Consumption Wedges: Measuring and Diagnosing Distortions

Ample empirical evidence documents deviations from the canonical consumption-savings model; yet, it remains difficult to assess the roles of different underlying distortions, such as financial constraints and behavioral preferences. We develop a sufficient-statistics approach that measures individual-level wedges between observed and counterfactual “frictionless” consumption. Since different distortions imply different wedge properties, wedges provide a diagnostic to distinguish between models. We measure wedges using administrative transactions data linked to surveyed expectations for a population of middle-income, low-liquidity US consumers. The expectations data allow us to distinguish wedges attributable to frictions and behavioral preferences from wedges driven by deviations from full-information rational expectations (FIRE). We find that consumption wedges are large and heterogeneous: the median wedge is 40% of frictionless consumption in absolute value, with 49% of consumers under-consuming (negative wedges) and 51% over-consuming (positive wedges). Borrowing constraints cannot rationalize this pattern because they only generate negative wedges. Models combining present bias with borrowing constraints, or featuring consumption adjustment costs, best account for the wedge properties we document.