Market analysts and strategists have offered incumbent retailers a myriad of strategies to position themselves for market turbulence from the entry of large-format retail stores. Yet, storeowners are still left scratching their heads, unable to launch any meaningful response to the strategic 'ploy' (Mintzberg, 1987) of 'everyday low prices' that large-formats place in the minds of consumers (Litz and Stewart, 1996; Peterson and McGee, 2000; Barber and Tietje, 2004). The Incumbent retailers struggle with the need to select a particular niche width that will make them competitive once again in the presence of large-format retailers. Should they specialize, offering a narrow but deep line in a particular area of retail home improvement products or should they generalize by offering a wide variety that caters to a broad array of home improvement items?

This study followed incumbent retailers of home improvement products over six years (1995-2001), comprising a period before, during, and after the entry of Home Depot and Revy to the Winnipeg, Manitoba market. In 1995, Winnipeg was a rather stable market, comprised of locally owned home improvement stores and national retailers of moderate size. There were no large-format sellers of home improvement products to contend with. Within two years of the arrival of the first large-format store only two-thirds of the incumbents remained. What characteristics did these surviving stores possess that enabled them to compete in the same market with the large-format retailers? This study utilizes literature from niche theory within population ecology to examine the differences between firms that survived and those that failed during the period of market turbulence experienced by incumbent retailers of home improvement in the Winnipeg market. A brief review of literature in population ecology and niche width are presented with context from prior research on large-formats in order to generate the hypothesis. Research methods, analysis and findings are then presented. The paper concludes with a discussion.

Population ecology has its origins in biology and ecology, a concept that regards natural selection or "survival of the fittest" as a means of explaining the success of one organism over another by virtue of having the right combination of characteristics to successfully adapt and thrive in their environment (Aldrich, 1979). The population ecology perspective fits well here since this study attempts to explain why certain organizational characteristics resulted in one-third of study participants failing. Fitness in organizations is defined as "the probability that a given form of organization would persist in a certain environment" (Hannan and Freeman, 1977; 937). Organizations that have a 'fit' with the environment are better suited and therefore 'selected' by the environment for survival while those that are less fit fail to continue existing in their current state and then are selected out. In the context of this study, incumbent retailers that did not have a fit with the retail home improvement market in Winnipeg failed.

Niche Theory

In niche theory, two similar organizations with the same environmental requirements cannot exist in the same time and space without competition. Competition between existing organizations in the same environment ensues when they compete for the same limited set of resources (Hawley, 1950; Hannan and Freeman, 1989; Baum and Korn, 1996). Organizations that are of greatest similarity will be in greatest competition. As a result, organizations will vie for areas of distinction that allow them to minimize any overlap, often exploiting a narrower set of resources not sought out by other organizations. This narrow set of resources has been identified in organizational ecology with terms such as a "niche" (Carroll, 1985), or "market" (Baum and Korn; 1996). While some organizations will pursue "specialization" via niche, others will look towards a "generalization" approach.

The distinction between specialism and generalism is often based upon the "width of the niche" -- the range of environmental dimensions across which a population exists (Carroll, 1985); or "market domain" -- the set of markets in which a firm operates (Baum and Korn, 1996; p. 256). Porter (1996) has characterized this activity as the "variety-based" approach for positioning an organization - drawing upon a specific subset of an industry's products and services. Specialists will maximize their focus in a particularly narrow niche. Incumbent retailers in home improvement that focus upon products such as paint or plumbing supplies are considered "specialists" in this study. By specializing, they reduce "niche overlap" (Baum and Singh, 1994) or "domain overlap" (Baum and Korn, 1996), thereby reducing exposure to competitive forces, but they assume greater risk of becoming irrelevant if the environment's resources were to shift completely beyond their niche (Hannan and Freeman, 1977). Generalists accept a lower level of risk from shifting resources by distributing their focus over a wider niche, thereby providing a wider spectrum of goods and services within a given environment (Hannan and Freeman, 1977). In return, generalists average their outcomes over a wide range of conditions. Incumbent retailers deemed generalists in this study are stores that offer a wide variety of products that cater to home improvements. "Generalist organizations compete in a variety of domains simultaneously, whereas specialists focus on only one or a limited few" (Carroll, 1985; p. 1266).

In a stable environment, generalists and specialists can find equilibrium: specialists relying upon a narrow band of resources within an environment, leaving the remaining environment for others; and generalists relying upon a broad spectrum of resources, but not so much in any specific area as to be in direct conflict with specialists. Prior to the entry of Home Depot and Revy, the Winnipeg home improvement market would have fit the definition of "stable" (Khandwalla, 1977). According to research in population ecology, in a stable market, specialists will be able to out-compete generalists in their given area (Hannan and Freeman, 1977). However, when the environment becomes unstable due to external shocks, then generalists and specialists alike are exposed to a more turbulent environment and become vulnerable to losing necessary resources to survive.

Increased competition in the market due to the entry of large-format retail to a region would deem an environment as turbulent leading to a reduction in performance or firm failure (Khandwalla, 1977). "The competitive environment is in continual flux, as customers, suppliers, and competitors enter and depart, as technologies evolve, and competitive initiatives are launched and countered in the arena" (Dess and Picken, 1999; p. 104). Heightened competition can increase the 'niche density' such that it affects the finite carrying capacity of an environment. The turbulence generated by the entry of competition leads to a "shake out" (Porter, 1985) where unfit firms struggle. This prompts an adjustment where incumbents either 'escape' from their original niche (Delacroix et al, 1989), or experience organizational 'death' (Hannan and Carroll, 1992). The incumbents that remain once equilibrium has been reestablished comprise a population that, in general, demonstrates the best fit with the new environment's competition and market demand (Aldrich, 1979; Delacroix et al, 1989; Hager, Galaskiewicz, Bielefeld, and Pins, 1999).

Carroll (1985, 1987) found in a study of daily newspapers that when competition was increased beyond the carrying capacity of the market, the death rate of generalist newspapers increased, while that of specialist newspapers decreased. "Increased concentration among general dailies enhances life chances of specialized newspapers" (Carroll, 1985; p. 1273). This would suggest that, from a pre-entry standpoint, incumbent retailers that specialize should enjoy better performance over generalists during the period of turbulence. However, Baum and Korn (1996) found in their study of competitive rivalry of California commuter airlines that when the degree of multi-market contact, or domain overlap between firms increased, competition in one market was attenuated by the potential for retaliation in another market similarly held by the two firms. This 'mutual forbearance' meant that close competitors in multi-markets may not actually be the most intense rivals. Instead, multi-market firms may focus upon single-market holders or specialists that possess a market domain overlap in a single area (p. 257). This finding suggests the opposite effect from an environmental shock from increased competition within a given ecological niche. So, mixed signals are received on this particular characterization of the firm.

However, in the context of retail, there appears to be greater consensus on the discussion of specialists or generalists in the presence of increased competition. Miller, Reardon, and McCorkle (1999) found that while the entry of large format retail may create initial market turbulence, the long term effects result in a positive market where small retailers and large formats may coexist. McGee and Rubach (1996) compared the effectiveness of different strategies when 238 small independent retailers perceived the environment as 'hostile' or 'benign' following the recent arrival of Wal-Mart in their trade area. They found that in terms of overall performance, the strategy of 'target marketer' (providing products that address a specific market segment or niche) was the best among four strategies to apply in a hostile environment (an environment where the presence of a Wal-Mart is deemed to have made an impact on their business). The strategy of 'variety' (emphasis placed upon product variety) was rated lowest in performance. When retailers perceive their environment as 'benign' (the presence of a Wal-Mart is deemed to not have made an impact on their business) the strategy of 'target marketer' (carrying relatively deep lines of unique and highly recognizable products) generated the greatest net income after tax of the three strategic approaches studied.

These findings suggest that, in both hostile and benign environments, retailers that focus upon product depth and specialization will generate greater performance than incumbents that engage in product breadth and variety. Litz and Stewart (1998) found that independent retailers tend to reduce product diversity, instead increasing upon service diversity or complexity. Furthermore, they parallel McGee and Rubach's (1996) findings that under high competition situations independents tend to become more niche-oriented in their products, opting for depth over breadth. Barber and Tietje (2004) also found that smaller stores, in the presence of large-formats, will find market opportunities by niching. It would appear that research in retail suggests a specialist approach in the presence of large-format retail. However, Litz and Stewart (1998) and McGee and Rubach (1996) did not utilize longitudinal methods that included pre-entry data collection in their research. As well, Barber and Tietje, 2004, focused on three retail stores in a given community, two incumbents and the entering Home Depot, while collecting consumer data two months prior to the entry of Home Depot and then six months post entry. The sample and data collection period may not have been sufficient to have witnessed any 'shake out' effects from the entering large format retailer.

Given the above discussion on the effect of the environment on organizations, and the difference in performance found in research that categorized organizations as "specialists" and "generalists" the following hypothesis is presented.

Hypothesis In the presence of market turbulence as the result of increased competition from entering large-format retailers, incumbent retailers that occupy a narrow niche will outperform incumbent retailers that occupy a wide niche.

A significant contributing element offered by this study is the rare look into the activities undertaken by numerous incumbent retailers at the pre-entry, entry, and post-entry stages of a large-format retailer. Based upon my review of the literature, no other study has looked at the impact of entering large-format retailers in a longitudinal analysis that includes pre-entry data with incumbent retailers as the unit of analysis. The data for this study was collected over six years in Winnipeg, Manitoba; a city rated as Canada's most economically diverse and closest in matching the overall economic diversity of Canada (Moody's Investors Services, 2000). The first wave of data collection, 1995, was prior to the physical presence of Home Depot or Revy in Winnipeg. However, rumours had already started that two large-format home improvement retailers would soon become part of the local retail market. The second wave occurred in 1997, a point at which Revy and Home Depot had entered the market. The third period of data collection occurred in 1999, which would reflect a point where both Revy and Home Depot had settled into the local retail market. A fourth and final collection of data occurred in 2001. This last wave completed three full periods of study and was limited to confirming whether or not the remaining incumbent home improvement stores in 1999 were still in existence, thereby determining the effect that their stated strategies in the 1999 data collection had on their operations two years hence.

The sampling frame applied for the first wave of collection (1995) was the local yellow pages, in which firms that met the sampling criteria were randomly selected for the study. The selection criteria had three requirements. First, businesses had to have a product line that would overlap with the products offered by Home Depot or Revy. The product sectors in home improvement included lumber/building supplies; paint/wall-paper/blinds; flooring/carpet/ vinyl; hardware; electrical/heating; lighting/fans; plumbing; kitchen/bathroom cabinets; and lawn/garden. Second, the business had to operate within the business trading area that Home Depot and Revy would be entering, which covers about 465 square kilometers, comprising the geographical area of the city of Winnipeg. Third, retail sales had to account for more than half of the business generated by the store.

In total, 72 retailers were contacted in T1 (1995) to participate in the study. Of the 72, data were received from 58 stores, for a response rate slightly over 80%. Sampling in subsequent waves of data collection (1997, 1999 and 2001) was limited to participants in the initial wave of data collection. Therefore, periods of data collection after the first wave (T1) were limited to the initial 58 participants. However, initial criteria still had to be maintained for a participant to remain in the study. If, for example, a study participant relocated to an area beyond the region in the study they would be deemed as having exited the competing market area and that would have been noted as a failure to compete with the large-format stores.

Over the span of the study, 20 of the 58 stores experienced firm failure, terminating their competition with the big box stores. At T2 (1997) twelve of the 58 incumbent stores had failed. By T3 (1999), eight more failed. A fourth survey in 2001 found that the incumbent population had stabilized, with total firm failure remaining at twenty. The total failure rate for the duration of the study was slightly under 35% (34.48%).

Data collection involved survey, interview, and observational methods. The application of the three methods allowed for triangulation of data (McGrath, 1982), generating support for findings collected in each of the three methods, as well as offering a richness of detail not necessarily found when limited to fewer methods. The results should increase the ability to draw conclusions from the study (Scandura and Williams, 2000). For example, statements made by storeowners could be compared against corresponding data found in the survey or by observations made onsite.

Dillman's Total Design method (1978) was utilized in order to develop and administer the survey instrument. With the help of a trained research assistant familiar with the home improvement industry, pilot testing of the 1995 survey was performed on a few Winnipeg businesses and later refined, dropping or adjusting questions as necessary. The survey instrument adopts the format designed by Litz and Stewart (1996) as a template, with the intent of gaining understanding of each retailer's products offered, customers served, and prior year's performance.

A One-Sample Kolmogorov-Smirnov Test (K-S Test) was conducted to test the distribution of each of the independent variables. Each test resulted in a significant finding (p >100), indicating that the independent measures are not normally distributed (see Table 1). Attempts to normalize the data, utilizing a variable's natural log or square root generated were not successful. Testing of the hypothesis therefore utilized non-parametric methods.

The first series of tests utilized the simple differentiation between 'specialists' and 'generalists,' followed by the categorization of 'limited-line,' broad-line,' and 'general merchandisers.' A third series examined the niche focus as outlined earlier, concluding with analysis that combined niche focus and retail store type for a measure labeled as 'niche width score.' The results of the test generated no significance in any of the outcome measures.

This research utilizes literature on niche theory to offer explanation on the performance of incumbent retailers in the presence of market turbulence due to competition by entering large-format retail. Suggestions that niching creates superior performance in the presence of market turbulence were the foundation for the hypothesis.

The results of the analysis generated findings, which were not fully anticipated, casting doubt on some of the research literature or possibly on the methods that were previously applied in order to generate their findings. Distinction of retailers by virtue of their niche width, as operationalized in the four different ways in this study, generated no significant difference in the success or failure of retailers. Furthermore, the fact that four approaches to describing the width of a retailer's niche all ended with no significant relationship to firm performance means that for this study, discussions on specializing based upon a variety-based position will not ensure good performance in a period of market turbulence.

Prior research in large-format retail has largely taken the stance that specializing, or exploiting a narrow band of resources provides optimal positioning in a period of turbulence. However, a key difference between the prior studies and this one is the inclusion of pre-entry incumbent store data to draw from in the analysis stage. If the environment had reached stability by T3 (which is assumed, since no firm failures occurred in the sample after T3, then market equilibrium has been reached. If that is the case, and if exit by incumbents prior to entry by large-formats is deemed avoidance of competition (Baum and Korn, 1996), then the least fit organizations may have actually exited the study market area prior to Home Depot and Revy entering. This would suggest that some of the richest information on firm failure and competition in the presence of large-formats occurred at pre-entry. In an attempt to replicate previous studies of entering large-formats, a Chi-square test was conducted on the hypothesis using only data that could have been gathered from incumbent retailers that were in existence at T2 and T3, leaving out all data from T1, including the twelve firms that had failed by the time Home Depot and Revy had entered the market. The tests generated significant results (p ≤ 100), suggesting that specializing would offer superior performance.

This exercise underscores the critical importance of longitudinal data collection when conducting research on organizational change at the population level. As Miller et al (1999) have stressed "competition is by its nature, a longitudinal construct" and "further examination [longitudinally] is needed" (p. 119). As well, because this study is longitudinal, it provided an opportunity to have an examination of incumbent retailers that includes firm failure as an outcome variable. This offered a look into the pre-existing conditions of incumbent retailers that resulted in firm failure. It also offered a response to prior calls for study on the impact of different strategic positions made by incumbent retailers and whether their actions (or inaction) has led to firm failure (Miller et al, 1999; McGee and Rubach, 1996). Furthermore, including firms that experience firm failure eliminates implicit bias of studying only surviving firms, which has been noted as a research limitation by authors in prior studies (Litz and Stewart, 1996; Peterson and McGee, 2000).

A major limitation of this study however is the low number of participants in the data. This could explain why it was not possible to generate a normal distribution in the independent variables used. However, where this data-set lacked in number of participants, it gained in having the data collected at four points over six years. Regarding construct measurement, the measures used in testing niche width may not have provided sufficient representation of a niche being narrow or wide. Still, some significant results were generated when the analysis did not include T1 data. However, assigning a niche width based upon coarse grain 'specialist' and 'generalist' categorizations; the consistency of the end use of products; a mathematically derived value of niche based upon the products sold, or any combination of the above may not be reliable measures of niche width.

As well, Generalizability of the findings must also be considered. While the niche dimensions described here could be applied in other retail environments the product variety and industry activities of the sector that the tests were undertaken in are fairly specific to the industry of home improvement.

Finally, we must accept that not all fit firms survive, and not all unfit firms fail. Aldrich (1979) stated "just as the fittest sometimes fail, because of random variations in the selection process, so patently maladaptive organizations sometimes survive" (p. 54). Here as well, some organizations that are not fit will close their doors for reasons not accountable in the hypothesis generated.

Although I have explored the position of retailers based upon variety-based positioning; one of three distinct sources as identified by Porter (1996), I have not looked at needs-based positioning - tailoring a set of products to best serve the needs of its target customers, or access-based positioning - configuring activities so that they are optimal in reaching the buyer. Initial study undertaken with this data set shows promise for significant differences in store performance based upon needs-based positioning. As well, prior research in retail has looked at access-based positioning, suggesting how stores may create advantages by increasing accessibility via location and store hours and therefore convenience to consumers. Longitudinal research that attends to the evaluation and interrelation of needs-based and access-based positioning of retailers is required so that we can confirm that positions recommended by past studies are still relevant.