Detecting Fraud With Data Mining Slides

Embed Size (px)

Citation preview

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    1/45

    CPAs & ADVISORS

    IDEA Webinar Series

    Detecting Fraud with Data Mining

    Presented by

    Jeremy Clopton, CPA, CFE, ACDA

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    2/45

    How Fraud is Detected

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    3/45

    Intriguing Quote on Big Data

    As of 2012, about 2.5 exabytes of data are

    created each day, and that number is doubling

    every 40 months or so.

    Harvard Business Review, Big Data: The

    Management Revolution

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    4/45

    Growth of Unstructured Data

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    5/45

    Paper-based & limitedelectronic testing

    (Sampling)

    Reactive ProactiveResponsiveness

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    6/45

    Problems with the old method

    Ineffective

    Inefficient

    Reactive Hindsight

    Prevalence of Big Data

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    7/45

    The new method: a wish list

    Do more with less

    100 percent coverage

    Increase effectiveness More insight

    Not overly complex

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    8/45

    The New Method: Data Analytics

    processes & activities designed to obtain & evaluatedata to extract useful information...

    Useful information includesConflicts of interest

    Unknown relationships

    Abnormal patterns of activity

    Errors in key processesControl weaknesses

    Hindsight, insight, foresight

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    9/45

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    10/45

    Common Data Mining Areas

    Vendors & accounts payable

    Employees & payroll

    Expense reimbursement

    Travel & entertainment

    General ledger

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    11/45

    Vendor Attribute Capture

    0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

    TIN

    Address

    Name

    TIN Address Name

    Attribute Present 29,276 68,804 69,535

    Attribute Missing 40,259 731 -

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    12/45

    Vendor Activity Assessment

    FY 2012 FY 2011 FY 2010 FY 2009 pre-FY 2009

    10,051 5,765 5,443 4,598 43,678

    -

    5,000

    10,000

    15,000

    20,000

    25,000

    30,000

    35,000

    40,000

    45,000

    50,000

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    13/45

    Name Mining

    Mick E. Mowse

    Princess Ariel

    George Ruth

    John Dough

    1. Acronym / Initials

    2. Anagrams

    3. Fictitious Names

    4. OthersSubstitution

    Insertion or Omission

    Transposition

    Numb3r Subst1tut10n

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    14/45

    Employee/Vendor Relationships

    Matching

    Attributes

    Employee

    ID

    First

    Name

    Middle

    Initial Last Name Vendor ID Name City State

    Total

    Payments

    Address 123456789 Jeremy R Clopton 987654321 Vendor Name Anytown MO 16,040

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    15/45

    Conflicts of Interest

    Matching

    Attributes Employee ID

    First

    Name

    Middle

    Initial Last Name Vendor ID Name City State

    Total

    Payments

    Address 131313131 Beth E DavisD58468431Davis Designs Anytown MO 5,768

    Address, TIN 687431598 George R Davis

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    16/45

    Address Mining Mailbox Services

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    17/45

    Address Mining Proximity

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    18/45

    Address Mining Proximity

    Employer

    UPS Store

    Employee

    Home

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    19/45

    Proximity Analysis

    Vendor (A)

    Jeremys Design Company, 123 5thStreet, Anytown, MO (Total Payments = $84,337)

    Employee (B)

    Jeremy Clopton, 4300 Oak Street, Anytown, MO

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    20/45

    Proximity Analysis

    AP Manager

    Vinnys

    Salvage

    Yard

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    21/45

    Proximity Analysis

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    22/45

    Vendor Trending Analysis

    Vendor: JLM Plumbing Authorized: Janice L. McPhearson

    Test phase

    Acceleration as

    confidence builds

    Getting

    Greedy

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    23/45

    Vendor Trending

    Acceleration Patterns:Vendors exhibiting a pattern of

    increased activity over multiple

    consecutive periods.

    Valley Patterns:Vendors exhibiting a pattern of activity

    characterized by long periods of

    inactivity between periods of activity.

    Spike Patterns:Vendors exhibiting a pattern of activity

    characterized by unusually high

    payments in a single period.

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    24/45

    Payment Trend Analysis

    By Day of Week

    By Day of Month

    By Month

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    25/45

    Benfords Law Analysis Expected

    FrequenciesFirst Digit Expected Frequency Second Digit Expected Frequency

    1 30.10% 0 11.97%

    2 17.61% 1 11.39%

    3 12.49% 2 10.88%

    4 9.69% 3 10.43%

    5 7.92% 4 10.03%

    6 6.69% 5 9.67%

    7 5.80% 6 9.34%

    8 5.12% 7 9.04%

    9 4.58% 8 8.76%

    9 8.50%

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    26/45

    Benfords Law A/P

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    27/45

    Benfords Law A/P

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    28/45

    Benfords Law Expense Accounts

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    29/45

    Check Sequence analysis

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    30/45

    Purchasing Cards Split Transactions

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    31/45

    Analysis of Overtime Hours (654 hrs)

    -

    20

    40

    60

    80

    100

    120

    140

    160

    180

    Regular Pay Overtime 1.5 Holiday Pay Vacation Pay Sick Leave

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    32/45

    Analysis of Vacation Hours (426 hrs)

    -

    20

    40

    60

    80

    100

    120

    Regular Pay Overtime 1.5 Holiday Pay Vacation Pay Sick Leave Personal Circumstance Leave

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    33/45

    Analysis Of Holiday Hours (182 hrs)

    -

    10

    20

    30

    40

    50

    60

    70

    80

    90

    Regular Pay Overtime 1.5 Holiday Pay Vacation Pay Sick Leave

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    34/45

    Other Areas of Application

    Access log controls testing

    Maintenance file analysis Vendors

    Customers Loans

    Credit cards & purchasing cards

    Employee expense reimbursements

    General ledger

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    35/45

    Whats Next?

    Automated testing

    Analytics at the speed of business

    Foresightin addition to hindsight & insight

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    36/45

    Paper-based & limitedelectronic testing

    (Sampling)

    Data Analytics

    (100% coverage, ad hoc

    electronic testing)

    Continuous Auditing

    (Automated analytics,100% coverage)

    Reactive ProactiveResponsiveness

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    37/45

    Example 1 Manufacturing

    Company Description

    Revenues: $7.9 billion

    Internal audit staff: 5

    Operating divisions: 20Vendors: 100,000

    Employees: 7,000

    Payments per year: 250,000

    Application of ContinuousAuditing

    Risk #1: Conflicts of interest

    Solution:Annual employee/vendormatching

    Risk #2:Duplicate payments

    Solution:Annual analysis of all invoicesfor potential duplicates

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    38/45

    Example 2 Public University

    Company Description

    Revenues: $1 billion

    Internal audit staff: 5

    Vendors: 83,000Employees: 3,900

    Purchasing card users: 1,100

    Application of ContinuousAuditing

    Risk #1:Duplicate payments

    Solution:Quarterly analysis of allinvoices for potentialduplicates

    Risk #2: Split transactionsSolution:

    Quarterly analysis ofcardholder transaction details

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    39/45

    Critical Information on Continuous

    Auditing

    Continuous auditing continuous monitoring

    Continuous auditingOwned by internal audit

    Risk & control assessmentProcess focused

    Continuous monitoring

    Owned by managementEffectiveness & adequacy of controls

    Control focused

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    40/45

    A PLAN TO GET YOU THERE

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    41/45

    Assess Risk

    DefineObjectives

    Obtain Data

    Develop & ApplyProcedures

    Analyze Results

    Manage Results

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    42/45

    Progression of Procedure

    Development

    Single-Purpose(Individual) Tests

    Groups of Similar

    Tests

    Repetitive

    Individual Tests

    Automation ofGroups of Tests

    Groups ofRepetitive Similar

    Tests

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    43/45

    Challenge to Consider Data Quality

    417-865-8701, (417)865-8701, 8658701, 417-8658701

    Missoura, MO, Mis, Miss, MZ, MS, Miz, Mizz

    PO Box 34, P.O. Box 34, Box 34, Bx 34, P.O Box 34

    Clopton, Clapton, Clompton, Clampton, Cloptin

    12345, 12345a, 12345-1, 012345

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    44/45

    Resources

    IIA Global Technology Audit Guides Continuous Auditing

    Fraud Prevention and Detection in an AutomatedWorld

    Data Analysis

    ISACA White Paper

    Data Analytics A Practical Approach

    http://www.audimation.com

  • 8/10/2019 Detecting Fraud With Data Mining Slides

    45/45

    Contact Information

    Jeremy Clopton, CPA, CFE, ACDA

    Managing Consultant | BKD, LLP Forensics & Valuation Services Group

    910 East St. Louis Street, Suite 200

    Springfield, Missouri 65806

    417.865.8701

    [email protected]

    Social Media

    Blog: bkdforensics.com

    Twitter: @j313

    LinkedIn: http://www.linkedin.com/in/jeremyclopton/