Where Mongolia Grew: A Decade of Population Change, Mapped

Mongolia

mapping

population

Mapping ten years of population change across Mongolia’s soums and Ulaanbaatar’s khoroos with two R packages: mongolstats for the data and mongolmaps for the boundaries.

Author

Temuulen Enebish

Published

October 1, 2026

← Back to Blog

Population is the denominator behind almost every public-health rate, and where people live decides where clinics, schools and services are needed. This post maps how Mongolia’s population shifted between 2015 and 2025, using two R packages I maintain:

mongolstats fetches tables from the National Statistics Office (NSO) of Mongolia.
mongolmaps provides the boundaries, from aimags (provinces) down to Ulaanbaatar’s khoroos (subdistricts), and joins NSO tables to them.

Setup

Update, October 25, 2026: Both packages are now on CRAN. The latest mongolmaps release is 0.2.0; the analysis below was run with 0.1.0, as recorded in the session information.

Show code

install.packages(c("mongolstats", "mongolmaps"))

Show code

library(mongolstats)
library(mongolmaps)
library(dplyr)
library(tidyr)
library(ggplot2)
library(scales)

Fetching the data

The NSO publishes mid-year resident population for every aimag, soum, bag and khoroo in a single table, DT_NSO_0300_002V4. One call fetches 2015 and 2025 for every unit:

Show code

tbl <- "DT_NSO_0300_002V4"
regions <- nso_dim_values(tbl, "Region")$code

pop <- nso_data(
  tbl,
  selections = list(Region = regions, Year = c("2015", "2025")),
  labels = "en"
)

pop_wide <- pop |>
  select(Region, year = Year_en, value) |>
  pivot_wider(names_from = year, values_from = value, names_prefix = "pop_") |>
  mutate(change = pop_2025 / pop_2015 - 1)

national <- filter(pop_wide, Region == "0")

Mongolia’s population grew from 2,964,085 to 3,437,000, a 16.0% increase. The NSO table mixes every level in one Region column, from the national total down to bags. mn_join() recognises the NSO codes, keeps only the level you ask for, and attaches the boundaries:

Show code

soums <- mn_join(pop_wide, by = "Region", level = "soum")

n_declined <- sum(soums$change < 0)

Growth was concentrated, decline was widespread

National growth hides a split. Of Mongolia’s 339 soums and Ulaanbaatar districts, 135 (40%) had fewer residents in 2025 than in 2015.

Show code

p <- mn_map(soums, fill = change, title = "Population change by soum, 2015 to 2025") +
  scale_fill_gradient2(
    name = NULL, low = "#a6611a", mid = "#f5f5f5", high = "#018571",
    midpoint = 0, limits = c(-0.4, 0.4), oob = squish,
    labels = label_percent(style_positive = "plus")
  )
p + labs(caption = paste0(p$labels$caption, "\nPopulation: NSO Mongolia, DT_NSO_0300_002V4"))

Map of Mongolia's soums shaded by population change from 2015 to 2025, brown for decline and teal for growth. Declining soums cluster in the centre and south, while most soums in the far west, north and east grew, and the strongest growth is in a few southern Gobi mining soums and around Ulaanbaatar.

Declines cluster in the central and southern soums of the Khangai and Gobi, while most of the far west, north and east grew modestly. The strongest growth is in Umnugovi in the south: Khanbogd and Tsogttsetsii, home to the Oyu Tolgoi and Tavan Tolgoi mines, and the aimag centre Dalanzadgad. No soum lost more than a fifth of its residents, but a handful grew by more than 40%; the colour scale is capped there, so they share the darkest shade. Ulaanbaatar’s districts are too small to see at this scale.

Summed by aimag, the pattern is clearer:

Show code

aimags <- mn_join(pop_wide, by = "Region", level = "aimag") |>
  sf::st_drop_geometry()

ggplot(aimags, aes(change, reorder(name, change))) +
  geom_vline(xintercept = national$change, linetype = "dashed", colour = "grey50") +
  geom_vline(xintercept = 0, colour = "grey80") +
  geom_point(aes(colour = change > national$change), size = 2.5, show.legend = FALSE) +
  annotate("text", x = national$change, y = 1.5, hjust = -0.05, size = 3,
           label = "National average", colour = "grey40") +
  scale_x_continuous(labels = label_percent(style_positive = "plus")) +
  scale_colour_manual(values = c(`TRUE` = "#018571", `FALSE` = "grey45")) +
  labs(x = "Population change, 2015 to 2025", y = NULL,
       caption = "Source: NSO Mongolia, DT_NSO_0300_002V4") +
  theme_minimal(base_size = 11) +
  theme(panel.grid.minor = element_blank(), panel.grid.major.y = element_blank())

Dot chart of population change from 2015 to 2025 for Mongolia's 21 aimags and Ulaanbaatar, sorted from highest to lowest, with a dashed line at the national average of 16%. Umnugovi and Ulaanbaatar are the only units above the national average; Selenge, Arkhangai and Bulgan were roughly unchanged.

Only Umnugovi and Ulaanbaatar grew faster than the country as a whole. Every other aimag grew more slowly, and Selenge, Arkhangai and Bulgan were essentially unchanged. Because Ulaanbaatar was already home to almost half the country, 74% of the national increase landed in the capital.

Inside Ulaanbaatar

Many of Ulaanbaatar’s khoroos were split or redrawn after 2015, so khoroo populations can’t be compared fairly over time. The 2025 figures show something else worth knowing: how unevenly people are spread across the six central districts.

Show code

central <- c("BGD", "BZD", "ChD", "KhUD", "SKhD", "SBD")

khoroos <- mn_join(pop_wide, by = "Region", level = "bag", within = central) |>
  mutate(density = pop_2025 / area_km2) |>
  arrange(desc(density)) |>
  mutate(
    pop_share = cumsum(pop_2025) / sum(pop_2025),
    half = lag(pop_share, default = 0) < 0.5
  )

dense <- filter(khoroos, half)
area_share <- sum(dense$area_km2) / sum(khoroos$area_km2)

Ranked by density, the 90 most crowded of the 189 khoroos hold half of the central districts’ 1,603,387 residents, on 1.2% of their land.

Show code

mn_map(
  mutate(khoroos, group = if_else(half, "Densest khoroos holding half of residents", "Other khoroos")),
  fill = group,
  title = "Half of central Ulaanbaatar lives here"
) +
  scale_fill_manual(name = NULL, values = c("#018571", "grey88")) +
  theme(legend.position = "bottom")

Map of central Ulaanbaatar's khoroos with the densest khoroos, which together hold half the population, highlighted in teal. They form a small cluster in the city core, while the much larger outer khoroos are shown in light grey.

For health planning, that matters in both directions. Services in the dense core reach many people within a short distance, while the outer khoroos and the many shrinking rural soums need ways of reaching people that don’t depend on density.

Caveats

These are NSO mid-year estimates of resident population, so they reflect where people are registered, which can differ from where they actually live.
Changes in a district or soum can include boundary or registration changes, not only migration and births. That is why I don’t compare khoroos over time.
Khoroo boundaries come from an open project, not an official release, and mongolmaps flags them as unofficial.

Appendix: Reproducibility

Show code

sessionInfo()
#> R version 4.6.1 (2026-06-24 ucrt)
#> Platform: x86_64-w64-mingw32/x64
#> Running under: Windows 11 x64 (build 26300)
#> 
#> Matrix products: default
#>   LAPACK version 3.12.1
#> 
#> locale:
#> [1] LC_COLLATE=English_Canada.utf8  LC_CTYPE=English_Canada.utf8   
#> [3] LC_MONETARY=English_Canada.utf8 LC_NUMERIC=C                   
#> [5] LC_TIME=English_Canada.utf8    
#> 
#> time zone: America/Vancouver
#> tzcode source: internal
#> 
#> attached base packages:
#> [1] stats     graphics  grDevices utils     datasets  methods   base     
#> 
#> other attached packages:
#> [1] scales_1.4.0      ggplot2_4.0.3     tidyr_1.3.2       dplyr_1.2.1      
#> [5] mongolmaps_0.1.0  mongolstats_0.2.0
#> 
#> loaded via a namespace (and not attached):
#>  [1] gtable_0.3.6       jsonlite_2.0.0     compiler_4.6.1     Rcpp_1.1.2        
#>  [5] tidyselect_1.2.1   yaml_2.3.12        fastmap_1.2.0      R6_2.6.1          
#>  [9] labeling_0.4.3     generics_0.1.4     classInt_0.4-11    curl_7.1.0        
#> [13] httr2_1.3.0        sf_1.1-2           knitr_1.51         htmlwidgets_1.6.4 
#> [17] tibble_3.3.1       units_1.0-1        DBI_1.3.0          pillar_1.11.1     
#> [21] RColorBrewer_1.1-3 rlang_1.3.0        stringi_1.8.7      xfun_0.60         
#> [25] S7_0.2.2           otel_0.2.0         viridisLite_0.4.3  cli_3.6.6         
#> [29] withr_3.0.3        magrittr_2.0.5     class_7.3-24       digest_0.6.39     
#> [33] grid_4.6.1         lifecycle_1.0.5    vctrs_0.7.3        KernSmooth_2.23-26
#> [37] proxy_0.4-29       evaluate_1.0.5     glue_1.8.1         farver_2.1.2      
#> [41] e1071_1.7-17       rmarkdown_2.31     purrr_1.2.2        tools_4.6.1       
#> [45] pkgconfig_2.0.3    htmltools_0.5.9

--- title: "Where Mongolia Grew: A Decade of Population Change, Mapped" author: "Temuulen Enebish" date: "2026-10-01" description: "Mapping ten years of population change across Mongolia's soums and Ulaanbaatar's khoroos with two R packages: mongolstats for the data and mongolmaps for the boundaries." categories: [R, Mongolia, mapping, population] summary: "Mongolia's population grew by about 16% between 2015 and 2025, but the growth was uneven: Ulaanbaatar and the mining soums of Umnugovi grew fastest, while roughly two in five soums and districts shrank. In central Ulaanbaatar, half of all residents live on about 1% of the land. The analysis takes a few dozen lines of R using mongolstats and mongolmaps." --- <a href="/blog" class="back-link">← Back to Blog</a> ```{r setup, include=FALSE} knitr::opts_chunk$set( collapse = TRUE, comment = "#>", fig.width = 7, fig.height = 4.5, fig.dpi = 150, out.width = "100%", warning = FALSE, message = FALSE ) ``` Population is the denominator behind almost every public-health rate, and where people live decides where clinics, schools and services are needed. This post maps how Mongolia's population shifted between 2015 and 2025, using two R packages I maintain: - [mongolstats](https://temuulene.github.io/mongolstats/) fetches tables from the National Statistics Office (NSO) of Mongolia. - [mongolmaps](https://temuulene.github.io/mongolmaps/) provides the boundaries, from aimags (provinces) down to Ulaanbaatar's khoroos (subdistricts), and joins NSO tables to them. ## Setup **Update, October 25, 2026:** Both packages are now on CRAN. The latest mongolmaps release is 0.2.0; the analysis below was run with 0.1.0, as recorded in the session information. ```{r install, eval=FALSE} install.packages(c("mongolstats", "mongolmaps")) ``` ```{r load-packages} library(mongolstats) library(mongolmaps) library(dplyr) library(tidyr) library(ggplot2) library(scales) ``` ## Fetching the data The NSO publishes mid-year resident population for every aimag, soum, bag and khoroo in a single table, `DT_NSO_0300_002V4`. One call fetches 2015 and 2025 for every unit: ```{r fetch} tbl <- "DT_NSO_0300_002V4" regions <- nso_dim_values(tbl, "Region")$code pop <- nso_data( tbl, selections = list(Region = regions, Year = c("2015", "2025")), labels = "en" ) pop_wide <- pop |> select(Region, year = Year_en, value) |> pivot_wider(names_from = year, values_from = value, names_prefix = "pop_") |> mutate(change = pop_2025 / pop_2015 - 1) national <- filter(pop_wide, Region == "0") ``` Mongolia's population grew from `r comma(national$pop_2015)` to `r comma(national$pop_2025)`, a `r percent(national$change, accuracy = 0.1)` increase. The NSO table mixes every level in one `Region` column, from the national total down to bags. `mn_join()` recognises the NSO codes, keeps only the level you ask for, and attaches the boundaries: ```{r join-soums} soums <- mn_join(pop_wide, by = "Region", level = "soum") n_declined <- sum(soums$change < 0) ``` ## Growth was concentrated, decline was widespread National growth hides a split. Of Mongolia's `r nrow(soums)` soums and Ulaanbaatar districts, `r n_declined` (`r percent(n_declined / nrow(soums))`) had fewer residents in 2025 than in 2015. ```{r soum-map, fig.height=4.2} #| fig-alt: "Map of Mongolia's soums shaded by population change from 2015 to 2025, brown for decline and teal for growth. Declining soums cluster in the centre and south, while most soums in the far west, north and east grew, and the strongest growth is in a few southern Gobi mining soums and around Ulaanbaatar." p <- mn_map(soums, fill = change, title = "Population change by soum, 2015 to 2025") + scale_fill_gradient2( name = NULL, low = "#a6611a", mid = "#f5f5f5", high = "#018571", midpoint = 0, limits = c(-0.4, 0.4), oob = squish, labels = label_percent(style_positive = "plus") ) p + labs(caption = paste0(p$labels$caption, "\nPopulation: NSO Mongolia, DT_NSO_0300_002V4")) ``` Declines cluster in the central and southern soums of the Khangai and Gobi, while most of the far west, north and east grew modestly. The strongest growth is in Umnugovi in the south: Khanbogd and Tsogttsetsii, home to the Oyu Tolgoi and Tavan Tolgoi mines, and the aimag centre Dalanzadgad. No soum lost more than a fifth of its residents, but a handful grew by more than 40%; the colour scale is capped there, so they share the darkest shade. Ulaanbaatar's districts are too small to see at this scale. Summed by aimag, the pattern is clearer: ```{r aimag-chart, fig.height=5.5} #| fig-alt: "Dot chart of population change from 2015 to 2025 for Mongolia's 21 aimags and Ulaanbaatar, sorted from highest to lowest, with a dashed line at the national average of 16%. Umnugovi and Ulaanbaatar are the only units above the national average; Selenge, Arkhangai and Bulgan were roughly unchanged." aimags <- mn_join(pop_wide, by = "Region", level = "aimag") |> sf::st_drop_geometry() ggplot(aimags, aes(change, reorder(name, change))) + geom_vline(xintercept = national$change, linetype = "dashed", colour = "grey50") + geom_vline(xintercept = 0, colour = "grey80") + geom_point(aes(colour = change > national$change), size = 2.5, show.legend = FALSE) + annotate("text", x = national$change, y = 1.5, hjust = -0.05, size = 3, label = "National average", colour = "grey40") + scale_x_continuous(labels = label_percent(style_positive = "plus")) + scale_colour_manual(values = c(`TRUE` = "#018571", `FALSE` = "grey45")) + labs(x = "Population change, 2015 to 2025", y = NULL, caption = "Source: NSO Mongolia, DT_NSO_0300_002V4") + theme_minimal(base_size = 11) + theme(panel.grid.minor = element_blank(), panel.grid.major.y = element_blank()) ``` Only Umnugovi and Ulaanbaatar grew faster than the country as a whole. Every other aimag grew more slowly, and Selenge, Arkhangai and Bulgan were essentially unchanged. Because Ulaanbaatar was already home to almost half the country, `r percent(with(filter(aimags, name == "Ulaanbaatar"), pop_2025 - pop_2015) / (national$pop_2025 - national$pop_2015))` of the national increase landed in the capital. ## Inside Ulaanbaatar Many of Ulaanbaatar's khoroos were split or redrawn after 2015, so khoroo populations can't be compared fairly over time. The 2025 figures show something else worth knowing: how unevenly people are spread across the six central districts. ```{r khoroos} central <- c("BGD", "BZD", "ChD", "KhUD", "SKhD", "SBD") khoroos <- mn_join(pop_wide, by = "Region", level = "bag", within = central) |> mutate(density = pop_2025 / area_km2) |> arrange(desc(density)) |> mutate( pop_share = cumsum(pop_2025) / sum(pop_2025), half = lag(pop_share, default = 0) < 0.5 ) dense <- filter(khoroos, half) area_share <- sum(dense$area_km2) / sum(khoroos$area_km2) ``` Ranked by density, the `r nrow(dense)` most crowded of the `r nrow(khoroos)` khoroos hold half of the central districts' `r comma(sum(khoroos$pop_2025))` residents, on `r percent(area_share, accuracy = 0.1)` of their land. ```{r khoroo-map, fig.height=5} #| fig-alt: "Map of central Ulaanbaatar's khoroos with the densest khoroos, which together hold half the population, highlighted in teal. They form a small cluster in the city core, while the much larger outer khoroos are shown in light grey." mn_map( mutate(khoroos, group = if_else(half, "Densest khoroos holding half of residents", "Other khoroos")), fill = group, title = "Half of central Ulaanbaatar lives here" ) + scale_fill_manual(name = NULL, values = c("#018571", "grey88")) + theme(legend.position = "bottom") ``` For health planning, that matters in both directions. Services in the dense core reach many people within a short distance, while the outer khoroos and the many shrinking rural soums need ways of reaching people that don't depend on density. ## Caveats - These are NSO mid-year estimates of *resident* population, so they reflect where people are registered, which can differ from where they actually live. - Changes in a district or soum can include boundary or registration changes, not only migration and births. That is why I don't compare khoroos over time. - Khoroo boundaries come from an open project, not an official release, and mongolmaps flags them as unofficial. ## Appendix: Reproducibility ```{r session-info} sessionInfo() ```