By Gopal “Sharath” Sharathchandra, Senior Vice President, PNC.
Gopal, can you please tell the Risk Insights readers a little bit about yourself, your experiences and what your current professional focus is?
I have over 20 years’ experience in the area of credit risk management with background in data, model development, credit securitization, portfolio management – monitoring and analytics – and loss forecasting. I work in the area of credit portfolio management at PNC where I am responsible for CCAR credit loss forecasting, credit risk appetite analytics, CECL implementation and impact analytics, economic capital and risk-adjusted return measurement.
At the CECL Congress 2018, you will be speaking on your insight regarding, ‘Developing effective forecasts that fulfil requirements by defining reasonable and supportable period and reversion to historical loss’ Why is this a key talking point in the industry right now?
In order to forecast lifetime expected credit losses, the CECL standard requires that the forecast be made over a period that is reasonable and supportable and to, thereafter, revert to a loss estimate based on historical loss information that is adjusted for differences in credit characteristics as of the reporting date from that of the historical period.
Within these parameters, banks have significant leeway to make choices regarding the length of the reasonable and supportable period, the decision to use multiple scenarios with weighting, the trajectory of reversion to the historical loss estimate as well as the determination of the historical loss estimate itself. This makes this topic a key talking point.
Can you outline some of the challenges in interpreting the standard?
As mentioned above, the standard is not very prescriptive and provides considerable flexibility to banks on choosing their approach to determining their loss allowance. However, the approach chosen will still need to meet the parameters laid out by the standard including the overarching requirement that the loss allowance be determined such that, when it is deducted from the amortized cost basis, the result is the net amount expected to be collected. The challenges lie in choosing from among the approaches that meet these requirements while understanding their implications for ease of forecasting and implementation, the requirements for data and modeling as well as the behavior of the loss allowance across the credit cycle.
What are the key considerations that need to be made when conducting historical loss construction?
The standard requires that historical loss information be used and that it be adjusted for credit quality differences between the historical period and the present. While the standard does not specify how the historical loss estimate needs to be determined, when one considers the CECL requirement to estimate expected loss, it seems reasonable that the historical loss estimate would not just use historical data over a favorable credit period but would represent an average over both favorable and unfavorable periods. One way to meet the requirement to adjust the historical loss estimate for current credit quality is to form the estimate at granular portfolio segments and to then weight these segments using the current mix, which can account for portfolio mix differences. Broader adjustments, say due to underwriting changes, can be applied, differentially if necessary, directly across the portfolio segments.
Finally, what challenges do you foresee with CECL implementation over the coming years and how can institutions best plan to meet deadlines?
CECL implementation can be complex even for banks that have implemented stress testing such as CCAR/DFAST due to the additional requirements of CECL as well as the controls needed for financial reporting. It is important to identify at the outset all of the calculations and processes that need to occur for the bank to produce a reserve that is compliant with the CECL standard. Once this discovery process is complete, it may make sense for execution to prioritize components in the rough order of their importance to the reserves i.e. focus on developing a complete CECL reserve with attention first on modeling the most material portfolios and calculations and, if necessary, using simplified modeling for the less material calculations. Once a complete reserve estimate has been developed, then work on improving the sophistication of the simpler calculations. This iterative approach to implementation, rather than a linear one, may help reduce execution risk given the tight timelines.
