Data Analysis in theTennessee
Division of State Audit
Software Used
ACL, currently version 8 Since 1996
SAS, both mainframe and PC Since the late 1980s
Culprit, mainframe Since the beginning of time . . . uh, I
mean since we started doingdata analysis!
Why ACL?
Selected because, particularly at that time, it seemed to manage the very large data sets better than its competitors.
Why SAS?
Mainframe: Allows the use of large mainframe data sets that cannot be managed in ACL.
PC: Allows the use of data sets that cannot be managed in ACL and provides higher-level statistical analysis.
Why Culprit?
Cullinane Corporation’s EDP Auditor application. Still in use because there are routine jobs working in Culprit that would not pay to re-write. New data analysis not written in Culprit.
Data Analysis Performed
Accounting System
Summary: tie data from an extract to reporting--ensure that data file is complete
Stratification: identify outliersSamples (per auditor request) Monthly reconciliation of the
month’s accounting system records to ensure data is complete.
Accounting System
Vendor-type payments to state employees
Reasonable date valuesPayments made within required
timeValid data in the fieldsValid social security numbers
Payroll
Duplicate checks Invalid SSNsSupplemental pay calculationsOvertimeReconcile gross pay by area to
payroll expenditures in the accounting system
Payroll
Recalculate FICA and net pay
Check number sequence (gap/duplicate)
Checks over a given dollar value considered excessive
Purchasing
Test percentage of minority businesses getting contracts
Online approvals by persons with appropriate authority
Fixed asset field is appropriate based on the item's commodity code
TennCare (Medicaid)
State employees receiving TennCare
TennCare enrollees with P.O. boxes
Out-of-state enrollees Incarcerated persons receiving
TennCareAnd the list goes on . . .
Authorization Controls
From a variety of systems, we have compared security (access) listings to payroll records to test that staff with access to the system are still employed by the state (or college, or university, or agency . . . .)
WIC
Test for clients meeting program parameters
Summarize vouchers issued by staffStratify results
Lottery
Calculation of sales, liabilities, and changes in each before and after a programming error caused no repeating numbers to be drawn
Unemployment Benefits
Recalculate Benefit AmountsVoluminous DataComplex DataUse of Mainframe SAS and ACL
The Good, The Bad,and the Promising
ACL: The Good
ACL is a strong performer, and can be effectively used by audit staff with realistic training
Continually updated with new features
Very responsive to customer comments
ACL: If only . . .
It would allow users more control over data imported from CSV and Excel files.
It had “prettier” output.They hadn’t gone and messed
up the log structure (from Version 7).
SAS: The Good
Haven’t found the data it won’t handle!
Output can be formatted to look like just about anything you want.
Can perform higher-level statistics than most audit tools.
SAS: If only . . .
If only staff auditors could readily learn the base product without having to buy a front-end!
If only it weren’t quite so resource intensive!
The Promising
We are currently wading into the world of GIS in auditing.
Alternative fuel stationsDollars generated by or provided to
individual counties Individuals served by countyPossible ARRA reporting tool