New Insights on Consumption and Income Dynamics from High-Frequency Transactions Data
Paper Session
Monday, Jan. 5, 2026 1:00 PM - 3:00 PM (EST)
- Chair: Christopher Carroll, Johns Hopkins University
Anticipatory Spending
Abstract
We leverage exogenous income changes from various state and federal transfer policies to demonstrate that U.S. consumers tend to increase spending in advance of income arrival. However, these anticipatory increases are relatively small compared to the substantial increases observed in the period following income receipt. Our analysis identifies two primary factors that jointly drive this behavior: present bias and salience. These factors explain both the small anticipatory response (excess smoothness) and the large spending response following the arrival of income (excess sensitivity). While liquid assets are associated with excess sensitivity in the post-income period, our findings suggest that this relationship is more likely driven by persistent household characteristics, such as present bias, rather than temporary conditions like binding credit constraints. Notably, individuals who respond strongly following income changes also tend to increase their spending the most in advance of the arrival of income. Tests of present bias are particularly powerful for repeated or otherwise highly salient income changes.Expecting More Tomorrow
Abstract
We study how employees respond to an expected and permanent salary increase. In standard economic models, agents would spend more before the salary increase and, potentially, utilize short-term unsecured debt. We first conduct a large-scale representative online survey and find that the reported anticipatory spending and borrowing responses to an expected and permanent salary increase is limited. Next, we use anonymized information on income, spending and account balances from a German bank to analyze by how much employees actually expand their spending and borrowing ahead of a permanent salary increase of at least 3% from the same employer. Consistent with the survey data, we find that the majority of employees wait until the actual salary increase before expanding their spending by any measurable amount. Furthermore, borrowing does not seem to be affected by the expected salary increase even though the bank offers overdraft facilities. By combining transaction-level consumption data with survey evidence, we show that consumption remains closely tied to realized income rather than expected permanent income changes, with little corresponding adjustment in borrowing behavior.Daily Consumption Smoothing? New Evidence from a Payment Diary
Abstract
This paper shows that consumption and income data in the Diary of Consumer Payment Choice (DCPC) cover high percentages of U.S. data, forecast well in real time, and replicate the literature's estimation and general rejection of PIH models with time-aggregated data. Novel estimates reveal evidence of daily consumption smoothing after accounting for three features of daily data: 1) discrepancy between consumption and expenditures (e.g., bill payments); 2) discrete daily expenditures and income; and 3) asynchronous consumption and income. Convenience samples used in the literature appear to reflect selection effects related to payment choices (cash or mobile) that affect model inference. Relative to bank transactions data, the DCPC is more representative, publicly available, and offers other advantages.Discussant(s)
Dirk Krueger
,
University of Pennsylvania
Jonathan Parker
,
Massachusetts Institute of Technology
Andrew Hertzberg
,
Federal Reserve Bank of Philadelphia
Orazio Attanasio
,
Yale University
JEL Classifications
- E2 - Consumption, Saving, Production, Investment, Labor Markets, and Informal Economy
- D1 - Household Behavior and Family Economics