The relationship between key demographic profile descriptors and the propensity for inshopping and outshopping by Sowetan residents

11Soweto was one of the largest disadvantaged townships in South Africa, and a unique pattern of outshopping originated due to the lack of retail investment and development in the area. After 1994, Soweto as a township benefited more than any other South African township through retail development, resulting in a major shift in shopping patterns towards buying inside the township (inshopping). This change in shopping pattern provides the focus of the research problem, namely to understand the changes in buying behaviour and certain retail patronage practices of Sowetan residents. The main aim of the study was to investigate, firstly, the profile of inshoppers and outshoppers and, secondly, to examine the relationship between certain profile components (education, income, car ownership and duration of residency) and (i) inshopping, and (ii) outshopping propensity as examples of changes in retail patronage in the Soweto township. In this study, a descriptive research design was used. A disproportionate stratified sample of Soweto households was selected and interviewed, consisting of 690 households spread over 11 sub-areas of Soweto. There is a marked difference between the profiles of inand outshoppers living in Soweto in terms of income and educational levels, car ownership and duration of residency in the area. The value and contribution of the study lies in the fact that some of these findings correlate with findings in other countries of the world; however, there are also a number of major differences in the profiles. South African investors and retailers should take cognisance of these differences and adapt their retail strategies accordingly in their efforts to market successfully in the Soweto market. Prof. J.W. Strydom is in the Department of Business Management, University of South Africa. Email: Strydjw@unisa.ac.za

1R etail innovation has always been seen as the way forward for emerging countries to uplift their economically and socially disadvantaged communities (Reinartz, Dellaert, Krafft, Kumar & Varadarajan 2011;Kurilla & Joshi 2010).One of the vehicles to effect this was through the development of large shopping malls, especially in disadvantaged areas.This was done in conjunction with the upliftment of these communities by improving job opportunities and through the investment in social infrastructure such as roads and other municipal amenities by private investors and the South African government.This was the case in South Africa and especially in Soweto where the residents have a vast array of new retail facilities available, as manifested in the opening of the Jabulani and Maponya shopping centres in the past few years.Gone are the days when residents of Soweto had to travel long distances to access modern retail facilities in the central city areas of Johannesburg or in upmarket suburbs such as Sandton.Sowetan residents at that stage had to travel far to buy products and services and thus spent a sizeable chunk of their disposable income outside the borders of Soweto.Retail activities that entail buying outside the borders of the community where they live are called 'outshopping' (Paddison & Calderwood 2007;Dunne & Lusch 2008).
Retailers would obviously like their customers to shop locally, which is termed 'inshopping', and implies that the customer deliberately patronises the retail institutions in the area in which he or she resides (Mullis & Kim 2011).The concepts of outshopping and inshopping form the focus of this article, examining the changing shopping patterns and retail patronage behaviour of Sowetan residents.There has been a major shift in shopping patterns towards buying inside the township (inshopping) that needs to be investigated, as it impacts on the retailers operating inside the township.The research problem focuses on this issue and tries to understand the changes in buying behaviour and retail patronage practices among Sowetan residents by examining the relationship between key demographic profile descriptors and the propensity for outshopping and inshopping.
Various reasons exist for the research focus on Soweto, the most important being that this township has been the biggest beneficiary of retail infrastructure development since 1994 and has the largest number of consumers (1.2 million).These consumers live within five minutes' drive from Soweto's major shopping malls such as the Maponya Mall, which is one of the largest malls in South Africa (Zondi 2011;Anonymous 2012).
The concept of retail patronage attempts to explain the process of retail institutional choice by the customer and can be defined as "all the possible inner features of dynamism around the shopping behaviour phenomenon in terms of store choice" (Lee 2009;Manana 2009).Customer profile is defined as the most important distinguishing characteristic of inshoppers and outshoppers (Jooste, Strydom, Berndt & Du Plessis 2012).The aim of this article is to investigate the profile of inshoppers and outshoppers and to examine the relationship between the key identified profile components (education, income, car ownership and duration of residency) and (i) inshopping and (ii) outshopping propensity of previously disadvantaged South African shoppers in general.More specifically, the aim in this article is to examine the previously disadvantaged township of Soweto, which has only recently acquired modern retail facilities (in the form of shopping centres and brand name stores).The discussion commences with a description of what outshopping and inshopping entail and refers to global trends related to these shopping behaviours.This will be followed by an explanation of the research methodology followed in this study.The results will be discussed and then compared with international shopping occurrences, after which certain conclusions are drawn.
The findings of the study are expected to contribute to the existing retailing literature on inshopping and outshopping practices in South Africa's townships, of which Soweto is the biggest.This will provide useful information for retailers doing business within the Soweto township and for national and international retail investors who are contemplating new retail infrastructures in the area.

The phenomenon of outshopping
1O utshopping, also known as market leakage, occurs in various arenas, and ranges from international outshopping to domestic outshopping (Piron 2002, Wayland, Simpson & Kemmerer 2003;Zondi 2011).International outshopping occurs across the borders of different countries.An example is Chinese citizens -it is estimated that about two million travel to Europe and one million to the Americas every year to buy products and services (Wang, Doss, Guo & Li 2010).However, the majority of Chinese shoppers who engage in outshopping travel mostly to neighbouring Asian destinations (referred to as transborder outshopping), such as Hong Kong, Singapore, South Korea and Japan.In various other countries, transborder outshopping occurs, for example, in Europe, America, Hong Kong, Singapore, South Korea (Wang et al. 2010) and Malaysia (Piron 2002).In Southern Africa, Zimbabwean citizens travel to South Africa to buy products and services to take back home (Anonymous 2007).In most cases, however, outshopping is seen as a more limited shopping phenomenon that occurs inside a country.The most obvious form is the age-old tradition of rural citizens travelling to larger towns and cities to trade and do business (Piron 2002;Wayland et al. 2003).Rural areas have fewer retail facilities and retail options, resulting in rural retailers that have served the local community for years being put under severe strain simply to survive when local residents start buying outside the area (outshopping).Rural retailers are also under threat from increased competition by national retailers and even global retail groups such as Wal-Mart (Worstall 2013).These chains develop innovative smaller versions of their successful hypermarkets and supermarkets to cater for customers in larger towns, thus attracting outshoppers in the surrounding smaller rural towns and further threatening the continued existence of rural retailers.

Profile of outshoppers
1Th e general profile of an outshopper has been researched by various authors (see Table 1 for a summary of the main findings of research by various authors) and found to be younger and more affluent.The type of product purchased has been found to be those with high status, and speciality products such as home furnishings, furniture, electrical appliances and jewellery (Ashley-Cotleur, Gaumer & Foltos 2009).Outshopping customers are described as decision-makers, mostly with higher educational qualifications (Jarratt 2000).Product-related factors were also indicated as an important indicator of outshopping behaviour.Piron (2002) states that shopping goods and speciality goods in particular were mostly bought inside the town or municipal area where the customer resides.As far as price is concerned, it was reported that higher-priced products were more readily purchased during outshopping trips and that the outshopper felt that the perceived process and selection options were better outside the local community (Wayland et al. 2003).Piron (2002) further states that consumers' dissatisfaction with local shopping conditions encouraged outshopping.By and large, outshoppers are not very loyal to local retailers and do not feel obliged to support local establishments (Qiu, Maksymiuk & Bruning 2008).Powe & Hart (2009) state that car ownership was also a significant contributor to outshopping activities.Length of residency in the community was not regarded as a determining factor in outshopping (Singh 2011).The literature review on outshopping can be summarised by stating that, in general, there is a positive association between income and education levels and outshopping behaviour, while there is generally a negative association between age and tenure in the community and the tendency to do outshopping (Jarratt 2000).Local retailers would obviously prefer customers to buy and support local businesses to ward off the encroachment of national and international retailers.This retail patronage behaviour is known as inshopping, and will now be discussed in more detail.
The phenomenon of inshopping 1I nshoppers are defined as those shoppers who patronise retail facilities in the area in which they live (Kim & Stoel 2010;Mullis & Kim 2011;Cole & Clowe 2011).Inshoppers are therefore loyal to local retailers and are welcomed by local retailers, as their shopping activities sustain the domestic retailing fraternity.Inshopping usually occurs in rural areas, where local retailers find themselves competing with regional shopping malls in the larger towns in the area (Ashley-Cotleur et al. 2009).This patronage behaviour occurs because of the worldwide trend towards retail concentration, where there has been a spatial and organisational move to geographically concentrated areas by a small group of large national and international retailers (Powe & Hart 2009).Rural retailers are especially disadvantaged, because this leaves them with fewer customers, and customers who less often shop locally and who spend less per visit (Powe & Hart 2009).This obviously is detrimental to the economy of the rural community, since these retailers are some of the main job creators in the community (Bhat & Fox 1996;Vias 2004).

Profile of inshoppers
1Th e profile of inshoppers is that of consumers who are less mobile, slightly older, and have an income and socioeconomic profile that is lower than the profile for outshoppers.Inshoppers often have positive feelings (store loyalty) towards local retailers (Qiu et al. 2008, Mullis & Kim 2011).It is reported that, in general, inshoppers either do not own a car or they find it more difficult to travel to the shops; inshoppers also tend to buy convenience products (Qiu et al. 2008).Jarratt (2000) refers to the negative association that exists between age and living in the area: inshoppers are older and have resided for a longer period in the area than outshoppers.
The same threats facing rural retailers worldwide are also prevalent in South Africa; in South Africa today, there is a marked outflow of retail purchasing from the rural to the urban areas, and this is also true of the township areas of South Africa (Ravhugoni & Ngobese 2010;Ruhigga 2011).For a summary of the findings of the theoretical discussion on outshopping and inshopping, see Table 1.
Having examined the retail profile of consumers in terms of outshopping and inshopping in other parts of the world, the focus is now on outshopping and inshopping patterns in South Africa.(Jarratt 2000) mmccxxxviiiS tore loyalty: mmccxxxixH old positive feelings towards local retailers and support these retailers (Qiu et al. 2008) mmccxlD issatisfaction with local shopping conditions (Piron, 2002;Qiu et al. 2008).mmccxliM ethod of transport: mmccxliiL ess mobile with lower levels of car ownership (Qiu et al. 2008) mmccxliiiH igher levels of car ownership (Powe & Hart 2009) mmccxlivL ength of residency: mmccxlvR esiding for a longer period in the area (Jarratt 2000) mmccxlvi No influence of the length of residency reported on outshopping (Singh 2011) Outshopping and inshopping patterns in South Africa 1O utshopping has always been of concern for the South African retail community, with the move towards urbanisation and the shift of buying power away from rural to urban and metropolitan areas.There has been a general decline in retail activities in rural areas, a decline that dates back decades (Du Plessis, Strydom & Jooste 2012).Evidence of this is the general decline in the number of wholesale and retail facilities in rural areas and increasing informalisation of retail activities, resulting in the growth of informal street trading (Du Plessis et al. 2012).The result has been the reduction of patronage levels in small towns and rural areas and the subsequent reinforcement of outshopping pattern by consumers (Ravhugoni & Ngobese 2010;Ruhigga 2011).In South Africa, there are special reasons for outshopping activities in the townships, driven by socio-political influences such as the previous political dispensation of apartheid.The development of separate black residential areas (known as townships) in South Africa (of which Soweto is one of the largest), and the restriction of formal retail business development in these areas left a void that was only rectified after the new democratic government came to power.It also resulted in township consumers being underserviced, with only rudimentary retail services being provided -this, in itself, has encouraged residents to engage in outshopping.Further reasons for outshopping in the townships, apart from the lack of alternative retail institutions, were a limited selection of merchandise, poor service and generally higher prices for products.This forced township customers to travel further distances to the surrounding cities to shop for the products and services they wanted (Strydom 2011).The limited number of formal retailers in South Africa's townships was either clustered in small formal neighbourhood centres or, alternatively, these retailers operated as general retailers.There were, however, also a growing number of informal retailers and informal retail structures, of which spaza shops (a convenience retailer operating from a room in a house), hawkers (selling mostly perishable products) and shebeens (selling beer and other forms of liquor) are the most visible examples.In the townships, retail businesses were scattered throughout the area (Adatia 2010).
The new government decided on a review of the commercial sectors in the townships so that these could be properly planned and developed.This opened up opportunities for the development of new retail infrastructure opportunities such as shopping centres and regional shopping malls.The development of these new retail facilities was of major importance also to the existing trend of outshopping by Sowetan residents.In a study conducted in 2004 in Soweto, research indicated that Sowetan households at the time spent only 25% of their retail expenditure at retail outlets situated in the township, while 75% was spent on outshopping outlets (Kloppers 2009).
Further research conducted in 2005 corroborated these figures, and estimated the demand for retail goods in the Soweto area during this period to be R4.2 billion, of which only R1.05 billion was spent inside Soweto (inshopping) by high-income groups.The results showed that it was the lower household income groups that spent the most of their disposable income inside Soweto.In fact, it was reported that, in 2005, households with a monthly income of R800 spent 49% of their income inside Soweto (Group 2005).This research preceded the opening of the Jabulani and Maponya shopping malls in Soweto, which proved to be of major importance in convincing residents to shop inside the Soweto area (inshopping).Maponya Mall is a major regional shopping centre that houses 170 shops with national chain anchor tenants such as Pick n Pay, Woolworths, the Foschini Group, Truworths and Ackermans.Jabulani is a smaller regional centre with 107 retail shops.By 2005, after the opening of these two centres, total retail floor space increased by 62% in this area (Zondi 2011).
As mentioned earlier, the new (post-1994) government focused particular attention on rectifying retail infrastructure deficiencies and addressing the needs of residents in the townships.The government's economic strategy was to increase black economic empowerment and black participation in South Africa's mainstream economy.This was reflected in the development of shopping infrastructure preferably by blackowned businesses in the townships, and included the major development of retail infrastructure in South African townships such as Soweto.
In Soweto, at least six shopping centres were developed after 2005, namely the Jabulani, Bara, Dobsonville, Maponya and Protea Gardens centres, and the Protea Glen Mall in 2012 (Sibanyoni 2012).These developments occurred in tandem with the increase in spending power of black township consumers and the fact that most middle-income township residents decided to remain in the township (Tustin & Strydom 2006;Kohler 2010).Ligthelm (2008) reported on two profile descriptors of Sowetans, namely income (that the townships have a fairly stable middle-income population) and duration of residence (that more than 60% have resided in the township for a period of more than ten years).
With the development of these new shopping centres, retail patronage patterns have also changed, and township malls have become favourite shopping destinations.A report prepared for Commuta-Net (Thys 2009: 5) indicated that township shopping centres in Pretoria attracted 56.25% of the local population's spending power, whilst the figure was 52.17% for Johannesburg.Research for Commuta-Net also found that it was primarily the Living Standards Measure (LSM) groupings of four to seven that supported these township retail facilities (income profile descriptor), and that the major reason for inshopping was convenience and the lack of transport, which made inshopping more attractive (car ownership profile descriptor).
From the discussion so far, it is clear that there has been a major shift in shopping patterns in the township areas, from buying mostly outside the townships (outshopping) towards buying inside the townships (inshopping).It was also mentioned that three profile descriptors could be identified, namely income, duration of residence and car ownership.Level of education was also implied in the international comparison by Jarratt (2000) as a profile descriptor for outshopping.From the summary in Table 1, it is clear that not all the profile descriptors discussed in the theory could be tested in Soweto, as the original survey design focused on the shopping behaviour of households (Tustin 2008).In order to better understand current retail patronage practices in the Soweto township, some profile descriptors (identified from the literature sources) and the effect of these profile descriptors on inshopping and outshopping propensity, need to be examined.Hence the following two research objectives: 1. To obtain a profile of inshoppers and outshoppers in terms of the key identified profile descriptors: level of income, level of education, car ownership and length of residency.
2. To investigate the relationship between these profile descriptors (income levels, levels of education, car ownership and length of residency) and the inshopping and outshopping propensity of Sowetan consumers.
1Th e next step is to develop the research hypotheses.The research hypotheses for inshopping can be formulated as follows: 1H 1: There is a relationship between the propensity for inshopping in Soweto by township residents and (i) Income levels (ii) Education levels (iii) Length of residency in the Soweto township (iv) Car ownership.
1Th e research hypotheses for outshopping can now be formulated.

Research design
1G iven the dynamic nature of retail developments in South African townships, a specific approach was needed in order to conduct this research.For this study, both exploratory and descriptive research designs were used.Soweto was chosen as a baseline and test market owing to the number of retail developments that have occurred in this geographic area (particularly since 2005).Since the research study is the first of its kind, it was decided to conduct exploratory research into the existing retail infrastructure in Soweto.This process involved, firstly, an exploration of the size of the Sowetan consumer market, focusing primarily on the number of people residing in Soweto.This process was further complemented by obtaining a spatial orientation of existing retail shopping malls/centres located within the demarcated geographic boundaries of Soweto.For this purpose, personal site visits were undertaken.
The exploratory research enabled the determination of a profile of the Soweto population in terms of certain socioeconomic variables, and the information served as input to the descriptive research phase in order to design the sampling plan.
The descriptive research phase of the project involved a quantitative survey.The target population for the research was all consumers residing in Soweto.A disproportionate stratified sample design was used.The final stratified sample consisted of 690 households as the sampling unit, with respondents spread over 11 subareas of Soweto (Tustin 2008).Sowetan consumers who purchase most in the household qualified as final sample elements/respondents.
The subareas of Soweto and the number of respondents per area are shown in Table 2.This study focuses on and utilises the information obtained in the descriptive research phase, as discussed below.
The descriptive research phase consisted of the design of an appropriate research instrument (questionnaire) and the appointment of interviewers and fieldwork managers who helped with the data-collection process.The questionnaire was a refined version of a previous survey in the area and it was pre-tested during a pilot study done in the area.The fieldworkers employed for this study were experienced interviewers with years of experience in fieldwork within the Soweto area.The fieldworkers' experience of conducting surveys in Soweto was further supported by their understanding of the language, culture and retail business environment in Soweto.These factors contributed considerably to the quality of the fieldwork, which was verified by fieldwork supervisors during the execution phase of this study.
The questionnaire used for the study consisted of questions relating to the following: • Household purchase patterns in Soweto • Household demographics (e.g.area, years of residence, employment and income) • Changes in purchase patterns in Soweto due to development of the retail infrastructure • Shopping expenditure • Shopping experience • Shopping information sources • Financial service needs of Soweto shoppers.
1C ontent validity was established in the research regarding the Soweto township as indicated in the above discussion (Tustin 2008).As no summated construct scale was used in this research, no internal consistency measurement test applied.Descriptive and inferential statistical analysis was used to explore and test the hypotheses.
Correlation analysis was used to test hypotheses H1 (i, ii and iii) and H2 (i, ii and iii).Statistical correlation evaluates the strength of the relationship between two variables.Spearman's Rho rank-order correlation coefficient was used since the profile variables (education, income, length of residence) are ordinal and the ratio of monthly average expenditure values is continuous.The Pearson Chi-square test for independence was used to test the relationship between car ownership (nominal variable) and the ratio of monthly expenditure values grouped as shown in the results below.The rule used when there are two variables that are measured at different levels is to use the method appropriate for the variable that is at the lower level of measurement.

Results
1F or the purposes of this study, a distinction was made between outshoppers and inshoppers, as defined in the theoretical discussion above.
Unfortunately, the data did not include respondents' age profile, since the aim of the initial research was the buying behaviour of households rather than the buying behaviour of individuals.
Education level was recorded in terms of number per type of qualification within the household.The researcher could only isolate the information of households where there was one individual with a post matric diploma/certificate, a bachelors degree or a postgraduate qualification.The ratio of average monthly household expenditure (in-and outshopping) to total average monthly household expenditure was used as a proxy for the propensity for inshopping and outshopping.

Profile of Sowetan in-and outshoppers
1Th e in-and outshopping monthly average expenditure patterns per profile descriptor are shown in Figure 1, while Table 3 shows the category within each group with the highest average expenditure per month.• The 'bachelors degree' group of respondents spends the most in both outshopping and inshopping areas.This finding is contrary to the argument that higher qualifications and advanced degrees will result in more disposable income and hence higher spending patterns.It could be postulated that these higher-qualified residents could be spending more of their disposable income on acquiring financial assets (e.g.buying a house or even making hire-purchase payments for cars and expensive white goods acquired for their households).Respondents who own a car spend substantially more than those who do not own a car in both outshopping and inshopping areas.• Respondents who have been living in Soweto longer than 20 years or less than 10 years tend to spend more within Soweto (inshopping) than outside Soweto (outshopping).• As expected, the higher the income level of the respondent, the higher the spending in both outshopping and inshopping areas.
Investigating the means of transport for both owners and non-owners of cars 1Of the 690 respondents, 70.9% indicated that they do not have their own transport.The question relating to the use of transportation to shop at the retail facilities was asked individually per shopping mall or for other type of outlet in terms of frequency of visits, distance and mode of transport used.The data structure thus did not make it possible to conduct statistical testing for differences between the average monthly expenditure based on different modes of transportation.Descriptive analyses, in terms of average monthly spending per mode of transport and business location, based on combining the relevant information described above, are shown in Table 4.For the purposes of this study, the group classified as inshoppers was further broken down into four sub-categories, namely: • Inshoppers visiting new/existing shopping complexes • Inshoppers visiting industrial areas • Inshoppers frequenting home-based businesses such as spazas • Inshoppers buying from street vendors/hawkers.1F rom Table 4, it can be deduced that own transportation is a preferred means of transport for outshopping and that the average monthly spending by this category of outshoppers is the highest.Taxis are also a major form of transport for outshoppers.
It must be stated that quite a number of Sowetans work outside the Soweto area, and these people do their shopping either during the course of the working day or while waiting for their taxis to arrive, which means this forms part of their daily shopping routine.It is also important to note the amount of expenditure on goods purchased from street vendors/hawkers, which can be classified as convenience shopping.

Correlation analysis
1Th e average monthly expenditure for shopping inside Soweto was calculated by summing the average monthly expenditure at malls within Soweto and at businesses within Soweto but outside the malls.The ratio of total average inshopping expenditure to total average monthly shopping expenditure, as well as the ratio of average monthly outshopping expenditure to total average monthly shopping expenditure, was then calculated.Table 5 shows the results for hypotheses H1 (i, ii and iii).A 5% level of significance was used (p ≤ 0.05).lthough the relationships between the ratio of average monthly inshopping expenditure and education, as well as length of residency, were not statistically significant and the strength of the relationships was weak (small effect), the negative sign indicated that the relationship is an inverse relationship.In other words, lower levels of education and shorter length of residency are weakly correlated with higher levels of average monthly inshopping expenditure.The null hypothesis associated with hypotheses H1 (ii) and H1 (iii) can therefore not be rejected.
The only statistically significant correlation was income level, at the 1% level of significance, with the ratio of average monthly inshopping expenditure (p = 0.000).Since the value was negative (Spearman's Rho = -.175), this indicates that lower levels of income were correlated with higher ratios of average monthly inshopping expenditure.This correlation was expected, given the low levels of income and the convenience factor for these low-income level consumers.
The null hypothesis associated with hypothesis H1 (i) could therefore be rejected, given that the results indicate the existence of a relationship between the ratio of average monthly inshopping expenditure and levels of income.
Because car ownership was measured on a nominal scale (yes/no), it was decided to test hypothesis H1 (iv) (that there is a relationship between those who own cars and those who do not and their ratio of monthly inshopping expenditure (Qiu et al. 2008;Powe & Hart 2009) by using the Pearson Chi-square test (see Table 6).No statistically significant association exists between the ratio of inshopping expenditure groups and car ownership (p = 0.249).
Table 7 shows the results for hypotheses H2 (i, ii and iii).1F urthermore, in the case of outshopping, the relationship between the ratio of average monthly outshopping expenditure, and education and income levels was not statistically significant at the 5% level of significance and the strength of the relationships was weak.However, the relationship between education and the ratio of outshopping was significant at the 10% level of significance (p = 0.090), and the positive sign indicates that the relationship between education and outshopping expenditure is a positive relationship.In other words, higher levels of education are weakly but positively correlated with higher levels of average monthly outshopping expenditure.This finding is in agreement with the literature (see Jarratt [2000] and Ashley-Cotleur et al. [2009]).The case of income levels is surprising, since the results indicate a weak negative relationship between income levels and the ratio of average monthly outshopping expenditure.Although there is no clear indication for the poor negative relationship between income levels and outshopping, it can be surmised that consumers who work in the city are more disposed to buy their products in the city at outlets that are near to their place of work.The null hypothesis associated with hypotheses H2 (i) and H2 (ii) could therefore not be rejected at the 5% level of significance.

N of valid cases 486
1Th e results indicated that a statistically significant association exists between the ratio of outshopping expenditure groups and car ownership (p = 0.011).

Research findings
1R esearch findings in other countries and the evidence from the literature review indicate that, in general, outshoppers are more inclined to have higher educational qualifications, and that outshoppers tend to buy more expensive products and speciality goods during outshopping trips.This finding could not be completely proven in this research as a result of the particular circumstances of the residents of Soweto who work in the city and other places outside Soweto.Retailers inside Soweto should take note of this occurrence and entice these consumers to spend more of their disposable income on inshopping for products and speciality goods.This is especially important for the Soweto shopping malls, which are comparable to those retail outlets outside Soweto where outshopping occurs.
As far as the mode of transport is concerned, car ownership in other countries is usually a precondition for outshopping.In the Soweto study, it was found that the situation is similar, but there are also some differences.While car ownership results in more outshopping opportunities and higher expenditure during such shopping trips (which is comparable with the finding for other countries), this study found that the outshopping category is linked to car ownership but is further supplemented by Sowetans who use taxis to go on outshopping excursions.This is understandable, given that a large number of consumers work in the central business districts (CBD) and in the suburbs around the CBD.They therefore buy from the shops and malls in the areas where they work.Market leakage from the Soweto township therefore occurs independently from car ownership, and depends directly on Soweto residents' places of work.
Research in other countries indicates that inshoppers tend to be older than outshoppers.The Soweto study found the length of residency, which could also be an indicator of the stage of respondents' life cycle, to be inconclusive.The results indicated that residents who have lived in the area for longer (and could therefore be classified as pensioners) tend to spend less at the Soweto malls, which is an accepted economic reality, because their disposable income stream is smaller than that of the younger group of residents.The results indicate that even older persons can be outshoppers in South African townships, thus undermining the belief that older people do the majority of inshopping at the retailers located in the areas in which they live.
Unlike findings in other countries, Sowetan residents with higher income levels do not seem to show a higher propensity for outshopping.It is of interest that these residents (i.e.higher-income residents) now support the new shopping malls located inside the Sowetan townships.There was, however, a higher propensity for lower income customers to buy from local retailers (inshopping).
Due to limitations in the development of the survey instrument, it was not possible to pinpoint any similarities with other studies done on age, types of products bought and store loyalty in other parts of the world.

Conclusions and implications
1R esearch in other countries has indicated that there are clear differences between the retail patronage of inshoppers and outshoppers.Inshoppers are seen as generally older with a lower income, not owning a car and having lived longer in the area in which they do their inshopping.Outshoppers are seen as generally younger, more affluent and better educated, and they usually own a car.Length of residency was not reported to influence their retail patronage behaviour.
Due to South Africa's unique socio-political past, Sowetan outshoppers were forced to pursue outshopping.In short, apartheid led to a lack of retail facilities, poor product assortment and poor service quality inside the townships.Since the establishment of democracy and the building of new retail facilities in the townships, these consumers have turned to inshopping -in other words, they now shop inside the township where they live.The research results show that, while education levels and car ownership are positively related to outshopping, the workplace has a more significant influence on Sowetan residents' buying decisions, ensuring that outshopping and market leakage will continue to be a major factor for the foreseeable future.
In summary, inshoppers in Soweto are described as middle-to lower-income people with lower education levels, who have been living in the township for more than ten years and rely mostly on public transport.It would therefore seem that there are some commonalities between inshoppers and outshoppers in the rest of the world and in South Africa.However, there are also major differences owing to the unique social and political circumstances that have prevailed in South Africa in the past.This suggests a fertile area for further research to understand the influence of these variables and to obtain a holistic view of the inshopping and outshopping debate in South Africa.

1H 2 :
There is a relationship between the propensity for outshopping in Soweto by township residents and (i) Income levels (ii) Education levels (iii) Length of residency in the Soweto township (iv) Car ownership.

Figure 1 :
Figure 1: Average monthly in-and outshopping expenditure per profile descriptor

Table 1 :
Comparison of key identified profile components between inshopping and outshopping customers

Table 2 :
Soweto sample distribution by household location

Table 3 :
Groups with the highest average monthly expenditure

Table 4 :
Average monthly spending per mode of transport and business location mmcccxviiB usiness

Table 5 :
Test results for hypotheses H1 (i, ii and iii)

Table 6 :
Pearson Chi-square test

Table 7 :
Test results for hypotheses H2 (i, ii and iii).