Who built Mega Bank?

Recently I read an article which tried to excuse the media for missing the GFC based on the premise that no one is really interested in this type of analysis or would believe it if they read a decent analysis of the risks of catastrophic events occurring. Instead of giving up on my controversial topic, this made me think about, why?  My conclusion is that controversial views on the risks within a system must be continually debated and the deep detail outlined.

So, to continue my previous work, this piece is a laypersons take on a very complicated area of financial regulation that highlights why Australia’s major banks have very thin capital buffers on residential mortgages. My strong belief is that this system is set up to fail or be rescued. To understand the detail may give clues and confidence in recognizing when this will occur.

As previously outlined, Australia’s big four banks, which I call “Mega Bank” are allowed to use an Internal Risk Based approach under Basel II to calculate risk weighted assets and minimum capital requirements. APRA follows the Basel II standards and regulates the approach under Prudential Standard 113 (APS 113).  APS 113 covers the full IRB approach to credit risk for all bank assets. My analysis today concentrates on residential mortgages as these make up the bulk of Mega Bank assets and have by far  the lowest capital requirements. Readers who want to go the extra mile should read the referenced document in full.

As a general overview APS 113 is a disjointed document and lacks accountability on whom or which organization is actually responsible for managing the risk of unforeseen consequences. It provides a formulaic approach to calculations without clear definitions and with no process for interpretation or responsibility for a process to ensure a reasonable approach to interpretation of definitions, statistics or external risks occurs. The almost 100 pages of APS 113 are more appropriate as a regulator’s guidance notes then a Prudential Standard. Far too much of, “ this is the way it should be done”: and far too little of “who’s responsible for getting it right or balanced”. This makes the standard perfectly gameable by banks, who are purely motivated by minimizing capital requirements and increasing returns on capital. Regulators actually creating moral hazard is the result with the risk born by the taxpayer.

My next main criticism is the fundamental flaw in the whole IRB approach to residential mortgages. Basel II and APS 113 treat all retail loans in almost  the same manner. That is residential mortgages, car loans, personal loans and credit cards are to use the same basic methodology for determining minmum capital requirements. I believe  this is a huge issue as all retail IRB approaches are based around historical loan performance and normal distributions. The difference with residential mortgages, besides being very long term, is that the provision of increasing credit availability and lowering underwriting standards can and does influence the value of the collateral supporting the loan. This distorts and decreases the expected loss statistics over” short” periods of time  for residential mortgages, whereas for other retail assets the opposite probably occurs. Neither APS113 nor the Basel standards acknowledge the possibility of this occurring, why would Mega Bank?

To state simply, the methodology used under APS 113 to calculate minimum capital requirements for residential mortgages encourages a bank to increase risk through the provision of increasing credit and lowering underwriting standards because the formulas will actually calculate low risk capital requirements whilst significantly increasing future systemic risk. The more a bank lends on housing the lower the risk calculated and the higher the return on capital. I think this is perverse, many others clearly think otherwise.

Now to the detail. The following is the actual formula that APS113 requires to be used:

  • where PD is the probability of default
  • LGD is the loss given default
  • R is the correlation, given as 0.15
  • N is the standard normal cumulative distribution function and
  • G is the inverse standard normal cumulative distribution function.

PD and LGD are taken to be the banks expected (ie unstressed) loss parameters, based on historical performance.  Their product is then the bank’s Expected Loss (EL).

The effect of the formula (making plenty of assumptions) is to attempt to insulate the bank against 99.9% of loss scenarios, and to then reduce this by the EL which is required to be covered by provisions. The formula is not incorrect and used appropriately would give a balanced conservative risk based approach, but it is gamed and here’s how.

APS 113 stipulates that Mega Bank must use at least 5 years performance data to establish PD. As if 5 years data is going to give the ability to create a normal distribution from the performance of loans during that period. As the bank’s only reliable data metric is LVR then that is used to map performance. Unless the bank’s other assumption is that house prices will fall, then the data will give a PD that is highly skewed to the left of a normal distribution and cannot be relied upon to be normal over changing economic conditions or decreasing house prices.

Simply, the methodology of using a normal distribution can only be valid if enough performance data is collected over a long enough time period in contrasting economic circumstances.

Mega Bank uses the formula and APS 113 to its advantage to minimise capital, all with the approval of APRA by using short term performance during rising house prices as normal. Surprisingly there is also no definition of “default” which leaves the whole  PD assessment open to gaming. Do mortgagors that qualify for the Borrower’s Assist program, fall within the definition of default? Readers may have their own views that Mega Bank and APRA make sufficient adjustments for the above issues to not put the bank or taxpayers at risk, or that under Basel III sufficient adjustment will be made to remove any systemic risk (see below).

The facts, however, speak for themselves. Megabank has less than 2% capital against the total residential mortgage book including mortgage insurance, with little capital adjustment in operating risk calculations to increase that amount of capital overall. Mega Bank with the approval of APRA is incentivized to write more and more residential mortgages in preference to all other loan classes. Whilst this is really a diabolical situation, I believe  APRA  staff do their best with what they have. Mega Bank follows the rules as set by Basel II and APRA standards and has massive resources to support their case. APRA do not have the resources to compete with Mega Bank. However, the RBA does.

Under the proposed  Basel III reforms, APRA will introduce a counter cyclical buffer of capital. This counter cyclical buffer would seem to be a system strengthening measure  that will be universally adopted across the globe. Although I agree with a carefully used counter cyclical buffer, I do not think it’s a solution to the issues concerning residential mortgages in this country. The base assumption of the counter cyclical buffer is that the IRB approach under Basel II is being used and applied appropriately. It’s not meant as a solution to inadequate methodology. However, the Basel III reforms for determining the counter cyclical buffer do suggest an interesting process, as follows:

APRA proposes to introduce the Basel III countercyclical buffer in its prudential capital regime. Broadly speaking, the application of the buffer would have the following main elements:

  • APRA will continuously review the need for the countercyclical buffer, in consultation with the Reserve Bank of Australia (RBA);
  • in addition to macroeconomic indicators of excessive credit growth such as the credit-to-gross domestic product (GDP) gap28, APRA’s review will be informed by input from the supervisory visits it conducts

Basel III reforms propose putting the RBA squarely at the front of controlling credit growth and systemic risk by being able to provide economic scenarios which add to Mega Bank’s capital requirements. Basel III reforms are to apply from January 2013, but what was stopping the RBA in assisting APRA with the Basel II advanced methodologies being proposed by Mega Bank? The RBA boffins would have more than enough ability to assist APRA in applying data performance adjustments which take account of enough cycles to determine a more risk averse PD over time.

The power of APRA and the RBA working together to ensure systemic risk doesn’t sit on Mega Bank’s balance sheet or in the hands of taxpayers is built into the system already. Although Basel III reforms will ensure that there is a focus on this power, its to our great detriment that it has not been used.


  1. “My strong belief is that this system is set up to fail or be rescued”

    To be rescued as the banks have the govt in their back pockets and they know it. They can take all the risks they want.

  2. It’s funny you say not many predicted the GFC, I got information from all different streams – Denning, Emery, Kiosakyi e.t.c. and I personally knew about a dozen businessman who sold all property & stocks prior to 2008. So the information was their. I think that what has confounded most is that the true effects of the GFC have been frozen or the ‘can kicked down the road’. If the crash had been allowed to take it’s full course then it would have cleansed the market of all the dross – yes, I know that the pain would have been enormous – but it will have to happen at some stage, why not sooner, rather than later with much greater pain? The same dozen businessmen now do not know what to make of the current situation – they got back into property & stocks and made killings…..but the uncertainty remains and they have less knowledge about a repeat GFC; but believe that it will take place, they just do not know when.

  3. A pretty good summary of what is wrong with the IRB approach IMO. I personally dont put any faith in the models that are being used; given how they are constructed they must understate the PD by a significant amount.

    The real issue I see around adjusting a PD is justifying it. I dont believe you could justify a significant enough adjustment to PD based on Aussie housing data to get the models somewhere close to reasonable. Overseas data is equally hard to justify since we are “different” here.

    The pillar 2 adjustment compensates a little but again how do you apply that correctly if you dont know what the model would output with significantly higher PD assumptions?

    Though i dont think a normal distribution will adequately capture risk no matter what data you use; to do that you need to rely on a scenario testing approach which overlays the model (this is somewhat done in the ICAAP, could be better). A VaR approach will only ever give you a minimum value for tail risk exposure (why we have seen some pretty spectacular blow-ups from the use of VaR).

    Once you adopt Basel rules you are pretty hamstrung though; you need to go by what the Basel committees decide. It would be nice if the banking standards were standalone, in a similar fashion to the insurance capital standards.

    • DT,

      Another fine contribution from you. Thank you.

      Crocodile Chuck,

      I read that Edge piece from Taleb which you linked. Very interesting. Thank you.

      This passage leapt off the screen at me:

      Clearly, with current International Monetary Fund estimates of the costs of the 2007-2008 subprime crisis, the banking system seems to have lost more on risk taking (from the failures of quantitative risk management) than every penny banks ever earned taking risks. But it was easy to see from the past that the pilot did not have the qualifications to fly the plane and was using the wrong navigation tools: The same happened in 1983 with money center banks losing cumulatively every penny ever made, and in 1991-1992 when the Savings and Loans industry became history.

      It prompted me to wonder if it is possible to make a sustainable profit from the existing banking model. Could it be that the certainty of loss simply grows over time until it becomes inevitable?

  4. “the media for missing the GFC based on the premise that no one is really interested in this type of analysis”

    Written by a banker or one of our more notable MSM so called guru’s. It’s disgraceful.

    However there were a few who did predict the GFC are were ridiculed.

    Didn’t the RBA/APRA argued against BIII, and I’d love to see the minutes of those meetings?

    I’ve just been reading the FED’s (BB) lecture notes from yesterday, and it looks like the FED can do whatever it wants so at the end of the day do the CB’s really need to follow BIII?

    Thanks for another top post.

  5. As per other posters here, thanks for the interesting post. What I find curious is the fact that there’s a rapidly diminishing source of those expotential profits for the megabank conglomerate. The last 15 years saw profits skyrocket on the back of jumbo mortgages and increasing fees to retail customers – it’s really hard to imagine where the new growth areas for the Big 4 are going to come from in the coming years to sustain their greedy billion plus profits. I think consumers have been gouged about as much as they possibly can be…however I could be wrong about that one… perhaps they could reintroduce the debtors prison or the poor house rather than let people go bankrupt?

    • Jumping jack flash

      That’s an excellent point.

      Extending that thought, I wonder what the rating agencies will do if any of the megabank conglomerate were to post a less than spectacular profit?

      • innocent bystander

        I think they will go hard wealth management – what with compulsory going form 9% to 12%

  6. And the probability of mortgage default used in the calculations is about 1% (source: http://www.rba.gov.au/publications/bulletin/2010/sep/6.html ). US experience would show that in the event of relatively high LTVs (> 90%), and high loan to income (>4.5), that the PD in these cases will be more like 0.13 (source: http://scholar.harvard.edu/campbell/files/mortdefault24112010.pdf, table 3). Plug PD=0.13 into the equation and see whether 2% capital buffer is adequate for Megabank! Oops, sorry I forgot. It’s different here.

    • Well, I think this is the problem. The PD are based on “historical data”, where the PD has been low. But as we all know, this figure will significantly spike in the next financial crisis. What they should be imposing is the usage of canned scenarios, which takes into account a stress situations. They have plenty of real life examples such as Ireland/US etc This should provide a “truer” reflection of PD and impact on the mortgage Book / capital etc.

      This type of stress testing should be done at least on a monthly basis in order banks to understand the delta of their capital requirements in the event of a downturn. They should make provisions/preparations as opposed to relying on a begging bowl from tax payers as and when the proverbial hits the fan.

      There really are no excuses. Once again our regulators are asleep at the wheels.

      • All the banks do internal stress testing on a regular basis as part of their economic capital and other risk management work.

  7. Excellent post.

    On Basel iii, I think it is too little and too late. Systemic risk & accounting control fraud has already been built into the existing edifice by IBG YBG (*) banksters.

    I don’t have too much confidence in the RBA/APRA or it’s staff, in doing anything right by the taxpayer. They are merely running a protection racket for private banks by way of invisipower.

    (*) I’ll Be Gone. You’ll Be Gone.

    • To be fair Mav, if you’ve dealt with APRA you would have the impression that they take their jobs seriously and are doing the best they can.

      I think the Invisipower thing is somewhat unfair when applied to a regulator. Some regulators in the US had to make their reports on companies public and it just led to rounds of negotiation to ensure that the company wasnt too unhappy with what was produced. This included companies that were later bailed out.

      If you have a regulator that cannot disclose it means that companies cant hide behind commercial-in-confidence and other such arguments, and the regulator has better access to information. I prefer that to a situation where the regulator is just about as in the dark as we are.

      • There is an important disclosure issue here. As I’ve pointed out previously Mega Bank do not meet their Pillar 3 disclosre requirements on what its IRB methodologies are. This is not APRA disclosing anything but the bank not follwoing the rules and those rules not being enforced, all on the basis of commercial in confidence.

        Its a sad state of affairs when how a bank games the system to minmise capital at the risk of the taxpayer is commercial in confidence

        • A good example of Regulatory forbearance – They have even invented a term to describe regulators not even enforcing EXISTING regulations.

        • Deep T, if we can use CBA as an example what havent they disclosed in their June 2011 disclosure that you would expect them to disclose?


          I agree that there are some things that could be described more clearly but to me this disclosure would meet the requirements of APS 330.

          You wont be getting enough information to replicate an IRB model, but there is enough there understand the approach as well as the mode outputs.

          • We’ll disagree on this. Enough disclosure should be made so that an analyst can replicate the calculation. How else could anyone understand the risk, not just the approach?

            The issue is exacerbated by the TBTF nature of Mega Bank. Its not good enough for APRA to simply say, “everything is under control”, when every taxpayer is carrying the risk and the facts clearly indicate that the risk game being played is a very dangerous one

          • Deep T,

            I would be very interested to hear your thoughts on that quote from Nassim Taleb above and my musing about the viability of banking per se.

  8. Interesting article. It’s a good point about using 5 years data. This was a wider problem for a lot of VAR and capital work in the run up to the GFC, leading to a general underestimate of tail risk masked by beautiful mathematical models.

    However, even using 40 or 50 years of Australia residential mortgage loss data would, I suspect not give a very different answer as they have been pretty trivial over this period (especially when clawbacks from mortgage insurers are allowed for, as they should be).

    Remember that securitisation was originally driven by Basel 1 which set a 4% capital requirement for residential mortgages. This was perceived (and was generally)to be much higher than the real risks. This lead to non-bank investors buying mortgages to get a better return and banks being relieved of excessive capital requirements.

    In my view this broke down as the securitisation industry commoditised and volumes ramped up. Investors stopped doing proper due diligence which allowed mortgage sellers to write any old crap mortgages. Fannie and Freddie provided support and one in the chain appeared to have much incentive to check things properly.

    In my view the purchasers of the securitised mortgages are at least as much to blame for the whole procss as greedy banks.

    It should also be noted that APRA does pretty stringent stress testing of banks and other financial institutions and their capital buffers have to withstand these too.

    Setting regulatory capital requirements is a difficult art. If they are too high compared to actual costs then they just drive activity out of the regulated sphere in shadow institutions.

    • As per above post, why not used real life “canned” data emanating from Ireland/US mortgages and its impact on mortage books/capital requiremnts etc to reflect what may actually happen to PD in the next financial crisis. Some people would say, its not relevant as Ireland/US is not aussie data, but globally we are so interconnected these days.

      I know that the FSA (english regulator) has mandated this (canned scenarios emanating from Lehman days) for VaR calculations of banks that it is supervising.

      To me see it seems reasonable

      • Stress testing is what you are thinking of. Personally I find stress testing and stochastic asset-liability modelling more useful in managing institutional financial risk than arbitrary capital ratios which generally do not fit the risk profile of an institution very well.

        The APRA stress testing uses something like a 25% peak to trough fall in residential house prices and 45% in commercial property and unemployment increasing to over 10%. In real terms (ie post inflation)the falls are around 30% and 50% respectively.

        Details of APRA stress testing can be found at


        These stress scenarios showed a reduction in tier 1 capital of about 1% with no mitigating actions (ie unchanged dividends and no repricing).

        Despite what’s said here Australian banks are finacially strong and resilient.

        The senior people I know at APRA are smart, experienced and well plugged into events around the world (they can also be very conservative and pig headed on some technical issues).

  9. Many years ago, in a place far far away, Australian banks had savings bank subsidiaries that kept residential mortgages apart from their trading banks. They were separately capitalized and scrutinized to prevent cross-contamination, at least in theory.

    We now have four parallel debt-manufacturing bureaucracies determied to maximize market share in all financial ‘products’ nation-wide.

    The macrobiz commentariat cannot simply wish away this anti-competitive oligarchy, but please, please be present when the RBA is recapitalizing and stripping bad loans from these financial parasites to protect innocent taxpayers from bearing the full cost and more.

    Australia needs and we must demand a genuinely competitive banking system.

  10. Totally agree David, unfortunatly Australia is going to relearn the lessons of the 1890’s all over again.

    The CBA that was established in 1913 did not lend for housing until the 1946.
    Just as a coincidence Australian housing did not recover until about 1950 from the effects of the 1890’2 depression.

    We need a new CBA trading bank so that capital can be directed in to productive income producing assets for the nation

  11. The other issue about who built Mega bank. I dont like to say it but i also think the compulsory super system has played its part in this.

    THe original intent by Keating was to create a pool of savings that could be used in the countries development and lessen the reliance on overseas debt. The funds invested into the bank shares push up values and allow greater leverage.

    • Note 20E of the Future Fund show as at 2010 they had approx 22B in domestic banks, and it would be good to know what the numbers are for the banks.

      • Probably a better way of putting it is that those super savings instead of being used for productive purposes have found there way into the non virtuous circle of the banking system. Certainly 2 billion a month into the super system is supporting both asset values and providing a source of funds for the mega bank.

  12. Diogenes the CynicMEMBER

    Great article.

    Yes the system will explode on current course projections.

    My view is somewhat Austrian here if we had a lot of small banks with no explicit govt backing, strict legislated caps on LVRs (60-70% max), nasty jail terms for bank CEOs when banks failed and access to take back previous bank bonuses, market audits by regulators at any time, no investment banking and retail lending together in one institution, investment banks are partnerships not incorporated we could eliminate a lot of this capital ratio stuff that few even in the industry have a good understanding on. Tier 1 should be 10% in gold/cash at Central bank not in the hands of the institution. Market scrutiny should take care of the rest and if a bank fails no problem it will not take down the entire payments system with it.

    Of course that would mean Megabank and its ridiculous oversized sector profits vis-a-vis the real economy would not be possible. Good. Perhaps they would then lend to small business who actually create jobs.

    • Agreed.

      Most rooves held up by four pillars will collapse if just one of the four pillars topples. The concept that it creates stability is garbage, what it fundamentally creates is the TBTF paradigm.

  13. A very interesting article and a few points
    -The pillar 3 disclosures are actually better than most overseas banks
    -the deduction from Tier 1/2 capital of the difference between expected losses and loan loss provision modifies some of the low PDs/LGDs in the models
    -interestingly NAB has a capital allocation against mortgages that is 50% higher than the other majors. Same book I would have thought.
    -The really interesting point is how affected by leverage and procyclicality it is. If changes in PD/LGD are linear rather than proportional, the low capital charge will increase materially ie a move in the RWA/Assets from 15% to 25% will increase WBCs capital requirements by $3b . Banks will be forced to be increasing capital at times when losses are increasing
    -Not sure how it is with the majors but APRA really do get stuck into smaller banks under Pillar 2. The effective capital held by even the larger regionals is 2-3 times the majors for similar credit risk. Of course diversification from size solves many problems but not by that much