ABSTRACT

Road traffic speed limits are suggested to be associated with, e.g.,

changes in travel times,
vehicle operating costs, accidents, noise and emissions. In this paper we
analyze the
impacts of speed limit policies, i.e. restricting the maximum permissible
road traffic speed,
on an urban economy. While most existing studies do only focus on the
effects of speed
limits on frequency and severity of accidents, we provide a more general
assessment of
speed limit policies by employing a spatial computable general
equilibrium model calibrated
to an average Kulim city area. It is shown that besides transport related
effects additional economic effects may influence the overall performance
of speed limit
policies significantly. Driven by spatial economic effects, tightening speed
limits on all
roads, i.e. setting a general urban speed limit of, e.g. 30 km/h, lowers
aggregate social welfare, although aggregate environmental and accident
costs decline. However, setting speed limits around the city center only
and not in suburban areas with access to beltways curtails negative
effects on the urban economy and, in the end, may result in overall
welfare
gains. Therefore, our results suggest that implementing a general speed
limit uniformly in
the entire urban area, thus paying no attention to the spatial shape of the
city and its road
network, is likely to be an inadequate measure to enhance social welfare.
However,
restricting speed limits locally, thus focusing on the design of a slow
zone, is essential
and, in the end, is a more promising speed regulation policy having more
likely the chance

to enhance social welfare.

1. Introduction
A major issue surrounding the effects of tightening road traffic speed
limits in urban areas concerns the impacts on mobility
and the environment. Speed limit policies either already implemented or
at least controversially discussed in cities or
countries around the world are suggested to be associated with, e.g.,
changes in travel times, congestion levels, vehicle
operating costs, the frequency and severity of accidents, noise and
emissions of air pollutants and carbon dioxide. Furthermore,
because car drivers seem to overestimate time benefits from speeding at
the expense of higher accident risks (see e.g.Elvik, 2010; Matsuki et al.,
2002), only consider private costs (ignore externalities) by their choice of
driving speed, and are
just inadequately informed on traffic conditions and their consequences,
regulating drivers speed choice may be a useful and essential traffic
managing instrument (see e.g. Archer et al., 2008). The suggested
positive impacts of speed limits have triggered European citizens to form
an initiative called 30 km/h making the streets liveable!.1 The vision
of the initiative is that a car speed of 30 km/h should no longer be limited
to single zones, but shall become the standard speed limit for villages,
towns and cities with local authorities being able to decide on exemptions.
To meet the subsidiarity principle, the local authorities should have the
final decision to set other speed limits on their roads and implement
equivalent alternatives to meet, e.g., environment related goals.
There are extensive research efforts towards the impacts of lowered
automobile travel speed on accidents, CO2 emissions, noise and air
pollution. In particular the relationship between driving speed and the risk
and severity of road crashes has been analyzed and reviewed to a large
extent (see e.g. Aarts and van Schagen, 2006; Aljanahi et al., 1999;
Archer et al., 2008; Baruya and Finch, 1994; BMJ, 2009; Elvik, 2009; Elvik

and Amundsen, 2000; Elvik et al., 2004; Garber and Graham, 1990;
Joksch, 1975; Kloeden et al., 1997, 2001, 2002; Lai et al., 2012; Nilsson,
1982, 2004; OECD/ECMT, 2006; Taylor et al., 2000; Wong et al., 2005).
Some studies figured out an evidence for an exponential function or a
power function between speed and accidents/crash rates. But almost all
studies conclude that the probability of being involved in a crash as well
as the severity of an accident increases with travel speed and that
lowering speeds improves the interaction between different road users.2
Furthermore, there is evidence that increasing speed differences between
vehicles (speed dispersion) increase the crash rate, too. The impact of
speed management policies on CO2 and air pollution emissions are
analyzed in detail as well (see e.g. Baldasano et al., 2010; Dijkema et al.,
2008; Gan et al., 2012; Madireddy et al., 2011; Int Panis et al., 2011,
2006; OECD/ECMT, 2006; Owen, 2005). These studies show that reducing
speed on urban ring highways/beltways significantly reduces emissions.
For local urban roads, however, this picture is less clear.4 Studies
examining the impact of reduced speeds on noise emissions (see e.g.
Amundsen and Klboe, 2005; den Boer and Schroton, 2007; Dora et al.,
2011; Freitas et al., 2012; Gan et al., 2012; Nijland and Van Wee, 2012;
OECD/ECMT, 2006) mainly conclude that lowering speeds reduces noise
emissions,5 but the potential of noise reduction is mainly influenced by
the speed level.
Further studies analyze the impacts on speed choice behavior (see e.g.
berg et al., 1997; Delhomme et al., 2010; Elvik, 2010;Elvik, 2009; Fuller
et al., 2009; Haglund and berg, 2000; Matsuki et al., 2002; Nilsson, 1991;
Schmid Mast et al., 2008; Tarko, 2009). Their main results can be
summarized as follows: first, most drivers choose speed above the limit
because they overestimate time profits as well as they underestimate
rising accident risks from speeding6; second, because drivers experience
social pressure from other road users they choose their speed according
to the speed of others even though the speed is above the limit (berg et
al., 1997); and third, although drivers are aware of the negative impact of
speed on noise and emissions, this knowledge affects the choice of speed
only to a little degree.

Considering the different research topics reveals that analyses regarding

road traffic speed limits mainly focus on environmental effects and
accidents caused by adjustments in transport. However, in addition to
pure transport related effects particularly on accidents or emissions,
speed limits may have various further impacts from an economic
perspective. Because urban areas constitute economies on a local scale,

additional economic effects arising in cities inhabited by workers or

consumers facing the necessity of being mobile, occupied by businesses
and firms which are reliant on commercial traffic, and governed by
local/federal authorities imposing, collecting, and redistributing fees and
taxes could influence the performance of speed limits to a similar degree
or even more. The objective of this paper is therefore to provide a more
general assessment of speed limit policies by examining their overall
impacts on a metropolitan area and its residents. The overall assessment
includes environmental, safety, economic, transport related and spatial
effects. In order to account for several interdependencies between
economic agents (households, firms, the public sector) and their decisions
on urban markets we develop and employ a spatial computable general
equilibrium model (CGE) that takes into account the endogeneity of
location decisions of households and firms, endogenous labor-leisure
choice, traffic congestion, fuel consumption, traffic related CO2 emissions
concerning private and commercial traffic, travel mode and route choice.
In addition, the presence of multiple distortionary taxes allows to account
for feedback effects on endogenous governmental tax revenues. All these
features are essential to study the impacts of speed limits on cities and, to
the best of our knowledge, have never been considered simultaneously in
speed limit policy analyses.
We consider differentiated speed limiting measures, i.e. either restricting
speeds on all roads (local city roads and bypassing beltways) or setting
speed limits on local roads around the city center only. Based on the
effects of reduced travel speeds reported in the literature, environmental
and accident costs can be expected to be reduced. Imposing a speed limit
is likely to make traveling more expensive accounting for time values
which might cause adjustment in individual behavior with respect to travel
demand (trip distance and/or frequency), travel mode and/or route choice.

These behavioral adjustments, in turn, may result in changes in emission

and accident costs. In addition to these travel related effects, changes in
relative prices might be able to drive spatial economic effects. Tightening
speed limits then may also affect leisure demand/labor supply and
associated with it the allocation of individual time budgets towards less
travel activities, income, commodity demand, spatial consumption
possibilities and location decisions of urban residents as well as economic
activity of firms in the city by affecting freight traffic. Moreover, behavioral
changes of residents might have impacts on governmental tax revenue
via tax interaction effect and on rent dividend income of landowners
working via the urban land/housing market. Consequently, additional

spatial economic effects are likely to constitute a countervailing force to

the potential welfare enhancement caused by a reduction in
environmental and accident costs. The spatial CGE simulations carried out
will reveal the importance of these differentiated effects and thus, the
potential of the speed limit policies to enhance social welfare.
The remainder of the paper is organized as follows: Section 2 presents the
main mathematical formulations of the spatial urban general equilibrium
model and describes its main characteristics. In Section 3 we give a
description of the model calibration according to Kulim data. Furthermore,
we present some results of the initial benchmark simulation and their
correlation with empirical and statistical evidence. Section 4 explains the
way speed limits are implemented as well as the scenario design of the
speed limit policies and, subsequently, presents and analyzes the results
of the simulations including sensitivity analyses. Eventually, Section 5
concludes.

2. The spatial urban CGE model

2.1. The general setting
Our calculations are generally based on the spatial urban general
equilibrium model originally developed by Anas and Kim (1996), Anas and
Xu (1999), and Anas and Rhee (2006). Here we employ the modified
version of Tscharaktschiew and Hirte (2010a,b, 2012) extended, first, by
the transportation part of the RELU-TRAN8 algorithm described in Anas
and Liu (2007)9 and, second, by the explicit consideration of
commercial/business travel. The model explicitly takes into account the
interactions between different markets (land, labor, commodities),
households and firms. All location decisions of households and firms are
endogenously determined. Households are characterized by idiosyncratic
tastes for locations within the urban area, for available travel modes and
for certain routes on the urban road network. As a result, the decisions of
households create mixed land use and various possible travel patterns.
This allows e.g. the reflection of employment decentralization that
characterizes todays urban areas (see e.g. Anas et al., 1998). Fig. 1
shows the spatial shape of the urban area, the spatial decision structure of
urban agents and basic interdependencies between urban residents,
firms, travel decisions and markets.

We consider a stylized metropolitan area encompassing I = 7 zones

(locations), where the innermost zone i = 4 is assumed to be the city

center.10 One can think of zones 35 as the city shaping zones of the
whole urban area, zones 2/6 as the surrounding inner suburbs, and zones
1/7 as the surrounding outer suburbs. The overall expansion of the urban
area is 27.5 km. In each zone i (i 2 I) there is a given homogeneous land
area, Ai, available for residences (housing), and establishments/firms
(production
and retail). As in real cities, land supply of defined urban districts
increases with distance from the city center.
All zones are linked via a transport network as shown in Fig. 1. Also
reflecting real cities patterns, in addition to local
routes, the road network also contains faster expressways (compared to
the slower local routes) that directly connect more
distant (non-neighboring) locations (called beltways in our case). The road
network is represented by links l and nodes. The
nodes correspond to the zone centroids and the links are an aggregate
illustration of a road system. The road (traffic) network
allows private car drivers (h = 1) and commercial travel (h = 2) to choose
among different routes for traveling implying
endogenous vehicle distances. By contrast, routes of public transport
users and pedestrians are fixed implying exogenous
travel distances within the urban area. Public transit routes are assumed
to run through the city on local roads. Trips that
originate and terminate within the same zone i.e. intra-zonal trips
utilize congestible local roads located in the corresponding
zone. According to Anas and Hiramatsu (2012), trips that originate and
terminate in different zones i.e. inter-zonal
trips utilize the intra-zonal local roads for access and egress and the
corresponding link(s)11 connecting those zones.12
Furthermore, each one-way link l with distance dl is defined by an
exogenous road capacity Kl that is used to calculate equilibrium
flow congestion and hence equilibrium monetary and time costs on each
link.
It is assumed that the urban economy is closed in the sense that the total
population in the urban area is fixed and exogenously

given. Hence, there is no interurban migration and utility levels of

households are endogenous. There are N households in the urban
economy containing one working member who supports n dependents.13
Households decide where to reside, where to work, where and how much
to shop, how much labor to supply and, thus, how often to commute, how
much
land to rent, which travel mode to use, and in the case of automobile
travel which route to choose for commuting and shopping trips. All
theses decisions are endogenously determined and implicitly determine
commuting and shopping trip distances,
frequencies and, along with travel speeds, travel times. Households face
monetary travel cost and opportunity cost of travel time. They maximize
utility subject to a monetary budget and time constraint. Income, general
consumption and fuel is taxed according to Malaysias tax system.
Concerning consumption it is assumed that residents have a preference to
go shopping at different urban locations implying that there is spatial
product differentiation.
Travel mode choice where m 2 {automobile, public transport,walking}
denotes the travel modes14 available for trips in th urban area as well
as route choice when driving by automobile where r 2 R denotes a
certain route within a possible route set Rij
15 are endogenously determined. The choice set of alternative routes is
illustrated in Fig. 1. For example, in order to travel from zone 1 to zone 2,
there is only one route. Traveling from zone 2 to zone 6, however, offers
several possible route options. First, driving on local roads (initial vmax =
50 km/h) via zone 3, zone 4, and zone 5. Second, using the outer beltway
(initial vmax = 70 km/h) to travel directly from location 2 to location 6.
Third, driving on local roads from zone 2 to zone 3, then using the inner
beltway (initial vmax = 60 km/h) from zone 3 to zone 5, and eventually,
driving on local roads from zone 5 to destination zone 6. Fourth, using
inner beltways only to travel first from zone 2 to zone 4 and then from
zone 4 to zone 6. Automobile travel times, gasoline consumption and CO2
emissions on a specific link l are all a function of the endogenous traffic
flow road capacity ratio and, thus, implicitly depend on link specific
travel speed (the higher the traffic flow road capacity ratio the lower the
travel speed). They are specified by empirically derived functional
relationships (see Appendix A). This implies that the individual travel
decision of a specific car driver slows down travel speed of all other
drivers being on the road, causing travel time delays (congestion), excess
gasoline consumption and excess CO2 emissions. In addition to

congestion and CO2 emissions, automobile travel causes environmental

costs in regard to air pollution and noise as well as accident costs. The
zone-specific valuation of air pollution and noise costs varies according to
the zone-specific population density.16 Automobile accident costs are
assumed to depend on car travel speed.17 In the case of public transport,
air pollution and noise also depend on zone-specific population density
while accident costs and CO2 emissions per vehicle kilometer (vkm) are
fixed. All in all, since all location and travel decisions are endogenous
urban residents may therefore respond to the speed limit/speed reduction
policies in a variety of ways.

For example, residents can change their place of residence and/or their
place of work, thus, lowering commuting distances; they can shop more
frequently at stores closer to home or less frequently at stores farther
away from home; and/or they can switch to alternative travel modes
and/or routes. On the production side we extend former approaches18 by
assuming that there are U = 2 basic industries which can potentially
produce in every urban location. This allows to take commercial/business
traffic explicitly into account. First, an industry u = 1 produces zonespecific intermediate inputs X1,i by applying a Cobb-Douglas constant
returns technology that combines land and labor inputs. Second, an
industry u = 2 produces and supplies zone-specific final
commodities/services X2,i by applying an overall Cobb-Douglas constant
returns technology that combines land, labor and within a CES.
subproduction function imperfectly substitutable intermediate inputs
supplied by industry u = 1. We assume that there is a spatial
differentiation between the zone-specific intermediate inputs and that
firms producing final commodities/services have a preference to use these
intermediate inputs from different locations. In each zone i a sufficiently
large number of firms of each industry produces their zone-specific
products Xu,i. Within each zone and industry there is competition implying
price taking behavior of firms which sell their products at the competitive
mill price (see Appendix B). The products of each industry are offered in
the zone of production. This, on the one hand, generates
commercial/business traffic since retail firms selling final products/services
pick up intermediate inputs for their own production at different locations
and, on the other hand, generates shopping trips because urban
consumers purchase/consume their products/services at local stores.

Commercial/business traffic uses the urban transport network too, and

therefore is potentially affected by congestion. The corresponding vehicles
consume diesel fuel and generate CO2 emissions depending on travel
speed (see Appendix A). Hence, accruing transport costs for retail firms
and the level of emissions are endogenous as in the case of private
automobile traffic. In addition, commercial/business traffic also generates
accident, air pollution and noise costs.
According to Tscharaktschiew and Hirte (2010a, 2012) the federal
government levies income, sales (consumption) and energy (fuel)
taxes.19 The federal tax revenues are used for public consumption
consisting of purchasing locally produced final commodities. Absentee
landowners use their rent income and an external transport sector
monetary travel costs (except for travel related taxes) accruing from
urban private and commercial travel activities to purchase urban
commodities. At spatial urban general equilibrium, endogenous land rents,
wages, intermediate input prices and final commodity prices clear the
spatially differentiated markets for land, labor, intermediate inputs and
commodities (see Appendix C). The city government decides on local
speed limit policies. A typical working household faces a three-stage
decision.20 In the first stage the household decides on demand for
commodities (shopping), housing and leisure for all feasible combinations
of residential location i and working location j (j 2 I).21

In the second stage, the household chooses its utility maximizing home
and work location, i.e. the preferred location within the location choice set
i = {i, j} given utility attached to i as well as idiosyncratic tastes reflected
by a stochastic utility component.22
With regard to the first two stages, for each location choice set i the
household takes the expected monetary travel costs and travel times over
all travel modes, trip patterns, and routes as given. In the third stage, the
traveler decides on travel mode m and given the choice of travel mode
automobile route of travel r for each location choice set i. In this stage
expected monetary travel costs and travel times are calculated
constituting input data for the first two stages.