The following tables represent, by loan portfolio segment, a summary of changes in the ACL on loans for the three and six months ended June 30, 2021 and 2020:


	Three Months Ended June 30, 2021
	Commercial		Real Estate		Mortgage Warehouse		Consumer		Total
Balance, beginning of period	$	42,980	$	4,229	$	1,163	$	8,814	$	57,186
Provision for credit losses on loans	(1,168)		(144)		(8)		(172)		(1,492)
PCD loan charge–offs	(6)		—		—		—		(6)
Charge–offs	(67)		—		—		(237)		(304)
Recoveries	27		23		—		215		265
Balance, end of period	$	41,766	$	4,108	$	1,155	$	8,620	$	55,649


	Three Months Ended June 30, 2020
	Commercial		Real Estate		Mortgage Warehouse		Consumer		Total
Balance, beginning of period	$	32,550	$	5,654	$	1,055	$	9,181	$	48,440

Provision for credit losses on loans	6,603		202		135		117		7,057
Charge–offs	(15)		(34)		—		(526)		(575)
Recoveries	9		10		—		149		168
Balance, end of period	$	39,147	$	5,832	$	1,190	$	8,921	$	55,090


	Six Months Ended June 30, 2021
	Commercial		Real Estate		Mortgage Warehouse		Consumer		Total
Balance, beginning of period	$	42,210	$	4,620	$	1,267	$	8,930	$	57,027


Provision for credit losses on loans	(240)		(600)		(112)		(173)		(1,125)
PCD loan charge–offs	(6)		—		—		—		(6)
Charge–offs	(263)		—		—		(472)		(735)
Recoveries	65		88		—		335		488
Balance, end of period	$	41,766	$	4,108	$	1,155	$	8,620	$	55,649


	Six Months Ended June 30, 2020
	Commercial		Real Estate		Mortgage Warehouse		Consumer		Total
Balance, beginning of period	$	11,996	$	923	$	1,077	$	3,671	$	17,667
Impact of adopting ASC 326	13,618		4,048		—		4,911		22,577
Provision for credit losses on loans	13,519		902		113		1,123		15,657
Charge–offs	(84)		(60)		—		(1,144)		(1,288)
Recoveries	98		19		—		360		477
Balance, end of period	$	39,147	$	5,832	$	1,190	$	8,921	$	55,090

The Company utilized the Cumulative Loss Rate method in determining expected future credit losses. The loss rate method measures the amount of loan charge–offs, net of recoveries, (“loan losses”) recognized over the life of a pool and compares those loan losses to the outstanding loan balance of that pool as of a specific point in time (“pool date”).

To estimate a CECL loss rate for the pool, management first identifies the loan losses recognized between the pool date and the reporting date for the pool and determines which loan losses were related to loans outstanding at the pool date. The loss rate method then divides the loan losses recognized on loans outstanding as of the pool date by the outstanding loan balance as of the pool date.

The Company’s expected loss estimate is anchored in historical credit loss experience, with an emphasis on all available portfolio data. The Company's historical look–back period includes January 2012 through the current period, on a monthly basis. When historical credit loss experience is not sufficient for a specific portfolio, the Company may supplement its own portfolio data with external models or data.

Qualitative reserves reflect management’s overall estimate of the extent to which current expected credit losses on collectively evaluated loans will differ from historical loss experience. The analysis takes into consideration other analytics performed within the organization, such as enterprise and concentration management, along with other credit–related analytics as deemed appropriate. Management attempts to quantify qualitative reserves whenever possible.

The Company’s CECL estimate applies to a forecast that incorporates macroeconomic trends and other environmental factors. Management utilized National, Regional and Local Leading Economic Indexes, as well as management judgment, as the basis for the forecast period. The historical loss rate was utilized as the base rate, and qualitative adjustments were utilized to reflect the forecast and other relevant factors.

The Company segments the loan portfolio into pools based on the following risk characteristics: financial asset type, loan purpose, collateral type, loan characteristics, credit characteristics, outstanding loan balances, contractual terms and prepayment assumptions, industry of the borrower and concentrations, and historical or expected credit loss patterns.