Expected Losses are Not Baked into the Risk Weights


Tim Bush and Kevin Dowd


On December 15th 2015 there was an interesting exchange on the Treasury Committee between Steve Baker MP and then Bank of England Deputy-Governor Andrew Bailey. In response to a question from Mr. Baker about the incurred loss accounting model, which does not take account of expected losses, Mr. Bailey responded that, in the standardised Basel II approach, the expected loss “is baked into the risk weights”.[1]


The claim that expected losses are built into the risk weights is however an extraordinarily ill-informed answer.


In fact, it is mathematically impossible for Basel risk weights to take account of expected losses.


The headline Basle capital ratio is capital divided by ‘risk-weighted’ assets. Each risk-weighted asset is equal to the asset size multiplied by a fixed ‘risk weight’ that is zero or positive, and typically in the range between 0 and 1.


The problem with Mr Bailey’s assertion is that applying positive risk weights to assets in the denominator of the capital ratio cannot create a negative number for the capital numerator however large expected losses might be. On the other hand, booking expected losses correctly by way of a deduction from capital can create a negative number for capital, which would be indicative of insolvency whereby liabilities exceed assets.


Dealing with losses through higher risk weights can make a bank look like it has not got much capital, whereas taking losses directly against capital can reveal whether a bank is insolvent, i.e. whether its capital might be zero or negative.


In short, the Basel risk weights do not take account of the possibility of insolvency and yet this possibility is the one that matters when considering bank capital adequacy.



[1] See

http://data.parliament.uk/writtenevidence/committeeevidence.svc/evidencedocument/treasury-committee/independent-review-of-the-report-into-the-failure-of-hbos/oral/26149.html, Q185.