Development Discussion Papers

Cross-Town Bus Routes as a Solution

for Decentralized Travel: A Cost-Benefit Analysis
for Monterrey, Mexico

M. Baher El-Hifnawi

Development Discussion Paper No. 660

October 1998

© Copyright 1998 M. Baher El-Hifnawi

and President and Fellows of Harvard College

Harvard Institute for

International Development
HARVARD UNIVERSITY
 HIID Development Discussion Paper no. 660

CROSS-TOWN BUS ROUTES AS A SOLUTION FOR

DECENTRALIZED TRAVEL: A COST-BENEFIT ANALYSIS
FOR MONTERREY, MEXICO

M. Baher El-Hifnawi*

Abstract

An urban transportation planning simulation model is used to analyze the viability and
sustainability of 16 cross-town routes introduced in Monterrey, Mexico to accommodate the
increasing decentralized employment and travel patterns. The model shows that the new route
reform results in net time and monetary savings for travelers during the morning peak hour, most
of which accrues to cross-town bus users. One of the main attractive features of the route reform
is the negligible cost associated with its implementation. No new investment in buses has been
made as some buses were withdrawn from the radial routes and redeployed to the cross-town
ones. As a result of the route reform, radial route users face slightly longer trip times;
automobile users were not affected in any significant way and bus operators stand to lose due to
the elimination of full-cost-transfers. The net welfare gain indicates that losers can be
compensated, if necessary.

M. Baher El-Hifnawi is a Project Associate at the Harvard Institute for International

Development and Academic Director of the Fulbright Economics Teaching Program in Vietnam.

*I wish to express my gratitude to Tony Gomez-Ibanez, Glenn Jenkins and John Kain for many comments on and
contributions to previous drafts of this paper. Special thanks to Socrates Rizzo, who as Governor of Nuevo
Leone initiated the project that resulted in this paper. Former Executive Director of Monterrey’s Consejo
Estatal Del Transporte (State Council for Transportation), Carlos Chavarria; Chief of Systems, Eduardo Zaragoza;
and Chief of Projects, Hector Salinas have been extremely helpful with their time and comments. Any errors are
entirely mine.
 HIID Development Discussion Paper no. 660

CROSS-TOWN BUS ROUTES AS A SOLUTION FOR DECENTRALIZED

TRAVEL: A COST-BENEFIT ANALYSIS
FOR MONTERREY, MEXICO

M. Baher El-Hifnawi

Travel patterns have become more decentralized as employment has dispersed throughout
metropolitan areas and has become less highly concentrated in central business districts (CBD).
Decentralized travel patterns have lead several transit authorities to complement traditional radial
networks, where all transit routes radiate from the CBD, with cross-town transit routes from one
suburb to another. The main characteristic of cross-town routes is that neither end of the bus line
originates in the CBD. Peripheral cross-town routes totally bypass the CBD.

The objective of this article is to use an urban transportation planning (UTP) model to
simulate the productivity gains or losses in the transit system in Monterrey, Mexico following
the introduction of 16 new cross-town routes. The study analyzes the financial productivity of
the transit lines as well as their economic viability by measuring welfare gains and losses to bus
users, automobile users and bus operators. Monterrey, capital city of the state of Nuevo Leone in
Mexico, is used as a case study because it is typical of a large number of developing cities in
many respects, particularly, the high rates of growth in population and auto ownership, and the
limited road capacity. Tripmakers, particularly those traveling to the CBD or passing through it,
continue to face high and increasing levels of congestion.

The following section provides a background of the bus routes in Monterrey both prior to
the introduction of the cross-town routes and after their introduction. Section II provides a
conceptual discussion of the nature of gains and losses associated with the new policy. Sections
III, IV and V present the methodology used, the simulation results and the conclusions
respectively.
 I) MONTERREY’S PUBLIC TRANSPORT SYSTEM1

Radial System (prior to introduction of cross-town bus routes)

Monterrey’s bus system consisted mainly of 114 bus routes prior to the introduction of
the 16 cross-town routes. Almost all the old routes were CBD-oriented but they were not
exclusively of the typical radial kind that has one end of the line in the CBD. A few routes could
be classified as cross-town routes in that they intersected, or totally bypassed, the CBD. With
the exception of the few routes that offered cross-town travel, under the radial system, a
passenger wishing to travel from one outlying area to another had to transfer at, or close to, the
CBD at full cost.

The 114 routes, constituting the Radial network, can be broadly divided into four
categories. Category I consists of pure radial lines that have one end in the CBD, and Category
II includes downtown routes that are entirely limited to the CBD. Categories III and IV may be
classified as cross-town routes. Category III consists of cross-town lines passing through the
CBD but having neither line end in it, and Category IV includes cross-town routes that do not
pass through the CBD. Table 1 presents the number of routes, line frequencies and lengths for
each of the four categories.

1
Data for this section were obtained through interviews with members of the Monterrey State
Council for Transportation; MSCT (Consejo Estatal Del Transporte) between 1993 and 1996 and from
two reports by MSCT: Consejo Estatal Del Transporte (1993), and (1992).

2
 Table 1
Characteristics of Radial Bus System in Monterrey

Route Type No. of bus Frequency Roundtrip Distance

routes (minutes) (km)
min max avg min max avg total

I) Pure Radial: One of line ends lies in CBD

East/CBD 27 3.3 30.0 7.9 14 49 32 824

West/CBD 14 1.8 14.3 6.6 18 48 38 505

North/CBD 16 2.9 6.6 6.9 17 43 31 496

South/CBD 8 3.6 6.9 6.2 19 31 24 193

Northeast/CBD 15 2.7 6.2 7.2 16 50 33 494

Northwest/CBD 13 1.4 7.2 3.5 23 40 29 382

Entire Category I 93 1.4 30.0 6.6 14 50 32 2,895

II) Downtown: Both line ends are in CBD

CBD Border 2 2.9 10.6 7.8 8 12 10 20

III) Cross-Town: Line passes through CBD

North/South 10 2.5 8.3 4.4 17 35 25 254

East/West 5 1.9 5.4 3.3 30 41 36 178

Entire Category III 15 1.9 8.3 4.0 17 41 29 432

IV) Peripheral Cross-Town: Line does not pass through CBD

East/West 3 2.3 6.7 4.2 38 43 41 124

North/ South 1 4.0 4.0 4.0 55 55 55 55

Entire Category IV 4 2.3 6.7 4.1 38 55 45 179

ALL TRANSIT LINES 114 1.4 30 6.2 8 55 31 3,526

Source: Spreadsheets from Monterrey State Council for Transportation (MSCT).

3
 Cross-Town Routes
In January 1993, the Monterrey State Council for Transportation (MSCT) introduced 16
cross-town bus routes to accommodate the increasing demand for suburban travel. Buses
required to serve the cross-town routes were obtained from the radial routes. In other words,
there was no new investment in buses to accommodate the new service and its introduction only
resulted in a reallocation of buses among all routes. Consequently, the total number of operating
hours by buses and the total operating costs did not change as a result of the new service.

Figure 1
Schematic Representation of Cross-Town Routes
with respect to CBD

Category D

Category C

Category B

Category A

CBD

The new routes can be classified on the basis of their orientation to the CBD into four
categories as shown in Table 2 and Figure 1. Category A includes pure circumferential routes:
these routes travel close to the periphery of the city with route ends on one side of the CBD and
the entire route alignment relatively far from the CBD. Category B consists of routes that cross
from one side of the CBD to another, but do not pass through the CBD.

4
 Table 2
Characteristics of Cross-Town Bus Routes in Monterrey

Route Frequency Roundtrip

Route Orientation1 Number (minutes) distance (km)
A) Pure circumferential routes
South/West 16S 6.5 47
Northwest/West 21S 6.2 53
Northwest/Northeast 24S 6.0 47
Northwest/East 34S 7.7 62
North/East 53S 6.0 60
Northeast/East 45S 14 46
Entire Category A (averages) 7.7 53

B) Routes going past, but not entering, CBD

North/West 13S 9.6 71
Northwest/East 52S 7.3 47
Entire Category B (averages) 8.5 59

C) Routes passing through, or on the border of, CBD

East/West-1 14S 5.9 67
East/West-2 51S 8.0 66
East/South 46S 8.1 57
Northwest/South 26S 5.3 50
North/South 36S 8.0 57
Entire Category C (averages) 7.1 59

D) CBD-oriented routes2
Northwest/CBD 32S 10 55
North/CBD 47S 26 35
Northeast/CBD 48S 25 42
Entire Category D (averages) 20.3 44

ALL CROSS-TOWN ROUTES 10.0 54

1. Directions are with respect to CBD. North/South means a route traveling between points
south and north of CBD.

2. Routes tending toward the CBD but stopping before reaching it.

Sources: Consejo Estatal Del Transporte (1993).

Spreadsheets provided by MSCT Chief of Systems.

5
Category C includes routes that pass through, or on the border of, the CBD, while Category D
includes transit lines that have a radial orientation, tending toward the CBD but stopping short of
it. Figure 1 provides a schematic representation of the four route categories, and Table 2
presents frequencies and lengths for the cross-town routes.

Most of the cross-town routes are substantially longer than the average route in the Radial
network due, in large part, to the inherent nature of cross-town travel which typically entails
covering longer distances. The long length is also because many of these lines involve more
travel within the suburban towns than radial routes in order to attract more passengers. The
average length of a cross-town route is 54 km compared to 31 km in the Radial network.

II) WHO GAINS AND WHO LOSES

The main benefits of the new cross-town routes are measured by the time savings to bus
users (and to auto users, if any), as well as any resource savings in operating buses and autos.
Net benefits can be estimated in two ways: either by adding up the value of net time savings and
net resources saved, or by adding up the value of all time and monetary savings and losses to all
involved stakeholders. This paper follows the second approach. To determine the benefits of the
new routes, one to has to first understand the nature of the new policy. An important feature of
this policy is that it hardly involves any change in bus operating costs. Since buses were taken
away from some routes and redeployed to new routes, the total number of bus operating hours
remained largely unchanged, and so did the resources used in operating the buses.

The introduction of the new routes is expected to decrease cross-town travel time by bus.
Cross-town travelers who used the bus prior to the introduction of the new routes will enjoy both
monetary and time savings. The reduction in bus travel time and cost for cross-town travel is
also likely to divert some auto travelers from auto to bus for cross-town travel. It is important to
realize that these ‘diverted’ cross-town bus travelers may face longer trip times when traveling
by bus, but will incur lower money costs and enjoy an overall net gain. Cross-town bus users are
divided for this analysis into ‘original’ users and ‘diverted’ users. Cross-town auto travelers may

6
 face slightly longer travel times and monetary cost if the introduction of the new routes
increases congestion and slows down cross-town travel.

The reduction in the number of buses on the radial routes could result in either time
savings or losses to the users of these routes. While the reduction in the number of buses will
result in an increase in average wait time, it could also result in lower levels of congestion and
faster travel towards, and away from, the CBD. Whether there are monetary gains to radial bus
users or not depends on whether the average number of transfers (all of which are at full cost)
made by radial bus users is reduced or not.2 For this analysis, radial bus users are divided into
two categories. The first category includes travelers who use radial bus routes both before and
after the new policy. The second category will depend on whether radial bus users enjoy a net
gain or a net loss as a result of the route reform. If they enjoy a net gain, the category will be
radial bus users diverted from auto to bus; and if they suffer a net loss, the category will be radial
bus users diverted from bus to auto. As far as radial auto users are concerned, there may be
some time and monetary savings if the direct cross-town travel results in less congestion in the
down town and faster trips to and from the CBD.

Table 3 shows the different stakeholders and how they may be impacted by the new
policy. Since this study is concerned with the economic viability and financial sustainability of
the new policy and its impact on bus users in particular, auto users are aggregated into one
category. Bus operators stand to lose from the new policy due to the reduction in the number of
boardings which results in a direct loss in farebox revenues without any corresponding savings in
operating costs. The decline in farebox revenues due to the reduction in transfers is slightly
offset by the diversion of some cross-town travelers from auto to bus.

2
The broadness or narrowness of definition of radial users will have an impact on the average
number of transfers. The definition is given in Section IV.

7
 Table 3
Winners and Losers of the New Policy

Cross-Town Bus Users Radial Bus Users Auto Bus

Original1 Diverted Non- Diverted Users Operators
from auto2 diverted3 to/from auto4
Time Savings (+) (-) (?) (?) (?) na

Monetary (+) (+) (?) (?) (?) (-)

Savings
Net Impact (+) (?) (?) (?) (?) (-)

(+) represents a gain; (-) represents a loss; (?) result is indeterminate; na: not applicable
1. Cross-town travelers using bus both before and after the new policy.
2. Cross-town travelers diverted from auto to bus as a result of the new policy.
3. Radial travelers using bus both before and after the new policy.
4. Radial travelers diverted to auto (or from auto) as a result of the new policy.

The economic viability of the new routes will be determined by adding up the gains and
losses to bus users, auto users and bus operators. However, even if the policy is economically
viable, another question remains regarding the financial sustainability of the new routes and the
system as a whole. Bus operators will lose as a result of the new policy as explained above.
While this loss is a mere transfer from bus operators to bus users and does not constitute any net
economic loss or gain, one needs to determine how the bus system will be financed following the
loss in revenues. If bus operators are currently earning economic rents, then it may be possible
to maintain fares at their current levels. If however, operators are only covering costs, then a
different fare structure and/or subsidy will need to be considered for the financial sustainability
of the system.

8
 III) METHODOLOGY
UTP models were initially developed for the assessment of comprehensive highway
plans, and later were used to assess new transit systems or combined highway and transit systems
as well. More recently, UTP models are becoming more popular for analyzing small-scale
transportation policies, such as the introduction of changes to public transportation networks (see
for example: Kain and Fauth, 1979; Kerin, 1990; and Liu, 1993). UTP models can be very
useful in appraising the impacts of such policies because they provide detailed simulations of
ridership on each route on the network for a given land-use pattern and transportation system.

Different Approaches for Evaluating Route Changes

Liu (1993) states that the impacts of new transit routes on ridership levels, route
productivity, and trip times are more commonly based on observations made before and after
implementation. There are two main problems with this approach. The first problem is a
practical one. It is simply not possible to obtain a clear understanding of the magnitude of the
impacts of the new service before its introduction. This might not be a problem if the service
being introduced has no disrupting impacts such as increased congestion in certain parts of town.
If however, the new service has some negative impacts, one could argue that a simulation of the
service may have pointed to these negative impacts, and that it may have been possible to avoid
them or mitigate them.

The second problem is a methodological one. The before-and-after results will only be
reliable if there have been no changes in any of the other service or socioeconomic variables that
affect travel behavior. But as it is common for new services to be accompanied by other service
cuts or changes in the tariff structure, the before-and-after estimates are not always adequate.
The impacts of changes in other variables in both the short and long run need to be isolated so
that the true impact of the new service could be determined.

The use of urban transportation planning simulation models would solve both of these
problems. The models can be used to vary one variable at a time while holding all other
variables constant, thus allowing for the analysis of a single change in a policy variable even
when several changes are introduced simultaneously. However, UTP simulation models are not

9
 without their problems. Building an accurate UTP model is fairly complicated and time-
consuming and one needs to keep in mind that results of these models are only simulations.

UTP Model Components used in Analysis

It is necessary to develop full-blown UTP models to analyze the impacts of large-scale
transportation investments or large urban development projects on land use patterns, trip
generation and distribution rates, modal choice and route choice. It might not be necessary,
however, to use all of the model’s components when analyzing certain transportation policies
with limited impacts such as the introduction of a new bus route or the modification of an
existing one. Figure 2 presents a conventional UTP model and highlights the model components
used for the analysis of the cross-town routes in this study. These components are trip
distribution matrices, a modal split model for the journey to work and a route assignment model.

10
 Figure 2
Conventional UTP Model highlighting
Components used in Analysis of Cross-Town Routes

Inventory and Data Collection

Socioeconomic Data
Transportation Land Use
Employment Data
System Data Data
Demographic Data

Land Use Forecasts

Submodel

Trip Generation Submodel

Trip Distribution Submodel

Modal Split Submodel

Travel Impedance by
mode (time & cost)
Route Assignment
Submodel

Enclosed within hatched box (above) are model components used in analysis

11
 Trip Distribution: In building a UTP model for Monterrey, the city was divided into 293
zones. Trip distribution matrices include trips made between every zone pair. The matrices used
in this study are for morning peak-hour passenger trips made by public bus or private
automobile.

Modal Split: Ideally one should use a different mode split model for each trip purpose: work
trips, school trips, shopping trips, social and recreational trips, etc. However, only a mode split
model for the journey to work developed in El-Hifnawi (1996) was used in this study.
Inaccurate reporting of non-work trips in household surveys precluded the development of mode
split models for other trip purposes. Two modes were used in estimating the mode split model:
private automobile and public bus. The mode split model was used to determine trips made by
private auto and trips made by public bus for all zone pairs.

Route Assignment: The Monterrey street network was coded and used with the computer
software program, EMME/2 (INRO consultants, 1992). EMME/2 determined the route
assignment for all bus and automobile trips. EMME/2 uses algorithms that determine an
equilibrium traffic assignment for automobiles based on shortest travel time. Upon reaching the
equilibrium assignment, no user can improve her travel time by selecting another route. Transit
assignment is also predicated on the behavioral assumption that the traveler seeks to minimize
generalized travel cost, which includes walk, wait and in-vehicle times in addition to bus fare.

Using an initial set of travel times, route choice for all zone pairs was determined using
EMME/2. A new set of travel times was determined on the basis of the route choice. The new
times were fed back into the mode split model to determine a new bus:auto split for each zone
pair. The process was repeated iteratively until an equilibrium was reached where there was very
little change in travel times from one run to another.

Underestimation of Benefits of New Routes

The model used in this study is likely to result in an underestimation of benefits on three
different accounts. First, the distribution matrices are for trips by private auto and public bus
only. These two modes account for 88% of all motorized linked trips during the morning peak

12
 hour in Monterrey. By ignoring 12% of the trips, we underestimate congestion levels and
possibly time savings following the introduction of the new routes.

Table 4
Distribution of Morning Peak-Hour Automobile and Bus Linked Trips
by Trip Purpose and Mode in Monterrey, 1993

Work School Other Purposes Total

Automobile 40% 9% 3.5% 52.5%
Bus 35% 9% 3.5% 47.5%
Total 75% 18% 7% 100%

Source: The 1993 household survey conducted by MSCT.

The second source of underestimation is the assumption that only work trips will be
affected by the introduction of the new routes. Table 4 (corresponding to the trip distribution
matrices) shows that 25% of linked trips made by automobile or bus are non-work trips; and of
these 50% are made by automobile and 50% are made by bus. As it was not possible to estimate
mode split models for non-work trips, the percentages above were taken to be constant for all
zone pairs. For example, if there were 100 motorized linked trips made by private automobile or
public bus from zone 1 to zone 2, 75 trips were taken to be work trips and 25 to be non-work
trips. Work trips were distributed using the modal split model for the journey to work. Out of
the 25 non-work trips, on average 12.5 trips were assumed to be made by automobile and 12.5
trips were assumed to be made by public bus both before and after the introduction of the cross-
town routes. One could argue that the faster and cheaper cross-town travel on the cross-town
routes might result in an increase in the share of non-work trips made by bus. Consequently one
has some reason to believe that benefits simulated using constant shares might be somewhat
understated. Since 72% of non-work trips are school trips (18% of total trips), it is plausible to
assume that a large proportion of these trips will not involve cross-town travel as the majority of
students tend to go to schools relatively close to their homes. That being the case, one should not

13
 expect a large change in the modal shares for school trips following the introduction of the
cross-town routes.

Thirdly, not using a trip distribution model is likely to result in an underestimation of

benefits as this would imply that a tripmaker can not change his trip origin or destination as a
result of the introduction of the new routes. The main beneficiaries of the new routes can be
classified into two groups. The first group consists of cross-town travelers who did not change
residence or workplace location following the introduction of the new routes. The second group
consists of cross-town travelers who were induced to change residence or workplace location and
use these routes to capture some of the gains in travel time and cost. Without a trip distribution
model, benefits accruing only to the first group of beneficiaries are captured.

To illustrate numerically, suppose there are 100 tripmakers whose trips, prior to the
introduction of the cross-town routes, originate and end in suburbs. Also, suppose that following
the introduction of the routes, each tripmaker saves 5 minutes for a total gain of 500 minutes.
This gain would be captured by the model in this study. But if the introduction of the new routes
results in a different trip distribution pattern with 105 tripmakers using cross-town routes, the
total gain would now exceed 500 minutes (see Figure 3). The additional gain resulting from the
5 induced tripmakers will not be captured without a trip distribution submodel, and benefits of
the new routes will, as a result, be underestimated.

14
 Figure 3
Benefits of Cross-Town Routes

Trip time and cost

Benefits to
original
cross-town
travelers

x Benefits to
induced
travelers
x-5

100 105 Linked trips

by all modes

A plausible assumption, however, is that the most significant impact of the cross-town
routes is likely to be on the choice of route and mode, and that the new policy is unlikely to
significantly alter land use, trip generation or trip distribution. One can reasonably argue that the
likelihood that travelers will change place of residence or employment as easily as they would
change route or mode following the introduction of the cross-town routes is small because time
and monetary gains from the new routes are unlikely to be large enough to offset the relatively
larger costs associated with changing residence or workplace. One can further argue that the
time and cost savings of the new policy will result in smaller, if any, changes in land use and trip
productions and attractions than in trip distribution.

IV) SIMULATION RESULTS

Two transit networks are simulated: the Radial system (prior to the introduction of the
cross-town routes), and the Hybrid system (following the introduction of the cross-town routes).
In some analyses, bus users were divided into radial users and cross-town users. Radial users
were defined as tripmakers with the trip origin or destination in the CBD. All other tripmakers
were in the cross-town group. Needless to say, some of the passengers classified as cross-town

15
 travelers travel on radial routes. For instance, a passenger traveling towards the CBD but
disembarking before reaching it would be classified as a cross-town traveler even if he was
traveling on a radial route. This may help explain the large number of trips classified as cross-
town trips in Table 6.

Table 5
Simulation Results of Network Service Levels for
the Morning Peak Hour in Monterrey

Radial Hybrid Hybrid -

Network Network Radial
Linked trips 293,900 294,100 200
Revenue hours1 2,890 2,885 (5)
Linked trips per revenue hour 102 102 0
Boardings 418,800 392,500 (26,300)
Transfers 124,900 98,400 (26,500)
Mean travel speed (km/hour) 18.927 18.943 0.016
Mean trip distance (km) 9.15 9.00 (0.15)

1. Since the number of buses operating during the morning peak hour does not change after the
introduction of the cross-town routes, there should be no change in revenue hours. The resulting
change is an outcome of the simulation process.

Source: Simulations carried out in this study.

Tables 5, 6 and 7 present the simulation results of service levels, mean trip times, and costs and
revenues for the two networks. Buses operating on the cross-town routes were drawn from radial
routes and no new buses were added. Consequently there was no significant change in the total
number of revenue hours following the introduction of the cross-town routes as shown in Table
5. The table also indicates no change in the transit system’s overall productivity as measured by
the number of linked trips per revenue hour. This outcome is expected given the small increase
in the number of linked trips and revenue hours.

16
 Table 6
Simulation Results of Network Trip Times for
the Morning Peak Hour in Monterrey
Radial and Cross-Town Users

(All times in minutes unless otherwise indicated)

Cross-Town Users Radial Users All Users

Radial Hybrid Hybrid - Radial Hybrid Hybrid - Hybrid -
Network Network Radial Network Network Radial Radial
Linked trips 200,409 201,455 1,046 93,491 92,645 (846) 200
Mean in-vehicle 28.27 27.70 (0.57) 30.40 30.30 (0.10) (0.50)
trip time
Mean walk time 12.13 12.10 (0.03) 11.30 11.39 0.09 0.02

Mean wait time 3.00 3.28 0.28 2.19 2.38 0.19 0.25
Mean boarding 2.94 2.69 (0.25) 2.68 2.62 (0.06) (0.18)
time
Unweighted 46.34 45.77 (0.57) 46.57 46.69 0.12 (0.43)
mean trip time
Weighted mean 81.52 80.93 (0.59) 78.25 78.76 0.51 (0.31)
trip time1
Boardings per 1.47 1.35 (0.12) 1.34 1.31 (0.03) (0.09)
linked trip

1. Weights for the wait, walk, and boarding times are 2.7, 3.0 and 3.0 respectively. Initial weights for
walk and wait times were obtained from the mode split model and were later fine-tuned in the calibration
process of the UTP model (El-Hifnawi, 1996). Boarding was treated the same as walking, and was given
a weight of 3.0.

Source: Simulations carried out in this study.

Table 6 shows that the introduction of the cross-town routes resulted in a small reduction
(0.31 minutes) in the weighted mean travel time by bus during the morning peak hour.3 The new

3
The close values of total trip time for cross-town and radial travel by bus can be explained by
the arbitrariness in the classification of cross-town and radial travel. The broad definition of cross-town
travel includes many short trips that were hardly affected by the route reform. These trips lowered the
average trip time for cross-town travelers.

17
 service also lead to an increase of 1,046 linked trips on the cross-town routes and a decrease of
846 linked trips on the radial routes, a net increase of 200 linked trips. The small increase in
linked trips (0.06%) is consistent with the small reduction in travel time (0.38%) and the
typically low elasticities of travel time.

The new service resulted in a mean reduction in in-vehicle and boarding times while
causing an increase in mean wait time. Mean walk time per linked trip hardly changed. The
reduction in in-vehicle time (30 seconds) accompanying the new routes was due to the small
increase in mean travel speed; and the reduction in boarding time per linked trip (11 seconds)
was due to the decline in the number of boardings per linked trip. The slight increase in mean
wait time (15 seconds) is because the introduction of the new routes was accompanied by a
reduction in frequencies for buses operating on the radial routes. (Recall that the buses operating
on the cross-town routes were drawn from the radial routes.) The reduction in frequencies
resulted in longer wait times for buses on the radial routes.

The reduction in mean travel time due to the more direct travel on the cross-town routes
caused some cross-town travelers to shift from auto to bus. Furthermore, the reduction in the
number of boardings lead to a decrease in the money cost per linked trip leading to a further
increase in the number of bus linked trips on the cross-town routes. The longer wait times on the
radial routes caused some travelers to shift from bus to auto for radial travel but this was partially
offset by the reduction in in-vehicle and boarding times per linked trip on the radial routes, as
shown in Table 6.

18
 Table 7
Simulation Results of Network Operating Costs and Revenues for the
Morning Peak Hour in Monterrey

Radial Hybrid Hybrid -

Network Network Radial

Linked trips 293,900 294,100 200

Total operating costs1 (US$) 33,325 33,178 (147)
Operating costs per linked trip (U.S. cents) 11.3 11.3 0
Boardings 418,800 392,500 (26,300)
Total farebox revenues2 (US$) 94,042 88,136 (5,906)
Revenue per linked trip (U.S. cents) 32.0 30.0 (2.0)
Surplus per linked trip (U.S. cents) 20.7 18.7 (2.0)

1. Operating costs are based on an the average cost per hour reported by MSCT (US$11.50). Since
the number of buses operating during the morning peak hour did not change after the introduction
of the cross-town routes, there should be no significant change in operating costs. The resulting
change is an outcome of the simulation process.

2. Revenues are based on the fixed fare of 0.75 new pesos and an exchange rate of 3.34 new pesos
per U.S. dollar.

Source: Simulations carried out in this study.

Table 7 shows no significant change in total operating costs or in the overall efficiency of
the system as measured by the operating cost per linked trip. There was a decrease in daily
farebox revenues from US$94,042 in the Radial system to US$88,136 in the Hybrid system as a
result of the new service. This decrease represents a loss to bus operators of 2.0 U.S. cents per
linked trip during the morning peak hour. Operating surplus (farebox revenues less operating
costs) dropped by about 10% from US$60,717 with the Radial network to US$54,958 with the
Hybrid network. This loss to bus operators is simply a financial transfer to bus users and will be
netted out of their gain when estimating the net benefits resulting from the introduction of the
new routes. According to MSCT, the loss is unlikely to affect the financial viability of the transit

19
 system because the bus routes are highly profitable, and the farebox revenue per linked trip is
about three times the operating cost per linked trip.4

An interesting situation arises if bus operators are earning no excess rents prior to the
introduction of the service and are expected to lose once it is introduced if the fare is not raised
or if the system is not subsidized. A subsidy seems highly unlikely for a number of reasons.
First, bus trips make up a large share of daily trips in most developing countries. Subsidizing the
transit system would place tremendous pressure on the city’s scarce resources, and it is highly
unlikely that a subsidized system would be sustainable in the long run. Second, subsidies may
create inefficiencies with some bus passengers valuing the service at less than what is costs to
provide it. Third, providing a subsidy might create incentives for the inefficient operation of the
transit system (Pickrell, 1985).

The alternative to subsidizing would be raising the fare. This requires a careful analysis
of the entire transit system to determine the operating costs on the different routes (or groups of
routes), the congestion costs associated with the radial and cross-town services, and the
elasticities and cross elasticities of auto and bus shares with respect to fare attributes.

Welfare Analysis
From an economic standpoint, it is necessary to analyze the impact of introducing the
routes on the three main groups affected by the new service--bus passengers, automobile users
and bus operators--to determine whether there is a net welfare gain or not. Table 8 presents a
summary of the gains and losses to bus users, auto users and bus operators.

Bus Users: Cross-town bus users are clear gainers as a result of the new routes. Each day,
cross-town travelers gain a total of US$8,872 in time and cost savings during the morning peak
hour (see Figure 4A). Table 6 shows that passengers on cross-town routes gain on average 35
seconds in weighted trip time per linked trip. Most of the gain accrues to the original cross-town

4
The high profitability of the system was confirmed by MSCT Executive Director (in 1993) and
its Chief of Systems.

20
 users who traveled by bus prior to the introduction of the new routes. This group of travelers
enjoys monetary savings (US$5,400) resulting from eliminated transfers as well as time savings
(US$3,449) due to the reduction in trip time. The 1,046 cross-town travelers who switched from
auto to bus perceive a small net benefit of US$23 per day during the morning peak hour.
Although these travelers are likely to face longer trip times after switching from auto to bus, the
savings in cost outweigh the value of time lost.

Figure 4A
Savings to Cross-Town Bus Travelers

Trip Cost Value of time savings accruing to original cross-

town travelers. (US$3,449=1.72¢/100* 200,409)

Cost savings to original cross-town travelers.

(US$5,400=2.70¢/100*200,409)
Average value of time
savings per trip C0 Net savings to travelers diverted from
(1.72 US cents) auto. (US$23={1.72¢+2.70¢}/100*
1,046 *0.5)
Average cost saving per
C1
trip (2.70 US cents)

Linked Trips
200,409 201,455

• Co and C1 are average trip cost before and after the route reform respectively.
• Average value of time savings per trip is estimated as follows:
= (Difference in weighted mean trip time before and after route reform) * value of travel time.
= (0.59 minutes)/60 {Table 6} * 1.75 $/hour (for value of travel time, see El-Hifnawi, 1996)
= US$0.0172.
• Average cost saving per trip is estimated as follows:
= Difference in number of boardings per linked trip before and after route reform {Table 6}* fare
= 0.12 boardings * 0.75 Mexican new pesos/3.34 Mexican new pesos per US$ = US$0.0270.

21
 Among cross-town route passengers, the largest beneficiaries are the users of Category
A routes (pure circumferential routes). An analysis of simulated travel times between zones
along the path of one of these routes indicates that savings in in-vehicle travel time along the
route were as high as 25 minutes. Moreover, these time savings are accompanied by cost savings
as result of the reduction in the average number of bus rides per linked trip.

Radial bus users lose in that some will now face longer wait times and experience an
overall increase in trip time. Table 6 shows that passengers on the radial routes lose an average
of 31 seconds as the longer wait and walk times outweigh the benefits of the shorter in-vehicle
times. Some of the time losses are offset, however, by cost savings due to eliminated transfers
by radial bus users (see Figure 4B). Most of the losses are borne by radial travelers who have
not switched to the auto mode after the introduction of the new routes. Each day, during the
morning peak hour, this group faces time losses valued at US$1,380, but enjoys cost savings
worth US$624 for a net loss of US$756. The longer trip times have lead 846 travelers to shift
from bus to auto for radial travel. The daily net loss to this group of diverted travelers during the
morning peak hour is quite low at US$3.5.

Monetary savings accruing to bus passengers represent approximately 74% of total

savings to these users with time savings representing the balance. If one were to take a
conservative approach and assume that the gains in the transit system occur during morning and
evening peak hours only, the total value of time savings accruing to bus passengers as a result of
the introduction of the cross-town routes would be about US$1.076 million per year (value of
time saved by cross-town travelers less value of time lost by radial travelers: [US$3,449-
US$1,380] * 2 peak periods * 260 days).

22
 Figure 4B
Losses/Savings to Radial Route Bus Travelers

Trip Cost
Value of time lost by radial travelers who still use
bus after the introduction of the new routes [abec1].
(US$1,380=1.49¢/100*92,645)

Cost savings to radial travelers who still use

bus after introduction of new routes [abdc0]
Average value of time (US$624= 0.67/¢100*92,645)
lost per trip c1 e
(1.49 US cents) Net loss to travelers diverted to auto.
c0 d (US$3.5={1.49¢-0.67¢}/100*846
*0.5)
Average cost saving per a b
trip (0.67 US cents)
Linked Trips
92,645 93,491

• co and c1 are average trip cost before and after the route reform respectively.
• Average value of time lost per trip is estimated as follows:
(Difference in weighted mean trip time before and after route reform) * value of travel time.
= (0.51 minutes)/60 {Table 6} * 1.75 $/hour (for value of travel time, see El-Hifnawi, 1996)
= US$0.0149.
• Average cost saving per trip is estimated as follows:
= Difference in number of boardings per linked trip before and after route reform {Table 6}* fare
= 0.03 boardings * 0.75 Mexican new pesos/3.34 Mexican new pesos per US$ = US$0.0067.

23
 Table 8
Simulated Distributional Impacts of Cross-Town Routes
for the Morning Peak Hour each Weekday in Monterrey

(All value are expressed in 1993 US dollars)

I) Bus Passengers
A) Original Cross-Town Travelers
1. Monetary savings due to eliminated transfers1 5,400
2. Monetary value of time savings1 3,449
B) Cross-Town Travelers Diverted from Autos
Net savings1 23.0
C) Remaining Radial Travelers
1. Monetary savings due to eliminated transfers2 624
2
2. Monetary value of time lost (1,380)
D) Radial Travelers Diverted to Autos
Net losses2 (3.5)
Total savings accruing to bus passengers 8,112.5

II) Auto Users

1. Savings in vehicle operating costs3 339
2. Monetary value of time savings4 275
Total gains to auto users 614

III) Bus Operators

1. Monetary loss due to eliminated transfers5 (5,906)
2. Change in operating costs6 0
Total losses to bus operators (5,906)

NET SAVINGS 2,820.5

1. See Figure 4A.
2. See Figure 4B
3. Number of auto linked trips * time saved per trip * operating cost per km * velocity:
314,700 * (0.03/60) hours * US$0.09 per km * 24 km/hour = US$339
(operating cost was obtained from MSCT: US$ 0.045 pays for gasoline and US$0.045 pays for
maintenance.)
4. Number of auto linked trips * time saved per trip * value of time per hour:
314,700 * (0.03/60) hours * US$1.75 = US$275.
5. See Table 7.
6. There should be no significant change in operating costs because there is no change in the number of
buses operating during the morning peak hour. The value (US$147) in Table 7 is an outcome of the
simulation process.

24
 Auto Users: The results of the simulation indicate that the average trip time by auto decreases
by a negligible 2 seconds following the introduction of the new routes from 24.29 minutes to
24.26 minutes. The total gain in time savings to auto users during the morning peak hour is 157
hours valued at US$275 (see Table 8). The reduction in operating costs resulting from the
slightly faster trips is estimated at US$339. This brings the total gain to auto users during the
morning peak hour to US$614 which is about 7.5% of the total gain to bus users.

Bus Operators: Based on the simulation results in Table 7, 26,300 boardings are eliminated
during the morning peak hour costing the operators US$5,906 in lost farebox revenues during
this period alone. There will be no significant change in bus operating costs because no
additional buses will be used for the new service. The loss in farebox revenues to bus operators
is slightly less than the total savings to bus users from eliminated transfers because the new
service results in a diversion of trips from auto to bus and a subsequent increase in bus ridership.

Adding the costs and benefits of all involved parties, and keeping in mind that this analysis is
based on morning peak-hour travel only, the introduction of the cross-town routes results in a net
gain. The total gains accruing to bus passengers and auto users exceed the losses to bus
operators by about US$2,820 for the peak hour per weekday or US$733,200 per year.

A comprehensive study would include similar analyses for off-peak weekday travel and for
weekend travel to obtain a more complete picture regarding the economic viability and financial
attractiveness of the new routes. However the lack of data did not make it possible to conduct
these analyses; and hence, the gains associated with morning peak-hour travel should be treated
as a fraction of total daily benefits. This is because it is reasonable to expect passengers to enjoy
time and monetary savings during other peak periods and off-peak periods as well due to the new
service. Moreover, even if the operation of the new routes proved to be economically and
financially desirable only during peak periods, the new service would still be desirable because
the new routes could be operated as designed during peak periods; and the transit network could
be designed differently for off-peak periods.

25
 V) CONCLUSIONS
The analysis of morning peak-hour travel indicates that the introduction of the cross-town
routes in Monterrey, Mexico has resulted in time and monetary savings for their users.
Redeploying buses from radial routes to cross-town ones reduced congestion and resulted in
slightly higher travel speeds. As a result, automobile users enjoy a reduction in vehicle operating
costs and average trip time. The main losers are bus operators who forego the revenues on
eliminated transfers without any corresponding reduction in operating costs. Many radial route
users also lose because of the reduced frequencies of buses on the radial routes and the
subsequent longer wait times.

The net gain resulting from the new service indicates that the losers can be compensated
if necessary. In the case of Monterrey where the concessionaires running the bus companies
enjoy economic rents, compensation may not be needed. In the more common cases where bus
companies are not earning economic rents, a redistribution of the gains may be required for the
sustainability of the new service. In such cases, a differentiated tariff that takes into account both
operating and congestion costs may be desirable.

There are large variations in productivity and the profitability among the cross-town
routes, perhaps suggesting that the design of some of them can be improved. A final decision on
whether to modify, or perhaps even eliminate, a route will have to be based on the overall
financial and economic impacts of the changes on the entire network. Routes operating close to
the periphery of the city and relatively far from the CBD have high levels of ridership and
productivity and their users enjoy the largest time savings.

The overall success of these cross-town routes, as exhibited by their financial and
economic viability, suggests that a policy of introducing such route even if it means cutting back
on the radial routes should be considered more in developing cities. Given that most cities are
likely to have, or should have, some form of a UTP model, the impacts of such a new service can
be analyzed with relative ease. This policy appears even more attractive once one realizes the
negligible investment levels associated with its implementation.

26
 REFERENCES

El-Hifnawi, M. Baher (1996). “Transportation Policies in Developing Cities, The Case of

Monterrey, Mexico.” Ph.D. Diss. Department of Urban Planning, Harvard University.

EMME/2 (1992). Computer software. Software release 6.0. Inro Consultants Inc. Montreal,
Canada.

INRO Consultants Inc. (1992). EMME/2 User’s Manual. Montreal, Canada.

Kain, John and Gary Fauth (1979). “Increasing the Productivity of Urban Expressways:
Combining TSM Techniques and Transit Improvements. Research Report 79-1. Harvard
University.

Kerin, Paul D. (1990). “Efficient Transit Management Strategies and Public Policies: Radial
Commuter Arteries.” Diss. Harvard University.

Liu, Zhi (1993). "Determinants of Public Transit Ridership: Analysis of Post World War II
Trends and Evaluation of Alternative Networks." Ph.D. Diss. Department of Urban Planning,
Harvard University.

Monterrey, Mexico. Consejo Estatal Del Transporte (1993). Sistema de Rutas Perificas, Bulletin
No.9. Monterrey, Nuevo Leone, Mexico. September.

Monterrey, Mexico. Consejo Estatal Del Transporte (1992). Sistema Integral de Transporte
Colectivo para el Area Metropolitana de Monterrey. Monterrey, Nuevo Leone, Mexico. April.

Pickrell, Don H. (1985). “Rising Deficits and the Use of Transit Subsidies.” Journal of
Transport Economics and Policy, September.

27