DOT Finance

Lowering CEI Guidelines Will Limit the FTA’s Effectiveness in Choosing Projects

» New transit projects are absolutely necessary, but what’s most needed is an increase in funding, not a change in regulations.

In reviewing transit capital projects to fund with New Starts grant money, the Federal Transit Administration evaluates proposals from a variety of perspectives. Since 2005, it has placed an overwhelming focus on one criterion, requiring a medium “cost-effectiveness” rating, which values predicted overall travel time saved by commuters likely to use the new service.

Transit agencies pushing big new line expansions have argued that this specific guideline is too strict, forcing unreasonable cutbacks in project costs and valuing suburban, long-distance lines over slow-moving urban ones. Now two Democratic congressmen are hoping to overturn the policy. But their efforts may compromise the FTA’s ability to pick the nation’s most important new transportation programs.

It has been easy enough to pick on the Bush-era rule, since it put major barriers on new projects and shut down projects in areas as disperate as Miami and North Carolina’s Triangle. Many cities were only able to get their medium-low-rated corridors into the pipeline because of exceptions written into the law. Under the Bush Administration, no streetcar projects were funded directly with New Starts money. Other agencies that were successful in winning grants were forced to be cost-cut to keep their corridor budgets within the established guidelines.

But the biggest problems with U.S. transit funding have little to do with the cost-effectiveness rule. Despite the good motivations of those that would change the game — Representatives Peter DeFazio (D-OR) and Keith Ellison (D-MN) — Jim Oberstar (D-MN), the chair of the House Transportation and Infrastructure Committee, has yet to sign on to the idea.

There are good reasons for him not to do so. The FTA must sort through dozens of projects every year, choosing only a few to move forward to full funding under the New Starts program. Plenty of these corridors meet the medium cost-effectiveness ratings, and those that have been rejected because of their low cost-effectiveness arguably do not deserve federal funding, at least when they’re put up to comparison with other planned corridors around the country that would serve more people taking more arduous commutes. And the FTA must have an adequate mechanism to establish the value of each new line.

The solution advanced by Mr. DeFazio and Mr. Ellison would instead increase the importance of such characteristics as land use and local support, both of which are important but neither of which is vital enough to discriminate between similar projects in different cities. In other words, with a limited pot of funds, the FTA can only choose to invest in so much. It’s not as if it’s operating with an overflow of cash right now.

This is not to suggest that the New Starts process is perfect. Rather, the program could be seriously reformed to ensure an objective, equitable standard to determine which projects should be moved forward. Cost-effectiveness guidelines such as those the FTA uses will be an necessary element in that game.

In the long term, transit proponents must attempt to dramatically increase the size of that pot but hold off on preventing the FTA from making discriminating choices in the short term. Current efforts to do the former have been lacking — the House passed a jobs bill this week that funded roads at a three-to-one ratio compared to transit, exactly as had the stimulus. The Senate is currently planning a climate bill that might add an average of $1.6 billion a year over the next ten.

These are baby steps. According to the most recent Census estimates, the American population may grow to 458 million by 2050 from 308 million today. How will those people move around in the cities in which they’ll live? The amount we’re currently committing to transit won’t be nearly enough to fulfill those needs.

This means we’ve got to use the resources we do have even more judiciously. Attempting to deny the FTA the ability to take advantage of the important cost-effectiveness guidelines would be no help.

30 replies on “Lowering CEI Guidelines Will Limit the FTA’s Effectiveness in Choosing Projects”

I don’t think anyone is opposed to the concept of cost-effectiveness, but the way the FTA has defined it and applied it makes no sense whatsoever. They’ve taken a rather expansive definition of costs and an exceedingly narrow definition of effectiveness.

I think back to Minneapolis’ Central Corridor, where I’ve gotten the distinct impression that local jurisdictions would be willing to pay the extra incremental cost to get a tunnel through the Univ. of Minn. campus, but that cost would put them over the CEI hump. Why must that cost be included? Why can’t the Feds pony up for the basic line, and let the locals tack on as many add-ons as they like at their own expense?

Same thing with Dulles Rail – they’ve stripped the stations down to bare bones and cut other costs with a chainsaw – and these will have a real impact on how the line functions, yet their value isn’t captured in the twisted view of what constitutes cost-effectiveness.

There’s no doubt we need more money for these programs, but you yourself outlined the poor outcomes this rule creates with the Southwest corridor in Minneapolis.

All three of these things need to happen:

1. The CEI needs some serious redefinition
2. More money for transit new starts
3. Less emphasis placed on the CEI in the selection process.

How is this post consistent with your previous posts regarding the Minneapolis Southwest LRT line? CEI kept the line out of densely-populated, transit-using districts of Minneapolis. CEI props up the suburbs at the expense of the core city neighborhoods. It’s gotta go, or at least change substantially.

“Plenty of these corridors meet the medium cost-effectiveness ratings, and those that have been rejected because of their low cost-effectiveness arguably do not deserve federal funding, at least when they’re put up to comparison with other planned corridors around the country that would serve more people taking more arduous commutes.”

This is the part where you’re basically saying that transit lines between outer suburbs and the CBD are the ones that deserve the limited federal support. It’s the whole premise of your argument, and I don’t think it holds any water.

“Time saved vs driving” is, for many projects, a worthless metric. It’s useful if what your are trying to do is, as the lead suggests, design a system to serve suburban commuters facing a clogged freeway network; but for many types of trips, making things quicker than the car (which often won’t happen) is irrelevant.

It’s a framing issue–that suggests that transit is only useful or desirable when it can make life easier for motorists.

Many transit projects have nothing to do with motorists.

Other corridors that do quite well under the CEI system: New York’s Second Ave Subway, Seattle’s University Link, LA’s Expo Line, among others. This system does not only prioritize suburban links — many corridors that only serve urban centers also do quite well. In terms of the Southwest Corridor on Minneapolis, the primary problem there is poor ridership estimates by local authorities, who are unable to demonstrate the attractability of Uptown as a destination. If the ridership estimates had been more accurate, the CEI on the SW line would have gone up.

SAS and the Expo Line would get high ratings under any metric. The important part for a metric is not just to fund those obvious extensions, but in general keep costs per rider down. You want a metric that would defund the Gold Line and its Foothills extensions, and instead fund something like the Calgary LRT. A good rule of thumb is that cost per weekday rider should be not much above $10,000, and preferably far lower (cf. Calgary at $2,200).

Is cost per boarding really a good metric? It seems that cost per passenger-mile is a far better metric. Consider the choice between operating the following two bus lines:

Line 1 runs picks up 40 people in the suburbs, delivers them downtown 60 miles away with no intermediate stops in 60 minutes, picks up 10 reverse commuters and heads back to the suburb to make another run.

Line 2 operates at 12 mph meandering along the edge of downtown and an inner suburb, picking up a person every 2 minutes who spends 20 minutes on the bus traveling 4 miles. The bus never has more than 10 people on it.

Assuming that the operating costs per hour are equal (which is a good approximation), by Alon’s metric, running the nearly empty line 2 is a far better choice because it generates 30 boardings per hour vs. the 25 boardings per hour that line 1 (which is always carrying more people than line 2) generates. Using a passenger mile based metric line 1 is a clear winner that generates 1500 pass-mi/hour vs. the 12 pass-mi/hour that line 2 generates.

If you would prefer to make this more relevant to FRA new starts money then multiply ridership numbers by 50 and call line 1 a commuter rail line and line 2 a light rail line.

Winston, operating costs are not equal. Longer lines require more maintenance and more fuel. And no, they don’t tend to be busier – they have long sections through the suburbs where the trains run empty.

Doing urban rail with passenger-miles is just wrong, and creates boondoggles like BART to Livermore while deprecating transit that could work like Geary.

Why is a longer trip more important than a shorter one? It isn’t.

If a transit line is moving people between places they need to go, then it is serving a use for those people. Though going further out may net you more possible destinations, the utility of a trip does not increase with its length. And being able to make short trips of high utility is exactly what effective urban environments are about.

I wrote a really long rant but deleted it because this post made me so angry. Over the last 4 years I’ve seen the CEI force cities to plan crappy lines or make stupid decisions that I will never think it is a good idea to allow the CEI to rule over ALL other metrics. Many of the posters above have hit on the points that I was going to make in a ten paragraph comment.

The only thing I will say is that if you keep the CEI, you can say goodbye to good projects like the Central Corridor in the future. There have been many cities that didn’t even try to get into the process because they started planning lines not based on need, but on what they could get into the federal process. That is not a way to plan transit. In fact, the Hiawatha Line which was a wild success and spurred the planning for more transit lines was a medium low in CEI. So was Charlotte. So was the Orange Line BRT. So was… you get my point.

The Orange Line BRT was the best possible project considering the political circumstances in Los Angeles at the time of its construction. All modes of rail transit except a deep bore subway were banned along the right of way, otherwise an elevated extension of the Red Line or a LRT line could have (and would have) been built.


I don’t have it in front of me, but the final CEI for the Minneapolis Southwest LRT 3C and Sub-Alts was astronomically high ($40 range?). Even if they had done a better job with ridership estimates–and I think they were deliberately skewed to favor 3A–the CEI for 3C would’ve probably been too high to get New Starts funding. Now there is almost zero chance of Minneapolis’ densest neighborhoods receiving any major transit improvements in the foreseeable future.

I don’t think anyone’s saying cost-effectiveness is a bad measure at a theoretical level. But the CEI itself is a flawed metric. The numbers are easily manipulated and, as we have seen, can lead to some pretty pro-sprawl decision making in otherwise-progressive cities (like Minneapolis).

Alon obviously it’s not my favorite project but yes given the political realities of a law that didn’t allow rail above ground in the valley as Karl mentions it was the best they could do under the circumstances.


Which US rail projects in the past 10 years would have passed your $10K per weekday boarding “rule of thumb”? Given labor costs and other factors that the transit agency can hardly change by itself, I think that standard would mean never seing another inch of frequent urban rail transit built in most American cities.


Not to pile on, but I also would like to know about your $10K per weekday rider number. Unless I’m missing something, that would–as the poster above noted–preclude the building of almost any presently proposed urban rail transit project.

Furthermore, how do we create a culture of transit use when we won’t invest in projects that relate strongly to smarter land use? Projects like the Mpls SW LRT continue to reward expansion in distant suburbs where it’s simply too late to meaningfully adopt dense, sustainable land use policies.

Calgary Transit has a neat little table (page 9) of light rail project cost per rider, as of 2000. Calgary’s cost $2,400. Edmonton and Sacramento came in at just under $10,000; Denver’s and St. Louis’s didn’t go over by much. Since then ridership has increased, lowering the cost per rider on the lines. The LA Blue Line is now down to $11,000.

Labor costs aren’t what makes projects expensive. Calgary had lower per-km costs than most other cities, but not by much, and that came from acquiring ROW in advance; Calgary actually has very high labor costs. The reason it has such a high cost-effectiveness is that it gets high ridership, due to proper planning.

In other developed countries, where transit is planned well and gets high ridership, even subway projects can come cheap. To put things in perspective: Tokyo’s Oedo Line cost $16,000 per weekday rider, and that’s after cost overruns and a shortage of ridership. Light rail should be able to come at much lower.

Someone can correct me if I’m wrong, but I’m under the impression that the reason the ridership estimates were so out of whack from the smell test in Minneapolis’ Southwest Corridor because that’s the exact methodology spelled out by the FTA for their CEI numbers.

Ridership was excessively underestimated for the Twin Cities SW LRT because Hennepin County & HDR (consultant) refused to consider the multitude of bus-to-rail transfers and modal switches.

First, they don’t believe that potential riders in the city would walk even one block out of their way.
Second, they wouldn’t let the operating costs for the urban route (3C2) look better by scaling back parallel bus service in the city that gets downtown at a far slower speed, but would eliminate nearly all of the suburban express buses that today reach downtown faster.

The longer, urban route would have reached the first of seven downtown stations faster (3.5 minutes) than the shorter, railroad-corridor route would have. While 3C2 would have navigated downtown Minneapolis in a circuitous fashion.

For greater depth, I have an ever more exhausting explanation below.

There are so many other things that the Metropolitan Council and Hennepin County did over the last few years that insured the southwest light rail line would not serve the city:

Marquette Avenue, having been one of the ideal north-south arteries for light rail downtown, was rebuilt for two bus lanes. The bus project has great merit, but LRT would have been much better. Nicollet Mall simply wouldn’t have worked for multiple reason.

Lyndale & Hennepin, being the other ideal north-south artery, was shot down, supposedly because the Basilica didn’t want rail so close to the church. It’s on a hill, and has about 60 feet setback from the wide avenue LRT would have run in, and vibration concerns would not be nearly as bad as for the U of Mn, NPR, or the churches next to NPR in St. Paul.

Lyndale Avenue, from the Greenway to Hennepin Avenue bottleneck, is being rebuilt next year, and light rail on the surface or in a tunnel could have been coupled with this route. Lyndale is more ideal than Hennepin or Nicollet because there is little bus usage south Lake Street, unlike the other two. The biggest legit problem for Lyndale is that it runs above a highway tunnel and next to on-ramps. The biggest problem downtown would be connecting with LRT at 5th Street. If the Central Corridor had been routed through Dinkytown and across the Hennepin Avenue bridge, this would be a moot point. Too many leaders in the Twin Cities are uninterested in building a network.

They built the Midtown Greenway bike path too close to the center of the right of way for rail. This will also be a problem on the far east end if we ever build a streetcar, which was another one of the obstacles with the citizenry. People really believed that a streetcar would get built here, and they really thought it would be harder to construct it, end to end, if we didn’t build it half of it with this project.

The race card was played in such a sick and asinine way that I wish not to discuss it. If citizens and leaders in Eden Prairie really believed this line was going to effectively serve North Minneapolis, do you think they’d support it?


Some great points. As for the race card point, I always thought it was a bit disingenuous when you consider that more minorities would be served by going through Whittier and up Nicollet than with the Kenilworth alignment. But facts are such troublesome things.

As for the streetcar on the Greenway: 3,300 daily riders compared with an estimated 8,000-9,000 on Hennepin, more on Chicago. Higher costs, too. I don’t have the Streetcar Feasibility Study in front of me for more exact figures, but the point is clear. Policy makers and many transit “advocates” in Minneapolis have their priorities seriously out of whack. The Southwest LRT bypassing the Uptown area will go down as one of the city’s biggest transportation mistakes since ripping out the old streetcar network.

Winston | December 18, 2009 at 12:23

Is cost per boarding really a good metric? It seems that cost per passenger-mile is a far better metric. Consider the choice between operating the following two bus lines:

They are both good metrics. Consider that if the Express Bus runs with 10 miles between stops and each stop is a local TOD center, a far greater variety of trips are available via the Express bus than if it is a non-stop commuter from park and ride lot to downtown destination. Each trip offered transport of value to someone, even if they only rode 1 mile, and that diversity of service is ignored in a passenger-miles only metric.

The problem with the CEI is that it is trying to do the analytically impossible – assign a single scalar index to a vector joint product. Certainly under the traditional neoclassical model its possible, but that is at the expense of accepting counter-factuals as model assumption. Out in the real world, its impossible. And failing at the impossible should not be tremendously surprising.

Better to have a Commuter’s support ranking, a Local Transport support ranking, and Energy efficiency support ranking, etc. and allow projects to gain a certain proportion of funding (10%, 20% or 40%, based on the weight given to the goal) for being a winning project in a particular specific goal, to a maximum of 80% total federal funding.

Discussion of the FTA’s CEI process should not even happen without mentioning the fact that no road project in the United States, regardless of its dubious merits, must pass through any similar type of scrutiny.

Yes, there will be always more potential projects than money. However, this post reinforces the modal tranches of federal investment that have been around for too long as making sense in the first place.

A new transportation bill where funding moves away from backwards state DOTs and towards MPOs, with a fixed percentage match for infrastructure across all modes rather than 80% for highways even if they generate low benefits and “50% if you’re lucky” for transit projects that generate massive benefits– is the key to getting this fixed, not nibbling around the edges of a poorly designed, elaborate Rube Goldberg device of an evaluation program whose sole purpose is to split hairs among even the most worthy projects to allocate an embarrassingly small pot of money across a nation of 300 million people.

There’s no such thing as a vector joint product. There are scalar products of vectors, but they’re not very useful here – they measure the angle between the vectors, which you don’t particularly care about.

What you say about having a commuter support ranking and an energy efficiency ranking is taking a vector in 3 or 4 dimensions, and then projecting it onto 1 dimension based on some weighting formula. Personally I think the primary metric should be “cost per rider” and the feds should say something like “We pay up to a flat $30 per annual ride.” That translates to about $9,000 per weekday boarding.

“There’s no such thing as a vector joint product”

I think that’s another name for the so-called “outer product”, where you “multiply” two vectors and get IIRC a matrix. It’s rarely used, but you can look it up.

And regarding the article, Yonah is totally, utterly wrong. The existing CEI just isn’t a good cost-effectiveness measure. Come up with a decent cost-effectiveness measure and we can encourage it — but one this stupid really needs to be made much less relevant.

I didn’t know it was used as a name for an outer product. But an outer product is just as inappropriate here – all it does is give you the coefficients you get from multiplying each coefficient of vector a with each coefficient of vector b.


-I’ll be clear on the race issue. While it wasn’t wholly explicit in the studies or by the Southwest LRT project management, they made sure that everybody knew that the Kenilworth corridor “served” North Minneapolis. Demographically, the neighborhood that is hypothetically served, Harrison, has nearly the same demographics as Whittier.

They also coaxed the numbers into appearing that large numbers of bus transfers from North Minneapolis (routes 5, 19, 22) would come to Royalston station. It’s a wash, and is a terrible blunder. Transit for Livable Communities’ founder, John DeWitt, along with many people in that organization, are definitely the “misguided advocates” you’re referring to.


I agree whole-heartedly. But then, TLC has become an institution dedicated to preserving unsustainable suburban sprawl by supporting suburban commuter lines for new dedicated-ROW transit. TLC–its founders included–has become a laughingstock when you contrast it to its far more progressive nationwide peers.

Wow a lot of Twin Cities folk here.

One bothersome part about the CEI is it generally seems to reward far suburban/exurb bus transit service well. It’s better to run a ‘slow bus’ than to operate a ‘fast bus’ — if its slow and has more ridership and the modal trip time difference is huge, you’ve scored! Skip the city core and go right downtown and still be ok on CEI numbers. Bypass the transit dependent (with exceedingly slow trip times with low ridership due to their slow trip times) and well operated lines (ie: dedicated guideway use through over congested roads like 252) even though they have more potential ridership nearby due to higher density, because we’ve got a pile of suburban bus riders to move to rail transit!

Heck, I guess BP is not looking well for Bottineau because they don’t have a good handle on where the Target project is at vs. Maple Grove who has a master plan for their gravel pits (who knows when those get developed, but they have a plan, BP does not have something solid yet it sounds)

Oh, and I’d be fine with MPO’s running the show if MetCouncil wasn’t appointed by the governor. The CTIB would be closer to what I’d like to see — elected officials serving on a joint powers board deciding on transit investments in their communities.

Leave a Reply