Risk management

6 Months Ended

Jun. 30, 2024

Basis of disclosures (*)

This risk management section contains an update of information relating to the nature and the extent of the risks arising from financial instruments as disclosed in the 2023 ING Group consolidated financial statements as included in the 2023 Annual Report on Form 20-F. These disclosures are an integral part of ING Group condensed consolidated interim financial statements and are indicated by the symbol (*). Chapters, paragraphs, graphs or tables within this risk management section that are indicated with this symbol in the respective headings or table header are considered to be an integral part of the condensed consolidated interim financial statements.

This risk management section also includes additional disclosures beyond those required by IFRS standards, such as certain legal and regulatory disclosures. Not all information in this section can be reconciled back to the primary financial statements and corresponding notes, as it has been prepared using risk data that differs to the accounting basis of measurement.

Credit risk

Loan loss provisioning (*)

ING recognises loss allowances based on the expected credit loss (ECL) model of IFRS 9, which is designed to be forward-looking. The IFRS 9 impairment requirements are applicable to on-balance sheet financial assets measured at amortised cost or fair value through other comprehensive income (FVOCI), such as loans, debt securities and lease receivables, as well as off-balance sheet items such as undrawn loan commitments, financial and non-financial guarantees issued.

ING distinguishes between two types of calculation methods for credit loss allowances:

•Collective 12-month ECL (Stage 1) and collective Lifetime ECL (Stage 2) for portfolios of financial instruments, as well as collective Lifetime ECL for credit impaired exposures (Stage 3) below €1 million.

•Individual Lifetime ECL for credit-impaired (Stage 3) financial instruments with exposures above €1 million.

Climate and environmental risks in IFRS 9 models (*)

Climate risk drivers (physical and transition risks) can reduce the ability of businesses and households to fulfil their obligations due on existing lending contracts. These may also lead to deprecation/ erosion of

collateral values which would translate into higher credit losses and loan-to-value ratios in the lending portfolio of ING.

At this point in time it is not yet possible to fully incorporate climate risk separately into IFRS 9 ECL models given the lack of sufficient empirical historical data and data limitations in the risk assessments on client level.

Where climate and environmental factors have impacted the economy in the recent past or present, these impacts will currently be implicitly embedded in ING's IFRS 9 ECL models through the projected macroeconomic indicators (e.g. indirectly via GDP growth and unemployment rates). We note however that our ECL models are primarily sensitive to the short-term economic outlook as we use a three-year time horizon for macroeconomic outlook, after which a mean reversion approach is applied.

With regard to our evaluation of climate-related matters, where such events have already occurred (e.g. floods, stranded assets etc.), the impact of such events is individually assessed in the calculation of Stage 3 Individual provisions or management adjustments to ECL models. For example, we consider whether affected assets have suffered from a significant increase in credit risk (or are credit impaired) and whether the ECL is appropriate.

Over the near-term horizon, ING plans to continue to refine its methodologies to evaluate climate risks. ING is working on putting into practice quantitative methodologies for climate and environmental (C&E) risk identification, materiality assessment and risk appetite setting. Going forward, ING aims to close the gaps on climate risk data, which will enable use to eventually more explicitly embed climate risks into the IFRS 9 ECL models.

Portfolio quality and concentration (*)

Our lending portfolio is diversified over various sectors and countries. The total gross carrying amount is composed of approximately 65% business lending and 35% consumer lending. For a detailed breakdown of ING’s credit risk portfolio by Sector and Geographical area, refer to the section “Credit Risk portfolio” reported in the ‘Risk management’ section of the 2023 Annual Report on Form 20-F.

The table below describes the portfolio composition over the different IFRS 9 stages and rating classes. The Stage 1 portfolio represents 92.2% (2023: 91.5%) of the total gross carrying amounts, mainly composed of investment grade, while Stage 2 makes up 6.5% (2023: 7.4%) and Stage 3 makes up 1.3% (2023: 1.2%) of the total gross carrying amounts, respectively.


Gross carrying amount per IFRS 9 stage and rating class (*) 2
in € million		12-month ECL (Stage 1)		Lifetime ECL not credit impaired (Stage 2)		Lifetime ECL credit impaired (Stage 3)		Total
30 June 2024
Rating class		Gross Carrying Amount	Provisions	Gross Carrying Amount	Provisions	Gross Carrying Amount	Provisions	Gross Carrying Amount	Provisions
Investment grade	1 (AAA)	97,215	1	61				97,276	1
	2-4 (AA)	137,427	6	1,512	1			138,939	7
	5-7 (A)	240,731	19	4,419	7			245,151	25
	8-10 (BBB)	312,399	63	18,634	24			331,033	88
Non-Investment grade	11-13 (BB)	158,219	218	15,684	78			173,903	296
	14-16 (B)	28,681	173	17,843	380			46,523	554
	17 (CCC)	851	15	4,231	201			5,082	216
Substandard grade	18 (CC)			4,617	318			4,617	318
	19 (C)			1,967	179			1,967	179
Non-performing loans	20-22 (D)					13,459	4,432	13,459	4,432
Total		975,522	494	68,969	1,190	13,459	4,432	1,057,950	6,117


Gross carrying amount per IFRS 9 stage and rating class (*)1,2
in € million		12-month ECL (Stage 1)		Lifetime ECL not credit impaired (Stage 2)		Lifetime ECL credit impaired (Stage 3)		Total
31 December 2023
Rating class		Gross Carrying Amount	Provisions	Gross Carrying Amount	Provisions	Gross Carrying Amount	Provisions	Gross Carrying Amount	Provisions
Investment grade	1 (AAA)	87,071	1	439				87,510	1
	2-4 (AA)	132,159	8	2,553	2			134,711	9
	5-7 (A)	231,018	24	6,188	6			237,206	30
	8-10 (BBB)	302,967	85	17,004	24			319,971	108
Non-Investment grade	11-13 (BB)	157,387	226	19,273	93			176,661	319
	14-16 (B)	26,414	164	19,336	455			45,750	618
	17 (CCC)	617	10	4,125	233			4,742	242
Substandard grade	18 (CC)			4,617	402			4,617	402
	19 (C)			1,919	221			1,919	221
Non-performing loans	20-22 (D)					11,956	3,887	11,956	3,887
Total		937,633	517	75,454	1,435	11,956	3,887	1,025,043	5,839

1 Includes impact from change in accounting policy as disclosed in table "Changes in gross carrying amounts and loan loss provisions"
2 Stage 3 lifetime credit impaired provision includes €7 million (2023: €11 million) on purchased or originated credit impaired

Changes in gross carrying amounts and loan loss provisions (*)

The table below provides a reconciliation by stage of the gross carrying amount and allowances for loans and advances to banks and customers, including loan commitments and guarantees issued (financial and non-financial). The transfers of financial instruments represent the impact of stage transfers on the gross carrying/nominal amount and associated allowance for ECL. This includes the net-remeasurement of ECL arising from stage transfers, for example, moving from a 12-month (Stage 1) to a lifetime (Stage 2) ECL measurement basis.

The net-remeasurement line represents the changes in provisions for facilities that remain in the same stage.

Please note the following comments with respect to the movements observed in the table below:

•Stage 3 gross carrying amount increased by €1.5 billion from €12.0 billion as at 31 December 2023 to €13.5 billion as at 30 June 2024, mainly as a result of €3.1 billion net inflow into NPL (credit impaired) in the first half of 2024 which is offset by €1.2 billion derecognitions and repayments and €0.4 billion write-offs. Partially as a result of the inflow into NPL, Stage 3 provisions increased by €0.5 billion.

•In the first 6 months of 2024 the stage 2 gross carrying amount decreased by €6.5 billion from €75.5 billion as at 31 December 2023 to €69.0 billion as at 30 June 2024, mainly as a result of repayments, outflow to Stage 3 and upgrades to Stage 1 largely driven by improved macroeconomic outlook. Stage 2 provisions decreased by €0.2 billion from €1.4 billion as at 31 December 2023 to €1.2 billion as at 30 June 2024, largely driven by improved macro-economic outlook, the release of management adjustments to Stage 2 and migration of files to Stage 3.

Information on macroeconomic scenarios is included in the section ‘Macroeconomic scenarios and sensitivity analysis of key sources of estimation uncertainty’.


Changes in gross carrying amounts and loan loss provisions (*)1
in € million	12-month ECL (Stage 1)		Lifetime ECL not credit impaired (Stage 2)		Lifetime ECL credit impaired (Stage 3)		Total
30 June 2024	Gross carrying amount	Provisions	Gross carrying amount	Provisions	Gross carrying amount	Provisions	Gross carrying amount	Provisions
Opening balance as at 1 January 2024	937,633	517	75,454	1,435	11,956	3,887	1,025,043	5,839
Transfer into 12-month ECL (Stage 1)	16,436	21	-16,237	-164	-199	-19		-162
Transfer into lifetime ECL not credit impaired (Stage 2)	-21,990	-32	22,489	291	-498	-65		194
Transfer into lifetime ECL credit impaired (Stage 3)	-1,457	-6	-2,360	-180	3,817	1,002		817
Net remeasurement of loan loss provisions		-72		-88		-3		-163
New financial assets originated or purchased	106,366	117					106,366	118
Financial assets that have been derecognised	-58,961	-45	-6,485	-93	-1,048	-95	-66,495	-232
Net drawdowns and repayments	-2,505		-3,893		-188		-6,585
Changes in models/risk parameters		-2		-15		-2		-19
Increase in loan loss provisions2		-19		-247		819		553
Write-offs					-379	-379	-379	-379
Disposals3		-1		-2				-3
Recoveries of amounts previously written off						23		23
Foreign exchange and other movements		-3		5		83		85
Closing balance	975,522	494	68,969	1,190	13,459	4,432	1,057,950	6,117

1Stage 3 Lifetime credit impaired provision includes €7 million on Purchased or Originated Credit Impaired.

2The addition to the loan loss provision in Profit or Loss amounts to €559 million of which €553 million relates to IFRS 9 eligible financial instruments and €6 million relates to the adjustments to CHF-indexed mortgages in Poland to reflect the changed expectation in future cash flows.

3Disposals reported relate to asset sales recorded during the first six months in 2024.


Changes in gross carrying amounts and loan loss provisions (*)1, 2
in € million	12-month ECL (Stage 1)		Lifetime ECL not credit impaired (Stage 2)		Lifetime ECL credit impaired (Stage 3)1		Total
31 December 2023	Gross carrying amount	Provisions	Gross carrying amount	Provisions	Gross carrying amount	Provisions	Gross carrying amount	Provisions
Opening balance as at 1 January	885,222	581	70,725	1,679	11,708	3,841	967,655	6,101
Impact of changes in accounting policies	37,078	9	4,704	13	158	73	41,939	95
Adjusted opening balance	922,300	590	75,429	1,692	11,866	3,914	1,009,595	6,196
Transfer into 12-month ECL (Stage 1)	11,832	28	-11,583	-239	-249	-36		-247
Transfer into lifetime ECL not credit impaired (Stage 2)	-29,470	-67	30,185	449	-716	-105		277
Transfer into lifetime ECL credit impaired (Stage 3)	-2,053	-10	-1,775	-115	3,828	978		853
Net remeasurement of loan loss provisions		-149		-94		59		-184
New financial assets originated or purchased	195,775	204					195,775	204
Financial assets that have been derecognised	-121,991	-72	-14,239	-215	-1,475	-266	-137,705	-553
Net drawdowns and repayments	-38,758		-2,525		-229		-41,511
Changes in models/risk parameters		8		10		84		102
Increase in loan loss provisions		-58		-204		714		452
Write-offs3	-3	-3			-787	-787	-790	-790
Disposals3			-38	-38	-283	-283	-321	-321
Recoveries of amounts previously written off						71		71
Foreign exchange and other movements		-12		-15		258		231
Closing balance	937,633	517	75,454	1,435	11,956	3,887	1,025,043	5,839

1Stage 3 Lifetime credit impaired provision includes €11 million on purchased or originated credit impaired.

2 The addition to the loan provision (in the consolidated statement of profit or loss) amounts to €520 million of which €483 million related to IFRS 9 eligible financial assets, €-31 million related to non-credit replacement guarantees and €68 million to modification gains and losses on restructured financial assets.

3Table was updated for presentation purposes to disaggregate utilisation of the provision between write-offs and disposals. Comparatives have been updated accordingly.

Macroeconomic scenarios and sensitivity analysis of key sources of estimation uncertainty (*)

Methodology (*)

Our methodology in relation to the adoption and generation of macroeconomic scenarios is described in this section. We continue to follow this methodology in generating our probability-weighted ECL, with consideration of alternative scenarios and management adjustments supplementing this ECL where, in management's opinion, the consensus forecast does not fully capture the extent of recent credit or economic events. The macroeconomic scenarios are applicable to the whole ING portfolio in the scope of IFRS 9 ECLs.

The IFRS 9 standard, with its inherent complexities and potential impact on the carrying amounts of our assets and liabilities, represents a key source of estimation uncertainty. In particular, ING’s reportable ECL numbers are sensitive to the forward-looking macroeconomic forecasts used as model inputs, the probability-weights applied to each of the three scenarios, and the criteria for identifying a significant increase in credit risk. As such, these crucial components require consultation and management judgement, and are subject to extensive governance.

Baseline scenario (*)

As a baseline for IFRS 9, ING has adopted a market-neutral view combining consensus forecasts for economic variables (GDP, unemployment) with market forwards (for interest rates, exchange rates and oil prices). Input from a leading third-party service provider is used to complement the consensus with consistent projections for variables for which there are no consensus estimates available (most notably house prices and – for some countries - unemployment), to generate alternative scenarios, to convert annual consensus information to a quarterly frequency and to ensure general consistency of the scenarios. As the baseline scenario is consistent with the consensus view it can be considered as free from any bias.

The relevance and selection of macroeconomic variables is defined by the ECL models under credit risk model governance. The scenarios are reviewed and challenged by two panels of ING experts. The first panel consists of economic experts from Global Markets Research and risk and modelling specialists, while the second panel consists of relevant senior managers.

Alternative scenarios and probability weights (*)

Two alternative scenarios are taken into account: an upside and a downside scenario. The alternative scenarios have statistical characteristics as they are based on the forecast deviations of the leading third-party service provider.

To understand the baseline level of uncertainty around any forecast, the leading third-party service provider keeps track of all its deviations (so called forecast errors) of the past 20 years. The distribution of forecast errors for GDP, unemployment, house prices and share prices is applied to the baseline forecast creating a broad range of alternative outcomes. In addition, to understand the balance of risks facing the economy in an unbiased way, the leading third-party service provider runs a survey with respondents from around the world and across a broad range of industries. In this survey respondents put forward their views of key risks. Following the survey results, the distribution of forecast errors (that is being used for determining the scenarios) may be skewed.

For the downside scenario, ING has chosen for the 90th percentile of that distribution because this corresponds with the way risk management earnings-at-risk is defined within the Group. The upside scenario is represented by the 10th percentile of the distribution. The applicable percentiles of the distribution imply a 20 percent probability for each alternative scenario. Consequently, the baseline scenario has a 60 percent probability weighting. Please note that, given their technical nature, the downside and upside scenarios are not based on an explicit specific narrative.

Macroeconomic scenarios applied (*)

The macroeconomic scenarios applied in the calculation of loan loss provisions are based on the consensus forecasts.

Baseline assumptions (*)

The general picture that the consensus conveys is that global economic growth is going through a weak spell but that the worst seems to be behind us. Inflation has come down rapidly and is currently still above but relatively close to central bank targets. Higher interest rates play their part in slowing demand. Central banks at this point are likely done with hiking interest rates and the effect of tight policy is expected to ease over time. GDP growth for the coming years is expected to remain relatively stable around 2.4%. The more upbeat economic outlook is also reflected in an expected recovery of house prices – which is already ongoing in quite a few countries – and declining unemployment.

The June 2024 consensus expects global output (as measured by the weighted average GDP growth rate of ING's 25 main markets) to slow from 2.6% in 2023, to 2.4% in 2024. For 2025-2026, economic growth is expected to come in at 2.3 and 2.4% respectively.

The ongoing resilience in the US has continued into 2024, supported by stronger than expected consumer spending and business investment. Still, economic activity is expected to gradually moderate given the recent tightening of financial market conditions and increase in oil prices, which could delay the Fed cutting rates. Overall, while no recession is expected anymore, the US is expected to see economic growth slow over

2024 and 2025. The consensus expects the growth rate of the US economy to fall from 2.5% in 2023 to 2.3% in 2024 and further to 1.7% in 2025, after that it is expected to recover to 2.1% in 2026.

The eurozone economy is just coming out of broad stagnation as consumers are regaining some purchasing power and financial conditions are becoming more favourable. This is boosted by a more benign inflation environment, which has allowed the ECB to cut interest rates for a first time after an aggressive series of hikes. For now, the recovery is very modest and is not expected to gain much strength either over the course of the year as the eurozone still faces significant headwinds. Think about energy insecurity and weak global demand for example. Consensus expects the eurozone to grow by only 0.6% in 2024, before recovering to 1.4% on average in 2025-2026.

Elsewhere in Europe, the outlook becoming more upbeat. Gradual recovery in the Polish economy is underway, led by household consumption. The upturn is so far gradual though, given external headwinds (particularly from Germany). The economy is expected to grow by 3% this year, picking up to 3.5% in 2025 and 2026. The consensus expectation for Türkiye is to see growth slow from 3.3% in 2024 to 3% in 2025 and picking back up again to 3.7% in 2026. Overall, consensus has become a lot more upbeat about the Turkish economy in recent months. The Russian economy transformed more successfully than expected into a war economy, which has boosted consensus economic expectations. Growth is expected to come in at 3% in 2024, before slowing to 1.6% and 1.4% in 2025 and 2026.

For China, economic underperformance continues as it still struggles with the impact of the real estate correction and weak domestic demand. Guidance from central authorities suggest that fiscal stimulus will likely be more accommodative than previously anticipated. There are growing signs that authorities are turning to a big manufacturing push through high-value technology investments and “new productive forces”. While this may help in the short-term, medium-term consensus projections are more downbeat. For 2024, consensus expects 5% growth, down to 4.4% in 2025 and 4.1% in 2026.

Economic momentum in Australia is expected to be soft, which reflects recent data, which indicates that consumer confidence is at a low point and household consumption growth remains weak. In addition, tight monetary policy and high inflation are squeezing real disposable incomes, further weighing on activity. After growing by 2.1% in 2023, the outlook sees a growth rate of 1.3% for 2024 and some pick-up to 2.2 and 2.5% for 2025-2026.

When compared to the December 2023 consensus forecast, the June 2024 forecast assumes significantly better economic circumstances in 2024. Global GDP is expected to increase by 2.4% in 2024 (compared to 1.8% assumed before) and is expected to grow by 2.3% in 2025 (2.3% assumed before). The upgrade for 2024 mainly reflects the better than expected economic performance of the US, which was once assumed to be in recession around the middle of the year. Other advanced markets have seen smaller upgrades to their outlooks.

Alternative scenarios and risks (*)
The baseline scenario assumes continued steady economic growth on the back of easing inflation and financial conditions becoming more favourable. However, a longer period of weakness, due to even more concerning geopolitical tensions, persistent elevated inflation and the outcome of US elections – for example related to trade escalations - could lead to a more protracted and deeper economic slowdown. As such, the balance of risks to the baseline outlook is negative and the alternative scenarios have a downward skew in line with the outcomes of Oxford Economics’ Global Risk Survey.

The downside scenario – though technical in nature – sees for most countries a recession in 2024. Unemployment increases strongly in this scenario and house prices in most countries show outright falls. The downside scenario captures the possible impact from geopolitical developments, persistent inflation and the outcome of US elections – for example related to trade escalations.

The upside scenario – while equally technical in nature – reflects the possibility of a better economic outturn because of a substantial loosening of monetary policy, buoyant consumer spending, and policy stimulus in China.

Management adjustments applied this reporting period (*)

In times of volatility and uncertainty where portfolio quality and the economic environment are changing rapidly, models alone may not be able to accurately predict losses. In these cases, management adjustments can be applied to appropriately reflect ECL. Management adjustments can also be applied where the impact of the updated macroeconomic scenarios is over- or under-estimated by the IFRS 9 models, as well as to reflect the impact of model redevelopment or recalibration and periodic model assessment procedures that have not been incorporated in the IFRS 9 models yet.

ING has internal governance frameworks and controls in place to assess the appropriateness of all management adjustments.


Management adjustments to ECL models (*)
in € million	30 June 2024	31 December 2023
Economic sector based adjustments	16	36
Inflation and Interest rate increases adjustments	233	351
Mortgage portfolio adjustments	128	126
Other post model adjustments	38	64
Total management adjustments	415	577

The economic sector based adjustments of €16 million as at 30 June 2024 (€36 million as at 31 December 2023) fully relates to Business Banking clients in the Netherlands that have benefited from government

support programmes in the Netherlands during the Covid-19 Pandemic and have relatively high taxes to be repaid. The management adjustment has been partially released in prior year and further released in the first six months of 2024.

The inflation and interest rate increases adjustments, amount to €233 million in total as at 30 June 2024 (31 December 2023: €351 million), of which €104 million (31 December 2023: €138 million) relates to Retail Banking segments and €129 million (31 December 2023: €213 million) to the Wholesale Banking segment. These management adjustments are reflected in Stage 1 and Stage 2.

As the credit risk models generally assume that inflation and interest rate increases risks materialise via other risk drivers such as GDP and unemployment rates with a delay, an overlay approach was determined to timely estimate the Expected Credit Losses related to reduced repayment capacity and affordability for private individuals and business clients in the Retail Banking segment.

In Wholesale Banking the IFRS 9 credit risk models mostly leverage on GDP growth as a generic macroeconomic variable. High inflation and rapidly increasing interest rates however trigger economic heterogeneity (i.e. some businesses benefit, while others suffer), as such the current circumstances are expected to cause more defaults than normally predicted using GDP growth. The overlay is focussed on the prevailing risks from interest and inflation and is based on a sector based calculation methodology. The methodology uses stage transition rates that are dependent on the risk classification of sectors vulnerable to inflation and interest rate increases.

The overall mortgage portfolio adjustment as at 30 June 2024 increased to €128 million (31 December 2023: €126 million). The management adjustment for the risk segmentation model that captures affordability, repayment and refinancing risk on performing mortgage customers with a bullet loan in the Netherlands was increased to €118 million (31 December 2023: €115 million). The mortgage portfolio adjustment that relates to the overvaluation of house prices was reduced to €10 million (31 December 2023: €11 million) and is recognised in Stage 2 and Stage 3 on the mortgage portfolio in Germany. The management adjustment is determined by calculating the impact of lower house prices on loan-to-value (LTV) and loss given default (LGD).

Other post model adjustments mainly relate to the impact of model redevelopment or recalibration and periodic model assessment procedures that have not been incorporated in the ECL models yet, the impact on total ECL can be positive or negative. These result from both regular model maintenance and ING’s multiyear programme to update ECL models for the definition of default. These adjustments will be removed once updates to the specific models have been implemented. The change in balance compared to previous reporting date is due to i) released PMAs because of model updates that have been implemented and ii) new PMAs recognized for new redevelopments and recalibrations.

Analysis on sensitivity (*)

The table below presents the analysis on the sensitivity of key forward-looking macroeconomic inputs used in the ECL collective-assessment modelling process and the probability-weights applied to each of the three scenarios. The countries included in the analysis are the most significant geographic regions, in terms of both gross contribution to reportable ECL, and sensitivity of ECL to forward-looking macroeconomics. Accordingly, ING considers these portfolios the most significant in terms of risk resulting in a material adjustment to the carrying amount of financial assets within the next financial year. ING also observes that, in general, the WB business is more sensitive to the impact of forward-looking macroeconomic scenarios.

The purpose of the sensitivity analysis is to enable the reader to understand the extent of the impact from the upside and downside scenario on model-based reportable ECL.

In the table below the real GDP is presented in percentage year-on-year change, the unemployment in percentage of total labour force and the house price index (HPI) in percentage year-on-year change.


Sensitivity analysis as at 30 June 2024 (*)
		2024	2025	2026	Un-weighted ECL (€ mln)	Probability-weighting		Reportable ECL (€ mln)1
Netherlands Upside scenario	Real GDP	1.0	3.2	3.1	171	20	%	251
	Unemployment	3.8	3.4	3.3
	HPI	9.0	18.8	5.8
Baseline scenario	Real GDP	0.6	1.5	1.6	231	60	%
	Unemployment	4.0	4.1	4.3
	HPI	6.2	5.3	3.7
Downside scenario	Real GDP	0.0	-1.2	-0.6	393	20	%
	Unemployment	5.0	6.6	7.7
	HPI	2.6	-11.1	0.8
Germany Upside scenario	Real GDP	0.7	3.2	2.4	494	20	%	543
	Unemployment	2.8	2.4	1.9
	HPI	0.7	6.4	7.9
Baseline scenario	Real GDP	0.2	1.1	1.3	533	60	%
	Unemployment	3.1	3.0	2.9
	HPI	-0.4	3.0	4.5
Downside scenario	Real GDP	-0.5	-2.0	-0.6	626	20	%
	Unemployment	3.8	5.0	5.5
	HPI	-1.8	-1.9	0.7
Belgium Upside scenario	Real GDP	1.5	2.8	2.4	549	20	%	599
	Unemployment	5.0	4.8	4.7
	HPI	4.0	4.7	4.5
Baseline scenario	Real GDP	1.2	1.5	1.7	588	60	%
	Unemployment	5.6	5.5	5.4
	HPI	3.5	3.9	4.0
Downside scenario	Real GDP	0.6	-0.8	0.6	683	20	%
	Unemployment	6.4	7.4	7.7
	HPI	2.4	1.9	2.6
United States Upside scenario	Real GDP	2.7	3.2	3.5	50	20	%	90
	Unemployment	3.8	2.9	2.3
	HPI	5.2	6.0	8.2
Baseline scenario	Real GDP	2.4	1.7	2.1	78	60	%
	Unemployment	4.1	4.1	4.1
	HPI	5.0	2.8	2.6
Downside scenario	Real GDP	1.7	-1.4	-0.6	169	20	%
	Unemployment	4.9	6.8	7.8
	HPI	3.8	-2.8	-3.9
1Excluding management adjustments.


Sensitivity analysis as at 31 December 2023 (*)
		2024	2025	2026	Un-weighted ECL (€ mln)	Probability-weighting		Reportable ECL (€ mln)1
Netherlands Upside scenario	Real GDP	1.3	3.3	2.8	214	20	%	310
	Unemployment	3.7	3.3	3.3
	HPI	10.4	11.2	4.0
Baseline scenario	Real GDP	0.8	1.6	1.5	282	60	%
	Unemployment	4.1	4.3	4.5
	HPI	0.9	3.0	3.9
Downside scenario	Real GDP	-1.7	-1.2	0.1	487	20	%
	Unemployment	5.9	7.2	8.1
	HPI	-10.9	-7.4	3.7
Germany Upside scenario	Real GDP	1.4	3.1	1.6	472	20	%	525
	Unemployment	2.6	2.0	1.7
	HPI	0.9	6.6	8.0
Baseline scenario	Real GDP	0.5	1.3	1.2	513	60	%
	Unemployment	3.0	3.0	3.0
	HPI	-1.4	3.4	4.5
Downside scenario	Real GDP	-2.4	-1.4	0.3	615	20	%
	Unemployment	4.5	5.2	5.5
	HPI	-6.0	-0.8	0.4
Belgium Upside scenario	Real GDP	1.5	2.7	2.3	568	20	%	619
	Unemployment	5.3	5.0	4.9
	HPI	1.3	5.6	4.5
Baseline scenario	Real GDP	0.9	1.5	1.8	604	60	%
	Unemployment	5.6	5.5	5.4
	HPI	0.4	5.2	3.9
Downside scenario	Real GDP	-1.3	-0.2	1.2	713	20	%
	Unemployment	7.3	8.0	7.9
	HPI	-2.2	3.9	2.6
United States Upside scenario	Real GDP	1.8	3.2	3.4	102	20	%	165
	Unemployment	4.1	3.3	3.1
	HPI	0.6	8.7	8.7
Baseline scenario	Real GDP	0.9	1.9	2.1	144	60	%
	Unemployment	4.5	4.5	4.4
	HPI	-0.7	3.5	3.3
Downside scenario	Real GDP	-1.3	-1.4	-0.1	292	20	%
	Unemployment	6.6	8.2	8.8
	HPI	-4.2	-2.7	-3.0
1 Excluding management adjustments.

When compared to the sensitivity analysis of 2023, the macroeconomic inputs are overall more favourable, driven by an improved macroeconomic outlook mainly because economies prove to be rather resilient to increased interest rates, particularly in US, as well as recovery in house prices in amongst others the Netherlands.

On a total ING level, the unweighted ECL for all collective provisioned clients in the upside scenario was €2,623 million, in the baseline scenario €2,895 million and in the downside scenario €3,579 million compared to €2,977 million reportable collective provisions as at 30 June 2024 (excluding all management adjustments). To perform the sensitivity analysis, a point in time reportable ECL is used as input which slightly deviates from the total Model ECL as reported below:


Reconciliation of model (reportable) ECL to total ECL (*)
in € million	30 June 2024	31 December 2023
Total model ECL	2,919	2,856
ECL from individually assessed impairments	2,783	2,406
ECL from management adjustments	415	577
Total ECL	6,117	5,839

Criteria for identifying a significant increase in credit risk (SICR) (*)

All assets and off-balance sheet items that are in scope of IFRS 9 impairment and which are subject to collective ECL assessment are allocated a 12-month ECL if deemed to belong in Stage 1, or a lifetime ECL if deemed to belong in Stages 2 or 3. An asset belongs in Stage 2 if it is considered to have experienced a significant increase in credit risk since initial origination or purchase. ING considers the credit risk of an asset to have significantly increased when either a threshold for absolute change in lifetime probability of default (PD) or a relative change in lifetime PD is reached.

It should be noted that the lifetime PD thresholds are not the only drivers of stage allocation. An asset can also change stages as a result of other triggers, such as having over 30 days arrears, being on a watch list or being forborne. Furthermore, this analysis is rudimentary in the sense that other parameters would change when an asset changes stages.

Absolute lifetime PD threshold

The absolute threshold is a fixed value calibrated per portfolio/segment and provides a fixed threshold that, if exceeded by the difference between lifetime PD at reporting date and lifetime PD at origination, triggers Stage 2 classification. The thresholds for the absolute change in lifetime PD vary between 75bps for Retail portfolios, 100bps for WB and 250bps for SMEs, based on the characteristics of the specific portfolio. ING is in

the process of refining the thresholds on a portfolio level. These have already been implemented for more than 25 percent of the portfolio, resulting in deviating absolute lifetime PD thresholds.

Relative lifetime PD threshold

The relative threshold defines a relative increase of the lifetime PD beyond which a given facility is classified in Stage 2 because of significant increase in credit risk. The relative threshold is dependent on the individual PD assigned to each facility at the moment of origination and a scaling factor calibrated in the model development phase that is optimised depending on the observed default rates and overall average riskiness of the portfolio. While the scaling factor is associated with a whole portfolio/segment, the PD at origination is facility-specific and, in this sense, the relative threshold may differ facility by facility.

Ultimately the relative threshold provides a criterion to assess whether the ratio (i.e. increase) between lifetime PD at reporting date and lifetime PD at origination date is deemed a significant increase in credit risk. If the threshold is breached, SICR is identified and Stage 2 is assigned to the given facility.

The threshold for the relative change in lifetime PD is inversely correlated with the PD at origination; the higher the PD at origination, the lower the threshold. The logic behind this is to allow facilities originated in very favourable ratings to downgrade for longer without the need of a Stage 2 classification. In fact, it is likely that said facilities will still be in favourable ratings even after a downgrade of a few notches. On the contrary, facilities originated in already unfavourable ratings grades are riskier and even a single-notch downgrade might represent a significant increase in credit risk and thus a tighter threshold will be in place. Still, the relative threshold is relatively sensitive for investment grade assets while the absolute threshold primarily affects non-investment grade assets.

Average threshold ratio

In the table below the average increase in PD at origination needed to be classified in Stage 2 is reported, taking into account the PD at origination of the facilities included in each combination of asset class and rating quality. In terms of rating quality, assets are divided into 'Investment grade' and 'non-investment grade' facilities. Rating 18 and 19 are not included in the table since facilities are not originated in these ratings and they constitute a staging trigger of their own (i.e. if a facility is ever to reach rating 18 or 19 at reporting date, it is classified in Stage 2). In the table, values are weighted by IFRS 9 exposure and shown for both year-end 2023 and mid-year 2024.

To represent the thresholds as a ratio (i.e. how much should the PD at origination increase in relative terms to trigger Stage 2 classification) the absolute threshold is recalculated as a relative threshold for disclosure purposes. Since breaching only relative or absolute threshold triggers Stage 2 classification, the minimum between the relative and recalculated absolute threshold is taken as value of reference for each facility.


Quantitative SICR thresholds (*)
	30 June 2024		31 December 2023
Average threshold ratio	Investment grade (rating grade 1-10)	Non-investment grade (rating grade 11-17)	Investment grade (rating grade 1-10)	Non-investment grade (rating grade 11-17)
Asset class category
Mortgages	2.6	2.3	2.5	2.3
Consumer Lending	2.9	2.2	2.9	2.1
Business Lending	2.7	2.1	2.7	2.1
Governments and fin. institutions	3.0	1.8	3.0	1.9
Other Wholesale Banking	2.7	1.9	2.8	1.8

As is apparent from the disclosures above, per ING’s methodology the threshold is tighter the higher the riskiness at origination of the assets, illustrated by the difference between the average threshold applied to investment grade facilities and non-investment grade facilities. In 3Q 2022, following up an ECB request, a new backstop trigger was implemented. This trigger forces Stage 2 classification in case the lifetime PD at reporting date has increased more than three times with respect to the origination, regardless of the actual staging thresholds in force for a given portfolio. The requirement entails that the threshold ratio in the table above is effectively capped at a threefold increase for both reporting dates. The effect of this trigger is especially apparent in the average thresholds ratio for investment grade facilities. When comparing the two snapshots it is clear the staging thresholds are stable over the last 6-months and limited changes are attributable to portfolios fluctuations in risk distribution.

Sensitivity of ECL to lifetime PD thresholds

The setting of PD threshold bands requires management judgement and is a key source of estimation uncertainty. On Group level, the total model ECL on performing assets, which is the ECL collective-assessment without taking management adjustments into account, is €1,343 million (31 December 2023: €1,412 million). To demonstrate the sensitivity of the ECL to these PD thresholds bands, hypothetically solely applying the upside scenario would result in total model ECL on performing assets of €1,034 million and a decrease in the Stage 2 ratio by 0.5%-point, while solely applying the downside scenario would result in total model ECL on performing assets of €2,078 million and an increase in the Stage 2 ratio by 2.7%-point.