In the midst of all the new regulations and new risk management process upgrades required to meet new regulations and increased scrutiny, we may have forgotten one of the most potent risk and left it behind.
Lately, I’ve been traveling and doing a fair amount of work around CCAR, Stress testing and Operational risk. The Basic Indicator and Standard Basel methodologies for operational risk have been left mostly unchanged in the new Basel accord but given the recent crisis and the breadth of application of the new Basel accord NPR in the US, does it make sense for the smaller US banks to keep applying an unchanged basic indicator and standardized approach for operational risk? Does it make sense to relate operational risk capital to the level of gross revenues of the past three years? What happens when one applies the same method during the capital stress testing process? Won’t Oprisk capital actually go down under stress given the relationship with gross revenue (even if averaged over three years)?
The OCC’s Thomas J. Curry points out during a financial services roundtable on September 20, in D.C. that “many of the biggest problems we are dealing with today – and the ones that are proving most damaging to large banks – have nothing to do with credit risk, liquidity or any of the other areas of risk that are the typical focus of stress testing. Instead, they involve operational risk.”
To be fair the remarks are directed towards the operational risk framework and the lack of robust controls and monitoring infrastructure but why then are we not seeing any of this in the Standardized NPR for the Basel Accord when that aspect of operational risk is addressed in all its glorious details in the Basel III NPR for the advanced approaches? Is the qualitative side of the operational risk framework considered an advanced method and therefore only applicable to large banks?
Another interesting fact, the Federal Reserve CCAR methodologies for assessing operational risk capital for each bank is computed using projected operational risk losses across the 19 banks. “The model used to generate projections of operational risk losses projects operational risk losses for the 19 BHCs as a group and then distributes these aggregate losses across the BHCs in proportion to their Q3 2011 tier 1 common capital. This approach reflects the idiosyncratic nature of operational risk events, as well as differences across the 19 BHCs in the way these events are captured and reported in the data submitted to the Federal Reserve, which suggests that industry-level projections might be more robust than firm-specific results.”
So we know the advanced measurement approach has its issues because of the limited data history and the effect of new losses added to the history every quarter. In my conversations across the country, I’ve not met one operational risk capital person who doesn’t have to re-calibrate parts – or the entire model – from quarter to quarter due to the effects of a given set of current losses of the distribution assumptions driving the current model. Enough about this topic; many others cover this in detail.
So what about the Basic indicator and standardized approach? How has average operational risk capital behaved on an aggregate basis for banks in the $10B range? Here is an example of the computation of Oprisk capital for a small $13B bank. The computation is based on the basic indicator and compared to the bank’s capital reserve and the apportioned share for operational risk, which is approximately 18 percent of total earnings volatility.
Basic Indicator approach definition:
The least risk-sensitive of the Basel approaches is the Basic Indicator Approach, in which capital is a multiple (capital factor = 15 percent) of a single indicator (base), which is the average annual gross income, where positive, over the previous three years for which gross income was positive. The regulators have postulated that gross income serves as a proxy for the scale of operational risk exposure. Gross income is defined as the sum of net interest and noninterest income.
 Based on the estimation of earnings volatility contribution by Kuritzkes and Schuermann – “what we know and don’t know and can’t know about bank risks: a view from the trenches.”
 Approximated by capital reserve for comparison purpose.
In the table above, we computed the three year average gross income and then took 15 percent to derive the Basel II – basic Indicator capital. On the other side, we took the total capital set aside by the bank as per RWA calculation (which includes operational risk capital), we used 18 percent as the apportioned share of capital to cover operational risk as an indicative estimate of the operational risk capital set aside.
The scale difference between the two numbers in the difference column is a factor of about 10+, which is interesting. The more interesting fact is that the basic indicator approach computation actually indicates that capital goes down as we move through a period of major stress whereas overall level of capital for the bank increases substantially. Given the recent attention that operational risk has been gathering, would this not be a compelling enough reason to revise the basic approaches for operational risk capital calculations?
What can a bank without an AMA process do – or be expected to do – with respect to stress testing operational risk capital for its balance sheet? Is it time to re-think the basic approach to include other factors that may represent better drivers of potential operational risk and yet simple enough for smaller banks to use effectively.
A more basic method could include looking at the banks history of major losses (top 50) and supplementing this with an external loss database that is analyzed to draw out potential scenarios that would apply to said bank and ensuring that these losses are scaled appropriately. Then do a basic analysis of the amount of capital needed to sustain an average loss based on that scenario.
This point of view is meant to stimulate discussion and raise the point that operational risk basic and standard approaches may have come to the end of their useful life given the focus on risk sensitivities and forward looking estimation of stressed capital numbers.
If you have comments or other suggestions for tackling this issue, let’s discuss them with us in the comments section below.