The Surplus Identification task and limits to multi-attribute consumer choice

We present a novel experimental method for investigating consumer choice. The Surplus Identification (S-ID) task is inspired by studies of detection in perceptual psychophysics. It employs a forced-choice procedure, in which participants must decide whether a novel product is worth more or less than the price at which it is being offered, that is, whether there is a positive or negative surplus. The SI-D task reveals how precision, bias and learning vary across attribute and price structures. We illustrate its use by testing for cognitive capacity constraints in multi-attribute choice in three separate experiments, with implications for models of bounded rationality and rational inattention. As the number of product attributes rises from one to four in the S-ID task (Experiment 1), participants cannot integrate additional information efficiently and they display systematic, persistent biases, despite incentivised opportunities to learn. Experiment 2 demonstrates how the S-ID task can be used to track learning and serves as a robustness check for the findings of Experiment 1. Experiment 3 adapts the S-ID task to test accuracy of surplus identification when multiple attributes are perfectly correlated. The S-ID task also has the potential to test multiple aspects of consumer choice models and to test specific hypotheses about the cognitive mechanisms behind surplus identification.