- Email: [email protected]
Site selection in the US retailing industry
Applied Mathematics and Computation 182 (2006) 1725–1734 www.elsevier.com/locate/amc
Site selection in the US retailing industry Michael Nwogugu P.O. Box 170002, Brooklyn, NY 11217, USA
Abstract This article critiques existing store-location models and explains the many mathematical and methodological problems inherent in these models. The key issues are that behavioral factors are completely omitted in these models, emphasis on distance in un-warranted and site-specific and retailer-specific characteristics are typically omitted in these model. The author introduces new site-selection models. The US retailing sector is used an example. 2006 Published by Elsevier Inc. Keywords: Retail stores; Facilities location; Complexity; Corporate strategy; Risk management; Dynamical systems; Supply chain
1. Introduction The store location decision has become perhaps the most important decision for many retailers—Internet retail sales and mail order retail sales account for less than 8% and 30% of total US retail sales respectively. Store location sometimes contributes as much to a retailer’s brand value as some forms of marketing. Real estate accounts for more than 30% of retailers’ operating expenses and total assets. Retailing remains the largest industry in the US – in 2005, at least 30% of US GDP was from retail sales. In 2004/2005, the US retail industry had more than 1.5 million establishments, and employed more than 20 million people (17% of the US workforce). (Standard & Poors). 1.1. The store location (site-selection) decision Given the structural changes that occurred in the US retailing sector during the last five years, and the fact that real estate constitutes a substantial portion of retailer’s assets and operating expenses, the store location decision has become probably the most important strategy decision for retailers. The literature on store site-selection is extensive, multi-disciplinary and covers various approaches in operations-research/management science, economics and marketing (See: [25,60,39,31,32,38,46,44,11,8,17,20,18,3,6,28,4,5,55,68,69,78,72,67,73,77,63,58, 52,50,59,62,34,33,41,54,57,63,66,9,21,1,10,2,14,4,15,7,6,22,24,36,42,45,27,80,86,83,70,84]). However, all existing
E-mail addresses: [email protected], [email protected] 0096-3003/$ - see front matter 2006 Published by Elsevier Inc. doi:10.1016/j.amc.2005.12.050
1726
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
store-location models are inaccurate and some of the problems problems inherent in the existing literature are explained as follows. 1. Distance Existing store-location models over-emphasize and don’t incorporate the distance element properly. Most models erroneously assume: – That distance from each community (in the trade area) to the store is constant, and remains constant for each customer, for all trips to the store, and for all time periods. – That all residents of each community travel the full distance between their community and the store location whenever they shop. Some models have erroneously tried to use Newton’s gravity model (laws of physics) for location analysis [25,26]. However, the Newton’s law is not applicable because its assumptions do not fit the reality of shopping, consumers’ travel patterns, consumer’s reaction to distances and retail activities. • That distance is not affected by any other factors (traffic jams, psychology of driving, environmental psychology, perceived value of time spent traveling to/from store, perceived value or trip (multipurpose trips vs. single-purpose trips), etc.) and is completely determined by a physical measure. Prior site selection studies and models omitted relevant transportation psychology and environmental psychology analysis. (See: [53,51,19,47,40,29,12,71,81,85,75,76,64,65,6]). These site selection models erroneously state and assume that distance is a major element of site selection models. However, in reality, the distance between the store location and various communities does not matter (See: [52]), because: • Most people do not work in their immediate communities. • Most people shop before going to work, during lunch, after work, or on weekends. • Most people go to the shopping mall on their way to other destinations; and prefer malls that are on the route to destinations that they regularly go to. • People may have psychological attachments to specific retail brands. Kuo et al. [52] presented evidence (questionnaire results) that customers will walk/drive more than 10–30 min past certain retail stores to go shop at other retail stores. • Many people in most communities drive to store locations. When driving, actual distance (between the community and the store) becomes less relevant, partly because (a) the consumer has reserved time for shopping which includes the travel, and thus is less sensitive to distance, (b) the consumer is willing to travel to various locations to shop, and hence is less sensitive to distance, (c) the consumer derives utility/disutility from driving, depending on the car, characteristics of the road, traffic, etc. • People may be more sensitive to the psychology of traveling to a store (traffic jams, landscaping, width of roads, signs, bill boards, etc.) than they are to the actual physical distance to the store location. • In many instances, the differences between products in each category (and products sold by retail stores in the same trade area) are not substantial enough to warrant inclusion of distance as a factor in store location models. Many retailers purchase products from the same or very similar manufacturers and distributors. Hence car/pedestrian traffic-counts and consumers’ psychological reactions (to distance) are better predictors of demand than actual physical distances from communities, CBDs or other locations (See: [87,82]). 2. Site-specific operating costs The models don’t consider the effect of site-specific operating costs on the location decision (these costs include insurance, maintainance, taxes, etc.); the economics of store operations, the retailer’s cost structure and leverage, the retailer’s cost of capital, and the nature of the retailer’s decision to close a store or to continue operations at the store location. Most models don’t incorporate the customers’ fixed and variable costs of shopping (See: [25]). 3. Models are based on improper probability distributions Some models use, or are based on specific probability distributions that often don’t match real world conditions and events.
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
1727
4. Models don’t consider competing stores Existing store location models typically don’t consider existing competing stores in the same and neighboring Trade Areas, and their potential reactions. The competitors’ sales-per-square foot number by itself, is misleading because some retailers may choose to operate at a loss at certain locations for specific strategic reasons; there may be short term factors in traffic, real estate development and or tenancy at competing shopping centers that may distort sales-per-SF data (See: [11]). 5. The models don’t incorporate the effects of the internet and mail order Most of the models don’t incorporate the effect of the Internet, internet-shopping, direct-mail/catalogues on the site selection decision. Most of the models don’t incorporate the effect of the retailer’s brand power on the site selection decision (See: [11]). 6. The models don’t incorporate the retailer’s method of financing the property The existing store location models don’t incorporate the retailer’s choice among building the physical store (improvements only, or land and improvements), leasing an existing store, and establishing an online store. Some store locations offer only certain types of occupancy/tenancy, and this choice affects the retailer’s cost structure and profits. 7. The models don’t incorporate the retailer’s probability of bankruptcy The models don’t consider the impact of the proposed store on the lessee/retailer’s probability of bankruptcy, and actual/perceived risk profile, and actual/perceived cost of capital. (See: [48,54,87]). 8. Economic value of customer’s time Most store-location models don’t consider the value of the consumer’s time (leisure time, paid time, multi-purpose time) and the impact of consumer’s time on his/her propensity-to-shop at a specific location (within the context of their ages) – in any trade area, the potential impact of consumers’ time is reflected in the distribution of resident’s ages and their incomes. Availability of time substantially influences how and when people shop, and what they buy, and where they are willing to buy (See: [61,35]). 9. In store-location analysis, the concept of ‘communities’ is meaningless In the site-selection context, the concept of ‘communities’ as a unit/basis of analysis is almost meaningless because ‘communities’ form and dissipate at various times of the day, within any given region or Trade Area. People travel to other ‘communities’ for work, leisure and education and other purposes. For analytical purposes, its more meaningful to use larger populations (towns, cities, etc.). For any given Trade Area or community, there are several time-dependent, transportation-dependent, informationdependent and knowledge-dependent ‘‘communities’’: a) ‘‘communities’’ at 7am, just before people leave for work; b) ‘‘communities’’ between 6.30–9.30am when people travel to work; c) ‘‘communities’’ between 9.30am–12noon, when people are at work; d) ‘‘communities’’ between 12noon–2.30pm when people are at lunch; e) ‘‘communities’’ between 3pm and 4pm, when children/teens leave high school and primary school for their homes; f) ‘‘communities’’ between 4.30–7.00pm, when people travel from work to their homes and for recreation; and g) ‘‘communities’’ between 7pm and 6.30am when people are resting at home (See: [34,36,57,54,67,66,74,87]). 10. ‘Multiplicative’ store location models are inaccurate Some store-location models use ‘multiplicative’ methods (such as the MCI model), and are inaccurate because they only capture the intersection of events and factors, don’t consider the economics of site-specific operations, and erroneously assume that distance is a major element in site selection, and that customers must travel the full length of the distance (to the store) each time they go shopping. 11. Data manipulation procedures affect model quality Data manipulation procedures greatly affect the quality of the models – this is especially true with neural networks. The neural network-based models, are only as good as the quality of data inputs (See: [49,52]).
1.2. New store-location models The following is a store location model. The relevant literature on Fuzzy systems include: [84,74,56,33,35,13,16]. The relevant literature on marketing issues include: [38,37,57,54]. Much of the data
1728
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
used can be collected using questionnaires, publicly available information and interviews. The Primary Trade Area is defined as an x-y mile radius around the selected site; and the Secondary Trade Area is defined as an a-b mile radius around the selected site (these demarcations can be changed depending on the product/ service). Let: P = # of individuals aged 18 years or more that earn more than $18,000 annually (or another threshold amount). Oi = monthly operating costs at location i. Oi includes rent, cost of capital and amortized capital expenses (capital expenditures amortized over the estimated useful lives using the firm’s cost of capital). Si = minimum monthly net-sales at location i, required to continue store operations. CFi = minimum monthly net-cash-flow to be earned at store at location i, in order to continue operations at that location. Oi, CFi 2 Si. OVi = corporate overhead allocated to location i. Ri = minimum monthly rent at location i, required to continue operations. Ri, Oi, CFi 2 Si. Tc = traffic count = # of cars that pass the location each day. Tp = # of pedestrians that pass the store each day. Bp = percent of pedestrian traffic that will purchase items from store. qp = average probability of product returns. Bc = percent of car traffic that will purchase items from store. Ep = percent of pedestrian traffic that will enter the store. Ec = percent of car traffic that will enter the store. Sa = average purchase per customer that buys (in dollars). Fc = total square footage of all direct-competitor retail stores in the trade area. Fa = square footage of proposed store at location i. Faverage = average square footage of all competing stores in the trade area. Sfb = average sales per SF for existing competing stores before proposed store is built. Sfa = average sales per SF for existing competing stores after proposed store is built. Sfr = average sales per SF for existing stores in the region/state. Sc = total actual sales for all direct-competitor retail stores in the primary trade area. Se = total estimated sales for all direct-competitor retail stores that will be built in the trade area during the next three years. Sta = total potential sales in trade area = disposable income of all households that earn more than $25,000 in the primary trade area. MU = migrative use factor, which measures the extent to which people aged 16 years or more, in communities in the primary and secondary trade areas migrate out of their own communities and the trade areas, between 6 am and 10 pm, in order to shop, spend leisure time or work in other areas. It also measures the characteristics and buying propensity of the population in the trade area in terms of household income, composition of household, disposable income, and household buying patterns. 0 < MU < 1. MU ! 0, as a higher percentage of percentage of the population in the primary and secondary trade areas aged 16 years or more, (a) work outside the primary trade area, (b) spend leisure time outside the primary trade area, (c) shop outside the primary trade area. TS = time factor, which measures the impact of availability of time (or lack thereof) on shopping habits of people aged 16–65 years in the primary trade area and the secondary trade area. Time also has an important effect on propensity to shop in retail locations. For young people, inadequate time, results in demands for packaged food, less trips to the mall, and other tendencies. For senior citizens, time is less important. Time also affects shopping habits of teens. Hence, the influence of time on shopping can be reflected in the segmentation of certain age groups above sixteen years, and the composition of households (i.e. DINKs will shop less frequently than traditional families with one working parent). 0 < TS < 1. TS ! 0, as (1) a higher percentage of the population in the primary and secondary trade areas aged 16 years or more, have more than X hours
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
1729
available for shopping and leisure each week, and (b) as a higher percentage of households (in the primary and secondary trade areas) are made up of more than two people. X = pricing of products in the store. 0 < X < 1. X ! 0, as the retailer’s prices become less than prices of competing retail stores in the primary trade area and the secondary trade area. D = new development activity within two miles of the site, which will generate additional traffic to the immediate neighborhood. 0 < D < 1. D ! 0, as probability of new development within two miles, increases. Y = percentage of households in which there are women/mothers and children. 0 < Y < 1. Y ! 0, as percentage of households with women/mothers and children increase. V = variety in the retailer’s store. 0 < V < 1. This factor measures the depth of variety in the retailer’s store. Some retailers sell a wider variety of products, and or have ancillary items such as ATM machines, restaurants, pharmacies, etc. Variety is also measured relative to competing store, and with respect to ambience of the stores. V ! 0, as variety in the retailer’s stores increases. A = attractiveness of retailer’s store. 0 < A < 1. This measures relative attractiveness—retailer’s brand, store layout, quality of products, etc. A ! 0, as attractiveness of store increases. M = proximity to major ‘attractors’ that exist or will exist within the next three years (such as the Central Business District, convention centers, universities, government institutions, shopping mall that sells other types of goods, etc.). 0 < M < 1. M ! 0, as proximity to major attractors is greater. T = measures attractiveness of traffic conditions to and from the location—measures traffic congestion, landscaping, presence/absence of offensive sites/images along the route, visibility of site from 500 to 900 m in each direction etc. 0 < T < 1. T does not measure distance. T ! 0, as attractiveness of traffic conditions increase. IA = percentage of population that have internet access at home or office. 0 < IA < 1. IC = the retailer’s incremental costs of building/opening the physical store, compared to its incremental costs of building an Internet retail store (this will include the costs of internet/physical marketing geared towards residents in the secondary trade area to encourage them to shop via the Internet). 0 < IC < 1. IC ! 0, as the retailer’s incremental costs of building the physical store becomes greater than the incremental costs of building internet store. S = suitability of the retailer’s products for sale through the Internet. 0 < S < 1. S ! 0, as the retailer’s products becomes more suitable for sale through the internet. U = community resistance to selection of site. This index measures resistance from municipal authorities and residents to selection of the site to for use as a store location. Such resistance results in additional costs that make the site more expensive and may negatively impact post-occupancy sales—the costs include litigation costs, public hearings, environmental impact studies, community meeting, impact fees, etc. 0 < U < 1. U ! 0, as the costs of resistance increases. H = an index of: (1) # of hours spent in the Internet by individuals aged 18 years or more, and in the Secondary Trade Area, compared to the national average; and (2) percentage of population in the Primary Trade Area that use the Internet to search for products/services to buy. 0 < H < 1. H ! 0, as the # of hours spent in the Internet by individuals aged 18 years or more, and in the Secondary Trade Area, becomes less than the national average; and as fewer people in the Primary Trade Area use the Internet to search for products/services to buy. S = the average amount of total monthly online purchases and average monthly catalogue retail purchases made in the primary trade area and the secondary trade area for the preceding twelve months, compared to the national average. 0 < $ < 1.$ ! 0, as the average amount of total monthly online retail purchases plus the average monthly catalogue retail purchases made in the secondary trade area for the preceding twelve months, becomes less than the national average. I = internet suitability factor which measures the suitability of the retailer’s products for sale on the Internet and other related causal factors. 0 < I < 1. Hence, a grocery chain store will have a lower impact than an electronics store. I ! 0, as it becomes more likely that individuals can purchase the retailer’s products using the Internet (either through the retailer’s website or the websites of other retailers). The Internet affects consumer psychology, consumer preferences, consumer knowledge, and consumer lifestyles, and consumers’ reaction to direct mail and catalogues. Hence, the internet suitability factor measures not only the consumer’s propensity to buy goods/services using the Internet, but also consumers’ preferences, knowledge, education,
1730
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
socialization modes, family structure and psychology, and also the consumers’ propensity to buy through direct-mail/catalogues. Internet retail sales and direct-mail/catalogue retail sales account for at least 6% and 15% of total US retail sales respectively. Because franchisees account for about 40% of total US retail sales, and the type of retailing for which site selection is most critical is provided by non-franchised companies, the internet and direct-mail/catalogues are relevant site-selection factors. Internet suitability factor (I): I ¼ efðIAÞþðICÞ
ðSÞþðH Þþð$ÞþðTSÞg
0 < I < 1:
The natural log scale is used because it reflects (but does not exactly match) the change in scale/magnitude of I, as various factors change. Brand factor (B): B ¼ eðaT þbAþkV þlMþpX þhY þhDþðdTSÞþð1MUÞÞ
0 < B < 1;
a; b; k; l; p; h; 1; d 2 ð0; 1Þ
a, b, k, l, p, 1, d and h are weights attached to T, A, V, M, X, Y, MU, TS and D, solely to reflect their relative importance with respect to each other. The natural log scale is used because it reflects (but does not exactly match) the change in scale/magnitude of I, as various factors change. Factors such ‘‘attractiveness’’, ‘‘variety’’, ‘‘pricing’’ refer to qualities of the retailer and its brand relative to competing retail stores in the trade area. For example, it may be more profitable to locate a store with highattractiveness, high-variety and medium-prices farther away from high-rent and high-labor-cost districts, and to locate low-attractiveness, low-variety and low-price retail stores in a high-traffic site (and still achieve the same profit from both stores). Similarly, its best to locate Walmart stores (high-attractiveness, high-variety, low-price) in low-rent areas outside CBDs, than to locate them in high-rent, high-labor-cost CBDs (See: [54,48,82]). A typical new retail store is not likely to carry any combination of goods that will drastically change traffic patterns and traffic counts (See: [22,24,35,37,56,57,62,67,79])—but the only likely exceptions are: (a) Walmart stores and category-killer stores, because they change the price dynamics and scope of choices, and in such instances, there is a traffic-count-multiplier (TCM) which is greater than the number one, and for all stores that are not super-centers or category killer, is approximately equal to one. (b) Primary trade areas where there are no chain stores, and or where there are relatively few mom-and-pop stores that sell mostly groceries. TCM ¼ X Max½fðoS c =oF c Þ LnðF a =F average Þg; 1 The basic assumptions in calculating the TCM are that: (a) total retail sales in the trade area are proportional to the existing total square footage of stores in the trade area, (b) big-box stores are very likely to attract more traffic because they typically offer lower prices and more variety, (c) LN, the natural log, provides a scale that approximates the rate of change of the impact of size of the proposed store on estimated incremental traffic. At each given potential site, the retailer has at least three alternatives, each of which results in drastically different capital commitments for the retailer: Alternative 1: Purchase land and then build a store—the retailer’s capital commitment will be the sum of the following: Rb = monthly interest for amount borrowed. E = equity invested. Ti = monthly amortization of transaction costs (amortized over time t). P c1 ¼ ðRb þ E þ T i Þ Alternative 2: Purchase land and building—the retailer’s capital commitment will be the sum of the following:
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
1731
Rb = monthly interest for amount borrowed. E = equity invested. Ti = monthly amortization of transaction costs (amortized over time t). P c2 ¼ ðRb þ E þ T i Þ Alternative 3: Have a developer purchase/lease land and build a store, and then lease the property from the developer—the retailer’s capital commitment will be the monthly lease payments over time t plus the amortized transaction costs (where t 2 T), and t is the average number of months that the retailer stays at each site, and T is the total lease term (typically 20–30 years), the sum of which is denoted Ld. Alternative 4: Lease an existing store—the retailer’s capital commitment will be the monthly lease payments over time t, plus the amortized transaction costs (where t 2 T), and t is the average number of months that the retailer stays at each site, and T is the total lease term (typically 20–30 years), the sum of which is denoted Ll. Its assumed that taxes, utilities and insurance costs are the same under both types of property rights. Then, typically: P c1 > P c2 > Ll > Ld Ll Lo Ld > Lo Hence, the component of the site selection decision that concerns property rights, results in an incremental cost = W ¼ fMaxðP c1 ; P c2 ; Ll ; Ld Þ Lo g,where Lo is the ‘normal’ lease payment which is already reflected in Oi, the operating expenses. The retailer’s multi-criteria objective function in site-selection will be Max½MaxfS a X ð1 qp Þ ðT c Bc Ec þ T p Ep Bp Þ 30 efððS c þS e Þ=S ta ÞMU g Ig; 1 Min½MinðfOi þ OVi þ CFi þ Wg; 0Þ (See: [30,23,43]). Hence, the retailer will select the site if it meets the following conditions: 1. [SaX(1 qp)(TcBcEc + TpEpBp) * 30 * efððS c þS e Þ=S ta ÞMUg * I * U] > Max[Si > {Oi + OVi + CFi + W}, 0] 2. [SaX(1 qp)(TcBcEc + TpBpEp) * 30 * {Ln((Sta/(Sc + Se)) * MU)} * I * B * U] > Max[Si > {Oi + OVi + CFi + W}, 0] 3. Max[{Oi + OVi + Ri + W},0] < Si 4. B > 0.5 5. (oSc/oSta) > 1 6. (oSc/oP) > 1 7. (oSta/oFc) < 1 8. (oSfb/oSfr) > 1 9. (oSfb/oSfr) > Max[(oSc/oSta), 1]
2. Conclusion The store-location decision remains critical to the survival of retailing companies—between 2002 and 2005, approximately 80% of all new retail chain stores in the US were single-tenant properties. The US Retailing industry is experiencing substantial and fundamental long-term changes that continue to affect the real estate industry. Most existing store location models are inaccurate and grossly mis-specified. References [1] D. Adams, A. Disberry, N. Hutchinson, T. Munjoma, Retail location, competition and urban re-development, The Service Industries Journal 22 (3) (2005) 135–140.
1732
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
[2] T. Arentze, H. Oppewal, H. Timmermans, A multipurpose shopping trip model to assess retail agglomeration effects, Journal of Marketing Research 42 (1) (2005) 109–115. [3] S. Balasubramian, Mail versus mall: a strategic analysis of competition between direct marketers and conventional retailers, Marketing Science 17 (3) (1998) 181–195. [4] O. Berman, Z. Drezner, G. Wesolowsky, The facility and transfer point location problem, International Transactions in Operations Research 12 (4) (2005) 387–397. [5] J. Brimberg, S. Salhi, A continuous location-allocation problem with zone-dependent fixed cost, Annals of Operations Research 136 (1) (2005) 99–109. [6] R. Betancourt, M. Malanoski, An estimable model of supermarket behavior: prices, distribution services and some effects of competition, Empirica 26 (1999) 55–73. [7] E. Brynjolfsson, M. Smith, Frictionless commerce? A comparison of internet and conventional retailers, Management Science (2000). [8] G. Cachon, M. Larivere, Supply chain coordination with revenue sharing contracts: strengths and limitations, Management Science 51 (1) (2005) 30–44. [9] B. Carlsson, The digital economy: what is new and what is not? Structural Change and Economic Dynamics 15 (3) (2004) 245–264. [10] C. Carter, K. Vandell, Store location in shopping centers: theory and estimates, The Journal of Real Estate Research 27 (3) (2005) 237–265. [11] M. Chang, J. Harrington, Centralization versus decentralization in a multi-unit organization: a computational model of a retail chain as a multi agent adaptive system, Management Science 46 (2000) 1427–1440. [12] C. Chen, T. Garling, R. Kitamura, Activity rescheduling: reasoned or habitual? Transportation Research Part F: Traffic Psychology & Behavior 7 (2004) 351–371. [13] J. Choi, L. Geistfield, A cross cultural investigation of consumer e-shopping adoption, Journal of Economic Psychology 25 (6) (2004) 821–838. [14] C. Chu, L. Huei-Chung, The multi-store location and pricing decisions of a spatial monopoly, Regional Science & Urban Economics 28 (3) (1998) 255–281. [15] R. Claycombe, Cournot retail chains, Journal of Regional Science 38 (3) (1998) 481–493. [16] O. Coban, G. Secme, Prediction of socio-economic consequences of privatization at the firm level with fuzzy cognitive mapping, Information Sciences 169 (2005) 131–154. [17] E. Colla, International expansion and strategies of discount grocery retailers: the winning models, International Journal of Retail and Distribution Management 31 (1) (2003) 55–66. [18] G. Coloma, An economic analysis of horizontal property, International Review of Law and Economics 21 (2001) 343–354. [19] E. Coluccia, G. Louse, Gender differences in spatial orientation: a review, Journal of Environmental Psychology 24 (2004) 329–340. [20] P. Colwell, Coping with technological change: the case of retail, Journal of Real Estate Finance and Economics 26 (1) (2003) 47–63. [21] W. Cook, R. Green, Selecting sites for new facilities using data envelopment analysis, Journal of Productivity Analysis 19 (1) (2003) 77–87. [22] P. Davis, Empirical models of demand for differentiated products, European Economic Review 44 (2000) 993–1005. [23] M. Dellnitz, O. Junge, On the approximation of complicated dynamical behavior, SIAM Journal of Numerical Analysis 36 (2) (1999) 491–515. [24] R. Docters, M. Reopel, J. Sun, S. Tanny, Price is a ‘language to customers’, Journal of Business Strategy (May/June) (2003) 31–35. [25] T. Drezner, Z. Drezner, S. Shiode, A threshold-satisfying competitive location model, Journal of Regional Science 42 (2) (2002) 287– 299. [26] T. Drezner, Z. Drezner, Validating the gravity-based competitive location model using inferred attractiveness, Annals of Operations Research 111 (1) (2002) 227–237. [27] T. Drezner, Z. Drezner, Retail facility location under changing market conditions, IMA Journal of Management Mathematics 13 (4) (2002) 283–293. [28] R. Easterlin, Locational restructuring and financial crisis, Structural Change and Economic Dynamics 11 (1–2) (2000) 129–138. [29] A. Ellaway, S. McIntyre, R. Hiscock, A. Kearns, In the driving seat: psycho-social benefits from private motor vehicle transport compared to public transport, Transportation Research Part F: Traffic Psychology & Behavior 6 (2003) 217–231. [30] J. Ferreira, J. Patricio, F. Oliviera, The computation of solutions of systems of interval polynomial equations, Journal of Computational & Applied Mathematics 173 (2) (2005) 295–302. [31] H. Fock, Retail outlet location decision-maker – franchisor or franchisee? Marketing Intelligence and Planning 19 (3) (2001) 171–177. [32] L. Friess, K. Klamroth, M. Sprau, A wavefront approach to center location problems with barriers, Annals of Operations Research 136 (1) (2005) 35–40. [33] A. Garcia, A. Berlanga, J. Molina, J. Davila, Optimization of airport ground operations integrating genetic and dynamic flow management algorithms, AI Communications 18 (2) (2004) 143–164. [34] Y. Ge, J. Yang, N. Prodlove, M. Spring, Systems dynamics modeling for supply-chain management: a case study on a supermarket chain in the UK, International Transactions in Operational Research 11 (5) (2004) 495–510. [35] K. Gehrt, R. Yan, Situational consumer and retailer factors affecting internet, catalogue and store shopping, International Journal of Retail and Distribution Management 34 (1) (2004) 5–15. [36] V. Gibson, R. Barkham, Corporate real estate management in the retail sector: an investigation of current strategy and structure, Journal of Real Estate Research 22 (1&2) (2001) 107–128. [37] R. Goldsmith, L. Flynn, Bricks, clicks and pix: apparel buyers use of stores, internet and catalogues compared, International Journal of Retail and Distribution Management 33 (4) (2005) 271–284.
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
1733
[38] O. Gonzalez-Benito, J. Gonzalez-Benito, The role of geo-demographic segmentation in retail location strategy, International Journal of Market Research 47 (3) (2005) 295–305. [39] P. Greistorfer, C. Rego, A simple filter-and-fan approach to the facility location problem, Computers & Operations Research 33 (9) (2006) 2590–2597. [40] J. Groeger, Trafficking in cognition applying cognitive psychology to driving, Transportation Research Part F: Traffic Psychology & Behavior 5 (2002) 235–248. [41] C. Howgego, Maximizing competitiveness through the supply chain, International Journal of Retail & Distribution Management 12 (2002) 603–609. [42] C. Huang, C. Chang, Perceived fairness of pricing on the internet, Journal of Economic Psychology 26 (2005) 343–361. [43] S. Iacus, Efficient estimation of dynamical systems, Nonlinear Dynamics and Econometrics 4 (4) (2001) 213–226. [44] N. Khanna, S. Tice, Pricing, exit and location decisions of firms: evidence on the role of debt and operating efficiency, Journal of Financial Economics 75 (2) (2005) 397–407. [45] H. Krishnan, R. Kapuscinski, D. Butz, Coordinating contracts for decentralized supply chains with retailer promotional efforts, Management Science 50 (1) (2001) 48–63. [46] T. Holmes, The location of sales offices and the attraction of cities, The Journal of Political Economy 113 (3) (2005) 551–581. [47] C. Jakobsson, Accuracy of household planning of car use: comparing prospective to actual car logs, Transportation Research Part F: Traffic Psychology & Behavior 7 (2004) 31–42. [48] M. Janssen, V. Karamychev, P. Van Reeven, Multi-store competition: market segmentation or interlacing, Regional Science & Urban Economics 35 (2005) 700–714. [49] C. Kahrahman, D. Ruan, I. Dogan, Fuzzy group decision making for facility location selection, Information Sciences 157 (2003) 135– 153. [50] K. Karande, J. Lombard, Location strategies of broad-line retailers, Journal of Business Research 58 (5) (2005) 687–697. [51] C. Klockner, E. Matthies, How habits interfer with norm directed behavior: a normative decision making model for travel mode choice, Journal of Environmental Psychology 24 (2004) 319–327. [52] R. Kuo, S. Chi, S. Kao, A decision support system for selecting convenience store location through integration of fuzzy AHP and artificial neural network, Computers in Industry 47 (2002) 199–214. [53] G. Kyles, A. Mowen, M. Tarrant, Linking place preferences with place meaning: an examination of the relationship between place motivation and place attachment, Journal of Environmental Psychology 24 (2004) 439–454. [54] L. Lambertini, Equilibrium locations in a spatial model with sequential entry in real time, Regional Science and Urban Economics 32 (1) (2002) 47–58. [55] M. Lee, K. Pace, Spatial distribution of retail sales, Journal of Real Estate Finance and Economics 31 (1) (2005) 53–63. [56] P. Leszcyzc, A. Sinha, A. Saghal, The effect of multipurpose shopping on pricing and location strategy for grocery stores, Journal of Retailing 80 (2) (2004) 85–95. [57] V. Lucas, Cross-border shopping in a federal economy, Regional Science and Urban Economics 34 (4) (2004) 365–385. [58] C. Manning, M. Rodriguez, J. Ghosh, Devising a corporate facility location strategy to maximize shareholder wealth, Journal of Real Estate Research 17 (3) (1999) 321–340. [59] L. Mejia, J. Benjamin, What do we know about the determinants of shopping center sales? spatial versus non-spatial factors, Journal of Real Estate Literature 10 (1) (2002) 3–26. [60] A. Mendes, I.H. Themido, Multi-outlet retail site location assessment, International Transactions in Operations Research 11 (2004) 1–18. [61] P. Messinger, C. Narasimhan, A model of retail formats based on consumers’ economizing on shopping time, Marketing Science 16 (1997) 1–23. [62] S. Michael, Competition in organizational form: mail order versus retail stores, 1910–1940, Journal of Economic Behavior and Organization (1994). [63] K. Minagawa, K. Sumiyoshi, Studies on the optimal location of commerce – basic laws considering the distribution of population, International Journal of Production Economics 60–61 (1999) 295–300. [64] C. Ng, Satisfying shopper’s psychological need: from public market to cyber-mall, Journal of Environemntal Psychology 23 (2003) 439–455. [65] H. Ohta, A phenomenological approach to natural landscape cognition, Journal of Environmental Psychology 21 (2001) 387–403. [66] M. O’Mara, Strategic drivers of location decision for information age companies, Journal of Real Estate Research 17 (3) (1999) 365– 387. [67] L. Pellegrini, Social fabric, and the efficiency of retail systems, International Review of Retail, Distribution & Consumer Research 15 (3) (2005) 328–335. [68] J. Puerto, M. Rodriguez-Chia, New models for locating a moving service facility, Mathematics of Operations Research 63 (1) (2006) 31–38. [69] C.S. Revelle, H. Eiselt, Location analysis: a synthesis and survey, European Journal of Operations Research 165 (1) (2005) 1–7. [70] D. Rogers, Developing a location research methodology, Journal of Targeting, Measurement & Analysis for Marketing 13 (3) (2005) 201–208. [71] T. Rosenbloom, D. Nemrodov, H. Barkan, For heaven’s sake follow the rules: pedestrian’s behavior in an ultra orthodox and a nonorthodox city, Transportation Research Part F: Traffic Psychology & Behavior 7 (2004) 395–404. [72] S. Roulac, Corporate real estate strategy is integral to corporate business strategy, Journal of Real Estate Research 22 (1&2) (2001) 129–152.
1734
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
[73] A. Schobel, Locating stops along bus or railway lines – a bicriteria problem, Annals of Operations Research 136 (1) (2005) 211–227. [74] P. Sinha, A. Banerjee, Store choice behavior in an evolving market, International Journal of Retail & Distribution management 32 (10) (2004) 482–492. [75] A. Siren, L. Hakamies-Blomqist, Private car as the grand equalizer? demographic factors and mobility in finnish men and women aged 65? Transportation Research Part F: Traffic Psychology & Behavior 7 (2004) 107–118. [76] V.P. Sisiopiku, D. Akin, Pedestrian behaviors at and perceptions towards various pedestrian facilities: an examination based on observation and survey data, Transportation Research Part F: Traffic Psychology & Behavior 6 (2003) 249–274. [77] D. Smith, S. Sanchez, Assessment of business potential at retail sites: empirical findings from a US supermarket chain, International Review of Retail, Distribution and Consumer Research 13 (1) (2003) 37–58. [78] L. Snyder, M. Daskin, Reliability models for facility location: the expected failure cost case, Transportation Science 39 (3) (2005) 400– 417. [79] S. Sulaiha, K. Hokao, Estimation of Traffic Generated by New Development: Current Practice and Possible Improvements Based on Bangkok Experience. Working Paper, Victoria Transport Policy Institute, Victoria, BC, Canada (May 2000). [80] T. Suzuki, J. Hodgson, Optimal facility location with multi-purpose trip making, IIT Transactions 37 (5) (2005) 481–486. [81] G. Underwood, P. Chapman, Z. Berger, D. Crundall, Driving experience, attention focusing, and the recall of recently inspected events, Transportation Research Part F: Traffic Psychology & Behavior 6 (2003) 289–304. [82] T. Valletti, Location choice and price discrimination in a duopoly, Regional Science and Urban Economics 32 (3) (2002) 339–358. [83] J. Van Ommeren, A.F. Bumb, A. Sleptchenko, Locating repair shops in a stochastic environment, Computers & Operations Research 33 (6) (2006) 1575–1594. [84] Y. Wang, C. Parkan, Multi-attribute decision-making based on fuzzy preference information on alternatives: ranking and weighting, Fuzzy Sets and Systems 153 (2005) 331–346. [85] J. Winer, A. Schnee, H. Malllot, Use an interaction of navigation strategies in regionalized environments, Journal of Environmental Psychology 24 (2004) 475–493. [86] J. Zhang, B. Chen, Y. Ye, A multiexchange local search algorithm for the capacitated facility location problem, Mathematics of Operations Research 30 (2) (2005) 389–400. [87] D. Zhou, I. Vertinsky, Strategic location decisions in a growing market, Regional Science and Urban Economics 31 (5) (2001) 523– 641.
Further reading [1] O. Berman, D. Krass, The generalized maximal covering location problem, Computers & Operations Research 29 (6) (2002) 563–581. [2] A. Dasci, G. LaPorte, A continuous model for multistore competitive location, Operations Research 53 (2) (2005) 263–280. [3] A. Gupta, B. Su, Z. Walter, An empirical study of consumer switching from traditional to electronic channels: a purchase decision process perspective, International Journal of Electronic Commerce 8 (3) (2004) 131–141. [4] C. Hallam, G. Pindar, Prediction of land use traffic impact, Transportation Research Record 940 (1983) 51–60. [5] T. Helstrup, S. Magnussen, The mental representation of familiar, long distance journeys, Journal of Environmental Psychology 21 (2001) 411–421. [6] T. Hernandez, D. Bennison, The art and science of retail location decisions, International Journal of Retail & Distribution Management 28 (8) (2000) 357–367. [7] M.C. Hidalgo, B. Hernandez, Place attachment: conceptual and empirical questions, Journal of Environmental Psychology 21 (2001) 273–281. [8] C. Hsu, Y. Hsieh, Travel and activity choices based on an individual accessibility model, Papers in Regional Science 83 (2004) 387– 406. [9] Y. Kato, Y. Takeuchi, Individual differences in way-finding strategies, Journal of Environmental Psychology 23 (2003) 171–188. [10] P. Kaufmann, F. Lafontaine, Costs of control: the sources of economic rents for Mcdonalds franchisees, Journal of Law & Economics 37 (1994) 417–453. [11] C. Keen, M. Wetzels, K. Ruyter, et al., ETailers versus retailers: which factors determine consumer preferences 2004.
Site selection in the US retailing industry Michael Nwogugu P.O. Box 170002, Brooklyn, NY 11217, USA
Abstract This article critiques existing store-location models and explains the many mathematical and methodological problems inherent in these models. The key issues are that behavioral factors are completely omitted in these models, emphasis on distance in un-warranted and site-specific and retailer-specific characteristics are typically omitted in these model. The author introduces new site-selection models. The US retailing sector is used an example. 2006 Published by Elsevier Inc. Keywords: Retail stores; Facilities location; Complexity; Corporate strategy; Risk management; Dynamical systems; Supply chain
1. Introduction The store location decision has become perhaps the most important decision for many retailers—Internet retail sales and mail order retail sales account for less than 8% and 30% of total US retail sales respectively. Store location sometimes contributes as much to a retailer’s brand value as some forms of marketing. Real estate accounts for more than 30% of retailers’ operating expenses and total assets. Retailing remains the largest industry in the US – in 2005, at least 30% of US GDP was from retail sales. In 2004/2005, the US retail industry had more than 1.5 million establishments, and employed more than 20 million people (17% of the US workforce). (Standard & Poors). 1.1. The store location (site-selection) decision Given the structural changes that occurred in the US retailing sector during the last five years, and the fact that real estate constitutes a substantial portion of retailer’s assets and operating expenses, the store location decision has become probably the most important strategy decision for retailers. The literature on store site-selection is extensive, multi-disciplinary and covers various approaches in operations-research/management science, economics and marketing (See: [25,60,39,31,32,38,46,44,11,8,17,20,18,3,6,28,4,5,55,68,69,78,72,67,73,77,63,58, 52,50,59,62,34,33,41,54,57,63,66,9,21,1,10,2,14,4,15,7,6,22,24,36,42,45,27,80,86,83,70,84]). However, all existing
E-mail addresses: [email protected], [email protected] 0096-3003/$ - see front matter 2006 Published by Elsevier Inc. doi:10.1016/j.amc.2005.12.050
1726
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
store-location models are inaccurate and some of the problems problems inherent in the existing literature are explained as follows. 1. Distance Existing store-location models over-emphasize and don’t incorporate the distance element properly. Most models erroneously assume: – That distance from each community (in the trade area) to the store is constant, and remains constant for each customer, for all trips to the store, and for all time periods. – That all residents of each community travel the full distance between their community and the store location whenever they shop. Some models have erroneously tried to use Newton’s gravity model (laws of physics) for location analysis [25,26]. However, the Newton’s law is not applicable because its assumptions do not fit the reality of shopping, consumers’ travel patterns, consumer’s reaction to distances and retail activities. • That distance is not affected by any other factors (traffic jams, psychology of driving, environmental psychology, perceived value of time spent traveling to/from store, perceived value or trip (multipurpose trips vs. single-purpose trips), etc.) and is completely determined by a physical measure. Prior site selection studies and models omitted relevant transportation psychology and environmental psychology analysis. (See: [53,51,19,47,40,29,12,71,81,85,75,76,64,65,6]). These site selection models erroneously state and assume that distance is a major element of site selection models. However, in reality, the distance between the store location and various communities does not matter (See: [52]), because: • Most people do not work in their immediate communities. • Most people shop before going to work, during lunch, after work, or on weekends. • Most people go to the shopping mall on their way to other destinations; and prefer malls that are on the route to destinations that they regularly go to. • People may have psychological attachments to specific retail brands. Kuo et al. [52] presented evidence (questionnaire results) that customers will walk/drive more than 10–30 min past certain retail stores to go shop at other retail stores. • Many people in most communities drive to store locations. When driving, actual distance (between the community and the store) becomes less relevant, partly because (a) the consumer has reserved time for shopping which includes the travel, and thus is less sensitive to distance, (b) the consumer is willing to travel to various locations to shop, and hence is less sensitive to distance, (c) the consumer derives utility/disutility from driving, depending on the car, characteristics of the road, traffic, etc. • People may be more sensitive to the psychology of traveling to a store (traffic jams, landscaping, width of roads, signs, bill boards, etc.) than they are to the actual physical distance to the store location. • In many instances, the differences between products in each category (and products sold by retail stores in the same trade area) are not substantial enough to warrant inclusion of distance as a factor in store location models. Many retailers purchase products from the same or very similar manufacturers and distributors. Hence car/pedestrian traffic-counts and consumers’ psychological reactions (to distance) are better predictors of demand than actual physical distances from communities, CBDs or other locations (See: [87,82]). 2. Site-specific operating costs The models don’t consider the effect of site-specific operating costs on the location decision (these costs include insurance, maintainance, taxes, etc.); the economics of store operations, the retailer’s cost structure and leverage, the retailer’s cost of capital, and the nature of the retailer’s decision to close a store or to continue operations at the store location. Most models don’t incorporate the customers’ fixed and variable costs of shopping (See: [25]). 3. Models are based on improper probability distributions Some models use, or are based on specific probability distributions that often don’t match real world conditions and events.
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
1727
4. Models don’t consider competing stores Existing store location models typically don’t consider existing competing stores in the same and neighboring Trade Areas, and their potential reactions. The competitors’ sales-per-square foot number by itself, is misleading because some retailers may choose to operate at a loss at certain locations for specific strategic reasons; there may be short term factors in traffic, real estate development and or tenancy at competing shopping centers that may distort sales-per-SF data (See: [11]). 5. The models don’t incorporate the effects of the internet and mail order Most of the models don’t incorporate the effect of the Internet, internet-shopping, direct-mail/catalogues on the site selection decision. Most of the models don’t incorporate the effect of the retailer’s brand power on the site selection decision (See: [11]). 6. The models don’t incorporate the retailer’s method of financing the property The existing store location models don’t incorporate the retailer’s choice among building the physical store (improvements only, or land and improvements), leasing an existing store, and establishing an online store. Some store locations offer only certain types of occupancy/tenancy, and this choice affects the retailer’s cost structure and profits. 7. The models don’t incorporate the retailer’s probability of bankruptcy The models don’t consider the impact of the proposed store on the lessee/retailer’s probability of bankruptcy, and actual/perceived risk profile, and actual/perceived cost of capital. (See: [48,54,87]). 8. Economic value of customer’s time Most store-location models don’t consider the value of the consumer’s time (leisure time, paid time, multi-purpose time) and the impact of consumer’s time on his/her propensity-to-shop at a specific location (within the context of their ages) – in any trade area, the potential impact of consumers’ time is reflected in the distribution of resident’s ages and their incomes. Availability of time substantially influences how and when people shop, and what they buy, and where they are willing to buy (See: [61,35]). 9. In store-location analysis, the concept of ‘communities’ is meaningless In the site-selection context, the concept of ‘communities’ as a unit/basis of analysis is almost meaningless because ‘communities’ form and dissipate at various times of the day, within any given region or Trade Area. People travel to other ‘communities’ for work, leisure and education and other purposes. For analytical purposes, its more meaningful to use larger populations (towns, cities, etc.). For any given Trade Area or community, there are several time-dependent, transportation-dependent, informationdependent and knowledge-dependent ‘‘communities’’: a) ‘‘communities’’ at 7am, just before people leave for work; b) ‘‘communities’’ between 6.30–9.30am when people travel to work; c) ‘‘communities’’ between 9.30am–12noon, when people are at work; d) ‘‘communities’’ between 12noon–2.30pm when people are at lunch; e) ‘‘communities’’ between 3pm and 4pm, when children/teens leave high school and primary school for their homes; f) ‘‘communities’’ between 4.30–7.00pm, when people travel from work to their homes and for recreation; and g) ‘‘communities’’ between 7pm and 6.30am when people are resting at home (See: [34,36,57,54,67,66,74,87]). 10. ‘Multiplicative’ store location models are inaccurate Some store-location models use ‘multiplicative’ methods (such as the MCI model), and are inaccurate because they only capture the intersection of events and factors, don’t consider the economics of site-specific operations, and erroneously assume that distance is a major element in site selection, and that customers must travel the full length of the distance (to the store) each time they go shopping. 11. Data manipulation procedures affect model quality Data manipulation procedures greatly affect the quality of the models – this is especially true with neural networks. The neural network-based models, are only as good as the quality of data inputs (See: [49,52]).
1.2. New store-location models The following is a store location model. The relevant literature on Fuzzy systems include: [84,74,56,33,35,13,16]. The relevant literature on marketing issues include: [38,37,57,54]. Much of the data
1728
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
used can be collected using questionnaires, publicly available information and interviews. The Primary Trade Area is defined as an x-y mile radius around the selected site; and the Secondary Trade Area is defined as an a-b mile radius around the selected site (these demarcations can be changed depending on the product/ service). Let: P = # of individuals aged 18 years or more that earn more than $18,000 annually (or another threshold amount). Oi = monthly operating costs at location i. Oi includes rent, cost of capital and amortized capital expenses (capital expenditures amortized over the estimated useful lives using the firm’s cost of capital). Si = minimum monthly net-sales at location i, required to continue store operations. CFi = minimum monthly net-cash-flow to be earned at store at location i, in order to continue operations at that location. Oi, CFi 2 Si. OVi = corporate overhead allocated to location i. Ri = minimum monthly rent at location i, required to continue operations. Ri, Oi, CFi 2 Si. Tc = traffic count = # of cars that pass the location each day. Tp = # of pedestrians that pass the store each day. Bp = percent of pedestrian traffic that will purchase items from store. qp = average probability of product returns. Bc = percent of car traffic that will purchase items from store. Ep = percent of pedestrian traffic that will enter the store. Ec = percent of car traffic that will enter the store. Sa = average purchase per customer that buys (in dollars). Fc = total square footage of all direct-competitor retail stores in the trade area. Fa = square footage of proposed store at location i. Faverage = average square footage of all competing stores in the trade area. Sfb = average sales per SF for existing competing stores before proposed store is built. Sfa = average sales per SF for existing competing stores after proposed store is built. Sfr = average sales per SF for existing stores in the region/state. Sc = total actual sales for all direct-competitor retail stores in the primary trade area. Se = total estimated sales for all direct-competitor retail stores that will be built in the trade area during the next three years. Sta = total potential sales in trade area = disposable income of all households that earn more than $25,000 in the primary trade area. MU = migrative use factor, which measures the extent to which people aged 16 years or more, in communities in the primary and secondary trade areas migrate out of their own communities and the trade areas, between 6 am and 10 pm, in order to shop, spend leisure time or work in other areas. It also measures the characteristics and buying propensity of the population in the trade area in terms of household income, composition of household, disposable income, and household buying patterns. 0 < MU < 1. MU ! 0, as a higher percentage of percentage of the population in the primary and secondary trade areas aged 16 years or more, (a) work outside the primary trade area, (b) spend leisure time outside the primary trade area, (c) shop outside the primary trade area. TS = time factor, which measures the impact of availability of time (or lack thereof) on shopping habits of people aged 16–65 years in the primary trade area and the secondary trade area. Time also has an important effect on propensity to shop in retail locations. For young people, inadequate time, results in demands for packaged food, less trips to the mall, and other tendencies. For senior citizens, time is less important. Time also affects shopping habits of teens. Hence, the influence of time on shopping can be reflected in the segmentation of certain age groups above sixteen years, and the composition of households (i.e. DINKs will shop less frequently than traditional families with one working parent). 0 < TS < 1. TS ! 0, as (1) a higher percentage of the population in the primary and secondary trade areas aged 16 years or more, have more than X hours
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
1729
available for shopping and leisure each week, and (b) as a higher percentage of households (in the primary and secondary trade areas) are made up of more than two people. X = pricing of products in the store. 0 < X < 1. X ! 0, as the retailer’s prices become less than prices of competing retail stores in the primary trade area and the secondary trade area. D = new development activity within two miles of the site, which will generate additional traffic to the immediate neighborhood. 0 < D < 1. D ! 0, as probability of new development within two miles, increases. Y = percentage of households in which there are women/mothers and children. 0 < Y < 1. Y ! 0, as percentage of households with women/mothers and children increase. V = variety in the retailer’s store. 0 < V < 1. This factor measures the depth of variety in the retailer’s store. Some retailers sell a wider variety of products, and or have ancillary items such as ATM machines, restaurants, pharmacies, etc. Variety is also measured relative to competing store, and with respect to ambience of the stores. V ! 0, as variety in the retailer’s stores increases. A = attractiveness of retailer’s store. 0 < A < 1. This measures relative attractiveness—retailer’s brand, store layout, quality of products, etc. A ! 0, as attractiveness of store increases. M = proximity to major ‘attractors’ that exist or will exist within the next three years (such as the Central Business District, convention centers, universities, government institutions, shopping mall that sells other types of goods, etc.). 0 < M < 1. M ! 0, as proximity to major attractors is greater. T = measures attractiveness of traffic conditions to and from the location—measures traffic congestion, landscaping, presence/absence of offensive sites/images along the route, visibility of site from 500 to 900 m in each direction etc. 0 < T < 1. T does not measure distance. T ! 0, as attractiveness of traffic conditions increase. IA = percentage of population that have internet access at home or office. 0 < IA < 1. IC = the retailer’s incremental costs of building/opening the physical store, compared to its incremental costs of building an Internet retail store (this will include the costs of internet/physical marketing geared towards residents in the secondary trade area to encourage them to shop via the Internet). 0 < IC < 1. IC ! 0, as the retailer’s incremental costs of building the physical store becomes greater than the incremental costs of building internet store. S = suitability of the retailer’s products for sale through the Internet. 0 < S < 1. S ! 0, as the retailer’s products becomes more suitable for sale through the internet. U = community resistance to selection of site. This index measures resistance from municipal authorities and residents to selection of the site to for use as a store location. Such resistance results in additional costs that make the site more expensive and may negatively impact post-occupancy sales—the costs include litigation costs, public hearings, environmental impact studies, community meeting, impact fees, etc. 0 < U < 1. U ! 0, as the costs of resistance increases. H = an index of: (1) # of hours spent in the Internet by individuals aged 18 years or more, and in the Secondary Trade Area, compared to the national average; and (2) percentage of population in the Primary Trade Area that use the Internet to search for products/services to buy. 0 < H < 1. H ! 0, as the # of hours spent in the Internet by individuals aged 18 years or more, and in the Secondary Trade Area, becomes less than the national average; and as fewer people in the Primary Trade Area use the Internet to search for products/services to buy. S = the average amount of total monthly online purchases and average monthly catalogue retail purchases made in the primary trade area and the secondary trade area for the preceding twelve months, compared to the national average. 0 < $ < 1.$ ! 0, as the average amount of total monthly online retail purchases plus the average monthly catalogue retail purchases made in the secondary trade area for the preceding twelve months, becomes less than the national average. I = internet suitability factor which measures the suitability of the retailer’s products for sale on the Internet and other related causal factors. 0 < I < 1. Hence, a grocery chain store will have a lower impact than an electronics store. I ! 0, as it becomes more likely that individuals can purchase the retailer’s products using the Internet (either through the retailer’s website or the websites of other retailers). The Internet affects consumer psychology, consumer preferences, consumer knowledge, and consumer lifestyles, and consumers’ reaction to direct mail and catalogues. Hence, the internet suitability factor measures not only the consumer’s propensity to buy goods/services using the Internet, but also consumers’ preferences, knowledge, education,
1730
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
socialization modes, family structure and psychology, and also the consumers’ propensity to buy through direct-mail/catalogues. Internet retail sales and direct-mail/catalogue retail sales account for at least 6% and 15% of total US retail sales respectively. Because franchisees account for about 40% of total US retail sales, and the type of retailing for which site selection is most critical is provided by non-franchised companies, the internet and direct-mail/catalogues are relevant site-selection factors. Internet suitability factor (I): I ¼ efðIAÞþðICÞ
ðSÞþðH Þþð$ÞþðTSÞg
0 < I < 1:
The natural log scale is used because it reflects (but does not exactly match) the change in scale/magnitude of I, as various factors change. Brand factor (B): B ¼ eðaT þbAþkV þlMþpX þhY þhDþðdTSÞþð1MUÞÞ
0 < B < 1;
a; b; k; l; p; h; 1; d 2 ð0; 1Þ
a, b, k, l, p, 1, d and h are weights attached to T, A, V, M, X, Y, MU, TS and D, solely to reflect their relative importance with respect to each other. The natural log scale is used because it reflects (but does not exactly match) the change in scale/magnitude of I, as various factors change. Factors such ‘‘attractiveness’’, ‘‘variety’’, ‘‘pricing’’ refer to qualities of the retailer and its brand relative to competing retail stores in the trade area. For example, it may be more profitable to locate a store with highattractiveness, high-variety and medium-prices farther away from high-rent and high-labor-cost districts, and to locate low-attractiveness, low-variety and low-price retail stores in a high-traffic site (and still achieve the same profit from both stores). Similarly, its best to locate Walmart stores (high-attractiveness, high-variety, low-price) in low-rent areas outside CBDs, than to locate them in high-rent, high-labor-cost CBDs (See: [54,48,82]). A typical new retail store is not likely to carry any combination of goods that will drastically change traffic patterns and traffic counts (See: [22,24,35,37,56,57,62,67,79])—but the only likely exceptions are: (a) Walmart stores and category-killer stores, because they change the price dynamics and scope of choices, and in such instances, there is a traffic-count-multiplier (TCM) which is greater than the number one, and for all stores that are not super-centers or category killer, is approximately equal to one. (b) Primary trade areas where there are no chain stores, and or where there are relatively few mom-and-pop stores that sell mostly groceries. TCM ¼ X Max½fðoS c =oF c Þ LnðF a =F average Þg; 1 The basic assumptions in calculating the TCM are that: (a) total retail sales in the trade area are proportional to the existing total square footage of stores in the trade area, (b) big-box stores are very likely to attract more traffic because they typically offer lower prices and more variety, (c) LN, the natural log, provides a scale that approximates the rate of change of the impact of size of the proposed store on estimated incremental traffic. At each given potential site, the retailer has at least three alternatives, each of which results in drastically different capital commitments for the retailer: Alternative 1: Purchase land and then build a store—the retailer’s capital commitment will be the sum of the following: Rb = monthly interest for amount borrowed. E = equity invested. Ti = monthly amortization of transaction costs (amortized over time t). P c1 ¼ ðRb þ E þ T i Þ Alternative 2: Purchase land and building—the retailer’s capital commitment will be the sum of the following:
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
1731
Rb = monthly interest for amount borrowed. E = equity invested. Ti = monthly amortization of transaction costs (amortized over time t). P c2 ¼ ðRb þ E þ T i Þ Alternative 3: Have a developer purchase/lease land and build a store, and then lease the property from the developer—the retailer’s capital commitment will be the monthly lease payments over time t plus the amortized transaction costs (where t 2 T), and t is the average number of months that the retailer stays at each site, and T is the total lease term (typically 20–30 years), the sum of which is denoted Ld. Alternative 4: Lease an existing store—the retailer’s capital commitment will be the monthly lease payments over time t, plus the amortized transaction costs (where t 2 T), and t is the average number of months that the retailer stays at each site, and T is the total lease term (typically 20–30 years), the sum of which is denoted Ll. Its assumed that taxes, utilities and insurance costs are the same under both types of property rights. Then, typically: P c1 > P c2 > Ll > Ld Ll Lo Ld > Lo Hence, the component of the site selection decision that concerns property rights, results in an incremental cost = W ¼ fMaxðP c1 ; P c2 ; Ll ; Ld Þ Lo g,where Lo is the ‘normal’ lease payment which is already reflected in Oi, the operating expenses. The retailer’s multi-criteria objective function in site-selection will be Max½MaxfS a X ð1 qp Þ ðT c Bc Ec þ T p Ep Bp Þ 30 efððS c þS e Þ=S ta ÞMU g Ig; 1 Min½MinðfOi þ OVi þ CFi þ Wg; 0Þ (See: [30,23,43]). Hence, the retailer will select the site if it meets the following conditions: 1. [SaX(1 qp)(TcBcEc + TpEpBp) * 30 * efððS c þS e Þ=S ta ÞMUg * I * U] > Max[Si > {Oi + OVi + CFi + W}, 0] 2. [SaX(1 qp)(TcBcEc + TpBpEp) * 30 * {Ln((Sta/(Sc + Se)) * MU)} * I * B * U] > Max[Si > {Oi + OVi + CFi + W}, 0] 3. Max[{Oi + OVi + Ri + W},0] < Si 4. B > 0.5 5. (oSc/oSta) > 1 6. (oSc/oP) > 1 7. (oSta/oFc) < 1 8. (oSfb/oSfr) > 1 9. (oSfb/oSfr) > Max[(oSc/oSta), 1]
2. Conclusion The store-location decision remains critical to the survival of retailing companies—between 2002 and 2005, approximately 80% of all new retail chain stores in the US were single-tenant properties. The US Retailing industry is experiencing substantial and fundamental long-term changes that continue to affect the real estate industry. Most existing store location models are inaccurate and grossly mis-specified. References [1] D. Adams, A. Disberry, N. Hutchinson, T. Munjoma, Retail location, competition and urban re-development, The Service Industries Journal 22 (3) (2005) 135–140.
1732
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
[2] T. Arentze, H. Oppewal, H. Timmermans, A multipurpose shopping trip model to assess retail agglomeration effects, Journal of Marketing Research 42 (1) (2005) 109–115. [3] S. Balasubramian, Mail versus mall: a strategic analysis of competition between direct marketers and conventional retailers, Marketing Science 17 (3) (1998) 181–195. [4] O. Berman, Z. Drezner, G. Wesolowsky, The facility and transfer point location problem, International Transactions in Operations Research 12 (4) (2005) 387–397. [5] J. Brimberg, S. Salhi, A continuous location-allocation problem with zone-dependent fixed cost, Annals of Operations Research 136 (1) (2005) 99–109. [6] R. Betancourt, M. Malanoski, An estimable model of supermarket behavior: prices, distribution services and some effects of competition, Empirica 26 (1999) 55–73. [7] E. Brynjolfsson, M. Smith, Frictionless commerce? A comparison of internet and conventional retailers, Management Science (2000). [8] G. Cachon, M. Larivere, Supply chain coordination with revenue sharing contracts: strengths and limitations, Management Science 51 (1) (2005) 30–44. [9] B. Carlsson, The digital economy: what is new and what is not? Structural Change and Economic Dynamics 15 (3) (2004) 245–264. [10] C. Carter, K. Vandell, Store location in shopping centers: theory and estimates, The Journal of Real Estate Research 27 (3) (2005) 237–265. [11] M. Chang, J. Harrington, Centralization versus decentralization in a multi-unit organization: a computational model of a retail chain as a multi agent adaptive system, Management Science 46 (2000) 1427–1440. [12] C. Chen, T. Garling, R. Kitamura, Activity rescheduling: reasoned or habitual? Transportation Research Part F: Traffic Psychology & Behavior 7 (2004) 351–371. [13] J. Choi, L. Geistfield, A cross cultural investigation of consumer e-shopping adoption, Journal of Economic Psychology 25 (6) (2004) 821–838. [14] C. Chu, L. Huei-Chung, The multi-store location and pricing decisions of a spatial monopoly, Regional Science & Urban Economics 28 (3) (1998) 255–281. [15] R. Claycombe, Cournot retail chains, Journal of Regional Science 38 (3) (1998) 481–493. [16] O. Coban, G. Secme, Prediction of socio-economic consequences of privatization at the firm level with fuzzy cognitive mapping, Information Sciences 169 (2005) 131–154. [17] E. Colla, International expansion and strategies of discount grocery retailers: the winning models, International Journal of Retail and Distribution Management 31 (1) (2003) 55–66. [18] G. Coloma, An economic analysis of horizontal property, International Review of Law and Economics 21 (2001) 343–354. [19] E. Coluccia, G. Louse, Gender differences in spatial orientation: a review, Journal of Environmental Psychology 24 (2004) 329–340. [20] P. Colwell, Coping with technological change: the case of retail, Journal of Real Estate Finance and Economics 26 (1) (2003) 47–63. [21] W. Cook, R. Green, Selecting sites for new facilities using data envelopment analysis, Journal of Productivity Analysis 19 (1) (2003) 77–87. [22] P. Davis, Empirical models of demand for differentiated products, European Economic Review 44 (2000) 993–1005. [23] M. Dellnitz, O. Junge, On the approximation of complicated dynamical behavior, SIAM Journal of Numerical Analysis 36 (2) (1999) 491–515. [24] R. Docters, M. Reopel, J. Sun, S. Tanny, Price is a ‘language to customers’, Journal of Business Strategy (May/June) (2003) 31–35. [25] T. Drezner, Z. Drezner, S. Shiode, A threshold-satisfying competitive location model, Journal of Regional Science 42 (2) (2002) 287– 299. [26] T. Drezner, Z. Drezner, Validating the gravity-based competitive location model using inferred attractiveness, Annals of Operations Research 111 (1) (2002) 227–237. [27] T. Drezner, Z. Drezner, Retail facility location under changing market conditions, IMA Journal of Management Mathematics 13 (4) (2002) 283–293. [28] R. Easterlin, Locational restructuring and financial crisis, Structural Change and Economic Dynamics 11 (1–2) (2000) 129–138. [29] A. Ellaway, S. McIntyre, R. Hiscock, A. Kearns, In the driving seat: psycho-social benefits from private motor vehicle transport compared to public transport, Transportation Research Part F: Traffic Psychology & Behavior 6 (2003) 217–231. [30] J. Ferreira, J. Patricio, F. Oliviera, The computation of solutions of systems of interval polynomial equations, Journal of Computational & Applied Mathematics 173 (2) (2005) 295–302. [31] H. Fock, Retail outlet location decision-maker – franchisor or franchisee? Marketing Intelligence and Planning 19 (3) (2001) 171–177. [32] L. Friess, K. Klamroth, M. Sprau, A wavefront approach to center location problems with barriers, Annals of Operations Research 136 (1) (2005) 35–40. [33] A. Garcia, A. Berlanga, J. Molina, J. Davila, Optimization of airport ground operations integrating genetic and dynamic flow management algorithms, AI Communications 18 (2) (2004) 143–164. [34] Y. Ge, J. Yang, N. Prodlove, M. Spring, Systems dynamics modeling for supply-chain management: a case study on a supermarket chain in the UK, International Transactions in Operational Research 11 (5) (2004) 495–510. [35] K. Gehrt, R. Yan, Situational consumer and retailer factors affecting internet, catalogue and store shopping, International Journal of Retail and Distribution Management 34 (1) (2004) 5–15. [36] V. Gibson, R. Barkham, Corporate real estate management in the retail sector: an investigation of current strategy and structure, Journal of Real Estate Research 22 (1&2) (2001) 107–128. [37] R. Goldsmith, L. Flynn, Bricks, clicks and pix: apparel buyers use of stores, internet and catalogues compared, International Journal of Retail and Distribution Management 33 (4) (2005) 271–284.
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
1733
[38] O. Gonzalez-Benito, J. Gonzalez-Benito, The role of geo-demographic segmentation in retail location strategy, International Journal of Market Research 47 (3) (2005) 295–305. [39] P. Greistorfer, C. Rego, A simple filter-and-fan approach to the facility location problem, Computers & Operations Research 33 (9) (2006) 2590–2597. [40] J. Groeger, Trafficking in cognition applying cognitive psychology to driving, Transportation Research Part F: Traffic Psychology & Behavior 5 (2002) 235–248. [41] C. Howgego, Maximizing competitiveness through the supply chain, International Journal of Retail & Distribution Management 12 (2002) 603–609. [42] C. Huang, C. Chang, Perceived fairness of pricing on the internet, Journal of Economic Psychology 26 (2005) 343–361. [43] S. Iacus, Efficient estimation of dynamical systems, Nonlinear Dynamics and Econometrics 4 (4) (2001) 213–226. [44] N. Khanna, S. Tice, Pricing, exit and location decisions of firms: evidence on the role of debt and operating efficiency, Journal of Financial Economics 75 (2) (2005) 397–407. [45] H. Krishnan, R. Kapuscinski, D. Butz, Coordinating contracts for decentralized supply chains with retailer promotional efforts, Management Science 50 (1) (2001) 48–63. [46] T. Holmes, The location of sales offices and the attraction of cities, The Journal of Political Economy 113 (3) (2005) 551–581. [47] C. Jakobsson, Accuracy of household planning of car use: comparing prospective to actual car logs, Transportation Research Part F: Traffic Psychology & Behavior 7 (2004) 31–42. [48] M. Janssen, V. Karamychev, P. Van Reeven, Multi-store competition: market segmentation or interlacing, Regional Science & Urban Economics 35 (2005) 700–714. [49] C. Kahrahman, D. Ruan, I. Dogan, Fuzzy group decision making for facility location selection, Information Sciences 157 (2003) 135– 153. [50] K. Karande, J. Lombard, Location strategies of broad-line retailers, Journal of Business Research 58 (5) (2005) 687–697. [51] C. Klockner, E. Matthies, How habits interfer with norm directed behavior: a normative decision making model for travel mode choice, Journal of Environmental Psychology 24 (2004) 319–327. [52] R. Kuo, S. Chi, S. Kao, A decision support system for selecting convenience store location through integration of fuzzy AHP and artificial neural network, Computers in Industry 47 (2002) 199–214. [53] G. Kyles, A. Mowen, M. Tarrant, Linking place preferences with place meaning: an examination of the relationship between place motivation and place attachment, Journal of Environmental Psychology 24 (2004) 439–454. [54] L. Lambertini, Equilibrium locations in a spatial model with sequential entry in real time, Regional Science and Urban Economics 32 (1) (2002) 47–58. [55] M. Lee, K. Pace, Spatial distribution of retail sales, Journal of Real Estate Finance and Economics 31 (1) (2005) 53–63. [56] P. Leszcyzc, A. Sinha, A. Saghal, The effect of multipurpose shopping on pricing and location strategy for grocery stores, Journal of Retailing 80 (2) (2004) 85–95. [57] V. Lucas, Cross-border shopping in a federal economy, Regional Science and Urban Economics 34 (4) (2004) 365–385. [58] C. Manning, M. Rodriguez, J. Ghosh, Devising a corporate facility location strategy to maximize shareholder wealth, Journal of Real Estate Research 17 (3) (1999) 321–340. [59] L. Mejia, J. Benjamin, What do we know about the determinants of shopping center sales? spatial versus non-spatial factors, Journal of Real Estate Literature 10 (1) (2002) 3–26. [60] A. Mendes, I.H. Themido, Multi-outlet retail site location assessment, International Transactions in Operations Research 11 (2004) 1–18. [61] P. Messinger, C. Narasimhan, A model of retail formats based on consumers’ economizing on shopping time, Marketing Science 16 (1997) 1–23. [62] S. Michael, Competition in organizational form: mail order versus retail stores, 1910–1940, Journal of Economic Behavior and Organization (1994). [63] K. Minagawa, K. Sumiyoshi, Studies on the optimal location of commerce – basic laws considering the distribution of population, International Journal of Production Economics 60–61 (1999) 295–300. [64] C. Ng, Satisfying shopper’s psychological need: from public market to cyber-mall, Journal of Environemntal Psychology 23 (2003) 439–455. [65] H. Ohta, A phenomenological approach to natural landscape cognition, Journal of Environmental Psychology 21 (2001) 387–403. [66] M. O’Mara, Strategic drivers of location decision for information age companies, Journal of Real Estate Research 17 (3) (1999) 365– 387. [67] L. Pellegrini, Social fabric, and the efficiency of retail systems, International Review of Retail, Distribution & Consumer Research 15 (3) (2005) 328–335. [68] J. Puerto, M. Rodriguez-Chia, New models for locating a moving service facility, Mathematics of Operations Research 63 (1) (2006) 31–38. [69] C.S. Revelle, H. Eiselt, Location analysis: a synthesis and survey, European Journal of Operations Research 165 (1) (2005) 1–7. [70] D. Rogers, Developing a location research methodology, Journal of Targeting, Measurement & Analysis for Marketing 13 (3) (2005) 201–208. [71] T. Rosenbloom, D. Nemrodov, H. Barkan, For heaven’s sake follow the rules: pedestrian’s behavior in an ultra orthodox and a nonorthodox city, Transportation Research Part F: Traffic Psychology & Behavior 7 (2004) 395–404. [72] S. Roulac, Corporate real estate strategy is integral to corporate business strategy, Journal of Real Estate Research 22 (1&2) (2001) 129–152.
1734
M. Nwogugu / Applied Mathematics and Computation 182 (2006) 1725–1734
[73] A. Schobel, Locating stops along bus or railway lines – a bicriteria problem, Annals of Operations Research 136 (1) (2005) 211–227. [74] P. Sinha, A. Banerjee, Store choice behavior in an evolving market, International Journal of Retail & Distribution management 32 (10) (2004) 482–492. [75] A. Siren, L. Hakamies-Blomqist, Private car as the grand equalizer? demographic factors and mobility in finnish men and women aged 65? Transportation Research Part F: Traffic Psychology & Behavior 7 (2004) 107–118. [76] V.P. Sisiopiku, D. Akin, Pedestrian behaviors at and perceptions towards various pedestrian facilities: an examination based on observation and survey data, Transportation Research Part F: Traffic Psychology & Behavior 6 (2003) 249–274. [77] D. Smith, S. Sanchez, Assessment of business potential at retail sites: empirical findings from a US supermarket chain, International Review of Retail, Distribution and Consumer Research 13 (1) (2003) 37–58. [78] L. Snyder, M. Daskin, Reliability models for facility location: the expected failure cost case, Transportation Science 39 (3) (2005) 400– 417. [79] S. Sulaiha, K. Hokao, Estimation of Traffic Generated by New Development: Current Practice and Possible Improvements Based on Bangkok Experience. Working Paper, Victoria Transport Policy Institute, Victoria, BC, Canada (May 2000). [80] T. Suzuki, J. Hodgson, Optimal facility location with multi-purpose trip making, IIT Transactions 37 (5) (2005) 481–486. [81] G. Underwood, P. Chapman, Z. Berger, D. Crundall, Driving experience, attention focusing, and the recall of recently inspected events, Transportation Research Part F: Traffic Psychology & Behavior 6 (2003) 289–304. [82] T. Valletti, Location choice and price discrimination in a duopoly, Regional Science and Urban Economics 32 (3) (2002) 339–358. [83] J. Van Ommeren, A.F. Bumb, A. Sleptchenko, Locating repair shops in a stochastic environment, Computers & Operations Research 33 (6) (2006) 1575–1594. [84] Y. Wang, C. Parkan, Multi-attribute decision-making based on fuzzy preference information on alternatives: ranking and weighting, Fuzzy Sets and Systems 153 (2005) 331–346. [85] J. Winer, A. Schnee, H. Malllot, Use an interaction of navigation strategies in regionalized environments, Journal of Environmental Psychology 24 (2004) 475–493. [86] J. Zhang, B. Chen, Y. Ye, A multiexchange local search algorithm for the capacitated facility location problem, Mathematics of Operations Research 30 (2) (2005) 389–400. [87] D. Zhou, I. Vertinsky, Strategic location decisions in a growing market, Regional Science and Urban Economics 31 (5) (2001) 523– 641.
Further reading [1] O. Berman, D. Krass, The generalized maximal covering location problem, Computers & Operations Research 29 (6) (2002) 563–581. [2] A. Dasci, G. LaPorte, A continuous model for multistore competitive location, Operations Research 53 (2) (2005) 263–280. [3] A. Gupta, B. Su, Z. Walter, An empirical study of consumer switching from traditional to electronic channels: a purchase decision process perspective, International Journal of Electronic Commerce 8 (3) (2004) 131–141. [4] C. Hallam, G. Pindar, Prediction of land use traffic impact, Transportation Research Record 940 (1983) 51–60. [5] T. Helstrup, S. Magnussen, The mental representation of familiar, long distance journeys, Journal of Environmental Psychology 21 (2001) 411–421. [6] T. Hernandez, D. Bennison, The art and science of retail location decisions, International Journal of Retail & Distribution Management 28 (8) (2000) 357–367. [7] M.C. Hidalgo, B. Hernandez, Place attachment: conceptual and empirical questions, Journal of Environmental Psychology 21 (2001) 273–281. [8] C. Hsu, Y. Hsieh, Travel and activity choices based on an individual accessibility model, Papers in Regional Science 83 (2004) 387– 406. [9] Y. Kato, Y. Takeuchi, Individual differences in way-finding strategies, Journal of Environmental Psychology 23 (2003) 171–188. [10] P. Kaufmann, F. Lafontaine, Costs of control: the sources of economic rents for Mcdonalds franchisees, Journal of Law & Economics 37 (1994) 417–453. [11] C. Keen, M. Wetzels, K. Ruyter, et al., ETailers versus retailers: which factors determine consumer preferences 2004.