Scholarly article on topic 'Mitigating the impact of the expected increase in the population, economy and urban footprint in Cities of the South on greenhouse gas emissions: The case of Cape Town'

Mitigating the impact of the expected increase in the population, economy and urban footprint in Cities of the South on greenhouse gas emissions: The case of Cape Town Academic research paper on "Social and economic geography"

Share paper
Academic journal
Transportation Research Procedia
OECD Field of science
{"Sufficient accessibility" / "motorized travel" / "urban structure" / "greenhouse gas emissions"}

Abstract of research paper on Social and economic geography, author of scientific article — Romano Del Mistro, Viola Proctor, Hazvinei Tsitsi Tamuka Moyo

Abstract The paper aims to demonstrate the potential of structuring the future growth of Cities of the South to reduce the expected growth in greenhouse gas emissions resulting from significant growth in urban population, economy, urban spatial footprint, and hence motorised travel. A situation that cannot be redressed by the typical responses of promoting non-motorised and public transport use because Cities of the South already display high levels of NMT and PT. The paper applies the findings of research aimed at determining whether increasing accessibility always increases utility to inform the planned location of projected economic and population growth for Cape Town. Alternative land use structures are devised in which future population growth (i.e. housing and community facilities) and related work opportunities are allocated in an attempt to minimise motorised travel but yet achieve “sufficient” accessibility for four income groups. The City of Cape Town has modelled the effect of applying a TOD urban land use and transport system in 2032. The paper allocates the changes in trip making between 2013 and 2032 in support of the concept of sufficient accessibility. This shows a significant reduction in motorised travel and greenhouse gas emissions when compared to the TOD approach.

Academic research paper on topic "Mitigating the impact of the expected increase in the population, economy and urban footprint in Cities of the South on greenhouse gas emissions: The case of Cape Town"

Available on line at



Transportation Research Procedía 25C (2017) 3515-3532 * * * **

World Conference on Transport Research - WCTR 2C16 Shanghai. 10-15 July 2C16

Mitigating the impact of the expected increase in the population, economy and urban footprint in Cities of the South on greenhouse gas emissions: The case of Cape Town

Romano Del Mistroa*, Viola Proctorb and Hazvinei Tsitsi Tamuka Moyoc

a Department of Civil Engineering, University of Cape Town, Ronde bosch, C ape Town 7701, South Africa.

bMott MacDonald PDNA, 2nd Floor, 5 S t Georges Building, St Georg es Mall; Wape Town, 8001 c Department ofCi vil Engineering, University of Cape Town, Rondeboschi Cape Town7701, SouthAfrica. *Corresponding author Phone: +27 (0)72 573 7478, Fax: ++27 (0)21 689 7471, E-mail address:


The paper aims to demonstrate the potential of etracturieg the future growth of Cities of the South to reducer the expected growth in greenhouse oas emissions resulting from significant growth in usbm population, economy, urb an spatial footprint, and henm motorised travel. A situation that cannot be redressed by the typical responses of non-motorised and public transport use because Cities of the South already display high levels of NMT

andTPhTe. paper applies the findings of research aimed at determining whether increasing accessibility always The p aper applies the findings of research aimed at determining whether increaiing accessibility always increases utihty to inform the pl anned location of projected economic and population growth for Cape Town. Alternative land usp structures are devised in which future populatfon growth (i.e. housing and community u^cilities) and related work opportunities are allocated in an attempt to minimise motorised travel but yet achieve "sufficient" accessibility oor foue incotne croups.

The City oo Cape Town has modelled the effect of applying a TOD U3ban land use and transport system in 2032. The paper alloeates the ch anges in trip) making b etween 20 13 and 2032 in suuport of the concept of sufficient taoc cthesesTibOilDitya. pTphroisa csh.ows a significant reduction in motorised travel and greenhouse gas emissions when compared to the TOD approach.

© 2017 The Authors. Pished by Elsevier B.V.

Peer-review urnder reepimeiaihfy of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY. Keywords: Sufficient accessibility; motorized travel; urban structure; greenhouse gas emissions;

2352-1465 © 2017 The Authors. Published by Elsevier B.V.

Peer-review under responsibility of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY. 10.1016/j.trpro.2017.05.272

1. Introduction

One of the primary objectives of urban transportation is to increase accessibility for persons living and working in the region. As cities grow and expand spatially transport authorities will attempt to provide easy access between all areas in the region. This is achieved by building more facilities on which high travel speeds can be attained; e.g. freeways, Bus Rapid Transit, Light Rail Transit and railways.

The populations of cities in developing countries, especially in Africa, are expected to double over the next 25 years (Demographia, 2010). The growth in population will result in the spatial expansion of major urban areas. When this is coupled with the expected increase in per capita income (Heerman, 2014), the spatial growth will be even greater as private car ownership rises and "suburbia" becomes the housing choice of the growing middle income population. This translates into longer motorized trips for the more affluent and the poor. While the more affluent can be expected to afford the increased cost of motorized travel and will coerce the transport authorities to provide faster travel options, three questions must be asked; namely:

a) Can the national, provincial or metropolitan governments afford these additional costs?

b) Can the poor afford the costs of longer travel distances?

c) What are the environmental consequences of increased motorized travel distances?

Faced with these realities, transport authorities should be implementing city structures that require less motorized travel rather than retrofitting "laissez faire" land use development with higher speed transport solutions. Transport authorities applying such an approach will face strong opposition unless they are able to convincingly reply that "sufficient" accessibility is being provided. From a research point of view, this can be restated as "Is there a level of accessibility beyond which additional accessibility does not increase benefit or utility?" A socio-economic analysis of costs and benefits (in the broadest sense) could provide some answers. This paper does not describe such an analysis; but attempts to develop an understanding of the perceptions of employees and employers of the benefits and costs of different levels of accessibility.

The consequences of excessive travel are well known. The environmental costs are well known with transport having contributed 23% of CO2 emissions in 2007 (WCTRS. 2011). There are examples of companies assisting employees to make commuting more sustainable by using public transport (e.g. Aspen Valley Hospital, 2010, Work, Job and Income, 2010).), incentivise carpooling ( 2010), and awareness programmes for staff encouraging modal shift and reducing carbon emissions (United Nations, (n.d.-b).).

The socio-economic costs are also well known with the poor spending more than 20% of their income on transport (Walters, 2008) and travelling for more than two hours to work. Some companies contribute to the costs of employee commuting in various ways that can overlap with the incentives to use public transport mentioned above (Shoup, 1997), private mass transit services and transport allowances separate from salary to reduce the burden of travel on staff. These incentives generally soften the cost of long distances, albeit in some cases through using mass transit; but do not encourage trip distances to be shortened.

Furthermore, employers admit to staff being late due to delays from traffic congestion and public transport inefficiencies (Coleman, 2000), as well as tiredness and reduced efficiency from long commuting distances. Yet it seems that decisions by employers on location choice and staff selection are based on minimising immediate financial cost and maximising short-term profit (Parr, 2002).

There are also examples of employers encouraging employees to travel less e.g. by employing locally or incentivising workers to move nearer to where they work (City of Trenton, 2011) while others try encourage working from home to reduce commuting (United Nations, n.d.-a).

In this paper we provide a brief review of the theory on "too much choice"; describe two studies in Cape Town to ascertain whether employees and employers perceive that too much choice of work opportunities or employees can bring negative benefits, and before testing the implications on motorised travel and greenhouse gas emissions of applying the concept of "sufficient" accessibility to land use and travel patterns in Cape Town in 2032.

2. Increasing accessibility increases choice

As mentioned earlier the purpose of increasing accessibility is to increase the number of destinations at which a trip purpose can be fulfilled or the catchment size of customers and employees; i.e. to increase choice. The

underlying hypothesis of the research on which this paper is based was that the relationship between accessibility and the benefits is not linear.

Since no literature could be found on the benefits of increasing accessibility in urban areas, one had to resort to the fields of psychology, marketing and manufacturing. The findings of the review can be grouped in three categories:

a) Benefits increased with choice because of the large assortment of products (Koelemeijer & Oppewal, 1999.

b) Benefits increase with choice but at a decreasing rate. (Coombs & Avrunin, 1977).

c) Benefits increase with choice until a point after which they decrease ((Reutskaja & Hogarth, 2009; Iyengar &

Lepper, 2000).

An aspect that is often ignored is the cost of more choice. (Scheibehenne, Greifeneder & Todd, 2009).. Speaking in Ted Talks, Dr Schwartz claimed: "There's no question that some choice is better than none, but it doesn't follow that more choice is better than some" (Schwartz, B. 2005).

The second aspect of cost increasing with choice arises from the "inventory" cost of providing more choice. in retailing and manufacturing (Benjaafar, Kim, & Vishwanadham, (2004).

3. Two exploratory studies

In both retail and manufacturing the cost of being offered more choice is paid by the customer as it is included in the price. In the case of transport, while the commuter will experience the benefit of more choice in where the trip function can be fulfilled, he/she does not always take all the costs into account e.g. the cost of additional wear and tear on private cars, the cost of infrastructure, disproportionate subsidies paid to long distance passengers, the external costs of using scarce resources, production of greenhouse gas emissions, etc..

There are many trip purposes that need to be fulfilled in an urban area. The journey to work is the trip most frequently made using motorized transport. Accessibility for this trip purpose can be measured as the number of suitable jobs that are accessible to employees and the number of suitable workers that are accessible to an employer.

Employees and employers in Cape Town were interviewed to ascertain how much they perceived the benefits of increased accessibility. Both studies followed the same methodology in that respondents were presented with pairs of accessibility options and asked to choose one (i.e. a forced choice). Employees were also offered the opportunity of retaining the status quo (unforced choice). The choice options were described using five attributes for employees and four attributes for employers. The data collected from the interviews were analyzed using a multinomial logit (MNL) model assuming that each respondent would be maximizing their utility from each choice.

3.1 Theoretical framework

Basic utility theory states that the choice between alternatives is made on the basis of the respondent's perceived utility. The respondent's utility has two components; namely: a deterministic component (which is a function of the observed attributes of the alternatives, respondent characteristics and economic variables such as income, price of goods, etc.); and an error term, (which is a function of unobserved characteristics that may influence the respondent's choice) (Del Mistro & Hensher, 2009). as shown in Equation 1.

Ult = Vt +eit (1)

where Uit is the total utility of choosing alternative i to the decision maker t,

Vit is the observable (deterministic) portion of the utility which is estimated by the analyst, and

eit is the error component representing the influences that are unobservable by the researcher, but known to the


The functional form of Vi can be expressed as linear, logarithmic or quadratic (Koppelman & Bhat, 2006). The two studies assumed a linear relationship (equation 2).

Vi = P01 + Pli/(Xli)+ P2i/(X2i)+ P3i/(X3i)+.... Pni/(X„i) (2)

p1i is the weight (or parameter) associated with attribute X1 and alternative i

p0i is the parameter not associated with any of the observed and measured attributes, called the alternative-

specific constant, which represents on average the role of all the unobserved sources of utility of the specific alternative.

The underlying assumption is that individuals will try to choose an alternative that awards them the highest utility. In other words, "alternative 'i' is chosen amongst a set of alternatives, if and only if the utility of alternative 'i', is greater than or equal to the utility of all alternatives, 'j' in the choice set" (Hensher, Rose & Greene, 2009).).

Assuming a Gumbel distribution of the error term, allows the probability of an option being chosen to be calculated from equation 3 (Hensher, et al, 2009).

pi = 3


The multi-nomial logit (MNL) model uses the theory of Maximum Likelihood to estimate the value of the parameters that will link the utility function to the choice data likelihood function and indicate the probability of choice.

The Likelihood ratio (p2) tests whether subsets of the bs (as functions of the UA and UB) are significant in the MNL function, compared to the null hypothesis and is generally expressed as:

p2 = 1- (LL(9)/LL(0)) (4)

To account for "small" samples the equation is adjusted as follows:

Adj-p2 test = 1- [{Nobs/(Nobs-Nvar)}* {LL(9)/LL(0)}] (5)

Where Nobs is the number of profile pairs Nvar is the number of variables in the utility equation It has been shown that -2((LL(0)-LL(9)) is c2 distri selected significance level, the null hypothesis can be rejected.

3.2 The value of accessibility to employees

3.2.1 F actors affecting choice

The factors identified in the literature that affect job choice were selected through four focus groups to determine the attributes to be used in the survey. These were:

a) Distance from home to work.

b) Peak period travel time.

c) Cost of transport -based on monthly fares for public transport users and average per kilometre rates provided by the AASA for private transport.

d) Change in Salary.

e) Number of Job Opportunities- Turnover of workers within each income band (Procter and Del Mistro, 2013).

It has been shown that -2((LL(0)-LL(6)) is c2 distributed, such that if this value exceeds the c2 for the pre-

3.2.2 Experimental design

To offer all combinations of the values of the five attributes (i.e. 6A1*2A3*3A1) would require 144 choice pairs. These were reduced to 36 experiments using a fractional factorial design (The R project, 2013). These were converted into 18 pairs and into 2 blocks of 9 choice pairs.

3.2.3 Selection of respondents

Respondents came from two groups; namely persons earning between US$330 and US$700, and using public transport, and persons earning between US$700 and US$1400 per month and using private transport either as drivers or passengers.

Suburbs that showed a high probability of finding respondents using the criteria were selected. One person from every eighth dwelling was selected for interview in each of the suburbs. 400 Persons were interviewed to achieve a 95 per cent confidence level; 200 in the lower income group and 200 in the lower middle income group.

Each respondent was offered nine pairs of choices and asked to make a forced and unforced choice. The data

obtained from the low and low-middle income respondents were analysed separately using Limdep. The best equations are shown in Annexure A. The four equations were selected because the P-values are less than 0.05 and the signs of the coefficients are logical. All the equations achieved an acceptable p2 value (i.e. between 0.2 and 0.4).

3.2.4 Findings of the study

From Annexure A it can be seen that the distance coefficients indicate that the part-worth utility declines with distance.

The total impact of distance derived from all the part worth utilities is shown in Figure 1. This shows that utility declines with distance. What is interesting is that utility actually becomes negative beyond 20km and 5km for forced and for unforced choice among low income respondents and 10km and 40km for forced and unforced choice among low-middle income respondents respectively. While these support the supposition that increase choice of trip ends (with distance) does not produce a related increase in utility, the range in values of distance where this utility becomes negative does not provide an obvious value for "sufficient" choice.



§-2 -3 -4

Forced choice

y = -0.1591x —3.3893 R2 = 0.7596


y = -0,0811x + 0,2233 R2 = 0,51328


1 0 -1 -2

i\ 1 ♦ : ♦

0 1 5 10 15 i 1 1 1 20 25 30 35

♦ ♦

x ♦ ♦

Unforced choice 4

-3 H y = -0,1493x + 1,4342

R2 = 0,79811

S 1 0 -1


y = -0,0753x + 2,9991 R2 = 0,45347


Figure 1. Employee's utility versus catchment size

3.2 The value of accessibility to employers

3.3.1 F actor affecting choice

From literature and conversations with employers, the following variables, were identified as influencing employment decisions:

a) The number of employees to choose from- The recruitment pool was calculated by factoring the number of workers within four catchment radii (i.e. <10km, <15km, <25km, <40km) using Census 2001 (Statistics South Africa, 2001) population per suburb, in each of two income groups (namely Low income: US$375 - US$1 250 per month (assumed to use Public Transport) and Low-Middle income: US$1 250 - US$3 750 (assumed to use Private Transport) by the historical staff turnover rates in each income group in the firm.

b) Transport subsidy- This attribute was applied only to Level 1 employees to reflect the concept (not currently applied in South Africa but in countries such as Brazil) that employers should be cognisant of the travel costs incurred by employees; and that they should pay the difference between what is "reasonable" and the actual fares. A value of eight per cent was assumed.)

c) CO2 emissions- The emissions from employee commuting should be considered when recruiting as it adds to the carbon footprint of the company; e.g. Nedbank (2011) found commuting contributes to approximately 20% of their total carbon footprint. Emission factors (DEFRA, 2011), combined with CO2 from the production of petrol/diesel (Sasol, 2011), were used to calculate CO2 emissions per employee to be 0,54 kgCO2e/passenger km (car) and 0,1 kgCO2e /passenger km (bus).

d) CO2 tax- Companies were to pay a "green tax" for their carbon emissions of staff commuting at $15 per ton CO2e (DNA Economics, 2012).

3.3.2 Experimental design

With one variable at 4 levels and the other three at two levels B and assuming that 2-factor interactions between the variables were not to be investigated; an orthogonal design was developed, using Kocur et. al. (1982), having 8 profiles paired into 4 questions for each income group.

3.3.3 Selection of respondents

Of the twenty randomly selected large firms, five agreed to participate in this study and 49 managers responsible for staff recruitment were interviewed. These firms included the service sector (hospital), retail sector, public institution, academic institution and consulting firm.

3.3.4 Findings of the study

Each respondent was offered four pairs of choices for each income group.

The data was analysed using Limdep. The best equations are shown in Annexure B. Only the equations for the low income employees achieved an acceptable X2 value (i.e. between 0.2 and 0.4). For both income groups, the coefficients reflect that benefits increase for employers with increased catchment size for both groups of employees.

Figure 2 shows the perceived utility derived by employers from increasing catchments for employees of each income group. This figure indicates that employers perceived that:

a) For low income employees, utility would increase as catchment size increased to 45% of total catchment and then decrease.

b) For low-middle income employees, utility would increase as catchment size increased to 55% of total catchment and then increase at a much reduced rate.


Figure 2: Employer's utility versus catchment size 4. Application of the findings

The two exploratory studies highlighted the difficulties in attempting to determine the benefits that employees and employers perceived they would derive from increased levels of accessibility. While the findings do not provide a value of accessibility at which the benefit is perceived to become negative or where the slope of the benefit curve becomes negative, they do prove that both employees and employers recognise that increase accessibility is not always accompanied by increased benefits.

4.1 Alternative city structures

Most cities have, historically had a prominent centre, as represented in monocentric city models such as those of Burgess and Hoyt. But as these grow into metropolitan areas the often incorporate other towns to create, a multi-nodal pattern emerges as described by Ulman and Harris. This is accompanied by major transport infrastructure to make the enlarged urban area more cohesive; at the same time making one area much more accessible than other parts of the region. While this significantly increase property values in this area it also increases congestion which in turn is "resolved" by more transport infrastructure that increases the spatial catchment area, and the necessity for and amount of motorised travel. The question is seldom raised about whether it is always beneficial to further increase accessibility.

4.2 Replacing physical connectivity with virtual connectivity.

Advances continue to be made in information and communication technology that provide more opportunities for the electronic transfer of information as opposed to the transfer of information kept in people's heads; which requires face to face communication (Hall and Pain, 2006). IT is unable to meet all the needs of information and knowledge transfer. Building relationships internally and externally requires physical interaction. Pain and Hall (2006, 108) explain how important face to face interaction is for Advanced Producer Services (APS) because "trust is an essential factor in cooperation in the high value exchanges which are knowledge intensive, individual and incapable of standardisation".

It is accepted that face to face communication remains an essential activity in economic and social interactions. What is not clear, is how often these physical interactions occur, how often they need to occur; the implications if they do not occur as frequently, how they are spatially bound, how much do the city structure and transport infrastructure need to be arranged to facilitate them, the implications if accessibility is sub optimum; and do other sectors (besides APS) of the economy have different needs? In essence, what is being asked is what level of accessibility is required, and what the costs are of achieving or not achieving this level.

4.3 Land use and transport structure; to reduce the amount of motorised travel.

Motorised travel, under existing land use/transport conditions, can be reduced by eliminating excess commuting/travel. This is travel that can be avoided by commuting or travelling to a destination that is closer to fulfil the purpose of a trip. There is considerable debate about the extent of "excess" travel (Ma & Banister, 2006).

While there does not appear a lot of scope for reduction in motorised travel from "excess" travel; the expected doubling of populations (and therefore work places) in cities in developing countries over the next 25 years provides the opportunity to locate new growth in residential, work and other land uses in such a way that it minimises the need for motorised travel.

What emerges from the preceding discussion is a recognition of the need to concentrate people and activities. This concentration occurs at locations with relatively high accessibility; producing a hierarchical system of centres in the case of a city; and cities in the case of mega city regions. This can be explained to some extent by theories such as Christaller's Central Place theory (Everson and Fitzgerald, 1971).

Accessibility in the case of "centres" is determined by the number of persons that can access the centre in a reasonable period of time and the attraction of the activities at the centre. Centres compete for new activities and for people to support the activities; and like cities they are

"..... in a permanent struggle between the economies of scale and scope (localisation advantages, economies of

density, etc.) and agglomeration diseconomies (congestion, pollution , criminality, etc.) ... "(Bannister et al., 2007, 7).

Theoretically, the diseconomies of growth can be expected to limit growth. Unfortunately decision makers are not always aware of all the costs of their decisions and as such the diseconomies might not be appreciated in individual decisions.

While one can debate whether the structure of a city can be left to the market or whether the planning and transport authority can shape it, there still remains the need to seek more sustainable urban structures and supporting transport solutions; one of which is a system of sub-areas of the city or region each focused on a major centre. In his book, Great Cities and their Traffic, J Michael Thomson (1977) discussed 5 archetypes; namely:

a) Full Motorisation; composed of low density development and grid of roads and freeways. The city is unable to achieve a centre of any stature; nor does it provide the opportunity for a good public transport system because of the low densities.

b) Weak-centre Strategy; in which the public transport system is not strong enough to allow the city centre to gain sufficient stature and applies to cities that are not too large.

c) Strong-centre Strategy;-evident in major cities where the city centre is powerful because it is served by radial rail services with a city centre public transport distribution system which further allows the city centre to grow still further. However, attempts are still made to provide more access for private motor vehicles into the centre leading to congested conditions.

d) Traffic limitation strategy; which is also a strong centre strategy but the emphasis here is to limit the amount of private car travel. In very simple terms cars are not to be encouraged to come into the city centre but can be used to go out of the city. This also requires a good radial public transport system to concentrate workers in the centre and a public transport distribution system to distribute them within the city centre.

e) The low cost strategy; which was proposed as a solution for cities, typically in developing countries, where resources are not available for major radial rail systems to the city centre and an underground rail system to permit its city centre to grow to the size of those in Paris or London. Instead the radial routes are served by buses or trams and nodes are developed along these radials to reduce the number of commuters travelling to the city centre. The strategy is shown diagrammatically in Figure 3.

Arterial road Bus priority road • Sub centre

Figure 3: The low cost strategy (after Thomson, 1977)

Thomson summarises the concept as:

"....... It is possible for 200 000 workers and their families to live within walking distance of the city centre,

without excessive overcrowding. (This assumes a resident population of 500 000 living at 20 000 per square km in a area of 5 km square.) Eight main radials each carrying 20 000 persons per hour, could bring in 320 000 commuters in a two-hour period.... " (Thomson, 1977,225) "... .sub-centres are situated along the main radial

roads. Their size be within walking distance (i.e. within 800 metres........accommodating ... .in a

low income city ... not more than 30 000 [jobs]" (Thomson, 1977, 227).

The low cost strategy perpetuates the importance of primate centre; which necessitates long distance travel to and from it. Its relatively high level of accessibility will produce high property values and unaffordability of accommodation for the poor; who then have to accept housing on the outskirts and the resultant high commuting cost.

Another approach could be a multi-centre city in which sub areas of the city are relatively self-contained. Figure 4 shows an example of the concept, which can be described as follows:

a) A system of sub-cities with a radius of say between 5 and 7km.

b) Each sub-city has:

i) A centre of 4 km2; accommodating 50 000 residents (30 dwellings/ha) and 50 000 work places; all within walking distance.

ii) 8 Development corridors 2km wide and 6km long outside the centre; each able to accommodate 100 000 residents (20 dwellings/ha) and 50 000 work places.

ii) 100km2 of less developed land for open space, urban agriculture, etc.; and even lower density development. The advantage of this urban structure is that if the majority of trip purposes can be satisfied within the sub-city itself; then travel distances can be expected to less than 5 km; many of which could be made walking or cycling. The fact that the city as whole will not have a traditional primate centre might be considered a disadvantage. But it can be expected that each of the sub-city would generate specific primate functions.

r . . . A sub-city with walkable centre and 8 radial

A system of sub-cities . . . .

mixed use corridors

Figure 4: A system of sub-cities

The fundamental issue that needs research is to determine the size of the sub-city that is necessary for it to be self-contained; i.e. provides "sufficient" accessibility.

While the scale of the proposed sub-city might be different, the concept is not new. For example, Sir Peter Hall (2010, 35) describes the Mark 2 new towns located as being located "... .outside the commuter belt .... 200 000 or

even 250 000 inhabitants so that they could support all the jobs and service many as 1/3 of the

inhabitants would find jobs in place where they lived".

While the concept has obvious advantages it might appear difficult to implement. Two reasons can be put forward for this; namely:

a) Decisions on development are made by property developers individually and not by a city collectively. Furthermore, the scale of existing development in the primate centre tends to attract new development than t newer and smaller centres. Nevertheless, it should be possible to induce developers to locate their development to towards achieving the goals of the city. Van der Weteren and Del Mistro (2003) used utility maximisation theory to explain the factors that affect the location decisions made by developers. These included three factors within the discretion of the city without additional cost; namely:

i) The city approves development more quickly in desired locations that meet the city's developmental goals than in others.

ii) The city provides the land for the development.

iii) The city leases rental space in the development.

b) The other reason is that a city lacks stature without a strong city centre (Thomson, 1977, 162). This view is supported by Pain and Hall (2006, 115) who refer to the city symbolism attached to the "First city" in the context of regions which equates to the city centre in the case of a city. They go on to reflect that locational decisions are not always "based on rational economic criteria, the personal preferences of senior decisions makers can be highly influential"; that "environmental quality and quality of life are highly regarded' yet "paradoxically, ... the centres attracting the largest firms .. are invariably those most compromised environmentally by traffic congestion and pollution'..

The questions that arise are:

a) How large should sub-cities be to be able to provide sufficient choice in internally to significantly reduce the need to travel beyond its limits?

b) What "distance or travel time" impediment is required to discourage travel/commuting to other sub-cities?

c) What are the consequences of deterring inhabitants of a sub-area from a wider choice of destinations to satisfy their job (and other activity) preferences?

These all lead to the need to understand the consequences of reducing choice of origin destination pairs. As mentioned earlier, transportation strategy generally attempts to improve accessibility; i.e. increase the choices available. In this pare we discuss the findings in respect to the last question.

5. Application of "sufficient" accessibility strategy to Cape Town 2032

5.1 Alternative city structures: Cape Town

The City of Cape Town has recently completed its land use / transport plan for 2032. One alternative is referred to as the 2032TOD alternative which attempts to develop a more energy efficient city; i.e. reduce motorised travel.

A spatial model, composed of seven sub-cities, was developed for the Cape Town for this paper (Figure 5). (It must be stressed that the arrangement has not been optimised.)

Figure 5: Application of "sufficient" accessibility spatial structure to Cape Town 4.2 Building the study model

Zonal trip generation data for our income groups (high, high middle, low middle and low) and three motorised transport modes (i.e. private vehicle driver, private vehicle passenger and public transport passenger) for the 2013 base year and the 2032TOD land use-transport alternative were provided by the City of Cape Town. These data were aggregated into 96 zones for this paper.

The increase in regional trips generated (by income group by mode) between 2013 and 2032TOD was purposely allocated to the seven sub-city regions (with the remainder being allocated to zones outside these sub-cities) to

improve the balance between trip productions and trip attractions (by income by mode) in each sub-city, as is required when applying the concept of "sufficient" accessibility.

A computer model was developed to calculate four trip distribution matrices (namely two gravity models using a high and a current coefficient for the friction of distance and two linear programming models to calculate the maximum and minimum travel distance in terms of the concept of excess commuting (Horner, 2002)) for each set of trip generation data. The trip distribution outputs for each combination of income group-transport mode-trip distribution models (i.e. 48 trip distribution matrices). were summarised into trip length distributions, average travel distance, and total travel distance. Greenhouse gas emissions were also calculated.

5.3 Spatial distribution of trip production

Table 1 summarises the performance of the 2013 Base Year (2013), the 2032 Transit Oriented Development (2032TOD) and the 2032 "Sufficient" Accessibility (2032SUFf) land use-transport models.

Table 1 Spatial generation of motorised trips

Trips Produced Trips Attracted

Sub-city 2013 2032T0D 2032SUFF 2013 2032T0D 2032SUFF

1 174042 309486 349121 323561 439389 344434

2 879939 823151 1093779 729941 956737 1095953

3 417889 830830 653646 520726 649427 659451

4 61235 146811 95254 63021 144149 96825

5 400789 125926 549749 422923 141345 539760

6 350336 405670 404952 248717 316470 392550

7 126911 190700 172438 114568 204204 176876

Other* 407736 1113879 594490 395421 1094734 607580

Total 2818877 3946454 3913428 2818877 3946454 3913428

* Other refers to areas outside the 7 sub city areas

Table 1 shows the trips produced and attracted by each of the sub-city areas for the 2013 Base Year (2013), the 2032 Transit Oriented Development (2032TOD) and the 2032 "Sufficient" Accessibility (2032SUFF) land use-transport models. It can be seen that in the 2032SUFF model, the trips produced and attracted have purposely been distributed to create a better balance between them within each sub-city area; as would be required when applying the concept of sufficient accessibility. In practical terms the growths in trips produced and attracted by each income mode combination between 2013 and 2032T0D models was distributed to improve the balance in each sub-city.

5.4 Implications of alternative city structure on motorised travel

Table 2 shows the average trip length by income and mode for the study area for the three alternative city structures. As expected the average trip length when a high friction factor is applied is at least less than half the average trip length when the current friction factor is applied. The 2032SUFF spatial structure produces the lowest average trip length; namely 14 percent less than 2013 and 9 per cent less than 2032T0D for the high friction factor respectively; but only 1 per cent less for both other models at the current friction factor. One needs to remember that the population of the city has increased by 40% between 2013 and 2032. There is some variability in average trip length between different income groups and modes.

Table 2 Alternative city structure: Average trip length (km).

High friction factor Current friction factor

2013 2032TOD 2032SUFF 2013 2032TOD 2032SUFF

High 8.30 8.86 7.70 17.97 18.18 17.87

High middle 8.42 8.37 8.03 18.68 19.03 18.74

Low middle 9.39 8.99 7.88 18.92 19.02 18.65

Low 8.51 7.99 7.60 18.00 18.08 17.96

PvT Driver 8.18 7.85 7.63 18.01 18.09 17.90

Pvt Passenger 8.04 8.42 8.06 18.78 19.18 18.85

PT Passenger 9.35 8.86 7.80 18.67 18.73 18.45

All 8.86 8.53 7.79 18.51 18.62 18.36

These findings indicate that city structure of itself is not sufficient to produce significantly less motorised travel. It is essential that it is accompanied by a higher travel impedance; i.e. limiting accessibility. This can be put into context of the concept of excess travel. Horner (2002) proposes that both minimum and maximum travel should be considered. Table 3 shows these values determined using linear programming. It shows that under current friction factor, trip end choices result in average trip lengths that are less than 40% of the maximum average travel distance derived using linear programming and substantially more than the minimum average travel distance derived through linear programming; suggesting that there is some scope to reduce motorised travel.

Table 3. Alternative city structure: Minimum and maximum average trip length (km)

2013 2032TOD 2032SUFF

Minimum 7.53 7.34 6.6

Maximum 47.02 47.66 47.82

Table 4 shows the percentage of trips that are longer than 20 km. As expected a high friction factor reduces the percentage significantly; from over 46 per cent to 10.9 per cent for 2032SUFF, 14.54 percent for 2032TOD and 15.42 for 2013. At current friction factor, the percentages of trips longer than 20km are almost the same across the three spatial structures.

Table 4. Alternative city structure: Percentage of trips longer than 20 km High friction factor Current friction factor

2013 2032TOD 2032SUFF 2013 2032TOD 2032SUFF

High 11.52 12.75 9.57 42.02 41.87 42.13

High middle 12.17 12.52 10.51 45.65 47.58 45.82

Low middle 17.36 16.06 11.31 48.76 49.02 47.77

Low 15.41 14.21 10.97 46.91 47.73 46.46

PvT Driver 12.98 12.02 10.43 45.24 45.59 44.87

Pvt Passenger 11.11 12.30 10.91 46.35 47.78 46.77

PT Passenger 17.49 16.22 11.12 48.11 48.65 47.16

All 15.42 14.54 10.91 47.11 47.71 46.49

The last aspect of trip making considered in this paper is the percentage of trips with both trip ends within one sub-city itself. This is shown in Table 5. From this table it can be seen that there is very little difference between the three city structures in their ability to satisfy both trip ends within individual sub-cities. The effect of a high friction factor is obvious. A significant difference is evident between the two levels of friction; with percentages improving

from approximately 55 percent to approximately 90 per cent; consistently across income groups and modes.

Table 5. Alternative city structure: Percentage of trips satisfied within sub-city

High friction factor Current friction factor

2013 2032TOD 2032SUFF 2013 2032TOD 2032SUFF

High 87.47 83.94 90.55 57.53 56.85 58.06

High middle 86.53 83.25 89.47 53.76 55.74 54.25

Low middle 82.69 81.56 88.92 54.14 51.42 55.00

Low 85.36 84.16 90.82 57.92 55.48 57.44

PvT Driver 86.84 85.21 90.69 57.28 55.19 56.57

Pvt Passenger 88.02 83.58 89.40 54.16 52.00 54.48

PT Passenger 82.76 81.78 89.52 55.14 52.79 56.09

All 84.56 82.95 89.82 55.57 53.32 55.99

This finding is important because it emphasises that even with the current and 2032TOD spatial structures there is a possibility to reduce the amount of motorised travel and still provide sufficient accessibility in terms of broad income groups and three motorised modes; provided the friction of distance is appropriate.

5.4.5 Implications of alternative city structure on greenhouse gas emissions.

Table 6 shows the estimated CO2 emissions produced in the peak period assuming that drivers of private transport produce 0.54 kg/km and passengers using public transport produce 0.1kg/km. From this table it can be seen that there is minimal difference in CO2 emissions from the two city structures for 2023. However there is a significant difference if the friction of distance is increased. The values in the table can be scaled to annual values assuming that the peak period motorised travel is 30 percent the daily travel and there are 300 equivalent days per year. Implementing a land use transport plan aimed at reducing motorised travel could achieve an annual reduction of 740 000 tons CO2.

Table 6. Alternative city structure: CO2 emissions in peak period (kg*10A6). High friction factor Current friction factor

2013 2032TOD 2032SUFF 2013 2032TOD 2032SUFF

High 0.58 0.91 0.76 1.21 1.83 1.78

High middle 0.52 0.68 0.61 1.08 1.53 1.51

Low middle 2.32 2.83 2.56 4.90 6.34 6.18

Low 1.46 2.09 2.17 3.22 4.92 4.89

PvT Driver 3.30 4.45 4.30 7.27 10.26 10.10

Pvt Passenger 0.00 0.00 0.00 0.00 0.00 0.00

PT Passenger 1.58 2.07 1.80 3.15 4.37 4.25

All 4.88 6.52 6.10 10.41 14.62 14.35

5.6 Discussion.

The effects of city structure or friction factor emerging from the modelling described above are summarised in Table 7. From this table it can be seen that any significant reduction in motorised travel can best be attained by increasing the friction of distance and not from spatial planning.

Table 7: Impacts of alternative city structure or friction of distance.

City Structure Friction factor

Average travel distance +9% +50%

Trips less than 20km +20% +30%

Percent within sub-city +10% +40%

CO2 emissions +3% +50%

The findings also indicate that there would be substantial social, economic and environmental benefits in respect of reduction in motorised travel, percentage of trips with longer travel distances and amount of greenhouse gas emissions.

In developing countries, most motorised trips are made using public transport, so a strategy to shift private to public modes is very limited. It should also be noted that travel distances in Cape Town, are longer than those in other cities in developing countries. The provision of improved public transport services (e.g. BRT) with their increased travel speeds and often highly subsidised fares can be expected to significantly increase motorised travel distances.

Increasing the friction of distance implies reducing accessibility. It can be expected that any land use transport plan that espouses this objective will be met with political opposition. TOD aims to reduce the effect of motorised travel by increasing the use of public transport. It is not intended to intended to limit accessibility in any way.

Sections 3.2 and 3.3 described two studies proved that utility is not linearly related to accessibility. They provided a very wide range of values at what catchment distance the utility became less than zero or at what percentage of total catchment the rate of increasing utility declined significantly. While the wide range does not provide a definitive value, one can use 25 percent of current Cape Town catchment as a "sufficient" catchment, (although this does depend on employee income). From Table 1, this would amount to 700 000 trip ends for Cape Town in 2013. If the trips generated to or from Sub-cities 2, 3 and 5 in the 2032SUFF model were more evenly redistributed, each would provide sufficient accessibility. Such an argument could be used by land use transport authorities having the objective to minimise the negative effects of motorised travel; and are prepared to encourage/ coerce future land use development to achieve the desired spatial balance between workers and jobs. This would be accompanied by the provision of the appropriate amount of transport infrastructure and services to significantly.

Benefits, besides the environmental benefits, that a transport authority can expect from implementing a "sufficient" accessibility approach; include:

a) Reduced capital expenditure on transport infrastructure; because of lower maximum traffic (vehicles and

passengers) to be coped with in any location.

b) Reduced public transport subsidy because of the smaller number of passengers being carried over longer


There might be some groups of the population for which accessibility is never "sufficient". These would be individuals with very specialised skills; i.e. usually very well remunerated and having the means to choose the locations of their trip ends.

6. Conclusions

There is a major concern about the contribution of motorised travel to greenhouse gases. Many approaches have been proposed to resolve this. However, they generally do not attempt to reduce distance travelled; preferring to not reduce accessibility; and often aimed at increasing accessibility.

This paper has proposed the need to re-examine the amount of accessibility that is necessary and beneficial. Two studies attempting to do this were described. While they did find that the relationship between increasing accessibility and benefits is not linear, nor always positive; they were unable to specify a value at which accessible is "sufficient". A value of 700 000 trip ends can be used to begin the discussion, but needs more research.

Trip generation data for Cape Town 2013 and 2032 studies was used to estimate the motorised travel that would result from an alternative allocation of the growth in trips by income group and by mode between the 2013 and 2032

models. The alternative allocation attempted to better balance trip production and trip attraction by 2032 within sub-cities within Cape Town. The city structure in itself had some effect on reducing motorised travel. The factor that was more effective in reducing motorised travel was an increase in the friction of distance. This could be created by a transport system that makes travel very easy within the sub-city itself and very "costly" to destinations outside. The application of such a spatial and transport model was found to have the following benefits:

a) Reduce the average travel distance from 18.5 to 7 km.

b) Reduce the number of motorised trips longer than 20 km from 47 to 11 percent.

c) Increase the percentage trips that could be satisfied within the sub-city area from 55 to 90 percent.

d) Reduce CO2 emission by almost 60%

Further research is required on the savings in transport infrastructure and operating costs that such an approach would produce the incentives required to coerce development in such a way to achieve the necessary land use patterns that provide "sufficient" accessibility and the quantum of "sufficient" accessibility itself.


This paper reports on work done for Project 12: City Restructuring at the African Centre of Excellence for Studies in Public and Non-motorised Transport (ACET); which is a joint research programme between the Universities of Cape Town, Dar es Salaam and Nairobi funded by the Volvo Research and Education Foundation (VREF).

Annexure A. Best equations of utility to Employees

Low income group Low middle income group

Forced Unforced Forced Unforced

LL Base -819.21 -1974.21 -1247.66 -1977.5

LL Model -811.87 -1579.90 -794.82 -1496.40

P2 adj 0.349 0.200 0.363 0.24

Observations 3600 5400 3600 5400

Utility Equation Coefficient P-value Coefficien P-value Coefficient P-value Coefficient P-value


Travel Time -0.023 0.000 -0.017 0.005 -0.005 0.003

Cost of transport -0.001 0.001 -0001 0.000 -0.001 0.011

Change in salary 0.002 0.000 1.976 0.000 0.001 0.000

Jobs Advertised -0.013 0.016

Distance (<5km) 1.528 0.000 0 0 1.605 0.000

Distance (<10km) 1.561 0.000 -0.301 0.002 1.681 0.000 1.020 0.000

Distance (<15km) 0.939 0.000 -0.444 0.000 0.575 0.019 0.512 0.000

Distance (<20km) 1.007 0.000 -0.495 0.000 0.892 0.000 0.442 0.000

Distance (<25km) -1.254 0.000 -0.521 0.000

Distance (<30km) -0.906 0.000 n/a 0

Distance (>20km) 0 0.281 0.097

ASC 0.74 0.006 2.359 0.000

Annexure B: Best equations of utility to Employers

Low income Low-middle income

Log likelihood Base -130.31 -130.31

Log likelihood Model -97.36 -115.00

P2 adj 0.24 0.10

Observations 376 376

Utility Equation Coefficient P-value Coefficient P-value

CO2 -6.46 0.01

Cost -0.05 0.02

Catchment <10km -9.51 -8.50

Catchment <15km +1.49 0.02 +2.38 0.00

Catchment <20km +3.91 0.01 +2.74 0.00

Catchment <40km +5.12 0.05 +4.38 0.01


Aspen Valley Hospital: Benefits Enrolment Guide. Retrieved November 23, 2011, from Summary_of_ Benefits_ Booklet.pdf

Banister, D., Marshall, S. & Blackledge, D.. 2007. Land Use and Transport: The Context in Marshall, S. and Banister D (ed) 2007. Land Use and transport. Elsevier. Oxford.

Benjaafar, S., Kim, J., & Vishwanadham, N. (2004). On the Effect of Product Variety in Production - Inventory Systems. Annals of Operations Research, 126, 71-101.

City of Trenton. (2011). Live Where You Work Trenton. Retrieved November 18, 2011, from

Coleman, C. (2000). Green commuter plans and the small employer: an investigation into the attitudes and policy of the small employer towards staff travel and green commuter plans. Transport Policy, 7(2), 139-148. doi:10.1016/S0967-070X(00)00003-2

Coombs,C. H.,& Avrunin,G. S. (1977). Single-peaked functions and the theory of preference. Psychological Review,84,216-230. Cited in Reutskaja, E., & Hogarth, R. M. (2009). Satisfaction in Choice as a Function of the Number of Alternatives : When " Goods Satiate ." Psychology & Marketing, 26(3), 197-203. doi:10.1002/mar

DEFRA (2011). 2011 Guidelines to Defra/DECC"s GHG Conversion Factors for Company 33 Reporting. Department of Energy and Climate Change.

Del Mistro, R., & Hensher, D. (2009). Upgrading Informal Settlements in South Africa: Policy, Rhetoric and what Residents really Value. Housing Studies, 24(3), 333-354.

Demographia (2010) Demographia World Urban Areas Population Projections, Supplement to 6th Annual Edition.

DNA Economics, & IMBEWU Sustainability Legal Specialists. (2012). Carbon Tax Discussion Forum (pp. 1-28).

Everson, J.A. and Fitzgerald, B.P.. 1971. Settlement Patterns. Longman, London.

Hall P.. 2010. Megacities, World Cities and Global Cities. in Buijs, S., Tan, W. and Tunas, D. 2010. Megacities: Exploring a Sustainable Future. 010 Publishers, Rotterdam.

Hall, P. & Pain, K.. 2006. The Polycentric Metropolis: Learning from Megacity Regions in Europe. Earthscan London.

Heerman K. (2014) Real Projected Gross Domestic Product (GDP) and Growth Rates of GDP for Baseline Countries/Regions (in billions of

2010 dollars) 2000-2030 sourceid=chrom-instant&rlz=1C1GGGE_ZA593ZA605&ion=1&espv=2&ie=

UTF-8#q=world%20bank%20gdp%20forecast%202030 Accessed 2015.09.15

Hensher, D. A., Rose, J. M., & Greene, W. H. (2009). Applied Choice Analysis. A Primer (pp. 1-742). New York: Cambridge University Press.

Horner, M. W.. 2002. Extensions to the concept of excess commuting, Environment and Planning A, 34, 543-566.

Iyengar, S. S., & Lepper, M. R. (2000). When Choice is Demotivating : Can One Desire Too Much of a Good Thing? Journal of Personality, 79(6), 995-1006.

Koelemeijer, K., & Oppewal, H. (1999). Assessing the Effects of Assortment and Ambience: A Choice Experimental Approach. Journal of Retailing, 75(3), 319-345

Koppelman, F. S., & Bhat, C. (2006). A Self Instructing Course in Mode Choice Modelling : 19 Multinomial and Nested Logit Models. (2010). Taxes & Qualified Transportation Fringe Benefits. Retrieved November 18, 2011, from

Ma, K.-R. and Banister, D.. 2006. Excess Commuting: A Critical Review Transport Reviews, 26 (6), 749-767

Nedbank. Integrated Report 2011. Retrieved 20 July 2012, from financial/Nedbank_ar2011/operational/environmental. asp

Pain, K. & Hall, P.. 2006. Flows and Relationships: Internal and External Linkages in Hall, P. and Pain, K. 2006. The Polycentric Metropolis: Learning from Megacity Regions in Europe. Earthscan, London.

Parr, J. B. (2002). The location of economic activity: Central place theory and the wider urban system. In P. McCann (Ed.), Industrial location economics. Cheltenham, UK & Northampton, MA, USA: Edward Elgar. Cited in Sridhar, K. S., & Wan, G. (2010). Firm location choice in cities : Evidence from China , India , and Brazil. China Economic Review, 21(1), 113-122. doi:10.1016/j.chieco.2009.11.003

Sasol (2011) Integrated Annual Report 2011. Retrieved June 2012, from 35 sasolinternet/downloads/Sasol_IR_2011_1328013405316.pdf

Scheibehenne, B., Greifeneder, R., & Todd, P. M. (2009). What Moderates the Too-Much-Choice Effect ? Psychology & Marketing, 26(3), 229253. doi:10.1002/mar

Schwartz, B. 2005b. Barry Schwartz: The Paradox of Choice [Video file]. Retrieved from

Shoup, D. C. (1997). Evaluating the effects of cashing out employer-paid parking : Eight case studies. Science, 4(4), 201-216.

Statistics South Africa (2001) South African National g Application of Compensatory Rules to Model Private Sector Land Use Investment. SA Transportation Conference, Pretoria. July 2003.

The R Project for Statistical Computing. Retrieved June 2013 from

Walters, J., 2008. Overview of Public Transport policy developments in South Africa. Research in Transportation Economics., Volume 22, pp. 98-108.

WCTRS. (2011) Putting Transport into Climate Change Agenda- Recommendations from WCTRS to COP. 17 December 2011.