According to analysts at Barclays Capital, the average risk weighting that European banks apply to their assets when calculating regulatory capital ratios has declined steadily from 49% in 1998 to 35% at the end of 2010.
Although it might well be that banks have deliberately reoriented their exposures in recent years by shifting towards asset classes that consume less regulatory capital because they are less risky, it would be touchingly naive for any investor to draw too much comfort from this. That decline in average risk weightings was gathering speed, hitting 45% in 2005, just as banks accelerated into a brick wall of asset write-downs.
As the reduction in risk weightings as a percentage of total assets increases, so do suspicions that banks are cooking the books. Many bank executives are fearful that equity investors won’t supply the higher amounts of capital regulators now demand because their returns are so low. Lowering the risk weighting is a good way to make a capital ratio that comes out only just above the regulatory minimum suddenly appear well above the minimum. So while it is possible that low risk weightings are entirely justified, investors and regulators would be remiss if they did not ask for more explanation from banks that report a combination of both low RWA percentages of total assets and a high reliance on internal risk models to calculate these scores across their balance sheets.
The dispersion of risk weights different banks apply to similar portfolios is striking. Analysts at KBW recently reviewed 27 banks across Europe and note a range in risk weightings for residential mortgages from 6% at Dexia to 30% at Erste, with an average of 16%. They suggest that UK and Italian banks appear to be more conservative on their risk weightings applied to retail exposures than for example the Nordic and the French banks, which tend to apply lower risk weightings.
Corporate loan exposures also reveal a wide range in risk weightings, with a low of 27% at Deutsche Bank rising to a high of 85% at Lloyds Banking Group and a mean risk weight of 55%. All this might be justified. Deutsche reports a large proportion of its corporate exposures to highly rated triple- and double-A rated companies with many others providing cash and non-cash collateral to the bank, which is a leading provider of FX and interest-rate swaps.
The good news is that KBW analysts find no strong correlation between low risk weightings and subsequent larger-than-expected actual losses. Indeed banks with low risk weightings on mortgages also reported low deviations from the model to actual loss experience. But only 11 banks provided sufficient data to allow for such comparisons. And banks’ models are less successful in generating reasonable expectations compared with actual losses on corporate exposures.
The broad dispersion of these ranges and the tendency for risk weightings to fall when banks shift more assets to have their risk weightings calculated according to internal ratings, at the very least raises concerns over model risk and whether and how regulators and investors should build in buffers against it. One obvious answer is for regulators and investors to demand higher capital buffers for banks reporting both low risk weightings and large portions of assets risk-assessed via internal models.
Danske Bank, which has such a low risk weighting for mortgages, kicked off the latest round of European bank capital raising with a $3.8 billion fully underwritten rights issue in April to strengthen its capital. This deal was announced in February just a month after the Danish regulator had observed that Danske already held capital well above the regulatory minimum, but that it should raise the estimated losses its models generate because the evidence available to the regulator suggests internal models often underestimate actual losses, especially on corporate exposures.
They might not say so explicitly but regulators might be asking for higher capital ratios to compensate for concerns about model risk. Is any of this getting us anywhere though? It was Basle II regulation, still very much alive inside Basle III, that promoted the use of internal risk models in the first place. Regulators are now caught. If they insist on hard floors on risk weights, say 10% for mortgages, might that discourage the development of better risk models because banks would derive little benefit from them?
A simple first step would be for European regulators to demand a common disclosure format to allow investors more easily to compare variances in the inputs banks are using to come up with their risk weights. A lot of people outside the banks are watching, not least banking rivals in the US and regulators from outside Europe.
