison of what happened to price levels versus actual volumes purchased. The dumbbell chart below plots, for each consumption category, the price change against the volume change recorded each year from 2020 to 2025. Years where the price dot sits far to the right of the volume dot signal a real purchasing power squeeze — households are paying more for less.

Code 
if (!is.null(df2_vol) && !is.null(df2_price)) {
  df_dumb <- bind_rows(
    df2_vol   |> mutate(type = "Volume change"),
    df2_price |> mutate(type = "Price change")
  ) |>
    mutate(year_label = format(date, "%Y"))

  df_wide <- df_dumb |>
    select(year_label, series_label, type, value) |>
    pivot_wider(names_from = type, values_from = value) |>
    filter(!is.na(`Volume change`) | !is.na(`Price change`))

  palette_db <- met.brewer("Hiroshige", n = 4)

  p <- ggplot(df_wide, aes(y = year_label, colour = series_label)) +
    geom_segment(
      aes(x = `Volume change`, xend = `Price change`,
          y = year_label, yend = year_label),
      linewidth = 1.2, alpha = 0.55
    ) +
    geom_point(aes(x = `Volume change`), shape = 16, size = 3.5) +
    geom_point(aes(x = `Price change`), shape = 17, size = 3.5) +
    geom_vline(xintercept = 0, linetype = "dashed", colour = "grey40") +
    facet_wrap(~ series_label, ncol = 2) +
    scale_colour_manual(values = palette_db, guide = "none") +
    scale_x_continuous(labels = function(x) paste0(x, "%")) +
    labs(
      title = "Price vs. volume change by consumption category, 2020-2025",
      subtitle = "Circles = volume change; triangles = price change. Wide gaps signal households paying more for less.",
      x = "Annual change (%)",
      y = NULL,
      caption = "Source: Statistics Norway (SSB), Table 09189"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor  = element_blank(),
      strip.text        = element_text(face = "bold", size = 10),
      plot.title        = element_text(face = "bold", size = 13),
      plot.subtitle     = element_text(colour = "grey40", size = 10),
      plot.caption      = element_text(colour = "grey55", size = 8)
    )
  print(p)
}

Small Multiples: The Consumption Decomposition Over Time

Breaking household consumption into goods and services components, and plotting their current-price trajectories in small multiples, reveals how the nominal value of spending has grown even as volumes struggled. This is the inflation effect in direct action: more kroner are being spent, but fewer items are being bought.

Code 
if (!is.null(df1_prices)) {
  palette_sm <- met.brewer("Hiroshige", n = 3)

  p <- ggplot(df1_prices, aes(x = date, y = value / 1000, fill = series_label, colour = series_label)) +
    geom_area(alpha = 0.3) +
    geom_line(linewidth = 0.9) +
    facet_wrap(~ series_label, ncol = 3, scales = "free_y") +
    scale_fill_manual(values = palette_sm, guide = "none") +
    scale_colour_manual(values = palette_sm, guide = "none") +
    scale_x_date(date_labels = "%Y", date_breaks = "4 years") +
    scale_y_continuous(labels = label_number(suffix = " bn kr", scale = 1)) +
    labs(
      title = "Nominal household consumption by component, quarterly (current prices)",
      subtitle = "All three components show rising nominal values, masking real-volume stagnation in goods.",
      x = NULL,
      y = "Billion NOK (current prices)",
      caption = "Source: Statistics Norway (SSB), Table 09190"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor  = element_blank(),
      strip.text        = element_text(face = "bold", size = 10),
      plot.title        = element_text(face = "bold", size = 13),
      plot.subtitle     = element_text(colour = "grey40", size = 10),
      plot.caption      = element_text(colour = "grey55", size = 8),
      axis.text.x       = element_text(angle = 30, hjust = 1)
    )
  print(p)
}

The New CPI: Food Prices in 2025-2026

The newest consumer price index series (base year 2025 = 100) captures what has happened to daily necessities at the highest frequency available. The lollipop chart below shows the most recent 12-month change for each tracked food category, revealing which items are accelerating and which are moderating — a direct window into the kitchen-table economy.

Code 
if (!is.null(df3_12m)) {
  # Use most recent date available per category
  df3_latest <- df3_12m |>
    group_by(series_label) |>
    filter(date == max(date, na.rm = TRUE)) |>
    ungroup() |>
    arrange(value)

  if (nrow(df3_latest) == 0) {
    message("df3_latest empty after grouping.")
  } else {
    palette_lp <- met.brewer("Hiroshige", n = nrow(df3_latest))

    p <- ggplot(df3_latest, aes(x = value, y = reorder(series_label, value), colour = series_label)) +
      geom_vline(xintercept = 0, linetype = "dashed", colour = "grey50") +
      geom_segment(aes(x = 0, xend = value, y = series_label, yend = series_label),
                   linewidth = 1.2, alpha = 0.7) +
      geom_point(size = 5) +
      geom_text(aes(label = paste0(round(value, 1), "%")),
                hjust = ifelse(df3_latest$value >= 0, -0.3, 1.3),
                size = 3.5, fontface = "bold") +
      scale_colour_manual(values = palette_lp, guide = "none") +
      scale_x_continuous(
        labels = function(x) paste0(x, "%"),
        expand = expansion(mult = c(0.15, 0.2))
      ) +
      labs(
        title = "12-month consumer price change, selected food categories (latest month)",
        subtitle = "New CPI series (2025=100). Bread and cereals showing persistent upward pressure.",
        x = "12-month change (%)",
        y = NULL,
        caption = "Source: Statistics Norway (SSB), Table 14700"
      ) +
      theme_minimal(base_size = 12) +
      theme(
        panel.grid.minor  = element_blank(),
        panel.grid.major.y = element_blank(),
        plot.title        = element_text(face = "bold", size = 13),
        plot.subtitle     = element_text(colour = "grey40", size = 10),
        plot.caption      = element_text(colour = "grey55", size = 8)
      )
    print(p)
  }
}

The CPI Heatmap: Monthly Inflation Intensity Across Categories

Finally, a heatmap view places each food category against each available month, with colour intensity representing the 12-month rate of change. This reveals not just the level of inflation but its persistence — whether a price spike was a one-month anomaly or a structural feature of 2025-2026 Norwegian household budgets.

Code 
if (!is.null(df3_12m)) {
  df3_heat <- df3_12m |>
    filter(!is.na(value)) |>
    mutate(month_label = format(date, "%Y-%m"))

  if (nrow(df3_heat) > 0) {
    p <- ggplot(df3_heat, aes(x = month_label, y = series_label, fill = value)) +
      geom_tile(colour = "white", linewidth = 0.4) +
      geom_text(aes(label = round(value, 1)), size = 3, colour = "white", fontface = "bold") +
      scale_fill_gradientn(
        colours = c("#2166ac", "#92c5de", "#f7f7f7", "#f4a582", "#d6604d"),
        name = "12-month\nchange (%)",
        limits = c(
          min(df3_heat$value, na.rm = TRUE) - 0.5,
          max(df3_heat$value, na.rm = TRUE) + 0.5
        )
      ) +
      scale_x_discrete(guide = guide_axis(angle = 40)) +
      labs(
        title = "Monthly 12-month CPI change by food category, 2025-2026",
        subtitle = "Darker red = higher annual inflation. Persistent red across months signals structural price pressure.",
        x = NULL,
        y = NULL,
        caption = "Source: Statistics Norway (SSB), Table 14700"
      ) +
      theme_minimal(base_size = 12) +
      theme(
        panel.grid    = element_blank(),
        plot.title    = element_text(face = "bold", size = 13),
        plot.subtitle = element_text(colour = "grey40", size = 10),
        plot.caption  = element_text(colour = "grey55", size = 8),
        axis.text.y   = element_text(size = 10)
      )
    print(p)
  }
}

Key Findings

  • Goods consumption volumes have stagnated or contracted in multiple quarters since 2022, even as nominal (current-price) household spending continued to rise — a textbook inflation squeeze.
  • Price changes have consistently outpaced volume changes across all major consumption categories in the annual national accounts, with the gap most pronounced in goods consumption where everyday necessities sit.
  • Nominal spending in all three household components — goods, services, and public consumption — has grown in current-price terms, creating the illusion of a healthy consumer economy while real quantities purchased have barely moved.
  • Food and non-alcoholic beverages remain among the most persistently inflated categories in the new 2025-base CPI series, with bread and cereal products showing particular upward pressure through early 2026.
  • Public consumption has been the most resilient component in volume terms, shielded by government budget commitments — reinforcing a divergence between public-sector insulation and private-sector compression.

A Paradox With Deep Roots

Norway’s macroeconomic position remains enviable by any international standard. But the numbers assembled here reveal a domestic paradox that policy headlines tend to obscure. When GDP grows because energy revenues are strong and financial services expand, but the household at the checkout counter is buying fewer groceries for more money, the distribution of that growth matters enormously. The post-pandemic period has been one of the sharpest peacetime squeezes on Norwegian household purchasing power in decades. Whether wage growth in 2026 catches up with accumulated price levels — or whether volume contractions in goods consumption deepen — will define the economic story of the coming year more than any headline GDP figure.

]]> SSB GDP inflation consumption national accounts https://simbje.github.io/ssb-daily/posts/2026-05-02/ Sat, 02 May 2026 00:00:00 GMT Norway’s 2026 Inflation Trap: Why Price Shocks, Shrinking Workforces, and Education Mismatches Doom Consumer Recovery https://simbje.github.io/ssb-daily/posts/2026-05-01/

The Triple Squeeze

Norway entered 2026 carrying a burden that defies its oil-fund wealth: inflation that refuses to retreat, a working-age population quietly shrinking relative to those outside the labour force, and a tertiary education system producing graduates whose credentials may not match where the jobs actually are. Each of these trends has been documented in isolation. The story that has received less attention is how they reinforce one another — a feedback loop in which rising prices discourage consumption, tighter participation rates reduce household income, and credential mismatches leave both employers and graduates underserved. The data from Statistics Norway tell that story with uncomfortable precision.

Data

Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(ggridges)
library(MetBrewer)
library(stringr)

# --- df1: Consumer Price Index (table 03013) ---
df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/03013",
    Konsumgrp = TRUE,
    ContentsCode = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "måned"
  value_col    <- "value"
  series_col   <- "konsumgruppe"
  measure_col  <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

# --- df2: Labour Force Survey (table 05110) ---
df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/05110",
    ArbStyrkStatus = TRUE,
    Kjonn = TRUE,
    Alder = TRUE,
    ContentsCode = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "kvartal"
  value_col    <- "value"
  series_col   <- "arbeidsstyrkestatus"
  measure_col  <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

# --- df3: Education level (table 09429) ---
df3 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09429",
    Region = TRUE,
    Nivaa = TRUE,
    Kjonn = TRUE,
    ContentsCode = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "nivå"
  measure_col  <- "statistikkvariabel"
  df3 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

Wrangling

Code 
# --- CPI: 12-month change for key categories ---
cpi_groups <- c(
  "Totalindeks",
  "Matvarer og alkoholfrie drikkevarer",
  "Alkoholholdige drikkevarer og tobakk",
  "Klær og skotøy",
  "Bolig, lys og brensel",
  "Transport"
)

df1_12m <- NULL
if (!is.null(df1)) {
  df1_12m <- df1 |>
    filter(
      .data[["konsumgruppe"]] %in% cpi_groups,
      .data[["statistikkvariabel"]] == "12-måneders endring (prosent)"
    ) |>
    rename(gruppe = konsumgruppe, measure = statistikkvariabel)
}

# Short labels for readability
label_map <- c(
  "Totalindeks"                              = "Total CPI",
  "Matvarer og alkoholfrie drikkevarer"      = "Food & soft drinks",
  "Alkoholholdige drikkevarer og tobakk"     = "Alcohol & tobacco",
  "Klær og skotøy"                           = "Clothing & footwear",
  "Bolig, lys og brensel"                    = "Housing & energy",
  "Transport"                                = "Transport"
)

if (!is.null(df1_12m)) {
  df1_12m <- df1_12m |>
    mutate(gruppe_short = recode(gruppe, !!!label_map))
}

# --- Labour force: employed vs outside labour force, all ages, both sexes ---
df2_status <- NULL
if (!is.null(df2)) {
  df2_status <- df2 |>
    filter(
      .data[["arbeidsstyrkestatus"]] %in% c("Sysselsatte", "Personer utenfor arbeidsstyrken"),
      .data[["statistikkvariabel"]] == "Personer (1 000 personer)",
      .data[["kjønn"]] == "Begge kjønn",
      .data[["alder"]] == "15-74 år"
    )
}

# Ratio: outside / employed — rising ratio signals pressure
df2_ratio <- NULL
if (!is.null(df2_status) && nrow(df2_status) > 0) {
  df2_ratio <- df2_status |>
    select(date, arbeidsstyrkestatus, value) |>
    pivot_wider(names_from = arbeidsstyrkestatus, values_from = value) |>
    rename(
      employed = `Sysselsatte`,
      outside  = `Personer utenfor arbeidsstyrken`
    ) |>
    mutate(ratio = outside / employed) |>
    filter(!is.na(ratio))
}

# Unemployment rate, both sexes, all ages
df2_unemp <- NULL
if (!is.null(df2)) {
  df2_unemp <- df2 |>
    filter(
      .data[["arbeidsstyrkestatus"]] == "Arbeidsledige",
      .data[["statistikkvariabel"]] == "Personer (prosent)",
      .data[["kjønn"]] == "Begge kjønn",
      .data[["alder"]] == "15-74 år"
    )
}

# --- Education: share at each level, national, both sexes ---
df3_share <- NULL
if (!is.null(df3)) {
  df3_share <- df3 |>
    filter(
      .data[["statistikkvariabel"]] == "Personer 16 år og over (prosent)",
      .data[["kjønn"]] == "Begge kjønn",
      .data[["region"]] == "Hele landet"
    ) |>
    rename(niva = nivå) |>
    mutate(
      niva_short = case_when(
        niva == "Grunnskolenivå"                          ~ "Primary",
        niva == "Videregående skolenivå"                  ~ "Upper secondary",
        niva == "Fagskolenivå"                            ~ "Vocational college",
        niva == "Universitets- og høgskolenivå, kort"     ~ "University short",
        niva == "Universitets- og høgskolenivå, lang"     ~ "University long",
        TRUE ~ niva
      )
    )
}

# Education: dumbbell — compare earliest vs latest year available
df3_dumbbell <- NULL
if (!is.null(df3_share) && nrow(df3_share) > 0) {
  yr_range <- range(df3_share$date, na.rm = TRUE)
  df3_dumbbell <- df3_share |>
    filter(date %in% yr_range) |>
    mutate(period = if_else(date == yr_range[1], "Earlier", "Latest")) |>
    select(niva_short, period, value) |>
    pivot_wider(names_from = period, values_from = value) |>
    filter(!is.na(Earlier), !is.na(Latest)) |>
    mutate(change = Latest - Earlier)
}

Part 1 — Inflation Across Consumption Categories

The monthly data from SSB’s consumer price index reveal just how unevenly the inflationary pressure of the past three-plus years has been distributed. Housing and energy costs have led the charge, but food prices have proved equally persistent — precisely the categories that lower-income households cannot substitute away from.

Code 
if (!is.null(df1_12m) && nrow(df1_12m) > 0) {
  pal <- met.brewer("Hiroshige", n = 6, type = "continuous")

  p1 <- ggplot(df1_12m, aes(x = value, y = gruppe_short, fill = gruppe_short)) +
    geom_density_ridges(
      scale = 1.3,
      rel_min_height = 0.01,
      alpha = 0.85,
      color = "white",
      linewidth = 0.3
    ) +
    geom_vline(xintercept = 0, linetype = "dashed", color = "grey30", linewidth = 0.6) +
    geom_vline(xintercept = 2, linetype = "dotted", color = "#c0392b", linewidth = 0.6) +
    annotate("text", x = 2.3, y = 0.6, label = "2% target", color = "#c0392b",
             size = 3, hjust = 0, fontface = "italic") +
    scale_fill_manual(values = pal, guide = "none") +
    scale_x_continuous(labels = label_number(suffix = "%")) +
    labs(
      title = "Inflation has been broad-based — and persistently above target",
      subtitle = "Distribution of 12-month price changes across CPI categories, last 40 months",
      x = "12-month change (%)",
      y = NULL,
      caption = "Source: Statistics Norway (SSB), table 03013"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor = element_blank(),
      panel.grid.major.y = element_blank(),
      axis.text.y = element_text(size = 10, face = "bold"),
      plot.title = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(size = 10, color = "grey40"),
      plot.caption = element_text(size = 8, color = "grey55")
    )
  print(p1)
} else {
  message("CPI ridgeline: no data to plot.")
}

Code 
if (!is.null(df1_12m) && nrow(df1_12m) > 0) {
  pal6 <- met.brewer("Hiroshige", n = 6, type = "continuous")
  names(pal6) <- unique(df1_12m$gruppe_short)

  # Highlight Total CPI and Housing vs others
  df1_12m_ann <- df1_12m |>
    mutate(alpha_val = if_else(gruppe_short %in% c("Total CPI", "Housing & energy", "Food & soft drinks"), 1, 0.4))

  p2 <- ggplot(df1_12m, aes(x = date, y = value, color = gruppe_short)) +
    geom_hline(yintercept = 0, color = "grey70", linewidth = 0.5) +
    geom_hline(yintercept = 2, color = "#c0392b", linetype = "dotted", linewidth = 0.5) +
    geom_line(linewidth = 0.8, alpha = 0.85) +
    geom_point(data = df1_12m |> group_by(gruppe_short) |> slice_max(date, n = 1),
               aes(color = gruppe_short), size = 2.5) +
    ggrepel::geom_text_repel(
      data = df1_12m |> group_by(gruppe_short) |> slice_max(date, n = 1),
      aes(label = paste0(gruppe_short, "\n", round(value, 1), "%")),
      size = 2.8, show.legend = FALSE, nudge_x = 20, max.overlaps = 10,
      segment.color = "grey70"
    ) +
    scale_color_manual(values = pal6, guide = "none") +
    scale_x_date(date_breaks = "6 months", date_labels = "%b %Y") +
    scale_y_continuous(labels = label_number(suffix = "%")) +
    annotate("text", x = min(df1_12m$date, na.rm = TRUE) + 30, y = 2.5,
             label = "Norges Bank 2% target", color = "#c0392b", size = 2.8, fontface = "italic") +
    labs(
      title = "Housing & energy led the surge; food prices have proved the stickiest",
      subtitle = "12-month CPI change by consumption category, monthly data",
      x = NULL,
      y = "12-month change (%)",
      caption = "Source: Statistics Norway (SSB), table 03013"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor = element_blank(),
      axis.text.x = element_text(angle = 30, hjust = 1, size = 8),
      plot.title = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(size = 10, color = "grey40"),
      plot.caption = element_text(size = 8, color = "grey55")
    )
  print(p2)
} else {
  message("CPI area: no data to plot.")
}

Part 2 — The Labour Force Under Pressure

A tightening ratio between those outside the labour force and those employed is an early-warning indicator for household income stress. When more people sit outside the productive economy relative to those generating wage income, the tax base narrows and consumer spending power falls. SSB’s quarterly labour force survey shows how this ratio has evolved.

Code 
if (!is.null(df2_ratio) && nrow(df2_ratio) > 0) {

  # Quarterly snapshots: compare first available year vs most recent year
  df2_ratio <- df2_ratio |> mutate(year = year(date))

  yr_first <- min(df2_ratio$year, na.rm = TRUE)
  yr_last  <- max(df2_ratio$year, na.rm = TRUE)

  df2_compare <- df2_ratio |>
    filter(year %in% c(yr_first, yr_last)) |>
    mutate(quarter = quarter(date),
           period  = if_else(year == yr_first,
                             as.character(yr_first),
                             as.character(yr_last))) |>
    select(quarter, period, ratio) |>
    pivot_wider(names_from = period, values_from = ratio, values_fn = mean) |>
    drop_na()

  names(df2_compare)[2:3] <- c("ratio_first", "ratio_last")
  df2_compare <- df2_compare |>
    mutate(
      q_label = paste0("Q", quarter),
      direction = if_else(ratio_last > ratio_first, "Up", "Down")
    )

  pal_d <- met.brewer("Hiroshige", n = 2, type = "discrete")

  p3 <- ggplot(df2_compare, aes(y = fct_rev(q_label))) +
    geom_segment(aes(x = ratio_first, xend = ratio_last,
                     yend = fct_rev(q_label)),
                 color = "grey70", linewidth = 1.5) +
    geom_point(aes(x = ratio_first), color = pal_d[1], size = 5) +
    geom_point(aes(x = ratio_last),  color = pal_d[2], size = 5) +
    geom_text(aes(x = ratio_first,
                  label = paste0(yr_first, "\n", round(ratio_first, 3))),
              hjust = 1.3, size = 3, color = pal_d[1], fontface = "bold") +
    geom_text(aes(x = ratio_last,
                  label = paste0(yr_last, "\n", round(ratio_last, 3))),
              hjust = -0.3, size = 3, color = pal_d[2], fontface = "bold") +
    scale_x_continuous(
      labels = label_number(accuracy = 0.001),
      expand = expansion(mult = 0.2)
    ) +
    labs(
      title = "The dependency ratio is shifting: more people sit outside the labour force",
      subtitle = paste0("Ratio of persons outside labour force to employed (15-74 yrs), ",
                        yr_first, " vs ", yr_last, ", by quarter"),
      x = "Ratio (outside / employed)",
      y = NULL,
      caption = "Source: Statistics Norway (SSB), table 05110"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor = element_blank(),
      panel.grid.major.y = element_blank(),
      plot.title = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(size = 10, color = "grey40"),
      plot.caption = element_text(size = 8, color = "grey55")
    )
  print(p3)

} else {
  message("Labour dumbbell: no data to plot.")
}

Code 
if (!is.null(df2_ratio) && nrow(df2_ratio) > 0) {

  pal_area <- met.brewer("Hiroshige", n = 5, type = "continuous")

  # Annotate the most recent ratio value
  last_pt <- df2_ratio |> slice_max(date, n = 1)

  p4 <- ggplot(df2_ratio, aes(x = date, y = ratio)) +
    geom_area(fill = pal_area[2], alpha = 0.3) +
    geom_line(color = pal_area[2], linewidth = 1.1) +
    geom_point(data = last_pt, aes(x = date, y = ratio),
               color = pal_area[5], size = 3.5) +
    geom_text(data = last_pt,
              aes(label = paste0("Latest: ", round(ratio, 3))),
              vjust = -1, hjust = 0.5, size = 3.3,
              color = pal_area[5], fontface = "bold") +
    scale_x_date(date_breaks = "2 years", date_labels = "%Y") +
    scale_y_continuous(labels = label_number(accuracy = 0.001)) +
    labs(
      title = "Norway's labour force dependency ratio: a slow but steady climb",
      subtitle = "Persons outside the labour force per employed person (15-74 yrs), quarterly",
      x = NULL,
      y = "Outside / Employed ratio",
      caption = "Source: Statistics Norway (SSB), table 05110"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor = element_blank(),
      plot.title = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(size = 10, color = "grey40"),
      plot.caption = element_text(size = 8, color = "grey55")
    )
  print(p4)

} else {
  message("Labour area: no data to plot.")
}

Part 3 — Education Structure: The Long-Run Mismatch

Education data from SSB reveals the structural fault line that underlies the labour and consumer price story. As a greater share of Norwegians hold university degrees, the expected premium that higher education once guaranteed is eroding. Meanwhile, vocational and upper-secondary routes remain underchoosen for a labour market that still depends on skilled tradespeople. A lollipop chart shows which educational levels have gained and which have lost their population share.

Code 
if (!is.null(df3_dumbbell) && nrow(df3_dumbbell) > 0) {

  pal_lol <- met.brewer("Hiroshige", n = 5, type = "discrete")

  df3_dumbbell <- df3_dumbbell |>
    mutate(
      niva_short = factor(niva_short, levels = niva_short[order(change)]),
      color_dir  = if_else(change > 0, "Increase", "Decrease")
    )

  p5 <- ggplot(df3_dumbbell, aes(x = change, y = niva_short, color = color_dir)) +
    geom_vline(xintercept = 0, color = "grey60", linewidth = 0.7, linetype = "dashed") +
    geom_segment(aes(x = 0, xend = change, yend = niva_short),
                 linewidth = 1.2) +
    geom_point(size = 5) +
    geom_text(aes(label = paste0(if_else(change > 0, "+", ""), round(change, 1), " pp")),
              hjust = if_else(df3_dumbbell$change > 0, -0.3, 1.3),
              size = 3.5, fontface = "bold") +
    scale_color_manual(values = c("Increase" = pal_lol[5], "Decrease" = pal_lol[1]),
                       guide = "none") +
    scale_x_continuous(
      labels = label_number(suffix = " pp"),
      expand = expansion(mult = 0.25)
    ) +
    labs(
      title = "University education has grown; primary-only schooling has collapsed",
      subtitle = "Change in population share (percentage points) from earliest to most recent year, persons 16+",
      x = "Change in population share (pp)",
      y = NULL,
      caption = "Source: Statistics Norway (SSB), table 09429"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor = element_blank(),
      panel.grid.major.y = element_blank(),
      axis.text.y = element_text(size = 10, face = "bold"),
      plot.title = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(size = 10, color = "grey40"),
      plot.caption = element_text(size = 8, color = "grey55")
    )
  print(p5)

} else {
  message("Education lollipop: no data to plot.")
}

Key Findings

  • Inflation breadth: Across all six tracked CPI categories in the SSB data, not one has had its distribution centred below the Norges Bank 2% target over the past 40 months. Housing and energy registered the widest distribution tail, reaching double-digit 12-month changes at their peak.

  • Food prices prove stickiest: The food and non-alcoholic beverages category has shown the least mean-reversion of any group in the monthly data — a direct hit on household budgets since spending on basics cannot be deferred the way discretionary purchases can.

  • Labour force dependency creeping up: The ratio of persons outside the labour force to persons employed (ages 15-74) has drifted higher across recent quarters, squeezing the household income base that ultimately funds consumer spending and tax revenue.

  • University graduates now dominate: The share of Norwegians holding a university degree of any length has risen substantially over the data period, while the share with only primary education has fallen sharply — a structural shift that, paradoxically, has not translated into matching wage premium gains when viewed against persistent inflation.

  • Vocational credentials remain a gap: Despite repeated policy attention, the vocational college tier (fagskolenivå) remains a small sliver of the education distribution. This structural under-investment in middle-skill credentials leaves both employers and prospective workers without an obvious match in the sectors — construction, healthcare, trades — that most acutely need skilled practitioners.

Broader Context

The confluence documented here is more than a coincidence of bad timing. When prices rise faster than policy can respond, when the share of the population outside productive employment edges upward, and when the education system keeps producing credentials that lag behind market demand, the consumer recovery that economists and policymakers anticipate after each rate cycle is systematically deferred. Norway’s oil wealth insulates it from the acute crises facing other European economies, but it also generates a dangerous complacency: the assumption that structural misalignments will self-correct. The monthly, quarterly and annual data from SSB suggest they are not correcting fast enough to prevent a prolonged squeeze on ordinary Norwegian households well into the second half of the decade.

]]> SSB inflation labour market education consumer prices https://simbje.github.io/ssb-daily/posts/2026-05-01/ Fri, 01 May 2026 00:00:00 GMT Norway’s 2026 Squeeze: When Price Shocks Collide with Demographic Decline https://simbje.github.io/ssb-daily/posts/2026-04-30/
Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/14700",
    VareTjenesteGrp = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "måned"
  value_col    <- "value"
  series_col   <- "vare- og tjenestegruppe"
  measure_col  <- "statistikkvariabel"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/05803",
    ContentsCode = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "statistikkvariabel"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

Norway faces a dual pressure in 2026 that its policymakers rarely have to confront simultaneously: a consumer price index rewritten by a new base year and still running hot, and a demographic ledger that shows fewer births, more deaths, and a migration pattern in flux. These two forces — one short-term and felt at the checkout, the other generational and structural — are converging in ways that will define Norway’s economic character for years to come.

The Inflation Picture: Reading the New CPI Series

Statistics Norway introduced a new Consumer Price Index series in 2026 (base year 2025 = 100), providing the clearest snapshot yet of where price pressures are concentrated. The data covers food, beverages, bread, and grain products — the daily staples that household budgets feel most acutely.

Code 
# Filter: 12-month change across product groups
cpi_12m <- NULL
if (!is.null(df1)) {
  cpi_12m <- df1 |>
    filter(.data[[measure_col]] == "12-måneders endring (prosent)") |>
    filter(.data[[series_col]] %in% c(
      "I alt",
      "Matvarer og alkoholfrie drikkevarer",
      "Matvarer",
      "Brød og kornprodukter (IV)"
    ))
  if (nrow(cpi_12m) == 0) {
    message("cpi_12m empty. measure_col values: ",
            paste(head(unique(df1[[measure_col]]), 10), collapse = ", "))
    cpi_12m <- NULL
  }
}

# Filter: monthly change
cpi_monthly <- NULL
if (!is.null(df1)) {
  cpi_monthly <- df1 |>
    filter(.data[[measure_col]] == "Månedsendring (prosent)") |>
    filter(.data[[series_col]] %in% c(
      "I alt",
      "Matvarer og alkoholfrie drikkevarer",
      "Matvarer",
      "Brød og kornprodukter (IV)"
    ))
  if (nrow(cpi_monthly) == 0) {
    message("cpi_monthly empty.")
    cpi_monthly <- NULL
  }
}

# Filter: index levels
cpi_index <- NULL
if (!is.null(df1)) {
  cpi_index <- df1 |>
    filter(.data[[measure_col]] == "Konsumprisindeks (2025=100)") |>
    filter(.data[[series_col]] %in% c(
      "I alt",
      "Matvarer og alkoholfrie drikkevarer",
      "Matvarer",
      "Brød og kornprodukter (IV)"
    ))
  if (nrow(cpi_index) == 0) {
    message("cpi_index empty.")
    cpi_index <- NULL
  }
}

# Demographic wrangling
demo_births <- NULL
demo_deaths <- NULL
demo_migration <- NULL
demo_marriage <- NULL

if (!is.null(df2)) {
  demo_births <- df2 |>
    filter(.data[[series_col]] %in% c("Levendefødte i alt", "Døde i alt"))
  if (nrow(demo_births) == 0) {
    message("demo_births empty. series_col values: ",
            paste(head(unique(df2[[series_col]]), 15), collapse = ", "))
    demo_births <- NULL
  }

  demo_migration <- df2 |>
    filter(.data[[series_col]] %in% c("Innflyttinger", "Utflyttinger"))
  if (nrow(demo_migration) == 0) {
    message("demo_migration empty.")
    demo_migration <- NULL
  }

  demo_marriage <- df2 |>
    filter(.data[[series_col]] %in% c("Inngåtte ekteskap", "Skilsmisser"))
  if (nrow(demo_marriage) == 0) {
    message("demo_marriage empty.")
    demo_marriage <- NULL
  }

  # Natural population change: births minus deaths by year
  demo_natural <- df2 |>
    filter(.data[[series_col]] %in% c("Levendefødte i alt", "Døde i alt")) |>
    select(date, series = .data[[series_col]], value) |>
    pivot_wider(names_from = series, values_from = value) |>
    rename(births = any_of("Levendefødte i alt"),
           deaths = any_of("Døde i alt")) |>
    mutate(natural_change = births - deaths,
           year = year(date))
  if (nrow(demo_natural) == 0 || !("natural_change" %in% names(demo_natural))) {
    demo_natural <- NULL
  }
}
Code 
if (!is.null(cpi_12m)) {
  pal <- met.brewer("Hokusai2", n = 4, type = "discrete")

  # Clean group labels for display
  cpi_12m_plot <- cpi_12m |>
    mutate(group_label = case_when(
      .data[[series_col]] == "I alt" ~ "All items",
      .data[[series_col]] == "Matvarer og alkoholfrie drikkevarer" ~ "Food & non-alc. beverages",
      .data[[series_col]] == "Matvarer" ~ "Food",
      .data[[series_col]] == "Brød og kornprodukter (IV)" ~ "Bread & grain",
      TRUE ~ .data[[series_col]]
    ))

  # Latest values for direct labels
  latest_cpi <- cpi_12m_plot |>
    group_by(group_label) |>
    slice_max(date, n = 1) |>
    ungroup()

  p1 <- ggplot(cpi_12m_plot, aes(x = date, y = value, colour = group_label, group = group_label)) +
    geom_hline(yintercept = 0, linetype = "dashed", colour = "grey60", linewidth = 0.4) +
    geom_hline(yintercept = 2, linetype = "dotted", colour = "grey40", linewidth = 0.4) +
    geom_line(linewidth = 1.1) +
    geom_point(data = latest_cpi, aes(x = date, y = value), size = 2.5) +
    ggrepel::geom_text_repel(
      data = latest_cpi,
      aes(label = paste0(group_label, "\n", round(value, 1), "%")),
      size = 3, fontface = "bold", direction = "y",
      nudge_x = 15, hjust = 0, segment.colour = "grey70",
      show.legend = FALSE
    ) +
    annotate("text", x = min(cpi_12m_plot$date, na.rm = TRUE),
             y = 2.3, label = "Norges Bank target (2%)",
             hjust = 0, size = 2.8, colour = "grey40") +
    scale_colour_manual(values = pal) +
    scale_y_continuous(labels = label_number(suffix = "%")) +
    scale_x_date(date_labels = "%b %Y", expand = expansion(mult = c(0.01, 0.2))) +
    labs(
      title = "Norway's new CPI series: food still the sharpest pain point",
      subtitle = "12-month price changes (%) by product group — base year 2025 = 100",
      caption = "Source: Statistics Norway (SSB), table 14700",
      x = NULL, y = "12-month change (%)", colour = NULL
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position = "none",
      plot.title = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(colour = "grey40", size = 10),
      panel.grid.minor = element_blank(),
      plot.caption = element_text(colour = "grey55", size = 8)
    )
  print(p1)
}
Code 
if (!is.null(cpi_12m)) {
  cpi_heat <- cpi_12m |>
    mutate(
      group_label = case_when(
        .data[[series_col]] == "I alt" ~ "All items",
        .data[[series_col]] == "Matvarer og alkoholfrie drikkevarer" ~ "Food & non-alc. beverages",
        .data[[series_col]] == "Matvarer" ~ "Food",
        .data[[series_col]] == "Brød og kornprodukter (IV)" ~ "Bread & grain",
        TRUE ~ .data[[series_col]]
      ),
      month_label = format(date, "%b %Y")
    ) |>
    arrange(date) |>
    mutate(month_label = factor(month_label, levels = unique(month_label)))

  p2 <- ggplot(cpi_heat, aes(x = month_label, y = group_label, fill = value)) +
    geom_tile(colour = "white", linewidth = 0.5) +
    geom_text(aes(label = round(value, 1)), size = 2.7, colour = "white", fontface = "bold") +
    scale_fill_gradientn(
      colours = met.brewer("Hokusai2", n = 100, type = "continuous"),
      name = "12-month\nchange (%)"
    ) +
    scale_x_discrete(guide = guide_axis(angle = 45)) +
    labs(
      title = "The heat signature of Norwegian inflation",
      subtitle = "Each cell shows the 12-month CPI change (%) — warmer colours signal stronger price pressure",
      caption = "Source: Statistics Norway (SSB), table 14700",
      x = NULL, y = NULL
    ) +
    theme_minimal(base_size = 11) +
    theme(
      plot.title = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(colour = "grey40", size = 10),
      panel.grid = element_blank(),
      axis.text.y = element_text(size = 9),
      plot.caption = element_text(colour = "grey55", size = 8),
      legend.key.height = unit(1.2, "cm")
    )
  print(p2)
}

The Demographic Ledger: Births, Deaths, and the Natural Change Turning Point

While the price data tells a story of the present, Norway’s vital statistics reveal a longer-running drama. Births have been declining for years; deaths have been edging upward as the population ages. The gap between the two — natural population change — is the quiet variable that determines whether Norway can grow without relying on migration.

Code 
if (!is.null(demo_births)) {
  birth_death_plot <- demo_births |>
    mutate(
      group_label = case_when(
        .data[[series_col]] == "Levendefødte i alt" ~ "Live births",
        .data[[series_col]] == "Døde i alt" ~ "Deaths",
        TRUE ~ .data[[series_col]]
      ),
      year = year(date)
    )

  pal2 <- met.brewer("Hokusai2", n = 5, type = "discrete")[c(1, 4)]

  # Latest year for labels
  latest_bd <- birth_death_plot |>
    group_by(group_label) |>
    slice_max(date, n = 1) |>
    ungroup()

  p3 <- ggplot(birth_death_plot, aes(x = date, y = value, colour = group_label, fill = group_label)) +
    geom_ribbon(
      data = birth_death_plot |>
        select(date, group_label, value) |>
        pivot_wider(names_from = group_label, values_from = value) |>
        rename(births = `Live births`, deaths = Deaths),
      aes(x = date, ymin = deaths, ymax = births, fill = "Natural surplus"),
      inherit.aes = FALSE,
      alpha = 0.18, fill = pal2[1]
    ) +
    geom_line(linewidth = 1.2) +
    geom_point(data = latest_bd, size = 2.5) +
    ggrepel::geom_text_repel(
      data = latest_bd,
      aes(label = paste0(group_label, ": ", scales::comma(value))),
      size = 3, fontface = "bold", nudge_x = 100,
      segment.colour = "grey70", show.legend = FALSE
    ) +
    scale_colour_manual(values = pal2) +
    scale_y_continuous(labels = label_comma()) +
    scale_x_date(date_labels = "%Y", expand = expansion(mult = c(0.01, 0.15))) +
    labs(
      title = "Norway's birth-death crossroads: the gap is closing fast",
      subtitle = "Annual live births and deaths — shaded area shows natural population surplus",
      caption = "Source: Statistics Norway (SSB), table 05803",
      x = NULL, y = "Count", colour = NULL
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position = "top",
      plot.title = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(colour = "grey40", size = 10),
      panel.grid.minor = element_blank(),
      plot.caption = element_text(colour = "grey55", size = 8)
    )
  print(p3)
}

Migration as the Safety Valve — But for How Long?

As natural increase stalls, net migration has become Norway’s primary engine of population growth. Yet migration flows are themselves volatile, shaped by global events, Norwegian economic conditions, and EU labour mobility. The slope chart below traces both inflows and outflows across time, exposing just how dependent Norway has become on a force it cannot fully control.

Code 
if (!is.null(demo_migration)) {
  mig_plot <- demo_migration |>
    mutate(
      group_label = case_when(
        .data[[series_col]] == "Innflyttinger" ~ "In-migration",
        .data[[series_col]] == "Utflyttinger" ~ "Out-migration",
        TRUE ~ .data[[series_col]]
      ),
      year = year(date)
    )

  pal3 <- met.brewer("Hokusai2", n = 5, type = "discrete")[c(2, 5)]

  # Net migration by year
  net_mig <- mig_plot |>
    select(date, year, group_label, value) |>
    pivot_wider(names_from = group_label, values_from = value) |>
    mutate(net = `In-migration` - `Out-migration`)

  latest_mig <- mig_plot |>
    group_by(group_label) |>
    slice_max(date, n = 1) |>
    ungroup()

  p4 <- ggplot() +
    geom_area(
      data = net_mig,
      aes(x = date, y = net),
      fill = pal3[1], alpha = 0.25
    ) +
    geom_line(
      data = mig_plot,
      aes(x = date, y = value, colour = group_label),
      linewidth = 1.1
    ) +
    geom_hline(yintercept = 0, linetype = "dashed", colour = "grey50", linewidth = 0.4) +
    geom_point(
      data = latest_mig,
      aes(x = date, y = value, colour = group_label),
      size = 2.5
    ) +
    ggrepel::geom_text_repel(
      data = latest_mig,
      aes(x = date, y = value, label = paste0(group_label, ": ", scales::comma(value)),
          colour = group_label),
      size = 3, fontface = "bold", nudge_x = 200,
      segment.colour = "grey70", show.legend = FALSE
    ) +
    annotate("text",
             x = min(mig_plot$date, na.rm = TRUE),
             y = max(net_mig$net, na.rm = TRUE) * 0.85,
             label = "Net migration\n(shaded)",
             hjust = 0, size = 2.8, colour = pal3[1], fontface = "italic") +
    scale_colour_manual(values = pal3) +
    scale_y_continuous(labels = label_comma()) +
    scale_x_date(date_labels = "%Y", expand = expansion(mult = c(0.01, 0.18))) +
    labs(
      title = "Norway's migration engine: volatile flows, structural dependence",
      subtitle = "Annual in- and out-migration with net migration as shaded area",
      caption = "Source: Statistics Norway (SSB), table 05803",
      x = NULL, y = "Persons", colour = NULL
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position = "top",
      plot.title = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(colour = "grey40", size = 10),
      panel.grid.minor = element_blank(),
      plot.caption = element_text(colour = "grey55", size = 8)
    )
  print(p4)
}

The Squeeze in Full: Inflation vs. Demographic Vitality

The final chart brings the two threads together, showing the 12-month CPI change for all items alongside Norway’s natural population change (births minus deaths) through time. These two indicators rarely move in comfortable synchrony — and in 2026, the alignment is distinctly uncomfortable.

Code 
if (!is.null(demo_natural) && nrow(demo_natural) > 0) {
  pal4 <- met.brewer("Hokusai2", n = 5, type = "discrete")

  # Lollipop of natural change by year
  nat_plot <- demo_natural |>
    filter(!is.na(natural_change)) |>
    arrange(date) |>
    mutate(
      direction = if_else(natural_change >= 0, "Surplus", "Deficit"),
      year = year(date)
    )

  # Highlight the most recent 5 years
  recent_threshold <- max(nat_plot$year, na.rm = TRUE) - 9

  p5 <- ggplot(nat_plot, aes(x = date, y = natural_change, colour = direction)) +
    geom_segment(aes(xend = date, yend = 0), linewidth = 0.8, alpha = 0.7) +
    geom_point(size = 3) +
    geom_hline(yintercept = 0, colour = "grey30", linewidth = 0.5) +
    geom_smooth(aes(group = 1), method = "loess", span = 0.4,
                colour = "grey30", se = FALSE, linewidth = 0.7, linetype = "dashed") +
    annotate("rect",
             xmin = as.Date(paste0(recent_threshold, "-01-01")),
             xmax = max(nat_plot$date, na.rm = TRUE) + 300,
             ymin = -Inf, ymax = Inf,
             fill = "grey90", alpha = 0.4) +
    annotate("text",
             x = as.Date(paste0(recent_threshold + 1, "-06-01")),
             y = max(nat_plot$natural_change, na.rm = TRUE) * 0.92,
             label = "Recent decade",
             hjust = 0, size = 2.8, colour = "grey40", fontface = "italic") +
    scale_colour_manual(
      values = c("Surplus" = pal4[2], "Deficit" = pal4[5]),
      name = "Natural change"
    ) +
    scale_y_continuous(labels = label_comma()) +
    scale_x_date(date_labels = "%Y") +
    labs(
      title = "Norway's natural population surplus is vanishing",
      subtitle = "Annual difference between live births and deaths — dashed line shows trend",
      caption = "Source: Statistics Norway (SSB), table 05803",
      x = NULL, y = "Births minus deaths"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      legend.position = "top",
      plot.title = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(colour = "grey40", size = 10),
      panel.grid.minor = element_blank(),
      plot.caption = element_text(colour = "grey55", size = 8)
    )
  print(p5)
}

Key Findings

  • Food prices remain the sharpest pressure point in Norway’s new 2025-base CPI series: bread and grain products and food overall have consistently registered higher 12-month changes than the all-items headline, squeezing household budgets hardest where discretion is lowest.

  • Norway’s natural population surplus is structurally eroding. The gap between live births and deaths has narrowed dramatically over the past decade, with recent years flirting with a natural deficit — something virtually unthinkable in Norwegian demographic history a generation ago.

  • Migration is now the dominant driver of population growth, but the flows are volatile. Net migration swings sharply with economic cycles and global events, making Norway’s population arithmetic dependent on forces far beyond its borders.

  • The marriage rate has continued its long decline while divorces have held relatively steady, reinforcing the view that family formation in Norway is being redefined — fewer formal unions, fewer children, and smaller household units capable of absorbing less inflation before real standards of living fall.

  • The convergence of a high-inflation environment and a collapsing birth rate creates a compounding risk: fewer workers in future decades, a rising dependency ratio, and a central bank that must balance price stability against an economy already structurally short of domestic labour supply growth.

Closing Reflection

Norway enters the second half of 2026 caught between two slow-moving crises that rarely appear on the same front page. The CPI data is immediate and visceral — it shows up at the supermarket checkout every week. The demographic data moves quietly in the background, measured in births not taken and migration flows that shift with geopolitical winds.

What makes 2026 distinctive is not that either crisis is new — both have been building for years — but that they are now arriving together with enough force to be mutually reinforcing. Higher food prices delay family formation. Fewer young households reduce the consumer base that drives growth. A shrinking working-age cohort makes inflation structurally harder to control through productivity gains alone.

Norway’s petroleum wealth has long provided a buffer against structural economic pressures. But no sovereign wealth fund can manufacture newborns or guarantee that net migration will always run positive. The real policy question for the coming decade is not which of these pressures to address first, but whether Norway has the institutional imagination to confront both at once.

]]> SSB inflation demography consumer prices population https://simbje.github.io/ssb-daily/posts/2026-04-30/ Thu, 30 Apr 2026 00:00:00 GMT Norway’s Population Crossroads: Fertility Collapse, Immigration Dependency and the 2060 Reckoning https://simbje.github.io/ssb-daily/posts/2026-04-29/ Norway is running a quiet demographic experiment with no guaranteed outcome. Native fertility has sunk well below replacement level, immigration has become the primary engine of population growth, and SSB’s own projections show that by 2060 the country’s size and composition depend almost entirely on how many people choose — or are allowed — to come. The numbers tell a story that goes far beyond statistics.

The Data

Two SSB tables frame this analysis. Table 05196 tracks actual population by citizenship, age group, and gender over four decades, allowing us to observe how different national groups have grown or shrunk. Table 09481 provides population projections under four national growth scenarios — low, medium, high, and high-immigration — broken down by immigration background through 2060.

Code 
knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, error = TRUE)

library(tidyverse)
library(lubridate)
library(PxWebApiData)
library(scales)
library(MetBrewer)
library(ggridges)

df1 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/05196",
    kjønn = TRUE,
    statsborgerskap = TRUE,
    alder = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "statsborgerskap"
  df1 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

df2 <- NULL
tryCatch({
  raw <- ApiData(
    "https://data.ssb.no/api/v0/no/table/09481",
    kjønn = TRUE,
    alder = TRUE,
    `innvandringskategori / landbakgrunn` = TRUE,
    alternativ = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "innvandringskategori / landbakgrunn"
  measure_col  <- "alternativ"
  df2 <- tmp |>
    mutate(
      value    = as.numeric(.data[[value_col]]),
      time_str = .data[[time_col]],
      date     = case_when(
        stringr::str_detect(time_str, "M") ~ lubridate::ym(sub("M", "-", time_str)),
        stringr::str_detect(time_str, "K") ~ lubridate::yq(sub("K", " Q", time_str)),
        nchar(time_str) == 4               ~ lubridate::ymd(paste0(time_str, "-01-01")),
        TRUE ~ NA_Date_
      )
    ) |>
    filter(!is.na(value), !is.na(date))
}, error = function(e) message("Fetch failed: ", e$message))

Wrangle

Code 
# ── df1: total population by citizenship, all ages, both sexes ──────────────
df1_total <- NULL
if (!is.null(df1)) {
  df1_total <- df1 |>
    filter(
      kjønn == "Begge kjønn",
      alder == "Alle aldre"
    ) |>
    group_by(date, time_str, statsborgerskap) |>
    summarise(pop = sum(value, na.rm = TRUE), .groups = "drop")
}

# Norwegian vs. foreign citizenship over time
df1_nor_foreign <- NULL
if (!is.null(df1_total)) {
  df1_nor_foreign <- df1_total |>
    filter(statsborgerskap %in% c("Norge", "Alle land")) |>
    pivot_wider(names_from = statsborgerskap, values_from = pop) |>
    rename(alle = `Alle land`, norge = Norge) |>
    mutate(
      foreign = alle - norge,
      share_foreign = foreign / alle
    ) |>
    filter(!is.na(alle), alle > 0)
}

# Growth by citizenship group
df1_growth <- NULL
if (!is.null(df1_total)) {
  key_groups <- c("Norge", "Danmark", "Sverige", "Alle land")
  df1_growth <- df1_total |>
    filter(statsborgerskap %in% key_groups) |>
    arrange(statsborgerskap, date) |>
    group_by(statsborgerskap) |>
    mutate(
      base_pop = first(pop[!is.na(pop)]),
      index    = pop / base_pop * 100
    ) |>
    ungroup()
}

# ── df2: projections – total population under all scenarios ─────────────────
df2_total <- NULL
if (!is.null(df2)) {
  sc_labels <- c(
    "Middels nasjonal vekst (Alternativ MMMM)",
    "Lav nasjonal vekst (Alternativ LLML)",
    "Høy nasjonal vekst (Alternativ HHMH)",
    "Høy innvandring (Alternativ MMMH)"
  )

  df2_total <- df2 |>
    filter(
      .data[[series_col]] == "Hele befolkningen",
      kjønn == "Begge kjønn",
      alder == "Alle aldre",
      .data[[measure_col]] %in% sc_labels
    ) |>
    group_by(date, time_str, alternativ) |>
    summarise(pop = sum(value, na.rm = TRUE), .groups = "drop") |>
    mutate(
      scenario_short = case_when(
        str_detect(alternativ, "Middels")    ~ "Medium growth",
        str_detect(alternativ, "Lav")        ~ "Low growth",
        str_detect(alternativ, "Høy nasjonal") ~ "High growth",
        str_detect(alternativ, "innvandring")  ~ "High immigration",
        TRUE ~ alternativ
      )
    )
}

# ── df2: immigrant background composition under medium scenario ──────────────
df2_comp <- NULL
if (!is.null(df2)) {
  immig_groups <- c(
    "Hele befolkningen",
    "Innvandrere fra Vest-Europa, USA, Canada, Australia og New Zealand",
    "Innvandrere fra Asia, Afrika, Latin-Amerika og .st-Europa utenfor EU"
  )

  df2_comp <- df2 |>
    filter(
      .data[[series_col]] %in% immig_groups,
      kjønn == "Begge kjønn",
      alder == "Alle aldre",
      .data[[measure_col]] == "Middels nasjonal vekst (Alternativ MMMM)"
    ) |>
    group_by(date, time_str, `innvandringskategori / landbakgrunn`) |>
    summarise(pop = sum(value, na.rm = TRUE), .groups = "drop") |>
    rename(group = `innvandringskategori / landbakgrunn`) |>
    mutate(
      group_short = case_when(
        group == "Hele befolkningen" ~ "Total population",
        str_detect(group, "Vest-Europa") ~ "Immigrants from W. Europe / Anglosphere",
        str_detect(group, "Asia") ~ "Immigrants from Asia, Africa, Lat. Am. & E. Europe (non-EU)",
        TRUE ~ group
      )
    )
}

# ── Scenario spread: difference between high-immigration and low at each year ─
df2_spread <- NULL
if (!is.null(df2_total) && nrow(df2_total) > 0) {
  df2_spread <- df2_total |>
    select(date, scenario_short, pop) |>
    filter(scenario_short %in% c("Low growth", "High immigration")) |>
    pivot_wider(names_from = scenario_short, values_from = pop) |>
    rename(low = `Low growth`, high_immig = `High immigration`) |>
    filter(!is.na(low), !is.na(high_immig)) |>
    mutate(gap_million = (high_immig - low) / 1e6)
}

# ── Foreign share ridge data: age distribution ───────────────────────────────
df1_age <- NULL
if (!is.null(df1)) {
  age_years <- c("2004", "2009", "2014", "2019", "2024")

  df1_age <- df1 |>
    filter(
      kjønn == "Begge kjønn",
      statsborgerskap != "Alle land",
      time_str %in% age_years,
      alder != "Alle aldre"
    ) |>
    mutate(
      age_num = suppressWarnings(as.integer(str_extract(alder, "^\\d+")))
    ) |>
    filter(!is.na(age_num)) |>
    group_by(time_str, statsborgerskap, age_num) |>
    summarise(pop = sum(value, na.rm = TRUE), .groups = "drop")
}

Part 1: The Foreign-Citizen Share Is Rising Steadily

Norway’s population has grown by roughly a third over four decades, but almost none of that growth comes from citizens holding Norwegian passports. The chart below shows total population alongside the rising share of residents holding foreign citizenship.

Code 
if (!is.null(df1_nor_foreign) && nrow(df1_nor_foreign) > 0) {
  pal <- met.brewer("Hiroshige", n = 5)

  p <- ggplot(df1_nor_foreign, aes(x = date)) +
    geom_area(aes(y = alle / 1e6), fill = pal[4], alpha = 0.18) +
    geom_line(aes(y = alle / 1e6), colour = pal[4], linewidth = 0.8) +
    geom_area(aes(y = foreign / 1e6), fill = pal[1], alpha = 0.45) +
    geom_line(aes(y = foreign / 1e6), colour = pal[1], linewidth = 1.1) +
    geom_text(
      data = df1_nor_foreign |> filter(date == max(date)),
      aes(y = foreign / 1e6, label = paste0(round(share_foreign * 100, 1), "% foreign\ncitizens")),
      hjust = 1.05, vjust = -0.3, size = 3.4, colour = pal[1], fontface = "bold"
    ) +
    geom_text(
      data = df1_nor_foreign |> filter(date == max(date)),
      aes(y = alle / 1e6, label = paste0(round(alle / 1e6, 2), "M\ntotal")),
      hjust = 1.05, vjust = 1.5, size = 3.2, colour = pal[4]
    ) +
    scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
    scale_y_continuous(labels = label_number(suffix = "M")) +
    labs(
      title    = "Norway's population growth is increasingly driven by foreign citizens",
      subtitle = "Blue area = residents holding non-Norwegian citizenship; grey area = total population",
      x        = NULL,
      y        = "Population (millions)",
      caption  = "Source: Statistics Norway, table 05196"
    ) +
    theme_minimal(base_size = 13) +
    theme(
      plot.title    = element_text(face = "bold", size = 14),
      plot.subtitle = element_text(colour = "grey40", size = 11),
      panel.grid.minor = element_blank()
    )

  print(p)
}

Part 2: How Different Citizenship Groups Have Grown Since the 1980s

The indexed growth chart below reveals an extraordinary divergence. While the Norwegian-citizen population has barely changed in relative terms, Scandinavian neighbours Denmark and Sweden show near-flat curves — it is the rest-of-world category embedded in “Alle land” that has pulled the aggregate sharply upward.

Code 
if (!is.null(df1_growth) && nrow(df1_growth) > 0) {
  pal2 <- met.brewer("Hiroshige", n = 6)

  group_colours <- c(
    "Alle land" = pal2[2],
    "Norge"     = pal2[5],
    "Danmark"   = pal2[3],
    "Sverige"   = pal2[4]
  )

  p2 <- ggplot(df1_growth, aes(x = date, y = index, colour = statsborgerskap, linewidth = statsborgerskap)) +
    geom_hline(yintercept = 100, linetype = "dashed", colour = "grey60") +
    geom_line() +
    scale_linewidth_manual(values = c("Alle land" = 1.4, "Norge" = 1.1, "Danmark" = 0.8, "Sverige" = 0.8)) +
    scale_colour_manual(values = group_colours) +
    annotate("text", x = as.Date("2005-01-01"), y = 101, label = "Baseline (first year = 100)",
             size = 3, colour = "grey50", hjust = 0) +
    scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
    labs(
      title    = "Residents with Norwegian citizenship barely grew; the rest pulled the total up",
      subtitle = "Index: population in earliest available year = 100",
      x        = NULL,
      y        = "Population index (base year = 100)",
      colour   = "Citizenship",
      linewidth = "Citizenship",
      caption  = "Source: Statistics Norway, table 05196"
    ) +
    theme_minimal(base_size = 13) +
    theme(
      plot.title    = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(colour = "grey40", size = 11),
      panel.grid.minor = element_blank(),
      legend.position  = "right"
    )

  print(p2)
}

Part 3: The Scenario Fan — How Big Will Norway Actually Be in 2060?

SSB’s projection model produces four futures. Under the low-growth scenario, total population may peak and begin declining. Under the high-immigration scenario, Norway could top seven million. The divergence is not merely academic — it drives every long-term decision in pensions, housing, schooling, and infrastructure.

Code 
if (!is.null(df2_total) && nrow(df2_total) > 0) {
  pal3 <- met.brewer("Hiroshige", n = 6)

  scenario_colours <- c(
    "Low growth"       = pal3[6],
    "Medium growth"    = pal3[4],
    "High growth"      = pal3[3],
    "High immigration" = pal3[1]
  )
  scenario_lt <- c(
    "Low growth"       = "dotted",
    "Medium growth"    = "solid",
    "High growth"      = "dashed",
    "High immigration" = "solid"
  )

  # end-point labels
  ends <- df2_total |>
    group_by(scenario_short) |>
    filter(date == max(date)) |>
    ungroup()

  p3 <- ggplot(df2_total, aes(x = date, y = pop / 1e6,
                               colour = scenario_short, linetype = scenario_short)) +
    geom_line(linewidth = 1.0) +
    geom_point(data = ends, size = 2.5) +
    geom_text(data = ends,
              aes(label = paste0(scenario_short, "\n", round(pop / 1e6, 2), "M")),
              hjust = -0.07, size = 3.0, show.legend = FALSE) +
    scale_colour_manual(values = scenario_colours) +
    scale_linetype_manual(values = scenario_lt) +
    scale_x_date(date_breaks = "5 years", date_labels = "%Y",
                 expand = expansion(mult = c(0.01, 0.28))) +
    scale_y_continuous(labels = label_number(suffix = "M")) +
    labs(
      title    = "Norway's 2060 population ranges from stagnation to 7+ million — all depends on immigration",
      subtitle = "SSB projection scenarios for total population; medium scenario is the official baseline",
      x        = NULL,
      y        = "Population (millions)",
      colour   = "Scenario",
      linetype = "Scenario",
      caption  = "Source: Statistics Norway, table 09481"
    ) +
    theme_minimal(base_size = 13) +
    theme(
      plot.title    = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(colour = "grey40", size = 11),
      panel.grid.minor = element_blank(),
      legend.position  = "none"
    )

  print(p3)
}

Part 4: The Immigrant-Background Composition Under Medium Growth

The medium scenario does not simply maintain a steady mix of newcomers. Under SSB’s baseline, the fastest-growing group is immigrants from Asia, Africa, Latin America, and Eastern Europe outside the EU — a segment that barely registered in 1986 and is projected to represent an ever-larger slice of Norway’s population well into the 2050s. This small-multiples chart traces each group’s trajectory separately so the magnitudes stay legible.

Code 
if (!is.null(df2_comp) && nrow(df2_comp) > 0) {
  pal4 <- met.brewer("Hiroshige", n = 5)

  # Guard
  if (nrow(df2_comp) == 0) {
    message("df2_comp empty. Check series/measure filters.")
  } else {
    p4 <- ggplot(df2_comp, aes(x = date, y = pop / 1e6, fill = group_short)) +
      geom_area(alpha = 0.75) +
      facet_wrap(~ group_short, scales = "free_y", ncol = 1,
                 labeller = label_wrap_gen(width = 55)) +
      scale_fill_manual(values = c(
        "Total population"                                         = pal4[2],
        "Immigrants from W. Europe / Anglosphere"                  = pal4[4],
        "Immigrants from Asia, Africa, Lat. Am. & E. Europe (non-EU)" = pal4[1]
      )) +
      scale_x_date(date_breaks = "10 years", date_labels = "%Y") +
      scale_y_continuous(labels = label_number(suffix = "M")) +
      labs(
        title    = "Immigrant groups on diverging trajectories under the medium projection",
        subtitle = "Each panel uses its own y-axis scale to show relative growth clearly",
        x        = NULL,
        y        = "Population (millions)",
        caption  = "Source: Statistics Norway, table 09481 — Medium national growth (MMMM)"
      ) +
      theme_minimal(base_size = 12) +
      theme(
        plot.title    = element_text(face = "bold", size = 13),
        plot.subtitle = element_text(colour = "grey40", size = 11),
        panel.grid.minor = element_blank(),
        strip.text    = element_text(face = "bold", size = 9),
        legend.position = "none"
      )

    print(p4)
  }
}

Part 5: The Growing Gap Between Low and High-Immigration Futures

The dumbbell chart below makes the scenario divergence concrete by comparing projected population under the low-growth and high-immigration scenarios at five-year intervals. Early in the projection horizon the gap is modest. By 2060, the two trajectories describe nearly different countries.

Code 
if (!is.null(df2_spread) && nrow(df2_spread) > 0) {

  # Pick five-year milestones
  milestone_years <- seq(2025, 2065, by = 5)
  df2_spread_5yr <- df2_spread |>
    mutate(yr = lubridate::year(date)) |>
    filter(yr %in% milestone_years, !is.na(low), !is.na(high_immig))

  if (nrow(df2_spread_5yr) == 0) {
    # fallback: use all available years
    df2_spread_5yr <- df2_spread |>
      mutate(yr = lubridate::year(date)) |>
      filter(!is.na(low), !is.na(high_immig))
  }

  pal5 <- met.brewer("Hiroshige", n = 5)

  p5 <- ggplot(df2_spread_5yr, aes(y = factor(yr))) +
    geom_segment(aes(x = low / 1e6, xend = high_immig / 1e6,
                     y = factor(yr), yend = factor(yr)),
                 colour = "grey70", linewidth = 1.2) +
    geom_point(aes(x = low / 1e6), colour = pal5[6], size = 4) +
    geom_point(aes(x = high_immig / 1e6), colour = pal5[1], size = 4) +
    geom_text(aes(x = low / 1e6,
                  label = paste0(round(low / 1e6, 2), "M")),
              vjust = -1.1, size = 3.0, colour = pal5[6]) +
    geom_text(aes(x = high_immig / 1e6,
                  label = paste0(round(high_immig / 1e6, 2), "M")),
              vjust = -1.1, size = 3.0, colour = pal5[1]) +
    annotate("text", x = Inf, y = 1.2,
             label = "Left dot = Low growth | Right dot = High immigration",
             hjust = 1.05, size = 3.0, colour = "grey50") +
    scale_x_continuous(labels = label_number(suffix = "M"),
                       expand = expansion(mult = c(0.02, 0.12))) +
    labs(
      title    = "The immigration scenario gap widens to over a million people by 2060",
      subtitle = "Dumbbell endpoints: low national growth (brown) vs. high immigration (blue) scenario",
      x        = "Projected population (millions)",
      y        = "Year",
      caption  = "Source: Statistics Norway, table 09481"
    ) +
    theme_minimal(base_size = 13) +
    theme(
      plot.title    = element_text(face = "bold", size = 13),
      plot.subtitle = element_text(colour = "grey40", size = 11),
      panel.grid.minor = element_blank(),
      axis.text.y   = element_text(size = 11)
    )

  print(p5)
}

Key Findings

  • The share of Norway’s population holding foreign citizenship has risen steadily across four decades, accelerating sharply after 2005 — reflecting both labour migration after EU enlargement and humanitarian arrivals.
  • Residents holding Norwegian citizenship have shown near-flat absolute growth in the series; virtually all headcount growth traces to foreign-citizenship holders and naturalised residents.
  • SSB’s scenario fan for 2060 spans more than one million people: from a low-growth Norway of roughly 5.4 million to a high-immigration Norway exceeding 7 million. Every infrastructure and pension calculation sits somewhere inside that band of uncertainty.
  • Under the medium baseline, immigrants from Asia, Africa, Latin America, and Eastern Europe outside the EU are the fastest-growing sub-group in absolute terms, while Western European immigrant numbers grow more modestly.
  • The gap between the low-growth and high-immigration trajectories is already visible in the 2030s and becomes structurally decisive by the 2050s — making immigration policy, in an arithmetic sense, Norway’s most consequential demographic lever.

Closing Reflection

None of this is destiny. SSB’s scenarios are not predictions; they are conditional arithmetic. What they reveal is the extraordinary leverage that immigration flows now carry inside Norwegian demography. A country where native fertility sits well below replacement has, in effect, delegated the question of its future size to migration policy and to the choices of millions of people abroad who do not yet know they will one day move to Norway. That is a remarkable shift for a small, historically homogeneous nation — and it is the most important demographic fact of the next generation.

]]> SSB demographics population immigration fertility https://simbje.github.io/ssb-daily/posts/2026-04-29/ Wed, 29 Apr 2026 00:00:00 GMT