Allowance for impairment losses

At 30 June 2020 the Group’s expected credit loss allowance (ECL) was £6,541 million (31 December 2019: £3,455 million), of which £6,040 million (31 December 2019: £3,278 million) was in respect of drawn balances.

The calculation of the Group's ECL allowances and its provisions against loan commitments and guarantees under IFRS 9 requires the Group to make a number of judgements, assumptions and estimates.

Forward-looking information

The measurement of expected credit losses is required to reflect an unbiased probability-weighted range of possible future outcomes. In order to do this, the Group has developed an economic model to project a wide range of key impairment drivers using information derived mainly from external sources. These drivers include factors such as the unemployment rate, the house price index, commercial property prices and corporate credit spreads. The model-generated economic scenarios for the six years beyond 2020 are mapped to industry-wide historical loss data by portfolio. Combined losses across portfolios are used to rank the scenarios by severity of loss.

Alongside a defined central economic scenario, reflecting the Group’s base case assumptions used for medium-term planning purposes, three further economic scenarios are generated to represent the range of future outcomes. The upside, downside and severe downside scenarios are produced by averaging across a group of constituent scenarios around the 15th, 75th and 95th percentiles of the estimated loss distribution around the central case, with the central case expected to lie in the vicinity of the 45th percentile. These locations correspond to scenario weightings that allow for the inclusion of a relatively unlikely severe downside scenario associated with relatively large credit losses. At 31 December 2019 and 30 June 2020, the base case, upside and downside scenarios each carry a 30 per cent weighting, while the severe downside scenario is weighted at 10 per cent. The weights reflect the location of the economic scenarios on the estimated loss distribution.

Following review of the severe downside scenario generated by the modelled approach described above, a judgement was made to increase the severity of GDP and unemployment dispersion from the base case. Whilst the modelled approach gives an unbiased method of creating a loss distribution, it is built on historic experience that does not yet fully capture the unprecedented complexities of the current economic environment and the risk of inflated near-term shocks. The impact of this change has been reflected as a central overlay to reflect the incremental ECL estimated outside the core ECL calculation process. The following economic assumptions include both the modelled severe scenario - used in portfolio level ECL and staging assessment, and the adjusted severe downside - used to generate the final ECL through a central overlay in recognition of more adverse economic outcomes.

The key UK economic assumptions made by the Group are shown below. Compounded growth rates have been calculated on a geometric average basis, they were previously calculated on an arithmetic average basis:

Impact of economic assumptions


				Modelled	Adjusted
	Base case	Upside	Downside	severe	severe
	%	%	%	%	%
At 30 June 2020
GDP	0.4	0.8	0.3	(0.4)	(0.8)
Interest rate	0.15	1.06	0.16	0.03	0.03
Unemployment rate	6.0	5.5	7.1	8.1	8.8
House price growth	0.4	4.7	(4.8)	(9.6)	(9.6)
Commercial real estate price growth	(0.6)	2.7	(3.5)	(8.0)	(8.0)

At 31 December 2019
GDP	1.4	1.7	1.2	0.5	n/a
Interest rate	1.25	2.04	0.49	0.11	n/a
Unemployment rate	4.3	3.9	5.8	7.2	n/a
House price growth	1.0	4.8	(3.2)	(7.7)	n/a
Commercial real estate price growth	0.0	1.8	(3.8)	(7.1)	n/a

The five year averages shown do not demonstrate the extent of peaks and troughs in the stated assumptions over the period. The tables below illustrate the variability of the assumptions from the start of the scenario period to the peak and trough.

Economic assumptions – start to peak


				Modelled	Adjusted
	Base case	Upside	Downside	severe	severe
	%	%	%	%	%
At 30 June 2020
GDP	1.9	4.0	1.7	(1.8)	(2.0)
Interest rate	0.25	1.50	0.21	0.10	0.10
Unemployment rate	9.0	8.6	9.2	9.7	12.5
House price growth	2.1	25.8	0.4	0.4	0.4
Commercial real estate price growth	(2.7)	14.8	(2.7)	(2.7)	(2.7)

At 31 December 2019
GDP	7.0	8.6	6.2	2.7	n/a
Interest rate	1.75	2.56	0.75	0.75	n/a
Unemployment rate	4.6	4.6	6.9	8.3	n/a
House price growth	5.2	26.3	(1.9)	(2.3)	n/a
Commercial real estate price growth	0.1	10.4	(0.6)	(1.1)	n/a

Economic assumptions – start to trough


				Modelled	Adjusted
	Base case	Upside	Downside	severe	severe
	%	%	%	%	%
At 30 June 2020
GDP	(19.7)	(19.5)	(19.8)	(20.2)	(26.1)
Interest rate	0.10	0.10	0.08	0.01	0.01
Unemployment rate	3.9	3.9	3.9	3.9	3.9
House price growth	(6.1)	(3.8)	(21.6)	(39.7)	(39.7)
Commercial real estate price growth	(20.0)	(11.5)	(27.2)	(42.3)	(42.3)

At 31 December 2019
GDP	0.4	0.7	0.2	(2.7)	n/a
Interest rate	0.75	0.75	0.35	0.01	n/a
Unemployment rate	3.8	3.4	3.9	3.9	n/a
House price growth	(2.7)	(0.8)	(14.8)	(33.1)	n/a
Commercial real estate price growth	(0.9)	0.3	(17.5)	(30.9)	n/a

The Group’s base case economic scenario has been materially revised in light of the impact of the COVID-19 pandemic in the UK and globally. The estimated impacts reflect judgments on the net effect of restrictions on economic activity unprecedented in peacetime, large-scale and previously untried government interventions, and lasting behavioural changes by households and businesses.

Although the UK economy has begun to recover as restrictions are eased, there is considerable uncertainty about the pace and eventual extent of the recovery. The Group’s base case assumptions reflect an expectation of some enduring scarring as the economy works through the sharp contraction in economic activity in 2020. Consistent with this, and despite the support provided by the UK Government's Coronavirus Job Retention Scheme and other income and lending assistance, the base case outlook entails a rise in the unemployment rate and weakness in residential and commercial property prices. The Group considers that risks to its base case economic view lie in both directions, reflecting both epidemiological and other developments, including vis-à-vis the UK’s transition to new trading arrangements with the European Union.

Scenarios by year


	2020	2021	2022	2020-22
	%	%	%	%
Base Case
GDP	(10.0)	6.0	3.0	(1.8)
Interest rate	0.10	0.10	0.10	0.10
Unemployment rate	7.2	7.0	5.7	6.7
House price growth	(6.0)	(0.1)	2.9	(3.3)
Commercial real estate price growth	(20.0)	10.0	4.0	(8.5)

Upside
GDP	(9.5)	7.5	3.1	0.3
Interest rate	0.21	1.15	1.42	0.92
Unemployment rate	7.1	6.2	4.9	6.1
House price growth	(3.7)	5.0	9.0	10.2
Commercial real estate price growth	(8.4)	18.6	3.4	12.4

Downside
GDP	(10.2)	5.8	3.1	(2.0)
Interest rate	0.09	0.12	0.19	0.13
Unemployment rate	7.3	7.7	6.8	7.3
House price growth	(8.0)	(6.1)	(4.5)	(17.5)
Commercial real estate price growth	(27.2)	4.0	2.9	(22.1)

Severe downside – Modelled
GDP	(10.9)	3.0	2.2	(6.2)
Interest rate	0.06	0.01	0.02	0.03
Unemployment rate	7.5	8.9	8.4	8.3
House price growth	(9.5)	(11.5)	(11.7)	(29.2)
Commercial real estate price growth	(36.2)	(7.8)	(1.4)	(41.9)

Severe downside - Adjusted
GDP	(17.2)	4.1	5.2	(9.4)
Interest rate	0.06	0.01	0.02	0.03
Unemployment rate	8.0	11.6	9.2	9.6
House price growth	(9.5)	(11.5)	(11.7)	(29.2)
Commercial real estate price growth	(36.2)	(7.8)	(1.4)	(41.9)

Base Case Scenario by Quarter


	2020	2020	2020	2020	2021	2021	2021	2021
	Q1	Q2	Q3	Q4	Q1	Q2	Q3	Q4
	%	%	%	%	%	%	%	%
Base Case
GDP	(1.6)	(19.3)	(10.9)	(8.1)	(4.7)	18.1	7.7	5.1
Interest rate	0.10	0.10	0.10	0.10	0.10	0.10	0.10	0.10
Unemployment rate	3.9	7.5	8.5	9.0	8.0	7.4	6.6	6.2
House price growth	2.8	0.9	(2.4)	(6.0)	(6.3)	(4.0)	(1.1)	(0.1)
Commercial real estate price growth	(5.0)	(12.3)	(19.9)	(20.0)	(14.4)	(3.7)	7.7	10.0

Impact of multiple economic scenarios

The following table 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 (MES). The Group’s probability-weighted ECL allowance continues to reflect a 30 per cent weighting of base case, upside and downside and a 10 per cent weighting of adjusted severe downside. The majority of post-model adjustments and all individually assessed provisions, although assessed on a range of multiple case specific outcomes, are reported flat under each economic scenario. At 30 June 2020 the impact of MES was an increase of £510 million to the base case (31 December 2019: £191 million).


	Probability				Severe
	-weighted	Upside	Base case	Downside	downside
	£m	£m	£m	£m	£m

UK Mortgages	1,111	773	929	1,264	2,214
Other Retail	2,404	2,208	2,383	2,510	2,741
Commercial Banking	2,763	2,416	2,656	2,954	3,553
Other	263	63	63	64	2,064
At 30 June 2020	6,541	5,460	6,031	6,792	10,572

UK Mortgages	569	317	464	653	1,389
Other Retail	1,521	1,443	1,492	1,564	1,712
Commercial Banking	1,315	1,211	1,258	1,382	1,597
Other	50	50	50	50	50
At 31 December 2019	3,455	3,021	3,264	3,649	4,748

Sensitivity of ECL to key economic variables

The table below shows the impact on the Group’s ECL resulting from a decrease/increase in loss given default for a 10 percentage point (pp) increase/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.


	At 30 June 2020		At 31 December 2019
	10pp increase	10 pp decrease	10pp increase	10 pp decrease
	in HPI	in HPI	in HPI	in HPI
ECL impact, £m	(149)	185	(110)	147

The table below shows the impact on the Group’s ECL resulting from a decrease/increase for a 1 percentage point (pp) increase/decrease in the UK unemployment rate. The increase/decrease is presented based on the adjustment phased evenly over the first ten quarters of the base case scenario.


	At 30 June 2020		At 31 December 2019
	1pp increase in	1pp decrease in	1pp increase in	1 pp decrease in
	unemployment	unemployment	unemployment	unemployment
ECL impact, £m	294	(276)	141	(143)

Post-model adjustments

Limitations in the Group’s impairment models or input data may be identified through the on-going 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 the overall provision adequately reflects all material risks. These adjustments are generally determined taking into account the particular attributes of the exposure which have not been adequately captured by the primary impairment models. At 30 June 2020 the incorporation of the changes in the economic outlook required an additional £636 million of post model adjustments; other adjustments increased to £346 million from £161 million at 31 December 2019.


		Economic
		outlook	Other
	Modelled	post-model	post-model	Total
	ECL	adjustments	adjustments	ECL
	£m	£m	£m	£m
UK Mortgages	803	50	258	1,111
Other Retail	2,008	358	38	2,404
Commercial Banking	2,685	28	50	2,763
Other	63	200	–	263
At 30 June 2020	5,559	636	346	6,541

At 31 December 2019	3,294	—	161	3,455

Post-model adjustments amounting to £636 million have been made to incorporate aspects of the updated economic outlook that have not been adequately captured by the models including adjustments to losses given default. The adjusted severe downside scenario has also been incorporated using a post model adjustment.

At 30 June 2020, other post-model adjustments amounted to £346 million of which £258 million relates to UK Mortgages. This comprises increases for the additional end of term risk on interest-only mortgages of £171 million (31 December 2019: £132 million); accounts in long-term default of £34 million (31 December 2019: £33 million); additional risk on forborne accounts, £21 million, and adjustments to possession rate levels, £32 million. In Other Retail post-model adjustments reflect the extension of modelled lifetime on revolving products of £38 million (31 December 2019: £36 million). All post-model adjustments are reviewed at least half-yearly and are subject to strict internal governance and controls.

Significant increase in credit risk

An assessment of whether credit risk has increased significantly since initial recognition considers the change in the risk of default occurring over the remaining expected life of the financial instrument. In determining whether there has been a significant increase in credit risk, the Group uses quantitative tests based on relative and absolute probability of default movements linked to internal credit ratings together with qualitative indicators such as watchlists and other indicators of historical delinquency, credit weakness or financial difficulty. These quantitative tests are carried out on both observed and forward-looking probabilities of default (PDs) to determine whether a customer has triggered the required deterioration appropriate for their PD at origination. For each major product grouping, models have been developed which utilise historical credit loss data to produce probabilities of default for each scenario; and it is the overall weighted-average forward-looking PD that is used to assist in determining the staging of financial assets.

There have been no changes to the quantitative or qualitative triggers used at 30 June 2020. The Group considers these to continue to perform adequately under the current economic conditions and notably with the widespread use of payment holidays. The use of a payment holiday in itself has not been judged to indicate a significant increase in credit risk, 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.

Definition of default

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 Group uses a 90 day past due backstop for all of its products except for UK mortgages wherein a backstop of 180 days past due is in place. The use of payment holidays is not considered to be an automatic trigger of regulatory default and therefore does not automatically trigger Stage 3. Days past due will also not accumulate on any accounts that have taken a payment holiday including those already past due.

Loss given default

The calculation of the ECL allowance also requires an estimate to be made of the loss that will be incurred in the event of a default. The loss given default (LGD) is based on market recovery rates and internal credit assessments. The LGD for customers utilising government funding schemes incorporates an appropriate level of recovery dependent upon the individual scheme and corresponding level of guarantee being used. The use of forecast collateral value indices in determining LGDs continues to be effective despite the temporarily low volumes of transactions upon which those indices are based.