The increase in personal loans and advances to customers was driven by other personal loans growth of $12.5bn, mainly in Hong Kong (up $13.0bn). Mortgages increased by $10.0bn, mainly in the UK (up 5.1bn), Hong Kong (up $2.9bn) and Canada (up $2.0bn). This was partly offset by a decrease of $1.0bn in credit cards, mainly in the US (down $0.4bn) and Hong Kong (down $0.2bn).

During the first six months of 2021, the Group experienced a release in allowances for ECL, which was driven by improving economic forecasts. Excluding foreign exchange movements, the allowance for ECL in relation to loans and advances to customers decreased by $1.6bn from 31 December 2020. This was attributable to:

•a $0.9bn decrease in wholesale loans and advances to customers, of which $0.7bn was driven by stages 1 and 2; and

•a $0.7bn decrease in personal loans and advances to customers, of which $0.6bn was driven by stages 1 and 2.

At 30 June 2021, the allowance for ECL of $13.8bn decreased by $1.9bn compared with 31 December 2020, including favourable foreign exchange movements of $0.1bn. The $13.8bn allowance comprised $13.1bn in respect of assets held at amortised cost, $0.6bn in respect of loan commitments and financial guarantees, and $0.1bn in respect of debt instruments measured at fair value through other comprehensive income (‘FVOCI’).

Stage 3 balances at 30 June 2021 remained broadly stable compared with 31 December 2020.

The ECL release for the first six months of 2021 was $719m, inclusive of recoveries. This comprised: $633m in respect of wholesale lending, of which the stage 3 and purchased or originated credit impaired ('POCI') charge was $196m; $116m in respect of personal lending, of which the stage 3 charge was $221m; and $26m in respect of debt instruments measured at FVOCI, partly offset by a charge of $56m in other financial assets measured at amortised cost. Uncertainty remains as countries emerge from the pandemic at different speeds, government

support measures unwind and new virus strains test the efficacy of vaccination programmes.

During 1H21, we continued to provide Covid-19-related support to customers under the current policy framework. For further details of market-specific measures to support our personal and business customers, see page 75.

Summary of credit risk

The following disclosure presents the gross carrying/nominal amount of financial instruments to which the impairment requirements in IFRS 9 are applied and the associated allowance for ECL.

The following tables analyse loans by industry sector and represent the concentration of exposures on which credit risk is managed. The allowance for ECL decreased from $15.7bn at 31 December 2020 to $13.8bn at 30 June 2021.


Summary of financial instruments to which the impairment requirements in IFRS 9 are applied
	At 30 Jun 2021		At 31 Dec 2020
	Gross carrying/ nominal amount	Allowance for ECL1	Gross carrying/ nominal amount	Allowance for ECL1
	$m	$m	$m	$m
Loans and advances to customers at amortised cost	1,072,375	(12,864)	1,052,477	(14,490)
– personal	482,447	(4,006)	460,809	(4,731)
– corporate and commercial	520,201	(8,640)	527,088	(9,494)
– non-bank financial institutions	69,727	(218)	64,580	(265)
Loans and advances to banks at amortised cost	86,905	(19)	81,658	(42)
Other financial assets measured at amortised cost	854,504	(224)	772,408	(175)
– cash and balances at central banks	393,562	(3)	304,486	(5)
– items in the course of collection from other banks	9,406	—	4,094	—
– Hong Kong Government certificates of indebtedness	41,880	—	40,420	—
– reverse repurchase agreements – non-trading	201,714	—	230,628	—
– financial investments	84,662	(88)	88,719	(80)
– prepayments, accrued income and other assets2	123,280	(133)	104,061	(90)
Total gross carrying amount on-balance sheet	2,013,784	(13,107)	1,906,543	(14,707)
Loans and other credit-related commitments	661,373	(530)	659,783	(734)
– personal	238,559	(23)	236,170	(40)
– corporate and commercial	288,414	(475)	299,802	(650)
– financial	134,400	(32)	123,811	(44)
Financial guarantees	27,274	(64)	18,384	(125)
– personal	919	(1)	900	(1)
– corporate and commercial	21,679	(58)	12,946	(114)
– financial	4,676	(5)	4,538	(10)
Total nominal amount off-balance sheet3	688,647	(594)	678,167	(859)
	2,702,431	(13,701)	2,584,710	(15,566)

	Fair value	Memorandum allowance for ECL4	Fair value	Memorandum allowance for ECL4
	$m	$m	$m	$m

Debt instruments measured at fair value through other comprehensive income (‘FVOCI’)	348,107	(111)	399,717	(141)

1 Total ECL is recognised in the loss allowance for the financial asset unless total ECL exceeds the gross carrying amount of the financial asset, in which case the ECL is recognised as a provision.

2 Includes only those financial instruments that are subject to the impairment requirements of IFRS 9. ‘Prepayments, accrued income and other assets’, as presented within the consolidated balance sheet on page 100, includes both financial and non-financial assets. The 30 June 2021 balances include $2,649m gross carrying amounts and $48m allowances for ECL related to assets held for sale due to the exit of domestic mass market retail banking in the US.

3 Represents the maximum amount at risk should the contracts be fully drawn upon and clients default.

4 Debt instruments measured at FVOCI continue to be measured at fair value with the allowance for ECL as a memorandum item. Change in ECL is recognised in ‘Change for expected credit losses and other credit impairment charges’ in the income statement.

Measurement uncertainty and sensitivity analysis of ECL estimates

There remains a high degree of uncertainty as countries emerge from the pandemic at different speeds, government support measures unwind and new virus strains test the efficacy of vaccination programmes. As a result of this uncertainty, management judgements and estimates reflect a degree of caution. Over 2020, stage 1 and stage 2 ECL provisions on loans increased by $3.9bn, reflecting mainly the evolution of the global pandemic, and while they reduced by $1.5bn in 1H21 as economic conditions recovered, $2.4bn of the 2020 uplift remained at 1H21 ($1.7bn wholesale and $0.7bn retail). Caution is reflected both in the selection of economic scenarios and their weightings, and in the management judgemental adjustments, which reflect how economic conditions interact with modelled outcomes, and are described in more detail below. The highest degree of uncertainty in ECL estimates relates to the UK.

The recognition and measurement of ECL involves the use of significant judgement and estimation. We form multiple economic scenarios based on economic forecasts, apply these assumptions to credit risk models to estimate future credit losses, and

probability-weight the results to determine an unbiased ECL estimate.

Methodology

Four economic scenarios have been used to capture the exceptional nature of the current economic environment and to articulate management’s view of the range of potential outcomes. Scenarios produced to calculate ECL are aligned to HSBC’s top and emerging risks. Three of these scenarios are drawn from consensus forecasts and distributional estimates. The Central scenario is deemed the ‘most likely’ scenario, and usually attracts the largest probability weighting, while the outer scenarios represent the tails of the distribution, which are less likely to occur. The Central scenario is created using the average of a panel of external forecasters, while consensus Upside and Downside scenarios are created with reference to distributions for select markets that capture forecasters’ views of the entire range of outcomes. Management has chosen to use an additional scenario to represent its view of severe downside risks. The use of an additional scenario is in line with HSBC’s forward economic guidance methodology and has been regularly used over the course of 2021. Management may include additional scenarios if it

feels that the consensus scenarios do not adequately capture the top and emerging risks. Unlike the consensus scenarios, these additional scenarios are driven by narrative assumptions, could be country-specific and may incorporate shocks that drive economic activity permanently away from trend.

Description of consensus economic scenarios

The economic assumptions presented in this section have been formed by HSBC, with reference to external forecasts specifically for the purpose of calculating ECL.

Global economic growth is experiencing a recovery in 2021, following an unprecedented contraction in 2020. Restrictions to mobility have started to ease across our key markets, aided in some cases by the successful roll-out of vaccination programmes. Data from vaccinated groups suggests vaccines provide a high level of immunity against the Covid-19 virus despite the emergence of more transmissible variants. To date, vaccinations have shown their effectiveness in lowering hospitalisations and deaths. A rapid roll-out of vaccination programmes has been a key factor enabling economies to reopen and some resumption of travel. The emergence of new variants that reduce the efficacy of vaccines remains a risk.

Economic forecasts are subject to a high degree of uncertainty in the current environment. While risks to the economic outlook are dominated by the progression and management of the pandemic and vaccine roll-out, geopolitical risks also present downside threats. These geopolitical risks include continued differences between the US and China over a range of issues, dampened business sentiment in Hong Kong, and the evolution of the UK’s relationship with the EU. Four global scenarios have been used for the purpose of calculating ECL at 30 June 2021. These are the consensus Central scenario, the consensus Upside scenario, the consensus Downside scenario and an additional Downside scenario.

The scenarios used to calculate ECL in the Interim Report 2021 are described below.

The consensus Central scenario

Following a severe and unprecedented drop in global economic activity in 2020, HSBC’s Central scenario features a sharp recovery in 2021, followed by a subsequent normalisation of growth. The
V-shape in activity over the course of 2020 and 2021 reflects the impact of the pandemic on our key markets, with restrictions to mobility and a reduction in activity resulting in a strong contraction in 2020, and an increase in mobility and resumption in activity in 2021 signalling a recovery.

The Central scenario further assumes that the stringent restrictions on activity, employed across several countries and territories in 2020 and the first half of 2021, will not be repeated. This will allow economic activity to first rebound and then revert to more normal long-run trend rates of growth. Minimal long-term damage to economic prospects is expected. Cross-region differences in the speed and scale of recovery across the forecast horizon reflect timing differences in the progression of the Covid-19 outbreak, different speeds of roll-out of vaccination programmes, national level differences in restrictions imposed and the scale of support measures.

Global GDP is expected to grow by 5.3% in 2021 in the Central scenario. The average rate of global GDP growth is expected to be 3.3% over the forecast period, which is higher than the average growth rate over the five-year period prior to the onset of the pandemic.

The unique circumstances surrounding the current fall in economic activity make it difficult to compare current prospects for global economic activity with previous recessions. However, we note that the depth of the contraction in economic activity and the subsequent recovery are both expected to be sharper than experienced during the last global economic downturn of
2008–2009 across our key markets.

Across the key markets, the Central scenario assumes the following:

•Economic growth is expected to increase sharply in 2021 as governments ease restrictions to mobility, encouraging consumers and firms to spend and invest. GDP is expected to grow across all our major markets in 2021. Country-specific measures aimed at supporting labour markets as economies reopen will affect the rate at which unemployment will decline.

•Inflation is expected to rise in 2021 in line with the economic recovery, before gradually converging back to central bank targets over the forecast period.

•Fiscal deficits are expected to reduce gradually over the course of the projection period from their peak in 2020 following a period where governments, in several of our key markets, provided extensive support to households and corporates. Sovereign indebtedness is expected to remain at high levels.

•Interest rate policy is expected to be highly accommodative over the projection horizon after major central banks lowered their main policy interest rates, implemented emergency support measures for funding markets, and either restarted or increased quantitative easing programmes, in order to support economies and the financial system.

•The West Texas Intermediate oil price is forecast to average $58 per barrel over the projection period.

The Central scenario was first created with forecasts available in May, and subsequently updated in June to reflect significant changes to forecasts. Probability weights assigned to the Central scenario reflect both the higher level of uncertainty in the current global economic environment and relative differences across markets. Weights assigned to the Central scenario vary from 45% to 80%.

The following table describes key macroeconomic variables and the probabilities assigned in the consensus Central scenario.


Central scenario 3Q21–2Q26
	UK	US	Hong Kong	Mainland China	Canada	France	UAE	Mexico
	%	%	%	%	%	%	%	%
GDP growth rate
2021: Annual average growth rate	6.1	6.1	5.2	8.5	5.8	4.9	2.5	4.9
2022: Annual average growth rate	5.5	4.0	3.2	5.5	3.9	3.9	3.8	2.9
2023: Annual average growth rate	2.2	2.3	2.7	5.3	2.3	2.1	3.0	2.3
5-year average	3.0	2.9	2.6	5.0	2.6	2.1	3.6	2.4
Unemployment rate
2021: Annual average rate	5.8	5.5	6.2	3.9	7.6	8.9	2.7	4.5
2022: Annual average rate	5.8	4.3	4.6	3.8	6.3	8.7	2.7	4.2
2023: Annual average rate	5.0	4.0	3.9	3.8	6.1	8.4	2.7	4.1
5-year average	5.1	4.1	4.0	3.8	6.1	8.3	2.7	4.2
House price growth
2021: Annual average growth rate	8.3	11.9	2.6	4.3	16.1	4.5	(3.9)	5.4
2022: Annual average growth rate	2.7	6.2	3.9	6.0	6.4	3.5	(0.7)	5.2
2023: Annual average growth rate	2.5	4.4	2.5	5.4	2.6	4.2	0.3	4.7
5-year average	3.0	5.1	2.9	4.9	4.7	3.5	0.8	4.6
Short-term interest rate
2021: Annual average rate	0.2	0.3	0.9	3.4	0.5	(0.6)	0.8	4.5
2022: Annual average rate	0.3	0.4	1.2	3.4	0.7	(0.6)	0.9	5.5
2023: Annual average rate	0.5	0.7	1.6	3.5	1.2	(0.5)	1.2	6.4
5-year average	0.6	1.1	1.9	3.5	1.4	(0.4)	1.5	6.4
Probability	50	75	70	80	70	45	65	65

The graphs comparing the respective Central scenarios in the second quarters of 2020 and 2021 reveal the extent of economic dislocation that occurred in 2020 and compare current economic expectations with those held a year ago.

GDP growth: Comparison of Central scenarios


UK

Note: Real GDP shown as year-on-year percentage change.


Hong Kong

Note: Real GDP shown as year-on-year percentage change.


US

Note: Real GDP shown as year-on-year percentage change.


Mainland China

Note: Real GDP shown as year-on-year percentage change.

The consensus Upside scenario

Compared with the consensus Central scenario, the consensus Upside scenario features a faster recovery in economic activity during the first two years, before converging to long-run trends.

The scenario is consistent with a number of key upside risk themes. These include the orderly and rapid global abatement of

Covid-19 via successful containment and prompt deployment of vaccines; de-escalation of tensions between the US and China;
de-escalation of political tensions in Hong Kong; continued support from fiscal and monetary policy; and smooth relations between the UK and the EU.

The following table describes key macroeconomic variables and the probabilities assigned in the consensus Upside scenario.


Consensus Upside scenario best outcome
	UK		US		Hong Kong		Mainland China		Canada		France		UAE		Mexico
	%		%		%		%		%		%		%		%
GDP growth rate	11.1	(1Q22)	11.0	(1Q22)	10.0	(2Q22)	13.5	(2Q22)	11.2	(2Q22)	8.3	(2Q22)	14.9	(1Q22)	8.6	(2Q22)
Unemployment rate	3.4	(2Q23)	2.2	(1Q22)	3.2	(2Q23)	3.6	(3Q22)	4.7	(1Q22)	7.2	(3Q22)	2.2	(2Q22)	3.0	(1Q22)
House price growth	9.1	(3Q21)	12.0	(3Q21)	8.9	(4Q21)	13.9	(2Q22)	20.2	(4Q21)	6.1	(3Q22)	21.1	(3Q22)	8.7	(3Q22)
Short-term interest rate	0.2	(3Q21)	0.5	(3Q21)	1.2	(3Q21)	3.4	(3Q21)	0.6	(3Q21)	(0.6)	(1Q22)	1.0	(3Q21)	5.0	(3Q21)
Probability	5		5		5		10		10		5		5		5

Note: Extreme point in the consensus Upside is ‘best outcome’ in the scenario, for example the highest GDP growth and the lowest unemployment rate, in the first two years of the scenario.

The consensus Downside scenario

In the consensus Downside scenario, economic recovery is considerably weaker compared with the Central scenario. GDP growth remains weak, unemployment rates stay elevated and asset and commodity prices fall before gradually recovering towards their long-run trends.

The scenario is consistent with the key downside risks articulated above. Further outbreaks of Covid-19, coupled with delays in vaccination programmes, lead to longer-lasting restrictions on economic activity in this scenario. Other global risks also increase and drive a rise in risk aversion in asset markets.

The following table describes key macroeconomic variables and the probabilities assigned in the consensus Downside scenario.


Consensus Downside scenario worst outcome
	UK		US		Hong Kong		Mainland China		Canada		France		UAE		Mexico
	%		%		%		%		%		%		%		%
GDP growth rate	0.4	(2Q23)	(0.6)	(2Q22)	(6.0)	(1Q22)	(0.7)	(4Q21)	(0.5)	(1Q22)	(1.6)	(3Q21)	(3.3)	(3Q22)	(2.7)	(1Q22)
Unemployment rate	7.3	(2Q22)	6.9	(1Q22)	7.1	(3Q21)	4.1	(4Q21)	8.3	(4Q21)	11.0	(4Q21)	3.2	(2Q22)	5.5	(4Q21)
House price growth	(3.7)	(4Q22)	2.7	(4Q22)	(8.0)	(2Q22)	0.8	(2Q22)	(2.3)	(4Q22)	0.3	(1Q22)	(17.0)	(4Q22)	2.3	(3Q22)
Short-term interest rate	0.2	(2Q23)	0.4	(1Q22)	1.2	(2Q23)	3.1	(3Q21)	0.4	(2Q23)	(0.6)	(3Q21)	0.9	(4Q21)	3.4	(3Q21)
Probability	30		15		20		8		10		35		25		25

Note: Extreme point in the consensus Downside is ‘worst outcome’ in the scenario, for example the lowest GDP growth and the highest unemployment rate, in the first two years of the scenario.

Additional Downside scenario

An additional Downside scenario that features a global recession has been created to reflect management’s view of severe risks. Such a scenario has been in use since 2Q20. In this scenario, infections rise over the second half of 2021, with setbacks to vaccine programmes such that it takes until the end of 2022 for the pandemic to come to an end. The scenario also assumes governments and central banks are unable to significantly increase

fiscal and monetary programmes, which results in a rise in unemployment and a fall in asset prices. In Hong Kong and France, the impacts on the unemployment rate are similar to those in the consensus Downside scenario, reflective of recent historical experiences. GDP growth is stronger in the additional Downside scenario compared with the other scenarios and this stronger bounce-back is a consequence of the deeper initial economic contraction.

The following table describes key macroeconomic variables and the probabilities assigned in the additional Downside scenario.


Additional Downside scenario worst outcome
	UK		US		Hong Kong		Mainland China		Canada		France		UAE		Mexico
	%		%		%		%		%		%		%		%
GDP growth rate	(2.1)	(2Q22)	(4.4)	(2Q22)	(10.6)	(1Q22)	(7.4)	(2Q22)	(4.6)	(2Q22)	(3.1)	(1Q22)	(11.6)	(2Q22)	(7.6)	(2Q22)
Unemployment rate	9.3	(3Q22)	11.0	(2Q23)	7.1	(3Q21)	5.7	(1Q23)	9.8	(2Q22)	11.1	(4Q21)	4.4	(3Q21)	6.1	(4Q22)
House price growth	(7.8)	(2Q22)	(5.7)	(2Q22)	(17.0)	(2Q22)	(20.7)	(2Q22)	(16.1)	(3Q22)	(5.9)	(2Q22)	(18.1)	(2Q22)	0.9	(4Q22)
Short-term interest rate	1.0	(4Q21)	1.3	(4Q21)	2.1	(3Q21)	4.8	(4Q21)	0.5	(3Q21)	0.3	(4Q21)	0.4	(4Q21)	7.2	(4Q21)
Probability	15		5		5		2		10		15		5		5

Note: Extreme point in the additional Downside is ‘worst outcome’ in the scenario, for example the lowest GDP growth and the highest unemployment rate, in the first two years of the scenario.

In considering economic uncertainty and assigning probabilities to scenarios, management has considered both global and country-specific factors. This has led management to assigning scenario probabilities that are tailored to its view of uncertainty in individual markets.

To inform its view, management has considered trends in the progression of the virus in individual countries, the expected reach

and efficacy of vaccine roll-outs over the course of 2021, the size and effectiveness of future government support schemes and the connectivity with other countries. Management has also been guided by the actual response to the Covid-19 outbreak and by the economic experience across countries in 2020. China’s visible success at containing the virus and its repeated rapid response to localised outbreaks, coupled with government support programmes and clear signs of economic recovery, have led

management to conclude that the economic outlook for mainland China is the least volatile out of all our top markets. The Central scenario for mainland China has an 80% probability while a total of 10% has been assigned to the two Downside scenarios. In Hong Kong, the combination of recurrent outbreaks in the recent past, delays to its vaccination programme, evidence of vaccine hesitancy which has delayed the original target of reaching widespread immunity by the end of the third quarter this year, and the other risks outlined above, have led management to assign 25% weight to the two Downside scenarios.

The UK and France face the greatest economic uncertainties in our key markets. In the UK, the discovery of more infectious strains of the virus and subsequent national restrictions on activity imposed before the end of 2020, as well as the current increase in infections, have resulted in considerable uncertainty in the economic outlook. In France, the increases in cases and hospitalisations in the first few months of 2021, the difficulties experienced with the launch of a national vaccination programme and the spread of a more infectious strain of the virus similarly affect the economic outlook. Given these considerations, the consensus Central scenarios for the UK and France have been assigned probabilities of 50% and 45% respectively, while the consensus Downside scenarios have been allocated 30% and 35%. The additional Downside scenario has been assigned 15% probability to each of these markets to reflect the view that the balance of risks is weighted to the downside, and the consensus Upside scenario for these countries has been given a 5% probability.

For the US, Canada and Mexico, connectivity across the three North American economies has been considered. In the UAE, the impact of the oil price on the economy and the ability of non-oil sectors to contribute to economic recovery have influenced the view of uncertainty. The Central scenario has been assigned between 65% and 75% weight for these four markets and, with risks perceived as being weighted to the downside, the two Downside scenarios have been given weights of between 20% and 30%.

The following graphs show the historical and forecasted GDP growth rate for the various economic scenarios in our four largest markets.


US


UK


Hong Kong


Mainland China

Note: Real GDP shown as year-on-year percentage change.

Critical accounting estimates and judgements

The calculation of ECL under IFRS 9 involves significant judgements, assumptions and estimates, as set out in the Annual Report and Accounts 2020 under ‘Critical accounting estimates and judgements’. The level of estimation uncertainty and judgement has remained high since 31 December 2020 as a result of the economic effects of the Covid-19 outbreak, including judgements relating to:

•the selection and weighting of economic scenarios, given rapidly changing economic conditions in an unprecedented manner, uncertainty as to the effect of government and central bank support measures designed to alleviate adverse economic impacts, and a wide distribution of economic forecasts. There is judgement in making assumptions about the length of time and severity of the economic effects of the pandemic and the shape of recovery;

•estimating the economic effects of those scenarios on ECL, when the volatility of economic changes associated with the pandemic are outside the observable historical trends that can be reflected in the models. Modelled assumptions and linkages between economic factors and credit losses may underestimate or overestimate ECL in these conditions, including the effect of real estate prices on modelled ECL outcomes; and

•the identification of customers experiencing significant increases in credit risk and credit impairment, where judgements are made about the extent to which government support programmes have deferred or mitigated the risk of defaults, and the effects once support levels are reduced, particularly in relation to lending in high-risk and vulnerable sectors. Where customers have accepted payment deferrals and other reliefs designed to address short-term liquidity issues, or have extended those deferrals, judgements include the extent to which they are able to meet their financial obligations on returning to their original terms. The use of segmentation techniques for indicators of significant increases in credit risk for retail customers involves estimation uncertainty.

How economic scenarios are reflected in ECL calculations

The methodologies for the application of forward economic guidance into the calculation of ECL for wholesale and retail loans and portfolios are set out on page 132 of the Annual Report and Accounts 2020. Models are used to reflect economic scenarios on ECL estimates. These models are based largely on historical observations and correlations with default rates.

We continue to observe volatility in macroeconomic variables as a result of the Covid-19 pandemic, which – together with significant governmental support programmes, forbearance and payment holidays – have impacted model performance and historical correlations between macroeconomic variables and defaults. As economic forecasts begin to improve, the level and speed of economic recovery remains outside the range of historical experience used to calibrate the models, and the timing of defaults has considerably shifted from the modelled assumptions. Management judgements have been used to overcome the limitations in the model generated outcome, increasing the ECL.

Management judgemental adjustments arise when data and model limitations are addressed in the short term using in-model and post-model adjustments. This includes refining model inputs and outputs and using post-model adjustments based on management judgement and higher level quantitative analysis for impacts that are difficult to model.

Management judgemental adjustments

In the context of IFRS 9, management judgemental adjustments are typically short-term increases or decreases to the ECL at either a customer or portfolio level to account for late-breaking events, model deficiencies and other assessments applied during management review and challenge.

At 30 June 2021, management judgements were applied to reflect credit risk dynamics not captured by our models. The drivers of the management judgemental adjustments continue to evolve with the economic environment. We have internal governance in place to monitor management judgemental adjustments regularly and, where possible, to reduce the reliance on these through model recalibration or redevelopment, as appropriate.

Wider-ranging model changes will take time to develop and need observable loss data on which models can be developed. Models will be revisited over time once the longer-term impacts of the Covid-19 outbreak are observed. Therefore, we continue to anticipate significant management judgemental adjustments for the foreseeable future.

Judgemental adjustments, which primarily relate to delays in the timing and extent of defaults, will likely cease to occur when macroeconomic forecasts have stabilised and move within the range of historical experience, portfolio impacts due to unwinding of government schemes become visible and the uncertainty due to Covid-19 reduces.

The wholesale and retail management judgemental adjustments are presented as part of the global business impairment committees with representation from Model Risk Management. This is in line with the governance process as set out on page 120 of the Annual Report and Accounts 2020.

Management judgemental adjustments made in estimating the reported ECL at 30 June 2021 are set out in the following table. The table includes adjustments in relation to data and model limitations resulting from the pandemic, and as a result of the regular process of model development and implementation. It shows the adjustments applicable to the scenario-weighted ECL numbers.


Management judgemental adjustments to ECL at 30 June 20211
	Retail	Wholesale	Total
	$bn	$bn	$bn

Low-risk counterparties (banks, sovereigns and government entities)	0.1	(0.8)	(0.7)
Corporate lending adjustments		1.4	1.4
Retail lending probability of default adjustments	(0.1)		(0.1)
Retail model default timing adjustments	0.4		0.4
Macroeconomic-related adjustments	0.6		0.6
Other retail lending adjustments	0.3		0.3
Total	1.3	0.7	2.0


Management judgemental adjustments to ECL at 31 December 20201
	Retail	Wholesale	Total
	$bn	$bn	$bn

Low-risk counterparties (banks, sovereigns and government entities)	—	(0.7)	(0.7)
Corporate lending adjustments		0.5	0.5
Retail lending probability of default adjustments	(0.8)		(0.8)
Retail model default timing adjustments	1.9		1.9
Macroeconomic-related adjustments	0.1		0.1
Other retail lending adjustments	0.3		0.3
Total	1.5	(0.2)	1.3

1 Management judgemental adjustments presented in the table reflect increases or (decreases) to ECL, respectively.

In the wholesale portfolio, management judgemental adjustments were an ECL increase of $0.7bn (31 December 2020: $0.2bn decrease).

The adjustments relating to wholesale low-credit risk exposures decreased ECL by $0.8bn at 30 June 2021 (31 December 2020: $0.7bn decrease). These were mainly to highly rated banks, sovereigns and US government-sponsored entities, where modelled credit factors did not fully reflect the underlying fundamentals of these entities or the effect of government support and economic programmes in the Covid-19 environment.

Adjustments to corporate exposures increased ECL by $1.4bn at 30 June 2021 (31 December 2020: $0.5bn increase). These principally reflected the outcome of management judgements for high-risk and vulnerable sectors in some of our key markets, supported by credit experts’ input, quantitative analyses and

benchmarks. Considerations include uncertainty around vaccine efficacy and risk of new variants and, uncertainty around timing and extent of defaults in some sectors due to government intervention. The increase in adjustment impact relative to
31 December 2020 was mostly driven by management judgements as a result of further improvement of macroeconomic scenarios and increased dislocation of modelled outcomes to management expectations for high-risk and vulnerable sectors.

In the retail portfolio, management judgemental adjustments were an ECL increase of $1.3bn at 30 June 2021 (31 December 2020: $1.5bn increase).

The retail model default timing adjustment increased ECL by $0.4bn (31 December 2020: $1.9bn increase). This was applied in several economies as customer relief and government support programmes continue to delay the emergence of defaults. The

level of adjustment decreased during the period given the improvement in macroeconomic forecasts and the unwinding in a number of markets as customer relief and government support concludes. Retail models are reliant on the assumption that as macroeconomic conditions deteriorate, defaults will crystallise. We will monitor the continuation of customer relief and government support programmes that have stabilised macroeconomic conditions and therefore the timing of retail model defaults.

The retail lending probability of default adjustments decreased ECL by $0.1bn (31 December 2020: $0.8bn decrease). These related to severe projections of macroeconomic variables that are outside the historical observations on which IFRS 9 models have been built and calibrated to operate. The majority of scenarios are now within historical observations leading to lower levels of adjustment.

Macroeconomic-related adjustments increased ECL by $0.6bn
(31 December 2020: $0.1bn increase). These were applied to reflect management’s expectation in regards to the extent of previously forecast defaults that have not yet emerged in the retail portfolio in the context of improvements in the macroeconomic forecasts.

Other retail lending adjustments increased ECL by $0.3bn (31 December 2020: $0.3bn increase), reflecting those who remain in or have recently exited customer support programmes and all other data and model adjustments.

Economic scenarios sensitivity analysis of ECL estimates

Management considered the sensitivity of the ECL outcome against the economic forecasts as part of the ECL governance process by recalculating the ECL under each scenario described above for selected portfolios, applying a 100% weighting to each scenario in turn. The weighting is reflected in both the determination of a significant increase in credit risk and the measurement of the resulting ECL.

The ECL calculated for the Upside and Downside scenarios should not be taken to represent the upper and lower limits of possible

ECL outcomes. The impact of defaults that might occur in the future under different economic scenarios is captured by recalculating ECL for loans in stages 1 and 2 at the balance sheet date. The population of stage 3 loans (in default) at the balance sheet date is unchanged in these sensitivity calculations. Stage 3 ECL would only be sensitive to changes in forecasts of future economic conditions if the loss-given default of a particular portfolio was sensitive to these changes.

There is a particularly high degree of estimation uncertainty in numbers representing tail risk scenarios when assigned a 100% weighting.

For wholesale credit risk exposures, the sensitivity analysis excludes ECL for financial instruments related to defaulted obligors because the measurement of ECL is relatively more sensitive to credit factors specific to the obligor than future economic scenarios. Therefore, it is impracticable to separate the effect of macroeconomic factors in individual assessments.

For retail credit risk exposures, the sensitivity analysis includes ECL for loans and advances to customers related to defaulted obligors. This is because the retail ECL for secured mortgage portfolios, including loans in all stages, is sensitive to macroeconomic variables.

Wholesale and retail sensitivity

The wholesale and retail sensitivity analysis is stated inclusive of management judgemental adjustments, as appropriate to each scenario. The results tables exclude portfolios held by the insurance business and small portfolios, and as such cannot be directly compared with personal and wholesale lending presented in other credit risk tables. Additionally, in both the wholesale and retail analysis, the comparative period results for additional/ alternative Downside scenarios are also not directly comparable with the current period, because they reflect different risk profiles relative to the consensus scenarios for the period end.

Wholesale analysis


IFRS 9 ECL sensitivity to future economic conditions1, 2
	Gross carrying amount	Reported ECL	Central scenario ECL	Upside scenario ECL	Downside scenario ECL	Additional Downside scenario ECL
By geography at 30 Jun 2021	$m	$m	$m	$m	$m	$m

UK	481,849	1,740	1,433	1,083	1,888	2,949
US	229,768	270	253	185	319	490
Hong Kong	436,443	429	377	211	557	892
Mainland China	124,547	98	75	16	224	1,192
Canada	84,398	146	115	66	176	361
Mexico	24,971	180	161	123	210	358
UAE	45,997	201	164	119	257	607
France	197,175	133	115	107	147	233




By geography at 31 Dec 2020

UK	430,555	2,077	1,514	1,026	2,271	3,869
US	201,263	369	314	219	472	723
Hong Kong	452,983	474	388	211	672	1,363
Mainland China	118,163	116	93	28	252	1,158
Canada	85,720	183	140	82	253	528
Mexico	25,920	246	222	177	285	437
UAE	44,777	250	241	190	330	536
France	164,899	117	109	97	131	238

1 ECL sensitivity includes off-balance sheet financial instruments that are subject to significant measurement uncertainty.

2 Includes low credit-risk financial instruments such as debt instruments at FVOCI, which have high carrying amounts but low ECL under all the above scenarios.

At 30 June 2021, the most significant level of ECL sensitivity was observed in the UK, mainland China and Hong Kong. This higher sensitivity was largely driven by significant exposure in these regions and more severe impacts of the Downside scenarios relative to the Central and probability-weighted scenarios. For

mainland China, the additional Downside scenario weighting of 2% reflected a scenario that is considered highly unlikely and is significantly more adverse compared with the Central scenario, resulting in a higher ECL estimate relative to the reported and Central scenarios.

Retail analysis


IFRS 9 ECL sensitivity to future economic conditions1,2
	Gross carrying amount	Reported ECL	Central scenario ECL	Upside scenario ECL	Downside scenario ECL	Additional Downside scenario
By geography at 30 June 2021	$m	$m	$m	$m	$m	$m
UK
Mortgages	151,435	205	199	193	210	221
Credit cards	7,563	678	625	533	767	945
Other	8,460	713	631	534	779	934
Mexico
Mortgages	4,427	112	104	93	125	142
Credit cards	1,044	180	175	163	198	214
Other	2,626	395	380	357	412	437
Hong Kong
Mortgages	93,283	—	—	—	—	—
Credit cards	7,414	244	231	213	269	384
Other	5,787	102	96	89	119	141
UAE
Mortgages	1,902	55	52	44	60	64
Credit cards	395	73	66	61	80	95
Other	613	28	28	26	29	31
France
Mortgages	23,583	61	60	60	61	62
Other	1,668	75	75	74	76	77
US
Mortgages	15,283	41	41	40	42	48
Credit cards	938	162	161	160	164	178
Canada
Mortgages	24,681	31	30	27	32	38
Credit cards	232	10	10	10	10	10
Other	1,699	21	21	20	23	25




By geography at 31 December 2020
UK
Mortgages	146,478	197	182	172	205	221
Credit cards	7,869	857	774	589	904	1,084
Other	9,164	897	795	471	1,022	1,165
Mexico
Mortgages	3,896	111	101	79	136	167
Credit cards	1,113	260	255	243	269	290
Other	2,549	436	428	411	451	491
Hong Kong
Mortgages	89,943	—	—	—	—	—
Credit cards	7,422	266	259	247	277	405
Other	6,020	112	105	102	115	130
UAE
Mortgages	1,889	66	63	53	73	78
Credit cards	426	92	81	62	107	126
Other	683	38	37	33	41	46
France
Mortgages	24,565	68	68	68	69	70
Other	1,725	88	87	85	88	91
US
Mortgages	15,399	41	39	38	41	53
Credit cards	570	86	84	81	88	119
Canada
Mortgages	22,454	31	30	29	31	36
Credit cards	260	9	9	8	9	9
Other	1,775	22	21	20	24	28

1 ECL sensitivities exclude portfolios utilising less complex modelling approaches.

2 At 30 June 2021, US sensitivity includes the implementation of an enhanced model.

At 30 June 2021, the most significant level of ECL sensitivity was observed in the UK, Mexico and Hong Kong. Mortgages reflected the lowest level of ECL sensitivity across most markets as collateral values remain resilient. Hong Kong mortgages had low levels of reported ECL due to the credit quality of the portfolio, and so presented sensitivity was negligible. Credit cards and other unsecured lending are more sensitive to economic forecasts, which have reflected improvements during the first half of 2021.

Group ECL sensitivity results

The ECL impact of the scenarios and management judgemental adjustments are highly sensitive to movements in economic forecasts, including the efficacy of government support measures. Based upon the sensitivity tables presented above, if the Group ECL balance (excluding wholesale stage 3, which is assessed individually) was estimated solely on the basis of the Central scenario, Upside scenario, Downside scenario or the additional Downside scenario at 30 June 2021, it would increase/(decrease) as presented in the below table.


	Retail1	Wholesale1
Total Group ECL at 30 June 2021	$bn	$bn
Reported ECL	3.9	3.7
Scenarios
100% consensus Central scenario	(0.2)	(0.6)
100% consensus Upside scenario	(0.5)	(1.4)
100% consensus Downside scenario	0.3	0.7
100% additional Downside scenario	1.1	4.5



Total Group ECL at 31 December 2020
Reported ECL	4.5	4.5
Scenarios
100% consensus Central scenario	(0.3)	(0.9)
100% consensus Upside scenario	(1.0)	(2.0)
100% consensus Downside scenario	0.3	1.0
100% alternative Downside scenario	1.3	5.9

1 On the same basis as retail and wholesale sensitivity analysis.

At 30 June 2021, Group ECL sensitivity decreased across all scenarios compared with 31 December 2020, driven by the improvement of macroeconomic forecasts.

There still remains a significant degree of uncertainty in relation to the UK economic outlook. If a 100% weight were applied to the consensus Downside and additional Downside scenario for the UK, respectively, it would result in an increase in ECL of $0.1bn and $1.2bn in wholesale and $0.2bn and $0.5bn in retail.

Reconciliation of changes in gross carrying/nominal amount and allowances for loans and advances to banks and customers

The following disclosure provides a reconciliation by stage of the Group’s gross carrying/nominal amount and allowances for loans and advances to banks and customers, including loan commitments and financial guarantees. Movements are calculated on a quarterly basis and therefore fully capture stage movements between quarters. If movements were calculated on a year-to-date basis they would only reflect the opening and closing position of the financial instrument.

The transfers of financial instruments represent the impact of stage transfers upon the gross carrying/nominal amount and associated allowance for ECL.

The net remeasurement of ECL arising from stage transfers represents the increase or decrease due to these transfers, for example, moving from a 12-month (stage 1) to a lifetime (stage 2) ECL measurement basis. Net remeasurement excludes the underlying customer risk rating (‘CRR’)/probability of default (‘PD’) movements of the financial instruments transferring stage. This is captured, along with other credit quality movements in the ‘changes in risk parameters – credit quality’ line item.

Changes in ‘New financial assets originated or purchased’, ‘assets derecognised (including final repayments)’ and ‘changes to risk parameters – further lending/repayments’ represent the impact from volume movements within the Group’s lending portfolio.


Reconciliation of changes in gross carrying/nominal amount and allowances for loans and advances to banks and customers including loan commitments and financial guarantees
	Non-credit impaired				Credit impaired
	Stage 1		Stage 2		Stage 3		POCI		Total
	Gross carrying/ nominal amount	Allowance for ECL	Gross carrying/ nominal amount	Allowance for ECL	Gross carrying/ nominal amount	Allowance for ECL	Gross carrying/ nominal amount	Allowance for ECL	Gross carrying/ nominal amount	Allowance for ECL
	$m	$m	$m	$m	$m	$m	$m	$m	$m	$m
At 1 Jan 2021	1,506,451	(2,331)	223,432	(5,403)	20,424	(7,544)	279	(113)	1,750,586	(15,391)
Transfers of financial instruments:	(9,572)	(993)	6,141	1,411	3,431	(418)	—	—	—	—
– transfers from stage 1 to stage 2	(88,050)	273	88,050	(273)	—	—	—	—	—	—
– transfers from stage 2 to stage 1	79,249	(1,239)	(79,249)	1,239	—	—	—	—	—	—
– transfers to stage 3	(1,092)	18	(3,248)	522	4,340	(540)	—	—	—	—
– transfers from stage 3	321	(45)	588	(77)	(909)	122	—	—	—	—
Net remeasurement of ECL arising from transfer of stage	—	625	—	(320)	—	(12)	—	—	—	293

New financial assets originated or purchased	224,124	(268)	—	—	—	—	1	—	224,125	(268)
Assets derecognised (including final repayments)	(143,261)	82	(15,521)	258	(1,559)	234	(8)	6	(160,349)	580
Changes to risk parameters – further lending/repayments	(25,456)	285	(3,440)	285	(808)	348	(44)	4	(29,748)	922

Change in risk parameters – credit quality	—	609	—	(397)	—	(1,229)	—	28	—	(989)
Changes to models used for ECL calculation	—	12	—	(12)	—	—	—	—	—	—
Assets written off	—	—	—	—	(1,356)	1,352	(9)	1	(1,365)	1,353
Credit-related modifications that resulted in derecognition	—	—	—	—	(1)	—	—	—	(1)	—

Foreign exchange	(3,130)	(7)	(91)	(19)	(92)	35	(1)	—	(3,314)	9
Other	(180)	7	(20)	5	(7)	2	—	—	(207)	14
At 30 Jun 2021	1,548,976	(1,979)	210,501	(4,192)	20,032	(7,232)	218	(74)	1,779,727	(13,477)
ECL income statement change for the period		1,345		(186)		(659)		38		538
Recoveries										209



Other										(41)
Total ECL income statement change for the period										706



	At 30 Jun 2021		6 months ended 30 Jun 2021
	Gross carrying/nominal amount	Allowance for ECL	ECL release/(charge)
	$m	$m	$m
As above	1,779,727	(13,477)	706
Other financial assets measured at amortised cost	854,504	(224)	(56)
Non-trading reverse purchase agreement commitments	68,200	—	—
Performance and other guarantees	—	—	43
Summary of financial instruments to which the impairment requirements in IFRS 9 are applied/Summary consolidated income statement	2,702,431	(13,701)	693
Debt instruments measured at FVOCI	348,107	(111)	26
Total allowance for ECL/total income statement ECL charge for the period	n/a	(13,812)	719

As shown in the previous table, the allowance for ECL for loans and advances to customers and banks and relevant loan commitments and financial guarantees decreased by $1,914m during the period, from $15,391m at 31 December 2020 to $13,477m at 30 June 2021.

This decrease was driven by:

•$1,353m of assets written off;

•$1,234m relating to volume movements, which included the ECL allowance associated with new originations, assets derecognised and further pending repayment;

•$293m relating to the net remeasurement impact of stage transfers; and

•foreign exchange and other movements of $23m.

This decrease was offset by $989m relating to underlying credit quality changes, including the credit quality impact of financial instruments transferring between stages.

The ECL release for the period of $538m presented in the previous table consisted of $1,234m relating to underlying net book volume and $293m relating to the net remeasurement impact of stage transfers. These were partly offset by $989m relating to underlying credit quality changes, including the credit quality impact of financial instruments transferring between stages.


Reconciliation of changes in gross carrying/nominal amount and allowances for loans and advances to banks and customers including loan commitments and financial guarantees (continued)
	Non-credit impaired				Credit impaired
	Stage 1		Stage 2		Stage 3		POCI		Total
	Gross carrying/ nominal amount	Allowance for ECL	Gross carrying/ nominal amount	Allowance for ECL	Gross carrying/ nominal amount	Allowance for ECL	Gross carrying/ nominal amount	Allowance for ECL	Gross carrying/ nominal amount	Allowance for ECL
	$m	$m	$m	$m	$m	$m	$m	$m	$m	$m
At 1 Jan 2020	1,561,613	(1,464)	105,551	(2,441)	14,335	(5,121)	345	(99)	1,681,844	(9,125)
Transfers of financial instruments:	(129,236)	(1,122)	116,783	1,951	12,453	(829)	—	—	—	—
– transfers from stage 1 to stage 2	(298,725)	947	298,725	(947)	—	—	—	—	—	—
– transfers from stage 2 to stage 1	172,894	(2,073)	(172,894)	2,073	—	—	—	—	—	—
– transfers to stage 3	(3,942)	30	(10,320)	986	14,262	(1,016)	—	—	—	—
– transfers from stage 3	537	(26)	1,272	(161)	(1,809)	187	—	—	—	—
Net remeasurement of ECL arising from transfer of stage	—	907	—	(1,158)	—	(750)	—	—	—	(1,001)
New financial assets originated or purchased	437,836	(653)	—	—	—	—	25	(1)	437,861	(654)
Assets derecognised (including final repayments)	(313,347)	160	(37,409)	464	(3,430)	485	(23)	2	(354,209)	1,111
Changes to risk parameters – further lending/repayment	(83,147)	157	29,092	85	(597)	248	(50)	(2)	(54,702)	488
Changes in risk parameters – credit quality	—	(408)	—	(4,374)	—	(4,378)	—	(39)	—	(9,199)
Changes to models used for ECL calculation	—	134	—	294	—	5	—	—	—	433
Assets written off	—	—	—	—	(2,946)	2,944	(30)	30	(2,976)	2,974
Credit-related modifications that resulted in derecognition	—	—	—	—	(23)	7	—	—	(23)	7
Foreign exchange	32,808	(47)	9,123	(223)	633	(163)	4	(3)	42,568	(436)
Other	(76)	5	292	(1)	(1)	8	8	(1)	223	11
At 31 Dec 2020	1,506,451	(2,331)	223,432	(5,403)	20,424	(7,544)	279	(113)	1,750,586	(15,391)
ECL income statement change for the period		297		(4,689)		(4,390)		(40)		(8,822)
Recoveries										326


Others										(84)
Total ECL income statement change for the period1										(8,580)



	At 31 Dec 2020		12 months ended 31 Dec 2020
	Gross carrying/nominal amount	Allowance for ECL	ECL charge
	$m	$m	$m
As above	1,750,586	(15,391)	(8,580)
Other financial assets measured at amortised cost	772,408	(175)	(95)
Non-trading reverse purchase agreement commitments	61,716	—	—
Performance and other guarantees	—	—	(94)
Summary of financial instruments to which the impairment requirements in IFRS 9 are applied/Summary consolidated income statement	2,584,710	(15,566)	(8,769)
Debt instruments measured at FVOCI	399,717	(141)	(48)
Total allowance for ECL/total income statement ECL charge for the period	n/a	(15,707)	(8,817)

1 The 31 December 2020 total ECL income statement change of $8,580m is attributable to $6,464m for the six months ended 30 June 2020 and $2,116m to the six months ended 31 December 2020.

Credit quality of financial instruments

We assess the credit quality of all financial instruments that are subject to credit risk. The credit quality of financial instruments is a point-in-time assessment of PD, whereas stages 1 and 2 are determined based on relative deterioration of credit quality since initial recognition. Accordingly, for non-credit-impaired financial instruments, there is no direct relationship between the credit quality assessment and stages 1 and 2, though typically the lower credit quality bands exhibit a higher proportion in stage 2.

The five credit quality classifications each encompass a range of granular internal credit rating grades assigned to wholesale and personal lending businesses and the external ratings attributed by external agencies to debt securities, as shown in the following table. Personal lending credit quality is disclosed based on a
12-month point-in-time PD adjusted for multiple economic scenarios. The credit quality classifications for wholesale lending are based on internal credit risk ratings.


Distribution of financial instruments to which the impairment requirements in IFRS 9 are applied, by credit quality and stage allocation

	Gross carrying/nominal amount						Allowance for ECL	Net
	Strong	Good	Satisfactory	Sub- standard	Credit impaired	Total
	$m	$m	$m	$m	$m	$m	$m	$m
Loans and advances to customers at amortised cost	528,036	232,101	258,457	34,496	19,285	1,072,375	(12,864)	1,059,511
– stage 1	522,137	204,332	164,251	4,826	—	895,546	(1,770)	893,776
– stage 2	5,899	27,769	94,206	29,670	—	157,544	(3,888)	153,656
– stage 3	—	—	—	—	19,069	19,069	(7,131)	11,938
– POCI	—	—	—	—	216	216	(75)	141
Loans and advances to banks at amortised cost	77,928	4,819	3,155	1,003	—	86,905	(19)	86,886
– stage 1	77,739	4,662	3,062	23	—	85,486	(16)	85,470
– stage 2	189	157	93	980	—	1,419	(3)	1,416
– stage 3	—	—	—	—	—	—	—	—
– POCI	—	—	—	—	—	—	—	—
Other financial assets measured at amortised cost	751,028	71,664	30,323	1,163	326	854,504	(224)	854,280
– stage 1	750,477	70,918	27,469	114	—	848,978	(130)	848,848
– stage 2	551	746	2,854	1,049	—	5,200	(38)	5,162
– stage 3	—	—	—	—	284	284	(47)	237
– POCI	—	—	—	—	42	42	(9)	33
Loan and other credit-related commitments	408,027	150,779	91,284	10,537	746	661,373	(530)	660,843
– stage 1	404,424	139,198	67,581	1,766	—	612,969	(184)	612,785
– stage 2	3,603	11,581	23,703	8,771	—	47,658	(266)	47,392
– stage 3	—	—	—	—	744	744	(80)	664
– POCI	—	—	—	—	2	2	—	2
Financial guarantees	15,693	5,018	5,125	1,216	222	27,274	(64)	27,210
– stage 1	15,592	4,178	3,151	248	—	23,169	(17)	23,152
– stage 2	101	840	1,974	968	—	3,883	(30)	3,853
– stage 3	—	—	—	—	221	221	(17)	204
– POCI	—	—	—	—	1	1	—	1
At 30 Jun 2021	1,780,712	464,381	388,344	48,415	20,579	2,702,431	(13,701)	2,688,730
Debt instruments at FVOCI1
– stage 1	317,105	13,039	11,126	—	—	341,270	(65)	341,205
– stage 2	791	93	321	741	—	1,946	(15)	1,931
– stage 3	—	—	—	—	226	226	(24)	202
– POCI	—	—	—	—	48	48	(7)	41
At 30 Jun 2021	317,896	13,132	11,447	741	274	343,490	(111)	343,379

1 For the purposes of this disclosure, gross carrying value is defined as the amortised cost of a financial asset, before adjusting for any loss allowance. As such, the gross carrying value of debt instruments at FVOCI will not reconcile to the balance sheet as it excludes fair value gains and losses.


Distribution of financial instruments to which the impairment requirements in IFRS 9 are applied, by credit quality and stage allocation (continued)
	Gross carrying/notional amount
	Strong	Good	Satisfactory	Sub- standard	Credit impaired	Total	Allowance for ECL	Net
	$m	$m	$m	$m	$m	$m	$m	$m
Loans and advances to customers at amortised cost	506,231	233,320	256,584	36,970	19,372	1,052,477	(14,490)	1,037,987
– stage 1	499,836	199,138	165,507	5,439	—	869,920	(1,974)	867,946
– stage 2	6,395	34,182	91,077	31,531	—	163,185	(4,965)	158,220
– stage 3	—	—	—	—	19,095	19,095	(7,439)	11,656
– POCI	—	—	—	—	277	277	(112)	165
Loans and advances to banks at amortised cost	71,318	5,496	3,568	1,276	—	81,658	(42)	81,616
– stage 1	71,126	5,098	3,357	73	—	79,654	(33)	79,621
– stage 2	192	398	211	1,203	—	2,004	(9)	1,995
– stage 3	—	—	—	—	—	—	—	—
– POCI	—	—	—	—	—	—	—	—
Other financial assets measured at amortised cost	683,231	61,768	26,581	611	217	772,408	(175)	772,233
– stage 1	682,412	61,218	24,532	54	—	768,216	(80)	768,136
– stage 2	819	550	2,049	557	—	3,975	(44)	3,931
– stage 3	—	—	—	—	177	177	(42)	135
– POCI	—	—	—	—	40	40	(9)	31
Loan and other credit-related commitments	400,911	157,339	90,784	9,668	1,081	659,783	(734)	659,049
– stage 1	396,028	143,600	63,592	1,265	—	604,485	(290)	604,195
– stage 2	4,883	13,739	27,192	8,403	—	54,217	(365)	53,852
– stage 3	—	—	—	—	1,080	1,080	(78)	1,002
– POCI	—	—	—	—	1	1	(1)	—
Financial guarantees	6,356	5,194	5,317	1,247	270	18,384	(125)	18,259
– stage 1	6,286	4,431	3,163	210	—	14,090	(37)	14,053
– stage 2	70	763	2,154	1,037	—	4,024	(62)	3,962
– stage 3	—	—	—	—	269	269	(26)	243
– POCI	—	—	—	—	1	1	—	1
At 31 Dec 2020	1,668,047	463,117	382,834	49,772	20,940	2,584,710	(15,566)	2,569,144

Debt instruments at FVOCI1
– stage 1	367,542	12,585	10,066	—	—	390,193	(88)	390,105
– stage 2	143	93	343	825	—	1,404	(20)	1,384
– stage 3	—	—	—	—	257	257	(23)	234
– POCI	—	—	—	—	49	49	(10)	39
At 31 Dec 2020	367,685	12,678	10,409	825	306	391,903	(141)	391,762


Trading VaR, 99% 1 day
	Foreign exchange and commodity	Interest rate	Equity	Credit spread	Portfolio diversification1	Total
	$m	$m	$m	$m	$m	$m
Half-year to 30 Jun 2021	13.6	33.5	15.8	18.3	(42.5)	38.7
Average	15.0	33.4	16.5	18.1	(46.2)	36.8
Maximum	31.8	50.4	24.3	29.4		48.2
Minimum	6.9	18.5	12.1	12.2		31.1


Half-year to 30 Jun 2020	10.4	36.8	26.3	18.7	(47.8)	44.4
Average	10.6	27.6	25.0	23.1	(36.2)	50.1
Maximum	19.9	43.5	41.3	44.1		69.3
Minimum	5.6	19.1	13.6	13.7		33.6


Half-year to 31 Dec 2020	13.7	20.3	21.5	24.3	(36.4)	43.4
Average	11.3	25.5	29.0	20.2	(40.2)	45.8
Maximum	25.7	38.1	40.5	28.6		62.2
Minimum	7.1	19.8	21.1	12.6		37.2

1 When VaR is calculated at a portfolio level, natural offsets in risk can occur when compared with aggregating VaR at the asset class level. This difference is called portfolio diversification. The asset class VaR maxima and minima reported in the table occurred on different dates within the reporting period. For this reason, we do not report an implied portfolio diversification measure between the maximum (minimum) asset class VaR measures and the maximum (minimum) total VaR measures in this table.


Non-trading VaR, 99% 1 day
	Interest rate	Credit spread	Portfolio diversification1	Total
	$m	$m	$m	$m
Half-year to 30 Jun 2021	193.7	73.8	(18.0)	249.5
Average	201.1	80.5	(31.2)	250.4
Maximum	248.7	99.3		298.8
Minimum	163.3	64.7		193.5

Half-year to 30 Jun 2020	184.3	83.2	(61.6)	205.9
Average	122.8	80.9	(60.7)	143.0
Maximum	190.1	133.4		219.7
Minimum	59.0	44.2		79.7

Half-year to 31 Dec 2020	166.6	87.0	(5.7)	247.8
Average	176.7	84.0	(23.7)	237.0
Maximum	196.4	102.1		274.6
Minimum	159.2	67.5		179.7



Balance sheet of insurance manufacturing subsidiaries by type of contract
	With DPF	Unit- linked	Other contracts1	Shareholder assets and liabilities	Total
	$m	$m	$m	$m	$m
Financial assets	87,984	9,134	19,402	9,238	125,758
– trading assets	—	—	—	—	—
– financial assets designated and otherwise mandatorily measured at fair value through profit or loss	29,111	8,788	3,651	1,907	43,457
– derivatives	146	1	6	2	155
– financial investments – at amortised cost	41,201	93	14,145	4,391	59,830
– financial investments – at fair value through other comprehensive income	11,851	—	490	1,700	14,041
– other financial assets2	5,675	252	1,110	1,238	8,275
Reinsurance assets	2,191	77	1,625	1	3,894
PVIF3	—	—	—	9,449	9,449
Other assets and investment properties	2,551	3	658	717	3,929
Total assets at June 2021	92,726	9,214	21,685	19,405	143,030
Liabilities under investment contracts designated at fair value	—	2,390	3,941	—	6,331
Liabilities under insurance contracts	87,685	6,726	16,225	—	110,636
Deferred tax4	180	8	19	1,407	1,614
Other liabilities	—	—	—	8,068	8,068
Total liabilities	87,865	9,124	20,185	9,475	126,649
Total equity	—	—	—	16,381	16,381
Total liabilities and equity at June 2021	87,865	9,124	20,185	25,856	143,030

Financial assets	84,478	8,802	18,932	8,915	121,127
– trading assets	—	—	—	—	—
– financial assets designated at fair value	26,002	8,558	3,508	1,485	39,553
– derivatives	262	3	13	3	281
– financial investments at amortised cost	39,891	30	13,984	4,521	58,426
– financial investments at fair value through other comprehensive income	12,531	—	459	1,931	14,921
– other financial assets2	5,792	211	968	975	7,946
Reinsurance assets	2,256	65	1,447	2	3,770
PVIF3	—	—	—	9,435	9,435
Other assets and investment properties	2,628	1	227	721	3,577
Total assets at December 2020	89,362	8,868	20,606	19,073	137,909
Liabilities under investment contracts designated at fair value	—	2,285	4,100	—	6,385
Liabilities under insurance contracts	84,931	6,503	15,827	—	107,261
Deferred tax4	145	5	25	1,400	1,575
Other liabilities	—	—	—	7,244	7,244
Total liabilities	85,076	8,793	19,952	8,644	122,465
Total equity	—	—	—	15,444	15,444
Total liabilities and equity at December 2020	85,076	8,793	19,952	24,088	137,909

1 Other contracts includes term assurance, credit life insurance, universal life insurance and certain investment contracts not included in the ‘Unit-linked’ or ‘With DPF’ columns.

2 Comprise mainly loans and advances to banks, cash and inter-company balances with other non-insurance legal entities.

3 Present value of in-force long-term insurance business.

4 Deferred tax includes the deferred tax liabilities arising on recognition of PVIF.

Market risk

Description and exposure

Market risk is the risk of changes in market factors affecting HSBC’s capital or profit. Market factors include interest rates, equity and growth assets, and foreign exchange rates.

Our exposure varies depending on the type of contract issued. Our most significant life insurance products are contracts with discretionary participating features (‘DPF’) issued in France and Hong Kong. These products typically include some form of capital guarantee or guaranteed return on the sums invested by the policyholders, to which discretionary bonuses are added if allowed by the overall performance of the funds. These funds are primarily invested in bonds, with a proportion allocated to other asset classes to provide customers with the potential for enhanced returns.

DPF products expose HSBC to the risk of variation in asset returns, which will impact our participation in the investment performance.

In addition, in some scenarios the asset returns can become insufficient to cover the policyholders’ financial guarantees, in which case the shortfall has to be met by HSBC. Amounts are held against the cost of such guarantees, calculated by stochastic modelling.

Where local rules require, these reserves are held as part of liabilities under insurance contracts. Any remainder is accounted for as a deduction from the present value of in-force (‘PVIF’)
long-term insurance business on the relevant product.

For unit-linked contracts, market risk is substantially borne by the policyholder, but some market risk exposure typically remains, as fees earned are related to the market value of the linked assets.

Sensitivities

Changes in financial market factors, from the economic assumptions in place at the start of the year, had a positive impact on reported profit before tax of $413m (1H20: $320m negative). The following table illustrates the effects of selected interest rate, equity price and foreign exchange rate scenarios on our profit for the period and the total equity of our insurance manufacturing subsidiaries.

Where appropriate, the effects of the sensitivity tests on profit after tax and equity incorporate the impact of the stress on the PVIF.

Due in part to the impact of the cost of guarantees and hedging strategies, which may be in place, the relationship between the profit and total equity and the risk factors is non-linear, particularly in a low interest-rate environment.

Therefore, the results disclosed should not be extrapolated to measure sensitivities to different levels of stress. For the same reason, the impact of the stress is not necessarily symmetrical on the upside and downside. The sensitivities are stated before allowance for management actions, which may mitigate the effect of changes in the market environment. The sensitivities presented allow for adverse changes in policyholder behaviour that may arise in response to changes in market rates. The differences between the impacts on profit after tax and equity are driven by the changes in value of the bonds measured at fair value through other comprehensive income, which are only accounted for in equity.


Sensitivity of HSBC’s insurance manufacturing subsidiaries to market risk factors

	At 30 Jun 2021		At 31 Dec 2020
	Effect on profit after tax	Effect on total equity	Effect on profit after tax	Effect on total equity
	$m	$m	$m	$m
+100 basis point parallel shift in yield curves	(42)	(188)	(67)	(188)
-100 basis point parallel shift in yield curves	(53)	97	(68)	58
10% increase in equity prices	366	366	332	332
10% decrease in equity prices	(372)	(372)	(338)	(338)
10% increase in US dollar exchange rate compared with all currencies	10	10	84	84
10% decrease in US dollar exchange rate compared with all currencies	(10)	(10)	(84)	(84)