Anatomy of Fraud Investigation and Detection - Part 1

While Anatomy of fraud investigation will lead one towards a structured approach towards detection and eventual prosecution, it is interesting to understand the mindset of fraudsters before deep dive into the details and nuances. More often than not it is the burning desire to get rich overnight that gets executive lean towards irregularities leading to organized financial frauds.

Let us closely look at some of the top corporate frauds of recent times

1. The Enron scandal – Enron misrepresented earnings, modified balance sheet to indicate favorable performance, was hiding huge debts from failed deals and projects utilizing accounting loopholes
2. The Cendant Scandal – CUC reports inflated the company’s revenue by $500 million over a period of three years.
3. Worldcom – Worldcom improperly recorded $3.3 Bn in profits on its books between 1999 to 2002
4. Healthsouth – Over Reported their earnings by $ 2.7 Bn engineering a massive company wide accounting fraud
5. Tyco International – CEO and CFO paid themselves $150 mn in unearned bonuses and loans.
6. Qwest Communications – Qwest was fined $250 Mn for accounting irregularities
7. Satyam Computer Services – CEO Resigned confessing that he manipulated accounts by USD $1.45 Bn
8. Roop Bhansali – CRB a top notch investment bank in 1992 collected money from investors through mutual funds, fixed deposits and debentures and invested in non existent shell companies. Investors lost Rs 1200 Crore
9. Home Trade – Home Trade borrowed approximately Rs 400 Crore from over 25 cooperative banks to buy government securities for these banks. However, home trade sold same securities to multiple banks.
10. Ringing Bells Private Limited – Ringing Bells announced a Freedom 251 smartphone for Rs 251 in india. They claimed they have delivered 70000 phones and will deliver remaining phones soon

We observe that most places people committed this frauds to over report Revenues Profits or to report high expenses to non existent companies, employees or other non existent entities to transfer money to their private accounts.

These frauds impact common people and their lives severely. Many governments have put regulatory bodies to monitor all corporate activities including financial details to protect against such frauds and sudden impact to people.

Indian government has taken some concrete steps to fight the black money menace. Aside of Demonetization, GST Implementation , one of the biggest and although still work in progress, most concrete step is going after shell companies.

The government has prepared a list of 16794 companies with inputs from all its investigative agencies. These companies will be investigated further and some of them may get branded as shell companies. SEBI has already identified 331 companies as possible shell companies and limited trading in these companies.

Such companies help holders of black money convert their unaccounted wealth into legally earned white money into regular banking channels. Most of these money gets transferred to other countries in tax safe havens.

As most of these shell companies are in business of converting black money and not really into doing any real transactions they work on inflated pricing and inflating prices over a period of time to save tax through long terms capital gains, there are several possible data mining techniques to fight this menace.

Each time fraudsters come up with a new idea of fraud to get away with what they are trying to do. Can we still find some pattern in these frauds and utilize artificial intelligence, machine-learning, data-analytics, deep-learning techniques to detect or predict such frauds thereby savings people from getting impacted?

Although it is not always possible to automate detecting fraud with 100% accuracy, we can use a combination of automated algorithms to flag probable patterns where risk of fraud is high and then use manual mechanism to detect fraudulent behaviour.

To help financial auditors detect fraud from huge amount of available data, implementing statistical methods to flag possibility of fraud is definitely be a big help where the auditors can go more deeper and look for patterns of fraudulent transactions.

These methods can be divided into two categories

1. Statistical Techniques to detect anomaly in financial data in financial statements. These methods typically rely on financial ratios, change in financial ratios over time and corresponding change in related parameters as associated cost
2. Text Analytics techniques. These techniques help cross reference data from financial data to management summary or various investors reports to substantiate management claims through financial parameters, also to find any anomaly in what is said in public against what is there in internal emails or social communications.

In the next part we will detail about Statistical techniques to determine fraud through financial ratios

About the Author : Gopi Agarwal is the Chief Technology Officer at Silfra Technologies; Gopi has varied interest in the areas of Machine Learning, Deep Learning and Artificial Intelligence. He is currently researching on application of Data Science in Financial Fraud Detection.

Want to know more? Contact us.