Bank of America: The challenge of unified accounting

If we asked you the value of your assets, how close would you get?  Okay, we’ll give you a few hours to figure it out, but you need to include the value of the house and car(s). Do you think you could estimate the value to within 0.2%?

That is the thought experiment you should run to put Bank of America’s $4 billion accounting error into perspective. With end-2013 balance sheet assets of roughly $2.1 trillion, that’s the size of the mistake: 0.2%. (It's also 1.8% of book capital.) Now, of course, banks have an army of accountants and auditors who are supposed to be computing this stuff. And, the error that led the bank to overestimate its regulatory capital was a simple one. But we would argue that the problem is both bigger and a bit different than what is commonly assumed.

To understand the challenge big banks face, take a look at this picture, which might be described as the reverse genealogy of the four largest financial holding companies in the United States since 1990. Looking just at the Bank of America, we see that 20 years ago, it was 13 significant institutions. (This next picture implies that 30 years ago, the number was nearly 40.)

Bank mergers are pretty complicated. Each entity comes not only with its own people and culture, but also with its own accounting, information technology (IT) and risk-management systems. And that’s where the problems start. Infrastructure is a cost center in a firm. Putting in new IT systems is expensive – expenses that need to be paid upfront and only recovered over time. What acquisitive, empire-building management has an incentive to reduce profits today to benefit future generations?

We saw a clear sign of this in the run-up to January 1, 2000. To address their “Y2K problem,” banks scoured the countryside for retired Cobol programmers to come in and patch their antiquated IT systems. The goal was to make sure the computers could tell the difference between the year 2000 and the year 1900.  These firms hadn’t rewritten their code for decades; all they’d done was add to it and patch it.

Scarce programming talent is usually reserved for more profitable activities – like building a new platform for a global derivatives business or developing and implementing new algorithms for high-frequency trading in the equities or foreign exchange markets. It is not usually devoted to upgrading things that look like they are working.

As we think about the Bank of America error, we are lead to wonder: do these large firms have reliable real-time enterprise-wide accounting, IT and risk-management systems that are fully integrated? If so, how could errors creep into regulatory accounting measures, but not into general accounting reports? If management cannot ensure reporting consistency, can shareholders and regulators detect it?

In short, can anyone get a complete and unified picture of what is going on across the entirety of a U.S. megabank each morning?  As a matter of public safety, we should ensure they can!

Note: Our original post made an incorrect comparison with net worth, rather than assets.