An experimental approach to valuing new differentiated products

Abstract

An experimental store was created to evaluate initial demand for locally produced and guaranteed tender steak products as a more realistic alternative to contingent valuation (CV) and dichotomous-choice experimental methods. Strengths of the approach are incentive compatibility, a realistic consumption set, and a familiar choice environment. Consumers selected among USDA Choice, premium quality, lean, guaranteed tender and locally produced strip steaks. A double-hurdle count data model indicated initial willingness-to-pay (WTP) for locally produced steak comparable to prior CV results, but demand was highly elastic. Demand for premium quality steak crowded out demand for the guaranteed tender product, contrasting with prior dichotomous-choice experimental results.

Introduction

One of the most striking trends in the US food system is the increasing emphasis on product differentiation at the farm level, and producers show heightened interest in identifying consumer willingness-to-pay (WTP) for prospective new products. Beef steak is a lightly processed food for which product differentiation often originates at the farm level. Within a given cut of steak, consumers can typically choose among quality grades, differentiated primarily by marbling, a measure of intramuscular fat content. Higher quality products are often branded (e.g., Certified Angus Beef). In some markets, lean and hormone-free attributes are bundled and sold at a premium (e.g., Laura's Lean Beef).

Two differentiated products of particular interest to the industry that are rarely available on supermarket shelves are certified locally produced beef (i.e., produced within the state) and ‘guaranteed tender’ steak. The hypothesised value-added attributes of locally produced beef are source verification (food safety, accountability, environmental stewardship) and support of local, independent, small-scale producers. Tenderness and uniformity are important steak quality attributes that, when bundled under the guaranteed tender label, are hypothesised to command premiums.

The goal of this study was to provide producers and potential investors with valid initial consumer demand assessments for locally produced steak and guaranteed tender steak. An experimental method was designed to avoid four potential problems common in food product valuation studies. First, most techniques elicit WTP for a single unit, but ignore diminishing marginal utility and interpret the result as WTP for multiple units. When a product is often purchased in multiple units for household consumption, as is steak, such methods can only exaggerate predicted commercial success. Second, demand for prospective products should be evaluated in the context of substitutes already occupying the market. Third, convincing evidence exists that hypothetical methods produce inflated WTP estimates (Cummings et al., 1995). Fourth, attempts to address the first two problems create a complex decision environment not well suited to choice-based conjoint (CBC) analysis with repeated scenarios. The guiding principle behind experimental design in this study was that context matters; any discrepancy between the experimental setting and a realistic shopping experience was expected to compromise the validity of information gathered.

The experiment illustrates a methodology that adds value because it returns data on both participation decisions (whether to purchase product j), and consumption decisions (how much of product j to purchase). Subjects faced a textbook utility maximisation problem allowing estimation of demand functions unencumbered by the aggregation, simultaneity and separability issues that accompany most demand analyses. Despite its strengths, the experimental method described here is not superiour to existing methods in every respect, and one purpose of this paper is to promote discussion of the relative merits of diverse elicitation methods.