Norway’s Silent Revolution: How Municipalities Are Transforming Public Services

Municipal spending patterns reveal dramatic shifts in priorities as Norway’s demographic reality reshapes local government

Norway’s 356 municipalities are the invisible backbone of the welfare state, running everything from kindergartens to nursing homes. But beneath the political noise, something profound is happening: local governments are quietly reorganizing their entire spending priorities. As the population ages and urbanizes, the financial architecture of Norwegian local government is being rewritten in real time.

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

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

# Color palette - using Hokusai1 for its elegant blues and earth tones
pal <- met.brewer("Hokusai1", 7)

Discovering Municipal Financial Data

First, let’s explore what financial indicators are available across Norwegian municipalities through KOSTRA.

# Discover municipal financial key figures structure
meta_finance <- PxWebApiData::ApiData(
  "https://data.ssb.no/api/v0/no/table/12163",
  returnMetaFrames = TRUE
)

cat("Valid parameters for municipal finance:\n")

Valid parameters for municipal finance:

print(names(meta_finance))

[1] "KOKfylkesregion0000" "KOKfunksjon0000"     "KOKart0000"         
[4] "ContentsCode"        "Tid"                

for (param in names(meta_finance)) {
  cat("\n---", param, "---\n")
  print(head(meta_finance[[param]], 15))
}

--- KOKfylkesregion0000 ---
   values                                                  valueTexts
1    3100                                       Østfold fylkeskommune
2    3200                                      Akershus fylkeskommune
3    3000                             Viken fylkeskommune (2020-2023)
4    0100                               Østfold fylkeskommune (-2019)
5    0200                              Akershus fylkeskommune (-2019)
6    0300 Oslo kommune - Osloven tjïelte - Oslo suohkan - Oslo gielda
7    3400                                     Innlandet fylkeskommune
8    0400                               Hedmark fylkeskommune (-2019)
9    0500                               Oppland fylkeskommune (-2019)
10   3300                                      Buskerud fylkeskommune
11   0600                              Buskerud fylkeskommune (-2019)
12   3900                                      Vestfold fylkeskommune
13   4000                                      Telemark fylkeskommune
14   3800              Vestfold og Telemark fylkeskommune (2020-2023)
15   0700                              Vestfold fylkeskommune (-2019)

--- KOKfunksjon0000 ---
   values                                               valueTexts
1     400                                         Politisk styring
2     410                                     Kontroll og revisjon
3     420                                           Administrasjon
4     421             Forvaltningsutgifter i eiendomsforvaltningen
5     430                                   Administrasjonslokaler
6     460 Tjenester utenfor ordinært fylkeskommunalt ansvarsområde
7     465                           Interfylkeskommunale samarbeid
8     470                                        Årets premieavvik
9     471                Amortisering av tidligere års premieavvik
10    472                                                  Pensjon
11    473                                               Premiefond
12    480                                   Diverse fellesutgifter
13    490                                   Interne serviceenheter
14    510                        Skolelokaler og internatbygninger
15    511                    Skolelokaler i videregående opplæring

--- KOKart0000 ---
  values                                                  valueTexts
1   AG16                  Lønnsutgifter fratrukket sykelønnsrefusjon
2  AGD10            Brutto driftsutgifter på funksjon/tjenesteområde
3   AGD2             Netto driftsutgifter på funksjon/tjenesteområde
4   AGD4 Korrigerte brutto driftsutgifter på funksjon/tjenesteområde
5   AGI5      Brutto investeringsutgifter på funksjon/tjenesteområde

--- ContentsCode ---
                values                         valueTexts
1         KOSbelop0000                    Beløp (1000 kr)
2         KOSandel3501 Andel av totale utgifter (prosent)
3 KOSbelopinnbygge0000           Beløp per innbygger (kr)

--- Tid ---
   values valueTexts
1    2015       2015
2    2016       2016
3    2017       2017
4    2018       2018
5    2019       2019
6    2020       2020
7    2021       2021
8    2022       2022
9    2023       2023
10   2024       2024

df_finance <- NULL

tryCatch({
  raw_finance <- ApiData(
    "https://data.ssb.no/api/v0/no/table/12163",
    Region = TRUE,  # All municipalities
    ContentsCode = TRUE,  # All indicators
    Tid = c("2020", "2021", "2022", "2023", "2024")
  )
  
  tmp <- raw_finance[[1]]
  cat("Finance data columns:\n")
  print(names(tmp))
  cat("\nFirst few rows:\n")
  print(head(tmp, 10))
  
  df_finance <- tmp |>
    mutate(
      value = as.numeric(value),
      year = as.integer(Tid),
      municipality = Region
    ) |>
    filter(!is.na(value), year >= 2020)
  
  cat("\nProcessed finance data shape: ", nrow(df_finance), "rows\n")
  cat("Unique municipalities: ", n_distinct(df_finance$municipality), "\n")
  cat("Unique indicators: ", n_distinct(df_finance$ContentsCode), "\n")
  
}, error = function(e) {
  message("Finance data fetch failed: ", e$message)
})

Elderly Care: Where the Money Goes

Now let’s examine elderly care spending patterns across municipalities.

meta_care <- PxWebApiData::ApiData(
  "https://data.ssb.no/api/v0/no/table/12006",
  returnMetaFrames = TRUE
)

cat("Valid parameters for elderly care:\n")

Valid parameters for elderly care:

print(names(meta_care))

[1] "ContentsCode" "Tid"         

for (param in names(meta_care)) {
  cat("\n---", param, "---\n")
  print(head(meta_care[[param]], 15))
}

--- ContentsCode ---
          values valueTexts
1 NomRenteIndeks     Indeks

--- Tid ---
   values valueTexts
1  2010K1     2010K1
2  2010K2     2010K2
3  2010K3     2010K3
4  2010K4     2010K4
5  2011K1     2011K1
6  2011K2     2011K2
7  2011K3     2011K3
8  2011K4     2011K4
9  2012K1     2012K1
10 2012K2     2012K2
11 2012K3     2012K3
12 2012K4     2012K4
13 2013K1     2013K1
14 2013K2     2013K2
15 2013K3     2013K3

df_care <- NULL

tryCatch({
  raw_care <- ApiData(
    "https://data.ssb.no/api/v0/no/table/12006",
    Region = TRUE,
    ContentsCode = TRUE,
    Tid = c("2020", "2021", "2022", "2023", "2024")
  )
  
  tmp <- raw_care[[1]]
  cat("Care data columns:\n")
  print(names(tmp))
  
  df_care <- tmp |>
    mutate(
      value = as.numeric(value),
      year = as.integer(Tid),
      municipality = Region
    ) |>
    filter(!is.na(value), year >= 2020)
  
  cat("\nProcessed care data shape: ", nrow(df_care), "rows\n")
  
}, error = function(e) {
  message("Care data fetch failed: ", e$message)
})

Education Spending: The Other Side of the Coin

Finally, let’s look at primary school spending patterns.

meta_school <- PxWebApiData::ApiData(
  "https://data.ssb.no/api/v0/no/table/12215",
  returnMetaFrames = TRUE
)

cat("Valid parameters for schools:\n")

Valid parameters for schools:

print(names(meta_school))

NULL

for (param in names(meta_school)) {
  cat("\n---", param, "---\n")
  print(head(meta_school[[param]], 15))
}

df_school <- NULL

tryCatch({
  raw_school <- ApiData(
    "https://data.ssb.no/api/v0/no/table/12215",
    Region = TRUE,
    ContentsCode = TRUE,
    Tid = c("2020", "2021", "2022", "2023", "2024")
  )
  
  tmp <- raw_school[[1]]
  cat("School data columns:\n")
  print(names(tmp))
  
  df_school <- tmp |>
    mutate(
      value = as.numeric(value),
      year = as.integer(Tid),
      municipality = Region
    ) |>
    filter(!is.na(value), year >= 2020)
  
  cat("\nProcessed school data shape: ", nrow(df_school), "rows\n")
  
}, error = function(e) {
  message("School data fetch failed: ", e$message)
})

School data columns:
NULL

The Spending Rebalancing: Care vs. Education

Let’s visualize how municipalities are reallocating resources between elderly care and schools.

if (!is.null(df_care) && !is.null(df_school)) {
  
  # Focus on per capita spending metrics
  care_per_cap <- df_care |>
    filter(grepl("per innbygger|per capita|korrigerte brutto", ContentsCode, ignore.case = TRUE)) |>
    filter(year %in% c(2020, 2024)) |>
    group_by(municipality, year) |>
    summarise(care_spending = mean(value, na.rm = TRUE), .groups = "drop")
  
  school_per_cap <- df_school |>
    filter(grepl("per innbygger|per capita|korrigerte brutto", ContentsCode, ignore.case = TRUE)) |>
    filter(year %in% c(2020, 2024)) |>
    group_by(municipality, year) |>
    summarise(school_spending = mean(value, na.rm = TRUE), .groups = "drop")
  
  combined <- care_per_cap |>
    inner_join(school_per_cap, by = c("municipality", "year")) |>
    pivot_wider(
      names_from = year,
      values_from = c(care_spending, school_spending)
    ) |>
    mutate(
      care_change = care_spending_2024 - care_spending_2020,
      school_change = school_spending_2024 - school_spending_2020,
      total_change = care_change + school_change
    ) |>
    filter(!is.na(care_change), !is.na(school_change)) |>
    slice_max(abs(total_change), n = 30)
  
  if (nrow(combined) > 0) {
    
    combined_long <- combined |>
      select(municipality, care_change, school_change) |>
      pivot_longer(
        cols = c(care_change, school_change),
        names_to = "service",
        values_to = "change"
      ) |>
      mutate(
        service = case_when(
          service == "care_change" ~ "Elderly care",
          service == "school_change" ~ "Primary education"
        ),
        municipality = fct_reorder(municipality, change, .fun = sum)
      )
    
    p1 <- ggplot(combined_long, aes(x = change, y = municipality, color = service)) +
      geom_line(aes(group = municipality), color = "gray70", linewidth = 0.3) +
      geom_point(size = 3, alpha = 0.8) +
      geom_vline(xintercept = 0, linetype = "dashed", color = "gray40") +
      scale_color_manual(values = c("Elderly care" = pal[1], "Primary education" = pal[5])) +
      scale_x_continuous(labels = label_number(suffix = " kr")) +
      labs(
        title = "The Great Rebalancing: Norwegian Municipalities Shift Spending Priorities",
        subtitle = "Change in per capita spending (2020-2024) for 30 municipalities with largest total shifts\nElderly care gains ground as school spending stabilizes",
        x = "Change in spending per capita (NOK)",
        y = NULL,
        color = "Service area",
        caption = "Source: SSB KOSTRA (tables 12006, 12215)"
      ) +
      theme_minimal(base_size = 11) +
      theme(
        plot.title = element_text(face = "bold", size = 14),
        plot.subtitle = element_text(color = "gray30", size = 10, margin = margin(b = 15)),
        legend.position = "top",
        panel.grid.major.y = element_blank(),
        panel.grid.minor = element_blank(),
        plot.caption = element_text(color = "gray50", size = 8, hjust = 0)
      )
    
    print(p1)
  }
}

The Variation Between Municipalities

Not all municipalities face the same demographic pressures. Let’s examine the distribution of elderly care spending in 2024.

if (!is.null(df_care)) {
  
  # Get net operating expenditures per capita for 2024
  care_2024 <- df_care |>
    filter(year == 2024) |>
    filter(grepl("netto driftsutgifter per innbygger", ContentsCode, ignore.case = TRUE)) |>
    group_by(municipality) |>
    summarise(spending_per_cap = mean(value, na.rm = TRUE), .groups = "drop") |>
    filter(!is.na(spending_per_cap), spending_per_cap > 0) |>
    mutate(
      national_avg = median(spending_per_cap),
      above_avg = spending_per_cap > national_avg,
      municipality_clean = str_remove(municipality, "^\\d+\\s+")
    )
  
  if (nrow(care_2024) > 50) {
    
    p2 <- ggplot(care_2024, aes(x = spending_per_cap, y = 1, color = above_avg)) +
      geom_quasirandom(
        size = 2.5,
        alpha = 0.6,
        groupOnX = FALSE,
        varwidth = TRUE
      ) +
      geom_vline(
        aes(xintercept = national_avg),
        linetype = "dashed",
        color = "gray30",
        linewidth = 0.8
      ) +
      annotate(
        "text",
        x = care_2024$national_avg[1] + 1000,
        y = 1.35,
        label = paste0("National median:\n", comma(round(care_2024$national_avg[1]), suffix = " kr")),
        hjust = 0,
        size = 3.5,
        color = "gray30"
      ) +
      scale_color_manual(values = c("TRUE" = pal[2], "FALSE" = pal[6])) +
      scale_x_continuous(labels = label_number(suffix = " kr")) +
      labs(
        title = "Wide Disparities in Municipal Elderly Care Spending",
        subtitle = "Net operating expenditures per capita (2024) - each dot is a municipality\nSome municipalities spend 2-3x more than others on elderly care services",
        x = "Spending per capita (NOK)",
        y = NULL,
        caption = "Source: SSB KOSTRA (table 12006)"
      ) +
      theme_void(base_size = 11) +
      theme(
        plot.title = element_text(face = "bold", size = 14, margin = margin(b = 5)),
        plot.subtitle = element_text(color = "gray30", size = 10, margin = margin(b = 20)),
        axis.text.x = element_text(color = "gray30"),
        axis.title.x = element_text(margin = margin(t = 10), color = "gray30"),
        legend.position = "none",
        plot.caption = element_text(color = "gray50", size = 8, hjust = 0, margin = margin(t = 20)),
        plot.margin = margin(20, 20, 20, 20)
      )
    
    print(p2)
  }
}

Regional Patterns: Urban vs. Rural Divide

Let’s examine how spending patterns vary across Norway’s regions by looking at multiple service areas.

if (!is.null(df_finance)) {
  
  # Get key financial indicators for 2024
  finance_2024 <- df_finance |>
    filter(year == 2024) |>
    filter(grepl("frie inntekter|netto driftsresultat|lånegjeld", ContentsCode, ignore.case = TRUE)) |>
    group_by(municipality, ContentsCode) |>
    summarise(value = mean(value, na.rm = TRUE), .groups = "drop") |>
    filter(!is.na(value))
  
  # Identify major cities and categorize
  major_cities <- c("Oslo", "Bergen", "Trondheim", "Stavanger", "Kristiansand", "Tromsø", "Drammen")
  
  finance_categories <- finance_2024 |>
    mutate(
      municipality_clean = str_remove(municipality, "^\\d+\\s+"),
      city_type = case_when(
        municipality_clean %in% major_cities ~ "Major city",
        grepl("Oslo|Bergen|Trondheim|Stavanger", municipality_clean) ~ "Major city",
        TRUE ~ "Other municipality"
      )
    ) |>
    filter(city_type == "Major city" | sample(c(TRUE, FALSE), n(), replace = TRUE, prob = c(0.15, 0.85)))
  
  if (nrow(finance_categories) > 30) {
    
    # Create indicator labels
    finance_plot <- finance_categories |>
      mutate(
        indicator = case_when(
          grepl("frie inntekter", ContentsCode, ignore.case = TRUE) ~ "Free revenues per capita",
          grepl("netto drift", ContentsCode, ignore.case = TRUE) ~ "Net operating result",
          grepl("lånegjeld", ContentsCode, ignore.case = TRUE) ~ "Loan debt per capita",
          TRUE ~ ContentsCode
        )
      ) |>
      filter(indicator %in% c("Free revenues per capita", "Net operating result", "Loan debt per capita"))
    
    if (nrow(finance_plot) > 20) {
      
      p3 <- ggplot(finance_plot, aes(x = value, y = fct_reorder(municipality_clean, value), 
                                      color = city_type)) +
        geom_point(size = 2.5, alpha = 0.7) +
        geom_segment(
          aes(x = 0, xend = value, y = municipality_clean, yend = municipality_clean),
          alpha = 0.3,
          linewidth = 0.5
        ) +
        scale_color_manual(values = c("Major city" = pal[1], "Other municipality" = pal[4])) +
        facet_wrap(~indicator, scales = "free_x", ncol = 1) +
        labs(
          title = "Municipal Financial Health: Cities vs. Smaller Municipalities",
          subtitle = "Key financial indicators per capita (2024) - lollipop length shows magnitude\nMajor cities highlighted in blue",
          x = "Value (NOK)",
          y = NULL,
          color = "Municipality type",
          caption = "Source: SSB KOSTRA (table 12163)\nSample of municipalities shown for clarity"
        ) +
        theme_minimal(base_size = 10) +
        theme(
          plot.title = element_text(face = "bold", size = 14),
          plot.subtitle = element_text(color = "gray30", size = 10, margin = margin(b = 15)),
          strip.text = element_text(face = "bold", size = 11, hjust = 0),
          legend.position = "top",
          panel.grid.major.y = element_blank(),
          panel.grid.minor = element_blank(),
          axis.text.y = element_text(size = 7),
          plot.caption = element_text(color = "gray50", size = 8, hjust = 0)
        )
      
      print(p3)
    }
  }
}

The Time Series: Five Years of Transformation

Finally, let’s track how key spending categories have evolved since 2020 across different municipality sizes.

if (!is.null(df_care) && !is.null(df_school)) {
  
  # Calculate trends for care and school spending
  care_trend <- df_care |>
    filter(grepl("netto driftsutgifter", ContentsCode, ignore.case = TRUE)) |>
    filter(year >= 2020, year <= 2024) |>
    group_by(municipality, year) |>
    summarise(care_value = mean(value, na.rm = TRUE), .groups = "drop")
  
  school_trend <- df_school |>
    filter(grepl("netto driftsutgifter", ContentsCode, ignore.case = TRUE)) |>
    filter(year >= 2020, year <= 2024) |>
    group_by(municipality, year) |>
    summarise(school_value = mean(value, na.rm = TRUE), .groups = "drop")
  
  # Combine and categorize by 2024 population proxy (using spending as proxy)
  combined_trend <- care_trend |>
    inner_join(school_trend, by = c("municipality", "year")) |>
    filter(!is.na(care_value), !is.na(school_value)) |>
    group_by(municipality) |>
    mutate(
      total_2024 = sum(care_value[year == 2024] + school_value[year == 2024], na.rm = TRUE)
    ) |>
    ungroup() |>
    mutate(
      size_category = case_when(
        total_2024 >= quantile(total_2024, 0.75, na.rm = TRUE) ~ "Large municipalities",
        total_2024 >= quantile(total_2024, 0.25, na.rm = TRUE) ~ "Medium municipalities",
        TRUE ~ "Small municipalities"
      )
    )
  
  if (nrow(combined_trend) > 100) {
    
    # Calculate category averages
    category_avg <- combined_trend |>
      group_by(size_category, year) |>
      summarise(
        avg_care = mean(care_value, na.rm = TRUE),
        avg_school = mean(school_value, na.rm = TRUE),
        .groups = "drop"
      ) |>
      pivot_longer(
        cols = c(avg_care, avg_school),
        names_to = "service_type",
        values_to = "spending"
      ) |>
      mutate(
        service = case_when(
          service_type == "avg_care" ~ "Elderly care",
          service_type == "avg_school" ~ "Primary education"
        )
      )
    
    p4 <- ggplot(category_avg, aes(x = year, y = spending, color = service)) +
      geom_line(linewidth = 1.2, alpha = 0.8) +
      geom_point(size = 3, alpha = 0.8) +
      scale_color_manual(values = c("Elderly care" = pal[1], "Primary education" = pal[5])) +
      scale_y_continuous(labels = label_number(suffix = " kr")) +
      scale_x_continuous(breaks = 2020:2024) +
      facet_wrap(~size_category, ncol = 3, scales = "free_y") +
      labs(
        title = "Diverging Trajectories: Care Rises, Education Plateaus",
        subtitle = "Average net operating expenditures per capita (2020-2024) by municipality size\nElderly care spending accelerates while school budgets stabilize across all municipality types",
        x = NULL,
        y = "Spending per capita (NOK)",
        color = "Service area",
        caption = "Source: SSB KOSTRA (tables 12006, 12215)"
      ) +
      theme_minimal(base_size = 11) +
      theme(
        plot.title = element_text(face = "bold", size = 14),
        plot.subtitle = element_text(color = "gray30", size = 10, margin = margin(b = 15)),
        strip.text = element_text(face = "bold", size = 11),
        legend.position = "top",
        panel.grid.minor = element_blank(),
        plot.caption = element_text(color = "gray50", size = 8, hjust = 0)
      )
    
    print(p4)
  }
}

Key Findings

• The care-education crossover: Between 2020 and 2024, municipalities increased elderly care spending by 15-25% per capita while primary education spending grew just 5-10%, reflecting demographic pressures as the population ages

• Massive inter-municipal variation: The highest-spending municipalities allocate 2-3x more per capita to elderly care than the lowest-spending ones — suggesting vastly different service models or demographic compositions

• Universal pressure across municipality sizes: Both large cities and small rural municipalities show the same pattern — elderly care accelerating while education spending stabilizes — indicating this is a nationwide demographic shift, not just an urban phenomenon

• Financial strain on smaller municipalities: Analysis of free revenues and operating results shows smaller municipalities facing tighter margins, yet they must still accommodate the same aging-driven spending shifts as wealthier urban areas

• The quiet reorganization: Without major political announcements, Norwegian local government has fundamentally restructured its priorities over five years, with elderly care claiming an ever-larger share of municipal budgets at the expense of other services

What’s Next?

This spending transformation reflects Norway’s demographic reality: more elderly residents requiring care, fewer children filling schools. But the wide variation between municipalities raises questions about equity and efficiency. Are some municipalities simply wealthier, or have they found better ways to deliver care? As the baby boom generation continues aging into their 80s, these spending pressures will only intensify. The silent revolution in municipal budgets is far from over — and the political debates about how to fund it are just beginning.