The regulatory post-mortems released today on the failures of SIVB and SBNY could be summarized as “round up the usual suspects.”

The reality is that in order to create a bank regulatory and supervisory structure that will improve significantly the stability and smooth functioning of our financial system over time, the inherent conflicts of interest and irreconcilable mandates that exist in bank oversite need to be resolved, and the FDIC needs to have the tools necessary to address the agency problem that results because regulations are inadequately enforced and so-called “prudential” supervision is functionally nonexistent.

This can be accomplished by correcting layered misalignments in the supervisory system by the following reforms:

• Reconstitute the FDIC’s Board of Directors to remove representatives from other bank supervisors and replace them with members deeply qualified in risk management, insurance/actuarial, and bank operations, and rotate a head of one of the Federal Reserve Banks.
• Consolidate bank and BHC supervision under the OCC, removing bank and BHC supervisory authority from the Federal Reserve and the FDIC. This is necessary, among other reasons, to remove the outsized influence that the Division of Supervision has within the FDIC.
• Move BHC stress testing out from under the Fed to the FDIC and maintain an oversight role for the Fed through the FSOC.
• Importantly, substantially enhance the FDIC’s backup supervisory role by giving them GAO-like audit authority to check the examination procedures and conclusions of OCC exams, authorize it to accompany the OCC on examinations when it sees fit, and vest the FDIC with full authority to conduct independent, at-will lateral issue-based examinations of groups of banks when it believes a particular set of risks in the sector pose a potential threat that need to be better monitored and measured.
• Finally, vest the FDIC with greater authority over the composition of the Call Reports and Y9-C forms–the offsite reporting mechanism available to bank regulators, and give the FDIC unfettered access to all data collected by the OCC from the banks and BHCs that it supervises.

Here’s why I believe that this set of reforms is superior to alternatives such as lowering the asset level of banks subject to stress tests:

The FDIC’s “agency” problem is a classic one.

In Part 1, I spent some time discussing how the various banking supervisors are funded. If you recall, the OCC is funded from examination fees that relate to the size of the bank being supervised. The loss of a charter to another regulator directly reduces the budget of the OCC. This funding structure renders the OCC and other supervisors inherently conflicted with their responsibilities as an “agent” of the FDIC because it is the case that chartering authorities compete on “laxity.”

This is more than a theory. It explains behavior.

To see the proof, see Gary Whalen’s 2002 paper on the subject. This paper focuses on identifying the factors that best explain the decision of national banks to convert to a state charter. The results showed that several indicators of bank risk significantly increase the likelihood of a national bank charter flip. He used several supervisory variables to demonstrate that banks are more likely to flip their charter the worse their supervisory ratings.

A classic charter flip occurred when Countrywide shed its OCC charter to become an FSB.

The OCC was tired of losing budget to laxer regulators and was known to have lengthy discussions with banks before they decided to change charters. At the risk of sounding clinical, I doubt that the argument to maintain a national bank charter was based on tougher supervision and higher capital requirements.

Subprime Lending: anatomy a poorly handled risk

Leading up to the GFC, my colleagues and I in the FDIC’s Division of Insurance (before it digested the Division of Research and Statics and became far less agile) were working hard to understand subprime lending. At the time, few in the FDIC’s Division of Supervision (DOS) believed that this was a “bank” problem.[1] But I was even more shocked when I learned that the OCC didn’t think it was much of a problem at all.

I remember one particular meeting vividly. We were in a conference room at FDIC headquarters where we had been discussing with the other regulators about amending the Call Reports (Y9-Cs for BHCs) to add an item requiring banks to report how much subprime lending they were doing. Of course, in retrospect, it sounds like an obviously good idea; but not everyone thought so then.

To this point in time, we didn’t know how widespread subprime lending was among insured depositories. We suspected that it was growing because we were monitoring the dispersion of yields of banks’ mortgage loans, and it was increasingly the case that banks were surfacing that were earning exceptionally high yields on their mortgage loan books. There are only two explanations for that: either some banks were providing services and an “experience” that were so exceptional that their customers were willing to pay significantly higher rates to access that experience, or they were issuing mortgages to significantly riskier borrowers who failed to qualify for the “going rate.”

We thought we had a strong argument to change the Call Report. The hesitance came from within and without. Our own division of supervision (DOS) leaned toward agreeing with the other supervisors that subprime lending was too hard to define to add it to the Call Report and that only examiners could identify the activity during an examination where “they’d know it when they saw it.”

In one of these meetings, we sat across the table from the OCC’s first “Risk Czar,” Scott Calhoun when he told my FDIC colleagues and me that no one had ever convinced him that subprime lending was inherently any riskier than any other kind of lending. To be fair, I think he was pushing back against an emerging idea (held strongly by, then-Director of Supervision for the FDIC, Nick Ketcha) that insured depositories should be banned from engaging in subprime lending.

That sounds brilliant now; but there wasn’t any evidence yet that the activities couldn’t be engaged in safely and soundly by a bank. The problem with Scott Calhoun’s statement is that it was already theoretically obvious that subprime lending was in fact riskier than other kinds of mortgage lending, even without a track record of loss data to prove it.

We had studied the characteristics of subprime borrows and at least knew that they were significantly more financially vulnerable to exogenous shocks to their financial worlds like a recession or high unemployment levels. At that point, we in the FDIC’s Division of Insurance knew that subprime lending was penetrating banks and we knew, at least, that if a bank were to engage in subprime lending, it needed to be done with significantly higher capital levels and under close monitoring. But, at the time, we had few allies, even internally, to implement the measures that we knew were prudent to manage the risk to the deposit insurance fund.

After the horse had left the barn, the FFIEC finally agreed to slam the barn door by defining, for Call Report purposes, subprime lending as loans made to borrows with a sub-660 FICO (or equivalent) score. It was too late. At the time, because of growing concerns about regulatory burden, the examination cycles were being lengthened—not shortened. So, we flew blind into a crisis.

Okay boomer, what’s your story telling have to do with what’s going on today?

It’s the same pattern that repeats with costly regularity.

The FDIC’s Division of insurance in its early days was focused on learning to measure and price the risk to the deposit insurance fund. We were doing things like identifying and using new offsite information to monitor and measure risks. We were building “dashboards” using the FDIC’s massive research information system (RIS) database. We were developing stress tests to assess capital adequacy at insured depositories. All this was in order for the FDIC’s senior management to understand how risks were moving and changing within the banking system. We were trying to build “macro” monitoring tools that would supplement and, at times, challenge the “micro” measures that we were receiving through the examination process, and it was beyond promising—it was working.

That is, until it didn’t.

More topically relevant to today’s crisis is a dispute that occurred within the FDIC about how to build a risk dashboard and, in particular, how to report about interest rate risk in the system. At the nimble programming hands of one, Mike Anas, we had used Call Report information from the RIS to build a model measuring trends in banks’ securities books and other characteristics of asset/liability management. It looked at readily available data like maturity characteristics of a bank’s balance sheet among other things.

What we had built showed that interest rate risk was rising at insured depositories.

Internally, however, our little division’s work was beginning to be scrutinized by the “gorilla in the room,” [2] the FDIC’s Division of Supervision (DOS). We were soon tasked to work with DOS to build the dashboard.

The dispute: our “rudimentary” measures of interest rate risk, which were on the rise, were at odds with trends in the component of banks’ CAMELS ratings that measure this risk: the “S” rating. “S” stands for sensitivity (generally interest rate and market sensitivity). We quickly lost a battle with DOS who wanted to include in the model a weighting for banks’ “S” scores that fully diluted the influence of our modeling efforts on the results that would be reported to FDIC’s risk committee.

In fact, we were never able to get traction around any risk that at the time seemed to be anything other than an imminent risk (i.e. low-probability events) even if these risks could prove catastrophic.

A solid concept of risk management requires evaluating catastrophic scenarios even if the probability of that scenario seems remote because a small number if it is multiplied by a big enough number results in a big number.

Being imaginative about risks in this way, though, is something that is easily bread out of bank examiners. While I’d like to think that I was different when I was a field examiner for the FDIC, even I succumbed to a review process that squelched creative bank examining. Quantity, not quality of examination is what is rewarded and results in promotions.

This is the nature of Bank Examination:

• During good times, your primary objective is to cut budgets so that you examine less.
• During bad times, you are understaffed because of the budget cuts that happened during good times.
• During good times, it’s a “banker’s market” where its easy to make things difficult for troublesome examiners who are raising too many “matters requiring the board’s attention.”
• In a “banker’s market,” bank lobbies are strong, and congress is eager to listen (and this is not a partisan statement).
• In a “banker’s market,” supervision backs off.

Here is another cynical observation: examiners can’t be promoted without problem bank experience. It shouldn’t be surprising that examiners that have experience dealing with the complexities of writing cease and desist orders or other consent orders have more to put on their resumes. Furthermore, it takes a crisis to both get this experience and to create positions higher in the organization to which to ascend.   

Until all this changes, we will repeat with some regularity the unchecked accumulation of risks that inevitably lead to banking crises.

What about the Federal Reserve?

That’s for part 3…

[1] My then boss, Kathy Kalser, and I presented at the Boston regional conference of FDIC examiners on the characteristics and risks of subprime lending. We were followed by a senior examination specialist from the FDIC’s DOS who told our audience that he “wasn’t going to waste their time talking about something that they’d never see in banks.”

[2] Because bank examination is highly labor intensive, supervising banks requires lots of people. Within any government agency, size matters, and numbers mean influence. The DOS of the FDIC is many orders of magnitudes larger than any other part of the FDIC.