UDAY RAJAN, University of Michigan at Ann Arbor - Stephen M. Ross School of Business
Email: urajan@bus.umich.edu
AMITABH SINHA, University of Michigan - Stephen M. Ross School of Business
Email: amitabh@umich.edu

We study a generalized Hotelling duopoly in which a consumer's net utility from a product depends on the location of product and consumer in product attribute space, a random utility term that captures idiosyncratic preferences, and the price of the product. Our model allows us to vary the relative impact of product attribute preferences (i.e., location) and idiosyncratic preferences on consumer utility. Since the model is analytically intractable, we computationally study equilibria under both simultaneous location and sequential location by the two firms, with prices decided simultaneously after locations have been chosen.

In our numerical analysis, the simultaneous game admits only symmetric equilibria. If product attribute preferences are weak (i.e., distance costs are unimportant) and idiosyncratic preferences are also weak, both firms locate in the interior of the feasible location space. With strong product attribute preferences and weak idiosyncratic preferences, price competition is at its most intense, leading to maximal differentiation. As idiosyncratic preferences become more important, price competition is mitigated, leading to both firms locating at the market center.

In the sequential game, we find that when product attribute preferences are strong and idiosyncratic preferences are weak, the follower has a strong desire to avoid price competition and the leader experiences a first-mover advantage. However, if idiosyncratic preferences are also strong, price competition is somewhat weaker, and the leader cedes a location and profit advantage to the follower. If product attribute preferences and idiosyncratic preferences are both weak, maximal differentiation results, and entry order is irrelevant. Finally, when idiosyncratic preferences dominate, price competition is a non-issue, and both players locate at the market center. Once again, the order of entry does not matter.

KENNETH C. WILBUR, University of Southern California - Marshall School of Business
Email: kwilbur@usc.edu
PAUL FARRIS, University of Virginia - Darden Graduate School of Business Administration
Email: farrisp@darden.virginia.edu

Launching a new stock-keeping unit (SKU) is a costly, common business practice with a high observed rate of failure. Market share forecasts for new SKUs are typically calibrated based on projections of consumer awareness, trial, repeat purchase, and retail distribution. Discussions with executives indicate that managers sometimes use distribution assumptions as a "tuning factor" to approve an otherwise unjustifiable go-to-market decision for a new SKU. We propose a two-step technique to assess the realism of distribution assumptions for new SKUs. The first step is to flexibly estimate the relationship between distribution and market share among the brand's existing SKUs. The second step is to invert this relationship and evaluate it at a new SKU's target market share, which allows for a statistical hypothesis test of a manager's distribution assumption at a prespecified confidence level. The primary advantages of the proposed technique are its speed, generality, and ease of implementation. We validate the method using brand-level holdout samples in 37 categories. We find that pre-existing SKUs' distribution-share relationships in 2004 explain at least half the variance in new SKUs' performance for 61 of 87 brands that introduced 5 or more new SKUs in 2005. For those brands where the model does a relatively poor job of explaining new SKUs' performance, successful new SKUs are very rare. These results suggest that our technique could have prevented the release of many new SKUs that ultimately failed.

BABU P. GEORGE, University of Southern Mississippi - College of Business Administration
Email: Babu.George@usm.edu

Available evidence from the literature tells that satisfaction depends upon the consistency between expectations and performance. When product performance is below expectations dissatisfaction is resulted, whereas satisfaction arises when performance equals or exceeds the performance expected. We argue that incorporating tourist's purchase involvement with the tourism product as a moderator in the above relationship can enrich the above model. Through an empirical study we establish that increased purchase involvement increases dissatisfaction in the event of disconfirmation and decreases satisfaction in the event of confirmation. Likewise, decreased purchase involvement decreases dissatisfaction in the event of disconfirmation and increases satisfaction in the event of confirmation.