The Impact of Unpredictable Work Schedules on Hourly Workers’ Ability to Predict Their Short-Term Liquidity Demand: Evidence from an Earned Wage Access Fintech Experiment

How to cite

Koepcke, Eric; Cefala, Luisa; Swanson, Nicholas. 2022. "The Impact of Unpredictable Work Schedules on Hourly Workers’ Ability to Predict Their Short-Term Liquidity Demand: Evidence from an Earned Wage Access Fintech Experiment." AEA RCT Registry. AEARCTR-0009362. https://doi.org/10.1257/rct.9362-1.2

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Eric Koepcke* , Luisa Cefala , Nicholas Swanson

* Primary investigator

Registry ID
AEARCTR-0009362
DOI
https://doi.org/10.1257/rct.9362-1.2
First registered
Last updated
Status
In development
Trial dates
2022-05-13 – 2023-05-15
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/9362

Abstract

We partner with an Earned Wage Access (EWA) fintech company to run a survey-based experiment through their app. EWA technology gives workers access to their wages as they are earned, rather than having to wait until payday. EWA serves as an alternative to high-interest forms of short-term credit (e.g., payday loans) and predominantly caters to lower-wage, hourly workers.

Our experiment seeks to answer three questions: (1) Can workers accurately predict how many hours they will work in the future? (2) Can workers accurately predict how much pay they will advance in the future? (3) How much of a role, if any, does misprediction of work hours play in misprediction of pay advanced? Our survey collects workers’ beliefs about work hours and pay advances. Using workers’ actual hours and pay advances, we can answer the first two questions. We develop and employ two treatments in order to answer the third.

Intervention

We run a survey-based experiment. An invitation to our survey is sent through the fintech company’s app.

Survey:
1. The survey collects workers’ beliefs about how many hours they'll work during their next pay period.
2. The survey collects workers’ beliefs about how much pay they'll have advanced to them during their next pay period.
3. In the vein of Allcott et al. (forthcoming), we introduce an incentive to advance less pay during the next pay period. We collect workers' beliefs about pay advances both without and with this incentive. We also use a multiple-price list (MPL) to measure workers’ valuation of this incentive.

The app's administrative data records workers actual work hours and pay advances during their next pay period.

Experimental design

We have three treatment groups in our experiment:
1. A control group is just given the survey described above.
2. An information treatment group is provided with information before they report their work hours and pay advance beliefs.
3. A fixed hours treatment group is asked to report their pay advance beliefs under different work hours outcomes.

Design details

Randomization method
Randomization done by computer.
Randomization unit
Individual
Clusters
3,000-6,000 individuals.
Observations
3,000-6,000 individuals.
Arms
1,000-2,000 individuals in the control group and each of the two treatment groups.
Primary outcomes
1. Beliefs: For future work hours, we are interested in how the information treatment affects workers’ hours beliefs. For future pay advances, we are interested in the effect the information treatment has on workers’ pay advance beliefs. For the fixed hours treatment, we are interested in how their pay advance beliefs differ under different work hours outcomes, as well as how they differ relative to the control and information treatment groups. 2. Misprediction: Using the administrative data, we can measure how accurate workers’ hours and pay advance beliefs are on average. This can be done for each treatment group. In addition to misprediction within each treatment group, we are interested in the differences in misprediction between each treatment group and under different work hours outcomes for the fixed hours treatment.
Secondary outcomes
-'Effect' of Incentive to Advance Less Pay: Within each treatment group, we are interested in comparing workers’ pay advance beliefs without the incentive to those with the incentive, i.e., the perceived change in future advanced pay the incentive would lead to. We are also interested in how much workers’ value this incentive. We will look at these outcomes within each treatment, as well as between treatments and under different work hours outcomes for the fixed hours treatment.

Cited by

No citation data available for this registration yet.

Keywords