Over the past week, I’ve managed to attain the Portland tax lot files, which hold data on the county’s assessment of land value. After clipping the data to Portland city boundaries, I divided the data on land value by tax lot by area. I further processed this data by removing tax lots associated with parks, cemeteries, government-owned land, and condominiums (as the city apparently registers all condos as having zero land and building value).
Land value per square foot is generally, as basic Urban Economics predicts, highest in the downtown core and descending away from the center. However, beyond simple proximity to downtown, land value appears significantly influenced by both urban form—Cully, pictured right as the pocket of “Eastern Neighborhoods” in the midst of Inner Neighborhoods and bordering the Industrial District in Portland’s designation of its urban form, has noticeable cheaper land values than the neighborhoods to its immediate west and south. Moreover, one can see small areas of relatively high land value both near stops along the Blue Line to Gresham in the neighborhoods to the east of I-205 and along frequent bus routes on 82nd Avenue and Barbur Boulevard. Finally, while there are obviously other factors contributing to the relative value of Inner Eastside neighborhoods, it’s also worth noting the spatial correlation between the robust transit network there and the high value per square foot.
To process this data further, I joined the tax lots by location with my prior block group shapefile. This produced an mean land value by census block. I consider this a reasonable compromise between geographical granularity and ease of working with the data. From here, I was able to create a centroid file of all the block groups and then measure the minimum distance between those block groups and a frequent transit stop. I was then able to create the scatterplot shown at the left. When fitted with a negative power trend line, this relationship has by far the greatest r-squared of any data I’ve investigated so far—0.29!
One major flaw of the tax lot shapefile data is its lack of historical information on land value—data which Multanomah County clearly maintains, considering the presence of land value data going back to 1997 in PortlandMaps.com. I may still need to trek down to the county assessor’s office to get that full information. Nevertheless, there is information on land values going back to 2013, mapped below at the census block level. As we can see, land is generally appreciating in value in areas in which land values were already relatively high, notably in the central city and around Alberta in North Portland. Overall, this data has significant advantages in being able to represent appreciation in land values downtown and in the Pearl District, where the dynamics of gentrification aren’t manifested in a change in residential socio-economic status or owner-occupied housing value.
During our meeting last week, Jim Proctor raised the issue that I hadn’t yet designated a formal focused research question, despite the amount of data I’d been sifting through. To address that, I’ve come up with a tentative research question: How does transit influence the landscape of gentrification in Portland? My interim framework for addressing this admittedly broad question is to consider transit as a partial driver of underlying land values, by providing commuting access to downtown; as an indicator of a walkable, prewar urban form; and as a justifier for upzoning and urban redevelopment, particularly in the context of the myriad urban corridors/villages designations which Portland uses to plan regional growth.
So far, I think my mapping work has provided some tentative, though not statistically definitive, evidence that transit is associated with higher land values and gentrification. I hope to develop this evidence further with more historical data on land value appreciation. Regardless, this evidence will need to be bolstered with a theoretical framework and reference to prior case studies, considering the time constraints preventing me from developing a comprehensive model of urban land values. On the second point, quantifying urban form is not exactly straightforward, though there are a couple possible approaches. Mapping the age of existing buildings and integrating this with historical street maps provides a good starting place. Additionally, I could make some sort of street connectivity/intersection density index. Finally, to address the issue of regional planning, I intend to do some sort of content analysis on regional plans and revisions, highlighting how transit is a definitional property of urban village designations, investigating the zoning changes which accompany those plans, and analyzing how zoning is correlated with land value.
I also attended another OPAL research meeting over the past week. The content of the meeting was roughly the same as the preceding meeting. OPAL is gearing up for a press conference this Friday on its upcoming fare structure campaign, with the emphasis remaining on advocating for a low income fare, multi-rider cards, and addressing HOP retail deserts. They also discussed the issue of Trimet’s position on the monthly $100 e-fare cap, which the agency is apparently representing as a $1 million cost for the benefit of low-income riders. The organizers took issue with this, speculating that most of the expected cost would be from people with a current monthly pass would don’t quite use transit twenty or more days per month, not from people currently paying over $100 in bus rides per month. They also are planning to do a literal bean-counting exercise next week of Trimet’s proposed budget, painstakingly allocating 1,100 beans, representing $1 million each, to help investigate possible bloat in the proposed projects or capital projects line items.