Systemic Risk in Australia's Major Banks

1. Systemic Risk in Australias Major Banks the impact of deposit insurance and capital raisings. Xiping Li, Chris Malone, David Tripe

2. Motivations The banking system has become safer and stronger since the financial crisis. Certain activities that people didnt like are no longer being done. James Dimon, Chairman, CEO, JPMorgan Chase the fundamental risks remained and the efforts of regulators and politicians were simply rearranging the deckchairs on the Titanic. Baily and Elliott, 2013, Brookings Institute There is very limited research on how a banks financial and market data may signal information about a banks risk exposure during a financial crisis Calluzzo & Dong, 2015, p. 235, JBF

3. Introduction Like the banking systems of many other countries around the world, the Australian banking system is quite concentrated, and dominated by a small number of banks ANZ, CBA, NAB, WBC. The business of the banks outside Australia means their combined size relative to their home market is very large. This level of concentration engenders the risk that if any of the banks got into difficulty it would be costly to rescue, and would cause widespread disruption to the broader economy. These effects would be exacerbated if the failure of one bank was a trigger for the failure of others contagion, systemic risk. Hence, there is a motivation in being able to measure systemic risk.

4. Measuring Tail Risk The traditional approach to measure an institutions downside risk lies in a sensitivity analysis of its balance sheet (BS) to key risk factors. J.P Morgan refined the idea into BS based Value-at-Risk (VaR) in the 1990s. Basel II (2004) recommended VaR for use in bank risk management . Given the prescience and poignancy of financial markets, Stock market based VaR is now commonly used. VaR is the dominant form of risk measurement in the financial sector Acharya et al. (2010) Other market based measures of risk include CAPMs -risk and Expected Shortfall (ES) both of which we look at. Basle III (2013) recommend the Expected Shortfall (ES) method (VaR extension).

5. Measuring Risk Using the stock market to measure bank risk has issues. During the financial crisis (FC) the market couldnt easily distinguish the good from the bad (i.e Akerloff, 1970). While all bank stocks plummeted during the FC, the variation in the Balance Sheet performance amongst these banks during the crisis was substantial. Some banks in the US and Europe obviously had BS failure (AIG, RBS, Lehman Bros and Bear Sterns being notable) But most banks did not. Australias four majors capitalisations dropped $115 billion (to $144.4 billion) at the depths, while BS losses turned out to be about $20 billion. Although things could have turned out differently under alternate policy settings, so who knows.

6. Measuring Risk The point is that risk has three components: unique firm risk, everyday systematic market risk, and rare systemic risk. Trying to disentangle them is keeping researchers busy. Bisias et al. (2012) in their review of the risk measurement literature outline 31 different quantitative measures of systemic risk. There is considerable confusion about terms. Consider this NASDAQ definition: Risk is either unique or systemic. Unique risk is exposure to a particular company, and is sometimes referred to as firm-specific risk. Systemic risk (or systematic risk) is the risk of a portfolio after all unique risk has been diversified away. Teoh & Roberts conclude (2015, thesis): Systematic risk is the variation due to movements in common risk factors. Systemic risk is the contagion spillover of extreme risk caused by a trigger event.

7. Research Questions 1. Is the Australian economy more exposed to the four major banks performance now compared to what it was 20 years ago? 2. How effective was a) beta-risk, b)VaR-risk, c) ES-risk, and d) VIX-risk in identifying the FC? 3. What was the risk response to the Deposit Guarantee (DWFG) policy? 4. What was the risk response to the Capital Raisings by the banks?

8. Reference Studies Key reference studies: VaR and ES: Artzner, Delbaen, Eber, and Heath (1997, Risk) Acharya et al (2010, WP) Tail-betas: De Jonghe (2010, J Fi Intermediation) Bollen, Skully, Tripe & Wei (2015, Int Rev Fin) Two factor CAPM: Agarwal and Naik (2004, RFS) FF3 + Put option premiums, Knaup and Wagner (2012, JFSR) MM + OTM Put option premiums VIX index: Index of implied volatility on 30-days to expiration at-the-money put and call options. The investor fear gauge, Whaley (2000, JPM)

9. Q1. Relative size of the sector Market Share and HHI

10. Q1 Relative size of the sector cont. Table 2 - Banking sector size over time Year 4-majors cap to market cap % 4-majors TA/GDP 1992 0.218 0.837 1995 0.246 0.834 2000 0.271 1.164 2005 0.245 1.258 2010 0.240 1.805 2014 0.273 1.946 This table provides statistics on two ratios: bank capitalisation to stock market capitalisation and total assets to nominal GDP. The ratios are based on the four major banks. Increase in relative size Reasons M&A, expansion in business.

11. Data Datastream: daily total return data for the four major banks bank and market capitalisations Australian All Ordinaries total return index September 13 1991 (when CBA listed) until October 15 2014 Bloomberg: daily data for the S&P ASX 200 VIX index, from 3 Jan. 2008 Annual reports

12. Regression Windows 1. From 17 Sept. 1991to obtain pre-estimation period statistics 2. From 3 January 2004 was a relatively stable time for the markets and we use this window to obtain risk estimates before the FC developed. 3. From 7 th September, 2006, Noriel Roubini, in an IMF presentation, warned that sub- prime defaults would become a crisis. 4. From January 21, 2008. Brunnermeier (2009) - key date when the major credit rating agencies downgraded monoline insurers. World equity markets fell and, in response, the Fed conducted its first emergency cut since 1982. 5. From 12 October, 2008. DWFG and CR. Deposit and Wholesale Funding Guarantee scheme. Bollen et al. (2014) linked it to a reduction in beta-risks. Bank Capital Raising (CR). 6. From Dec. 2009. The date when the four majors had finished their CRs. Each bank engaged in at least three CRs during the GFC period, including both IPPs and SPPs. For these events we examine the abnormal return when the shares were allocated to shareholders.

14. Augmented CAPM Augmented Systemic Risk measures Models

15. Results--Capital Raisings Table - Capital Raising event wealth effects Variable Parameter Estimate t-value Pr > |t| Intercept - period 1991 - 2004 0 0.0002 3.22 *** R M (All Ords) 0 1.0514 138.95 *** CR1 anounced 1 0.0015 0.94 CR1 closed 2 0.0033 1.47 CR2 anounced 3 -0.0006 -0.45 CR2 closed 4 0.0084 3.90 *** CR3 anounced 5 0.0001 0.07 CR3 closed 6 0.0036 1.51 Adj R-square 0.4492 Observations 23720 CRs consisted of IPP, SPP, Convertible Preference Shares and PERL issues. CRs commenced on 28 August 2008 and were completed by 21 December 2009

16. Results

17. Conclusions R M (beta-risk) and % VaR (or % VIX ) (v-risk) are so highly correlated that including the two measures in the same regression negates the power of systemic risk measures. Models without systematic risk included as a variable highlight the usefulness of VAR and VIX as systemic risk indicators. % VIX is expected to have a positive sign but the sign flips during systemic risk events. Represents an obvious issue for the Black Scholes OPM. Volatility can be bad! R M (beta-risk) is surprisingly useful in indicating systemic risk episodes. Although it is a lagged indicator compared to fast responding VIX and VAR measures.

18. Conclusions Capital Raising activities by the banks were associated with significant reductions in market and systemic risk in the banks. This indicates that concerns about capital adequacy were very real. The deposit guarantee scheme (DWFG) was associated with a significant reduction in market and systemic risk in the banks. This indicates that concerns about liquidity were very real. While the market apparently over-reacted to the FC by plunging bank values by as much as $115 billion, when losses were just a fraction of this amount, it appears that regulator actions (i.e. DWFG) and bank decision making (i.e. CRs) mitigated what could have been much worse.

19. We appreciate your support THANKS

20. 4.2 Did VaR risk effectively identify the GFC event? VaR analysis- 180-day rolling window estimates of the 1% VaR percentile. 20 Readily apparent is how VaR reveals the GFC systemic event between Jan 2008 and mid 2009. The four majors portfolio has similar VaR characteristics to the lowest risk bank over the majority of the period of coverage diversification effect.

21. 180-day rolling daily 1% VaRs 1991-2014 for the four majors portfolio and for the All Ords index 21 The banks generally carry more VaR risk than the broad market

22. 12 consecutive VaR down-steps after Nov 2007 22 The portfolio VaR measure is revised downwards 12 times in a row, from -2% to -7%. Chance of 12 downgrades in a row: 0.5 12 ? 1 in 4096

23. Did beta-risk effectively identify the GFC? 23 180-day rolling beta-risk

24. Did volatility-risk effectively identify the GFC? 24 When the VIX index went above 30 on January 22 nd , 2008, it was well into the upper quartile of observations.

25. References JPMorgan Chase Fails to Impress Wall Street by Maureen Farrell, CNN Money, April 12, 2013 . (Retrieved from http://money.cnn.com/2013/04/12/investing/jpmorgan-earnings/index.html ). 25