Does small store location matter? A test of three classic theories of retail location.
Litz, Reginald A. ; Rajaguru, Gulasekaran
"It is not possible to determine how many businesses with
positive potential fail because of the entrepreneur's failure to
find a location compatible with the business. The choice of location can
make or break any business venture."
Scarborough and Zimmerer (2003: 468)
One of the most critical decisions a small retail establishment
makes is choice of location (Kuo et al., 2002). As Levy and Weitz note,
"while retailers can change their pricing, service, and merchandise
assortments in a relatively short time ... location decisions are harder
to change because retailers frequently have to either make substantial
investments to buy and develop real estate or commit to long-term leases
with developers" (2004: 217).
Accordingly, and as this study's title attests, some see
location as everything. However, others are not so persuaded and provide
convincing evidence for the strategic viability of less convenient
locations (Woodside and Trappey, 2001; Jones et al., 2003). In this
study we seek to shed light on these conflicting perspectives by
exploring the strategic significance of retail location as a physical
resource (Barney, 1991). We undertake this task by reporting findings
from our study of the relationship between retail store location
characteristics and small retailer performance.
Our paper proceeds as follows. First, we review insights from three
seminal works on location theory and advance a set of hypotheses linking
specific locational attributes to firm performance. Next, drawing on
insights from a group of small retail managers, we identify two
alternative location-related hypotheses. We then describe the research
design utilized in testing this set of hypotheses. Following a
description of our analysis strategy we report our findings from our
sample of over 340 small hardware stores. Our paper concludes with a
discussion of the implications of these findings for small-firm
researchers and practitioners.
Theoretical Framework and Hypotheses
Just over two millennia ago the Chinese strategist Sun Tzu asserted
that terrain plays a potentially critical role in shaping a
battle's outcome. In his treatise, The Art of War (1983), he also
offered a preliminary typology of six basic types of terrain: (1)
accessible ground, that is, ground that can be freely traversed by both
sides, (2) entangling ground, which can be abandoned but hard to
reoccupy, (3) temporizing ground, where neither side gains advantage by
making the first move, (4) narrow passes and (5) precipitous heights,
both of which were ideal for waiting for one's enemy, and finally,
(6) positions at a great distance from the enemy. As the previous
descriptions suggest, some types of terrain, such as accessible ground,
are more amenable to offensive manoeuvres; others, such as narrow passes
and precipitous heights, favour more inherently defensive tactics.
Sun Tzu's assertions also raise an important question of
special interest to retailers battling for customer patronage: To what
extent does retail store location influence retail store performance?
During the past century there has been significant attention devoted to
this question. In the review that follows, we revisit three classic
perspectives focusing on the importance of proximity to customers and
competitors, respectively, and also on the non-location related role of
product- and service-mix.
Retail Location Theory: Seminal Insights
During the past century retail researchers advanced several seminal
theories concerning the nature and significance of retail location.
These included central place theory (Christaller, 1933), spatial
interaction theory (Reilly 1929, 1931), and the principle of minimum
differentiation (Hotelling, 1929). As Brown's (1993) review of
retail location theory demonstrates, each perspective offers a
different, and complementary, perspective on the nature of locational
advantage.
Central place theory (Christaller, 1933) focuses on the role of
transportation costs and predicts that demand for a good or service
declines with distance from the source of supply. While this theory
fails to address the two possibilities of heterogenous goods, such as
expensive and infrequently purchased wares, and multi-purpose shopping
trips, it is nonetheless seen as possessing significant predictive power
to the extent its core assumption, of single-purpose shopping trips,
corresponds with customer demand. Accordingly, locations closer to the
centre of customer demand warrant a rent premium, compared to less
central locales.
In contrast, spatial interaction theory (Reilly, 1929, 1931)
assumes the possibility of customers making tradeoffs between
store-specific product- and service-related differences vis-a-vis the
shopping location's attractiveness, with what is transacted for
being more important than where the transaction occurs. Jones et al.
(2003) provide empirical support for this perspective with their finding
that location was less important for small convenience stores offering
less standardized and more personalized goods and services.
A third theory of retail location, advanced by Hotelling (1929), is
the principle of minimum differentiation. Whereas other perspectives
focus on locational centrality, Hotelling asserted that not every
transaction depended upon access to an entire market. Instead, what
mattered was relative proximity vis-a-vis other sources of the same
product or service. In other words, proximity to rivals is more critical
than proximity to customers. The practical implication of this principle
is evidenced by the "automobile row" phenomenon found in many
metropolitan areas, where several different automobile retailers
congregate in close proximity. Nelson (1958) advanced an accompanying
principle of "cumulative attraction" that articulates the
underlying logic of minimal differentiation. According to this
principle, "a given number of stores dealing in the same
merchandise will do more business if they are located adjacent, or in
proximity to each other than if they are widely scattered" (58).
In some respects, all three theories share at least one common
characteristic: the theories emanate from "positivist premises,
which presuppose, essentially, that there is an identifiable order in
the material world, that people are rational, utility-maximizing
decision makers and that economic activity takes place in freely
competitive, equilibriumseeking contexts or settings" (Brown, 1993:
186).
However, these commonalities notwithstanding, the theories also
differ in their predictive implications. While central place theory
places a priority on the relative density of a store's trading
area, spatial interaction theory posits performance-related differences
emanating from the specific competitor's product and service
offering. In contrast, the principle of minimum differentiation
hypothesizes that proximity to rivals enhances performance. Accordingly,
three divergent hypotheses follow:
H1. Small retailer performance is positively related to the store
location's surrounding customer density (Central place theory).
H2. Small retailer performance is unrelated to location but
determined by the small retailer's product and/or service mix
(Spatial interaction theory).
H3. Small retailer performance is positively related to the store
location's surrounding competitor density (Principle of minimum
differentiation).
Practitioner Propositions: Exploring Theories-in-Use
Recently, Pearce (2004) called on management researchers to listen
more closely to practicing managers and subject their 'folk
wisdom' to "careful analysis and ... more systematic reality
checks that could improve its veracity" (2004: 178). In the spirit
of Pearce's exhortation and other recent efforts to integrate
managerially grounded insights (Clarke et al., 2000; Pioch and Byrom,
2004), we "listened" to a group of retail store managers
concerning the nature of effective retail store management. While a full
description of our research design and sample is presented in the
methodology section that follows, we follow a slightly non-conventional
format here by presenting highlights of these managers'
"theories in practice" (Argyris and Schon, 1982) as they
concern the significance of store location. (1)
Of the 348 stores in our sample, 133 (38.2%) offered voluntary
comments. Our analysis of the comments proceeded in two stages. First,
we sorted the comments by whether they contained the keyword
"location." Using this heuristic we identified 11 comments
(8.3% of the total) that explicitly mentioned the word
"location." In the second stage, we undertook a content
analysis (Kolbe and Burnett, 1991) of the remaining 122 comments and
looked for any phrase or words pertaining to their specific store's
location. Following this strategy we found an additional nine responses
for a total of 20 comments (or 15% of all comments offered).
Two clusters of insights emerged from analysis of the comments. The
first set of insights concerned the extent to which managers either
supported, or refuted, one or more of the seminal perspectives advanced
previously. The second set of insights concerned additional hypotheses
involving the nature of location-based advantage.
Concerning mention of the seminal theories, the most frequently
mentioned perspective was Hotelling's principle of minimum
differentiation. Exemplar comments supporting the importance of locating
near competitors included the following:
"If I were starting a new business I would look for a location
as close to a Big Box as possible, preferably across the street, or even
next door." (California manager)
"We built at the same time next door to a Walmart. Strategy
was traffic. We get sales when Walmart is out of something."
(Minnesota manager)
"[We] wish our store was located right next door to the Big
Box." (Illinois manager)
However, several small store managers advocated a contrarian view,
preferring to be as far away from competitors as possible:
"My location away from Big Boxes makes it easier to
compete!" (North Carolina manager)
"Being on an island helps to insulate us from the Big
Boxes." (Washington manager)
A final insight, resonating with Hotelling's principle of
minimum differentiation, emanated from the mixed blessing of retail
malls, particularly as it concerned the small tenant's dependence
on anchor stores:
"When we learned that Home Depot and Menards were coming to
town there was a bit of nervous anticipation. At the same time, the
anchor grocery store in our strip mall was moving to a new location near
Home Depot. We have had only two years of declining sales (modest) since
our Big Box retailer opened, but our sales increases have been single
digit. Prior to their entry into our market, we experienced several
years of double-digit growth. In many ways I believe these competitors
have generated new customers for our business." (Illinois manager)
Several store managers also voiced hybrid perspectives integrating
central place theory and spatial interaction theory. For these managers
location was a key strategic element, but only if strategically
leveraged by offering unique and valuable products and services. For
example:
"Concentrate on your store's strengths. Ours is our
convenient location, our service level and our friendliness. We no
longer try to compete on price. We feel that we are worth more."
(Georgia manager)
"Our location, nothing else around within 12 miles. Our
dedication to greet each customer, personalized repairs especially in
the plumbing section. Our specialization in contractor sales selling to
the electrical, plumbing areas." (California manager)
A second cluster of insights suggested hypotheses not mentioned by
any of the three classic perspectives. Several managers, for example,
stressed the importance of transactional convenience. One Texas manager
articulated the essence of this view:
"I think convenience to the consumer is the only major
competitive advantage we [that is, small retailers] have. Understand
that convenience includes everything in the shopping experience from
location (access), parking, inviting store environment, helpful
knowledgeable sales staff when needed, quick checkout and competitive
pricing (not equal or better) ..." (Texas manager)
Another factor not expressly mentioned by location theorists
concerned the importance of location-specific embeddedness (Granovetter,
1985). The following comment encapsulates the essence of such
continuity-based advantage:
"Operating at the same location for 131 years we have
developed a strong loyal customer base. This fact was proven in the last
2 years. On Father's Day 2001 we had a fire that destroyed our
store, killed 3 firemen and hurt 50 other people. We thought of closing.
The community and customer base would not hear of it, within 2 days we
were up and running, our trucks were rolling the very next day. It was
nothing short of amazing. Despite not having a retail store we still had
an increase of sales in 2001." (New York manager)
[FIGURE 1 OMITTED]
We integrated these practitioner insights into our theoretical
framework in two ways. First, we expanded our hypothetical framework to
include two additional possibilities, specifically:
H4. Small retailer performance is positively related to the store
location's transactional convenience.
H5. Small retailer performance is positively related to the
store's locational continuity. Second, in consideration of the
distinctions made between certain retail contexts, such as strip malls
and freestanding locations, we controlled for building-specific
characteristics in our research design.
In summary, our research seeks to test the applicability of three
classic theories of retail location for a group of small retailers. In
an effort to include these same managers' theories-in-use we also
integrated two additional location-related hypotheses focusing on
transactional convenience and locational continuity (Figure 1).
Research Design
Sample
Our study was conducted in the United States retail hardware
industry, an industry that includes thousands of small retail
establishments. The research was conducted under the auspices of the
Russell R. Mueller Retail Hardware Foundation, the research arm of the
North American Retail Hardware Association (NRHA). The association
maintains an active database of 40,000-plus hardware/home improvement
retailers in the United States, each of which is periodically invited to
provide the trade association with store-specific information including
self-categorization data as either a hardware store, a home improvement
centre, or a lumberyard, and annual sales levels.
Our sample included all retail hardware stores in the NRHA's
database in the states of Texas, North Carolina, Washington, northern
California and Iowa, and the metropolitan areas of Chicago, New York
City/Long Island, Atlanta, Minneapolis-St. Paul, Kansas City and San
Diego that satisfied two criteria: first, they identified themselves as
hardware stores, and second, they reported 2002 sales of between
$500,000 and $5,000,000. We did not discriminate based on whether the
stores were single- versus multiple-unit operations; likewise, our
design did not control for membership in a retail hardware trade-name
franchise (Litz and Stewart, 1998).
Survey
The survey was developed using Dillmans' Total Design
Technique (1978), together with counsel from industry experts at the
NRHA. The instrument was pre-tested on representative hardware stores in
North Dakota and Minnesota in the spring of 2003. Following revision of
the survey we began contacting the stores in the fall of 2003 and
completed our calling by the spring of 2004.
Response
We made up to three attempts to contact the 1,971 stores in our
sample; 118 (6%) of the stores were no longer in business; 92 (4.7%)
were categorized as misclassified; we encountered a wrong number for 4
(0.2%) of the stores; for 437 (22.2%) the store manager was
inaccessible, that is, either out of the store or serving a customer at
the time we called. We did, however, succeed in speaking to managers of
1,320 (67%) stores. After introducing ourselves and explaining the
purpose of our study, we asked each store manager to participate in the
study by completing a mail survey; 355 (18%) managers declined to
participate, while 965 (49%) agreed to complete the survey. We
subsequently followed up with up to three telephone calls to confirm the
survey's arrival and request its timely completion and return. Our
final tally counted 346 returned surveys. Compared to the total sample
this constituted a response rate of 17.6%; compared to the 1,320 stores
where we made contact, the response rate was 26.2%; finally, relative to
the 965 who agreed to participate, the effective response rate was
35.9%.
Operationalizations
Dependent Variables
We utilized the standard retail productivity measure of
sales-per-square-foot (Levy and Weitz, 2004) as our performance measure,
with 2002 sales divided by total square footage. In order to more
closely approximate a normal distribution we utilized a logarithmic
transformation.
Independent Variables
Customer density (Central place theory). We asked each respondent
to identify their store's zip code. We then located the most recent
(2001) census estimate for relative population density for the
particular zip code area.
Product mix and service mix (Spatial interaction theory). Given
spatial interaction theory's focus on the individual store's
specific product and service mix, we operationalized the theory in terms
of product and service mix diversity. The survey included a list of 22
different product categories (see Table 1) and 36 different specialized
services (see Table 2).
Each respondent was asked to identify which products and services
were included in its product and service mix, respectively. We then
summed the total number of product categories included in the
store's product mix, and total number of services included in its
service mix, as measures of product and service diversity, respectively.
Competitor density (Principle of minimum differentiation). Using
each respondent's zip code, we located census data on the total
number of retail establishments operating in the same zip code as the
respondent.
Transactional convenience. The survey asked respondents to estimate
the number of free parking spots located within 50 yards of the
store's front entrance, and also the total number of hours of
operation during each day of a typical week. We then summed the number
of hours across all seven days and multiplied by 52 weeks to generate
the total number of hours per year the store was accessible to
customers.
Locational continuity. We operationalized continuity for both the
store and its customers. In terms of store continuity, we asked
respondents how many years their store had been operating at its current
location; in terms of customer continuity, we included census data on
the percentage of residents living in the same house for at least five
years in the store's zip code area.
Control Variables
Intraorganizational. We controlled for store size as
operationalized by number of square feet. Also, given the aforementioned
differences in store type, respondents were asked to categorize their
store's location using Bell's (1994) six-level hierarchy of
locations, which included: (1) free-standing stores, (2) downtown
shopping districts, (3) strip malls with no anchor store, (4) local
malls with one large anchor store, (5) regional malls with two or three
anchor stores, and (6) super-regional malls with four or more anchors.
Just over one-half (212 stores or 61.5%) of respondents identified their
store as freestanding entities; 61 (17.7%) categorized their store as
part of a downtown shopping district; 55 (15.9%) as strip malls with no
anchor store; 14 (4.1%) as local malls with one large anchor store, and
3 (0.9%) as regional malls with two or three anchor stores. No stores
identified themselves as operating within super-regional malls. We then
converted stores operating as part of a downtown shopping district,
stores in strip malls with no anchor store, and stores in local malls
with one large anchor store into dummy variables (that is, Building
types 2, 3, and 4, respectively) using free-standing entities as the
reference category. Given the small number of stores identifying
themselves as operating in regional malls with two or three anchor
stores, we did not include them as a dummy category.
Extraorganizational. We also controlled for house age, per capita
income, and rate of population growth.
Validity and Reliability Assessment
We sought to assess the validity and reliability (Carmines and
Zeller, 1979) of our data in three ways. First, we compared the sales
levels reported in our survey to those reported to the North American
Retail Hardware Association. Consistent with Dess and Robinson's
(1984) observation concerning the problem of securing performance data
from privately-held operations, 59 stores (23.1%) did not share their
2002 sales data. However, of the remaining 287 stores, 264 stores (92%)
reported sales of $500,000 to $5,000,000. Of those falling outside the
range, 20 (5.5%) stores reported sales below $500,000 and three (0.87%)
stores reported sales above $5,000,000. A second validity check involved
the number of stores excluded from participation in the study because
they did not consider themselves hardware stores; as mentioned earlier,
less than 5% of the sample was categorized as misclassified. Taken
together, based on size-related sales and self-categorization data, we
accepted the data as suitable for the purposes of our study.
In order to assess the reliability of our data we resurveyed a
subset of the respondents (105 stores, or 30.3% of the original sample)
in the fall of 2005 during which we asked each store manager to complete
virtually the identical version of the survey, excepting an update of
annual sales data for 2004. We followed up our mailing with a phone call
to request timely completion and return of the survey. Just under
one-third (32 stores or 30.5%) returned the survey within two months; in
all but one case the same individual completed the survey (the sole
exception involved the spouse of the original respondent completing the
survey). A comparison of the two data points across the 32 stores showed
reported square feet increasing over the two-year period by 5.4% (9,787
square feet for 2003 versus 10,343 for 2005). Likewise, mean reported
sales increased from $1,131,540 to $1,425,293 (or an average annual
increase of 8% over the three-year period). The mean number of reported
parking spots increased slightly, from 42.9 in 2003 to 45.8 in 2005.
While differences may stem from either manager recall error, and/or bona
fide changes to either store square footage or available parking, we
accepted the data as sufficiently reliable for the purposes of our
investigation.
A summary of descriptive statistics and correlation matrix for the
key variables are presented in Table 3. The regression results from the
five different regression models are presented in Table 4.
Model 1 examines the variability in sales-per-square-foot
(dependent variable) as a linear function of variables relating to all
five of our hypotheses. In addition, we also considered several
disaggregated measures for number of retailers (Hypothesis 2) to examine
for effects related to specific types of retail establishments including
(i) retail flooring stores, (ii) retail home furnishing stores, (iii)
retail hardware stores, (iv) retail building materials stores, (v)
retail gardening stores and (vi) general merchandise stores. Only
general merchandise stores achieved significance, which we include in
Model 2. In Models 3 through 7 we include the variables related to each
of the five hypotheses one at a time to examine the stability of the
parameter estimates established in Models 1 and 2. If the sign and the
significance of the parameter estimates do not change significantly in
Models 1 and 2, compared to the other more restricted models, we say
that the parameter estimates are stable and robust. In each model, we
also include appropriate control variables as indicated in the
operationalization section.
In general, we find all parameter estimates are robust. The
F-statistic provides support for the significance of each regression
model insofar as it indicates that at least one of the parameter
estimates is statistically different from zero. Though the correlation
matrix reported in Table 2 indicates some minimal level of collinearity
among the independent variables, the formal measure for
multicollinearity (i.e., Variance inflation factor (VIF)) did not show
any problems of multicollinearity between the independent variables. For
the sake of brevity, the VIFs to examine the multicollinearity are not
reported here. (2) The estimated Durbin-Watson statistics, being close
to 2, indicate that the errors are serially uncorrelated. In addition,
the White heteroskedasticity test for constant variances indicates that
the residual variances are constant across each observation and hence
safe to use. The major problem from residual diagnostics comes from the
JB test on normality. In all models, we reject the null of normality at
1% level of significance as it indicates that errors were not normally
distributed; hence we did not compare the estimated t-statistic with
tabulated student t-tables to make the inference about the statistical
significance of estimated parameters. To overcome this problem, we
obtained the critical values through bootstrap sampling of 10,000
replications.
The regression results based on bootstrap sampling reported in
Table 4 show that the customer density has a positive relationship with
store performance, lending strong support for Central Place Theory. On
the other hand, our analysis failed to reject the absence of
relationship between store performance and product/service mix,
indicating evidence against Spatial Interaction Theory. Excepting the
number of general merchandise stores, the number of retailers was found
to be statistically insignificant in determining store performance, thus
lending only limited support for the Principle of Minimum
Differentiation.
We also find that the transactional convenience measures, such as
presence of free parking spots within 50 yards of the store's front
entrance and the number of hours of operation during each day of a
typical week, positively influence store performance at the 1% level of
significance in all regressions. Moreover, these regressions also show
that locational continuity, such as store continuity (number of years of
operations in the same area) and customer continuity (percentage of
people living in the same house for at least five years), also
positively enhances store performances at the 5% level of significance.
In summary, in terms of our hypothesized relationships (Figure 1),
we find support for Hypothesis 1 (Central Place Theory). Conversely,
excepting Model 4 that showed a positive relationship between diversity
of service mix and performance, we find no evidence in support of the
Spatial Interaction Theory (Hypothesis 2) and only limited support for
the Principle of Minimum Differentiation (Hypothesis 3). However,
significant support is evident for both manager-grounded hypotheses
linking performance to locational continuity, both for the store and its
residents, and transactional convenience, both for the number of free
parking spots and total number of hours of operation. No significant
relationships were evident for any of the different types of retail
locations, nor any of the environmental control variables.
Discussion
This research seeks to advance understanding concerning the
relevance of classic theories of retail location for small retail
establishments. We undertook this task in two stages; first, we compared
longstanding theories, such as central place theory, spatial interaction
theory, and the principle of minimum differentiation, with
"theories in practice" offered by a group of hardware store
managers. In terms of anecdotal commentary, the principle of minimum
differentiation was mentioned most frequently; however, comments were
not unanimously positive, with voices both supporting and refuting
Hotelling's principle. Also, the managers offered two additional
hypotheses: the first concerned the importance of locational continuity;
the second, the critical role of convenient access.
We then tested all five hypotheses statistically on a sample of
retail hardware stores. As the preceding analysis reports, significant
support was evident for central place theory and both managerial
hypotheses. In essence, small retailers' performance prospects are
significantly enhanced when they operate in more densely populated areas
over longer periods of time, and offer their customers easy access, both
in terms of longer hours of operation and more accessible parking
facilities. At the same time, while more diverse product mixes did not
achieve a statistically significant relationship with performance, more
diverse service mixes did, but only when considered apart from the other
predictors (Model 3). Likewise, no discernible advantage was
attributable to operating in more sophisticated retail venues, as might
be suggested by conventional retail logic (Bell, 1994: Levy and Weitz,
2004).
Implications
Several implications follow from these findings. While their
potential generalizability must be tempered by the recognition that
these findings emanate from a very specialized subcategory of retail
establishments, they nonetheless point to some guiding principles that
managers of new, established, closing and stagnant retail establishments
might consider.
The first implication is relevant for managers contemplating
launching a retail startup. According to these findings, locational
centrality should be prioritized, even if it means forgoing a more
sophisticated retail environ. In the spirit of this recommendation,
managers of new start-ups might also be well advised to consider buying
out a well-located establishment, particularly one operating at the same
location for a significant period of time.
A second implication is of relevance for managers of growing
operations. Rather than simply following another conventional maxim,
namely that "If you're not growing you're dying,"
managers of expanding retail firms (Bitner and Powell, 1987) might
carefully reflect on the reasons for their success. Specifically, to
what extent is their success due to locational, rather than operational,
characteristics?
A third implication, essentially the converse of the first, is for
managers considering selling their retail establishments. These managers
need to be especially careful not to overlook the strategic value of a
well-situated, easily-accessed store, particularly where book value of
an enterprise's physical facilities significantly understates
current market value.
A final implication is of interest for retail managers facing flat
sales, whose store may be in the right place but does not offer their
customers sufficient accessibility. One strategic tactic these managers
might consider is simply increasing their store's hours of
operation. While this decision will obviously also increase operating
expenses, the right location may generate bottom-line benefits that more
than offset these costs.
Limitations
These results must also be qualified by several caveats relating to
the study's conceptual framework, sample selection, variable
operationalization and implicit treatment of outlier cases.
First, concerning our study's conceptual framework, it should
be recognized that we have drawn on only two sources: classic retail
theory and an industry-specific community of retail managers. We
recognize that this implicitly overlooks work in other disciplines, such
as physical geography, and also the perspectives of other competent
professionals, such as real estate developers and commercial realtors.
A second cluster of caveats concerns our sample's focus on
retail hardware stores. While we have treated all stores
indistinguishably, given the diversity of the afore-cited comments, the
findings reported here may be contingent on finer-grained differences,
such as whether the store manager perceives their store as a specialty,
shopping, or convenience store. Likewise, given the industry-specific
nature of this sample, further research on other types of small retail
establishments, such as corner grocery stores, neighborhood bookshops
and local sporting good establishments, needs to be carried out.
Finally, our study did not consider the distinctive dynamics of service
businesses (Gilmore, 2003; Jones et al., 2003). Do the patterns observed
here apply to small service establishments, such as dry cleaners, hair
stylists and restaurants?
A third set of caveats relates to an implicit assumption embedded
in our operationalization of central place theory as customer
density--namely, that each store's customers came from the
immediate surrounding area. However, one or more of the stores sampled
here could conceivably include customers from outside the immediate area
who see their hardware store as a "mini-destination" shopping
experience. Such occurrences would, therefore, support spatial
interaction theory, rather than central place theory, as reported here.
A final set of caveats concern our research design's omission
of several variables that might influence store performance. At the
extra-organizational level these could include regional quantitative
and/or qualitative change in environmental munificence (Castrogiovanni,
1991); at the intra-organizational level these might include the likes
of store signage, store layout and selling behaviours, not to mention
alternative measures of store performance, such as employment growth
(Hoogstra and van Dijk, 2004). Taking an even finer-grained and
case-specific perspective, we also recognize the omission of
consideration for what might be called "the cursed location,"
that is, a location where one business after another opens and closes
until one finally moves in and succeeds. However, we have also excluded
the converse possibility, that is, the case-specific situation where a
store thrives despite appearing to violate many of the classic rules of
retail location noted here.
Further Research
While this study helps advance our understanding of the role
location plays for retail success, there are several research questions
future efforts on small retailer location might consider exploring. For
example, how does membership in a buying group cooperative, such as Ace,
Do-It-Best, or True Value, which typically provides its members with
access to more sophisticated location analysis, influence location
choice (Litz and Stewart, 1998)? A second question concerns the role of
owning versus renting one's facilities: to what degree does owning
one's store building impact on the merchandising and selling
strategies employed? Another related question concerns the potential
interaction between physical and virtual terrains: Simply said, to what
extent is the relationship between an establishment's physical
presence moderated by its web-based presence? Finally, our study only
sampled stores currently in operation; that is, we did not receive input
from any former owners or managers whose stores had closed. This
suggests one final question; specifically, how does a failed
venture's locational characteristics contribute to the post-failure
sensemaking processes of small-business operators (Shepherd, 2003)?
Conclusion
This study seeks to advance understanding concerning the importance
of small retail store location. In the spirit of the earlier cited
writings of Sun Tzu, the findings reported here support the conjecture
that terrain, or in this case, location, does matter. In addition, this
study also provides specific guidance as to the type of location small
firms might seek out--specifically, easily accessible positions at close
distance to stable customers.
Acknowledgements
The authors wish to thank Frank Hoy for helpful comments provided
on an earlier version of this paper. In addition, the first author
gratefully acknowledges the support received from the Russell R. Mueller
Retail Hardware Research Foundation for this research.
References
Argyris, C. and D.A. Schon. 1982. Theories in Practice: Increasing
Professional Effectiveness. San Francisco: Jossey-Bass.
Barney, J.B. 1991. "Firm Resources and Sustained Competitive
Advantage," Journal of Management 17: 99-120.
Bell, D. 1994. "Note on Store Location," Harvard Business
School Publishing, case number 9-593-112.
Bitner, L. and J. Powell. 1987. "Expansion Planning for Small
Retail Firms," Journal of Small Business Management 25, no. 2:
47-54.
Booth, J.G., R.W. Butler and P. Hall. 1994. "Bootstrap Methods
for Finite Sample Populations," Journal of the American Statistical
Association 89: 1282-89.
Brown, S. 1993. "Retail Location Theory: Evolution and
Evaluation," International Review of Retail Distribution and
Consumer Research 3, no. 2: 185-229.
Carmines, E. and H. Zeller. 1979. Reliability and Validity
Assessment. Beverly Hills: Sage Publications.
Castrogiovanni, G. 1991. "Environmental Munificence: A
Theoretical Assessment," Academy of Management Review 16, no. 3:
542-65.
Christaller, W. 1933. Central Places in Southern Germany,
translated by C. Baskin (1966). Englewood Cliffs, NJ: Prentice-Hall.
Clarke, I., H. Masahide and W. MacKaness. 2000. "The Spatial
Knowledge of Retail Decision Makers: Capturing and Interpreting Group
Insight Using a Composite Cognitive Map," International Review of
Retail, Distribution and Consumer Research 10, no. 3: 265-85.
Davison, A.C. and D.V. Hinkley. 1997. Bootstrap Methods and Their
Application. New York: Cambridge University Press.
Dess, G.G. and R.B. Robinson. 1984. "Measuring Organizational
Performance in the Absence of Objective Measures: The Case of the
Privately-held and Conglomerate Business Unit," Strategic
Management Journal 5: 265-73.
Dillman, D. 1978. Mail and Telephone Surveys: The Total Design
Method. New York: John Wiley and Sons.
Efron, B. and R. Tibshirni. 1993. An Introduction to the Bootstrap.
New York: Chapman and Hall.
Gilmore, A. 2003. Services Marketing and Management. London: Sage.
Granovetter, M. 1985. "Economic Action and Social Structure:
The Problem of Embeddedness," American Journal of Sociology 91:
481-510.
Hoogstra, G. and J. van Dijk. 2004. "Explaining Firm
Employment: Does Location Matter?," Small Business Economics 22,
nos. 3-4: 179-92.
Hotelling, H. 1929. "Stability in Competition," Economic
Journal 39 (March): 41-57.
Jones, M., D. Mothersbaugh and S. Beatty. 2003. "The Effects
of Locational Convenience on Customer Repurchase Intentions Across
Service Types," Journal of Services Marketing 17, no. 7: 701-12.
Kolbe, R. and M. Burnett. 1991. "Content-analysis Research: An
Examination of Applications with Directives for Improving Research
Reliability and Objectivity," Journal of Consumer Research 18:
243-50.
Kuo, R., S. Chi and S. Kao. 2002. "A Decision Support System
for Selecting Convenience Store Location through Integration of Fuzzy
AHP and Artificial Neural Network," Computers in Industry 47, no.
2: 199-214.
Levy, M. and B. Weitz. 2004. Retailing Management. Boston: McGraw
Hill Irwin.
Litz, R. and A. Stewart. 1998. "Franchising for Sustainable
Advantage? Comparing the Performance of Independent Retailers and
Trade-name Franchisees," Journal of Business Venturing 13, no. 2:
131-50.
Nelson, R.L. 1958. The Selection of Retail Locations. New York:
Dodge.
Pearce, J. 2004. "What Do We Know and How Do We Really Know
It?," Academy of Management Review 29, no. 2: 175-79.
Pindyck, R.S. and D.L. Rubinfeld. 1991. Econometric Models and
Economic Forecasts. New York: McGraw-Hill.
Pioch, E. and J. Byrom. 2004. "Small Independent Retail Firms
and Locational Decision-making: Outdoor Leisure Retailing by the
Crags," Journal of Small Business and Enterprise Development 11,
no. 2: 222-32.
Reilly, W.J. 1929. Methods for the Study of Retail Relationships.
Austin: University of Texas.
--. 1931. The Law of Retail Gravitation. New York: W.J. Reilly.
Scarborough, N. and T. Zimmerer. 2004. Effective Small Business
Management. Upper Saddle River, NJ: Prentice Hall.
Shepherd, D. 2003. "Learning from Business Failure:
Propositions of Grief Recovery for the Self-employed," Academy of
Management Review 28, no. 2: 318-28.
Tzu, S. 1983. The Art of War (edited by J. Clavell). New York: Dell
Publishing.
Woodside, A. and R.J. Trappey, III. 2001. "Learning Why Some
Customers Shop at Less Convenient Stores," Journal of Business
Research 54: 151-59.
(1.) We extracted these working hypotheses from an open-ended
question found in the final section of our survey instrument. The item,
which formed part of a larger study on the challenges faced by small
firms competing against big box retailers, stated: "We welcome any
additional comments you might have on the general topic of competing
with 'Big Box' stores or retailing in general. Please write
your comments in the space below."
(2.) Available from the second author upon request.
Contact
For further information on this article, contact:
Reginald A. Litz, Asper School of Business, University of Manitoba,
Winnipeg, Canada, R3T 2N2
Telephone: (204) 474-9406/Fax: (204) 474-7544
E-mail: rlitz@cc.umanitoba.ca
or
Gulasekaran Rajaguru, Bond University, Gold Coast, Queensland,
Australia, 4226
Telephone: +61 7 5595 2049/Fax: (07) 5595-1160
E-mail: rgulasek@bond.edu.au
Reginald A. Litz, Asper School of Business, University of Manitoba,
Winnipeg, Manitoba, Canada/Bond University, Gold Coast, Queensland,
Australia (Visiting Professor)
Gulasekaran Rajaguru, Bond University, Gold Coast, Queensland,
Australia
Table 1: List of Product Categories
* Hardware and Fasteners
* Hand Tools
* Power Tools
* Power Tool Accessories
* Electrical, Lighting and Fans
* Plumbing, Heating and Cooling
* Paint Sundries, Home Decor
* Glass and Screens
* Lawn and Garden Tools
* Outdoor Power Equipment
* Lawn and Garden Chemicals
* Green Goods (live plants)
* Barbeque Grills and Accessories
* Outdoor Furniture
* Housewares and Small Appliances
* Major Appliances
* Sporting Goods
* Pet Supplies
* Automotive
* Toys
* Giftware
* Others
Table 2: List of Specialized Services
* Precision sharpening
* General sharpening
* Lock repairs/rekeying locks
* Screen cutting and screen window repairs
* Glass cutting and glass window repairs
* Faucet rebuilding
* Electric power tools repairs
* Small electrical appliance repairs
* Small engine repairs
* VCR/TV repairs
* Bicycle repairs
* Paint mixing
* Computer color matching
* Pipe cutting
* Key cutting
* Door sizing
* Rope/chain cutting
* Engraving
* Licenses (fishing, etc.)
* Window shade cutting
* Custom millwork
* Free delivery to customers' homes
* Special delivery services for the elderly and disabled
* Delivery-at-a-cost to customers' homes
* Assembling/installing store products in homes
* Repair jobs not requiring store's products
* Construction jobs not requiring store products
* Cleaning equipment
* Painting equipment
* Plumbing tools
* Small construction tools
* Heavy duty power tools
* Garden/outdoor tools
* Party equipment
* Vehicle rental services
* Accept credit/debit cards
* Trade accounts for local businesses/organizations
* Trade accounts for private individuals
Table 3. Descriptive Statistics and Correlation Matrix
Mean
s.d. 1 2 3
1. Performance 2.22 1.0 0.240 * -0.232 *
0.346
2. Customer density 2622.2 1.0 0.016
7163.8
3. Product mix 17.1 1.0
5.14
4. Service mix 16.8
6.34
5. Number of retailers 97.85
82.18
6. Number of retailers 2.43
(GM) 2.80
7. Number of hours 3509
524.8
8. Parking 29.27
24.18
9. Same house 0.543
0.071
10. Location duration 31.77
22.98
11. Square feet 8260.4
6246.9
12. House ownership 0.668
0.117
13. Per capita income 22387.9
5545.3
14. Population growth 0.171
0.154
Mean
s.d. 4 5 6
1. Performance 2.22 -0.174 * 0.160 * 0.21 *
0.346
2. Customer density 2622.2 -0.04 0.39 * 0.19 *
7163.8
3. Product mix 17.1 0.350 * -0.061 -0.06
5.14
4. Service mix 16.8 1.0 -0.06 -0.08
6.34
5. Number of retailers 97.85 1.0 0.534 *
82.18
6. Number of retailers 2.43 1.0
(GM) 2.80
7. Number of hours 3509
524.8
8. Parking 29.27
24.18
9. Same house 0.543
0.071
10. Location duration 31.77
22.98
11. Square feet 8260.4
6246.9
12. House ownership 0.668
0.117
13. Per capita income 22387.9
5545.3
14. Population growth 0.171
0.154
Mean
s.d. 7 8 9
1. Performance 2.22 0.044 -0.080 0.144 *
0.346
2. Customer density 2622.2 -0.051 -0.020 0.167 *
7163.8
3. Product mix 17.1 0.227 * 0.109 0.003
5.14
4. Service mix 16.8 0.322 * 0.125 ** 0.057
6.34
5. Number of retailers 97.85 0.015 0.070 -0.102
82.18
6. Number of retailers 2.43 -0.015 -0.027 0.06
(GM) 2.80
7. Number of hours 3509 1.0 0.308 * -0.089
524.8
8. Parking 29.27 1.0 -0.12 **
24.18
9. Same house 0.543 1.0
0.071
10. Location duration 31.77
22.98
11. Square feet 8260.4
6246.9
12. House ownership 0.668
0.117
13. Per capita income 22387.9
5545.3
14. Population growth 0.171
0.154
Mean
s.d. 10 11 12
1. Performance 2.22 0.165 ** -0.331 * 0.218 *
0.346
2. Customer density 2622.2 0.073 -0.007 0.415 *
7163.8
3. Product mix 17.1 -0.104 0.143 ** -0.046
5.14
4. Service mix 16.8 -0.168 * 0.016 -0.058
6.34
5. Number of retailers 97.85 0.059 0.015 0.299 *
82.18
6. Number of retailers 2.43 0.054 0.043 -0.109
(GM) 2.80
7. Number of hours 3509 -0.224 * 0.285 * -0.013
524.8
8. Parking 29.27 -0.13 ** 0.413 * -0.019
24.18
9. Same house 0.543 0.099 -0.166 * -O.OOl4
0.071
10. Location duration 31.77 1.0 -0.096 0.066
22.98
11. Square feet 8260.4 1.0 -0.055
6246.9
12. House ownership 0.668 1.0
0.117
13. Per capita income 22387.9
5545.3
14. Population growth 0.171
0.154
Mean
s.d. 13 14
1. Performance 2.22 -0.014 -0.070
0.346
2. Customer density 2622.2 0.226 * -0.265 *
7163.8
3. Product mix 17.1 -0.073 0.071
5.14
4. Service mix 16.8 0.079 -0.078
6.34
5. Number of retailers 97.85 0.295 * -0.080
82.18
6. Number of retailers 2.43 -0.233 * -0.015
(GM) 2.80
7. Number of hours 3509 0.180 * -0.019
524.8
8. Parking 29.27 0.076 0.050
24.18
9. Same house 0.543 -0.166 * -0.512 *
0.071
10. Location duration 31.77 0.006 -0.184 *
22.98
11. Square feet 8260.4 0.063 0.151 *
6246.9
12. House ownership 0.668 0.5901 * 0.0860
0.117
13. Per capita income 22387.9 1.0 0.094
5545.3
14. Population growth 0.171 1.0
0.154
Note: * and ** denote the rejection of null of no contemporaneous
correlation at 1% and 5% levels of significance respectively.
Table 4: Dependent Variable: Log (Sales per Square Feet)
Model 1 Model 2
Independent Variables
Customer density 933E-06 ** 1.19E-05 *
(4.66E-06) (4.66E-06)
Product mix -0.008 -0.0007
(0.004) (0.001)
Service mix -0.004 -0.016
(0.003) (0.010)
Number of 3.86E-05
retailers (0.0002)
Number of 0.01 *
retailers (GM) (0.006)
Number of hours 00002 * 0.0001 *
(0.000001) (0.00002)
Number of 0.001 * 0.001 **
parking spots (0.0004) (0.0005)
Same house 0.73 ** 0.70 **
(0.31) (0.30)
Location duration 0.002 ** 0.002 *
(0.0008) (0.0006)
Control Variables
Building 2 -0.078 -0.09
(0.05) (0.05)
Building 3 -0.04 -0.04
(0.06) (0.05)
Building 4 -0.09 -0.1
(-0.09) (0.09)
Square feet -0.000161 * -1.52E-05 *
(0.00002) (2.91E-06)
House age 1.25E-07 -8.22E-08
(4.67E-07) (4.64E-07)
Per capita income -6.35E-06 -6.68E-06
(5.69E-06) (5.56E-06)
Population growth 0.18 0.16
(0.14) (0.14)
Constant 1.61 * 1.73 *
(0.25) (0.25)
Residual Diagnostics and Goodness of Fit
F-statistic 6.41 * 6.85 *
[0.00001] [0.00001]
Adjusted [R.sup.2] 0.32 0.29
Durbin-Watson 2.23 2.22
statistic
White 7.35 7.32
heteroskedasticity [0.88] [0.88]
Jarque-Bera 41.1 * 34.56 *
[0.00001] [0.00001]
Mode 3 Mode1 4
Independent Variables
Customer density 1.358-05 *
(5.08E-06)
Product mix -0.0068
(0.005)
Service mix 0.01 **
(0.004)
Number of
retailers
Number of
retailers (GM)
Number of hours
Number of
parking spots
Same house
Location duration
Control Variables
Building 2 0.07 0.08
(0.06) (0.06)
Building 3 0.02 0.03
(0.06) (0.06)
Building 4 -0.17 -0.19
(0.11) (0.11)
Square feet -1.63E-05 * -1.75E-05 *
(3.09E-06) (3.22E-06)
House age -4.93E-07 7.63E-07
(5.54E-07) (3.91E-07)
Per capita income 5.47E-06 7.09E-07
(6.84E-06) (5.86E-06)
Population growth 0.10 0.05
(0.15) (0.14)
Constant 2.21 * 2.32 *
(0.11) (0.13)
Residual Diagnostics and Goodness of Fit
F-statistic 6.79 * 5.75 *
[0.00001] [0.00001]
Adjusted [R.sup.2] 0.26 0.25
Durbin-Watson 1.86 1.98
statistic
White 7.28 7.17
heteroskedasticity [0.89] [0.95]
Jarque-Bera 40.2 * 49.3 *
[0.00001] [0.00001]
Mode1 5 Mode1 6
Independent Variables
Customer density
Product mix
Service mix
Number of 0.0002
retailers (0.0002)
Number of 0.008 *
retailers (GM) (0.0007)
Number of hours
Number of
parking spots
Same house
Location duration
Control Variables
Building 2 0.09 -0.09
(0.06) (0.06)
Building 3 0.02 -0.03
(0.06) (0.06)
Building 4 -0.15 -0.04
(0.11) (0.10)
Square feet -1.7E-05 * -1.52E-05 *
(3.21E-06) (2.92E-06)
House age 4.83E-07 1.06E-06
(4.07E-07) (2.84E-06)
Per capita income -3.51E-06 -1.13E-06
(5.94E-06) (5.29E-06)
Population growth 0.05 -0.11
(0.15) (0.14)
Constant 2.30 * 2.51 *
(0.11) (0.10)
Residual Diagnostics and Goodness of Fit
F-statistic 5.77 * 7.48 *
[0.00001] [0.00001]
Adjusted [R.sup.2] 0.19 0.18
Durbin-Watson 2.3 2.17
statistic
White 4.84 4.77
heteroskedasticity [0.98] [0.98]
Jarque-Bera 34.15 * 32.21 *
[0.00001] [0.00001]
Mode1 7
Independent Variables
Customer density
Product mix
Service mix
Number of
retailers
Number of
retailers (GM)
Number of hours 0.0001 *
(0.000004)
Number of 0.001 **
parking spots (0.0005)
Same house 0.79 **
(0.31)
Location duration 0.002 **
(0.0008)
Control Variables
Building 2 -0.10
(0.05)
Building 3 -0.05
(0.06)
Building 4 -0.11
(0.09)
Square feet -1.71E-05 *
(2.95E-05)
House age 1.00E-07
(2.63E-07)
Per capita income -1.16E-06
(4.96E-06)
Population growth 0.13
(0.14)
Constant 1.49 *
(0.25)
Residual Diagnostics and Goodness of Fit
F-statistic 7.36 *
[0.00001]
Adjusted [R.sup.2] 0.28
Durbin-Watson 2.18
statistic
White 7.15
heteroskedasticity [0.95]
Jarque-Bera 36.8 *
[0.00001]
Notes:
* and ** denotes the rejection of null hypothesis at 1% and 5% levels
of significance respectively. All inferences are based on the
bootstrap sampling of 10,000 replications.
Values in parentheses (.) are standard errors and the values in
square parentheses [.] are p-values.
Estimation of complete models is denoted by Model 1and Model 2,
while Models 3 through 7 represent the robustness of the
estimated parameters.
Number of retails (GM) refers to number of general merchandise
retail stores in the zip code area.
