39
was 12.1%-21.0% for credit unions without opt-in, 13.1%-21.3% for credit unions with opt-in,
4.5%-13.1% for banks without opt-in, and 14.3%-19.8% for banks with opt-in.
For sake of comparison, the last panel of Table 27 provides the distribution of overdraft fees for
all institutions with an overdraft program among the credit unions and banks in the core
processor sample
42
to the large study banks.
43
As these statistics demonstrate, a lower share of
accounts were assessed overdraft fees at the core processor credit unions and banks than at the
large study banks; however, the share of accounts with more than 10 overdraft fees was the
highest at the core processor credit unions. It is important to keep in mind that the statistics in
Panel C are conditional on an FI having an overdraft program and that the 39.1% of credit
unions and 7.1% of banks in the core processor sample that did not have an overdraft program
had no accounts with overdraft fees, by definition.
Overdraft revenue per account, as reported in Table 28, amounted to $27.36 per year on average
among observed credit unions with an overdraft program without opt-in and $52.78 among
those with opt-in, underscoring that whether an institution had opt-in is an important correlate
of overdraft revenue per account.
44
Within the dataset, per-account revenues were lower among
banks with an overdraft program, on average, than at credit unions, at least among institutions
for which we have opt-in policy information. In the dataset, banks without opt-in averaged
$22.80 in overdraft revenue per account per year while banks with opt-in averaged $44.42.
42
Note that when calculating these overall statistics in Panel C, we include institutions for which we lack opt-in policy
information, so these statistics are not simply weighted averages of the statistics in Panels A and B. This is
especially relevant for credit unions since 207 credit unions lack opt-in policy information, while only five banks
lack this information. Institutions in the dataset for which we lack opt-in policy information appear to have had a
substantially lower share of accounts with overdraft fees –4.8% of accounts at such institutions were assessed an
overdraft fee in a given year. It is possible that the core processor sample is not truly indicative of the policies and
performance of these institutions. Or, these institutions could have had a very limited overdraft program either
through their policies or by setting low overdraft coverage limits. We have no way of testing whether either of these
scenarios explains the reported results. Institutions for which we lack opt-in policy information also appear to have
substantially lower overdraft revenue per account which is consistent with the low incidence of fees.
43
Recall that the study bank data are from 2011-2012 while the core processor sample data are from 2014. The
calculations using the study bank data have been performed for this report to ensure comparability to the core
processor data and are different from the calculations in the 2014 Data Point. In the 2014 Data Point, we reported
annualized numbers for active accounts and focused on all overdrafts regardless of whether they incurred a fee. The
statistics reported by the core processors in this report are not annualized, do not use active accounts as the base,
and count only overdrafts for which a fee was assessed. For these reasons, the study bank statistics reported here
are not directly comparable to those published in the 2014 Data Point.
44
Revenue per account for an FI is calculated by taking total revenues of a given type and dividing by the total
number of accounts, here and subsequently. This means that all accounts are in the denominator, including those
that did not incur a fee of a given type.