Region

Jurisdiction

Regulator

2022-05-17

KPMG Examination of Reserve Uncertainty for ACS NZ at 31 December 2011

KPMG Actuarial Pty Ltd, acting at the direction of the Reserve Bank of New Zealand, examined the outstanding claims and claims handling expense reserve uncertainty for ACS (NZ) Limited as of 31 December 2011. The report reveals extremely high uncertainty in reserves, with KPMG's central estimates significantly exceeding ACS's own provisions and projecting that the Lyttelton earthquake event will exhaust all available reinsurance, including the Adverse Development Cover. Key risks include substantial potential shortfalls in reinsurance recoveries due to exchange rate fluctuations and commercial disputes, alongside significant uncertainty regarding the final resolution of specific coverage and policy wording issues.

New Zealand

Reserve Bank of New Zealand

KPMG Actuarial Pty Ltd ABN: 91 144 686 046 Australian Financial Services Licence No. 392050 10 Shelley Street Sydney NSW 2000 PO Box H67 Australia Square NSW 1215 Australia Telephone: +61 2 9335 7000 Facsimile: +61 2 9335 7001 DX: 1056 Sydney www.kpmg.com.au

© 2012 KPMG, an Australian partnership and a member firm of the KPMG network of independent member firms affiliated with KPMG International Cooperative (“KPMG International”), a Swiss entity. All rights reserved. Printed in Australia. KPMG and the KPMG logo are registered trademarks of KPMG International. 29 May 2012 Mr Andrew Moon Mr Toby Fiennes ACS (NZ) Limited Head of Prudential Supervision Level 12 Reserve Bank of New Zealand Ansvar House 2 The Terrace 432 St. Kilda Road, Wellington 6011 Melbourne VIC 3004 New Zealand EXAMINATION OF RESERVE UNCERTAINTY AT 31 DECEMBER 2011 Please find enclosed my report in respect of the examination of reserve uncertainty for the outstanding claims and claims handling expense liabilities of ACS (NZ) Limited (“ACS”) as at 31 December 2011. I have prepared central estimates and estimates at the 75% and 90% Probability of Sufficiency (“PoS”). Yours sincerely Jefferson Gibbs Executive, KPMG Actuarial Pty Ltd Fellow of the Institute of Actuaries of Australia Fellow of the New Zealand Society of Actuaries

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 i Table of Contents Executive Summary ..................................................................................................... iii 1 Terms of reference and limitations ............................................................... 1 1.1 Terms of reference .......................................................................................... 1 1.2 Items out of scope ........................................................................................... 1 1.3 Prudential and professional standards ............................................................ 3 1.4 Distribution and use ........................................................................................ 4 1.5 Reliances ........................................................................................................ 5 1.6 Inherent uncertainty ........................................................................................ 5 1.7 Emerging information ...................................................................................... 6 1.8 Limitations ....................................................................................................... 7 2 Nature of business ......................................................................................... 8 2.1 Overview of ACS ............................................................................................. 8 2.2 Other material issues relevant to the ACS earthquake liabilities .................... 9 3 Data ................................................................................................................ 12 3.1 Data supplied ................................................................................................ 12 4 Approach ....................................................................................................... 13 4.1 Outstanding claims liability ............................................................................ 13 4.2 Analysis Segments ....................................................................................... 13 4.3 Currency ....................................................................................................... 13 4.4 Non–reinsurance recoveries ......................................................................... 14 4.5 Discounting ................................................................................................... 14 5 Earthquake liabilities .................................................................................... 15 5.1 Overview ....................................................................................................... 15 5.2 Reinsurance .................................................................................................. 16 5.3 ACS methodology and results ...................................................................... 22 5.4 KPMG methodology and central estimate results ......................................... 26 5.5 Scenarios and discussion on individual large claims .................................... 35 5.6 Sensitivity testing on the Lyttelton event ....................................................... 46 5.7 Risk margins ................................................................................................. 47 5.8 Earthquake results ........................................................................................ 52 5.9 Uncertainties ................................................................................................. 52 6 Non earthquake liabilities ............................................................................ 55 6.1 Overview ....................................................................................................... 55 6.2 Approach ....................................................................................................... 56 6.3 Risk margins at 75% PoS ............................................................................. 57 6.4 Risk margins at 90% PoS and uncertainty .................................................... 59 6.5 Non earthquake provisions ........................................................................... 61 7 Claims Handling Expenses .......................................................................... 62 7.1 Overview ....................................................................................................... 62 7.2 Approach ....................................................................................................... 62 7.3 Key assumptions ........................................................................................... 62

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 ii 7.4 Comparison to alternate estimates ............................................................... 64 7.5 Risk margins ................................................................................................. 64 7.6 Uncertainties ................................................................................................. 65 8 Aggregate results ......................................................................................... 67 8.1 Outstanding claims provision ........................................................................ 67 Appendices A Data ................................................................................................................ 69 B Lyttelton earthquake incurred cost development...................................... 72 C General discussion on uncertainty – Earthquakes ................................... 74 D Actuarial techniques for non earthquake ................................................... 82 E Rate of currency exchange (net impacts on Lyttelton earthquake loss) . 84

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 iii Executive Summary Important Note: Basis of report This report has been prepared by KPMG Actuarial Pty Ltd (“KPMG”) at the request of, and for the benefit of, ACS, its management and directors and the Reserve Bank of New Zealand. This report is not intended to be used for any other purpose and may not be suitable, and should not be used, for any other purpose. Opinions and estimates contained in the report constitute our judgment as of the date of the report. The information contained in this report is of a general nature and is not intended to address the objectives, financial situation or needs of any particular policyholder, other individual or any other entity. The report is provided for information purposes only and does not constitute, nor should it be regarded in any manner whatsoever, as advice and is not intended to influence a person in making a decision in relation to any financial product or an interest in a financial product. No one should act on the information contained in this report without obtaining appropriate professional advice after a thorough examination of the accuracy and appropriateness of the information contained in this report having regard to their objectives, financial situation and needs. In preparing the report, KPMG has relied on information supplied to it from various sources and has assumed that the information is accurate and complete in all material respects. KPMG has not independently verified the accuracy or completeness of the data and information used for this report. Except insofar as liability under statute cannot be excluded, KPMG, its executives, directors, employees and agents will not be held liable for any loss or damage of any kind arising as a consequence of any use of the report or purported reliance on the report including any errors in, or omissions from, the valuation models. The report must be read in its entirety. Individual sections of the report, including the Executive Summary, could be misleading if considered in isolation. In particular, the opinions expressed in the report are based on a number of assumptions and qualifications which are set out in the full report To the extent permitted by law, neither KPMG nor its Executives, directors nor employees will be responsible to any third parties for the consequences of any actions they take based upon the opinions expressed with this report, including any use of or purported reliance upon this report not contemplated in our scope as set out in Section

Any reliance placed is that party’s sole responsibility. Where distribution of this report is permitted by KPMG, the report may only be distributed in its entirety. Judgments about the conclusions and comments drawn from this report should only be made after considering the report in its entirety and with necessary consultation with KPMG.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 iv Our report includes comparisons to results and financial figures from earlier analysis as at 31 December 2011 completed by ACS and its Appointed Actuary. We note that this report is not a review of those results. The basis of our analysis is different to this earlier analysis in many respects, for example what to allow for within claims handling expenses, has referenced data that was not available at the time or includes development to a later date and provides results at a higher probability of sufficiency that were not considered in earlier analysis by ACS or its Appointed Actuary. We have had the “benefit of hindsight” in setting figures as at 31 December 2011. We have not reviewed what reserves may have been reasonable to establish in the formal reporting of the ACS position as at 31 December 2011 and the reader of this report should not infer any opinion on that topic as we note such comment would require a different scope – in particular it would exclude consideration of any information not available or reasonably available at the time of the earlier analysis. The Earthquakes present substantial complexity. The range of possible final results is wide. The final outcome may fall above our 90% Probability of Sufficiency or may fall below the central estimate provided. The results of this report should not be interpreted as providing the range of possible outcomes. Over the coming months new information will emerge and work will be completed that may have a material bearing on the valuation of the ACS liabilities. We therefore note that future valuations may show materially different results to those tabled within this report. Please note this report does not include any legal, accounting or taxation opinion or advice. As set out in the scope of our engagement, noting the substantial uncertainty in liabilities at this time, our report does not provide any guarantee of the current, ongoing or final solvency position of ACS. Solvency is a matter for the recipients of our report to consider in light of available information.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 v Introduction and scope ACS decided to implement a managed withdrawal from New Zealand following the Canterbury Earthquakes. ACS (NZ) Limited (“ACS”), formerly Ansvar Insurance Limited (New Zealand), has proposed a Contingent Scheme of Arrangement (“the Scheme”). A review of the Scheme is outside the scope of this report. As part of the Scheme, ACS, at the direction of the Reserve Bank of New Zealand (“RBNZ”) under a Section 126 Notice, have requested KPMG Actuarial Pty Ltd (“KPMG”) to provide an examination of outstanding claims and claims handling expense reserve uncertainty as at 31 December 2011. The purpose of this report is limited by the terms of reference discussed in Section 1.1, and as set out in our letter of engagement, and should not be used for any other purpose. Key observations Our key findings are as follows:  Our results show an extremely high level of uncertainty in reserves at this time at both a Gross and Net of reinsurance level.  Our results are higher than the provisions of ACS as at 31 December 2011. We have had access to new information and have a different basis for Claims Handling Expenses.  The outcome on the Lyttelton (February 2011 Earthquake) event is absolutely key to final results. Our central estimate of the gross ultimate cost of Lyttelton is $573.8 million. This projects ongoing positive development of the reported incurred. This projected ultimate position exhausts reinsurance covers, including the Adverse Development Cover (“ADC”).  At higher Probability of Sufficiency (“PoS”) levels our results for Lyttelton significantly exceed the available reinsurance including the ADC.  The cost of the claim outcomes that ultimately occur are highly uncertain, with both positive and negative outcomes possible. The final claims outcome is highly dependent on the resolution of a number of specific issues on coverage and policy wording. For example, the allocation of claims between events is a major driver of uncertainty and the application to major claims of a recently developed ACS allocation model is incomplete at this time.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 vi  The ultimate development of claims may largely depend on the outcome of commercial settlements around material claims. The top 25 claims make up 74% of the total incurred costs as at 31 December 2011. As many factors will be at play in the eventual settlement of these claims, we have included scenario testing as part of our work (as traditional actuarial estimation techniques may not reflect the ultimate claim progression). No commercial settlements have occurred to date, and so the likely impact is subject to a wide range of possible outcomes.  Reinsurance assets are material under all scenarios. Our review highlights significant uncertainty in the way the reinsurance may respond due to exchange rate issues. Our results include no explicit allowance for reinsurance dispute or default. Reinsurance dispute or default would have a material adverse impact to the net liabilities.  The ultimate reinsurance recoveries similarly may depend on the outcome of commercial agreements between ACS and the reinsurers. The characteristics of the claims and the wording of the reinsurance policy terms and conditions may mean that there is a shortfall against what recoveries were originally understood to be in place.  Our basis for claims handling expenses considers all non-allocated costs. Our results, at $7.3 million discounted, therefore significantly exceed the provision held as at 31 December 20111 which was not prepared on that basis.  Claims handling costs are directly correlated to the duration of the run-off. The time taken for the liabilities and reinsurance to reach finality is expected to be a lengthy process and is likely to be associated with ongoing management costs and the potential for increased legal costs around disputes.  Many of the uncertainties are beyond the control of ACS, for example, a court case where ACS is not a participant but which sets a precedent. Similarly for the reinsurance recoverable prior to the ADC, the Pound to NZ dollar exchange rate, which may be influenced by the current Euro debt issues, influences the amount actually received.  Over the coming months new information will emerge and work will be completed that may have a material bearing on future liability valuations, and final outcomes, of the ACS liabilities.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 vii Results The following tables summarise the KPMG provision for outstanding claims liabilities at the central estimate, 75% and 90% probabilities of sufficiency at 31 December 2011. KPMG outstanding claims central estimate ($’m) KPMG provision for outstanding claims 75% PoS ($’m) KPMG provision for outstanding claims 90% PoS ($’m)

Segment Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Earthquake 793.8 24.5 5.9 704.8 19.3 5.2 Non earthquake 10.3 4.1 4.1 9.8 3.9 3.9 Expenses 7.8 7.3 Total 811.9 36.4 17.8 721.9 30.4 16.4 Undiscounted Discounted Segment Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Earthquake 856.1 64.5 45.9 760.1 49.4 35.3 Non earthquake 12.2 4.9 4.9 11.6 4.6 4.6 Expenses 8.8 8.2 Total 877.1 78.2 59.6 779.9 62.2 48.2 Undiscounted Discounted Segment Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Earthquake 912.1 100.5 81.8 809.8 76.5 62.4 Non earthquake 14.5 5.8 5.8 13.8 5.5 5.5 Expenses 10.9 9.7 Total 937.4 117.1 98.5 833.3 91.7 77.7 Undiscounted Discounted

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 viii The following table shows the ACS adopted discounted central estimate and the estimate at 75% PoS as at 31 December 2011. Our Net outstanding claims at the central estimate level and higher PoS levels suggest that the Lyttelton event will exhaust the reinsurance program, inclusive of the proposed Adverse Development Cover (“ADC”). We note the high level of uncertainty in the reserves at this time and the potential that the final outcome, overall and on Lyttelton in isolation, may fall below our best estimate or conversely above our 90% PoS figure. ACS adopted provision for outstanding claims ($’m) Inherent uncertainty The estimation of insurance liabilities is subject to uncertainty due to the number of assumptions that have to be made and because liabilities are influenced in some cases, by events that have yet to occur. It should therefore be expected that the actual emergence of the liabilities will vary from those estimated in this report. It follows that ACS’s actual liabilities may ultimately exceed or fall short of those estimated. The final outcome may fall above our 90% PoS or may fall below the central estimate provided. The results of this report should not be interpreted as providing the range of possible outcomes. The uncertainties in the context of the ACS run-off position and Canterbury Earthquakes are very material at this time. The Earthquakes present substantial complexity. General uncertainties relating to the earthquakes are set out in Appendix C and specific matters are discussed in the body of the report. The range of possible final results is wide. We draw your attention to the uncertainties covered in Section 1.6, Section 1.7 and as discussed in the body of the report. Segment Gross outstanding claims Net outstanding claims before ADC Gross outstanding claims Net outstanding claims before ADC Earthquake 624.3 2.1 653.2 2.1 Non earthquake 9.2 4.0 10.3 4.8 Expenses 2.0 2.0 2.2 2.2 Total 635.4 8.1 665.7 9.2 Discounted central estimate Discounted 75% PoS

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 ix Limitations The scope of our work and limitations is set out in Section 1 of the report. We note that our report does not consider the potential outcome from an individual policyholder perspective. For example, we have not modelled or considered the potential impacts of the Western Pacific judgement explicitly. In the event that this judgement is upheld and ACS reinsurance is exhausted the outcome from an individual policyholder perspective is very complicated and uncertain. Executive summary not report The report must be read in its entirety. Individual sections of the report, including the Executive Summary, could be misleading if considered in isolation. In particular, the opinions expressed in the report are based on a number of assumptions and qualifications which are set out in more detail in the report.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 Terms of reference and limitations 1.1 Terms of reference ACS decided to implement a managed withdrawal from New Zealand following the Canterbury Earthquakes. ACS, formerly Ansvar Insurance Limited (New Zealand), has proposed a Contingent Scheme of Arrangement. As set out in the ACS media release of 18 April 2012, the Scheme is described as contingent as it would only be triggered in situations where the company would be insolvent. The media statement also notes that the Scheme is intended to provide better outcomes for claimaints than under a disruptive insolvency process. A review of the Scheme is outside the scope of this report. As part of the Scheme, ACS, at the direction of the Reserve Bank of New Zealand (“RBNZ”) under a Section 126 Notice, have requested KPMG Actuarial Pty Ltd (“KPMG”) to provide an examination of outstanding claims reserve uncertainty as at 31 December 2011. This report sets out the examination KPMG has performed for ACS. The Appointed Actuary (“AA”), Greg Moran of AM Actuaries, prepared a valuation of the outstanding claims liabilities as at 31 December 2011 with the results included in an insurance liability valuation report (“ILVR”) dated 9 March 2012. The AA performed the valuation of the non-earthquake liabilities and adopted the results of the internal ACS Actuary for the earthquake liabilities. Our engagement letter of 11 May 2012 includes details of the scope of the actuarial services provided to ACS. In particular, the examination is to be completed as at 31 December 2011, based upon the data available that was used by ACS in completing the valuation at that date updated for data that is material that has emerged after that date to the date of this report. In recognition of the timetable, it is not anticipated that an updated claims position for all claims will be provided. However, in addition to the data used in the valuation as at 31 December 2011, it will be the responsibility of ACS to provide any further specific data that is, in its view, material to the outcome. 1.2 Items out of scope The scope of the engagement has not extended to:  Independent review of the reinsurance arrangements.  Independent review of case estimates or underlying information (e.g. Loss Adjustor reports, Engineering reports). This includes the basis of any estimation of Business Interruption Claims. 1

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 2  Any independent legal interpretations, such as how many claims the earthquakes have triggered by property and which policies may respond. Where different outcomes, with materially different liabilities resulting, have been identified, these will be addressed within the engagement to the extent possible.  Any independent assessment of any matters relating to potential or expected Constructive Total Loss claims.  Any examination of liquidity issues or asset positions, including assessment of credit risk on reinsurances. We have therefore not recognised the cost of any liquidity funding.  Any independent review of the information provided on claims handling expenses (including both historic and forecast levels).  Any examination of liabilities other than Outstanding Claims liabilities and Claims Handling Expenses  We understand that there are limited unearned premium exposures and that these are substantially reinsured. In the event that policies remain exposed to risk, we note that an examination of the premium liabilities associated with such policies is outside of our scope. We have not reviewed the reinsurances applying to unearned premiums. Furthermore:  We have not independently reviewed the terms and conditions of any run-off administration arrangements proposed or in place.  We have not reviewed and provide no direct comment on the proposed Scheme.  We have not completed any independent interpretations of reinsurances in place or proposed.  We have not reviewed the individual circumstances of individual policyholders or sub-sets of policyholders.  We have not independently reviewed the allocation of claims between events.  Our report does not consider the potential outcome from an individual policyholder perspective. We have not modelled or considered the potential impacts of the Western Pacific judgement explicitly. A potential by-product of the judgement may be additional legal and other costs. Such costs could be material to ACS on a net basis if they are not recoverable from reinsurers. The impact may or may not be material overall; however, the judgement may significantly impact the claim outcome of individual policyholders. In the event that this judgement is upheld and ACS

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 3 reinsurance is exhausted the outcome from an individual policyholder perspective is very complicated and uncertain.  We understand that the claims data and projected liabilities of ACS are net of any GST and that ACS has assumed that any GST payable will be recoverable by ACS. We note that we have not reviewed that position and offer no opinion on the taxation position of ACS and the impact that may have on final liabilities. We note further that we have not allowed for the impact on the net results in the potential timing delay between the payment of GST amounts and recovery.  We have formed judgements based on outcomes that will be influenced or finally resolved by legal processes. We note that, in doing so, we offer no legal opinions in providing this report. Noting the substantial uncertainty in liabilities at this time, our report does not provide any guarantee of, or comment on the current, ongoing or final solvency position of ACS. 1.3 Prudential and professional standards This report has been prepared to comply with the Professional Standard 4 (“PS4”) of the New Zealand Society of Actuaries. We note however, that certain constraints should be noted as limitations to our review:  The timetable of the exercise has been very short. It has not been possible to gather specific data, complete more detailed data checking and analysis, and complete certain dialogues as a result of the timeframe. Material examples include the gathering of data relating to allocation, spot checking data for consistency with claims files or aggregate data with past positions and dialogues with ACS relating to adverse scenarios on the February 2011 event.  Furthermore, certain processes planned by ACS, which will provide relevant data that may be material to the liability valuation are not complete at this time. In particular we note that a process has been designed to re-examine issues of allocation within the estimates on large claims. The application of that process is yet to begin in earnest.  The work refers to the valuation results adopted by ACS as at 31 December 2011, however, we have not completed an actual versus expected analysis relative to those updated claims positions provided.  We note that our scope does not include and examination of assets and we have not completed a determination of the financial soundness of ACS. Therefore Section 9 of PS4 is not applicable to this review.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 4 This report does not include premium liabilities although we understand the gross amount is small and that the liabilities are substantially reinsured. We have reviewed the progress of the policies exposed to risk from 1 January 2012 to the date of this review and, based on the material provided concluded that these do not appear to be material to the results of this review. 1.4 Distribution and use The purpose of this report is limited by the terms of reference discussed above, and should not be used for any other purpose. This report is addressed to ACS and RBNZ. Our report may not be disclosed by ACS or the RBNZ to any other parties without our prior written consent. We provide consent to ACS release within the Ecclesiatical group; we note that all Ecclesiatical group entities outside ACS are considered to be Third Parties. In our letter of engagement KPMG acknowledged that this work has been sought in the context of the proposed Scheme. The letter of engagement provided consent to the RBNZ to disclose our report in the context of the Scheme. We note that the report contains discussion that may be commercially sensitive to ACS and that implicit in the structure of this engagement is our understanding that ACS accepts the release planned by RBNZ. In any event, upon any release to third parties by either RBNZ or ACS, KPMG will accept no liability for any adverse outcome to ACS from any such release, for example where the detail contained in this report may influence a commercial discussion with specific policyholders, service providers or reinsurers. In all cases:  KPMG will not be liable for the consequences of any third party acting upon or relying upon any information or conclusions contained in this report.  Each third party that receives this report recognises that the furnishing of this report is not a substitute for its own due diligence.  This report should only be distributed in its entirety. In the event that ACS or the RBNZ seeks to disclose a sub-set or extract of our report the detail and extent of any extract must be pre-agreed with KPMG.  Judgements about the conclusions and comments drawn in this report should be made only after considering the report in its entirety.  Each third party receiving the report recognises that KPMG is available, at the expense of ACS, to answer questions concerning the report. However, it is also noted that we have no duty to provide additional information or to update this report.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 5  Each third party acknowledges that this report may contain information that is proprietary or confidential to KPMG or ACS. Each third party agrees to hold such information in strict confidence, and not to divulge such information except as may be required by law or judicial process, by any persons or bodies responsible for regulating that third party’s business (including any regulatory or accounting profession supervisory authorities in Australia or elsewhere).  Each third party receiving the report understands that such recipient is deemed to have accepted these terms and conditions by retaining a copy of this report.  Except as necessary under the role of the third party under its regulatory obligations in respect of ACS, the third party agrees not to reference or distribute the report to any other party. Outside of the releases described above, no copy or abstract from this report is to be distributed to third parties without our prior written consent. We assume that the user of this report will seek such explanation as required from us on any matter in question. 1.5 Reliances In preparing this report, full reliance was made on data supplied as outlined in Section 3.1. The accuracy of the results therefore depends upon the accuracy and completeness of the data. We have also relied upon information provided by, and representations made by, the management of ACS and ACS service providers acting on its behalf. We have not independently audited or verified the data with external sources. Should ACS discover material errors in the data, this should be advised to KPMG so that the report can be amended if required. 1.6 Inherent uncertainty The estimated insurance liabilities documented in this report prior to risk margins are intended to be central estimates in the sense that they are intended to represent the expected (i.e. mean) outcome of future claims experience. A risk margin is estimated on top of this to increase the reserves to a level that is expected to provide a 90% PoS. The report also shows 75% PoS positions for further reference, and to document our 90% PoS result where the 75% PoS is an interim step. The process of estimating outstanding claims liabilities is uncertain by nature due to the difficulty of estimating outcomes of events that will occur in the future. As a result, it is still possible that adverse results will impact the results going forward. Some of the main sources of uncertainty are:

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 6  Claims experience may be impacted by the state of the economy including unemployment rates.  There is considerable scope for potential variation in future claim payments.  We have not considered nor allowed for types of claims or events that are not yet apparent or adequately represented in the data. In the event that new types of claims do arise, it is very possible that the estimated reserves will not have adequate provision for these new claims.  Reinsurance dispute or default would have a material impact to the net liabilities.  Whilst every effort is made to reduce the risk through an appropriate range of checks and reviews, there is always the possibility of mis-estimation as a result of data or modelling errors. Key controls include data reconciliation, model reviews, analysis of profit and internal quality assurance review of the overall result.  The timetable available to complete our analysis and prepare this report is noted as a limitation to the scope of the exercise. Key uncertainties specific to each of the earthquake, non-earthquake and expense liabilities is included in the relevant section of the report. 1.7 Emerging information The earthquakes present substantial complexity. The range of possible final results is wide. The final outcome may fall above our 90% PoS or may fall below the central estimate provided. The results of this report should not be interpreted as providing the range of possible outcomes. Over the coming months new information will emerge and work will be completed that may have a material bearing on future liability valuations, and final outcomes, of the ACS liabilities. Key examples include:  Any completed commercial settlements as these may provide a clearer indication of the likely outcomes of negotiation;  Any legal precedent or negotiated precedent in respect of the issues relating to reinstatement amounts, building codes and margin issues as discussed within this report;  The completion of the allocation review proposed and the resulting update to case estimates;  Any dialogue or announcement from the Christchurch City Council in respect of Building codes applicable for repairs;  Greater clarity on the extent of any actual delay in claims notification on the Sumner and 23rd December event;

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 7  The out-workings of the Western Pacific judgement;  Any clarification provided by reinsurers as regards the issue of currency and the attachment point of global covers, or conversely the emergence of any dispute on this matter; and  The resolution of administration arrangements and clarity around ACS legal rights or obligations in respect of ex -gratia support provided to date and potential reinsurance reinstatement liabilities currently provided for outside of ACS. In addition we note that there is ongoing risk of further earthquakes in the region. Although ACS has minimal active policies any future earthquakes have the potential to further complicate and delay the resolution of the ACS liabilities. We also note the ongoing risk that certain properties move or settle further due to the liquefaction that has occurred. That risk leads to the potential for further damage to insured buildings. We therefore note that future liability valuations may show materially different results to those tabled within this report. 1.8 Limitations The limitations of the data used in this report are discussed in section 3. It is noted that the figures provided in this report, being as at 31 December 2011 with allowance for information after that date, are not prepared for the purposes of regular financial reporting. Due to time constraints we have not implemented a full set of controls around our process; specifically we have not performed our own reconciliation of the data provided to an external source or to data used for previous valuations. We are also aware that much of the data provided has not been subject to any audit processes by the ACS auditor.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 8 2 Nature of business 2.1 Overview of ACS ACS, is a member of the Ecclesiastical Insurance Office plc (“Ecclesiastical”) group and as at 31 December 2011 was a wholly owned subsidiary of Ansvar Insurance Limited (Australia). Registration of Ansvar Insurance Limited (New Zealand) occurred on 30 August 1982. On 24 November 2011 ACS announced it will be ceasing to provide insurance cover in New Zealand and will undertake a managed exit from the market quoting “This is a result of the prohibitive cost of reinsurance following the devastating series of earthquakes in Canterbury in 2010 and 2011.” ACS cancelled all policies from 31 December 2011 with the exception of a small amount of engineering business. We have been advised that as at 31 December 2011 there are limited unearned premium exposures and that these are substantially reinsured. We have not reviewed the reinsurances applying to unearned premiums. Prior to being placed into run-off, ACS was an insurer of faith organisations, educational facilities, charitable organisations, community groups, heritage buildings and care facilities. It also offered a range of personal insurance products for individuals and families, protecting homes, their contents and cars. Following the Christchurch earthquakes, ACS has received claims for a number of heritage and other buildings made of unreinforced masonry which experienced significant damage. The Earthquake Royal Commission commissioned a report on the performance of unreinforced masonry buildings during the Canterbury earthquakes. The report discusses the architectural characteristics and seismic vulnerability of unreinforced masonry buildings in New Zealand, makes observations about the performance of such buildings in the Canterbury earthquakes and available techniques for seismic upgrading. The report recommends certain structural elements of all unreinforced masonry buildings be improved to meet the standard for new buildings and that other elements be improved to meet at least 67% of the standard required for new buildings. The authors also recommend that there be one national standard instead of policies being set by individual territorial authorities. We understand that following the earthquakes the Christchurch City Council (“CCC”) has set a target for structural strengthening for the repair of damaged commercial buildings to 67% of the Building Code standards for new buildings and the target was previously set at 33% (some references quote 34%). 2

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 9 ACS has received legal advice on the rebuilding standard with the conclusion stating that “In my view therefore, insurers are only liable to pay for repairs to earthquake damaged buildings to restore them to 34% of the structural integrity of a new building.” However the legal advice also notes that “Ansvar does not yet know the extent to which the CCC may try to enforce the new standard.”. Other legal advice recommends a formal dialogue with the CCC to resolve this issue. At the time of this report no formal advice from the CCC was available. We have been advised that case estimates are currently set assuming that 67% of the building code applies. 2.2 Other material issues relevant to the ACS earthquake liabilities Appendix C provides a general discussion on uncertainty relating to the earthquakes. Further detail of specific relevance to ACS is given below. 2.2.1 Allocation between events As noted by Fitzgerald Consulting Limited (“FCL”) in their March 2012 update, allocation of loss between earthquake events is a key challenge facing the insurance market. FCL have devised a methodology for allocating the losses arising from the major earthquakes, focusing initially on the top 10 cases by value. To date only the allocation exercise for the Majestic Church has been completed. Following the allocation exercise the reported incurred cost for this location increased from $6.8m at 31 December 2011 ($10.6m in the TPAG system and $11.3m in the FCL review) to $16.2m at 30 April 2012 and the allocation between the events was modified significantly. The remainder of the top 10 cases are expected to be completed by 30 September 2012. We understand that the Earthquake Commission is in the process of completing a template that can be used for the allocation of household losses, however this had not been completed at the date of this report. An allocation review is proposed by ACS but incomplete. This has the potential to materially alter both overall case estimates and the allocation between events. 2.2.2 Reinstatement of policy limits Many buildings have suffered damage from more than one earthquake. The typical commercial property policy is annual and has a defined sum insured or policy limit. It is also not uncommon in New Zealand that the value of the property sits above the sum insured. Legal advice provided to ACS notes that the principle of average is not common to New Zealand property insurance.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 10 In the circumstance of the earthquakes it is highly possible that the accumulated damage to a property exceeds policy limits and was, arguably at least, caused by two or more separate events during the policy year. The potential exists that policyholders may pursue claims for the policy limit on an event basis for damage caused during the policy year, i.e. may claim for a multiple of the Total sum insured. Legal advice suggests that the risk that this will apply under law is limited. We can offer no legal comment or opinion on this, but note the uncertainty this gives on whether policy limits may in fact provide an upper limit when considering extreme scenarios. We further note that in the event that policyholders explore this issue formally, even if unsuccessfully, this would add to legal and other costs. Given the figures involved on larger properties the potential for additional claims handling costs or higher final claims (through legal or negotiated resolution) cannot be ruled out. 2.2.3 Margin clause Another feature of many commercial policies is a ‘Margin Clause’ designed to recognise the impact of inflation from the date of the event (or later valuation of claim) and the final settlement date. We understand that at this time the case estimates on many larger commercial policy claims are set at the Total Sum Insured multiplied by the Margin set out in the contract’s Margin Clause. ACS has sought legal input, based on a specific large policy, on whether the margin clause might apply to the total sum insured on the policy. KPMG has not independently reviewed policy wordings. The advice provides the view that ‘the margin clause does not override the Total sum insured …’. It is possible that the amounts above policy limits will not be paid out and that the case estimates on these claims may resolve favourably. As with other legal uncertainties the potential that if policyholders explore this issue formally, even if unsuccessfully, this would add to legal costs. Given the figures involved on larger properties the potential for additional claims handling costs cannot be ruled out. In the case of claim amounts, where case estimates include a Margin allowance already, the indemnity element of adverse outcomes relative to the advice received has already been taken into account. 2.2.4 Reinsurance program non New Zealand Dollar layers The Lyttelton event is expected to draw on reinsurance recoveries from the highest layers of the ACS catastrophe reinsurance program. Not all layers are denominated in New Zealand dollars.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 11 Cat Layer 5 is denominated in British Pounds (“GBP”). In respect of its application to the New Zealand exposures the policy includes a defined rate of exchange of 2.05. In contrast, the Global 1 policy and those above it are denominated in GBP but have no apparent mechanism to confirm conversion to New Zealand dollars that is consistent with Cat Layer 5. This creates uncertainty in how the reinsurance program will ultimately respond. The first element to examine is coverage: in that the coverage of layers denominated in GBP may increase or contract relative to that allowed for dependent on the exchange rate ultimately applicable to gross payments. The second element to examine is attachment points and limits. Certain attachment points are in New Zealand dollars, or in GBP with defined rates of exchange, whereas others are defined in GBP only. The potential exists for material shifts (up and down) in the reinsurance available and for gaps or overlaps in the program to emerge. Commensurate with these uncertainties are the potentials for additional legal and other costs in resolving the issues and the potential for reinsurance dispute. At this time that uncertainty also has the potential to cause dispute with the reinsurer providing the Adverse Development Cover. This risk issue is discussed further in Section 5.2.2. 2.2.5 Interaction with EQC Many domestic claims have a direct interaction with the New Zealand Earthquake Commission. A number of uncertainties that arise from that interaction are set out in our general discussion of uncertainty in Appendix C. For ACS the liabilities for domestic policies are much smaller than those relating to commercial properties. Therefore, those uncertainties are less significant in relevant terms when discussing the uncertainty in ACS’ gross liabilities. On the Lyttelton event the potential exists for exhaustion of the catastrophe reinsurance program. In that context relatively small increases on the domestic policies impacted by that event may remain material in absolute and relative terms, to the final net result. We therefore note that the uncertainty on the domestic policies should not be overlooked and, therefore, the interaction of claim amounts with the EQC is a material consideration when considering the net liabilities.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 12 3 Data 3.1 Data supplied In preparing this report, we have been provided with the items and spreadsheets detailed in Appendix A. We note that ACS are in the process of gathering further data that will be of significant benefit in clarifying exposures and beneficial in defining the potential outcomes and liabilities covered in this report. Of particular note:  ACS has completed a detailed study of issues relating to allocation between earthquake events in respect of the Majestic Church in order to establish a process for the re-examination of this key claims aspect. The application of the process to other claims is planned but incomplete at this time. The data resulting from that review could have a material impact on the valuation of liabilities.  ACS is in the process of negotiation with other entities within the Ecclesiastical group for administrative support. This process is incomplete at this time. We also understand that the downsize of floor space is still subject to negotiations.  ACS is in the process of discussion / negotiation with a number of policyholders to complete a negotiated settlement across policies and damaged sites (‘global settlements’). Although several large positions are under negotiation no material contracts have been completed and no data, as a result, is available at this time from which to extrapolate the probable or likely ACS outcome on open positions. We were provided with spreadsheet triangulations of the paid and incurred position of non earthquake liabilities as at 31 December 2011 that was used in the ILVR. We were provided with an extract of claim transactions for all earthquake events to 31 December 2011 that was used in the Appointed Actuary’s analysis as at 31 December 2011. We were subsequently provided with an extract of claim transactions for all events to 30 April 2012. We understand that the 30 April 2012 extract has not been reconciled to other sources of information. Should ACS discover material errors in this, or any other data, this should be advised to KPMG so that the report can be amended if required. ACS have advised that there have been no major claims settled between 31 December 2011 and the date of this report. 3

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 13 4 Approach 4.1 Outstanding claims liability We have applied actuarial valuation techniques to produce an estimate of the outstanding claims liabilities (at the central estimate, 75% and 90% probability of sufficiency levels). These results in turn illustrate the reserving uncertainty at this time. This specifically includes allowances for:  Outstanding amounts on notified claims;  Claims not yet notified;  Inflation of claims and expenses costs including superimposed inflation;  Expenses in respect of the future administration and establishment of claims;  Future investment earnings; and  Risk margins. 4.2 Analysis segments We have separately performed analysis for:  Earthquake liabilities;  Non earthquake liabilities; and  Expenses associated with the run-off of ACS (“Claims Handling Expenses” or “CHE”). We have not separately considered sub-claim types in our analysis. For example we have not split commercial claims into property damage and business interruption components or split household claims into property damage and contents components. 4.3 Currency Throughout this report, unless otherwise specified, numbers are in New Zealand dollars. 4

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 14 4.4 Non–reinsurance recoveries 4.4.1 Non-reinsurance recoveries We have been advised that the data received is net of GST recoveries. We understand that GST will be payable but recoverable; the amounts will be material. To the extent that there is an asset arising from GST recoveries, we have made no allowance for any non-recoverability of this asset. Non reinsurance recoveries are recognised implicitly within the non-earthquake liability projections. 4.5 Discounting The outstanding claim liabilities are discounted using risk free rates as at 31 December 2011, based on a yield curve derived from New Zealand Government bonds. The annual forward discount rates assumed to apply from 31 December 2011 are shown in the following table. Table 4.1: Assumed forward discount rates We note that our approach to the discounting of gross and net liabilities implicitly assumes that recoveries, including GST, reinsurance and non-reinsurance, are concurrent with gross payments. As a result of this approach, the costs of liquidity funding, if required, are not considered explicitly or implicitly within our results. 2012 2.6% 2013 2.3% 2014 3.3% 2015 3.9% 2016 4.4% 2017 3.9% 2018 4.1% 2019 4.3% 2020 4.5% 2021 4.7% Period ending 31 December Forward discount rate

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 15 5 Earthquake liabilities 5.1 Overview The liabilities for the following Christchurch earthquakes have been valued separately from non-earthquake liabilities:  Darfield (catastrophe code 21) on 4 September 2010.  Boxing Day (catastrophe code 25) on 26 December 2010.  Lyttelton (catastrophe code 29) on 22 February 2011.  Sumner (catastrophe code 32) on 13 June 2011.  23rd December (catastrophe code 34) on 23 December 2011. All other catastrophes are valued as part of the non-earthquake liabilities. This is the same approach as adopted by the Appointed Actuary in his 31 December 2011 ILVR. This section summarises the reinsurance in place for the earthquakes, the approach adopted by ACS to determine the provision for earthquake liabilities as at 31 December 2011 and our approach to estimating the earthquake liabilities. The following table compares our gross and net discounted central estimate of outstanding claims for the earthquake liabilities as at 31 December 2011 with those adopted by ACS. The ADC represents the Adverse Development Cover discussed in Section 5.2.1. Table 5.1 Summary of discounted results ($’m) Event Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Darfield 104.5 0.0 N/A 163.7 0.0 0.0 Boxing Day 5.2 0.4 N/A 5.6 0.4 0.4 Lyttelton 406.0 0.1 N/A 438.2 17.1 3.1 Sumner 104.8 0.7 N/A 89.5 0.8 0.8 23rd Dec 11 3.8 0.9 N/A 7.8 1.0 1.0 Total 624.3 2.1 N/A 704.8 19.3 5.2 ACS adopted KPMG adopted 5

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 16 5.2 Reinsurance We were provided with a spreadsheet summarising the reinsurance arrangements over 2010 and 2011. Where the details appeared inconsistent with the Reinsurance Arrangements Statement for 2010 and 2011 we confirmed the final application with ACS. A summary of our understanding of the reinsurance event limits for non catastrophe cover is shown in the following table. Table 5.2 Summary of event limits for non catastrophe cover ($’m) The following table shows our understanding of the retention and capacity of the catastrophe reinsurance arrangements. Table 5.3 Summary of catastrophe reinsurance ($’m) The exchange rate of 1.9922 is the rate of exchange effective 31 December 2011 as advised by ACS. We understand that the limit on the 2011 cat layer 5 has a fixed exchange rate of 2.05 but that all other exchange rates are floating. We have also been advised that ACS expects no gap in coverage as a result of the differing exchange rates that will apply between Cat Layer 5 and Global 1. We have adopted the ACS view but note the uncertainty in this position. We discuss the uncertainty in Section 5.2.2 below. Type Event limit 2010 CY Event limit 2011 CY Quota share 75.0 40.0 Surplus 1st 115.0 60.0 Surplus 2nd 30.0 30.0 Engineering 1,000.0 10.0 Per risk excess of loss 3.5 1.5 Type Retention 2010 CY Capacity 2010 CY Retention 2011 CY Capacity 2011 CY Buy down 0.5 0.5 1.0 1.5 Cat layer 1 1.0 4.0 2.5 2.5 Cat layer 2 5.0 Cat layer 3 10.0 40.0 10.0 20.0 Cat layer 4 50.0 199.0 1 30.0 50.0 Cat layer 5 80.0 150.6 2 Global 1 249.0 149.4 3 230.6 49.8 4 Global 2 280.4 149.4 5 1 GBP 125m converted to NZD using exchange rate of 1.9922 less retention 2 GBP 112.5m converted to NZD using exchange rate of 2.05 less retention 3 GBP 75m converted to NZD using exchange rate of 1.9922 4 GBP 25m converted to NZD using exchange rate of 1.9922 5 GBP 75m converted to NZD using exchange rate of 1.9922

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 17 The reinsurance provides for 1 reinstatement for catastrophe layer 3 and above in 2011; the Lyttelton and Sumner events are already into this layer. We have been advised that additional backup cover has been purchased for 2011 and in the event that recoveries are expected in this layer for the 23rd December 2011 event the backup covers will respond. We have been provided with details of the backup arrangements. Some of the catastrophe layers are shared with Australia. The following Figure shows the impact of the 2011 catastrophes for both Australia and New Zealand. For illustrative purposes we have shaded cells with horizontal stripes to represent the progression of the Lyttelton event through the layers e.g. as the Brisbane floods consumed all of the 1st cover of cat layer 2, the Lyttelton event claimed under the first reinstatement. Figure 5.4 Summary of 2011 catastrophe treaty response We have been advised the following regarding reinstatement premiums: As at 30.04.2012 Original Cat Treaty Back-up Cover Cost (M&D) Limits 1st Cover 2nd Cover 3rd Cover 1st Cover 2nd Cover Free RIP @ 100% Free RIP @ 100% Free Free RIP @ 100% Free 1,250,960 GBP 261.5M (AU 418m) Purchased eff 21.05.2011 AUS AUS 960,000 GPB 212.5m (AU 340m, NZ 434m) COMBINED 1 NZ 81.9% COMBINED 504,000 GBP 137.5m (AU 220m, NZ 281.25m) COMBINED 1,468,000 GBP 112.5m (AU 180m, NZ 230m) COMBINED COMBINED 1,555,000 80m COMBINED COMBINED 1,569,000 30m COMBINED 1 NZ 12% COMBINED 1,072,000 10m COMBINED 1 NZ 100% COMBINED 2 NZ 18.1% COMBINED COMBINED 1,142,000 5m AUS Qld 100% NZ 1 NZ 100% AUS NZ 2 NZ 100% AUS NZ NZD 324,000 2.5m AUS NZ 1 NZ 100% AUS NZ 2 NZ 100% AUS NZ AUS NZ NZ AUS NZ 1m AUS NZ AUS NZ AUS NZ AUS NZ NZ AUS NZ COMBINED Qld 100% COMBINED 1 NZ 100% COMBINED COMBINED 1 NZ 100% COMBINED 1 NZ 100% COMBINED Qld 12% 1 NZ 88%

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 18  In the event that Global 2 is fully utilised a reinstatement premium will be due of $0.192m (assuming 20% of the premium continues to be allocated to New Zealand) as payments have not yet been made for this layer. We understand that Ansvar Australia has provisioned for that amount.  There is a possible premium of $0.26m for cat layer 2 in the event that the 23rd December event requires this layer (assuming 20% of the premium continues to be allocated to New Zealand). ACS advised that a provision for future payment of the Global 2 premium was made in the 31 December 2011 financial statements of Ansvar Australia with no charge to ACS. We have made no allowance for future payment of this premium in our results. We note that we have no visibility of any legal exposure ACS has to the potential it will be required to meet some or all of this liability if it falls due. We have not projected claims into cat layer 2 from the 23rd December event and have made no allowance for future payment of this premium in our results. The data provided for our investigations includes proportional reinsurance recoveries on payments and case estimates before application of the event limits. In estimating proportional reinsurance recoveries on outstanding claim estimates we have applied the proportion implied by the claims reported to date, subject to the event limits. In line with the ACS approach we have assumed no development on claims impacted by per risk excess of loss reinsurance, however contrary to the ACS approach we have allowed for development on claims impacted by surplus 2nd reinsurance. 5.2.1 Adverse development cover (“ADC”) On the 29 March 2012 an agreement was reached between ACS and Ecclesiastical whereby Ecclesiastical will pay the difference between $570m ultimate net loss for the Lyttelton event and the sum of all third party reinsurance protection available in respect of the Lyttelton event and the net catastrophe retention. The premium for coverage is $4.5m, the first instalment of $100,000 payable within 14 days and the remainder payable on the date that a scheme of arrangement (as described in the ADC agreement) becomes effective. However, if a scheme of arrangement has not been approved and become effective on or before 1 July 2012, Ecclesiastical may any time prior to 30 September 2012 cancel the policy.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 19 We understand that although the premium is to be met by ACS in the event that the premium is payable it is the intention that Ecclesiastical will make a capital injection to ACS which will then be used to make the premium payment. We have made no allowance for the payment of premium for the ADC in our results. We have determined reinsurance recoveries both before and after application of the ADC. 5.2.2 Rate of currency exchange The discussion below examines the potential impacts of exchange rate movements on the net position of the Lyttelton event. The first element to examine is reinsurance coverage: in that the coverage of layers denominated in British Pounds (“GBP”) may increase or contract relative to that allowed for dependent on the exchange rate ultimately applicable to gross payments. The second element to examine is reinsurance attachment points and limits. For all Cat Layers 1-5, the attachment points and limits are in New Zealand dollars, or with defined rates of exchange to the GBP, whereas Global 1 and the above policies are defined in GBP only. The limit on Cat Layer 5 is denominated in GBP, and in respect of its application to the New Zealand exposures the policy includes a defined rate of exchange of 2.05 to the GBP. In contrast, the Global 1 policy and those above it (excluding the ADC) are denominated in GBP but have no apparent mechanism to confirm conversion to New Zealand dollars that is consistent with the limit on Cat Layer 5. This circumstance gives rise to two explicit risks in respect of the response of the program as follows:  If the NZD/GBP exchange rate is greater than 2.05 when applied to gross payments a gap between Layer 5 and Global 1 will exist resulting in an unrecovered net liability to ACS. The coverage above will however shift upwards – potentially therefore available to respond to a higher gross position.  If the NZD/GBP exchange rate is less than 2.05 when applied to gross payments an overlap will emerge between the limit of Cat Layer 5 and the attachment point on the Global 1 policy. The covers above will shift downwards, reducing the top of the cover available. The Global 1 Policy contains a clause – condition f) net retained lines that suggests that the overlap would not be met by that policy – and that, given the way the limit is defined, the effect of that is a reduced cover overall equal to the overlap. In this case it would appear that available reinsurances will fall from the level

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 20 assumed by ACS, and that assumed in our calculations, given that we have followed the ACS interpretation at this stage. ACS has sought confirmation of their interpretation of the Global 1 policy (and policies above it) from their reinsurance broker Aon Benfield. A confirmation has not been available within the timeframe of this review. Comment to date, and our review of the Global 1 policy wording, shows a very complex situation with the potential for reinsurance disagreement, dispute and lower available coverage. A further risk is that reinsurers on Cat Layer 5 and reinsurers on Global 1 dispute the overlap – further complicating discussions, adding to costs and potentially leading to even lower reinsurance assets overall – i.e. neither Cat Layer 5 or Global 1 cover the overlap. It is of note that the ADC is defined in New Zealand dollars. Therefore, these issues should not directly impact the position after the ADC. However, in the event that legal and other costs are incurred in seeking clarity or resolution this may impact the net position to ACS even after application of the ADC. A further potential for dispute exists between ACS and the ADC reinsurer so this interpretation also assumes that the provider of the ADC does not dispute cover as a result of a gap, overlap or material shift in exchange rates increasing the payments otherwise due. We have completed sensitivity testing below to illustrate the financial quantum of these issues. The testing does not include allowance for additional legal costs and is applied prior to the ADC. An additional complexity, set out in the testing, is that exchange rates will also change the cover of each GBP denominated layer – in addition to the potential impacts of a gap or overlap. The net impact, relative to the result from ADC’s existing interpretation, will differ depending on whether the gross position is expected to sit within or above the top of the current interpretation of reinsurance cover. The table below helps to illustrate the potential impact of these issues relative to the current interpretation and relative to KPMG’s adopted net results on the Lyttelton earthquake (prior to the application of the ADC cover). All figures are undiscounted and ignore the further potential impacts of legal costs.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 21 Table 5.5 Potential impact of exchange rate issues ($’m) We note in the event of an upward movement in currency (i.e. appreciation of the GBP against the New Zealand dollar), the resulting uncovered gap between Layer 5 and Global 1 is offset by the expansion in reinsurance coverage and upward shifts in the limits for Global 1 policy and those above it. On the other hand, a downward shift in the currency rate will result in material deteriorations in the net position on the Lyttelton earthquake (prior to application of the ADC). This is due to both a contraction in reinsurance coverage and downward shifts in the limits for Global 1 policy and those above it. As noted, there could be further risk of reinsurer disputes in respect of the payments on the overlapped layers. If all three aspects result the reduction in effective reinsurance coverage relative to that currently interpreted will be very material to the net position before the ADC. In the event that the ADC responds to cover these items the ADC recovery will be much greater than would otherwise be the case. Appendix E shows more detail behind the scenario analysis for the Lyttelton event. 5.2.3 Western Pacific judgement Under this New Zealand High Court judgement there is a statutory charge over reinsurance assets payable to the reinsured which is enforceable by Canterbury policy holders. This decision may be appealed. Currency Scenarios +/-5% currency change +/-10% currency change Base adopted rate - 1.9922 (NZD/GBP) Upward currency shift Gap between layers 4.7 15.9 Expansion in coverage (10.0) (19.9) Increase in limit (4.7) (2.5) (10.0) (6.5) Downward currency shift Dispute due to overlap between layers 17.7 28.9 Contraction in coverage 10.0 19.9 Decrease in limit 17.7 28.9 45.4 77.8 positive figures denote adverse net impacts (negative) figures denote favourable net impacts Net impacts (prior to allowance for ADC) NZ$m

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 22 Our report does not consider the potential outcome from an individual policyholder perspective. We have not modelled or considered the potential impacts of the Western Pacific judgement explicitly. A potential by-product of the judgement may be additional legal and other costs. Such costs could be material to ACS on a net basis if they are not recoverable from reinsurers. The final impact on ACS reinsurance arising from the NZ Court’s rulings to date relating to ACS reinsurance assets, (referred to as the Western Pacific Judgement) is not clear. The impact may or may not be material overall; however, the judgement may significantly impact the claim outcome of individual policyholders. 5.3 ACS methodology and results In a draft report dated 27 February 2012, the ACS Actuary described the approach and summarised the results of the ACS valuation of the earthquake liabilities as at 31 December 2011. Our understanding of the approach and results was supplemented by discussions with the ACS Actuary and a review of the following spreadsheets:  Christchurch IBNR 2012-03-30.xlsx  Analysis of Transactional Claims Data as at 20111231.xlsx The ACS Actuary provided an update to the valuation effective 31 March 2012 in a report dated 14 May 2012. Our understanding of the methodology adopted to estimate the gross ultimate claims costs for each earthquake is summarised below. 5.3.1 Darfield For the 31 December 2011 valuation, the central estimate gross ultimate cost was estimated to be the reported incurred cost at 31 December 2011 of $135m. No risk margin was added at the gross level to bring the results to the 80% probability of sufficiency. In the 31 March 2012 valuation update, the gross ultimate cost was increased to $200m and the risk margin set to $15m at the 80% probability of sufficiency for gross claims. The increase in the central estimate was in response to a difference in opinion between Fitzgerald Consulting Ltd and the local loss adjustors on the estimate of the September 2010 loss at three locations:  Arts Centre: $31m higher;  Catholic Cathedral: $20m higher; and

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 23  Durham St Methodist Church: $10m higher. The revised adopted ultimate is greater than the reported incurred cost in recognition of the Fitzgerald Consulting Ltd estimates. 5.3.2 Boxing Day For the 31 December 2011 valuation, the gross ultimate cost was estimated to be the reported incurred cost at 31 December 2011 of $7m. As for Darfield, no risk margin was added to bring the results to the 80% probability of sufficiency. No change was made to the results in the 31 March 2012 valuation update. 5.3.3 Lyttelton The ultimate gross position for Lyttelton is expected to be the largest of the earthquake events. This is a material point in respect of this exercise as it is this event that is of greatest concern as regards the vertical reinsurance protections. For the 31 December 2011 valuation, four methods were used in the estimation of the gross ultimate cost as follows: a) Market share Based on an assumed insurance market loss of $16bn and a share of 3.2% (based on the share of earthquake premium in New Zealand) this methodology suggested an ultimate cost of $506.5m. No change was made to the results in the 31 March 2012 valuation update under this method. b) Claims development projections Various benchmark incurred cost development patterns were applied to the reported incurred cost at 31 December 2011, with additional percentage delay added to standard development curves in order to allow for the slower incurred cost development relative to other earthquakes. The ACS Actuary has noted the involvement of the Earthquake Commission (“EQC”), the multiple events, limited access to restricted areas, uncertainty around building codes, resource constraints and continuing movement of properties as contributing to slower incurred code development. This methodology suggested an ultimate cost of $504.7m. In the 31 March 2012 valuation update, the ultimate cost was revised to $509.9m from this method. c) Unassessed claims

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 24 This method adjusts the case estimate for claims that have not had a revision to the initial estimate by year-end to a figure closer to the average claim size typical for these types of claims. It also makes an allowance for some claims currently below the EQC cap of $100,000 (plus GST) to deteriorate and for ACS to pay that portion of the claim above the EQC cap. This methodology suggested an ultimate cost of $503.3m. In the 31 March 2012 valuation update, the ultimate cost was revised to $510.3m from this method. d) Fitzgerald Consulting Ltd (“FCL”) review This method considers the results of the FCL review adjusted for the following:  A reduction of 5.5% (applied to total results) to allow for application of 33% of the building code rather than 67% of the building code (derived by considering an estimate from FCL and looking at a selection of claims).  A reduction of 3% (applied to total results) to allow for some clients settling on a lower indemnity basis rather than the replacement value basis (selected by looking at a spread of indemnity values against replacement values and considering the likely negotiating position of ACS and the claimant, cross checked against values to date on the limited number of claims settled). This methodology suggested an ultimate cost of $529.2m. No change was made to the results in the 31 March 2012 valuation update. A summary of the results from the different methodologies and those adopted by ACS are shown in the following table (note the risk margins are shown at the 80% PoS level adopted by ACS at that time): Table 5.6 Summary of ACS adopted Lyttelton earthquake results ($’m) Method Ultimate 31/12 Risk margin Ultimate 31/3 Risk margin Market share 506.5 506.5 Claims development 504.7 509.9 Unassessed claims 503.3 510.3 FCL review 529.2 529.2 Adopted 510.0 25.0 510.0 25.0

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 25 5.3.4 Sumner For the 31 December 2011 valuation, the gross ultimate cost was estimated by applying various benchmark incurred cost development patterns to the reported incurred cost at 31 December 2011. As per the Lyttelton event, an additional percentage delay was added to standard development curves. This methodology suggested an ultimate cost of $100.4m. The ACS Actuary adopted a gross ultimate cost of $115m (allowing for a margin on the incurred cost development results) and a risk margin of $10m to bring the results to the 80% probability of sufficiency. No change was made to the results in the 31 March 2012 valuation update. 5.3.5 23rd December For the 31 December 2011 valuation, the gross ultimate cost was estimated by applying various benchmark incurred cost development patterns to the reported incurred cost at 31 December 2011. As per the Lyttelton event, an additional percentage delay was added to standard development curves. This methodology suggested an ultimate cost of $3.6m. The ACS Actuary adopted a gross ultimate cost of $4m and a risk margin of $1m to bring the results to the 80% probability of sufficiency. No change was made to the results in the 31 March 2012 valuation update. 5.3.6 Total A summary of the gross results adopted by ACS are shown in the following table: Table 5.7 Summary of ACS gross adopted earthquake results ($’m) There has been a small amount of payments made to 31 December 2011 relative to the incurred and ultimate positions and there is significant uncertainty as to the ultimate claims outcomes. A summary of the results net of reinsurance adopted by ACS as at 31 December 2011 are shown in the following table: Event Date Paid Case estimate Incurred Adopted ultimate 31/12 Adopted risk margin 31/12 Revised adopted ultimate 31/3 Revised adopted risk margin 31/3 Darfield 04-Sep-10 24.4 110.2 134.6 135.0 0.0 200.0 15.0 Boxing Day 26-Dec-10 1.5 5.5 7.0 7.0 0.0 7.0 0.0 Lyttelton 22-Feb-11 80.2 404.3 484.5 510.0 25.0 510.0 25.0 Sumner 13-Jun-11 4.0 73.1 77.1 115.0 10.0 115.0 10.0 23rd Dec 11 23-Dec-11 0.0 0.3 0.3 4.0 1.0 4.0 1.0 Total 110.1 593.3 703.5 771.0 36.0 836.0 51.0

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 26 Table 5.8 Summary of ACS net adopted earthquake results ($’m) As the catastrophe reinsurance is not projected to be exhausted (including after application of the risk margin) on any event, ACS has adopted a net incurred amount of the catastrophe reinsurance retention for each event. No adjustment is made for any other items. 5.3.7 Discounting The Appointed Actuary discounted the reserves assuming an average delay to payment of 2 years and an interest rate of 2.5% p.a. 5.4 KPMG methodology and central estimate results 5.4.1 Analysis of transactional claims experience The following table compares some statistics between events based on transactions to 30 April 2012 for the 23rd December event and transactions to 31 December 2011 for earlier events. (In the table below Transaction development month refers to the month end position from the date of the specific event.) Event Date Net incurred Darfield 04-Sep-10 0.5 Boxing Day 26-Dec-10 0.5 Lyttelton 22-Feb-11 1.0 Sumner 13-Jun-11 1.0 23rd Dec 11 23-Dec-11 1.0 Total 4.0

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 27 Table 5.9 Statistics from the different events  As different events caused different amounts of damage, differences in average claim size are not necessarily unexpected between events.  The reported average claim size from the 23rd December event is lower than that for the earlier events at the same period of development.  There is a higher proportion of claims from the 23rd December event with only one transaction recorded compared with the earlier events.  The Sumner event also shows a comparatively high proportion of claims with only one transaction and similarly shows a lower reported average claims size compared with the earlier events.  The Lyttelton event has a significantly higher reported average claim size than the other events.  The claim above $1m reported for the 23rd December event appeared to take a comparatively longer period of time to be recognised as large compared with the earlier events. We understand this claim is the Majestic Church claim and the increase in incurred cost in the 4th development month followed a detailed exercise to allocate the cost between events (no allocation was made to the 23rd December event before this exercise).  ISR policies have a higher reported average claim size than household policies for all events. Event 01234 Current all products Current ISR only Reported average claim size (claims <$1m) Darfield 77,550 $ 66,082 $ 60,349 $ 56,634 $ 54,716 $ 49,874 $ 61,180 $ Boxing Day 245,652 $ 44,894 $ 52,470 $ 53,007 $ 52,871 $ 48,288 $ 67,892 $ Lyttelton 149,257 $ 116,479 $ 103,923 $ 90,810 $ 87,895 $ 75,295 $ 107,211 $ Sumner 49,969 $ 44,998 $ 39,926 $ 34,732 $ 33,406 $ 32,576 $ 46,127 $ 23rd Dec 11 34,234 $ 10,748 $ 10,224 $ 11,809 $ 21,750 $ 21,750 $ 28,446 $ Reported average claim size (claims >$1m) Darfield $6.1m $6.2m $6.2m Boxing Day $2.2m $2.1m Lyttelton $13.5m $9.3m $9m $8.6m Sumner $9.7m 23rd Dec 11 $0m $0m $1.5m $1.5m % of claim reports with one transaction Darfield 0% 7% 5% 9% 9% 3% Boxing Day 50% 33% 24% 30% 10% Lyttelton 39% 46% 37% 18% 17% 13% Sumner 72% 64% 53% 43% 43% 44% 23rd Dec 11 60% 75% 78% 66% 47% 47% Transaction development month

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 28 The following table shows the cumulative claim reports by development month for each event and the calculated chain ladder incurred development factor. Our projected ultimate number of claims has been determined based on the reporting pattern for Darfield, assuming there are an additional 10 reports to arise for Darfield beyond the current position (1.2% of the current reported number). Table 5.10 Cumulative claim reports by event The proportion of ISR claims reported for the Lyttelton event is significantly lower than those for the Darfield and Boxing Day events, whereas the proportion reported for the 23rd December event is significantly higher than these events. However, despite the higher proportion of ISR claims there is also a lower reported average claim size for the 23rd December event compared with earlier events. The following table shows the reported incurred cost development for claims below $1m based on transactions to 30 April 2012 for the 23rd December event and transactions to 31 December 2011 for earlier events. Event 0 1 2 3 4 5 6 Current Ultimate proj Current % claims

$1m Current % ISR Cumulative claim reports Darfield 436 555 622 671 728 776 791 851 861 1.8% 69.2% Boxing Day 2 13 33 35 44 51 54 60 63 3.3% 70.0% Lyttelton 149 502 650 846 894 977 1,019 1,108 1,185 4.2% 51.3% Sumner 89 174 199 237 256 270 284 284 315 2.5% 69.0% 23rd Dec 11 10 25 46 61 74 74 89 1.4% 91.9% Chain ladder development factor Darfield 1.3 1.1 1.0 1.1 Boxing Day 6.5 2.5 1.1 1.3 1.2 1.1 1.1 Lyttelton 3.4 1.3 1.3 1.1 1.1 1.0 1.1 Sumner 2.0 1.1 1.2 1.1 23rd Dec 11 2.5 1.8 1.3 1.2 Report development month

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 29 Table 5.11 Incurred cost development by event for claims < $1m  There has been little incurred cost development for claims below $1m after development month 6. However, there has been relatively little payments to date on claims so the gross ultimate cost continues to be uncertain for the earlier events.  Incurred cost recognition has been slower for the 23rd December event compared with the earlier events. 5.4.2 Update on reported incurred cost development Subsequent to the initial data provided, we were provided with an updated transactions file for all events to 30 April 2012 as well as weekly catastrophe reports showing the reported incurred cost position to 4 May 2012. Effective 4 May 2012, the reported incurred position was $139.8m for Darfield, $527.5m for Lyttelton and $89.1m for Sumner. Effective 30 April 2012, the reported incurred position was $6.4m for the Boxing Day event and $3.1m for the 23rd December 2011 event. The following graphs show the progression of the payments and reported incurred cost over time for the three major catastrophes to 30 April 2012 from the updated transaction file provided. Event 0 1 2 3 4 5 6 Current Cumulative incurred cost ($'m) Darfield 32.8 35.8 36.8 37.3 39.0 40.6 40.7 41.7 Boxing Day 0.5 0.5 1.6 1.7 2.2 2.8 Lyttelton 18.8 53.7 63.0 72.6 74.4 76.6 78.2 79.9 Sumner 4.1 7.5 7.7 8.0 8.3 8.9 9.0 9.0 23rd Dec 11 0.3 0.3 0.5 0.7 1.6 1.6 Chain ladder development factor Darfield 1.09 1.03 1.01 1.05 1.04 1.00 1.02 Boxing Day 1.10 3.02 1.08 1.27 1.25 1.00 1.00 Lyttelton 2.86 1.17 1.15 1.03 1.03 1.02 1.02 Sumner 1.83 1.02 1.04 1.04 1.07 1.01 23rd Dec 11 0.78 1.71 1.54 2.24 Transaction development month

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 30 Figure 5.12 Darfield Figure 5.13 Lyttelton Figure 5.14 Sumner 0 20 40 60 80 100 120 140 160 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 $'m Development month from event Reported incurred cost development (Darfield) Paid Case estimates 0 100 200 300 400 500 600 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 $'m Development month from event Reported incurred cost development (Lyttelton) Paid Case estimates 0 20 40 60 80 100 0 1 2 3 4 5 6 7 8 9 10 $'m Development month from event Reported incurred cost development (Sumner) Paid Case estimates

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 31 5.4.3 Darfield In the 31 March 2012 valuation update ACS adopted a central estimate for the gross ultimate cost of $200m. This compares to the reported incurred cost of $139.8m at 4 May 2012 and the increase to $200m represents allowance for the Fitzgerald Consulting Ltd opinion on three locations as described in Section 5.3.1. We have adopted the ACS estimate of $200m. 5.4.4 Boxing day The Darfield event showed some development between 31 December 2011 and 4 May 2012. We applied the development pattern of Darfield to the reported incurred cost of the Boxing Day event at 31 December 2011 and adopted a central estimate for the gross incurred cost of $7.5m compared with $7.0m adopted by ACS. We were subsequently provided with the reported incurred cost at 30 April 2012 which showed a reduction to $6.4m however we have maintained our result of $7.5m based on the 31 December 2011 analysis. 5.4.5 Lyttelton ACS adopted a central estimate for the gross ultimate cost of $510m. This is lower than the reported incurred cost of $527.7m at 4 May 2012. The reported incurred position included loss adjuster estimates based on repair to 67% of the building code. A reduction for repairs to 33% of the buildings code appears to be a key factor in the $510m selected by ACS. Based on the recent incurred cost experience of the Darfield event we have adopted a central estimate for the gross incurred cost of $550m. We note that the development factor applied is relatively modest relative to the development seen to date and implicitly acknowledges that some policies are at or near policy limits. Our primary methodology was the application of an incurred development pattern from the Darfield event to date. The pattern of development of Darfield suggests that there may be further development – however we have not added that to the development factor selected for Lyttelton. We have performed a high-level check of the adopted incurred cost development pattern by fitting statistical curves to the incurred cost development experience of both the Darfield and Lyttelton events and concluded that the allowance is within the range suggested by the curves. This is shown in Appendix B. As discussed in Section 5.5 we have also examined scenarios completed on the largest claims/locations to complete a high-level check of this result.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 32 As discussed in Section 5.4.8, to the $550m we have added an additional explicit allowance for inflation noting the ongoing delay in finalising claims, the risk of additional localised demand surge inflation and the clear complexity in loss adjusting and other activity that will add to claims costs in the settlement process. Our methodology implicitly allows for late reporting of household claims that may occur as a result of these claims developing above the EQC cap of $100,000. 5.4.6 Sumner For the Sumner event we have considered the results of an incurred cost development method applied to the incurred position at 4 May 2012 (using development patterns from both the Darfield and Lyttelton experience to 4 May 2012) and ACS’ adopted results. The central estimate of the gross ultimate cost is shown in the following table. Table 5.15 Sumner event: gross ultimate cost from different methods ($’m) We have adopted a central estimate of $100m allowing for some increase in the results of the incurred cost development methods as the Darfield and Lyttelton events continue to show development and the ultimate cost of these events is still uncertain. 5.4.7 23rd December At the time of the 31 December 2011 valuation little data had emerged on this event. The gross ultimate cost figure adopted by ACS at 31 December 2011 was $4.0m. As there appears to be an increased delay to recognition of reported incurred cost (both in terms of the number of claims with only one transaction and the observed incurred development pattern) for the 23rd December event compared with the earlier events, a method that relies on reported incurred cost development is expected to understate the ultimate incurred cost. We have therefore adopted an average claim size and numbers approach for this event. The following table shows the adopted central estimate for the gross ultimate cost for the 23rd December event. Method Gross ultimate cost Incurred cost development (Darfield) 96.6 Incurred cost development (Lyttelton) 96.8 ACS adopted 115.0 KPMG adopted 100.0

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 33 Table 5.16 Gross ultimate cost December 23rd event A gross ultimate cost of $8.4m has been adopted. Where policyholders, domestic and commercial, already have claims in place, there is a strong potential of under reporting of claims for this event. We also note that the higher proportion of commercial claims may further suggest that domestic claims may be under-represented to date in the data. Given these two drivers of a slower development we observe that outcomes materially higher than our projected gross ultimate cost remain possible. We understand anecdotally that as claims assessments occur they tend to get allocated to the Lyttelton event rather than the later events. Once the allocation exercise is rolled out (following completion of the Majestic Church test case), this could lead to a decrease in the proportion allocated to the Lyttelton event and an increase in the proportion allocated to the Sumner and 23rd December events. We also note that there may be additional notifications in respect of this event once allocation issues have been finalised 5.4.8 Total Two separate inflation related pressures are noted in the circumstance of the Christchurch earthquake. The first is often described as “Demand Surge” and relates to localised inflation due to high demand for services, building materials etc. relating to the catastrophe. In addition in this circumstance we have ongoing uncertainty and delay in completing building repairs, claims assessments and, ultimately, slower settlement. This second element will also lead to higher ultimate claims as general inflation will continue and influence final settlements. In order to allow for demand surge inflation and the inflation impacts of ongoing delay we have increased the outstanding claims component for each event described above by 5%. The 5% is a general allowance noting that for policies at limits these inflation pressures will not impact indemnity amounts; the pressures will in that case still impact loss adjuster and other legal costs. We note that this allowance may still understate this risk as few claims are settled and industry wide uncertainty remains. Number Average claim size Gross ultimate cost Claims < $1m 87 $50,000 $4,350,000 Claims > $1m 2 $2,000,000 $4,000,000 Total 89 $2,050,000 $8,350,000

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 34 We also note that the longer claims remain open associated costs would also typically rise – through for example the need for additional reports or more detailed reports being required. We have made no explicit adjustment for that potential impact of delay on the final gross position. We observe that for commercial claims policy limits will reduce that exposure. Although this has been applied to all policies even if they are already at policy limits we have considered scenarios for large claims allowing claiming to policy limits only in Section 5.5. We have set out results before and after this inflation pressure allowance. Our net results are based on the gross results inclusive of this adjustment. A summary of our gross and net central estimate results are shown in the following table: Table 5.17 Summary of KPMG gross adopted earthquake results ($’m) The central estimate for the net ultimate cost of the catastrophes has been determined as $26.4m before allowance for the ADC and $7.8m after allowance for the ADC. When considering the Adopted Net after ADC the ADC premium expense of $4.5m also requires recognition. A summary of the assumed ultimate reinsurance recoveries is shown in the following table: Table 5.18 Summary of KPMG adopted reinsurance recoveries ($’m) Event KPMG adopted gross ultimate before inflation adjustment KPMG adopted gross ultimate after inflation adjustment KPMG adopted net ultimate before ADC KPMG adopted net ultimate after ADC Darfield 200.0 208.8 0.5 0.5 Boxing Day 7.5 7.8 0.5 0.5 Lyttelton 550.3 573.8 23.4 4.8 Sumner 100.0 104.8 1.0 1.0 23rd Dec 11 8.4 8.8 1.0 1.0 Total 866.2 904.0 26.4 7.8 Event Quota share Surplus 1st Surplus 2nd Engine ering Per risk Xol buy down Per risk Xol Cat buy down Cat layer 1 Cat layer 2 Cat layer 3 Cat layer 4 Cat layer 5 Global 1 Global 2 Total before ADC ADC Darfield 43.0 115.0 4.1 0.2 0.5 1.7 0.5 4.0 5.0 34.2 0.0 208.3 0.0 Boxing Day 0.6 6.5 0.0 0.1 0.0 7.3 0.0 Lyttelton 40.0 60.0 11.1 8.9 0.0 1.5 1.5 2.5 5.0 20.0 50.0 150.6 49.8 149.4 550.4 18.6 Sumner 7.7 60.0 9.1 0.0 0.0 0.6 1.5 2.5 5.0 17.3 0.0 103.8 0.0 23rd Dec 11 4.4 0.0 1.5 1.9 0.0 7.8 0.0 Total 95.8 241.5 24.4 9.2 0.5 3.8 5.1 10.9 15.0 71.5 50.0 150.6 49.8 149.4 877.6 18.6

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 35 We expect the payment pattern of the earthquake liabilities to be ‘lumpy’ with occasional large payments as settlements or progress payments are made for large claims. We have discounted the gross outstanding claims component assuming an average delay to payment of 4 years. We have applied no discounting to the net outstanding claims apart from the Lyttelton event where we have assumed a delay of 8 years for amounts in excess of the catastrophe retention of $1.0m. The undiscounted and discounted central estimate of the outstanding claims is shown in the following table. Table 5.19 Summary of adopted outstanding earthquake results ($’m) The gross and net payments to 31 December 2011 are shown in the following table. Table 5.20 Summary of earthquake payments ($’m) 5.5 Scenarios and discussion on individual large claims There continues to be significant uncertainty surrounding the gross ultimate cost of each event. To illustrate this uncertainty and to help assess the adequacy of the proposed central estimate, we have performed a number of scenarios on individual locations as described below. As these locations are impacted by a number of events, we have presented the analysis by location rather than by event. The reported incurred noted in the tables that follow is as at 31 December 2011. The policy information was provided by ACS for the top 25 claim locations. Event Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Darfield 184.4 0.0 0.0 163.7 0.0 0.0 Boxing Day 6.3 0.4 0.4 5.6 0.4 0.4 Lyttelton 493.6 22.4 3.8 438.2 17.1 3.1 Sumner 100.8 89.5 0.8 0.8 23rd Dec 11 8.8 1.0 1.0 7.8 1.0 1.0 Total 793.8 24.5 5.9 704.8 19.3 5.2 Undiscounted Discounted Event Gross payments Net payments Darfield 24.4 0.5 Boxing Day 1.5 0.1 Lyttelton 80.2 1.0 Sumner 4.0 0.2 23rd Dec 11 0.0 0.0 Total 110.1 1.9

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 36 The following table shows the reported incurred cost at 31 December 2011 of each location discussed. Table 5.21 Reported incurred cost for large locations ($’m) 5.5.1 Arts Centre When the policy was written the buildings were valued at $100m. At the end of December 2010 the building was professionally valued again at $120m. Following adherence to the risk management policy of ACS, ACS renewed insurance of $100m from November 2010. A further $20m was insured by Lumley. The reported incurred cost allows for a 15% margin clause, however, legal advice suggests that this applies only to sub-limits and that the overall sum insured is maintained. The total cost of repairs has been quoted as $240m by the Arts Centre. Currently the claim from the second policy year is allocated 50% to each of the Lyttelton and Sumner events. Location Reported incurred cost 31 December 2011 ($'m) Cumulative % Arts Centre 138.1 20% Catholic Cathedral 67.2 29% St Margaret's College 49.2 36% Anglican Cathedral 39.6 42% Rangi Ruru Girls School 30.9 46% St Andrews College 29.9 50% Nazareth House 28.7 55% Churchill Courts Complex 19.1 57% Durham St methodist church 13.8 60% Bishops Court Aged Care 8.6 60% Majestic Church 6.8 61% Other locations in top 25 88.2 74% Other claims 183.1 100% Total 703.5 100% Start date End date Sum insured Margin Reported incurred Any Disputes on the claim Dar Box Lytt Sum 23rd 1-Jan-10 31-Dec-10 $97,230,000 15% $26,181,478 100% 0% 0% 0% 0% 1-Jan-11 31-Dec-11 $100,000,000 15% $111,956,336 0% 0% 50% 50% 0% Allocation

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 37 It appears plausible that this allocation could change to 100% Lyttelton if damage caused by this event had already reached the sum insured. The impact of this scenario is an increase in the gross central estimate of the Lyttelton event by $59m (allowing for an overall limit of $115m) and a corresponding decrease in the gross central estimate of the Sumner event. The remaining surplus 2nd cover firstly reduces the gross impact of the amount transferred to the Lyttelton event. (As the Lyttelton event is already at limits for quota share and surplus 1st coverage these do not affect the net impact of this change.) After allowance for recoveries on surplus 2nd the recoveries due from the catastrophe program increase by $51.7m. At the central estimate level the catastrophe program before and after ADC is already exhausted for this event, so the net incurred cost increases $51.7m (both before and after ADC) under this scenario based upon KPMG’s adopted gross ultimate after inflation allowance. We also understand the potential for Commercial policyholders to present an argument that the sum insured is available more than once in a year for different events due to a Reinstatement Amount clause. After allowance for the inflation adjustment it is estimated that the central estimate for the Arts Centre implicitly includes a total gross ultimate of $188.5m. This estimate allows for the $31m adjustment described in Section 5.3.1 and assumed development. Assuming the estimated total costs are claimed of $240m and that ACS takes 83.3% of the deterioration which is fully allocated to the Lyttelton event, the gross ultimate cost increases by $43m and the net ultimate cost increases by $37.8m (recognising additional recoveries available on the surplus 2nd treaty) due to the increase to the gross position on the Lyttelton event. We have also prepared a scenario providing no allowance for claim development above the policy limit of $100m in 2011 (assuming the decrease occurs equally on the Lyttelton and Sumner events) which effectively removes the 15% margin clause and any development on these events implicitly allowed for in our central estimate basis above $100m. Table 5.22 Arts Centre scenarios on KPMG central estimate ($’m) Scenario Increase to gross ultimate Increase to net ultimate before ADC Increase to net ultimate after ADC Allocate from Sumner to Lyttelton 0.0 51.7 51.7 Claim $240m 43.0 37.8 37.8 Claim $100m in 2011 (27.6) (12.1) (3.8)

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 38 5.5.2 Catholic Cathedral We note that there has been no incurred cost recorded for the Sumner or 23rd December events so there is the possibility of additional claims registered for these events. The reported incurred cost for the 2nd policy year is currently above the sum insured, however, it is below what the sum insured would be if the 10% margin was applied ($62.7m). We are not aware of the current estimated repair costs, although we understand the case estimates have been set on the basis of 67% of the building code and that the building has been declared a total loss. As the majority of this claim is already allocated to the Lyttelton event, there appears to be minimal risk of additional cost allocated to this event increasing the net ultimate incurred cost. However, it is possible that the policyholder argues that the sum insured is available more than once in a policy year for different events. After allowance for the inflation adjustment it is estimated that the central estimate implicitly includes a total gross ultimate of $89.5m for this location (allowing for the $20m adjustment described in Section 5.3.1 and assumed development). Assuming a total claim of $100m with the deterioration fully allocated to the Sumner event (as the Lyttelton event is already close to the limit), the gross ultimate cost increases by $10.5m, but the net ultimate cost is unchanged as the deterioration is picked up by the reinsurance in place. We understand there have been some discussions suggesting that a lower payout amount may eventuate if one building is built to replace three buildings as the indemnity payment would be less than the full sum insured in this situation. We have not been advised what this payment would be. We have performed a scenario assuming the ultimate payment is $65m, with the decrease allocated to the Lyttelton event. Start date End date Sum insured Margin Reported incurred Any Disputes on the claim Dar Box Lytt Sum 23rd 30-Nov-09 30-Nov-10 $48,415,310 10% $10,000,000 Floating sums insured, Indemnity v Replacement Values, Building 100% 0% 0% 0% 0% 30-Nov-10 31-Dec-11 $56,959,802 10% $57,226,594 Code issues, Building Act strengthening,Unda maged foundations. 0% 4% 96% 0% 0% Allocation

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 39 We have also prepared a scenario providing no allowance for claim development above the policy limit of $57m in 2011. This effectively removes the allowance of 10% for the margin clause and any development on these events implicitly allowed for in our central estimate basis above $57m. The scenario apportions the resulting decrease by event in the same proportion as the reported incurred. (We note that this scenario does not allow for loss adjuster and other expenses incurred on this claim.) Table 5.23 Catholic Cathedral scenarios on KPMG central estimate ($’m) 5.5.3 St Margaret’s College Management have noted that they expect that this claim will proceed to settlement shortly with a potential reserve release. Details of the proposed settlement are not available and no formal correspondence with the policyholder has been provided. The reported incurred cost increased from $49.2m at 31 December 2011 to $54.3m at 30 April 2012. After allowance for the inflation adjustment it is estimated that the central estimate implicitly includes a total gross ultimate of $59.1m for this location (allowing for assumed development). As this is slightly greater than the sum insured (with no margin clause) we have performed a scenario at the sum insured, on the same allocation to event as shown above. An alternative scenario has been performed assuming that a saving compared with the reported incurred cost at 30 April 2012 is achieved on settlement of 5.5% (in line with the ACS adjustment to case estimates to allow for a building code of 33% vs 67%), producing a gross ultimate cost of $51.3m. Scenario Increase to gross ultimate Increase to net ultimate before ADC Increase to net ultimate after ADC $100m ultimate cost 10.5 0.0 0.0 $65m ultimate cost (24.5) (24.5) (3.8) Claim $57m in 2011 (6.2) (6.0) (3.8) Start date End date Sum insured Margin Reported incurred Any Disputes on the claim Dar Box Lytt Sum 23rd 1-Aug-10 31-Jul-11 $58,007,264 0% $49,173,647 23% 4% 61% 12% 0% 1-Aug-11 31-Dec-11 $58,007,264 0% $0 0% 0% 0% 0% 0% Repair solutions for a number of buildings to be resolved and agreed Allocation

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 40 Table 5.24 St Margaret’s College scenarios on KPMG central estimate ($’m) 5.5.4 Anglican Cathedral We note that there has been no incurred cost recorded for the Sumner or 23rd December events which appear to fall within a separate policy year to the Lyttelton event so there is the possibility of additional claims registered for these events. However, noting the assessment of a constructive total loss following the Lyttelton event this is not expected despite media reports of further damage from later events. We have been advised that settlement of this claim has been agreed at $38.0m. The reported incurred cost increased from $38.9m at 31 December 2011 to $39.5m at 30 April 2012. After allowance for the inflation adjustment it is estimated that the central estimate implicitly includes a total gross ultimate of $43.7m for this location (allowing for assumed development) which is slightly less than the sum insured plus 15% margin. We have done a scenario of a settlement value of $38m on the same allocation to event shown above. (We note that this scenario does not allow for loss adjuster and other expenses incurred on this claim which may result in a claim above the limit.) Table 5.25 Anglican Cathedral scenario on KPMG central estimate ($’m) 5.5.5 Rangi Ruru Girls School Scenario Increase to gross ultimate Increase to net ultimate before ADC Increase to net ultimate after ADC $58m ultimate cost (1.1) (0.7) (0.7) $51.3m ultimate cost (7.8) (4.7) (3.8) Start date End date Sum insured Margin Reported incurred Any Disputes on the claim Dar Box Lytt Sum 23rd 1-Apr-10 31-Mar-11 $39,364,229 15% $39,647,316 1% 1% 98% 0% 0% 1-Apr-11 31-Dec-11 $39,364,229 15% $0 0% 0% 0% 0% 0% Allocation Application of Margins clause to demolition and inflation provisions Scenario Increase to gross ultimate Increase to net ultimate before ADC Increase to net ultimate after ADC $38m ultimate cost (5.7) (5.6) (3.8) Start date End date Sum insured Margin Reported incurred Any Disputes on the claim Dar Box Lytt Sum 23rd 1-Aug-10 1-Aug-11 $56,700,520 0% $30,918,851 2% 2% 94% 2% 0% 1-Aug-11 31-Dec-11 $56,800,520 0% $0 0% 0% 0% 0% 0% Allocation

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 41 After allowance for the inflation adjustment it is estimated that the central estimate includes a total gross ultimate of $39.9m for this location (allowing for assumed development) which is significantly lower than the sum insured. Although no details are available it appears possible that there could be additional future development and the claims managers have noted that it is a multiple building campus, with various repair/valuation methodologies to address. We have performed scenarios assuming the ultimate payment is $45m, both assuming the allocation to event is as per the reported incurred cost and assuming the increase is all allocated to the Sumner event. We have also performed scenarios assuming the ultimate payment is at the limit of $56.7m, both assuming the allocation to event is as per the reported incurred cost and assuming the increase is all allocated to the Sumner event. Table 5.26 Rangi Ruru School scenario on KPMG central estimate ($’m) 5.5.6 St Andrews College After allowance for the inflation adjustment it is estimated that the central estimate implicitly includes a total gross ultimate of $31.5m for this location (allowing for assumed development) which is significantly lower than the sum insured. Although no details are available it appears possible that there could be additional future development. We note the insured is continuing with their repair schedule. We have performed scenarios assuming the ultimate payment is $40m, both assuming the allocation to event is as per the reported incurred cost (allowing some recoveries from the surplus 2nd reinsurance for the Lyttelton event) and assuming the increase is all allocated to the Sumner event. We have also performed scenarios assuming the ultimate payment is at the limit of $77.5m, both assuming the allocation to event is as per the reported incurred cost and assuming the increase is all allocated to the Sumner event. Scenario Increase to gross ultimate Increase to net ultimate before ADC Increase to net ultimate after ADC $45m ultimate cost 5.1 4.8 4.8 $45m ultimate cost Sumner alloc 5.1 0.0 0.0 $56.7m ultimate cost 16.8 15.7 15.7 $56.7m ultimate cost Sumner alloc 16.8 0.0 0.0 Start date End date Sum insured Margin Reported incurred Any Disputes on the claim Dar Box Lytt Sum 23rd 1-Aug-10 1-Aug-11 $77,471,582 0% $29,910,940 6% 0% 89% 5% 0% 1-Aug-11 31-Dec-11 $66,890,582 0% 0% 0% 0% 0% 0% Allocation

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 42 Table 5.27 St Andrews College scenario on KPMG central estimate ($’m) 5.5.7 Nazareth House As the majority of this claim is already allocated to the Lyttelton event, there appears to be minimal risk of additional cost allocated to this event increasing the net ultimate incurred cost. However, it is possible that the policyholder argues that the sum insured is available more than once in a year for different events. After allowance for the inflation adjustment it is estimated that the central estimate implicitly includes a total gross ultimate of $31.2m for this location (allowing for assumed development). We have performed a scenario assuming a total claim of $35m with the deterioration fully allocated to the Sumner event (as the Lyttelton event is already close to the limit). We have also prepared a scenario providing no allowance for claim development above the policy limit of $28.8m in 2011 (assuming the decrease occurs on each event in the same proportion as the reported incurred) which effectively removes the 10% margin clause and any development on these events implicitly allowed for in our central estimate basis above $28.8m. Table 5.28 Nazareth House scenario on KPMG central estimate ($’m) Scenario Increase to gross ultimate Increase to net ultimate before ADC Increase to net ultimate after ADC $40m ultimate cost 8.5 6.8 6.8 $40m ultimate cost Sumner alloc 8.5 0.0 0.0 $77.5m ultimate cost 46.0 37.1 37.1 $77.5m ultimate cost Sumner alloc 46.0 0.0 0.0 Start date End date Sum insured Margin Reported incurred Any Disputes on the claim Dar Box Lytt Sum 23rd 30-Nov-09 30-Nov-10 $25,887,700 10% $70,725 Floating sums insured, Indemnity v Replacement Values, Building 100% 0% 0% 0% 0% 30-Nov-10 31-Dec-11 $28,848,700 10% $28,666,070 Code issues, Building Act strengthening,Unda maged foundations. 0% 0% 98% 2% 0% Allocation Scenario Increase to gross ultimate Increase to net ultimate before ADC Increase to net ultimate after ADC $35m ultimate cost 3.8 0.0 0.0 $28.8m ultimate cost (2.4) (2.3) (2.3)

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 43 5.5.8 Churchill Courts Complex (Anglican) We note that there has been no incurred cost recorded for the Sumner or 23rd December events which appear to fall within a separate policy year to the Lyttelton event so there is the possibility of additional claims registered for these events. After allowance for the inflation adjustment it is estimated that the central estimate implicitly includes a total gross ultimate of $22.2m for this location (allowing for assumed development). We understand that there is a possibly of a cash settlement. As the Darfield and Lyttelton events fall within the same policy year, we have not considered the possibility of a greater allocation to the Lyttelton event although this is a possible scenario. We have prepared a scenario providing no allowance for claim development above the policy limit of $17m in the policy year ended 30 April 2011 (assuming the decrease occurs on each event in the same proportion as the reported incurred) which effectively removes the 10% margin clause and any development on these events implicitly allowed for in our central estimate basis above $22.2m (although we note that reported incurred is currently higher than the policy limit). Table 5.29 Churchill Courts scenario on KPMG central estimate ($’m) Start date End date Sum insured Margin Reported incurred Any Disputes on the claim Dar Box Lytt Sum 23rd 1-Apr-10 30-Apr-11 $17,028,642 15% $19,141,991 67% 0% 33% 0% 0% 1-Apr-11 31-Dec-11 $19,464,542 15% $0 0% 0% 0% 0% 0% Allocation Application of Margins clause to demolition and inflation provisions Scenario Increase to gross ultimate Increase to net ultimate before ADC Increase to net ultimate after ADC $17m ultimate cost (5.1) (1.7) (1.7)

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 44 5.5.9 Durham St Methodist Church We have not performed any additional scenarios for this location as the Darfield event includes a $10m increase to reported incurred cost (as described in Section 5.3.1) and the reported incurred is already above the sum insured. 5.5.10 Bishops Court Aged Care (Anglican) After allowance for the inflation adjustment it is estimated that the central estimate implicitly includes a total gross ultimate of $9.2m for this location (allowing for assumed development) which is significantly lower than the sum insured. Although no details are available it appears possible that there could be additional future development. We understand that there is a possibly of a cash settlement along with the Churchill Courts Complex. We have performed scenarios assuming the ultimate payment is at the limit of $16m in the first policy year assuming the allocation to event is as per the reported incurred cost. Table 5.30 Bishops Court scenario on KPMG central estimate ($’m) Start date End date Sum insured Margin Reported incurred Any Disputes on the claim Dar Box Lytt Sum 23rd 1-May-10 1-May-11 $12,000,000* 0% $13,820,265 86% 0% 14% 0% 0% 1-May-11 31-Dec-11 N/A 0% $3,031 0% 0% 0% 100% 0%

this is the global sum insured rather than for the site. Allocation Additional payments clause could apply where any building damaged beyond 80% of it's reinstatement value ACS will pay full reinstatement even if the insured chooses not to reinstate. Brokers consider this clause triggered on this claim Start date End date Sum insured Margin Reported incurred Any Disputes on the claim Dar Box Lytt Sum 23rd 1-Apr-10 30-Apr-11 $16,340,961 15% $8,444,699 4% 0% 96% 0% 0% 1-Apr-11 31-Dec-11 $19,989,372 15% $174,653 0% 0% 0% 100% 0% Application of Margins clause to Allocation Scenario Increase to gross ultimate Increase to net ultimate before ADC Increase to net ultimate after ADC $16m ultimate cost 7.2 6.9 6.9

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 45 5.5.11 Majestic Church The Majestic Church claim was a “test case” for a detailed allocation exercise that is expected to be rolled out for other large claims. Following the allocation exercise the reported incurred cost for this location increased to $16.2m at 30 April 2012 and the allocation between events was modified to 10%, 4%, 64%, 12% and 10% (in chronological order). Despite the increase in reported incurred cost we have been advised that there is some scope for a reduction on settlement. As the reported incurred cost is close to the sum insured (before application of the margin clause) we have not performed any additional scenarios. 5.5.12 Summary of range of outcomes from scenario analysis The following table summarises the range of increases and decreases to the gross and net ultimate cost from the scenario analysis performed on the central estimate for the locations discussed above. Table 5.31 Summary of scenario analysis on KPMG central estimate ($’m) Relative to the implied central estimate of the locations forming the scenario analysis, the unfavourable scenarios suggest an increase to the gross ultimate of 31% and the favourable scenarios a decrease of 18%. 5.5.13 Other locations in top 25 For other locations in the top 25 claims, the following table shows the sum insured (for the policy year that includes the Lyttelton event) and the reported incurred cost as at 31 December 2011 (again for the policy year that includes the Lyttelton event). Start date End date Sum insured Margin Reported incurred Any Disputes on the claim Dar Box Lytt Sum 23rd 8-May-10 8-May-11 $17,355,000 15% $6,830,396 2% 0% 98% 0% 0% 8-May-11 31-Dec-11 $17,355,000 15% $0 0% 0% 0% 0% 0% Allocation ACS relying on pol. condition reverting to indemnity payment - current potential for a significant reserve saving. KPMG gross ultimate after inflation adjustment KPMG net ultimate before ADC KPMG net ultimate after ADC Adopted central estimate 904.0 26.4 7.8 Favourable scenarios 830.9 4.0 4.0 Unfavourable scenarios 1,031.1 137.8 119.2

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 46 It appears that there is an additional $26.5m of sum insured available for these claims (before allowance for any margin clauses) that could potentially be claimed for the Lyttelton event. 5.6 Sensitivity testing on the Lyttelton event To illustrate the sensitivity of results to key assumptions, we have performed a number of sensitivities around the Lyttelton event estimates as follows: a) Catastrophe reinsurance exhausted – this shows the point at which recoveries are estimated to begin from the ADC based upon the current interpretation of the catastrophe programme and an exchange rate of 1.9922 as per the central estimate. b) ADC exhausted – this shows the point at which the limit of the ADC cover is exhausted. c) No further development in the reported incurred (i.e. no IBNR/IBNER) from 4 May 2012 – this shows the impact of adopting the reported incurred cost at 4 May 2012 as the central estimate. d) Adopt additional tail of 1.5% - this shows the impact of adopting an additional 1.5% incurred cost development to Lyttelton in addition to that of our central estimate assumption. The additional 1.5% is based upon the fitting of statistical curves to the Darfield reported incurred development to date. e) 25% of Lyttelton re-allocated to Darfield – this shows the impact of modifying the allocation between these two key events. The adopted results suggest 63% of total costs allocated to the Lyttelton event which is similar to the results of the Majestic test case (64%). In contrast, this sensitivity is based upon a much lower allocation to Lyttelton of 48%. The net results of Darfield are unchanged under this scenario as there continues to be a significant amount of catastrophe protection remaining on this event. f) 33% building code on settlements – this shows the impact of adopting an ultimate result 5.5% lower than the adopted result which is based on the ACS adjustment to the FCL results described in Section 5.3.3. $109,950,770 $83,457,235 Sum insured (covering Lyttelton) Reported incurred (covering Lyttelton)

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 47 g) 1% saving on settlement - this sensitivity was suggested by ACS management and shows the impact of adopting an ultimate result 1% lower than the adopted result to reflect the possible saving on settlement as a result of settling on indemnity or reductions due to the time value of money. h) Remove margin clause for large claims – this sensitivity reduces the reported incurred cost for the individual claims discussed in Section 5.5 to a maximum of the sum insured excluding margin clause before applying the assumed development factors. i) CoV +/-5% - this shows the impact of modifying the key risk margin assumptions by +/- 5% in the 90% PoS results. The sensitivities are shown in the following table. Table 5.32 Sensitivity testing on the Lyttelton event ($’m) The sensitivity tests further assist in demonstrating the wide range of potential outcomes. Many show exhaustion of the ADC. Others show the ADC remaining intact. 5.7 Risk margins Given the recency of the earthquake events, and the lack of actual experience on settled claims, KPMG’s analysis of risk margins for earthquake claims is necessarily highly judgemental. A coefficient of variation (CoV), a measure of volatility of the central estimate, is selected by earthquake event. The gross risk margin is determined assuming a Normal distribution. The net risk margin is determined by applying the reinsurance arrangements to the adopted gross ultimate inclusive of risk margins. Sensitivity Gross central estimate Gross 90% PoS Net central estimate before ADC Net 90% PoS before ADC Net central estimate after ADC Net 90% PoS after ADC Adopted 493.6 569.5 22.4 98.3 3.8 79.7 a) Catastrophe reinsurance exhausted 470.4 N/A 0.0 N/A 0.0 N/A b) ADC exhausted 489.8 N/A 0.0 N/A 0.0 N/A c) No IBNR/ER from 4 May 2012 447.5 514.6 0.0 20.3 0.0 21.0 d) Adopt additional tail of 1.5% 502.2 579.8 31.0 114.5 12.4 98.6 e) 25% of Lyttelton re-allocated to Darfield 350.2 350.2 0.0 f) 33% building code on settlements 462.0 531.9 0.0 37.0 0.0 38.3 g) 1% saving on settlement 487.9 562.7 16.7 83.5 0.0 69.1 h) Remove margin clause for large claims 480.6 556.2 9.4 73.4 0.0 66.2 i) CoV + 5% 493.6 594.8 22.4 120.6 3.8 105.0 j) CoV - 5% 493.6 544.2 22.4 71.1 3.8 54.4 Lyttelton sensitivities (undiscounted outstanding claims)

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 48 The COV is selected noting that many large claims are reserved at or near policy limits. However, we have, in addition, considered the potential impact of not applying a risk margin at the 75% and 90% PoS to the large claims which, on a central estimate basis, are projected to be at or close to the applicable policy limits. This results in an adjustment to give a lower risk margin. The adjustments relate specifically to the Arts Centre, Catholic Cathedral, St Margaret College, Anglican Cathedral, Nazareth House, and Churchill Court. Our adopted gross and net risk margins are selected to be the middle point between the corresponding risk margins inclusive and exclusive of the adjustments for the limits on the large claims. The analysis of risk margins has been performed separately for each event and there is no allowance for diversification between events. Although the total gross risk margin may be less than the sum of the individual event gross risk margins, as there is a net risk margin for the Lyttelton event only the total net risk margin is therefore also the sum of the individual net risk margins. 5.7.1 Assumed coefficient of variation In selecting the CoV assumption, in light of uncertainty in the claims data from the earthquake, we have considered the gross outcomes from the scenario analysis in the previous section as a starting point. The following table resummarises the impact from favourable and unfavourable scenarios on the gross central estimate. We have not estimated or allocated a probability measure to each of the scenarios presented, however, we infer that the probability of all of the unfavourable scenarios occurring at the same time would be commensurate to a level in excess of a 90% PoS outcome. We have selected our final individual CoVs based on this high-level reasonableness check. The table below summarises the selected CoV for gross earthquake claims by event. Note that the differences in the selected CoV for each event are mainly a consideration to allow for the “passage of time” since the event date. Table 5.33 Selected CoV for Earthquake Claims Event Event date Selected CoV Darfield 4-Sep-2010 7.5% Boxing Day 26-Dec-2010 7.5% Lyttelton 22-Feb-2011 15.0% Sumner 13-Jun-2011 25.0% 23rd Dec 11 23-Dec-2011 45.0%

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 49 In arriving at the assumed CoV, we have considered at a high-level the potential impact of further adverse development in the underlying claims, which may exhaust the sums insured and/or annual policy limits on events for a particular policy or group of policies. However, we have not modelled the complexity of tracking the detailed position of each individual claim or policy. In light of this approximation, it is important to note that the assumed CoVs for earthquake claims remain highly judgemental, and further analysis of the policy limits as well as monitoring of unfolding claims experience on earthquake claims, will be required to refine the risk margin assumptions. Given the high degree of uncertainty, also the nature and close proximity of the earthquake events, we have not allowed for any benefits from diversification in the risk margins on grounds of materiality. 5.7.2 KPMG risk margins at 75% PoS The table below summarises the gross and net risk margins at 75% probability of sufficiency (PoS) for earthquake claims grouped by event. Table 5.34 Earthquake Claims Gross Risk Margins at 75% PoS ($’m) Event Gross risk margin at 75% PoS inclusive of claims at limits Gross risk margin at 75% PoS exclusive of claims at limits Selected gross risk margin at 75% PoS Darfield 9.3 8.4 8.9 Boxing Day 0.3 0.2 0.3 Lyttelton 49.9 30.0 40.0 Sumner 17.0 3.9 10.5 23rd Dec 11 2.7 Total 79.2 45.2 62.2 Table 5.35 Earthquake Claims Net Risk Margins at 75% PoS ($’m) Event Net risk margin at 75% PoS inclusive of claims at limits Net risk margin at 75% PoS exclusive of claims at limits Selected net risk margin at 75% PoS Darfield 0.0 Boxing Day 0.0 Lyttelton 49.9 30.0 40.0 Sumner 0.0 23rd Dec 11 0.0 Total 49.9 30.0 40.0 It is of note from the above table that for the Lyttelton event, as it is already projected to ultimately develop beyond the limit of the catastrophe reinsurance protection at the central estimate, the dollar gross risk margin directly translates to the dollar net risk margin. Therefore it is of note that the selected CoV for the Lyttelton event is the most significant assumption for the net margin results.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 50 5.7.3 Impact of adopting AA’s central estimate As a point of comparison, we have applied our risk margin assumptions to the AA’s central estimate of the earthquake claims. Note that only the 80% PoS risk margins are available from the AA’s results to provide a comparison. Table 5.36 Inferred Earthquake Claims Risk Margins at 80% PoS using AA’s Central Estimates and KPMG’s Risk Margins Assumptions ($’m) Event AA gross risk margin at 80% PoS Implied AA gross risk margin at 80% PoS (KPMG basis) AA net risk margin at 80% PoS Implied AA net risk margin at 80% PoS (KPMG basis, before ADC) Implied AA net risk margin at 80% PoS (KPMG basis, after ADC) Darfield 0.0 7.0 0.0 Boxing Day 0.0 0.3 0.0 Lyttelton 25.0 54.3 0.0 13.2 0.0 Sumner 10.0 23.4 0.0 23rd Dec 11 1.0 1.5 0.0 Total 36.0 86.5 0.0 13.2 0.0 The above table shows that applying KPMG’s risk margin assumptions to the AA’s central estimate will result in higher gross risk margins for all events. This is not a surprising result given KPMG’s central estimates are higher than those of the AA’s. On a net basis, notwithstanding the AA’s central estimate for the earthquake claims are below that of KPMG’s overall, adopting KPMG’s risk margin assumptions results in the Lyttelton event exhausting the limits of the catastrophe reinsurance protection at the 75% PoS both before the ADC, and a gap of about $5.8m after ADC before exhausting reinsurance cover. 5.7.4 Risk margins at 90% PoS In calibrating the risk margins at a higher probability of sufficiency (e.g. 90%), our experience suggests a simple extrapolation without adjustment of the 75% PoS model, may overstate the true level of probability of sufficiency of the resulting higher PoS risk margin. Or phrased another way, may understate the margin required. See the following section for examples of the key additional adjustments to consider. The following table extrapolates the methodology and assumptions for the 75% PoS to the higher PoS of 90%, again selecting the mid-point of the results inclusive and exclusive of the further adjustment for large claims projected at or close to limits, and the uplift implied by this extrapolation.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 51 Table 5.37 Earthquake Claims Risk Margins at 90% PoS and Uplifts from 75% PoS ($’m) Event Gross risk margin at 90% PoS Uplift from gross risk margin at 75% PoS Net risk margin at 90% PoS Uplift from net risk margin at 75% PoS Darfield 16.9 8.0 0.0 0.0 Boxing Day 0.5 0.3 0.0 0.0 Lyttelton 75.9 36.0 75.9 36.0 Sumner 19.9 9.4 0.0 0.0 23rd Dec 11 5.1 2.4 0.0 0.0 Total 118.3 56.0 75.9 36.0 As a reasonableness check, we have compared the AA’s and KPMG’s risk margins with the dollar impact from the scenario analysis. Note that the comparison is not like-with-like, as the scenario analysis considered selected locations only Table 5.38 Check of risk margins against scenario analysis for large policies ($m) Summary of impact of scenarios to gross ultimate Favourable scenarios (73.1) Unfavourable scenarios 127.1 Risk margins at selected PoS AA 80% PoS risk margin 36.0 KPMG 75% PoS risk margin 62.2 KPMG 90% PoS risk margin 118.3 From this check, we conclude that the risk margin assumptions overall do not appear to result in a 90% PoS gross risk margin that is clearly either conservative or understated. We note that ACS and RBNZ acknowledge that a 90% PoS does not consider extreme outcomes. It is important to note that the above calculation is approximate, but should provide an indication of the 90% PoS outcome, considering the impact if policy limits were to be taken into account more explicitly. There remains significant uncertainty around the possibility to claim beyond the stated limits, multiple claims, how individual policy limits may apply, and that a different allocation of earthquake claims can lead to a materially different outcome on a net basis. We therefore note, that outcomes above those illustrated could emerge.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 52 5.8 Earthquake results The following tables summarise the KPMG results for earthquake outstanding claims liabilities at the 75% and 90% probabilities of sufficiency. Table 5.39 KPMG results for outstanding claims 75% PoS ($’m) Table 5.40 KPMG results for outstanding claims 90% PoS ($’m) 5.9 Uncertainties Our assessment of ACS gross and net liabilities is subject to significant uncertainty. Some are related to the situation of insurers exposed to the Earthquake events. Others are related to the specific circumstances of ACS, for example, its run-off circumstance, reinsurances, available data and case estimating approach. 5.9.1 Earthquake events – general uncertainty Insurance liabilities relating to the Canterbury Earthquake are subject to significant uncertainty. There are many open issues that exacerbate the normal level of uncertainty in valuing insurance liabilities. These include, but are not limited to:  The limited extent of rebuilding to date;  The extent of delay that may exist before rebuilding can commence; Event Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Darfield 193.2 0.0 0.0 171.6 0.0 0.0 Boxing Day 6.6 0.4 0.4 5.9 0.4 0.4 Lyttelton 533.6 62.4 43.7 473.7 47.2 33.2 Sumner 111.2 0.8 0.8 98.8 0.8 0.8 23rd Dec 11 11.4 1.0 1.0 10.1 1.0 1.0 Total 856.1 64.5 45.9 760.1 49.4 35.3 Undiscounted Discounted Event Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Darfield 201.2 0.0 0.0 178.7 0.0 0.0 Boxing Day 6.9 0.4 0.4 6.1 0.4 0.4 Lyttelton 569.5 98.3 79.7 505.7 74.3 60.3 Sumner 120.6 0.8 0.8 107.1 0.8 0.8 23rd Dec 11 13.8 1.0 1.0 12.3 1.0 1.0 Total 912.1 100.5 81.8 809.8 76.5 62.4 Undiscounted Discounted

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 53  The potential impact of building code revisions;  Other potential legal or regulatory changes that may arise in relation to the Earthquakes (noting the materiality of the overall impact in New Zealand);  The interactions and complications arising from the series of Earthquakes that may have caused damage, impacting, for example, the number of claims and limit of ACS exposure and, for domestic policies, the amounts to be paid by the EQC;  The impact on reinsurance from the allocation of claims costs to the individual Earthquake Events;  Ongoing movement in properties due to the impact of liquefaction during the earthquakes; and  The potential for “demand surge” inflation to impact final claims costs;  The potential for further earthquakes to cause ongoing delay and complexity and commensurate increase in claims handling costs and inflation on claim amounts. The impact on ACS liabilities on the final outcomes of potentially very large claims relating to a small number of major properties insured by ACS. This impacts gross and net positions and the ability to discount liabilities noting the large cash flows that may ultimately result. A broader discussion of the marketwide uncertainties associated with the Canterbury Earthquakes can be found in Appendix C. 5.9.2 ACS specific uncertainties There are a number of key uncertainties that are specific to the circumstances of ACS, or where we note the perspectives relevant to the ACS valuation. Many of these have been highlighted in the discussions of this report. The list below summarises key ACS earthquake uncertainties. These include, but are not limited to:  Very few claims have formally settled. A significant element of the amounts paid to date relates to advance payments. The outstanding claim amount may be less indicative of the uncertainty in the estimate of outstanding claim amounts than the, greater figure, of the reported incurred position of claims that are open at 31 December 2011;  The potential impact of building code revisions noting that case estimates are currently set assuming a 67% of building code costing – whereas the ACS actuarial results adjust for this to recognise an expected 33% of building code outcome;

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 54  Interplay with policy limits on commercial covers and in particular the risks and uncertainties relating to the Margin Clause and Policy Reinstatement that may give rise to claims above policy limits, and or legal costs associated with these matters;  Allocation of costs and claims between events is a material driver of net liabilities. The impact of allocation to or from the Lyttelton event in particular may have a material impact to ACS’ final net claims position;  Significant uncertainty exists in how the reinsurance programme will respond to exchange rates and whether issues relating to this will lead to dispute and unrecoverable legal costs;  The impact, if any, on the value obtained from reinsurance due to ACS entering “run-off”, and the typically stricter approach that reinsurers take in that situation to enforcing policy terms and conditions. Noting the issues discussed relating to potential gaps or overlaps in coverage the potential for reinsurance dispute is a material uncertainty in terms of both reinsurance coverage and legal costs;  The final impact on ACS reinsurance arising from the NZ Court’s rulings to date relating to ACS reinsurance assets, (referred to as the Western Pacific Judgement) is not clear. The impact may or may not be material overall, however, this may significantly impact the claim outcome of individual policy holders. Our report does not consider the potential outcome from an individual policyholder perspective; and  Further development from domestic properties is implicit in our modelling of the earthquakes. However, a specific scenario of a major emergence of further claims, for example where these previously sat below or just below the EQC cap, has not been examined. Although liabilities on domestic properties are not a substantial proportion of the gross liabilities, the uncertainty on domestic properties, including uncertainty through the interaction with the EQC, may be material to the net liabilities.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 55 6 Non earthquake liabilities 6.1 Overview For non-earthquake claims liabilities, we have adopted the same groupings as per ACS’ current data groupings, i.e.:  Commercial Liability;  Motor;  House;  Commercial Property. The following table compares the KPMG gross and net undiscounted case estimates and central estimate results as at 31 December 2011 with the results of the AA for non-earthquake claims liabilities. The results exclude claims handling expenses. Table 6.1 Summary of Gross Central Estimate for Non-Earthquake ($’000) Gross Gross Central Est KPMG NZ $'000 Case est KPMG AA less AA Liability 163 521 561 -40 Motor 315 547 559 -12 Household* 1,259 1,776 1,444 332 Comm prop* 6,253 7,446 6,230 1,216 TOTAL 7,990 10,290 8,794 1,497

note: AA's figures approximated from discounted figures Table 6.2 Summary of Net Central Estimate for Non-Earthquake ($’000) Net Net Central Estimate KPMG NZ $'000 Case est KPMG AA less AA Liability 163 388 564 -176 Motor 231 201 243 -42 Household 624 851 661 190 Comm prop 2,449 2,611 2,971 -360 TOTAL 3,467 4,051 4,439 -388 Overall, KPMG’s projected gross estimate is above AA’s estimate, but on a net basis KPMG is below the AA results. An update of the reported positions up to April 2012 was also provided on 23 May 2012. We have not examined the updated data in detail, but presented the following table to summarise the data movements. 6

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 56 Table 6.3 Incurred Development December 2011 to April 2012 ($’000) Non-Earthquake Gross Incurred movements Dec11 to Apr12 Net Incurred movements Dec11 to Apr12 Liability 78 78 Motor 80 39 House 432 221 Comm Prop 187 -434 TOTAL 777 -96 The post-balance date incurred developments do not appear to post significant challenge to our projected IBNR/ER, and therefore we have selected the results based on data up to December 2011 without adjustment. 6.2 Approach The key points of our approach to the non-earthquake claims are as follows:  ACS provided run-off data triangles on gross and net payments and reported incurred, in yearly cohorts and development, up to December 2011;  KPMG figures are based on a high-level re-projection relying on the gross and net data provided by ACS NZ without independent verifications;  We have applied four standard actuarial techniques to gross and net claims data to project the ultimate position, namely the chain ladder (CL) method as well as the Bornhuetter-Furguson (BF) method on payments and reported incurred data separately;  In arriving at the net central estimate, given the lower materiality compared to the size and uncertainty of the earthquake losses at this examination, we have performed the projections using data net of recoveries, and did not separately project the reinsurance and non-reinsurance recoveries;  For all four classes of business, we have selected the BF Incurred method for the 2009-2011 accident years, and selected the CL Incurred method for the prior accident years, given their more advanced stage of development. These methods recognise the impact of reopened claims implicitly to the extent that reopened claims are present in historic data;  For the BF method we have also considered the ultimate loss ratios for past accident years from the chain ladder methods to inform the assumptions for the initial expected loss ratios;  For commercial property, we have also judgementally allowed for an explicit loading for IBNR (for accident years 2007 and 2009) due to a different pattern being observed and large open claim positions relative to other

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 57 years. The adjustment is a proxy to a projected case estimate method and establishes a “year-specific” IBNR without creating a load on all years.  By projecting the net position we have allowed for reinsurance and nonreinsurance recoveries implicitly. We have examined the resulting ratio of the net to gross position by accident year for earned premium, current paid and reported incurred positions, projected IBNR and projected ultimates.  We have discounted the gross and net outstanding claims component assuming an average delay to payment of 2 years. 6.3 Risk margins at 75% PoS The table below summarises the net diversified risk margins at 75% probability of sufficiency (PoS) from the perspective of non-earthquake claims in isolation of other liabilities. Table 6.4 Non-Earthquake Risk Margins at 75% PoS ($’000) KPMG (non-earthquake only) AA (non-earthquake only) Net Central Estimate Risk Margin 75% PoS Risk Margin Net Central Estimate Risk Margin 75% PoS Risk Margin Liability 388 45% 176 564 11% 62 Motor 201 45% 91 243 6% 15 House 851 16% 133 661 7% 46 Comm Prop 2,611 16% 408 2,971 14% 416 TOTAL 4,051 20% 809 4,439 12% 539 Table 6.5 Non-Earthquake Net Results at 75% PoS ($’000) Non-Earthquake Net Provision KPMG 75% PoS AA 75% PoS KPMG less AA Liability 564 626 -62 Motor 292 258 35 House 984 708 277 Comm Prop 3,019 3,387 -367 TOTAL 4,860 4,978 -118 Overall, the KPMG diversified net risk margin is approximately 20% for the portfolio. This is higher than the AA’s overall risk margin of 12%. The key sources of difference appear to be the use of company’s own data versus benchmarks and the differing recognition of the small size of the portfolios. However, notwithstanding the differences in the central estimates and risk margins, we note that KPMG’s overall net provision at 75% PoS for nonearthquake claims is in line with the AA’s net provision.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 58 6.3.1 Risk margins at 75% PoS KPMG’s high-level analysis is based on an application of the Mack method on gross/net paid and incurred data as a starting point. The coefficient of variation (CoV), a measure of observed volatility in past data, is summarised below. Given the low volume of claims data and small size of the portfolios, assumption selection is necessarily based on a high degree of judgement. Table 6.6 Selected Co-efficient of Variation Mack CoV Gross Paid Gross Incurred Net Paid Net Incurred Selected Net CoV Liability 66% 68% 74% 77% 75% Motor 22% 16% 48% 74% 75% House 34% 26% 49% 26% 26% Comm Prop 31% 8% 5% 13% 26% To allow for the effect of diversification, an overall correlation of 70% is assumed across all classes. This selection reflects the small size of the portfolios. This results in an overall diversification discount to the total risk margins of about 10%, compared to the AA’s assumption of 20%. The Normal distribution is assumed. Table 6.7 Undiversified and Diversified Risk Margins at 75% PoS Net Risk Margins Undiversified risk margin Diversified risk margin Liability 51% 45% Motor 51% 45% House 17% 16% Comm Prop 17% 16% The chart below compares the risk margins at 75% probability of sufficiency (PoS) level from our analysis (blue columns), compared with the AA’s results (red square dots). We have also collated available benchmarks by class of business (yellow diamonds). We note that the size of the benchmark portfolios are typically larger than that of ACS’, therefore care is needed when interpreting the differences.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 59 Figure 6.8 Diversified Net Risk Margins at 75% PoS 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0% 40.0% 45.0% 50.0% Liability Motor House Comm Prop diversified risk margins comparison KPMG Benchmarks AA The following observations are of note:  KPMG’s analysis suggests a higher risk margin than the AA’s results for all classes, except for Commercial Property where the risk margins are at similar levels.  The AA’s results appear to be among the lower ends of the benchmarks. On this, we note the much smaller size of the company’s portfolios compared to the benchmark portfolios, therefore adjusting the benchmarks for the difference in portfolio size may further exacerbate the divergence.  From KPMG’s analysis, the company’s own data indicates a high level of net volatility in the Liability and Motor class (and substantially above the benchmarks). This can also be impacted by the small size of these classes, with a low capacity to absorb the past volatility.  In discussions with the AA he has noted patterns of favourable patterns of run-off relative to his past central estimate figures. Noting the higher net central estimate position of the AA it is important that the combined impact of the central estimates and risk margins must be taken into account when considering these results. 6.4 Risk margins at 90% PoS and uncertainty In calibrating the risk margins at higher probability of sufficiency (e.g. 90%), our experience suggests a simple extrapolation without adjustment of the 75% PoS model, may overstate the true level of probability of sufficiency of the resulting higher PoS risk margin. Or phrased another way, may understate the margin required.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 60 Examples of the key additional adjustments to consider include:  The lack of observable large claims and/or catastrophe claims, and the extent these claims may be under-represented in the data history or open claims. Therefore an explicit allowance may be required.  Allowance for claims that exhibit a much longer delay between the insured event occurring and the notification of the claim to the insurer. We have not considered or made any allowance for the risk of claims of a latent nature. We have discussed the potential for Leaky Building related claims and sexual abuse / molestation claims with ACS. It is noted that the ACC is expected to cover the latter and the nature of the portfolio means that Leaky Building exposures are limited. In both cases ACS is not aware of any claims to date. ACS noted that reinsurances in place should limit the net impact of any one claim. However, the concern with latent claims is more an issue relating to a moderate frequency of claims emerging from many years so the net exposure may still be significant if a latent claim issue does emerge.  Scenarios pertaining to the higher probability of sufficiency often present an increased tendency for outcomes to be linked and develop adversely at the same time. This is often known as “tail dependency”.  Data and parameter error will have a more material impact when extrapolating higher probabilities of sufficiency. Tail dependency is one such example. Explicit adjustments for data and parameter error are often made when modelling higher PoS positions; typically set by testing the sensitivity of results to key assumptions. Given the time constraint in completing this examination, we have not explicitly examined or modelled each of these issues. We have allowed for a combined data and parameter error uplift factor to address these uncertainties overall, namely:  Increase the CoV used in the 75% PoS model for all classes by a multiplicative loading of 10%; and  Increasing the correlation between the four classes from 70% to 80%. The resulting 90% PoS risk margins for non-earthquake claims before and after the above illustrative adjustments are summarised in the table below.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 61 Table 6.9 Non-Earthquake Risk Margin at 90% PoS and Model Adjustments ($’000) Net Risk Margins Net Central Estimate Risk Margin at 75% PoS Risk Margin at 90% PoS Uplift from 75% PoS Adjusted Risk Margin at 90% PoS Uplift from 90% PoS Model Adj Liability 388 176 334 158 383 48 Motor 201 91 173 82 198 25 House 851 133 253 119 289 36 Comm Prop 2,611 408 775 366 886 112 TOTAL 4,051 809 1,535 726 1,756 221 6.5 Non earthquake provisions The following tables summarise the KPMG provision for non earthquake outstanding claims liabilities at the 75% and 90% probabilities of sufficiency. Table 6.10 KPMG results for outstanding claims 75% PoS ($’m) Table 6.11 KPMG results for outstanding claims 90% PoS ($’m)

Class of business Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Liability 0.8 0.6 0.6 0.7 0.5 0.5 Motor 0.8 0.3 0.3 0.8 0.3 0.3 House 2.1 1.0 1.0 2.0 0.9 0.9 Comm Prop 8.6 3.0 3.0 8.2 2.9 2.9 Total 12.2 4.9 4.9 11.6 4.6 4.6 Undiscounted 75% PoS Discounted 75% PoS Class of business Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Liability 1.0 0.8 0.8 1.0 0.7 0.7 Motor 1.1 0.4 0.4 1.0 0.4 0.4 House 2.4 1.1 1.1 2.3 1.1 1.1 Comm Prop 10.0 3.5 3.5 9.5 3.3 3.3 Total 14.5 5.8 5.8 13.8 5.5 5.5 Undiscounted 90% PoS Discounted 90% PoS

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 62 7 Claims Handling Expenses 7.1 Overview Based on our assessment of the expense budget supplied by ACS, our claims handling expense (“CHE”) results are as follows: Table 7.1 Claims handling expense liabilities ($’000s)

7.2 Approach Our approach to estimating the claims handling expense liabilities has been as follows:  Review of the internal expense budget over four years with questions and clarifications put to ACS management.  Restatement of the internal analysis for a number of areas based on our review and clarification.  The expenses have then been projected beyond 2015, using the initial projection as a base and with reference to the schedule of services.  Bringing the projected expense cash-flows to a present value at a discount rate consistent with our review of the claims liabilities.  This analysis has been completed at the central estimate and 90% PoS. We have not examined this issue from the perspective of what an independent third party might charge to take on the management of the claims run-off. This is an equally valid approach to the issue, but it has not been possible to seek quotes or indications in the timeframe available to us. 7.3 Key assumptions The key assumptions made in determining the central estimate claims handling provision are set out below:  Noting the run-off status of ACS, in the context of this examination, CHE will include an estimate of all run-off administration expenses that do not sit within the outstanding claims liabilities. This differs from the situation of an insurer writing new business. This treatment is consistent with our Inflated and discounted Inflated undiscounted Central estimate 7,289 7,796 75% PoS estimate 8,214 8,819 90% PoS estimate 9,737 10,851 7

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 63 experience of many other run-off insurers, where the view is often taken that all expenses are incurred to support the management and run-off of claims. This is particularly the case where there are limited invested assets generating a return for shareholders. In the case where surplus assets are being managed to provide a return for shareholders we have observed insurers maintaining a blend between the standard approach and ‘full cost’ approach as followed here.  Salary expenses and employee costs have been based on the information supplied by ACS, and we have adopted the provided view as to the duration of each role and expected reductions in staffing during the period of projection.  We have adopted the rental costs without amendment, which assume a stepdown in rental from 2012, and a further stepdown in 2014. Motor vehicle leases have also been adopted without amendment, noting a stepdown in 2014 and 2015. We have not reviewed rental agreements directly. Any charges for breaking agreements early would need to be provided for in addition to the costs projected.  We have adopted the ACS estimates for professional costs without adjustment. We have added the costs estimated in the Schedule of Services in relation to operational tasks to be provided and charged by Ansvar going forward.  Other areas such as printing, stationery, information technology etc have been based on ACS projections without amendment.  The cost of additional staff or contractors that might be employed to manage earthquake claims are assumed to be recoverable from reinsurance programmes. Such costs sit within our claims liabilities. We have modelled those costs implicitly within our projection of the gross Earthquake claims. No allowance for such costs sits within the expense provision results.  Beyond 2015, we have assumed that the operations continue on for a further six years (ten years in total), based on our experience that run-off situations can be protracted, with a period of about ten years before a scheme of arrangement quite typical. The costs are assumed to reach a “fixed cost minimum” during this period, based on a continuation of “professional costs” only. This assumes that Ansvar or another organisation provides operational services in accordance with the schedule of services. We have not inflated these costs, assuming that inflationary impacts are offset by small reductions each year in the costs being charged to ACS.  At the end of the ten year period, additional costs of $518,000 are assumed, based on additional costs of winding up the operation of approximately

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 64 $350,000 in current values, inflated by ten years. This is a high level estimate only and in time should be reviewed in more detail. 7.4 Comparison to alternate estimates The below table sets out our estimated claims handling expense relative to those provided by ACS NZ and the AA. (We note the estimates are not directly comparable as discussed below.) Table 7.2 Comparison of CHE allowances ($’000) The key sources of difference between the three estimates are as follows:  Relative to the ACS internal estimate, we have included non-claims expenses, added additional professional fees representing charges that will no longer be absorbed by Ansvar, and extended the term of the projection.  The AA estimate was based on a loading of 16% of the net central estimate with an additional loading for short term additional expenses. As the estimate was not based on a detailed consideration of line by line expenses and expectations of how that would change over the run-off period, an attribution of difference to specific assumptions is not possible. 7.5 Risk margins In forming a view as to the 75% PoS and 90% PoS for the claims handling expense results, we have considered variations in two key variables – the length of time to run-off, and the degree to which costs can be scaled down during the run-off. There are a number of other uncertainties which have not been quantified at this time, and are discussed in the following subsection. We have assumed at the 75% PoS:  A total of 10 years of run-off, unchanged from the central estimate.  Costs from 2016 to 2018 continue at the same level as for 2015, before reverting to the “fixed cost minimum” for the remainder of the run-off. For a 1 in 4 adverse outcome, it appeared reasonable to assume delays in making alternate arrangements, rather than a more protracted duration of claims. At the 90% PoS, we have assumed: Inflated and discounted Inflated undiscounted KPMG Central estimate 7,289 7,796 ACS internal estimate 2,512 2,783 AA estimate 1,985 n/a

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 65  A total of 15 years of run-off.  Costs from 2016 to 2018 continue at the same level as for 2015, before reverting to the “fixed cost minimum” for the remainder of the run-off. The 90% PoS is expected to involve ongoing uncertainty and complications in settling earthquake claims. Therefore, we have adopted an extended duration of claims and resulting run-off costs. 7.6 Uncertainties While the claims handling expense loading is usually a relatively minor part of an outstanding claims provision, in the case of a run-off, the uncertainties of these loadings become more pronounced. The following areas have been identified as the main sources of uncertainty:  ACS run-off costs will be subject to commercial negotiation and choices made to put in place final arrangements. It is not clear at this time what those arrangements will be. While we might expect that the current claims handling operations continue until such time that the remaining claims volume suggest that outsourcing would be more economical, the timing and associated cost of this cross-over point is an unknown.  Duration will be a clear driver of overall costs. Regardless of the arrangements made, the total duration of run-off is uncertain. While management have some expectation that the bulk of the liability (being earthquake claims) will be settled quickly, it cannot be ruled out that the difficulties and delays associated with assessment and settlement of claims will lead to a more protracted run-off than expected. Market commentary at this time suggests ongoing delays in the settlement of the earthquake claims.  The degree to which costs can be reduced over time is uncertain. If claims run-off more slowly than expected, the speed at which positions are made redundant may be slower than expected, with assumed step downs in staff and rental costs in particular continuing at higher levels than anticipated. Late in the process, we were advised that the ability to step down the rental costs immediately is in doubt, and dependent on negotiations with the landlord. If we assumed no step down in rental for the next two years (i.e. until the next renewal date, at which point it might be assumed that the agreement could be renegotiated or alternate premises found for lower cost) the impact would be an increase of about $0.2m in the claims handling provision above our results.  We understand that a number of past expenses have been met by Ecclesiastical. There has been no suggestion that recovery of these will be

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 66 sought, but this is an additional area of uncertainty. We have not further considered this possibility in our estimates. It would provide greater certainty in this regards were ACS to seek and receive a formal waiver of any rights other parties may have to recover ex-gratia support provided to ACS to date.  We have not explicitly included additional costs introduced by the run-off status e.g. professional costs in negotiating new arrangements. We observe that the actual costs to date in 2012 are higher than budget in most areas, particularly for professional costs but also for areas such as rental, travel and other items. Although we have adopted a higher level of professional costs in our provision, in accordance with the schedule of services, we would not expect that our current allowance would be sufficient to cover the current level of expenses on an ongoing basis. This observations highlights the uncertainties in this provision.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 67 8 Aggregate results 8.1 Outstanding claims provision The following tables summarise the KPMG results for outstanding claims liabilities at the central estimate, 75% and 90% probabilities of sufficiency at 31 December 2011. Table 8.1 KPMG results for outstanding claims central estimate ($’m) Table 8.2 KPMG results for outstanding claims 75% PoS ($’m) Table 8.3 KPMG results for outstanding claims 90% PoS ($’m) The following table shows the ACS adopted discounted central estimate and the estimate at 75% PoS as at 31 December 2011. Segment Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Earthquake 793.8 24.5 5.9 704.8 19.3 5.2 Non earthquake 10.3 4.1 4.1 9.8 3.9 3.9 Expenses 7.8 7.3 Total 811.9 36.4 17.8 721.9 30.4 16.4 Undiscounted Discounted Segment Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Earthquake 856.1 64.5 45.9 760.1 49.4 35.3 Non earthquake 12.2 4.9 4.9 11.6 4.6 4.6 Expenses 8.8 8.2 Total 877.1 78.2 59.6 779.9 62.2 48.2 Undiscounted Discounted Segment Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Gross outstanding claims Net outstanding claims before ADC Net outstanding claims after ADC Earthquake 912.1 100.5 81.8 809.8 76.5 62.4 Non earthquake 14.5 5.8 5.8 13.8 5.5 5.5 Expenses 10.9 9.7 Total 937.4 117.1 98.5 833.3 91.7 77.7 Undiscounted Discounted 8

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 68 Table 8.4 ACS adopted provision for outstanding claims ($’m) Segment Gross outstanding claims Net outstanding claims before ADC Gross outstanding claims Net outstanding claims before ADC Earthquake 624.3 2.1 653.2 2.1 Non earthquake 9.2 4.0 10.3 4.8 Expenses 2.0 2.0 2.2 2.2 Total 635.4 8.1 665.7 9.2 Discounted central estimate Discounted 75% PoS

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 69 A Data A.1 Data provided We have been provided with the following documents and spreadsheets: No. Document 1 11511_2010 NZ PER RISK and CAT (Signed).pdf 2 11512_2010 CAT (Signed).pdf 3 11514_2010 PROPERTY PER RISK (Signed).pdf 4 11915_2010 SURPLUS (Signed).pdf 5 11917_2010 PROP QS MAPFRE CIAR (Signed).pdf 6 11918_2010 PROP QS HANNOVER (Signed).pdf 7 11919_2010 PROP QS MUNICH (Signed).pdf 8 11919_2011 PROPERTY QUOTA SHARE (Signed).pdf 9 11919_2011 PROPERTY QUOTA SHARE Endorsement iro Munich Re (Signed).pdf 10 11921_2010 SECOND SURPLUS (Signed).pdf 11 13477_2011 PROPERTY PER RISK (Signed).pdf 12 13478_2011 NZ CAT BUY-DOWN (Signed).pdf 13 13479_2011 CAT (Signed).pdf 14 13480_2011 SURPLUS (Signed).pdf 15 13744_2011 CAT Back-up 20110304-20111231 (Signed).pdf 16 13833_2011 CAT Back-up 20110406-20111231 (Signed).pdf 17 1997 Motor and General Third Party Liability Excess of Loss 19970101.19971231 18 2009 2010 Combined PDBI policy - Signed 08.01.2010.pdf 19 2010 2011 Combined PDBI policy - Signed 30.12.2010.pdf 20 2010 Engineering Quota Share Addendum No 2.pdf 21 2010 Global Cat Reinstatement Premium Letter (Signed).pdf 22 2010 Global Cat Treaty Wording (Signed).pdf 23 2010 NZ Contingency Agreement (Signed).pdf 24 2010 RI Arrangement Statement (APRA) v2 July.pdf 25 2010-2011 signed wording.pdf 26 2011 Engineering Quota Share and Surplus Signed Wording.pdf 27 2011 GLOBAL CAT LAYER 1 (Signed).pdf 28 2011 GLOBAL CAT LAYER 2 (Signed).pdf 29 2011 Reinsurance Arrangement Statement - July.pdf 30 2011 SURPLUS TREATY 2 (Signed).pdf 31 20111012113748856.pdf 32 ACS Claim Experts.pdf 33 Actual expenses in ACS Jan - Mar 2012.xlsx 34 AIB - BI policy .pdf 35 All non EQ claims above NZ$200k as at end of April.xlsx 36 Allocatio Update end March - final final Friday (2) (3) (2).doc 37 ALLOCATION OF LOSS OVER EVENTS - end-March Guideline (3) (2) (2).doc 38 Analysis of Transactional Claims Data as at 20111231.xlsx 39 Anglican Churches - AIB Placing & Terms 2010-2011.pdf 40 Anglican Churches - AIB Placing & Terms 2011-2012.pdf A

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 70 No. Document 41 Ansvar_NZ Risk margins Dec11.xlsx 42 Ansvar_NZ Tillinghast_Dec_2011.xlsx 43 Ansvar_NZ Trowbridge_Dec_2011.xlsx 44 Ansvar_NZ_Dec_2011_House.xlsx 45 Ansvar_NZ_Dec_2011_Liability.xlsx 46 Ansvar_NZ_Dec_2011_Motor.xlsx 47 Ansvar_NZ_Dec_2011_Property.xlsx 48 App 1 Financial summary.doc 49 App 2 Small Claims Report Commentary.doc 50 App 3 Small Claims Report Bordereau.doc 51 App 4 Medium Claims Repoirt Schedule.xlsx 52 App 5 Medium Claims Report Commentary.docx 53 App 6 Large Claims Report Top 20.xls 54 App 7 Large Claims Report Schedule £1m+.xlsx 55 App 8 Large Claims Report Commentary.docx 56 B08752011BGY5092 ADDITIONAL CAT PURCHASE Ren Re.pdf 57 B08752011BGY5093 ADDITIONAL CAT PURCHASE Hannover Re.pdf 58 B08752011BGY5094 ADDITIONAL CAT PURCHASE Hiscox.pdf 59 B08752011BGY5095 ADDITIONAL CAT PURCHASE HCC.pdf 60 B08752011BGY5096 ADDITIONAL CAT PURCHASE MS Frontier.pdf 61 B08752011E2F5146 ADDITIONAL CAT PURCHASE TMRE 62 BI Policy - Jan 09.pdf 63 C33 - MD Ex Fire - WEF 01.04.09.pdf 64 C34 Fire IV - Incl Deductible Schedule - WEF 01.04.09.pdf 65 C46 - Fire Excess of indemnity Policy - WEF 01.04.09.pdf 66 CAL Report 30-Mar-12.pdf 67 Cat 21 - High Level Summary 20120504.pdf 68 Cat 29 - High Level Summary 20120504.pdf 69 Cat 32 - High Level Summary 20120504.pdf 70 Cat_21_-High_Level_Summary_20111231.pdf 71 Cat_29-High_Level_Summary_20111231.pdf 72 Cat_32-_High_Level_Summary_20111231[1].pdf 73 Catastrophe Update Australia New Zealand 201204.xlsx 74 Catholic Cathedral (1).pdf 75 Catholic Cathedral (2).pdf 76 Catholic Cathedral (3).pdf 77 Catholic Diocese of Christchurch - Broker Placing & Terms 2009-2010.pdf 78 Catholic Diocese of Christchurch - Broker Placing & Terms 2010-2011.pdf 79 Christchurch Earthquakes IBNR December 2011.docx 80 Christchurch Earthquakes IBNR March 2012.pdf 81 Christchurch IBNR 2012-01-27.xlsx 82 Christchurch IBNR 2012-03-30.xlsx 83 Copy of Ansvar reserve review Stage 2 with history (DD insert).xlsx 84 Copy of Top 25 as at 27 01 12.xlsx 85 COSA agreement (signed).pdf 86 Currency_Risk.xlsx 87 Data - Transactions for EQ claims 20120430.xlsx 88 Durham Street Methodist.pdf 89 EIG REPORT Final.doc 90 Engagement Letter signed by Andrew Moon.pdf

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 71 No. Document 91 FINAL Expense Budget template - NZ Runoff.xlsx 92 Fire Excess of Indemnity Policy - Jan 09.pdf 93 Fire Indemnity Only Policy - Jan 09.pdf 94 FW ALLOCATION OF LOSS OVER EVENTS - UPDATE.msg 95 img-Z16125436-0001.pdf 96 Largest Claims Listing for KPMG (20120523).xlsx 97 Level 2 report Dec2010.pdf 98 Marsh e-Dit wording v2.pdf 99 MD - Excluding Fire Policy - 1 Mar 09.pdf 100 media release_905190.pdf 101 Methodist Churches - Broker Placing & Terms 2010-2011.pdf 102 Methodist Churches - Broker Placing & Terms 2011-2012.pdf 103 Monitoring Event Reporting (end of April 2012).xlsx 104 NZ Annual Report - FY2011 FINAL (20120426).pdf 105 NZ Lease.pdf 106 NZ Non - Life Insurance Solvency Standard - Run Off (Working File).xlsx 107 NZ_Claims_Triangulations_Q4_(With Cats Removed).xlsx 108 Op MR 270212 - allocation for adjusters.pdf 109 Open Liability Files April 2012 for CSofA (2012-04-30) as text.xlsx 110 PCANZ - Broker Placing & Terms 2009-2010.pdf 111 PCANZ - Broker Placing & Terms 2010-2011.pdf 112 Position of Largest Claims Locations From Analysis of Transactional Claims Data as at 20111231.xlsx 113 Presentation_to_Reinsurers_March_2012_update_final_pptx.pptx 114 R_Ansvar_31 Dec 2010.pdf 115 R_Ansvar_31 Dec 2011_Final.pdf 116 R_Ansvar_NZ Dec 2010.pdf 117 R_Ansvar_NZ Dec 2011_Final.pdf 118 RBNZ Canterbury Earthquake Claims April 2012.pdf 119 RBNZ Canterbury Earthquake Claims February 2012.pdf 120 RBNZ Canterbury Earthquake Claims January 2012.pdf 121 RBNZ Canterbury Earthquake Claims March 2012.pdf 122 RBNZ_Canterbury_Earthquake_Claims_December_2011.pdf 123 Schedule of Services provided to ACS Draft.docx 124 Screen Shot of E-mail from AB on Currency Risk.docx 125 Signed Copy of ADC 2012-05-08-180532.pdf 126 St Margarets College - Broker Placing & Terms 2009-2010.pdf 127 St Margarets College - Broker Placing & Terms 2010-2011.pdf 128 Summary for KPMG of Non Earthquake Claims Jan-Apr 2012.xlsx 129 The Arts Centre - Broker Placing & Terms 2009-2010.pdf 130 The Arts Centre - Schedule 2010-2011.pdf 131 The Arts Centre.pdf 132 Transparency No.14.pdf 133 Transparency No.15.pdf 134 Transparency No.16.pdf 135 Transparency No.17.pdf 136 Transparency NZ Earthquake Update February 2012.pdf 137 Transparency Special Edition November 2011.pdf 138 Transparency Special Edition November No2 2011.pdf 139 Transparency Special Edition September 2011.pdf

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 72 B Lyttelton earthquake incurred cost development B.1 Introduction As described in Section 5.4.5 we performed a high level check of the allowance for incurred cost development from 4 May 2012 for the Lyttelton event by fitting statistical curves to the incurred cost development experience of both the Darfield and Lyttelton events to 30 April 2012. The fitted curves to the Lyttelton and Darfield experience are shown in the following graphs. The range of tail development factors implied from the Lyttelton curves is shown in the following table along with the adopted tail development factor (which was set with further reference to the development to date of the Darfield event without further adjustment). 1.00 1.05 1.10 1.15 1.20 1.25 1.30 1.35 1.40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Chain ladder incurred factor Development month Fitting curves to selected data - Lyttelton Graph data Weibull Exponential Power Inverse Power 1.00 1.05 1.10 1.15 1.20 1.25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Chain l;adder incurred development factor Development month Fitting curves to selected data - Darfield Graph data Weibull Exponential Power Inverse Power Lyttelton tail development factor Weibull 1.035 Exponential 1.026 Power 1.029 Inverse Power 1.069 Adopted 1.043 B

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 73 The adopted development of 1.043 is within the range of results from the fitted curves. Our curve fitting for Darfield suggests that there may be an additional 1.5% to 3.1% of incurred cost development on this event compared with no future development assumed when applying Darfield experience to Lyttelton recognising only actual development to date.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 74 C General discussion on uncertainty – Earthquakes C.1 Introduction This note starts with an acknowledgement that the size of the insured losses associated with the Canterbury Earthquake is unprecedented for New Zealand. There are many issues associated with estimating gross liabilities. In most circumstances the final liability outcome for insurers could differ significantly from that implied by the current reserves. We have identified and discussed a number of specific items that drive uncertainty in this note. The discussion touches on uncertainties that may involve both, or one of, commercial and domestic insurances. The uncertainties listed may not consider all uncertainties; further issues may emerge as the situation develops and individual insurers and insureds may have specific uncertainties that require separate consideration. The note provides no legal opinion and is not intended to provide any specific advice. The Canterbury earthquakes and aftershocks are exceptional in both claim numbers and claim costs. There have been five significant events in Canterbury:  A magnitude 7.1 earthquake on 4 September 2010 (Darfield);  A magnitude 4.9 earthquake on 26 December 2010 (Boxing Day);  A magnitude 6.3 earthquake on 22 February 2011 (Lyttelton);  A magnitude 6.4 earthquake on 13 June 2011 (Sumner); and  A magnitude 6.0 earthquake on 23 December 2011 (23rd December). These multiple events significantly impact on the insurance industry across New Zealand. C.2 Uncertainty There is still substantial uncertainty in the ultimate costs of the NZ Canterbury earthquakes. A number of key uncertainties remain within the valuation around the appropriateness of the liabilities. Key uncertainties at this time are noted below (the items are covered in no particular order): C

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 75 C.2.1 Potential for further aftershocks A long-term forecast by GeoNet scientists (in collaboration with the Earthquake Commission and GNS Science) indicates the potential for further aftershocks from the Canterbury events anywhere in the entire Canterbury region aftershock zone. As time passes the likelihood that further aftershocks occur becomes smaller, but any further significant earthquakes that do occur will change this view. The impact on existing outstanding claims and costs inflation is unclear. C.2.2 Lack of data The lack of earthquake experience from previous events and lack of relevant industry data adds difficulty and uncertainty when assessing and benchmarking the appropriateness of assumptions and methodology adopted in reserving. C.2.3 Depopulation A cordon area was established around the worst damaged area of the CBD following the Earthquake of 22 February 2011 and is still in place. Businesses within the cordon have not been able to trade as usual and as such at the extreme may not have generated any turnover. If insurers follow the ‘Orient Express’ decision, effectively arguing provision of limited indemnity only, then the insurance industry is likely to be faced with litigation and highly adverse publicity. (The evacuation of New Orleans as a result of serious damage from Hurricane Katrina in September 2005 was a special circumstance for which the historical Standard Turnover had to be adjusted to zero because but for the damage it would not have generated turnover. Consequently only limited indemnity was provided under policy extensions). C.2.4 Uninhabitable land There is a high probability that areas are declared uninhabitable due to instability of the land following the earthquakes and aftershocks. This would directly increase exposure to the Canterbury losses where a full loss of property results. Similar to other issues surrounding the Canterbury events, this may result in legal uncertainties for insurers and litigation on this issue may follow. The Government made an offer on 22 June 2011 to buy 5,100 Christchurch homes at 2007 rateable values. Under the terms of the offer, residents had nine months in which to consider the Government’s purchase offer. With respect to the Government offer for properties in the red zone, insured home owners had two options:

Sell the entire property to the Crown at the 2007 rateable value; or
Sell the land only to the Crown and continue to deal with their insurance company regarding the property damage.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 76 Owners who have spent heavily on their dwellings since 2007 could sell just the land to the NZ Government, and claim back the value of the building from their insurer. This optionality continues to complicate the modelling of the gross claims cost. C.2.5 TC3 and zoning references Zones as a result of the Canterbury events are constantly under review. It remains possible that current non-red zones may be re-classed as red-zones in the future. The red zone issue may impact on the payment pattern of claims. The further aftershocks and earthquakes, including those of 23 December, extend the uncertainty around the extent of red zoning. Local media shows that public interest is following this issue directly and pressure for additional red zoning may be correlated to the duration of ongoing shocks and the degree, in particular, of liquefaction in residential areas from subsequent events. In addition to impacting the final overall costs this further complicates aspects of allocation as previously discussed. C.2.6 General difficulties around consequential loss and how this is measured under these circumstances Given that the earthquakes have destroyed many buildings and the sums insured have not always proved adequate to rebuild or rebuilding is not yet possible, this has created the potential for large business interruption insurance and consequential loss claims. Given the potential size of these claims and that these are previously un-encountered circumstances there is significant uncertainty surrounding these claims. C.2.7 Processing delays The potential exists that claim processing demands impacts claims information. As insurers’ workloads have increased significantly, claims size might increase as time available to pro-actively manage the claims may have decreased, increasing the risk of claims leakage. Additionally, given the delay experienced across the industry in loss adjustors and engineers assessing all properties, this delay means it may not be possible to obtain accurate assessments at this point in time. The earthquake complexities and work volumes for insurers across the industry has resulted in a backlog of work for earthquake teams, loss adjusters and insurers. As a result, the volume of claims that needs to be dealt with will create resource pressures, delays for insurers and may diminish the accuracy/ reserving value of claims data.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 77 C.2.8 Inconsistencies in case estimates / claims handling process Actuarial processes often assume consistency or stability in claims handling processes and application between claims and over time. The high demands of the earthquakes in terms of claims volumes and the mix of handling old events and the claims of new events would suggest that claim handling patterns will be subject to disruption and distortion. This will add uncertainty to any projection in this circumstance. We also note that, unlike more typical situations, claims handlers may not have the time to revisit and revise case estimates to the degree planned or otherwise expected – particularly given the series of events. Therefore caution will be required when examining stable reported incurred positions to confirm whether this is indicative of a stabilising situation or a pause in case file review and data updating. C.2.9 EQC payouts and other potential legal or regulatory change that may arise in relation to the Earthquakes The Earthquake Commission and Insurance Council went to New Zealand’s High Court to get a declaratory judgement on how the New Zealand legislation around the Earthquake Commission (‘EQC’) Act should be interpreted and the potential for regulatory and legislative developments around such issues remain material. The High Court ruled on 2 September 2011 that the $100,000 limit on houses and the $20,000 limit on contents applied for each earthquake, and not only the total loss. That is, EQC cover is automatically reinstated for each event. At the time of this file note the EQC has not lodged an appeal against the decision. However, the potential for an appeal is an example of the potential legal change that may arise in relation to the earthquakes. C.2.10 The potential impact of building code revision Following the Canterbury events of 2010 and 2011, the Christchurch City Council is likely to require repairs and reinstatements to restore damaged buildings to a more stringent building standard. A specific aspect is in respect of repairs and the threshold of seismic strength required relative to the codes for new buildings. The target level articulated by the Christchurch City Council (“CCC”) is 67% of the building code; whereas the apparent trigger requiring a remediation plan is 33%. This would have implications on insurers’ claim costs and as such case estimates are likely to remain volatile for the Canterbury claims and any claims arising from subsequent aftershocks. The issue relating to repair standards creates uncertainty on repair costs and also raises uncertainty around legal costs (where dispute with either the CCC or insured arises) and the potential for inflation to add further costs during delays due to disagreement. Revisions to building codes could have an impact on rebuilding costs. Furthermore, in a general sense the examination of building codes may cause delay in repairs, rebuilding, negotiations and claim settlement.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 78 C.2.11 Demand surge inflation This continues to be a risk particularly if there is a continuation of aftershocks that impact the rebuild and repair works being undertaken and/or are still required; which may lead to demand surge inflation. For domestic properties, Fletcher Construction will manage the repairing costs from $10,000 up to $100,000 and will not work with sub-contractors. This may slow settlement but should serve to ease ‘demand surge’ inflation. However, commercial insurers will handle claims above EQC’s cap ($100,000 plus GST) and building costs are estimated to have increased to $1,500 per m2, which is approximately a 25% to 40% increase. It is still unclear about the impact on claims costs post the 22 February 2011 event given the time taken to scope and cost these damages however the impact of inflated labour rates and inflated material rates could be material. C.2.12 Building delay We are aware that there is a considerable building delay associated with properties damaged in the Canterbury events as little to no building work has taken place to date. This creates significant uncertainty surrounding the payment pattern and discounting to be adopted. Building work commenced in February 2012 and so some feedback to valuation assumptions will continue to emerge over time. The extent of delay that may exist before rebuilding can commence adds significant uncertainty. C.2.13 EQC interaction For domestic properties, a major uncertainty surrounding the events is the interaction with the EQC. This may impact reporting patterns and the final payments by the EQC will impact the overall sharing of market losses. This may be a specific issue for reserving where market loss models are used and EQC payments therefore need to be offset from the liabilities that result. The potential EQC liabilities are significant but are also highly uncertain at this stage. The interactions and complications arising from the series of Earthquakes may have caused damage impacting for example the number of claims and limit of an insurer’s exposure and the amounts to be paid by the EQC. Accordingly, the amount of EQC recoveries will have a material impact. Specific interactions that may affect reporting patterns include: situations where policyholders are awaiting EQC inspection (and as a result non EQC covered aspects such as paths are not reported); and ‘reporting fatigue’ i.e., where policyholders, particularly those with other open claims reported to the insurer, are awaiting EQC assessment prior to updating the commercial insurer. Reporting fatigue is a particular uncertainty for the 30 June 2011 and 23 December 2011 events.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 79 C.2.14 Additional uncertainty associated with EQC payments There are additional uncertainties that arise in relation to the EQC payment given that the EQC pays for the first $100k on an indemnity basis. If an insurance policy cover is on a replacement value basis this can lead to some additional uncertainty regarding who pays for what. For example, if the replacement value of a house is agreed to be $150k (on whichever agreed building code standard) but the indemnity value is assessed at $90k then the insurer might have only expected to be have to pay $50k (ie excess of EQC $100k) but due to the different covers provided would potentially have a liability of $60k. C.2.15 Impact of possible underinsurance If a risk affected by more than one event is underinsured the total claim cost for that risk could exceed the sum insured. This could result in numerous claims equal to the sum insured, even if the damage to the building is only repaired once. We understand that the application of the ‘average’ principle (whereby the claim is reduced in circumstances of under-insurance) is not a common practice in New Zealand. C.2.16 Claims handling, loss adjuster and legal expenses There is still substantial uncertainty surrounding the ongoing duration and final resolution of the Canterbury events. Complex issues are arising that will require the involvement of expert investigations and reports, and take the investment of time to resolve. With ongoing delays in rebuilding inflation continues to impact both the potential final claims costs and the extent and level of expenses incurred. The longer market issues remain open and cause delay in settlement the greater the potential for claims handling and other costs to increase. C.2.17 Fraud There has been market commentary noting identified cases of fraudulent claims. The potential exists for undetected fraud to impact liabilities. C.2.18 Commercial settlement of large claims Particularly on commercial policies it is likely that the industry will seek settlement of protracted cases through commercial negotiation. This may provide savings in some cases dependent on the extent to which case estimates recognise such outcomes.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 80 C.2.19 The final impact on reinsurance arising from the NZ Court’s rulings to date relating to reinsurance assets, (referred to as the Western Pacific Judgement) Under this New Zealand High Court judgement there is a statutory charge over reinsurance assets payable to the reinsured which is enforceable by Canterbury policy holders. This decision may still go to appeal if any creditors (Nonearthquake or those not protected by specific reinsurance) are prepared to fund it. Specific uncertainties arising from the judgement may be material to some insurers, for example, credit risks, the impact to final reinsurance allocations and resulting recoveries, and the additional legal complexity and potential costs that may arise. C.2.20 Reinsurance risks including disputes, credit risk and potential gaps in cover We note that a combination of circumstances creates risk that reinsurance does not respond as expected. Those circumstances include, but may not be limited to: the allocation of costs and claims to events, the size of the events testing the wording and operation of very high layers, interactions between local and group programs in some cases particularly at higher layers. C.2.21 Ex gratia settlements Where ex gratia payments have been made or advance payments not confirmed against policy terms the potential exists that reinsurance will not respond to those amounts.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 81 C.2.22 The impact on policy limits, costs and reinsurance from the allocation of claims costs to the individual Earthquake Events The damage from the earthquake events has been concentrated around the Christchurch region and this has added to the complexity of assigning the portion of costs attributable to each event for insurance and reinsurance purposes. The apportionment of claims on properties damaged by more than one event creates considerable complexity in loss adjusting. The resolution on some policies will come down to negotiation and the assessment and input will require highly specialised input from engineers and others that will add to costs. This in turn adds significantly to the complexity of determining reinsurance recoveries. Noting the many separate earthquakes a major uncertainty in final net results is the allocation of claims costs between the events. Across the industry it is at present difficult to ascertain when damage belongs to one event over any other. With each additional event it becomes increasingly more difficult to treat the events as separate occurrences. In particular, these difficulties have been compounded in cases where a claim has occurred but prior to a claims assessor visiting the site, a subsequent event has occurred, causing further damage. The difficulty associated with allocation claims costs between events will impact reinsurance recoveries for the individual earthquake events. C.2.23 Market/Industry resolution The concept of a market / industry resolution of apportionment has been raised. Such a resolution would help facilitate claim settlement and avoid major dispute between the insurers, reinsurers and the EQC but is uncertain at this stage. C.2.24 Exchange rate movements Many goods, building materials, expertise and services needed to deal with the earthquakes will be sourced from overseas. Exchange rate fluctuations will impact those costs. The likely term of rebuilding increases the uncertainty in allowing for this. C.2.25 Development periods For the earthquake claims longer than normal development periods are expected, which leads to greater uncertainty when projecting further into the future.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 82 D Actuarial techniques for non earthquake D.1 Introduction There is an inevitable delay between the occurrence and payment of a claim. There are two main components of the delay:  Delay in reporting claims; and  Delay in settling reported claims. An insurer must therefore estimate its liability to make future claim payments. There exist several approaches used by actuaries to estimate outstanding claims liabilities. This appendix summarises the techniques used in the examination of the non earthquake liabilities. In particular, the steps involved in each of the methods are described below. D.2 Chain Ladder Chain ladder approaches are based upon the assumption that from one development period to the next, the relative change in the variable observed is similar across all incident years. Chain ladder approaches involve the following steps:  Period–to–period development factors are calculated for each incident period by comparing the observed variable from one period to the next.  These are used to project the incomplete event periods to ultimate figures. It requires the earliest period of origin to be fully run-off or at least that the final outcome for that year can be estimated with confidence. D.3 Bornhuetter Ferguson (BF) on Paid or Incurred The Bornhuetter Ferguson (BF) method is based on a blend of actual claims experience, alongside an initial expectation of the ultimate experience pertaining to the incurred but not reported (IBNR) component. The BF method utilises information from both the claims and exposure data, hence is predominately used for the recent event periods where actual claims experience to date has been limited. The Bornhuetter Ferguson involves the following steps:  The expected IBNR proportion for each event period is derived as 1 minus the inverse of the cumulative development pattern (of the variable being estimated) selected from a chain ladder method.  The prior loss ratio assumption for the recent event periods is selected. D

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 83  The IBNR component for each event period is then calculated by multiplying the corresponding IBNR proportion to the respective exposure amount and the prior loss ratio assumption.  The reported payments or incurred costs for each event period are then added to the corresponding IBNR component to derive the ultimate claims cost. The prior loss ratio assumption is generally based on an average of the projected ultimates from earlier event periods. These estimates are checked for reasonableness using our knowledge of similar portfolios in the market.

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 84 E Rate of currency exchange (net impacts on Lyttelton earthquake loss) Lyttleton EQ - KPMG examination of currency movements and impacts on RI layers +/- 5% movements Impact relative to current adopted net incurred position (prior to allowance for ADC) Current Interpretations Scenario 1a) - roe 2.10 Scenario 1b) - roe 1.89 Scenario 2a) - roe 2.10 Scenario 2b) - roe 1.89 Scenario 3 Adopted Coverage change only Coverage change only Coverage and limit change Coverage and limit change Coverage, limit change and overlap dispute Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) ACS - CE Gap - Gap - Gap - Gap 4.7 Gap - Gap - $NZDm Overlap - Overlap - Overlap - Overlap - Overlap - Overlap 17.7 391.6 Coverage - Coverage - Coverage - Coverage - Coverage - Coverage - (net of prop RI) Limit - Limit - Limit - Limit - Limit - Limit - Total - Total - Total - Total 4.7 Total 0.0 Total 17.7 Remaining cover (G2) 38.2 Remaining cover (G2) 48.2 Remaining cover (G2) 28.3 Remaining cover (G2) 52.9 Remaining cover (G2) 10.6 Remaining cover (G2) 10.6 Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) KPMG - CE Gap - Gap - Gap - Gap 4.7 Gap - Gap - $NZDm Overlap - Overlap - Overlap - Overlap - Overlap - Overlap 17.7 452.2 Coverage - Coverage -10.0 Coverage 10.0 Coverage -10.0 Coverage 10.0 Coverage 10.0 (net of prop RI) Limit - Limit - Limit - Limit -4.7 Limit 17.7 Limit 17.7 Total - Total -10.0 Total 10.0 Total -10.0 Total 27.7 Total 45.4 Remaining cover (G2) 0.0 Remaining cover (G2) 0.0 Remaining cover (G2) 0.0 Remaining cover (G2) 0.0 Remaining cover (G2) 0.0 Remaining cover (G2) 0.0 150.6 49.8 149.4 ACS - CE $388.5m KPMG - CE $452.2m Current Interpretations NZD/GBP 1.99 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 429.8m 150.6 52.3 156.9 ACS - CE $388.5m KPMG - CE $452.2m Scenario 1a) NZD/GBP 2.09 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 439.8m 150.6 52.3 156.9 ACS - CE $388.5m KPMG - CE $452.2m Scenario 2a) NZD/GBP 2.09 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 444.5m 235.3m Coverage Gap 150.6 47.3 141.9 ACS - CE $388.5m KPMG - CE $452.2m Scenario 1b) NZD/GBP 1.89 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 419.9m 150.6 47.3 141.9 ACS - CE $388.5m KPMG - CE $452.2m Scenario 2b) NZD/GBP 1.89 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 402.2m 212.9m Coverage Overlap 150.6 47.3 141.9 ACS - CE $388.5m KPMG - CE $452.2m Scenario 3) NZD/GBP 1.89 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 402.2m 212.9m Coverage Overlap E

ACS (NZ) Limited Examination of reserve uncertainty at 31 December 2011 85 Lyttleton EQ - KPMG examination of currency movements and impacts on RI layers +/- 10% movements Impact relative to current adopted net incurred position (prior to allowance for ADC) Current Interpretations Scenario 1a) - roe 2.10 Scenario 1b) - roe 1.89 Scenario 2a) - roe 2.10 Scenario 2b) - roe 1.89 Scenario 3 Adopted Coverage change only Coverage change only Coverage and limit change Coverage and limit change Coverage, limit change and overlap dispute Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) ACS - CE Gap - Gap - Gap - Gap 15.9 Gap - Gap - $NZDm Overlap - Overlap - Overlap - Overlap - Overlap - Overlap 28.9 391.6 Coverage - Coverage - Coverage - Coverage - Coverage - Coverage - (net of prop RI) Limit - Limit - Limit - Limit - Limit 10.6 Limit 10.6 Total - Total - Total - Total 15.9 Total 10.6 Total 39.5 Remaining cover (G2) 38.2 Remaining cover (G2) 58.2 Remaining cover (G2) 18.3 Remaining cover (G2) 74.1 Remaining cover (G2) 0.0 Remaining cover (G2) 0.0 Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) Net Impact ($NZDm) KPMG - CE Gap - Gap - Gap - Gap 15.9 Gap - Gap - $NZDm Overlap - Overlap - Overlap - Overlap - Overlap - Overlap 28.9 452.2 Coverage - Coverage (19.9) Coverage 19.9 Coverage (19.9) Coverage 19.9 Coverage 19.9 (net of prop RI) Limit - Limit - Limit - Limit (2.5) Limit 28.9 Limit 28.9 Total - Total (19.9) Total 19.9 Total (6.5) Total 48.8 Total 77.8 Remaining cover (G2) 0.0 Remaining cover (G2) 0.0 Remaining cover (G2) 0.0 Remaining cover (G2) 13.4 Remaining cover (G2) 0.0 Remaining cover (G2) 0.0 150.6 49.8 149.4 ACS - CE $388.5m KPMG - CE $452.2m Current Interpretations NZD/GBP 1.99 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 429.8m 150.6 54.8 164.4 ACS - CE $388.5m KPMG - CE $452.2m Scenario 1a) NZD/GBP 2.19 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 449.8m 150.6 54.8 164.4 ACS - CE $388.5m KPMG - CE $452.2m Scenario 2a) NZD/GBP 2.19 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 465.7m 246.5m 150.6 44.8 134.5 ACS - CE $388.5m KPMG - CE $452.2m Scenario 1b) NZD/GBP 1.79 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 409.9m 150.6 44.8 134.5 ACS - CE $388.5m KPMG - CE $452.2m Scenario 2b) NZD/GBP 1.79 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 381.0m 201.7m 150.6 44.8 134.5 ACS - CE $388.5m KPMG - CE $452.2m Scenario 3) NZD/GBP 1.79 CAT Layer 1-4 CAT Layer 5 Global 1 Global 2 230.6m 80m 381.0m 201.7m Coverage Gap Coverage Overlap Coverage Overlap