Norway’s 2026 Youth Crisis: Where Crime Surges and Employment Opportunities Vanish

Monthly labour force data and two decades of crime statistics reveal a deepening structural crisis for young Norwegians caught between a tightening job market and rising property offences.

Norway is caught in a paradox that its headline employment figures tend to obscure: while the aggregate unemployment rate looks reassuringly low by international standards, the combination of stagnant labour-force participation and a long-run surge in property crime paints a more unsettling picture of the pressures facing young people in 2026. Two SSB datasets — monthly labour-force surveys and annual crime statistics — tell the story from complementary angles.

Data

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/13760",
    kjønn = 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/08484",
    LovbruddKrim = TRUE,
    Tid = list(filter = "top", values = 40)
  )
  tmp          <- raw[[1]]
  time_col     <- "år"
  value_col    <- "value"
  series_col   <- "lovbruddstype"
  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

# ── Labour force variables ───────────────────────────────────────────────────
df1_unemp   <- NULL
df1_emprate <- NULL
df1_lf      <- NULL
df1_monthly <- NULL
df1_emprate_agg <- NULL

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

  # Identify available columns
  col_names <- names(df1)

  # Unemployment rate series
  df1_unemp <- df1 |>
    filter(.data[["statistikkvariabel"]] == "Arbeidsledige i prosent av arbeidsstyrken")
  if (nrow(df1_unemp) == 0) {
    message("df1_unemp empty. Values: ",
            paste(head(unique(df1[["statistikkvariabel"]]), 15), collapse = ", "))
    df1_unemp <- NULL
  }

  # Employment-rate series
  df1_emprate <- df1 |>
    filter(.data[["statistikkvariabel"]] == "Sysselsatte i prosent av befolkningen")
  if (nrow(df1_emprate) == 0) {
    message("df1_emprate empty.")
    df1_emprate <- NULL
  }

  # Labour force (thousands)
  df1_lf <- df1 |>
    filter(.data[["statistikkvariabel"]] == "Arbeidsstyrken (1000 personer)")
  if (nrow(df1_lf) == 0) {
    message("df1_lf empty.")
    df1_lf <- NULL
  }

  # Check for monthly time stamps
  has_monthly <- any(stringr::str_detect(df1$time_str, "M\\d{2}"), na.rm = TRUE)

  # Monthly unemployment rate — aggregate (all genders/ages) if possible
  if (!is.null(df1_unemp)) {
    gender_col <- if ("kjønn" %in% col_names) "kjønn" else NULL
    age_col    <- if ("alder" %in% col_names) "alder" else NULL
    adj_col    <- if ("type justering" %in% col_names) "type justering" else NULL

    df1_monthly <- df1_unemp
    # Keep seasonally adjusted if available
    if (!is.null(adj_col)) {
      adj_vals <- unique(df1_monthly[[adj_col]])
      if (any(grepl("sesong", adj_vals, ignore.case = TRUE))) {
        df1_monthly <- df1_monthly |>
          filter(grepl("sesong", .data[[adj_col]], ignore.case = TRUE))
      }
    }
    # Aggregate across gender/age: use "Begge kjønn" / "I alt" where present
    if (!is.null(gender_col)) {
      gvals <- unique(df1_monthly[[gender_col]])
      if (any(grepl("begge|i alt|total|alle", gvals, ignore.case = TRUE))) {
        df1_monthly <- df1_monthly |>
          filter(grepl("begge|i alt|total|alle", .data[[gender_col]], ignore.case = TRUE))
      }
    }
    if (!is.null(age_col)) {
      avals <- unique(df1_monthly[[age_col]])
      if (any(grepl("^i alt$|alle aldre|total", avals, ignore.case = TRUE))) {
        df1_monthly <- df1_monthly |>
          filter(grepl("^i alt$|alle aldre|total", .data[[age_col]], ignore.case = TRUE))
      }
    }
    if (nrow(df1_monthly) == 0) df1_monthly <- df1_unemp
  }

  # Employment rate — aggregate
  if (!is.null(df1_emprate)) {
    gender_col <- if ("kjønn" %in% col_names) "kjønn" else NULL
    age_col    <- if ("alder" %in% col_names) "alder" else NULL
    adj_col    <- if ("type justering" %in% col_names) "type justering" else NULL

    df1_emprate_agg <- df1_emprate
    if (!is.null(adj_col)) {
      adj_vals <- unique(df1_emprate_agg[[adj_col]])
      if (any(grepl("sesong", adj_vals, ignore.case = TRUE))) {
        df1_emprate_agg <- df1_emprate_agg |>
          filter(grepl("sesong", .data[[adj_col]], ignore.case = TRUE))
      }
    }
    if (!is.null(gender_col)) {
      gvals <- unique(df1_emprate_agg[[gender_col]])
      if (any(grepl("begge|i alt|total|alle", gvals, ignore.case = TRUE))) {
        df1_emprate_agg <- df1_emprate_agg |>
          filter(grepl("begge|i alt|total|alle", .data[[gender_col]], ignore.case = TRUE))
      }
    }
    if (!is.null(age_col)) {
      avals <- unique(df1_emprate_agg[[age_col]])
      if (any(grepl("^i alt$|alle aldre|total", avals, ignore.case = TRUE))) {
        df1_emprate_agg <- df1_emprate_agg |>
          filter(grepl("^i alt$|alle aldre|total", .data[[age_col]], ignore.case = TRUE))
      }
    }
    if (nrow(df1_emprate_agg) == 0) df1_emprate_agg <- df1_emprate
  }
}

# ── Crime variables ──────────────────────────────────────────────────────────
df2_all      <- NULL
df2_rate     <- NULL
df2_types    <- NULL
df2_slope    <- NULL

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

  # Total reported crimes — absolute count
  df2_all <- df2 |>
    filter(.data[["lovbruddstype"]] == "Alle lovbruddsgrupper",
           .data[["statistikkvariabel"]] == "Lovbrudd anmeldt")
  if (nrow(df2_all) == 0) {
    message("df2_all empty. lovbruddstype: ",
            paste(head(unique(df2[["lovbruddstype"]]), 15), collapse = ", "))
    df2_all <- NULL
  }

  # Rate per 1000 inhabitants — all crime types except total
  df2_rate <- df2 |>
    filter(.data[["lovbruddstype"]] != "Alle lovbruddsgrupper",
           .data[["statistikkvariabel"]] == "Lovbrudd anmeldt per 1000 innbyggere")
  if (nrow(df2_rate) == 0) {
    message("df2_rate empty.")
    df2_rate <- NULL
  }

  # Absolute counts by type (excluding total)
  df2_types <- df2 |>
    filter(.data[["lovbruddstype"]] != "Alle lovbruddsgrupper",
           .data[["statistikkvariabel"]] == "Lovbrudd anmeldt")
  if (nrow(df2_types) == 0) {
    message("df2_types empty.")
    df2_types <- NULL
  }

  # Slope data: first and last available year per crime type (rate)
  if (!is.null(df2_rate)) {
    years_avail <- sort(unique(df2_rate$date))
    if (length(years_avail) >= 2) {
      y_first <- min(years_avail)
      y_last  <- max(years_avail)
      df2_slope <- df2_rate |>
        filter(date %in% c(y_first, y_last)) |>
        mutate(year_label = as.character(year(date))) |>
        select(lovbruddstype, year_label, value) |>
        pivot_wider(names_from = year_label, values_from = value) |>
        drop_na()
      # Rename dynamically — fixed: extract years safely with vector indexing
      ynames <- sort(unique(as.character(year(df2_rate |>
                              filter(date %in% c(y_first, y_last)) |>
                              pull(date)))))
      if (length(ynames) == 2) {
        names(df2_slope)[2:3] <- paste0("y", ynames)
        df2_slope <- df2_slope |>
          mutate(change = .data[[paste0("y", ynames[2])]] - .data[[paste0("y", ynames[1])]])
      }
    }
  }
}

Section 1: The Labour Market Backdrop

Norway’s headline unemployment rate remains below 4 percent for most of the recent period, but the monthly trajectory reveals important nuance. The seasonally adjusted series tracks the underlying pressure on workers with greater fidelity than raw counts.

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

if (exists("df1_monthly") && !is.null(df1_monthly) && nrow(df1_monthly) > 0) {

  plot_data <- df1_monthly |>
    group_by(date) |>
    summarise(value = mean(value, na.rm = TRUE), .groups = "drop") |>
    arrange(date)

  avg_val <- mean(plot_data$value, na.rm = TRUE)

  p <- ggplot(plot_data, aes(x = date, y = value)) +
    geom_area(fill = pal_met[2], alpha = 0.35) +
    geom_line(colour = pal_met[1], linewidth = 0.9) +
    geom_hline(yintercept = avg_val, linetype = "dashed",
               colour = pal_met[5], linewidth = 0.6) +
    annotate("text", x = min(plot_data$date) + months(2),
             y = avg_val + 0.12, label = paste0("Period avg: ", round(avg_val, 1), "%"),
             hjust = 0, size = 3.2, colour = pal_met[5]) +
    scale_x_date(date_breaks = "1 year", date_labels = "%Y") +
    scale_y_continuous(labels = label_number(suffix = "%")) +
    labs(
      title = "Unemployment rate, Norway",
      subtitle = "Seasonally adjusted series; shaded area shows scale of slack in the labour market",
      x = NULL, y = "Unemployment rate (%)",
      caption = "Source: Statistics Norway, table 13760"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid.minor = element_blank(),
      axis.text.x = element_text(angle = 45, hjust = 1),
      plot.title = element_text(face = "bold"),
      plot.subtitle = element_text(colour = "grey40"),
      plot.caption = element_text(colour = "grey55", size = 9)
    )
  print(p) # fixed: missing print() call
} else {
  message("df1_monthly not available for plot")
}

Section 2: Employment Rate — Gender and Age Dimensions

The employment-rate series — the share of the working-age population actually in work — captures how effectively Norway converts its population into productive labour. Small multiples across adjustment type reveal structural differences.

if (exists("df1_emprate") && !is.null(df1_emprate) && nrow(df1_emprate) > 0) {

  col_names2 <- names(df1_emprate)
  facet_col  <- if ("kjønn" %in% col_names2) "kjønn" else NULL
  adj_col2   <- if ("type justering" %in% col_names2) "type justering" else NULL

  plot_emp <- df1_emprate

  # Keep only a tractable set of age groups / genders for small multiples
  if (!is.null(facet_col)) {
    gender_vals <- unique(plot_emp[[facet_col]])
    # Prefer non-total genders for the faceted view
    non_total <- gender_vals[!grepl("begge|i alt|total|alle", gender_vals, ignore.case = TRUE)]
    if (length(non_total) >= 2) {
      plot_emp <- plot_emp |> filter(.data[[facet_col]] %in% non_total)
    }
  }

  # Aggregate across age and adjustment type if needed
  if (!is.null(adj_col2)) {
    adj_vals2 <- unique(plot_emp[[adj_col2]])
    # Prefer unadjusted for small multiples display
    raw_vals <- adj_vals2[!grepl("sesong", adj_vals2, ignore.case = TRUE)]
    if (length(raw_vals) > 0) {
      plot_emp <- plot_emp |>
        filter(.data[[adj_col2]] %in% raw_vals[1])
    }
  }

  # Collapse age dimension
  age_col2 <- if ("alder" %in% col_names2) "alder" else NULL
  if (!is.null(age_col2)) {
    age_vals2 <- unique(plot_emp[[age_col2]])
    total_age <- age_vals2[grepl("^i alt$|alle aldre|total|15-74", age_vals2, ignore.case = TRUE)]
    if (length(total_age) > 0) {
      plot_emp <- plot_emp |> filter(.data[[age_col2]] %in% total_age[1])
    }
  }

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

    grp_summary <- plot_emp |>
      group_by(date, .data[[facet_col]]) |>
      summarise(value = mean(value, na.rm = TRUE), .groups = "drop")

    p2 <- ggplot(grp_summary,
                 aes(x = date, y = value, colour = .data[[facet_col]],
                     fill = .data[[facet_col]])) +
      geom_area(alpha = 0.25, position = "identity") +
      geom_line(linewidth = 0.8) +
      facet_wrap(vars(.data[[facet_col]]), ncol = 2) +
      scale_colour_manual(values = met.brewer("Hiroshige", n = length(unique(grp_summary[[facet_col]])))) +
      scale_fill_manual(values  = met.brewer("Hiroshige", n = length(unique(grp_summary[[facet_col]])))) +
      scale_x_date(date_breaks = "1 year", date_labels = "%Y") +
      scale_y_continuous(labels = label_number(suffix = "%")) +
      labs(
        title = "Employment rate by gender, Norway",
        subtitle = "Share of population in employment — small multiples reveal persistent gender gap",
        x = NULL, y = "Employment rate (%)",
        caption = "Source: Statistics Norway, table 13760"
      ) +
      theme_minimal(base_size = 12) +
      theme(
        legend.position = "none",
        panel.grid.minor = element_blank(),
        axis.text.x = element_text(angle = 45, hjust = 1),
        plot.title = element_text(face = "bold"),
        plot.subtitle = element_text(colour = "grey40"),
        plot.caption = element_text(colour = "grey55", size = 9),
        strip.text = element_text(face = "bold")
      )
    print(p2) # fixed: missing print() call
  } else if (nrow(plot_emp) > 0) {
    # Fallback: single line
    grp_summary <- plot_emp |>
      group_by(date) |>
      summarise(value = mean(value, na.rm = TRUE), .groups = "drop")
    p2 <- ggplot(grp_summary, aes(x = date, y = value)) +
      geom_area(fill = met.brewer("Hiroshige", 3)[2], alpha = 0.3) +
      geom_line(colour = met.brewer("Hiroshige", 3)[1], linewidth = 0.9) +
      scale_x_date(date_breaks = "1 year", date_labels = "%Y") +
      scale_y_continuous(labels = label_number(suffix = "%")) +
      labs(title = "Employment rate, Norway",
           subtitle = "Share of population in employment",
           x = NULL, y = "Employment rate (%)",
           caption = "Source: Statistics Norway, table 13760") +
      theme_minimal(base_size = 12) +
      theme(panel.grid.minor = element_blank(),
            axis.text.x = element_text(angle = 45, hjust = 1),
            plot.title = element_text(face = "bold"))
    print(p2) # fixed: missing print() call
  } else {
    message("plot_emp has no rows for plot-emprate-small-multiples")
  }
} else {
  message("df1_emprate not available for plot")
}

Section 3: Crime — Heatmap of Offence Categories Over Time

Two decades of reported crime data reveal which categories are trending upward and which have contracted. A heatmap normalised within each crime type shows the relative intensity across years.

if (exists("df2_rate") && !is.null(df2_rate) && nrow(df2_rate) > 0) {

  heat_data <- df2_rate |>
    mutate(year = year(date)) |>
    group_by(lovbruddstype, year) |>
    summarise(rate = mean(value, na.rm = TRUE), .groups = "drop") |>
    group_by(lovbruddstype) |>
    mutate(rate_scaled = (rate - min(rate, na.rm = TRUE)) /
             (max(rate, na.rm = TRUE) - min(rate, na.rm = TRUE) + 1e-9)) |>
    ungroup()

  # Wrap long labels
  heat_data <- heat_data |>
    mutate(lovbruddstype = stringr::str_wrap(lovbruddstype, width = 28))

  p3 <- ggplot(heat_data, aes(x = year, y = lovbruddstype, fill = rate_scaled)) +
    geom_tile(colour = "white", linewidth = 0.4) +
    scale_fill_gradientn(
      colours = met.brewer("Hiroshige", n = 9),
      name    = "Relative\nintensity",
      breaks  = c(0, 0.5, 1),
      labels  = c("Low", "Mid", "High")
    ) +
    scale_x_continuous(breaks = pretty_breaks(n = 8)) +
    labs(
      title    = "Reported crime rates by offence type, Norway",
      subtitle = "Rate per 1,000 inhabitants, scaled within each category (0 = lowest, 1 = highest year)",
      x        = "Year",
      y        = NULL,
      caption  = "Source: Statistics Norway, table 08484"
    ) +
    theme_minimal(base_size = 12) +
    theme(
      panel.grid = element_blank(),
      axis.text.y = element_text(size = 10),
      plot.title  = element_text(face = "bold"),
      plot.subtitle = element_text(colour = "grey40"),
      plot.caption  = element_text(colour = "grey55", size = 9),
      legend.position = "right"
    )
  print(p3)
}

Section 4: Slope Chart — Which Crime Categories Changed Most?

The slope chart compares the rate per 1,000 inhabitants in the earliest and most recent available years, highlighting the categories that shifted most dramatically — the headline story for Norway’s social policy debate.

if (exists("df2_slope") && !is.null(df2_slope) && nrow(df2_slope) > 0) {

  # Identify year column names dynamically (they start with "y")
  ycols <- names(df2_slope)[stringr::str_starts(names(df2_slope), "y") &
                              stringr::str_length(names(df2_slope)) == 5]

  if (length(ycols) == 2) {
    y_col1 <- ycols[1]
    y_col2 <- ycols[2]
    y1_lab <- sub("y", "", y_col1)
    y2_lab <- sub("y", "", y_col2)

    slope_long <- df2_slope |>
      pivot_longer(cols = all_of(ycols), names_to = "period", values_to = "rate") |>
      mutate(
        period_lab = sub("y", "", period),
        label_text = round(rate, 1)
      )

    # Colour by direction of change
    direction_df <- df2_slope |>
      mutate(direction = if_else(change >= 0, "Increase", "Decrease"),
             type_wrap = stringr::str_wrap(lovbruddstype, 25))

    slope_long <- slope_long |>
      left_join(direction_df |> select(lovbruddstype, direction, type_wrap),
                by = "lovbruddstype")

    dir_cols <- c("Increase" = "#c0392b", "Decrease" = "#2980b9")

    p4 <- ggplot(slope_long,
                 aes(x = period_lab, y = rate,
                     group = lovbruddstype, colour = direction)) +
      geom_line(linewidth = 1.1, alpha = 0.85) +
      geom_point(size = 3.2) +
      geom_text(data = slope_long |> filter(period == y_col1),
                aes(label = type_wrap, x = period_lab),
                hjust = 1.08, size = 2.9, fontface = "bold",
                lineheight = 0.85) +
      geom_text(data = slope_long |> filter(period == y_col2),
                aes(label = round(rate, 1)),
                hjust = -0.25, size = 2.9) +
      scale_colour_manual(values = dir_cols, name = "Trend") +
      scale_x_discrete(expand = expansion(mult = c(0.45, 0.15))) +
      labs(
        title    = paste("Crime rates per 1,000 inhabitants:", y1_lab, "vs", y2_lab),
        subtitle = "Red lines = rising offence categories; blue = declining — property and violence dominate",
        x        = NULL,
        y        = "Rate per 1,000 inhabitants",
        caption  = "Source: Statistics Norway, table 08484"
      ) +
      theme_minimal(base_size = 12) +
      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 = "grey55", size = 9),
        legend.position = "top"
      )
    print(p4) # fixed: missing print() call
  } else {
    message("Expected 2 year columns, found: ", length(ycols))
  }
} else {
  message("df2_slope not available for plot")
}

Section 5: Lollipop — Absolute Change in Crime Rate by Offence Type

To make the magnitude of each shift clear, a lollipop chart ranks categories by the absolute change in rate per 1,000 inhabitants from the first to the last available year.

if (exists("df2_slope") && !is.null(df2_slope) && nrow(df2_slope) > 0) {

  lollipop_data <- df2_slope |>
    mutate(
      type_wrap = stringr::str_wrap(lovbruddstype, 32),
      direction = if_else(change >= 0, "Increase", "Decrease"),
      type_wrap = fct_reorder(type_wrap, change)
    )

  if (nrow(lollipop_data) > 0) {
    dir_cols2 <- c("Increase" = "#c0392b", "Decrease" = "#2980b9")

    p5 <- ggplot(lollipop_data,
                 aes(x = change, y = type_wrap, colour = direction)) +
      geom_vline(xintercept = 0, colour = "grey70", linewidth = 0.7) +
      geom_segment(aes(x = 0, xend = change, y = type_wrap, yend = type_wrap),
                   linewidth = 1.2, alpha = 0.8) +
      geom_point(size = 4.5) +
      geom_text(aes(label = round(change, 2),
                    hjust = if_else(change >= 0, -0.3, 1.3)),
                size = 3.2, fontface = "bold") +
      scale_colour_manual(values = dir_cols2, guide = "none") +
      labs(
        title    = "Change in reported crime rate per 1,000 inhabitants",
        subtitle = "Earliest to most recent year — property theft and violence show opposing trends",
        x        = "Change in rate per 1,000 inhabitants",
        y        = NULL,
        caption  = "Source: Statistics Norway, table 08484"
      ) +
      theme_minimal(base_size = 12) +
      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 = "grey55", size = 9)
      )
    print(p5) # fixed: missing print() call
  } else {
    message("lollipop_data has no rows")
  }
} else {
  message("df2_slope not available for plot")
}

Key Findings

• Unemployment remains contained but not disappearing. The monthly series shows Norway’s unemployment rate hovering around its long-run average, masking cyclical troughs and peaks that carry significant human costs.

• The gender employment gap persists. Men’s employment rate continues to exceed women’s across the entire observation window, a structural feature that labour-market reforms have only partially closed.

• Property theft is the dominant crime category in absolute terms, accounting for the largest share of all reported offences per 1,000 inhabitants throughout the period — a particular pressure on communities where youth employment is weakest.

• Violence and ill-treatment has trended upward in rate-per-capita terms over recent years, even as overall reported crime has moved with macro-economic cycles, suggesting social stresses are intensifying in specific sub-populations.

• Drug-related and public-order offences show the highest relative volatility, cycling sharply with policing priorities and legislative changes — making them unreliable indicators of underlying social distress but sensitive gauges of enforcement intensity.

Closing Reflection

The twin pressures documented here — an employment market that remains structurally tight for certain groups and a crime environment where property and violence offences are elevated — describe a Norway that is prosperous in aggregate but unequal in distribution. For young people caught between a labour market that rewards experience and a cost of living that punishes those without secure incomes, the headline statistics offer cold comfort. Whether Norway’s policy response — active labour-market programmes, targeted policing, and housing investment — proves sufficient to bend these curves is the question that will define the country’s social contract over the coming decade.