FDIC Mandate Boosts Data Quality In World's Largest XBRL Project

Although data tagging with Extensible Business Reporting Language produces accurate, transmittable and researchable financial data, it has languished as a largely theoretical improvement as agencies have balked at mandating use of the XML-based format — until now.

The FDIC mandated last year that banks use XBRL, which tags individual data cells and each line of content in financial statements so they can be deciphered by computers and are easier to query and analyze. Two of the four other agencies in the interagency Federal Financial Institutions Examination Council (FFIEC), the Federal Reserve and the OCC, quickly followed suit.

FFIEC regulators share several auditing challenges, but chief among them is the need to quickly identify errors and problems within Call Reports — banks' quarterly filings of balance sheet and income information. Roughly 8,200 financial institutions with a combined $9 trillion in assets supply FFIEC members with reports that cover 1,200 data points such as total assets, net income and equity.

Until recently, a full third of banks' reports were tripping red flags, indicating possible inaccuracies or discrepancies in data. If a bank showed a total assets boost of more than 200 percent from quarter to quarter, for example, that would trip an audit. The regulator would then inquire with the bank in question — by phone, e-mail or fax — to see if, for instance, it had made an acquisition. Sometimes simple reporting errors occur. Sometimes reports don't meet fundamental validation criteria, such as total assets equaling liabilities plus equity.

"You'd be surprised," says Martin Henning, associate director for the statistics branch at the FDIC. "Some straightforward accounting relationships that should always hold true didn't all the time."

At the FDIC, a staff of 20 to 25 was handling "interventions," checking on a whopping 13,000 to 14,000 reporting irregularities every quarter. Moreover, the FDIC must check reporting data against the Federal Reserve's separate pool of reported financial information. The two massive data sets reside in separate mainframes and use different data-quality standards and data-processing systems.

Enter XBRL. The FFIEC launched a modernized Call Report process called the Central Data Repository (CDR) in October 2005. It's the first broad-scale XBRL project in the United States and marks the largest use of the XBRL standard worldwide, according to the FFIEC.

The FDIC now works directly with seven firms that specialize in making Call Report software, pushing updated XBRL schema to them on a quarterly basis. The software companies then push the XBRL changes to their banking clients — something they did on a quarterly basis to account for small Call Report reporting changes. The process is transparent to banks when they submit Call Report data to the CDR electronically.

The improvement to reporting data quality has been drastic. The FFIEC reports that only five percent of Call Reports in the most recently finished quarter — the first using XBRL tagging — set off audit flags. That compares with about 33 percent in prior quarters.

"That's the 'Wow,'" says Dan Roberts, chair for the XBRL-U.S. Steering Committee Consortium. "The FDIC is only now entering its second reporting cycle using this new environment. They went from 66 percent data accuracy to 95 percent data accuracy within the first quarter it was implemented."