Fraud and Forensics:

radiographerfictionΔιαχείριση Δεδομένων

31 Οκτ 2013 (πριν από 4 χρόνια και 6 μήνες)

131 εμφανίσεις

© 2010 IBM Corporation

© 2010 IBM Corporation

August 10, 2010

Fraud and Forensics:

New Techniques, Better Results

National Association of State Auditors, Comptrollers, and Treasurers

© 2010 IBM Corporation

Fraud, waste, and abuse costs governments hundreds of
billions in revenues each year

Medicaid fraud increasing throughout state, but
economy forcing cutbacks in investigations… losses
could be
$2 billion

a year
Naples (FL) Daily News, 1/4/2009

Medicare &

Tax &



Food & Nutrition

the FTB has pegged California’s tax gap
associated with the PIT and CT at
$6.5 billion


California Legislative Analyst’s Office, 2008
09 Budget

workers’ compensation fraud is the fastest
growing insurance scam in the nation. Today, 10
cents out of each premium dollar is wasted on
Connecticut Conference of Municipalities, 11/12/2008

“…sophisticated shell games are costing Michigan's
unemployment trust fund up to an estimated

a year.

Michigan Unemployment Insurance Agency

“Every year, food stamp recipients exchange
hundreds of millions of dollars

in benefits for cash
instead of food with retailers across the country...”
Lincoln (NC) Tribune, 10/21/2006


“…agencies reported a total improper payment
estimate of about
$55 billion

for fiscal year 2007
US Government Accountability Office, 1/23/2008

© 2010 IBM Corporation

Date provided to OMB by the Federal Departments and Agencies

Fraud, waste, and abuse costs governments hundreds of
billions in revenues each year

© 2010 IBM Corporation

The problem of fraud and abuse has process, organization,
technology, and analytics dimensions

Fraud &





Process Dimension

Are proper business controls and
procedures in place to detect and deter
fraud and abuse?

Do laws and policies constrain fraud
related processes?

Are fraud detection and investigation
processes optimally aligned to
organizational goals?

Organization Dimension

Are there organizational barriers to
implementing an effective fraud
detection and investigation

Is there a single focal point (person
or team) that is responsible for
fraud and abuse activities?

Are key performance indicators in
place that measure and promote
excellence in fraud and abuse

Analytics Dimension

What detection analytics, if any,
are in place?

Do detection analytics allow for
both immediate and
retrospective analysis?

Are analytics used to control
and optimize fraud workload?


Technology Dimension

What technologies are used to
support fraud detection and

Does fragmented data create
challenges in having a complete
picture of behavior?

Does technology support
measurement and reporting of
fraud exposure?

© 2010 IBM Corporation


Who is behaving well?

Which entities are likely to behave “badly” in the

What are the indicators that an entity’s behavior
is getting “better” over time? “Worse” over

Data Mining &

What are patterns of non
compliant (and
criminal) behavior that I don’t know about?

If I catch a “bad” entity, how can I find out who
else is behaving like that?

Are there groups of entities who behave the
same way?

Which entities are behaving differently than
others (in a suspicious way)?

How “good” or “bad” is a entity behaving,
relative to other entities?

What is “normal” behavior?


Analytics are transforming how governments are tackling
fraud, waste, and abuse

© 2010 IBM Corporation

New York State Department of Taxation and Finance

Income tax refund fraud and abuse

Debt collection

North Carolina Department of Health and Human Services

Medicaid fraud detection

Case Studies

© 2010 IBM Corporation

5,000 total employees

Approximately $60 billion collected annually (2009)

Highly sophisticated taxpayer population (most Fortune 500
companies have a presence in New York)

Wide range of

taxpayers (demographic and cultural)

New York State Department of Taxation and Finance

© 2010 IBM Corporation

The project objective was to build a system to enhance existing audit
case selection methods for fraud detection of pre
processed Returns

The Questionable Refund Detection unit wanted…

A better way to identify questionable returns

To question suspect returns before issuing refunds

To improve collect ability of audit cases

To issue refunds in a timely manner

To make program management more flexible

To leverage investments in data warehousing and business intelligence

To scientifically predict good audit candidates utilizing return filing patterns, case
history, and other external indicators

To improve their ability to detect new areas of fraud

New York State Department of Taxation and Finance

© 2010 IBM Corporation


State of New York

Predicting tax compliance

The Case Identification and Selection System (CISS) applies
business rules and predictive models to categorize and score
returns nightly and identifies the ‘next best case’ for audit
selection. In addition, a separate web based portal provides
screening and resolution of cases.




New York wanted to enhance current audit case selection
methods for detection of audit issues at the time a return is
processed. Specific audit programs include Earned Income
Credit, Dependent Child Care Credit, Itemized Deductions,
Wage/Withholding, and Identity Theft.

$889 million

increase in revenue in the first five years

Increased screener and auditor productivity

Enhanced taxpayer correspondence

Improved audit program management


© 2010 IBM Corporation

100 employees in Program Integrity Unit

Approximately $14 billion in annual paid claims (2009)

$25 million in recoupment letters issued annually

North Carolina Department of Health and Human Services

© 2010 IBM Corporation

Solution components:

IBM Fraud Analysis Center

InfoSphere Identity Insight

IBM Global Business Services:
Business Analytics and

IBM Software Group: Lab

“I think we are going
to save tens of
millions of dollars.”

Beverly Perdue, Governor
State of North Carolina

The Need:

This large state social services agency faces a significant exposure
to healthcare fraud and abuse. The current business process and
technology used to fight fraud, waste, and abuse in Medicaid is

producing only around $25 million annually in recoveries.
This leakage, combined with a significant state budget deficit,
motivated the state to aggressively pursue cost takeout projects.

The Solution:

The state implemented a comprehensive health analytics solution.
This solution examines claims for suspicious patterns of behavior,
quickly identifying providers and recipients for investigation. In
additional, the solution identifies organized criminal rings and
collusive behaviors by uncovering suspicious relationships among
providers and between providers and recipients.




in recoupments in a 12 month period (expected)


State of North Carolina

Detecting and pursuing Medicaid fraud

© 2010 IBM Corporation

Claim Data Identified $140M = 18% in suspect claims

Personal Care Services ($86M of $555M = 15%)

Schemes identified within PCS include:

High payments per patient, high home health aid visit utilization

Billing for services on Sundays/Holidays that may not have been

Out of sequence billing

Durable Medical Equipment ($55M of $235M = 23%)

Schemes identified within DME suppliers include:

Expensive orthotics for amputees

Respiratory equipment unbundling

North Carolina Department of Health and Human Services

© 2010 IBM Corporation

PCS Analysis Results

North Carolina Department of Health and Human Services

© 2010 IBM Corporation

DME Investigation

Provider 1122233

Andy Griffith Medical Equipment

Andy Griffith Medical Equipment claims reveal the following patterns:

Claim submission delay which may indicate bill fabrication

Many patients with only 1 visit or service and over $5,000 charged

High reimbursement per patient in traction equipment, beds and power

Billing high charges for only a short window of time

appears to be a
store front scheme

Relationship analytics discovered that these providers are related to
Andy Griffith Medical Equipment:

112244, 112255, 112266

each or these DME suppliers exhibit the
same behaviors

Biilling for items for the same patients in the following pattern:

1 DME supplier bills the wheelchair

2 months later, another of the suppliers bills repair and additional

© 2010 IBM Corporation


Predicting the “next best”
case and technique for debt collections

The Collections Optimizer applies business rules and
predictive models to prioritize the ‘next best debt’ for
collection. The solution will also create a customized
“collection technique map” for each individual collection
case, rescoring as new events occur.





New York assigns collections cases based on dollar value
and uses a standard series of techniques to pursue
payment. This approach has led to a substantial backlog of
collections cases. As new events occur that affect
collectability, cases are not reprioritized. As a result,
collectors may spend time on cases that have low
probability of collection.

$100 million

increase in revenues collected over a 3 year

Substantial reduction in backlog of delinquent debts owed
to the state


New York State Department of Taxation and Finance

© 2010 IBM Corporation

What lessons can you take away from these case studies?

We are in the “early adopter” phase of sophisticated, real
time analytics in government

Learn more about these methods and technologies

Helps to provide proper oversight

Can recommend proactively in your audits

Analytics uses math and statistics to examine your existing data sources smarter, faster, and

Not an exercise in “magical” data sources

Privacy concerns (if any) can be overcome

Using analytics is plain ol’ good government

Great approach for helping to close budget gaps and be responsible stewards of public funds

Can enable positive ROI in same budget cycle

Allows you to do more with the same or less

Sentinel effect

Adopt analytics in YOUR audits and day
day work

Ways to get started

Pick a business risk and jump in

Government Commission

Fraud Club

© 2010 IBM Corporation

Frieda Yueh

(914) 474

Shaun Barry

(516) 203

Contact Information