Valuing walkability: New evidence from computer vision methods. (December 2019)

• Record Type: Journal Article
• Authors: Yencha, Christopher
• Is Part Of: Transportation research. Volume 130(2019)
• Page Start: 689
• Page End: 709
• Publication Date: 2019-12
• Subjects: C81 -- R21 -- R40; Walkability -- Land use -- Residential real estate -- Hedonic regression -- Computer vision; Transportation -- Research -- Periodicals; 388.011
• DOI: 10.1016/j.tra.2019.09.053
• Languages: English
• ISSNs: 0965-8564

Abstract

Highlights: Estimates the capitalized value of pedestrian infrastructure and amenity access. Supplements walkability measures with new data gathered from images. Demonstrates application of computer vision in gathering street-level data. Value of proximal amenities reliant on personal access to pedestrian infrastructure. Abstract: Walkability describes the efficiency and pleasure of walking in an area and is an aspect of urban design that has received much attention. Frequently used measures of walkability largely ignore the quality of nearby pedestrian pathing, such as sidewalks, in quantifying walkability. This paper expands upon the literature's understanding of walkability by supplementing current measures of walkability with data gathered from street-level images using computer vision techniques. Using hedonic methods and a sample of almost 60, 000 house transactions in Ohio, I find that nearby establishments are an amenity capitalized into home prices only when there also exists access to adequate pedestrian pathing, and that walkability measures derived from computer vision methods contain information not found in other commonly used measures of walkability. With cities increasingly pushing towards creating more walkable neighborhoods, this paper provides evidence that walkable space is indeed valued by residents.