Stefano DellaVigna and Ulrike Malmendier tracked 7,752 members across three US health clubs for three years, and published what they found in the American Economic Review in 2006. People who signed up for a flat monthly contract, paying more than $70 a month on average, went to the gym about 4.3 times a month. That works out to more than $17 per visit, at clubs that also offered a pay-per-visit pass for $10 a visit. Only 24 percent of monthly members paid less than that $10 rate. Averaged across a membership, the researchers calculated the typical monthly member was forgoing about $600 in savings they could have kept simply by paying per visit instead. This is a report on a published study, not financial advice about any particular gym contract, which will vary by club and by how often any one person genuinely expects to go.
The part that looks like an obvious mistake
On paper, this looks like a straightforward error that anyone could correct by doing basic arithmetic. Multiply your actual visits by $10 and compare it to your monthly fee, and the better deal is usually obvious. But DellaVigna and Malmendier weren’t just documenting a math mistake. They were documenting a mismatch between how often people expected to go and how often they actually went, sustained over months at a time, by people who had every opportunity to notice the gap and switch.
A second finding that runs the wrong direction
Here’s the detail that made the study more than just a story about bad math. Members who chose the monthly contract were 17 percent more likely to still be enrolled a year later than members who’d committed to an annual contract upfront, even though the monthly option cost more per period and could be cancelled at any time. If people were simply price-sensitive and reasonably accurate about their own attendance, the annual contract, cheaper per month and requiring no repeated decision to keep paying, should have kept people enrolled longer, not shorter. The pattern only makes sense if a meaningful share of monthly members kept expecting, month after month, that they’d cancel soon, and kept not doing it.
The researchers also had access to something most attendance data doesn’t include: members’ own forecasts of how often they expected to visit, collected as part of a survey when they joined. Comparing those forecasts against three years of actual swipe-card attendance records let DellaVigna and Malmendier show the gap wasn’t just a retrospective pattern in the data. People were making a specific, measurable prediction about their future selves at the moment they signed the contract, and that prediction was the thing that turned out to be wrong.
Paying for a membership after the visits stop
The data showed something else that fits the same pattern. Members frequently kept paying for a period of time after their last actual gym visit before formally cancelling their membership, rather than cancelling right after deciding they were done going. Nothing about the club’s cancellation process explained this delay on its own. It reads less like inertia and more like a genuine, if temporary, belief that the next visit was still coming.
What the researchers think is actually going on
DellaVigna and Malmendier’s own explanation centers on overconfidence, specifically overconfidence about future self-control and about how efficiently a future version of yourself will manage to get to the gym. That’s a more specific claim than simply saying people are irrational about money. Members weren’t guessing randomly about their attendance. They were consistently, systematically optimistic about a future self that would show up more often than the present self actually demonstrated it would, and they kept paying for that imagined future self well past the point where the numbers stopped supporting it.
What I’d take from this
I don’t think this study is really about gyms, even though gyms happen to be where it’s easiest to measure. A gym membership is a bet on a specific future version of yourself, one that goes consistently enough to make the monthly rate worth it, and the study is really a record of how often that bet doesn’t pay off, and how long people keep renewing it anyway before admitting the version of themselves they were counting on hasn’t shown up. The interesting question isn’t why people are bad at gym math. It’s why the belief in that future, more consistent self is so durable that people will pay hundreds of dollars to keep betting on it.
