Evaluation of the application of Regulation E to EBT systems final report

              * volH-^'Ol fl%2 '<€{/ //*/?**>*<-
USDA
United States
Department of
Agriculture
Food and
Consumer
Service
Office of
Analysis and
Evaluation
Evaluation of the
Application of
Regulation E to EBT
Systems
Final Report
July 1997
I g^i^ggg
USDA
Unitad States
Dapartmant of
Agriculture
Food and
Consumer
Service
3101 Park
Center Drive
Alexandria. VA
22302 1500
Enclosed for your information is a copy of Evaluation ofthe Application ofRegulation
E to Electronic Benefit Transfer (EBT) Systems. It provides the first empirical data on the
likely costs of Regulation E for food stamp and other cash assistance EBT systems.
While legislation (P.L. 104-193) subsequent to the study exempts the Food Stamp
Program from Regulation E, the information remains pertinent to direct Federal benefit
programs, like Supplemental Security Insurance. There are also some implications for
how existing client protections are administered in EBT systems.
If you have questions about the content of this report or need additional copies, please
contact the following:
Office of Analysis and Evaluation
Food and Consumer Service
3101 Park Center Drive
Alexandria, Virginia 22302
Phone (703) 305-2133
Enclosure
AN EQUAL OPPORTUNITY EMPLOYER
USDA United States
Department of
Agriculture
Food and
Consumer
Service
Evaluation of the Application of
Regulation E to EBT Systems
Office of
Analysis and
Evaluation
FINAL REPORT
July 1997
Authors:
John Kirlin
Nancy Burstein
Stefanie Gluckman
Julie Masker
Laura Peck
Submitted by:
Abt Associates Inc.
55 Wheeler Street
Cambridge. MA 02138
Project Director: John Kirlin
Submitted to:
Office of Analysis and Evaluation
USDA Food and Consumer Service
3101 Park Center Drive, Rm. 214
Alexandria, VA 22302
Project Officer: Carol Olander
This study was conducted under Contract No. 63-3198-4-027 with tha Food and Consumer Sorvico, United States
Department of Agricutturo. under the authority of the Food Stamp Act of 1977, as emended. Points of view or opinions
statad in this report do not naressarty rapresant tha official position of tha Food and Consumer Servica.
ACKNOWLEDGEMENTS
On behalf of the authors, I would like to thank the many individuals who assisted us
in the preparation of this report. They represent many organizations and have contributed in
many different ways.
First and foremost, the officials in each demonstration site have been very cooperative
and patient with the evaluation team. I therefore wish to thank John Waller, Dick Woods, and
Marlee Torres in New Mexico; David Heins, Bonnie Mecanko, and Bob Knapp in New Jersey;
Nancy Celaya and David Larsen at Deluxe Data Systems; and John Simeone and Alma Parrish
at Citibank EBT Services. They and their colleagues have assisted us a great deal.
Two individuals at the Food and Consumer Service deserve special mention. Carol
Olander, the evaluation's Project Officer, and Karen Walker, the demonstrations' Project
Officer, have been most helpful to us throughout the evaluation.
Special thanks also go to members of the Reg E/EBT Stakeholder Panel, which assisted
the evaluation in identifying research questions, helping interpret evaluation findings, and
reviewing and commenting on a draft of this report. The panel members were Dr. Howard
Bloom, Ms. Dale Brown, Mr. Gerald Hurst, Ms. Adele Latourette, Ms. Carrie Lewis, Ms.
Barbara Leyser, Mr. Thomas McLaughlin, Ms. Michele Meier, and Ms. Margaret Philben. In
addition, Ms. Brown and Ms. Philben, project directors for EBT projects in Maryland and
Ramsey County, Minnesota, have been most valuable in helping the evaluation obtain
information on claims of EBT loss and outcomes in their respective sites.
Mr. George White, president of White Papers Inc. (a subcontractor for the evaluation),
has been most helpful with his industry perspective of Regulation E and his review of the draft
of this report.
Finally, and with apologies to many other individuals whom I have surely overlooked,
I wish to thank the other members of the evaluation's project team who contributed to this
report. The project's technical reviewer was Chris Hamilton, and Diane Stoner surved as the
project's survey director. Carlos Gandiaga provided computer programming support, and
Jonathan Yeung supervised many aspects of data receipt and coding. Carissa Climaco analyzed
the Ramsey County ATM misdispense data referenced in Chapter Three and discussed in
Appendix F. Finally, Susan Byers, Jan Nicholson, and Stefanie Falzone produced this report
and its graphics.
John Kirlin
Project Director
Prepared by Abt Associates Inc.
TABLE OF CONTENTS
EXECUTIVE SUMMARY i
Chapter One: INTRODUCTION 1
1.1 EBT Systems and Client Protections Against Benefit Loss 3
1.2 Reg E and EBT Systems 5
1.3 The Reg E Demonstrations 7
1.4 The Evaluation of the Demonstrations 10
1.5 Organization of the Report 12
Chapter Two: PLANNING FOR AND IMPLEMENTING REGULATION E 13
2.1 Establishing General Policy 13
2.2 Claim Handling Procedures 21
2.3 Implementation "«sks 28
2.4 Conclusions 32
Chapter Three: THE IMPACT OF REGULATION E ON BENEFIT CLAIMS 35
3.1 Data Sources and Research Approach 36
3.2 Components of Reported and Actual Loss 38
3.3 Frequency of Claims of Lost Benefits 41
3.4 Unreported Incidents of Lost Benefits 48
3.5 Conclusions 56
Chapter Four THE IMPACT OF REGULATION E ON BENEFIT REPLACEMENT 59
4.1 Disposition of Claims 60
4.2 Reasons for Denial 64
4.3 Provisional Credits 71
4.4 Liability from Benefit Replacements 76
4.5 Conclusions 80
Chapter Five THE IMPACT OF REGULATION E ON ADMINISTRATIVE COSTS 83
5.1 Data Sources and Research Approach 84
5.2 Administrative Costs of EBT and Reg E Claims 88
5.3 Reg E Start-up Costs 104
5.4 Conclusions 105
Prepared by Abt Associates Inc.
)(/
Table of Contents
Chapter Six PROJECTIONS OF DEMONSTRATION RESULTS 107
6.1 Total Claim-Related Costs 108
6.2 Projections of Claim Rates 109
6.3 Projections of Liability 120
6.4 Projections of Administrative Costs 121
Chapter Seven LESSONS FROM THE REG E DEMONSTRATIONS 131
7.1 Proposed Principles and Goals 131
7.2 Lessons Learned 134
Appendix A: Procedures for New Jersey's Regulation E Demonstration
Appendix B: Procedures for New Mexico's Regulation E Demonstration
Appendix C: Procedures for Citibank's Regulation E Demonstration
Appendix D: Claim Handling Procedures in Camden County
Appendix E: Impact of Reg E on Benefit Replacement: Supplementary Exhibits
Appendix F: Incidence of ATM Misdispenses in Ramsey County, 1992-1994
Appendix G: Analysis of New Mexico's EBT Project Problem Reports
Appendix H: Reg E Claimant Survey
Appendix I: The Survey of Unreported Loss
Appendix J: Administrative Cost Methods and Data
Appendix K: Caseworker Survey Results
Prepared by Abt Associates Inc.
EXECUTIVE SUMMARY
Electronic benefit transfer (EBT) systems have been implemented in a number of
different states across the country. These systems deliver benefits electronically for a number
of state, state-administered federal, and direct federal programs. State programs using EBT to
deliver benefits include General Assistance, and direct federal programs using EBT include
Social Security (Old Age, Survivors, and Disability Insurance, or OASDI), Supplemental
Security Income (SSI), and other federal retirement and disability programs. By far the largest
users of EBT systems to date, however, are the state-administered assistance programs; these
include the Food Stamp Program (FSP), the Aid to Families with Dependent Children (AFDC)
program, and the new Temporary Assistance for Needy Families (TANF) programs.
EBT systems work very much like commercial bank card networks. Program
participants receive an EBT card and select a personal identification number, or PIN. The EBT
card is functionally similar to a bank debit card. Using the EBT card and PIN, the EBT
cardholder can access cash assistance program benefits either by withdrawing them from an
automated teller machine (ATM) or by using them at the point of sale (POS) to make purchases
or to receive cash back. For the FSP, the EBT card can be used to access food stamp benefits
to pay for purchases in program-authorized food retail outlets.
REGULATION E AND EBT SYSTEMS
The Electronic Fund Transfer Act governs the operations of commercial debit card
networks. A regulation commonly referred to as "Regulation E" implements the provisions of
the Act. Regulation E (or simply "Reg E") establishes a framework of legal rights and
responsibilities for card issuers and card holders in electronic fund transfer systems. In March
1994, the Board of Governors of the Federal Reserve ruled that Reg E must be applied to all
EBT systems by March 1997. Although EBT systems serving beneficiaries of direct federal
programs have always operated under the provisions of Reg E, the Board's ruling had several
major implications for EBT systems delivering state-administered prograr i benefits. Specifi-cally,
these EBT systems would now have to:
Prepared by Abt Associates Inc.
W
Executive Summary
• Cap a client's liability for benefits lost through unauthorized use of the EBT card
at $50 if the client reported the loss within two days of discovery. (As card
issuer, the state would be liable for the remaining lost benefits.)
• Issue a provisional credit for the loss amount (minus any client liability) if a claim
could not be fully investigated within a specified time period. If the claim was
subsequently denied, the state would have to initiate recoupment proceedings to
recover the provisional credit.
• Issue a disclosure statement explaining the rights and responsibilities of the state
and the client in an EBT system, and explaining how to go about filing a claim for
lost benefits.
The Board's decision to extend the provisions of Reg E to EBT was controversial.
Client advocates supported the decision, asserting that households receiving public assistance
should have the same protections against debit card loss as anyone. Many federal and state
proponents of EBT systems, however, believed that regular program protections against EBT
loss were sufficient. These protections reimbursed clients for losses due to ATM misdispenses
and many system or procedural errors. The EBT proponents worried that the potential cost of
replaced benefits and claims processing under Reg E would increase the overall cost of EBT
services 10 the point where EBT would no longer be a cost-effective alternative to paper benefit
delivery.
With passage of the Personal Responsibility and Work Opportunity Reconciliation Act
of 1996,' Congress overturned the Board's decision and exempted most EBT systems from the
provisions of Reg E. Systems delivering direct federal program benefits still must operate
under Reg E, but any EBT system operated by a state or county unit or delivering benefits for
a state or state-administered program is exempt from the regulation. These state-administered
sy'terns, however, continue to provide client protections against loss due to ATM misdispenses
and many system or procedural errors, as before.
THE REG E DEMONSTRATIONS
In an effort to provide empirical evidence on the impacts of applying Reg E to EBT
systems, federal and state agencies used the Federal Reserve's three-year implementation period
1 Public Law 104-193.
Prepared by Abt Associates Inc. u
Executive Summary
to conduct a series of demonstrations in which several sites with EBT systems operated under
Reg E provisions for 12 months. The purposes of the demonstrations were to:
• learn more about the likely impacts of Reg E on administrative costs and benefit
replacements;
• assess the effectiveness of different strategies for implementing Reg E and control-ling
claims of benefit loss; and
• prepare funding plans for any costs associated with the application of Reg E.
Exhibit 1 provides an overview of the demonstration sites. As one reads down the
exhibit, the sites are listed in general ascending order of the protections they provided clients
against loss of benefits. For instance, the comparison site of Camden County, New Jersey, did
not offer any "Reg E" protections against losses due to unauthorized card usage. As in all
current EBT sites, however, Camden's EBT operating policy was to reimburse clients fully for
any verified losses they incurred due to ATM misdispenses or system errors.
Exhibit 1
OVERVIEW OF DEMONSTRATION SITES
Site
Level or
Protection
Programs
Served
Average
Monthly
Caseload*
Camden County, NJ Regular EBT AFDC. FSP 22.740
San Juan County, NM Responsibility
Standard
AFDC. FSP 3.514
Citibank DPC System (TX) Full Reg E OASDI, SSI, others 12.405
Hudson County, NJ Full Reg E AFDC, FSP 28.456
Bemalillo County. NM Full Reg E AFDC. FSP 24.703
Dona Ana County, NM Full Reg E AFDC. FSP 10.259
* Unduplicated case count (i.e., households receiving both food stamps and AFDC are counted just once).
The protections offered clients in San Juan County, New Mexico, were nearly identical
to those offered in Camden County. San Juan County, however, participated in the Reg E
demonstrations as a "responsibility standard" site. This meant that losses due to unauthorized
card usage were not reimbursed if the transaction in question was initiated with a valid EBT
Prepared by Abt Associates Inc. ui
4///{
Executive Summary
card and PIN.2 Reg E provisions regarding how quickly claims of loss must be investigated,
however, were in effect in San Juan County, as was the requirement that provisional credits be
granted if investigations could not be completed before the Reg E deadlines.
The last four sites—the Citibank DPC system in Texas, Hudson County in New Jersey,
and Bernalillo and Dona Ana counties in New Mexico—operated under "full" Reg E protections
during the demonstration periods. In these four sites, losses due to unauthorized card usage
were reimbursable if the client cooperated with the investigation and the circumstances of the
loss could be verified. Furthermore, provisional credits were granted when investigations could
not be completed widiin 10 days (for losses at an ATM) or 20 days (for losses at a POS
device). Citibank's DPC system was the only demonstration site serving direct federal
programs like Social Security and SSI. As such. Citibank was the only system operating with
previous Reg E experience.
KEY FINDINGS
This report presents the findings from an evaluation of these demonstrations. The
principal findings are:
• Reg E had no consistent impact on the number of claims submitted. Claim
submission rates, although generally low in all sites, were higher in some Reg E
sites than in Camden County, but lower in other Reg E sites.
• Reg E's impact on liability arising from replaced benefits was quite small. For
both the cash assistance and food stamp programs, liability averaged $0.03 or less
per case month in all but one demonstration site, where it averaged $0.09 per case
month for cash assistance claims.3 Liability arising from unrecovered provisional
credits (for claims subsequently denied) was even smaller. This liability averaged
$0.01 or less per case month in each demonstration site.
2 As in all sites, any losses due to unauthorized card usage after the card had been reported as lost or
stolen were reimbursable.
3 Cost impacts are measured "per case month" throughout the report. They are calculated by dividing
the impact (here, total replaced benefits in a site) by the sum of the number of cases active during each month
of the demonstration.
Prepared by Abt Associates Inc. iv
Executive Summary
• If factors affecting liability rates remained the same nationwide as in the demon-strate
.1 sites, projected AFDC/TANF liability for replaced benefits across all SO
states would be $1.8 million annually, and food stamp liability would be
$722,000 annually.
• Reg E increased AFDC administrative costs considerably. Administrative costs
were $0.11 to $0.63 per case month higher in the Reg E sites than in Camden
County, where the cost of existing client protections was $0.37 per case month.
(As a point of comparison, total monthly costs to operate an EBT system have
ranged between $3.00 and $4.50 per case, depending on the state.)
• Reg E's impact on food stamp administrative costs was smaller and less consis-tent.
Compared to food stamp administrative costs oi $0.33 per case month in
Camden County, Reg E costs ranged from being $0.10 per case month higher to
$0.23 per case month lower.
• Total projected Reg E administrative cost for all SO states varies between $14-$22
million annually for the AFDC/TANF programs and $21-$42 million annually
for the FSP. These projections use the demonstration sites as alternative models
for nationwide implementation of Reg E, with some recommended changes in
staffing patterns to reduce costs.
With Congress' exemption of many EBT system from the provisions of Reg E, the
usefulness of the Reg E demonstrations may appear limited. EBT systems serving direct federal
programs continue to be covered by Reg E, however, and all EBT systems offer some
protections against benefit loss to clients. In addition, it may be possible to offer some added
protections to clients without substantial increases in administrative costs. Thus, a need still
exists to be able to process claims of loss effectively and efficiently. We therefore point out
below several other lessons from the Reg E demonstrations:
• The cost of client protections need not be as expensive as the administrative costs
incurred in the Reg E sites. Different staffing patterns and organizational
structure could reduce costs substantially, while still maintaining service levels.
• Even without a Reg E requirement, EBTsystems can impose substantial costs on
local offices in unexpected ways. For instance, instead of reporting EBT account
problems to the EBT Help Desk (as instructed), many recipients in Hudson and
Camden County contacted their caseworkers to ask questions. These contacts
imposed large costs on local office operations.
• More than SO percent of the losses reported by clients could, in theory, be
avoided. In particular, if recipients were more careful about protecting their EBT
Prepared by Abt Associates Inc.
X
Executive Summary
cards and PINs, then losses due to unauthorized card usage could be reduced.
This would not only help clients, but it would also reduce sites' costs to investigate
the losses.4
The following sections provide more detailed discussion of what has been learned from
the Reg E demonstrations with regard to Reg E's impacts on benefit liability and administrative
costs, as well as lessons for providing client protections in the future.
IMPACT OF REG E ON BENEFIT LIABILITY
The evaluation grouped nearly all claims of lost or stolen benefits into three main
categories:
(1) claims arising from non-receipt offunds (i.e., ATM misdispenses);
(2) claims arising from unauthorized usage of a client's EBT card; and
(3) claims arising from system or procedural errors (e.g., a transaction mistakenly
entered twice at a store POS terminal).
Within each of these categories, the study examined the rate at which claims were submitted
(expressed as the number of claims submitted per 1,000 ewe months of benefit receipt), their
disposition, reasons for deniJ, and the resulting impact on lability due to replaced benefits.
A major concern prior to the demonstrations was that Reg E would increase state or
county financial liabilities by an amount sufficient to render EBT systems no longer cost-effective.
Increased liabilities could arise from two sources: program benefits replaced
following approval of a claim of lost benefits, and unrecovered provisional credits. As shown
in Exhibit 2, however, liabilities from approved claims and unrecovered provisional credits
were quite low in all demonstration sites.
There are four reasons why the benefit liabilities shown in the exhibit were generally
less than $0.03 per case month. First, claim submission rates were low in all sites. Second,
most approved claims of benefit loss do not impose a financial liability on the state or county.
4 Even when losses due to unauthorized card usage are not reimbursable, some administrative time (and
cost) is incurred to deter nine that the loss is indeed due to unauthorized card use and not some other factor.
Prepared by Abt Associates Inc. vi
A
Executive Summary
Exhibit 2
FINANCIAL LIABILITY FROM CLAIMS OF LOST BENEFITS
(dollars per case month)'
Level of
Protection
Regular
EBT
Responsi-bility
Standard Full Reg E
Site
Camden
County
(NJ)
San Juan
County
(NM)
Citibank
DPC
System
(TX)
Hudson
County
(NJ)
Beraalillo
County
(NM)
Dona Ana
County
(NM)
All Full
RegE
Sites
Cash Anrrtmrr Benefits
Approved
claims
0 0 .016 .006 .088 .028 .027
Provisional
credits
0 .001 0 .001 .016 .009 .004
Food Stamp Benefits
Approved
claims
0 0
a/a
.000 .017 .000 .007
Provisional
[credits
0 0 0 .001 0 .000
* A value of "0" indicates zero cost. A value of '.000* indicates a positive cost equal to less than SO.0005 (1 -0th of a
cent) per case month.
n/a Not applicable Food stamp benefits are not issued through the Citibank DPC system.
Only approved claims of unauthorized usage impose additional liability.5 Third, approval rates
for claims of unauthorized usage were low. Finally, exposure from provisional credits was low
because relatively few provisional credits were granted. If these factors affecting liability rates
remained the same nationwide as in the demonstration sites, then projected liability for replaced
5 Approved claims of ATM misdispenses do not generate a financial liability because the credit to the
client's account is offset by a credit from the ATM owner. Similarly, approved claims of system or
procedural error usually do not create a financial liability for the state or county agency or EBT vendor. In
those few instances in which an approved claim arising from a system or procedural error does create a
financial liability, the liability would have been incurred under standard EBT operating rules as well as under
Reg E, so Reg E generates no additional liability. An example would be transactions approved by the system
after the client had properly informed system representatives that his or her EBT card had been lost or stolen.
Prepared by Abt Associates Inc.
jai
Executive Summon/
AFDCTANF benefits would be $1.8 milboD annually across all SO stales, and food stamp
liability at the national level would be $722,000 annually.6
IMPACT OF REG E ON ADMTSTSTRATTVF. COSTS
The administrative costs of investigating and processing Reg E claims in the Reg E
demonstration sites were compared to the cost of investigating and processing claims of lost
EBT benefits in Camden County. The Reg E administrative costs in each site were substantial,
especially when compared to the site's costs of benefit replacements and unrecovered provi-sional
credits. The cost of helping recipients in Camden County with their EBT account
problems, however, also was substantial.
Demonstration Costs
For both the cash assistance and food stamp programs. Exhibit 3 presents the evalua-tion's
estimates of average administrative costs per submitted claim and per case month during
the demonstration periods. Average cost per claim is high in each site, but vanes substantially
across sites. At an average cost of $98 per cash claim. Citibank has the lowest cost per claim.
perhaps due to its previous experience in handling Reg E claims, but also because participants
in the DPC system do not have access to other staff (e.g.. caseworkers) to help with EBT
problems. The three counties in New Mexico had higher costs, with average per-claim costs
for cash assistance claims ranging from $188 to $357. Average costs te food stamp claims in
New Mexico ranged from $168 to $831 per claim.
The two New Jersey counties had the highest average costs per claim, but for different
reasons. Caseworkers in both Camden and Hudson County spent considerable time In \\m%
clients with real or perceived problems of lost benefits. For Camden County, this caseworker
time was the major contributor to average per-claim costs of $437 and $1,020 for claims
involving AFDC and food stamp benefits, respectively Caseworkers in Hudson County had
lower average salaries than their counterparts in Camden County, so the costs vi Hudson
6 These projections are based on a projected national AFDC claim rate dm is 18 percent higher than the
average demonstration rase, and a projected national food stamp claim rate that is 12 percent lover. The
claim rate projecoons adjust for differences in caseload composition between the dexnonstranon naes and U.S.
averages.
Prepared by Abt Associates Inc.
Executive Summary
Exhibit 3
AVERAGE ADMINISTRATIVE COSTS
Level of
Protection
Ziegular
EBT
Responsi-bility
Standard Full RegE
Site
Caraden
County
San Juan
County
Citibank
DPC
System
Hudson
County
Bernalillo
County
Dona
Ana
County
All Full
RegE
Sites
Cask Assistance Benefits
| Cost per claim
Q (actual)
$437 $188 $98 $1,144 $342 $342 $357
U Cost per case
| month (actual)
$0,369 $0,587 $0,330 $0,999 $0,733 $0,478 $0,691
| Cost per case
|| month (projected)' $0,369 $0,305 $0,330 $0,262 $0,417 $0,307 $0,319
Food Stamp Bent fits
Cost per claim
(actual) $1,051 $168
n/a
$1.3)7 $378 $831 $582
Cost per case
month (actual)
$0,326 $0,101 $0,328 $0,426 $0,184 $0,344
Cost per case
month (projected)' $0,326 $0,059 $0,164 $0,263 $0,161 $0,203
* Projected costs in New Mexico assume the Reg E project director and Reg E coordinator are replaced by a full-time staff
member at the Help Desk. Projected costs in Hudson County assume that investigators' time spent waiting for clients to
arrive can be spent productively on non-Reg E activities. No changes in staffing or procedures are assumed for Camden
County or Citibank's DPC system.
n/a Not applicable. Food sump benefits are not issued through the Citibank DPC system.
County are driven instead by investigation time. Reg E investigators in Hudson County worked
an average of 30 hours per claim, compared to fewer than six hours per Reg E claim elsewhere.
A possible reason for this differential effort is that Hudson County was the only site to require
a face-to-face meeting with claimants. Although this policy may have contributed to the low
claim submission and approval rates in Hudson County, it also required the full-time presence
of an investigator to meet with clients. This time-consuming approach contributed to average
administrative costs in Hudson County of $1,144 and $1,317 per AFDC and food stamp claim,
respectively.
Prepared by Abt Associates Inc. IX
Executive Summary
Some of the administrative costs in Hudson County were relatively fixed with respect
to number of claims submitted; this is true of investigation costs because staffing levels were
established before the start of the demonstrations. This fixed component helps explain the high
average cost per claim in Hudson County, where claim rates were low. Indeed, when adminis-trative
costs are measured on a per-case-month basis rather than per claim, cost differences
across sites decline. As shown in Exhibit 3, Citibank still has the lowest average cost for
claims involving cash assistance benefits, but now Hudson County's AFDC claim cost of $0,999
per case month is only three times higher than Citibank's cost of $0,330 per case month, not
close to 12 times higher, as was the case with per-claim costs.
When compared to costs in Camden County, Reg E increased administrative costs for
the cash assistance programs more so than for the Food Stamp Program. This differential
impact on program costs is due to two reasons. First, claim rates for food stamp benefits were
much lower than for cash assistance benefits in all sites, but more so in the Reg E sites than in
Camden County. Compared to the Reg E sites, therefore, Camden County had relatively more
food stamp claims to handle and investigate, narrowing the cross-site difference in food stamp
administrative cosis. Second, caseworker costs have been allocated across programs in
proportion to caseload size. Because caseworker costs represented a larger share of Camden
County's administrative costs than in any other site, this increased food stamp costs in Camden
more than in the Reg E sites, again reducing cross-site differences in administrative costs.
Projected Costs
The last row of each section of Exhibit 3 presents projected Reg E costs under different
assumptions about staffing plans. As shown in the exhibit, the projected costs for the Citibank
DPC system do not change; Citibank's staffing patterns and claim investigation procedures were
already responsive to workload charges. The lower projected costs for the three New Mexico
counties arise from the EBT project director's plan to more thoroughly integrate Reg E
processing with Help Desk procedures had Reg E become mandatory for all EBT systems.
Finally, although Hudson County staff indicated that staffing patterns would not have changed
had Reg E become permanent, the projected costs in the exhibit assume that the previously-mentioned
time spent waiting to meet with clients could have been spent productively on non-
Reg E tasks.
Prepared by Abt Associates Inc.
Executive Summary
If the per-case-month projections in Exhibit 3 are adjusted to a common claim rate and
projected nationally, the projected total administrative costs for cash assistance claims vary
between $6.4 and $22.3 million annually, with the lower projection based on the San Juan
County and Citibank DPC system experiences. Total projected food stamp administrative costs
vary from $6.0 to $41.5 million annually. Again, the San Juan County model has the lowest
administrative costs. As a "responsibility standard" site, the San Juan County model has lower
projected administration costs because claims of unauthorized card usage generally do not need
to be investigated.
The upper end of the administrative cost projections in both programs assumes that
caseworker costs increase in proportion to the number of claims filed. If caseworker costs are
instead assumed to be fixed with regard to claim rate, the projected annual Reg E administrative
costs for the Hudson County and New Mexico models (excluding San Juan County) would be
about $15-$17 million for the AFDC/TANF programs and $21-$22 million for the FSP. The
evidence from the demonstrations is not sufficient to identify whether caseworker costs are fixed
or variable with respect to claim rate; most of the caseworker time was spent with clients with
account problems, but who did not file Reg E claims.
LESSONS FROM THE REG E DEMONSTRATIONS
The Reg E demonstrations provided a wealth of information concerning the provision
to clients of extra protections against EBT loss. In addition to the key findings regarding claim
rates, liability, and administrative costs, a number of interesting lessons emerged concerning
EBT staffing and organization, types of loss incurred, and communicating information to
clients, each of which is described below.
EBT StafTing and Organization
As noted in the discussion of administrative costs, these costs probably could have been
reduced substantially had several sites changed the way their Reg E units were organized.
Based on an analysis of cost components across the sites, the study can offer three organiza-tional
strategies for keeping claim handling costs low:
Prepared by Abt Associates Inc. XI
Jfrt
Executive Summary
(1) Integrate claim handling and investigation procedures as much as possible with the
EBT system's general approach for handling all system problems (e.g., through
use of specially trained "Help Desk" staff).
(2) At the local office level, concentrate the job of helping clients with EBT card or
account problems to a few staff.
(3) Keep claim tracking and management systems simple.
Avoiding Benefit Loss
Approximately 48 percent of all claims submitted during the demonstration were for
losses due to unauthorized card usage. Another 11 percent were for losses resulting from
system or procedural error, many of which occurred when store clerks mistakenly submitted an
EBT transaction twice for system processing. Both types of loss are avoidable, in theory.
What is needed is improved training techniques for both clients and store clerks and, for store
clerks, better supervision by management. Whether improved training and management would
be cost-effective in reducing loss is not known at this point. With client protections against loss
resulting from unauthorized card usage now eliminated, however, helping clients avoid such
losses would be most beneficial to them.
Communicating Information to Clients
In accordance with Reg E policy, the demonstration sites prepared lengthy disclosure
notices informing clients of their rights and responsibilities in an EBT system. Program
administrators generally agree that the disclosure notices were too long, not formatted in an
attractive and easily-readable manner, and perhaps too complicated. Many believe that few
clients took the time to read the notices.
Even in the absence of a Reg E requirement there is a need to communicate to clients
information regarding how to report incidents of benefit loss to the agency. Furthermore, as
discussed in the previous section, there is a need to help clients learn how to avoid losses in the
first place. Thus, although no longer required, it may still be useful to develop an EBT notice
or brochure and to distribute it to existing and new EBT clients. To be successful, however,
any new notices will have to be concisely written, attractive, and informative.
Prepared by Abt Associates Inc. xii
Executive Summary
CONCLUSIONS
Findings from the Reg E demonstrations confirm some concerns program administra-tors
had about applying Reg E to EBT systems, but fail to support other concerns. In particu-lar,
Reg E administrative costs were high in some sites, much higher than liability costs and
often equal to or greater than the savings in issuance costs that states expect when they convert
from paper issuance to EBT. Thus, if Reg E had become mandatory, its administrative costs
might have been large enough to change some states' minds about converting to EBT. It does
appear, however, that the observed Reg E administrative costs could have been reduced
substantially—through changes in staffing structures—if Reg E had continued.
Somewhat surprisingly, the demonstration sites did not experience a large number of
Reg E claims of lost benefits. Furthermore, the sites ended up denying most claims of
unauthorized card usage because clients often failed to provide requested documentation. In
addition, relatively few claims with provisional credits were subsequently denied. Taken
together, these three factors explain why concerns over Reg E's impacts on financial liability
were not realized.
Prepared by Abt Associates Inc. xni
yi/i'i
CHAPTER ONE
INTRODUCTION
Under the sponsorship of the U.S. Department of Agriculture's Food and Consumer
Service (FCS) and other federal agencies, electronic benefit transfer (EBT) systems have been
implemented in a number of different sites across the country. These systems deliver benefits
electronically for a number of state, federal (but state-administered), and direct federal
programs. State programs using EBT to deliver benefits include General Assistance, and direct
federal programs using EBT include Social Security (Old Age, Survivors, and Disability
Insurance, or OASDI), Supplemental Security Income (SSI), and other federal retirement and
disability programs. By far the largest users of EBT systems to date, however, are the state-administered
assistance programs; these include the Food Stamp Program (FSP), the Aid to
Families with Dependent Children (AFDC) program, and the new Temporary Assistance for
Needy Families (TANF) programs.1
EBT systems use either of two alternative technologies—on-line or off-line. On-line
EBT systems work very much like commercial bank card networks. Program participants
receive an EBT card and select a personal identification number, or PIN. The EBT card, which
has a magnetic stripe on the back encoded with identifying information, is functionally similar
to a bank debit card. Using the EBT card and PIN, the EBT cardholder can access cash
assistance program benefits either by withdrawing them from an automated teller machine
(ATM) or by using them at the point of sale (POS) to make purchases or to receive cash back.
For the FSP, the EBT card can be used to access food stamp benefits to pay for purchases in
program-authorized food retail outlets. Whether the transaction is initiated at an ATM or POS
terminal, the device must establish an on-line telecommunications connection to a central
computer to check the cardholder's remaining balance before the transaction can oe authorized.
Off-line systems, in contrast, store information about the client's remaining balances
and the encoded PIN in the EBT card itself, thereby avoiding the need to establish contact with
a central computer for transaction authorization. To date, two off-line EBT systems using stored
1 The Personal Responsibility and Work Opportunity Reconciliation Act of 1996 eliminates the AFDC
program and provides block grant funding for states implementing TANF programs.
Prepared by Abt Associates Inc. 1
Chapter One: Introduction
value cards have been tested. Both have issued FSP benefits; one has also issued benefits in the
Special Supplemental Nutrition Program for Women, Infants, and Children (WIC).
Evaluations of several demonstration on-line EBT systems have shown that they can be
a cost-effective alternative to the issuance of government assistance checks and paper food stamp
coupons,2 and Congress—with passage of the Personal Responsibility and Work Opportunity
Reconciliation Act of 1996 (welfare reform legislation)—now requires that states implement EBT
systems before October 1, 2002.3 Prior to passage of the Act, however, there was considerable
discussion and debate over whether these EBT systems should be subject to the provisions of
"Regulation E."4
Regulation E (or simply "Reg E") implements the provisions of the Electronic Fund
Transfer Act (EFTA), which establishes a framework of legal rights and responsibilities for
participants in electronic fund transfer systems.5 Several provisions within the regulation
protect clients against loss associated with use of their debit cards. In March 1994, the Board
of Governors of the Federal Reserve voted to extend the provisions of Reg E to all on-line EBT
systems, with a three-year implementation period. Congress eventually decided to exempt
certain EBT systems from the provisions of Reg E as part of its welfare reform legislation,6
but only after several demonstrations had been conducted to determine the impacts of applying
Reg E to EBT svstems. This report presents the findings of an evaluation of those demonstra-tions.
2 Sec. for example, John A. Kirlin. The Evaluation of the Expanded EBT Demonstration in Maryland:
Summary of Findings, Cambridge, MA: Abt Associates Inc., 1994. Evaluations to determine the cost-effectiveness
of off-line EBT systems are still underway.
3 With regard to using EBT systems to deliver food stamp benefits, the Act does allow the Secretary of
the U.S. Department of Agriculture to grant waivers to states facing 'unusual barriers to implementation."
4 The debate was framed solely in terms of on-line EBT systems for two reasons: few states were
considering off-line systems at the time, and Regulation E itself was designed to cover commercial on-line
systems.
5 15 U.S.C. § 1693.
6 The Act exempts from Reg E requirements any EBT programs established under state or local law or
administered by a state or local government.
Prepared by Abt Associates Inc. 2
Chapter One: Introduction
1.1 EBT SYSTEMS AND CLIENT PROTECTIONS AGAINST BENEFIT LOSS
All EBT systems provide their users with some protections against benefit loss. For
instance, if an EBT cardholder reports that an ATM dispensed fewer AFDC benefits than
requested and debited from her EBT account, the EBT system vendor will typically initiate an
investigation; the missing benefits will be credited to the account if the ATM misdispense is
verified. Most EBT systems, however, will not reimburse losses due to unauthorized card
usage. An example would be when a lost or stolen EBT card is used to withdraw funds from
an ATM or to buy groceries at a store equipped to accept EBT transactions at the POS. Part
of the reasoning for not reimbursing such losses is that cardholders can prevent unauthorized use
of a lost or stolen caid if they keep their PINs a secret; an EBT transaction cannot be completed
without knowledge of the cardholder's PIN. Although all EBT systems provide clients some
protection against benefit loss, the nature of the protection varies by program, as described
below.
Food Stamp Program
F'SP regulations require that state agencies be liable for benefits lost or stolen as a result
of EBT system error or fraud. Some client advocates have suggested that the rules are not as
specific as they should be with regard to when recipient benefits should be replaced by the state
agency. Towards that end, the Department of Agriculture may in the future propose regulatory
changes for the purpose of clarification. For now, existing food stamp regulations specify:
• "Once a household reports that their EBT card has been lost or stolen, the State
agency shall assume liability for benefits subsequently drawn from the account and
replace any lost or stolen benefits to the household" (7 CFR § 274.12(0(5)(iv)).
• "Errors (i.e., problem transactions) shall be resolved in a timely manner" (7 CFR
§274.12(h)(2)(iii)).
• "The State agency shall be strictly liable for manual transactions that result in
excess deductions from a household's account" (7 CFR § 274.12(1X3)).
• "State agencies shall be held strictly liable for overissuances resulting from
Electronic Benefit Transfer system errors and unauthorized account activities" (7
CFR § 276.2(b)(7)).
Prepared by Abt Associates Inc.
Chapter One: Introduction
These program regulations do not stipulate how quickly benefits need to be replaced in the above
situations. They also make no provision for providing an interim (or provisional) credit to the
food stamp household during any period of investigation.
Cash Assistance Programs
State EBT systems also deliver benefits for one or more cash assistance programs. The
most notable of these programs has been AFDC (now being replaced by the TANF programs),
but other programs include General Assistance (GA) and Refugee Assistance (RA).
Currently, no federal rules govern the replacement of lost cash assistance benefits within
a state's EBT system. States and their EBT vendors have generally applied the same benefit
replacement policy, however, for both food stamp and cash assistance benefits. Thus, the
protections outlined above for lost food stamp benefits have applied to lost cash assistance
benefits as well. In addition, recipient claims of ATM misdispenses are typically investigated
by the EBT vendor and ATM owner. If an ATM misdispense is verified, the missing benefits
are reimbursed to the client.
Direct Federal Programs
Unlike the EBT systems providing benefits for state-administered programs, EBT
systems providing direct federal program benefits have always operated under the provisions of
Reg E. With respect to client claims of lost benefits, the provisions of Reg E require:
• that clients receive i disclosure notice each year summarizing their liability for
unauthorized card usage and detailing error resolution procedures (12 CFR §
205.7(a)(10));
• that a client's liability for unauthorized usage of his or her card be limited to $50
if the loss is reported within two days of discovery7 (12 CFR § 205.6(b)); and
• that card issuers complete their investigation within 45 days (for losses at an ATM)
or 90 days (for losses at a POS device), but that the amount of the loss be credited
to the cardholder within 10 days (ATM) or 20 days (POS) if the investigation is not
complete by that time.
7 Liability increases to up to $500 if the loss is reported more man two days after the loss if discovered,
but within 60 days. If a loss is not reported within 60 days of its discovery, the cardholder assumes full
liability for the loss.
Prepared by Abt Associates Inc.
Chapter One: Introduction
Welfare reform legislation does not exempt EBT systems providing direct federal program
benefits from the provisions of Reg E, nor does it address EBT systems providing both state and
federal benefits.
EBT vs. Reg E Protections Against Loss
The differences between standard EBT protections against loss and Reg E protections
can now be better defined. First, losses due to unauthorized card usage are not reimbursable
in state-administered EBT systems (unless the recipient had already reported the card as lost or
stolen); they are reimbursable under Reg E, although the client does bear some liability.
Second, Reg E requires investigations to be completed within 45 to 90 days; there is no defined
timeliness standard for investigation and processing of EBT claims of loss. Third, unlike
systems operating under Reg E, EBT systems do not provide for interim crediting of the loss
amount. Fourth and finally, state agencies are not required to provide disclosure notices to
recipients. (Instead of disclosure notices, however, nearly all recipients receive special EBT
training. During each training session, recipients are told to protect their cards and their PINs
and to whom account problems should be reported.)
1.2 REG E AND EBT SYSTEMS
The Federal Reserve Board's 1994 decision to extend the provisions of Reg E to EBT
was controversial. Client advocates supported the decision, asserting that households receiving
public assistance should have the same protections against loss as anyone using a debit card to
withdraw funds from a bank account or to purchase goods or services at a POS terminal. Many
federal and state proponents of EBT systems serving state-administered programs, however,
argued against the regulation's application to EBT. These proponents believed that existing
program protections included in EBT systems were both sufficient and appropriate, especially
given that risk controls available to the private sector (e.g., revoking use of a bank card) were
either not available to public programs or were difficult to implement.
Program administrators also were concerned about the requirement to replace benefits
(above the client's $50 liability) for losses associated with unauthorized card usage. They
reasoned that if clients followed prudent procedures in keeping their PINs secret, most types of
unauthorized transactions could not occur. Thus, administrators believed that introducing a
Prepared by Abt Associates Inc. 5
Chapter One: Introduction
replacement policy for such losses reduced the incentive to safeguard one's PIN, which could
increase incidents of loss. In addition, program officials were concerned about fraudulent claims
of loss and the possible difficulty in identifying such claims during investigations.
Finally, given the lack of generalizable Reg E experience in EBT applications and the
consequent uncertainty over the cost impacts of applying Reg E to EBT systems, EBT
proponents feared the Board's decision would substantially delay or even halt EBT system
development and expansion. This would prevent program participants from enjoying the positive
features of EBT systems that had been documented during previous demonstrations of the
technology.
In the midst of this controversy stood officials of the U.S. Department of Treasury's
Financial Management Service (FMS). FMS supports EBT systems as a means of providing
direct federal payments to clients without bank accounts. Because Treasury uses financial
institutions to deliver EBT benefits, it has never tried to make a distinction between private and
public sector applications of EBT. Instead, FMS has required the incorporation of Reg E
protections in these systems from their inception. The most notable example is a direct federal
payment system called the Direct Payment Card (DPC) system, operated by Citibank EBT
Services (Citibank) and serving clients in Texas. Reg E protections have not led to large levels
of replaced benefits in the DPC system.8 Opponents of applying Reg E to state-administered
programs, however, believed that the direct federal and sttte-administered programs and clientele
were sufficiently different from one another that Reg E experience with direct federal programs
could not be generalized to predict what might happen if Reg E were applied to state-administered
programs.
Faced with these uncertainties and opposing viewpoints, the Board of Governors
provided a three-year implementation period for applying Reg E to EBT systems. The purpose
of the three-year implementation period was to give federal and state agencies time to:
• learn more about the likely impacts of Reg E on administrative costs and benefit
replacements;
8 From the start of DPC system operations in April 1992 through June 1994, the number of approved
claims of unauthorized usage averaged just over one per 10,000 case months (with a "case month"
representing one month of EBT participation by one cardholder). The total cost of replaced benefits averaged
just $0.03 per case month. See "Direct Payment Card: Expansion Evaluation," Citibank EBT Services
(undated).
Prepared by Abt Associates Inc.
Chapter One: Introduction
• assess the effectiveness of different strategies for implementing Reg E and
controlling claims of benefit loss; and
• prepare funding plans for any costs associated with the application of Reg E.
This three-year period was also used by Congress to consider the advantages and disadvantages
of applying Reg E to EBT systems. It was after several hearings, including one in which
preliminary findings from the Reg E demonstrations described below were presented,9 that
Congress elected to exempt state- and county-administered EBT systems from the provisions of
Reg E. One reason for exempting these EBT systems was to foster the widespread acceptance
and implementation of EBT by 2002.
1.3 THE REG E DEMONSTRATIONS
Not knowing whether legislative action would be forthcoming, and in an effort to
provide empirical evidence on the impacts of applying Reg E to EBT systems, federal and state
agencies used the three-year implementation period to conduct a series of demonstrations in
which several sites with EBT systems operated under Reg E provisions for 12 months.
Interested states with EBT systems were invited to submit proposals for implementing Reg E.
Two states—New Jersey and New Mexico—were selected for the demonstrations. Within these
two states, four sites (Hudson County in New Jersey, and Bernalillo, Doha Ana, and San Juan
counties in New Mexico) implemented Reg E protections. One county (Camden in New Jersey)
continued to operate under regular EBT protections and thereby served as a comparison site for
the evaluation. Citibank's DPC system in Texas was also included as a Reg E demonstration
site. The DPC system has been operating under the provisions of Reg E since the system's
inception in 1992.
As shown in Exhibit 1-1, the six demonstration sites varied in a number of important
ways. For instance, with respect to program mix, the EBT systems in New Jersey10 and New
9 These findings were presented to the Subcommittee on Financial Institutions and Consumer Credit,
House Committee on Ranking and Financial Services, on June 19, 1996. The findings are reported in The
Evaluation ofthe Application ofRegulation E to EBT Systems: PreliminaryFindings, Cambridge. MA: Abt
Associates Inc., June 1996.
10 New Jersey calls its EBT system the "Families First" system. For ease of exposition, this report will
use the more generic "EBT" terminology in most instances.
Prepared by Abt Associates Inc.
Chapter One: Introduction
Exhibit 1-1
OVERVIEW OF DEMONSTRATION SITES
She
Programs
Served
Avenge
Monthly
Caseload* Percent Urban"
Combined
Larceny and
Robbery Ratesc
Camdcn County, NJ
AFDC. FSP 22.740 97.5% 49.8
(comparison site)
San Juan County. NM
(responsibility standard site)
AFDC. FSP 3.514 61.1% 64.5
Citibank DPC System (TX)
(full Reg E site)
OASDI. SSI.
others
12.405 80.3% 49.1
Hudson County, NJ
(full Reg E site)
AFDC. FSP 28.456 100.0% 36.1
Bernalillo County, NM
(full Reg E site)
AFDC. FSP 24.703 95.6% 59 8
Dona Ana County. NM
1 (full Reg E site)
AFDC. FSP 10.259 73.9% 59.5
* Unduplicated case count (i.e.. households receiving both food sumps »nd AFDC ire counted jusi once) Average monthly
caseloads, by program, are presented in Appendix E.
" Census of Population and Housing, 1990: Summary Tape Files on CD-ROM, Bureau of the Census. U.S. Department of
Commerce. 1992. Percent urban is defined as the percentage of an area's population living w«hm an 'urbanized area."
which is any central place and densely-settled fringe w*h a minimum of 30.000 persons.
c Source: Crime in the United Slates. Federal Bureau of Investigation. U.S. Department of Justice. 1995 Rales are defined
as number of offenses per 1.000 persons.
Mexico served clients receiving food stamps and/or AFDC benefits. In contrast. Citibank's
DPC system served clients receiving either Social Security or SSI payments, or one of several
different federal retirement or disability programs.
The demonstration sites also varied in what types of loss were reimbursable and claim
handling procedures. As one reads down Exhibit 1-1, the sites generally are listed in ascending
order of the protections they provided clients against toss of benefits. For instance, the
comparison site of Camden County offered only regular EBT system protections against loss;
there were no protections against losses due to unauthorized card usage, and there were no
interim credits or deadlines for claim investigations.
San Juan County, in contrast, participated as a Reg E demonstration site, and Reg E
provisions regarding how quickly claims of loss bad to be investigated were in effect there, as
Prepared by Abt Associates Inc.
Chapter One: Introduction
was the requirement that provisional credits be granted if investigations could not be completed
before the Reg E deadlines. New Mexico wanted to explicitly test a different set of client
protections than offered by Reg E, however, so San Juan County participated in the Reg E
demonstrations as a "responsibility standard" site. This meant that losses due to unauthorized
card usage were not reimbursed if the transaction in question was initiated with a valid EBT card
and PIN. Therefore, with respect to which types of loss were reimbursable, the piotections
offered clients in San Juan County were nearly identical to those ffered in CmAm County.
The last four sites were Citibank's DPC system. Hud: . County in New Jersey, and
Bernalillo County and Doha Ana County in New Mexico. All four sites operated under "mil"
Reg E protections during the demonstration periods. Thus, unlike Camden and San Juan
counties, losses due to unauthorized card usage were reimbursable in these four sites if the client
cooperated with the investigation and the circumstances of the loss could be verified.
In terms of planned procedural differences in Reg E practices across the sites, all claims
of loss due to unauthorized card usage were investigated in Hudson County, whereas New
Mexico officials—in an effort to control uncertain administrative costs—had discretion over
whether to investigate or simply approve claims involving "small" loss amounts.11 The sites'
planned administrative controls for reducing future losses also varied, ranging from additional
training in bow to use and protect one's card and PIN. to conversion from EBT back to check
issuance. (The latter control was used only in the Citibank DPC system, where participation is
voluntary.) In addition, all offices in Bernalillo County initiated, as part of the demonstration,
a policy of charging clients S2 for replacement cards (beyond the first card lost in any calendar
quarter), in an effort to reduce card loss. Another administrative control, implemented in March
1996, was the issuance of photo EBT cards to new applicants and those recipients needing
replacement cards in two of the four local offices in Bernalillo County. New Mexico expects
that use of a photo EBT card will reduce the frequency of card loss and any associated
unauthorized transactions.
The sites' geographic settings were quite different as well. Camden. Hudson, and
Bernalillo counties represented highly urbanized areas with large central cities (Newark, Jersey
1: In practice, however. New Mexico officials did not exercise this discretion, m large part
volume of Haunt was small MMMfc dm all H»m»c could be investigated wnh available
Prepared by Abt Associates Inc. 9
Chapter One: Introduction
City, and Albuquerque, respectively) and large caseloads, whereas Dona Ana and San Juan
counties in New Mexico were less urbanized and had smaller central cities (Las Cruces and
Farmington, respectively) and smaller caseloads. The DPC system in Texas, on the other hand,
covered a variety of both urban and rural environments; its caseload was small (for a statewide
system) because, unlike the other EBT systems, participation in the DPC system was voluntary.
As shown in Exhibit 1-1, crime rates, as measured by the annual number of robberies and
larcenies per 1,000 population, also varied across the sites, although perhaps not to as great an
extent as might have been expected.
Finally, the demographics of the EBT caseloads in the New Jersey and New Mexico
sites varied considerably, especially in the area of race and ethnicity.12 For instance, the
percent of Hispanic recipients in the site caseloads varied from 12 percent in San Juan County
to 79 percent in Dona Ana County, whereas the percent of African-Americans ranged from five
percent or less in all three New Mexico counties to 45 percent in Camden County. Native
Americans, who represented less than one percent of the New Jersey and Doha Ana County
caseloads, represented 7 percent of the Bernalillo County caseload and 57 percent of the San
Juan County caseload.
1.4 THE EVALUATION OF THE DEMONSTRATIONS
The purpose of the Reg E demonstrations was to gain experience in how to implement
Reg E and various claims control strategies in an EBT environment, and to learn about the
administrative and benefit replacement costs that arise when implementing the regulation.
Specifically, the evaluation of the demonstrations has four primary objectives:
(1) To describe systematically how Reg E was implemented in each demonstration site
and to compare protection and claims procedures across the Reg E sites and
Camden County;
(2) To assess and compare the frequency and dollar value of benefit claims and
replacements in each Reg E site, in Camden County, and in other EBT sites not
operating under Reg E;
12 Demographic information was not available for participants in Citibank's DPC system. Appendix I
presents a more detailed comparison of the caseload demographics in the five New Jersey and New Mexico
counties.
Prepared by Abt Associates Inc. 10
Chapter One: Introduction
(3) To measure and compare the administrative costs of processing reported EBT
losses in each Reg E site and in Camden County; and
(4) To elicit comments from stakeholders on Reg E policy and operational changes
that, within the context of EBT, offer a better balance between recipient protection
and program accountability.
The research design for the evaluation is essentially cross-sectional, with Camden
County serving as the comparison site, San Juan County representing one particular type of Reg
E treatment, and the remaining four sites representing full Reg E treatments. In addition to
using Camden County as a formal comparison site (with pre-arranged plans for collecting data
on claims of lost benefits and the administrative costs of processing these claims), the evaluation
used extant data on claims of loss from other EBT sites to broaden the base of comparison data.
This was possible because all EBT sites offer clients protection against certain types of loss,
especially those due to ATM misdispenses (when an ATM disburses fewer funds than requested
and debited from the client's account) or errors in system operating procedures or processing.
Specifically, information on claims of lost AFDC and food stamp benefits was gathered from
the EBT systems in Maryland, Ramsey County (Minnesota), and New Mexico.13 The New
Mexico comparison data include claims filed before the introduction of Reg E and, for counties
not participating in the Reg E demonstrations, claims filed during the demonstration period.
Another issue concerning research design is the relative timing of Reg E and EBT
implementation. In planning for the Reg E demonstrations, both FCS and the Administration
for Children and Families (ACF) realized that implementing Reg E protections in a site already
on EBT would not replicate many future implementations of Reg E, and that this might affect
the generalizability of demonstration results. The specific concern was that clients in sites
already on EBT might not be as aware of the new Reg E protections as would clients in sites
with simultaneous implementation. Consequently, FCS and ACF wanted one or more
demonstration sites in New Mexico or New Jersey to implement EBT and Reg E concurrently.
This was not possible in New Mexico, where the Reg E demonstrations ran from May 1995 to
April 1996 in all three counties. EBT had been fully implemented in Bernalillo County since
13 Except for South Carolina, no other reasonably-sized sites were issuing benefits through on-line EBT
systems at the time of data collection. The South Carolina system issues food stamp benefits, but not AFDC
benefits, so no attempts were made to collect comparable data on benefit loss.
Prepared by Abt Associates Inc. 11
Chapter One: Introduction
March 1992 and, as part of a statewide expansion of EBT, San Juan County finished its
conversion to EBT by July 1994. It had been hoped that EBT and Reg E could be implemented
simultaneously in Dona Ana County, where EBT conversion started in October 1994, but delays
in the start of the Reg E demonstrations precluded this.
Simultaneous implementation of EBT and Reg E was possible in New Jersey. The 12-
month Reg E demonstration began in Hudson County in March 1995, which is the same time
the county began converting to EBT. EBT conversion was completed in Hudson County by May
1, 1995. The comparison site of Camden County has been operating under EBT since April
1994.
The Citibank DPC system also had simultaneous implementation of EBT and Reg E.
Citibank first implemented its DPC system, with full Reg E protections, in the Houston area in
April 1992. The system then expanded to the Dallas-Fort Worth area in November 1993; it is
now available statewide. The Reg E demonstration period for the DPC system was the same
as for Hudson County—March 1995 through February 1996.
1.5 ORGANIZATION OF THE REPORT
Chapter Two of this report, together with appendices A through D, addresses the
evaluation's first objective—describing how Reg E was implemented in the demonstration sites
and comparing protection and claims procedures across the demonstration sites. Evidence on
Reg E's impacts on frequency of benefit claims is presented in Chapter Three, and Chapter Four
discusses what impact those claims had on levels of replaced benefits. The administrative costs
of processing EBT claims in Camden County and Reg E claims in the other sites are examined
in Chapter Five.
Chapter Six presents the results of efforts to project the likely impacts of implementing
Reg E in other state-administered EBT systems, given what has been learned from these
demonstrations. The report concludes in Chapter Seven by examining what lessons the
demonstrations offer in terms of protecting clients against benefit loss.
Prepared by Abt Associates Inc. 12
CHAPTER Two
PLANNING FOR AND IMPLEMENTING REGULATION E
The process of planning for and implementing the Reg E demonstrations involved
representatives from federal and state government agencies, EBT vendors, and client advocacy
groups. These representatives worked together to develop a series of demonstrations that would
provide the information needed for implementation of Reg E protections in all EBT sites by
March 1997—the end of the Board of Governors' intended three-year implementation period.
With Congress' exemption of state-administered EBT systems from the provisions of Reg E, this
demonstration experience is no longer directly relevant to state efforts to implement EBT
systems by 2002. Nevertheless, the Reg E demonstrations offer lessons to be learned for the
provision of EBT client protections more generally. For instance, they provide the first state
and county experience investigating claims of unauthorized card usage. They also represent the
first systematic documentation of Reg E costs in an EBT environment. The demonstrations
should also offer particularly valuable insights for states implementing EBT systems that include
both state-administered and direct federal programs. These states need to operate the direct
federal program portion of their EBT systems under Reg E protections.
This chapter details how the demonstration sites, the federal agencies, the Federal EBT
Task Force, and a number of client advocacy groups addressed the numerous issues that arose
in establishing Reg E policy for the demonstrations and in implementing the policy and
associated Reg E procedures in each site.
2.1 ESTABLISHING GENERAL POLICY
Planning for the demonstrations began with a broad policy discussion of how the
provisions of Reg E could be applied in an EBT environment. Reg E has been applied in the
banking industry since 1979, but applying it to public sector assistance programs posed new
policy issues, especially when the protections embodied in the regulation overlapped with
existing program rules and protections. Program administrators, with input from client
advocates, also had to define exactly what would constitute a Reg E-covered loss, and when and
how clients would be liable for a portion of the loss. Finally, decisions had to be made about
Prepared by Abt Associates Inc. 13
Chapter Two: Planning for and Implementing Regulation E
the proper content and format for a disclosure statement for an EBT system operating under Reg
E protections. This section discusses each of these three broad issues.
Applying Reg E to Public Sector Assistance Programs
During the summer and fall of 1994 there were many meetings among demonstration
planners to discuss and interpret the provisions of Reg E in the context of EBT. Two broad
policy issues were addressed: what to do when program regulations offered greater protections
than did Reg E, and whether claims could be categorically denied for any reason.
EBT Versus Reg E Protections. The planning meetings raised a number of
fundamental questions regarding the relative level of protection offered by program regulations
and Reg E. For instance, if a single claim of unauthorized use of an EBT card involved loss
in two or more programs, should the client's $50 liability—as defined by Reg E—be applied just
once, or separately for each program? Or, given that program benefits are generally issued for
a specified household "unit," should an "unauthorized" transaction by a household member other
than the cardholder be treated as an unauthorized usage of the card, subject to reimbursement?
Or again, if a provisional credit were granted for a Reg E claim, could the entire credit be
immediately debited from the client's EBT account if the claim were subsequently denied (as is
done in the private sector)?
During these meetings, and after discussions with the Federal Reserve Board and the
Department of Agriculture's Office of General Counsel (OGC), a general consensus arose—when
Reg E protections and program policy do not match, follow the policy providing the greatest
level ofprotection to the cardholder. Thus, it was decided that cardholders would be liable for
just $50, even if unauthorized transactions were made against both their AFDC and food stamp
benefits.1 Similarly, unauthorized transactions made by other household members would be
reimbursable, although the cardholder would be expected to cooperate with the investigation and
be willing to prosecute. For claims denied after a provisional credit had been granted, recovery
of funds would follow established program rules for recoupment of benefits rather than an
immediate debit for the full amount. In addition, program regulations regarding the client's
1 The $50 maximum for liability holds only if the client reports the loss within two days of its discovery.
After two days the maximum liability increases to $500.
Prepared by Abt Associates Inc. 14
Chapter Two: Planning for and Implementing Regulation E
right to a fair hearing following any adverse action would be retained, another protection not
included in Reg E. In this regard, then, New Mexico and New Jersey implemented what might
be called "super" Reg E protections—not only did clients in these states' demonstration sites
have greater protections against loss than other EBT participants, they had greater protections
than are available under Reg E in the private sector, including those clients participating in
Citibank's DPC system.
Categorical Denials. Another issue regarding the application of Reg E to public sector
programs arose during the early planning months. According to staff at the Federal Reserve
Board, Reg E claims are supposed to be reviewed and acted upon on a case-by-case basis by
weighing all available evidence. That is, a decision to deny a claim should not be based on a
single criterion that disregards other information about the claim. This discussion followed a
query by program staff as to whether claims could be denied on a categorical basis (e.g.,
categorically deny the second claim from the same client in a given time period).
Acting on all claims on a case-by-case basis clearly increases the administrative cost of
the review process (one of the concerns of program administrators). Moreover, it also
introduces some subjectivity into the process of deciding whether to approve or deny a claim;
that is, the importance of various circumstances of the claim need to be "weighed" (a subjective
determination) before making a decision. Many public sector programs seek to avoid such
subjectivity, inasmuch as it is difficult to ensure equal treatment of all cases in such an
environment. Thus, some program administrators planning tor the Reg E demonstrations felt
uncomfortable with this case-by-case approach.
Demonstration planners handled this dilemma in two ways. First, they generally agreed
that all available evidence pertaining to a claim would be considered when making a decision
whether to approve or deny the claim. This is how Citibank had been handling its claims in the
DPC system in Texas. Second, however, planners for the New Jersey and New Mexico
demonstrations argued that a claim should be denied outright if the claimant failed to meet
procedural requirements of the claim's investigation (e.g., by failing to file a police report, if
requested). The rationale for this categorical basis for denial was that pre-existing program
regulations required such cooperation in other (non-Reg E) investigations. Federal Reserve
Board staff agreed that such categorical denials would be appropriate given the program
regulations.
Prepared by Abt Associates Inc. 15
Chapter Two: Planning for and Implementing Regulation E
Although not discussed during the planning stage of the demonstrations, Reg E officials
in New Jersey and New Mexico added two other categorical reasons for denying a claim during
the demonstrations. Claims were denied outright if the type of benefit loss was not reimbursable
(as defined in the next section), and they were denied outright if the amount of loss was less than
the claimant's maximum liability under Reg E (e.g., $30 for loss due to unauthorized card
usage).
Determine What Constitutes a Reg E-Covered Incident
Together with discussion of how to apply Reg E in EBT systems, the states had to
decide what types of loss would be reimbursable under Reg E. For losses already deemed
reimbursable under current EBT applications, this was not a problem. Adhering to the decision
that the demonstrations should include all protections already in place in EBT systems, these
losses continued to be treated as reimbursable (assuming verification of the circumstances of the
reported loss). Thus, as shown in Exhibit 2-1, any losses reimbursable in Camden County,
which operated under regular EBT protections, were treated as Reg E-covcred losses in both San
Juan County (the responsibility standard site) and the four full Reg E protection sites. These
losses included ATM misdispenses, losses due to three types of system or procedural error, and
losses due to employee theft. The three types of system or procedural error were: (1) state or
EBT vendor staff fail to disable an EBT card after it has been reported as lost or stolen;2 (2)
a system-processing error resulting from software problems or incorrect operating procedures;
and (3) a single transaction at the store being debited twice against a client's account (usually
the result of clerk error). Employee theft covers EBT vendor, state, county, and retail store
staff.
With a few exceptions (discussed later in this section), this left claims of unauthorized
usage as the major type of loss that the sites had to decide bow to handle, and this is where the
San Juan County protections diverged from the other Reg E sites. In almost all situations of
2 Although a loss that occurred after an EBT card was reported as lost or stolen was likely due to
unauthorized usage of the card, such loss is treated as a "system or procedural error" because, after a card
is reported as lost or stolen, an instruction to prevent further system authorization of transactions initiated with
the card was supposed to be entered into the computer system. Therefore, if a loss occurred after a card was
reported as lost or stolen, there was either failure to follow system operating procedures or an error in the
software governing which cards could access the system.
Prepared by Abt Associates Inc. 16
Chapter Two: Planning for and Implementing Regulation E
unauthorized usage, the loss was not reimbursable in San Juan County. (The only exception was
incidents of "shoulder surfing," in which somebody determines a client's PIN and card number
and then initiates a transaction using a counterfeit card.) In contrast, losses due to unauthorized
usage were reimbursable in all four sites offering full Reg E protections. As indicated in the
exhibit, however, Citibank did not reimburse a loss involving unauthorized usage if the client
ever knowingly gave the card and PIN to another person.3 Officials in the other sites said that
failure to protect one's PIN would enter into their decision about whether or not to approve a
claim of unauthorized usage, but that they would not necessarily deny a claim solely for this
reason.
Turning to the bottom of Exhibit 2-1, officials in New Mexico and Hudson County said
that, because merchants are not allowed to charge fees to EBT cardholders, imposition of such
fees would be reimbursable under Reg E. There is no prohibition against merchant fees in
Citibank's DPC system, so this was not a reimbursable loss there. State officials in New Jersey
said that although fees are not allowed in Camden County, and that action would be taken
against merchants who charge fees, fees already paid by clients would not be reimbursed by the
state. They also said that, unlike the Reg E sites, a loss resulting from a "forced transaction"
(defined as a situation in which the client is forced to withdraw and turn over funds) would not
be reimbursable. In this respect Camden County treats a forced transaction just like a robbery,
in which funds are stolen from a client after a valid and uncoerced transaction has been made.
All sites view robberies as a police matter; any EBT funds lost in a robbery were not reimburs-able
under Reg E.
Finally, although not shown in the exhibit, FCS and the sites agreed that calls from
clients who believed that benefits were missing would not be treated as Reg E claims if, while
speaking with the client, it became apparent that the client's monthly benefits had not yet been
posted to his or her EBT account. Clients often make such calls shortly before benefits are
actually posted, and treating such calls as Reg E claims would have placed an undue administra-tive
burden on the sites for situations in which no loss occurred. A similar policy was followed
when, during a call to report missing funds, the client remembered making a transaction for the
amount in question.
3 This might be viewed as a categorical denial, although not on procedural grounds.
Prepared by Abt Associates Inc. 17
Chapter Two: Planning for and Implementing Regulation E
Exhibit 2-1
REIMBURSABLE LOSSES, BY SITE
Lerd of Protection
Regular
EBT
Respomii-biHty
Stan-dard
Full Reg E
Site
Camden
County
(NJ)
San Juan
County
(NM)
Citibank
DPC System
(TX)
Hudson
County
(NJ)
Bernalilk)
County
(NM)
Dona Ana
County
(NM)
Card lost or stolen,
user unknown
No No Yes Yes Yes Yes
Card stolen, client
knows who used card No No
Yes"
Yes1 Yes* Yes1
Nob
Client still has card No No Yes Yes Yes Yes
Shoulder surfing0 No Yes Yes Yes Yes Yes
Nm r*o+tefrm*
ATM misdispense Yes Yes Yes Yes Yes Yes
Sp$em or ProcodmrmJ Error
Loss occurs after card
reported as lost or
stolend
Yes Yes Yes Yes Yes Yes
Processing error*' Yes Yes Yes Yes Yes Yes
Double debit at storee Yes Yes Yes Yes Yes Yes
m*
Merchant charges fee Nof Yes na* Yes Yes Yes
Employee theft*1 Yes Yes Yes Yes Yes Yes
Forced transaction No Yes Yes Yes Yes Yes
1 Robbery (after with
|drawal)
No No No No No No
Prepared by Abt Associates Inc. 18
Chapter Two: Planning for and Implementing Regulation E
Exhibit 2-1 (continued)
REIMBURSABLE LOSSES, BY SITE
NOTES.
* This loss was reimbursable under Reg E, but Reg E officials required that the client file a police report and be willing to
prosecute.
b Citibank would not reimburse the loss if the cardholder knowingly give the card and PIN to this person at any earlier time.
c Shoulder surfing refers to a situation in which someone looks over a client's 'shoulder" and sees the PIN as it is being
entered. If that person can then determine the client's EBT card number (e.g., by picking up a thrown-away receipt), it is
possible for a counterfeit card to be made up and used—with the PIN—to steal funds from the account.
6 Liability would fall on whichever organization was responsible for the error. For cards reported as lost or stolen, the
organization receiving the report is supposed to enter the information into the system's computer, thereby preventing further
use of the card.
' The presumption is that, upon notification of the error by the EBT vendor, the store would process a refund for the client.
' The state would notify the merchant that, per their contract, fees are not allowed on EBT transactions If the merchant
continued to charge fees, the contract would be canceled and the EBT equipment removed. Any clients who paid fees,
however, would not be reimbursed.
* "Not applicable"; the Reg E staff know of no restrictions against merchant fees in the DPC system.
k The presumption is that the employer (EBT vendor, state, county, or retail store) would make restitution
Prepare (and Provide) a Reg E Disclosure Notice
One of the requirements of Reg E—as it applies to both EBT systems and the private
sector—is that a card issuer provide disclosure statements to system participants. Thus, federal
and state officials recognized early in the planning stages of the demonstrations that an EBT
disclosure notice describing EBT card use and the new Reg E protections and procedures would
need to be prepared and made available to EBT clients. This was especially important for those
clients already on EBT, inasmuch as the Reg E demonstrations would change their protections
under EBT. Citibank already had a disclosure statement in use for its DPC system in Texas (see
Appendix D for a copy); thus, the bank did not have to participate in this process.
A representative from FCS prepared a first draft of an "EBT Agreement and Disclosure
Statement" in September 1994. The six-page document adhered closely to Reg E requirements
Prepared by Abt Associates Inc. 19
Chapter Two: Planning for and Implementing Regulation E
covering disclosure of terms by financial institutions to customers receiving debit cards.4 The
document defined terms, explained bow to use an EBT card, stressed the importance of keeping
one's PIN number a secret, explained how and when to report errors or lost or stolen benefits,
described procedures that would be taken to investigate and process claims of lost or stolen
benefits, and listed when and under what circumstances information about an EBT account could
be disclosed by the government to others. The final page provided space for disclosure
agreement signatures by the client and card issuer. The reason for obtaining the client's
signature was to document that the client had indeed received a copy of the disclosure notice.
This draft was distributed to FCS, ACF, the Federal EBT Task Force, ..ew Mexico,
New Jersey, and representatives from client advocacy organizations. The document was revised
(and expanded) several times following a series of meetings in the fall of 1994. Throughout this
process, the major concern of the advocates was ensuring that the disclosure clearly specified
the cardholder's rights and responsibilities in an EBT system. FCS, on the other hand, was
concerned that too much emphasis was being placed on clients' rights without enough being said
about clients' responsibilities. Also of major concern to all parties was the resulting length of
the document; many believed that EBT clients would not take the time to read a 16 page
disclosure notice. There was also concern about the clients being able to read and comprehend
die concept of Regulation E. The purpose of the client acknowledgement form was also ques-tioned,
as it would only indicate that someone received the disclosure statement, not that they
understood the information.
FCS then prepared a final draft of a generic EBT disclosure statement; each state took
this model and made changes to reflect its specific situations. In New Mexico the only major
change was the deletion of the signature panel acknowledging the cardholder's receipt of the
notice (see Appendix B for a copy of the notice). New Mexico planned to mail the disclosure
statements to its (already trained) EBT clients, and officials saw no feasible way to ensure that
clients would return a signed acknowledgement form. The New Mexico disclosure notice was
not translated to any other languages, as it was felt that literacy would be the barrier to
understanding the statement, regardless of the language.
4 12 CFR 205.7.
Prepared by Abt Associates Inc. 20
Chapter Two: Planning for and Implementing Regulation E
State staff finalized New Jersey's disclosure notice, going through several drafts in an
effort to make the document's language as understandable as possible. The Hudson County
disclosure notice was eight pages long and covered essentially the same information as the New
Mexico disclosure (see Appendix A for a copy of the notice). Hudson County produced both
an English version and a Spanish version of the disclosure notice, which was distributed to
clients during their EBT training session.5
After the Hudson County notice was finished, state staff used it to draft English and
Spanish versions of a notice for Camden County (see Appendix D). The Camden County notice
was created in response to the concerns of client advocates. The advocates were worried that
clients in Camden County would be less likely to report any loss, due to lack of information
about what to do if an unauthorized transaction was experienced. The notice was mailed to all
clients. At only three pages, the Camden notice is shorter than the Hudson County notice for
several reasons, but primarily because the protections are different. The Hudson County notice
includes more detailed information on reporting a lost or stolen card, including how to report
the loss, getting a claim number, and filing a police report.
2.2 CLAIM HANDLING PROCEDURES
An obvious step in planning for the introduction of Reg E was determining exactly how
Reg E claims would be accepted, investigated, and administratively processed. This step was
fundamental for the demonstration sites: not only did procedures have to be established before
any claims could be processed, but the procedures themselves would affect levels of replaced
benefits, administrative costs, and interactions with clients. The demonstration experience
exemplifies the tradeoffs the sites had to address. For example. New Jersey implemented a
system of very thorough investigations, based on already established staffing patterns. Although
this model may have increased accuracy and reduced the likelihood of replacing benefits never
lost, it also increased administrative costs (as discussed in Chapter Five). The very thorough
investigations may have also imposed greater burdens on claimants.
3 Recall that Reg E and EBT were implemented simultaneously in Hudson County, precluding the need
to distribute disclosure notices to clients who had already been to an EBT training session.
Prepared by Abt Associates Inc. 21
Chapter Two: Planning for and Implementing Regulation E
The following framework divides the sites' operating procedures into a series of steps
that generally conforms with the order in which chums were processed. The steps are
filing a Reg E claim;
follow-up contact with claimant;
further processing and investigation;
notifying claimant of decision;
providing provisional credit;
recovering a provisional credit;
handling client appeals;
administering corrective actions; and
tracking Reg E claims.
Each step is discussed below. More detailed descriptions of each site's Reg E operating
procedures are provided in Appendices A through C. Camden County's claim investigative
procedures are described in Appendix D.
Filing a Reg E Claim
Demonstration officials had to decide to whom clients should report incidents cf loss.
New Jersey decided that all claims from recipients in Hudson County should be reported to the
Help Desk operated by the EBT system vendor. Deluxe Data Systems. This provided a single
point of "entry" for all claims, making it easier to ensure that all claims were tracked and that
all necessary information was collected in a consistent manner. This approach came closest to
matching how reports of benefit loss were handled elsewhere in the state. Citibank also utilized
a central Help Desk to which all claims were reported.
New Mexico adopted a more decentralized approach for filing claims. Prior to Reg E
most clients reported problems with their EBT card or account to an "EBT specialist" at then-local
welfare office. Sometimes, however, they would call the state's central EBT Help Desk,
which operated mainly to answer questions from merchants and the EBT specialists. New
Mexico's disclosure notice instructed clients to report any problems with their EBT account to
the Help Desk, but New Mexico also trained the ymalwrs to handle calls involving Reg E-covered
losses (as well as calls to report lost or stolen cards). In these situations the specialists
Prepared by Abt Associates Inc.
Chapter Two: Planing for and Implementing Regulation E
were to fill out a special form and relay the information to the Help Desk, which forwarded the
information to the Reg E unit.
A second issue for this initial step in the process was determining what infounation
should be collected from the client. Generally, all the demonstration sites tried to have the client
describe the circumstances of the loss in some detail, both to aid further investigation and to
establish a "benchmark" explanation for later verification. During this initial contact the sins
also used administrative terminals to access their compeer system's history file to identify any
transactions in question. This sometimes refreshed the client's memory about a fotgotten
transaction, and the client realized that no loss has actually occurred. In such cases a forma]
Reg E claim was never filed.
Follow-up Contact with Claimant
In most instances the sites endeavored to have the client report what happened several
times during the overall investigation. If the client's story changed over time, the sites were less
likely to approve the claim. Citibank and New Mexico therefore asked the client to send in a
written statement of what happened, within specified time periods.6 In contrast, clients in
Hudson County were told to go to the Hudson County Investigative Unit (HCIU) to fill out and
sign a written affidavit of what happened. Clients in all sites were also asked to submit
supporting documentation (e.g.. receipts from transactions m question), if available If the claim
involved an unauthorized transaction, clients were often asked to file a police report and to
submit a copy of the report.
Further Processing and Investigation
After the client's written report of what happened was submitted, the sues conducted
further investigation if necessary. If the claim involved an ATM misdispense. the sacs—or their
EBT vendors—requested a report from the ATM network; the report verified whether a
misdispense actually occurred and. if so, the amount of the nusdwpeme
The requirement for a written suf-mrm is fully rouuf win normal Reg E wocedies ■ the prrvar
Prepared by Abt Associates Inc. 23
Chapter Two: Planning for and Implementing Regulation E
Other types of claims evoked yet another contact with the client, usually by telephone,
to request again a description of what happened or to pursue any apparent inconsistencies in the
client's previous descriptions. If the claim involved a disputed transaction at a POS terminal,
the sites sometimes interviewed the store clerk who handled the transaction. For clients who
claimed that they did not make the ATM transaction in question (as opposed to claims of ATM
misdispense), the sites sometimes sought a copy of any photograph taken by the ATM at the
time of the transaction. Such requests were rare, however, in part because photographs often
are not available.
Due to their different reimbursement policies for claims of unauthorized card usage,
Camden County and San Juan County did not investigate such claims as thoroughly as did the
full Reg E sites. When the Deluxe Help desk received calls from Camden County clients about
unauthorized card usage, clients were told that such losses were not reimbursable. For claims
of unauthorized card usage from San Juan County, the state's Reg E unit carried out an initial
investigation to determine whether the disputed transaction had been completed with a valid card
and PIN entry. If a valid card with PIN entry had been used, the loss was not reimbursable and
the investigation ended.
Notifying Claimant of Decision
After a claim was fully investigated, the sites sent a letter to the client indicating
whether the claim was approved or denied and, if approved, for what dollar value. The
demonstration sites varied in the practice of including reason for denial in their notice. New
Mexico and New Jersey usually indicated the reason for denial; Citibank usually did not,
although its notice indicated that clients could request copies of any documentation used by the
bank in making its decision.
Notice letters in both New Mexico and New Jersey indicated that the client had a right
to a fair hearing or appeal.
Providing Provisional Credit
Reg E stipulates that if a loss involving a transaction at an ATM cannot be fully
investigated within ten business days, the card issuer must grant a provisional credit to the client
for the full amount of the claimed loss pending a final decision. For losses involving
Prepared by Abt Associates Inc. 24
Chapter Two: Planning for and Implementing Regulation E
transactions at a POS terminal, the card issuer has 20 business days to complete an investigation
before a provisional credit must be granted. If a claim involves both ATM and POS
transactions, the 20-day timeframe applies.
All the demonstration sites (except Camden County) implemented procedures for
providing provisional credits to clients' EBT accounts when claims could not be fully
investigated within the Reg E deadlines.
Recovering a Provisional Credit
The sites varied in the procedures they followed to recover a provisional credit if the
claim was subsequently denied. EBT systems serving food stamp and AFDC clients have rules
specifying how quickly overpayments can be recovered; these rules conform to program
regulations governing benefit "recoupment." Whether the "overpayment" arises from a duplicate
or incorrectly calculated issuance, some other error, or a Reg E provisional credit, no more than
$10 or 10 percent (whichever is greater) of the client's monthly food stamp or AFDC allotment
can be recouped. If the client leaves the program before an overpayment is completely
recovered, there are three possible outcomes. First, the missing funds may never be recovered,
which leaves the agency with an unrecovered liability—one of the concerns of opponents of Reg
E. Second, if the client re-enters the program at a later date, the recoupment process can
continue; and third, an agency can initiate more traditional credit recovery procedures (e.g.,
turning the claim over to a commercial collection agency).
The programs participating in Citibank's DPC system do not have regulations analogous
to the recoupment procedures for food stamps and AFDC. Thus, if a provisional credit needs
to be recovered in the DPC system, Citibank can immediately debit the client's account for the
full reimbursement. If sufficient funds are not available for an immediate debit, Citibank is
allowed to debit the client's account after the clients' next issuance is posted to his or her DPC
account.
Handling Client Appeals
The food stamp and AFDC programs have procedures in place in which a client may
appeal an "adverse action," which may include a reduction in authorized benefits or a suspension
or termination of eligibility. One of the general Reg E policy decisions made prior to the start
Prepared by Abt Associates Inc. 25
Chapter Two: Planning for and Implementing Regulation E
of the demonstrations was that denial of a Reg E claim was an adverse action as well. The New
Mexico disclosure notice, therefore, indicated that a client had the right to appeal a denied
claim. The appeal initiated an administrative review of the circumstances of the claim, which
was conducted by the EBT project director in consultation with the Reg E investigator. If the
administrative review upheld the initial reason for denial, clients in New Mexico could request
a fair hearing. They could also request a fair hearing directly after receiving notice of the
denied claim (thereby skipping the less formal administrative review of the claim).
If a client requested a fair hearing, a date was set by the Hearings Bureau of the State
Human Services Department in Santa Fe. Fair hearings were held in person in Santa Fe or via
telephonic conference with a Reg E staff representative, the client, and a state hearing officer.
Both parties could present information about the claim during the hearing. The hearing officer
would then consider this information and make a determination either to support the original
decision or to overturn it.
There was no formal administrative review process in Hudson County; the disclosure
notice instructed clients to request a fair hearing if they disagreed with the decision to deny their
claim. No requests for fair hearings regarding claim decisions were made during the
demonstration.
Appeals in the DPC system were handled by the Citibank project manager; there was
no recourse to a formal fair bearing. Clients who were dissatisfied with the administrative
review of their claim could seek judicial relief. Clients in New Jersey and New Mexico, of
course, could also turn to the judicial process if they were not satisfied with a ruling by the Fair
Hearing Officer.
Administering Corrective Actions
The demonstration sites implementing Reg E wanted to incorporate actions to reduce
subsequent losses by a client. Each of the following potential corrective actions was available
for use by at least one of the demonstration sites: requiring additional training in how to use
one's card and keep one's PIN secure; restricting cash withdrawals to POS locations where a
store clerk can assist the client; requiring use of an authorized representative; bypassing the EBT
account tor some program funds by making direct restrictive payments (e.g., to a landlord); or—
Prepared by Abt Associates Inc. 26
Chapter Two: Planning for and Implementing Regulation E
where EBT participation was not mandatory—returning the client to a paper issuance system.
The only action actually taken by the demonstration sites, however, was additional EBT training.
In addition to corrective actions, New Mexico instituted what it hoped to be two
"preemptive" actions. Throughout Bernalillo County, clients paid $2 for each replacement card
they needed (beyond the first in any calendar quarter) due to a loss or theft of their old card.
In addition, near the end of the demonstration period, two of the four county offices in Bernalillo
also began issuing EBT cards containing the client's photograph. State officials hoped that each
of these measures would reduce rates of card loss, which in turn might also reduce unauthorized
usage of the cards.
Tracking Reg E Claims
New Jersey, New Mexico, and Citibank all used specially-designed and relatively
complex tracking systems to record detailed information about Reg E claims filed during their
demonstrations. New Jersey and New Mexico used PC-based systems that, for the most part,
collected information similar to that of the paper Reg E job ticket that Citibank used (see
Appendix C). Both the PC-based and paper-tracking systems provided the information required
by the evaluation to assess the impacts of Reg E on benefit replacements. There was no
evidence that any of the sites would maintain such detailed tracking systems in the absence of
the evaluation's need for the data.
In addition to the Reg E job tickets. Citibank continued using the same tracking system
that was in place before the start of the demonstration. All written documentation of claim
receipt and claim investigation was maintained in claim files, organized by the claimant's last
name. The EBT project manager also maintained an electronic spreadsheet of all claims of
unauthorized card usage; the spreadsheet maintained the following information: client's name
and system identification number, dates that oral and written notifications were received from
the client; dates that the investigation was completed and that a notice was sent to the client;
dollar amount reimbursed; and, if the claim was denied, the reason for denial. This information
was used to generate monthly reports concerning the number of unauthorized claims received
and the dollars of benefits replaced as a result of these claims.
Although Hudson County entered data about each claim onto the PC-based tracking
system developed for the demonstration's evaluation, the county did not use the information in
Prepared by Abt Associates Inc. 27
Chapter Two: Planning for and Implementing Regulation E
the database as a tracking system or management tool. The state's Division of Family
Development, however, had access to the Hudson County Reg E database and used it to generate
summary reports about Reg E claims. The reports indicated, by month and cumulative since
the start of the demonstration, the number of claims submitted by program, reasons for claims,
their disposition, and the dollar amounts of provisional credits, approved claims, and credits
being recouped.
The Reg E unit in New Mexico used its PC-based Reg E Tracking System (RETS) to
generate daily reports of pending claims, and monthly reports of the number of claims filed and
whether investigations were being completed within the 10- and 20-day time periods before
provisional credits needed to be granted. The Reg E project manager expressed a desire for a
tracking system that encompassed all EBT-related problems, with Reg E claims being an
identifiable subset of the entire database.
2.3 IMPLEMENTATION TASKS
After addressing the general policy issues related to implementing Reg E, and having
defined specific operating procedures, New Mexico and New Jersey were ready to implement
their demonstrations. Implementation tasks included:
• preparing a public notice and holding public hearings;
• developing necessary forms;
• establishing final staffing arrangements;
• obtaining required space and equipment;
• modifying client training materials; and
• training clients, Reg E staff, and other welfare office staff.
Prepare Public Notice and Hold Public Hearing
Depending on state law, proposed changes in policy, such as the implementation of Reg
E, might require public notice. Both New Mexico and New Jersey were required to publish the
proposed changes to their EBT systems and to allow for public comment. No comments were
received in response to New Mexico's public notice. Although a few comments were received
Prepared by Abt Associates Inc. 28
Chapter Two: Planning for and Implementing Regulation E
in response to New Jersey's notice, they pertained to EBT generally and not to Reg E.7 Thus,
this was a task that required relatively little time from demonstration staff and resulted in no
change in the proposed implementation of Reg E protections.
Develop Necessary Forms and Notices
Different forms and notices to support the implementation of Reg E or Reg E-like
protections were necessary. Examples of forms developed by New Mexico or New Jersey
include:
• a claim report to record information provided by the client when a loss was first
reported;
• an affidavit that a loss occurred (used only in Hudson County);
• notices to claimants indicating either that credit was being granted provisionally,
that the claim was approved, or that the claim was denied (and, possibly, reason
for denial); and
• internal forms for notifying other departments of actions that needed to be taken
(e.g., grant a provisional credit, initiate recoupment procedures).
The demonstration sites worked to integrate the Reg E operating documents with existing EBT
system operating procedures to avoid replication and incompatibility with existing system
documents and procedures.
Establish Staffing Arrangements
New Jersey, New Mexico, and the Citibank DPC system provide different models of
staffing for Reg E operations. Not surprisingly, the division of labor in the Reg E demonstra-tion
sites reflected the respective roles taken on by the states and their EBT vendors under
general EBT operations. For example, county agencies in New Jersey have greater autonomy
than their counterparts in New Mexico. Thus, in the Hudson County demonstration, county staff
had primary responsibility for investigating and processing Reg E claims. In New Mexico, on
7 New Jersey's public notice covered die state's overall implementation of its Families First system, with
special (i.e., Reg E) provisions noted for Hudson County.
Prepared by Abt Associates Inc. 29
Chapter Two: Planning for and Implementing Regulation E
the other hand, all EBT and Reg E functions were handled through a central, statewide EBT
unit.
In terms of the roles that the vendors played, the Texas DPC system was administered
by Citibank, which handled all facets of the operations including card distribution and
replacement, training, problem resolution, and investigations. In New Jersey, Deluxe Data
Systems was more involved in ongoing operations of the Reg E demonstration than was First
Security Bank, the vendor for the New Mexico EBT system. Again, this reflects the roles that
the vendors take under existing EBT operations. First Security Bank continued to operate the
New Mexico EBT system in much the same way as it did prior to the demonstration—handling
all EBT processing, and researching claims of ATM misdispenses. Nearly all other
investigations were handled by the Reg E unit. For the New Jersey system. Deluxe handled
initial EBT training and Deluxe staff performed all Help Desk functions.
Obtain Necessary Equipment
The demonstration sites in New Jersey and New Mexico found it necessary to purchase
computers and computer accessories to operate their tracking systems, and a phone line had to
be installed. There was also a need to obtain other types of equipment in order to implement
the demonstration interventions, such as the photo identification equipment in Bernalillo County
and a fax machine in Hudson County. A clear lesson from the demonstrations was that
acquisition of equipment can be a long process within state bureaucracies. The sites encountered
unanticipated delays in this step, resulting in frustration and delay of the demonstrations.
Modify EBT Client Training Materials and Train Clients
Citibank's DPC system training incorporated Reg E rights and responsibilities prior to
the demonstration; thus, no modifications to materials were necessary. Both New Jersey and
New Mexico had to revise their EBT training materials and procedures to include or expand
information related to Reg E issues (e.g., the importance of protecting one's PIN and reporting
any suspected loss immediately, how to report a loss, and to whom).
Train Clients. In terms of client training, the demonstration sites used two methods:
m-person and via mail. Clients of Citibank's DPC system were trained via the mail, receiving
a pamphlet explaining their rights and responsibilities. The pamphlet explained that if clients
Prepared by Abt Associates Inc. 30
Chapter Two: Planning for and Implementing Regulation E
reported a lost/stolen card within two days, their liability was limited to $50. The brochure also
provided an 800 number to call to report a loss. There was no retraining for clients with lost/
stolen cards.
Clients in both Hudson and Camden counties received EBT training in group sessions.
Those clients in Camden County who had already been trained on EBT received the disclosure
statement in the mail. The basic message in the training was that if clients believed they had
experienced a loss, they should file a claim as soon as possible.
In New Mexico, new clients received both EBT and Reg E information during in-person
training sessions conducted by their office's EBT specialist. Shortly after the demonstration
began, disclosure notices were mailed to all households already on EBT. The notices were also
distributed to all new clients and to clients comir into the offices to obtain replacement EBT
cards or for recertifications.
Both New Mexico and Hudson County found that Reg E added about five to ten minutes
to the client training.
Train Reg E Staff. Because Reg E had never been applied to state EBT systems
before. New Mexico and New Jersey had no formal training model to follow. (The Citibank
DPC system was not considered «i applicable model due to the difference in benefit systems and
clientele involved.) State staff in New Mexico, and state and county staff in New Jersey,
therefore, learned Reg E procedures over time as they made plans for their own demonstrations
and began operations.
The EBT specialists in the three demonstration counties in New Mexico did receive
formal training by the Reg E manager and coordinator in the month prior to demonstration start-up.
Although the EBT specialists were not officially "Reg E" staff, the specialists often served
as the fust point of contact with clients reporting losses. Thus, it was a must that the EBT
specialists be very clear on the policies regarding what qualified as a Reg E claim and on the
process for filing a claim.
Train Other Staff. Although Reg E tasks were handled by designated groups of staff
in the demonstrations, the importance of Reg E training for all staff who had contact with clients
became evident during the demonstrations. For example, in New Mexico, even though the Reg
E staff trained the EBT specialists, local office supervisors and caseworkers were never formally
trained on Reg E demonstration procedures. There were some instances in which a claim was
Prepared by Abt Associates Inc. 31
Chapter Two: Planning for and Implementing Regulation E
reported to the caseworker, but it was not forwarded to the Reg E staff in a timely fashion,
requiring a provisional credit to be issued.
County supervisors and case workers in Hudson County attended a general training
presentation when EBT was implemented. Both at this presentation and in follow-up memoranda
and meetings with the administrative supervisors, the chief investigator explained Reg E policy
and procedures to income maintenance staff.
2.4 CONCLUSIONS
This chapter has focused on efforts undertaken by federal and state administrators to
plan for and implement the Reg E demonstrations. In part, this material provides important
contextual information for understanding the demonstration impacts described in the next three
chapters. It was also meant, originally, to help states as they prepared to implement Reg E as
part of their EBT system operations.
Congressional action exempting state-administered EBT systems from the provisions of
Reg E, of course, reduces interest in the latter objective. Program administrators and client
advocates, however, are still interested in reducing benefit loss associated with EBT and in
helping those who incur loss. In Chapter Seven, we address some of the lessons from the
demonstrations that can help programs meet these goals. This current chapter, however, also
points out issues germane to helping clients avoid or respond to benefit loss, even in the absence
of Reg E requirements. In particular, the relevant questions that program administrators might
want to consider with EBT are:
• What types of benefit loss should be reimbursable under EBT? For losses not
reimbursable, what is the rationale for not offering clients some protection against
such losses?
• What is the most cost-effective way to inform clients of their rights and responsibil-ities
under EBT? Should a disclosure notice be provided and, if so, what
information should it provide?
• What is the most cost-effective way to collect information about claimed losses,
and how should such claims be handled? Within what timeframes (explicitly
published or not) should claims be processed?
• What actions (e.g., additional training, issuing photo EBT cards) can be taken to
reduce the likelihood or magnitude of benefit loss?
Prepared by Abt Associates Inc. 32
Chapter Two: Planning far and Implementing Regulation E
• On what basis should claims of benefit loss be approved or denied? What is an
appropriate level of documentary evidence?
The report will return to these and other related issues in Chapter Seven.
Prepared by Abt Associates bic. 33
BUNSfSiS
M
CHAPTER THREE
THE IMPACT OF REGULATION E ON BENEFIT CLAIMS
One of the biggest concerns program administrators had about Reg E was that it might
lead to a large—and perhaps huge—increase in submitted claims. This chapter explores the
impact of Reg E on the frequency of reported claims of loss. It also looks at the obverse
situation—the impact of Reg E on incidents of loss thai were not reported. If Reg E increased
reports of lost benefits, it presumably should have reduced the number of unreported losses as
well. This is particularly so for losses due to unauthorized card usage, a loss type not covered
by regular EBT protections.
If the introduction of Reg E led more clients to report incidents of lost benefits, one
would expect Camden County—as the only non-Reg E site in the demonstrations—to have the
lowest rate of reported loss across the six sites (and the highest rate of unreported loss).
Conversely, the four full Reg E sites would be expected to have the highest rates of reported
loss (and the lowest rates of unreported loss). San Juan County, with its intermediate
protections, would be expected to have a claim rate higher than Camden County, but lower than
the full Reg E sites. Similarly, its rate of unreported loss would be lower than Camden
County's, but higher than that found in the four full Reg E sites.
Demonstration data from the six sites are inconclusive with regard to whether Reg E
increased the number of claims submitted. Claim rates (measured as the number of submitted
claims per 1,000 cases per month) in some full Reg E sites were higher than in Camden County,
as expected, but other Reg E sites had claim rates lower than Camden's. Similarly, the
percentage of clients with unreported losses in some Reg E sites was lower than in Camden
County (again, as expected), but higher elsewhere.
What is lbsohitely clear, however, is that none of the sites experienced a large number
of claims during the demonstrations. Indeed, in what is probably the demonstrations' most
important finding, claim rates in the full Reg E sites and the responsibility standard site were
generally low. For claims involving lost cash assistance benefits, claim rates varied from 0.77
(Hudson County) to 3.38 (the Citibank DPC system) claims per 1,000 case months. Claim rates
for lost food stamp benefits were considerably lower; they varied from 0.22 (Dona Ana County)
Prepared by Abt Associates Inc. 35
Chapter Three: The lm\, -a of Regulation E on Benefit Claims
to 1.12 (Bernalillo County). For comparison, cash assistance and food stamp claim rates in
Camden County were 0.84 and 0.32, respectively. The most surprising feature of the Camden
experience is that, across the two programs, two thirds of all claims were for unauthorized card
usage, even though such losses were not reimbursable under regular EBT protections.
Of course, factors other than Reg E could have affected claim rates in the demonstration
sites. The six sites differed in terms of geography, urbanization, programs served, client
demographics, underlying crime rates, and a host of other factors. To help control for the
(unknown) effects of these factors on claim rates, we would have liked to compare claim rates
during the demonstration periods with claim rates from the same sites prior to the introduction
of Reg E. In general, this was not possible. Two of the six sites—Hudson County and the DPC
system in Texas—implemented EBT and Reg E simultaneously; thus, there is no "pre-Reg E"
claim experience in these two sites. In addition, historical data on claim rates in Camden
County were not available. Such data are available for the three sites in New Mexico, and they
initially suggest that Reg E did increase the number of claims submitted there. Problems with
comparability of the New Mexico data across time periods, however, reduce the strength of
these pre-post comparisons. Thus, although it is possible that Reg E increased claim rates in
New Mexico and elsewhere, an equally plausible interpretation of the data is that site differences
unrelated to Reg E account for die variation in claim rates.
3.1 DATA SOURCES AND RESEARCH APPROACH
Data Sources
This evaluation of Reg E's impact on claims of benefit loss is based on data from a
variety of sources. The majority of the data comes from claim tracking systems that were in
place in each of the sites during their 12-month demonstration periods. Although the design of
these systems varied, they collected very similar information. Citibank's DPC system used a
paper "job ticket" to track information regarding claims of benefit loss. (A sample job ticket
is included at the end of Appendix C to illustrate the type of information collected at each site.)
New Mexico and Hudson County used PC-based tracking systems to collect and store
information on claims of lost benefits. Finally, the claims data from the comparison site of
Camden County came from forms filled out by customer service representatives at Deluxe Data
Prepared by Abt Associates Inc.
Chapter-nine: The Impact of Regulation E am Btmejk Claims
Systems, New Jersey's EBT vendor. (A copy of the form is included a the end of Appendix
A.)
Another major source of data is a survey of EBT clients in each of the six sacs The
survey asked clients whether they had ever exneiieuted a loss of benefits and, if so, whether
they had always reported the loss. If any losses bad not been reported, the survey asked for
details about the loss (e.g., date it occurred, program involved, size of loss) and why k bad not
been reported.
Other sources of data used in this analysis include:
• Caseload counts from monthly program authorization files and from Citibank's
monthly EBT payment files (to enable computation of chum nates);
• Interviews with state, county, and vendor personnel;
• ATM misdispense records, caseload counts, and benefit replacement data from
Ramsey County;
• New Mexico EBT problem reports; and
• Interviews with a sample of claimants in each Reg E site.
Research Approach
Throughout this study of Reg E's impact on claims of lost benefits, separate analyses
are conducted for cash assistance and food stamp benefits. In addition, the analysis of claims
hy Mnjm— program k broken cut hy the four types of claims mmmmma m the previous chapter
(Exhibit 2-1):
(1) Claims arising from unauthorized usage of a client's EBT card;
(2) Claims arising from mom-receipt offunds (i.e., ATM nusdispenses).
(3) Claims arising from system or mmturn) ml errors (e.g., a transaction miMatraiy
entered twice at a store's EBT terminal); and
(4) Claims arising from other —— (e.g., employee theft of client benefits, or
forced transaction).
1 AddmooaJ infonnaooo about am survey is iwhriril m Appendix
Prepared by Abt Associates Inc. 37
L*J
Chapter Three: The Impact of Regulation E on Benefit Claims
The unit of observation used throughout is the individual claim, with claims that involve
losses of both AFDC and food stamp benefits treated as two separate claims. Finally, the basic
unit of measure used herein is the "rate" of claim submissions, which is the number of claims
submitted per 1,000 cases per month. This measure facilitates comparison of claim experiences
across sites and programs with different numbers of cases.
3.2 COMPONENTS OF REPORTED AND ACTUAL LOSS
It is instructive to begin by considering the components of reported and actual benefit
loss. This framework will help in formulating hypotheses about the possible impacts of Reg E
on benefit loss. (It should also help in understanding the uncertainty faced by demonstration
staff as they investigated reports of benefit loss.)
Consider the diagram in Exhibit 3-1. The horizontal bar (blocks A, B, and C)
represents all reported incidents of loss. The vertical bar (blocks B, D, and E) represents all
actual incidents of loss. Ideally, for any type of loss that is deemed reimbursable if verified,
one would like recipients to be reimbursed for all losses in blocks B, D, and E, but not for any
claims of loss in blocks A or C.
Block B represents all actual losses that are reported. Block C represents losses that
clients mistakenly believe occurred. (An example would be "loss" caused by an ATM
withdrawal or POS purchase that the client has forgotten.) Block A represents fraudulent reports
of loss; the client knows the loss did not occur, but reports a loss anyway in hopes of receiving
additional benefits.
Together, blocks D and E represent losses that occurred but which are not reported.
Clients know that some of these occurred (block D), but fail to report the losses for any number
of reasons, e.g., believed loss was not reimbursable, loss too small to bother with reporting, did
not know procedures for reporting loss. Block E represents losses that the client does not realize
occurred.
The prevalence of unreported losses and the reasons why they are not reported are
discussed at the end of this chapter. For now, it is sufficient to say that one objective of Reg
E is to minimize the size of block D by informing clients about which losses are reimbursable
and letting them know how to report these losses. The mechanism for doing so is the disclosure
notice that Reg E requires of all card issuers.
Prepared by Abt Associates Inc. 38
Chapter Three: The Impact of Regulation E on Benefit Claims
Exhibit 3-1
COMPONENTS OF REPORTED AND ACTUAL LOSS
B
KEY.
Reported Lou
Actual Lou
A + B + C - Reported loss of benefits.
B + D + E - Actual loss of benefits.
A - No loss occurred, but client fraudulently reports a loss.
B - Loss occurred and is reported.
C » No loss occurred, but client thinks it has and reports it.
D - Client knows loss occurred, but does not report the loss.
E - Client fails to recognize that loss occurred (and so does not report it).
For claims of loss that are filed, investigators have to decide which losses actually
occurred (block B) and which did not (blocks A and C). The appropriate action for both honest
mistakes and fraudulent claims is to deny the claim. If fraud is strongly suspected, of course,
program administrators may turn the case over to the Office of Inspector General (OIG) for
further investigation and possible sanctions against the claimant.
Given this structure, we can now identify some of the hypothesized impacts on benefit
loss and claim rates where Reg E protections are introduced (Exhibit 3-2). In general, one
would expect actual loss rates to be largely unaffected (except, perhaps, for losses due to
unauthorized card usage), whereas reports of loss would be higher, on average, than in areas
without Reg E protections. Expected rates of unreported losses would be lower. These impacts
are, of course, only hypotheses. To the extent that clients do not hear about or pay attention
to Reg E information about reimbursable claims or how to file a claim, any potential impact on
recognized but unreported losses will not materialize. Similarly, fraudulent claims will not
increase if hitherto honest clients are not enticed by the increased opportunities for filing claims,
nor will they increase if clients believe that fraudulent claims will be uncovered during claim
investigations.
Prepared by Abt Associates Inc. 39
Chapter Three: The Impact of Regulation E on Benefit Claims
Exhibit 3-2
HYPOTHESIZED IMPACTS OF REG E
Impacts on Actual Loss Rates
► The incidence of actual losses due to non-receipt of funds or system or procedural error should
not differ systematically between Reg E and non-Reg E locations. The factors affecting the
incidence of these types of losses (e.g., ATM reliability, double debits at stores) should not
be affected by the introduction of Reg E.
► The incidence of actual losses due to unauthorized card usage in Reg E locations might be
higher, lower, or the same as elsewhere. The new protections offered by Reg E could reduce
clients' incentive to take care of their cards, inasmuch as a portion of any resulting loss now
would be reimbursable. On the other hand, the sites' disclosure notices about Reg E, which
included reminders about taking care of EBT cards, could increase some clients' care of their
cards, reducing the opportunity for loss. Finally, both effects could be at work within
different segments of the caseload, leading to no net effect; or neither effect could materialize.
Impacts on the Reporting of Actual Losses
ft With increased emphasis on explaining which losses are reimbursable and how to file a claim,
the claim rate for losses due to non-receipt of funds and system or procedural error might be
higher in Reg E locations than elsewhere, even if the underlying rates of loss were the same.
► Assuming an increased awareness that losses due to unauthorized card usage are reimbursable,
the claim rate for such losses should be higher in Reg E locations than elsewhere.
► Claim rates, especially for losses due to unauthorized card usage, might be higher in sites
where EBT and Reg E are introduced simultaneously (e.g.. Citibank's DPC system and the
EBT system in Hudson County) than in sites where Reg E follows EBT implementation.
When implementation is not concurrent, some system participants may never learn of the new
protections offered by Reg E.
Impacts on Recognized, but Unreported, Losses
» With increased emphasis on explaining which losses are reimbursable and how to file a claim,
the frequency of recognized, but unreported, losses should be lower in Reg E locations than
elsewhere.
Impacts on the Incidence of Honest Mistakes
» The incidence of claims arising from honest mistakes might be higher for all types of loss,
with Reg E reminding clients that losses are reimbursable and explaining how losses are to be
reported.
Impacts on the Incidence of Fraudulent Claims
► The incidence of fraudulent claims of loss due to unauthorized card usage should be higher in
Reg E locations than elsewhere, solely due to such losses being reimbursable.
» The incidence of fraudulent claims of loss due to any reason might be higher in Reg E
locations than elsewhere, due to the availability of provisional credits if investigations cannot
be completed within required tuneframes.
Prepared by Abt Associates Inc. 40
Chapter Three: The Impact of Regulation E on Benefit Claims
3.3 FREQUENCY OF CLAIMS OF LOST BENEFITS
When the demonstrations began mere was concern that Reg E would lead to a very
large (but never specified) number of claims of loss being submitted to the welfare offices,
especially claims involving unauthorized card use. Contrary to these expectations, the total
numb