Modeling patronage shift to a new entrant for predicting disproportionate losses for incumbent outlets
Introduction
Understanding store choice behavior is a fundamental step in the process of predicting the effect of the introduction of a new outlet. Bucklin (1967) and Geisel, Narasimhan, and Sen (1993) collect individual-level data to measure the effect of a new store on the market shares of existing stores. Drezner (1994), Ghosh and Craig (1983) and Kaufmann and Rangan (1990) develop normative models based on store choice models in order to provide insights into the best location for a new store. Recently, Singh, Hansen, and Blattberg (2006) analyzed a frequent-shopper database, investigating the effect of a Wal-Mart supercenter on the inter-purchase time and purchase amount of an incumbent store’s customers. Pancras, Sriram, and Kumar (2009) develop a dynamic demand model for addressing whether new chain stores just cannibalize sales from the chain’s existing stores or bring in additional sales.
While these prior studies draw the insightful conclusion that the impact of a new outlet on the market shares of incumbent outlets is disproportionate, they overlook the patronage-shifting behavior that underlies retail competition. Most of the studies use store choice models which assume that consumers are most likely to visit the store that is most accessible. While this probabilistic nature of spatial behavior is still widely used by studies of retail competition (e.g., Davis, 2006, and Thomadsen, 2007), the models do not guarantee a proper representation of patronage-shifting behavior. This assumption, which was proposed by Huff (1964), has the Independence-from-Irrelevant-Alternatives (IIA) property at the consumer level. If the assumption of proportionate substitution patterns at the consumer level is not valid, it may cause the prediction of market-level substitution patterns to be inaccurate. This inaccuracy can be more severe in retail markets where the consumer choice set of the market is a nested structure (Gonzalez-Benito, Munoz-Gallego, & Kopalle, 2005) or spatial competition is asymmetric (Zhu and Singh, 2009, Zhu et al., 2009).
Our research objective is to provide insights into how consumers shift their patronage in a retail market which is facing growing competition. To achieve this goal, the study develops a store choice model that accounts for patronage-shifting behavior. Under the assumption that the store choice set plays a key role in consumers’ patronage shift decisions, outlet characteristics which affect the choice set formation of consumers are incorporated into the proposed model. The proposed model is then applied to the motion picture exhibition industry. Movie theater market share data are used for model calibration. Our underlying premise for model assessment is that actual patronage-shifting behavior is represented more closely by a store choice model that more accurately predicts changes in market share and estimates consumer-level store choice patterns. The factors of interest here are accessibility of the new outlet, outlet awareness, and inter-outlet substitutability.
The primary contribution of this study is to expand our knowledge of patronage-shifting behavior. Prior studies only suggest that consumers are more likely to shift their patronage to a new outlet if it is geographically more accessible to them. This study shows that the decreasing rate of patronage of existing outlets is disproportionate because of the asymmetric substitutability of a new entrant for incumbents, which depends on its proximity to the incumbents, their differences in size, and the asymmetric substitution relationship. In addition, this study finds that the probability of a consumer being aware of a new outlet increases gradually over each time period. Our findings are demand-side insights into the patronage-shifting behavior that underlies asymmetric spatial competition, which is a supply-side implication of the findings of Zhu and Singh (2009) and Zhu et al. (2009).
Another contribution of this study is the identification of the importance of outlet characteristics in improving prediction accuracy at the market level. Most prior studies have presented the disproportionate changing patterns of market share by using store choice models with the IIA property. However, these studies infer a market-level phenomenon from consumer-level data without making any attempt to justify the accuracy of their inferences using actual market response data. The present study empirically demonstrates that the IIA assumption at the consumer level is not valid for representing retail competition at the market level. The proposed model, with both the substitutability and accessibility factors, predicts market share more accurately than the benchmark model, which only has the accessibility factor. In particular, substitutability is more crucial than accessibility in capturing the varying impacts of the new entrant on different existing theaters. In addition, modeling the probability that consumers will be aware of new outlets is important in improving the prediction accuracy. The proposed model with the awareness factor represents these actual market responses more accurately than the benchmark model, which only has the outlet age variable for capturing the dynamics.
Our secondary contribution is to show that the proposed approach using aggregate data can estimate the patronage patterns of consumers properly. Studies of location research generally propose an approach using consumer survey data for building a store choice model, and then use the model to predict market share losses (Geisel et al., 1993, Ghosh and Craig, 1983). However, there are two limitations when utilizing survey data: first, the prediction error is often large due to the responses being biased against behavioral frequency (e.g., Blair & Burton, 1987, and Lee, Hu, & Toh, 2000), or due to inaccurate questions about patronage behavior, without considering store-switching (Leszczyc & Timmermans, 1997); second, it is impractical to conduct consumer surveys frequently when competing stores are entering the market continuously, owing to time and budget constraints. Our empirical study reveals that the patterns of theater patronage estimated from aggregate data are consistent with survey responses from moviegoers. This consistency implies that the proposed approach using aggregate data should enable retailers to predict changes in the patronage patterns of consumers in a timely manner prior to a new entry. Such opportune predictions will allow timely decisions on the part of retail and customer management.
The remainder of this paper consists of four sections. The next section suggests the major characteristics of a new store which will affect consumer choice set formation, and explains the procedure of the model specification. The following sections present the data and the results of an empirical analysis. Finally, the paper concludes with a summary and suggestions for further research.
Section snippets
Model
Choice set plays a key role in modeling patronage-shifting behavior. As consumers generally avoid evaluating all of the alternatives in the markets (e.g., Shocker, Ben-Akiva, Boccara, & Nedungadi, 1991; and Roberts & Lattin, 1997), they will only shift their patronage to a new outlet if it is a part of their choice set. In two-stage or hierarchical spatial choice processes (Black, 1984, Fotheringham, 1988, Spiggle and Sewall, 1987), patronage of existing outlets can also change if the new
Market
This study applies the proposed approach to the motion picture exhibition industry in South Korea. The market environment of the industry was very stable between the early 1980s and the late 1990s, but a retail revolution started in 1998 because of the introduction of the multi-screened complex theater (so called multiplex). Exhibitors maintain their market expansion strategy continuously, and consequently the market competition among theaters has intensified. The market we use in our empirical
Benchmark model
The estimation and prediction results of the proposed model are compared with those of the benchmark model, Eq. (16). Eq. (16) is given in Box I, where and . The benchmark model is the competing destinations model (CDM) proposed by Fotheringham (1988). He represents patronage as a function of the degree of clustering of stores and the attractiveness of a store, considering the distance of the store from
Conclusion
The entrance of competing outlet entries into the marketplace poses a substantial threat to existing outlets. Incumbents become extremely concerned about how much their market shares will decrease, which factors have a significantly effect on customer behavior, and which customers will change their patronage to a new outlet. This study analyzes outlet market share data in order to investigate the patronage-shifting behavior that underlies retail competition. The results of our empirical
Duk Bin Jun is a Professor at KAIST Business School. He received his doctorate from the U.C. Berkeley. He has published in the International Journal of Forecasting, Journal of Forecasting, Technological Forecasting and Social Change, Marketing Letters, Telecommunications Policy, Telecommunication Systems, and other journals. His research interests include adaptive forecasting, structural changes in time series analysis, business cycle forecasting, new product diffusion and choice processes, and
References (34)
Optimal continuous location of a retail facility, facility attractiveness, and market share: an interactive model
Journal of Retailing
(1994)- et al.
Shopping-center patronage models: fashioning a consideration set segment solution
Journal of Business Research
(1990) - et al.
Quantifying the competitive impact of a new entrant
Journal of Business Research
(1993) - et al.
Asymmetric competition in retail store formats: evaluating inter- and intra-format spatial effects
Journal of Retailing
(2005) - et al.
Two-stage discrete choice models for scanner panel data: an assessment of process and assumptions
European Journal of Operational Research
(1998) - et al.
Constructive consumer choice processes
Journal of Consumer Research
(1998) Choice-set definition in patronage modeling
Journal of Retailing
(1984)- et al.
Cognitive processes used by survey responses to answer behavioral frequency questions
Journal of Consumer Research
(1987) How to standardize regression coefficients
The American Statistician
(1994)- et al.
Limited choice sets, local price response, and implied measures of price competition
Journal of Marketing Research
(1996)
Competitive impact of a new supermarket
Journal of Marketing Research
Bringing stores to shoppers–not shoppers to stores
International Journal of Retail and Distribution Management
Spatial competition in retail markets: movie theaters
The RAND Journal of Economics
Maximum likelihood from incomplete data via the EM algorithm
Journal of the Royal Statistical Society, Series B
The role of spatial demand on outlet location and pricing
Journal of Marketing Research
Consumer store choice and choice set definition
Marketing Science
Formulating retail location strategy in a changing environment
Journal of Marketing
Cited by (0)
Duk Bin Jun is a Professor at KAIST Business School. He received his doctorate from the U.C. Berkeley. He has published in the International Journal of Forecasting, Journal of Forecasting, Technological Forecasting and Social Change, Marketing Letters, Telecommunications Policy, Telecommunication Systems, and other journals. His research interests include adaptive forecasting, structural changes in time series analysis, business cycle forecasting, new product diffusion and choice processes, and telecommunication forecasting.
Jungki Kim is a senior researcher at the Research Institute of Insurance and Finance at Samsung Life Insurance. He received his doctorate from KAIST in 2010. His research interests include predictive models for changes in market demand in a changing environment.
Myoung Hwan Park is a professor in the Department of Industrial Engineering at Hansung University. He received his doctorate from KAIST in 1993. His research interests include telecommunications forecasting and new product forecasting. He has published in Telecommunication Systems, Computers and Operations Research, and other journals.
Kyoung Cheon Cha is an assistant professor in the Department of Business Administration at Dong-A University. He received his doctorate from KAIST in 2004. His research interests include demand forecasting and marketing dynamics. He has published in Telecommunications Policy and other journals.