Performance audit, Department of Economic Security, Division of Benefits and Medical Eligibility

              PERFORMANCE AUDIT
DEPARTMENT OF ECONOMIC SECURITY
Division of Benefits and Medical Eligibility
Report to the Arizona Legislature
By the Auditor General
October 1 993
93- 7
DOUGLAS R. NORTON, CPA
lVDlTOR GENERAL
STATE OF ARIZONA
OFFICE OF THE
AUDITOR GENERAL
DEBRA K. DAVENPORT, CPA
DEPUTY I Y O I T O I GENErll\ L
October 19, 1993
Members of the Arizona Legislature
The Honorable Fife Symington, Governor
Dr. Linda J. Blessing, Director
Department of Economic Security
Transmitted herewith is a report of the Auditor General, A Performance Audit of the
Arizona Department of Economic Security ( DES), Division of Benefits and Medical
Eligibility. Ths report is in response to a December 13, 1991 resolution of the Joint
Legislative Oversight Committee.
DES is taking a number of important steps to improve efficiency and reduce errors
in its Public Assistance programs. Although DES can further improve on these efforts
by greater use of automation, it is making progress in these areas. However, DES'
current programs to address fraud and misrepresentation appear destined for limited
success at best. To more effectively address fraud, DES needs to change its focus from
attempting to detect fraud after it has occurred and place more emphasis on
prevention.
My staff and I will be pleased to discuss or clarify items in the report.
This report will be released to the public on October 20, 1993.
Sincerely,
DO&&& R. Norton
Auditor General
291 0 NORTH 44TH STREET . SUITE 410 9 PHOENIX, ARIZONA 85018 * ( 602) 553- 0333 FAX ( 602) 553- 0051
SUMMARY
The Office of the Auditor General has conducted a performance audit of the Arizona
Department of Economic Security ( DES), Division of Benefits and Medical Eligibility,
pursuant to a December 13, 1991, resolution of the Joint Legislative Oversight
Committee. This performance audit, the thrd in a series of audits of DES, was
conducted as part of the Sunset Review set forth in Arizona Revised Statutes ( A. R. S.)
§ § 41- 2951 through 41- 2957.
The Division of Benefits and Medical Eligibility ( DBME) is responsible for determining
eligibility for Federal and State public assistance programs. The most notable of the
programs administered by DBME include Aid to Families with Dependent Children
( AFDC), and Food Stamps, for whch over $ 650 million in benefits was issued in fiscal
year 1992- 1993. DBME also spends considerable resources determining eligibility for
medical assistance services through the Arizona Health Care Cost Containment System
( AHCCCS).
Arizona's public assistance programs have experienced tremendous growth over the
past five years. In fiscal year 1988- 1989, 800 eligibility interviewers handled
approximately 166,930 cases ( 209 cases per worker). By the end of fiscal year 1992- 1993,
1,415 eligibility interviewers handled over 377,000 cases ( 266 cases per worker). As its
programs have grown, DBME has had an increasingly difficult time providing timely
and accurate service. Program error rates ( resulting from both agency mistakes and
intentional or unintentional misrepresentation by clients) have steadily increased,
resulting in millions of dollars in welfare benefits issued erroneously and Federal
financial sanctions. In the last 3 Federal fiscal years, over $ 113 million in benefits were
erroneously issued.
DBME Is Taking Steps To Reduce
Error Rates ( see pages 11 through 16)
During the course of our audit, DBME began a series of internal studies and quality
initiatives to reduce agency- caused errors. Several of these initiatives focus on
important problems that our audit work also identified. For example, the Division is
attempting to improve the effectiveness of its quality assurance and quality control
systems by reviewing cases for errors prior to issuance of benefits. It is also
strengthening its ability to hold employees accountable for work quality and job
performance.
As DBME continues its quality improvement efforts, it should also consider other steps
that would further assist workers in the eligibility determination process. For instance,
DBME should consider pursuing an " expert" automated system that would reduce the
need for workers to make judgments and interpret policy. By consistently applying
appropriate program policies automatically, expert systems make the workers' jobs
easier and reduce errors. DBME should also consider enhancing its training program
for line workers and supervisors.
DES Needs To Strengthen Efforts
To Prevent And Detect Fraud ( see pages 17 through 27)
DES needs to strengthen its programs and activities for fraud prevention and detection.
Although no hard data exists on fraud in Arizona, official estimates of fraud range
from 2 to 5 percent of total cases. However, studies in other states and a survey of
Arizona eligibility workers suggest that fraud is much more prevalent. For example,
eligibility workers estimate 28 percent of their cases contain fraud. Regardless of which
estimate is used, the cost of fraud is significant. A conservative estimate of a 5 percent
fraud rate has an annual cost of $ 19 million. On the other hand, if eligibility workers'
estimates of a 28 percent fraud rate are correct, fraud could cost as much as $ 107
million per year. Further, once these dollars are spent, they probably cannot be
recovered.
DES's current approach to combating fraud, however, provides little deterrence to those
intent on committing fraud and does not place sufficient emphasis on prevention. The
Office of Special Investigations ( OSI), a unit within DES specifically charged with
investigating fraud, is not an effective deterrent to fraud because few of its
investigations result in any kind of penalty or sanction. In fiscal year 1991- 1992, OSI
opened over 17,000 cases for investigation, yet only 318 cases were referred for
prosecution and only 111 individuals were ultimately convicted and sentenced in that
year.
DES needs to strengthen its fraud prevention and detection activities to minimize the
amount of benefits paid out in error. Workers on the front- line need to improve their
investigative interviewing techruques. Workers can also conduct more thorough
verification to " double- check" the information provided by the client. For instance, one
worker regularly uses such techniques as checking job service records, calling
landlords, and checking addresses and telephone numbers.
In addition, OSI should take a more proactive role in fraud detection. Rather than
relying primarily on referrals from eligibility interviewers, OSI should also consider
utilizing additional methods of fraud detection, such as computer matching, error-prone
profiles, and random case checking.
TABLE OF CONTENTS
Page
INTRODUCTION AND BACKGROUND . . . . . . . . . . . . . . . . . . . . .
FINDING I: DBME IS TAKING STEPS
TOREDUCEERRORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Division Proposals Should
Impact Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Enhanced Automation And
Training Should Be Considered . . . . . . . . . . . . . . . . . . . . .
Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
FINDING II: DES NEEDS TO STRENGTHEN EFFORTS
TO PREVENTAND DETECTFRAUD . . . . . . . . . . . . . . . . . . . 17
Fraud Rate May Be Much
Higher Than Official Estimates . . . . . . . . . . . . . . . . . . . . . . . . 17
Current Approach To Fraud
Is Not Effective . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Arizona Should Refocus
Its Fraud Prevention And
Detection Efforts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
TECHNICAL APPENDIX
AGENCY RESPONSE
TABLES & FIGURES
Page
TABLE 1: Department Of Economic Security
Division Of Benefits And Medical Eligibility
Estimated Expenditures For Fiscal Year 1992- 93 . . . . . .
TABLE 2: Department Of Economic Security
Division Of Benefits And Medical Eligibility
Dollars Issued In Error
Federal Fiscal Years 1990 Through 1992 . . . . . . . .
TABLE 3: Eligibility Interviewers' Estimates Of
Percentage Of Cases
Containing Fraud/ Misrepresentation . . . . . . . . . . . . . .
TABLE 4: Eligibility Interviewers' Perception Of
Fraud Status In Arizona . . . . . . . . . . . . . . . . . . .
FIGURE 1: Food Stamp Error Rates
Federal Fiscal Years 1988 Through 1993 . . . . . . . .
FIGURE 2: AFDC Error Rates
Federal Fiscal Years 1988 Through 1993 . . . . . . . . . . .
FIGURE 3: Caseload Growth In AFDC, Food Stamps And
Medical Assistance Programs
For Fiscal Years 1989 Through 1993
And Projected For
Fiscal Years 1994 Through 1995 . . . . . . . . . . . . . . . . .
FIGURE 4: The " Funnel Effect" Of Fraud Investigations
And Prosecutions During
Fiscal Year 1992 . . . . . . . . . . . . . . . . . . . . . . . . . . .
INTRODUCTION AND BACKGROUND
The Office of the Auhtor General has conducted a performance audit of the h z o n a
Department of Economic Security ( DES), Division of Benefits and Mehcal
Eligibhty, pursuant to a December 13, 1991, resolution of the Joint Legislative
Oversight Committee. This performance auht, the third in a series on the
Department of Economic Security, was conducted as part of the Sunset Review set
forth in Anzona Revised Statutes ( A. R. S.) $ 541- 2951 through 41- 2957.
The Division of Benefits and Mehcal Eligibihty ( DBME) is responsible for
determining eligibihty for Federal and State public assistance programs aimed at
assisting inhviduals and families in meeting their immehate basic needs and
promoting self- sufficiency. The major public assistance programs for which DBME
determines eligibihty include:
AIDT O FAMILIEWSI TH DEPENDENCTH ILDRE( NA FDC) - The U. S. Department
of Health and Human Services ( HHS) regulates this program that provides cash
benefits to single- parent famhes with children under the age of 19. Cash
assistance, as a temporary means of support, continues until the family can
become independent.'
TWO- PARENETM PLOYMENPTR OGRAM( T PEP) - Part of the AFDC program,
TPEP provides cash benefits to two- parent families that need assistance in
meeting their needs until the parents' transition back into the labor force. Cash
benefits are limited to 6 months of payments in a 12- month period and are
issued on a semimonthly basis after work assignments are completed.
FOODS TAMPS( F S) - The U. S. Department of Agriculture, Food and Nutrition
Services ( FNS) establishes requirements for the food stamp program. Eligibhty
for benefits is based on resources, income, and other requirements such as
residence, citizenship or legal resident status, and cooperation with the DES
Employment and Training program.
MEDICALA SSISTANCE( M A) - Individuals approved for AFDC or TPEP are
automatically eligible for medical assistance and enrolled for a variety of mehcal
services through the h z o n a Health Care Cost Containment System ( AHCCCS).
AFDC- ineligible households may also q u a y for medical assistance upon a
review and eligibihty determination for Medical Assistance Only ( MAO).
1. As a condition of receiving benefits, recipients must participate in educational or work- related
programs that help them find and retain employment. However, some recipients are exempt from
participation for such reasons as: caring for a child under the age of 2, attending school full- time,
or being under the age of 16 or over the age of 59.
GENERAALS SISTANC( EG A) - State- funded General Assistance provides financial
assistance to individuals who are physically, mentally, and/ or socially
incapacitated to the degree they are unemployable. Inhviduals required to live
in the same home and provide custohal care to a ksabled person may also
qualify for General Assistance.
DISABILITY- F ederal financial assistance is also available to the disabled. The
Social Security Administration ( SSA) forwards claims for Supplemental Security
Income ( SSI) and Social Security Disabhty Insurance ( SSDI) to the State for
eligibhty determination. Once eligibility is determined, the claim is returned to
SSA for determination of benefit amount.
Organization And Staffing
Headed by an assistant director, the Division has an authorized full- time staffing
level of 2,709. The Division operates three sections:
FAMILYA SSISTANCAED MINISTRATIO( FNA A) - maintains responsibhty for all
program eligibhty determinations, except for potential SSI and SSDI recipients.
Its 2,492 staff operate in approximately 70 field offices across the State and a
central office located in Phoenix.' The majority of these employees ( 1,451) are
eligibhty interviewers ( EI's), who are responsible for interviewing clients,
processing cases, and determining program eligibility and benefit amounts.
DISABILITDYE TERMINATISOENR VICEA DMINISTRATIO( DND SA) - determines
eligibhty, based on hsabhties, for SSI and SSDI recipients. The administration
maintains a staffing level of 156 employees allocated between Phoenix and
Tucson.
OFFICEO F PROGRAMEV ALUATIO( ONP E) - conducts the quahty control function
for FAA client cases to identlfy the causes of repetitive errors and to eliminate
those causes through appropriate corrective action plans. Results are transmitted
to the Federal agencies responsible for determining program error rates and are
subsequently used to determine sanctions or incentive payments to the State.
OPE maintains a staffing level of 61 employees located among Phoenix, Tucson,
and Flagstaff.
1. Itinerant visits are also made to 100 rural sites across the State on a regular basis.
Funding
The Division receives both State and Federal fundmg. The Federal government
provides approximately 50 percent of the necessary fundmg for AF'DC and Food
Stamp program administrative costs; however, funhng for actual recipient benefits
varies by program. For AFDC, the U. S. Department of Health and Human Services
provides approximately 66 percent of the funchng for benefits, while the Department
of Agnculture, Food and Nutrition Services fully funds food stamps. As shown in
Table 1, the cost of administering and funhng public assistance programs in
Anzona is projected to total nearly $ 754 mlllion in fiscal year 1992- 1993. Of this,
nearly $ 674 mdhon actudy goes to welfare recipients.
TABLE 1
Department Of Economic Security
Division Of Benefits And Medical Eligibility
Estimated Budget For Fiscal Year 1992- 1 993
( unaudited)
FAA State Federal Total
- Food Stamps $ 11,377,438 $ 405,033,019 $ 41 6,410,457
- AFDC 99,467,953 185,729,743 285,197,696
- Medical 9,739,250 9,739,250 19,478,500
- GA 17,422,887 0 17,422,887
- Other ( a) 648,174 53 648,227
Subtotal $ 1 38.655.702 $ 600.502.065 $ 739.157.767
OPE 932,100 1,381,000 2,313,100
DDS A 185,576 12,189,556 12,375,132
Total $ 1 39,773,378 $ 61 4.072.621 $ 753.845.999
( a) " Other" public assistance programs provide benefits to individuals needing emergency
assistance for circumstances such as eviction or utility shutoff. In addition, individuals
certified as having tuberculosis may also receive cash assistance.
Source: Division of Benefits and Medical Eligibility, Operations Support Budget Report
for fiscal year 1992- 1993, Office of Program Evaluation, and the Disability
Determination Services Administration.
Poor Performance
Has Costly Impact
Historically, Arizona has experienced problems in providing accurate and timely
welfare benefits. Errors in the issuance of welfare benefits are attributed to either
the agency or the client. Agency errors are often due to misapplication of pohcies,
fdure to act on information provided by the clients, and fdure to gather all the
necessary information when determining elig- lbhty. Client errors result when the
client intentionally or unintentionally fads to report or misrepresents information
at the time of application and/ or when changes in status occur. Based on quahty
control reviews from October 1991 through March 1993, 70 percent of case errors
were agency caused and 30 percent were client caused.
As shown in Figure 1, Food Stamp program error rates for Federal fiscal years 1988
through 1993 have increased, while national tolerance levels ( i. e., error rate
standards) have decreased and/ or remained constant. Figure 2 ( see page 5)
Illustrates AFDC program error rates for the same time period. While AFDC error
rates have generally decreased, they stdl remain well above tolerance levels.
Figure 1
Food Stamp Error Rates
Federal Fiscal Years 1988 Through 1993
14.00 1 I I I I
I Nat'l Tolerance ~ e v e l ( a ) l l 10.97 ) 10.80 1 10.80 / 10.31 1 10.31 10.31 1 Arizona Error Ratem
( a) The National Tolerance Level is the national error rate ( as determined by FNS)
+ 1 percent; however, the tolerance level cannot increase past a previously established
national error rate.
( b) The error rate for fiscal year 1993 is an estimate as of June 21, 1993.
Source: Division of Benefits & Medical Eligibility and the U. S. Department of Agriculture, Food and ,
Nutrition Services.
FYI988
9.85
FYI989
10.58
FYI990
10.93
FY1991 FY1992
11.23 / 13.35
FY1993( b)
10.89
Figure 2
AFDC Error Rates
Federal Fiscal Years 1988 Through 1993
( a) Health and Human Services( HHS) projects fiscal year 1991 Tolerance Level results by the
Summer of 1993.
( b) The fiscal year 1991 and fiscal year 1992 results are not yet final pending official release from
HHS. The error rate for fiscal year 1993 is an estimate as of June 21, 1993.
( c) National average ( Total Dollars in Errorfrotat Dollars Issued) or 4 percent, whichever is higher.
Arizona Error RateCl
Nat'l Tolerance Level( c) l
Source: Division of Benefits and Medical Eligibility and U. S. Department of Health and Human
Services.
These high error rates produce a signficant amount of misspent dollars annually.
Some eligible clients received too much in benefits, while others received benefits
for which they were not eligible. As shown in Table 2 ( see page 6), over $ 113
mfion was misspent in both the Food Stamp and AFDC programs during Federal
fiscal years 1990- 92.
10.93
6.79
Federal monetary sanctions add further to the cost of Arizona's high error rates.
The Division was sanctioned $ 311,295 in Federal fiscal year 1989- 1990 and
$ 2,848,469 in Federal fiscal year 1990- 1991 for high error rates in its food stamp
program.' Adhtionally, the Division recently received notice of a nearly $ 11.5
mfion sanction for its continually high food stamp error rate for Federal fiscal year
199 1- 1992. Additional dollar sanctions could be forthcoming for Federal fiscal year
1992- 1993 based on the error rate to date.
1. The State and FNS reached a negotiated settlement of $ 473,965 to be paid by the State over the next
5 years. These monies will pay for enhancement to the current automated system and quality
assurance staff to identify how and why errors occur. The Division has avoided sanctions associated
with its hgh AFDC error rates due to waivers from the Federal government for fiscal years 1988,
1989, and 1990.
9.89
5.70
7.81
5.98
8.34b)
NIA
6.2qb)
NIA
9.25 ( b)
NIA
-
Table 2
Department Of Economic Security
Division Of Benefits And Medical Eligibility
Dollars Issued In Error
Federal Fiscal Years 1990 Throuah 1992
Federal Food
Fiscal Year AFDC Stamps Total
1990 $ 1 0,570,391 $ 17,389,793 $ 27,960,184
1991 13,183,662 23,747,675 36,931,337
1992 14,901,542 33,384,075 48,285,617
Total $ 38.655.595 $ 74.521.543 $ 113.177.138
Source: Ad Hoc report on error dollars paid during Federal fiscal years 1990- 1992, prepared by
Division of Benefits and Medical Eligibility, Operation Support
Factors Impacting
Error Rates
Tremendous caseload growth, the Division's focus on timeliness, lagging staff levels,
and program complexity combine to dramatically impact program error rates and the
Division's ability to efficiently and effectively serve clients.
CASELOADGR OWTH - As portrayed in Figure 3 ( see page 7), Food Stamp and AFDC
caseloads have doubled, while Medical Assistance caseloads have more than tripled
from fiscal years 1989 through 1993. This level of growth, which is expected to
continue during fiscal years 1994 and 1995, places greater strain on limited staff
resources and impacts the Division's ability to give clients high- quality, timely
service.
TIMELINES- S Further compounding Division efforts to improve quality is the
Division's focus on timeliness. This focus evolved from court actions brought
against the State in 1981 and 1982 mandating compliance with Federal timeliness
requirements1 As a result, the Division's emphasis has been on case timeliness,
often to the detriment of quality case preparation.
Figure 3
Caseload Growth In AFDC, Food Stamps And
Medical Assistance Programs
Actuals For Fiscal Years 1989 Through 1993
And Projected For
Fiscal Years 1994 Through 1995
Source: Division of Benefits and Medical Eligibility, Operation Support.
Note: Fiscal years 1994 and 1995 are estimates.
I. Food stamp applications must be processed withn 30 days, although in cases deemed to be
emergencies, applications must be processed and benefits received withn 5 days. AFDC applications
must be processed within 45 days.
STAFFING- While the Division has experienced dramatic caseload growth and
requirements to remain timely, staffing levels have failed to keep pace, thus placing
greater demands on the eligibility interviewer ( EI). In fiscal years 1991- 1992 and
1992- 1993, growth in EI positions averaged 9.4 percent, well below the caseload
growth experienced in each of the programs. Tlus trend is expected to continue. No
additional funding was appropriated for staff by the Legislature for fiscal year
1993- 1994 to handle future increases in caseload growth.
PROGRAMC OMPLEXIT- Y F inally, the complexity associated with the programs
further contributes to high program error rates. Although many individuals apply
and are eligible for both AFDC and food stamp benefits, over 50 differences in
policy between the 2 programs complicate and lengthen the eligibility determination
process. Constant changes in policy further add to the problem by requiring
workers to read, understand, and apply all policy revisions. In 1992 alone, EI's
received over 100 documents outlining either new or revised policies.
Audit Scope
Prior to initiation of this audit, a new DES Director and Assistant Director for the
Division were appointed. Both have aggressively pursued changes to better serve
clients and reduce program error rates. For example, several quality improvement
committees were established and actively working during our audit. This report
acknowledges the efforts made to effect change within DBME, and proposes additional
actions that might positively impact Division performance.
Our report, focusing on the sources of program error rates ( the agency and the client)
presents findings and recommendations in two areas:
Additional steps DBME should consider taking to reduce agency- caused errors, and
The need to develop more effective methods to prevent and detect client fraud.
To help us determine how well DBME has assisted persons needing welfare benefits,
we contacted other offices within DES, as well as the Attorney General's office. The
offices contacted were: the Office of Accounts Receivable and Collections ( OARC),
which maintains responsibility for collecting overpayments made to welfare recipients;
the Office of Special Investigations ( OSI), which maintains responsibility for
investigating alleged perpetrators of welfare fraud; and the Office of Appeals, which
hears appeals on behalf of welfare recipients.
The audit was conducted in accordance with government auditing standards.
The Auditor General and staff express appreciation to the Director of the Arizona
Department of Economic Security and the Assistant Director and staff of the Division
of Benefits and Medical Eligibility, as well as the numerous staff in other DES
Divisions and offices for their cooperation and assistance during the audit.
FINDING I
DBME IS TAKING STEPS
TO REDUCE ERRORS
Recently, DBME has initiated significant steps to reduce the number of agency errors.
The Division is redesigning its quality control/ case review system, strengthening
accountability for individual worker performance, and streamlining the client
application process. To supplement these efforts, the Division should also consider
pursuing an " expert" automated system that will simplify and expedite the eligibility
interviewer's job, and strengthening its employee training program.
Early in our audit, we spent considerable time reviewing operations at the Division,
district, and local office levels. We visited numerous local offices, interviewing various
staff and observing EI's performing their jobs. Based on ths work, we identified and
began detailed audit work in a number of areas, including the effectiveness of quality
assurance and quality control systems, internal efforts to address program error rates,
and the impact of employee discipline and accountability on job performance.
However, at the same time, DBME established internal review committees to study,
develop, and implement recommendations for improvement in many of these same
areas. Therefore, to avoid duplication of effort, we reduced the scope of our work and
focused on areas that would provide additional opportunities for improved
performance and error reduction.
Division Proposals Should
Impact Performance
Facing hgh program error rates, past and pending Federal monetary sanctions,
millions in misspent dollars, and dissatisfied clients, DBME has developed several
initiatives to address its historically poor performance. These initiatives seek to reduce
error rates by refocusing quality review and assurance, strengthening employee
accountability, and improving client service.
Qualim review and assurance - The Division has refocused its efforts in the quality
control/ case review process on error prevention. This process consists of daily
supervisory reviews of randomly selected client case files worked by EI's for
completeness and accuracy. In the past, these quality reviews were performed after
benefits had already been issued. Under the new process, the focus of quality review
and assurance will be on EI's and the expectation that they will prepare a quality
product prior to benefit issuance. The following activities should assist EI's in this
objective:
SUPERVISOWRSIL LR EVIEW CASESP RIORT O BENEFIT ISSUANCE - In addition to
preventing benefits from being issued erroneously, the immediate feedback from
these reviews is intended to help the EI understand the error and the reason it was
made.
FOCUS ON MOST COMMON ERRORS - case reviews will focus on the four most
common errors in determining client eligibility, such as improper calculation of
wages and salaries and inaccurate accounting of all eligible members.' This will
allow supervisors to focus their attention on problem areas.
AN ERROR PREVENTION TEAM - will be created for the purpose of determining and
addressing the root causes of errors.
AN ERROR PREVENTIOSNT EERINCGO MMITTEE was formed to monitor error
reduction initiatives, review error root cause information and analysis, and review
recommendations for further improvement of procedures and processes.
Emplwee accountabiliht - The Division has also developed and implemented several
initiatives designed to increase employee accountability for Division performance and
program error rates:
A progressive disciplinary action plan was developed that empowers supervisors
and management to more effectively deal with problem employees. DBME
previously had difficulty dealing with employees who were not performing
adequately ( i. e., were not processing cases timely, creating errors, etc.). Additionally,
DOA Personnel and the Attorney General's office were heavily involved with
disciplinary actions, which created an extremely bureaucratic and lengthy process.
DBME has now streamlined the process, reducing the time required to implement
discipline from 120 to 75 days.
The Division will now require strict adherence to its operating manual for field
offices. The Local Office Procedures ( LOP) manual provides standardized practices
for processing cases. The Division has found that strict LOP compliance contributes
to improved performance, while deviations from LOP negatively impact
performance and impair employee accountability.
1. Previously, supervisory review of cases encompassed 33 case characteristics. However, DBME found
that by focusing on only a few case characteristics that yield the most errors, the same benefits of
a full review can be realized.
Client service - Finally, the Division has developed a plan to streamline the client
application process in an effort to improve performance and simplify the EI's job. The
objective of this plan is to reduce client visits from 4 days to 1 day and average
processing time from 26 days to 1 day.' In order to meet this goal, the following
adjustments to the current process will be made:
Appointment scheduling will be eliminated, allowing clients to be seen the same
day they complete an application.
More experienced staff will be available in office lobbies to assist clients in the
application process.
8 The EI's job will be simplified and accelerated by reducing paperwork, reorganizing
its flow, and allowing immediate input of client information on the computer
during the interview. Clients will be informed after the interview as to their
program eligibility and amount of benefits, subject to verification of information
provided by the client.
The Division will undertake an intensive client education campaign, with the
assistance of community leaders and organizations. The campaign will stress the
importance of providing the necessary documentation at the time of the interview
to facilitate the application and eligibility determination process.
These initiatives are in various stages of implementation. While some will be piloted
in a local office prior to implementation, others have already been implemented
Statewide. According to DBME management, these initiatives are expected to impact
error rate performance indicators beginning in late 1993.
Initiatives should be properlv implemented and fullv supported - While these initiatives
are promising, the Division will need to manage and monitor their implementation to
ensure success. According to both DES and Federal officials, previous efforts to
improve performance and impact program error rates have been unsuccessful or
abandoned due to lack of commitment at all levels of the organization, lack of
organizational support, and lack of staff and management accountability. To avoid
these pitfalls, management needs to secure organizational commitment, support, and
accountability by continually soliciting staff input. In addition, DBME needs to monitor,
evaluate, and revise its initiatives based on interim results.
1. We anticipate some additional processing time would be required to implement the
recommendations in Finding I1 ( see pages 17 through 27).
13
Enhanced Automation And
Training Should Be Considered
We identified other efforts DBME should also consider to help reduce agency errors.
For example, automation and training present additional opportunities to further
enhance the EI's ability to handle their growing and complex caseloads more
effectively.
Automation - Upgrading current automation capabilities may give EI's the ability to
better handle increasing caseloads and produce quality work. Although DBME has
several automated systems to assist the EI in the eligibility determination process, they
are somewhat duplicative, difficult to use, and do not necessarily simplify the EI's
work. Even with these various systems, the Division remains dependent on EI's to
know and correctly apply all of the rules and regulations for each program. Ths often
results in the inconsistent and/ or incorrect application of policy, ultimately leading to
errors.
Recent technology advancements in automated systems present a unique option to
expand the skills of EI's and provide them with a tool to further impact program error
rates. Currently, there are " expert" systems in various stages of use and development
that are designed to simplify the EI's job, reduce the propensity for EI errors, and
accelerate the eligibility determination process. These systems all but eliminate the need
for EI's to remember, interpret, and appropriately apply overwhelming and complex
program policy. For instance, Texas utilizes an " expert" system that prompts the
eligibility worker through the interview process, applies appropriate program policy,
determines client program eligibility, and computes budget and benefit levels. Texas
has found that its system increased staff productivity by 20 percent despite dramatic
increases in size and compIexity of caseloads. The system has also reduced
administrative costs, and enhanced fraud detection.
Similarly, both Merced and Tulare counties in California have implemented " expert"
systems. Merced County implemented its " expert" system in response to an increasing
caseload, high staff turnover, and dependence on eligibility workers to manage more
than 6,000 eligibility rules. Based on applicant responses, the system will apply
eligibility rules, determine follow- up questions, and ensure that all necessary
information is obtained to make an accurate and appropriate eligibility decision. While
Merced County has experienced some difficulties with the system in performing all of
the functions for which it was designed, the concept remains viable and appears
promising.
Although different from the Merced system, Tulare County also uses an " expert"
system to facilitate the eligibility determination process. The Tulare system allows
potential clients to do much of the up- front work without relying on state eligibility
workers. Clients are provided access to computer " touch" screens on which they
answer questions that will determine if they are eligible for public assistance. This
allows eligibility workers to spend more time verifying information provided by
applicants.
Tulare County estimates an annual cost savings of nearly $ 2 million as a result of
implementing ths system. Savings have resulted from both reductions in program
errors and staffing and administrative costs. Other benefits that have been realized
include significant enhancement in the timeliness and efficiency of case processing,
uniform application of program policies and regulations, and better fraud prevention
and detection.
Cost estimates for implementing " expert" systems range from $ 3.2 to $ 6 million.
However, the cost to the State could be effectively reduced to $ 1.6 million to $ 3
million, based on a Federal participation rate of 50 percent in the project.
Training - DBME should consider enhancing its training for new eligibility
interviewers. Currently, training consists of classroom lectures during the course of the
EI's first year on the job. The training program focuses on the major public assistance
programs, automated systems, interviewing, and client application processing.
However, many EI's have expressed concern that tlus training curriculum fails to
address the realities of the job, such as high caseloads, complex programs, and constant
time pressures. They want something that more closely resembles actual working
conditions.
Other states' training programs combine classroom lectures with " real life experience"
to better prepare workers for the casework they will be expected to perform. San
Diego County, California, incorporates on- the- job training with classroom- type learning.
New workers actually process cases, interview clients, and input information on
computers while being supervised. According to a San Diego County official, the
advantage of this type of training is the preparation provided to new workers for what
to expect on the job. They are not so overwhelmed, and function better when they first
start handling cases in the " real world.'' Texas also utilizes a combined
classroom/ on- the- job training approach. According to a Texas official, staff training has
been a very effective tool in reducing errors. Training evaluations indicate that the
most effective training occurs when a trainer or facilitator is involved and a protected
environment is provided.
Some districts and offices in Arizona have taken similar approaches by providing
additional training for new EI's to better prepare them for the responsibilities,
expectations, and pressures of the job. For instance, one district has established a
transitional training program consisting of on- the- job training and further enhancement
of topics covered during initial training. Under close supervision, training participants
interview clients, process cases, and determine client program eligibility. Several other
local offices employ " baby units," which provide new EI's the opportunity to work with
smaller caseloads under closer supervision.
Supervisory training could also be improved. Current supervisory training, offered two
to three months after a supervisor has been on the job, misses some important content
areas and skills important to the supervisor's job. Both supervisors and management
would like supervisory training to become more timely ( offered to EI's before or upon
promotion). In addition, they recommended that topics relevant to the job, such as
problem solving, motivation of employees, interpersonal and organizational skills, and
time management be added to the core curriculum.
RECOMMENDATIONS
1. DBME should continue its efforts to address and impact program error rates. In
doing so, Division management should secure organizational commitment and
support for initiatives and should continuously monitor, evaluate, and analyze the
impact these initiatives are having on division performance.
2. DBME should consider pursuing upgraded automation capabilities such as an
" expert" system to simplify the job of the eligibility interviewer and allow workers
to handle their increasing caseloads more effectively.
3. DBME should revise training for both EI's and supervisors to incorporate on- the- job
training, make classroom training timely, and expand training content areas.
FINDING II
DES NEEDS TO STRENGTHEN EFFORTS
TO PREVENT AND DETECT FRAUD
While DES has begun major efforts to address its errors, it needs to strengthen
programs designed to combat client fraud. Studies conducted in other states and a
survey of h z o n a eligibhty workers suggest that fraud is much more prevalent
than official estimates indicate. DES's current approach to fraud provides little
deterrence, and does not place sufficient emphasis on prevention by line workers.
Workers can do much more to prevent and detect fraud before benefits are paid in
error.
Arizona criminal statutes define benefits fraud as " knowingly obtaining benefits by
means of false or fraudulent pretenses, representations, promises or material
omissions. " Common types of fraud or misrepresent ation include not fully reporting
all income and assets avdable to the client, or not accurately cLzsclosing the number
of household members eligible to receive benefits. Federal regulations require states
to pursue action against individuals intentionally committing such fraudulent acts
by either administrative cLzsqualification from the program, or criminal prosecution.
States are also expected to recover the dollars paid out in error.
Fraud Rate May Be Much
Higher Than Official Estimates
Although no studies have been done specifically to establish the prevalence of fraud
in h z o n a or nationwide, fraud may be much more widespread than has been
previously inchcated by social service authorities. Both Federal and State official
estimates suggest that fraud occurs in only 2 to 5 percent of welfare cases.
However, stuches in other states and a survey of h z o n a eligibihty workers inhcate
that the occurrence of welfare fraud may be considerably higher.
Official estimates - Federal and State officials estimate low rates of fraud. Federal
officials we contacted estimated that, nationwide, about 2 to 5 percent of welfare
cases contain some form of fraud or misrepresentation. These estimates are based
on the Federally mandated annual Quality Control ( QC) reviews designed to
determine program error rates and identify their causes. Likewise, Arizona QC data
reveal a relatively low fraud rate. Based on QC reviews from February 1991 to
January 1993, less than 2 percent of the client- caused errors identfied were
attributed to clients' M u 1 misrepresentation ( i. e., fraud).
However, there are serious limitations to relying on QC data to determine the
extent of fraud. First, the intent of a QC review is to ensure that the work done by
the eligibihty workers is accurate; the purpose of the review is not to detect fraud.
Moreover, many Arizona QC reviewers were initially trained to use the error
codes identifying fraud. One reviewer indicated that she has been doing QC reviews
for eight years and has never coded a client error as willful misrepresentation.
Finally, accorchng to some QC reviewers, they rarely document fraud, unless it is
blatant. According to a QC administrator, errors identified ( both agency and client)
are passed on to the local office, and it is then the local office's responsibihty to
refer the case for fraud investigation.
Other state studies and EI estimates - Information obtained from several states
that have stuched fraud and a survey of Anzona eligibhty workers suggest that the
prevalence of fraud may be much higher than official estimates. For example:
Nevada randomly selected 100 cases to investigate in a large, metropolitan office
to get an idea of the amount of fraud present. As a result of these investigations,
57 cases were ultimately closed or denied due to client fraud, misrepresentation,
or fdure to report changes.
The Florida Auchtor General's Office, charged with the responsibility to
investigate welfare fraud, periochcally reviews cases within the food stamp
program to determine a probable fraud rate. In its most recent statistical
analysis, a probable fraud rate of 11.3 percent for food stamp cases was
projected.
A 1992 San Diego County, Cahfornia grand jury report conservatively estimated
fraud at 10 percent, at a cost of $ 70 maon per year to San Diego County. The
report also found that the county social services agency failed to determine
accurately the level of welfare fraud and consistently reported rates of fraud
proven to be erroneously low.
In Arizona, many EI's believe fraud is prevalent. Because DES has not attempted
to quantify the amount of fraud in Anzona, we surveyed EI's and their supervisors
who interact with clients virtually on a dady basis to obtain the front- line workers'
perception of fraud from within the welfare system.' As shown in Table 3 ( see page
19), over 50 percent believe more than 1 out of every 4 cases involve some form of
fraud or misrepresentation. Moreover, as shown in Table 4 ( see page 20), an
overwhelming majority of the workers ( 68 percent) perceive the amount of fraud to
be increasing.
1. We conducted a survey of all EI's and EI supervisors located in field offices across the State
( approximately 1,517 full- time positions). Three hundred forty- nine responses were received. We
then contacted 70 of the nonrespondents to check for nonresponse bias and found their responses
to be slightly lower -- a median fraud estimate of 23 percent.
Regardless of which estimate is used, the cost of fraud is significant. A conservative
5 percent fraud rate would mean that approximately 12,500 cases are fraudulent,
costing $ 19 million annually. On the other hand, if EI estimates of a 28 percent
fraud rate are correct, we estimated that 70,346 cases, costing as much as $ 107
mdhon, may contain some form of fraud or misrepresentation. ( See Technical
Appendix).
Table 3
Eligibility Interviewers' Estimates
Of Percentage Of Cases Containing
Number of El Percent of El's
Percent of Cases Responses Respondinq
0 - 10% 5 1 15%
11 - 25% 112 32 %
26 - 50% ( a) 107 3 1 ' 10
51 - 75% 55 16%
76 - 100% - 20 6%
Total 345 100%
( a) The median estimate is 28 percent.
Source: Auditor General's March 1993 survey of Eligibility Interviewers and Eligibility
Supervisors.
Table 4
Eligibility Interviewers' Perception
Of Fraud Status In Arizona
7
Number of Percentage of
Status of Fraud El Responses Respondents
Increasing 232 68%
Unchanged 97 29%
Decreasing - 11 3%
Total = 340 100%
Source: Auditor General's March 1993 Survey of Eligibility Interviewers and Eligibility
Interviewer Supervisors.
Current Approach To
Fraud Is Not Effective
DES's current approach to combating fraud provides little deterrence to those intent
on committing fraud and does not place sufficient emphasis on prevention. Fraud
deterrence is limited because only a small proportion of cases investigated result in any
kind of penalty or sanction. Further, DBME does not place sufficient responsibility on
line workers for preventing fraud.
The Office of Special Investigations ( OSI), a separate unit within DES, has primary
responsibility for investigating welfare fraud. OSI has 35 investigators who pursue
cases after they are referred by EI's in the local offices.' If an EI encounters conflicting
or suspicious information in the course of processing a case, the EI may prepare a
referral to OSI for an " early fraud" investigation. OSI investigators attempt to either
confirm or deny the presence of fraud or misrepresentation, and then return the case
to the EI for further action. OSI also investigates on- going cases for potential criminal
prosecution.
While many cases investigated by OSI are found to contain inconsistent or inaccurate
information, the number of cases in which fraud can eventually be substantiated is
significantly less. This is due to several factors. First, once the investigator identifies
1. OSI has a total of 86 full- time equivalent ( FTE) positions. The 35 fraud detection investigators are
located in local offices, primarily in Maricopa and Pima Counties. The remaining 51 FTEs are
located in Phoenix and include administrative and support staff, as well as investigative staff who
handle forgery cases and further investigation of cases being prepared for prosecution.
a problem, the client must be notified and given an opportunity to clarify the
conflicting information. If the client disputes what the investigator found, or provides
information different from that whch led to the investigation, the case remains open.
Second, if a client cannot be contacted after three attempts, the investigation is closed.
Finally, an investigator may have valid reasons to suspect a case is fraudulent, but is
unable to obtain sufficient documentation to confirm his or her suspicions. Therefore,
further action on the case is unlikely.
Ineffective deterrence - Despite the commitment of significant resources, few
investigations result in any kind of penalty or sanction. There are two avenues for
pursuing welfare fraud: criminal and administrative. The Arizona Attorney General's
Office is responsible for criminal prosecution. Currently, it limits its cases to those
where client error has resulted in $ 1,500 or more in overpaid benefits.' As shown in
Figure 4 ( see page 22) this produces a funnel like- effect on welfare fraud cases. For
instance, while in 4,327 fraudulent cases an action against the client was pursued in
fiscal year 1991- 92, only 318 cases were referred to the Attorney General's Office for
prosecution. Additionally, only 111 individuals were convicted of welfare fraud and
sentenced in that year. Each of these cases averaged about $ 3,500 in client- caused
overpayments. Typically, criminal sanctions imposed on these individuals included
three years' probation, restitution, and community service hours.
Fraud cases handled administratively also result in few significant consequences. DES
pursues administrative disqualification from future program participation after a client
has received benefits and an overpayment has been identified.' While DES took steps
to administratively disqualify 1,669 clients in fiscal year 1991- 1992, only 449 were
eventually disqualified from future program participation. Because administrative cases
are often so poorly prepared, about half are decided on appeal in favor of the client.
The relatively few clients who are eventually disqualified are required to repay the
amount of benefits received in error; however, there are no additional financial
penalties or fees assessed.
While some individuals experience the consequences of committing fraud, many
workers believe it happens too infrequently to be much of a deterrent. In responding
to our survey, several EI's frustrated with the current system discussed client attitudes
and the lack of an effective system to deter fraud. They offered the following
comments:
" Clients are not deterred fiom fiaud because they do not perceive the penalties to be harsh."
1. Using this dollar amount allows the Attorney General's Office to prosecute welfare fraud as a Class
3 felony which can result in five years' imprisonment.
2. Food Stamp regulations allow for clients to be disqualified from participating in the Food Stamp
program if they have committed an intentional program violation. A client can be disqualified for
6 months upon a first offense, 12 months for a second offense, and permanently on a third offense.
Additionally, HHS recently established regulations allowing similar administrative disqualifications
for AFDC clients who have committed intentional program violations.
" Fraud is on the rise but we won't do anything but slap their hands when they cornrtrit
fraud. "
" I feel that the penalties fm fraud are too removed in time from the act of fraud and too
lenient to act as a deterrent. There is no deterrent efect with the current system. "
Prevention is limited - The current approach to fraud also limits prevention efforts.
Relying on OSI to investigate fraud sfufts responsibility for preventing fraud away
FIGURE 4
The " Funnel Effect" Of Fraud Investigations
And Prosecutions During
Fiscal Year 1992
31 8 cases referred for prosecution
4
.............. .
111 individuals sentenced
Source: OSI and Attorney General Fiscal Year 1992 statistics.
from line workers, since EI's are not required to detect or investigate fraud. OSI, on
the other hand, has only limited ability to prevent fraud. OSI conservatively estimates
its early fraud investigations saved over $ 1.1 million in fiscal year 1991- 1992. However,
many if not most fraudulent cases are never referred to OSI. Because EI's are
constantly faced with time pressures to process cases quickly, an EI may decide to
disregard the conflicting information and forgo referring the case for further
investigation, thus avoiding additional paperwork. One EI indicated that she is aware
of many EI's who simply do not refer cases because of the time it takes to prepare a
referral and because, even if the effort is made, nothing happens to the clients who get
caught. As a result, there could potentially be thousands of unreferred cases containing
some element of fraud or misrepresentation.
Arizona Should Refocus
Its Fraud Prevention
And Detection Efforts
As an alternative to its current approach, DES should refocus its efforts on more
effective up- front prevention and enhanced detection activities. First, more emphasis
should be placed on the initial interview and more thorough verification of information
prior to making an eligibility determination. Second, OSI should take a more proactive
role in identifying and pursuing fraudulent cases. Finally, DES should strengthen its
use of administrative sanctions to make it more difficult for clients to commit fraud.
In our opinion, simply increasing resources allocated to the criminal investigation and
prosecution of welfare fraud cases would not be cost- effective. Typically, welfare fraud
cases involve low dollar amounts and are not likely to be prosecuted. Therefore, we
attempted to identify ways that EI's, who have front- line contact with clients and make
the initial eligibility determination decisions, could better detect and prevent fraud. We
also attempted to identify ways to enhance the effectiveness of OSI's investigative
resources, and the role of administrative sanctions.
More effective interuiewing- and verification - By employing more effective
investigative interviewing and verification techmques, EI's can do more up front to
prevent fraud. Currently, because of the pressure to process cases quickly, there is
often insufficient time devoted to interviewing clients and obtaining all the necessary
documentation. Interviews may vary in length depending on the complexity of the
case, but in every case, the EI must review the 11- page application with the client,
asking over 100 questions. Common weaknesses revealed by EI's in the interviewing
process include confusion on what information to obtain from the client, use of closed-ended
or leading questions, and failure to use inquisitive or follow- up questions. For
example, some EI's do not inquire further into the existence of bank accounts when
there is evidence to suggest that accurate bank account information has not been
disclosed.
However, some EI's we contacted apply interviewing techniques that they claim reduce
their error rates and prevent fraud:
Some EI's repeat questions, or ask the same question in a different way to ensure
that they get consistent and complete answers.
One has found that by conveying a nonthreatening, " help us help you" approach,
many clients have changed the information they had originally written on the
application to the correct information.
One EI asks clients to draw maps showing where they live. If they cannot, or have
trouble, it is an indication that they may not be living where reported.
Other EI's check for wedding rings, quality of clothing, and cars as an indication
of income level.
EI's should also ensure that clients understand their rights and responsibilities and the
consequences of providing incorrect or misleading information. In its handbook for
eligibility workers, the California State Fraud Bureau states that such understanding
is basic to fraud prevention. Despite the importance of this task, only half of the EI's
responding to our survey indicated that they always explain the consequences of fraud
and misrepresentation to clients.
Finally, EI's can do much more to verify information provided by clients. While time
pressures often prohibit the EI's from conducting comprehensive verification checks,
some EI's responding to our survey stated that the EI can and should do more
verification. For instance, one EI said that she checks job service records to verify
employment for clients and noncustodial parents, calls landlords, uses an automated
system to check addresses and telephone numbers through the city library, and calls
the jail and probation offices, if necessary. Another EI stated that she typically uses
several innovative techniques to locate people and to identify or verify employment.
Those EI's indicated that, at times, they will work extra hours in order to conduct
comprehensive verification.
OSI could take a more proactive role - In addition, OSI could more effectively use
its resources to detect and pursue fraudulent cases on its own. As previously
mentioned, OSI uses much of its resources to investigate fraud referrals from the EI's.
However, Federal regulations for AFDC specifically identify certain other periodic
support activities states should consider taking to detect fraudulent applications for
AFDC prior to establishing eligibility for such aid. Examples of such activities include
automated data matches, use of error- prone profiles, home visits or collateral contacts,
credit bureau inquiries, and training on investigative interviewing techniques.
We contacted several states and found that many use the same techniques to identify
fraud:'
COMPUTEMRA TCHING- Five of the six states we contacted use computer matching
as a method of identifying potential fraudulent cases. For instance, Florida has used
computer matchng involving multiple states in an attempt to identify clients who
are receiving benefits from more than one state, or who owed money to one state
and were receiving benefits from another state. Other states conduct computer
matches of post office boxes, other states' welfare rolls, and social security number
checks. Matchng has proven so successful in Michigan that 95 percent of its
investigations come from this method of case identification.
PROFILE- S Some states use error- prone profiles to identify potential fraudulent
cases. From past experience, Orange County, California, has found that clients tend
to follow the same patterns when committing fraud. Therefore, it has developed
profiles that indicate situations most susceptible to fraud. Iowa's Food Stamp
Investigations Unit has also developed a list of indicators that the eligibility worker
must identify prior to referring a case for investigation. Finally, Texas investigators
use profiles as a means of picking cases most likely to result in fraud detection.
RANDOMC HECKINGOR TARGETING- A few states have randomly selected cases
or targeted certain offices in an effort to determine if adequate referrals are being
made. For example, Tennessee conducts random sampling in smaller counties where
there is limited investigative presence. It also targets counties that have not sent
many referrals. In addition, San Diego County, California, has proposed random
checking to enhance its efforts to detect fraud. Finally, as mentioned previously,
Nevada has conducted random checks as a means of determining the extent of
fraud.
ENHANCEDT RAININGFO R ELIGIBILITWYO RKER- S C alifornia has developed a Fraud
Prevention Training Program stressing such worker techniques as ensuring client
understanding, building a good client- worker relationship, and projecting the proper
agency image.
The eligibility workers are required to receive eight hours of investigative
interviewing training each year.
INTAKEF RAUDW ORKER- S O range County, California identified the use of intake
fraud investigators in every local office as its most effective method for deterring
fraud or misrepresentation. These investigators are able to go immediately to the
applicant's home and verify the accuracy of information provided on the
1. Based on discussions with Federal and State officials and a national fraud association, we identified
the following states to contact for further information on their fraud programs: Florida, Iowa,
Nevada, Michigan, Orange and San Diego Counties, California, and Tennessee.
application. By devoting staff for this activity, one Orange County official claims
that 70 percent of the applications are denied. Although OSI does not have enough
investigative positions to allow one investigator in each office, a similar approach
to intake fraud workers has been used in the past. DBME assigned EI's as
" verification specialists" responsible only for documentation verification, which
assisted in preventing ineligible cases from being approved. This process was
discontinued, however, due to complaints from client advocates opposed to EI's
working outside of the local office.
Use of administrative penalties - More effective use of administrative penalties could
also help reduce the amount of fraud. First, to increase the success rate of
administrative disqualifications, DES should ensure that sufficient evidence is presented
at the hearings. As mentioned previously, the success rate of administrative
disqualifications is hampered by the lack of adequate evidence provided. If a client
appeals the disqualification, he or she stands a good chance of prevailing, and will
continue to receive benefits. On the other hand, two states we contacted maintain at
least a 90 percent success rate in upholding the local offices' decision to pursue
program disqualification.
In addition, DES should disqualify individuals who attempt to obtain benefits
fraudulently. Currently, there is no penalty for those individuals who attempt to
commit fraud during the application process. OSI officials believe the lack of penalties
for fraudulent activities prior to benefit issuance diminishes the effectiveness of its
early fraud detection program. They state that if a client applies for benefits in one
office, but is denied for inaccurate information, there is nothing to stop the same client
from applying at another office the next day.
Other states, such as Texas and Nevada, disqualify clients caught attempting to obtain
benefits fraudulently even though no money was ever paid out. The same sanctions
apply to these individuals as to those who actually receive benefits fraudulently.
RECOMMENDATIONS
1. As a means of refocusing on up- front fraud prevention, DBME should enhance its
training for EI's to further develop and emphasize thorough and effective
investigative interviewing techniques and verification methods.
2. OSI should consider devoting more of its resources to supplemental fraud detection
methods, such as:
Computer matching
Profiles
Random checking or targeting
Increased involvement in training on fraud prevention and detection principles
Intake fraud workers
3. DES should take steps to strengthen the administrative disqualification process by:
Ensuring that local offices are properly trained on how to prepare a quality case
for administrative disqualification; and
Pursuing the use of administrative sanctions for those individuals attempting to
commit fraud prior to benefit issuance.
TECHNICAL APPENDN
( This Page Intentionally Left Blank)
TECHNICAL APPENDIX
Our estimate that as many as 70,346 cases may contain fraud or misrepresentation at
a potential cost of approximately $ 107 million is based on the following assumptions
and calculations:
STEP 1: First, we calculated the number of cases estimated to contain fraud or
misrepresentation. This was done by multiplying the median estimate of fraud
( 28 percent) provided by EI's by the number of AFDC ( 70,252) and Food
Stamp cases ( 179,814) as of March 31, 1993.
Food Stamp Cases 179,814 ( 72%)
AFDC Cases 70,252 ( 28%)
Total Cases 250.066 [ loo%)
STEP 2: Next, using the proportion of Food Stamp ( 179,814 or 72%) and AFDC cases
( 70,252 or 28%) to the total number of cases as of March 31, 1993, ( 250,066),
we determined how many of the estimated fraudulent cases were Food Stamp
cases and how many were AFDC cases:
L
Total Estimated Percentage of Total Cases
Fraud Cases That Are Food Stamp
70,346 X .72 - 50,349
Total Estimated Percentage of Total Cases
Fraud Cases That Are AFDC
70,346 X .28 -- 19,697
Total Cases Fraud Estimate Estimated No. of Cases
Containing Fraud
250,066 X .281308 -
70,346
Summary: 28% Fraud or Misrepresentation Produces:
Fraudulent FS Cases 50,349
Fraudulent AFDC Cases 19.697
Total Estimated Fraudulent Cases 70,346
STEP 3: Next, we identified the total amount of AFDC and FS benefits issued in
fiscal year 1991- 92.
FS
AFDC
Total
STEP 4: We then estimated the average annual dollars in benefits per case by case
type.
Food Stamps
STEP 5: Next, we estimated the dollars associated with the estimated fraudulent
cases in each of the two programs:
AFDC
No. Fraud. Avg. Expenditure Total Fraud
Cases Per Case Dollars
FS 50,649 X $ 2,128 -- $ 107,781,072
AFDC 19,697 X $ 3,398 -- $ 66,930,406
$ 2,128 Per Case
FS Dollars ($ 382,565,000)
FS Cases ( 179,814)
--
$ 3.398 Per Case
AFDC Dollars ($ 238.697,631)
AFDC Cases ( 70,252)
--
STEP 6: Because we cannot assume that every case is 100 percent fraudulent ( for
instance, a client may be eligible to receive benefits, but based on the
information provided at the time of eligibility determination, the amount
of benefits may not be correct), we used the OPE codes for " client W u l
misrepresentation" ( i. e. the client willfully misrepresented the information
provided at the time of eligibility determination) to calculate the average
percentage of dollar error associated with a case. Based on 596 error cases
examined by OPE in Federal fiscal year 199 1- 1992, 17 Food Stamp and 15
AFDC cases were identified as having errors associated with client W u l
misrepresentation.
The dollars associated with those error cases and the percentage of each
type of case that could be assumed to contain fraud or misrepresentation
is presented below:
Total Avg. Total Percentage of Fraud
DollarsICase Fraud Dollars Dollars To Dollars Issued
FS $ 3,824 $ 1,632 42.68%
AFDC $ 3,266 $ 3,009 92.13%
STEP 7: Finally, using the assumed percentage of fraudlmisrepresentation in each
type of case, we calculated a revised estimate of the dollars associated with
the estimated fraudulent AFDC and FS cases:
Percentage of Revised Total
Total Fraud Fraud Dollars to Fraud
Expenditures Total DollarsICase Expenditures
( FS) $ 107,781,072" X .4268 -- $ 46,000,962
( AFDC) 66,930,406" X ,9213 -- $ 61.662.983
Total Fraudulent
Dollars $ 107.663,945
( a) See Step 5.
I Fife Symington
Governor
17 17 W. Jefferson - P. O. Box 6 123 - Phoenix, AZ 85005
Linda J. Blessing, DPA
Director
October 15, 1993
Mr. Douglas R. Norton
Auditor General
2910 North 44th Street, Suite 410
Phoenix, Arizona 85018
Re: Report of Performance Audit on the Arizona Department of
Economic Security's Division of Benefits and Medical
Eligibility.
Dear Mr. Norton:
The Department of Economic Security ( DES) has reviewed the
Auditor General's report of the performance audit of the DES
Division of Benefits and Medical Eligibility. I know this report
represents many hours of concerted effort by you, your staff, and
the Division. I appreciate the audit team's efforts in analyzing
this large and complex DES Division.
We are pleased that the report recognizes the intensive effort
the Department has made in the last year to improve the quality
of performance in this Division. As noted in the report, the
Department has " aggressively pursued changes to better serve
clients and reduce program error rates." I1[ T] he Division is
attempting to improve the effectiveness of its quality assurance
and quality control systems by reviewing cases for errors prior
to issuance of benefits ...[ and] strengthening its ability to hold
employees accountable for work quality and job performance. I1
The Department is keenly aware of the need to reduce errors and
strengthen efforts to prevent and detect fraud. In furtherance
of these objectives, the Department has instituted several
quality improvement initiatives in this Division during the past
year. The Department agrees with the report's recommendations
concerning additional measures the Division can implement to
improve services and performance, and reduce error rates. Our
response to the audit report, which is enclosed with this letter,
addresses those measures and the Department's intent to implement
them.
Letter to Douglas R. Norton
Page 2
The Department is mindful of its role as a public servant and
takes seriously the duty to handle public funds in a fiscally
responsible manner. The quality initiatives already underway,
and the additional measures we will implement based on the
suggestions contained in the report, are all aimed at fiscal
responsibility and accountability. I am strongly committed to
continuing the process improvement in the Division of Benefits
and Medical Eligibility, as well as in other areas throughout the
Department.
Despite our agreement with much of the information contained in
the report, we strongly disagree with one statement: the
suggestion that the cost of fraud and misrepresentation is as
high as $ 107 million annually. While the Department concurs with
the conclusion that fraud is a problem, the Department is very
disturbed that the Auditor General's office has supported that
conclusion with unsubstantiated opinions. The report implies
that the $ 107 million figure is based upon empirical data, when,
in fact, it is based upon the subjective observations of less
than one fourth of the Department's eligibility interviewers
( llEI'slf).
As noted in the report, the Auditor General's office computed
this $ 107 million estimate based on a survey of EIfs. It is
important to remember that the EIfs are not the Department
employees who investigate fraud and misrepresentation in the
public assistance programs. Trained investigators in the
Department's Office of Special Investigations ( OSI) handle this
function. As the audit report indicates, OSI is unable to
document fraud in all the cases referred to it for investigation,
which could suggest a lower rate of fraud than that indicated in
the EI survey. By comparison, the EIsl opinions about fraud
rates are anecdotal and unverified.
Another flaw is the lack of information concerning the survey
respondents. The audit staff did not assess the characteristics
of the responding population. The audit staff made no attempt to
determine if the EIfs who responded had been on the job for ten
years, or just for three months; or whether they were from rural
areas or from metropolitan areas. These factors and others would
significantly affect the opinions as to the rate of fraud and
misrepresentation.
The audit staff then took this subjective fraud rate and
multiplied it by the total number of cases and benefits, to
arrive at a figure of $ 107 million, which is presented in the
report as solid empirical data. The figure lacks validity and
credibility, and may distort the extent of the problem.
Letter to Douglas R. Norton
Page 3
The report indicates that other states have determined fraud
estimates by analyzing a random sample of cases. The Department
believes that this is the proper methodology for arriving at a
solid empirical figure. Accordingly, I have instructed the
Division to do such a study. The Divisionrs study will produce a
more reliable figure, and, more importantly, illuminate the root
causes of the problems. With a better understanding of the
causes, we will make the most beneficial, systematic
improvements.
~ otwithstanding our disagreement with the Auditor General's
estimate of the upper range of the annual cost of fraud and
misrepresentation, we believe the report will serve as a useful
tool for improvement. We again extend the Departmentrs gratitude
to you and your staff for the time and effort invested in the
audit.
Sincerely,
Director
Enclosure
LJB : vtb
I DIVISION OF BENEFITS AND MEDICAL ELIGIBILITY
RESPONSE TO THE AUDITOR GENERAL'S REPORT
FINDING I - Recommendations: DBME IS TAXING STEPS TO REDUCE ERRORS
( 1. DBME should continue its efforts to address and impact proqram
error rates. In doins so, Division manaqement should secure
orqanizational commitment and support for initiatives and should I continuousl~ monitor, evaluate and analyze the impact these
initiatives are havinq on division performance.
I We concur with this recommendation and have implemented the
following:
o In order to secure organizational support and commitment
throughout the division, three employee teams have been
established through the Excellent Service Through People ( ESP)
Initiative. These teams first met in October 1992. The teams
completed their initial assessment phase and made
recommendations to improve the division's processes.
In order to implement the Quality Assurance Team's
recommendations, mid- level managers are involved in developing
and providing training to local office managers. These
managers are, in turn, responsible for training the
supervisors in their local office. This approach to training
promotes commitment to the process.
I o To monitor the initiatives, we have established an Error
Prevention Team ( EPT) and an Error Prevention Steering
Committee ( formerly known as the Continuous Assessment Team).
The EPT includes staff from the Office of Program Evaluation
( OPE) and Family Assistance Administration ( FAA). The
membership is as follows: FAA district program manager, local
office manager, public assistance local office representative,
eligibility interviewer ( EI), policy specialist, trainer and
Office of Program Evaluation ( OPE) quality control and
management evaluation reviewer.
This Team will analyze the data through review of case records
which are performed by FAA Supervisors and Quality Control
staff. Using the total quality management tools, such as;
brainstorming, cause and effect and force field analysis, root
cause of the errors will be determined and recommendations for
district and statewide solutions will be elevated to the Error
Prevention Steering Committee.
The EPT Steering Committee is comprised of key FAA and DBME
Management staff which will review the EPT recommendations and
monitor state and district quality initiatives, so that we can
maintain a focused approach. These individuals include
DBMEfs Response
Page 2
managers from: FAA, the Program Administrator, Field
Operations, two district program managers, Systems, Operations
Support, both policy units and an eligibility interviewer
( EI); OPE Chief; DBME Assistant Director and Special Quality
Advisor. This committee has been meeting since March 1993 to
oversee all of the divisional error reduction initiatives.
2. DBME should consider ~ ursuins upsraded automation capabilities
such as an " expert" system to simplify the iob of the elisibilitv
interviewer and allow workers to handle their increasins caseloads
more effectively.
DBME is currently evaluating the feasibility and cost- benefit of
implementing an " expert" automated system. To assist in this
evaluation, DBME is conducting a national survey and the Division
of Data Administration is reviewing all of the automation systems
currently available.
Through the ESP Initiative, the DBME will continue improving the
eligibility determination processes prior to implementing any
" expert" automation system.
3. DBME should revise traininq for both EIs and supervisors to
incorporate on- the- job trainins, make classroom traininq timely,
and expand traininq content areas.
We concur with the recommendation. We recognize the need for
staff training at all levels and are in the process of
implementing several items.
o When an EI attends Food Stamp Basic Skills training session, a
passing score of 80 must be achieved prior to reporting to a
local office to interview clients. Those individuals who do
not pass this training session are dismissed from state
employment or returned to their prior position. This change
was implemented effective April 1993.
o In an effort to increase the knowledge of EIs, the Food Stamp
Basic Skills training class was expanded from three weeks to
four. This class was validated by a Quality Control reviewer,
an FAA supervisor and Field Representative. This change was
made effective November 1992.
o Development of supervisor/ manager policy refresher training,
so that we can give management the appropriate tools to
perform their job. After this initial training, an assessment
will be conducted to determine other types of training which
may be needed by a supervisor.
DBME1s Response
Page 3
o In August 1993, FAA and OPE staff, including the EPT and EPT
Steering Committee, underwent an in depth training on root
cause analysis.
o Other districts and local offices continue to assess the need
to establish training units for new EIs within their offices.
o We will continue to explore training methods from other states
and expand on the refresher training for all levels of staff.
FINDING I1 - Recommendations: DES NEEDS TO STRENGTHEN EFFORTS TO
PREVENT AND DETECT FRAUD
1. As a means of refocusins on up front fraud prevention, DBME should
enhance its trainins for EIs to further develop and emphasize
thorouqh and effective investisative interviewins techniques and
verification methods.
We concur with this recommendation. As a part of their program
training, EIs attend food stamp basic skills class for four weeks.
During this four- week timeframe, they receive two days training on
interviewing skills. The FAA Central Office recently expanded
this module of the training to provide EIs with interviewing
skills which are used by Quality Control staff.
FAA overpayment staff recently underwent a training course on the
preparation of overpayments which included how to effectively
prepare cases for administrative disqualification hearings.
~ ffective July 1993, ~ uality Control Reviewers have been directed
to use criteria which provide them with fraud indicators. When
the Quality Control findings are reported to the FAA office, the
reviewer reports to the local office a potential fraud. FAA will
review the case and if appropriate refer the case to OSI for an
investigation. In addition, effective 10/ 01/ 93, the intentional
program violation process was expanded to include AFDC client
fraud.
1 2. OSI should devote more resources to fraud detection.
We concur that more resources should be devoted to fraud
I detection. The Office of Special ~ nvestigations has begun
discussions with California and other neighboring states regarding
computer matching. In addition, training is being conducted with
I ~ ligibility Interviewers as requested by local office management.
A typical subject at this training is detection and telltale signs
of fraud during client interviews. OSI will continue to explore
the ability to formalize fraud training. Further, OSI I
DBMEfs Response
Page 4
investigators are outstationed in local offices with a strong
focus on precertification investigations. This program will
continually be evaluated and strengthened to assist in early fraud
detection.
DES should strensthen the administrative disqualification process.
Regarding the process of disqualifying individuals who have never
received a benefit from our programs, DES will be contacting other
states. However, prior to implementation of such a process, a
data- based decision is required, which must include a cost benefit
analysis of costs associated with investigating, alleging and
imposing the administrative sanction. This cost benefit analysis
will include staff from the FAA, Office of Special Investigations,
Office of Accounts Receivable and Collections, and Office of
Appeals.