The first keynote was delivered by Eric Rosengren, President and CEO of the Boston Fed. Rosengren opened by showing an interesting chart on the LIBOR to Overnight Swap spread, which jumped last summer and has been very volatile ever since, evidence of the reluctance of banks willingness to lend to each other.
Rosengren covered the role of liquidity in risk modeling, which he noted was largely underestimated in many models over the last year. He also noted that other fundamental assumptions were wrong, like the one that housing prices across the US are not correlated (he showed a chart of regional housing data over the last five years that looked highly correlated.)
Rosengren also spoke about the impact of rogue trading and legal settlements. Many institutions think these losses are 1 in a 1000 year events, but as we get more data, it’s emerging that these events are much more common than previously thought.
Regarding scenarios analysis and stress testing, Rosengren asked how much confidence should we put into this? In many cases, the stress tests did not accurately take into account the risks. He noted that the effect of falling housing pricing was not accurately assessed. He also noted that the impact of mortgage defaults on liquidty was universally missed.
In the Q&A period, he went on to say that we need to be more humble about the effect of some of these unexpected events and that we need to broaden our thinking about what could possibly happen.
A key theme of Rosengren’s talk is that organizations are too willing to ignore what they consider 1 in a 1000 year events, when in fact these events are turning out to be quite frequent. For instance, last year there were 14 losses over $1 billion reported. He reinforced this notion in the Q&A session that extreme losses have occurred much more frequently than we would have assumed a couple years ago.
Rosengren was followed by Randall Kroszner, Member of the Board of Governors, Federal Reserve. Kroszner took a broader perspective on Basel II, and the enhancements the framework committee is considering. He noted that banks pursuing AMA qualification need strong senior management and board oversight. He also noted that senior management can create an AMA that’s reflective of organizational realities.
Kroszner noted that Basel II has been the official regulation for just one month, but the implementation will take some time. Implementation must be taken “thoughtfully and deliberately” by individual banks which should first start with a sober and frank appraisal of their current state.
The core banks will have to plan in place for AMA qualification by Oct 1, and Kroszner noted that this will require buy-in and resource commitment from the top.
Kroszner also noted that their hope is to provide more information over the next couple months but provided some initial thoughts on what the plan will have to cover:
- Gaps between existing practice and AMA
- Objective and measurable milestones
- Planning and governance process for meeting qualification requirements fully
He noted that the final rule allows 36 months before exiting the parallel run phase.
After some discussion of upcoming improvements to the Basel II framework, Kroszner addressed the standardized approach for non-core banks. He stated that the Fed expects that Basel II (referring to both the AMA and standardized approaches) will make the US banking system more resilient.
A key theme that emerged from Kroszner’s talk and the subsequent Q&A period was that a one size fits all approach is probably not best for the range of institutions we have in the US. Rosengren noted in the Q&A period that the final rule is more of a principles-based than a rules-based document and repeated that “it’s not clear that one size fits all.” He also noted that there’s already a wide range of practices in play right now.
Someone asked if Basel II make us more vulnerable to systemic risk because of model convergence? Kroszner responded that the flexibility of the final rule and the judgement afforded by the icap process should mitigate systemic risk. Rosengren said that oprisk has enough variety in the modeling, but that credit risk calculations over the last year may have been too reliant on the same historical data.