oversight

Student Aid Fraud Ring Assessment. X18M0001, Date Issued: 01/17/2013 PDF (2.12M)

Published by the Department of Education, Office of Inspector General on 2013-01-17.

Below is a raw (and likely hideous) rendition of the original report. (PDF)

      -
                   UNITED STATES DEPARTMENT OF EDUCATION
                                      OFFICE OF INSPECTOR GENERAL

                                                    Information Technology Audits and Computer Crime Investigations
                                                                  Computer Assisted Assessment Techniques Division



FINAL MANAGEMENT INFORMATION REPORT

DATE: 	          1117/2013

TO: 	           James W. Runcie
                Chief Operating Officer
                Federal Student Aid

                David A. Bergeron 

                Acting Assistant Secretary 

                Office of Postsecondary Educatio 


FROM: 	         Charles E. Coe, Jr.
                Assistant Inspector General
                Information Technology Audits and Computer Crime Investigations

SUBJECT: Management Information Report
         Student Aid Fraud Ring Assessment
         Control No. ED-OIG/Xl8M0001


The purpose of this management information report is to provide the Office of Federal Student
Aid (FSA) and the Office of Postsecondary Education (OPE) with the results of our risk analysis
regarding one type of student aid fraud ring activity associated with the electronic processiny of
Federal student aid applications. We refer to this type of fraud ring activity as school-based
because we developed our methodology by focusing on student aid recipient activity within each
school. Although most of these identified recipients relate to distance education, our
methodology assessed all school-based recipients.

We are attempting to identify other types of student aid fraud ring activity and may be able to
report on such activity in the future. As this report will demonstrate, this school-based student
aid fraud ring activity is a rapidly growing problem. The population of school-based recipients
considered as potentially participating in this activity has increased 82 percent from award year
(AY) 2009 (18,7 19 students) to A Y 2012 (34,007 students). We identified a total of over 85,000
recipients who may have participated in this type of student aid fraud ring activity and who
received over $874 million in Federal student financial aid under Title IV of the Higher
Education Act of 1965, as amended (Title IV funds or programs) during the period AY 2009
through A Y 2012. Further, applying a statistical model, we estimated that $187 million of this
$874 million in Title IV funds are probable fraud loss.

1
    School-based reflects the focus ofthe fraud assessment and does not intend to imply that we have any indication
that school officials or anyone working at the institution was involved in the fraud ring activity.

                             'The Department of Education's mission is to promote student achievement and preparation
                             for global competitiveness by fostering educational excellence and ensuring equal access.
Final Report
ED-OIG/Xl8M0001                                                                                        Page 2 of6




The Office ofinspector General (OIG) Office of Investigation Services issued Investigative
Program Advisory Report (IPAR) "Distance Education Fraud Rings" (Control No. ED-OIG/L42­
LOOO1) in 2011. The IPAR alerted FSA and OPE to significant fraud vulnerability in distance
education programs and provided recommendations that, if implemented, would mitigate future
risks of fraud ring activity in Title IV programs.

Although the IPAR described the problem of an increasing number of cases involving fraud
rings-large, loosely affiliated groups ofpeople who conspire to defraud Title IV programs
through distance education programs-it did not identify the magnitude of the overall activity.

We now have quantified the risk of one type of fraud ring activity with a statistically supportable
dollar estimate. In this report, we present our results in terms of Title IV funds for the entire
85,000 recipients who may have participated in the school-based student aid fraud ring activity,
our supportable estimate of probable fraud loss, and our methodologies.

Identification of One Type of Fraud Ring Activity

For A Y 2009 through A Y 2012, more than 53.8 million recipients 2 received more than $509.9
billion in Federal student financial aid under Title N. 3 As we noted in our "Distance Education
Fraud Rings" IPAR, the growth in distance education has made Title IV programs more
vulnerable because all aspects of it- admission, student financial aid, and instruction- take
place on-line, and the students are not required to present themselves in person at any point. In
addition, access to personal information using conventional and electronic means has made it
easier for criminals to assume the identity of fictitious recipients and fraudulently apply for and
obtain these funds.

To address this growing risk, the OIG developed a model of student aid fraud rings to assist in
cases. 4 The overall objective of this project was to develop an analytical model that would focus
investigators' attention as early as possible on the activities most at risk for being fraudulent.
This model would also provide the capability of an automated, ongoing analytical process within
the OIG to assist with audit planning and investigation resource management. The first phase of
this project was to develop a model that would successfully detect, with a fairly high degree of
probability, fraudulent or suspicious activities relating to school-based student aid fraud ring
activity. The primary goal included identifying suspicious data associated with fraud ring
participants enrolling at a school.

To identify the suspicious data, we developed a rule-based filter, which is a set of qualifying
determinants that would eliminate those school-based recipients that do not meet minimum
2
  Though the 53.8 million student count represents a one-time count of students within each award year, there may 

be double counting of those students receiving awards in more than one year between AY 2009 thru AY 2012. 

3
  The Title IV programs included in our review were Wmiam D. Ford Federal Direct Loan Program loans, Federal 

Family Education Loan Program loans, Perkins loans, and Pel! grants. 

4
  We used AY 2010 in which 19 million recipients' received more than $129.8 billion in Federal student financial 

aid under Title IV as our base year within the development process. 

Final Report
ED-OIG/X18M0001                                                                                           Page 3 of6

threshold school-based fraud patterns established within the model. That is, we filtered out those
recipients who did not exhibit characteristics of an established school-based fraud pattern. We
then applied this filter to student aid recipients from 14 schoollocations 5 with suspected
fraudulent activity in A Y 2010, which resulted in a subset of recipients flagged as potential fraud
risks. Investigators reviewed those identified and confirmed that the filter (1) identified all open
fraud ring cases and all known participants within those cases for each school, (2) identified new
potential fraud rings, and (3) added additional suspects to fraud rings the OIG already was
investigating. To affirm the validity of this model, we performed a further study on four
additional schools, and the results were identical to the 14 schools.

We presented an overview of our methodology and some initial results to FSA representatives on
February 10,2012. This management information report presents the final results of this aspect
of our project. We will also provide a description of the details of our model and statistical
validation methodologies to FSA and OPE officials.

Process to Quantify Funds Lost to Fraud Rings

We developed a separate statistical model to confirm and quantify the estimated student aid
dollars associated with the fraud we had identified in our school-based model. We constructed a
recipient-level data set combining Free Application for Federal Student Aid reported data,
supplemental demographic data, Pell grant and loan history, and other data sources. Pairing this
information with known fraud cases in the 14 schools described above, we constructed a
predictive statistical model that estimated a recipient's probability of participating in a school­
based student aid fraud ring. We evaluated the accuracy of the model's predictions using a
measure called a concordance rate, which is a measure that compares the model predictions to
the known behavior. We found our statistical model predicted the presence or absence of fraud
with a high degree of accuracy, as measured by a concordance rate of 92.9 percent. The model
created the resulting mathematical equation that allowed us to calculate a recipient's estimated
probability of participating in a school-based student aid fraud ring given his or her personal­
level data.

While we found the statistical model's accuracy to the training data (data derived from the 14
selected schools) to be acceptable, we had concerns with the model's applicability to the general
population. Because these 14 schools were subjectively identified by our investigators, we
wanted to test the model before applying it to the national recipient population. Consequently,
we performed two validation checks.

First, a school that was identified with fraud ring cases but was excluded from the original
model-building was now used to test the model predictions. This school had a much larger
recipient population and a lower known incidence of fraud than the other schools. We used the
model to score the school's recipients and found that the estimated probability of school-based
fraud corresponded with the actual known incidence of fraud. The second approach used a
resampling technique referred to as a "jack-knife" procedure, which reconstructed the model

5
    Initially, we were supplied with 15 school locations, but one school had a very low count ofapplicants that were
flagged by our rule-based filter, which would indicate a low probability offraud ring activity and, therefore, was
dropped from the study.
Final Report
ED-OIG/X 18MOOO I                                                                                                        Page 4 of6

using random subsets ofthe schools and recipients. We estimated probabilities of school-based
fraud ring activity and dollars associated with that fraud ring activity using over 1,000 different
subsets. This approach resulted in a wider range of estimated fraud dollars where 90 percent of
the repetitions did not alter the dollar amount estimate by more than +/- 7 percent. We have
concluded that both of these additional tests have validated the results of our school-based
model.

Results of OIG Analysis of Risk Rela ting to School-based Fraud Ring Activity
Based upon the success of the validated rules-based filter in identifying fraudulent or suspicious
activities, we then incorporated the filter into the school-based student aid fraud ring model.
When we applied the filter to all students receiving Title IV disbursements between A Y 2009 to
A Y 2012, 6 it identified 85,000 7 recipients that may have participated in school-based fraud ring
activity. As depicted in FIGURE 1 below, these 85,000 recipients received a total of $874
million of the $509.9 billion of Title IV aid that was distributed AY 2009 through A Y 2012.
While our school-based model identifed $874 million in awards at risk, we could not state with
assurance which of these recipients' Title IV dollars were actually associated with a school­
based fraud ring and, therefore, would be a probable fraud loss.

Consequently, we used the statisical model discussed in the previous section to estimate each of
these at-risk recipients' probability of fraud based on his or her personal data. We then
multiplied the total student aid received by each recipient by their estimated fraud probability to
estimate the total fraud Title IV dollars obtained by that recipient. Based on this analysis, we
estimated as depicted in FIGURE 2 below that $ 187 million of Title IV funds for the period A Y
2009 through A Y 2012 are probable fraud losses.8
                       FSA Funding                      Total FSA
                                                                                          FSA Funding
                           At Risk                       Fund"ng                      Probab e Fraud Loss
                                                                               ~


                                                                          .                            $52.7M
                                                                                                        AY2012
                                                                                                                 I


                                                        I $142.481    $187M
                                                                                                         $57.8M      I
                                                                                                        AY2011

                                                        I S129.ss I                              $45.8M   I
                                                                                                   AY2010]

                                                                          ..             $30.9M
                                                                                          AY2009
                                                                                                   I


                                                                                          ~              ~                 

                • AY 2012 Effect1ve Through July 31st                          •Av 2012 Effective Through July 31st

                         I FIGURE 1 I                                                     I FIGUR£2 1
6
  Award Year Period- Though an award year covers a July-to-June time period, the actual processing time duration
for all related funding documentation and associated modifications ends on December 31st of the award year.
7
  The 85,000 student count represents a one-time count of students and negates double counting of those students (as
reflected in the graph) with grants or loans awarded in more than one year between AY 2009 thru AY 20 12.
8
    Using the 95 percent confidence level for each student's estimated fraud probability, the model estimate ranged
from -11 percent below to + 13 percent above the $187 million.
Final Report
ED-OIG/Xl8MOOOI                                                                               Page 5 of 6




As shown in FIGURE 3, our school-based
model determined that 18,719 recipients were
considered at risk in AY 2009. This
population increased to 34,007 by A Y 2012.
                                                                                          3
This increase of 82 percent indicates that                                      3         4
                                                                                2
school-based fraud ring activity is a growing                          2
                                                                                6
                                                                                          0
                                                                       6                  0
problem.                                                               7
                                                                                8         7
                                                                                6
                                                                       1
                                                               7       1
                                                               1
                                                               9




                                                                           I FIGURE 3 I

Suggestions

OIG's IPAR recommended the establishment of application edits in both the Central Processing
and the National Student Loan Data Systems to flag potential fraud ring participants, as well as
to identify practices for institutions to detect and prevent distance education fraud. In its
response to the previous report, the Department of Education (Department) acknowledged the
need to reassess the aid application and delivery system controls. The Department also stated
that it recognized an opportunity to develop continuing and more comprehensive measures to
minimize the potential for fraud.

As a continuation of these efforts, we suggest that the Department consider incorporating the
fraud detection processes referenced in this report into FSA processes.


Objective, Scope, and Methodology
The objective of this project was to develop a model that successfully detects with a high degree
of probability school-based student aid fraud ring activity, within the Federal student aid arena.
We selected A Y 2010 as the most recently completed school year to conduct this assessment and
identified 15 9 schools with suspected fraudulent activity as the area of focus.

The project team (I) identified potential fraud indicators based on experience and subject matter
expertise; (2) assessed the accumulated indicators in terms of how best to logically separate,
converge, and display student versus school anomalies; (3) captured an end-of-year snapshot of
all of the related data points from the related Department systems for each of the risk indicators;
(4) assessed whether the results met initial expectations (that is, did the data say what we
expected it to say using known high-risk entities for reasonableness testing within the identified

9
    See footnote 5.
Final Report
ED-OIG/Xl8M0001                                                                       Page 6 of6

high-risk pool); and (5) analyzed the model and finalized it by adding, deleting, and modifying
original fraud indicators, to include appropriate weighting if needed.

We conducted our work in accordance with the OIG quality standards for a Management
Information Report.

Department Response

A draft of this report was provided to FSA and OPE on November 20,2012. FSA and OPE
reviewed the draft report and provided OIG with a formal, written response on January 11, 2013.
We are including that response as an attachment to this report.
                          UNITED STATES DEPARTMENT OF EDUCATION
                                             WASHINGTON, D.C. 20006- - - ­




MEMORANDUM

·nATE:          January 11, 2013

TO: 	           Charles E. Coe, Jr. 

                Assistant Inspector General 


                                                                        {2
                Information Technology Audits and Computer Crime Investigations 

                Office oflnspector General

FROM: 	         JamesW.Runcie         _
                Chief Operating Officer             /~~
                                                     'J.- - ~~~
                Federal Student Aid
             ~:.. 1	 d_,                     ilL. •- .-­
              '1)avid A. Be~eron ,.._- ~ - - - - -·
               Acting Assistant Secretary
               Office of Postsecondary Education

SUBJECT: 	      Draft Management Information Report
                Student Aid Fraud Ring Assessment
                Control No. ED-OIG/X18M0001

Thank you for providing us with an opportunity to respond to the Office oflnspector General's
(OIG} draft Management Information Report entitled, "Student Aid Fraud Ring Assessment"
(Control Number ED-OIG/Xl8M0001). The draft report provided Federal Student Aid (FSA)
and the Office of Postsecondary Education (OPE) with the results of your risk analysis regarding
one type of student aid fraud ring activity associated with the electronic processing of Federal
student aid applications, expanded on the OIG's Investigative Program Advisory Report (!PAR),
"Distance Education Fraud Rings," issued in the fall of2011, and discussed the magnitude of
probable fraud losses. We share your concerns about the potential growth of fraud in the student
aid programs, and we continue to work aggressively to analyze, identify and establish plans to
implement a series of new policies, controls and processes to combat this problem.

As an initial matter, describing the type of fraud that you have identified as "school-based" is
somewhat misleading and could cause confusion with readers who are not familiar with the
analyses you have conducted. Specifically, the analysis is "school-based" because you have
looked to identify instances where there were individuals that share certain common
characteristics at the same institution. The description of the fraud as "school-based" is likely to
result in some readers believing that the institution has played some _role in the fraud itself rather

                                                                1




              Ourrnissicm. is to ensure equal access to education and to promote eduoational excellence throughout the Naticn.
than seeing it simply as a description of how the analysis was performed. For this reason, we
believe the draft report should not use "school-basedH as a descriptor of the fraud.

Upcoming Regulatory Activities

As part of our response and corrective actions taken in response to the OIG IPAR, on May 1,
2012, we published a notice in the Federal Register announcing our intention to establish a
negotiated rulemaking committee to prepare proposed regulations for the Federal student aid
programs authorized under Title IV of the Higher Education Act of 1965, as amended (HEA). In
that notice, we referenced the OIG IPAR and indicated that we were interested in suggestions for
regulatory changes to further help institutions combat fraud and protect students and taxpayers
from fraudulent activity. We held two public hearings in May 2012 in Phoenix, Arizona and
Washington, D.C. and also solicited written comments from the public. A number of external
parties submitted comments during this process that should be helpful as we proceed.
Combating fraud continues to be a major component of our regulatory agenda, and we are
hoping to move forward with the regulatory process in early 2013.

We have also reviewed existing regulations to identify where mechanisms for preventing
distance education fraud and obtaining repayment of fraudulently received aid could be added or
made tighter, and have also investigated whether it would be helpful to limit the method by
which schools release federal student aid credit balances to only the use of electronic fund
transfers to student bank accounts. The regulatory solutions ultimately adopted must be
developed through the negotiated rulemaking process.

Procedures for Enforcement

As recommended by the OIG in Finding 7 of its IPAR, we have put in place procedures to
assess repayment liabilities against individuals whom the Department determines to have
fraudulently applied for and obtained Title IV funds. Under these procedures, the perpetrators
are subjected to collection and ''flagged" as ineligible for further aid until repayment in full has
been received. Likewise, as recommended in Finding 9 ofthe IPAR, we have established
procedures by which individuals who enter into pretrial diversions are required to repay the
Department and are flagged as "ineligible, pending repayment.

Additional Verification Selection Criteria and Documentation Requirements

The Program Integrity regulations, published on October 29, 2010, contained significant
modifications to the verification regulations to allow the Secretary to specify annually the items
that an applicant and an institution may be required to verify each year and the acceptable
documentation for those items. The Department's long-range goal for verification is to develop
a highly customized selection and verification process that targets applicants for review based on
the data provided by each applicant on the FAFSA. This process will identify, for a selected
applicant, only the F AFSA information that requires verification based upon that applicant's
data. As a step toward this long-term goal, we published a Federal Register notice on July 12,


                                                  2
2012 that specified the verification items for the 2013-2014 award year. This notice included two
additional items that were recommended in the OIG IPAR:

    I. High School completion status; and
    2. Proof of identity/Statement of Educational Purpose.

Beginning with the 2013-2014 FAFSA application cycle, the verification selection process has
been enhanced to identify and select applicants who, based on a statistical risk model, might be
attempting to obtain Title IV student aid funds either fraudulently or without serious educational
intent It is important to note that even under this statistical selection process, it is possible that
selected applicants are serious financial aid applicants. This is especially true for this first year
where analytic data used for selection is limited.

As announced in Dear Colleague Letter GEN-12-11, there will be five Verification Tracking
Groups for 2013-2014. The items that an institution must verify are based upon the Verification
Tracking Group to which the applicant is assigned. Applicants who meet the possible fraud risk
model will be assigned to either Verification Tracking Group IV or V. Both of these Verification
Tracking Groups will require the applicant to provide documentation of their high school
completion status. Applicants selected for these groups also will be required to present
themselves in person to the institutional financial aid office with a government issued photo ID.
These applicants must execute, in front of the institutional representative, a Statement of
Educational Purpose. The Dear Colleague Letter provides an alternative for applicants who are
unable to appear in person at the institution that allows them to meet the proofof identity
requirement by executing the Statement of Educational Purpose in person before a notary public.

Requiring these selected applicants to appear in person (either at the institution or in front of a
notary public) should address instances where applicants were using other persons' identifiers to
apply for student aid as well as for admission and registration at the institution. Requiring these
applicants to execute a new Statement of Educational Purpose adds evidence for enforcement
(including repayment of aid and/or prosecution) if the applicant fails to fully participate in the
educational program for which the aid was provided.

Through this process, we expect to address at least some of the fraud that would otherwise have
occurred while we monitor whether these measures have unintended consequences such as
whether they deter applications from legitimate students.

Enhanced Scrutiny of Applicants with Unusual Enrollment History

Beginning with the 2013-2014 FAFSA application cycle, applicants who appear to have an
unusual enrollment history will have their records ''flagged" to require the institution to
determine ifthe student's prior academic record supports a serious academic intent. Specifically,
applicants who received Pell Grant funding from multiple institutions over a period of time will
be selected for this increased scrutiny. An example might be an applicant who received Pell
Grant funds for attendance at multiple institutions in the same award year would be flagged.
Because there could be legitimate reasons such an applicant had multiple enrollments over a

                                                  3

relatively short period of time, the guidance we will provide to institutions will be to analyze
academic records and/or documentation from the applicant before making a determination of the
applicant's continued eligibility.




Analysis of Multiple Email Addresses and Multiple IP Addresses

Beginning with the 2013-2014 FAFSA application cycle, FSA will collect email addresses and
IP address information when applicants apply electronically using FAFSA on the Web (FOTW).
FSA will analyze these data, both individually and combined, (1) to determine if there are
instances that raise the possibility of fraud, and (2) to establish an enhanced risk model that
would be used to select applicants for increased verification or other scrutiny.

Probable Fraud Loss

In your report you estimate $187 million in probable fraud loss. We would greatly appreciate the
opportunity to review the list of individuals who contribute to this total so that, in coordination
with our on-going efforts to bolster our analytical procedures and preventative controls, we may
take more inunediate action to research and resolve these specific risks for these specific
individuals. If, ultimately, we determine that these individuals were engaged in fraud, we will
take steps to collect the funds improperly disbursed as required by the Office of Management
and Budget when improper payments have been made.

We appreciate the work that the OIG has conducted in this area and we look forward to
continuing to work together to address this important issue. We would like to have a continuing
dialogue about further expeditious actions that can be taken to enhance this effort. It would be
particularly helpful to us if the OIG would repeat the analysis used for the draft report once the
2013-14 award year is underway. This would give us an early indication about the effectiveness
ofthe new selection and verification process. If you have any significant concerns or questions
about our comments and planned actions, please let us know, and we would be glad to further
discuss these matters with you.

Thank you again for the opportunity to comment on your draft Management Information Report.




                                                4