Note 5

Loans⁽¹⁾⁽²⁾


	In accordance with IFRS 9										In accordance with IAS 39
$ millions, as at October 31									2018										2017
	Gross amount		Stage 3 allowance		Stages 1 and 2 allowance		Total allowance		Net total		Gross amount		Individual allowance		Collective allowance		Total allowance		Net total
Residential mortgages ⁽³⁾	$	207,749	$	143	$	71	$	214	$	207,535	$	207,271	$	2	$	201	$	203	$	207,068
Personal ⁽⁴⁾		43,058		109		372		481		42,577		40,937		7		488		495		40,442
Credit card		12,673		–		418		418		12,255		12,378		–		386		386		11,992
Business and government ⁽³⁾		109,555		230		296		526		109,029		97,766		183		351		534		97,232
	$	373,035	$	482	$	1,157	$	1,639	$	371,396	$	358,352	$	192	$	1,426	$	1,618	$	356,734

(1)	Loans are net of unearned income of $421 million (2017: $376 million).

(2)	Includes gross loans of $61.0 billion (2017: $53.1 billion) denominated in U.S. dollars and $4.8 billion (2017: $4.8 billion) denominated in other foreign currencies.

(3)	Includes $12 million of residential mortgages (2017: $12 million) and $16,424 million of business and government loans (2017: $14,010 million) that are measured at FVTPL (2017: Trading loans).

(4)

Includes $42 million (2017: $47 million) related to loans provided to certain individuals while employed by CIBC to finance a portion of their participation in funds which make private equity investments on a side-by-side basis with CIBC and its affiliates. These loans are secured by the borrowers’ interest in the funds. Of the total amount outstanding, $41 million (2017: $47 million) relates to individuals who are no longer employed by CIBC.

Allowance for credit losses⁽¹⁾

The following table provides a reconciliation of the opening balance to the closing balance of the ECL allowance under IFRS 9:


$ millions, as at or for the year ended October 31	2018
	Stage 1			Stage 2			Stage 3			In accordance with IFRS 9
	Collective provision 12-month ECL performing			Collective provision lifetime ECL performing			Collective and individual provision lifetime ECL credit-impaired ⁽²⁾			Total
Residential mortgages
Balance at beginning of year	$	28		$	43		$	151		$	222
Originations net of repayments and other derecognitions		7			(6	)		(13	)		(12	)
Changes in model		(2	)		1			22			21
Net remeasurement ⁽³⁾		(25	)		13			60			48
Transfers ⁽³⁾
– to 12-month ECL		20			(16	)		(4	)		–
– to lifetime ECL performing		(1	)		9			(8	)		–
– to lifetime ECL credit-impaired		–			(2	)		2			–
Provision for (reversal of) credit losses ⁽⁴⁾		(1	)		(1	)		59			57
Write-offs ⁽⁵⁾		–			–			(54	)		(54	)
Recoveries		–			–			–			–
Interest income on impaired loans		–			–			(10	)		(10	)
Foreign exchange and other		–			2			(3	)		(1	)
Balance at end of year	$	27		$	44		$	143		$	214
Personal
Balance at beginning of year	$	164		$	202		$	110		$	476
Originations net of repayments and other derecognitions		34			(22	)		(5	)		7
Changes in model		(2	)		–			–			(2	)
Net remeasurement ⁽³⁾		(116	)		148			299			331
Transfers ⁽³⁾
– to 12-month ECL		151			(148	)		(3	)		–
– to lifetime ECL performing		(40	)		49			(9	)		–
– to lifetime ECL credit-impaired		–			(31	)		31			–
Provision for (reversal of) credit losses ⁽⁴⁾		27			(4	)		313			336
Write-offs ⁽⁵⁾		–			–			(368	)		(368	)
Recoveries		–			–			58			58
Interest income on impaired loans		–			–			(3	)		(3	)
Foreign exchange and other		(1	)		1			(1	)		(1	)
Balance at end of year	$	190		$	199		$	109		$	498
Credit card
Balance at beginning of year	$	101		$	413		$	–		$	514
Originations net of repayments and other derecognitions		–			(24	)		–			(24	)
Changes in model		–			2			–			2
Net remeasurement ⁽³⁾		(143	)		370			145			372
Transfers ⁽³⁾
– to 12-month ECL		179			(179	)		–			–
– to lifetime ECL performing		(35	)		35			–			–
– to lifetime ECL credit-impaired		–			(247	)		247			–
Provision for (reversal of) credit losses ⁽⁴⁾		1			(43	)		392			350
Write-offs ⁽⁵⁾		–			–			(512	)		(512	)
Recoveries		–			–			120			120
Interest income on impaired loans		–			–			–			–
Foreign exchange and other		–			–			–			–
Balance at end of year	$	102		$	370		$	–		$	472
Business and government
Balance at beginning of year	$	234		$	150		$	204		$	588
Originations net of repayments and other derecognitions		19			(10	)		(15	)		(6	)
Changes in model		(11	)		(7	)		1			(17	)
Net remeasurement ⁽³⁾		(109	)		72			187			150
Transfers ⁽³⁾
– to 12-month ECL		66			(60	)		(6	)		–
– to lifetime ECL performing		(21	)		25			(4	)		–
– to lifetime ECL credit-impaired		(1	)		(24	)		25			–
Provision for (reversal of) credit losses ⁽⁴⁾		(57	)		(4	)		188			127
Write-offs ⁽⁵⁾		–			–			(116	)		(116	)
Recoveries		–			–			12			12
Interest income on impaired loans		–			–			(10	)		(10	)
Foreign exchange and other		3			1			(48	) ⁽⁶⁾		(44	)
Balance at end of year	$	180		$	147		$	230		$	557
Total ECL allowance ⁽¹⁾	$	499		$	760		$	482		$	1,741
Comprises:
Loans	$	450		$	707		$	482		$	1,639
Undrawn credit facilities and other off-balance sheet exposures ⁽⁷⁾		49			53			–			102

(1)

See Note 4 for the ECL allowance on debt securities measured at FVOCI. The ECL allowances for other financial assets classified at amortized cost were immaterial as at October 31, 2018 and were excluded from the table above. Other financial assets classified at amortized cost are presented on our consolidated balance sheet net of ECL allowances.

(2)	Includes the ECL allowance for purchased credit-impaired loans from the acquisition of The PrivateBank.

(3)

Transfers represent stage movements of prior year ECL allowances to the current year stage classification. Net remeasurement represents the current year change of ECL allowances for transfers, net write-offs, changes in forecasts of forward-looking information, parameter updates, and partial repayments in the year.

(4)	Provision for (reversal of) credit losses for loans and undrawn credit facilities and other off-balance sheet exposures is presented as provision for (reversal of) credit losses on our consolidated statement of income.

(5)	We generally continue to pursue collection on the amounts that were written off. The degree of collection efforts varies from one jurisdiction to another, depending on the local regulations and original agreements with customers.

(6)	Includes ECL of $48 million relating to Barbados loans that were derecognized in the fourth quarter of 2018 as a result of a debt restructuring agreement completed with the Government of Barbados.

(7)	Included in other liabilities on our consolidated balance sheet.

Allowance for credit losses

The following table provides a reconciliation of the opening balance to the closing balance of allowance for credit losses under IAS 39:


$ millions, as at or for the year ended October 31	2017
							In accordance with IAS 39
	Individual allowance			Collective allowance			Total
Residential mortgages
Balance at beginning of year	$	1		$	220		$	221
Provision for (reversal of) credit losses		–			39			39
Write-offs		–			(38	)		(38	)
Recoveries		–			–			–
Interest income on impaired loans		–			(8	)		(8	)
Foreign exchange and other		1			(12	)		(11	)
Balance at end of year	$	2		$	201		$	203
Personal
Balance at beginning of year	$	8		$	489		$	497
Provision for (reversal of) credit losses		–			308			308
Write-offs		–			(359	)		(359	)
Recoveries		–			54			54
Interest income on impaired loans		–			–			–
Foreign exchange and other		(1	)		(4	)		(5	)
Balance at end of year	$	7		$	488		$	495
Credit card
Balance at beginning of year	$	–		$	386		$	386
Provision for (reversal of) credit losses		–			410			410
Write-offs		–			(529	)		(529	)
Recoveries		–			119			119
Interest income on impaired loans		–			–			–
Foreign exchange and other		–			–			–
Balance at end of year	$	–		$	386		$	386
Business and government
Balance at beginning of year	$	249		$	460		$	709
Provision for (reversal of) credit losses		61			11			72
Write-offs		(107	)		(24	)		(131	)
Recoveries		15			5			20
Interest income on impaired loans		(18	)		–			(18	)
Foreign exchange and other		(17	)		18			1
Balance at end of year	$	183		$	470		$	653
Total allowance for credit losses	$	192		$	1,545		$	1,737
Comprises:
Loans	$	192		$	1,426		$	1,618
Undrawn credit facilities and other off–balance sheet exposures ⁽¹⁾		–			119			119

(1)	Included in other liabilities on our consolidated balance sheet.

Inputs, assumptions and model techniques

Our ECL allowances are estimated using complex models that incorporate inputs, assumptions and model techniques that involve a high degree of management judgment. In particular, the following ECL elements are subject to a high level of judgment that can have a significant impact on the level of ECL allowances provided:

•

Determining when a significant increase in credit risk of a loan has occurred;

•

Measuring both 12-month and lifetime credit losses; and

•

Forecasting forward-looking information for multiple scenarios and determining the probability weighting of the scenarios.

In addition, the interrelationship between these elements is also subject to a high degree of judgment which can also have a significant impact on the level of ECL recognized.

Determining when a significant increase in credit risk has occurred

The determination of whether a loan has experienced a significant increase in credit risk has a significant impact on the level of ECL allowance as loans that are in stage 1 are measured at 12-month ECL, while loans in stage 2 are measured at lifetime ECL. Migration of loans between stage 1 and stage 2 can cause significant volatility in the amount of the recognized ECL allowances and the provision for credit losses in a particular period.

For the majority of our retail loan portfolios, we determine a significant increase in credit risk based on relative changes in the loan’s lifetime PD since its initial recognition. The PDs used for this purpose are the expected value of our upside, downside and base case lifetime PDs. Significant judgment is involved in determining the upside, downside and base case lifetime PDs through the incorporation of forward-looking information into long run PDs, in determining the probability weightings of the scenarios, and in determining the relative change in PDs that are indicative of a significant increase in credit risk for our various retail products. Increases in the expected PDs or decreases in the thresholds for changes in PDs that are indicative of a significant increase in credit risk can cause significant migration of loans from stage 1 to stage 2, which in turn can cause a significant increase in the amount of ECL allowances recognized. In contrast, decreases in the expected PDs or increases in the thresholds for changes in PDs that are indicative of a significant increase in credit risk can cause significant migration of loans from stage 2 to stage 1.

For the majority of our business and government loan portfolios, we determine a significant increase in credit risk based on relative changes in internal risk ratings since initial recognition. Significant judgment is involved in the determination of the internal risk ratings. Deterioration or improvement in the risk ratings or adjustments to the risk rating downgrade thresholds used to determine a significant increase in credit risk can cause significant migration of loans and securities between stage 1 and stage 2, which in turn can have a significant impact on the amount of ECL allowances recognized.

While potentially significant to the level of ECL allowances recognized, the thresholds for changes in PDs that are indicative of a significant increase in credit risk for our retail portfolios and the risk rating downgrade thresholds used to determine a significant increase in credit risk for our business and government loan portfolios are not expected to change frequently.

All loans on which repayment of principal or payment of interest is contractually 30 days in arrears and all business and government loans that have migrated to the watch list are automatically migrated to stage 2 from stage 1.

As at October 31, 2018, if the ECL for the stage 2 performing loans were measured using stage 1 ECL as opposed to lifetime ECL, the expected credit losses would be $273 million lower than the total recognized IFRS 9 ECL on performing loans.

Measuring both 12-month and lifetime expected credit losses

Our ECL models leverage the PD, LGD, and EAD parameters, as well as the portfolio segmentation used to calculate Basel expected loss regulatory adjustments for the portion of our retail and business and government portfolios under the advanced internal ratings-based (AIRB) approach. Adjustments are made to the Basel parameters to meet IFRS 9 requirements, including the conversion of through-the-cycle and downturn parameters used in the Basel regulatory calculations to point-in-time parameters used under IFRS 9 that considers forward-looking information. For standardized business and government portfolios, available long-run PDs, LGDs and EADs are also converted to point-in-time parameters through the incorporation of forward-looking information for the purpose of measuring ECL under IFRS 9.

Significant judgment is involved in determining which forward-looking information variables are relevant for particular portfolios and in determining the extent by which through-the-cycle parameters should be adjusted for forward-looking information to determine point-in-time parameters. While changes in the set of forward-looking information variables used to convert through-the cycle PDs, LGDs and EADs into point-in-time parameters can either increase or decrease ECL allowances in a particular period, changes to the mapping of forward-looking information variables to particular portfolios are expected to be infrequent. However, changes in the particular forward-looking information parameters used to quantify point-in-time parameters will be frequent as our forecasts are updated on a quarterly basis. Increases in the level of pessimism in the forward-looking information variables will cause increases in expected credit losses, while increases in the level of optimism in the forward-looking information variables will cause decreases in expected credit losses. These increases and decreases could be significant in any particular period and will start to occur in the period where our outlook of the future changes.

With respect to the lifetime of a financial instrument, the maximum period considered when measuring ECL is the maximum contractual period over which we are exposed to credit risk. For revolving facilities, such as credit cards, the lifetime of a credit card account is the expected behavioural life. Significant judgment is involved in the estimate of the expected behavioural life. Increases in the expected behavioural life will increase the amount of ECL allowances, in particular for revolving loans in stage 2.

Forecasting forward-looking information for multiple scenarios and determining the probability weighting of the scenarios

As indicated above, forward-looking information is incorporated into both our assessment of whether the financial asset has experienced a significant increase in credit risk since its initial recognition and in our estimate of ECL. From analysis of historical data, our risk management function has identified and reflected in our ECL allowance those relevant forward-looking information variables that contribute to credit risk and losses within our retail and business and government loan portfolios. Within our retail loan portfolio, key forward-looking information variables include unemployment rates, housing prices and GDP growth. In many cases these variables are forecast at the provincial level. Housing prices are also forecasted at the municipal level in some cases. Within our business and government loan portfolio, key drivers that impact the credit performance of the entire portfolio include S&P 500 growth rates, business credit growth rates, unemployment rates and credit spreads, while forward-looking information variables such as commodity prices are significant for certain portfolios.

Our forecasting process leverages the process used prior to the adoption of IFRS 9. For the majority of our loan portfolios, our forecast of forward-looking information variables is established from a “base case” or most likely scenario that is used internally by management for planning and forecasting purposes. For most of the forward-looking information variables related to our Canadian businesses, we have forecast scenarios by province. In forming the “base case” scenario, we consider the forecasts of monetary authorities such as the Organisation for Economic Co-operation and Development (OECD), the International Monetary Fund (IMF), and the Bank of Canada, as well as private sector economists. We then derive reasonably possible “upside case” and “downside case” scenarios using external forecasts that are above and below our “base case” and the application of management judgment. A probability weighting is assigned to our “base case”, “upside case” and “downside case” scenarios based on management judgment.

The following table provides the base case, upside case and downside case scenario forecasts for select forward-looking information variables used to estimate our October 31, 2018 ECL. The base case amounts shown represent the average value of the forecasts over the respective projection horizons. The upside case and downside case amounts shown represent the average value of the forecasts over the entire projection horizon.


	Base case						Upside case			Downside case
$ millions, as at October 31, 2018	Average value over the next 12 months			Average value over the remaining forecast period			Average value over the forecast period			Average value over the forecast period
Canadian GDP year-over-year growth ⁽¹⁾		1.9	%		1.4	%		2.3	%		1.2	%
Canadian unemployment rate ⁽¹⁾		5.8	%		6.0	%		5.3	%		6.4	%
Canadian Housing Price Index growth ⁽¹⁾		2.2	%		2.3	%		6.4	%		(1.2	)%
S&P 500 Index growth rate		4.6	%		(1.4	)%		11.3	%		(10.8	)%
West Texas Intermediate Oil Price ($US)	$	67		$	65		$	78			$ 52

(1)

Federal level forward-looking forecasts are presented in the table above, which represent the aggregation of the provincial level forecasts used to estimate our ECL. Housing Price Index growth rates are also forecasted at the municipal level in some cases. As a result, the forecasts for individual provinces or municipalities reflected in our ECLs will differ from the federal forecasts presented above.

The forecasting process is overseen by a governance committee consisting of internal stakeholders from across our bank including Risk Management, Economics, Finance and the impacted SBUs and involves a significant amount of judgment both in determining the forward-looking information forecasts for our various scenarios and in determining the probability weighting assigned to the scenarios. In general, a worsening of our outlook on forecasted forward-looking information for each scenario or an increase in the probability of the “downside case” scenario occurring will both increase the number of loans migrating from stage 1 to stage 2 and increase the estimated ECL allowance. In contrast, an improvement in our outlook on forecasted forward-looking information or an increase in the probability of the “upside case” scenario occurring will have the opposite impact. It is not possible to meaningfully isolate the impact of changes in the various forward-looking information variables for a particular scenario because of both the interrelationship between the variables and the interrelationship between the level of pessimism inherent in a particular scenario and its probability of occurring.

As indicated above, forecasting forward-looking information for multiple scenarios and determining the probability weighting of the scenarios involves a high degree of management judgment. If we were to only use our downside case scenario for the measurement of ECL for our performing loans, our ECL allowance would be $241 million higher than the recognized ECL as at October 31, 2018. This sensitivity is isolated to the measurement of ECL and therefore did not consider the additional migration of exposures to stage 2 from the additional significant increase in credit risk that would have resulted in a 100% downside scenario.

Management overlays to ECL allowance estimates are adjustments which we use in circumstances where we judge that our existing inputs, assumptions and model techniques do not capture all relevant risk factors. The emergence of new macroeconomic, microeconomic or political events, along with expected changes to parameters, models or data that are not incorporated in our current parameters, internal risk rating migrations, or forward-looking information are examples of such circumstances. The use of management overlays requires the application of significant judgment that may impact the amount of ECL allowances recognized.

The following tables provide the gross carrying amount of loans, and the contractual amounts of undrawn credit facilities and other off–balance sheet exposures based on the application of our 12-month point in time PDs under IFRS 9 to our risk management PD bands for retail exposures, and based on our internal risk ratings for business and government exposures. Refer to “Credit risk” section of the MD&A for details on the CIBC risk categories.

Loans⁽¹⁾

$ millions, as at October 31

2018

Stage 1

Stage 2

Stage 3 ^(2)(3)(4)

Total

Residential mortgages

– Exceptionally low

141,556

–

141,556

– Very low

40,225

–

40,225

– Low

15,321

798

–

16,119

– Medium

859

4,905

–

5,764

– High

–

996

–

996

– Default

–

510

– Not rated

2,163

249

167

2,579

Gross residential mortgages ⁽⁵⁾⁽⁶⁾

200,124

6,948

677

207,749

ECL allowance

143

214

Net residential mortgages

200,097

6,904

534

207,535

Personal

– Exceptionally low

23,808

–

23,808

– Very low

3,813

1,374

–

5,187

– Low

5,954

702

–

6,656

– Medium

4,428

1,151

–

5,579

– High

245

691

–

936

– Default

–

142

– Not rated

677

750

Gross personal ⁽⁶⁾

38,925

3,951

182

43,058

ECL allowance

176

196

109

481

Net personal

38,749

3,755

42,577

Credit card

– Exceptionally low

3,405

–

3,405

– Very low

1,747

–

1,797

– Low

3,809

710

–

4,519

– Medium

1,011

1,241

–

2,252

– High

528

–

538

– Default

–

– Not rated

162

–

162

Gross credit card

10,144

2,529

–

12,673

ECL allowance

330

–

418

Net credit card

10,056

2,199

–

12,255

Business and government

– Investment grade

42,532

221

–

42,753

– Non-investment grade

68,798

3,818

–

72,616

– Watchlist

145

1,120

–

1,265

– Default

–

504

– Not rated

2,397

168

117

2,682

Gross business and government ⁽⁵⁾⁽⁷⁾

113,872

5,327

621

119,820

ECL allowance

159

137

230

526

Net business and government

113,713

5,190

391

119,294

Total net amount of loans

362,615

18,048

998

381,661

(1)	Other financial assets classified at amortized cost were excluded from the table above as their ECL allowances were immaterial as at October 31, 2018. In addition, the table excludes debt securities measured at FVOCI, for which ECL allowances of $23 million were recognized in AOCI.

(2)	Includes purchased credit–impaired loans from the acquisition of The PrivateBank.

(3)	Excludes foreclosed assets of $14 million which were included in Other assets on our consolidated balance sheet.

(4)	As at October 31, 2018, 89% of stage 3 impaired loans were either fully or partially collateralized.

(5)	Includes $12 million of residential mortgages and $16,424 million of business and government loans that are measured at FVTPL.

(6)

The internal risk rating grades presented for residential mortgages and certain personal loans do not take into account loan guarantees or insurance issued by the Canadian government (federal or provincial), Canadian government agencies, or private insurers, as the significant increase in credit risk of these loans is based on relative changes in the loans’ lifetime PD without considering collateral or other credit enhancements.

(7)	Includes customers’ liability under acceptances of $10,265 million.

Undrawn credit facilities and other off-balance sheet exposures

$ millions, as at October 31

2018

Stage 1

Stage 2

Stage 3

Total

Retail

– Exceptionally low

100,772

–

100,772

– Very low

10,217

1,014

–

11,231

– Low

7,873

1,612

–

9,485

– Medium

1,729

1,188

–

2,917

– High

234

417

–

651

– Default

–

– Not rated

348

–

381

Gross retail

121,173

4,264

125,450

ECL allowance

–

Net retail

121,145

4,221

125,379

Business and government

– Investment grade

78,672

390

–

79,062

– Non-investment grade

41,727

1,198

–

42,925

– Watchlist

402

–

477

– Default

–

– Not rated

735

–

786

Gross business and government

121,209

2,041

123,257

ECL allowance

–

Net business and government

121,188

2,031

123,226

Total net undrawn credit facilities and other off-balance sheet exposures

242,333

6,252

248,605

The following tables provide the credit quality of business and government loans and acceptances and retail loans by carrying value as at October 31, 2017. For details on the CIBC rating categories and PD bands, see the “Credit risk” section of the MD&A.

Net business and government loans and acceptances


$ millions, for the year ended October 31									2017
Grade	CIBC rating		Corporate		Sovereign		Banks		Total
Investment grade		00 – 47	$	37,800	$	1,943	$	719	$	40,462
Non-investment grade		51 – 67		38,946		472		188		39,606
Watch list		70 – 80		745		–		–		745
Default		90		257		–		–		257
Total AIRB exposure			$	77,748	$	2,415	$	907	$	81,070
Strong			$	765	$	–	$	–	$	765
Good				126		–		–		126
Satisfactory				9		–		–		9
Weak				–		–		–		–
Default				4		–		–		4
Total slotted exposure			$	904	$	–	$	–	$	904
Standardized exposure			$	23,761	$	213	$	451	$	24,425
			$	102,413	$	2,628	$	1,358	$	106,399
Less: collective allowance on performing loans									$	343
Net business and government loans and acceptances ⁽¹⁾									$	106,056

(1)	Includes customers’ liability under acceptances of $8,824 million.

Net retail loans


$ millions, for the year ended October 31									2017
Risk level	PD bands		Residential mortgages		Personal		Credit cards		Total
Exceptionally low		0.01% – 0.20%	$	158,372	$	22,384	$	3,257	$	184,013
Very low		0.21% – 0.50%		22,512		4,107		1,767		28,386
Low		0.51% – 2.00%		19,223		6,307		4,031		29,561
Medium		2.01% – 10.00%		3,076		6,222		2,482		11,780
High		10.01% – 99.99%		340		895		686		1,921
Default		100%		176		8		–		184
Total AIRB exposure			$	203,699	$	39,923	$	12,223	$	255,845
Strong			$	104	$	–	$	–	$	104
Good				4		–		–		4
Satisfactory				12		–		–		12
Weak				–		–		–		–
Default				1		–		–		1
Total slotted exposure			$	121	$	–	$	–	$	121
Standardized exposure			$	3,306	$	873	$	155	$	4,334
			$	207,126	$	40,796	$	12,378	$	260,300
Less: collective allowance on performing loans			$	58	$	354	$	386	$	798
Net retail loans			$	207,068	$	40,442	$	11,992	$	259,502

Impaired loans


	In accordance with IFRS 9						In accordance with IAS 39
$ millions, as at October 31					2018								2017
	Gross impaired		Stage 3 allowance		Net impaired		Gross impaired		Individual allowance		Collective allowance ⁽¹⁾		Net impaired
Residential mortgages	$	677	$	143	$	534	$	513	$	2	$	143	$	368
Personal		182		109		73		171		7		134		30
Business and government		621		230		391		626		183		8		435
Total impaired loans ⁽²⁾⁽³⁾	$	1,480	$	482	$	998	$	1,310	$	192	$	285	$	833

(1)	Includes collective allowance relating to personal, scored small business and mortgage impaired loans that are greater than 90 days delinquent. In addition, we have a collective allowance of $1,260 million on balances and commitments which are not impaired.

(2)	Average balance of gross impaired loans was $1,333 million (2017: $1,376 million).

(3)	Foreclosed assets of $14 million (2017: $21 million) were included in Other assets on the consolidated balance sheet.

Purchased credit-impaired loans

The following table provides details of our purchased credit-impaired loans resulting from the acquisition of The PrivateBank:


$ millions, as at October 31	2018			2017
Unpaid principal balance ⁽¹⁾	$	20		$	81
Credit related fair value adjustments		(3	)		(15	)
Time value of money		(1	)		(3	)
Carrying value		16			63
Stage 3 allowance (2017: Individually assessed allowance)		(2	)		(2	)
Carrying value net of related allowance	$	14		$	61

(1)	Represents principal amount owed net of write-offs since the acquisition of the loan.

Contractually past due loans but not impaired

This comprises loans where repayment of principal or payment of interest is contractually in arrears. The following table provides an aging analysis of the contractually past due loans:

$ millions, as at October 31

Less than
31 days

31 to

90 days

Over

90 days

2018 ⁽¹⁾

Total

2017

Total

Residential mortgages

2,505

849

–

3,354

3,546

Personal

751

186

–

937

915

Credit card

547

172

103

822

853

Business and government

525

158

–

683

811

4,328

1,365

103

5,796

6,125

(1)

Effective November 1, 2017, all loans that are contractually 90 days in arrears are automatically classified as impaired and as stage 3 under IFRS 9, except for credit card loans which are classified as impaired and are fully written off when payments are contractually 180 days in arrears or at the earlier of the notice of bankruptcy, settlement proposal, or enlistment of credit counselling services. The determination of impairment was generally the same under IAS 39, except (i) residential mortgages guaranteed or insured by a Canadian government (federal or provincial) or a Canadian government agency were not classified as impaired until payments were contractually 365 days in arrears, and (ii) residential mortgages guaranteed or insured by a private insurer, or loans that were fully secured and in the process of collection were not classified as impaired until payments were contractually 180 days in arrears.

During the year, gross interest income that would have been recorded if impaired loans were treated as current was $81 million (2017: $78 million), of which $27 million (2017: $23 million) was in Canada and $54 million (2017: $55 million) was outside Canada. During the year, interest recognized on impaired loans was $23 million (2017: $26 million), and interest recognized on loans before being classified as impaired was $59 million (2017: $45 million), of which $41 million (2017: $35 million) was in Canada and $18 million (2017: $10 million) was outside Canada.

Net interest income after provision for credit losses


$ millions, for the year ended October 31	2018		2017		2016
Interest income	$	17,505	$	13,593	$	12,092
Interest expense		7,440		4,616		3,726
Net interest income		10,065		8,977		8,366
Provision for credit losses		870		829		1,051
Net interest income after provision for credit losses	$	9,195	$	8,148	$	7,315

Modified financial assets

From time to time, we may modify the contractual terms of loans classified as stage 2 and stage 3 for which the borrower has experienced financial difficulties, through the granting of a concession in the form of below-market rates or terms that we would not otherwise have considered. Changes to the present value of the estimated future cash payments through the expected life of the modified loan discounted at the loan’s original effective interest rate are recognized through changes in the ECL allowance and provision for credit losses. During the year ended October 31, 2018, loans classified as stage 2 with an amortized cost of $133 million and loans classified as stage 3 with an amortized cost of $119 million, in each case before the time of modification, were modified through the granting of a financial concession in response to the borrower having experienced financial difficulties. In addition, the gross carrying amount of previously modified stage 2 or stage 3 loans that have returned to stage 1 during the year ended October 31, 2018 was $42 million.