Interest in preventing waste, fraud, and abuse runs high in the federal government these days. Yet fraudulent contractors and healthcare providers continue to get paid so financial functions within agencies are always looking for a better mousetrap. For instance, The Association of Government Accountants released a study this May of best practices for scanning large data sets using the latest data analytics technologies for identifying potential improper payments.
Health and Human Services and the Defense Department account for most of the federal government’s annual improper payment tab. The 2011 figure was down a little, to $115 billion, thanks to some concerted agency work and prodding by the Office of Management and Budget.
One source of improper payments is poor reconciliation among invoices generated by a buying entity like a naval component. Payment authorizations from a service agency like the Defense Finance and Accounting Service (DFAS) and Treasury fund balances. The payment processes themselves, or course, have been automated for many years. But process automation doesn’t ensure only the right payments are made. In fact, it speeds up the payment of mistaken or fraudulent claims.
The volume of transactions makes it impossible for human financial analysts to check them all. Checking each transaction requires analysis of the large amounts of sub-transaction level data generated by the whole process.
One of the companies in this market space that I’m familiar with is Oversight Systems. Its software falls into the category of big data analyzers. It examines large data sets in financial settings to look for anomalies. According to Chris Rossie, Oversight’s vp for public sector and international markets, Oversight has been ingesting large data sets for the Navy at DFAS for several years, and it’s been getting results.
Analyzing the data, Rossie told me, involves taking in 2,400 files monthly from the Navy, DFAS, and Treasury. Those files come in over a five-to-seven-day period. But here’s the rub: The 2,400 files contain data on 350,000 monthly transactions-about 4 billion a year. The transactions break down into billions and billions of data elements because a single transaction might contain a half dozen pieces of discrete information and metadata. It all amounts to a treasure trove-several terabytes a year-that can be sifted and analyzed for anything that doesn’t quite add up.
By looking at the most granular data at high speed, the software can reconcile 90% of these three agencies’ transactions in the first pass. DFAS calls the process Business Activity Monitoring, or BAM. Rossie says that each month, the BAM reduces to a manageable number the unreconciled transactions that are then examined by humans. Built-in workflow tools route the exceptions to DFAS analysts for manual inspection. Over the past few years, the number of transactions reconciled automatically has increased by 50%, Rossie says, which gives analysts more time to fix whatever is wrong with those that get flagged.
Since DFAS has been using Oversight, it has prevented $1 billion in improper payments annually. Improper can be wrong amounts, the wrong vendor, or even late payments which, for small and disadvantaged businesses, are statutorily considered improper.
Rossie points out that while matching is critical, so is high speed. You need both in order to enable the government to close out its books soon after the end of a month. A third benefit is the ability to declare audit readiness, pursuant to the DoD’s larger goal of a clean financial audit. DFAS uses Oversight for all of the armed services plus the Defense Logistics Agency. The Navy uses Oversight additionally for audit readiness.
“The Navy scenario is about as complex as will happen in government,” Rossie says. “If one-to-one transactions are the simplest, the Navy’s are many-to-many, with multiple contractors, obligated funds crossing many components, payments spread out over years against an original appropriation or obligation.” Rossie says he’s amazed at how accurately the government actually completes its financial reconciliations, given that they are far more complex than those of the average large corporation.
This seems like a successful example of the government’s use of the latest technology to me.