skip to content
Data for Afghanistan

How we identify marketplace sites

1. Where people live

Province totals for 2025–26 (EUAA, from the September 2025 estimate) were split across 398 districts in proportion to NSIA's district estimates, then spread within each district using Kontur's 400-meter population hexagons (145,000 cells). Each district's urban share from NSIA was assigned to its densest cells. The result is a 2025 population surface that matches official totals at province level.

2. How long it takes to reach a market

A 1-kilometer travel-cost surface was built from 283,000 OpenStreetMap road segments (October 2026), with speeds from 70 km/h on motorways to 15 km/h on tracks and 4 km/h off-road. Travel time was computed with a least-cost path algorithm from three sets of existing markets: provincial centers, cities, and WFP-monitored markets.

3. Who is underserved

Rural cells more than 60 minutes from any existing market are underserved. Cells 30 to 60 minutes away are partly served; cells within 30 minutes are served.

4. Choosing sites

Candidates are the 367 official district centers and 268 OpenStreetMap towns, excluding those within 20 km or 30 minutes of an existing market. For each province, a greedy maximal-covering search picked up to three sites, each maximizing the underserved rural population newly brought within 60 minutes. Served people count at a reduced weight, and sites must be at least 25 km apart (15 km in small provinces).

5. From people to trade

Producer households were estimated per cell using FAO DIEM's 89.2% agricultural-activity rate for rural households, adjusted for settlement density and calibrated to that national rate, with 10% for urban cells. Only part of them will use a new market. Addressable households are weighted down where an existing market is already close, and by distance within the catchment. Turnover and jobs follow from documented assumptions.

Turnover = addressable producer HH × participation × sales per HH
+ addressable rural HH × participation × US$ 270 purchases

Full formulas, parameters and limitations are in the assumptions section.

Assumptions and calculations

Core formulas

Producer HH within 60 min (gross) = Σ cells ≤60 min [ population × producer share ] ÷ 7.7

Addressable producer HH = Σ cells ≤60 min [ population × producer share × competition weight × distance decay ] ÷ 7.7

Addressable rural HH = Σ rural cells ≤60 min [ population × competition weight × distance decay ] ÷ 7.7

Producer saless = addressable producer HH × participations × sales per participating HH

Other tradings = addressable rural HH × participations × US$ 270

Total turnovers = producer saless + other tradings    (s = low, base, high)

Direct jobss = 20 × sites + total turnovers ÷ 45,000    Indirect jobss = 1.6 × direct jobss

Definitions. A cell is one 400-meter population hexagon. Rural cells exclude each district's urban population (NSIA urban share, assigned to its densest cells). Competition weight: 1.0 if the cell is more than 60 minutes from an existing market, 0.5 if 30 to 60 minutes, 0.15 if within 30 minutes. Distance decay: 1.0 if the cell is within 30 minutes of the candidate site, 0.6 if 30 to 60 minutes. Addressable rural HH counts all rural households, producers or not, who could buy at the marketplace.

Parameters

ParameterValueSource or basisStatus
Population baselineProvince totals 2025–26EUAA (2026) from the September 2025 estimateStrong secondary
District distributionNSIA 1398 (2019–20) district shares; unmatched new districts allocated within provinceCSO/NSIA via CookmanStrong secondary
Within-district distributionKontur 400 m H3 population (2023) as weightsKontur (2023)Estimated
Household size7.7 peopleALCS 2016–17Strong secondary
Producer share, rural cells95% (<500/km²), 92% (500–3,000), 88% (3,000–6,000), 82% (>6,000); calibrated to 89.2% nationalFAO DIEM Round 8 (2024), 9,494 rural householdsEstimated
Producer share, urban cells10%AssumptionAssumption
Road speedsMotorway 70; trunk 60; primary 50; secondary 40; tertiary 30; unclassified 25; residential 20; track and service 15; paths 5 km/h; ×0.75 if unpaved; off-road 4 km/hAdapted from Weiss et al. (2018)Assumption
Existing markets34 provincial centers, OSM cities, 40 WFP-monitored marketsCOD-AB v03, OSM, WFPDirect
Underserved thresholdMore than 60 minutes to an existing marketARMO designAssumption
Competition weight1.0 if >60 min from existing market; 0.5 if 30–60; 0.15 if ≤30ARMO designAssumption
Distance decay1.0 within 30 min of site; 0.6 for 30–60 minARMO designAssumption
Participation10% low, 20% base, 30% high of addressable householdsWorkbookAssumption
Sales per participating HHUS$ 600 livestock and cereal; 750 mixed; 950 irrigated horticulture; 1,150 high-value tree cropsBuilt from WFP prices (see below)Assumption
Other tradingUS$ 270 per participating rural household per year25% of a staple basket (12-month median WFP prices) bought at the marketplaceEstimated
JobsDirect: 20 core staff per site + 1 job per US$ 45,000 turnover. Indirect: 1.6 × direct jobsPlanning benchmarksAssumption

Deriving sales per household

Sales per participating household are built up from typical quantities sold and current unit values, so every figure can be checked and replaced with field data. All conversions use AFN 65 per US$ (WFP, October 2026).

Component (unit value)Livestock & cerealMixed farmingIrrigated horticultureHigh-value tree crops
Small ruminants sold
US$ 140 per sheep or goat		3 head = 4202 head = 2801 head = 1401 head = 140
Dairy and livestock products
qurut, butter, wool, eggs; fixed at US$ 50		50505050
Wheat or other cereal surplus
US$ 420 per metric ton		0.31 t = 1300.5 t = 2100.31 t = 1300.2 t = 84
Fresh fruit and vegetables
US$ 300 per metric ton (farm gate)		none0.7 t = 2102.1 t = 6300.6 t = 180
Nuts, dried fruit or saffron
US$ 4 per kg (farm gate)		nonenonenone174 kg = 696
Total, US$ per year6007509501,150
  • Sheep value. WFP reports livestock terms of trade of 341 kg of wheat per one-year-old female sheep (June 2024). At the October 2026 wheat price of AFN 27 per kg, that is 341 × 27 = AFN 9,207, or about US$ 140 per head.
  • Wheat value. AFN 27 per kg ÷ 65 = US$ 0.42 per kg, or US$ 420 per metric ton (WFP, October 2026). Most rural households are net wheat buyers, so the marketed surplus is small.
  • Fruit, vegetable and nut values. Farm-gate prices are planning assumptions set well below retail and export prices. Afghanistan exported US$ 643 million of fresh and dried fruit and nuts in 2024.
  • How to read the table. Each column adds the values in its rows. For example, a livestock and cereal household selling 3 sheep (US$ 420), US$ 50 of dairy products and 0.31 metric tons of wheat (US$ 130) sells US$ 600 a year in total.

Other trading activity

A staple basket for a 7.7-person household (89 kg wheat flour, 21 kg rice, 9 kg pulses, 7 kg oil, 5 kg sugar and 1 kg salt a month) costs about AFN 5,900 a month at median WFP prices over October 2025 to October 2026 (AFN 6,050 in October 2026 alone), or about US$ 1,090 a year. Assuming a quarter of it is bought at the new marketplace gives US$ 270 per participating household.

Double counting

Site figures include everyone within 60 minutes, including people across provincial borders. Province totals take each cell once, at its best distance-decay value. National totals de-duplicate across provinces, which is why they are lower than the sum of province totals.

Limitations

  • No elevation model was available, so off-road speed is flat at 4 km/h. Travel in steep terrain and winter is slower, and catchments in the highlands are likely optimistic.
  • OpenStreetMap completeness varies. Unmapped tracks make some catchments too small.
  • Only provincial centers, cities and WFP-monitored markets count as existing markets. Smaller district bazaars exist, so "underserved" means far from a major market.
  • Kontur's 2023 layer is uneven in some provinces (Sar-e Pul notably). It is used only to distribute official totals within districts.
  • Product profiles are province-level. Site-level product mix needs field confirmation.
  • Security, land tenure, flood risk and community acceptance were not modeled.

Bibliography

  1. European Union Agency for Asylum (2026). Afghanistan: Country Focus. Provincial population table for 2025–26 reproduced from the de facto authorities' September 2025 population estimate. https://www.euaa.europa.eu/print/pdf/node/29177
  2. Central Statistics Organization / National Statistics and Information Authority. Estimated Population of Afghanistan 2019–20 (1398), district rural and urban estimates; machine-readable compilation by C. Cookman, afghanistan_district_population_data (GitHub). https://github.com/colincookman/afghanistan_district_population_data
  3. Kontur (2023). Afghanistan: Population Density for 400m H3 Hexagons, release 2023-11-01. Humanitarian Data Exchange. https://data.humdata.org/dataset/kontur-population-afghanistan
  4. OCHA Afghanistan and AGCHO (2025). Afghanistan Subnational Administrative Boundaries (COD-AB), version 03, valid from 1 June 2025. Humanitarian Data Exchange. https://data.humdata.org/dataset/cod-ab-afg
  5. OCHA Afghanistan and NSIA (2021). Afghanistan Subnational Population Statistics (COD-PS). Humanitarian Data Exchange. https://data.humdata.org/dataset/cod-ps-afg
  6. Runfola, D. et al. (2020). geoBoundaries: A global database of political administrative boundaries. PLoS ONE 15(4): e0231866. Afghanistan ADM2 layer (source: AIMS), build December 2023. https://www.geoboundaries.org
  7. Humanitarian OpenStreetMap Team (2026). Afghanistan Roads (OpenStreetMap Export), snapshot 7 October 2026. OpenStreetMap contributors, ODbL. Humanitarian Data Exchange. https://data.humdata.org/dataset/hotosm_afg_roads
  8. Humanitarian OpenStreetMap Team (2026). Afghanistan Populated Places (OpenStreetMap Export), snapshot 7 October 2026. OpenStreetMap contributors, ODbL. Humanitarian Data Exchange. https://data.humdata.org/dataset/hotosm_afg_populated_places
  9. Natural Earth (2024). 1:10m Cultural Vectors: Roads. Used for national cartography only. https://www.naturalearthdata.com
  10. World Food Programme (2026). Afghanistan – Food Prices, market-level retail prices, wages and exchange rates to October 2026. Humanitarian Data Exchange. https://data.humdata.org/dataset/wfp-food-prices-for-afghanistan
  11. World Food Programme (2024). Afghanistan Monthly Market Report, Issue 49, June 2024 (livestock terms of trade: 341 kg wheat per one-year-old female sheep; wages; exchange rate). https://reliefweb.int/report/afghanistan/afghanistan-monthly-market-report-issue-49-june-2024
  12. FAO (2025). Data in Emergencies Monitoring (DIEM) Household Survey, Afghanistan, Round 8, 2024. 9,494 rural households across all 34 provinces, representative at province level. Variable hh_agricactivity: 89.2% of sampled households produced crops and/or livestock in the past 12 months (24.6% both, 51.1% crops only, 13.5% livestock only; unweighted). https://microdata.fao.org/index.php/catalog/2895/variable/F1/V13?name=hh_agricactivity
  13. FAO (2025). Data in Emergencies Monitoring (DIEM) Household Survey, Afghanistan, Round 7, 2023. 9,560 rural households. https://microdata.fao.org/index.php/catalog/2945
  14. Central Statistics Organization (2018). Afghanistan Living Conditions Survey 2016–17: Analysis Report. Kabul: CSO (household size 7.7; 63% of rural population within 2 km of an all-season road; 44% of jobs in agriculture). https://washdata.org/report/afghanistan-alcs-2016-17-analysis-report
  15. World Bank (2023). Afghanistan Welfare Monitoring Survey, Round 3. Washington, DC. https://thedocs.worldbank.org/en/doc/975d25c52634db31c504a2c6bee44d22-0310012023/original/Afghanistan-Welfare-Monitoring-Survey-3.pdf
  16. Pajhwok Afghan News (29 July 2025). Omar: 40pc of Afghanistan's population enjoy electricity (statement by the Chief Executive of Da Afghanistan Breshna Sherkat; about 80% of electricity imported). https://pajhwok.com/2025/07/29/omar-40pc-of-afghanistans-population-enjoy-electricity/
  17. DataReportal (2025). Digital 2025: Afghanistan (22.3 million cellular connections, 51.6% of population; 30.5% internet penetration). https://datareportal.com/reports/digital-2025-afghanistan
  18. Innovators Magazine (2021). Afghanistan has huge solar potential (about 300 days of sunshine a year). Cross-check: World Bank/ESMAP Global Solar Atlas, Afghanistan Global Horizontal Irradiation layer. https://energydata.info/dataset/afghanistan-global-horizontal-irradiation
  19. KabulNow (May 2025). Afghanistan exported $643 million worth of fresh and dried fruit in 2024 (solar year 1403), citing NSIA and the Office of the Deputy Prime Minister for Economic Affairs. https://kabulnow.com/2025/05/taliban-says-afghanistan-exported-643-million-worth-of-fruit-last-year/
  20. Ministry of Rural Rehabilitation and Development / NABDP. Provincial Development Profiles (34 provinces), and Regional Rural Economic Regeneration Strategies (RRERS) provincial profiles, e.g. Ghazni (2006). Archived at ecoi.net. https://www.ecoi.net/en/file/local/1098341/1222_1197552083_ghazni-provincial-profile.pdf
  21. Naval Postgraduate School, Program for Culture and Conflict Studies. Provincial overviews (agriculture and economy sections), e.g. Paktya. https://nps.edu/web/ccs/paktya
  22. FEWS NET (2024). Afghanistan Data Book: province-level crop production series (wheat, maize, rice, barley, cotton, fruit). https://help.fews.net/fde/v3/afghanistan-data-book
  23. FAO. Subnational Agricultural Production, Admin level 1 (province production and harvested area). https://data.fao.org/catalog/dataset/sub-national-agricultural-production-admin-level-1
  24. FAO (2025). Population counts, Afghanistan 2025, 100 m (derived from WorldPop). Recommended for the next update. https://data.apps.fao.org/catalog/iso/b0a91b97-6d58-4f2f-ac3e-c150747c0980
  25. Weiss, D.J. et al. (2018). A global map of travel time to cities to assess inequalities in accessibility in 2015. Nature 553: 333–336. (Friction-surface method for travel-time modeling.) https://doi.org/10.1038/nature25181
  26. Church, R. and ReVelle, C. (1974). The maximal covering location problem. Papers of the Regional Science Association 32: 101–118. (Site-selection method.) https://doi.org/10.1007/BF01942293
  27. van der Walt, S. et al. (2014). scikit-image: image processing in Python. PeerJ 2: e453 (minimum-cost-path routine used for travel-time surfaces). https://doi.org/10.7717/peerj.453

Map data © OpenStreetMap contributors, ODbL. Boundaries are for analytical purposes and imply no position on legal status.