Banking? Keep it simple stupid
In 1873, Walter Bagehot wrote that “the business of banking ought to be simple; if it is hard it is wrong.” He would have struggled to recognize today’s banking system.
It is not just ever more ornate derivatives that bend the mind. Financial firms themselves have become fabulously complicated. Citigroup lists 2,061 subsidiaries and affiliates while the institutional chart of JPMorgan Chase is 267 pages long.
Complexity — as Bagehot predicted — has become a curse. If nobody can understand financial firms, they will become ever more accident prone.
The crisis that exploded a year ago offered a salutary lesson in the dangers of complexity. Many shareholders and creditors simply did not fully comprehend their investments. Instead they were forced to trust managers and the rating agencies.
Regulators too could be forgiven for scratching their heads.
“Supervisors are at a decided disadvantage in understanding risk-taking and compliance for firms that might involve dozens of jurisdictions, hundreds of legal entities and thousands of contractual relationships,” former Fed official Vincent Reinhart has written.
Indeed Basel II — the international capital code — was an admission of defeat by regulators. The message from the banking accord was that institutions had become so convoluted that only they were able to understand the risks they were taking.
Yet many intelligent executives of these same institutions failed spectacularly. It is no mean feat keeping tabs on an army of specialized financial engineers, lawyers and accountants.
As Robert Rubin, the former Treasury Secretary and Citigroup executive, acknowledged last year on the Charlie Rose show: “Unless you are either running the trading operations or running the independent risk management, you are not going to know the risk well enough to have a real sense of where those risks are.”
Making financial firms simpler will be far from simple.
One approach is coercive. Regulators can make it very uncomfortable to be big. Capital requirements that ratchet up with size would encourage firms to split themselves up into their component parts, giving managers and regulators a better shot at following what is going on. On the whole, smaller firms tend to be more straightforward.
Failing this, Reinhart has proposed a Lego model, in which financial firms would be composed of “well-defined modules.” A company made of units that can be easily disconnected from the whole would be easier to manage, with individually simple parts. Regulators can foster this model by insisting on a “living will,” complete with plans for how companies would salvage their firm in the event a single unit implodes.
Regulators need to make it much easier to understand financial statements. First, they should impose a strict consolidation of bank balance sheets, forcing them to incorporate all special purpose vehicles.
In addition, more information should be made available about banks’ risk-taking. Firms should be compelled to publish monthly indicators drawn up by regulators, including a measure of the relationship between short-term borrowing and long-term lending.
This would enable creditors to exercise proper discipline over the banks by pushing up their borrowing costs if they become too reckless. The notion that only banks themselves can understand their own risk-taking needs to be jettisoned.
Lastly, the government should also reduce the incentives for complexity. Financial institutions mirror the Byzantine structure of regulation, tax and accounting rules. They become complicated in order to shop for the most lenient regulator, lightest capital requirements and most tax efficient structure.
Paring down these rules and structures should be an underlying goal of any regulatory overhaul.
The first step to reducing the magnitude of future mishaps is to ensure that we can make sense of our financial institutions. The respect and awe often accorded to “black box” financial institutions is misplaced and dangerous. Instead we need to embrace simplicity.
The Year Since Lehman — related columns: