Memo to Oz-owned banks: resistance on IRB models is futile

OPINION: Memo to Oz-owned banks: resistance on IRB models is futile

The big four Australian-owned banks ought to have realised in February that it was futile to resist the Reserve Bank's intention to severely restrict the competitive advantage they gain from using their own internal models to calculate capital.

Perhaps they thought the subject of IRB, or internal ratings-based, models is so technical and esoteric that the average New Zealander would never understand that the big four banks have enjoyed a significant cost advantage since 2008 when the models were introduced.

Since then, the smaller, mostly New Zealand-owned banks have been forced to compete while straight-jacketed by having to use standardised models which require much higher levels of capital.

Then in February, RBNZ spelt out just how great the IRB advantage is.

It revealed that New Zealand's largest bank, ANZ Bank, needs to hold slightly more than half the capital that the government-owned Kiwibank needs to back each $100 of mortgage lending.

Kiwibank is New Zealand's fifth-largest bank but its $22.7 billion in total assets at June 30 is dwarfed by ANZ's $165 billion.

It beggars belief that ANZ's mortgage book is only half as risky as Kiwibank's and that's clearly the RBNZ's view because in May it ordered ANZ to increase its capital levels against both housing and farm loans.

It's equally as clear that the RBNZ regarded the measure as urgent because it wasn't content to wait until making the final decisions coming out of its bank capital review.

Sputtering about the RBNZ's role being the prudential regulator, not the competition regulator, as Westpac did, was never going to cut it when the numbers showed such an uneven competitive playing field stemming from past RBNZ decisions.

RBNZ governor Adrian Orr was already worried about how aggressively the big four banks have been using their IRB models to drive down how much capital they need.

As early as May last year, just a couple of months after his appointment, Orr expressed his discomfort about the models' real-world impacts.

The central bank had asked the big four to estimate the risks of the same hypothetical dairy loan portfolio and the outcome was an enormous 40-percentage point variance between the highest and lowest ratings.

By July last year, Orr had made up his mind that the 'big four' would have to report both the outcomes of their IRB models and what the outcomes would be using the standardised models that other banks have to use.

And he went even further in mid-December last year when the RBNZ released its proposed changes to bank capital rules.

Not only would the big four have to near double the minimum amount of equity they hold to 16 percent of risk-weighted assets, but their advantage from using IRB models would be reduced to no more than 90 percent of the outcome using standardised models.

While the smaller banks will have to continue using standardised models, their minimum capital was to be raised to a slightly lower 15 percent.

And, as per the earlier decision, the big four will also have to report both their IRB outcomes and what the outcomes would be using standardised models.

We'll have to wait until the final decisions, expected in early December, to see whether the RBNZ sticks to its guns, but the indications are that any moderation is unlikely to extend to the IRB models.

RBNZ's views on IRB models were strongly endorsed by the three independent academics that assessed its proposals.

David Miles from London's Imperial College noted that "many central banks and other supervisors share the uneasiness about the IRB approach, which seem to generate some surprisingly low assessments of risk-weighted assets for some banks."

Ross Levine of the University of California, Berkeley said: "I am wary of using IRB models to calculate RWAs because (a) the IRB banks have strong incentives to manipulate their models to reduce RWAs for regulatory purposes, (b) international evidence suggests that banks use their IRB models to generate lower RWAs, and (c) the RBNZ does not have all of the data and programs to vet fully each bank’s IRB mode."

And James Cummings of Macquarie University in Australia noted the "empirical evidence about the extent of inconsistent modelling practices between banks" and suggested that dual reporting of both the IRB and standardised outcomes would help fill a gap in the international literature.

But, as the Irish say, God loves a trier, because the Australian-owned banks have by no means accepted that they're on a losing wicket.

Westpac, for example, said the RBNZ was “creating some simplistic public support" with the IRB aspects of its capital proposals.

ANZ tried to argue that the outcomes from its IRB models "would be more conservative than the standardised model in an extreme downturn" while Commonwealth Bank of Australia-owned ASB Bank argued for a lower floor.

National Australia Bank-owned Bank of New Zealand argued that IRB models reduce the chances of a bank crisis and, moreover, the RBNZ's approach is out of step with the Australian Prudential Regulation Authority's approach.

It's worth remembering the RBNZ only reluctantly agreed to let the big four use IRB models and that Orr was deputy governor with responsibility for banking policy and regulation at around the time that decision was made – he left to head the New Zealand Superannuation Fund in 2007.

Despite the central bank's deep-seated preference for a one-size-fits-all approach, it was swayed by the fact that the Australian parents had been given permission to use IRB models and they didn't want to be forced to treat their New Zealand operations differently.

The real-world outcomes demonstrate the RBNZ's reluctance was warranted.

ends

© Scoop Media