National Fraud Initiative
Kevin Boon
IT Performance Specialist
Audit Commission
IDEA User Group – November 27 2008
Agenda
• Background to NFI
• Successes NFI 2006/07
• New Areas
• Statutory Powers & Data Protection
• NFI Process
• NFI Process and IDEA
Background to NFI
What is it?
• Data Matching Exercise
When is it carried out?
• Every two years
• but moving to more frequently
What does it do?
• Prevents and detects fraud
Background to NFI
Who is involved?
• The public sector
– Local councils
– NHS bodies
– Emergency services
– Government departments
• Plus the private sector (voluntary)
– Financial Institutions
– Social housing bodies
Background to NFI
Background to NFI
What data is involved?
• Student loans (absconders/Benefits/visas etc)
• Payroll (multiple employment/H & S/visas)
• Benefits (undeclared income)
• Pensions (deads)
• Rents/arrears (absconders)
• Right to buys (undeclared capital/ money laundering)
• Visas/asylum seekers (deportation/working)
• Licences (undeclared income, visas etc)
• Care homes (deads)
• Concessionary fares/parking (abuse)
• Council Tax (Single Persons discount)
• Creditors (duplicate payments)
• Companies House (conflict of Interest)
Background to NFI - 2006/07 Outcomes
Background to NFI - National Report
• Overpayments/fraud of £140 million detected (overall
figure since 1996 nearly £450 million)
• 157 public sector staff ineligible to work
• 69 council properties were recovered.
• 2,819 cases where pension continued to be paid after
the death of the pensioner
• 16,102 deceased persons’ blue badges were cancelled
• duplicate payments of £1.75 million detected
• £24 million+ housing benefit overpayments detected
• £4 million Income Support (IS) and Job Seeker
Allowance (JSA) fraud and overpayments
• 4,310 cases IS/JSA benefits reduced/ceased
• 31 prosecutions, 22 administrative penalties and 76
cautions issued by JCP/TPS
New areas 1 Expansion of existing work
• Introduce government
departments and agencies
– HB, Payroll, Procurement etc
• Council of Mortgage Lenders
• Cross border matching
• Housing Associations
New areas 2 Non fraud
• Council Tax arrears
• Rent arrears
• Other debts
New areas 3 Public safety
• Child protection
– List 99
– Sex offenders register
• Absconders from justice
• Identity fraud
• Counter terrorism
Data protection and Statutory powers
• Audit Commission Act
• Data Protection Act
– Information Commissioner
– Code of Data Matching Practice
– Parliamentary approval
– Fair processing
• Serious Crime Act
NFI Process
Specification
Extraction
HousekeepingSubmissionChecking
Manipulation
Results
Investigation
Prioritisation
Reporting
Matching
Cleansing
NFI Process and IDEA
Specification
Extraction
HousekeepingSubmissionChecking
Manipulation
Results
Investigation
Prioritisation
Reporting
Matching
Cleansing
NFI Process and IDEA
Manipulation/Pre-filtering Example Specification
• Print reports
• Joining databases
• Aligning/Appending databases
• Virtual fields
– Padding out
– Combining/Separating
– Changing formats e.g. dates
– Adding indicators
• Discard unwanted records/fields
• Reordering fields
NFI Process and IDEA
Checking/Validation
• Date sorting for completeness
• Control totals for reasonableness
• Alpha sorting for completeness
• Format checking for data quality
• Record count – truncation
• Gap detection for quality
NFI Process and IDEA
Prioritisation
Mostly available online (filtering, sorting)
not suitable for Trade Creditors duplicate payments
WHY?
• Periodic dataset used
• Multi-level data required
• Non trade creditors included
• Lack of pre-filtering
• Duplicate or regular payment?
• LARGE VOLUME OF MATCHES
But
Leading to:-
NFI Process and IDEA
Prioritisation
Trade Creditor duplicate payments
Example - On basis of same creditor and invoice amount
View matches in IDEA
Use IDEA functions:-
– Extraction by equations
– Summarisation
– Extraction by range
– Joining
– Sampling
NFI Process and IDEA
Prioritisation – using IDEA
Suggested process:-
• Extraction by equations (to discard unwanted records)
- view
• Summarise by Cred Ref/Amount (to show number
of records in each match)
• Extract on number of records (to exclude regular
payments and singletons) - view
• Extract on Amount range (to exclude outliers) - view
• Random sample (to select for checking) - view
• Joining (to reconstruct matches) – view
Next stage
NFI Process and IDEA
Housekeeping
• Producing subsets for email lists (exception reports v
contacts list)
• Updating contacts lists from external sources
NFI Process and IDEA
Other possible uses
• Benford’s law
• Statistical analysis
• Piloting new matches
Questions
Payroll Specification Back
Field name Data format
Employee reference number Character
Employee post number Character
Department Character
Title or Sex Character
Surname Character
Forename(s) Character
Address line 1 Character
Address line 2 Character
Address line 3 Character
Address line 4 Character
Post code Character
Date of birth Date
Date started Date
Date left Date
Leaver indicator Character
National insurance number Character
Gross pay to date Numeric
Date last paid Date
Sort code Character
Bank account Character
Building society roll number Character
Print Report Back Import
Print Report import Back
Trade Creditors – payments history Back
NFI Process and IDEA
Prioritisation – using IDEA
Equations:-
• Non numeric invoice numbers – view
• Include only paid invoices - view
• Exclude specific payment types - view
• Exclude general types - view
Next stage
Trade Creditors – payments history Back
Example
One of the last two characters of the invoice number is
numeric:-
@right(@strip(supplier_invoice_number) ,1) <= “9” .OR
@mid(@right(@strip(supplier_invoice_number) ,2),1,1) <= “9”
Trade Creditors – payments history
Back
Example
Default date imported as „Error‟:-
@IsFieldDataValid ("PDATE") = 1
Trade Creditors – payments history
Back
Example
Advances included in dataset:-
.not. @isini(„advance‟ , supplier_invoice_number)
Trade Creditors – payments history
Back
Example
Housing benefit payments included in error, the invoice number
refers to the HB number e.g. HB123456:-
.not. @regexpr(Supplier_invoice_number , “HB[0-9]”)
Trade Creditors – payments history
Back
Example
Potential quarterly and monthly payments excluded. Singletons
excluded where other sides of the match no longer there.
.not. (no_of_recs = 1 .or. no_of_recs = 4 .or. no_of_recs = 12)
Trade Creditors – payments history Back
Example
Risk areas seen to be between £100 and £5,000.
Total_Invoice_Amount > 100 .and. Total_Invoice_Amount < 5000
Trade Creditors – payments history
Back
Biased or random sampling
technique could be used to
select matches for checking
Trade Creditors – payments history
Back
New target population selected, the other side of the
matches need to be rejoined.
Joining required
Primary database = original file
Secondary database = sample file
Key = Credref/Amount
All match records

National Fraud Initiative using IDEA

  • 1.
    National Fraud Initiative KevinBoon IT Performance Specialist Audit Commission IDEA User Group – November 27 2008
  • 2.
    Agenda • Background toNFI • Successes NFI 2006/07 • New Areas • Statutory Powers & Data Protection • NFI Process • NFI Process and IDEA
  • 3.
    Background to NFI Whatis it? • Data Matching Exercise When is it carried out? • Every two years • but moving to more frequently What does it do? • Prevents and detects fraud
  • 4.
    Background to NFI Whois involved? • The public sector – Local councils – NHS bodies – Emergency services – Government departments • Plus the private sector (voluntary) – Financial Institutions – Social housing bodies
  • 5.
  • 6.
    Background to NFI Whatdata is involved? • Student loans (absconders/Benefits/visas etc) • Payroll (multiple employment/H & S/visas) • Benefits (undeclared income) • Pensions (deads) • Rents/arrears (absconders) • Right to buys (undeclared capital/ money laundering) • Visas/asylum seekers (deportation/working) • Licences (undeclared income, visas etc) • Care homes (deads) • Concessionary fares/parking (abuse) • Council Tax (Single Persons discount) • Creditors (duplicate payments) • Companies House (conflict of Interest)
  • 7.
    Background to NFI- 2006/07 Outcomes
  • 8.
    Background to NFI- National Report • Overpayments/fraud of £140 million detected (overall figure since 1996 nearly £450 million) • 157 public sector staff ineligible to work • 69 council properties were recovered. • 2,819 cases where pension continued to be paid after the death of the pensioner • 16,102 deceased persons’ blue badges were cancelled • duplicate payments of £1.75 million detected • £24 million+ housing benefit overpayments detected • £4 million Income Support (IS) and Job Seeker Allowance (JSA) fraud and overpayments • 4,310 cases IS/JSA benefits reduced/ceased • 31 prosecutions, 22 administrative penalties and 76 cautions issued by JCP/TPS
  • 9.
    New areas 1Expansion of existing work • Introduce government departments and agencies – HB, Payroll, Procurement etc • Council of Mortgage Lenders • Cross border matching • Housing Associations
  • 10.
    New areas 2Non fraud • Council Tax arrears • Rent arrears • Other debts
  • 11.
    New areas 3Public safety • Child protection – List 99 – Sex offenders register • Absconders from justice • Identity fraud • Counter terrorism
  • 12.
    Data protection andStatutory powers • Audit Commission Act • Data Protection Act – Information Commissioner – Code of Data Matching Practice – Parliamentary approval – Fair processing • Serious Crime Act
  • 13.
  • 14.
    NFI Process andIDEA Specification Extraction HousekeepingSubmissionChecking Manipulation Results Investigation Prioritisation Reporting Matching Cleansing
  • 15.
    NFI Process andIDEA Manipulation/Pre-filtering Example Specification • Print reports • Joining databases • Aligning/Appending databases • Virtual fields – Padding out – Combining/Separating – Changing formats e.g. dates – Adding indicators • Discard unwanted records/fields • Reordering fields
  • 16.
    NFI Process andIDEA Checking/Validation • Date sorting for completeness • Control totals for reasonableness • Alpha sorting for completeness • Format checking for data quality • Record count – truncation • Gap detection for quality
  • 17.
    NFI Process andIDEA Prioritisation Mostly available online (filtering, sorting) not suitable for Trade Creditors duplicate payments WHY? • Periodic dataset used • Multi-level data required • Non trade creditors included • Lack of pre-filtering • Duplicate or regular payment? • LARGE VOLUME OF MATCHES But Leading to:-
  • 18.
    NFI Process andIDEA Prioritisation Trade Creditor duplicate payments Example - On basis of same creditor and invoice amount View matches in IDEA Use IDEA functions:- – Extraction by equations – Summarisation – Extraction by range – Joining – Sampling
  • 19.
    NFI Process andIDEA Prioritisation – using IDEA Suggested process:- • Extraction by equations (to discard unwanted records) - view • Summarise by Cred Ref/Amount (to show number of records in each match) • Extract on number of records (to exclude regular payments and singletons) - view • Extract on Amount range (to exclude outliers) - view • Random sample (to select for checking) - view • Joining (to reconstruct matches) – view Next stage
  • 20.
    NFI Process andIDEA Housekeeping • Producing subsets for email lists (exception reports v contacts list) • Updating contacts lists from external sources
  • 21.
    NFI Process andIDEA Other possible uses • Benford’s law • Statistical analysis • Piloting new matches
  • 23.
  • 24.
    Payroll Specification Back Fieldname Data format Employee reference number Character Employee post number Character Department Character Title or Sex Character Surname Character Forename(s) Character Address line 1 Character Address line 2 Character Address line 3 Character Address line 4 Character Post code Character Date of birth Date Date started Date Date left Date Leaver indicator Character National insurance number Character Gross pay to date Numeric Date last paid Date Sort code Character Bank account Character Building society roll number Character
  • 25.
  • 26.
  • 27.
    Trade Creditors –payments history Back
  • 28.
    NFI Process andIDEA Prioritisation – using IDEA Equations:- • Non numeric invoice numbers – view • Include only paid invoices - view • Exclude specific payment types - view • Exclude general types - view Next stage
  • 29.
    Trade Creditors –payments history Back Example One of the last two characters of the invoice number is numeric:- @right(@strip(supplier_invoice_number) ,1) <= “9” .OR @mid(@right(@strip(supplier_invoice_number) ,2),1,1) <= “9”
  • 30.
    Trade Creditors –payments history Back Example Default date imported as „Error‟:- @IsFieldDataValid ("PDATE") = 1
  • 31.
    Trade Creditors –payments history Back Example Advances included in dataset:- .not. @isini(„advance‟ , supplier_invoice_number)
  • 32.
    Trade Creditors –payments history Back Example Housing benefit payments included in error, the invoice number refers to the HB number e.g. HB123456:- .not. @regexpr(Supplier_invoice_number , “HB[0-9]”)
  • 33.
    Trade Creditors –payments history Back Example Potential quarterly and monthly payments excluded. Singletons excluded where other sides of the match no longer there. .not. (no_of_recs = 1 .or. no_of_recs = 4 .or. no_of_recs = 12)
  • 34.
    Trade Creditors –payments history Back Example Risk areas seen to be between £100 and £5,000. Total_Invoice_Amount > 100 .and. Total_Invoice_Amount < 5000
  • 35.
    Trade Creditors –payments history Back Biased or random sampling technique could be used to select matches for checking
  • 36.
    Trade Creditors –payments history Back New target population selected, the other side of the matches need to be rejoined. Joining required Primary database = original file Secondary database = sample file Key = Credref/Amount All match records