Minneapolis officials were already unhappy when they saw the first list of proposed cuts to the Southwest Light Rail Transit project this summer. As part of an effort to slash the budget for the $1.77 billion extension of the Green Line, three stations located in the city were on the chopping block.
But when those officials saw the revised ridership forecasts for one of those stations — numbers that were significantly lower than previous estimates — they went from unhappy to angry. How could estimates drop from around 1,600 riders per day at the Royalston Station to less than 200? Were the ridership forecasts being manipulated to support the proposed cuts?
“The list of potential cuts, like the process to develop it, is driven not by objective criteria — since none were ever identified — but by politics,” Mayor Betsy Hodges and City Council Member Kevin Reich wrote to the Metropolitan Council in June.
Metropolitan Council staff disagreed, saying the changes were technical, not political. The lower estimate did, in fact, undercount riders from area bus routes and were added back in after the city pointed out the error.
But other changes were due to more complicated requirements. Specifically, that once another light rail extension — this one to the northwest suburbs — became an official part of the regional transportation plan, the effect on rider behavior had to be factored in. In other words, the projections had to account for the planned Bottineau Line possibly intercepting some bus riders from North Minneapolis before they would ever reach the Royalston station.
When Royalston and the other Minneapolis stations remained in the revised alignment, the issue of ridership was set aside. But it provided a glimpse into a perennial conflict in the planning of big and expensive transportation projects: How do public officials figure out how many people will use them? And do the numbers reflect reality — or simply a desire to fulfill the wishes of the project’s supporters?
The Pickrell Effect
Since 1964, when federal money became available under the Urban Mass Transit Act, the federal government has required local agencies to estimate the costs and other effects of mass transit lines. Since there are more projects than dollars, competition among urban areas is supposed to be decided on two basic criteria: impact and cost effectiveness. Complex modeling forecasting project costs, ridership, cost effectiveness and economic development is then used to assign ratings to the competing projects.
In a seminal 1992 study of the effects of the federal funding system, however, transportation economist Don H. Pickrell found that cities favored expensive capital projects over improved bus systems, largely because that was where the money was. The funding system had produced an incentive to overestimate ridership and underestimate costs. And why not: local agencies weren’t held accountable for the accuracy of their forecasts.
“You had grade inflation,” said David Levinson, the CTS Chair in Transportation Planning at the University of Minnesota. “Forecasts were being abused to achieve a particular outcome because there was a lot of money at stake.”
The so-called Pickrell Effect eventually led to both more scrutiny by the the Federal Transit Administration (FTA) and a measureable improvement in forecasts, with mixed results.
In 2003, an FTA-sponsored study looked at 19 mass transit projects completed between 1990 and 2003. It found that “ridership forecasts … improved since the 1990 study, with a number of projects’ actual ridership close to and even higher than predicted.” But the study also found such success to be relative. “All told,” the agency concluded, “eight of the 19 projects included in that study either achieved or had a good chance of coming within a reasonable range (plus or minus 20 percent)” of initial forecasts. The other 11, though, remained well below their forecasts.
Levinson agrees that ridership forecasting is better today than it was. “There are probably still a few places you could stick your thumb on the scale, but not as many,” he said.
“They are better, without doubt, because the grossest abuses have been exposed and people don’t want to be embarrassed by having that be done again,” said Martin Wachs, professor emeritus in urban and regional planning at UCLA. “The boundaries are more demanding today than they were a decade or two ago.”
‘Ridership is our bread and butter’
Wachs was the chair of a special committee that looked into forecasting practices for the Transportation Research Board of the National Academies in 2007. He thinks the current tools and methods of forecasting are sound, but added: “I find a lot of questions to be asked about particular circumstances when the modelers are under pressure by clients to support conclusions already reached for political reasons.”
The variables that can produce significant changes in final ridership forecasts include demographic information, economic growth, gas prices, regional planning, and surveys on how people actually travel.
“What is really important about forecasting is assumptions,” Wachs said. “Forecasts are more dependent on assumptions than they are the methods and the data. … [and] if you change the assumptions, you can get the outcomes to change dramatically.” And by the time assumptions become reality, once the project is up and running, it is too late, he added. “By then, the assumptions have done their job.”
The FTA rules and guidelines also have the ability to produce underestimates of ridership, something that isn’t exactly unwelcome for transit agencies. “Agencies will use high forecasts to sell the project, then lower the forecasts so they can exceed that forecast when it opens,” Levison said.
Met Council has seen its ridership exceed its forecasts on both of its existing light rail lines. When the first environmental impact statement was done for the Blue Line, in 1985, ridership was estimated at 37,000 average daily riders. But when a revised environmental assessment was done in 1999, ridership projections had fallen to 24,600 by the year 2020. Yet just two years after its 2004 opening, the line was exceeding those levels. The most-current estimate of average daily riders for the Blue Line: 34,584.
Mark Fuhrmann, the deputy general manager for Metro Transit who has worked on the Blue Line, the Green Line and now on SWLRT, said one explanation for the ridership jump is that the Twin Cities did not experience a marked drop in weekend ridership that occurs in other cities. Rather than a 50 percent drop on Saturdays and a 60-70 percent drop on Sundays, the Blue Line has seen Saturday traffic comparable to weekdays. Meanwhile, the Green Line is now at 45,644 average daily riders, or about 5,000 more than its forecast was — for 2030.
“Ridership is our bread and butter, it is why we make we make these large, significant and beneficial investments,” Fuhrmann said. But Ridership estimates also flow into different measurements used to convince the FTA to approve funding. More riders boosts the environmental benefits of a line, for example. In addition, more riders gives a project higher marks for congestion relief and mobility options for residents, Fuhrmann said.
And these days, the FTA does care whether ridership is on track, even after the project is running. It now require agencies to produce Before and After reports that compare forecasted ridership and budgets with actuals, Fuhrmann said. “Whenever Chair [Adam] Duininck and I talk with the FTA administrator, we always start out with ‘how’s the ridership doing.’”
But it’s also true that staff and political supporters of light rail like to brag about ridership numbers that exceeded forecasts. The Met Council sent out press releases in August when both the Blue and Green lines exceeded 1 million riders. And again in September when the Green Line set a monthly ridership record of 1.2 million.
“So sometimes we like to pound our chests a little,” Fuhrmann said. To those who think the numbers are manipulated to get a winning result, he said: “The reality is FTA carefully reviews and vets our ridership forecasts so it is based on the most-current input data.”
So how are the numbers actually calculated?
Methods for estimating how proposed transportation systems would be used first emerged in the 1950s, amid the massive effort to build the Interstate Highway System. Back then, the purpose was to decide how many lanes were needed for specific segments of the freeway.
And while the methods have evolved over the last 60 years, they still use the same types of information. Mark Filipi, who conducts transportation forecasting for the Met Council, said most of the work today is based on what’s called the Travel Behavior Inventory: a battery of surveys conducted once a decade to measure where people in the region go; when they go; how often; and by which modes of travel. Some of the information comes from travel diaries kept by residents (though some cities now use smartphone apps that record trips more accurately).
Like many agencies, and as recommended by Martin Wachs’ National Academies committee, the Met Council is working to improve the forecasting by doing surveys more frequently than once a decade, and by moving to what is called activity-based models.
Under that method, forecasters look at how people actually live and travel. Rather than count a trip from home to work as one trip and a trip from home to school as another, for example, the new method recognizes that people might go from home to work to lunch to the dentist and then to home. “We’re modeling what people actually do,” Filipi said.
Levinson, who is one of the experts conducting a peer review of the new models (which won’t be in place until after SWLRT is funded), agrees, saying the activity-based method more accurately capture how people actually use transit. “You can’t choose a car for going from work to lunch if you didn’t take a car to work,” he said.
The FTA wants estimates of average weekday riders for the end of the first year of service, and then for a future date — 2040 in the case of the SWLRT project. For the estimate, the FTA does not allow projections for weekends, evening and special events like sports.
The weekday ridership forecast also can’t include a project’s impact on future redevelopment, though if development is expected whether the light rail line is built or not, it can be counted. While the area around the Royalston Station has high development potential, for instance, riders that might be produced from that development can’t yet be included in ridership forecasts.
All of that modeling information, as well as cost estimates, local matching funds, the project’s impact on land use, economic development, and affordable housing and its overall cost-effectiveness goes into a rating by the FTA. Those ratings are used to distribute federal mass transit funds among competing projects around the nation.
During the meetings of local government officials who advise the Met Council on SWLRT, there was a lot of concern about how proposed cuts would affect both ridership and economic development projections. If cuts would reduce daily ridership projections too much, after all, the project might fall out of the running for a hoped-for 50 percent federal contribution. And so, with each proposed cut, the computer models were used to determine how many fewer riders would be attracted. For example, cutting the station at 21st Street would have cost the system 1,660 riders. Cutting a parking facility at Southwest Station by 165 spaces would have come at a loss of 250 riders.
In the end, the bulk of the cuts came in Eden Prairie, where the terminus was moved from Mitchell Road to Southwest Station and the Eden Prairie Center Station was scrapped. Along with some reduction in parking at various stations, the projected ridership for SWLRT is now 34,074.
When the new budget and new alignment was approved, project staff expressed confidence that the project will still compete well for the next round of funding set to be announced by the Obama Administration in February. That would keep it on schedule to win its full funding grant agreement (FFGA) and federal matching money before Obama leaves office.
A political process?
For all the sophisticated modeling, though, both Wachs and Levinson concede that it’s impossible to separate the technical aspects of forecasting from the politics. Levinson, for example, said forecasts of costs and ridership are not used to determine which localities win federal money but to justify those decisions.
“If [Rep. Martin] Sabo and [Rep. James] Oberstar are well placed, their projects will get funded because it’s not just on cost effectiveness or ridership,” Levinson said of former congressmen who backed the region’s first light rail line.
“Certainly the FTA bureaucrats see their job as being objective and their purpose as being honest,” Wachs said. “But they are strongly influenced by politically elected officials whose goal is to bring money to their project rather than see it go to another state.”
Local agencies should have an interest in not overbuilding or not building at all if the ridership isn’t there. And general managers who have to operate those systems do worry about unsustainable projects, Wachs said. But the elected officials are more interested in “ribbon-cutting” and figure that by the time the operating shortfalls appear, they’ll be out of the picture.
Fuhrman has been involved in building new rail lines for 30 years, first in Washington, D.C. and then locally. He said his experience with the FTA tells him the analysis is vital to winning funding. “The FTA doesn’t want frou-frou projects that they are going to fund and then become a miserable failure,” Fuhrmann said.
Fuhrmann acknowledged that politics do play a role in the appropriation process — from the president deciding which projects to fund to members of Congress making final decisions. “But the project metrics and the project justification had to stand on its own merit. Once we made that compelling case to FTA that [Southwest LRT] was truly a project worth federal investment, our congressional delegation would then help with the annual appropriation process.
The region received $334 million for the Blue Line, $478 million for the Green Line and is hoping for funding in the $850 million range for SWLRT.
“It truly is a competition,” Fuhrmann said. “And what I put under my pillow every night is a little map to see how we’re doing against other projects in other regions. And we’re doing pretty well — we have two projects in project development and there are a lot of states that don’t have any.”