The last thing in the world I need right now is to detect more fraud.
-- Senior Health Care Financing Administration (HCFA) official
Given the current state of affairs within the industry, many officials would prefer not to detect more fraud. They have far too many cases and far too few resources. They see too many investigations and prosecutions abandoned for lack of capacity in the court system and see little value in exacerbating the problem. Moreover, they suspect that a significant increase in the detection rate (which might result from improved detection systems) might be misinterpreted by many as yet more evidence of poor performance. As a matter of course, therefore, few within the industry show much eagerness to upgrade detection capabilities. For many practitioners, existing incentives push the opposite way: Their lives would be much easier if the level of detected frauds went down. Hopefully, a better understanding of the nature of the fraud-control business will eventually produce a greater enthusiasm for improved detection systems within the industry. The inadequacy of existing detection systems will become obvious when insurers start measuring the level of fraud systematically.
The IRS, in tackling tax-refund fraud, had no idea just how bad their detection systems were until they carried out their first random sampling in 1994 of claims based on the earned income tax credit (EITC). During the 1994 filing season, they detected 160 million dollars' worth of fraudulent refunds. The statistical measurement of the EITC fraud problem, through random sampling, indicated 3 billion dollars' worth of fraudulent refunds passing through the system. IRS detection systems were uncovering only a fraction (roughly 5 percent) of the total. Suddenly, the need for the IRS to upgrade its detection tools became obvious. Also, as the policy emphasis shifted from fraud investigation to fraud control, im-