Evaluation Scotland Wales
The UK Strategy for Financial Wellbeing is taking forward the work of the Financial Capability Strategy Opens in a new window


Balancing savings and debt: Findings from online experiment

Evidence type: Insight i


Most U.S. consumers hold savings and debt at the same time, even when the interest they are paying on the debt is higher than the interest they are earning. Previous work from an economic perspective and using observational data to explain this seemingly irrational behaviour, suggests that it may be due to a number of factors, such as the need for a savings cushion, concerns about unexpected reductions in credit card limits, or the desire to limit further spending. In contrast, this study is the first of its kind to take an experimental approach, using hypothetical scenarios, to explore possible psychological explanations for this behaviour.

The study

Using an online survey platform, participants were presented with an experimental scenario that randomly assigned them to one of ten conditions - ten savings amounts ranging between $1,000 and $10,000, belonging to a hypothetical ‘Mr Green’ - and asked how much savings they wanted to apply toward reducing his $5,000 in credit card debt.

The study used a convenience sample (recruited and compensated through Amazon’s Mechanical Turk, a crowdsourcing service) comprising 551 study participants, with 49 to 61 study participants randomly assigned to each of the ten savings scenarios.

Participants were also asked questions to assess their knowledge of the relative interest rates between savings accounts and credit cards, to test whether interest rate knowledge played a role in participants’ choices.

The experiment was conducted by the Research Office of the US Consumer Financial Protection Bureau, a government agency that makes sure banks, lenders, and other financial companies treat consumers fairly.

Key findings

The study found that the participants chose to preserve a savings cushion while also reducing debt.

  • Preserve a savings cushion: Most participants chose to continue holding some credit card debt to preserve more savings. In nine of the ten savings scenarios, fewer than half of participants put the maximum amount of savings toward debt reduction. Only in the scenario where savings was double the size of the debt ($10,000 versus $5,000), did even a majority of participants—77 percent— eliminate credit card debt.
  • Reduce debt: The vast majority of participants—over 90 percent in each of the savings scenarios—used at least some of the savings to pay down credit card debt. Further, on average, participants allocated more than half of the savings they could to debt reduction, even among those in the lowest savings groups.

There was no evidence that participants’ knowledge of relative interest rates between savings and credit cards influenced their decisions about how much savings to put toward credit card debt reduction.

The authors conclude that the study provides a valuable first step toward understanding how consumers preserve a savings cushion while pursuing debt reduction, but that there are many other questions about the motivation and behaviours of consumers that need to be explored.

Points to consider

  • Methodological strengths/weaknesses: The authors compared the sample to the nationally representative 2019 American Community Survey (ACS; U.S. Census Bureau) and the 2016 Survey of Consumer Finances (SCF; Federal Reserve Board), and find some differences. We can’t know the impact of these differences on how respondents answered the experiment. However the authors do conduct analyses that control for participant characteristics and find similar results.
    • They also explore effect sizes relating to the large variation in savings amounts. They state that they anticipated fairly large effect sizes (Cohen’s d of between 0.60 to 0.80, which translates to a 20 to 30 percent difference in the amount of savings used), and that power analyses suggested that between 42 and 72 participants per condition would be sufficient to detect differences for this range of effect sizes. The study achieves 49-61 participants per condition.
    • They also test whether participant demographic and economic characteristics vary across the 10 condition groups, and find roughly the same number of statistically significant differences as would occur by chance.
    • There is evidence of quality control: 179 participants were removed due to data quality concerns.
  • Generalisability/ transferability: The study took place in the US, but the findings about the motivations behind not paying off debt in full are likely to apply in other developed countries and credit markets.
  • Relevance: Topical and relevant, contributing to the understanding of how people make financial decisions.
    • The study will be of use to anyone with an interest in consumer decision making around savings and debt.

Key info

Client group
Year of publication
United States
Contact information