Mapping attractive urban areas

Instead of asking what parameters that are important for the population, the “Mapping urban attractivity” project takes a quantitative and geographical approach to these questions, testing a methodology which at a European level might be used as a supplement to the survey. The aim is to aid the survey in “which questions to ask?”, “where should we ask them?” and interpretation of results.

The probed methodology uses house prices as a proxy for attractivity, as they are a reflection on a kind of attractivity. Both “Total sales prices” and “Price per m2” are explored, with all point georeferenced dwelling sales throughout a year as the data which we wish to explain. We have focused on Norway’s largest cities, using Ordinary Least Square Regression analysis tools to correlate price and place with factors such as m2 floor space, mean income or education level of the adult population in a buffer zone around each dwelling, or other types of variables. The variable types tested for are: 1. Intrinsic characteristics of a dwelling, 2. Population characteristics 3. Employment, 4. Distance to geographic entities, 5. Distance to buildings, 6. Environmental.

A focus of the project is variation within cities, making comparisons between cities on to which degree our variables explain variation. This brings city intrinsic differences into an equation to a large degree lacking in the “Perception survey”. Our approach touches therefore into whether city planners have been successful in distributing important city services and amenities in an even fashion.

Dwelling-intrinsic characteristics such as m2 floor space clearly count for a vast amount of price variation. There is however variation in how true this is throughout different cities, leaving more explanatory power to non-dwelling-intrinsic characteristics. Of these variables, our findings are that “Education level” and “Household income” are the best indicators of variation in neighbourhood attractivity. Also “urban pull” variables can count heavily, with variables on distances to town centre, restaurants and higher education facilities.

The strengths of these correlations vary between the cities, playing in to a general picture that capital and largest city Oslo is the most socioeconomically divided of Norway’s largest cities. However, results show that size isn’t all, history and socioeconomic issues clearly matter. The potential in calculating these same correlations for a different year, or for creating a time series, is apparent. Results would pick up on nuances of correlation values in and between Norway’s cities.

The OLS-analysis produces results on which combination of variables that best correlate to our house price attractivity variable, producing coefficients on the strength of each variable. In the project, we reuse this output to produce predicted attractivity datasets, generating 500m X 500m attractivity grids for each city, scaling from “least attractive” to “most attractive”. The variation in these predictions are by definition an expression of variation in attractivity. Potential lies in locking these coefficients and creating a time series, mapping expected changes in attractivity, correlating this again to observed house price changes.

A lot of common sense and logic can be read from the resulting correlations. The conclusion of the project group is that findings and methodology definitely have a potential as supplement to the Europe’s “Perception survey”, making more out of time and resources invested in this important undertaking.