The preparation of the Group’s financial statements in accordance with IFRS requires management to make judgements, estimates and assumptions in applying the accounting policies that affect the reported amounts of assets, liabilities, income and expenses. Due to the inherent uncertainty in making estimates, actual results reported in future periods may be based upon amounts which differ from those estimates. Estimates, judgements and assumptions are continually evaluated and are based on historical experience and other factors, including expectations of future events that are believed to be reasonable under the circumstances.

The significant judgements made by management in applying the Group’s accounting policies and the key sources of estimation uncertainty in these financial statements, which together are deemed critical to the Group’s results and financial position, are as follows:


Key judgements:			Determining an appropriate definition of default against which a probability of default, exposure at default and loss given default parameter can be evaluated
			The appropriate lifetime of an exposure to credit risk for the assessment of lifetime losses, notably on revolving products
			Establishing the criteria for a significant increase in credit risk
			The use of management judgement alongside impairment modelling processes to adjust inputs, parameters and outputs to reflect risks not captured by models
Key estimates:			Base case and Multiple Economic Scenarios (MES) assumptions, including the rate of unemployment and the rate of change of house prices, required for creation of MES scenarios and forward looking credit parameters

The Group recognises an allowance for expected credit losses for loans and advances to customers and banks, other financial assets held at amortised cost, financial assets measured at fair value through other comprehensive income and certain loan commitment and financial guarantee contracts. At 31 December 2020 the Group’s expected credit loss allowance was £6,247 million (2019: £3,455 million), of which £5,788 million (2019: £3,278 million) was in respect of drawn balances.

The calculation of the Group’s expected credit loss (ECL) allowances and provisions against loan commitments and guarantees under IFRS 9 requires the Group to make a number of judgements, assumptions and estimates. The most significant are set out below.

The probability of default (PD) of an exposure, both over a 12 month period and over its lifetime, is a key input to the measurement of the ECL allowance. Default has occurred when there is evidence that the customer is experiencing significant financial difficulty which is likely to affect the ability to repay amounts due. The definition of default adopted by the Group is described in note 2(H) Impairment of financial assets. The Group has rebutted the presumption in IFRS 9 that default occurs no later than when a payment is 90 days past due for UK mortgages. As a result, at 31 December 2020, approximately £0.6 billion of UK mortgages (2019: £0.6 billion) were classified as Stage 2 rather than Stage 3; the impact on the Group’s ECL allowance was not material.

A range of approaches, segmented by product type, has been adopted by the Group to estimate a product’s expected life. These include using the full contractual life and taking into account behavioural factors such as early repayments and refinancing. For non-revolving retail assets, the Group has assumed the expected life for each product to be the time taken for all significant losses to be observed. For retail revolving products, the Group has considered the losses beyond the contractual term over which the Group is exposed to credit risk. For commercial overdraft facilities, the average behavioural life has been used. Changes to the assumed expected lives of the Group’s assets could impact the ECL allowance recognised by the Group. The assessment of SICR and corresponding lifetime loss, and the PD, of a financial asset deemed to be Stage 2, or Stage 3, is dependent on its expected life.

Performing assets are classified as either Stage 1 or Stage 2. An ECL allowance equivalent to 12 months expected losses is established against assets in Stage 1; assets classified as Stage 2 carry an ECL allowance equivalent to lifetime expected losses. Assets are transferred from Stage 1 to Stage 2 when there has been a significant increase in credit risk (SICR) since initial recognition. Credit impaired assets are transferred to Stage 3 with a lifetime expected losses allowance. The Group uses both quantitative and qualitative indicators to determine whether there has been a SICR for an asset. For Retail, the following tables set out the Retail Master Scale (RMS) grade triggers which result in a SICR for financial assets and the PD boundaries for each RMS grade. Credit cards SICR triggers have been refined in 2020 following a review of sensitivity to changes in economic assumptions, 2019 triggers were previously aligned to Loans and overdrafts. The impact of this has been approximately £1.4 billion of additional assets being classified as Stage 2 at 31 December 2020, with a corresponding increase in the ECL of £48 million resulting from the transfer to a lifetime expected loss.

The Group uses the internal credit risk classification and watchlist as qualitative indicators to identify a SICR. The Group does not use the low credit risk exemption in its staging assessments. The use of a payment holiday in and of itself has not been judged to indicate a significant increase in credit risk, nor forbearance, with the underlying long-term credit risk deemed to be driven by economic conditions and captured through the use of forward-looking models. These portfolio level models are capturing the anticipated volume of increased defaults and

therefore an appropriate assessment of staging and expected credit loss. During 2020, the Group has granted payment holidays on Retail loans and advances, £6.4 billion remained in place at 31 December 2020, £4.3 billion of these balances were classified as Stage 1. If all of these assets were classified as Stage 2, the Group's ECL would have been less than £50 million higher.

All financial assets are assumed to have suffered a SICR if they are more than 30 days past due; non-mortgage Retail financial assets are also assumed to have suffered a SICR if they are in arrears on three or more separate occasions in a rolling twelve month period. Financial assets are classified as credit impaired if they are 90 days past due except for UK mortgages where a 180 days backstop is used.

A Stage 3 asset that is no longer credit-impaired is transferred back to Stage 2 as no cure period is applied to Stage 3. If an exposure that is classified as Stage 2 no longer meets the SICR criteria, which in some cases include a minimum cure period, it is moved back to Stage 1.

The setting of precise trigger points combined with risk indicators requires judgement. The use of different trigger points may have a material impact upon the size of the ECL allowance. The Group monitors the effectiveness of SICR criteria on an ongoing basis.

The measurement of expected credit losses is required to reflect an unbiased probability-weighted range of possible future outcomes. The Group considers the choice of approach used to generate the range of economic outcomes to be judgemental, given several methods can be adopted. In addition to a defined base case, as used for planning, the Group’s approach relies on model-generated scenarios, reducing scope for bias in the selection of scenarios and their weightings. The conditioning assumptions underpinning the base case scenario reflect the Group’s best view of future events. Where outcomes materially diverge from the conditioning assumptions adopted, the base case scenario is updated. The base case is therefore central to the range of outcomes created as no alternative conditioning assumptions are factored into the model-generated scenarios.

The Group models a full distribution of economic scenarios around this base case, ranking them using estimated relationships with industry-wide historical loss data. The full distribution is summarised by a practical number of scenarios to run through ECL models representing four sections: an upside, the base case, and a downside scenario weighted at 30 per cent each, with a severe downside scenario weighted at 10 per cent. With the base case already pre-defined, the other three scenarios are constructed as averages of constituent modelled scenarios around the 15th, 75th and 95th percentiles of the distribution. The scenario weights therefore represent the allocation to each summary segment of the distribution and not a subjective view on likelihood. The inclusion of a severe downside scenario with a smaller weighting but relatively large credit losses, ensures the non-linearity of losses in the tail of the distribution is captured when ECL based on the weighted result of the four scenarios is calculated.

A committee under the chairmanship of the Chief Economist meets at least quarterly to review and, if appropriate, recommend changes to the method by which economic scenarios are generated; for approval by the Chief Financial Officer and Chief Risk Officer. In 2020, a change was made to the way in which the distribution of scenarios is created. This change allows for a greater dispersal of economic outcomes in the early periods of the forecast, to recognise the increased near-term profile of risks present since the onset of the coronavirus pandemic. This change allows for a wider distribution of losses both on the upside and downside, although is most evident in the severe downside scenario, given it represents a more adverse segment of the distribution. The change is estimated to have driven an additional £200 million of ECL resulting from the inclusion of more adverse economic outcomes.

The Group’s base case economic scenario has continued to be revised in light of the impact of the coronavirus pandemic in the UK and globally. The scenario reflects judgements of the net effect of government-mandated restrictions on economic activity, large-scale government interventions, and behavioural changes by households and businesses that may persist beyond the rollout of coronavirus vaccination programmes.

Despite large-scale vaccination efforts commencing in the UK and globally, there remains considerable uncertainty about the pace and eventual extent of the post-pandemic recovery. The Group’s current base case scenario builds in three key conditioning assumptions. First, the UK vaccine rollout successfully protects the elderly, key workers and the clinically vulnerable by mid-2021. Second, national lockdowns end by April 2021, allowing a phased return to a tiered system of restrictions that are progressively eased in the second quarter and second half of 2021, leaving only limited restrictions in place by the end of 2021. Third, government policy measures including specifically the furlough scheme continue to provide support for the duration of severe economic restrictions, through to mid-2021.

Conditioned on the above assumptions and despite the recovery in economic activity resuming from the second quarter of 2021, the Group’s base case outlook assumes a rise in the unemployment rate and weakness in residential and commercial property prices. Risks around this base case economic view lie in both directions and are partly captured by the MES generated. But uncertainties relating to the key conditioning assumptions, including epidemiological developments and the efficacy of vaccine rollouts, are not specifically captured by the MES scenarios. These specific risks have been recognised outside the modelled scenarios published below.

The Group has accommodated the latest available information at the reporting date in defining its base case scenario and generating the MES. The scenarios include forecasts for key variables in the fourth quarter of 2020, for which actuals may have since emerged prior to publication.


	First quarter 2020	Second quarter 2020	Third quarter 2020	Fourth quarter 2020	First quarter 2021	Second quarter 2021	Third quarter 2021	Fourth quarter 2021
Base case	%	%	%	%	%	%	%	%
Gross domestic product	(3.0)	(18.8)	16.0	(1.9)	(3.8)	5.6	3.6	1.5
UK Bank Rate	0.10	0.10	0.10	0.10	0.10	0.10	0.10	0.10
Unemployment rate	4.0	4.1	4.8	5.0	5.2	6.5	8.0	7.5
House price growth	2.8	2.6	7.2	5.9	5.5	4.7	(1.6)	(3.8)
Commercial real estate price growth	(5.0)	(7.8)	(7.8)	(7.0)	(6.1)	(2.9)	(2.2)	(1.7)

Key annual assumptions made by the Group are shown below. Gross domestic product is presented as an annual change, house price growth and commercial real estate price growth are presented as the growth in the respective indices within the period. UK Bank Rate and unemployment rate are averages for the period.


	2020	2021	2022	2023	2024
	%	%	%	%	%
Upside
Gross domestic product	(10.5)	3.7	5.7	1.7	1.5
UK Bank Rate	0.10	1.14	1.27	1.20	1.21
Unemployment rate	4.3	5.4	5.4	5.0	4.5
House price growth	6.3	(1.4)	5.2	6.0	5.0
Commercial real estate price growth	(4.6)	9.3	3.9	2.1	0.3
Base case
Gross domestic product	(10.5)	3.0	6.0	1.7	1.4
UK Bank Rate	0.10	0.10	0.10	0.21	0.25
Unemployment rate	4.5	6.8	6.8	6.1	5.5
House price growth	5.9	(3.8)	0.5	1.5	1.5
Commercial real estate price growth	(7.0)	(1.7)	1.6	1.1	0.6
Downside
Gross domestic product	(10.6)	1.7	5.1	1.4	1.4
UK Bank Rate	0.10	0.06	0.02	0.02	0.03
Unemployment rate	4.6	7.9	8.4	7.8	7.0
House price growth	5.6	(8.4)	(6.5)	(4.7)	(3.0)
Commercial real estate price growth	(8.7)	(10.6)	(3.2)	(0.8)	(0.8)
Severe downside
Gross domestic product	(10.8)	0.3	4.8	1.3	1.2
UK Bank Rate	0.10	0.00	0.00	0.01	0.01
Unemployment rate	4.8	9.9	10.7	9.8	8.7
House price growth	5.3	(11.1)	(12.5)	(10.7)	(7.6)
Commercial real estate price growth	(11.0)	(21.4)	(9.8)	(3.9)	(0.8)

The key UK economic assumptions made by the Group averaged over a five-year period are shown below. The five-year period reflects movements within the current reporting year such that 31 December 2020 reflects five years 2020 to 2024. The prior year comparative data has been re-presented to align to the equivalent period, 2019 to 2023. The inclusion of the reporting year within the five-year period reflects the need to predict variables which remain unpublished at the reporting date, and recognises that credit models utilise both level and annual change in calculating ECL. The use of calendar years also maintains a comparability between tables disclosed.


	At 31 December 2020				At 31 December 2019
	Upside	Base case	Downside	Severe downside	Upside	Base case	Downside	Severe downside
	%	%	%	%	%	%	%	%
Gross domestic product	0.3	0.1	(0.4)	(0.8)	1.6	1.3	1.0	0.3
UK Bank Rate	0.98	0.15	0.05	0.02	1.87	1.15	0.51	0.17
Unemployment rate	5.0	5.9	7.1	8.8	3.9	4.3	5.5	6.7
House price growth	4.2	1.1	(3.5)	(7.5)	5.1	1.4	(2.5)	(7.0)
Commercial real estate price growth	2.1	(1.1)	(4.9)	(9.7)	1.6	(0.3)	(3.9)	(7.3)


	At 31 December 2020				At 31 December 2019
	Upside	Base case	Downside	Severe downside	Upside	Base case	Downside	Severe downside
	%	%	%	%	%	%	%	%
Gross domestic product	1.4	0.8	(1.7)	(3.0)	8.4	6.6	5.5	1.8
UK Bank Rate	1.44	0.25	0.10	0.10	2.56	1.75	0.75	0.75
Unemployment rate	6.5	8.0	9.3	11.5	4.4	4.6	6.9	8.3
House price growth	22.6	5.9	5.6	5.3	28.3	7.1	2.7	2.7
Commercial real estate price growth	11.0	(2.7)	(2.7)	(2.7)	8.8	(0.8)	(0.8)	(0.8)


	At 31 December 2020				At 31 December 2019
	Upside	Base case	Downside	Severe downside	Upside	Base case	Downside	Severe downside
	%	%	%	%	%	%	%	%
Gross domestic product	(21.2)	(21.2)	(21.2)	(21.2)	0.3	0.3	0.3	(2.4)
UK Bank Rate	0.10	0.10	0.01	0.00	0.75	0.75	0.35	0.01
Unemployment rate	4.0	4.0	4.0	4.0	3.4	3.8	3.8	3.8
House price growth	(0.5)	(0.5)	(16.4)	(32.4)	1.5	0.0	(12.0)	(30.3)
Commercial real estate price growth	(6.9)	(9.0)	(22.2)	(39.9)	(1.4)	(2.3)	(17.9)	(31.4)

The table below shows the extent to which a higher ECL allowance has been recognised to take account of forward looking information from the weighted multiple economic scenarios. A significant difference between these bases arises on UK mortgages as the probability-weighted ECL includes the impact of house price movements on the loss given default (LGD). Commercial Banking also reflects movements in the loss given default, whereas for Other Retail portfolios only the probability of default responds to changes in the economic outlook. ECL applied through individual assessments and post-model adjustments is reported flat against each economic scenario, reflecting the basis on which they are evaluated. Judgements applied through changes to inputs are reflected in the scenario sensitivities.


	At 31 December 2020			At 31 December 2019
	Base case	Probability- weighted	Difference	Base case	Probability- weighted	Difference
Impact of multiple economic scenarios	£m	£m	£m	£m	£m	£m
UK mortgages	804	1,027	223	464	569	105
Other Retail	2,310	2,368	58	1,492	1,521	29
Commercial Banking	2,177	2,402	225	1,258	1,315	57
Other	450	450	—	50	50	—
ECL allowance	5,741	6,247	506	3,264	3,455	191

The table below shows the Group’s ECL for the upside, base case, downside and severe downside scenarios. The stage allocation for an asset is based on the overall scenario probability-weighted PD and, hence, the Stage 2 allocation is constant across all the scenarios. ECL applied through individual assessments and post-model adjustments is reported flat against each economic scenario, reflecting the basis on which they are evaluated. Judgements applied through changes to inputs are reflected in the scenario sensitivities.


	At 31 December 2020					At 31 December 2019
	Probability- weighted	Upside	Base case	Downside	Severe downside	Probability- weighted	Upside	Base case	Downside	Severe downside
	£m	£m	£m	£m	£m	£m	£m	£m	£m	£m
UK mortgages	1,027	614	804	1,237	2,306	569	317	464	653	1,389
Other Retail	2,368	2,181	2,310	2,487	2,745	1,521	1,443	1,492	1,564	1,712
Commercial Banking	2,402	1,910	2,177	2,681	3,718	1,315	1,211	1,258	1,382	1,597
Other	450	448	450	450	456	50	50	50	50	50
ECL allowance	6,247	5,153	5,741	6,855	9,225	3,455	3,021	3,264	3,649	4,748

The table below shows the Group’s ECL for the upside, base case, downside and severe downside scenarios, with stage allocation based on each specific scenario. ECL applied through individual assessments and post-model adjustments is reported flat against each economic scenario, reflecting the basis on which they are evaluated. Judgements applied through changes to inputs are reflected in the scenario sensitivities. A probability-weighted scenario is not shown as this does not reflect the basis on which ECL is reported.


	At 31 December 2020				At 31 December 2019
	Upside	Base case	Downside	Severe downside	Upside	Base case	Downside	Severe downside
	£m	£m	£m	£m	£m	£m	£m	£m
UK mortgages	602	797	1,269	2,578	311	461	670	1,667
Other Retail	2,154	2,299	2,509	2,819	1,435	1,486	1,570	1,740
Commercial Banking	1,892	2,157	2,738	4,155	1,206	1,254	1,387	1,625
Other	448	449	450	457	49	50	50	51
ECL allowance	5,096	5,702	6,966	10,009	3,001	3,251	3,677	5,083

The table below shows the percentage of assets that would be recorded in Stage 2 for the upside, base case, downside and severe downside scenarios, if stage allocation was based on each specific scenario. Given additional data has been generated to support this new disclosure the prior year comparatives are not available.


	At 31 December 2020				At 31 December 2019
	Upside	Base case	Downside	Severe downside	Upside	Base case	Downside	Severe downside
	%	%	%	%	%	%	%	%
UK mortgages	6.9	8.9	11.8	16.7
Other Retail	12.6	13.5	15.2	17.9
Commercial Banking	8.2	10.9	17.5	24.9
Other	—	—	—	—
Percentage of assets in Stage 2	7.0	8.7	11.8	16.4

The impact of changes in the UK unemployment rate and House Price Index (HPI) have also been assessed. Although such changes would not be observed in isolation, as economic indicators tend to be correlated in a coherent scenario, this gives insight into the sensitivity of the Group’s ECL to gradual changes in these two critical economic factors. The assessment has been made against the base case with the reported staging unchanged.

The table below shows the impact on the Group’s ECL in respect of UK mortgages resulting from a decrease/increase in loss given default for a 10 percentage point (pp) increase or decrease in the UK House Price Index (HPI). The increase/decrease is presented based on the adjustment phased evenly over the first ten quarters of the base case scenario.

The table below shows the impact on the Group’s ECL resulting from a 1 percentage point (pp) increase or decrease in the UK unemployment rate. The increase or decrease is presented based on the adjustment phased evenly over the first ten quarters of the base case scenario. An immediate increase or decrease would drive a more material ECL impact as it would be fully reflected in both 12 month and lifetime PDs.


	At 31 December 2020		At 31 December 2019
	1pp increase in unemployment	1pp decrease in unemployment	1pp increase in unemployment	1pp decrease in unemployment
	£m	£m	£m	£m
UK mortgages	25	(23)	33	(34)
Other Retail	54	(54)	39	(54)
Commercial Banking	125	(112)	68	(54)
Other	1	(1)	1	(1)
ECL impact	205	(190)	141	(143)

Stage 3 ECL in Commercial Banking is largely assessed on an individual basis using bespoke assessment of loss for each specific client. These assessments are carried out by the Business Support Unit based on detailed reviews and expected recovery strategies. While these assessments are based on the Group’s latest economic view, the use of group-wide multiple economic scenarios and weightings is not considered appropriate for these cases due to their individual characteristics. In place of this a range of case specific outcomes are considered with any alternative better or worse outcomes that carry a 25 per cent likelihood taken into account in establishing a probability-weighted ECL. At 31 December 2020 individually assessed provisions for Commercial Banking were £1,222 million (2019: £890 million) which reflected a range of £982 million to £1,548 million (2019: £515 million to £1,183 million), based on the range of alternative outcomes considered.

Impairment models fall within the Group’s Model Risk framework with model monitoring, periodic validation and back testing performed on model components (i.e. probability of default, exposure at default and loss given default). Limitations in the Group’s impairment models or data inputs, may be identified through the ongoing assessment and validation of the output of the models. In these circumstances, management make appropriate adjustments to the Group’s allowance for impairment losses to ensure that the overall provision adequately reflects all material risks. These adjustments are determined by considering the particular attributes of exposures which have not been adequately captured by the impairment models and range from changes to model inputs and parameters, at account level, through to more qualitative post-model overlays.

Judgements are not typically assessed under each distinct economic scenario used to generate ECL, but instead are applied on the basis of final modelled ECL which reflects the probability weighted view of all scenarios. All adjustments are reviewed quarterly and are subject to internal review and challenge, including by the Audit Committee, to ensure that amounts are appropriately calculated and that there are specific release criteria within a reasonable timeframe.

At 31 December 2020 the coronavirus pandemic and the various support measures that have been put in place have resulted in an economic environment which differs significantly from the historical economic conditions upon which the impairment models have been built. As a result there is a greater need for management judgements to be applied alongside the use of models. At 31 December 2020 management judgement resulted in additional ECL allowances totalling £1,383 million (2019: £153 million). This comprises judgements added due to COVID-19 in the year and other judgements not directly linked to COVID-19 but which have increased in size under the current outlook. The table below analyses total ECL allowance at 31 December 2020 by portfolio, separately identifying the amounts that have been modelled, those that have been individually assessed and those arising through the application of management judgement.


	Modelled ECL	Individually assessed	Judgements due to COVID-191	Other judgements	Total ECL
	£m	£m	£m	£m	£m
At 31 December 2020
UK Mortgages	481	—	36	510	1,027
Other Retail	2,060	—	321	(13)	2,368
Commercial Banking	1,051	1,222	131	(2)	2,402
Other	50	—	400	—	450
Total	3,642	1,222	888	495	6,247
At 31 December 2019
UK Mortgages	386	—	—	183	569
Other Retail	1,531	—	—	(10)	1,521
Commercial Banking	445	890	—	(20)	1,315
Other	50	—	—	—	50
Total	2,412	890	—	153	3,455

The use of payment holidays along with subdued levels of consumer spending is judged to have temporarily reduced the flow of accounts into arrears and default and to have also improved average credit scores across portfolios. Management believes that the resulting position does not fully reflect the underlying credit risk in the portfolios. Adjustments have therefore been made to increase expected future rates of default and remove the impact of the observed improvement in average credit scores.

Modelled LGDs in non-mortgage Retail portfolios are predominantly based on observed customer behaviour and resulting incurred losses. Management believes that this may not be representative of future experience, given the current economic outlook, and consequently an adjustment has been made to increase forward-looking LGDs to reflect a deterioration in cure and recovery rates. The impact has been estimated by using experience of losses in previous downturns and management’s view of relative comparability of anticipated economic scenarios.

Management does not believe that the observed corporate insolvency rates used as an input to Commercial default models adequately reflect the current economic situation and outlook given the temporary government support. As a result, the observed reductions in the rate of insolvencies have been replaced with an increase proportionate to that seen in unemployment to generate a level of predicted defaults.

An important element of the methodology used to calculate the Group’s ECL allowance is the determination of a base case economic scenario, predicated on certain conditioning assumptions, from which alternative scenarios are derived using stochastic shocks. The rapid evolution of the pandemic and significant changes that this has brought about could continue into 2021 and may partially invalidate the conditioning assumptions that underpin the Group’s base case scenario. Management believes that the risks to the conditioning assumptions around the base case scenario are markedly to the downside, reflecting notably the potential for a material delay in the vaccination programme or reduction in its effectiveness from further virus mutation and the corresponding delayed withdrawal of restrictions on social interaction or introduction of further lockdowns. The Group's ECL allowances are required to reflect an unbiased probability-weighted view of all possible future outcomes and therefore management believes that an adjustment is required to capture these additional risks.

An adjustment of £400 million has been made to increase the Group’s ECL allowances to reflect this increased uncertainty around the conditioning assumptions. This equates to a 1 percentage point increase in unemployment allied with a 5 per cent lower HPI in 2021, reflecting a more immediate and therefore greater ECL impact than the gradual increase reflected in the stated univariate sensitivity. It is proportionate to the level of volatility seen in forecasts as the pandemic has unfolded and is also equivalent to a 10 per cent re-weighting from the upside to the severe downside scenario. The adjustment, which has not been allocated to a specific portfolio, has been allocated against Stage 1 assets given the downside risks are largely considered to relate to exposures with currently low default probabilities, the majority of which are in Stage 1. Through 2021 the scale of the uncertainty is expected to diminish and the need for this adjustment will then be reassessed.

Adjustments have been required to the estimated defaults used within the ECL calculation for UK mortgages following the adoption of new default forecast models. Forecast models which predict quarterly defaults based on several economic variables have been developed using the response from the previous recession, as per usual modelling best practice. However, management believe further adjustments are necessary when the results of these models have been benchmarked to observed levels, given the atypical nature of the current economic outlook. These were derived using historical observed default rates under previous downturn conditions to ensure that the resulting forecast best reflected management’s view given the current economic outlook. The adjustment to forward looking parameters prior to their use in ECL calculations ensures that all downstream account level calculations reflect the Group’s best view of credit losses in respect of the economic scenarios stated. As such this in-model adjustment is reflected within all scenarios, assessment of staging and in subsequent assessment of all post-model adjustments.

The current definition of default used in the UK mortgages impairment model excludes past term interest only accounts that continue to make interest payments but have missed their capital payment upon maturity of the loan. This adjustment therefore mitigates the risk that the model understates the credit losses associated with interest-only accounts which have missed, or will potentially miss, their final capital payment. For those accounts that have reached end of term this adjustment manually overwrites PDs to 70 per cent or 100 per cent, thereby moving them into Stage 2, or Stage 3, depending on whether they are deemed performing, or non-performing respectively. For interest-only accounts with six years or less to maturity an appropriate incremental PD uplift is made to PDs based on the probability of missing a future capital payment, assessed through segmentation of behaviour score, debt-to-value and worst ever arrears status. The increase in the judgement in 2020 is primarily driven by an increase in the stock of long-term defaults following COVID-19 related litigation suspension.

The Group suspended mortgage litigation activity between late 2014 and mid 2018 as changes were implemented to the treatment of amounts in arrears, interrupting the natural flow of accounts to possession. An adjustment is made to ensure adequate provision coverage considering the resulting build-up of accounts in long term default. Coverage is uplifted to the equivalent levels of those accounts already in repossession on an estimated shortfall of balances expected to flow to possession. A further adjustment is made to mitigate for the risk that credit model provision understates the probability of possession for accounts which have been in default for more than 24 months, with an arrears balance increase in the last 6 months. These accounts have their probability of possession set to 95 per cent based on observed historical losses incurred on accounts that were of an equivalent status. The increase in judgement in 2020 is primarily driven by an increase in the stock of long-term defaults following COVID-19 related litigation suspension.

Unsecured revolving products use a model lifetime definition of three years based on historic data which shows that substantially all accounts resolve in this time. An adjustment is made to extend the lifetime used for Stage 2 exposures to six years by adding incremental probability of default through the extrapolation of the default trajectory observed throughout the three years and beyond. The resulting additional ECL allowance is added to Stage 2 accounts proportionate to the modelled three year PD. The increase in the judgement in 2020 is driven by growth in Stage 2 assets and their coverage, rather than any change to the lifetime assumption.

Due to a shortcoming in the models, it is not possible to retrieve relevant credit data for a number of accounts and therefore no PD is available and no assessment of whether there has been a SICR can be carried out. The model defaults these accounts to Stage 2 and a proxy ECL allowance calculated based on similar accounts within the portfolio. The deterioration in the economic outlook and growth in the number of accounts subject to this proxy have resulted in this approach having a more significant effect and an exercise has been carried out to identify and adjust those accounts which should not have been allocated to Stage 2.

The MBNA impairment model was developed using historical MBNA data. Following the acquisition of the business and the subsequent migration of this portfolio to Lloyds Banking Group collections strategies an adjustment is required to reflect the recent improvement in cure rates now evident as collections strategies harmonise, which are not captured by the original MBNA model development data.

The value of in-force business asset represents the estimated present value of future profits expected to arise from the portfolio of in-force life insurance and participating investment contracts. The valuation of this asset requires assumptions to be made about future economic and operating conditions which are inherently uncertain and changes could significantly affect the value attributed to this asset. The methodology used to value this asset and the key assumptions that have been made in determining the carrying value of the value of in-force business asset at 31 December 2020 are set out in note 23.

At 31 December 2020, the Group carried total liabilities arising from insurance contracts and participating investment contracts of £116,060 million (2019: £111,449 million). Elements of the valuations of liabilities arising from insurance contracts and participating investment contracts require management to estimate future investment returns, future mortality rates and future expenses. These estimates are subject to significant uncertainty. The methodology used to value these liabilities and the key assumptions that have been made in determining their carrying value are set out in note 30.

The effect on the Group’s profit before tax and shareholders’ equity of changes in annuitant mortality and other key assumptions used in determining the life insurance assets and liabilities is set out in note 31. Reducing annuitant mortality rates to 95 per cent of the expected rate would reduce profit before tax by £333 million (2019: £293 million).

The net asset recognised in the balance sheet at 31 December 2020 in respect of the Group’s defined benefit pension scheme obligations was £1,578 million comprising an asset of £1,714 million and a liability of £136 million (2019: a net asset of £550 million comprising an asset of £681 million and a liability of £131 million). The Group’s accounting policy for its defined benefit pension scheme obligations is set out in note 2(K).

The accounting valuation of the Group’s defined benefit pension schemes’ liabilities requires management to make a number of assumptions. The key areas of estimation uncertainty are the discount rate applied to future cash flows and the expected lifetime of the schemes’ members.

The discount rate is required to be set with reference to market yields at the end of the reporting period on high quality corporate bonds in the currency of and with a term consistent with the defined benefit pension schemes’ obligations. The average duration of the schemes’ obligations is approximately 19 years. The market for bonds with a similar duration is limited and, as a result, significant management judgement is required to determine an appropriate yield curve on which to base the discount rate. Assuming that there is no change in other assumptions or in the value of the schemes' assets, the effect on the net accounting surplus at 31 December 2020 of a decrease of 10 basis points in the discount rate would be a reduction of £890 million (2019: £784 million). To the extent that changes in the discount rate arise from changes in gilt yields, rather than credit spreads, the impact is largely mitigated by the schemes' asset-liability matching strategies.

The cost of the benefits payable by the schemes will also depend upon the life expectancy of the members. The mortality assumptions used by the Group are based on standard industry tables for both current mortality rates and the rate of future mortality improvement, adjusted in line with the actual experience of the Group's schemes. Assuming that there is no change in other assumptions or in the value of the schemes' assets, the effect on the net accounting surplus at 31 December 2020 of an increase in one year in the average life of scheme members would be a reduction of £2,146 million (2019: £1,636 million). The Group has in place a longevity swap, as described in note 34, to partially mitigate mortality risk.

At 31 December 2020 the Group carried deferred tax assets on its balance sheet of £2,741 million (2019: £2,666 million) principally relating to tax losses carried forward. Further information on the Group's deferred tax assets and uncertain tax positions is provided in notes 35 and 46 respectively.

Estimation of income taxes includes the assessment of recoverability of deferred tax assets. Deferred tax assets are only recognised to the extent that they are considered more likely than not to be recoverable based on existing tax laws and forecasts of future taxable profits against which the underlying tax deductions can be utilised. The Group has recognised a deferred tax asset of £4,064 million (2019: £3,611 million) in respect of trading losses carried forward. Substantially all of these losses have arisen in Bank of Scotland plc and Lloyds Bank plc, and they will be utilised as taxable profits arise in those legal entities in future periods.

The Group’s expectations of future UK taxable profits require management judgement, and take into account the Group’s long-term financial and strategic plans and anticipated future tax-adjusting items. In making this assessment, account is taken of business plans, the Board-approved operating plan and the expected future economic outlook as set out in the strategic report, as well as the risks associated with future regulatory change, in order to produce a base case forecast of future UK taxable profits. Under current law there is no expiry date for UK trading losses not yet utilised, and given the forecast of future profitability and the Group’s commitment to the UK market, it is more likely than not the

value of the losses will be recovered at some point in the future. Banking tax losses that arose before 1 April 2015 can only be used against 25 per cent of taxable profits arising after 1 April 2016, and they cannot be used to reduce the surcharge on banking profits. These restrictions in utilisation means that the value of the deferred tax asset in respect of tax losses is only expected to be fully recovered by 2049 (2019: 2039) in the base case forecast. The rate of recovery of the Group’s tax loss asset is not a straight line, being affected by the relative profitability of the legal entities in future periods, and the relative size of their tax losses carried forward. It is expected in the base case that 60 per cent of the value will be recovered by 2034, when Bank of Scotland plc will have utilised all of its available tax losses. It is possible that future tax law changes could materially affect the timing of recovery and the value of these losses ultimately realised by the Group. The value of the deferred tax asset in respect of tax losses increased by £420 million in 2020 as a result of the change in UK tax rates (see note 35).

As disclosed in note 46, the Group has an open matter in relation to a claim for group relief of losses incurred in its former Irish banking subsidiary, where the final tax position will remain uncertain until the matter is finally determined by judicial process.


Key judgements:			Determining the scope of reviews required by regulators
			The impact of legal decisions that may be relevant to claims received
Key estimates:			The number of future complaints
			The proportion of complaints that will be upheld
			The average cost of redress

Determining the amount of the provisions, which represent management’s best estimate of the cost of settling these issues, requires the exercise of significant judgement and estimation. It will often be necessary to form a view on matters which are inherently uncertain, such as the scope of reviews required by regulators, and to estimate the number of future complaints, the extent to which they will be upheld, the average cost of redress and the impact of decisions reached by legal and other review processes that may be relevant to claims received. Consequently the continued appropriateness of the underlying assumptions is reviewed on a regular basis against actual experience and other relevant evidence and adjustments made to the provisions where appropriate.

At 31 December 2020, the carrying value of the Group’s financial instrument assets held at fair value was £248,385 million (2019: £234,467 million), and its financial instrument liabilities held at fair value was £49,959 million (2019: £47,265 million).

The valuation techniques for level 3 financial instruments involve management judgement and estimates the extent of which depends on the complexity of the instrument and the availability of market observable information. In addition, in line with market practice, the Group applies credit, debit and funding valuation adjustments in determining the fair value of its uncollateralised derivative positions. A description of these adjustments is set out in note 48. Details about sensitivities to market risk arising from trading assets and other treasury positions can be found in the risk management section on page 57.


Key judgements:			Assessing future trading conditions that could affect the Group’s business operations
			Assessing whether certain of the Group’s purchased brands have an indefinite life
Key estimates:			The value-in-use calculations require management to estimate future cash flows, appropriate discount rates for those cash flows and long-term sustainable growth rates
			Estimated useful life of internally generated capitalised software

At 31 December 2020, the carrying value of the Group’s other intangible assets was £4,140 million (2019: £3,808 million), including capitalised software enhancements of £3,309 million (2019: £2,907 million) and brands of £380 million (2019: £380 million).

In determining the estimated useful life of capitalised software enhancements, management consider the product's lifecycle and the Group's technology strategy; assets are reviewed annually to assess whether there is any indication of impairment and to confirm that the remaining estimated useful life is still appropriate. For the year ended 31 December 2020, the amortisation charge was £590 million and at 31 December 2020, the weighted-average remaining estimated useful life of the Group’s capitalised software enhancements was 4.9 years (2019: 4.7 years). If the Group reduced by one year the estimated useful life of those assets with a remaining estimated useful life of more than two years at 31 December 2020, the 2021 amortisation charge would be approximately £175 million higher.

Brands arising from the acquisition of Bank of Scotland in 2009 are recognised on the Group's balance sheet and have been determined to have an indefinite useful life. The carrying value at 31 December 2020 was £380 million (2019: £380 million). The recoverable amount has been based on a value-in-use calculation. The calculation uses post-tax projections of the income generated by the brands, a discount rate of 9.31 per cent and a future growth rate of 2.5 per cent . Management estimates that if the growth rate were decreased by 1 per cent there would have been an impairment charge of £50 million.


		2020		2019
	Change in variable	Increase (reduction) in profit before tax	Increase (reduction) in equity	Increase (reduction) in profit before tax	Increase (reduction) in equity
		£m	£m	£m	£m
Non-annuitant mortality and morbidity1	5% reduction	16	13	19	16
Annuitant mortality2	5% reduction	(333)	(270)	(293)	(243)
Lapse rates3	10% reduction	70	57	107	89
Future maintenance and investment expenses4	10% reduction	332	269	299	248
Risk-free rate5	0.25% reduction	37	30	33	28
Guaranteed annuity option take up6	5% addition	(2)	(1)	(1)	(1)
Equity investment volatility7	1% addition	(2)	(2)	(2)	(1)
Widening of credit default spreads8	0.25% addition	(467)	(378)	(424)	(352)
Increase in illiquidity premia9	0.10% addition	219	178	191	159

8This sensitivity shows the impact of a 25 basis point increase in credit default spreads on corporate bonds and the corresponding reduction in market values. Swap curves, the risk-free rate and illiquidity premia are all assumed to be unchanged.

9This sensitivity shows the impact of a 10 basis point increase in the allowance for illiquidity premia. It assumes the overall spreads on assets are unchanged and hence market values are unchanged. Swap curves and the non-annuity risk-free rate are both assumed to be unchanged. The increased illiquidity premium increases the annuity risk-free rate.