[sustran] Evaluating Rail Transit Benefits (was UK Funding switch from Light Rail to Busway)

[Todd Alexander Litman]

Dear Colleagues,

It is unfortunate that the debate over the merits of rail transit has 
become so polarized. I urge all of you to read my report and contrast it 
with O'Toole's paper before you make any judgments one way or the other.

My report is not simply a list good things about rail and bad things about 
bus or automobile travel (in contrast with O'Toole's document, which only 
discusses the disadvantages of rail). I have tried to provide a balanced 
discussion of the advantages and disadvantages of each mode, and therefore 
the situations in which each is most appropriate. Although you might place 
more emphasis on one factor or another, I have yet to have anybody 
challenge the basic concepts or the categories of advantages and 
disadvantages in my paper.

I am certainly not saying that rail transit is always a cost-effective 
investment, or that all rail projects are optimal. I have, however, 
identified a number of benefits in terms of transportation system 
performance (per capita transportation costs, congestion delays, fatal 
traffic accidents, etc.) which seem to result from the leverage effects 
rail can have on a city's transportation and land use patterns. There are 
legitimate questions as to whether these advantages are simply legacies of 
older cities that cannot be achieved with new rail (experience in cities 
such as Portland indicate that new rail systems can reduce per capita 
vehicle ownership and mileage, but more research is needed to determine the 
overall magnitude of these impacts and the total potential market for such 
housing), and different people may place different values on these 
benefits, but it would be unfair to a community to evaluate rail system 
benefits, without considering these additional impacts. They are 
potentially very large and so have significant implications for project 
evaluation.


Below are my responses to Mr. O'Toole's comments about my paper.

>1. NEW YORK DISTORTS DATA
>I like to say that the U.S. has two kinds of urban areas: New York
>and everywhere else. Nowhere else has a Manhattan with 52,000 people
>per square mile and (more important) 80,000 jobs per square mile. New
>York transit has more than twice the market share of the next leading
>region.
>
>Lumping New York in the transit data for any other group of urban
>areas (as you do in your discussion of "Large Rail cities" along with
>tables 1, 6, 7, and elsewhere in your report) produces distorted
>results that are not reflective of other regions. Because New York is
>so large and because it produces more than 5 times as many transit
>rides as the next-highest urban area (and 38 percent of all transit
>rides in the U.S.), the averages you get from lumping it with other
>regions will be unrealistically high for any other region.

Todd Litman: I have recalculated the data in my spreadsheet 
(http://www.vtpi.org/rail.xls) excluding New York, and the results changed 
little, indicating that other "Large Rail" cities have per capita 
transportation system performance similar to that of New York, despite the 
fact that they are significantly smaller. Below are a few examples to 
illustrate the point.

Annual Per Capita Transit Trips
                                         Large Rail      Small 
Rail      Bus Only
50 largest U.S. Cities With New 
York              589             176             118
50 largest U.S. Cities W/O New 
York               520             176             118


Annual Traffic Fatalities Per 100,000 Population
                                                 Large Rail      Small 
Rail      Bus Only
50 largest U.S. Cities With New 
York                       7.46                  9.99           11.72
50 largest U.S. Cities W/O New 
York                        7.90                  9.99           11.72


Per Capita Annual Congestion Costs
                                                 Large Rail      Small 
Rail      Bus Only
50 largest U.S. Cities With New York                       $551 
$466            $397
50 largest U.S. Cities W/O New York                        $561 
$466            $397

Percent Income On Household Expenditures
                                                 Large Rail      Small 
Rail      Bus Only
50 largest U.S. Cities With New 
York                       12.04%       15.81%          14.89%
50 largest U.S. Cities W/O New 
York                        12.02%       15.81%          14.89%


In each of these cases, excluding New York reduces the relative advantage 
of Large Rail cities, but by a modest amount, indicating that other Large 
Rail cities also enjoy significant benefits.


>Similarly, the other six "Large Rail cities" are all older cities
>with high-density cores that have not been built elsewhere in the
>last century. While it is amazing that these regions have such low
>transit ridership compared with New York, any results for these six
>regions cannot be applied to newer regions such as Atlanta, Phoenix,
>and San Jose. These newer regions are just never going to look like
>Chicago or San Francisco. This is why I compared each region individually 
>and didn't try to
>lump them together.

Todd Litman: That is a key issue discussed in my paper, that is, whether 
new rail systems can achieve the land use impacts of older rail systems 
(see the "Counter Arguments" section). I think there is evidence that it 
can, provided it is supported with appropriate transport and land use 
policies. The question is not whether Atlanta can become Chicago, but 
whether some parts of Atlanta can become like some parts of Chicago, and 
whether rail projects that leverage such land use patterns provide more 
benefits than alternative transportation improvements in that area or on 
that corridor.

O'Toole misrepresents his paper when he claims that it does not try to lump 
cities with rail transit together. His report claims that, "The 
twenty-three urban areas with rail transit collectively lost more than 
33,000 transit commuters during the 1990s, while the twenty-five largest 
urban areas without rail transit collectively gained more than 27,000 
transit commuters." Not only is this an example of broad statements he 
makes comparing rail and non-rail cities, but he admitted to me more than a 
month ago that the numbers are incorrect (it should be a 14,097 loss in 
rail cities and a 1,388 gain in bus cities), yet he has not changed the 
wording of the reports posted on his website.


>2. PER CAPITA TRANSIT RIDERSHIP IS NOT AN INDICATOR OF LIVABILITY
>Much of your report focuses on the allegedly high per capita transit
>ridership in rail regions. But why is this important? Even the
>fastest transit tends to be slower and (because it is not
>door-to-door) less convenient than the automobile. High levels of per
>capita ridership thus suggest lower levels of mobility.
>
>Perhaps this is because the city is so well designed that people
>don't need that mobility -- the Robert Cervero argument for
>accessibility rather than mobility. In fact, no urban area, with the
>possible exception of New York (really, only parts of New York) is
>designed to give people accessibility through transit.
>
>This means that high levels of per capita transit ridership probably
>mean lower levels of mobility, which in turn means higher housing
>costs, consumer costs, and other costs.

Todd Litman: I agree that transit ridership is an objective, not a goal. 
The goal is to improve transportation system performance and provide 
benefits to consumers. Most urban mobility is a derived demand, to provide 
access to goods, services and activities. Few people would choose a more 
round-about commute trip simply because they enjoy the extra driving. 
Relative speed is just one aspect of access. For example, some people 
prefer commuting by transit, particularly rail, because it is less 
stressful than driving, even if it takes more time.

Out of ignorance or intent, O'Toole evaluates transportation only in terms 
of mobility, not accessibility, and therefore ignores any benefits that may 
result when transit provides a catalyst for more accessible land use 
patterns which reduce per capita vehicle mileage. He makes no mention of 
this issue or its implication in terms of transit evaluation. For more 
information see my article "Measuring Transportation: Traffic, Mobility and 
Accessibility", published last year in the ITE Journal 
(http://www.vtpi.org/measure.pdf), which discusses the difference between 
measuring mobility and accessibility.

I don't claim that every rail project significantly increases land use 
accessibility, or that only rail transit investments can achieve these 
changes, but my study suggests that when they occur they can provide large 
benefits to a community, including benefits to people who do not currently 
ride transit.

An interesting finding of my research is that even relatively modest 
increases in the portion of regional trips made by transit can result in 
relatively large reductions in per capita transportation costs, congestion 
costs and traffic deaths. I think this occurs because the regional data 
hide quite large impacts at the local level. For example, here in Victoria, 
BC, residents of some neighborhoods, those that reflect transit-oriented 
development attributes, tend to own fewer cars, drive less and walk more 
than residents of more suburban neighborhoods. In addition, the trips that 
tend to shift from automobile to transit tend to be the ones that provide 
the greatest benefit (i.e., peak-period travel to major centers).

O'Toole's claim that higher levels of accessibility and lower levels of 
mobility cause higher housing and transportation costs is typical of the 
claims made by smart growth critics without analysis or evidence. My 
research indicates that residents of cities with large, well-established 
rail systems spend significantly less on transportation than residents of 
other types of cities (12% versus 16% for Small Rail and 15% for Bus Only 
cities), representing about $450 annual savings per capita, despite the 
fact that Large Rail city residents earn substantially more, and so would 
otherwise be expected to spend more on their vehicles. These values do not 
change significantly when New York is excluded. In addition, reductions in 
per capita vehicle ownership in Large Rail cities can translate into 
significant residential parking cost savings. As mentioned in my paper, 
there is considerable uncertainty as to the impacts that smart growth and 
transit-oriented development have on housing affordability - in some ways 
they reduce affordability (reduced per capita land supply) and in other 
ways they increase it (increased density allowances and housing options, 
reduced building setback requirements, reduced per capita parking costs, 
transportation costs, etc.). It is simply wrong to claim that smart growth 
and transit oriented development necessarily reduce housing affordability 
(see "Parking Requirement Impacts on Housing Affordability" at 
http://www.vtpi.org/park-hou.pdf).


>3. THE REPORT EMPHASIZES POINTS IN TIME WITHOUT SCRUTINIZING TRENDS
>Most of the indicators in Great Rail Disasters were trends: typically
>1990 to 2000. Most of the indicators in your report represent single
>points in time. Rail regions, for example, may have high per capita
>transit ridership, but if transit commuting is declining while it is
>increasing in bus regions, then that high ridership is pretty
>meaningless. Rail cities may have slightly lower per capita driving,
>but if per capita driving is increasing faster in those cities, it
>will not do them much good.
>
>Of your "large rail cities," Boston is the only one that is showing
>much transit growth. Of your "small rail cities," Portland and, to a
>lesser degree, San Diego are the only ones showing much transit
>growth. That is hardly indicative of rail's great success.

Todd Litman: I hope to expand my research to include time series analysis. 
As stated in my paper, I agree that low transit ridership in U.S. cities 
(both with and without rail) is a reason for concern, but I don't think 
that proves that rail is necessarily a failure or a bad investment. A 
number of cost-effective strategies described in my paper can help increase 
transit ridership and attract discretionary riders (e.g., parking cash out, 
improved walkability, pay-as-you-drive vehicle insurance, congestion 
pricing, etc.). The question therefore shifts from whether transit is good 
or bad, to how to implement transit in conjunction with other 
transportation and land use policies to optimize benefits. The concerns 
O'Toole raises can therefore justify MORE rather than LESS support for rail.


>4. RAIL COST EFFECTIVENESS IS GREATLY OVERESTIMATED
>The report says that "rail transit is generally constructed in the
>densest part of a city where capacity expansion is most costly." It
>is equally true that rail transit is generally constructed in the
>slowest growing part of a city where capacity expansion is least
>needed. In any case, we have several examples of parallel rail and
>highway construction where the rail cost per passenger mile was far
>greater than the highway cost (I-25 in Denver, I-15 in Salt Lake City).
>
>Table 4 of your analysis compares user costs without mentioning the
>huge subsidies for rail transit. Through gas taxes, U.S. highways pay
>for themselves. Total subsidies to auto users are little more than
>0.3 cents per passenger mile. Subsidies to the average transit rider
>are around 60 cents per passenger mile, and subsidies to rail riders
>are greater.
>
>Your analysis also compares operating costs, when in fact capital
>costs (when annualized using a standard amortization formula as
>required by the FTA) greatly outweigh operating costs for rail
>transit. That is like comparing the costs of housing but leaving out
>the costs of the walls and roof!

Todd Litman: I agree that transit cost analysis should include both capital 
and operating costs, and I will incorporate capital costs in future 
analysis. I left them out because it requires more detailed research, since 
capital costs can fluctuate significantly from one year to the next, so 
considerable data is needed for accurate analysis.

Rail project budgets incorporate nearly all costs. Buses require highways; 
and automobile travel requires vehicles, highways and parking facilities, 
costs ignored in O'Toole's analysis. It is therefore not appropriate to 
compare rail capital costs with bus capital costs (ignoring roadway costs), 
or with highway costs (ignoring vehicle and parking costs) when evaluating 
the cost effectiveness of these options. Although it can be argued that 
vehicle fuel taxes fund most roadway costs, there are a number of 
additional costs and subsidies. In particular, peak-period urban-highway 
drivers are cross-subsidized by motorists who seldom or never use such 
facilities, land used for road rights-of-way pays no rent or taxes, and 
parking facilities are a large subsidy of automobile travel.

I have performed considerable research on the proper evaluation of 
different modes of transportation (see "Transportation Cost and Benefit 
Analysis at http://www.vtpi.org/tca and "Evaluating Public Transit Benefits 
and Costs at http://www.vtpi.org/tranben.pdf, which will soon be updated), 
in order to insure that all of these impacts are considered. My goal is not 
to make one mode look better than another, but rather to create a 
comprehensive and accurate evaluation framework.


>5. HOUSING AFFORDABILITY DISCUSSION WRONG
>The report says that "rail transit projects and smart growth policies
>are generally implemented in rapidly growing cities where property
>values are rising due to increasing demand." That is not necessarily
>true. The fastest growing cities in the U.S. have no rail transit and
>little smart-growth planning and their housing remains very
>affordable. It is only in cities such as San Jose and Portland, where
>planners have attempted to create a transit utopia by increasing
>population densities that housing prices have become dramatically
>unaffordable.

Todd Litman: As mentioned above, critics of smart growth have 
misrepresented the issues regarding housing affordability (See Danielle 
Arigoni, "Affordable Housing and Smart Growth: Making the Connections," 
Subgroup on Affordable Housing, Smart Growth Network (www.smartgrowth.org) 
and National Neighborhood Coalition (www.neighborhoodcoalition.org), 2001; 
Edward Carman, Berry Bluestone and Eleanor White, "Building on our 
Heritage: A Housing Strategy for Smart Growth and Economic Development," 
Center for Urban and Regional Policy, Northwestern University 
(www.curp.neu.edu), for the Commonwealth Housing Task Force, October 2003; 
Arthur C. Nelson, Rolf Pendall, Casy Dawkins and Gerrit Knaap, The Link 
Between Growth Management and Housing Affordability: The Academic Evidence, 
Brookings Institution Center on Urban and Metropolitan Policy 
(www.brook.edu/dybdocroot/es/urban/publications/growthmang.pdf), 2002). 
Housing unaffordability tends to occur in any desirable neighborhood. 
Within such areas, smart growth and transit-oriented development can reduce 
per capita land requirements, parking requirements and transportation 
costs, making them more affordable.


>6. SAFETY DISCUSSION USES WRONG MEASURE
>I compared the safety of various forms of transport in terms of
>fatalities per passenger mile. You compare it in terms of fatalities
>per capita. If it is true that smaller cities have higher per capita
>driving, then they can have lower fatalities per passenger mile yet
>higher fatalities per capita.
>
>Which is the right measure? If you value mobility, as I do, then
>fatalities per passenger mile is the correct measure. Though
>regrettable, fatalities result from almost anything we do. The
>question is whether what we do is worth the risk. Is getting to work
>on time worth the tiny and declining risk of getting killed in
>traffic? Apparently it is because most people drive.
>
>If you don't value mobility, then fatalities per capita may be
>adequate. But then you have to ask what the people in your smaller
>rail and bus cities are getting for their mobility. I suspect they
>are getting lower housing prices and other consumer costs, a wider
>range of job opportunities, access to more recreation, etc.

Todd Litman: As mentioned above, most urban mobility is a derived demand, 
not an end in itself. Few people drive across town just for the fun of it, 
and few people would prefer to live in a community with 11.7 annual traffic 
fatalities 100,000 residents if they can enjoy a similar level of 
accessibility while living in a community with only 7.5 fatalities per 
100,000 population (for even more dramatic evidence see the differences in 
traffic deaths between most- and least-sprawled cities in "Evaluating 
Criticism of Smart Growth, http://www.vtpi.org/sgcritics.pdf). Large rail 
cities, by the way, tend to have higher average incomes, suggesting more 
rather than less access to employment and higher levels of productivity.


>7. THE COST OF SPRAWL IS EXAGGERATED
>Your report says that I "favor automobile-oriented sprawl." Nothing
>could be further from the truth. I favor freedom of choice and I
>oppose government manipulation of people to get some predefined (and
>ineptly designed) goal.

Todd Litman: There are more regulations supporting sprawl and automobile 
dependency than supporting smart growth, including minimum parking 
requirements, single-use zoning, restrictions on density and multi-family 
housing, building setbacks and generous road standards, yet O'Toole only 
argues against one set of regulations. Analysis by Dr. Donald Shoup found 
that parking costs are 4.4 times higher than the average total of all other 
development fees, including roads, schools, parks, water, sewage and flood 
control.


>Cities without zoning (e.g. Houston) have demonstrated that, in the
>absence of regulation, people prefer to drive and to live in
>low-density, single-use developments. Cities with high degrees of
>regulation and restrictions on driving and low-density development
>(e.g., Paris, Amsterdam, and almost any other major European city)
>show that people still prefer to live in low densities and to drive,
>as driving is rapidly increasing and densities declining in almost
>all European cities.

Todd Litman: The evidence is quite mixed, and it misrepresents the issue to 
claim that it proves any single thing. For example, many cities in North 
America and Europe are experiencing redevelopment and population growth. It 
is simply wrong to claim that people prefer to drive and live in 
low-density areas; people are diverse and at least some prefer very 
different types of housing. Studies described in my paper indicate that 
many households are willing to pay a premium for New Urbanist housing and 
proximity to rail transit. Whether this market segment is a minority of a 
majority of all consumers is irrelevant, as long as there is a sufficient 
demand (I estimate 15-25% of households in a typical urban region), then 
this is a market segment that deserves to be served with quality transit 
services and transit-oriented development.


>What is wrong with what you call "sprawl"? The Russians say that
>"Americans don't have real problems, so they make them up." Sprawl is
>one of those made-up problems. Pollution from auto driving is rapidly
>declining even though we drive more every year. Auto fatalities are
>also declining. Lower densities translate to lower housing and
>consumer costs, lower taxes, and less congestion.
>If people decide to move to higher densities, that is up to them. I
>only oppose subsidies and regulation designed to promote higher
>densities and discourage lower densities.

There is considerable literature on the costs of sprawl and benefits of 
smart growth (see references in http://www.vtpi.org/tdm/tdm20.htm). Smart 
growth is supported by a variety of mainstream organizations including the 
Institute of Transportation Engineers (ITE, 2003), the International 
City/County Management Association (ICCMA, 1998) and the American 
Governor's Association (Hirschhorn, 2001), because of the cost savings and 
other benefits it provides.

As a person of Russian descent I recommend against making general 
statements about what Russians say: two Russians, three opinions. All the 
claims O'Toole makes are only partly true. Some U.S. cities are 
experiencing increasing air pollution problems as increased vehicle mileage 
offsets reductions in per-vehicle mileage. Although per-mile traffic 
fatalities are declining, per capita trends have leveled off and may be 
increasing (it is striking that the U.S. has one of the lowest per-mile 
traffic fatality rates, but one of the higher per-capita among developed 
countries due to high per capita vehicle mileage). Lower density housing 
often leads to HIGHER housing costs (due to higher infrastructure costs) 
and higher total housing and transportation costs.

As mentioned above, the test of whether O'Toole really opposes subsidies 
and regulations is whether he applies the same effort to reducing those 
that favor automobile dependency and sprawl.


>9. LACK OF REFERENCES A VALID CRITICISM
>You accurately point out that I failed to provide adequate references
>to some of my statements. I still stand behind those statements. In
>one case, I said that most rail cities are spending over half their
>transportation capital funds on transit. You can find the references
>at http://ti.org/vaupdate24.html . I will send you the list of EISs
>that I used to review rail costs and ridership soon.

That is helpful. However, the evidence presented misrepresents the issue. 
Rail transit projects show up in regional capital budgets, so they may 
appear proportionately large, but regional capital budgets are only a small 
portion of total transportation expenditures in any region. There are also 
regional operation and maintenance budgets, local and state capital and 
operating budgets, expenditures by businesses on parking facilities, and 
expenditures by consumers on vehicles and residential parking, all of 
which  should be considered when calculating and comparing costs of 
different modes.

Regional mode share is simply not an appropriate basis for evaluating 
investments. In most urban regions motorists have little trouble driving to 
just about any destination with reasonable speed, comfort and safety, 
provided it isn't during peak periods. The major transportation problems 
facing most urban regions include traffic and parking congestion on major 
urban corridors, and inadequate mobility for non-drivers. These are exactly 
the problems that transit improvements are most appropriate for addressing. 
Transit improvements can help achieve a variety of objectives, including 
reduced consumer costs, reduced parking costs, reduced accidents and 
reduced pollution emissions, all costs that tend to be exacerbated by 
highway capacity expansion and increased automobile dependency. As a 
result, transit investments that improve service quality and attract 
discretionary travelers who would otherwise drive on major urban corridors 
can be the most cost effective way to improve transportation, even if 
transit represents a small portion of total regional travel.


Please let me know if you have questions or comments about my report.


Best wishes,
-Todd Litman


At 11:32 PM 4/26/2004 +0700, Jonathan E. D. Richmond wrote:
>The situation Brendan describes in Ireland seems to be almost universal.
>As I found in my work (and as Don Pickrell established before me),
>forecasts are overestimated, costs underestimated and -- more important --
>there is a lack of realistic analysis of alternatives, which are assumed
>to be inferior from the start. Such analysis as takes place is usually
>biased: For example, a rail plus feeder bus network is compared to a bus
>only network, but the cost of the feeder buses is left out!
>
>In case any of you have not yet seen Randy O'Toole's response to Todd
>Litman, I am posting it below. I differ from Randy on his views of sprawl,
>and support the provision of public transport with public subsidies where
>appropriate, but I believe his analysis should be noted. My own analysis,
>(which I will be pleased to send anyone interested), shows that new rail
>projects in the US are a failure in every case except San Diego, and we do
>need to look at the record of such new projects to understand what is
>happening and not look at data for New York which really is not going to
>tell us much about the likelihood of success in rather different
>situations.
>
>                                --Jonathan
>
>-----
>
>Jonathan E. D. Richmond                               02 524-5510 (office)
>Visiting Fellow                               Intl.: 662 524-5510
>Transportation Engineering program
>School of Civil Engineering, Room N260B               02 524-8257 (home)
>Asian Institute of Technology                 Intl.: 662 524-8257
>PO Box 4
>Klong Luang, Pathumthani 12120                        02 524-5509 (fax)
>Thailand                                      Intl:  662 524-5509
>
>e-mail: richmond at ait.ac.th               Secretary:  Ms. Nisarat Hansuksa
>         richmond at alum.mit.edu                         02 524-6051
>                                               Intl:  662 524-6051
>http://the-tech.mit.edu/~richmond/


Sincerely,
Todd Litman, Director
Victoria Transport Policy Institute
"Efficiency - Equity - Clarity"
1250 Rudlin Street
Victoria, BC, V8V 3R7, Canada
Phone & Fax: 250-360-1560
Email: litman at vtpi.org
Website: http://www.vtpi.org