Bill Moreland is a partner in a company/website called BankRegData.com. The Austin, Texas company BankRegData started in 2010 to provide a web interface to aggregated Federal Depository Insurance Company (FDIC) insured Bank Quarterly Call Report data and additionally offer a narrative/interpretation on the bank book dynamics (see 2016 Q4 Asset Review or Winners & Losers: Citigroup). Since 2012 this data is taken from the FFIEC Central Data Repository. FFIEC is the acronym for the Federal Financial Institution Examination Council. The FFIEC is an interagency standards committee across five banking regulators including: Federal Reserve Board of Governors , Federal Deposit Insurance Corporation, National Credit Union Administration, Office of the Comptroller of the Currency, and the Consumer Financial Protection Bureau (note the Securities Exchange Commission is not listed as one of the regulators in the FFIEC.). What BankRegData does is aggregate, and to some degree normalize, the Quarterly Call Report Data. BankRegData will get you (to a good first approximation) the distribution of securities ( deposits, mortgages, credit card accounts, loans, and tradable securities) at a quarterly aggregation level across the Banking Book of the major Bank Holding Companies. Jesse Eisinger outlined some of the problems with the consistency of this data relative to Wells Fargo SEC filings (for example) in a 2011 DealBook article, Tackling Reams of Bank Data Can Take Diligence, and Trust. Obvious error sources include: correct and complete linkage of banks to BHCs, correct and complete enumeration of BHC accrual portfolios, incomplete individual security information, and aggregated/individual security cash flow quantitative modeling error. On the other hand BankRegData is placing the location of 16.779T USD Assets that don’t move all that much and produce cashflows according to fairly standard approximation algorithms on a relatively small set of market and econometric levels. Think of it like Thorp counting cards at the casino back in day, but in this new case of “allBHC” simulation the casino has limited ability to shuffle the cards or even change the decks as play continues. BankRegData is counting the cards for you in allBHC (with the assistance of the FFIEC and each of the individual banks) and gives you a web interface to check the current and historical card counts.
Understand that each BHC prepares regulatory reports by monitoring the individual accounts holding each of the bank book securities. In a bank the size of JPM there may be Retail data files disclosing the actual monthly realized cash flows, late fees accrued, and default write downs for several 100 million loan accounts. This account (or security level) information is generally aggregated for regulatory reporting, but it is known at a very granular account level within the BHC itself. So within the BHC the error sources can be limited to the the individual account quantitative modeling error. For the lack to a better name let’s call this the perBHC problem. All the other error sources can be eliminated (or controlled) for the BHCs accrual inventory, otherwise the BHC has big internal control issues. The BHC does not know the other BHC positions at the same account granularity as their own internal positions. The PBC hypothesis is that the error in allBHC problem can be controlled or smoothed out in computing NIM explanatories of forward simulation so you can see a fit with the perBHC data.