To qualify for hedge accounting, certain criteria must be met including documenting the nature and purpose of the hedge and performing regular testing over its effectiveness.

Due to the complex nature of the hedge accounting rules this is often an area of significant risk for banks.

In our prior report to you, we noted that audit testing had identified a number of instances where hedging was applied, but the accounting rules had not been adequately met. This resulted in the remediation of existing controls and the implementation of new controls in the last quarter of 2015.

In light of the prior year matters, we determined this to be an area of significant audit risk.

We discussed with the GAC during the year, the progress made by management in the implementation of the new controls.

During December 2016, management in France identified a further issue with an established hedging relationship, which resulted in a partial discontinuation of the hedge. A discussion was held with the GAC regarding both the root cause of the matter, the period in which the adjustment should be recognized and over which controls that had not operated effectively. As indicated by the above matter, not all of the hedge accounting controls operated effectively in the year. The exceptions noted were limited to France.

• A sample of new hedging relationships was examined and the relationships assessed to determine if they had been appropriately designated. This included consideration of the hedge objectives and specific compliance with IFRS.
• Management’s hedge effectiveness reviews, and the measurement and recording of hedge ineffectiveness, were tested for a sample of hedge relationships.
• Understood and tested controls over the documentation and review of the hedge relationships and their initial and ongoing effectiveness.
• Additional substantive audit procedures were performed over the partial discontinuation of the hedging relationship in France.

The Group had goodwill of $15.5bn from a number of historical acquisitions across cash-generating units (CGUs).

An assessment is required annually to establish whether a CGU’s goodwill should continue to be recognized, or if any impairment exists. At each reporting period, management is also required to identify any potential indicators, and to perform an impairment assessment if any are identified.

The impairment assessment calculation used for the tests were based on estimated future cash flows for each CGU discounted at an appropriate cost of equity rate. HSBC used its Annual Operating Plan as the basis for the first five years of cash flows and then extrapolated returns into perpetuity using a terminal growth factor. Cost of equity rates were based on the investment rates used within the global business and approved by the Board.

The estimation of future cash flows and the level to which they are discounted is inherently uncertain and requires significant judgement. The extent of judgement and the size of the goodwill, resulted in this matter being identified as an area of focus.

We discussed the conclusions of goodwill assessments with the GAC when they considered the annual test and at each reporting period when they considered whether indicators of impairment existed.

At 30 June, indicators of impairment were identified in GPB – Europe and GB&M – Europe, which prompted a full impairment test for these two CGUs. This led to an impairment of $800m of goodwill in GPB – Europe.

The annual assessment was performed in the third quarter based on 1 July data. This assessment concluded no further impairment of goodwill was required. The discussions with the GAC focused on the key assumptions, both individually and when combined together. During these discussions, management confirmed their view that the forecasts for each CGU remained appropriate.

Subsequently, we discussed with the GAC the impact of changing segments on the CGUs, particularly the decision to change the CGU associated with GB&M, as disclosed on page 240. The discussion also covered the decision not to change the other CGUs.

At 31 December, management identified further indicators of impairment in the GPB – Europe CGU. A retest was performed and it was concluded that all remaining goodwill for the CGU should be written off. In reaching this conclusion, a view was taken on the future performance of the business, and the risk associated with these forecasts. We discussed the approach and adjustments with the GAC.

• PwC’s independent valuation experts critically assessed the discount rate and terminal growth rates used in the discounted cash flow models. The focus was on the methodology used to estimate discount rates of a CGU; and whether the use of the nominal GDP growth rates was the most appropriate in estimating the terminal growth rates into perpetuity for each CGU.
• The calculations used in the model were re-performed to check accuracy and the key inputs in the model were agreed to underlying sources.
• Management’s future cash flow forecasts used in the model were assessed by:
• testing that the forecasts agreed to the latest Annual Operating Plan approved by management;
• considering current year performance against plan and the reasons for any deviation, and key drivers or strategies underlying the plan. These were discussed with management of the Global Businesses for each sensitive CGU;
• reviewing the historical achievement of the Annual Operating Plan. Given the uncertainties in forecasting, this identified that forecasts have been less accurate for prior periods, and we considered if this was appropriately factored into the valuation model;
• independent sensitivity analysis was performed to identify any further CGUs with a risk of impairment. The reasonableness of management’s threshold of sensitive CGUs was assessed; and
• the appropriateness of disclosures made in relation to goodwill was also considered.

HSBC, like other global banking institutions, is exposed to a significant number of open legal cases and regulatory investigations in a number of its markets. Given the business is geographically dispersed; the same matter could be subject to investigation in multiple jurisdictions.

Provisions of $2.4bn have been established to account for legal settlements and regulatory fines. The most significant provisions relate to tax-related investigations and foreign exchange market manipulation.

There are a number of legal and regulatory matters for which no provision has been established, as discussed on page 257.

There is an inherent risk that legal exposures are not identified and considered for financial reporting purposes on a timely basis. Importantly, the decision to recognize a provision and the basis of measurement are judgmental.

Group Legal provided to each GAC meeting an update on the status of legal cases. These updates considered whether all related litigation or investigations about a specific matter had been identified.

Material matters were discussed during each meeting and the need for changes to provisions considered. We participated in these discussions, including consideration of whether any constructive obligation had arisen in individual cases.

• Open legal cases were discussed with Group Legal and in certain instances we obtained and reviewed the relevant regulatory and litigation documents in order to assess the facts and circumstances.
• The range of reasonably possible outcomes was considered for material provisions to independently assess the appropriateness of the judgement made by HSBC.
• The disclosures of legal exposures and provisions were assessed for completeness and accuracy.

HSBC has $39.8bn of pension liabilities as a result of defined benefit pension schemes.

The calculation of these pension liabilities is complex and HSBC uses third party actuaries to provide support in the process to ensure appropriate expertise is applied to the calculation. The use of these actuaries also increases the risk of error as data is passed to third parties for analysis and calculation purposes.

Considering all of these factors, our initial assessment of the risk of misstatement did not identify pension liabilities as an area of significant focus as there was no history of error and the pension funds were in surplus reducing the risk of fraud.

During the year management identified errors in the transfer and use of data by third parties for one of the schemes in the US. As a result of this error, we reconsidered our assessment of the audit risk surrounding pension liability valuations and increased our scope of testing in this area.

The change in the assessment of risk was discussed and agreed with the GAC in December 2016.

We focused our testing response and our discussions with GAC on the largest schemes in the UK and US, which made up 84% of the overall liability balance at 31 December 2016. Our increased testing was focused on the transfer and use of data by third parties to form the calculation.

• Controls over the third party vendors were tested and the third party assurance reports covering controls operated by the vendors were reviewed.
• The output from external actuaries was inspected and an independent view was formed of key actuarial assumptions.
• Data used by the actuary in the calculation and the system to ledger reconciliations was independently tested.

Impairment allowances represent management’s best estimate of the losses incurred within the loan portfolios at the balance sheet date. They are calculated on a collective basis for portfolios of loans of a similar nature and on an individual basis for significant loans. The calculation of both collective and individual impairment allowances is inherently judgmental for any bank.

Collective impairment allowances are calculated using models which approximate the impact of current economic and credit conditions on large portfolios of loans. The inputs to these models are based on historical loss experience with judgement applied to determine the assumptions used to calculate impairment. Model overlays are applied where data driven parameters or calculations are not considered representative of current risks or conditions of the loan portfolios.

For specific impairments, judgement is required to has occurred and then to estimate the expected future cash flows related to that loan. The audit was focused on impairment due to the materiality of the loan balances and associated impairment allowances and the subjective nature of the impairment calculation.

The largest loan portfolios are in Europe and Asia with the more significant impairment allowances being in Europe, North America and Latin America.

At each GAC and Group Risk Committee meeting there was a discussion on changes to risk factors and other inputs within the collective allowance models as well as discussions on individually significant loan impairments.

We discussed a number of specific risks that changed or emerged during the course of the year, including the impact of the UK’s decision to leave the European Union; the economic slowdown in China; volatility in the oil price which impacted individual credits; and the increased macroeconomic uncertainty in North America. In all of these cases we discussed the performance of the existing credit exposures, and the potential need for changes to modelling approaches.

We also discussed any significant changes made to the inputs or models impacting the collective impairment allowance as well as changes in the control environment. These included key assumptions over the retail impairment models and improvements in the way higher risk loans were identified and escalated within the organization.

The controls management has established to support their collective and specific impairment calculations were tested.

• For collective impairment, this included controls over the appropriateness of models used to calculate the charge, the process of determining key assumptions and the identification of loans to be included within the calculation.

• For specific impairment charges on individual loans, this included controls over the monitoring of the credit watch list, credit file review processes, approval of external collateral valuation vendors and review controls over the approval of significant individual impairments.

• For collective allowances, the appropriateness of the modelling policy and methodology used for material portfolios was independently assessed by reference to the accounting standards and market practices. Model calculations were tested through re-performance and code review. Specifically with respect to the collective impairment models for the retail portfolios, we reviewed the enhancements made to the models and methodology to ensure they were appropriate.

• The appropriateness of management’s judgements was also independently considered in respect of calculation methodologies, segmentation, economic factors and judgmental overlays, the period of historical loss rates used, loss emergence periods, cure rates for impaired loans, and the valuation of recovery assets and collateral.

• For specific allowances, the appropriateness of provisioning methodologies and policies was independently assessed for a sample of loans across the portfolio selected on the basis of risk. An independent view was formed on the levels of provisions booked based on the detailed loan and counterparty information in the credit file. Calculations within a sample of discounted cash flow models were re-performed.

All banks are highly dependent on technology due to the significant number of transactions that are processed daily. The audit approach relies extensively on automated controls and therefore procedures are designed to test access and control over IT systems.

As reported in the prior year, controls over individuals’ access rights to operating systems, applications and data used in the financial reporting process required improvement. Access rights are important as they ensure that changes to applications and data are authorized and made in an appropriate manner. Ensuring staff only have appropriate access, and that the access is monitored, are key controls to mitigate the potential for fraud or error as a result of a change to an application or underlying data.

A number of enhancements to the control environment have been made by management since our last report but some controls were not fully remediated by the year end and we continued to assess the risk of a material misstatement arising from access to technology as significant for the audit.

The original approach discussed with the GAC was based on the control enhancements proposed by management, and involved the testing of new and improved control processes. This was supplemented with other control and substantive procedures required for the periods of the year when the changes would not yet have been effective. As the timing of the enhancements to controls changed during the year, we reflected this in the nature and extent of testing, and our final approach was discussed with the GAC in October.

At each GAC meeting, there was a discussion on the status of the control remediation programme, work performed by management and results of testing performed.

Access rights were tested over the various aspects of technology relied upon for financial reporting. Specifically, the audit tested that:

• new access requests for joiners were properly reviewed and authorized;
• application user access rights were removed on a timely basis when an individual left or moved role;
• access rights to applications were periodically monitored for appropriateness; and
• highly privileged access was restricted to appropriate personnel.

Other areas that were independently assessed included password policies, security configurations, controls over changes to applications and databases and that business users, developers and production support did not have access to change applications, the operating system or databases in the production environment.

As a consequence of the deficiencies identified a range of other procedures were performed:

• where inappropriate access was identified, we understood the nature of the access, and, where possible, obtained additional evidence on the appropriateness of the activities performed;
• additional substantive testing was performed on specific year-end reconciliations (i.e. custodian, bank account and suspense account reconciliations) and confirmations with external counterparties;
• testing was performed on other compensating controls such as business performance reviews; and
• a list of users with access to systems was obtained and manually compared to other access lists where segregation of duties was deemed to be of higher risk, for example users having access to both core banking and payments systems.

Refer to page 72 (Audit Committee Report), page 186 (Accounting Policies), and page 261 (Note 52).

The Group enters into derivative contracts in order to manage and hedge risks such as interest rate and foreign exchange rate risk. These arrangements create accounting mismatches which are addressed through hedge accounting, predominantly fair value hedges or cash flow hedges.

The application of hedge accounting and ensuring hedge effectiveness can be highly judgmental and operationally cumbersome, and requires close monitoring from management.

We understood and tested key controls over the designation and ongoing management of hedge accounting relationships, including testing of hedge effectiveness as well as the controls around the preparation and review of hedging strategy and related documentation prior to the implementation of new hedges. We found the key controls were designed, implemented and operated effectively, and therefore we determined that we could place reliance on these controls for the purposes of our audit.

We examined hedge documentation to assess whether the documentation complied with all IAS 39 requirements. We tested key year-end reconciliations between underlying source systems and spreadsheets used to manage hedging models, including testing of hedging capacity after considering the impact of structural reform, designation of hedges and the measurement and recording of hedge effectiveness adjustments. In monitoring hedging effectiveness against stresses, we noted that despite significant market uncertainty and volatility during the year, all significant hedge accounting relationships continued to be effective. We also tested a sample of manual adjustments posted to hedge reserves relating to hedge ineffectiveness arising in cash flow hedging models. We specifically considered the re-designations that were required as a result of the reclassification of gilts from held-to-maturity to available-for-sale. We found that hedge accounting methodology was appropriately applied.

Refer to page 72 (Audit Committee Report), page 186 (Accounting Policies), and page 228 (Note 37 and Critical Accounting Estimates and Judgements).

The recognition of a deferred tax asset in respect of tax losses is permitted only to the extent that it is probable that future taxable profits will be available to utilize the tax losses carried forward.

When considering the availability of future taxable profits, judgement is required when assessing projections of future taxable income which are based on approved business plans/forecasts.

The allocation of forecast profits is also judgmental when considering the utilization of the deferred tax assets in the separate legal entities where the assets reside. Furthermore, there have recently been changes to legislation in relation to the level of profits which banking entities may offset with brought forward tax losses.

We understood and tested key controls over the production and approval of the forecast taxable profits used to support the recognition of various deferred tax assets. We found the key controls were designed, implemented and operated effectively, and therefore we were able to place reliance on these controls for the purposes of our audit.

We assessed whether the forecast profits were appropriate by challenging both the underlying and economic assumptions, focusing on those directly impacting the adjusted profit figures, for example interest rates and gross domestic product. We used our independent benchmarking data to benchmark a number of the economic assumptions to external data sources where possible, and also assessed previous forecasts for evidence of bias.

We also reviewed management’s basis for allocating forecast profits between legal entities, challenging significant assumptions and using our experience of the Group’s activities.

We have evaluated the impact of recent tax law changes on the calculation of the Group’s deferred tax balances, including confirming that the loss restriction rules, the banking surcharge and the restriction of deductions for certain customer compensation payments have been correctly applied. In addition, we ensured that management’s forecasts considered the impacts of structural reform.

We found that the both the utilization period and the carrying value of the deferred tax asset together with the related disclosures are reasonable.

Significant provisions have been made in respect of conduct matters in recent years, reflecting customer redress payments, operational costs and regulatory fines.

The most significant provisions have related to past sales of payment protection insurance policies, arrears handling activities, packaged bank accounts and insurance products of the German branch of Clerical Medical Investment Group Ltd (now Scottish Widows Ltd).

Given the number and volume of products sold by the Group historically, and the continued regulatory and public focus on the banking industry, there is a continuing risk that new conduct issues will emerge. Therefore, there is a financial reporting risk that such emerging risks and exposures are not appropriately identified and provided for.

In relation to known issues, the measurement of provisions is highly judgmental and involves the use of several management assumptions including volume of future complaints and related redress costs.

Furthermore, there is a risk that these known and emerging issues are not appropriately disclosed in the financial statements.

We understood and tested the key controls and management’s processes for:

–identifying emerging conduct risk exposures and assessing whether provisions or disclosures were necessary; and

–the calculation and review of conduct provisions including governance processes and approvals of model assumptions and outputs.

We found these key controls were designed, implemented and operated effectively and therefore we determined that we could place reliance on these key controls for the purposes of our audit.

In addition we have performed the following substantive procedures:

We met with Divisional and Group management to understand the emerging and potential issues that they had identified.

We independently assessed emerging and potential areas where exposures might have arisen based upon our knowledge and experience of emerging industry issues and the regulatory environment. We used this to challenge the completeness of the issues identified by management and whether a provision was required.

We understood customer complaints received, and assessed the trends. We used this analysis to understand whether there were indicators of more systemic issues being present for which provisions or disclosures may have needed to be made in the financial statements.

We read the Group’s correspondence with the Financial Conduct Authority and Prudential Regulation Authority and discussed the output of any meetings held. We met on a trilateral basis with the Financial Conduct Authority, Prudential Regulation Authority and the Chair of the Audit Committee. We also met on a bilateral basis with each regulator.

We read the minutes of key governance meetings including those of the Board, and of various management committees, as well as attending Audit Committee and Board Risk Committee meetings. We also understood the key activities of the Conduct and Compliance function.

The majority of our detailed audit work was on the significant conduct provisions in relation to past sales of payment protection insurance policies, arrears handling activities, packaged bank accounts and insurance products in the German branch of Clerical Medical Investment Group Ltd (now Scottish Widows Ltd). We also examined other areas of compensation payments made to customers.

For significant provisions made, we understood and challenged the provisioning methodologies and underlying assumptions used by management. For example, we challenged the basis that management used for forecasting the number of PPI complaints that will be received in the future. We also considered regulatory developments and management’s interactions with regulators.

For those assumptions based on historic information, we challenged whether this was appropriate for future experience and challenged the appropriateness of any adjustments made by management. We also independently performed sensitivity analysis on the key assumptions.

Given the inherent uncertainty in the calculation of conduct provisions and their judgmental nature, we evaluated the disclosures made in the financial statements. In particular, we focused on challenging management that the disclosures were sufficiently clear in highlighting the exposures that remain, significant uncertainties that exist in respect of the provisions and the sensitivity of the provisions to changes in the underlying assumptions.

No additional material conduct issues that would require either provision or disclosure in the financial statements were identified as a result of the audit work performed.

Refer to page 72 (Audit Committee Report), and page 242 (Note 48).

The Group has a number of open tax matters, for which management is required to make certain judgements as to the likely outturn for the purposes of calculating the Group’s tax liabilities.

The most significant uncertainties at present relate to a claim for group relief of group losses incurred in its former Irish banking subsidiary and the tax treatment of certain costs arising from the divestment of TSB Banking Group plc.

We understood and tested key controls surrounding the governance procedures in evaluating such uncertain exposures as well as performed an assessment over the experience of management in evaluating these exposures. We found the key controls were designed, implemented and operated effectively, and therefore we were able to place reliance on these controls for the purposes of our audit.

We examined and challenged the analyses performed by management which set out the basis for their judgements in respect of the material tax exposures identified, together with relevant supporting evidence such as correspondence with tax authorities and legal opinions obtained. We used our understanding of the business and also read correspondence with tax authorities to challenge the completeness of identified exposures and the need for provisions.

We made our own assessment of the likelihood of the tax exposures occurring based on our knowledge of tax legislation and applicable precedent. In making our assessment we considered the range of interpretation of the applicable tax legislation in the relevant jurisdictions. We also evaluated the calculation of the exposures and agreed these to the financial statements.

We assessed whether the extent of the disclosures made, in particular, in relation to contingent liabilities and judgements was appropriate.

We found management’s judgements in respect of the Group’s positions on uncertain tax items to be reasonable.

Refer to page 72 (Audit Committee Report), page 186 (Accounting Policies) and page 210 (Note 21 and Critical Accounting Estimates and Judgements).

The determination of impairment provisions remains a highly subjective and judgmental area. Furthermore, the Group is subject to significant regulatory scrutiny with respect to provisioning levels.

Our work covered impairment of loans and advances to customers within Retail, Consumer Finance and Commercial Banking.

We assessed the use of historic experience to estimate impairment events which have been incurred but not reported and to derive estimates of future cash flows.

We also focused on the calculation of required impairment provisions, including the use of models, and in particular the critical assumptions used in those models and calculations.

The models rely upon the accuracy of underlying data including the delinquency status of the borrower. Our work therefore focused on agreed customer treatments such as forbearance, including the presentation of such information in the financial statements, in order to validate the accuracy of such delinquency status markers.

Management also applies adjustments, or overlays, where they believe the data driven parameters and calculations are not appropriate, either due to emerging trends or models not capturing the risks in the loan portfolio. An example of this is an overlay for the current low interest rates which management apply on top of the impairment model output in Retail. These overlays require significant judgement and are therefore a main area of focus.

• the identification of impairment events and classification of forborne loans;
• the governance over the impairment processes, including controls over unauthorized modifications to the models and the re-assessment by management that impairment models are still calibrated in a way which is appropriate for the impairment risks in the Group’s loan portfolios;
• the transfer of data between underlying source systems and the impairment models that the Group operates; and
• the review and approval process that management have in place for the outputs of the Group’s impairment models, and the adjustments and overlays that are applied to modelled outputs.

We found these key controls were designed, implemented and operated effectively, and therefore we determined that we could place reliance on these key controls for the purposes of our audit.

In addition we have performed the following substantive procedures:

We tested the completeness and accuracy of relevant data from underlying systems and data warehouses that is used in those models.

We understood and critically assessed the models used. Modelling assumptions and parameters, such as probability of default, are based on historic data. We challenged whether historic experience was representative of current circumstances and of the recent losses incurred in the portfolios. Where changes had been made in model parameters and assumptions, we understood the reasons why changes had taken place and used our industry knowledge and experience to evaluate the appropriateness of such changes. We performed sensitivity analysis, and for certain portfolios, re-performed the provision calculation using our own independent models. We understood and corroborated any material differences identified. We also performed testing to obtain evidence over the existence and valuation of collateral.

In considering the appropriateness of provisions, we have assessed whether higher risk concentrations (e.g. past term interest only loans, forborne loans) have been appropriately considered and captured in the modeled provision, and where not, whether overlays to modelled calculations appropriately reflected those risks. We challenged management over the completeness of overlays and to provide objective evidence to support the adjustments made to the modelled provision and performed substantive testing over certain overlays.

Based on the evidence obtained we found that the impairment model assumptions, data used within the models and overlays to modelled outputs are reasonable and therefore concluded that provisions are appropriate.

We understood and evaluated the processes for identifying impairment events within the loan portfolios, as well as the impairment assessment processes for loans within the Business Support Unit and run-off portfolio.

We critically assessed the criteria for determining whether an impairment event had occurred and therefore whether there was a requirement to calculate an impairment provision. We tested a sample of performing loans with characteristics that might imply an impairment indicator existed (e.g. a customer experiencing financial difficulty or in breach of covenant) as well as an additional sample of haphazardly selected performing loans to assess whether these loans had any impairment indicators that management had not identified.

For a sample of individually impaired loans we understood the latest developments at the borrower and the basis of measuring the impairment provisions and considered whether key judgements were appropriate given the borrowers’ circumstances. We also re-performed management’s impairment calculation, testing key inputs including the expected future cash flows, discount rates and the valuation of collateral held. Our testing of collateral valuation specifically considered whether valuations were up to date, consistent with the strategy being followed in respect of the particular borrower and assessed the appropriateness and sensitivities of key assumptions. We back-tested previous provisions by comparing the gains or losses crystallized when impaired loans were sold or exited.

For the collective unidentified impaired provision, which reflects losses incurred but not yet identified, we tested the completeness and accuracy of the underlying loan information used in the impairment models by agreeing details to the Group’s source systems as well as re-performing the calculation of the modelled provision. For the key inputs and assumptions in the model, we obtained and tested objective evidence that supported their appropriateness. For overlays to the modelled output, we challenged management to provide objective evidence that the overlays were appropriate.

We also considered whether certain recent events and macro-economic factors (e.g. continued volatility and uncertainty around commodity prices, sterling exchange rate movements and further reduction in interest rates) had been appropriately considered and captured.

Based on the procedures performed and evidence obtained, we found management’s assumptions to be reasonable and therefore consider provisions to be appropriate.

Refer to page 72 (Audit Committee Report), page 186 (Accounting Policies) and page 226 (Note 36 and Critical Accounting Estimates and Judgements).

The valuations of the retirement benefit schemes are calculated with reference to a number of actuarial assumptions including discount rate, rate of inflation and mortality rates.

Because of the size of the schemes, small changes in these assumptions can have a significant impact on the financial statements.

We understood and tested key controls over the completeness and accuracy of data extracted and supplied to the Group’s actuary, which is used in the valuation of the Group’s defined benefit obligations. We also tested the controls for determining the actuarial assumptions and the approval of those assumptions by senior management.

We found the key controls were designed, implemented and operated effectively, and therefore we determined that we could place reliance on these controls for the purposes of our audit.

We engaged our actuarial specialists and met with management and their actuary to understand the judgements made in determining key economic assumptions used in the calculation of the liability. We assessed the reasonableness of those assumptions by comparing to our own independently determined benchmarks and concluded that the assumptions used by management were appropriate.

We tested the consensus and employee data used in calculating the obligation. Where material, we also considered the treatment of curtailments, settlements, past service costs and measurements, and any other amendments made to obligations during the year.

Based on the evidence obtained, we found that the data and assumptions used by management in the actuarial valuations for pension obligations are within a range we consider to be reasonable.

We also read and assessed the disclosures made in the financial statements, including disclosures of the assumptions, and found them to be appropriate.

The recognition and carrying value of deferred tax assets, goodwill and, in the parent company accounts, investments in subsidiaries are based on estimates of future profitability which require significant management judgement. At year end the Group had reported Goodwill of £5.6 billion (2015: £5.6 billion) and deferred tax assets of £1.8 billion (2015: £2.6 billion). The parent company has reported investments in subsidiaries of £44.6 billion (2015 – £52.1 billion).

In testing for impairment, the Group estimates the value in use of its cash generating units. Key judgements in determining the value in use include:

• Revenue forecasts impacted by business and strategic changes underway and the changing competitive environment.
• Cost forecasts impacted by the Group’s transformation programme and re-organization.
• Key assumptions used in the recoverability and valuation assessments (discount rates, growth rates, macroeconomic assumptions, etc.).
• Interpretation of recent changes to tax rates and laws for deferred tax assets.
• Assumptions regarding the economic consequences of the EU referendum and other political developments over an extended period.

We tested the design and operating effectiveness of the Group’s key controls around the preparation and review of budgets and forecasts. We tested the controls over the value in use model including the significant assumptions, inputs, calculations, methodologies and judgements.

With the support of our valuation specialists, we tested whether key macroeconomic assumptions used in the Group’s forecasting process were in line with our expectations. We assessed how these forecasts impacted the carrying value of deferred tax, goodwill and investments.

With the support of our taxation specialists, we assessed the estimate of future taxable profits to calculate the level of deferred tax assets recognized on the balance sheet including the time horizon used for recoverability of losses and other temporary differences.

We assessed how management considered alternative assumptions and performed sensitivity analysis on the assumptions used. We considered how key events, such as banking Structural Reform and the EU referendum impacted management’s estimates. We performed our own scenario analysis for certain assumptions we considered could have a significant impact on the results of the impairment tests.

We evaluated how the discount rates and long term growth rates used by management compared to peer practice. We also tested how previous management forecasts compared to actual results to evaluate the accuracy of the forecasting process.

In addition, we attended and observed management meetings where key judgements were discussed, including the value in use model and the carrying value of deferred tax assets. We also reviewed Board and Executive Committee minutes to assess the effectiveness of management’s review process and the appropriateness of the conclusions reached.

We are satisfied that the carrying values of deferred tax assets, goodwill and, in the parent company accounts, investments in subsidiaries are reasonable and in compliance with IFRS. The recoverable amounts exceed the carrying amounts when key assumptions are stressed.

We highlighted the results of our sensitivity analysis of the carrying values to key assumptions in the forecasts including the long term growth rate, discount rate and factors impacting the underlying level of profitability both at a Group level and for individual franchises. We noted the inherent uncertainty of the five year forecasts and the difficulty predicting revenue and costs over this period, particularly with respect to the impact of banking Structural Reform and the economic consequences of the EU referendum and other political developments over an extended period. We also noted the independent review of the forecasts undertaken by the Risk function and management’s challenge of the carrying values and key assumptions in relevant executive committees of the Group.

We considered the key assumptions impacting the forecast recovery period for material deferred tax assets and the extension of the recovery period for its subsidiary, National Westminster Bank Plc, from 5 to 6 years given the underlying profitability of this business.

Report of the Group Audit Committee (page 71) Accounting policies (page 297)
Note 15 (page 348) and Note 22 (page 354) of the financial statements, and Note 7 (page 396) of the Parent company financial statements.

The continued litigious environment and the heightened regulatory scrutiny gives rise to a high level of judgement in determining appropriate provisions and disclosures. At the year end, the Group has reported £12.8 billion (2015: £7.4 billion) of provisions for liabilities and charges, including £11.0 billion for conduct and litigation claims.

The most significant items at year end were:

• RMBS – A provision of £3.4 billion was recorded in 2016 in addition to the £3.4 billion already recorded for a total provision of £6.8 billion at 31 December 2016.
• UK shareholder – After mediation in July 2016, an agreement was reached with four of the five groups of claimants.
• PPI – Provision increased during the year from £1.0 billion to £1.3 billion to reflect an extension of the period in which claims could be made.
• FCA review of RBS’s treatment of SMEs – The Bank raised a provision of £400 million for costs associated with a new complaints review process and the automatic refund of complex fees.
• Business formerly known as Williams & Glyn – a provision of £750 million was recorded following the revised proposal presented in February 2017 by HM Treasury to the European Commission for RBS to meet its remaining State Aid commitments by alternative means.

These and other legacy matters have historically resulted in significant charges being taken to reflect the Group’s best estimate of costs relating to these areas.

The appropriateness of provisions for these matters is also impacted by developments, for example the settlement of RMBS claims by other banks, the partial settlement of claims made by certain shareholders with respect to the UK 2008 rights issue shareholder litigation and the judgement from Plevin v Paragon Personal Finance Ltd with respect to PPI.

Management judgement is needed to assess whether an obligation exists and a provision should be recorded at 31 December 2016 in accordance with the accounting criteria. This includes determining if

• It is likely that an economic outflow such as a payment will occur; and
• The amount of the payment (or other economic outflow) can be estimated reliably.

The measurement of the provision is based on the best estimate of the expenditure required to settle the present obligation.

The most significant areas of judgement are:

• Completeness of provisions recognized: judgement in the determination of whether an outflow in respect of identified material conduct matters are probable or can be estimated reliably.
• Measurement of provisions recognized: Integrity and completeness of data, and the appropriateness of assumptions and judgements used in the estimation of material provisions.
• Adequacy of disclosures of contingent liabilities

We tested the design and operating effectiveness of the Group’s key controls over the identification, estimation and monitoring of provisions considering the potential for management override of controls. The controls tested included those implemented by management to identify and monitor claims, assess the completeness and accuracy of data used to estimate provisions, and ensure the accuracy and completeness of disclosures made in accordance with accounting standards.

Audit procedures over the completeness of provisions recognized

We examined regulatory correspondence to assess developments in key cases, the Group’s complaint handling reports and RBS litigation reports to identify potentially material cases. We also confirmed the details of significant cases directly with the Group’s external legal representatives. For cases where a provision was not recognized, we considered whether the outcome was probable and reliably estimable in accordance with the accounting criteria.

Audit procedures over the measurement of provisions recognized.

We assessed the provisions recorded by testing the data and assumptions used in the calculation of the provisions recorded, including expected claim rates, legal costs, and the timing of settlement. This included
comparing the assumptions to peers which were involved in similar matters.

We also considered historical data and whether this supported current estimates. Our conduct risk specialists were involved to consider how the Group’s provisions compare to the latest industry developments in key matters.

In addition, we attended and observed key management meetings and reviewed the minutes of the legal provision committee meetings to assess the effectiveness of management’s review process and the appropriateness of the conclusions reached.

Audit procedures over disclosures

We also assessed the disclosure provided on conduct, litigation and regulatory provisions to determine whether it complied with accounting standards.

We are satisfied that the Group’s provisions for conduct, litigation and regulatory matters, customer remediation and claims are within a reasonable range and recognized in accordance with IFRS. We did not identify any material unrecorded provisions.

We highlighted the following matters:

• In considering the total U.S. RMBS provision of £6.8 billion, management and the Board considered other industry settlements and claims. It is a reasonable reflection of the Group’s current position given developments in Q4 given the prevailing uncertainty as to the outcome of the discussions with the U.S. Department of Justice. The risk of future substantial additional charges and costs is appropriately disclosed in the financial statements.

• The provision for the remaining claimants for the UK shareholder action incorporates assumptions which are reasonable for the purpose of determining a provision under IAS37.

• The PPI provision remains sensitive to key assumptions, such as future complaint volumes, time barring and Plevin outcomes. Management’s estimate was within our range of possible outcomes based on reasonable alternative assumptions.

• The provision related to the FCA review of RBS’s treatment of SMEs is sensitive to a number of assumptions. We consider it is reasonable based on the information available at this time.

The Group Audit Committee also discussed and approved the recording of a provision of £750m in the 2016 annual accounts, as a consequence of a revised proposal presented by HM Treasury to the European Commission in February 2017. This proposal envisages that RBS will deliver a package of remedies to promote competition in the market for banking services to SMEs in the UK. It remains subject to a consultation exercise by the European Commission and approval by the College of Commissioners as well as a market testing exercise by HM Treasury. We discussed and agreed the quantum of the provision, based on management’s assessment of the cost to deliver the package of remedies and their conclusion that it met the criteria to be recorded as an adjusting post balance sheet event.

The valuation of financial instruments with higher risk characteristics involves both significant judgment and risk of inappropriate revenue recognition through mis-marking. The judgement in estimating fair value of these instruments can involve complex valuation models and significant fair value adjustments both of which may be reliant on data inputs where there is limited market observability. At year end the Group reported level 3 assets £4.5 billion (2015 – £3.9 billion) and level 3 liabilities £3.0 billion (2015 £2.7 billion).

The potential risk of inappropriate recognition of revenue is most likely to arise through the valuation of these instruments given the level of management judgement involved.

The key judgements and estimates are:

• Complex model-dependent valuations, which are aligned with material pricing models as defined by the RBS Modelled Product Review Committee. These include interest-rate swaps linked to pre-payment behavior and interest rate and foreign exchange options with exotic features such as those having multiple call dates or with a variable notional.
• Instruments valued using illiquid pricing inputs, which are aligned with material positions defined as level 3 within RBS’s IFRS 7 fair value hierarchy disclosure. These include interest rate derivative instruments whose valuation is dependent upon the correlation between certain interest rates and rarely traded debt securities.
• Fair value adjustments made to uncollateralized derivatives to reflect funding risk and counter party credit risk. These include RBS’s Funding Valuation Adjustments (FVA) and Counterparty Valuation Adjustments (CVA) relating to derivative counterparties whose credit spread is less readily able to be determined.

We performed trade life-cycle product walkthroughs to confirm our understanding of RBS’s process and controls in the area of revenue recognition relating to financial instruments with higher risk characteristics.

We tested the design and operating effectiveness of the Group’s controls over financial instrument valuations, including independent price verification, model governance, and P&L analysis and reporting.

We performed further procedures as set out below for each risk characteristic with involvement from our financial instrument valuation and modelling specialists. These procedures were performed at multiple points in the year to validate the appropriateness of revenue recognition.

• Our testing on complex model-dependent valuations involved the specialist review of detailed model documentation and the building of a number of bespoke EY challenger models to analyze and challenge judgements and assumptions applied within each relevant model.
• Our re-pricing of instruments valued using illiquid pricing inputs covered material products associated with this risk and the results were compared to the valuations recorded by management. For derivatives our valuation used EY models and independent data and for illiquid cash positions, prices of comparable positions and other data points were used.
• Our testing on fair value adjustments for counterparty credit and funding risk on uncollateralized derivatives involved: (i) comparing valuation judgements applied by management to our knowledge of current industry practice through benchmarking exercises (ii) re-valuing a sample of counterparty level FVA and CVA calculations using independent models, (iii) testing funding spreads to third party data, including comparison with recent trade activity and (iv) independently recalculating illiquid CVA inputs.
• Where differences were identified between our independent valuation and management’s valuation, we performed additional testing over each variance to support our assessment of the appropriateness of the fair value. This work included our own analysis of: (i) recent trade activity, involving new trades and trade exits to back-test key valuation judgements, (ii) collateral disputes to compare to counterparty valuations, (iii) P&L attribution, particularly unexplained P&L in the year for the relevant instruments and (iv) associated valuation adjustments e.g. model fair value adjustments to reflect the associated uncertainty given lack of market data.

We are satisfied that the fair value of financial instruments with higher risk characteristics and the recognition of related income is reasonable and in accordance with IFRS.
We highlighted the following:

• Our independent valuation of a sample of derivatives were either within our threshold or, where initially outside, were corroborated by other data, for example, trade exit activity, valuation adjustments for model or data limitations, or benchmarking to peer practice. Valuations of hard-to-price cash positions were within our thresholds.

The Group’s recognition of fair value adjustments on uncollateralized derivatives is within a reasonable range of outcomes based upon our testing procedures which included revaluation exercises, benchmarking to peer practice and experience from new trades or trade exists.

Our audit procedures have a focus on IT. Our areas of audit focus included user access management, developer access systems and controls due to the pervasive nature and complexity of the IT environment, the large volume of transactions processed in numerous locations daily and the reliance on automated and IT dependent manual controls to the production environment and changes to the IT environment. These are key to ensuring IT dependent and application based controls are operating effectively.

We tested the design and operating effectiveness of the Group’s IT access controls over the information systems that are critical to financial reporting. We tested IT general controls (logical access, changes management and aspects of IT operational controls). This included testing that requests for access to systems were appropriately reviewed and authorized. We tested the Group’s periodic review of access rights. We inspected requests of changes to systems for appropriate approval and authorization. We considered the control environment relating to various interfaces, configuration and other application layer controls identified as key to our audit.

Where deficiencies were identified, we tested compensating controls or performed alternate procedures. In addition, we sought to understand where relevant, changes were made to the IT landscape during the audit period.

We are satisfied that IT controls relevant to financial reporting operated effectively as at year-end.

A number of user access related deficiencies were identified. Management identified compensating controls to mitigate these findings and undertook additional work to evidence that access was not used inappropriately.

We tested the work undertaken by management and performed additional audit procedures over impacted balances.

RBS operates a number of defined benefit schemes which in total are significant in the context of the overall balance sheet. At year end the Group reported a net pension liability of £87 million (2015: £3.6 billion).

The valuations of the retirement benefit liabilities are calculated with reference to a number of actuarial assumptions and inputs including discount rate, rate of inflation and mortality rates. Small changes in assumptions can impact the pension liability and asset disclosures.

The pension schemes hold certain complex and illiquid assets for which there are no quoted prices.

We tested the design and operating effectiveness of key controls over the completeness and accuracy of data extracted and supplied to the Group’s actuaries, which is used to calculate the pension schemes’ surplus or deficit. We also tested the controls associated with the measurement of the fair value of the schemes’ assets and the actuarial assumptions and valuations.

With the support of our actuarial specialists we determined whether the assumptions used to value the retirement benefit liabilities met the requirements of accounting standards and were in line with market practice, as well as the specific circumstances of the schemes and their participants. This included a comparison of life expectancy assumptions with relevant mortality tables, benchmarking inflation and discount rates against external market data, considering changes in historical assumptions and evaluating the independence, qualifications and results of work performed by management’s experts involved in the valuation process.

We tested the fair value of scheme assets by independently calculating a fair value for a sample of the assets held. Our sample included cash, equity instruments, and derivative financial instruments. We also tested the existence of the sampled pension assets by obtaining written confirmation from the pension asset custodian and by examining the relevant legal documentation.

We also reviewed Board and Executive Committee minutes where the pension valuation was discussed to assess the effectiveness of management’s review process and the appropriateness of the conclusions reached.

We are satisfied that the pension valuation and retirement benefit obligations recorded at 31 December 2016 are reasonable and in accordance with IFRS.

We highlighted the results of our internal benchmarking of key actuarial assumptions including the discount rate, inflation, mortality, and the rate of increase of salary and pension payments. We noted that assumptions tested are within a reasonable range. We also presented the results of our independent valuation of a sample of pension assets.

In determining the accounting for the pension scheme we considered the impact of IFRIC 14 on the amount of the surplus recognized and compliance with IFRS.

A significant degree of judgement is required to determine the timing and amount of impairment to recognize with respect to loans and advances. At year end the group reported total gross loans and advances of £386.5 billion (2015: £371.7 billion) and impairment provisions of £4.5 billion (2015: £7.1 billion).

We have focused on the following critical judgements and estimates which could give rise to material misstatement or are potentially subject to management bias:

• Completeness and timing of recognition of loss events in accordance with criteria set out in IAS 39.
• For individually assessed provisions, the measurement of the provision is dependent on the valuation of collateral, the timing of cash flows and realizations.
• For modelled provisions, the measurement is dependent upon key assumptions relating to probability of default and recovery rates.
• Completeness and measurement of post model adjustments and overlays.

We tested the design and operating effectiveness of key controls to identify loss events and assess and determine the extent to which impairments should be recognized considering the potential for management override of controls. These include:

• Manual and automated monitoring of loans with higher risk of default.
• Annual loan credit reviews.
• Assessment and approval of material impairment provisions including valuation of collateral.
• Governance over the impairment process, including assessment of suitability of models and assumptions.
• Model validation and challenge of assumptions and calculation accuracy.
• Completeness and accuracy of data input into models.

In addition, we attended and observed key management meetings.

For modelled provisions, we tested the inputs and used our credit risk specialists to test the assumptions and calculations. We examined the methodology to establish model parameters and assessed the appropriateness of the models used. Where possible, assumptions were benchmarked against pillar 3, EBA stress tests and our internally developed ranges. Where overlays were made as a result of limitations in existing models, we confirmed the extent of the model shortcoming, recalculated the overlay and assessed the appropriateness of the adjustment. Based on current economic events, we considered the need for sector or systemic overlay adjustments.

We performed a sample of loan reviews on performing loans to establish our own view as to whether any IAS 39 loss indicators were present.

For non-performing loans, we tested a sample of loan reviews, focusing on high risk sectors such as shipping, oil and gas and commercial real estate. With the support of our valuation specialists, we assessed the measurement of the provision by testing the valuation of collateral where relevant. We examined other cash flow assumptions where the level of provision is not dependent on collateral values. We also assessed the reasonableness of the timing of the cash flows estimated

We assessed the appropriateness and presentation of disclosures with relevant accounting standards.

We are satisfied that impairment provisions were reasonable and in compliance with IFRS. We highlighted the following:

• We discussed our benchmarking comparison for certain models which indicated that the assumptions were within an acceptable range of outcomes. The data used as inputs into the models and calculators is materially complete and accurate.
• For individually assessed impairments we did not identify material differences of judgement in respect of the provision and we were satisfied with the completeness of the identification of loss events.

Refer to pages 115 and 118 (Accounting policies) and page 150 (notes).

The Group has defined benefit pension plans with net post –retirement liabilities of £457m which is significant in the context of the overall balance sheet of the Group.

The valuation of the pension liabilities requires significant levels of judgement and technical expertise in choosing appropriate assumptions. Changes in a number of the key assumptions (including salary increases, inflation, discount rates and mortality) can have a material impact on the calculation of the net liability, particularly for the Airlines Germany pension schemes which are unfunded.

There is also an element of judgement in the measurement of fair value of pension assets due to the nature of financial investments.

We used our pension specialist to evaluate the assumptions used in the valuation of the liabilities and assets in the Group’s pension plans.

We also focused on the valuations of the pension plan liabilities and the pension assets and our procedures included the following:

• We agreed the discount and inflation rates used in the valuation of the pension liability to our internally developed benchmarks
• We assessed salary increase and mortality rate assumptions against national and industry averages
• We obtained third-party confirmations on the ownership and valuation of pension assets

We checked that there was no impact of specific events, such as changes to schemes and redundancies that should have been incorporated into the calculation.

We tested underlying inputs, such as employees in the scheme, to the liability valuation used by the scheme actuary. We also evaluated and tested management’s controls and processes over pension data such as leavers to the plan.

The assumptions used by the Directors’ were in line with our independently determined expectations and management’s disclosures in respect of the schemes are appropriate.

Refer to pages 113 to 114 (Accounting policies) and pages 138 to 141 (notes).

The Group uses a number of hedging structures including options to manage its exposure to adverse movements in fuel prices and foreign exchange rates.

The accounting for options and related derivatives is complex and we therefore focused on this area to assess whether hedge accounting had been appropriately applied and the impact of hedging had been properly presented.

We used our specialist treasury knowledge to test the valuations for fuel and foreign currency derivatives by recalculating their fair value using observable market data.

We evaluated the values held in the hedging reserve and tested manual adjustments made to correct for timing differences between the maturity of the hedging instrument and the underlying exposure.

We examined the hedge documentation and the hedging structures in place to check whether they had been accounted for in accordance with the Group’s accounting policies and presented appropriately in the Annual Report and Accounts. Our work performed did not identify any material misstatements.

Refer to page 112 (Accounting policies) and pages 128 and 144 (notes).

The Group holds significant goodwill and deferred tax assets on the balance sheet of £2,595m and £228m respectively.

In particular, in respect of goodwill we focused on the value in use of the UK, Continental Europe and Northern Europe Cash Generating Units (“CGUs”) which account for 99% of the total goodwill balance. Similarly, for deferred tax we focused on certain countries including the UK, Germany and Spain.

Determining the carrying value of these assets is dependent on complex and subjective judgements by the Directors about the future results of the business.

The value of these assets is highly the successful implementation dependent upon the Directors’ views of the future results and prospects of the business including of the ongoing UK transformation and business development and restructuring initiatives that form part of the New Operating Model.

If forecast results are not achievable, the valuation of the goodwill and the recognition of the deferred tax assets may not be appropriate.

We evaluated the process by which the Directors prepared their cash flow forecasts and confirmed that they reflected the latest Board approved three-year plans. We performed a critical review of the historical accuracy of budgets and forecasts by, for example, comparing the actual performance of the business in the current year against the board approved budgets. These procedures enabled us to determine the quality and accuracy of the forecasting process.

The key assumptions within the forecasts are the continuing successful implementation of the business model and profit improvement initiatives which drive the resulting growth rates. In assessing the appropriateness of the Directors’ assumptions we benchmarked certain external data used in the discount rate calculation against rates used by comparable companies. We also considered factors such as independent forecast growth rates for Thomas Cook and the wider travel industry.

We applied sensitivity analysis to the Directors calculations to ascertain the extent to which reasonably possible changes would, either individually or in aggregate, require the impairment of goodwill. We have assessed the recognition of deferred tax assets on losses against forecast future profits.

As a result of our work, we concurred with the Directors’ conclusion that no goodwill impairment charges were required and that it was appropriate to recognize deferred tax assets.

We focused on this area because the Group is subject to challenge in respect of a number of legal, regulatory and competition matters, many of which are beyond its control. Consequently, the Directors make judgements about the incidence and quantum of such liabilities arising from litigation and regulatory or competition claims which are subject to the future outcome of legal or regulatory processes.

See Note 29 to the financial statements on pages 330 to 338.

We assessed and tested the design and operating effectiveness of the controls over the identification, evaluation, provisioning and reporting of legal, regulatory and competition matters. We determined that we could rely on these controls for the purposes of our audit.

In view of the significant judgements required, we examined the more material provisions in detail and sought additional evidence. We evaluated the Group’s assessment of the nature and status of litigation, claims and provision assessments, if any, and discussed with internal counsel to understand the legal position and the basis of material risk positions. We received legal letters from the Group’s external counsel setting out their views in some cases.

Specifically, we challenged the timing of recognition for cases where there is potential exposure but it is not clear that a provision should be raised e.g. when obtaining reliable estimates are not considered possible.

We reviewed regulatory correspondence with US and UK regulators to identify potential regulatory investigations that could lead to the need for potential new provisions.

As set out in the financial statements, the outcome of litigation and regulatory claims are dependent on the future outcome of continuing legal and regulatory processes and consequently the calculations of the provisions are subject to inherent uncertainty.

We evaluated whether the disclosures within the financial statements appropriately address the significant uncertainties that exist around the outcome of litigation and regulatory claims. In our view, the provisions had been arrived at based on the information currently available to the Group and after consideration of the legal advice received by the Group.

We identified IT systems, and automated and IT dependent manual controls over financial reporting as an area of focus and an area that required more effort. The purpose of the work was to support our ability to rely on controls for the purpose of this report, as the Group’s financial accounting and reporting systems are heavily dependent on complex systems. There is a risk that automated accounting procedures and related IT dependent manual controls are not designed and operating effectively.

A particular area of focus related to logical access management including privileged access and developer access to the production environment. During the year, we identified a number of control deficiencies through the audit procedures performed in relation to access to financially significant infrastructure.

We evaluated the design and tested the operating effectiveness of the controls over the continued integrity of the IT systems that are relevant to financial reporting. We examined the framework of governance over the Group’s IT organization and the controls over access to programs and data, program changes, IT operations and program development. Where necessary, we carried out direct tests of certain aspects of the security of the Group’s IT systems including logical access management and segregation of duties.

Where controls failed either in design or operating effectiveness, we altered our audit approach and performed additional procedures, as follows:

• Identified compensating controls and tested that they operated effectively;
• Increased substantive procedures to mitigate the deficiencies found.

These additional procedures mitigated the deficiencies found and provided the additional audit evidence required.

We focused on this area because the Directors make complex and subjective judgements over both timing of recognition of impairment and the estimation of the size of any such impairment.

In wholesale loans and advances, the material portion of impairment is individually calculated. For retail loans and advances, the material portion of the impairment is calculated on a modeled basis for portfolios of loans and advances.

We focused our audit on the following areas of impairment specifically relating to:

The key assumptions and judgements made by the Directors that underlie the calculation of modeled unsecured retail impairment (including in relation to a number of model methodology changes in 2016). Key assumptions and judgements include the emergence period used for unidentified impairment and the probability of default calculation.

The post model adjustments recorded in response to a range of identified internal factors, such as known data and system issues impacting specific impairment models, and external factors such as the persistently low interest rate environment in the UK.

The completeness of the customer accounts that are included in the impairment calculation, including how unidentified impairment (customers that have had a loss event that has not yet manifested itself in a missed payment or other indicator) and forbearance are taken account of.

In addition for wholesale, we considered the impact of continued low oil, gas and commodity prices on the creditworthiness of relevant counterparties.

See Notes 7 and 20 to the financial statements on pages 291 to 292 and 322 and the relevant parts of the Risk review to which they are cross referred.

We evaluated the design and tested the operating effectiveness of the controls over impairment data and calculations. These controls included those over the identification of which loans and advances were impaired, the granting and monitoring of forbearance, the transfer of data (source systems to credit systems, credit systems to impairment models and model output to the general ledger) and the calculation of the impairment provisions. In addition we tested IT controls over the operation of the models, including restricted access and change management controls. We determined that we could rely on these controls for the purposes of our audit.

We tested the entity and business unit level controls over the end to end model process including in relation to model monitoring and validation, model inputs and outputs and management adjustments. We determined that we could rely on these controls for the purposes of our audit.

In addition, we performed detailed testing on a sample of new and existing models used to calculate both unidentified and identified impairment. This testing varied by portfolio, but typically included some combination of independent model rebuild, testing of the coding used in impairment models, re-performance of the calculation, testing the extraction of data used in the models including the ‘bucketing’ into delinquency bandings, and testing and applying sensitivities to the underlying critical assumptions.

We tested a sample of post model adjustments, including considering the basis for the adjustment, the logic applied, the source data used, the key assumptions adopted and the sensitivity of the adjustment to these assumptions. We assessed the completeness of post model adjustments. We found no material exceptions during our testing of the models and relevant controls.

Where impairment was individually calculated, we tested controls over the timely identification of potentially impaired loans. We determined that we could rely on these controls for the purposes of our audit. We tested a sample of loans and advances to ascertain whether the loss event (that is the point at which impairment is recognized) had been identified in a timely manner including, where relevant, how forbearance had been considered. Where impairment had been identified, we examined the business valuation or forecasts of future cash flows prepared by management to support the calculation of the impairment, challenging the assumptions and comparing estimates to external evidence where available. We found no material exceptions in these tests.

We examined a sample of loans and advances which had not been identified by management as potentially impaired and formed our own judgement as to whether that was appropriate including using external evidence in respect of the relevant counterparties. We found no material exceptions in these tests.

For customers with exposure to the oil, gas and commodity prices, we increased our sample testing of cases individually assessed for impairment, including those customers identified on the watch list, and those that remained in the ‘good book’. In addition we tested relevant post model adjustments held and considered the completeness of the unidentified impairment provision for these customers.

In the case of some impairment provisions, we formed a different view from that of management, but in our view the differences were within a reasonable range of outcomes in the context of the overall loans and advances and the uncertainties disclosed in the financial statements.

We focused on this area because the Group is subject to taxation in many jurisdictions and, in many cases, the ultimate tax treatment is uncertain and is not determined until resolved with the relevant tax authority. Consequently, the Directors make judgements about the incidence and quantum of tax liabilities which are subject to the future outcome of assessments by the relevant tax authorities and potentially associated legal processes.

See Note 10 to the financial statements on pages 294 to 297.

Our tax specialists examined the correspondence between the Group and the relevant tax authorities and between the Group and its external advisers. We examined the matters in dispute and used our knowledge of the law of the relevant tax jurisdictions and other similar taxation matters to assess the available evidence and the provisions made by Directors.

As set out in the financial statements, since the settlement of the Group’s tax position is subject to future negotiation with various tax authorities, the calculations of the provisions are subject to inherent uncertainty. In our view, the provisions were within a reasonable range of outcomes in the context of that uncertainty. We evaluated whether the disclosures within the financial statements appropriately address any significant uncertainties that exist around determining the provisions.

One of the key areas of focus is the Education, Social Housing and Local Authority (ESHLA) loan books (£8.9bn) which require significant judgement in the valuation methodology due to limited availability of observable market data used to calibrate the funding and credit spreads forming the valuation.

During the year the Directors unilaterally waived certain complex option features on all ESHLA loans with such features. The Directors made a significant accounting judgement around the implication of this action and concluded it was a de-recognition event for those loans. The Directors concluded the new loans should be booked at amortized cost.

The Directors make significant judgements because of the complexity involved in valuing some of these assets and liabilities. These judgements include:

• The determination of significant unobservable inputs and the categorization of financial instruments based on the IFRS 13 fair value hierarchy;
• The calculation methodology and inputs of valuation adjustments including Funding Fair Value Adjustment (FFVA), although the Judgement in this area has decreased as industry practice has become more standardized; and
• Where a long standing significant valuation disparity exists between the Group and counterparty in relation to a specific long-dated derivative portfolio.

The Directors established that the Methodology used in the above areas of judgement is appropriate.

See Notes 14 to 18 to the financial statements on pages 299 to 320.

In auditing the ESHLA loan portfolio that is carried at fair value we assessed the appropriateness of the valuation methodology and the level of observability in the market, including performing comparable credit spread research and comparing any observations to those used by management within the Credit Valuation Adjustment. We independently performed the valuation of a sample of loans and evaluated the appropriateness of ESHLA specific fair value adjustments.

We tested the Director’s judgements regarding the ESHLA loan de-recognition event and evaluated management’s technical accounting analysis. From our testing, we determined that the waiving of the complex options was a de-recognition event. We tested the recording of the new loans at amortized costs and tested the controls applied to the new hedges for these loans and found we could rely on them for the purposes of our audit.

For other judgmental or complex valuations, in general, we evaluated the design and tested the operating effectiveness of the key controls supporting the identification, measurement and oversight of valuation and accounting risk of financial instruments. We examined the key controls, including the independent price verification process, model validation and approval process, controls over data feeds and inputs to valuation, the fair value hierarchy categorization policy, and governance and reporting controls.

We performed substantive testing over the following key areas.

We utilized internal valuation specialists to perform independent valuations to determine if management’s valuations fell within a reasonable range. The revaluation covered a range of product classes and was performed across level 1, 2 & 3 of the Group’s IFRS 13 fair value hierarchy.

For valuations which depend on unobservable inputs, we evaluated the assumptions, methodologies and models used by the Directors.

We performed an independent valuation of a sample of positions including those relating to the significant valuation disparity.

Methodology and underlying assumptions of key valuation adjustments, including the Credit Valuation Adjustment, Debt Valuation Adjustment and Funding Fair Value Adjustment, were assessed, and compared with our knowledge of current industry practice. For FFVA, we evaluated the extent to which funding costs are incorporated into derivative valuation with reference to the limited observable transaction and other market data available.

We performed testing to validate that management had allocated financial instruments to the correct level (1, 2 or 3) within the fair value hierarchy in line with the established policy, and that the policy classifications were appropriate.

We examined of collateral disputes, gains and losses on disposals and other events which could provide evidence about the appropriateness of the Director’s valuations. We examined the valuation disparity in relation to the long-dated derivative portfolio.

Overall, in our view, in the context of the inherent uncertainties as disclosed in the financial statements, the accounting was appropriate and the valuations were within a reasonable range of outcomes.

Valuation of level 3 positions comprising $2.3 billion, 1.2% of total fair value financial instruments (2015 $3.0 billion, 1.5%), including unlisted Investments in the Principal Finance business and derivatives with significant unobservable pricing inputs, Modelling of and key inputs into the valuation of derivative and other instruments.

Appropriateness of and inputs into valuation adjustments including those relating to counter party credit, funding and own credit risk.

Disclosures of the fair value hierarchy of assets and liabilities held at fair value, including the assessment of observability of pricing inputs.

Our procedures included:

• Valuation controls: we tested the Group’s controls over the identification, measurement and management of valuation risk including independent price verification control, governance over valuation models, model validation and management reporting of valuation risk

• Modelling methodologies: we assessed whether, the model valuation methodologies used for material valuation risks are appropriate, utilizing our valuation specialists

• Price test work: we tested, (or a selection of pricing inputs, that they were externally sourced and were correctly input into pricing models. We independently valued a selection of the Group’s debt and equity securities and compared these to the Group’s valuation. For a selection of derivative positions, we obtained the valuation of each position submitted by the counterparty to a common collateral platform used in the Group’s collateral management process and compared these to the Group’s own valuation for the same position

• Level 3: we assessed, for a selection of level 3 positions, the valuation methodology, model and assumptions used, considering potential alternatives and sensitivities to key factors. Independently vouched key pricing inputs to source data and assessed the appropriateness of assumptions used in relation to unobservable inputs. We challenged the appropriateness and completeness of the model reserves held against instruments where elements of risk could not be modelled

• Valuation adjustments: utilising our valuation specialists, we assessed the appropriateness of the methodologies and models used to estimate valuation adjustments. We assessed for a selection of transactions the appropriateness of key assumptions and inputs, in particular counterparty credit risk, For a sample of counterparties agreed credit risk inputs to directly observable sources and. where proxies were used, we assessed the appropriateness of the proxy, considering the availability of alternatives

• Disclosures: we tested a selection of instruments to establish the observability of key pricing inputs with reference to active markets and, where relevant, the term of the instrument. We assessed the appropriateness of the fair value hierarchy assigned to these instruments. Particularly for level 3 positions, we assessed whether the financial statement disclosures, including sensitivity to key inputs, appropriately reflect the Group’s exposure to valuation risk.