My last post on Anglo-Irish Bank got me thinking some more about the question of trust and banking. The subject came up in conversations I recently had with senior bankers – BNP Paribas co-chief operating officer Philippe Bordenave and Deutsche Bank CFO Stefan Krause. I asked them about the declining trust in bank risk models, a factor behind moves to impose leverage ratios by regulators.

The subject also came up in a conversation with the treasurer of one of Europe’s biggest non-financial companies. He told me that he no longer deposited cash with European banks out of concern that wholesale deposits would be ‘bailed-in’ if a bank went bust. Instead, he used a portfolio of funds to deposit the company’s cash ““ an example of shadow banking.

I also received a comment on my Anglo-Irish post from Dan Davies, a financial sector analyst at Exane BNP Paribas. Although he agreed with much of my post, Davies took me to task for what he felt was an overly cynical conclusion about banks and bankers in general. He argued that banks were a lot safer today than ‘adrenaline junkie’ press coverage suggested, and that the majority of ‘law-abiding and sensible’ bankers had no incentive to fake the inputs of their risk models.

To this I’d say that I don’t think that using a risk model to allocate capital to a portfolio of loans is necessarily a bad thing. If a bank has better-than-random skill at determining the default risk of borrowers, then shareholders might expect to see the bank using that judgement when it allocates the capital they provide. I think that’s what Davies meant when he said that fiddling the inputs of such risk models was akin to cheating at patience (See Note 1).

However, it’s also a fact that risk models for trading books and securitised assets were heavily ‘gamed’ before the financial crisis with catastrophic consequences. And in countries like Spain or Ireland, risk models for loan portfolios failed to account for real estate bubbles.

While I don’t think that either Bordenave or Krause can deny such problems with models, both bankers resist the suggestion that risk models are untrustworthy in principle.

Bordenave says the problem was that Basel rules were misapplied because lax supervisors didn’t check bank models in the way they were supposed to.

As a counter-example, Bordenave points to BNP’s French regulator, which he says often rejected his bank’s internal models (forcing BNP to use more capital-heavy standardised models instead), delayed their implementation or demanded tweaks.

In other words, if regulators treated banks sceptically from the beginning, assuming their models were unreliable until proved otherwise, then banks that passed inspection ought to be solvent and the Basel risk-weighting system could be trusted to do its job. Don’t trust me, trust the regulators who supervise me is Bordenave’s message.

But who checks that the regulators are doing the checking? What happens with those regulators for whom checking the models was “too difficult”, in Bordenave’s words?

This brings us back to the leverage ratio debate. Whether it’s an untrustworthy bank or captured, lazy regulator, failures in models are amplified by leverage. Benchmarking studies carried out by the Basel Committee show that the current ten percent ratio of equity to risk-weighted assets could easily disappear at a bank whose aggressive modelling choices for lending, trading and operational risks were suddenly made more conservative.

When bankers such as Deutsche’s Anshu Jain decry leverage ratios as “simplistic” compared with Basel risk models, they are dodging this issue (Note 2). The fact that some of Europe’s biggest companies privately say that they plan to avoid banks as much as possible is evidence that the issue is important.

Ideally, banks should have a higher proportion of equity in their capital mix than they do now. Then, risk modelling choices would be the problem of bank shareholders alone, helped by beady-eyed analysts such as Davies. Failing that, limits on leverage ratios, which throw out risk models and limit the absolute size of bank balance sheets compared to equity, are the next best thing .

Leverage ratios would also go some way to solving the problem of mistrust of banks and lazy or inattentive regulators, restoring confidence among corporate treasurers that the banks need for their funding and derivatives businesses. Maybe that’s why Krause made his comment to me, heralding a retreat from Jain’s position. Perhaps  he understood that if bankers want to keep their risk models, then leverage ratios may be the best way of convincing others that they can do it safely.

Note (1) In support of Davies, a recent Basel Committee study shows that a sample of banks computed higher internal risk weights for sovereigns on average than standardized models which permit OECD countries, including Spain and Italy, to be risk-weighted at zero.

Note (2) In his speech, Jain implies that the old regulatory capital arbitrages of the late 1990s and early 2000s might return if banks were constrained by leverage ratios. This is an empty threat, since regulators can simply ban such practices if they want to.