Abstract
I implement a DCE to collect data with which I explore the variation across households in preferences for seven attributes of solar photovoltaic (PV) systems: upfront cost; savings on purchase of electricity; capitalisation into house price; fit with house; confidence in system performance; impact on household habits; and lock-in with one electricity retailer. The web-based survey software, called 1000Minds, works on an algorithm designed to specifically to elicit preferences at an individual level; the preference data consist of estimates of the relative strength of preference each participant places on each level of each attribute. The participants consist of 132 Dunedin owner-occupiers (50% response rate) who vary considerably in their preferences and their household characteristics. Cluster analysis reveals three distinct and informative clusters or market segments, each dominated by one or two major attributes of PV systems: a group especially keen to avoid a large upfront expenditure, a group keen to avoid a long-term commitment to a single electricity retailer, and a group relatively concerned about practical aspects of the performance of the PV system. This information about ‘market segments’ has implications for the design of public policy and commercial practice intended to increase the uptake of rooftop PV systems.
Of interest is how well preferences, as revealed in the choice survey, correlate with easier-to-observe house and householder characteristics. Multinomial logistic regression results indicate correlation sufficient to predict the allocation of participant households to clusters fairly accurately. However, the relationships between cluster membership and demographics are somewhat complex in that they involve a relatively large number of house and householder characteristics that interact to a significant extent.