The financial services industry moves as at a frantic pace. While you’re just getting to grips with one regulatory requirement, another is sure to spring up and need your attention. To help you to make sense of the latest PRA consultation paper for residential mortgage risk weights (CP29/16), our team of heavyweight risk experts have summarised the new requirements; the implications to mortgage lenders; and crucially what may be next.
Background
In July, the PRA published consultation paper CP29/16 titled “Residential mortgage risk weights”, following concerns raised about mortgage risk weights in the UK stress test. To help our clients respond to this, several of our experienced risk and regulatory experts established a working group to discuss the proposed changes to IRB rating practices and to fully understand the implications to banks and building societies with IRB status (as well as those seeking to gain it in the coming years).
In the IRB approach to capital assessment, institutions calculate risk weights for their assets by using historical data to produce estimates of PD, EAD and LGD (plus CFs as appropriate). Since the inception of the IRB approach, the following high-level principles have applied:
- PD estimates should represent the average PD experienced over an economic cycle
- PD estimates should not be procyclical - that is vary in step with the economic cycle
- LGD estimates should represent those experienced in an economic downturn
These are designed to ensure that enough capital is held to withstand a downturn at all times, without the need to raise capital in the onset of a downturn (when it will be unavailable other than via Central Bank bail outs).
There has been a spectrum of approaches - from Point in Time (PiT) to Through The Cycle (TTC) - to generating PD estimates in accordance with those principles. The Bank of England describes them as follows:
- PiT: firms seek explicitly to estimate default risk over a fixed period, typically one year. Under such an approach the increase in default risk in a downturn results in a general tendency for migration to lower grades. When combined with the fixed estimate of the long-run default rate for the grade, the result is a higher capital requirement. Where data are sufficient, grade level default rates tend to be stable and relatively close to the PD estimates
- TTC: firms seek to remove cyclical volatility from the estimation of default risk, by assessing borrowers’ performance across the economic cycle. TTC ratings do not react to changes in the cycle, so there is no consequent volatility in capital requirements. Actual default rates in each grade diverge from the PD estimate for the grade, with actual default rates relatively higher at weak points in the cycle and relatively lower at strong points.
In our experience, the majority of institutions adopting the IRB approach for Residential Mortgages use either a highly PiT approach or Variable Scalars - a TTC approach.
The consultation paper
The consultation paper covers changes in the PRA’s position on what constitute acceptable methodologies for calculating PD and LGD for Residential Mortgages. The key aspects of this are:
- The use of highly PiT PD approaches will be explicitly forbidden
- The use of Variable Scalars will be explicitly forbidden
- A quantitative measure of procyclicality is introduced along with a 30% upper threshold on it - in other words approaches going forward must be “nearly TTC”
- They’re explicit that an appropriate economic cycle in the UK must contain conditions equivalent to those experienced in the early 1990s
- Additional guidance on LGD is added stating that a house price decrease of at least 25% must be used
While this amounts to a change in their regulatory stance, it does not represent a shift away from the principles noted above. In fact, the updated rules will force closer adherence to them and the result will undoubtedly be a more resilient banking system and economy. So, we welcome the changes and believe they’re an important step in raising standards.That’s not to say that becoming compliant will be straightforward!...
The implications
It is clear that the majority of institutions who have adopted (or are about to adopt) the IRB approach will be affected by the changes - since they use methodologies that will be explicitly prohibited - and will need to re-visit their approach. However, it is not clear to us at this stage - in part because we have some questions on the paper (see below) - how to create an approach that satisfies all of the new requirements. There is a tension between requiring fixed LRA PDs per grade and restricting procyclicality that is difficult to address without re-developing the PD models themselves (the PRA states that they do not expect organisations to need to do this). Furthermore, some avenues that might have been opened by adopting a more recent economic cycle including the recent downturn - where account level performance data is more readily available - have been closed by the requirement that the early 90s should be included.
So, further thinking is likely required across the industry and clarification on some points of detail is needed from the PRA. Many UK financial institutions have strengthened their ability to understand the impact of economics on their portfolios as a result of new regulation on Stress Testing and IRFS 9 accounting standards and this can only help in tackling CP29/16. All of these things have significant commonalities and a well-designed common model suite that addresses all of them is clearly highly beneficial. This is an approach that we’ve advocated for some time and will take as a design principle in solving CP29/16 for our clients.
What next?
Jaywing’s working group are in the process of compiling some questions to submit directly to the PRA in response to this consultation paper. In the meantime, our experts will be exploring modelling approaches to address these new guidelines. We have recognised the need for pragmatism given the parallel regulatory change activities in full swing between now and the March 2019 deadline.