Density and the distribution of household car ownership

9 touko 2018
Matti Lindholm

This blog post presents analyses previously discussed in an earlier blog post written in Finnish. However, since there probably is wider interest on the topic, this post summarizes the results in English focusing on the relationship between household car ownership and density. This blog post also provides the results from the whole range of densities (graph 1), which was not included in the original post.

The relationship between household car ownership and density becomes evident when analyzed with detailed GIS data. The below graphs illustrate the relationship between combined population and job density and the distribution of households into three groups based on their car ownership status: Carless, one-car households and multi-car households (2 cars or more).

We used register-based 250×250 m grid dataset to get accurate interpretation of job and population density. Data is from the year 2015 and covers the area of 14 municipalities in the Helsinki region [1]. We calculated density using land-area of the grid cells and applied neighborhood method [2], which includes surrounding cells into calculation.  This way the variable represents density in the close proximity of each location. The household car ownership data is also from the 250×250 grid dataset. The data does not allow household level analysis, because the information is summed to each cell.

The first bar graph shows the overall distribution including the entire scale of densities.

First notion: Major changes in the distribution occur in the densities below 100 population+jobs/ha.

In the second graph we zoomed into the densities below 100 population+jobs/ha and changed the visualization. The graph shows the observed change in the distribution against one unit change in the density variable.

Second notion: When the density exceeds 50 population+jobs/ha carless households are the most common household type. In densities greater than 63-66 population+jobs/ha over half of the households are carless and the share of the multi-car households drops below 10%.

Third notion: The share of carless households exceeds multi-car households in the density range of 21-26 population+jobs/ha. In the urban structure of Helsinki region this can be regarded as a threshold density for proper public transport supply and walking/bicycle infrastructure.

In the densities below approximately 7 population+jobs/ha multi-car households are the majority, but still even in the most sparse areas about 10% of the households are carless.

This analysis illustrates a clear connection based on accurate data that is not tied to predefined administrative or statistical areas. Analysis does not consider other evident factors influencing car ownership like household income or household structure among others. However, these other factors are often connected to density. For example, household size undoubtedly explains car ownership of households and households of different size have different spatial distribution in urban region.

We utilize this analysis as a part of the research to determine threshold values to identify and delineate the areas of three urban fabrics: walking, transit and automobile fabrics. Analysis has been conducted in two projects: BEMINE and Urban fabric analysis project for Helsinki Region MAL 2019 Plan.

Ville Helminen, Finnish Environment Institute, SYKE

Cover photo: Riku Lumiaro

¹ Helsinki region municipalities: Helsinki, Espoo, Vantaa, Kauniainen, Hyvinkää, Järvenpää, Kerava, Kirkkonummi, Nurmijärvi, Sipoo, Tuusula, Vihti, Mäntsälä and Pornainen

² Read more about neighborhood method (focal statistics tool applied here) at:


Sähköpostiosoitettasi ei julkaista. Pakolliset kentät on merkitty *