Anda di halaman 1dari 12

ARTICLE IN PRESS

Journal of Retailing and Consumer Services 12 (2005) 273284


www.elsevier.com/locate/jretconser

Supermarkets vs. traditional retail stores: diagnosing the barriers to


supermarkets market share growth in an ethnic minority community
Arieh Goldman, Hayiel Hino
Jerusalem School of Business Administration, Hebrew University of Jerusalem, Jerusalem 91905, Israel

Abstract
We study the state of food retail system serving an ethnic minority community. This group, Israeli Arabs, enjoys a relatively high
standard of living but continues to make many food purchases in a variety of small, specialized retail food formats. In contrast, the
surrounding Jewish population is mostly shopping in supermarkets.
Data from a survey of consumer shopping behavior across formats of different product lines are used to identify the barriers to
the advancement of the supermarket format in this minority sector. Our study shows that socioeconomic factors, found in earlier
supermarket diffusion studies to be the main barrier, have no impact in this case. We identify the tendency to purchase perishable
food items in traditional outlets and the geographical diffusion barrier (distance of supermarket formats) to be the main limitation
on supermarkets market share growth. Further, we nd that both these factors are inuenced by underlying cultural and ethnic
factors characterizing the study population.
r 2004 Elsevier Ltd. All rights reserved.
Keywords: Retail modernization; Supermarkets; Retail formats; Cross format shopping; Marketing to ethnic minority community; The role of
ethniccultural factors; Food shopping patterns; Israel; Israeli Arabs

1. Introduction
The replacement of small family owned, traditional
food stores with supermarkets is a universal phenomenon. This process of retail modernization has been the
subject of a large number of studies. Many have focused
on less developed countries (LDCs) (Goldman, 1981;
Kaynak and Cavusgil, 1982; Slater and Henley, 1969),
others, on emerging economies (Findlay et al., 1990;
Kaynak, 1985; Samiee, 1993). Typically, these studies
describe the weaknesses of the traditional retail system,
and analyze the limitations on the supermarket formats
acceptance.
Traditional food retail systems are not only typical to
LDCs and to emerging economies. They exist also in
Corresponding author. Tel.: +972 2 651 1981; fax: +972 2 654
0152.
E-mail addresses: msgold@mscc.huji.ac.il (A. Goldman),
hayiel@mscc.huji.ac.il (H. Hino).

0969-6989/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jretconser.2004.10.002

developed economies where traditional food retail


formats often operate alongside modern supermarkets.
For example, consumers in the highly developed Asian
economies of Japan, Hong Kong, Taiwan and South
Korea regularly utilize traditional formats and supermarkets market share there has peaked at less than the
50% level (e.g. Goldman et al., 2002).
A number of researchers have recently drawn attention to yet another case of the coexistence of traditional
and modern food retail formats, this time in the highly
developed economies of Western Europe and North
America. In these cases ethniccultural minorities such
as Muslims in the UK (Jamal, 1995, 2003, 2005;
Penaloza and Gilly, 1999) and Chinese and Mexicans
in the USA (Ackerman and Tellis, 2001; Lavin, 1996;
Miller, 1998; Penaloza, 1994) make many of their food
purchases in traditional formats.
This last phenomenon is at the center of this paper.
We report the ndings of a study of the cross format
food shopping patterns of a large ethniccultural

ARTICLE IN PRESS
274

A. Goldman, H. Hino / Journal of Retailing and Consumer Services 12 (2005) 273284

minority: Israeli Arabs. While members of this group


can easily shop in the modern supermarkets serving the
mainstream Israeli population, they continue to patronize traditional outlets. We test hypotheses relating to the
factors that might explain this type of shopping
behavior and thus act as possible barriers to supermarkets market share growth.
Our study is positioned within two research traditions. The rst is food retail modernization which looks
at the barriers to supermarket format diffusion. The
second focuses on the shopping and consumption
patterns characterizing ethniccultural minority groups
residing in developed countries.
Our theoretical framework integrates prior work on
food retail modernization and consumer shopping
patterns in both developed and developing countries.
It is based on the work of Goldman et al. (2002) who
model the market share change process whereby supermarkets gain market share from traditional formats.
The theory identies three general components of
market share change: diffusion of supermarket use
across geography, across socioeconomic segments and
by product categories. Spatial accessibility of modern
formats, consumer ability variables, culturally determined behaviors and perceived characteristics of modern and traditional food retail formats are viewed as
factors limiting or enhancing each of these processes.
Our measurement approach is based on data from
consumer survey which is used to summarize the state of
competition among retail formats. In our particular
application we use a hierarchical series of discrete choice
models covering the various possible explanations. This
enables us to assess the relative impact of these factors
on the retail modernization processes.
The present study highlights the role played by
cultural variables and explains how they impact
consumers retail format choice.

2. Theoretical framework
The process of food retail modernization involves an
increase in the market share of supermarkets. The rich
literature on supermarket diffusion emphasizes the role
played by geographic and socioeconomic factors. To
increase supermarket market share retailers must
penetrate new segments and increase supermarket
accessibility. Studies (Appel, 1972; Findlay et al., 1990;
Goldman, 1981; Kaynak and Cavusgil, 1982; Kumcu
and Kumcu, 1987) show higher socioeconomic status
consumers to be more likely to switch to modern
supermarkets. The reason: a higher opportunity cost of
time is making multi-stop shopping in many small stores
more costly than the one-stop shopping associated with
the supermarket (Betancourt and Gautschi, 1986, 1990).
In addition, these households have access to better

transport and can store larger food quantities, making


one-stop shopping even more cost effective.
When a spatial separation of socioeconomic groups
exists, i.e. geographic and economic segments coincide,
we typically see the coincidence of diffusion through
economic and geographic segments: supermarkets rst
open in higher-income areas and later in lower-income
ones. This pattern was observed in many developing
economies (Slater and Henley, 1969).
Where no spatial separation of economic groups
exists diffusion by income may lag geographic diffusion.
In this case we nd supermarkets becoming widely
accessible even to lower-income consumers, but their use
is unequal. This scenario too has been widely observed.
Specically, studies in developing economies reported
cases where in spite of easy accessibility to supermarkets
consumers prefer to continue and purchase their food in
traditional formats. Many researchers attributed supermarkets failure in these cases (e.g. Goldman, 1974,
1981; Kaynak, 1985) to economic factors: higher-income
consumers getting higher benets from switching from
traditional stores to supermarkets. The underlying
reason is their higher opportunity costs of time, more
storage space and transportation possibilities enabling
less frequent shopping for food. In contrast, lowerincome consumers, who purchase small amounts and
shop frequently, get lower benets from shopping in
supermarkets.
Some researchers, however, also view the problem as
cultural. For example, lower-income consumers in the
USA residing in inner city or ghetto areas have
developed a subculture of poverty (e.g. Andreasen,
1972; Caplovitz, 1967). They put high value on
attributes associated with small traditional outlets such
as personal attention by store owners, social interactions
during shopping and being part of an informal economy
centered around these stores.
The product category dependent diffusion, i.e. the use
of the supermarket only for selected product categories,
is another barrier to supermarket market share growth.
This selective adoption phenomenon has been documented in both developing economies (Goldman, 1982;
Kaynak, 1985; Miossec, 1990; Yavas et al., 1981; Zain
and Rejab, 1989; Othman, 1990) and in developed ones
(Goldman et al., 2002). Many of the documented cases
of selective adoption involve the perishables category.
Consumers were found to systematically divide their
food purchases: shop regularly for perishable items in
traditional formats while purchasing other food lines in
supermarkets.
One explanation for the existence of this behavior
involves supply side factors. These relate to the underdeveloped and fragmented supply system for the perishable food categories and/or to supermarkets space
limitations. These may result in supermarkets perishable offering being inferior in variety, quality and price

ARTICLE IN PRESS
A. Goldman, H. Hino / Journal of Retailing and Consumer Services 12 (2005) 273284

to that of traditional outlets (Goldman et al., 1999). A


second explanation relates to behavioral patterns
reecting cultural values. These often characterize the
behavior of ethniccultural minority subgroups. (Ackerman and Tellis, 2001; Herche and Balasubramanian,
1994; Penaloza, 1994; Oswald, 1999; Jamal and Chapman, 2000; Jamal, 2003).
These values determine attitudes towards food, the
meaning and importance attached to certain food
products and categories, the nature of a proper meal,
the type of food products to be served in various meal
occasions, appropriate meal preparation modes, the
types and sources of required ingredients and the
denition of their quality and freshness (e.g. meats
from animals killed according to strict rituals or freshly
killed (warm, not chilled or frozen) meat or sh)
(Jamal, 1998, 2005; Mennel et al., 1992; Murcott, 1983;
Wallendorf and Reilly, 1983; Hirschman, 1981, 1983;
McCracken, 1986). These, in turn, lead these consumers
to prefer shopping in traditional outlets that are viewed
as better able to handle these requirements than the
impersonal supermarkets (Goldman et al., 1999).
Finally, another barrier to supermarkets market
share growth is adherence to traditional values associated with the nature of shopping activities. Consumers
may attach less importance to outputs typically associated with modern supermarkets (e.g. cleanliness,
variety, self-service, order and atmosphere) and place
high value on outputs in which traditional formats excel
such as personal relationships and being served by
members of the same ethnic community.
Studies of consumers food store choice (e.g. Arnold
et al., 1983) found that location, assortment price,
checkout time and shopping environment determine
store choice. These variables relate to our discussion as
follows: location affects geographic segment diffusion;
assortment decisions affect product category diffusion,
and price, service and shopping environment variables
affect economic segment diffusion.

3. Barriers to supermarkets market share growth


On the basis of the discussion above, we identify a
number of scenarios that may describe the state of a
food retail system where supermarkets encounter factors
hindering their attempts to bring members of relevant
groups (e.g. ethnic minorities) to switch from traditional
food outlets to them.
3.1. Incomplete geographical diffusion
This scenario addresses the spatial accessibility of
food outlets and its impact on consumers format choice
decision. Low accessibility of supermarkets, reected in

275

relatively long travel time, may restrict their adoption.


In these cases, we expect to observe:
H1. The greater the relative travel distance to supermarkets (travel time difference between travel to
traditional formats and to supermarkets), the greater is
the probability of shopping at traditional food outlets.
3.2. Socioeconomic factors
A variety of socioeconomic variables were shown to
impact consumers ability to shop in supermarkets.
Specically studies (e.g. Betancourt and Gautschi, 1986;
Messinger and Narasimhan, 1997; Goldman, 1981)
suggested that a households opportunity costs and
storage costs inuence consumers tendency to shop in
supermarkets.
H2. The ability to purchase larger food quantities is
positively related to the probability of shopping at
supermarkets.
3.3. The perishable food category
This scenario attributes supermarkets inability to
increase their market share to consumers tendency to
divide their food purchases between traditional and
modern formats, purchasing perishable food items in
traditional formats and non-perishable items in supermarkets (Goldman et al., 2002). This behavior may
reect cultural factors associated with the particular
ethnic minority group studied.
H3. The probability of shopping for a given food
product at supermarkets is greater if the product is
non-perishable.
3.4. Store outputs
Supermarkets are generally viewed as providing
outputs superior to those of traditional formats (Goldman, 1981). However, some consumers may view
supermarkets as providing similar or inferior outputs
than traditional stores (e.g. higher prices). These views
may reect cultural factors associated with the ethnic
minority studied or real difculties.
H4. Consumers who view traditional retail formats as
providing the same outputs as supermarkets or superior
ones are less likely to shop at supermarkets.
The quality and freshness of perishable food items
was found to be of special importance to ethnic
communities (Ackerman and Tellis, 2001; Goldman et
al., 1999; Jamal, 2003; Lavin, 1996). To the extent
differences in freshness levels exist between supermarkets and traditional formats they may be regarded
as another output barrier.

ARTICLE IN PRESS
276

A. Goldman, H. Hino / Journal of Retailing and Consumer Services 12 (2005) 273284

H5. Consumers who view traditional food outlets as


providing higher levels of freshness and quality in
perishable food items will be less likely to shop for
these in supermarkets.
3.5. Food preparation
Consumers who emphasize meals prepared at home,
take a longer time to prepare them and put a high value
on the use of fresh ingredients in these meals are less
likely to buy perishable food items in supermarkets. The
reason is the superior ability of traditional retailers to
cater to specialized needs reecting cultural, religious or
ethnic habits (Goldman et al., 1999; Jamal, 2003).
H6. Consumers who prepare more meals at home and
emphasize the use of fresh items in these meals are less
likely to shop for perishables at supermarkets.

4. Application: the Israeli Arabs


The Arab community in Israel is the setting to this
study. Israel is a developed country with an advanced
market economy and modern infrastructures (World
Bank, 1999/2000).
The Arab population comprises 20% (1.2 million) of
Israels citizens. While this group still lags the Jewish
population, its socioeconomic status is high, both in
absolute terms and in comparison to most Arab
countries.1 Specically, Israeli Arabs GNP per capita
is 16,000 US $ and the yearly per capita income is
4200 US $. Many (56%) Arab households own at least
one car, 97% own their residence, 95% a color TV,
99.5% a refrigerator, 44% at least one cell phone and
24% a personal computer (Israel Central Bureau of
Statistics, 1999; Tables: 13 and 6). However, family size
is large leading to a high housing density level (1.3
persons per one room).
The Israeli Arabs constitute an ethniccultural
minority living in a largely Jewish country. They
concentrate in few regions and in Arab cities, towns
and villages. Furthermore, even Arabs living in Jewish
areas often reside in their own neighborhoods. Also,
most Arabs maintain their culture; speak Arabic, read
Arabic language books and newspapers and view Arab
TV stations (Rinnawi, 2003).
Family relationships are very strong and of central
importance. The extended family serves as a foundation
of social, economic and political activities and continues
to be a key factor in their lives. The importance of the
extended family is enhanced by residential proximity.
1
Israels Arab citizens are often classied as similar in their
socioeconomic status to the Arabs residing in the most developed
Arab economies of Kuwait, Qatar and United Arab Emirates (The
World Bank, 1999/2000).

Many families reside near their relatives, often visit one


another, socialize, and frequently meet for family meals
and social activities (Rinnawi, 2003).
In addition, the low level of labor force participation
of Arab women (22%) enables them to allocate much
more time than Jewish ones to family activities including
shopping and home meal preparation (Jaffa Institute,
1997).
Almost 3000 small, family owned, grocery food stores
operate in the Arab areas. These stores carry mostly
non-perishable groceries and also chilled dairy items.
Fresh and perishable food lines are sold exclusively in
specialized traditional food formats. These included 400
meat, poultry and sh stores and around 400 fruit and
vegetables stores and stalls. These stores locate either in
the residential neighborhoods or in open markets
(concentrations of stores and stalls) that exist in most
Arab cities and towns (Israel Customs Authority, 1997).
One hundred Arab owned supermarkets serve the
Arab areas, most are small or medium size. They carry a
full assortment of packaged and processed food lines
and of dairy products but a limited selection of
perishable and frozen food lines. In addition, a number
of large supermarkets (many owned by Arabs and some
by Israeli chains) are located in or near the Arab areas.
These supermarkets carry a full assortment of fresh and
perishable food. Many of the Arab owned supermarkets
also serve Jewish consumers mostly because they open
on Saturdays when Jewish owned stores are closed.
The suppliers and supply arrangements used by the
Arabs supermarkets are identical to those used by Israeli
supermarkets. Also, they carry the same brands.

5. Research design and methodology


We analyzed the state of modernization of the food
retailing system serving the Israeli Arab population
through a survey of shopping patterns. Information was
collected about the store type (format) in which Arab
consumers shopped for each of the 14 product lines
studied. These included packaged food, packaged nonfood and perishable items (meat and poultry, sh, fruits
and vegetables, fresh bakeries). We selected these
product lines on the basis of family expenditure studies
conducted by the Central Bureau of Statistics (Israel
Central Bureau of Statistics, 1999).
The food outlets used by the interviewees included
Arab owned supermarkets, mini-supermarkets, neighborhood grocery stores, vegetable and fruit stores,
market stalls, sh stores, butcher shops and Israeli
owned chain supermarkets. We classied these into two
groups: modern (supermarkets) and traditional (all
other formats).
In addition, we asked about travel time to the
stores, frequency of shopping, food preparation habits,

ARTICLE IN PRESS
A. Goldman, H. Hino / Journal of Retailing and Consumer Services 12 (2005) 273284

perceptions of different store formats and collected


socioeconomic and demographic data. Also, we conducted discussions with the respondents asking them to
explain the reasons for their shopping patterns and
discuss meal preparation activities.
A stratied sample of 511 households was selected for
the study from the 11 geographical areas where Israeli
Arabs reside. The number of households selected in each
area was proportional to its relative size. Also, the
distribution of the sample among Arab cities, towns,
villages and large Jewish cities reected the distribution
of the Arab population.
Trained interviewers conducted face to face interviews
in Arabic with the person in each household responsible
for most food shopping. In 76% of the cases these were
woman. Given the cross cultural nature of the study we
followed the recommended procedure for such studies
(e.g. Craig and Douglas, 2000; Hui and Triandis, 1985).
Specically, the questions were rst written in Hebrew,
translated into Arabic, then into Hebrew again by a
professional translator and by one of the authors whose
native language is Arabic.
The questionnaire was pre-tested on 15 consumers
leading to modications and renements. The actual
interviews took between 20 and 40 min to complete,
depending on the number of stores shopped and the
amount of probing required.
Subsequent analysis indicated that the samples
socioeconomic and demographic prole closely resembled that of the Arab population.

277

to the nearest specialized traditional outlet (10.6 to the


nearest butcher, 10.8 to the nearest fruit and vegetables
store and 7.4 to the nearest open market).

7. Hypotheses tests
Following Goldman et al. (2002), we specify format
choice as a binomial logit model across all product
categories. Format choice for each product for each
household was treated as an individual observation, Yi
taking the value 1 if a household purchased a product
mainly at a supermarket, and 0 if mainly at a traditional
format. We estimated the contribution of each of the
explanatory variables discussed above to the overall
goodness of t of the binomial logit model. Thus, y is
modeled directly as the dependent variable in a model
incorporating the logistic function specication of the
form:
PYi 1

1
1 exphbxii

where xi contains the explanatory variables that are


expected to inuence each choice and b is a set of
parameters to be estimated by Maximum Likelihood
Estimation. We estimated the effects of the explanatory
variables in a nested fashion. First, we included
geographical diffusion variables, then socioeconomic
ones, then perishability, then format perception variables and nally food preparation ones.

6. Descriptive statistics
8. Measures
Respondents average age was 42.3 years. A household had an average 5.47 persons with an average of 1.4
fully employed members (65% one member employed,
25% two and 10% three or more). A households
average monthly income was 1195$ and 44% of the
respondents had high-school education and 8% a
college one. The average number of cars owned by a
household was 0.69 (37% had no car, 58% owned 1 car
and 5% owned 2 cars). In addition, over 10% of the
sample enjoyed the use of an employers car. Most
interviewees owned their house/apartment. The average
home size was 3.84 rooms with an average density of
1.53 persons per room.
The average number of weekly shopping trips by a
household to all food stores was 12.2. Of these 6.3 were
to neighborhood groceries, 0.97 to supermarkets, 1.1 to
fruit and vegetables stores, 0.7 to butchers, 1.8 to sh
stores and 1.4 to open markets (mostly for fruit and
vegetables and for fresh bakery products).
The average distance to the nearest supermarket was
14.7 min (21.4 min to the nearest Jewish chain supermarket and 11.7 min to an Arab owned one) and 7.6 min

Travel time was used to measure the accessibility of


food stores. High accessibility of supermarkets is dened
as travel time that is less or equal to travel time to the
traditional stores. We measured the distance effect as the
difference between the time to the nearest traditional
outlet used for the purchase of the product category and
the time to the nearest supermarket. The contribution of
this variable to the overall model t captures the effect
of location and accessibility on supermarkets market
share growth (Goldman et al., 2002).
The evaluation of the effect of socioeconomic factors
on market share requires measures of consumers ability
to purchase in supermarkets and to transport and store
large quantities of groceries at home. We included four
socioeconomic variables: monthly income per person
(income divided by family size) which measures consumers opportunity costs of time; car ownership and/or
the use of an employers car which measures mobility;
living density (persons per room) which measures
storage space; and education, a general consumer ability
measure (Hoch et al., 1995; Goldman et al., 2002).

ARTICLE IN PRESS
278

A. Goldman, H. Hino / Journal of Retailing and Consumer Services 12 (2005) 273284

Perishability was measured as a dummy variable


taking a value of 1 for perishable products and 0 for
non-perishable ones.
To measure the benets to be derived from shopping
in supermarkets, we used consumers perceptions of the
relative outputs of supermarkets and traditional formats. We averaged, for each respondent, the difference
in perceptions between supermarkets and traditional
outlets on three attributes shown in earlier surveys
conducted among Arabs in Israel as the most important
store choice characteristics: assortment variety, cleanliness and product quality.2 The actual measure used was
the average of the three differences (Cronbach alpha was
0.7). We also measured consumers assessment of the
freshness of the perishable assortment carried by supermarkets and traditional formats averaging, for each
respondent, the difference in perceptions of the two
formats.
Finally, we used three variables to assess the impact of
meal preparation habits: the number of meals a household prepares at home; the average time spend on
cooking the central daily meal; and the emphasis on the
use of fresh, rather than frozen, chilled or processed
ingredients in preparing home cooked meals. As the rst
two variables were strongly correlated, we use only the
rst of these two.

r2 increased to 8% and the hit rate from 63.6% in the


base model to 68.1%. Accordingly, hypothesis 1 is
accepted. The longer travel time to supermarkets (an
average of 14.7 min, 11.7 to the smaller Arab supermarkets and 21.4 to larger chain supermarkets, vs. an
average of 7.6 min to traditional formats) clearly
restricts their adoption.
There were major differences in relative travel
distance between supermarkets and the traditional
outlets specializing in perishable and between supermarkets and the grocery stores.3 Consequently, we
conducted a further analysis of the data. It indicated
that the impact of the distance factor is very strong in
the case of non-perishable items but weak in perishable
ones. To evaluate these differences quantitatively we ran
the same logit analysis but this time only for nonperishable products (Table 2). We found a major
increase (model 1 vs. 2) in both r2 (12% at present vs.
8% previously) and in the predictive hit rate (almost
17% compared to 4.5% previously).
These results mean that the opening of more supermarkets closer to the Arab population is expected to
lead to a major increase in the purchase there of nonperishable items but to a much more minor change in
the purchase patterns of perishable food lines.
9.2. Socioeconomic factors

9. Findings
We estimated the effect of the different variables in a
nested fashion. We rst included travel time, then
socioeconomic variables, next the perishability variable,
then the perceptions of store attributes and nally added
the meal preparation variables. The hypotheses were
evaluated in the general model. Also, the sign and
signicance level of the coefcients in all six types of
variables and the differences in goodness of t, as each
variable set is incrementally added, indicate the relative
contribution of each to the overall explanatory power of
the model. We also included a constant to capture the
residual preference for supermarkets after controlling
for all six variable groups. Estimates for the six models
and for the base model appear in Table 1.

All four socioeconomic variables (Table 1, model 2)


were expected to positively inuence consumers ability
to shop in supermarkets. The coefcient signs of all four
are in the expected direction and signicant. However,
in three cases the signicance level is low. The
incremental contribution to the overall explanatory
power of the model was evaluated by comparing model
3 to model 2. It was quite small (r2 increased to 12%
and predictive hit rate was marginally higher than in
model 2). The same results were also obtained when we
ran the analysis only for non-perishable items (Table 2).
Thus H2 is rejected and we conclude that socioeconomic
factors do not constitute a barrier to supermarkets
market share growth. This is true for both perishable
and non-perishable food lines.
9.3. The perishable category

9.1. Geographical diffusion


The travel time difference coefcient is signicant and
negative. Also, the marginal contribution of this
variable to the overall goodness of t is relatively high:

Model 4 tests the hypothesis relating to the expected


negative impact of the perishable food category on the
likelihood of buying in supermarkets. Our data supports
H3. The coefcient for perishability is strongly negative

2
We also collected information about perceptions of price levels. We
found that while supermarkets were perceived as somewhat less
expensive than traditional formats, these differences were not
signicant. Consequently, this variable was not included in the
analysis.

3
The average distance difference between neighborhood grocery
stores (non-perishables) and supermarkets is 10 min vs. an average
distance difference of only 4 min between supermarkets and the
perishable stores (vegetable, butcher, sh stores) and 7 min to open
markets.

Table 1
Maximum likelihood estimates of logit models of format choice (all product lines)a
Independent variables

Expected
coefcient sign

Distance variable
Travel time difference

Model 3

Model 4

Model 5

Model 6

Model 7

Base model

Model 1+travel
time distance

Model 2+consumer
socioeconomic
characteristics

Model
3+perishability

Model 4+store
attributes

Model
5+freshness

Model 6+food
preparation

0.56
(o.0001)b

0.17
(o.0001)b
0.07
(o.0001)

0.55
(0.0002)b
0.07
(o.0001)b

0.04
(0.82)
0.08
(o.0.0001)b

1.21
(o.0001)b
0.06
(o.0001)b

0.88
(o.0001)b
0.05
(o.0001)b

4.54
(o.0001)b
0.04
(o.0001)b

0.07
(0.09)c
0.35
(o.0001)b
0.07
(0.13)c
0.27
(0.0001)b

0.06
(0.17)
0.4
(o.0001)b
0.11
(0.03)d
0.36
(0.0001)b
3.33
(o.0001)b

0.01
(0.83)
0.3
(0.0002)b
0.13
(0.014)d
0.27
(0.0001)b
3.57
(o.0001)b

0.1
(0.08)b
0.3
(0.0005)b
0.15
(0.0009)d
0.26
(o.0001)d
4.12
(o.0001)b

0.05
(0.35)
0.17
(0.07)c
0.07
(0.3)
0.19
(0.007)b
4.33
(o.0001)b

1.56
(o.0001)b

1.18
(o.0001)b
1.1
(o.0001)b

Socioeconomic
characteristics
Income per person
Mobility

+
+

Living density

Education

Perishability

Store attributese

Assortment freshnessf

Food preparation
Meals prepared at
home

0.7
(o.0001)b
0.19
(0.006)b
1.15
(o.0001)b

0.17
(0.001)b

Fresh ingredients in
cooking
2 log-likelihood
r2 (AIC adjusted)g
Hit rate (% correct
predictions)

7167
63.6

6578
0.08
68.1

6287
0.12
68.2

5111
0.29
77.2

4644
0.35
79.3

3985
0.44
82.5

3615
0.494
85

(pr4chi-square in parenthesis).
po0.01.
c
po0.10.
d
po0.05
e
Store attribute variable is an average of consumers perceptions of the difference in three store characteristics (quality of products, assortment variety, in store cleanliness), between modern formats
and traditional groceries.
f
Assortment freshness calculated as the average difference in consumers perceptions between modern formats and traditional perishable food formats.
g
Calculated relative to the base model: r2=1 [(LL (model) number of additional parameters)/LL (base model)].
b

ARTICLE IN PRESS

Model 2

A. Goldman, H. Hino / Journal of Retailing and Consumer Services 12 (2005) 273284

Constant

Model 1

279

ARTICLE IN PRESS
280

A. Goldman, H. Hino / Journal of Retailing and Consumer Services 12 (2005) 273284

Table 2
Maximum likelihood estimates of logit models of format choice: (only non-perishable food lines)a
Independent variables

Expected
coefcient sign

Constant

Distance variable
Travel time difference

Socioeconomic characteristics
Income per person
Mobility

+
+

Living density

Education

Store attributese

2 log-likelihood
r2 (AIC adjusted)f
Hit rate (% correct
predictions)

Model 1

Model 2

Model 3

Model 4

Base model

Model 1+travel time


distance

Model 2+consumer
socioeconomic
characteristics

Model 3+store
attributes

0.14
(o.0001)b

0.85
(o.0001)b
0.09
(o.0001)b

0.02
(o.0001)b
0.09
(o.0001)b

1.35
(o.0001)b
0.07
(o.0001)b

0.05
(0.32)
0.41
(o.0001)b
0.09
(0.068)c
0.38
(0.0001)b

0.04
(0.43)
0.30
(0.0002)b
0.13
(0.013)d
0.3
(o.0001)b

4769
0.16
70.4

1.7
(o.0001)b
4269
0.25
74.1

5712
53.6

5025
0.12
70.2

(pr4chi-square in parenthesis).
po0.01.
c
po0.10.
d
po0.05.
e
Store attribute variable is an average of consumers perceptions of the difference in three store characteristics (quality of products, assortment
variety, in store cleanliness), between modern formats and traditional groceries.
f
Calculated relative to the base model: r2 1 [(LL (model) number of additional parameters)/LL (base model)].
b

and signicant showing that perishable food items are


more likely to be purchased in traditional outlets. The
comparison of the goodness of t measures (both r2 and
hit-rates) between models 3 and 4 shows that r2
increased from 12% to 29% and predictive hit rates
from 68.2% to 77.2%. We conclude that Arab
consumers tendency to continue and purchase perishable food in traditional outlets is a major limiting factor
on supermarkets ability to increase their market share.
9.4. Retail outputs
Model 5 evaluates the impact of consumers perceptions of the level of retail outputs (variety, cleanliness
and quality) provided by supermarkets and traditional
formats. The summary variable was signicant (at the
0.01 level) with the expected positive sign of the
coefcient. This nding supports H4 indicating that
consumers perceive supermarkets to provide superior
benets and that these perceptions contribute to an
increase in their market share. The overall contribution
of this variable to the overall explanation is quite high

(see r2 and hit rate improvements). While its impact is


stronger in the case of non-perishable food it has a
strong impact also in the case of perishable items. This
means that further improvements in these factors are
likely to yield meaningful market share increase for
supermarkets.
The addition of the variable measuring the perception
of freshness (model 6) contributes to a further improvement in the overall t of the model. The coefcient is
signicant with the expected negative sign. This nding
supports H5 indicating that the perception of traditional
stores superiority in the freshness of their perishable
assortment serves as an additional barrier to supermarkets market share increase. Consumers rated the
freshness level of the products carried by the traditional
formats as 4.6 (SD 0.47), vs. 3.1 (SD 0.79) for the
supermarket (scale 15).
9.5. Food preparation factors
Model 7 (Table 1) captures the added impact of the
underlying cultural and ethnic factors inuencing food

ARTICLE IN PRESS
A. Goldman, H. Hino / Journal of Retailing and Consumer Services 12 (2005) 273284

preparation and consumption habits. Hypothesis 6 is


supported by the negative and signicant coefcients of
the two food preparation variables. The incremental
contribution of this factor to the explanation (model 7
vs. model 6) is, however, not high. It is important to
point out that all respondents prepared a very large
number of meals at home (an average of 6.64 meals per
week, SD 1.16) and spent a large amount of time in
preparing them (an average of 2.6 h on each).

10. Discussion of ndings


The overall goodness of t of the nal model is high.
When we contrast goodness of t measures of models 7
and 1 (base model), we nd that r2 increased to almost
50% and the predictive hit rate to 85%. Coefcients for
all variables (except income per person in models 57)4
were in the predicted direction. This means that our
model was able to capture most of the factors impacting
supermarket diffusion in this population.
We interpret the signicant and positive constant
(intercept) of 4.54 in the overall model (model 7) as
indicating that after controlling for all the variables
included in the model there is a general underlying
preference among the study population for the supermarket. This conclusion is supported by the perception
of the supermarket viewed as providing superior outputs
compared to traditional outlets.5
The travel distance and the tendency to purchase
perishable products in traditional formats were found to
slow down supermarkets diffusion. In contrast, the
socioeconomic variables had no impact. This reects the
overall high socioeconomic level of the Israeli Arab
population. In other words, Israeli Arab consumers who
wanted to shop in supermarkets were not restricted by
the ability variables captured by the socioeconomic
measures.
Our study found the preference for purchasing
perishable food items in traditional outlets to be strong
and entrenched. Arab consumers purchased 95% of
4
The negative coefcient in these cases is explained by the manner in
which the income variable was constructed leading to the observed
distortion. we divided household income by the number of persons in
the household. Checks revealed that households with higher income
under-reported income. Because of the variables construction this
effect is most severe in larger households. On the other hand,
household size is positively correlated with the variables covered in
models 57.
5
Consumers ratings of modern and traditional outlets (scale of 15),
respectively, were as follows (standard deviation in parenthesis):
quality of assortment 4.8 (0.36) and 4.2 (0.8); assortment variety 4.8
(0.3) and 3.9 (0.97); store cleanliness 4.76 (0.4) and 4.1 (0.9). Only in
regard to the perishable food lines consumers perceived supermarkets
assortment freshness to be inferior: 3.5 (0.79) vs. 4.6 (0.47),
respectively.

281

these items in specialized traditional outlets in spite of


the fact that the travel distance difference between these
and Arab supermarkets, (which carried these perishable
lines), was small (an average of 10.8 min vs. 11.7,
respectively). These preferences reect the central role
these items play in the daily life of the Arab consumers.
Arab families tend to cook elaborate family meals at
home almost daily and to use fresh ingredients in
preparing these meals. These meals are a key feature of
the daily routine in most Arab homes and homemakers
invest a large amount of time and effort in preparing
them. Arab consumers believe that only fresh and
natural ingredients should be used in these meals and
look suspiciously at processed, frozen, chilled and
precooked foods. Also, they prefer to buy fresh red
meat and sh from a known and trusted source. Meals
prepared from natural and fresh ingredients are viewed
as healthier and less risky and as the hallmark of a good
homemaker who cares for the well-being of her children
and family (IMA, 1999, 2003).
The geographic barrier to market share growth,
travel time distance difference between traditional
formats and supermarkets, was found to play a much
stronger role in the non-perishable food categories.
The traditional grocery stores were, on average, only
4.7 min away from consumers vs. an average distance of
14.7 min for supermarkets (11.7 for Arab owned
and 21.4 for larger chain supermarkets). The high
frequency of the shopping trips to the neighborhood
groceries (an average of 6.3 trips per week) further
strengthens the distance advantages of the traditional
groceries.
Further analysis of the data show that Arab
consumers tended to split their non-perishable purchases between neighborhood groceries (70%) and
supermarkets (30%). Discussions with the respondents
revealed that in the neighborhood grocery stores they
purchased mostly milk, dairy products, soft drinks,
other daily necessities and snacks. It emerged that
cultural norms dictate that shopping trips made by
woman unaccompanied by their husbands should be
made close to home and involve buying only from
known and trusted retailers. Consequently, purchases
made by unaccompanied women or by children were
mostly in the neighborhood groceries. In contrast, the
husbands participated in almost all purchases made in
supermarkets.
In addition, we learned from these discussions that
respondents distinguished between two types of shopping trips. They viewed shopping in the neighborhood
grocery stores in strict functional terms. In contrast,
shopping in supermarkets, especially in large chain ones,
was viewed more as a family event containing elements
of a family outing, entertainment and an opportunity to
observe new things. Consequently, the whole family
often participated in these shopping trips.

ARTICLE IN PRESS
282

A. Goldman, H. Hino / Journal of Retailing and Consumer Services 12 (2005) 273284

11. Conclusions
A number of earlier studies documented the tendency
of consumers belonging to ethniccultural minority
groups to display food shopping patterns that differ
from those of mainstream consumers (e.g. Lavin, 1996;
Herche and Balasubramanian, 1994; Kaufman and
Hernandez, 1991; Jamal and Chapman, 2000; Jamal,
2003). They shop frequently, or exclusively, in traditional food retail stores rather than in the mainstream
supermarkets. The researchers conducting these studies
were mostly interested in acculturation, cultural diversity, the role of marketers and implications for marketers and retailers (e.g. Ackerman and Tellis, 2001; Jamal,
2003, 2005; Lavin, 1996; Wallendorf and Reilly, 1983).
Our interest in this paper is different. We focus on the
structure of the food retail system serving these
consumers and on its modernization: the move from a
system characterized by small, traditional, retail formats
to one dominated by supermarkets.
We employ here the conceptual framework and
management approach used by Goldman et al. (2002)
in their study of retail modernization in Hong Kong.
While earlier modernization studies emphasized the role
of socioeconomic variables, Goldman et al. (2002) found
that these factors had no effect. They identied
consumers tendency to make their perishables food
purchases in traditional retail formats as the main
barrier to supermarkets market share advance. Our
study nds similar results. However, we advance their
work by explicitly studying the role of cultural factors
which to underlie these shopping patterns.
We nd that the impact of these factors manifests
itself through two shopping behavior patterns: the rst
is the purchase of perishable food items in small,
specialized stores and the second is the tendency of
Israeli Arab woman to conne some non-perishable
shopping trips to neighborhood grocery stores. Underlying the perishables shopping pattern is the daily
practice of preparing a cooked, traditional, family meal
at home. Preparation of these meals often involves the
use of fresh, traditionally dictated, ingredients. Since
consumers view traditional stores as superior to supermarkets in the quality and variety of their perishable
assortment and in the freshness of perishable items they
prefer buying these items there rather than in the
supermarkets. Cultural factors also impact supermarkets market share through the shopping trip. Cultural
norms dictate that women should not venture unaccompanied out of the safe radius around the home.
As a result unaccompanied women buy only in the
neighborhood stores, not in the more distant supermarkets. Supermarket trips often involve husbands
accompanied by their wives. These practices have led
Israeli Arab households to split their non-perishable
food purchases between supermarkets and neighbor-

hood stores. This scenario partly explains the effect of


travel distance identied in the study as a barrier to
supermarket market share growth.

12. Implications
12.1. Implications for supermarket managers
Both Jewish and Arab supermarket executives interested in increasing market share in the Arab sector
believe that the key to supermarkets market share
growth is further improvements in Arab consumers
socioeconomic status. The present study indicates that
this is not the case. Changes in Arab consumers
socioeconomic status will have no impact on supermarkets position. Locating supermarkets nearer to
consumers and further improvements in assortment,
cleanliness and product quality will have a positive
impact. However, it will mostly be conned to the nonperishable food categories.
The study highlights the role of perishables. Underlying the tendency to buy these products in traditional
outlets are culturally determined food preparation and
consumption patterns. Also, the study nds that an
additional important barrier to supermarkets market
share growth is the culturally determined tendency of
woman to buy non-perishables items in the small
neighborhood grocery stores.
These ndings pose a major strategic challenge for
supermarket executives. They need to develop new
supermarket formats that will cater better than the
present supermarket formats to the particular needs of
these consumers. The design of these formats should be
based on the identication of the relevant attributes of
the specialized traditional formats that make them so
attractive to Arab consumers. The new formats will
need to incorporate these attributes combining them
with those elements that make the supermarkets an
effective retail mechanism. Also, supermarket executives
should rethink their strategy of a focus on large-scale
supermarket formats. An opportunity exists in the Arab
sector for a small scale, neighborhood supermarket
format that will be able to effectively compete with the
traditional neighborhood grocery stores.
12.2. Implications for researchers
Researchers studying the shopping behavior of ethnic
minority consumers residing in developed economies
realized before that these consumers tend to purchase in
both traditional and modern food retail formats.
However, this behavior was not studied before in a
systematic fashion. In this study, we demonstrate the
usefulness of a procedure that can be effectively used to
analyze this type of cross-format shopping behavior.

ARTICLE IN PRESS
A. Goldman, H. Hino / Journal of Retailing and Consumer Services 12 (2005) 273284

Data generated from a survey of shopping behavior


across formats are used to evaluate the barriers to
supermarkets market share growth. Specically, the
procedure involves the collection of detailed information as to which product line is purchased in which
format. These data provide researchers, public policy
makers and retail managers with rich, diagnostic
information about the distribution of food purchases
across formats and about the factors that limit supermarkets ability to gain market share.

283

information generated from the respondents through


open-ended questions and through observations. Also,
we did not investigate the impact of ethnicity and
culture directly nor did we identify its components. In
the case of Israeli Arabs, these components may include
an Arab identity, a religious component (Islamic,
Christian, Druze) and a city vs. village cultural
component.

Acknowledgments
13. Future research and limitations
13.1. Future research
Researchers should continue to explore the impact of
cultural and ethnic factors on the structure of the food
retail system serving ethnic consumer groups. These
factors inuence retail structure through their effect on
shopping behavior which, in turn, is determined by
variables related to food preparation and consumption
habits, and to the nature of family relationships.
Consequently, researchers should make these variables
an integral part of their research agenda.
An important issue that needs to be studied is the
degree of persistence of these behaviors over time and
the forces working for and against their continued
existence.
The possible existence of subgroups among these
ethnic minorities should also be studied. For example, in
our particular case the Israeli Arab population consists
of Muslim, Christian and Druze Arabs. It is not clear if
the cultural factors we discuss impact all these groups in
the same manner and what are the differences.
The process of retail modernization takes place in a
large number of Arab countries and in countries where
large Arab populations exist. The question that needs to
be studied is whether the shopping patterns we identied in this study and the underlying cultural
factors inuencing them are shared across other Arab
populations.
13.2. Limitations
Many of the early studies of ethnic minority groups
collected observations and employed qualitative, ethnographic, descriptive approaches to interpret the data. In
contrast, the present study uses survey data and
quantitative methods to test a series of hypotheses. We
feel, however, that given the complexity of the phenomenon studied and the state of knowledge in this area, a
purely quantitative research approach is limited in its
effectiveness. Further studies should use a combination
of data collection and analysis approaches. Survey
methods should be supplemented with qualitative

We acknowledge support of the Kmart Center for


International Retailing and of the Davidson Center for
Agribusiness, Hebrew University. The second author
acknowledges the support of a grant from The Israel
Ministry of Science, Culture and Sport.

References
Ackerman, D., Tellis, G., 2001. Can culture affect prices? A crosscultural study of shopping and retail prices. Journal of Retailing 77
(1), 5782.
Andreasen, A.R. (Ed.), 1972, Improving Inner-City Marketing. AMA,
Chicago.
Appel, D., 1972. The supermarket: early development of an institutional innovation. Journal of Retailing 48 (Spring), 3952.
Arnold, S.G., Oum, T.H., Tigert, D.J., 1983. Determinant attributes in
retail patronage: seasonal, temporal, regional and international comparisons. Journal of Marketing Research 20 (May),
149157.
Betancourt, R., Gautschi, D.A., 1986. The evolution of retailing: a
suggested economic interpretation. International Journal of
Research in Marketing 3 (4), 217232.
Betancourt, R., Gautschi, D.A., 1990. Demand complementarities
household production and retail assortments. Marketing Science 9
(Spring), 146161.
Caplovitz, D., 1967. The Poor Pay More: Consumer Practices of Low
Income Families. The Free Press, New York.
Craig, C.S., Douglas, S.P., 2000. International Marketing Research,
Second ed. Wiley, New York.
Findlay, A., Paddison, R., Dawson, J. (Eds.), 1990, Retailing
Environments in Developing Countries. Routledge, London.
Goldman, A., 1974. Outreach of consumers and the modernization of
urban food retailing in developing countries. Journal of Marketing
38 (October), 816.
Goldman, A., 1981. Transfer of a retailing technology into less
developed countries: the supermarket case. Journal of Retailing 57
(2), 529.
Goldman, A., 1982. Adoption of supermarkets shopping in a
developing country: the selective adoption phenomenon. European
Journal of Marketing 16/1, 1726.
Goldman, A., Krider, R.E., Ramaswami, S., 1999. The persistent
competitive advantage of traditional food retailers in Asia: wet
markets continued dominance in Hong Kong. Journal of
Macromarketing 19 (2), 126139.
Goldman, A., Ramaswami, S., Krider, R.E., 2002. Barriers to the
advancement of modern food retail formats: theory and measurement. Journal of Retailing 78 (4), 281295.
Herche, J., Balasubramanian, S., 1994. Ethnicity and shopping
behavior. Journal of Shopping Center Research 1, 6580.

ARTICLE IN PRESS
284

A. Goldman, H. Hino / Journal of Retailing and Consumer Services 12 (2005) 273284

Hirschman, E.C., 1981. AmericanJewish ethnicity: its relationship to


some selected aspects of consumer behavior. Journal of Marketing
45, 102110.
Hirschman, E.C., 1983. Cognitive structure across consumer ethnic
subcultures: a comparative analysis. Advances in Consumer
Research 10, 197202.
Hoch, S.J., Kim, B., Montgomery, A.L., Rossi, P.E., 1995. Determinants of store-level price elasticity. Journal of Marketing Research
32 (February), 1729.
Hui, C.H., Triandis, H.C., 1985. Measurement in cross cultural
psychology: a review and comparison of strategies. Journal of
Cross Cultural Psychology 16 (January), 131152.
IMA (Israel Manufacturers Association-Food Division) and GeoCartography, 1999. Food Purchasing Patterns in the Arab Sector,
Tel Aviv (Hebrew).
IMA (Israel Manufacturers Association-Food Division) and GeoCartography, 2003. Food Purchasing Patterns in the Arab Sector,
Tel Aviv (Hebrew).
Israel Central Bureau of Statistics, 1999. Family expenditure
survey1997/8general summary, Series No. 1147, Jerusalem.
Israel Customs Authority (Value Added Tax), 1997. Internal Documents, Jerusalem, Israel.
Jaffa Institute, 1997. Survey of Leisure and Employment among Arab
Woman in Israel, Nazareth, February (Hebrew).
Jamal, A., 1995. Food consumption among ethnic minorities: the case
of BritishPakistanians in Bradford, UK. British Food Journal 100
(5), 221228.
Jamal, A., 2003. Marketing in a multicultural world: the interplay of
marketing ethnicity and consumption. European Journal of
Marketing 37 (11/12), 15991620.
Jamal, A., 2005. Playing to win: An explorative study of marketing
strategies of small ethnic retail entrepreneurs in the UK. Journal of
Retailing and Consumer Services 12 (1), 113.
Jamal, A., Chapman, M., 2000. Acculturation and inter-ethnic
consumer perceptions: can you feel what we feel? Journal of
Marketing Management 16, 365391.
Kaufman, C.J., Hernandez, S., 1991. The role of the bodega in a US
Puerto Rican community. Journal of Retailing 67 (Winter),
375396.
Kaynak, E., 1985. Global spread of supermarkets: some experiences
from Turkey. In: Kaynak, E. (Ed.), Global Perspectives in
Marketing. Praeger Publishers, New York, pp. 7793.
Kaynak, E., Cavusgil, T., 1982. The evolution of food retailing
systems: contrasting the experience of developed and developing
Countries. Journal of the Academy of Marketing 10 (3), 249269.
Kumcu, E., Kumcu, M., 1987. Determinants of food retailing in
developing countries: the case of Turkey. Journal of Macromarketing 7 (fall), 2640.

Lavin, M., 1996. Ethnic/racial segmentations: insights from theory and


practice. Journal of Retailing and Consumer Services 3 (2), 99105.
McCracken, G., 1986. Culture and consumption: a theoretical account
of the structure and movement of the cultural meaning of consumer
goods. Journal of Consumer Research 13 (June), 7181.
Mennel, S., Murcolt, A., Otterloo, V., 1992. The Sociology of Food:
Eating, Diet and Culture. Sage, London.
Messinger, P.R., Narasimhan, C., 1997. A model of retail formats
based on consumers economizing on shopping time. Marketing
Science 16 (1), 123.
Miller, D., 1998. A Theory of Shopping. Cornell University Press,
Ithaca, NY.
Miossec, J.M., 1990. From Suq to supermarket in Tunis. In: Findlay,
A., et al. (Eds.), Retailing Environments in Developing Countries.
Routledge, London, pp. 227242.
Murcott, A., 1983. The Sociology of Food and Eating. Gower,
Aldershot.
Oswald, L., 1999. Cultural swapping: consumption and the ethno
genesis of middel-class Haitian immigrants. Journal of Consumer
Research 25, 303318.
Othman, K., 1990. Patterns of supermarket use in Malaysia. In:
Findlay, A., et al. (Eds.), Retailing Environments in Developing
Countries. Routledge, London, pp. 205215.
Penaloza, L., 1994. Atravesando fronteras/border crossings: a critical
ethnographic exploration of consumer acculturation of Mexican
immigrants. Journal of Consumer Research 21, 3254.
Penaloza, L., Gilly, M.C., 1999. Marketer acculturation: the changer
and the changed. Journal of Marketing 63, 84104.
Rinnawi, K., 2003. The Palestinian Society in Israel: An Ambivalent
Agent. College of Management, Academic Track, Rishon Lezion
(Hebrew).
Samiee, S., 1993. Retail and channel consideration in developing countries: a review. Journal of Business Research 27 (2),
103130.
Slater, Charles, C., Henley, D., 1969. Market processes in La Paz.
Michigan State Uuiversity, Latin American Studies Center,
Bolivia, East Lansing.
Wallendorf, M., Reilly, M.D., 1983. Ethnic migration, assimilation
and consumption. Journal of Consumer Research 10 (December),
292302.
World Bank, World Development Report, 1999/2000. New York.
Yavas, U., Kaynak, E., Borak, E., 1981. Retailing institutions
in developing countries: determinants of supermarket patronage
in Istanbul, Turkey. Journal of Business Research 9 (4),
367379.
Zain, O., Rejab, I., 1989. The choice of retail outlets among urban
Malaysian Shoppers. International Journal of Retailing 4 (2),
3545.