Trustmark adopted the amendments of FASB ASU 2016-13, on January 1, 2020. The amendments of ASU 2016-13 created FASB Accounting Standards Codification (ASC) Topic 326, “Financial Instruments – Credit Losses,” which, among other things, replace much of the guidance and disclosures previously provided in FASB ASC Topic 310, “Receivables.” The guidance in FASB ASC Topic 326 replaces the incurred loss methodology with a methodology that reflects expected credit losses and requires consideration of a broader range of reasonable and supportable information to determine credit losses. In accordance with FASB ASC Subtopic 326-20, “Financial Instruments – Credit Losses – Measured at Amortized Cost,” Trustmark has developed an ACL methodology effective January 1, 2020, which replaces its previous allowance for loan losses methodology. See the section captioned “Allowance for Credit Losses (ACL)” within this note for additional information regarding Trustmark’s ACL. Trustmark adopted FASB ASC Topic 326 using the modified retrospective approach prescribed by the amendments of ASU 2016-13; therefore, prior period balances are presented under legacy GAAP and may not be comparable to current period presentation.

At September 30, 2020 and December 31, 2019, LHFI consisted of the following ($ in thousands):

Accrued interest receivable is not included in the amortized cost basis of Trustmark’s LHFI. At September 30, 2020, accrued interest receivable for LHFI totaled $37.4 million with no related ACL and was reported in other assets on the accompanying consolidated balance sheet.

Trustmark does not have any loan concentrations other than those reflected in the preceding table, which exceed 10% of total LHFI. At September 30, 2020, Trustmark’s geographic loan distribution was concentrated primarily in its five key market regions: Alabama, Florida, Mississippi, Tennessee and Texas. Accordingly, the ultimate collectability of a substantial portion of these loans is susceptible to changes in market conditions in these areas.

Past due LHFI are loans contractually past due 30 days or more as to principal or interest payments. A LHFI is classified as nonaccrual, and the accrual of interest on such loan is discontinued, when the contractual payment of principal or interest becomes 90 days past due on commercial credits and 120 days past due on non-business purpose credits. In addition, a credit may be placed on nonaccrual at any other time Management has serious doubts about further collectibility of principal or interest according to the contractual terms, even though the loan is currently performing. A LHFI may remain in accrual status if it is in the process of collection and well secured. When a LHFI is placed in nonaccrual status, interest accrued but not received is reversed against interest income. Interest payments received on nonaccrual LHFI are applied against principal under the cost-recovery method, until qualifying for return to accrual status. Under the cost-recovery method, interest income is not recognized until the principal balance is reduced to zero. LHFI are restored to accrual status when the obligation is brought current or has performed in accordance with the contractual terms for a reasonable period of time and the ultimate collectibility of the total contractual principal and interest is no longer in doubt.

No material interest income was recognized in the income statement on nonaccrual LHFI for each of the periods ended September 30, 2020 and 2019.

The following table provides the amortized cost basis of loans on nonaccrual status and loans past due 90 days or more still accruing interest as of September 30, 2020 ($ in thousands):

The following table provides an aging analysis of the amortized cost basis of past due LHFI at September 30, 2020 ($ in thousands):

	September 30, 2020
	Past Due
	30-59 Days		60-89 Days		90 Days or More		Total Past Due		Current Loans		Total LHFI
Loans secured by real estate:
Construction, land development and other land	$	867	$	262	$	186	$	1,315	$	493,204	$	494,519
Other secured by 1-4 family residential properties		906		664		908		2,478		544,670		547,148
Secured by nonfarm, nonresidential properties		172		93		881		1,146		2,706,481		2,707,627
Other real estate secured		15		107		—		122		887,670		887,792
Other loans secured by real estate:
Other construction		—		—		—		—		891,428		891,428
Secured by 1-4 family residential properties		2,882		1,034		3,857		7,773		1,220,479		1,228,252
Commercial and industrial loans		740		340		1,922		3,002		1,395,466		1,398,468
Consumer loans		899		226		124		1,249		162,684		163,933
State and other political subdivision loans		—		—		177		177		935,172		935,349
Other commercial loans		46		520		401		967		592,245		593,212
Total	$	6,527	$	3,246	$	8,456	$	18,229	$	9,829,499	$	9,847,728

The following table provides an aging analysis of past due and nonaccrual LHFI by loan class at December 31, 2019 ($ in thousands):

	December 31, 2019
	Past Due
	30-59 Days		60-89 Days		90 Days or More (1)		Total		Nonaccrual		Current Loans		Total LHFI
Loans secured by real estate:
Construction, land development and other land	$	380	$	256	$	—	$	636	$	897	$	1,161,258	$	1,162,791
Secured by 1-4 family residential properties		5,254		940		211		6,405		16,810		1,832,698		1,855,913
Secured by nonfarm, nonresidential properties		1,698		—		—		1,698		7,700		2,465,847		2,475,245
Other real estate secured		8		—		—		8		1,032		723,440		724,480
Commercial and industrial loans		617		12		39		668		21,775		1,455,453		1,477,896
Consumer loans		2,208		380		392		2,980		108		172,650		175,738
State and other political subdivision loans		76		—		—		76		4,079		963,789		967,944
Other loans		152		4		—		156		825		494,640		495,621
Total	$	10,393	$	1,592	$	642	$	12,627	$	53,226	$	9,269,775	$	9,335,628

Prior to the adoption of FASB ASC Topic 326, Trustmark’s individually evaluated impaired LHFI included all commercial nonaccrual relationships of $500 thousand or more, which were specifically reviewed for impairment and deemed impaired, and all LHFI classified as TDRs in accordance with FASB ASC Subtopic 310-10-50-20 “Impaired Loans.” Once a LHFI was deemed to be impaired, the full difference between book value and the most likely estimate of the collateral’s net realizable value was charged off or a specific reserve was established.

No material interest income was recognized in the income statement on impaired LHFI for the periods ended September 30, 2019.

At December 31, 2019, individually evaluated impaired LHFI consisted of the following ($ in thousands):

	December 31, 2019
	LHFI
	Unpaid Principal Balance		With No Related Allowance Recorded		With an Allowance Recorded		Total Recorded Investment		Related Allowance		Average Recorded Investment
Loans secured by real estate:
Construction, land development and other land	$	926	$	610	$	16	$	626	$	—	$	1,089
Secured by 1-4 family residential properties		6,513		2,104		3,360		5,464		35		4,713
Secured by nonfarm, nonresidential properties		7,295		1,462		5,255		6,717		2,355		8,096
Other real estate secured		69		—		68		68		—		158
Commercial and industrial loans		27,178		19,374		4,084		23,458		707		27,088
Consumer loans		22		—		21		21		—		11
State and other political subdivision loans		4,079		—		4,079		4,079		1,809		6,337
Other loans		1,207		—		784		784		553		1,033
Total	$	47,289	$	23,550	$	17,667	$	41,217	$	5,459	$	48,525

A TDR occurs when a borrower is experiencing financial difficulties, and for related economic or legal reasons, a concession is granted to the borrower that Trustmark would not otherwise consider. Whatever the form of concession that might be granted by Trustmark, Management’s objective is to enhance collectibility by obtaining more cash or other value from the borrower or by increasing the probability of receipt by granting the concession than by not granting it. Other concessions may arise from court proceedings or may be imposed by law. In addition, TDRs also include those credits that are extended or renewed to a borrower who is not able to obtain funds from sources other than Trustmark at a market interest rate for new debt with similar risk.

At September 30, 2020 and 2019, LHFI classified as TDRs totaled $25.4 million and $30.5 million, respectively, and were primarily comprised of credits with interest-only payments for an extended period of time and credits renewed at a rate that was not commensurate with that of new debt with similar risk which totaled $13.3 million and $20.2 million at September 30, 2020 and 2019, respectively. The remaining TDRs at September 30, 2020 and 2019 resulted from bankruptcies or from payment or maturity extensions. Trustmark had $4.2 million of unused commitments on TDRs at September 30, 2020 compared to $7.9 million of unused commitments on TDRs at September 30, 2019.

At September 30, 2020, TDRs had a related ACL of $2.6 million, compared to a related allowance for loan loss of $2.9 million at September 30, 2019, and resulted in charge-offs of $2.3 million and $472 thousand for the nine months ended September 30, 2020 and 2019, respectively.

The following table illustrates the impact of modifications classified as TDRs for the periods presented ($ in thousands):

	Nine Months Ended September 30,
	2020						2019
	Number of Contracts		Pre-Modification Outstanding Recorded Investment		Post-Modification Outstanding Recorded Investment		Number of Contracts		Pre-Modification Outstanding Recorded Investment		Post-Modification Outstanding Recorded Investment
Loans secured by real estate:
Other secured by 1-4 family residential properties		11	$	794	$	800		11	$	992	$	980
Secured by nonfarm, nonresidential properties		1		139		139		1		5,055		5,055
Commercial and industrial loans		3		1,630		1,629		8		9,167		9,054
Consumer loans		6		26		26		2		30		30
State and other political subdivision loans		2		3,902		3,872		—		—		—
Total		23	$	6,491	$	6,466		22	$	15,244	$	15,119

The table below includes the balances at default for TDRs modified within the last 12 months for which there was a payment default during the periods presented ($ in thousands):

Trustmark’s TDRs have resulted primarily from allowing the borrower to pay interest-only for an extended period of time and credits renewed at a rate that was not commensurate with that of new debt with similar risk rather than from forgiveness. Accordingly, as shown above, these TDRs have a similar recorded investment for both the pre-modification and post-modification disclosure. Trustmark has utilized loans 90 days or more past due to define payment default in determining TDRs that have subsequently defaulted.

The following tables detail LHFI classified as TDRs by loan class at September 30, 2020 and 2019 ($ in thousands):

On March 27, 2020, the CARES Act, a stimulus package intended to provide relief to businesses and consumers in the United States struggling as a result of the pandemic, was signed into law. Section 4013 of the CARES Act also addressed COVID-19- related modifications and specified that COVID-19 related modifications executed between March 1, 2020 and the earlier of (i) 60 days after the date of termination of the national emergency declared by the President and (ii) December 31, 2020, on loans that were current as of December 31, 2019 are not TDRs. Additionally, under guidance from the federal banking agencies, other short-term modifications made on a good faith basis in response to COVID-19 to borrowers that were current prior to any relief are not TDRs under ASC Subtopic 310-40, “Troubled Debt Restructuring by Creditors.” These modifications include short-term (e.g., up to six months) modifications such as payment deferrals, fee waivers, extensions of repayment terms, or delays in payment that are insignificant. Borrowers considered current are those that are less than 30 days past due on their contractual payments at the time a modification program is implemented. Trustmark modified 2,812 individual loans with aggregate principal balances totaling $1.327 billion through September 30, 2020 without treating such modifications as TDRs. Commercial concessions were primarily either interest only for 90 days or full payment deferrals for 90 days. Consumer concessions were 90-day full payment deferrals.

The following table presents the amortized cost basis of collateral-dependent loans by class of loans and collateral type as of September 30, 2020 ($ in thousands):

	September 30, 2020
	Real Estate		Equipment and Machinery		Inventory and Receivables		Vehicles		Miscellaneous		Total
Loans secured by real estate:
Construction, land development and other land	$	1,024	$	—	$	—	$	—	$	—	$	1,024
Other secured by 1-4 family residential properties		474		—		—		—		—		474
Secured by nonfarm, nonresidential properties		13,527		—		—		—		—		13,527
Other real estate secured		61		—		—		—		—		61
Other loans secured by real estate:
Other construction		—		—		—		—		—		—
Secured by 1-4 family residential properties		1,453		—		—		—		—		1,453
Commercial and industrial loans		96		798		5,174		155		8,660		14,883
Consumer loans		—		—		—		—		—		—
State and other political subdivision loans		4,019		—		—		—		—		4,019
Other commercial loans		623		—		1,983		—		3,120		5,726
Total	$	21,277	$	798	$	7,157	$	155	$	11,780	$	41,167

A loan is collateral dependent when the borrower is experiencing financial difficulty and repayment of the loan is expected to be provided substantially through the sale of the collateral. The following provides a qualitative description by class of loan of the collateral that secures Trustmark’s collateral-dependent LHFI:

Trustmark’s LHFI portfolio credit quality indicators focus on six key quality ratios that are compared against bank tolerances. The loan indicators are total classified outstanding, total criticized outstanding, nonperforming loans, nonperforming assets, delinquencies and net loan losses. Due to the homogenous nature of consumer loans, Trustmark does not assign a formal internal risk rating to each credit and therefore the criticized and classified measures are primarily composed of commercial loans.

In addition to monitoring portfolio credit quality indicators, Trustmark also measures how effectively the lending process is being managed and risks are being identified. As part of an ongoing monitoring process, Trustmark grades the commercial portfolio segment as it relates to credit file completion and financial statement exceptions, underwriting, collateral documentation and compliance with law as shown below:

Trustmark has established a loan grading system that consists of ten individual credit risk grades (risk ratings) that encompass a range from loans where the expectation of loss is negligible to loans where loss has been established. The model is based on the risk of default for an individual credit and establishes certain criteria to delineate the level of risk across the ten unique credit risk grades. Credit risk grade definitions are as follows:

By definition, credit risk grades special mention (RR 7), substandard (RR 8), doubtful (RR 9) and loss (RR 10) are criticized loans while substandard (RR 8), doubtful (RR 9) and loss (RR 10) are classified loans. These definitions are standardized by all bank regulatory agencies and are generally equally applied to each individual lending institution. The remaining credit risk grades are considered pass credits and are solely defined by Trustmark.

To enhance this process, Trustmark has determined that loans will be individually assessed, and a formal analysis will be performed and based upon the analysis the loan will be written down to net realizable value. Trustmark will individually assess and remove loans from the pool in the following circumstances:

Each loan officer assesses the appropriateness of the internal risk rating assigned to their credits on an ongoing basis. Trustmark’s Asset Review area conducts independent credit quality reviews of the majority of Trustmark’s commercial loan portfolio both on the underlying credit quality of each individual loan class as well as the adherence to Trustmark’s loan policy and the loan administration process.

In addition to the ongoing internal risk rate monitoring described above, Trustmark’s Credit Quality Review Committee meets monthly and performs a review of all loans of $100 thousand or more that are either delinquent thirty days or more or on nonaccrual. This review includes recommendations regarding risk ratings, accrual status, charge-offs and appropriate servicing officer as well as evaluation of problem credits for determination of TDRs. Quarterly, the Credit Quality Review Committee reviews and modifies continuous action plans for all credits risk rated seven or worse for relationships of $100 thousand or more.

In addition, periodic reviews of significant development, commercial construction, multi-family and nonowner-occupied projects are performed. These reviews assess each particular project with respect to location, project valuations, progress of completion, leasing status, current financial information, rents, operating expenses, cash flow, adherence to budget and projections and other information as applicable. Summary results are reviewed by Senior and Regional Credit Officers in addition to the Chief Credit Officer with a determination made as to the appropriateness of existing risk ratings and accrual status.

Consumer LHFI that do not meet a minimum custom credit score are reviewed quarterly by Management. The Retail Credit Review Committee reviews the volume and percentage of approvals that did not meet the minimum passing custom score by region, individual location, and officer to ensure that Trustmark continues to originate quality loans.

Trustmark monitors the levels and severity of past due consumer LHFI on a daily basis through its collection activities. A detailed assessment of consumer LHFI delinquencies is performed monthly at both a product and market level by delivery channel, which incorporates the perceived level of risk at time of underwriting.

The tables below present the amortized cost basis of loans by credit quality indicator and class of loans based on the most recent analysis performed ($ in thousands):

	Term Loans by Origination Year
	2020		2019		2018		2017		2016		Prior		Revolving Loans		Total
As of September 30, 2020	Commercial LHFI
Loans secured by real estate:
Construction, land development and other land:
Pass - RR 1 through RR 6	$	226,639	$	96,572	$	33,263	$	4,709	$	3,424	$	4,336	$	29,238	$	398,181
Special Mention - RR 7		268		—		—		—		—		—		—		268
Substandard - RR 8		6,304		4,194		1,372		34		680		23		—		12,607
Doubtful - RR 9		—		—		—		—		—		42		—		42
Total		233,211		100,766		34,635		4,743		4,104		4,401		29,238		411,098

Other secured by 1-4 family residential properties:
Pass - RR 1 through RR 6	$	28,504	$	23,022	$	18,840	$	10,616	$	11,869	$	6,610	$	8,288	$	107,749
Special Mention - RR 7		46		43		—		—		—		—		—		89
Substandard - RR 8		934		82		981		353		305		613		4,150		7,418
Doubtful - RR 9		31		—		—		—		—		—		—		31
Total		29,515		23,147		19,821		10,969		12,174		7,223		12,438		115,287

Secured by nonfarm, nonresidential properties:
Pass - RR 1 through RR 6	$	545,273	$	503,516	$	435,877	$	326,971	$	274,113	$	333,242	$	110,298	$	2,529,290
Special Mention - RR 7		6,330		6,916		13,487		4,551		5,053		15,227		—		51,564
Substandard - RR 8		14,093		22,207		3,515		13,092		35,939		34,326		2,130		125,302
Doubtful - RR 9		55		167		—		—		217		318		—		757
Total		565,751		532,806		452,879		344,614		315,322		383,113		112,428		2,706,913

Other real estate secured:
Pass - RR 1 through RR 6	$	107,988	$	206,213	$	303,645	$	107,141	$	110,083	$	27,682	$	11,336	$	874,088
Special Mention - RR 7		—		—		—		—		—		858		—		858
Substandard - RR 8		11,076		148		18		—		566		501		—		12,309
Doubtful - RR 9		—		—		—		—		—		—		—		—
Total		119,064		206,361		303,663		107,141		110,649		29,041		11,336		887,255

	Term Loans by Origination Year
	2020		2019		2018		2017		2016		Prior		Revolving Loans		Total
As of September 30, 2020	Commercial LHFI
Other loans secured by real estate:
Other construction
Pass - RR 1 through RR 6	$	113,727	$	564,093	$	194,012	$	13,346	$	1,605	$	—	$	4,060	$	890,843
Special Mention - RR 7		—		—		—		—		—		—		—		—
Substandard - RR 8		585		—		—		—		—		—		—		585
Doubtful - RR 9		—		—		—		—		—		—		—		—
Total		114,312		564,093		194,012		13,346		1,605		—		4,060		891,428

Commercial and industrial loans:
Pass - RR 1 through RR 6	$	433,771	$	184,698	$	104,647	$	83,887	$	74,401	$	49,508	$	394,527	$	1,325,439
Special Mention - RR 7		844		—		177		—		—		—		3,000		4,021
Substandard - RR 8		5,824		1,731		15,519		3,781		2,286		5,800		33,604		68,545
Doubtful - RR 9		178		219		—		—		38		4		24		463
Total		440,617		186,648		120,343		87,668		76,725		55,312		431,155		1,398,468

State and other political subdivision loans:
Pass - RR 1 through RR 6	$	141,729	$	84,369	$	51,181	$	116,529	$	135,190	$	396,607	$	832	$	926,437
Special Mention - RR 7		—		—		—		—		—		4,000		—		4,000
Substandard - RR 8		—		—		—		293		—		4,619		—		4,912
Doubtful - RR 9		—		—		—		—		—		—		—		—
Total		141,729		84,369		51,181		116,822		135,190		405,226		832		935,349

Other commercial loans:
Pass - RR 1 through RR 6	$	74,539	$	76,935	$	23,491	$	11,708	$	56,530	$	51,725	$	265,683	$	560,611
Special Mention - RR 7		7,917		—		—		—		—		—		11,333		19,250
Substandard - RR 8		127		2,146		686		—		893		—		9,473		13,325
Doubtful - RR 9		—		3		—		—		—		23		—		26
Total		82,583		79,084		24,177		11,708		57,423		51,748		286,489		593,212

Total commercial LHFI	$	1,726,782	$	1,777,274	$	1,200,711	$	697,011	$	713,192	$	936,064	$	887,976	$	7,939,010

	Term Loans by Origination Year
	2020		2019		2018		2017		2016		Prior		Revolving Loans		Total
As of September 30, 2020	Consumer LHFI
Loans secured by real estate:
Construction, land development and other land:
Current	$	32,812	$	34,108	$	8,758	$	2,382	$	1,576	$	3,347	$	—	$	82,983
Past due 30-89 days		—		—		241		—		—		74		—		315
Past due 90 days or more		—		—		—		—		—		—		—		—
Nonaccrual		—		—		—		—		19		104		—		123
Total		32,812		34,108		8,999		2,382		1,595		3,525		—		83,421

Other secured by 1-4 family residential properties:
Current	$	18,709	$	13,768	$	13,212	$	4,295	$	2,142	$	14,300	$	360,724	$	427,150
Past due 30-89 days		126		297		—		15		58		186		538		1,220
Past due 90 days or more		—		15		12		—		—		—		272		299
Nonaccrual		8		33		110		436		—		462		2,143		3,192
Total		18,843		14,113		13,334		4,746		2,200		14,948		363,677		431,861

Secured by nonfarm, nonresidential properties:
Current	$	698	$	—	$	—	$	4	$	—	$	12	$	—	$	714
Past due 30-89 days		—		—		—		—		—		—		—		—
Past due 90 days or more		—		—		—		—		—		—		—		—
Nonaccrual		—		—		—		—		—		—		—		—
Total		698		—		—		4		—		12		—		714

Other real estate secured:
Current	$	139	$	—	$	41	$	38	$	102	$	217	$	—	$	537
Past due 30-89 days		—		—		—		—		—		—		—		—
Past due 90 days or more		—		—		—		—		—		—		—		—
Nonaccrual		—		—		—		—		—		—		—		—
Total		139		—		41		38		102		217		—		537

	Term Loans by Origination Year
	2020		2019		2018		2017		2016		Prior		Revolving Loans		Total
As of September 30, 2020	Consumer LHFI
Other loans secured by real estate:
Secured by 1-4 family residential properties
Current	$	217,950	$	226,130	$	194,593	$	102,945	$	119,707	$	352,569	$	—	$	1,213,894
Past due 30-89 days		256		—		71		274		1,127		1,180		—		2,908
Past due 90 days or more		7		—		—		—		181		171		—		359
Nonaccrual		148		839		2,051		439		241		7,373		—		11,091
Total		218,361		226,969		196,715		103,658		121,256		361,293		—		1,228,252

Consumer loans:
Current	$	54,252	$	32,198	$	16,771	$	4,956	$	1,625	$	617	$	52,170	$	162,589
Past due 30-89 days		329		40		70		168		10		5		503		1,125
Past due 90 days or more		24		1		19		—		—		—		79		123
Nonaccrual		11		6		59		16		3		—		1		96
Total		54,616		32,245		16,919		5,140		1,638		622		52,753		163,933

Total consumer LHFI	$	325,469	$	307,435	$	236,008	$	115,968	$	126,791	$	380,617	$	416,430	$	1,908,718

Total LHFI	$	2,052,251	$	2,084,709	$	1,436,719	$	812,979	$	839,983	$	1,316,681	$	1,304,406	$	9,847,728

The table below presents LHFI by loan class and credit quality indicator at December 31, 2019 ($ in thousands):

	December 31, 2019
		Commercial LHFI
		Pass - Categories 1-6		Special Mention - Category 7		Substandard - Category 8		Doubtful - Category 9		Subtotal
Loans secured by real estate:
Construction, land development and other land		$	1,075,146	$	—	$	15,726	$	42	$	1,090,914
Secured by 1-4 family residential properties			116,592		45		6,355		41		123,033
Secured by nonfarm, nonresidential properties			2,430,761		—		44,001		328		2,475,090
Other real estate secured			721,238		—		2,547		—		723,785
Commercial and industrial loans			1,407,837		909		68,262		888		1,477,896
Consumer loans			—		—		—		—		—
State and other political subdivision loans			957,948		4,650		5,346		—		967,944
Other loans			469,095		3,445		16,926		30		489,496
Total		$	7,178,617	$	9,049	$	159,163	$	1,329	$	7,348,158

	Consumer LHFI
	Current		Past Due 30-89 Days		Past Due 90 Days or More		Nonaccrual		Subtotal		Total LHFI
Loans secured by real estate:
Construction, land development and other land	$	71,413	$	332	$	—	$	132	$	71,877	$	1,162,791
Secured by 1-4 family residential properties		1,710,930		5,922		211		15,817		1,732,880		1,855,913
Secured by nonfarm, nonresidential properties		155		—		—		—		155		2,475,245
Other real estate secured		695		—		—		—		695		724,480
Commercial and industrial loans		—		—		—		—		—		1,477,896
Consumer loans		172,649		2,588		393		108		175,738		175,738
State and other political subdivision loans		—		—		—		—		—		967,944
Other loans		6,125		—		—		—		6,125		495,621
Total	$	1,961,967	$	8,842	$	604	$	16,057	$	1,987,470	$	9,335,628

LHFS past due 90 days or more totaled $121.3 million and $41.6 million at September 30, 2020 and December 31, 2019, respectively. LHFS past due 90 days or more are serviced loans eligible for repurchase, which are fully guaranteed by the Government National Mortgage Association (GNMA). GNMA optional repurchase programs allow financial institutions to buy back individual delinquent mortgage loans that meet certain criteria from the securitized loan pool for which the institution provides servicing. At the servicer’s option and without GNMA’s prior authorization, the servicer may repurchase such a delinquent loan for an amount equal to 100% of the remaining principal balance of the loan. This buy-back option is considered a conditional option until the delinquency criteria are met, at which time the option becomes unconditional. When Trustmark is deemed to have regained effective control over these loans under the unconditional buy-back option, the loans can no longer be reported as sold and must be brought back onto the balance sheet as loans held for sale, regardless of whether Trustmark intends to exercise the buy-back option. These loans are reported as held for sale with the offsetting liability being reported as short-term borrowings.

Trustmark did not exercise its buy-back option on any delinquent loans serviced for GNMA during the first nine months of 2020 or 2019.

Trustmark’s ACL methodology for LHFI is based upon guidance within FASB ASC Subtopic 326-20 as well as regulatory guidance from its primary regulator. The ACL is a valuation account that is deducted from the loans’ amortized cost basis to present the net amount expected to be collected on the loans. Credit quality within the LHFI portfolio is continuously monitored by Management and is reflected within the ACL for loans. The ACL is an estimate of expected losses inherent within Trustmark’s existing LHFI portfolio. The ACL for LHFI is adjusted through the provision for credit losses (PCL) and reduced by the charge off of loan amounts, net of recoveries.

The loan loss estimation process involves procedures to appropriately consider the unique characteristics of Trustmark’s LHFI portfolio segments. These segments are further disaggregated into loan classes, the level at which credit risk is monitored. When computing allowance levels, credit loss assumptions are estimated using a model that categorizes loan pools based on loss history, delinquency status and other credit trends and risk characteristics, including current conditions and reasonable and supportable forecasts about the future. Evaluations of the portfolio and individual credits are inherently subjective, as they require estimates, assumptions and judgments as to the facts and circumstances of particular situations. Determining the appropriateness of the allowance is complex and requires judgement by Management about the effect of matters that are inherently uncertain. In future periods, evaluations of the overall LHFI portfolio, in light of the factors and forecasts then prevailing, may result in significant changes in the allowance and credit loss expense.

Trustmark estimates the ACL using relevant available information, from internal and external sources, relating to past events, current conditions and reasonable and supportable forecasts including the COVID-19 pandemic effects. Trustmark uses a third-party software application to calculate the quantitative portion of the ACL using a methodology and assumptions specific to each loan pool. The qualitative portion of the allowance is based on general economic conditions and other internal and external factors affecting Trustmark as a whole as well as specific LHFI. Factors considered include the following: lending policies and procedures, economic

conditions and concentrations of credit, nature and volume of the portfolio, performance trends, and external factors. The quantitative and qualitative portions of the allowance are added together to determine the total allowance for credit losses, which reflects Management’s expectations of future conditions based on reasonable and supportable forecasts.

The methodology for estimating the amount of expected credit losses reported in the ACL has two basic components: a collective, or pooled, component for estimated expected credit losses for pools of loans that share similar risk characteristics, and an asset-specific component involving individual loans that do not share risk characteristics with other loans and the measurement of expected credit losses for such individual loans. In estimating the allowance for credit losses for the collective component, loans are segregated into loan pools based on loan product types and similar risk characteristics.

The loans secured by real estate and other loans secured by real estate portfolio segments include loans for both commercial and residential properties. The underwriting process for these loans includes analysis of the financial position and strength of both the borrower and guarantor, experience with similar projects in the past, market demand and prospects for successful completion of the proposed project within the established budget and schedule, values of underlying collateral, availability of permanent financing, maximum loan-to-value ratios, minimum equity requirements, acceptable amortization periods and minimum debt service coverage requirements, based on property type. The borrower’s financial strength and capacity to repay their obligations remain the primary focus of underwriting. Financial strength is evaluated based upon analytical tools that consider historical and projected cash flows and performance in addition to analysis of the proposed project for income-producing properties. Additional support offered by guarantors is also considered. Ultimate repayment of these loans is sensitive to interest rate changes, general economic conditions, liquidity and availability of long-term financing.

The commercial and industrial LHFI portfolio segment includes commercial loans made to many types of businesses for various purposes, such as short-term working capital loans that are usually secured by accounts receivable and inventory, equipment and fixed asset purchases that are secured by those assets and term financing for those within Trustmark’s geographic markets. Trustmark’s credit underwriting process for commercial and industrial loans includes analysis of historical and projected cash flows and performance, evaluation of financial strength of both borrowers and guarantors as reflected in current and detailed financial information and evaluation of underlying collateral to support the credit.

The consumer LHFI portfolio segment is comprised of loans which are underwritten after evaluating a borrower’s capacity, credit and collateral. Several factors are considered when assessing a borrower’s capacity, including the borrower’s employment, income, current debt, assets and level of equity in the property. Credit is assessed using a credit report that provides credit scores and the borrower’s current and past information about their credit history. Property appraisals are obtained to assist in evaluating collateral. Loan-to-value and debt-to-income ratios, loan amount and lien position are also considered in assessing whether to originate a loan. These borrowers are particularly susceptible to downturns in economic trends such as conditions that negatively affect housing prices and demand and levels of unemployment.

The state and other political subdivision LHFI and the other commercial LHFI portfolio segments primarily consist of loans to non-depository financial institutions, such as mortgage companies, finance companies and other financial intermediaries, loans to state and political subdivisions, and loans to non-profit and charitable organizations. These loans are underwritten based on the specific nature or purpose of the loan and underlying collateral with special consideration given to the specific source of repayment for the loan.

The following table provides a description of each of Trustmark’s portfolio segments, loan classes, loan pools and the ACL methodology and loss drivers:

Portfolio Segment	Loan Class	Loan Pool	Methodology	Loss Drivers
Loans secured by real estate	Construction, land development and other land	1-4 family residential construction	DCF	Prime Rate, National GDP
		Lots and development	DCF	Prime Rate, Southern Unemployment
		Unimproved land	DCF	Prime Rate, Southern Unemployment
		All other consumer	DCF	Prime Rate, Southern Unemployment
	Other secured by 1-4 family residential properties	Consumer 1-4 family - 1st liens	DCF	Prime Rate, Southern Unemployment
		All other consumer	DCF	Prime Rate, Southern Unemployment
		Nonresidential owner-occupied	DCF	Southern Unemployment, National GDP
	Secured by nonfarm, nonresidential properties	Nonowner-occupied - hotel/motel	DCF	Southern Vacancy Rate, Southern Unemployment
		Nonowner-occupied - office	DCF	Southern Vacancy Rate, Southern Unemployment
		Nonowner-occupied- Retail	DCF	Southern Vacancy Rate, Southern Unemployment
		Nonowner-occupied - senior living/nursing homes	DCF	Southern Vacancy Rate, Southern Unemployment
		Nonowner-occupied - all other	DCF	Southern Vacancy Rate, Southern Unemployment
		Nonresidential owner-occupied	DCF	Southern Unemployment, National GDP
	Other real estate secured	Nonresidential nonowner -occupied - apartments	DCF	Southern Vacancy Rate, Southern Unemployment
		Nonresidential owner-occupied	DCF	Southern Unemployment, National GDP
		Nonowner-occupied - all other	DCF	Southern Vacancy Rate, Southern Unemployment
Other loans secured by real estate	Other construction	Other construction	WARM	Prime Rate, National Unemployment
	Secured by 1-4 family residential properties	Trustmark mortgage	WARM	Southern Unemployment
Commercial and industrial loans	Commercial and industrial loans	Commercial and industrial - non-working capital	DCF	Trustmark historical data
		Commercial and industrial - working capital	DCF	Trustmark historical data
		Credit cards	WARM	Trustmark call report data
Consumer loans	Consumer loans	Credit cards	WARM	Trustmark call report data
		Overdrafts	Loss Rate	Trustmark historical data
		All other consumer	DCF	Prime Rate, Southern Unemployment
State and other political subdivision loans	State and other political subdivision loans	Obligations of state and political subdivisions	DCF	Moody's Bond Default Study
Other commercial loans	Other commercial loans	Other loans	DCF	Prime Rate, Southern Unemployment
		Commercial and industrial - non-working capital	DCF	Trustmark historical data
		Commercial and industrial - working capital	DCF	Trustmark historical data

In general, Trustmark utilizes a DCF method to estimate the quantitative portion of the allowance for credit losses for loan pools. The DCF model consists of two key components, a loss driver analysis (LDA) and a cash flow analysis. For loan pools utilizing the DCF methodology, multiple assumptions are in place, depending on the loan pool. A reasonable and supportable forecast is utilized for each loan pool by developing a LDA for each loan class. The LDA uses charge off data from Federal Financial Institutions

Examination Council (FFIEC) reports to construct a periodic default rate (PDR). The PDR is decomposed into a probability of default (PD). Regressions are run using the data for various macroeconomic variables in order to determine which ones correlate to Trustmark’s losses. These variables are then incorporated into the application to calculate a quarterly PD using a third-party baseline forecast. In addition to the PD, a loss given default (LGD) is derived using a method referred to as Frye Jacobs. The Frye Jacobs method is a mathematical formula that traces the relationship between LGD and PD over time and projects the LGD based on the levels of PD forecasts. This model approach is applicable to all pools within the construction, land development and other land, other secured by 1-4 family residential properties, secured by nonfarm, nonresidential properties and other real estate secured loan classes as well as the all other consumer and other loans pools.

For the commercial and industrial loans related pools, Trustmark uses its own PD and LGD data, instead of the macroeconomic variables and the Frye Jacobs method described above, to calculate the PD and LGD as there were no defensible macroeconomic variables that correlated to Trustmark’s losses. Trustmark utilizes a third-party Bond Default Study to derive the PD and LGD for the obligations of state and political subdivisions pool. Due to the lack of losses within this pool, no defensible macroeconomic factors were identified to correlate.

The PD and LGD measures are used in conjunction with prepayment data as inputs into the DCF model to calculate the cash flows at the individual loan level. Contractual cash flows based on loan terms are adjusted for PD, LGD and prepayments to derive loss cash flows. These loss cash flows are discounted by the loan’s coupon rate to arrive at the discounted cash flow based quantitative loss. The prepayment studies are updated quarterly by a third-party for each applicable pool.

An alternate method of estimating the ACL is used for certain loan pools due to specific characteristics of these loans. For the non-DCF pools, specifically, those using the weighted average remaining maturity (WARM) method, the remaining life is incorporated into the ACL quantitative calculation.

Trustmark determined that reasonable and supportable forecasts could be made for a twelve-month period for all of its loan pools. To the extent the lives of the loans in the LHFI portfolio extend beyond this forecast period, Trustmark uses a reversion period of four quarters and reverts to the historical mean on a straight-line basis over the remaining life of the loans. The econometric models in production today reflect segment or pool level sensitivities of probability of default (PD) to changes in macroeconomic variables. By measuring the relationship between defaults and changes in the economy, the quantitative reserve incorporates reasonable and supportable forecasts of future conditions that will affect the value of its assets, as required by FASB ASC Topic 326. Under stable forecasts, these linear regressions will reasonably predict a pool’s PD. However, due to the COVID-19 pandemic, the macroeconomic variables used for reasonable and supportable forecasting have changed rapidly. At the current levels, it is not clear that the models currently in production will produce reasonably representative results since the models were originally estimated using data beginning in 2004 through 2017. During this period, a traditional albeit severe economic recession occurred. Thus, econometric models are sensitive to similar future levels of PD.

During the second quarter of 2020, Trustmark revised its ACL methodology in order to prevent the econometric models from extrapolating beyond reasonable boundaries of their input variables. Trustmark chose to establish an upper and lower limit process when applying the periodic forecasts. In this way, Management will not rely upon unobserved and untested relationships in the setting of the quantitative reserve. This approach applies to all input variables, including: Southern Unemployment, National Unemployment, National GDP, Southern Vacancy Rate and the Prime Rate.

The upper and lower limits are based on the distribution of the macroeconomic variable by selecting extreme percentiles at the upper and lower limits of the distribution, the 1st and 99th, respectively. These upper and lower limits are then used to calculate the PD for the forecast time period in which the forecasted values are outside of the upper and lower limit range. For the current period, the forecast related to the macroeconomic variables used in the quantitative modeling process were positively impacted due to the updated forecast effects related to COVID-19, causing an overall decrease in quantitative reserve levels.

While all these factors are incorporated into the overall methodology, only three are currently considered active: economic conditions and concentrations of credit, performance trends and external factors.

Two of Trustmark’s largest loan classes are the loans secured by nonfarm, nonresidential properties and the loans secured by other real estate. Trustmark elected to create a qualitative factor specifically for these loan classes which addresses changes in the economic conditions of metropolitan areas and applies additional pool level reserves. This qualitative factor is based on third-party market data and forecast trends and is updated quarterly as information is available, by market and by loan pool.

For the performance trends factor, Trustmark uses migration analyses to allocate additional ACL to non-pass/delinquent loans within each pool. In this way, Management believes the ACL will directly reflect changes in risk, based on the performance of the loans within a pool, whether declining or improving.

The external factors qualitative factor is Management’s best judgement on the loan or pool level impact of all factors that affect the portfolio but cannot be accounted for using any other part of the ACL methodology, e.g., natural disasters, changes in legislation, impacts due to technology and pandemics. During the third quarter of 2020, Trustmark activated the External Factor – Pandemic to ensure reserve adequacy for collectively evaluated loans most likely to be impacted by the unique economic and behavioral conditions created by the COVID-19 pandemic. Additional qualitative reserves are derived based on two principles. The first is the disconnect of economic factors to Trustmark’s modeled probability of default (the econometric models underpinning the quantitative pooled reserves). During the pandemic, extraordinary measures by the federal government were made available to consumers and businesses, including COVID-19 loan payment concessions, direct transfer payments to households, tax deferrals, and reduced interest rates, among others. These government interventions may have extended the lag between economic conditions and default, relative to what was captured in the model development data. Because Trustmark’s econometric PD models rely on the observed relationship from the economic downturn from 2007 to 2009 in both timing and severity, Management does not expect the models to reflect these current conditions. For example, while the models would predict contemporaneous unemployment peaks and loan defaults, this cannot occur when borrowers can request payment deferrals. Thus, for the affected population, economic conditions are not fully considered as a part of Trustmark’s quantitative reserve. The second principle is the change in risk that is identified by rating changes. As a part of Trustmark’s credit review process, loans in the affected population have been given more frequent screening to ensure accurate ratings are maintained through this dynamic period. Trustmark’s quantitative reserve does not directly address changes in ratings, thus a migration qualitative factor was designed to work in concert with the quantitative reserve. In a downturn, the qualitative factor is inactive for most pools because changes in ratings are congruent with changes in macroeconomic conditions, which directly influence the PD models in the quantitative reserve. As discussed above, the disconnect of economic factors means that changes in rating caused by deteriorating and weak economic conditions as a result of the pandemic are not being captured in the quantitative reserve. To dimension the additional reserve, Management uses the sensitivity of the quantitative commercial loan reserve to changes in macroeconomic conditions. In addition, to account for the known changes in risk, a weighted-average of the commercial loan portfolio loss rate, derived from the performance trends qualitative factor, is used to dimension additional reserves for downgraded credits.

The ACL for individual loans that do not share risk characteristics with other loans is measured as the difference between the discounted value of expected future cash flows, based in the effective interest rate at origination, and the amortized cost basis of the loan, or the net realizable value. The ACL is the difference between the loan’s net realizable value and its amortized cost basis (net of previous charge-offs and deferred loan fees and costs), except for collateral-dependent loans. A loan is collateral dependent when the borrower is experiencing financial difficulty and repayment of the loan is expected to be provided substantially through the sale of the collateral. The expected credit loss for collateral-dependent loans is measured as the difference between the amortized cost basis of the loan and the fair value of the collateral, adjusted for the estimated cost to sell. Fair value estimates for collateral-dependent loans are derived from appraised values based on the current market value or the ‘as is’ value of the collateral, normally from recently received and reviewed appraisals. Current appraisals are ordered on an annual basis based on the inspection date or more often if market conditions necessitate. Appraisals are obtained from state-certified appraisers and are based on certain assumptions, which may include construction or development status and the highest and best use of the property. These appraisals are reviewed by Trustmark’s Appraisal Review Department to ensure they are acceptable, and values are adjusted down for costs associated with asset disposal. If the calculated expected credit loss is determined to be permanent or not recoverable, the amount of the expected credit loss is charged off.

LHFI are charged off against the ACL, with any subsequent recoveries credited back to the ACL account. Expected recoveries may not exceed the aggregate of amounts previously charged off and expected to be charged off. Trustmark’s loan policy dictates the guidelines to be followed in determining when a loan is charged off. Commercial purpose LHFI are charged off when a determination is made that the loan is uncollectible and continuance as a bankable asset is not warranted. Consumer LHFI secured by 1-4 family residential real estate are generally charged off or written down to the fair value of the collateral less cost to sell at no later than 180 days of delinquency. Non-real estate consumer purpose LHFI, including both secured and unsecured loans, are generally charged off by 120 days of delinquency. Consumer revolving lines of credit and credit card debt are generally charged off on or prior to 180 days of delinquency.

The following table disaggregates the ACL and the amortized cost basis of the loans by the measurement methodology used at September 30, 2020 ($ in thousands):

	September 30, 2020
	ACL						LHFI
	Individually Evaluated for Credit Loss		Collectively Evaluated for Credit Loss		Total ACL		Individually Evaluated for Credit Loss		Collectively Evaluated for Credit Loss		Total LHFI
Loans secured by real estate:
Construction, land development and other land	$	—	$	10,905	$	10,905	$	1,024		493,495	$	494,519
Other secured by 1-4 family residential properties		—		11,358		11,358		474		546,674		547,148
Secured by nonfarm, nonresidential properties		—		43,762		43,762		13,529		2,694,098		2,707,627
Other real estate secured		—		7,172		7,172		61		887,731		887,792
Other loans secured by real estate:
Other construction		—		10,209		10,209		—		891,428		891,428
Secured by 1-4 family residential properties		—		8,734		8,734		1,453		1,226,799		1,228,252
Commercial and industrial loans		578		12,580		13,158		14,882		1,383,586		1,398,468
Consumer loans		—		6,040		6,040		—		163,933		163,933
State and other political subdivision loans		1,749		1,209		2,958		4,019		931,330		935,349
Other commercial loans		2,295		5,419		7,714		5,725		587,487		593,212
Total	$	4,622	$	117,388	$	122,010	$	41,167	$	9,806,561	$	9,847,728

The following table disaggregates the allowance for loan losses and LHFI balances by the impairment evaluation methodology used at December 31, 2019 ($ in thousands):

	December 31, 2019
	Allowance for Loan Losses						LHFI
	Individually Evaluated		Collectively Evaluated		Total		Individually Evaluated		Collectively Evaluated		Total
Loans secured by real estate:
Construction, land development and other land	$	—	$	8,260	$	8,260	$	626	$	1,162,165	$	1,162,791
Secured by 1-4 family residential properties		35		8,897		8,932		5,464		1,850,449		1,855,913
Secured by nonfarm, nonresidential properties		2,355		23,803		26,158		6,717		2,468,528		2,475,245
Other real estate secured		—		4,024		4,024		68		724,412		724,480
Commercial and industrial loans		707		25,285		25,992		23,458		1,454,438		1,477,896
Consumer loans		—		3,379		3,379		21		175,717		175,738
State and other political subdivision loans		1,809		420		2,229		4,079		963,865		967,944
Other loans		553		4,750		5,303		784		494,837		495,621
Total	$	5,459	$	78,818	$	84,277	$	41,217	$	9,294,411	$	9,335,628

The following table details changes in the ACL by loan class for the period presented ($ in thousands):

	Three Months Ended September 30, 2020
	Balance at Beginning of Period		Charge-offs			Recoveries		PCL			Balance at End of Period
Loans secured by real estate:
Construction, land development and other land	$	11,940	$	—		$	443	$	(1,478	)	$	10,905
Other secured by 1-4 family residential properties		12,716		(18	)		75		(1,415	)		11,358
Secured by nonfarm, nonresidential properties		36,417		(115	)		18		7,442			43,762
Other real estate secured		7,600		—			42		(470	)		7,172
Other loans secured by real estate:
Other construction		10,803		—			30		(624	)		10,209
Secured by 1-4 family residential properties		10,899		—			18		(2,183	)		8,734
Commercial and industrial loans		12,550		(71	)		447		232			13,158
Consumer loans		6,397		(384	)		375		(348	)		6,040
State and other political subdivision loans		3,414		—			—		(456	)		2,958
Other commercial loans		6,452		(675	)		877		1,060			7,714
Total	$	119,188	$	(1,263	)	$	2,325	$	1,760		$	122,010

The decreases in the PCL for loans and other loans secured by real estate during the three months ended September 30, 2020 were primarily due to improvements in the macroeconomic forecasting variables used in the ACL modeling, such as National and Southern Unemployment, National GDP and Prime Rate.

During the third quarter of 2020, Trustmark conducted a review of significantly impacted borrowers that received one or more payment concessions and other borrowers in industries significantly impacted by COVID-19. The increases in the PCL for loans secured by nonfarm, nonresidential properties and other commercial loans during the three months ended September 30, 2020 were primarily due to downgrades that resulted from the in-depth portfolio review for those loans affected by the COVID-19 pandemic.

The following table details changes in the ACL by loan class for the period presented ($ in thousands):

	Nine Months Ended September 30, 2020
	Balance at Beginning of Period		FASB ASU 2016-13 Adoption Adjustment			Charge-offs			Recoveries		PCL			Balance at End of Period
Loans secured by real estate:
Construction, land development and other land	$	6,371	$	(188	)	$	(7	)	$	629	$	4,100		$	10,905
Other secured by 1-4 family residential properties		5,888		4,188			(118	)		221		1,179			11,358
Secured by nonfarm, nonresidential properties		26,158		(8,179	)		(2,563	)		524		27,822			43,762
Other real estate secured		4,024		(765	)		(8	)		60		3,861			7,172
Other loans secured by real estate:
Other construction		1,889		3,202			—			70		5,048			10,209
Secured by 1-4 family residential properties		3,044		2,891			(19	)		124		2,694			8,734
Commercial and industrial loans		25,992		(8,964	)		(1,350	)		1,180		(3,700	)		13,158
Consumer loans		3,379		2,059			(1,745	)		1,316		1,031			6,040
State and other political subdivision loans		2,229		2,455			—			—		(1,726	)		2,958
Other commercial loans		5,303		2,084			(2,868	)		2,978		217			7,714
Total	$	84,277	$	(1,217	)	$	(8,678	)	$	7,102	$	40,526		$	122,010

The increases in the PCL for loans and other loans secured by real estate and consumer loans during the nine months ended September 30, 2020 were primarily due to the negative impact of COVID-19 on the macroeconomic forecasting variables used in the ACL modeling, such as National and Southern Unemployment, National GDP, Prime Rate and South Vacancy Rate.

The PCL for the commercial and industrial loan portfolio decreased $3.7 million during the nine months ended September 30, 2020 primarily due to loans that had been specifically reserved for being charged down, upgrades on loans from substandard to pass, paydowns as well as a slight decrease in the calculated PD and LGD, which uses Trustmark’s historical data. The decrease in the PCL

for state and other political subdivision loans of $1.7 million was primarily due to a decrease in reserves based on routine updates to the qualitative portion of the allowance calculation.

The following table details changes in the allowance for loan losses, LHFI by loan class for the period presented ($ in thousands):

	2019
	Balance January 1,		Charge-offs			Recoveries		Provision for Loan Losses			Balance September 30,
Loans secured by real estate:
Construction, land development and other land	$	7,390	$	(35	)	$	807	$	(461	)	$	7,701
Secured by 1-4 family residential properties		8,641		(310	)		530		(417	)		8,444
Secured by nonfarm, nonresidential properties		22,376		(261	)		285		2,802			25,202
Other real estate secured		3,450		—			22		82			3,554
Commercial and industrial loans		27,359		(3,090	)		980		3,618			28,867
Consumer loans		2,890		(1,712	)		1,432		702			3,312
State and other political subdivision loans		990		—			—		(518	)		472
Other loans		6,194		(4,454	)		2,606		1,328			5,674
Total	$	79,290	$	(9,862	)	$	6,662	$	7,136		$	83,226