Does Paying by Card Make You Spend More?

Key takeaways

The advice comes from a 2001 auction for Boston Celtics tickets in which bidders told they would pay by card offered roughly twice as much as bidders told they would pay cash.
Pooled across 392 effect sizes from 71 papers, the cashless effect is reliably positive and small, invisible inside any one shopping trip.
It concentrates where other people can see you: 0.551 for conspicuous purchases against 0.218 for everything else, and 0.023 for tips and donations, indistinguishable from nothing.
Not supported at this level of evidence: that cards specifically drive junk-food buying, that the danger is paying later, or that people tip or donate more by card.
The effect has declined gradually over the decades. The clearest measured benefit of paying cash is not restraint but attachment: greater post-transaction connection to what was bought.

In 2001 two MIT professors ran a sealed-bid auction for tickets to a Boston Celtics game. Real tickets, real money, genuine bids. Half the bidders were told that if they won they would pay in cash. The other half were told they would pay by credit card.

The card bidders offered roughly twice as much for the same seats.

That result — an effect the authors described as potentially as large as 100 per cent — is the origin of nearly every piece of advice you have read about cutting up your cards and switching to an envelope of notes. It is a real experiment and it found what it found. It is also one small study from a quarter of a century ago, and the twenty-five years since have produced a far more useful answer.

Why paying at all is supposed to hurt

The idea underneath it is called coupling, set out by Drazen Prelec and George Loewenstein three years before the auction.

Their argument is that a purchase is two events, not one — getting the thing, and paying for it — and that the pleasure of the first is reduced by how vividly the second is attached to it. Handing over notes is maximally attached: you see the money leave, the amount is exact, and it happens at the moment of getting the thing. A card breaks that attachment. The payment is vaguer, later, and pooled with everything else you bought this month.

Less attachment, less of what the literature calls the pain of paying, and — the theory goes — a higher price you are willing to accept.

It is an elegant explanation, and the reason it caught on is that it matches something everyone has felt. The question is how much of your actual spending it accounts for.

What seventy-one papers say when you pool them

In 2024 a meta-analysis in the Journal of Retailing gathered 392 effect sizes from 71 papers on what has become known as the cashless effect, covering cards, mobile payments and everything in between.

g = 0.135

The pooled effect of paying cashlessly rather than with cash on how much people spend, across 392 effect sizes — statistically solid, with a confidence interval of 0.068 to 0.203, and small. It is the size of difference that shows up dependably across thousands of people and is invisible inside any one shopping trip.

So the direction of the famous finding survives. Its magnitude does not. The honest sentence is not “cards make you spend twice as much”; it is “cards make you spend slightly more, on average, and you would never notice it in your own life.”

Two further things in that paper matter more than the headline number, because they tell you where the effect actually lives.

What this meta-analysis can and cannot say

The heterogeneity is very high — an I² of 86 per cent, meaning most of the variation between studies is systematic rather than noise. A pooled average across studies that disagree that much is a summary, not a law, and the moderators below are the more informative output.

What it does settle is publication bias, which the authors tested three ways and found unlikely, and the direction and rough size of the average effect. It also finds no difference between hypothetical scenarios and real purchases, which is a genuine reassurance about a literature built substantially on the former.

The effect concentrates where other people can see you

The single largest moderator is not what you are buying or how much you enjoy it. It is whether the purchase is conspicuous — visible to other people, and carrying something about you.

For conspicuous purchases the effect more than doubles, to a predicted 0.551. For everything else it sits at 0.218.

And in the other direction, the effect essentially disappears in pro-social situations — donations, tips, charitable giving — where the predicted value is 0.023, statistically indistinguishable from nothing.

That pair of findings does more to the theory than the average did. If the mechanism were purely the pain of paying, it should apply about equally to a jacket and to a tip, because the money leaves your hand the same way. It does not. The effect swells where the purchase is a statement and vanishes where it is a gift, which points at something social sitting alongside the arithmetic.

Three things people repeat that the pooled evidence does not support

Because the small-study era produced so many memorable results, a set of claims has hardened into common knowledge. The meta-analysis tested them.

“Cards make you buy junk food.” A well-known 2011 study found exactly that. Pooled across all 71 papers, though, there is no difference between hedonic purchases and utilitarian ones — the effect is the same whether you are buying dessert or dish soap. The vice-specific version is not supported at this level of evidence.
“It’s about paying later — that’s why cards are dangerous.” The meta-analysis separated payments that are decoupled in time, like a credit card billed next month, from those that are not, like a debit tap. It found no significant difference between them. Whatever is happening, it is not mainly about the delay.
“People tip and donate more with a card.” This is the pro-social finding above, and it runs the other way: cashless payment showed no reliable advantage for tips or donations at all.

None of those corrections make the underlying effect fake. They narrow it, which is what a quarter-century of accumulating evidence is supposed to do to a striking early result.

And it is getting smaller

The last finding is the one that should shape what you do about any of this. The meta-analysis tested whether the cashless effect has changed over the decades it has been measured, and found a gradual, linear decline. It is weaker now than when it was discovered.

Which is not surprising. The effect was always defined against cash as the default and the card as the departure. For most people that relationship has now inverted: tapping is the ordinary act and producing notes is the unusual one. An effect that depends on one payment method feeling more abstract than the other has less to work with every year.

The one thing cash reliably does is not about the money

There is a separate finding worth knowing, because it is the strongest practical argument for cash and it has nothing to do with restraint.

Research published in the Journal of Consumer Research found that how people paid changed their relationship to the thing afterwards: paying with cash produced greater post-transaction connection to the product than paying by card. Same object, same price, different attachment to it later.

So the useful version of “use cash” is not that it stops you buying. It is that it appears to make what you buy matter more to you once you own it — which, if you are trying to own fewer and better things, is arguably the more interesting effect.

Try this

Do not switch to cash. Instead, pick the category where the evidence says the effect is largest: something visible, something that says something about you. Clothes, a watch, a bag, the round of drinks.

For one month, pay for those in cash and everything else however you normally would.

You are not testing whether you spend less overall — the effect is too small to see in a month. You are testing whether the deliberate friction changes which of those purchases you still want two weeks later.

The parts of this that are genuinely hard

A small pooled effect is easy to misread in both directions. It does not mean the effect is unimportant: across an entire retail economy, a systematic nudge of that size moves real money, which is precisely why payment companies care about it. It does mean that it will not explain your own overspending, and that a personal budget built on the assumption of a large card premium is built on a number the evidence does not support.

The heterogeneity is also a genuine limit rather than a footnote. With an I² of 86 per cent, individual studies disagree substantially, and a pooled figure conceals that disagreement rather than resolving it. The moderators explain some of it, not all.

And none of this touches the mechanism that actually empties accounts, which is not the pain of paying but access to credit. Spending money you do not have is a different phenomenon from spending money slightly more freely, it is much better documented, and switching to cash addresses one of them.

The bottom line

How you pay does change what you are willing to pay. The direction of the famous result holds up across seventy-one papers, and the size of it does not: the pooled effect is small, it is concentrated in purchases other people can see, it is absent when you are giving rather than buying, and it has been shrinking for decades.

Which means cash is not a spending control. It is a friction you can apply selectively, in the one place the evidence says it does the most — and its clearest measured benefit is not that you buy less, but that you seem to care more about what you did buy.

Sources (5)

Every factual claim in this article comes from one of these sources.