Earlier today, Project Connect released a methodology memo (PDF version) providing an overview of their approach to estimating ridership. The memo raises three major concerns, which I describe below.
1. Ridership estimates use population growth assumptions unbound from empirical reality
The Project Connect team continues to model population growth based on where our planning would politically prefer growth happen instead of where growth empirically happens. An excerpt from the memo follows:
We do not know if the team’s parcel attractiveness approach actually predicts observed growth. Proponents of this route have faith that rail can “shape” growth, but that is a concept with only anecdotes and ideology to support it. If the model was validated, Project Connect would have released an analysis of the model’s past accuracy in estimating parcel changes between 1990 and 2010 and/or 2000 and 2010. I suspect that such a validation will not be undertaken because it would prove there’s no meaningful contribution from the parcel attributes chosen by the modeling team that would satisfy the unrealistic projected growth for Highland.
2. The methodology triple counts ‘special’ markets
The actual forecast daily ridership for 2030 is not the much publicly discussed 18,000 figure but 15,580. To boost this total, the methodology arbitrarily tacks on 25 days to the standard FTA 300 day weekday and weekend service plan (why not 20 days or 30 days?). The reasoning is the special markets served (i.e. special events). This boosts the annual ridership from 4.674 million to 5.06 million.
A second events boost is tacked on based on observed higher uses of MetroRail, which gets annual ridership to 5.20 million or 16,000 daily riders. It’s not clear why those wouldn’t be included in the rationale for the boost from 300 to 325. Finally, the document indicates that the consultants believe an additional 2,000 riders are not being captured from ACC and the medical district (why not 1,000 or 3,000?). Obviously the demographic allocation tool is modeling population and employment growth in those areas.
And that’s how we get to 18,000. The use of the lower daily ridership estimate worsens the productivity of an already suspect route; understandably, proponents would prefer to avoid that number. We all should probably stop citing the 18,000 figure and stick to the actual model result: 15,580.
3. The medical school is not a major ridership source
Much of the public discourse about the advantages of this route are tied to unleashing an innovation district around the medical school. The daily boardings (both south- and northbound) at the medical school are estimated at 860 in 2030. That’s 6% of the 15,580 daily ridership. This reaffirms the importance of evaluating the service as a mobility investment to the many other already-existing employment destinations downtown. I am excited about an innovation district, but at this time, it seems serving it with high frequency bus or Bus Rapid Transit is a better use of scarce transit dollars.
On Point 1, you have recently been using the word “faith” to diminish the validity of rail as a tool to direct growth and density to particular areas. This sets up a false sense that your opinion is based on science while those in opposition are just feel that their unsubstantiated position is correct. In fact, numerous academic studies provide context to examine this issue without resorting to without resorting to a fact vs faith argument.
Let’s take a closer look.
Your statement: “We do not know if the team’s parcel attractiveness approach actually predicts observed growth. Proponents of this route have faith that rail can ‘shape’ growth, but that is a concept with only anecdotes and ideology to support it.”
Transit Cooperative Research Program (TCRP) report 1995 “urban rail transit investments rarely ‘create’ new growth, but more typically redistribute growth that would have taken place without the investment.” [My interpretation: growth can be ‘shaped’ or ‘redistributed’ by rail placement]
(Knight and Trygg 1977) “It seems from the evidence available that rapid transit improvements can provide an impetus toward generation of new nearby development. However, transit alone seems no longer enough to insure such development”.
“Market forces – primarily the availability of land for development – may significantly affect the location and degree of development above and beyond other influences…” The availability of “suitable, assemblable land” is an obvious prerequisite for development.
“…local government policies are important factors affecting development, with transit being an important but not sufficient condition for such development.”
“When the general character of the area is favorable toward development… transit may further enhance such development.”
(Vesalli 1996) “almost exclusively, transit system’s impacts on land use are limited to rapidly growing regions with a healthy underlying demand for high-density development”.
“these land use impacts of transit are not accidental, nor automatic… the only substantial impacts of transit on land use are those that have been planned, and this planning entails a substantial investment of public sector resources and coordination”
(Qingyun Shen 2013) “densification effects are more likely to be observed near a new urban rail transit station/line, if it is part of an established rail network, located in moderate-income neighborhoods with a low poverty rate, not too far away from downtown, with a pre-existing pattern of compact development and not dominated by single-family housing units”
Most studies on this topic look at the effects on density subsequent to the arrival of transit not its announcement.
Vessalli’s work is great. It’s a meta-study. I don’t think your excerpts reflect the profound uncertainty captured in the actual paper. The excerpts you are quoting I believe are from sections discussing ‘themes’. It’s not an actual crunching of data, that’s why the language is a bit platitudinous. I’d actually count this paper as supporting my point-of-view and it’s one I’ve encountered before. Here’s a link
I haven’t read the TCRP report. I haven’t read Knight and Trygg myself, but it’s oft-cited, but hardly as a defense of certainty around shaping. I’d suspect that if I could get access to a copy (do you have a link to an open version) that it would be like all other rail and land use studies – a big list of potential factors and with a cautious endorsement of their causal power based on some limited sample.
Shen’s is fantastic. But again, I think we are reading them differently. Her summary of the literature flatly points out the state of contradictory conclusions and lack of consensus around the impacts of land use from transit. I found her methodology (4 mini studies of LA, Chicago, Denver, and DC) fascinating. Most interesting is the variation within those four communities. Your excerpt comes from her pointers to planners and policymakers, which is a grab-bag of sorts based on the noisy findings from the four communities studied. This is what she writes in the concluding section on the densification relationship:
“Just as previous studies suggest, there is no consistent finding on the direct impact of a new rail transit system on density change in the surrounding neighborhoods. Among the four case regions this study selects, only the Green Line in Washington, D.C. shows statistically significant direct densification effect on population densification. The Orange Line in Chicago seems marginally significant in imposing population densification effect. In the other two case regions, Denver and Los Angeles, the new investments on urban rail transit systems are missing evidence on their impacts on directing density increases. ”
It’s a great document. Again, I’d count it as upholding my POV. Here’s the link.
I don’t think that Highland or the area between it and UT would pass the Shen quality test, by the way. The density is not already there and it’s far away from downtown. In the high r-squared mini-studies that latter one was pretty powerful. It’s why she cut out Denver’s downtown in her modeling.
I don’t think there’s a social science consensus or even clarity on light rail shaping development in the run-up to or immediately after its launch. There’s not even consensus on the long-term effects. That’s why proponents rely of anecdotes from their travels (New York!), misuse of case studies (Denver!), or flat out ideology (planning works!).
I don’t have a link to an online copy of Knight and Trygg. At several hundred pages, it’s really more of a book than a paper. Here’s a good executive summary:
The 2nd section on “Growth Focusing” should be read in the context of the Highland site and (in my interpretation) shows how favorable that site is for successful densification effects.
I agree with you Julio that there is not a consensus on the shaping effects of rail, but I think there are better ways to discuss it than in terms of “faith” – that was the point I was trying to make.
Shen points out the lack of academic consensus on the issue as the primary motivation for her paper but she continues on to propose several factors that can contribute to densification around a new rail stop. We are definitely reading her conclusions differently. In the quote you provide “there is no consistent finding on the direct impact of a new rail transit system on density change in the surrounding neighborhoods”, she is saying there is no DIRECT impact – that is that simply adding a rail line does not guarantee densification. She spends the majority of her conclusion after that quote detailing the other conditions that can be combined with a new rail line to promote densification. We cannot just ignore that as it is the bulk of the paper.
Section 3 of her conclusion deals specifically with neighborhood conditions. “This study finds four factors that can promote or hinder the densification effects of urban rail transit: the remoteness of the neighborhood, the pre-transit density, the income level, and the power of single-family housing owners.”
Taking these factors one by one.
the remoteness of the neighborhood – Shen classifies distances 6 miles as suburban. Google maps gives a distance for US 290 at Airport to I-35 at 12th Street as 3.9 miles. So Highland is either at the top limit of Urban or bottom limit of Urban Fringe depending on where you measure.
the pre-transit density – Here’s a good map of residential density by census tract. Note the trend to a dense yellow along the proposed Highland route – http://spatialdatanerds.com/wp-content/uploads/2013/11/Block-Boundaries-Austin.jpg
the income level – Mixed results on this one. Along St Johns and in the triangle formed east of I-35 by 183 and 290, there is a lot of poverty. However to the south of Airport it’s mostly middle income.
the power of single-family housing owners – With the exception of the small area around Hancock Golf Course, this route is not known for powerful neighborhood associations (in stark contrast to some other choices for a northern route).
Because of all the above, I have to continue to disagree with your assessment of Highland Mall’s potential for densification not on faith, but using available academic studies.
Your using her summary as if it were a Pronovost checklist, which it isn’t. Shen provides coefficients – plug those in. The DC mini study had the strongest effects, but IIRC there was population LOSS in the rail nodes, just less loss than the control. You’re interpreting ‘denser than the control’ as ‘denser than the starting point’. And again, these effects are expected a decade after launch; the growth modeling for Highland requires immediate, dramatic changes in the growth rate nearly a decade before service operation. I’m not sure you heard this critical temporal subtlety in my previous response.
Sorry, that line above “Shen classifies distances 6 miles as suburban” should read “Shen classifies distances less than 4 miles from downtown as urban, 4 to 6 miles as urban fringe, and greater than 6 miles as suburban.”
“Finally, the document indicates that the consultants believe an additional 2,000 riders are not being captured from ACC and the medical district (why not 1,000 or 3,000?). Obviously the demographic allocation tool is modeling population and employment growth in those areas.”
The CAMPO model predates the plans for both ACC Highland and the medical school. In the case of ACC Highland, that’s 20,000 students at buildout that are missing from the model. Because people now know that ACC is under construction, consultants seem to be compensating for this new information by using “special generators”. http://youtu.be/oHbz2V5zVTI?t=18m46s
Pingback: Let’s Go do rail like Houston! | Keep Austin Wonky
Pingback: Rail Risks | Keep Austin Wonky
Pingback: Pro-Transit, Anti-Proposition 1 | Keep Austin Wonky
Pingback: Minimum viable product won't work with Austin urban rail bond - Local Mustard