Professional Documents
Culture Documents
AAF Friedman Final
AAF Friedman Final
of
All Aboard Florida
February 2015
John N. Friedman
Brown University
Table of Contents
Executive Summary of Key Findings
I. Transportation Market Analysis
A. Analysis of Ticket Prices
B. Analysis of Ridership
II. Economic Feasibility of All Aboard Florida
A. Revenue Estimates
B. Operating Cost Estimates
C. Capital Cost Estimates
D. Overall Feasibility
III. Government Subsidy of All Aboard Florida
A. Taxpayer Costs from Subsidized Bonds
B. Economic Inefficiency of Tax-Exempt Bonds
C. Other Government Subsidies
About the Author
All Aboard Florida will generate an average ticket price of $34 and attract
between 1.5 and 2 million riders.
Train fares are limited for business travelers by the relatively cheap airfare for the
one-hour flight between Miami and Orlando, as well as the fact that unlike other
long-distance trains All Aboard Florida will arrive at the Orlando airport rather
than downtown, which is more advantageous to passengers. Fares for personal
travelers will be limited by the relative ease of car travel. Train ridership will also
be limited by high levels of urban sprawl and the lack of connecting public transit.
All Aboard Florida will generate annual losses of more than $100 million and
will be unable to service its large debt burden.
Under optimistic projections, this report projects annual revenues of $95.8
million, operating costs of $81.5 million, and debt-service costs of $125 million,
for annual losses of $110.7 million. AAF has no way to pay the resulting annual
deficit, except perhaps by non-train-related business such as real estate profits.
All Aboard Florida would have to charge $273 per train ticket one-way, even
under unrealistically optimistic assumptions, in order to service its debt.
The revenue required to service its debt is more than twice as large as actual
projected revenues. The implied fare of $273 required to raise this revenue
even holding ridership constant would be $145 more expensive than airfare on
the Miami to Orlando route. But of course ridership would fall drastically at such
high prices.
All Aboard Florida will benefit from between $50 and $73 million of annual
taxpayer subsidies.
All Aboard Florida plans to issue $1.75 billion of Private Activity Bonds, which
are tax-exempt bonds similar to municipal bonds. The foregone tax revenue is an
inefficient taxpayer subsidy of $37-$60 million for a private enterprise. AAF will
also benefit from another $13 million of annual subsidies from the State of
Florida and local governments in the form of state-funded station in Orlando and
safety upgrades and maintenance along the rail line.
Note that both air travel and the AAF rail line will bring passengers to Orlando Airport, and so the modes
of travel are equivalent from the perspective of traveling within the Orlando area. This aspect of travel is
therefore omitted from this analysis.
savings from the U.S. Department of Transportations analysis, which is $57.20 for
business travelers. 2 This implies that the average traveler values the 45 minutes saved
through air travel at $42.90. Including the higher cost of parking at the train station
increases this benefit to $46.90.
This analysis suggests that the average passenger is willing to pay $46.90 extra for air
travel relative to the proposed AAF rail line. Average one-way tickets from Miami to
Orlando cost $127.50. 3 Therefore, the market should be able to support one-way AAF
train fares between Miami and Orlando of roughly $81.
Leisure and Personal Travelers
Leisure and personal travelers choose between train and car. In order to better define the
costs of each, I analyze a hypothetical trip from Miami to Disneyworld Resort. The
direct car trip is 227 miles, which I assume would take 4 hours. The train would take 3
hours from station to station, plus an additional 15 minutes to reach the station in Miami
and an additional 30 minutes to reach Disneyworld from Orlando Airport (the site of the
Orlando AAF station). 4 Therefore, the train trip would save just 15 minutes relative to
the car trip. The U.S. Department of Transportation recommends a value of time of
$16.70 for intercity personal auto travel, so the extra 15 minutes is worth just $4, which is
just enough to offset the assumed $4 parking fee at the train station.
The only other cost of car travel is the use of the car itself; AAA estimated in 2014 that
the cost of driving each additional mile is $0.19, including gas, maintenance, and tires,
for a total of $43 for the entire one-way trip. Thus, personal travelers who drive by
themselves will pay up to $43 for a train ticket. This number is lower for families who
travel in larger groups; a similar analysis that accounts for the cost of time for each
traveler but allows parties to share other costs suggests that families of 4 will pay just $19
per person for a train ticket. I take the average of $31 between these figures for the single
projected ticket price for personal and leisure travelers. 5
Estimated Fares for Other Station Pairs
Based on theoretical AAF fares of $81 and $31 between Miami and Orlando for business
and personal travelers, respectively, I assume a constant per-mile train fare to project
fares for a one-way ticket to Orlando from Fort Lauderdale and West Palm Beach. For
business travelers, the fares would be of $71 and $56, respectively; for personal travelers
it would be $27 and $22. 6 For business travelers between Miami and West Palm Beach, I
Table 3, DOT Departmental Guidance for the Valuation of Travel Time in Economic Analysis, Revision
2, September 28, 2011.
3
Data analysis from www.farereport.com, based on Office of Aviation Analysis data from the U.S.
Department of Transportation.
4
I assume no cost other than time between Orlando Airport and Disneyworld due to the prevalence of free
shuttles.
5
An alternative interpretation of this analysis is that, at higher fares, AAF will fail to attract many personal
travelers, and especially those traveling in large groups. In this case, fares would be higher and ridership
lower, so revenue would be roughly similar to the case I present below.
6
Throughout this analysis, I assume that AAF can charge distinct fares to business and personal travelers.
In practice this separation is not exact, but one can do quite a bit through differential pricing across
departure times and other price discrimination.
then take AAFs own projections of a $24 fare. 7 I assume a $18 fare from Miami to Fort
Lauderdale and Fort Lauderdale to West Palm Beach. Relative to my market-based
projections for the Orlando routes, the AAF projections are slightly higher fares-per-mile,
though the gradient is roughly similar to that seen in Amtrak fares between shorter and
longer routes. These fares are clearly too high to attract personal travelers, so I reduce
them by 33% to $16 and $12, respectively. Table 1 presents the full set of fares. The
average one-way fare is $30.50, based on population-weighted ridership shares between
city-pairs and assuming that half of riders are business riders and half are personal
riders. 8
Table 1: Projected Fare Prices
Arrival Station
Average Fare: $30.50
Miami
Departure
Station
Departure
Station
Fort
Lauderdale
West
Palm
Beach
Orlando
Miami
Fort Lauderdale
West Palm Beach
Orlando
$81
$71
$56
$18
$24
$81
Miami
Fort Lauderdale
West Palm Beach
Orlando
York of course comes on top by far with 30.5% usage. Amtrak maintains 13 stations
along this route for the highest-speed Acela service, as well as a number of additional
stations served by its Northeast Regional service. With all of these advantages, in
FY2014 Amtrak ridership on this route grew to 11.6 million, its highest ridership ever. 11
All Aboard Florida proposes to link two combined metropolitan areas with total
population of 9.4 million people. Just scaling ridership for the Northeast Corridor down
to account for population would predict 2.27 million riders annually along the entire
route. This clear overestimate is still considerably smaller than AAFs projection of 3.47
million passengers once fully operational. 12 Clearly these AAF projections do not line up
with the experience of long-distance rail in the United States.
There are a number of other factors that drive any reasonable prediction below this basic
estimate. First, AAFs route would include just four stops; as compared with the
Northeast Corridor, AAF simply generates fewer routes to attract riders. Second,
Orlando and Miami do not have nearly the existing local public transit links to connect
AAF stations to other destinations. The combined public transit usage among workers in
Miami and Orlando is just 3%. This lack of public transit links is compounded by the
fact that the Orlando station is not in the city center but instead at the airport.
Finally, businesses and individuals are located far more diffusely around the city centers
in Miami and Orlando, as compared with the relatively compact cities along the
Northeast Corridor. For example, U.S. Census records show that 40% of the population
of cities along the Northeast Corridor live within 10 miles of the city center (where the
train stations are located), as compared with just 23% in Miami. 13 Similarly, cities along
the Northeast Corridor have 27% of jobs located within 3 miles of the central business
district, as compared with just 21% of jobs in Miami and Orlando. This situation is also
not improving; over the past decade, Orlando has had one of the fastest growing outer
rings, with the share of jobs outside 10 miles from the city center increasing by 7
percentage points from 38% to 45%. 14
Put simply, even after accounting for their size, Miami and Orlando are not cities with
characteristics in which passenger trains can thrive. Taking this into account, a reasonable
forecast would have somewhere between 1.5 and 2 million AAF riders annually; I
conduct the rest of this analysis using both of these figures as alternative scenarios, with 2
million as the optimistic case and 1.5 million as the realistic case.
I then assume that that these riders are distributed along the line proportionally with
population that is, if Fort Lauderdale has 50% higher population than West Palm
Beach, then there will be 50% more riders embarking at Fort Lauderdale for each
destination than at West Palm Beach. The route-specific ridership shares appear below in
Table 2; my assumptions imply symmetry between departure-destination pairs. Roughly
60% of the ridership is between the three cities in the Greater Miami area; the remaining
40% of journeys connect the Miami area with Orlando.
11
Amtrak Ridership and Revenues Continue Strong Growth in FY2014, October 27, 2014.
http://www.amtrak.com/ccurl/238/481/Amtrak-FY2014-Ridership-and-Revenue-ATK-14-096%20.pdf
12
AAF Draft Environmental Impact Statement, p. 3-47.
13
U.S., Census, Distance Profiles for U.S. MSA in 2000 and 2010.
14
Brookings Institute Report, Job Sprawl Stalls, April 2013.
York of course comes on top by far with 30.5% usage. Amtrak maintains 13 stations
along this route for the highest-speed Acela service, as well as a number of additional
stations served by its Northeast Regional service. With all of these advantages, in
FY2014 Amtrak ridership on this route grew to 11.6 million, its highest ridership ever. 11
All Aboard Florida proposes to link two combined metropolitan areas with total
population of 9.4 million people. Just scaling ridership for the Northeast Corridor down
to account for population would predict 2.27 million riders annually along the entire
route. This clear overestimate is still considerably smaller than AAFs projection of 3.47
million passengers once fully operational. 12 Clearly these AAF projections do not line up
with the experience of long-distance rail in the United States.
There are a number of other factors that drive any reasonable prediction below this basic
estimate. First, AAFs route would include just four stops; as compared with the
Northeast Corridor, AAF simply generates fewer routes to attract riders. Second,
Orlando and Miami do not have nearly the existing local public transit links to connect
AAF stations to other destinations. The combined public transit usage among workers in
Miami and Orlando is just 3%. This lack of public transit links is compounded by the
fact that the Orlando station is not in the city center but instead at the airport.
Finally, businesses and individuals are located far more diffusely around the city centers
in Miami and Orlando, as compared with the relatively compact cities along the
Northeast Corridor. For example, U.S. Census records show that 40% of the population
of cities along the Northeast Corridor live within 10 miles of the city center (where the
train stations are located), as compared with just 23% in Miami. 13 Similarly, cities along
the Northeast Corridor have 27% of jobs located within 3 miles of the central business
district, as compared with just 21% of jobs in Miami and Orlando. This situation is also
not improving; over the past decade, Orlando has had one of the fastest growing outer
rings, with the share of jobs outside 10 miles from the city center increasing by 7
percentage points from 38% to 45%. 14
Put simply, even after accounting for their size, Miami and Orlando are not cities with
characteristics in which passenger trains can thrive. Taking this into account, a reasonable
forecast would have somewhere between 1.5 and 2 million AAF riders annually; I
conduct the rest of this analysis using both of these figures as alternative scenarios, with 2
million as the optimistic case and 1.5 million as the realistic case.
I then assume that that these riders are distributed along the line proportionally with
population that is, if Fort Lauderdale has 50% higher population than West Palm
Beach, then there will be 50% more riders embarking at Fort Lauderdale for each
destination than at West Palm Beach. The route-specific ridership shares appear below in
Table 2; my assumptions imply symmetry between departure-destination pairs. Roughly
60% of the ridership is between the three cities in the Greater Miami area; the remaining
40% of journeys connect the Miami area with Orlando.
11
Amtrak Ridership and Revenues Continue Strong Growth in FY2014, October 27, 2014.
http://www.amtrak.com/ccurl/238/481/Amtrak-FY2014-Ridership-and-Revenue-ATK-14-096%20.pdf
12
AAF Draft Environmental Impact Statement, p. 3-47.
13
U.S., Census, Distance Profiles for U.S. MSA in 2000 and 2010.
14
Brookings Institute Report, Job Sprawl Stalls, April 2013.
ridership by projected ticket prices for each route yields projected farebox revenue of
$61.0 million annually under the optimistic ridership scenario, and $45.7 million annually
under the realistic scenario.
AAF will also generate revenue from ancillary sources such as advertising or
sponsorship. 15 AAF projects these revenues at $21 million for the Miami-WPB section
of the line. In order to present an optimistic case for AAF, I accept the AAF forecasts for
the Miami area and then I assume that these revenues scale with ridership to account for
additional ancillary revenues from routes to and from Orlando. Section I.B projects that
ridership within the Miami area accounts for 60% of total ridership; therefore I assume
that the additional line to Orlando would increase these revenues by 66%, from $21 to
$34.2 million.
Putting these two sources of revenue together yields $95.2 million annual revenue for the
full line under the optimistic scenario, and $80.0 million under the realistic scenario.
B. Operating Costs
Operating costs include fuel, labor, rolling stock maintenance, and any other expense that
AAF will incur in the course of running the train line. In order to estimate operating
costs, I extrapolate from AAFs own estimates contained in various documents. Once at
full operating capacity in 2018, AAF projects annual operating expenses of $30.3 million
for 16 round-trip train trips (and 5 physical trains) on the Miami-WPB segment (66.5
miles). The full route from Miami-Orlando will also have 16 round-trips (and 10
physical trains), covering 220 miles.
For most expenses, including on-train labor, fuel, maintenance of equipment and
maintenance of way, it is most appropriate to scale by the total miles traveled. These
expenses are expected to total $19.0 million for the Miami area alone, and so I project
they will increase to $62.7 billion for the full line. Certain other expenses, such as offtrain labor, will scale by a smaller percentage; as with ancillary revenues in the previous
section, I assume that these other expenses scale with ridership and so increase by 66%
when moving to the full line. 16 These expenses increase from $11.4 million to $18.9
million. Putting these numbers together, I project annual operating costs of $81.5 million
for the full line.
C. Capital Costs
The operating costs discussed in Section II.B do not yet include the very large expenses
of acquiring the land, constructing the line, and acquiring rolling stock. AAF plans to
raise $1.75 billion in Private Activity Bonds (PABs) under authority from the Department
of Transportation. AAFs application for the PABs suggested that total start-up costs
would include an additional $1.4 billion, for a total of $3.1 billion in capital to fund
construction, land acquisition, rolling stock purchases, and other expenses. I assume that
15
Ibid. Footnote 7.
Starting with AAFs 2013 cost projection for 2017, I assume that 25% of labor costs are on-train and
75% of labor costs are off-train. I assume that 50% of Other Expenses are on-train and 50% off-train. All
other operating expenses are on-train.
16
50% of the additional funds are raised via equity. The total debt will then be roughly
$2.5 billion.
The cost of this capital depends on the interest rate demanded by the market to finance
the bonds. In June 2014, AAF raised $405 million in bonds at a 12% coupon rate. As I
discuss below, this is equivalent to roughly a rate of 9.5% once factoring in the
government tax subsidy for the Private Activity Bonds. Furthermore, some of the debt
will be secured on rolling stock or other assets and will require a lower interest rate.
I make the optimistic assumption that AAF will be able to achieve a weighted average
cost of capital for debt of 5%. I also assume no dividend or other service costs of equity.
This implies annual capital costs of at least $125 million. If interest rates rise above this
optimistic benchmark for instance due to the recent downgrade of AAFs parent
company Fortress Investment Group then costs would increase. For instance, a
weighted average cost of capital of 8% would generate $200 million in annual debt
service. Cost overruns in construction that require additional borrowing would also
increase capital costs. 17
D. Overall Feasibility
Table 3 puts together the numbers from the previous three sections, under each scenario.
Under the optimistic scenario, AAF will generate $95.8 million in annual revenue, face
$81.5 million in annual operating costs, plus $125 million in interest costs for tax-exempt
bonds and other debt. Total annual losses will be $110.7 million. Under the realistic
scenario, losses will be even larger at $125.9 million. In either case, AAF has annual
losses that are greater than annual revenues put another way, one could arbitrarily
double revenue projects and still not project that AAF would be able to pay off its debt.
This problem will become even worse if AAF were to experience delays in the
construction process than would require more years of debt service before generating
revenue.
Table 3: Project AAF Annual Profits
Optimistic
Scenario
Realistic
Scenario
Revenue
$95.8
$80.6
Operating Cost
$81.5
$81.5
$125.0
$125.0
($110.7)
($125.9)
Annual Profits
17
For example, the Draft Environmental Impact Statement responses included significant evidence that the
planned line will utilize 80 year old railroad bridges. If these antiquated facilities need to be replaced or
upgraded in order to obtain EIS and Coast Guard approval, it would add significantly to construction costs.
10
Another way to look at this financing problem is to ask the question: how much would
AAF need to charge in order to break even? After accounting for ancillary revenue
sources, this analysis shows that AAF would need to raise $171.7 million in ticket
revenue, more than a tripling of the revenue projection in the optimistic scenario. Even
holding ridership constant, this would imply charging $273 for a one-way ticket from
Miami to Orlando, about $145 more than the average airline fare on this route. But of
course ridership would in fact plummet at such high prices, necessitating even higher
fares. Although it is difficult to project exactly how such a negative feedback loop would
end, it is likely that there exists no fare structure with which AAF could generate revenue
merely to cover its operating and capital expenses.
This forecasts pessimism on AAFs ability to service its debt is shared by the bond
market. In June 2014, AAF sold $405 million of five-year bonds at 12%; this high
interest rate is striking given the historically low spreads that characterized the high-yield
debt market during that period. Data from the St. Louis Federal Reserve Bank measured
the average US high-yield bond option-adjusted spread over treasuries at just 3.5
percentage points in June 2014, lower than at other point since 2007.
A similarly negative assessment may also underlie the U.S. Department of
Transportations decision to offer AAF support for Private Activity Bonds rather than
accept their initial application for a loan through the Railroad Rehabilitation and
Improvement Financing (RRIF) program. This program offers subsidized direct loans for
the development of railroad infrastructure but require the payment of a Credit Risk
Premium that is assessed as a percentage of the total loan amount and varies by the
overall risk of each transaction based on repayment and other non-payment risk. 18
Although these calculations are not public, the analysis in this report suggests that this
Credit Risk Premium may have been extremely high.
18
11
The tax-exempt status generates a significant subsidy for borrowers, as well as significant
costs for taxpayers. Table 3 demonstrates these subsidies in the case of AAF. Column 1
gives the baseline case, in which a firm borrows $1.75 billion of corporate bonds that are
not tax exempt. For the purposes of this example, I assume optimistically that the bonds
would carry a 7.5% coupon. Note that at higher interest rates such as the 12% coupon
rate on the private bond notes previously issued by AAF on July 1, 2014 these numbers
would be correspondingly larger.
Paying 7.5%, AAF would pay $131 million in interest each year, on which bond holders
would owe approximately $37 million in taxes (assuming an average tax rate of 28%). 19
Investors would receive net interest payments of $94.5 million.
Table 4: Consequences of Tax-Exempt Bond Issuance
Corporate
Bond
Private
Activity Bond
$1,750
$1,750
Interest Rate
7.5%
5.9%
$131
$104
$36.8
Investor Proceeds
$94.5
$104
Principal
Now consider the effects if AAF raised the same amount of money through a tax-exempt
Private Activity Bond. Because investors pay no tax on interest received, AAF need not
offer the full 7.5% coupon; in fact, research suggests that coupon rates fall by 21% on
equivalent tax-exempt bonds. 20 In this case, the rate falls from 7.5% to 5.9%, and interest
costs for the firm fall to $104 million, all of which investors keep. 21
Comparing the two methods of finance, tax-exempt bonds represent roughly a $37
million annual subsidy for the project, of which roughly 75% accrues to AAF itself and
25% accrues to investors. If AAF were to pay a tax-exempt equivalent to the 12%
coupon rate on its PABs, the tax-payer subsidy would rise to $60 million per year.
19
28% is at the mid-point of estimates of the federal tax cost of tax-exempt bonds. Treasury Analysis of
Build America Bonds Issuance and Savings, May 16, 2011, U.S. Treasury.
20
Subsidizing Infrastructure Investment with Tax-Preferred Bonds, CBO/JCT, October 2009.
21
The coupon rate for the PABs is above the 5% weighted average cost of capital because some of the debt
not issued through PABs will be secured against the rolling stock. In contrast the PABs are unsecured.
12
13
22
23
14
15