Chapter 4. Conceptual and methodological framework
4.1 A conceptual framework
Chapter 3 revealed major incongruities in the results of the small number of previous studies that have explored active transport use in relation to income.
Further exploration of this relationship is clearly needed at both the conceptual and methodological level. This chapter will state what my prior expectations are regarding income’s influence on the use of active modes. I start with the historical evidence that suggests a negative relationship between active commuting and income. I then introduce spatial considerations, which lead me to quite a different consideration of the relationship.
4.1 A conceptual framework
How are workers’ travel-‐to-‐work decisions affected by their income? Other things equal, one might expect additional income would buy more comfort and ease of travel, and raise the opportunity cost of time.
If everyone lived the same distance from work and had equal access to transport modes, the national evidence would suggest the motorised transport option would be more attractive the higher the income. As successively higher incomes are encountered, I would expect to see walking and cycling replaced by motorised commutes.
The above interpretation implies that active transport is negatively elastic in income; its consumption falls as income rises. It is, in the language of economics, an inferior good.15 There is historical evidence for such a negative relationship between income and active commuting in the aggregate case (Jacobson et al, 2011; Milne and Abley, 2011). There is certainly strong evidence for the positive relationship between income and car use from panel surveys (Dargay, 2001, 2007). Far less attention has been paid to the cross sectional relationship between income and active commuting within the context of the local labour market.
15 This very example is used in a popular dictionary of economics: “…as people become richer, they may substitute more cars for bicycles, and bicycles would be regarded as an inferior good”
(Baxter and Rees, 1972, p.215).
Complicating such an interpretation of active transport are the demographics of work and income. Generally speaking, wages rise with age so higher income earners tend to be older. To the extent that age might impose more effort and cost on the active transport option, we might expect commuters to opt for non-‐
active modes as they age quite independently of their higher income.
There are further considerations such as hours of work. Part-‐time workers earn lower incomes and may for this reason be more likely to take the cheaper, active modes. They may also have more time and be willing to commute for longer. At the same time, part-‐time workers are much more likely to be women, which may influence the choice of active transport options especially if there are child care responsibilities to be considered, which entail both time and feasibility issues.
This brings me to an additional aspect to consider: the structure and dynamic of the household that a person lives in. The options for travel may be more highly constrained in a household of two young adults and several young children, compared to that of a single male or older couple without dependents. There are at least three primary constraints associated with the former. The first is time, the second is perceived safety and logistics associated with young children, and the third is resources. Take a young couple both working full time to pay off a mortgage in a suburban property twenty or more kilometres from both work places who also need to drop off and pick up children, do the shopping and run other errands but with only one car. In summary, discretion and choice over transport options are likely to vary markedly across households quite apart from the various characteristics of individuals, including their income level.
The issue is even more complicated when viewed in a wider context. When it comes to the choice of active transport, probably the most important of these wider decisions is where to live. This has several components. The first is the choice of the local labour market -‐ whether it is a major metropolitan centre, a medium sized town or a small village in a largely rural area.
The second component is residential location at a regional level. Regional differences are likely to affect the choice of commute mode in a number of ways.
Differences in average air temperatures, rainfall, wind speed and sunshine hours
will impact the desirability of engaging with the elements on foot or bicycle.
There are also the geographical differences between various regions, such as whether the commuter faces predominantly hilly or flat terrain. Varying levels of investment in pedestrian and cycling infrastructure across regions will also play a role in facilitating or constraining the use of active modes. Availability of public transport options (buses and trains) in different regions may also be relevant, as use of public transport often involves some degree of walking at either end.
Thirdly, those who have chosen to work in a large dense labour market face several distance-‐to-‐work options. Faced with a downtown work location and an a priori preference for a commute of say 20 minutes, one can choose to live within a few blocks and walk to work or live some 20 minutes drive away in a suburban location. Such decisions cannot be made independent of income of course and, other things equal, it is the higher income person who is most likely to be able to afford to live very close to a downtown work place.16 Their location decision at the same time opens up commuting alternatives not available to the lower income suburban dweller. For this reason, we may find that walking or cycling to work may rise with income, largely because higher income individuals can access residential locations closer to their workplace.
The decision to choose active transport therefore is not simply one of economics, but of economic geography. Making the picture even more complex are a myriad of other facets, among them culture, habit, autonomy and control, and relative preferences for physical health. All of this means there are two ways of looking at the choices involved in the use of active transport. The first is the standard modal choice framework, which is about the daily choice on what transport mode to use to get to and from work. This is the classic modal choice problem.
The second set of choices are the prior long term ones, but they may be just as crucial in framing the daily commute mode decision. Those decisions made some time ago effectively set the wider context in which the daily transport decision is made: the decision on the type of household, life style, region of residence, type of settlement and location within large settlements. In other
16Assuming they live in a city where employment is concentrated in the centre.
words, the whole gamut of past choices lead to the socio-‐economic and geographic context that frame the daily choice of commuting mode. These earlier choices, on where to live in relation to work, between urban centres and within large centres have a major constraining influence on the relative costs that feature as constraints in the typical model of modal choice.
Therefore, modal choice models as such do not capture the embedded nature of the commuting decision. Because modal choices are made in the context of a broader set of prior situational factors, they often result in habitual choices, which may not adjust for new information. When performing repetitive behaviours such as commuting to work, people may be likely to ignore new information, even when the information may rationally be deemed to be a relevant input in the decision-‐making process. To ignore the repetitive nature of commute mode choices may result in the formation of unrealistic assumptions about the reasoning that precedes such choices. To some extent, habit helps explain the observed predictive importance of situational variables such as socioeconomic characteristics and car ownership (Diana, 2010).
Also, the endogenous nature of underlying residential self-‐selection processes can make it tricky to evaluate causation among locational, temporal and individual elements, and associated outcomes. For example, a researcher might observe that suburban dwellers walk to work less and drive a car more than their urban counterparts. However what is difficult to determine is the extent to which the observed patterns of travel behaviour can be attributed to the settlement type itself, as opposed to the prior self-‐selection of residents into a built environment that is consistent with their predispositions toward certain travel modes and land use configurations (Mokhtarian and Cao, 2008).
Assertions regarding such causal mechanisms will always be questionable unless data is available that maps both the residential and commute mode choices of the same individuals over time. Typically, this type of longitudinal data is not available and certainly not in New Zealand.
Whether a negative relationship between active commuting and income applies to individuals over the course of their own working lives is not easily discernable
from the literature, as I demonstrated in chapter 3. Without the available longitudinal samples in New Zealand we are forced to rely on cross sectional evidence, and to look at the propensity to actively commute across a range of incomes at a point in time.
What is of immediate interest is the degree to which the available cross-‐sectional evidence from the NZHTS is consistent with a conceptual position that argues for a negative relationship between active commuting and income. In addition to the geographical considerations that can alter the way active commuting might relate to income, there are several other possibilities. For example, active transport may rise with income because of the impact of education. To the extent that higher incomes are associated with higher levels of education (about, for instance, the health benefits of active transport), healthy options can be expected to play an increasingly important role in people’s decisions about life style as their incomes rise. In this respect, one might also want to add a social consciousness and concern for environmental sustainability, both of which might be expected to rise, at least with the education component of rising income.
There is some support for this line of thought in the literature, though the role of affluence in explaining socially and environmentally-‐motivated actions is quite contentious. According to the affluence hypothesis, environmental quality is a luxury good that becomes of concern only when basic needs have been met (Duroy, 2008). It is thus assumed that income is the most important determinant and that affluent nations are more likely to display greater demand for environmental quality than developing nations (Meyer and Liebe, 2010). This argument is reminiscent of Maslow’s hierarchy of needs theory (1954), and also Inglehart’s Theory of Post-Materialist Values (1990, 1997), which postulates that, with growing prosperity in post-‐industrialized nations, people are freed from burdensome economic concerns and able to pursue other goals such as improved health and environmental sustainability (Duroy, 2008; Meyer and Liebe, 2010).
But the view that rising social and environmental concern are the result of
economic affluence is rejected by many authors (e.g. Martinez-‐Alier, 1995; Shiva and Jafri, 1998; Escobar, 2006), who have noted that, while concern for global issues such as climate change is higher in developed nations, grassroots movements and action at the local and community level are negatively correlated with GNP per capita -‐ i.e. stronger in poorer countries (Dunlap and Mertig, 1995;
Duroy, 2008).
Add to this the argument that affluence, which necessitates greater levels of production and consumption, is itself a major cause of environmental degradation. This could provide an explanation for why environmental concern might increase along with it. By this rationale, an increase in environmentally-‐
friendly behaviours, such as the use of non-‐motorised transport modes, among better-‐educated individuals could be expected. According to this argument active commuting will decline with income up to a point (as car ownership becomes possible), after which it will begin to increase, as people become better educated and more socially and environmentally-‐responsible.
In summary, modeling modal choice only involves modeling the immediate daily decision on how to get to work. As such, it ignores or takes as a given those prior decisions made at previous junctures in people’s lives. Many of those choices point to the crucial nature of earlier decisions that have nothing to do with active commuting per se. The simple act of deciding where to live in relation to the workplace (or, conversely, where to work in relation to the residential location) is possibly the most important of these ‘non active commuting’ decisions. Although the residential location decision certainly locks many individuals into particular commuting options, there usually remains some choice within these ‘external constraints’. I contend that the choice made within those constraints will be influenced by income.
But more importantly, I am interested in income because of its link to economic growth, whose primary purpose is to raise incomes. If raising incomes also lowers the propensity to use active transport for the daily commute then we may not have an economic growth model that is sustainable, either in terms of public health or environmental impact. If, however, I find that other characteristics of
income actually encourage active commuting (e.g. one in which settlement type and proximity to workplace are more closely associated with income) then we might be closer to a more sustainable type of economic growth. From the urban, local labour market perspective, the empirical relationship between active commuting and income becomes quite central.
Figure 4.1 attempts a more structured approach to displaying the complex web of variables surrounding the central relationship between income and active commuting that have been identified in my conceptual framework. I have created a directed acyclic diagram, which is an instrument useful for clarify thinking and making explicit underlying assumptions (Greenland, 1999).
Figure 4.1. Directed acyclic diagram outlining factors related to the relationship between income and active commuting.