Box C: Results of the 2015 common scenario ICAAP stress test
Stress tests play two important roles in the Reserve Bank’s prudential framework. First, stress tests help the Reserve Bank identify and assess risks to the financial system. Second, stress tests are used to identify and manage risks to individual institutions’ capital and liquidity buffers. Individual institutions are expected to use stress tests to assess the viability of current business plans, including as part of their Internal Capital Adequacy Assessment Process (ICAAP). Stress test outcomes are also an important input into supervisory discussions with participating institutions.
In late 2015, the four largest banks in New Zealand participated in a common scenario ICAAP test. This test was a hybrid between an internal test (conducted regularly with each institution choosing their own scenarios) and a regulator-led stress test (occurring every 2-3 years with common scenarios and assumptions). Due to the use of a common scenario across banks, the results of the test provided insights for the financial system as a whole. However, the test featured less standardisation of methodology than a full regulator-led exercise. For example, there was no ‘phase 2’ where loss rates were standardised.
As with previous regulator-led tests, the stress scenario was a severe macroeconomic downturn. Over a three-year period, real GDP fell by 6 percent, unemployment rose to 13 percent, and dairy incomes remained at low levels. Residential property prices fell by 40 percent (with a more severe fall of 55 percent assumed for Auckland); and both commercial and rural property values fell by 40 percent. Finally, the 90-day interest rate fell by about 3 percentage points due to monetary policy easing, although banks typically assumed a partially offsetting increase in funding spreads above risk-free rates. Banks were asked to simulate the impact of this scenario on loan portfolio performance, and to trace through the implications for their balance sheet and profit and loss statement.
Banks reported a steady increase in bad debt expenses throughout the scenario, as would be expected under a severe economic contraction. This reflected a combination of deteriorating credit quality (increasing collective provisions) and rising defaults (increasing specific provisions). The cumulative hit to profits averaged around 4 percent of initial assets (figure C1), which is a similar outcome to phase 2 of the full regulator-led exercise conducted in late 2014.1 About 30 percent of total losses were related to mortgage lending, with half of this due to the Auckland property market. SME and rural lending accounted for most of the remainder of financial system losses. Loss rates for mortgage lending were around 2 percent, significantly lower than the 5 percent loss rate observed for most other sectors.2
Figure C1: Cumulative bad debt expense (% of initial assets)
Note: Other includes consumer lending, lending to financial institutions, and holdings of financial securities.
The underlying profitability of banks – earnings from core activities, prior to accounting for bad debt expenses – is a first line of defence against rising loan losses. On average, banks assumed a moderate decline in net interest margins during the scenario, reflecting rising defaults reducing interest income and the assumed increase in funding spreads. However, in line with the experience of the GFC, banks expect to be able to maintain net interest margins at around 2 percent by eventually passing on higher funding spreads to customers (rather than reducing mortgage rates by the same amount as the OCR). As a consequence, underlying earnings during the scenario were of a similar magnitude to reported credit losses, so that return on assets averaged around zero.
Although projected credit losses were largely absorbed with underlying profitability, capital ratios were expected to decline throughout the scenario. This reflected an increase in the average risk weight from around 50 to 70 percent, due to negative ratings migrations (rising probability of borrower defaults) and falling collateral values (rising losses given default). Although remaining well above the regulatory minimum, the average Common Equity Tier 1 (CET1) capital ratio declined from 10.3 to 8 percent as a result. The total capital ratio came under more pressure, due to smaller initial buffers to the regulatory minimum. As a result, the average bank reported falling into the upper end of the capital conservation buffer in the final year of the test, which would trigger restrictions on dividend payments to shareholders (figure C2). The results of the test suggest that individual institutions would remain well away from the point of economic failure. However, the results also suggest that the financial system would be far from fully functioning in a way that would support a swift economic recovery. The combined impact of bank responses to the scenario appears to risk a material worsening in the economic downturn, or to push the boundaries of realism:
- Deleveraging: Banks assumed that they could reduce credit exposures by around 11 percent (amounting to a decline of around 8 percent of nominal GDP over the scenario). While this could reflect a reduction in customer demand, there is a risk that such a sharp decline could be associated with a range of feedback effects not captured in the tests. For example, the associated increase in stressed sales and tightening in origination standards could further reduce prices and liquidity in property markets. Alternatively, less aggressive deleveraging would tend to lower reported capital ratios.
- Increased reliance on retail deposits: All banks assumed that they could achieve an increase in the proportion of funding sourced from retail deposits, with the average retail deposit share increasing from 62 to 70 percent. While this can be rationalised as a flight to safety in a crisis, a more limited availability of retail deposits is possible. For example, the scenario could generate concern among depositors about bank safety, or there could be more significant reductions in deposits associated with weak new lending activity. These alternative assumptions would increase the risk of a tightening in credit supply in the event that banks’ ability to raise wholesale funding was impaired.
- Issuance of Tier 2 capital: All banks assumed that the market environment would allow them to issue new Tier 2 capital instruments, both to replace expiring instruments and to increase the total amount outstanding. In the scenario, this may not be plausible, particularly with most banks operating within their capital conservation buffer.
Figure C2: Capital ratios relative to respective minimum requirements (% of risk-weighted assets)
These issues highlight the significant uncertainties involved in assessing the implications of the stress scenario for the financial system as a whole. A careful consideration of recent stress test results will be an input into the upcoming review of bank capital requirements.
2 Relatively low loss rates on mortgages are common in international crises, and the Reserve Bank considers them credible in this scenario. Because homeowners are personally liable for mortgages and lose their own home when in default, they generally keep servicing their mortgages if they can, even if they are in negative equity.