Increasing savings at tax-time

Tax pre-commitment
Experiment Type
Field Experiment
Save more
Increase short-term savings
Focus Areas
Marketing & messaging
Behavioral Concepts
Default bias Pre-commitment
Partner Type

What Happened

It worked. By implementing pre-commitment into the tax refund process, this nearly doubled savings rate to 22% (Experimental condition) from 12% (Control condition). Additionally, this led to an average savings rate of 15% (Experimental condition) from 10% (Control condition).

Lessons Learned

These results suggest that implementing pre-commitment into the tax refund process may increase savings from those respective tax refunds.


One of the biggest opportunities to help Americans save is when they receive their tax refunds. Roughly 83% of tax filers receive a tax refund, in the order of $3,120 on average.

We partnered with San Francisco fin-tech startup Digit, which aims to increase short-term savings by automatically withdrawing small amounts of money on a frequent basis from users’ checking accounts. Unlike other financial management applications, Digit primarily communicates with its users through SMS texting. Users link their primary checking account to Digit, and Digit sends users regular updates, including deposit alerts and daily balance status updates.

Prior behavioral research has shown that people are likely to make better long-term decisions if they pre-commit to a decision before they have to face the consequences of a decision. In short, we can commit to start saving $1,000 a year from now, because we don’t feel the pain of that commitment today. However, committing to starting to save $1,000 today is very difficult.


We leveraged the behavioral principle of pre commitment to get people to commit to save before they received their tax refund. To check if this principle would work, we conducted a randomized controlled trial in which users received one of two conditions.

In our control condition, we asked users to tell Digit what percentage of their tax refund they would like to save at the time Digit detected that they received a federal income tax refund in their checking account.

In our experiment condition, we asked users what percentage of their tax refund they would like to save before they received their tax refund.

In each case, Digit automatically saved the self-reported percentage when the funds were deposited into the users’ checking account.


The results of this simple text were astonishing. Of everyone who responded to the text, on average people wanted to save 10% of their tax refund in the control condition, and 15% in the experiment condition. When we looked at those that responded positively to the text and wanted to save, our experiment doubled the savings rate to 22% from 12% in the control condition.

This short low-cost text intervention helped thousands of people save part of their tax-refund, totaling over $1 million in savings. Three months after the intervention, roughly 85% of the savings were still in the savings accounts.