Symposium 2009 DISCUSSANT’S COMMENTS - Data Mining Journal Entries for Fraud Detection: A Pilot Study – R S Debreceny & Glen L Gray Eckhardt Kriel I want to congratulate the authors on a very interesting and topical paper. It is well researched, documented and while I agree with the paper and its conclusions I ask myself: Does it go far enough? Auditors have been doing this for over 5 years There is a massive amount of data - results and experiences Its perhaps time to ask – “How may frauds have been uncovered and how useful really is this exercise”? As the Authors state: “There is, however, very little knowledge of the efficacy of this important class of audit procedures.” My experiences: During 2002 to 2006 I led a team who performed JE analysis for roughly 500 listed clients in Canada every year. Frequent interaction with other areas and firms. In that period millions of journal entries were analysed. Spent thousands of hours of work. Tests complied with SAS 99 and more. We found many of the strange anomalies described in paper. We found many control exceptions and issues but; NO SIGNIFICANT FRAUDS WERE UNCOVERED. 2 MOTIVATION - JUSTIFICATION MOTIVATION - JUSTIFICATION CHALLENGES IN PROCESS CHALLENGES ITS NOT EASY! Tough challenges that are not mentioned in the paper. Accessing and extracting the data. Understanding unique client environment and FCP. Data Completeness verification. Are the appropriate tests being run? 3 Our procedures included those mentioned in the paper plus: Data Completeness Trial Balance Roll-Up Data Anomaly Tests Blank Date Fields Zero Dollar Items Blank Account Numbers Unbalanced Journal Entries Blank transaction description Blank Preparer ID Foreign Currency Adjustments Unusual Currencies Key Transaction Tests Accounts not in the Chart of Accounts Line items greater than the absolute value of a dollar threshold Back Dated Journal Entries greater than the absolute value of a dollar threshold Digital Filter Tests Benford Tests on leading and trailing digits Round Number testing Additional Testing Procedures: • Modified Standard Account/Period/Amount Cross Tabulation • Identify any Journal Entry exceeding the average daily posting amount for that account by x% • Identify any Journal Entry exceeding the average daily number of transactions for that account by x% • Identify Journal Entries with identical dollar amounts • Account Combination Testing Debits to Income Accounts and Credits to Expense Accounts Debits to Liability Accounts and Credits to Income Accounts Debits to Asset Accounts and Credits to Income Accounts Debits to Fixed Assets and Credits to Expenses • Identify Journal Entries with key words in description field – “professional fees”, “litigation”, “reserve”, etc. • Identify journal entries passed by unauthorized personnel etc., etc., etc. 4 STANDARD PROCEDURES Standard Procedures Data Trending by Source Code – Cross Tabulation of key data fields, cross comparison of linked items January Health Care Client - extract CASH AND CASH EQUIVALENTS February March April May June Total ($190,471) ($170,670) ($135,814) ($72,919) ($110,761) ($156,791) ($837,426) $20,218,188 $19,753,469 $19,109,002 $20,801,779 $22,089,963 $25,057,885 $127,030,286 ALLOWANCE FOR UNCOLLECTIBLES $573,675 $500,398 $440,131 $319,390 $470,985 $702,050 $3,006,629 ALLOWANCE FOR CONTRACTUALS $874,739 $627,713 $1,222,530 $1,476,653 $1,338,183 $1,219,811 $6,759,629 ($4,748,761) ($4,472,640) ($4,850,452) ($5,068,811) ($5,453,126) ($5,319,425) ($29,913,215) ($1) ($1) $0 ($1) ($1) $0 ($4) PATIENT A/R GROSS RECEIVABLE INPATIENT DAILY HOSPITAL REVENUE Total Comments: Patient revenues of $ 136 million were accrued, resulting in a corresponding increase in AR. PATIENT A/R GROSS RECEIVABLE INPATIENT DAILY HOSPITAL REVENUE 30,000,000 6,000,000 25,000,000 5,000,000 20,000,000 4,000,000 15,000,000 3,000,000 10,000,000 2,000,000 5,000,000 1,000,000 0 0 January February March April May June 5 CROSS TABULATION EXAMPLE Cross Tabulation Example SAS 99 details a number of procedures that auditors can follow to respond to the objective of consideration of fraud in F/S audit. Journal entry testing is one of these. I have reservations that, on its own, journal entry testing, is effective. So any paper or article on the subject must include it as one of a combination of tests. To detect potential irregularity in financial statements any analysis must be focused: It must contemplate fraudulent misstatement and profile its characteristics; It must search for the characteristics; It must be broadly based. 6 LIMITATIONS Limitations of Journal Entry Testing POSSIBLE FUTURE RESEARCH AREAS Expanding on SAS 99 Paragraphs 28/29/30 7 TEXTUAL MINING Possible future research Textual Mining 8 BENFORD ON COMPANY RESULTS Possible future research Benford on Listed Company Results – AD Saville1 1. Reference: SAJEMS NS 9 (2006) No 3 341 9 Advanced Financial Reporting Analysis – Example MICROSTRATEGY – Application of Different tests 1997 Q1 1997 Q2 1997 Q3 1997 Q4 1998 Q1 1998 Q2 1998 Q3 1998 Q4 1999 Q1 1999 Q2 1999 Q3 1999 Q4 Total Revenue 8,137 11,875 14,751 18,794 19,895 23,790 27,014 35,731 35,784 45,638 54,555 69,352 Gross Profit 5,981 73.5% 9,420 79.3% 11,855 80.4% 15,185 80.8% 16,194 81.4% 19,125 80.4% 21,770 80.6% 29,560 82.7% 28,655 80.1% 37,169 81.4% 44,655 81.9% 57,774 83.3% (941) 199 498 616 711 1,744 2,685 4,186 2,541 4,573 5,531 5,674 (1,003) 122 486 516 542 942 1,928 2,766 1,859 3,211 3,794 4,044 (994) 3,723 1,306 (310) (881) (1,230) (3,975) 3,538 (1,161) 529 (1,827) 1,252 Operating Earnings (loss) Net Income Net Cash provided by (used in) operations Deferred Revenue Bal at Dec 31 9,387 11,478 80,000 80,000 70,000 70,000 60,000 60,000 16,782 50,000 50,000 Total Revenue 40,000 40,000 30,000 20,000 Total Revenue 30,000 Gross Profit 20,000 Net Cash provided by (used in) operations 10,000 10,000 1999 Q4 1999 Q3 1999 Q2 1999 Q1 1998 Q4 1998 Q3 1998 Q2 1998 Q1 1997 Q4 1997 Q3 1997 Q2 1997 Q1 1997 Q2 1997 Q3 1997 Q4 1998 Q1 1998 Q2 1998 Q3 1998 Q4 1999 Q1 1999 Q2 1999 Q3 1999 Q4 -10,000 1997 Q1 0 0 10 ADVANCED FINANCIAL REPORTING ANALYSIS Possible future research CONCLUSION 11 TOO MUCH INFORMATION ? Too much Information ? M Gladwell Stephen Few’s2 Commentary on Gladwell: “Modern problems, on the other hand, are not the result of missing or hidden information, Gladwell argues, but the result, in a sense, of too much information and the complicated challenge of understanding it. The problems that we face today do not exist because we lack information, but because we don’t understand it. They can be solved only by developing skills and tools to make sense of information that is often complex. In other words, the major obstacle to solving modern problems isn’t the lack of information, solved by acquiring it, but the lack of understanding, solved by analytics.” 2. Visual Business Intelligence September-21-09 12 QUESTIONS Eckhardt Kriel CA (SA) E Kriel & Associates Inc. 1148 Forest Trail Place Oakville ON L6M 3H7 www.krielassoc.com www.d3cifer.com Mobile: +1. 416 451-3919 Direct: +1. 416 451-3919 Email: ekriel@cogeco.ca 13