Home › What the research says

What Research Says About How People Spend a Tax Refund

The label changes the outcome

Nicholas Epley, Dennis Mak, and Lorraine Chen Idson published this in the Journal of Behavioral Decision Making in 2006. Their reasoning is that all income raises absolute wealth, but consumption decisions run more on perceived change, and change is computed by comparing a current state to a former one. So money framed as a gain from where you are now should get spent more freely than the same money framed as a return to where you were. Four experiments supported that prediction, covering memory for spending of a government tax rebate, unobtrusive self-report of spending an unexpected windfall, and actual purchases of items for sale in a lab. Same amount, different description, different behavior.

What people said they would do

Matthew Shapiro and Joel Slemrod studied the 2001 US tax rebate in the American Economic Review in 2003 using survey responses. Only 22 percent of households receiving the rebate said they would mostly spend it, with the rest reporting they would save it or use it to pay down debt. They describe this as a striking break from what past behavior would have predicted. This is stated intention, not measured spending, which is the standard limitation of survey-based estimates and worth holding on to. What it establishes is that the money was not universally treated as free spending, and that people's stated plan for a windfall is a real object worth studying.

What the spending data showed

Jonathan Parker, Nicholas Souleles, David Johnson, and Robert McClelland measured the 2008 economic stimulus payments in the American Economic Review in 2013, using questions added to the Consumer Expenditure Survey and the randomized timing of disbursement. Households spent 12 to 30 percent of their payments on nondurable goods during the three-month period of receipt, depending on specification, and significantly more on durables, primarily vehicles, bringing the total response to 50 to 90 percent of the payments. Responses were substantial for older, lower-income, and home-owning households. Notably, spending did not vary significantly with the method of disbursement, check or electronic transfer.

Reading the three together

These studies are not measuring the same thing, and pretending otherwise is how bad advice gets written. One is a framing experiment, one is stated intention about a 2001 rebate, one is measured spending on a 2008 stimulus payment. Different years, different policies, different methods. What they jointly support is modest and useful: a windfall is not automatically saved, a substantial share of it does get spent, and how the money is described changes the share. Nothing here tells you what to do with a refund, and none of it is tax advice. It tells you that the decision is more malleable than the amount suggests, and that the labeling happens early.

The envelope change to make

Split the money into named envelopes the day it arrives, before anything gets a chance to relabel it as extra. A refund is your own withheld income coming back, which is the rebate framing, and it will not stay that way once it is sitting in a checking account next to your regular balance. Decide the split in advance if you can, in whole percentages across envelopes you already have, and write the decision down before the deposit. The mechanism you are using is the only one this literature actually demonstrates, which is that the name attached to money moves it.

Sources

Every source below was retrieved and checked. Findings are reported as the source states them.

• Bonus of rebate?: The impact of income framing on spending and saving — Journal of Behavioral Decision Making, 2006
  People were more likely to spend income framed as a gain from a current wealth state than income framed as a return to a prior state, confirmed across four experiments covering memory for spending of a government tax rebate, self-report measures of spending an unexpected windfall, and actual spending in a laboratory experiment.
• Consumer Response to Tax Rebates — American Economic Review, 2003
  In survey data on the 2001 US tax rebate, only 22 percent of households receiving the rebate said they would mostly spend it, with most reporting they would save it or use it to pay down debt, a low rate relative to what past behavior would have predicted.
• Consumer Spending and the Economic Stimulus Payments of 2008 — American Economic Review, 2013
  Using questions added to the Consumer Expenditure Survey and randomized disbursement timing, households spent 12 to 30 percent of their payments on nondurable goods during the three-month period of receipt and significantly more on durables, primarily vehicles, bringing the total response to 50 to 90 percent of the payments; spending did not vary significantly with check versus electronic disbursement.

Common questions

Is a tax refund a bonus or a return of my own money?

Factually it is a return of income that was already withheld from you, so the rebate description is the accurate one. Epley, Mak, and Idson found that describing identical money as a return to a prior state rather than as a gain led people to spend less of it. That is a research finding about framing, not permission to trick yourself into a specific decision. If you want the money to behave like recovered income rather than found income, describing it accurately is the cheapest thing you can do.

What percentage of a windfall do people typically save?

There is no single number, and any source giving you one is overreaching. Shapiro and Slemrod found 22 percent of households said they would mostly spend the 2001 rebate. Parker and colleagues measured 12 to 30 percent of the 2008 stimulus payments going to nondurables within three months, with the total response reaching 50 to 90 percent once durables are included. Those are different years, different policies, and different measurement approaches, so they cannot be averaged into one figure.

Should I split a refund or put it all toward one thing?

The research does not answer that, and it depends on facts about your situation that no page can know. What the research supports is the timing rather than the split: assign it before it lands in a general balance, because that is when the label is still attached. Whether the assignment is one envelope or five is a decision about your priorities. Writing it down before the money arrives is the part with evidence behind it.

Related