You are on page 1of 1

VOL. 96 NO.

3 DELLAVIGNA AND MALMENDIER: PAYING NOT TO GO TO THE GYM 695

TABLE 1-EMPIRICAL FEATURES AND POSSIBLE EXPLANATIONS

Time
Trans. costs Membership inconsist. Overestimation
Standard of payment benefits per Limited with Time inconsist. of future
model per usage usage memory sophistication with naivet6 efficiency Persuasion

Finding 1
Price per average attendance Distaste of Membership Commitment Commitment, Overestimation Pressure of
= $17.27 pay per benefits overestimation of attendance salesman
usage of attendance
Finding 2
Average attendance in months Sorting at Sorting at Sorting at Sorting at Sorting at Sorting at Sorting at Sorting at
2-4 higher in annual than enrollment enrollment enrollment enrollment enrollment enrollment enrollment enrollment
monthly contract
Finding 3
Users predict 9.50 monthly Overestimation Overestimation
visits; actual monthly visits of attendance of attendance
are 4.17
Finding 4
Interval between last Distaste of Membership Forget to Overestimation Overestimation Pressure of
attendance and termination pay per benefits cancel of cancellation of cancellation salesman
2.31 full months usage
Finding 5
Survival probability after 14 Forget to Overestimation Overestimation Pressure of
months 17 percent higher cancel of cancellation of cancellation salesman
for monthly than for annual
contract

Finding 6
Average attendance 27 percent Learning, Learning, Learning, Learning, Learning, Learning, Learning, Learning,
higher in second year for sorting out sorting out sorting out sorting out sorting out sorting out sorting out sorting out
annual contract

Finding 7
Decreasing average attendance Forget to Overestimation Overestimation Pressure of
over time in monthly cancel of cancellation of cancellation salesman
contract

Finding 8
Positive correlation of price Heterogeneity in Heterogeneity in
per average attendance and naivet6 overconf.
interval between last
attendance and termination

ior. The dataset contains information both on of visits. Eighty percent of the monthly mem-
the type of membership and the day-to-day at-bers would have been better off had they paid
tendance decisions of 7,752 health club mem- per visit for the same number of visits.
bers over three years. We find that consumers Second, consumers who choose the monthly
choose a contract that appears suboptimal given contract are 17 percent more likely to stay en-
their attendance frequency. In addition, low-rolled beyond one year than users choosing the
attendance consumers delay cancelling this con- annual contract. This is surprising because
tract despite small transaction costs. monthly members pay higher fees for the option
Our empirical analysis exploits the presence to cancel each month. This result occurs even
of a contractual menu. Consumers can choose though high-attendance users sort into the an-
between two flat-rate contracts-a monthlynual contract at enrollment.
contract and an annual contract-and a pay-per- These and additional empirical findings
visit option. The monthly contract is automati- (summarized in Table 1) are hard to reconcile
cally renewed from month to month until thewith standard preferences and beliefs. We ex-
consumer cancels. The annual contract, instead,plore potential explanations, including high
expires after 12 months unless the consumertransaction costs of payment per usage, risk
explicitly renews it. The variation in the per- aversion, underestimation of costs of attendance
usage pricing and in the renewal proceduresand of cancellation, time inconsistency, naivet6
allows us to identify several puzzling features ofabout the time inconsistency, and persuasion by
consumer behavior. health club employees.
First, consumers who choose a monthly In our view, the most parsimonious explana-
membership of over $70 per month pay on tions are those allowing for overconfidence (na-
average 70 percent more than they would under ivetd). Consumers overestimate, for example,
the pay-as-you-go contract for the same number their future self-control or their future efficiency

This content downloaded from 190.119.190.104 on Thu, 06 Dec 2018 17:42:32 UTC
All use subject to https://about.jstor.org/terms

You might also like