Banks have models for everything these days but, says Brian Caplen, they should not be fooled into thinking their problems are over.

Take a complex organisation such as an international bank, add in hundreds of new regulations and reporting requirements, and then graft on big data and advanced analytics. What do you have? An institution that is only manageable by models, and lots of them.

The number of models banks use is rising in leaps and bounds every year. There are models for calculating capital and liquidity, for determining the prices of products, for credit scoring, for targeting customer segments, for valuing financial instruments and for trying to prevent money laundering and fraud.

But it would be a very short-sighted management who concluded that once set up the models can take care of business. There are echoes in this approach to the misguided faith in risk management systems prior to the financial crisis.

In fact there are two big flaws with models – one is that the model is too liberal, doesn’t properly calculate the risks and leads to unexpected counterparty and hedging losses. The other is that the model is too conservative and so business is not carried out as efficiently as it could be.

Enter model risk management, or the practice of managing models effectively. McKinsey has just released a paper on The Evolution of Model Risk Management in which the consultant declares: “The stakes in managing model risk have never been higher. When things go wrong, consequences can be severe… the risk lies equally in defective models and model misuse.”

What needs to be done is hardly rocket science. The assumptions made in creating a model need to be well documented and everyone in the bank needs to understand them. The process needs to be better automated in a standardised way without overlap. Banks need to set up a “center of excellence for model development”, says McKinsey.

Banks can do all this and will surely see gains in the way the bank is managed. But they also need to do something else – retain a healthy scepticism for models. Only in a culture where challenging the model is encouraged can banks guard themselves from the dangers of model mishap. 

Brian Caplen is the editor of The BankerFollow him on Twitter @BrianCaplen

Register to receive my blog and in-depth coverage from the banking industry through the weekly e-newsletter. 

Order The Banker July edition

FREE trial access to Top 1000 World Banks

Join our community

View from Sibos: cybersecurity

Swift CEO Gottfried Leibbrandt reviews the various industry initiatives to prevent cyberattacks.

The Banker on Twitter

By continuing to use this site you consent to the use of cookies on your device as described in our cookie policy unless you have disabled them. You can change your cookie settings at any time but parts of our site will not function correctly without them.