Embrace internal model diversity

In the 1980s the world of modelling in the fashion industry was uniform. All commercial and fashion models that walked the runway or were shown in magazines represented the same standard of beauty. But that didn’t reflect real life, did it? We are diverse and there is no one size that fits all. Fortunately, this is increasingly being celebrated with models of all body shapes, sizes and colours/ethnicity.
In the banking industry there are models too. Not the human kind though. Under the Basel II banking standards, banks were stimulated to build sophisticated internal models - not for advertising but to measure their risk exposures. Careful measurement of key risk parameters is crucial to ensure strong risk management.

Unfortunately, the recent proposals for the application of the revised Basel standards in the UK (Basel III finalisation) and in the US (Basel III Endgame) favour the standardised approach and include strict restrictions on the use of internal models for calculating the regulatory capital. This is bad news for risk management, which is also bad news for our investments in Credit Risk Sharing (“CRS”).

Our concern is twofold. The first is more overarching – under these proposals banks are incentivised, or even required, to use the same, standardised view of risk. When everyone measures risk the same way, they will all mitigate the same risks, but will also all ignore the same risks. This greatly increases the probability that one such risk materialising will affect a large cross-section of banks in the financial system – the definition of systemic risk.

The second focuses on CRS. By sharing the credit risk of banks’ core lending activities, we also share in the quality of their risk measurement and management. The better banks measure and manage their credit risk, the more comfort we have and the better we can price a transaction. Conversely, when banks become strongly disincentivised to maintain adequate risk measurement tools, such as internal rating models, investors will have to become more conservative in their assessment, if they get sufficient comfort to invest at all.

Standardised risk weights oversimplify the diversity in the real world. It is true that risk models and risk metrics are imperfect and outcomes from one model might not be the same as those from another model. However, this is no different when it comes to clothing size in fashion. And ‘one size’ fits (almost) no one.

Please see here to read in more (technical) detail why we consider internal models crucial for CRS.