SBIR Phase I:Advanced Irregularity Prediction System (AIPS) to identify Accounting Errors and Financial Fraud in Small & Medium Businesses

This Small Business Innovation Research (SBIR) Phase I project aims to assess the feasibility of using statistical methods such as Hierarchical Clustering and Binary Classification to predict accounting errors and financial fraud amongst small and medium businesses (SMBs). Checks and balances in place to protect larger businesses from accounting errors and financial fraud are too complex, time consuming, and costly for SMBs. As a result, small and medium businesses suffer greater losses from accounting errors and financial fraud than any other sized businesses. Advanced data mining and analysis techniques may offer a simple and cost effective solution to the problem. The proposed research includes data collection and statistical assessment to predict accounting errors and fraud in financial systems. While the techniques proposed have been successfully used in fields such as information security to determine threats and prevent risks (e.g. intrusion prevention, anti-virus, anti-spyware, web content filtering), they have not been applied nor tested to transactional accounting data sets. Fraud is a serious problem. The Association of Certified Fraud Examiners (ACFE) estimates that organizations lose 7% of their revenues to fraud. Applied to the projected 2008 U.S. GDP, this 7% equals approximately $994 billion in fraud losses. Assuming an average corporate tax rate of 40%, results in uncollected tax revenue of nearly $398 billion. The "clean up" costs related to errors and fraud are also significant. Forensic audits can easily exceed $100,000. In many cases, businesses affected by fraud are forced to layoff employees, stop payments to suppliers, decrease quality levels, or reduce other vital spending in hope of recouping losses. In the worst cases, companies may even be forced to close. According to the ACFE, nearly half of all fraud in the U.S. occurs in businesses with less than 100 employees. The ability to detect errors and fraud using predictive techniques could provide benefit to the millions of small businesses fueling the U.S. economy.