The How and Why of Household Reactions to Income Shocks

Last registered on March 26, 2025
View Trial History

Pre-Trial

Trial Information

General Information

Title
The How and Why of Household Reactions to Income Shocks
RCT ID
AEARCTR-0015623
Initial registration date
March 21, 2025

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
March 26, 2025, 9:31 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country
United States of America
Region

Primary Investigator

Name
Roberto Colarieti
Affiliation
Harvard University
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Stefanie Stantcheva
PI Affiliation
Harvard University
Contact Investigator
PI Name
Pierfrancesco Mei
PI Affiliation
Harvard University
Contact Investigator

Additional Trial Information

Status
Completed
Start date
2021-05-01
End date
2023-01-31
Keywords
Other
Additional Keywords
Spending, saving, credit, heterogeneous agents, surveys, household finance
JEL code(s)
D12, D15, E21, E6, G51, H31
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
This paper studies how and why households adjust their spending, saving, and borrowing in response to transitory income shocks. We leverage new large-scale survey data to first quantitatively assess households' intertemporal marginal propensities to consume (MPCs) and deleverage (MPDs) (the "how"), and second to dive into the motivations and decision-making processes across households (the "why"). The combination of the quantitative estimation of household response dynamics with a qualitative exploration of the mental models employed during financial decisions provides a more complete view of household behavior. Our findings are as follows. First, we validate the reliability of surveys in predicting actual economic behaviors using a new approach called cross-validation, which compares the responses to hypothetical financial scenarios with observed actions from past studies. Participants' predicted reactions closely align with real-life behaviors. Second, we show that MPCs are significantly higher immediately following an income shock and diminish over time, with cumulative MPCs over a year showing significant variability. However, MPDs play a critical role in household financial adjustments and display significantly more cross-sectional heterogeneity. Neither is easily explained by socioeconomic or financial characteristics alone, and the explanatory power is improved by adding psychological factors, past experiences, and expectations. Third, using specifically-designed survey questions, we find that there is a broad range of motivations behind households' financial decisions and identify four household types using machine learning: Strongly Constrained, Precautionary, Quasi-Smoothers, and Spenders. Similar financial actions stem from diverse reasons, challenging the predictability of financial behavior solely based on socioeconomic and financial characteristics. Finally, we use our findings to address some puzzles in household finance.
External Link(s)

Registration Citation

Citation
Colarieti, Roberto, Pierfrancesco Mei and Stefanie Stantcheva. 2025. "The How and Why of Household Reactions to Income Shocks." AEA RCT Registry. March 26. https://doi.org/10.1257/rct.15623-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
We run separately two survey waves.
The main survey waves takes place between November 2022 and January 2023. The primary sample comes from this wave.
An additional survey waves takes place between May and October 2021.
Intervention (Hidden)
Intervention Start Date
2021-05-01
Intervention End Date
2023-01-31

Primary Outcomes

Primary Outcomes (end points)
The key outcome variables are: (1) spending and deleveraging propensities of surveyed households in response to hypothetical income shocks; (2) classification of households based on self-reported reasons to spend or deleverage.
Primary Outcomes (explanation)
Spending and deleveraging propesities: fraction of reported spending and deleveraging over the hypothetical income shock received.
Classification: based on machine learning algorithm, groups together households who report similar patterns of reasons for spending and deleveraging.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
For the primary survey wave (November 2022 - January 2023):
(A) We elicit spending and deleveraging propensities to income shocks through hypothetical scenarios. Respondents are asked to allocate hypothetical income shocks to additional spending, debt repayments, and savings over four quarters. We adopt the following randomizations. (i) The shock size (fixed and worth $1000, or 10% of the household’s total annual net income); (ii) The shock sign, either positive (unexpected one-time payment) or negative (unexpected one-time expense); (iii) The timing of the shock: each respondent is presented with two scenarios. In the first scenario, common to all respondents, the income shock occurs right away. In the second scenario, a randomly selected half of the sample is asked to consider an income shock occurring in one quarter, and the other half a shock occurring in two quarters in the future. Treatments (i)-(ii)-(iii) are assigned independently across them.
(B) We also elicit qualitative responses to income shocks. First, we elicit adjustment margins used by the household in response to both positive and negative income shocks of the same size as described in (A). Then, conditional on selecting or not selecting a given adjustment margin (i.e., spending, debt, savings, working hours), respondents are asked a set of questions about why they are using or not using that given margin.

For the first wave of our survey (May - October 2021): we additionally randomize across the source and the horizon of the income shock. In particular, the source of the shock can either be a direct Federal transfer such as a stimulus check, or a generic non-government transfer such as a bonus, gift or win. The horizon over which respondents allocate the income shock is instead either four or eight quarters.

Overall, the goal of our randomizations is to allocate different types of income shocks across respondents.
Experimental Design Details
Randomization Method
Randomization is done automatically by Qualtrics (the program we used to run the survey). The survey is conducted online.
Randomization Unit
The randomization is at the individual level. The individuals are U.S. residents who are in the labor force at the time of the interview, and are aged between 25 and 65.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
We do not have clusters.
Sample size: planned number of observations
The primary wave (November 2022 - January 2023) has approximately 3000 respondents. The first wave of the survey (May - October 2021) has approximately 1200 respondents.
Sample size (or number of clusters) by treatment arms
For the primary sample (Nov 2022 - Jan 2023), we allocate half of the sample in each treatment arm. Treatments (i)-(ii)-(iii) are assigned independently across them. So:
(i) 1500 respondents receive a fixed income shock and 1500 respondents receive a proportional income shock;
(ii)1500 respondents receive a positive income shock and 1500 respondents receive a negative income shock;
(iii) 1500 respondents receive a shock that occurs in 1 quarter and 1500 respondents receive a shock that occurs in 2 quarters.

For the first wave (May - October 2021) we allocate:
(a) 600 respondents to the government transfer income source vs 600 respondents to the non government transfer income shocks;
(b) 840 respondents are allocated to the 4 quarter horizon vs 360 respondents are allocated to the 8 quarter horizon.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Harvard University-Area IRB, Committee on the Use of Human Subjects
IRB Approval Date
2021-03-22
IRB Approval Number
IRB21-0279
Analysis Plan

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
Yes
Intervention Completion Date
January 31, 2023, 12:00 +00:00
Data Collection Complete
Yes
Data Collection Completion Date
January 31, 2023, 12:00 +00:00
Final Sample Size: Number of Clusters (Unit of Randomization)
No clusters.
Was attrition correlated with treatment status?
No
Final Sample Size: Total Number of Observations
For the primary wave of the survey (Nov 2022 - Jan 2023): 2923 respondents.
The first wave of the survey (May - October 2021): 1293 respondents.
Final Sample Size (or Number of Clusters) by Treatment Arms
For the primary sample (Nov 2022 - Jan 2023), we allocate half of the sample in each treatment arm. Treatments (i)-(ii)-(iii) are assigned independently across them. So: (i) 50% of respondents receive a fixed income shock and 50% of respondents receive a proportional income shock; (ii) 50% of respondents receive a positive income shock and 50% of respondents receive a negative income shock; (iii) 50% of respondents receive a shock that occurs in 1 quarter and 50% of respondents receive a shock that occurs in 2 quarters. For the first wave (May - October 2021) we also allocate: (a) 50% of respondents to the government transfer income source vs 50% of respondents to the non government transfer income shocks; (b) 70% of respondents are allocated to the 4 quarter horizon vs 30% of respondents are allocated to the 8 quarter horizon.
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
No
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials