For this week in my independent study, I delved deep into GIS, mapping out transit mode shares, lengths of transit and car commutes, the change in percent of people with a Bachelor’s degree or higher, and percent increase in rent and home value since 2000, all measured at the level of the census block. Census blocks provide noticeable finer-grained data, allowing for a more careful analysis of the possible socioeconomic and modal effects of transit in Portland. I began with maps on transit’s mode share in Portland and the percent of drivers and transit riders whose commute is under thirty minutes. I was expecting to find long transit times for East Portlanders, but was surprised to find that few areas of the city have a reliable transit commute under 30 minutes. Even for most block groups in the Center City, fewer than 60% of transit riders have a commute as short as the vast majority of drivers in most sections of the city.
The spatial modal data provides modest evidence that light rail leads to more people using transit, with the block groups surrounding rail stations in East Portland having noticeably higher rates of transit use than those around frequent bus service. The stations around 60th St also seemingly encourage higher rates of transit use than would be expected based on a comparison to the blocks further from MAX, though a comparison with the rail mode share map surprisingly reveals that the bulk of this ridership involves buses. The yellow line through North Portland appears substantially less effective at attracting total above the baseline seen in that district of the city. The green line along 205 to Clackamas also appears to attract higher ridership than one might expect based on land use patterns, but the rail mode share map reveals that this ridership is primarily on a combination of more direct downtown-bound buses or simply the 72 running on 82nd. The City Center features, as one would expect, relatively high transit use, with rail use being particularly concentrated along the light rail and streetcar lines.
To get the percent increase in rent and home values since 2000 required a rather painful process of manual alterations. The U.S. Census modestly revises the boundaries of both census tracts and census block groups during every decennial census, to reflect population changes in the interim. With census tracts, I was able to find a wonderful resource updating the data to present boundaries, but I found no such luck with the blocks. While most block groups were unchanged, approximately twenty had been either split apart or had been slightly rearranged and recategorized. To find where the geographies had to be adjusted, I first pored through the spreadsheet, adjusting the block group data to line up as best it could and moving mismatched cells onto separate lines. I then mapped data from the 2000 census with my 2014 block geographies and searched for gaps. After finding these blank spots on the map, I queried both the 2014 and 2000 block group geographies to find how the numbering and boundaries had changed. In the case of minor reorganizations of boundaries, on the order of a block or so, I chose to simply pair the two comparable geographies in my spreadsheet. For split tracts, I elected to assign the same 2000 value to both block groups to make it backwards compatible. While obviously not ideal, I made the assumption that variation within a pre-existing block would be relatively insignificant when examining a city as a whole.
Below are four maps on the absolute change in percent of people with a Bachelor’s degree or higher, the percent of residential structures built since 2000, and similar median gross rent and median home value data I produced last week, but made for the census block data. I’ve included the census tract maps for comparison.
I’m a little suspicious of the gross rent map, based on both how checkerboarded the spatial patterns are and how dissimilar it is to the census tract-based map I made last week. While it is possible that rents have genuinely risen across Portland in such an uneven manner (and it would be interesting to explore the relationship between initial relative rents and subsequent rises), I remain skeptical. The samples of this data are small and the margins of error within the data are consequently quite large, typically on order of one-third of the total rent and sometimes reaching even higher. Many block groups were also missing 2014 data and I elected to fill in those blank spaces with the median gross rents for their census tract. Combining or referencing this information with the 2010 Census would be one way to minimize the potential margin of error and scalar inadequacy of this method, though it is worth noting that I saw massive variances in the 2014 rent between census blocks within the same tract (in several cases near $1,000).
The map of housing value by census block is much more in accordance with my earlier census tract map, reflecting the general pattern of a rapidly appreciating Inner Portland housing market and an East Portland market that is actively depreciating in real terms. The housing value dataset was also far more complete than the rental assessments. Determining a relationship between transit and rising housing prices is difficult to determine at this scale—the effects of rapid, racialized gentrification of long-neglected North/Northeast Portland and, more generally, a tightly constrained Inner Portland housing market (see the image on housing units built since 2000) are far more apparent, especially considering the placement of MAX lines in the vicinity of highways or directly beside them. At the very least, there is a clear spatial correlation between an increase in degree-holders and changes in home values, with a modest relationship between transit in general and these socio-economic indicators of gentrification (though both could well be influenced primarily by proximity to downtown). Moreover, outside the Pearl District and South Waterfront, there is a pretty clear inverse relationship between recent construction and both these socio-economic indicators and transit use.
Moving forward, I intend to map the relationship between the 2000 median value of homes in census blocks and their subsequent appreciation. I also will attempt to do some statistical analysis of these visual correlations. My post next week will integrate these spatial findings with both Portland’s development plan and broader theories on gentrification. I’ve also come up with an idea for potentially measuring the rent gap in Portland, after digging through a map of all the tax lots in the city—using vacant land and surface parking lots to see the amount of capitalized value that a current building on a comparable lot adds and comparing that value against a newly developed lot. At any rate, I should read carefully through some of the existing literature on quantifying “potential land value,” an inherently difficult endeavor.
Lastly, I didn’t end up meeting with OPAL this week—their system of communicating about upcoming events seems rather uncoordinated, with the online calendar only bearing a rough resemblance to reality. They cancelled the action planning meeting and chose to instead do a workshop on public testimony, in preparation for a TriMet board meeting. I will, however, be attending their general meeting this Friday and doing some bus organizing this Saturday, so I should have plenty of reflections on OPAL for next week.
Leave a Reply