Automated Teller Machines: Survey Results Indicate Banks' Surcharge Fees
Have Increased (Letter Report, 04/24/98, GAO/GGD-98-101).

Pursuant to a congressional request, GAO provided information on
fee-charging practices of banks and thrifts with regard to automated
teller machines (ATM) used by individuals who do not hold accounts at
the institutions owning the ATMs, focusing on GAO survey data on changes
over the past year in: (1) ATM deployment by banks in the United States;
and (2) the number of banks that surcharge noncustomers and the amounts
surcharged.

GAO noted that: (1) its survey results indicated that approximately the
same percentage of banks in the United States operated ATMs as of
February 1, 1998, as operated them as of February 1, 1997--just over 70
percent; (2) however, during this 1-year period the number of ATMs that
banks operated increased about 13 percent, to an estimated 132,000; (3)
about 66 percent of this growth was in ATMs located off bank premises;
(4) GAO's survey results also indicated that, for the month of January,
the average number of customer withdrawal transactions per ATM declined
15 percent from 1997 to 1998, while the average number of noncustomer
withdrawal transactions per ATM declined 24 percent during this period;
(5) according to GAO's survey results, the percentage of banks operating
ATMs that assessed surcharges increased from 39 percent as of February
1, 1997, to 64 percent as of February 1, 1998; (6) also, the estimated
number of ATMs with surcharges increased by more than 50 percent during
the same period; (7) GAO's estimates indicated that the average
surcharge fee assessed by banks, including in GAO's calculation the ATMs
without surcharges, increased from $0.69 as of February 1, 1997, to
$1.00 as of February 11, 1998; (8) additionally, the estimated average
surcharge fee, calculated for those ATMs with surcharge only, increased
from $1.17 to $1.27 during this period; and (9) for ATMs with
surcharges, the most typical fee as of February 1, 1998, was $1.50,
compared to $1.00 the previous year.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  GGD-98-101
     TITLE:  Automated Teller Machines: Survey Results Indicate Banks' 
             Surcharge Fees Have Increased
      DATE:  04/24/98
   SUBJECT:  Savings and loan associations
             Insured commercial banks
             Fees
             Electronic funds transfer
             Computer networks
             Surveys

             
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.  Major          **
** divisions and subdivisions of the text, such as Chapters,    **
** Sections, and Appendixes, are identified by double and       **
** single lines.  The numbers on the right end of these lines   **
** indicate the position of each of the subsections in the      **
** document outline.  These numbers do NOT correspond with the  **
** page numbers of the printed product.                         **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************


Cover
================================================================ COVER


Report to the Chairman, Committee on Banking, Housing, and Urban
Affairs, U.S.  Senate

April 1998

AUTOMATED TELLER MACHINES - SURVEY
RESULTS INDICATE BANKS' SURCHARGE
FEES HAVE INCREASED

GAO/GGD-98-101

Banks' ATM Surcharge Fees Have Increased

(233550)


Abbreviations
=============================================================== ABBREV

  ATM - automated teller machine

Letter
=============================================================== LETTER


B-279227

April 24, 1998

The Honorable Alfonse M.  D'Amato
Chairman, Committee on Banking,
 Housing, and Urban Affairs
United States Senate

Dear Mr.  Chairman: 

This report responds to your request that we update our May 1997
report Automated Teller Machines:  Banks Reported That Use of
Surcharge Fees Has Increased (GAO/GGD-97-90).  Automated teller
machine (ATM) surcharge fees are assessed by some banks and thrifts
(hereinafter referred to as banks) on noncustomers--individuals who
do not hold an account at their institution.  Our 1997 report
provided information on ATM deployment and surcharge fees. 

As agreed with your office, the objectives of this report were to
summarize our survey data on changes over the past year in (1) ATM
deployment by banks in the United States and (2) the number of banks
that surcharge noncustomers and the amounts surcharged.  We also
provide similar information on ATMs owned by selected nondepository
institutions (nonbanks). 

To address these two objectives, we conducted a statistically
representative survey of about 500 randomly selected banks throughout
the United States.  All the estimates presented in this report are
based on responses to this survey, which had a response rate of 90
percent.  In an effort to obtain more precise estimates than we were
able to include in our 1997 report, we expanded our sample from about
250 banks to about 500 banks.  We categorized banks as large, medium,
and smaller and defined these categories as banks with assets of over
$10 billion, $1 billion to $10 billion, and less than $1 billion,
respectively.  We collected information from all the banks responding
in this year's sample as of February 1, 1998.  We also collected
February 1, 1997, information from the banks we surveyed this year
and used it to revise our 1997 estimates, taking advantage of the
larger sample to provide more precise estimates.  All estimates are
subject to sampling errors, which we reported if the errors are
greater than plus or minus 10 percent.\1 In addition, reported
changes in estimates from 1997 to 1998 are statistically significant
unless otherwise noted.  (See app.  II for details on our objectives,
scope, and methodology.) We did our work between December 1997 and
April 1998 in accordance with generally accepted government auditing
standards. 

As agreed with your office, we did not include credit unions or
nonbank ATM operators in our survey.  However, we gathered
information comparable to that for banks from 7 of the 10 largest
nonbank ATM owners.  In this report, nonbank ATM owners are
nondepository institutions that own ATMs--such as grocery store
chains or other specialized companies. 


--------------------
\1 Sampling errors are the maximum amount by which the estimates from
a statistical sample can be expected to differ from the true
population values being estimated. 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :1

Our survey results indicated that approximately the same percentage
of banks in the United States operated ATMs as of February 1, 1998,
as operated them as of February 1, 1997--just over 70 percent. 
However, during this 1-year period, the number of ATMs that banks
operated increased about 13 percent, to an estimated 132,000.\2

About 66 percent of this growth was in ATMs located off bank
premises.  Our survey results also indicated that, for the month of
January, the average number of customer withdrawal transactions per
ATM declined 15 percent from 1997 to 1998, while the average number
of noncustomer withdrawal transactions per ATM declined 24 percent
during this period. 

According to our survey results, the percentage of banks operating
ATMs that assessed surcharges increased from 39 percent as of
February 1, 1997, to 64 percent as of February 1, 1998.  Also, the
estimated number of ATMs with surcharges increased by more than 50
percent during the same period.  Our estimates indicated that the
average surcharge fee assessed by banks, including in our calculation
the ATMs without surcharges, increased from $0.69 as of February 1,
1997, to $1.00 as of February 1, 1998.  Additionally, the estimated
average surcharge fee, calculated for those ATMs with surcharges
only, increased from $1.17 to $1.27 during this period.  For ATMs
with surcharges, the most typical fee as of February 1, 1998, was
$1.50, compared to $1.00 the previous year. 


--------------------
\2 Sampling error is plus or minus 12 percent. 


   BACKGROUND
------------------------------------------------------------ Letter :2

According to a Federal Reserve study\3 and other industry sources,
some banks began deploying ATMs in the early 1970s to gain an
advantage over their competitors by offering customers 24-hour access
to their funds.  Industry sources indicated that, by the early 1980s,
consumers were substantially increasing their use of ATMs.  The
Federal Reserve study found that, today, many ATMs are located at
sites where consumers would most likely want to have them; these
sites are frequently off the premises of banks, such as at grocery
and convenience stores, shopping malls, and movie complexes. 
Nonbanks are also operating ATMs at such locations. 

According to industry sources, the majority of ATM transactions
continue to be for cash withdrawals from personal checking, savings,
and other deposit accounts.  However, technological advances have
made an array of additional services available to ATM users.  Some
ATMs enable users to complete financial transactions, such as
obtaining product information, making loan payments, purchasing
insurance and investment products, accessing lines of credit, and
making payments on credit cards. 

Regional, national, and/or international ATM networks link ATMs and
computer systems, allowing the customers of one network member to
access their accounts through other network members' ATMs.  These
networks are designed to enable users to obtain ATM services over a
much wider geographical area than that covered by their own
institution. 

Until April 1996, the two primary national networks banned the
practice of surcharging noncustomers for use of their networks' ATMs. 
However, the bans could not be implemented in 15 states that
statutorily prohibited ATM networks from enforcing them.  With other
states considering similar prohibitions, the two primary networks
lifted their bans in April 1996.  Since then, some ATM operators in
states where surcharges were formerly banned by the networks have
begun assessing surcharges. 

ATM operators assess surcharge fees on the user's bank; the bank in
turn pays the fees and then assesses them directly on the user.  The
national networks require that, if the surcharge fee is assessed on
the user, the fee appear on the receipt at the cash machine as well
as on the user's bank statement. 

In addition to surcharge fees, switching and interchange fees are
generated by ATM use.  A switching fee is assessed by the ATM network
on the user's home bank to pay for processing each of its network
transactions and to defray other network operating costs, such as
advertising and network security.  Banks may absorb this fee or pass
it on to the ATM user.  An interchange fee may be assessed by ATM
operators on a nonaccount holder's home bank for handling a
transaction.  If the home bank passes this fee on to the ATM user,
the Federal Reserve's Regulation E requires the user's home bank to
inform the user--typically at the time an account is opened--that the
assessment will appear on the user's bank statement. 


--------------------
\3 Payments Primer:  Traditional Payments, No.  6, Automated Teller
Machine, Federal Reserve System, December 1997. 


   ATM DEPLOYMENT BY BANKS
------------------------------------------------------------ Letter :3

Our survey results indicated that approximately the same percentage
of banks operated ATMs as of February 1, 1998, as operated them as of
February 1, 1997.  However, during this 1-year period, the estimated
number of ATMs that banks operated increased about 13 percent.  Our
survey results also indicated that, for the month of January, the
average number of withdrawal transactions per ATM decreased from 1997
to 1998. 


      THE PERCENTAGE OF BANKS
      OPERATING ATMS REMAINED
      CONSTANT, WHILE THE NUMBER
      OF ATMS INCREASED
---------------------------------------------------------- Letter :3.1

According to our survey results, just over 70 percent of banks in the
United States operated ATMs as of both February 1, 1998, and February
1, 1997.  We estimated that approximately 8,010 of 10,960 banks
active at the time of our survey operated ATMs as of February 1,
1998, and approximately 8,210 of 11,110 banks operated ATMs as of
February 1, 1997.  The percentage of large and medium banks that had
at least one ATM was about 90 percent as of these dates.  For smaller
banks, just over 70 percent had at least one ATM as of both dates. 

According to our survey results, banks operated an estimated
132,000\4 ATMs in the United States as of February 1, 1998, a 13
percent increase over the estimated 117,000\5 ATMs they operated as
of February 1, 1997.\6 Our survey results indicated that about 66
percent of this growth was in ATMs located off bank premises.  Large
banks operated more ATMs than either medium or smaller banks as of
February 1, 1998.  (See fig.  1.)

   Figure 1:  Percentages of ATMs
   Operated by Large, Medium, and
   Smaller Banks as of February 1,
   1997, and February 1, 1998

   (See figure in printed
   edition.)

Note:  Percentages may not add to 100 due to rounding. 

Source:  GAO analysis of survey results. 

According to our survey results, for those large banks that operated
ATMs, the median\7 number of ATMs operated increased by about 100
from February 1, 1997, to February 1, 1998.  Medium and smaller banks
operated approximately the same median number of ATMs as of both
these dates.  (See table 1.)



                                Table 1
                
                Median Number of ATMs Operated by Large,
                Medium, and Smaller Banks as of February
                     1, 1997, and February 1, 1998

                                        Median number of ATMs
                                --------------------------------------
Bank size                         February 1, 1997    February 1, 1998
------------------------------  ------------------  ------------------
Large                                          345                 440
Medium                                          45                  43
Smaller                                          2                   3
----------------------------------------------------------------------
Source:  GAO analysis of survey results. 

As indicated in figure 2, an estimated 59 percent of banks with ATMs
had three or fewer ATMs as of February 1, 1998.  Our survey results
indicated that these ATM operators were nearly all smaller banks. 

   Figure 2:  Percentages of Banks
   That Operated ATMs, by Number
   of ATMs, as of February 1,
   1997, and February 1, 1998

   (See figure in printed
   edition.)

Note:  Percentages do not add to 100 due to rounding. 

Source:  GAO analysis of survey results. 

The seven nonbanks that provided us information reported that, in
aggregate, the number of ATMs they deployed increased over the past
year.  These nonbanks reported an 11 percent increase in the number
of ATMs they owned--from 9,223 ATMs to 10,277 ATMs--between February
1, 1997, and February 1, 1998. 


--------------------
\4 Sampling error is plus or minus 12 percent. 

\5 Sampling error is plus or minus 11 percent. 

\6 Point estimates are subject to a sampling error that differs from
the sampling error on our estimates of the change between 1997 and
1998.  While the confidence intervals on the point estimates may
overlap, the estimate of the change between the two years is
statistically significant.  (See app.  II for details.)

\7 The median is the middle observation--that is, half the
observations lie above this value and half lie below it. 


      THE AVERAGE NUMBER OF
      WITHDRAWAL TRANSACTIONS PER
      ATM DECREASED
---------------------------------------------------------- Letter :3.2

Our survey results indicated that, for the month of January, the
average number of customer withdrawal transactions per ATM declined
15 percent from 1997 to 1998, while the average number of noncustomer
withdrawal transactions per ATM declined 24 percent during this
period.  (See table 2.) Overall, the seven nonbanks reported little
change in withdrawal transaction volume per ATM during the same
period. 



                                Table 2
                
                    Average Customer and Noncustomer
                  Withdrawal Transactions per ATM for
                     January 1997 and January 1998

                                 Withdrawal transactions per ATM
                           -------------------------------------------
                                                               Percent
Transaction type            January 1997   January 1998         change
-------------------------  -------------  -------------  -------------
Customer                         2,551\a        2,163\b         -15%\c
 withdrawal
Noncustomer                        1,343          1,023            -24
 withdrawal
----------------------------------------------------------------------
\a Sampling error is plus or minus 16 percent. 

\b Sampling error is plus or minus 12 percent. 

\c Point estimates are subject to a sampling error that differs from
the sampling error on our estimates of the change between 1997 and
1998.  While the confidence intervals on the point estimates may
overlap, the estimate of the change between the two years is
statistically significant.  (See app.  II for details.)

Source:  GAO analysis of survey results. 


   MORE BANKS ASSESSED SURCHARGES,
   AND FEES INCREASED
------------------------------------------------------------ Letter :4

Our survey results indicated both that a greater percentage of banks
assessed surcharges as of February 1, 1998, than assessed them a year
before and that surcharge fees increased during this period.  In
particular, the percentage of ATMs that had surcharge fees of $1.50
or more increased from 21 to 40 percent. 


      MORE BANKS SURCHARGED
---------------------------------------------------------- Letter :4.1

According to our survey results, almost 64 percent of banks that
operated ATMs as of February 1, 1998, assessed surcharges on at least
some of their ATMs, while about 39 percent of the banks that operated
ATMs as of February 1, 1997, imposed surcharges.  We found that large
and smaller banks increased their use of surcharging during this
period.\8 (See fig.  3.)

   Figure 3:  Percentages of
   Large, Medium, and Smaller
   Banks That Assessed Surcharges
   on at Least Some of Their ATMs
   as of February 1, 1997, and
   February 1, 1998

   (See figure in printed
   edition.)

\a Sampling error is plus or minus 11 percent.

Source:  GAO analysis of survey results. 

The estimated number of ATMs with surcharge fees also increased.  Our
survey results indicated that from February 1, 1997, to February 1,
1998, the number of ATMs with surcharge fees increased by more than
50 percent--from about 69,100\9 to about 104,100\10 ATMs.  This
represents an increase from 59 percent to 79 percent of ATMs.  The
increase included newly deployed or acquired ATMs, as well as
existing ATMs that previously did not have a surcharge fee. 
Similarly, during the same period, the seven nonbanks we contacted
reported a 61 percent increase--from 5,056 to 8,120--in the number of
ATMs they owned that charged users an access fee.\11


--------------------
\8 The increase from 1997 to 1998 in the percentage of medium banks
that surcharged was not statistically significant. 

\9 Sampling error is plus or minus 15 percent. 

\10 Sampling error is plus or minus 15 percent. 

\11 When referring to nonbanks, we use the generic term access fee
rather than surcharge fee.  Surcharge fees refer to the access fees
that banks charge individuals who do not hold an account at their
institution.  However, nonbanks do not have deposit account holders. 


      ATM SURCHARGE FEES INCREASED
---------------------------------------------------------- Letter :4.2

According to our 1998 survey results, the average ATM surcharge fee
has increased since February 1, 1997.\12 Our estimates indicated that
the average surcharge fee assessed by banks, including in our
calculations the ATMs without surcharges, increased from $0.69 as of
February 1, 1997, to $1.00 as of February 1, 1998.\13 The average
surcharge fee as of December 31, 1995, was $0.17.\14 This was before
the two primary national networks lifted their ban on surcharge fees. 
(See fig.  4.) The maximum ATM fee reported by banks in our sample as
of both dates was $3.00. 

   Figure 4:  Average ATM
   Surcharge Fees Assessed by
   Large, Medium, and Smaller
   Banks, Including ATMs Without
   Surcharges, as of December 31,
   1995; February 1, 1997; and
   February 1, 1998

   (See figure in printed
   edition.)

Note 1:  We collected December 1995 data during our 1997 survey.  We
did not collect additional 1995 data during our 1998 survey. 
December 1995 estimates were based on the responses of banks we
surveyed in February 1997. 

Note 2:  Although some of the sampling errors for the data for
smaller, medium, and large banks were greater than plus or minus 10
percent, all of the differences between the 1995 and 1997 data within
the three size groups were statistically significant, and the
differences for the smaller and large banks between 1997 and 1998
were also statistically significant. 

Note 3:  The average ATM surcharge fees were calculated for all ATMs,
including those that did not surcharge. 

Source:  GAO analysis of survey results. 

In addition to the previously discussed average fees, the estimated
average surcharge fee assessed by banks, excluding ATMs that did not
surcharge, increased from $1.17 as of February 1, 1997, to $1.27 as
of February 1, 1998. 

Our survey results indicated that for the ATMs with surcharges, the
most typical fee as of February 1, 1998, was $1.50 compared to $1.00
for the previous year.  These results also indicated that the number
of ATMs with surcharge fees of $1.50 or more had increased since
1997.  Specifically, an estimated 21 percent of ATMs had fees of
$1.50 or more as of February 1, 1997, and an estimated 40 percent of
ATMs had such fees as of February 1, 1998.  (See fig.  5.)
Additionally, as shown in table 3, the percentage of ATMs with
surcharge fees of $1.50 or more increased for large banks but
remained relatively constant for smaller banks.\15

   Figure 5:  Percentages of ATMs
   With Various Surcharges as of
   February 1, 1997, and February
   1, 1998

   (See figure in printed
   edition.)

\a Includes all other fee levels combined.

Source:  GAO analysis of survey results. 



                                     Table 3
                     
                         Percentages of ATMs With Various
                     Surcharges, by Bank Size, as of February
                          1, 1997, and February 1, 1998

        Large bank ATMs           Medium bank ATMs         Smaller bank ATMs
    ------------------------  ------------------------  ------------------------
Su
rc
ha
rg
e          1997         1998         1997         1998         1997         1998
--  -----------  -----------  -----------  -----------  -----------  -----------
$0          40%          17%        37%\a          26%          45%          24%
 .
 0
 0
0.            3           <1            3           <1            2            1
 50
1.           31           24         40\a         42\a         31\a         47\a
 00
1.           24         58\a           16           26           19           17
 50
2.           <1           <1            3            3            1            2
 00
Al            1            1            1            4            1            8
 l
 o
 t
 h
 e
 r
 f
 e
 e
 l
 e
 v
 e
 l
 s
 c
 o
 m
 b
 i
 n
 e
 d
--------------------------------------------------------------------------------
Note:  Totals may not add to 100 percent because of rounding. 

\a Sampling errors exceed plus or minus 10 percent. 

Source:  GAO analysis of survey results. 

Consistent with the results of our bank survey, the seven large
nonbank ATM owners generally reported that they increased their
access fees.  The average fee, including in our calculations the ATMs
without access fees, increased from $0.48 on February 1, 1997, to
$0.81 on February 1, 1998.  The maximum access fee reported by these
nonbanks was $8.00.  The average fee reported, excluding ATMs that
did not charge a fee, increased from $0.87 as of February 1, 1997, to
$1.03 as of February 1, 1998. 


--------------------
\12 We calculated average surcharge fees on a per ATM basis rather
than on a per bank basis. 

\13 Actual surcharge fees tend to occur in 50-cent increments (e.g.,
$1.00, $1.50, and $2.00).  However, a commonly used measure of
surcharge fees throughout the industry is a simple average.  We
calculated the average surcharge fee in two ways:  for all ATMs,
including those with no surcharge fees, and for only those ATMs with
surcharges. 

\14 Sampling error exceeds plus or minus 10 percent. 

\15 The increase in the percentage of ATMs with surcharge fees of
$1.50 or more was not statistically significant for medium banks. 


---------------------------------------------------------- Letter :4.3

As agreed with your office, unless you announce the contents of this
report earlier, we plan no further distribution until 30 days after
the date of this letter.  At that time, we will send copies of this
report to the Ranking Minority Member of your Committee, the Chairmen
and Ranking Minority Members of other interested congressional
committees, individual members, federal agencies, and the public on
request. 


This report was prepared under the direction of Cecile O.  Trop,
Assistant Director, Financial Institutions and Markets Issues.  Other
major contributors are listed in appendix III.  Please contact either
Ms.  Trop on (312) 220-7600 or me on (202) 512-8678 if you have any
questions about this report. 

Sincerely yours,

Susan S.  Westin
Associate Director
Financial Institutions
 and Markets Issues




(See figure in printed edition.)Appendix I
SURVEY INSTRUMENT USED TO COLLECT
DATA FROM A SAMPLE OF BANKS
============================================================== Letter 



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)


OBJECTIVES, SCOPE, AND METHODOLOGY
========================================================== Appendix II

The objectives of this report were to summarize our survey data on
the changes over the past year in (1) ATM deployment by banks and (2)
number of banks (including thrifts) in the United States that
surcharge noncustomers and the amounts surcharged.  We also provide
limited information on ATMs owned by selected nondepository
institutions (nonbanks). 


   SURVEY SAMPLE
-------------------------------------------------------- Appendix II:1

To address our objectives, we estimated the number of (1) banks that
operate ATMs; (2) ATMs operated by banks; and (3) withdrawal
transactions as of February 1, 1997, and February 1, 1998.  We also
estimated the number of banks that surcharge account holders and the
amounts surcharged as of those dates.  Finally, we estimated the
change in these numbers between February 1, 1997, and February 1,
1998.  We did not include credit unions or nonbanks in our survey. 

To obtain our estimates, we conducted statistically representative
surveys of banks in both February 1997 and February 1998.  In
February 1998 we surveyed a random sample of 501 banks and thrifts
throughout the United States.  Some of these 501 institutions had
been sampled in our 1997 survey.  We asked these banks to answer the
same questions again for 1998 (and for 1997, if they had not answered
them during the 1997 survey).  We also asked a supplemental sample of
banks, drawn for the first time in 1998, to provide data for both
years.  Using the additional data covering 1997 that we obtained in
1998, we were able to make more precise, revised estimates of 1997
ATM operations for this year's report.  All of the estimates made in
this report are representative of the entire banking industry as it
existed in February 1997 and February 1998. 


   SURVEY SAMPLE DESIGN
-------------------------------------------------------- Appendix II:2

We recontacted the 197 banks that were still active from the original
stratified probability sample of 246 banks drawn in 1997.  (The bank
charters of 49 of last year's sample had become inactive, mostly
through their being acquired by other institutions.) We drew a
supplemental stratified random sample of 304 additional institutions,
for a total of 501 institutions.  The 1997 responses of the 49
inactive banks, as available, contributed only to the 1997 estimates. 

We allocated proportionately more of our sample to banks in the
population with greater assets because they, on average, operated
more ATMs.  This allowed us to cover as many of the banks operating
ATMs, and the ATMs themselves, as possible.  Because we sampled large
banks at a higher rate than other banks, the estimates for large
banks are more precise--that is, they have smaller sampling errors. 
(Sampling errors are discussed later in this appendix.) The
allocation of this year's random sample across 11 strata--defined by
the total assets that banks reported in their September 1997 Federal
Financial Institutions Examination Council Call Report (or the Office
of Thrift Supervision's Thrift Financial Reports)--is described in
the first four columns of table II.1. 



                                    Table II.1
                     
                        Disposition of Bank Survey Sample

                                            Sample disposition
                                     ---------------------------------
Bank                   1998    1998                        No  Useable
size    Sample     populati  sample  Ineligib  Refusa  respon  respons  Response
catego  strata      on size    size      le\a      ls      se      e\b    rate\c
ry      ---------  --------  ------  --------  ------  ------  -------  --------
Large   Over $10         81      81         2       1      10       68     86.1%
         billion
Medium  Over $8          16      12         0       1       2        9      75.0
         billion,
         up to
         $10
         billion
        Over $6          28      17         1       1       2       13      81.3
         billion,
         up to $8
         billion
        Over $4          54      24         0       1       1       22      91.7
         billion,
         up to $6
         billion
        Over $2         123      31         2       1       5       23      79.3
         billion,
         up to $4
         billion
        Over $1         205      28         2       1       1       24      92.3
         billion,
         up to $2
         billion
Smalle  Over $500       425      33         0       3       3       27      81.8
 r       million,
         up to $1
         billion
        Over $300       546      28         0       1       1       26      92.9
         million,
         up to
         $500
         million
        Over $150     1,431      52         1       0       3       48      94.1
         million,
         up to
         $300
         million
        Over $90      1,730      50         0       5       3       42      84.0
         million,
         up to
         $150
         million
        $90           6,374     145         0       2       4      139      95.9
         million
         or less
================================================================================
Total                11,013     501         8      17      35      441     89.5%
--------------------------------------------------------------------------------
\a Not a depository institution, no longer in business, merged, or
otherwise closed. 

\b Useable responses included not only completed questionnaires from
banks with ATMs, but also the determination during telephone
precontact that a bank had no ATMs. 

\c The response rate was calculated as the number of banks completing
useable questionnaires divided by the number of eligible banks in the
sample (original sample minus ineligibles). 

Source:  GAO survey. 

The samples in both years were drawn from the populations of banks
identified as active institutions in the September 1996 and September
1997 Call Report databases obtained from the Federal Deposit
Insurance Corporation.  Although neither we nor the agencies that
produced the source data have fully assessed the reliability of this
database, Call Report data are widely used by researchers in
academia, government, and private industry. 


   QUESTIONNAIRE DESIGN
-------------------------------------------------------- Appendix II:3

We administered the same questionnaire that we used last year, after
making some minor modifications and adding questions on ATM customer
transactions and merger activity.  Each questionnaire was to be
filled out by a separate, individually chartered institution. 
Respondents were instructed to provide answers, to the extent
possible, for the offices of the one institution named on the
questionnaire only, and were instructed not to include any bank
holding company parent or other subsidiaries.  Because we extensively
pretested the questionnaire before administering the survey last
year, we asked only a small number of banks that were to be in our
sample to review our new questions before we administered the revised
questionnaire. 


   SURVEY ADMINISTRATION
-------------------------------------------------------- Appendix II:4

During the first 2 weeks of January 1998, we contacted all sampled
institutions by phone.  Following a written calling protocol, we
determined whether each of the 501 sampled banks was a currently
active institution and, if so, whether it operated any ATMs.  For
banks not operating at least one ATM, this information constituted a
complete response.  For those banks operating ATMs, we then
identified the most appropriate respondent and mailed that person a
questionnaire at the end of January 1998.  After making multiple
calls to encourage the return of all outstanding questionnaires and
following up on some questionnaires that contained missing or
inconsistent data, we ended our fieldwork in the second week of March
1998. 


   SURVEY RESPONSE
-------------------------------------------------------- Appendix II:5

We received useable responses (an indication that no ATMs were
operated, or answers to at least some of the questions on the
questionnaire mailed to ATM operators) from 441 sampled banks, for a
response rate of approximately 90 percent.  (See table II.1.) No
systematic differences in the rate of survey participation across
different bank size strata were detected. 


   CALCULATION OF SURVEY ESTIMATES
-------------------------------------------------------- Appendix II:6

After weighting survey responses to account for selection
probabilities and nonresponse, we were able to make estimates of the
number of banks operating ATMs, the number of ATMs, and surcharging
characteristics for the entire population of banks active as of
February 1, 1998.  Using 1997 survey data and retrospective reports
of other banks responding for the first time this year, we were also
able to update estimates for the entire population of banks as of
February 1, 1997. 

Although the sample design of this survey incorporates elements of a
panel survey design, where individual sample elements are tracked
over time, estimates of change between the two years are based on the
differences between the overall estimates for each of the two years
and not on changes within particular institutions.  However,
information about change at the individual bank level, where 1997 and
1998 data for one bank active in both years exists, helped us to more
accurately determine the precision of these change estimates, as
described later in this appendix. 


   SAMPLING ERROR
-------------------------------------------------------- Appendix II:7

Because we reviewed a statistical sample of banks, each estimate
developed from the sample (for either the 1997 or 1998 point
estimate, or the change between these two estimates) has a measurable
precision, or sampling error.  The sampling error is the maximum
amount by which the estimate obtained from a statistical sample can
be expected to differ from the true population value being estimated. 
Sampling errors are stated at a certain confidence level--in this
case, 95 percent.  This means that the chances are 19 out of 20 that,
if we surveyed all of the banks in the population, the true value
obtained for a question on this survey would differ from the estimate
obtained from our sample by less than the sampling error for that
question. 

The sampling errors account for the stratified sample design and are
reported for all of our 1997 and 1998 estimates unless they are less
than plus or minus 10 percentage points, or 10 percent of the
quantity estimated.  All differences cited are statistically
significant unless otherwise noted.  That is, any differences in ATM
characteristics between subgroups of banks, or in some industry
characteristic over time, can be considered present in the underlying
population and not simply due to chance introduced by relying on one
sample of the whole population. 

Our analysis also accounted for the association between the status of
individual banks in 1997 and 1998--that is, it accounted for the fact
that bank characteristics, such as the number of ATMs operated, tend
to change in a predictable way for individual banks.  As a result, we
could more precisely estimate changes in bank characteristics than
would be expected given the large sampling errors of some of the
point estimates.  Thus, the point estimates of a characteristic in
1997 and 1998 might have large sampling errors, while the estimate of
change for that characteristic between the two years might have a
relatively small sampling error. 


   NONSAMPLING ERROR
-------------------------------------------------------- Appendix II:8

In addition to the reported sampling errors, the practical
difficulties of conducting any survey may introduce other types of
"nonsampling" errors.  For example, differences in how a particular
question is interpreted, the sources of information that are
available to respondents, or the types of banks that do not respond
can all introduce unwanted variability into the survey results. 
Although we did not verify respondents' answers, we did include steps
in the data collection and analysis processes to minimize nonsampling
errors.  Specifically, we modified our questions based on pretests
and our experiences from the 1997 survey to make them more
understandable and easier to answer.  We also checked respondent's
answers for internal consistency and out-of-range values, and we
attempted to obtain missing data.  Finally, we conducted callbacks to
encourage a high level of response that would reduce the effects of
any nonresponse bias. 

In addition, our surveys were conducted in February 1997 and February
1998, while the Call Report data that we used to draw our sample was
dated 4 months earlier.  Although we incorporated revised Call Report
data released by the Federal Deposit Insurance Corporation in
December of 1997 into our 1998 sample frame, changes to the banking
industry (such as mergers resulting in the discontinuance of some
bank charters and the issuance of new charters) between the dates our
sample frames were defined and the dates our surveys were
administered did occur.  Those changes resulting in a decrease in the
number of chartered banks in the population were reflected in our
survey analysis by those banks we sampled and found ineligible due to
recent mergers or closings.  However, to the extent that newly
chartered institutions had been created in the period between the
definition of our sample frame and the survey, those increases in the
population would not be reflected in our sample, resulting in a small
but unknown amount of under coverage in our sample of the true
population of banks in February 1997 and February 1998.  We performed
tests on two revisions of the September 1997 Call Report database to
assess the magnitude of changes in listings over time. 

Also, banks that were sampled for the first time in 1998 and reported
that they did not operate ATMs were not asked for 1997 data.  We
assumed that these 64 banks did not operate ATMs in 1997, based on
reports from another group of 27 banks that did not operate ATMs in
1998 but had been sampled the year before.  Only 2 of those 27 banks
had operated ATMs the previous year. 


   INFORMATION ON ATMS OWNED BY
   NONBANKS
-------------------------------------------------------- Appendix II:9

We gathered information comparable to that for banks from 7 of the 10
largest nonbank owners, excluding credit unions, to obtain an
indication of nonbank practices.\16 This information could not be
projected to the universe of nonbank ATM owners because we did not
draw a statistical sample from this universe.  We were unable to draw
a statistical sample because the complete data necessary for defining
the universe was not available. 


--------------------
\16 The 10 largest nonbank ATM owners were identified from the top
350 ATM owners listed in Credit Card Management:  Debit Card
Directory, 1998 Edition (New York, NY:  Faulkner and Gray, Inc.,
1997).  Seven of the 10 provided us with information on their ATM
operations. 


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================= Appendix III

GENERAL GOVERNMENT DIVISION,
WASHINGTON, D.C. 

Kristi A.  Peterson, Evaluator-in-Charge
Carl M.  Ramirez, Senior Social Science Analyst
James M.  Fields, Senior Social Science Analyst
Stuart M.  Kaufman, Senior Social Science Analyst
Joanne D.  Meikle, Senior Social Science Analyst
George H.  Quinn, Jr., Computer Specialist
Sidney H.  Schwartz, Senior Social Science Analyst

CHICAGO FIELD OFFICE

Barry A.  Kirby, Evaluator

SAN FRANCISCO FIELD OFFICE

Tre M.  Forlano, Evaluator
May M.  Lee, Evaluator
Grace K.  Sakoda, Evaluator

*** End of document. ***