Allowance for Loan Losses

6 Months Ended

Jun. 30, 2020

Accounts, Notes, Loans and Financing Receivable, Gross, Allowance, and Net [Abstract]

Allowance for Loan Losses

Allowance for Loan Losses

As discussed in Note 1 - Summary of Significant Accounting Policies, the ALLL estimation process was revised on January 1, 2020 to reflect the adoption of ASU 2016-13. All information contained in the following disclosures reflects the application of requirements from the adoption of ASU 2016-13 for periods after 2019. Information for periods prior to 2020 has been retained with the content consistent with prior disclosures.

Periods after 2019

The ALLL has been determined in accordance with ASC 326-20, which requires a recognition of current expected credit losses on the amortized cost basis of loans. During the first half of 2020, expected credit loss estimates were adversely affected across all portfolio segments due to the steep decline in macroeconomic forecasts due to the actual and projected effects of the COVID-19 pandemic. To a lesser extent, loan growth also resulted in a higher ALLL as those increased balances received a full life-of-loan allowance based on current macroeconomic projections.

For all portfolio segments, FHN has selected a 4-year reasonable and supportable forecast period which reflects a 3-year period during which macroeconomic variables are used to estimate expected credit losses. This is followed by a 1-year, time-weighted reversion to historical loss factors with weights assigned to macroeconomic variables diminishing, and weights assigned to historical loss averages increasing, pro rata as months lapse during the 1-year period. Thereafter, FHN immediately reverts to historical loss averages over the remaining estimated life of loans.

In developing credit loss estimates for its loan portfolio, FHN evaluated multiple macroeconomic forecasts provided by Moody’s. FHN selected Moody’s baseline forecast as the primary source for its macroeconomic inputs which are inclusive of the following assumptions related to the economic effects of the COVID-19 pandemic:

• 1.6mm confirmed COVID-19 cases, begin to abate by early July. Assumes no second wave.

• Unemployment peak at 20mm in 2Q20

• Annualized GDP growth rate of -33% in 2Q20, 16% in 3Q20, & 0.6% in 4Q20

• Quicker recovery than Great Recession

• V-shaped recovery for housing including residential construction and sales

• Manufacturing should have a stronger initial bounce

• More caution due to COVID-19 drives a slower and longer recovery for the service sector; may depend on vaccine.

Where macroeconomic forecast variables used in the models do not take into effect the impact of federal stimulus and bank-supported payment deferral and forbearance programs on the timing of grade migration and recognition of loss content, management adjusted model inputs qualitatively to account for this assistance.

FHN also utilized targeted reviews of higher stressed loan portfolio (industries) that are most exposed to the effects of the COVID-19 pandemic, including Franchise Finance, Energy, Non-Profit, Arts & Entertainment, Restaurants outside of Franchise Finance, Nursing/Assisted Living and Hospitality within the C&I segment and CRE-Hospitality and CRE-Retail within the Commercial Real Estate segment. This analysis reviewed the level of impact from COVID-19 and the likelihood of additional financial assistance needed beyond 180 days. This analysis was utilized in developing qualitative adjustments to increase the recorded ALLL attributable to these components beyond the modeled results. FHN reviewed consumer deferrals and forbearance payment rates to analyze the likelihood customers will have difficulty making payments after the deferral or forbearance period ends. The analysis was utilized to develop an additional qualitative adjustment to increase the recorded ALLL for consumer real estate loans. Management also made qualitative adjustments to reflect estimated recoveries based on a review of prior charge off and recovery levels, for default risk associated with large balances with individual borrowers, for estimated loss amounts not reflected in historical factors due to specific portfolio risk and for instances where limited data for acquired loans is considered to affect modeled results.

Typically commercial loans in C&I and CRE have shorter expected lives, based on the contractual term of loan agreements, prepayment estimates and a limited amount of renewal or extension options that are not unconditionally cancellable by FHN. Estimated weighted average lives are normally under 3 years. TRUPs loans are an exception due to longer contractual lives, beneficial borrower terms and balloon payoff structure. Consumer HELOC and installment loans tend to have significantly longer lives based on their contractual terms which is reduced somewhat by estimated prepayments with estimated weighted average lives normally 5 years or less. Credit card loans have shorter estimated lives approximating 1 year based on customer payment trends and because the revolving lines are unconditionally cancellable by FHN.

As of June 30, 2020, FHN had General C&I loans with amortized cost of approximately $65 million that was based on the value of underlying collateral. At a minimum, the estimated value of the collateral for each loan equals the current book value. The collateral for these loans

generally consists of business assets including land, buildings, equipment and financial assets. During the three and six months ended June 30, 2020, respectively, FHN recognized charge-offs of approximately $1.5 million and $13.0 million on these loans related to reductions in estimated collateral values.

Consumer HELOC and installment loans with amortized cost based on the value of underlying real estate collateral were approximately $10 million and $26 million, respectively, as of June 30, 2020. At a minimum, the estimated value of the collateral for each loan equals the current book value. Charge offs during the three and six months ended June 30, 2020 were not significant for either portfolio segment.

Unfunded Commitments

The measurement of expected credit losses for unfunded commitments mirrors that of loans with the additional estimate of future draw rates (timing and amount). Consistent with the ALLL, the decline in macroeconomic forecasts during the first half of 2020 resulted in higher credit expense for unfunded commitments. However, in Q1 2020 this effect of higher loss forecasts was offset somewhat because many borrowers drew on available lines prior to the end of the 1st quarter which resulted in higher loan balances (and ALLL). During the second quarter, borrowers paid down on available lines which combined with higher loss rates caused credit loss expense to increase. Total credit loss expense for unfunded commitments was $11.2 million and $20.4 million, respectively, for the three and six months ended June 30, 2020.

Periods prior to 2020

The ALLL included the following components: reserves for commercial loans evaluated based on pools of credit graded loans and reserves for pools of smaller-balance homogeneous consumer loans, both determined in accordance with ASC 450-20-50, and to a lesser extent, reserves determined in accordance with ASC 310-10-35 for loans determined by management to be individually impaired and an allowance associated with PCI loans.

For commercial loans, ASC 450-20-50 reserves were established using historical net loss factors by grade level, loan product, and business segment. The ALLL for smaller-balance homogeneous consumer loans was determined based on pools of similar loan types that have similar credit risk characteristics. ASC 450-20-50 reserves for the consumer portfolio were determined using segmented roll-rate models that incorporated various factors including historical delinquency trends, experienced loss frequencies, and experienced loss

severities. Generally, reserves for consumer loans reflected inherent losses in the portfolio that were expected to be recognized over the following twelve months. The historical net loss factors for both commercial and consumer ASC 450-20-50 reserve models were subject to qualitative adjustments by management to reflect current events, trends, and conditions (including economic considerations and trends), which were not fully captured in the historical net loss factors. The pace of the economic recovery, performance of the housing market, unemployment levels, labor participation rate, the regulatory environment, regulatory guidance, and portfolio segment-specific trends, were examples of additional factors considered by management in determining the ALLL. Additionally, management considered the inherent uncertainty of quantitative models that were driven by historical loss data. Management evaluated the periods of historical losses that were the basis for the loss rates used in the quantitative models and selected historical loss periods that were believed to be the most reflective of losses inherent in the loan portfolio as of the balance sheet date. Management also periodically reviewed an analysis of the loss emergence period which was the amount of time required for a loss to be confirmed (initial charge-off) after a loss event had occurred. FHN performed extensive studies related to the historical loss periods used in the model and the loss emergence period and model assumptions were adjusted accordingly.

Impairment related to individually impaired loans was measured in accordance with ASC 310-10. For all commercial portfolio segments, commercial TDRs and other individually impaired commercial loans were measured based on the present value of expected future payments discounted at the loan’s effective interest rate (“the DCF method”), observable market prices, or for loans that are solely dependent on the collateral for repayment, the net realizable value (collateral value less estimated costs to sell). Impaired loans also included consumer TDRs. Generally, the allowance for TDRs in all consumer portfolio segments was determined by estimating the expected future cash flows using the modified interest rate (if an interest rate concession), incorporating payoff and net charge-off rates specific to the TDRs within the portfolio segment being assessed, and discounted using the pre-modification interest rate. The discount rates of variable rate TDRs were adjusted to reflect changes in the interest rate index to which the rates are tied. The discounted cash flows were then compared to the outstanding principal balance in order to determine required reserves. Residential real estate loans discharged through bankruptcy were considered collateral-dependent and were charged down to net realizable value (collateral value less estimated costs to sell).

The following table provides a rollforward of the allowance for loan losses by portfolio segment for the three and six months ended June 30, 2020 and 2019:

(Dollars in thousands)	C&I			Commercial Real Estate			Consumer Real Estate (a)			Credit Card and Other			Total
Balance as of April 1, 2020	$	254,516		$	47,625		$	123,022		$	19,327		$	444,490
Charge-offs	(18,201		)	(61		)	(1,968		)	(2,677		)	(22,907		)
Recoveries	1,073			156			3,987			1,082			6,298
Provision/(provision credit) for loan losses	81,334			9,565			18,716			385			110,000
Balance as of June 30, 2020	318,722			57,285			143,757			18,117			537,881
Balance as of January 1, 2020	122,486			36,112			28,443			13,266			200,307
Adoption of ASU 2016-13	18,782			(7,348		)	92,992			1,968			106,394
Charge-offs	(24,952		)	(642		)	(4,278		)	(6,488		)	(36,360		)
Recoveries	2,008			729			7,542			2,261			12,540
Provision/(provision credit) for loan loses	200,398			28,434			19,058			7,110			255,000
Balance as of June 30, 2020	318,722			57,285			143,757			18,117			537,881
Allowance - individually evaluated for impairment	16,577			—			13,275			352			30,204
Allowance - collectively evaluated for impairment	302,145			57,285			130,482			17,765			507,677
Loans, net of unearned as of June 30, 2020:
Individually evaluated for impairment	140,526			160			148,543			688			289,917
Collectively evaluated for impairment	21,253,367			4,813,181			5,903,850			448,622			32,419,020
Total loans, net of unearned income (b)	$	21,393,893		$	4,813,341		$	6,052,393		$	449,310		$	32,708,937
Balance as of April 1, 2019	$	103,713		$	34,382		$	34,154		$	12,662		$	184,911
Charge-offs	(6,590		)	(121		)	(1,714		)	(3,798		)	(12,223		)
Recoveries	519			(88		)	5,525			1,105			7,061
Provision/(provision credit) for loan losses	18,454			(1,220		)	(6,433		)	2,199			13,000
Balance as of June 30, 2019	116,096			32,953			31,532			12,168			192,749
Balance as of January 1, 2019	98,947			31,311			37,439			12,727			180,424
Charge-offs	(9,691		)	(555		)	(4,518		)	(7,986		)	(22,750		)
Recoveries	1,348			(31		)	9,566			2,192			13,075
Provision/(provision credit) for loan losses	25,492			2,228			(10,955		)	5,235			22,000
Balance as of June 30, 2019	116,096			32,953			31,532			12,168			192,749
Allowance - individually evaluated for impairment	8,484			—			22,255			442			31,181
Allowance - collectively evaluated for impairment	106,758			32,953			8,199			11,669			159,579
Allowance - purchased credit-impaired loans	854			—			1,078			57			1,989
Loans, net of unearned as of June 30, 2019:
Individually evaluated for impairment	115,808			1,777			182,442			699			300,726
Collectively evaluated for impairment	18,904,621			3,852,027			6,093,863			492,774			29,343,285
Purchased credit-impaired loans	33,840			7,227			26,829			903			68,799
Total loans, net of unearned income	$	19,054,269		$	3,861,031		$	6,303,134		$	494,376		$	29,712,810

a) In first quarter 2020, the Permanent Mortgage portfolio was combined into Consumer Real Estate portfolio, all prior periods were revised for comparability.

b) C&I loans as of June 30, 2020 include $2.1 billion in PPP loans which due to the government guarantee and forgiveness provisions are considered to have no credit risk and therefore have no allowance for loan losses.

In accordance with its accounting policy elections, FHN does not recognize a separate allowance for expected credit losses for AIR and records reversals of AIR as reductions of interest income. FHN reverses previously accrued but uncollected interest when an asset is placed on nonaccrual status. As of June 30, 2020, FHN recognized approximately $860 thousand in allowance for expected credit losses on COVID-19 deferrals that do not qualify for the election.

The total amount of interest reversals from loans placed on nonaccrual status during the three and six months ended June 30, 2020 was not material. In addition, the amount of income recognized on nonaccrual loans for the three and six months ended in June 30, 2020 was not material.