Skip to main content

The Australian banking system under stress – again?

Thursday 8 November 2012

John Laker, Chairman - AB+F Randstad Leaders Lecture 2012, Brisbane

I am pleased to have this opportunity to take part again in the AB+F Leaders Lecture series, this time in Brisbane.

Two years ago, in front of a different audience, I posed a question.[1] Would the Australian banking system cope if it were to be confronted with much greater adversity than it had weathered to that point in the global financial crisis? As I said then, for our banking institutions the crisis had not been the ‘near death’ experience that many banks abroad had faced, and to which some succumbed. I answered my question by releasing the results of a macroeconomic stress test that APRA had conducted, built on the hypothetical scenario of a substantial economic downturn in Australia driven by a marked slowdown in China.

The stress test results were reassuring, and reinforced our confidence in the resilience of the Australian banking system.

Two years on, the crisis is still with us. Global market confidence may have been boosted recently by additional monetary policy measures in major advanced economies, but fundamental economic concerns remain. Tangible progress in the resolution of the eurozone’s sovereign debt and banking problems has been slow to materialise, the United States is approaching a ‘fiscal cliff’ and the trajectory of the US and Chinese economies is unclear. The IMF has recently concluded that downside risks to global economic forecasts have increased and are considerable. The Australian economy has continued to perform strongly but it, too, is facing a softer growth outlook.

What more opportune time to pose the question, again. Would the Australian banking system cope if the global and domestic economic environment were to turn much gloomier? The answer, again, can be found in the results of another macroeconomic stress test that APRA has recently completed. And this time the hypothetical was tougher than our 2010 test — it involved a much sharper slowdown in China and a disorderly resolution of eurozone problems leading to a freeze in global funding markets.

Exercises of this type confirm the growing importance of stress testing in the toolkit of prudential supervisors. Admittedly, the technique has had something of a chequered history. Stress tests undertaken by global banks, supervisory authorities and international organisations before the crisis largely failed to send the right warnings. No more so than in Iceland, where stress tests confirmed the solvency of the Icelandic banking system not long before it collapsed. Since the crisis began, a substantial intellectual, IT and resource effort has been made to strengthen the rigour and the credibility of stress tests. And we now see highly publicised stress tests being employed in a number of major jurisdictions to help restore market confidence in individual banks and banking systems.

My talk today focuses on two particular dimensions of stress testing. The first is the use of stress testing by regulated institutions to identify vulnerabilities in their operations or solvency. My remarks will address:

  • the benefits and challenges of stress testing as a risk management tool; and
  • some ‘best practice’ aspirations for regulated institutions in further developing their stress-testing capabilities.

The second dimension is the use of supervisor-led stress tests to analyse system-wide vulnerabilities. To put our own work into an international context, I will provide a brief overview of recent industry-wide or macroeconomic stress tests conducted in the United States, Europe and Spain.

And then the punchline — the results of APRA’s recent macroeconomic stress test!

Stress testing as a risk management tool

Stress testing is a quantitative ‘what if’ exercise aimed at assessing vulnerabilities and resilience in the face of ‘severe but plausible’ shocks, to use the accepted parlance. That is, severe enough to be meaningful yet plausible enough to be taken seriously. At the level of a regulated institution, stress testing can be used to estimate the impact of shocks on cash flows, profits and capital positions.

APRA views stress testing as a critical risk management tool for regulated institutions and we have embedded this view in the recent strengthening of our capital adequacy framework. New prudential standards are now in place for authorised deposit-taking institutions (ADIs), implementing the Basel III reforms, and for life and general insurers.

Central to this framework are expanded requirements for capital planning and management in institutions, which are encapsulated in the ‘Internal Capital Adequacy Assessment Process’ or ICAAP. An ICAAP involves an integrated and documented approach to risk management and capital management. Properly developed, an ICAAP should assess the level of, and appetite for, risk in the institution and establish the level and quality of capital that is appropriate to support that risk profile.

The ICAAP must be approved by the board and must include a strategy for ensuring adequate capital is maintained over time. The ICAAP must also include stress testing and scenario analysis. To assist institutions in developing their ICAAPS, APRA has released for consultation a draft prudential practice guide on this topic.[2] As we make clear in the guidance, stress testing is a core part of the ICAAP process and can be used in the formulation of capital targets and trigger levels.

Capital targets should be in line with the board’s risk appetite and should ensure that the institution has a sufficient capital buffer to be able to absorb losses in stressed conditions, without breaching minimum prudential requirements. Capital adequacy cannot be demonstrated by ‘just meeting’ prudential requirements; institutions need to hold an adequate buffer above those requirements to minimise the risk of a breach in normal and stressed circumstances.

In its ICAAP, an institution should use its own stress tests, as well as any led by APRA, to help set appropriate capital targets and buffers. There should be a clear explanation for the board, and supervisors, of how stress tests have been factored into capital planning. Of course, stress testing is not limited to assessing capital adequacy — it also has a wider role. Its value can and should stretch far beyond showing that capital levels are sufficient. It can be applied at an enterprise-wide level, portfolio-specific or risk-specific level to inform risk appetite, set risk limits, provide early warning indicators and identify potential mitigating actions.

For APRA supervisors, the results of stress tests are used to anchor expectations for the level of capital that an institution should hold in normal times, to provide a sufficient buffer to withstand a challenging environment. The results are also used to inform our risk assessment of institutions and as part of the development of supervisory action plans. Where issues or concerns are identified, supervisors follow-up to ensure rectification action is taken. This can include steps to strengthen capital positions but is not limited to capital initiatives.

Stress testing is not a panacea. Stress testing is often described as an art, not a science. It is the combination of statistical modelling and expert judgement, based necessarily on simplifying assumptions. Hence, stress testing should not be relied upon in isolation and without recognition of the margin for error. It is part of the risk manager’s and supervisor’s toolkit, but not the only tool. It is a component of, but not a replacement for, robust and comprehensive risk management and supervision. To quote a recent International Monetary Fund (IMF) paper on stress-testing principles and practice:

‘No matter how much a stress tester tries, stress tests always have margins of error. Their results will almost always turn out to be optimistic or pessimistic ex post. In addition, there will always be model risk, imperfect data access, or underestimation of the severity of the shock. One should therefore set stress test results in a broader context.’[3]

This is not to dismiss stress testing as a critical tool because it is not exact. No forward-looking analysis can be. As Keynes is reputed to have said, it is better to be roughly right than precisely wrong. But to be most effective, stress testing in an institution should be used in concert with other forms of financial analysis, informed discussion and risk assessment.

Some ‘best practices’ in stress testing

In 2009, the Basel Committee on Banking Supervision reviewed the performance of stress testing practices before and during the crisis and was, frankly, less than impressed. To encourage global banking institutions to lift their game, the Basel Committee released its Principles for sound stress testing practices and supervision.[4] APRA requested that the five Australian banks using the ‘advanced’ Basel II approaches (‘advanced banks’) conduct a self-assessment of their compliance against these Principles, and I discussed the learnings in my earlier speech. Over 2011, a working group of the Basel Committee undertook a peer review of the implementation of the Principles by supervisory authorities, via an off-site survey.[5] A senior APRA executive coordinated this survey. The review found the Principles to be generally effective, although countries were at various stages of maturity in implementing them.

While not a primary focus of the review, many countries provided views on areas for improvement in stress-testing practices in banking institutions. I would rather turn the language around and describe these areas in terms of ‘best practice’ aspirations. The aspirations are equally relevant to ADIs and to insurers.

Five key areas of ‘best practice’ can be highlighted. Firstly, stress test results should be integrated into decision-making within the institution. In APRA’s view, this is not only obvious but fundamental, and the theme is enshrined in our ICAAP requirements. Stress testing should directly inform business and strategic decisions and, where necessary, trigger action. It should be an integral part of risk management, assisting in understanding the institution’s risk profile and testing its risk appetite, not seen as some sort of regulatory compliance exercise. The board should be clear on how the results of internal and supervisor-led stress tests have been factored into risk settings, such as the planning of appropriate capital buffers. In our view, this has not always been the case and we will be closely monitoring the use of stress tests in our ongoing reviews of capital plans and capital management.

Secondly, and notwithstanding the important role of stress testing, boards and senior management, as well as other users, need to have a strong understanding of the limitations, assumptions and uncertainties associated with stress tests.

APRA expects that enterprise-wide stress test results would be routinely reported to board risk committees, and be challenged by them. If the challenges are to be effective, there needs to be an adequate level of detail in the stress test results reported and an appropriate coverage of critical assumptions. Documentation should include, for example, the key macroeconomic parameters of the stress test scenarios, the impact on capital and profitability, and discussion of key modelling assumptions. Sensitivity analysis can help to highlight particular central assumptions or specific vulnerabilities, and communicate the range of uncertainty around the results. The limitations of models should also be recognised, with a significant additional margin for uncertainty built into the analysis of stress test results.

Where mitigating actions are factored in, stress test results should be presented both before and after such actions. This enables a clear view on the ‘worst case’ outcome and the value of proposed management responses. Relevant feedback from APRA on internal and supervisor-led stress tests should also be reported to and considered by the board, and issues addressed where relevant.

Thirdly, stress tests need to confront the institution with realistic adversity. The 'what if’, particularly the underlying economic scenario, must be demanding. Common scenarios set by APRA are one way of guaranteeing severity and they enable peer comparisons to be easily drawn. But institutions should also routinely design scenarios that are particularly relevant to their specific risk profile, updated and refreshed as that profile shifts over time.

The global financial crisis has redefined the ‘severe but plausible’ benchmark. It has shown that the Golden Decade that preceded the crisis is no guide to plausibility; the scale of recent financial system shocks in many countries is now a better guide. Institutions should not be boxed in by history in designing suitably challenging stress tests.

This is a particularly important issue for regulated institutions in Australia, given the good performance of the Australian economy over a long period, including through most of the crisis. It may be tempting for these institutions to succumb to what has been described as ‘disaster myopia’ — that is, the difficulty of imagining appropriately severe economic conditions after a long period of stability. The lack of severe stress experience can lead to reluctance by institutions to contemplate their own mortality and a willingness to dismiss as implausible scenarios that would drive financial losses. Scenarios built on benign experience will under-estimate potential stress and provide false confidence.

A couple of years ago, a senior Bank of England official aptly highlighted the constraints of history. He commented that the worst GDP growth outcome in the United Kingdom over the preceding 25-year period leading up to 2007 was a decline of 1.4 per cent.[6] In Australia, the equivalent worst case GDP outcome for a 1-in-25 year stress, based on the last 25-year period, would be even less than this. This is in stark contrast with the experience of major economies in the global financial crisis, where the decline in GDP for the G7 countries in 2009 averaged around four per cent. Great care is necessary, therefore, in drawing conclusions about longer-term risks based on recent history. As with other parts of stress testing, scenario development is as much about judgement as it is statistics.

There is also a secondary but equally critical challenge for regulated institutions in Australia after a long period of relative stability: viz., a lack of meaningful historical data in downturn economic conditions on which to base stress-testing models. Institutions have to look beyond recent history and beyond domestic precedent. Beware a failure of imagination!

The fourth area of best practice relates to data and IT infrastructure. Stress tests are a technical challenge, requiring co-ordination across a range of different teams within an institution. Risk, finance, treasury and strategy teams all need to be engaged. Enterprise-wide stress test exercises will typically rely on a number of different models, projecting forward the balance sheet, profitability and capital position at a granular level.

To give you a sense of this, APRA’s macroeconomic stress test this year involved financial and regulatory capital information from the five advanced banks covering nine different credit portfolios and many thousands of cells of data — and they are just the outputs from the process. Naturally, the reliability of the data is paramount. As stress testing becomes a regular and routine process, IT systems need to be sufficiently flexible and integrated to ensure information can be aggregated across the institution, and produce quick and reliable results. Aggregation is very important to providing a complete picture of the stress test. There needs to be appropriate checks and oversight to ensure the results are accurate and intuitive.

The final area of ‘best practice’ relates to modelling issues. Translating the scenarios into stressed losses is often seen as a black box but it is the critical part of the process. Weaknesses in modelling reduce the credibility of the exercise, but these weaknesses can often be hidden beneath the surface of the results. APRA supervisors, supported by our risk specialists, will be spending more time understanding institutions’ stress-testing models, peering into the black box to better understand the methodologies and assumptions it contains.

In the recent APRA macroeconomic stress test, the wide range of estimates from institutions’ own stress test models indicated that there is still some way to go in developing the required technology and capabilities in this area. For example, the default rates projected on lower-rated corporate exposures varied from 50 per cent to 90 per cent over the stress test period, contributing to a wide range of loss outcomes. This was a consequence of different modelling practices rather than differences in underlying risk. Given the use of a common scenario, similarities in the composition of loan books and relative credit quality, it would be reasonable to expect a much narrower range in default rates than the individual models implied. This underlines the importance both of internal governance of the stress-testing process and of supervisory challenge. Before APRA finalised its stress test results, the banks involved were asked to apply common APRA-determined credit risk factors to estimate loan losses. This helped to remove, or at least quieten, some modelling noise from the process.

The modelling challenge is not restricted to forecasting loan losses, although this is where stress-testing techniques are now being applied more intensely. Projecting changes in interest income, trading income and funding costs is also important. Differences in the ability to generate income in stressed conditions to offset loan losses can be a key driver of stressed losses across different institutions. Over-optimistic or unrealistic forecasts for stressed income can lead to inflated estimates of capital. Caution is required here. The assumed ratings migration of assets in stressed conditions is another topic where careful analysis is required.

The five areas of ‘best practice’ that I have outlined provide a high-level ‘to do’ list for our regulated institutions. In each area, of course, there is considerable detailed work to be done. In reviewing progress, APRA will be looking for a combination of investment in the supporting infrastructure, and effective oversight, judgment and challenge.

International context

Often building on exercises within regulated institutions, stress testing by supervisory authorities is gaining greater prominence as a tool for analysing system-wide risks. These stress tests are sometimes described as macroprudential stress tests but I have used the more familiar ‘macroeconomic’ term. These sorts of stress tests have become much more rigorous than some Pollyannaish pre-crisis tests and, in some major jurisdictions, the results of industry-wide stress tests have been eagerly anticipated by policymakers, analysts and investors as confirmation (or otherwise) of bank soundness.

Not surprisingly, stress testing has become a key feature of supervisory activity in Europe, where the European Banking Authority (EBA) conducts EU-wide stress tests. Its 2011 stress test was coordinated on an impressive scale, with 90 banks involved in 21 different countries. A common scenario was used to enable benchmarking and peer comparisons, supported by a common methodology and underpinning assumptions. The stress test was praised for its risk coverage and detailed disclosure of results, but also criticised for its treatment of sovereign debt exposures and the muted severity of the common scenario.[7] Following the publication of the stress-test results, the EBA issued a recommendation that national supervisory authorities should ensure that capital positions were strengthened at specific banks where shortfalls or concerns were identified. Later in 2011, as the eurozone sovereign debt crisis escalated, the EBA followed up with the EU ‘capital exercise’, in which it recommended that all large European banks build a temporary capital buffer to reach a nine per cent Core Tier 1 ratio.

Earlier this year, the US Federal Reserve conducted stress tests involving 19 large US bank holding companies as part of its Comprehensive Capital Analysis and Review (CCAR) process. Since the unprecedented Supervisory Capital Assessment Program (SCAP) in 2009, stress testing of major US banks has been a regular and critical part of the US supervisory process; it is central to the US Federal Reserve’s assessment of capital adequacy and its evaluation of major US banks’ proposals to make capital distributions. In the 2012 CCAR exercise, the aggregate Tier 1 common ratio was assessed to be higher in the stress test after hypothetical losses than actual capital levels held in 2008, testament to the steps taken to strengthen capital positions of US banks since the crisis began.

The most recent major stress test was that of the Spanish banking system in mid 2012, conducted on behalf of the authorities by independent consultants. This was a thorough and very detailed evaluation of the risks and capital needs of the main banking groups in Spain. The stress test program combined top-down model-based approaches with a forensic bottom-up bank-by-bank assessment. Given the importance of the exercise, it was in effect a full financial audit, involving the analysis of some 36 million loans and 8 million guarantees. Data quality was verified by more than 400 auditors. The process identified capital shortfalls at several banks under both baseline and adverse scenarios; in the adverse scenario, additional capital needs were estimated at around €60 billion for the system overall.

These industry-wide stress tests share several key characteristics. They are supervisor-led exercises carried out at a granular level of detail in the context of significant uncertainty, at a time when the domestic banking systems involved have been undergoing substantial restructuring. They are what have been termed ‘crisis management stress tests’, with a very distinct purpose in mind: to assess whether key banking institutions need to be recapitalised and, if so, by how much. As such, they focus on a critical capital ratio benchmark to assess capital adequacy, backed by recapitalisation plans where needed. Given their objective and context, public disclosure of the stress tests has been extensive. In addition to aggregate results, detailed results have been provided for individual banks and for specific outcomes, such as loss rates by portfolio.

The context for macroeconomic stress testing in Australia is a different one. The economic environment stands in marked contrast to that of the United States and Europe and the Australian banking system is not facing fundamental restructuring or needing reinforcement. Like many other prudential supervisors, APRA does not publish its stress test results for individual institutions. All this, however, does not make stress testing any less important. Indeed, as the Basel Committee noted in its Principles for sound stress testing practices and supervision:

“Stress testing is especially important after long periods of benign economic and financial conditions, when fading memory of negative conditions can lead to complacency and the underpricing of risk.”[8]

Stress testing by APRA

Let me turn now to the results of the macroeconomic stress test that APRA conducted this year for the five advanced banks. These banks account for around 80 per cent of total banking system assets in Australia.

The three-year macroeconomic scenario for the stress test was developed in conjunction with the Reserve Bank of Australia and the Reserve Bank of New Zealand. As with the 2010 exercise, coverage of the stress test extended to the New Zealand operations of the major banks.

The ‘what if’ scenario was built around a further deterioration of global economic conditions, with a disorderly resolution of the fiscal problems in Europe triggering a dislocation in global debt markets and a sharp downturn in the North Atlantic economies. China is assumed to be unable to fully offset the decline in its exports with domestic spending and, as a result, the rate of growth of the Chinese economy slows sharply. The implied reduction in Chinese demand for minerals lowers commodity prices significantly, with a consequent deterioration in the exchange rate for the Australian dollar. Domestically, households and businesses respond to the external shock by reducing consumption and investment expenditure. As a result, GDP falls and unemployment rises substantially, which feeds back into rising defaults and sharp falls in house prices and commercial property prices.

In this scenario, the key macroeconomic parameters for Australia used as the basis for the stress test were:

  • a sharp (5 per cent) contraction in real GDP in the first year;
  • a rapid rise in the unemployment rate to a peak of 12 per cent;
  • a peak-to-trough fall in house prices of 35 per cent; and
  • a fall in commercial property prices of 40 per cent.

This is a tougher stress test than the one APRA undertook in 2010. The projected economic contraction is deeper and more prolonged, with a weaker recovery and a longer period before return to growth. The rise in unemployment is higher and the impact on the housing market therefore more pronounced; there is a greater peak-to-trough fall in house prices. This time, the stress test also addressed liquidity consequences. The dislocation in global debt markets results in the largest banks being unable to access global funding markets for six months. The consequence is more intense competition for deposit funding and an increase in funding costs, weighing on lending margins and acting as a drag on revenues.

Remember, this is a hypothetical. It is in no way a forecast or a central expectation for the course of the Australian economy. Rather, the stress test was intended to test the boundaries of ‘severe but plausible’, especially given the current relatively strong position of the Australian economy. Benchmarked against recent industry-wide stress tests in other countries, the severity is confirmed by the fact that the GDP shock is more than four standard deviations based on the annual volatility of GDP in Australia since 1960; the shock was one-to-three standard deviations in other major tests. As a test of plausibility, the macroeconomic scenario would be comparable with the actual experience of the United Kingdom, United States and some European countries during the global financial crisis.

Although the macroeconomic scenario was tougher than in the 2010 exercise, the actual mechanics of the stress test were largely the same. The advanced banks were asked to apply the macroeconomic scenario in their own models and provide their assessment, in quite granular detail, of the impact on the ratings migration of assets, default behaviour, profitability and capital. After analysing this information, APRA then determined a common set of portfolio-specific risk measures that were applied to the banks’ loan portfolios.

Reflecting the severity of the scenario, the advanced banks all reported significant losses, driven by much higher bad debt expenses. Credit loss rates in aggregate were comparable with the experience in the early 1990s, although not quite as high as the peaks then reached. As expected, total losses were larger than in the 2010 exercise.

Despite the deterioration in labour market conditions and the projected stress on the housing market, residential mortgages, which account for nearly half of the advanced banks’ credit exposures, contributed only a fifth of total losses. The mortgage portfolio alone was not the principal driver of losses, a reflection of the structure of the domestic mortgage market as well as the general tightening in lending standards following the crisis. Losses were realised across a range of loan portfolios, particularly corporate, SME and commercial property portfolios. Losses on these business portfolios were more frontloaded, materialising earlier in the scenario than losses on residential mortgage portfolios, which tended to lag the increase in unemployment.

The main results of the stress test for the five advanced banks, taken as a group, are as follows:

  • none of the banks would have failed under the downturn macroecnomic scenario;
  • none of the banks would have breached the four per cent minimum Tier 1 capital requirement of the Basel II Framework in any year of the stress test; and
  • the weighted average reduction in Tier 1 capital ratios over the three-year stress period was 3.8 percentage points.

This is a very positive result. It reflects the efforts of the advanced banks to strengthen their Tier 1 capital positions since the crisis began through ordinary equity issues and profit retention. It leaves these banks well positioned to transition to the new Basel III capital regime.

The weighted average reduction in Total Capital ratios over the three-year stress period was 4.1 percentage points. This took the five advanced banks, as a group, marginally below the eight per cent minimum Total Capital requirement of the Basel II Framework by the end of the stress period. This result was not unexpected and nor does it raise undue concern on our part. These banks had been running down their Tier 2 capital levels ahead of finalisation of the Basel III capital standards, which impose stricter eligibility criteria for Tier 2 capital instruments. APRA has now released these eligibility criteria in final form and APRA supervisors will be monitoring plans to replenish Tier 2 capital holdings.

The reduction in capital ratios in the stress test is calculated gross, before consideration of any mitigating actions that management could take to respond to the stressed conditions. This enables APRA and the boards concerned to understand the ‘raw’ outcome and ensures that stress test results are not clouded by actions that may or may not be achievable, depending on market conditions.

There were a range of mitigating actions proposed by the banks. These included repricing of risk, tighter underwriting standards, cost-cutting, cutbacks in the provision of credit and the issuance of new capital. Some of these actions would have second-round effects on economic activity, although a further iterative phase was not factored into the stress test. After allowing for mitigating actions, the weighted average capital position of the advanced banks as a group returned to pre-stress levels. APRA has provided feedback to the banks concerned where it felt there was an element of over-optimism built into some of the proposed mitigating actions.

The liquidity consequences of the freeze in global funding markets did not, in the stress test, prove to be systemically challenging. Net interest margins narrowed before partially recovering towards the end of the stress period and this, combined with subdued credit demand, dampened income growth. Liquidity was also put under pressure. However, cash outflows from wholesale funding maturities were largely offset by growth in domestic deposits, while collateral flows associated with the projected weakening in the Australian dollar also alleviated funding pressure. The stress test set a benchmark that there be no reliance on the Reserve Bank of Australia for ongoing liquidity support beyond its current repo arrangements (including self-securitisations) by the end of the stress period. That benchmark was met.

This result is testament, in part, to the stronger funding positions that the advanced banks have adopted since the crisis. However, it is also an illustration of the difficulty in testing both capital and liquidity through a single macroeconomic scenario. Liquidity shocks tend to be most concerning when prompted by a deposit run, a relatively short sharp crisis event often measured in days. Shocks to solvency can build up over a longer horizon. Nonetheless, an integrated and comprehensive approach to stress testing that covers liquidity and solvency impacts remains an important objective for supervisor-led stress tests.

APRA’s macroeconomic stress test was a ‘bottom up’ stress test. In other words, the exercise was implemented by the individual banks involved using their advanced modelling and internal data, but based on a common set of risk estimates determined by APRA. A bottom-up approach builds on granular detail, particularly on credit risk exposures to which loss outcomes can be very sensitive. It can also consider institution-specific management responses to scenarios. However, it is a very resource-intensive exercise and the results can be affected by differences in modelling practices between institutions.

A ‘top-down’ stress test, in contrast, is one implemented fully by a supervisory authority on the basis of a uniform methodology and consistent modelling. Top-down stress testing is more flexible and enables greater scope for sensitivity analysis, alternative scenarios and different assumptions on risk. It also avoids variations in modelling across institutions. However, the approach lacks richness of detail and the results are contingent on a single model, or suite of models, and the data set that supports them.

A sensible approach, where it can be done, is to run both a bottom-up and a top-down stress test in parallel, providing validation and analysis through the process. This is, indeed, what has happened in Australia. As part of its 2012 Financial Sector Assessment Program (FSAP) review of Australia, the IMF conducted a systematic top-down stress test for the five advanced banks, based on the macroeconomic scenario used by APRA. The results will be published by the IMF shortly. Suffice to say that, despite the differences in approach, the IMF’s results are consistent with the more granular approach of APRA’s stress test. This is a reassuring cross-check of our conclusions.

Concluding comments

APRA’s recent macroeconomic stress test, together with the IMF’s assessment, provides further confirmation that the Australian banking system has the capital strength to cope not just with the real-world stress test of the crisis, now stretching beyond five years, but with much greater adversity. This was APRA’s view in my first speech on stress testing. It remains APRA’s view today.

However, nothing stands still. Risk profiles of regulated institutions are dynamic, not static. Management strategies and risk appetites constantly evolve. The operating environment will always be clouded with uncertainties. Perhaps the only certainly in this whole area is that the hypothetical shocks in stress testing exercises will not be the shocks that materialise in fact!

In my first speech, I emphasised that both APRA and our regulated institutions are on a journey to improve stress-testing capabilities. That journey continues. APRA is expanding its commitment to, and resourcing for, stress testing as a core area of frontline supervision. Though there is still much work to be done, we welcome improvements made by institutions to advance their stress-testing programs and their plans for further investment. We also welcome the greater engagement of boards and senior management. That engagement is a key element in the development and use of stress testing as an integrated part of risk management.

And, be assured, APRA will return to the question in the title of my speech, again and again!



  1. J.F. Laker, ‘The Australian Banking System Under Stress?’, Address to Australian Business Economists, Sydney, 9 June 2010.
  2. Draft Prudential Practice Guide CPG 110, Internal Capital Adequacy Assessment Process and supervisory review, APRA, September 2012.
  3. International Monetary Fund, Macrofinancial Stress Testing – Principles and Practices, August 2012.
  4. Basel Committee on Banking Supervision, Principles for sound stress testing practices and supervision, Bank for International Settlements, May 2009.
  5. Basel Committee on Banking Supervision, Peer review of supervisory authorities’ implementation of stress testing principles, Bank for International Settlements, April 2012.
  6. Andrew G Haldane, Why Banks Failed the Stress Test, Bank of England, 13 February 2009.
  7. International Monetary Fund, Macrofinancial Stress Testing – Principles and Practices, August 2012.
  8. Basel Committee on Banking Supervision, Principles for sound stress testing practices and supervision, Bank for International Settlements, May 2009.

The Australian Prudential Regulation Authority (APRA) is the prudential regulator of the financial services industry. It oversees banks, credit unions, building societies, general insurance and reinsurance companies, life insurance, private health insurers, friendly societies, and most members of the superannuation industry. APRA currently supervises institutions holding $6 trillion in assets for Australian depositors, policyholders and superannuation fund members.