The second major section of this review explores the literature on per capita income in relation to private motorised mobility at the city level. Firstly, a broad overview is presented that describes three prominent urban research studies on the demographics, economics, urban form, transportation modes and transportation networks of a wide selection of world cities (Newman and Kenworthy, 1989; Kenworthy et al., 1999; Kenworthy and Laube, 2001). An important aspect is the historical nature of the data collected, over several decades from the 1960s to the 1990s, that allows researchers to undertake time-series trend analysis on a large variety of robust data. The importance of these comprehensive urban studies is that they have encouraged further research, and through the development of new methodologies of collecting and standardisation of data, they have allowed examination of cross-city macro-level indicators or measures that drive a city’s automobile dependence.
These studies collected vast quantities demographic, economic, urban form, and transportation data from a wide selection of cities in more developed and less developed countries from 1960 to 1995. In developing a model that predicts private motorised mobility it is essential to gain a clear understanding of these data and their relationships to private motorised mobility.
The following discussion provides a brief outline of three urban studies that bring together the data just described. First, a study by Newman and Kenworthy (1989) collected these data from a selection of 32 cities in Australia, North America, Europe and wealthy parts of Asia between 1960 and 1980 and published them in ‘Cities and Automobile Dependence, An International Sourcebook’. The importance of this study was that it provided a somewhat groundbreaking quantitative examination and analysis of a series of urban measures that offer insights into urban automobile dependence. These researchers describe a series of bivariate linear regressions between a variety of physical factors and private passenger vehicle ownership and use in cities. The changes suggested to assist in lowering automobile dependence in cities were to increase their land use intensity, orientate their transportation infrastructure more to public transport, walking and bicycling; strengthen their degree of centralisation; expand public transport; services and restrain high-speed traffic flows.
The second study of 46 international cities from 1960 to 1990 by Kenworthy et al. (1999), updated and expanded the initial work of Newman and Kenworthy (1989) into a series of lower income cities in South-East and East Asia. It also improved the understanding of economic measures that may help to explain automobile dependence. The study suggests that, while being
important, economic measures such as per capita income, may not be as important or robust in explaining the level of automobile dependence in cities as other more physical measures noted in the earlier work.
The third major urban research study was the ‘Millennium Cities Database for Sustainable Transport’ (Kenworthy and Laube, 2001). This study presents a comprehensive collection of city level data in 1995-1996 from a selection of 100 cities in the United States, Australia and New Zealand, Canada, Western Europe, Asia (high and low-income cities), Eastern Europe, the Middle East, Latin America, Africa and China (Kenworthy and Laube, 2001). A doubling of cities together with a more extensive data set especially greatly expanded economic data and a broader geographical coverage, provides an ideal platform to further examine relationships between a wide variety of urban factors and automobile ownership and use at the city level. The urban data from more developed and less developed countries reveals a vast difference in the use of all transportation modes between cities, including private automobiles, motorcycles, taxis and shared taxis, public transport modes and walking and bicycling. The Millennium Cities study, while enhancing and broadening the findings of the previous two urban studies, again relies on bivariate linear regression analyses to establish statistical relationships between the various factors. One difference is that, this time, the authors fitted the best-fit curves to their regressions (linear, logarithmic, exponential or power functions). This study delivers a series of similar findings, but importantly develops these findings in cities with a wider disparity in all facets of population, wealth, urban form and transportation modes. In part, it asserts that wealth (GDP per capita) does not alone provide a consistent or satisfactory measure capable of explaining urban automobile dependence.
These three urban studies are essential to this thesis, not only because of the comprehensive and wide ranging nature of the data they contain, but more importantly in the standardised data collection methodologies and procedures used that allow direct comparisons between cities and across time. The data sets in each study were compiled by the study authors who visited most study cities at least once, a uniform definition of terms was applied through a data methodology handbook and strict data quality control and cross checking procedures were implemented (Kenworthy et al., 1999:9-15). No data were accepted into the data set without rigorous reality checking against other relevant factors or results from other similar cities. Much incorrect data were found and returned to suppliers in each city for clarification and correction.
The recent urban studies described here strongly suggest that, unlike the national level analyses where per capita income seems to be very dominant, automobile dependence is not driven by a
sole measure, but by a number of measures mutually acting in combination. After an investigation of what determines automobile dependence in the urban environment of Boston, USA, Zhang’s (2002) contribution to this premise was to conclude that measures do not act alone in driving automobile dependence. Zhang notes that Boston’s suburban population shows an overwhelming dependence on private automobiles for their daily activities. He also notes that the factors driving automobile dependence were a combination of complex and diverse factors that vary among and within differing sectors of Boston’s population. For example, automobile dependence appeared to be influenced by a combination of per capita income and residential location. This study revealed that lower income drivers were more dependent on automobiles than the higher income drivers because their jobs demand frequent and flexible travel typically in the non-peak hours when public transportation services, if any, are less likely to be available. Another important factor driving this high level of automobile dependence was the configuration of Boston’s alternative transport routes. This transportation pattern was predominantly radial between the suburbs and the downtown or city centre. As there was no viable non-automobile alternative transportation available for suburb-to-suburb travel, Boston’s population had no other means of travelling than in their automobiles.
Litman (1999 and 2002) also makes a significant contribution to the literature by proposing that no single measure drives automobile dependence, but that private motorised mobility levels result from various cumulative effects of transportation systems and land-use patterns. These measures tend to be mutually supportive or synergistic, so it is generally inappropriate to consider any one measure as the prime cause of automobile dependence. Litman’s (2002) measures associated with automobile dependence are shown in Table 2.11.
Table 2.11 – Litman’s measures that drive and qualify automobile dependence at the city level
Qualifier of automobile dependence Measure Description
Low Medium High
Vehicle ownership
Per capita motor vehicle ownership (per 1,000 persons)
Less than 250 per 1,000 person
250 to 450 Plus 450
Vehicle use Per capita annual
motor vehicle mileage (VMT) Less than 4,000 miles (6,500 km) 4,000-8,000 miles (6,500 to 13,000 km) 8,000 Plus (13,000 km)
Vehicle trips Automobile trips as a
portion of total personal trips
Less than 50% 50 to 80% Plus 80%
Quality of alternatives
Convenience, speed, comfort, affordability
and prestige of walking, cycling and public transit relative
to driving Alternative modes are of competitive quality Alternative modes are somewhat inferior Alternative modes are very inferior
Relative mobility of non-drivers Mobility of personal travel by non-drivers compared with drivers Non-drivers are not severely disadvantaged Non-drivers are moderately disadvantaged Non-drivers are severely disadvantaged Market distortions favouring automobile use Relative advantage provided to automobiles over other modes from planning, funding, tax
policy etc Minimal bias favouring automobile travel Moderate bias favouring automobile travel Significant bias favouring automobile travel Source: Litman (2002).
Table 2.11 provides a summary of the measures that Litman identified as drivers of automobile dependence. One of Litman’s premises is that automobile dependence is a matter of degree. The move from low through to high automobile dependence is qualified by a change in impact of the measure, and it appears that several measures need to act in combination for this change to occur. The preceding overview describes a series of relationships between automobile ownership and use and a variety of other factors. In the main, the factors were a city’s demographic structure, its economics (per capita income) and its urban form. However, automobile ownership and use should not be considered as being solely driven by a combination of such factors. Factors such as culture, status, image, symbolism and gender and other socio-economic variables also contribute to the picture, though many of these are much more difficult to specify or utilise at a macro-city level (Stokes and Hallett, 1990; Stokes et al., 1991; OECD, 1996 and 1997; Pucher and Lefèvre, 1996; Battelle; 2000). It is also a fact that automobile dependence is often ‘enforced’ by the lack of quality alternatives to the automobile, as suggested in Table 2.11. In this environment, people generally see automobiles as providing independence with personal advantages of immediate convenience, increased personal security and control over their personal mobility (Goodwin, 1995; Gray et al., 2001).