The Perils of Ignoring the Budget Constraint in Single-Unit Demand Models

Abstract

The problem of inferring unrealistically high prices from choice-based conjoint optimization exercises that lack face validity is widely known among applied researchers in industry and academia. We identify the omission of budget constraints in the standard single-unit demand discrete-choice framework as a hitherto unaccounted reason for optimization results that lack face validity, and present estimation and simulation methods that account for unobserved budget constraints. Our findings are complementary to two existing approaches that link reported choices to economic consequences (incentive-aligned conjoint analysis) or discard respondents from the sample that appear not to provide usable information, or simply answer randomly (sample censoring). We motivate budgets from theory and show that - by disentangling price-sensitivity within a budget from the budget constraint itself - accounting for unobserved budgets substantially increases the face validity of implied competitive prices in an industry-grade discrete-choice experiment. Finally, we leverage another large scale discrete-choice experiment to investigate how inferred budgets relate to a rich set of observed consumer characteristics. We find that ignoring budgets may obfuscate systematic relations between consumer characteristics and choice behavior.