Go to content

6. Exploring the spatial determinants of voting behaviour

Based on an Ordinary Least Squares (OLS) regression analysis, we have examined the factors influencing the likelihood of voting for populist parties in national elections across Nordic municipalities. As detailed in Section 4, Nordic parties were classified as populist according to the well-established PopuList classification (Rooduijn et al. 2024). Box 2 outlines the social, economic, demographic, and territorial variables considered in the analysis. Most of these variables are described in Section 5, where a spatial analysis revealed significant rural-urban differences for almost all of them. Additionally, the analysis includes variables indicating the presence of polluting facilities in or around municipalities. The study covers 1,013 municipalities in Denmark, Finland, Norway, and Sweden
Iceland was not included because of lack of municipal data for several explanatory factors.
, using standardized variables at the municipal level as predictors. All indicators were collected at the municipal (LAU2) level, except variables informing on the sectoral composition of employment, which are only available at the regional (NUTS-3) level.
Box 2: Independent variables considered in the regression analysis
  1. Green transition indicators: Share of green-task jobs (2022), greenhouse gas (GHG) emissions per employee in tonnes of CO2eq per person employed, presence of facilities included in the European Pollutant Release and Transfer Register (E-PRTR) in the municipality and within a 25-km radius. (Source: own, based on classification by Vona et al. (2019) and on National GHG inventories and employment by sector.)
  2. Electoral participation: Voting turnout for national elections, 2022 or latest election. (Source: own, based on data from NSIs and classification of populist parties in PopuList (Rooduijn et al. 2024))
  3. Territorial diversity: Urban-rural category based on the seven class, population-based Nordic urban-rural typology, reference year: 2022. (Source: own, based on data from NSIs and Eurostat. For information on the typology, see Annex 1 and Stjernberg et al. (2024))
  4. Demographic and compositional factors: Population aged 60 and over, gender ratio (male to female) and share of foreign-born individuals in the total population (2011-2022). (Source: own, based on data from NSIs.)
  5. Educational attainment: Share of population with tertiary education as the highest level of educational attainment (2011-2022). (Source: own, based on data from NSIs.)
  6. Employment evolution and structure: Sectoral distribution of employment by economic sector (NACE, lev. 1) and rolling average of the change rate for the previous four years (Source: ARDECO database and own, based on ARDECO database). Unemployment rate for ages 20-64 (2011-2022). (Source: own, based on data from NSIs.)
  7. Household income and inequality: Equivalised household income and rolling average of the change rate for the previous four years, and Gini coefficient on equivalised household income (2011-2022). (Source: own, based on data from NSIs.)
All variables are numeric, except for territorial diversity and the presence of polluting facilities, which are categorical. Numeric variables were standardized before fitting the model to ensure comparability.
The results of the analysis indicate that several predictors are significantly associated with the propensity to vote for a populist party.

Green transition factors

Higher levels of green task jobs in a municipality do not seem to be significantly associated with the proportion of vote for a populist political option. Nonetheless, municipalities with high levels of GHG emissions per employee show a smaller percentage of votes for populist parties. Such a negative association between populist voting and GHG emissions is further corroborated by the presence of polluting EPRTR
The European Pollutant Release and Transfer Register (EPRTR) Regulation requires facilities that exceed certain thresholds to report their pollutant emissions every year, starting in 2007. The Regulation specifies reporting thresholds for 91 different substances that have been classified as pollutants. Facilities which that meet or exceed the threshold for any of these pollutants are required to register their emissions.
facilities within municipalities or nearby areas, which is also negatively and significantly associated with populist vote share.
These somewhat surprising patterns might be related to existing working conditions in communities with prevalence of traditional industrial activities. In the Nordic context, where renewable energy sources dominate the energy production landscape, the majority of industrial emissions originate from traditional industrial facilities. These facilities are predominantly found in industries such as cement production, steel manufacturing, and the chemical industry. Emissions in these sectors are often characterised as difficult-to-abate, meaning they are challenging to reduce or eliminate using current technologies and practices. This is due to the inherent nature of the production processes involved, which often rely on high-temperature processes, chemical reactions, and the use of fossil fuels as feedstocks or energy sources (Kumar et al. 2024).
Industries in these sectors are often major local employers, providing jobs not only in the production process but also in supporting services like maintenance, transportation, and administration (Ostermeyer 2023). They source goods and services from local suppliers, which can have a positive impact on the local economy, and generate significant tax revenue for local municipalities, supporting public services and infrastructure. The industries can have a multiplier effect on the local economy, as employees spend their wages locally, supporting other businesses and services (Moritz et al. 2017). Additionally, compared to other activities, the cement, steel, and chemical industries in the Nordic countries tend to offer higher wages, more comprehensive benefits packages, greater job security, and more opportunities for professional advancement (Marginson & Dølvik 2020).  
Building on these considerations, we hypothesize that indirect factors may contribute to strengthening local communities and reducing perceptions of economic insecurity, which are often drivers of populist sentiment. This interpretation might explain why the presence of large industrial facilities in a municipality, often leading to greater GHG intensity of local employment, is often concurrent with a smaller percentage of votes for populist parties in those areas, as workers are more likely to feel invested in the existing economic and social order. By providing a sense of stability and security to local communities, traditional industries may indirectly reduce the appeal of populist movements, which can find momentum in feelings of economic anxiety and disillusionment with current development. Further research would be needed to delve deeper into these mechanisms and confirm whether the presence of large, traditional industries in the Nordic municipalities does in fact contribute to reduce the level of political discontent at the local level.

Other factors and controls

  • Electoral participation: The results reveal a significant negative correlation between voter turnout and populist voting. This implies that greater electoral participation seems to be associated with a lower propensity to vote for a populist party.
  • Territorial diversity: The analysis highlights the influence of the degree of urbanization on populist vote share. Municipalities categorised as intermediate – including peri-urban areas, rural areas close to urban areas and local centres in rural areas – show significantly higher populist voting compared to urban cores and remote rural areas.
  • Demographic and compositional factors: Demographic factors significantly affect voting for a populist party. Municipalities with a higher proportion of men relative to women exhibit a stronger propensity toward populist voting. When controlling for other factors, the share of the foreign-born population does not seem to be significantly associated with the propensity to vote for populist parties in the Nordic context, while the share of the population aged 60 years and over shows a negative coefficient, but is only marginally significant.
  • Educational attainment: Educational attainment is a relevant predictor of populist vote. A higher share of individuals with tertiary education as their highest educational attainment is strongly associated with a lower propensity to vote for populist parties. This implies that municipalities with an overrepresentation of individuals with less than tertiary education tend to cast higher share of votes for populist parties.
  • Employment structure and unemployment: A higher municipal unemployment rate is strongly associated with a greater propensity to vote for a populist party. This suggests that greater job insecurity and economic uncertainty may drive individuals towards populist political alternatives. In terms of economic specialization, a higher proportion of employment in mining, manufacturing and construction seems to be positively associated with populist vote. However, the evolution of job distribution across economic sectors shows a very weak association with voting for a populist party. This suggests that, at a Nordic level, the stability of the sectoral composition of employment does not have any significant effect on the propensity to vote for a populist party.  
  • Household income and inequality: There seems to be a weakly positive and non-significant correlation between average equivalised household income in Nordic municipalities and voting for a populist party in national elections. Income inequality at municipal level does not seem to be an important factor either. However, the rolling average of the change rate in equivalized household income for the previous four years is negatively associated with higher populist vote share. This suggests that the evolution of household income, rather than the current level of economic affluence, influences populist voting.
Overall, the regression findings suggest that the causes and drivers of discontent expressed as voting for populist parties are multiple and complex. Territorial and demographic factors, labour market conditions and educational aspects significantly influence voters' preferences for populist parties in the Nordic Region. Additionally, green transition factors play a role in voting for populist parties. The model's reasonable predictive power of data collected at the municipal level supports the notion that all these dimensions are important determinants of discontent and populist voting in the Nordic Region.
Related project publication:
Tapia, C. et al. (forthcoming). Mapping the uneven landscape of Nordic discontent: From regional disparities to the green transition.