
	September 30, 2020		Percentage of Total		December 31, 2019		Percentage of Total
Commercial and industrial	$	4,820,018	26.4	%	$	5,187,466	35.2	%
Energy:
Production	1,108,753		6.1		1,348,900		9.2
Service	163,197		0.9		192,996		1.3
Other	91,418		0.4		110,986		0.8
Total energy	1,363,368		7.4		1,652,882		11.2
Paycheck Protection Program	3,226,980		17.7		—		—
Commercial real estate:
Commercial mortgages	5,412,731		29.7		4,594,113		31.1
Construction	1,274,080		7.0		1,312,659		8.9
Land	320,407		1.8		289,467		2.0
Total commercial real estate	7,007,218		38.5		6,196,239		42.0
Consumer real estate:
Home equity loans	340,226		1.9		375,596		2.6
Home equity lines of credit	434,171		2.4		354,671		2.4
Other	533,356		2.9		464,146		3.1
Total consumer real estate	1,307,753		7.2		1,194,413		8.1
Total real estate	8,314,971		45.7		7,390,652		50.1
Consumer and other	498,540		2.8		519,332		3.5
Total loans	$	18,223,877	100.0	%	$	14,750,332	100.0	%

Concentrations of Credit. Most of our lending activity occurs within the State of Texas, including the four largest metropolitan areas of Austin, Dallas/Ft. Worth, Houston and San Antonio, as well as other markets. The majority of our loan portfolio consists of commercial and industrial and commercial real estate loans. As of September 30, 2020, there were no concentrations of loans related to any single industry in excess of 10% of total loans. The largest industry concentration was related to the energy industry, which totaled 7.4% of total loans, or 9.1% excluding PPP loans. Unfunded commitments to extend credit and standby letters of credit issued to customers in the energy industry totaled $910.3 million and $62.6 million, respectively, as of September 30, 2020.

COVID-19 Loan Deferments. Certain borrowers are currently unable to meet their contractual payment obligations because of the adverse effects of COVID-19. To help mitigate these effects, loan customers may apply for a deferral of payments, or portions thereof, for up to 90 days. After 90 days, customers may apply for an additional deferral, and a small proportion of our customers have requested such an additional deferral. In the absence of other intervening factors, such short-term modifications made on a good faith basis are not categorized as troubled debt restructurings, nor are loans granted payment deferrals related to COVID-19 reported as past due or placed on non-accrual status (provided the loans were not past due or on non-accrual status prior to the deferral). At September 30, 2020, there were approximately 300 loans in COVID-19 related deferment with an aggregate outstanding balance of approximately $157.2 million.

Non-Accrual and Past Due Loans. Loans are considered past due if the required principal and interest payments have not been received as of the date such payments were due. Loans are placed on non-accrual status when, in management’s opinion, the borrower may be unable to meet payment obligations as they become due, as well as when required by regulatory provisions. In determining whether or not a borrower may be unable to meet payment obligations for each class of loans, we consider the borrower’s debt service capacity through the analysis of current financial information, if available, and/or current information with regards to our collateral position. Regulatory provisions would typically require the placement of a loan on non-accrual status if (i) principal or interest has been in default for a period of 90 days or more unless the loan is both well secured and in the process of collection or (ii) full payment of principal and interest is not expected. Loans may be placed on non-accrual status regardless of whether or not such loans are considered past due. When interest accrual is discontinued, all unpaid accrued interest is reversed against income. Interest income on non-accrual loans is recognized only to the extent that cash payments are received in excess of principal due. A loan may be returned to accrual status when all the principal and interest amounts contractually due are brought current and future principal and interest amounts contractually due are reasonably assured, which is typically evidenced by a sustained period (at least six months) of repayment performance by the borrower.


	September 30, 2020				December 31, 2019
	Total Non-Accrual		Non-Accrual with No Credit Loss Allowance		Total Non-Accrual		Non-Accrual with No Credit Loss Allowance
Commercial and industrial	$	21,566	$	6,583	$	26,038	$	13,266
Energy	54,041		33,324		65,761		3,281
Paycheck Protection Program	—		—		—		—
Commercial real estate:
Buildings, land and other	13,525		6,754		8,912		6,558
Construction	680		680		665		665
Consumer real estate	1,748		1,748		922		922
Consumer and other	18		—		5		—
Total	$	91,578	$	49,089	$	102,303	$	24,692


	2020		2019		2018		2017		2016		Prior		Revolving Loans		Revolving Loans Converted to Term		Total
Commercial and industrial	$	9,317	$	3,743	$	1,886	$	1,263	$	110	$	51	$	1,879	$	3,317	$	21,566
Energy	24,149		7,021		2,204		—		—		—		15,054		5,613		54,041
Paycheck Protection Program	—		—		—		—		—		—		—		—		—
Commercial real estate:
Buildings, land and other	1,426		6,014		169		1,372		840		3,590		114		—		13,525
Construction	680		—		—		—		—		—		—		—		680
Consumer real estate	—		—		421		211		350		382		261		123		1,748
Consumer and other	—		—		—		—		—		—		18		—		18
Total	$	35,572	$	16,778	$	4,680	$	2,846	$	1,300	$	4,023	$	17,326	$	9,053	$	91,578

In the table above, loans reported as 2020 originations were, for the most part, first originated in various years prior to 2020 but were renewed in the current year. Had non-accrual loans performed in accordance with their original contract terms, we would have recognized additional interest income, net of tax, of approximately $670 thousand and $2.3 million for the three and nine months ended September 30, 2020, and approximately $937 thousand and $3.0 million for the three and nine months ended September 30, 2019.


	Loans 30-89 Days Past Due		Loans 90 or More Days Past Due		Total Past Due Loans		Current Loans		Total Loans		Accruing Loans 90 or More Days Past Due
Commercial and industrial	$	36,067	$	25,115	$	61,182	$	4,758,836	$	4,820,018	$	19,392
Energy	31,071		9,588		40,659		1,322,709		1,363,368		559
Paycheck Protection Program	—		—		—		3,226,980		3,226,980		—
Commercial real estate:
Buildings, land and other	25,113		19,571		44,684		5,688,454		5,733,138		11,676
Construction	—		—		—		1,274,080		1,274,080		—
Consumer real estate	11,054		4,761		15,815		1,291,938		1,307,753		3,406
Consumer and other	6,629		673		7,302		491,238		498,540		673
Total	$	109,934	$	59,708	$	169,642	$	18,054,235	$	18,223,877	$	35,706


	Nine Months Ended September 30, 2020				Nine Months Ended September 30, 2019
	Balance at Restructure		Balance at Period-End		Balance at Restructure		Balance at Period-End
Commercial and industrial	$	3,660	$	1,415	$	3,845	$	2,188

Commercial real estate:
Buildings, land and other	6,606		6,585		9,456		9,494
Construction	1,192		1,181		—		—
Consumer real estate	—		—		124		123
Consumer and other	1,104		104		—		—
	$	12,562	$	9,285	$	13,425	$	11,805

Loan modifications are typically related to extending amortization periods, converting loans to interest only for a limited period of time, deferral of interest payments, waiver of certain covenants, consolidating notes and/or reducing collateral or interest rates. The modifications during the reported periods did not significantly impact our determination of the allowance for credit losses on loans.

Credit Quality Indicators. As part of the on-going monitoring of the credit quality of our loan portfolio, management tracks certain credit quality indicators including trends related to (i) the weighted-average risk grade of commercial loans, (ii) the level of classified commercial loans, (iii) the delinquency status of consumer loans (iv) non-performing loans (see details above) and (vi) the general economic conditions in the State of Texas.

We utilize a risk grading matrix to assign a risk grade to each of our commercial loans. Loans are graded on a scale of 1 to 14. A description of the general characteristics of the 14 risk grades is set forth in our 2019 Form 10-K. We monitor portfolio credit quality by the weighted-average risk grade of each class of commercial loan. Individual relationship managers, under the oversight of credit administration, review updated financial information for all pass grade loans to reassess the risk grade on at least an annual basis. When a loan has a risk grade of 9, it is still considered a pass grade loan; however, it is considered to be on management’s “watch list,” where a significant risk-modifying action is anticipated in the near term. When a loan has a risk grade of 10 or higher, a special assets officer monitors the loan on an on-going basis.


	2020		2019		2018		2017		2016		Prior		Revolving Loans		Revolving Loans Converted to Term		Total		W/A Risk Grade
Commercial and industrial
Risk grades 1-8	$	1,101,917	$	625,692	$	340,799	$	251,180	$	120,777	$	129,757	$	1,788,761	$	43,468	$	4,402,351	6.12
Risk grade 9	69,846		27,843		36,271		21,352		9,910		1,087		77,802		4,462		248,573		9.00
Risk grade 10	4,394		7,722		13,867		6,875		812		824		60,028		4,347		98,869		10.00
Risk grade 11	3,675		4,507		6,720		4,422		1,896		167		17,909		9,363		48,659		11.00
Risk grade 12	6,917		2,791		1,544		708		110		51		1,350		3,022		16,493		12.00
Risk grade 13	2,400		952		342		555		—		—		529		295		5,073		13.00
	$	1,189,149	$	669,507	$	399,543	$	285,092	$	133,505	$	131,886	$	1,946,379	$	64,957	$	4,820,018	6.42
W/A risk grade	6.12		6.81		7.20		6.32		6.37		5.87		6.32		7.92		6.42
Energy
Risk grades 1-8	$	454,901	$	25,589	$	10,956	$	6,732	$	1,481	$	4,709	$	407,478	$	24,936	$	936,782	6.18
Risk grade 9	108,859		5,542		3,190		—		—		—		90,292		14,974		222,857		9.00
Risk grade 10	486		4,113		1,461		—		907		—		7,837		2,105		16,909		10.00
Risk grade 11	78,832		15,724		3,280		1,210		—		1,066		32,398		269		132,779		11.00
Risk grade 12	24,149		4,829		714		—		—		—		10,629		5,613		45,934		12.00
Risk grade 13	—		2,192		1,490		—		—		—		4,425		—		8,107		13.00
	$	667,227	$	57,989	$	21,091	$	7,942	$	2,388	$	5,775	$	553,059	$	47,897	$	1,363,368	7.39
W/A risk grade	7.43		9.19		8.78		7.83		7.93		8.06		6.98		8.76		7.39


	2020		2019		2018		2017		2016		Prior		Revolving Loans		Revolving Loans Converted to Term		Total		W/A Risk Grade
Commercial real estate:
Buildings, land, other
Risk grades 1-8	$	1,167,872	$	1,010,418	$	829,640	$	660,047	$	430,971	$	752,461	$	62,672	$	53,480	$	4,967,561	6.98
Risk grade 9	26,056		93,113		67,546		66,365		45,477		92,060		4,413		1,042		396,072		9.00
Risk grade 10	1,362		26,080		29,357		39,853		65,702		42,528		4,147		2,876		211,905		10.00
Risk grade 11	7,416		7,108		13,425		42,660		10,487		59,730		3,038		211		144,075		11.00
Risk grade 12	1,226		5,214		169		1,372		840		3,340		51		—		12,212		12.00
Risk grade 13	200		800		—		—		—		250		63		—		1,313		13.00
	$	1,204,132	$	1,142,733	$	940,137	$	810,297	$	553,477	$	950,369	$	74,384	$	57,609	$	5,733,138	7.34
W/A risk grade	7.09		7.34		7.46		7.48		7.73		7.26		7.41		6.84		7.34
Construction
Risk grades 1-8	$	272,025	$	480,166	$	224,389	$	753	$	1,164	$	1,786	$	160,654	$	—	$	1,140,937	7.01
Risk grade 9	31,402		8,370		—		39,688		—		—		14,608		—		94,068		9.00
Risk grade 10	6,133		—		27,479		—		—		—		3,838		—		37,450		10.00
Risk grade 11	—		—		—		—		—		945		—		—		945		11.00
Risk grade 12	680		—		—		—		—		—		—		—		680		12.00
Risk grade 13	—		—		—		—		—		—		—		—		—		13.00
	$	310,240	$	488,536	$	251,868	$	40,441	$	1,164	$	2,731	$	179,100	$	—	$	1,274,080	7.25
W/A risk grade	6.87		7.18		7.74		8.98		7.29		8.02		7.00		—		7.25
Total commercial real estate	$	1,514,372	$	1,631,269	$	1,192,005	$	850,738	$	554,641	$	953,100	$	253,484	$	57,609	$	7,007,218	7.33
W/A risk grade	7.04		7.29		7.52		7.55		7.73		7.26		7.12		6.84		7.33


	Commercial and Industrial			Energy			Commercial Real Estate - Buildings, Land and Other			Commercial Real Estate - Construction			Total Commercial Real Estate
	W/A Risk Grade	Loans		W/A Risk Grade	Loans		W/A Risk Grade	Loans		W/A Risk Grade	Loans		W/A Risk Grade	Loans
Risk grades 1-8	6.17	$	4,788,857	5.90	$	1,488,301	6.78	$	4,523,271	7.25	$	1,274,098	6.88	$	5,797,369
Risk grade 9	9.00	247,212		9.00	32,163		9.00	163,714		9.00	21,509		9.00	185,223
Risk grade 10	10.00	71,472		10.00	51,898		10.00	103,626		10.00	15,243		10.00	118,869
Risk grade 11	11.00	53,887		11.00	14,760		11.00	84,057		11.00	1,144		11.00	85,201
Risk grade 12	12.00	18,189		12.00	45,514		12.00	8,529		12.00	665		12.00	9,194
Risk grade 13	13.00	7,849		13.00	20,246		13.00	383		13.00	—		13.00	383
Total	6.44	$	5,187,466	6.39	$	1,652,882	7.01	$	4,883,580	7.31	$	1,312,659	7.07	$	6,196,239


	2020		2019		2018		2017		2016		Prior		Revolving Loans		Revolving Loans Converted to Term		Total
Consumer real estate:
Past due 30-89 days	$	787	$	850	$	1,434	$	858	$	540	$	4,558	$	1,535	$	492	$	11,054
Past due 90 or more days	—		139		894		542		351		2,028		600		207		4,761
Total past due	787		989		2,328		1,400		891		6,586		2,135		699		15,815
Current loans	253,392		188,770		106,947		93,143		75,216		142,983		414,331		17,156		1,291,938
Total	$	254,179	$	189,759	$	109,275	$	94,543	$	76,107	$	149,569	$	416,466	$	17,855	$	1,307,753
Consumer and other:
Past due 30-89 days	$	1,328	$	326	$	288	$	75	$	2	$	1	$	4,313	$	296	$	6,629
Past due 90 or more days	69		189		3		—		—		1		372		39		673
Total past due	1,397		515		291		75		2		2		4,685		335		7,302
Current loans	38,114		32,214		7,985		3,131		2,070		950		376,762		30,012		491,238
Total	$	39,511	$	32,729	$	8,276	$	3,206	$	2,072	$	952	$	381,447	$	30,347	$	498,540

In assessing the general economic conditions in the State of Texas, management monitors and tracks the Texas Leading Index (“TLI”), which is produced by the Federal Reserve Bank of Dallas. The TLI, the components of which are more fully described in our 2019 Form 10-K, totaled 113.1 at September 30, 2020 and 127.9 at December 31, 2019. A lower TLI value implies less favorable economic conditions.

Allowance For Credit Losses - Loans. The allowance for credit losses on loans is a contra-asset valuation account, calculated in accordance with ASC 326, that is deducted from the amortized cost basis of loans to present the net amount expected to be collected. The amount of the allowance represents management's best estimate of current expected credit losses on loans considering available information, from internal and external sources, relevant to assessing collectibility over the loans' contractual terms, adjusted for expected prepayments when appropriate. The contractual term excludes expected extensions, renewals and modifications unless (i) management has a reasonable expectation that a trouble debt restructuring will be executed with an individual borrower or (ii) such extension or renewal options are not unconditionally cancellable by us and, in such cases, the borrower is likely to meet applicable conditions and likely to request extension or renewal. Relevant available information includes historical credit loss experience, current conditions and reasonable and supportable forecasts. While historical credit loss experience provides the basis for the estimation of expected credit losses, adjustments to historical loss information may be made for differences in current portfolio-specific risk characteristics, environmental conditions or other relevant factors. The allowance for credit losses is measured on a collective basis for portfolios of loans when similar risk characteristics exist. Loans that do not share risk characteristics are evaluated for expected credit losses on an individual basis and excluded from the collective evaluation. Expected credit losses for collateral dependent loans, including loans where the borrower is experiencing financial difficulty but foreclosure is not probable, are based on the fair value of the collateral at the reporting date, adjusted for selling costs as appropriate.

Credit loss expense related to loans reflects the totality of actions taken on all loans for a particular period including any necessary increases or decreases in the allowance related to changes in credit loss expectations associated with specific loans or pools of loans. Portions of the allowance may be allocated for specific credits; however, the entire allowance is available for any credit that, in management’s judgment, should be charged off. While management utilizes its best judgment and information available, the ultimate appropriateness of the allowance is dependent upon a variety of factors beyond our control, including the performance of our loan portfolio, the economy, changes in interest rates and the view of the regulatory authorities toward loan classifications.

In calculating the allowance for credit losses, most loans are segmented into pools based upon similar characteristics and risk profiles. Common characteristics and risk profiles include the type/purpose of loan, underlying collateral, geographical similarity and historical/expected credit loss patterns. In developing these loan pools for the purposes of modeling expected credit losses, we also analyzed the degree of correlation in how loans within each portfolio respond when subjected to varying economic conditions and scenarios as well as other portfolio stress factors. For modeling purposes, our loan pools include (i) commercial and industrial and energy - non-revolving, (ii) commercial and industrial and energy - revolving, (iii) commercial real estate - owner occupied, (iv) commercial real estate - non-owner occupied, (v) commercial real estate - construction/land development, (vi) consumer real estate and (vii) consumer and other. We periodically reassess each pool to ensure the loans within the pool continue to share similar characteristics and risk profiles and to determine whether further segmentation is necessary.

For each loan pool, we measure expected credit losses over the life of each loan utilizing a combination of models which measure (i) probability of default (“PD”), which is the likelihood that loan will stop performing/default, (ii) probability of attrition (“PA”), which is the likelihood that a loan will pay-off prior to maturity, (iii) loss given default (“LGD”), which is the expected loss rate for loans in default and (iv) exposure at default (“EAD”), which is the estimated outstanding principal balance of the loans upon default, including the expected funding of unfunded commitments outstanding as of the measurement date. For certain commercial loan portfolios, the PD is calculated using a transition matrix to determine the likelihood of a customer’s risk grade migrating from one specified range of risk grades to a different specified range. Expected credit losses are calculated as the product of PD (adjusted for attrition), LGD and EAD. This methodology builds on default probabilities already incorporated into our risk grading process by utilizing pool-specific historical loss rates to calculate expected credit losses. These pool-specific historical loss rates may be adjusted for current macroeconomic assumptions, as further discussed below, and other factors such as differences in underwriting standards, portfolio mix, or when historical asset terms do not reflect the contractual terms of the financial assets being evaluated as of the measurement date. Each time we measure expected credit losses, we assess the relevancy of historical loss information and consider any necessary adjustments to address any differences in asset-specific characteristics. Due to their short-term nature, expected credit losses for overdrafts included in consumer and other loans are based solely upon a weighting of recent historical charge-offs over a period of three years.

The measurement of expected credit losses is impacted by loan/borrower attributes and certain macroeconomic variables. Significant loan/borrower attributes utilized in our modeling processes include, among other things, (i) origination date, (ii) maturity date, (iii) payment type, (iv) collateral type and amount, (v) current risk grade, (vi) current unpaid balance and commitment utilization rate, (vii) payment status/delinquency history and (viii) expected recoveries of previously charged-off amounts. Significant macroeconomic variables utilized in our modeling processes include, among other things, (i) Gross State Product for Texas and U.S. Gross Domestic Product, (ii) selected market interest rates including U.S. Treasury rates, bank prime rate, 30-year fixed mortgage rate, BBB corporate bond rate, among others, (iii) unemployment rates, (iv) commercial and residential property prices in Texas and the U.S. as a whole, (v) West Texas Intermediate crude oil price and (vi) total stock market index.

PD and PA were estimated by analyzing internally-sourced data related to historical performance of each loan pool over a complete economic cycle. PD and PA are adjusted to reflect the current impact of certain macroeconomic variables as well as their expected changes over a reasonable and supportable forecast period. We have determined that we are reasonably able to forecast the macroeconomic variables used in our modeling processes with an acceptable degree of confidence for a total of two years with the last twelve months of the forecast period encompassing a reversion process whereby the forecasted macroeconomic variables are reverted to their historical mean utilizing a rational, systematic basis. The macroeconomic variables utilized as inputs in our modeling processes were subjected to a variety of analysis procedures and were selected primarily based on statistical relevancy and correlation to our historical credit losses. By reverting these modeling inputs to their historical mean and considering loan/borrower specific attributes, our models will yield a measurement of expected credit losses that reflects our average historical loss rates for periods subsequent to the twelve-month reversion period. The LGD is based on historical recovery averages for each loan pool, adjusted to reflect the current impact of certain macroeconomic variables as well as their expected changes over a two-year forecast period, with the final twelve months of the forecast period encompassing a reversion process, which management considers to be both reasonable and supportable. This same forecast/reversion period is used for all macroeconomic variables used in all of our models. EAD is estimated using a linear regression model that estimates the average percentage of the loan balance that remains at the time of a default event.

Management qualitatively adjusts model results for risk factors that are not considered within our modeling processes but are nonetheless relevant in assessing the expected credit losses within our loan pools. These qualitative factor (“Q-Factor”) adjustments may increase or decrease management's estimate of expected credit losses by a calculated percentage or amount based upon the estimated level of risk. The various risks that may be considered in making Q-Factor adjustments include, among other things, the impact of (i) changes in lending policies and procedures, including changes in underwriting standards and practices for collections, write-offs, and recoveries, (ii) actual and expected changes in international, national, regional, and local economic and business conditions and developments that affect the collectibility of the loan pools, (iii) changes in the nature and volume of the loan pools and in the terms of the underlying loans, (iv) changes in the experience, ability, and depth of our lending management and staff, (v) changes in volume and severity of past due financial assets, the volume of non-accrual assets, and the volume and severity of adversely classified or graded assets, (vi) changes in the quality of our credit review function, (vii) changes in the value of the underlying collateral for loans that are non-collateral dependent, (viii) the existence, growth, and effect of any concentrations of credit and (ix) other factors such as the regulatory, legal and technological environments; competition; and events such as natural disasters or health pandemics.

In some cases, management may determine that an individual loan exhibits unique risk characteristics which differentiate the loan from other loans within our loan pools. In such cases, the loans are evaluated for expected credit losses on an individual basis and excluded from the collective evaluation. Specific allocations of the allowance for credit losses are determined by analyzing the borrower’s ability to repay amounts owed, collateral deficiencies, the relative risk grade of the loan and economic conditions affecting the borrower’s industry, among other things. A loan is considered to be collateral dependent when, based upon management's assessment, the borrower is experiencing financial difficulty and repayment is expected to be provided substantially through the operation or sale of the collateral. In such cases, expected credit losses are based on the fair value of the collateral at the measurement date, adjusted for estimated selling costs if satisfaction of the loan depends on the sale of the collateral. We reevaluate the fair value of collateral supporting collateral dependent loans on a quarterly basis. The fair value of real estate collateral supporting collateral dependent loans is evaluated by our internal appraisal services using a methodology that is consistent with the Uniform Standards of Professional Appraisal Practice. The fair value of collateral supporting collateral dependent construction loans is based on an “as is” valuation.


	Commercial and Industrial		Energy		Commercial Real Estate		Consumer Real Estate		Consumer and Other		Total
Historical valuation allowances	$	29,015	$	7,873	$	21,947	$	2,690	$	7,562	$	69,087
Specific valuation allowances	7,849		20,246		383		—		5		28,483
General valuation allowances	9,840		5,196		4,201		904		(409)		19,732
Macroeconomic valuation allowances	4,889		4,067		4,506		519		884		14,865
Total	$	51,593	$	37,382	$	31,037	$	4,113	$	8,042	$	132,167

The following table details activity in the allowance for credit losses on loans by portfolio segment for the three and nine months ended September 30, 2020 and 2019. Allocation of a portion of the allowance to one category of loans does not preclude its availability to absorb losses in other categories. No allowance for credit losses has been recognized for PPP loans as such loans are fully guaranteed by the SBA.


	Commercial and Industrial		Energy		Commercial Real Estate		Consumer Real Estate		Consumer and Other		Total
Three months ended:
September 30, 2020
Beginning balance	$	98,536	$	40,817	$	93,425	$	8,998	$	8,285	$	250,061
Credit loss expense	(18,547)		13,814		25,368		1,794		1,161		23,590
Charge-offs	(8,605)		—		(242)		(1,088)		(4,219)		(14,154)
Recoveries	1,121		—		42		573		2,242		3,978
Net charge-offs	(7,484)		—		(200)		(515)		(1,977)		(10,176)
Ending balance	$	72,505	$	54,631	$	118,593	$	10,277	$	7,469	$	263,475
September 30, 2019
Beginning balance	$	57,714	$	25,818	$	35,914	$	5,637	$	9,846	$	134,929
Credit loss expense	(3,527)		8,788		(607)		(650)		3,997		8,001
Charge-offs	(2,705)		(2,000)		—		(557)		(6,357)		(11,619)
Recoveries	1,185		740		46		454		2,823		5,248
Net charge-offs	(1,520)		(1,260)		46		(103)		(3,534)		(6,371)
Ending balance	$	52,667	$	33,346	$	35,353	$	4,884	$	10,309	$	136,559
Nine months ended:
September 30, 2020
Beginning balance	$	51,593	$	37,382	$	31,037	$	4,113	$	8,042	$	132,167
Impact of adopting ASC 326	21,263		(10,453)		(13,519)		2,392		(2,248)		(2,565)
Credit loss expense	10,737		96,478		104,716		3,716		8,096		223,743
Charge-offs	(14,815)		(68,842)		(3,826)		(1,508)		(13,402)		(102,393)
Recoveries	3,727		66		185		1,564		6,981		12,523
Net charge-offs	(11,088)		(68,776)		(3,641)		56		(6,421)		(89,870)
Ending balance	$	72,505	$	54,631	$	118,593	$	10,277	$	7,469	$	263,475
September 30, 2019
Beginning balance	$	48,580	$	29,052	$	38,777	$	6,103	$	9,620	$	132,132
Credit loss expense	9,999		7,478		(2,972)		859		10,040		25,404
Charge-offs	(8,782)		(4,000)		(617)		(2,936)		(17,157)		(33,492)
Recoveries	2,870		816		165		858		7,806		12,515
Net charge-offs	(5,912)		(3,184)		(452)		(2,078)		(9,351)		(20,977)
Ending balance	$	52,667	$	33,346	$	35,353	$	4,884	$	10,309	$	136,559


	September 30, 2020				December 31, 2019
	Loan Balance		Specific Allocations		Loan Balance		Specific Allocations
Commercial and industrial	$	22,930	$	5,256	$	24,360	$	7,849
Energy	56,420		8,357		65,244		20,246
Paycheck Protection Program	—		—		—		—
Commercial real estate:
Buildings, land and other	25,611		1,313		8,609		383
Construction	680		—		665		—
Consumer real estate	1,335		—		570		—
Consumer and other	18		18		5		5
Total	$	106,994	$	14,944	$	99,453	$	28,483