• No results found

2.2 Food Retailers and socio-economic deprivation

2.2.1 Food and grocery retailers (FGRs) and SEDs

In Canada, there is strong evidence that deprived areas have better access to supermarkets and multiple retailers, bringing to question the validity of the notion of food deserts (Smoyer-Tomic et al., 2006; Apparicio et al., 2007; Gould et al., 2012; Lu and Qiu, 2015). These studies measured accessibility using presence or absence of retail outlets and various measures of distances to assess provisioning while socio-economic status was calculated

21 using socio-economic variables. Lu and Qiu (2015) found a positively significant association between children, senior citizens and low-income areas and food markets, with densely populated and inner-city localities having the best provisioning. In addition, Smoyer-Tomic et al. (2006), in their study in Edmonton, allude that although there are cases of under-provisioning of grocers in some poor neighbourhoods and inner-city localities, areas that are largely occupied by low-income households with no car have better access to supermarkets.

In contrast, similar studies found a somewhat different relationship between area SECs and food availability. A longitudinal study by Larson and Gilliland (2008) in Ontario found a complex relationship between food accessibility and a deprivation index created using lone parent family, incidence of low education, low education attainment and unemployment rate.

The study found that impoverished neighbourhoods had the poorest access to supermarkets while the mid group had the best access. Further analysis of public transport availability and food accessibility revealed similar trends, with the worst access in the most deprived and most affluent areas. Likewise, Black et al. (2011) discovered minority ethnic areas (predominantly African American and Hispanic neighbourhoods) had the worst access to supermarkets. However, overall, the study found that an increase in income reduced the likelihood of supermarkets. Black et al. (2011) further examined urban land systems, neighbourhoods and accessibility and concluded that the contextual explanation for the spatial distribution of food retailers lies in the prevailing land use regulations, as results show that the disparity between affluent and impoverished neighbourhoods can largely explained by zoning regulations. This introduces a new perspective in the literature on spatial access to food and neighbourhood characteristics i.e. urban land use and planning policies further shape retail provisioning in Canada.

In the US, there is a divide in the literature on food provisioning. Zenk et al. (2005), in their study which examined the relationship between area racial composition, poverty, and supermarket provisioning, found more limited accessibility in poor areas compared to affluent neighbourhoods. They further alluded that even within poor neighbourhoods, area racial characteristics further influenced supermarket accessibility, with areas having large minority ethnic groups (African-Americans) having the poorest access by having to commute farther distances to a supermarket compared to the impoverished white neighbourhoods. Similarly, another study that explored the notion of ‘food deserts’ in New York using income and racial composition divides, found a negative relationship between area distress and supermarket availability (Gordon et al., 2011). Predominantly white

22 middle-income areas had the best access to supermarkets compared to Black and Hispanic neighbourhoods. Furthermore, Bower at al. (2014), in their nationwide analysis of food availability in relation to area racial composition, poverty and urban/rural characteristics, revealed a significant negative relationship, with fewer supermarkets as area SECs worsened.

More importantly, ethnic composition plays a significant role in supermarket access with areas dominated by persons of Black ethnic origin having the lowest access to supermarkets at all levels of socio-economic classification while areas dominated by Hispanic and White groups had better access across the country. Lamichhane et al. (2013) further stress the presence of food deserts, with supermarket provisioning limited in ethnic minority and deprived communities. The study further identified that affluent areas and areas with low minority representation have better access to supermarkets.

Conversely, Sharkey and Horel (2008) examined the relationship between socio-economic deprivation and ethnic minority composition in relation to disparity in food access in Texas.

The study revealed that neighbourhoods with the highest levels of socio-economic deprivation and racial composition had the best access to food retailers, even after stratifying with population density. Likewise, Raja et al. (2008) examined the relationship between the spatial distribution of supermarkets and area racial composition, highlighting that disparities exist in food access across ethnic lines with predominantly white neighbourhoods having better access to supermarkets compared to ethnic minority neighbourhoods. Findings also indicated that despite the disparities, these ethnic minority neighbourhoods did had access to smaller retailers and stores which compensated for the absence of supermarkets. Therefore, it would be inaccurate to conclude that inequality exists based on racial differences. A more recent study (Myers et al., 2016) found that co-location is a major factor that accounts for retail distribution patterns. This study found better access to supermarket in very deprived neighbourhoods, although these areas also had the highest presence of fast food, recreation centres and parks.

New Zealand and Australia have also had a fair share of studies attempting to examine clear cut linkages between area deprivation and retail provisioning. Wiki et al. (2019) examined whether or not the patterning of food provisioning conforms to a social gradient. Results reveal dense food provisioning (fast food and supermarkets) in deprived areas. The study further implies that central areas with high population density, large commercial spaces and businesses had the best provisioning. The study concludes that although socio-economic deprivation drives retail location preferences, other factors such as accessibility, urban

23 zoning regulations and land values are also determinants of the observed patterns. In addition, another comparative study (Pearce et al., 2007) which compared the patterns of fast food and super market provisioning in New Zealand along socio-economic disparities detected a similar pattern, with inhabitants of deprived neighbourhood travelling the shortest distances to food retailers, even after rural urban stratification. The authors further explained that higher consumer demand and resistance to retail establishments in deprived and affluent areas, respectively, might be one of the explanatory mechanisms. It was further conjectured that land values, planning, historical neighbourhood antecedents and population density might be other possible influences.

Similar evidence was uncovered by Pearce et al. (2008) in New Zealand which further examined food patterning and other retail uses along different socio-economic criteria. The study dismissed the notion that deprived neighbourhood had limited access to food retailers.

Rather, it discovered the best provisioning in the most deprived neighbourhoods and also concluded that similar factors identified in previous literature (Pearce et al., 2007; Wiki et al., 2019) exert considerable pressure on patterns of food retailers. A comparative study (Burns and Inglis, 2007) in Melbourne, which also compared food provisioning (fast food and supermarkets), discovered a different situation compared to the New Zealand studies (Wiki et al., 2019). In this study (Burns and Inglis, 2007), access to supermarket largely depended on car ownership, with deprived neighbourhoods having to travel farther distances compared to affluent communities. This disparity in findings might relate to definition of food provisioning as the study did not include independent supermarket retailers.

Irrespectively, this raises salient questions for health practitioners.

Evidence in the UK is mixed, based on findings from different studies. Cummins and Macintyre’s (1999; 2002) empirical research in Glasgow on price and food availability along areas’ socio-economic gradient, identified an even geographical distribution in food provisioning across central parts of the study area. Statistical findings further evidenced a concentration of discounters and multiple supermarkets in poor communities compared to affluent areas, contrary to the notion of “food deserts” amplified by many scholars.

Notwithstanding, the authors called for similar studies in other areas in the UK to further confirm the results of their study and further enlighten the food desert debate. Likewise, Smith et al. (2009) carried out an analysis of food environment and area deprivation by classifying areas based on socio-economic deprivation and along an urban and rural divide in Scotland. The study found a contrary situation to the “deprivation amplification”

24 hypothesis which emphasises that inhabitants of impoverished communities have limited access to grocery stores. Stratified analysis across the rural/urban divide evidenced that urban areas across all socio-economic groups had better access to food retailers compared to their counterparts in rural area neighbourhoods, leading to the conclusion that linkages between SED and food provisioning also varies along environmental characteristics. In addition, McDonald et al. (2009) found patterns in their study, with deprived localities having best access to different ranges of food retailers. Furthermore, important site location determinants such as accessibility and distance to central business districts and other establishments seem to play a role in the observed patterns.

A repeated cross-sectional study (Maguire et al., 2015) which examined supermarket outlets density in Norfolk over an 18-year period (1990 – 2008) in relation to area SED, found that supermarket density increased over time in both impoverished and affluent neighbourhoods and there were no significant linkages between SED and food provisioning. Black et al.

(2012) examined food varieties and neighbourhood SED in southern England. To measure deprivation, the study reclassified the income domain of the index of multiple deprivation data (2007) from deciles 1 – 10 deciles to 1 – 5 quintiles and counts of food outlets (supermarkets, world stores, forecourts and convenience stores) per LSOA to measure food availability. Interestingly, the results of the study revealed that the mid-deprived LSOAs had the highest supermarket and food outlet provision. Clarke et al. (2002) identified that, although food provisioning was accessible in many parts of Cardiff, the study unearthed poor provisioning in some deprived localities in Leeds and Bradford. Clearly put, “the series of

‘what if’ analyses have shown that planners must consider the full set of implications when attempting to alleviate ‘food deserts’ because of the trade-off between increased provision per household and deflections from existing stores. Problems could be increased in small areas of deprivation and low mobility causing closure of local shops” (Clarke et al., 2002, p.

2059).

To follow up this study, Guy et al. (2004) explored the notion of “food deserts” using longitudinal data (1989 – 2001), by examining changes in the retail floor space per household ratio, new store openings and deprivation in Cardiff. Low socio-economic status areas had better access to food retailers and, in respect to store closures, deprived areas experienced more store closures which had little or no impact on food provisioning in these areas. In addition, Guy et al. (2004) used a spatial interaction model, which incorporated demand, supply and interaction to model consumer behaviour. The results of the analysis showed that

25 although provision greatly increased for low socio-economic status areas, affluent areas seem to have better provisioning over time, which further emphasises linkages in food provisioning and neighbourhood characteristics. A recent paper (Burgoine et al., 2017) provides further evidence to suggest that systematic patterns of “food deserts” are present within the UK food landscape, with systematic absence in deprived localities. The study identified that education deprivation is a major determinant of access to supermarkets, with inhabitants of areas with high proportions of persons with low education having to travel the farthest distances to supermarkets.

From the above review of the literature on food availability and socio-economic deprivation, it can be concluded that evidence around the world (i.e. findings from the UK, Canada, New Zealand and Australia) revealed mixed findings. Some literatures debunk the notion of ‘food deserts’ or the ‘deprivation amplification’ hypothesis and allege that no systematic patterns of poor access to food retailers were identified in deprived areas, while others raise questions and suggest that within deprived areas, level of provisioning varies, with some deprived neighbourhoods having challenges and limited options to food retailers thus calling for intervention to improve food provisioning in poor areas.

The studies reviewed above have several limitations which are discussed here. These studies failed to carry out a critical comparative analysis of the location preferences of these FGR retailers across different localities by comparing FGRs with other retail types (Guy et al., 2004; Zenk et al., 2005; Smoyer-Tomic et al., 2006; Apparicio et al., 2007; Larson and Gilliland, 2008; McDonald et al., 2009; Black et al., 2011; Gordon et al., 2011; Gould et al., 2012; Black et al., 2012; Burgoine et al., 2017). In addition, some of the studies examined the location characteristics of different retail types but in a single geographical location (Pearce et al., 2007; Smoyer-Tomic, 2008; Pearce et al., 2008; Lamichhane et al., 2013;

Maguire et al., 2015; Myers et al., 2016; Wiki et al., 2019). A comparative analysis of retail preferences looking at different geographical locations and different retail types would further unearth any clear linkages between food provisioning and area deprivation because of seemingly complex patterns.

In addition, the deprivation measures adopted by some of the studies such as Zenk et al., 2005; Smoyer-Tomic et al., 2006; Larsen and Gilliland, 2008; MacDonald et al 2009; Bower et al., 2014; Lu and Qiu et al., 2015; Gould et al., 2015; and Wiki et al., 2019had limited variables. For example, Lu and Qiu et al., 2015 only considered senior citizens, children

26 under 15 and low income, whereas, Smoyer-Tomic et al., 2006 adopted low income, low vehicle ownership, low income and high proportions of elderly. From the literature review, deprivation is an all-encompassing term and it relates to lack of access to resources brought about by SECs such as employment, income, ethnicity and education. In addition, adoption of composite measures (Gould et al., 2012; Maguire et al., 2015; Wiki et al., 2019;

Macdonald et al, 2009) do not allow for a clear understanding of the influence of individual characteristics on FGR locations.

Location plays a pivotal role in retail success and optimum locations are those that offer accessibility, demand, favourable regulations and, especially, vacant premises.

Unfortunately, many of the studies (Cummins and Macintyre,1999; Zenk et al., 2005;

Smoyer-Tomic et al., 2006 ; Apparicio et al., 2007; Larsen and Gilliland, 2008; Macdonald et al., 2009 ; Maguire et al., 2015; Lu and Qiu, 2015) did not attempt to examine the influence of commercialisation and land use patterns on FGR location, although the studies acknowledged that other factors such as proximity to central business district and planning regulations might influence FGR locations.