The following table summarizes activity in the allowance for loan and lease losses for the periods indicated:

	Nine Months Ended September 30,
	2020									2019
(in thousands)	Mortgage			Other			Total			Mortgage			Other			Total
Balance, beginning of period	$	122,695		$	24,943		$	147,638		$	130,983		$	28,837		$	159,820
Impact of CECL adoption		99			1,812			1,911			—			—			—
Adjusted balance, beginning of period		122,794			26,755			149,549			130,983			28,837			159,820
Charge-offs		(2	)		(15,398	)		(15,400	)		(1,386	)		(15,010	)		(16,396	)
Recoveries		817			1,504			2,321			43			563			606
Provision for (recovery of) credit losses on loans		47,143			4,694			51,837			(6,673	)		12,076			5,403
Balance, end of period	$	170,752		$	17,555		$	188,307		$	122,967		$	26,466		$	149,433

ASU No. 2016-13 replaces the incurred loss methodology with an expected loss methodology that is referred to as the current expected credit loss (“CECL”) methodology. The measurement of expected credit losses under CECL is applicable to financial assets measured at amortized cost, including loan receivables. It also applies to off-balance sheet exposures not accounted for as insurance and net investments in leases accounted for under ASC Topic 842.

The allowance for loan and lease losses is deducted from the amortized cost basis of a financial asset or a group of financial assets so that the balance sheet reflects the net amount the Company expects to collect. Amortized cost is the principal balance outstanding, net of purchase premiums and discounts, fair value hedge accounting adjustments, and deferred fees and costs. Subsequent changes (favorable and unfavorable) in expected credit losses are recognized immediately in net income as a credit loss expense or a reversal of credit loss expense. Management estimates the allowance by projecting probability-of-default, loss-given-default and exposure-at-default depending on economic parameters for each month of the remaining contractual term. Economic parameters are developed using available information relating to past events, current conditions, and economic forecasts. The Company’s economic forecast period reverts to a historical norm based on inputs after 24 months. Historical credit experience provides the basis for the estimation of expected credit losses, with adjustments made for differences in current loan-specific risk characteristics such as differences in underwriting standards, portfolio mix, delinquency levels and terms, as well as for changes in environmental conditions, such as changes in legislation, regulation, policies, administrative practices or other relevant factors. Expected credit losses are estimated over the contractual term of the loans, adjusted for forecasted prepayments when appropriate. The contractual term excludes potential extensions or renewals. The methodology used in the estimation of the allowance for loan and lease losses, which is performed at least quarterly, is designed to be dynamic and responsive to changes in portfolio credit quality and forecasted economic conditions. Each quarter, we reassess the appropriateness of the economic period, the reversion period and historical mean at the portfolio segment level, considering any required adjustments for differences in underwriting standards, portfolio mix, and other relevant data shifts over time.

The allowance for loan and lease losses is measured on a collective (pool) basis when similar risk characteristics exist. Management believes the products within each of the entity’s portfolio classes exhibit similar risk characteristics. Loans that are determined to have unique risk characteristics are evaluated on an individual basis by management. If a loan is determined to be collateral dependent, or meets the criteria to apply the collateral dependent practical expedient, expected credit losses are determined based on the fair value of the collateral at the reporting date, less costs to sell as appropriate. The macroeconomic data used in the quantitative models are based on an economic forecast period of 24 months. The Company leverages economic projections including property market and prepayment forecasts from established independent third parties to inform its loss drivers in the forecast. Beyond this forecast period, we revert to a historical average loss rate. This reversion to the historical average loss rate is performed on a straight-line basis over 12 months.

The portfolio segment represents the level at which a systematic methodology is applied to estimate credit losses. Smaller pools of homogenous financing receivables with homogeneous risk characteristics were modeled using the methodology selected for the portfolio segment to which factors in the qualitative scorecard include: concentration, modeling and forecast imprecision and limitations, policy and underwriting, prepayment uncertainty, external factors, nature and volume, management, and loan review. Each factor is subject to an evaluation of metrics, consistently applied, to measure adjustments needed for each reporting period.

Loans that do not share risk characteristics are evaluated on an individual basis. These include loans that are in nonaccrual status with balances above management determined materiality thresholds depending on loan class and also loans that are designated as TDR or “reasonably expected TDR” (criticized, classified, or maturing loans that will have a modification processed within the next three months). In addition, all taxi medallion loans are individually evaluated.

The Company maintains an allowance for credit losses on off-balance sheet credit exposures. At September 30, 2020 and December 31, 2019, the allowance for credit losses on off-balance sheet credit exposures was $12.3 million and $461,000, respectively. Included in the September 30, 2020 amount was a $12.5 million adjustment related to the adoption of CECL. We estimate expected credit losses over the contractual period in which the Company is exposed to credit risk via a contractual obligation to extend credit, unless that obligation is unconditionally cancellable by the Company. The allowance for credit losses on off-balance sheet credit exposures is adjusted as a provision for credit losses expense. The estimate includes consideration of the likelihood that funding will occur and an estimate of expected credit losses on commitments expected to be funded over their estimated life. The Company examined historical credit conversion factor (“CCF”) trends to estimate utilization rates, and chose an appropriate mean CCF based on both management judgment and quantitative analysis. Quantitative analysis involved examination of CCFs over a range of fund-up windows (between 12 and 36 months) and comparison of the mean CCF for each fund-up window with management judgment determining whether the highest mean CCF across fund-up windows made business sense. The Company applies the same standards and estimated loss rates to the credit exposures as to the related class of loans.

We charge off loans, or portions of loans, in the period that such loans, or portions thereof, are deemed uncollectible. The collectability of individual loans is determined through an assessment of the financial condition and repayment capacity of the borrower and/or through an estimate of the fair value of any underlying collateral. For non-real estate-related consumer credits, the following past-due time periods determine when charge-offs are typically recorded: (1) closed-end credits are charged off in the quarter that the loan becomes 120 days past due; (2) open-end credits are charged off in the quarter that the loan becomes 180 days past due; and (3) both closed-end and open-end credits are typically charged off in the quarter that the credit is 60 days past the date we received notification that the borrower has filed for bankruptcy.

The following table presents additional information about the Company’s nonaccrual loans at September 30, 2020:

(in thousands)	Recorded Investment		Related Allowance		Interest Income Recognized
Nonaccrual loans with no related allowance:
Multi-family	$	—	$	—	$	—
Commercial real estate		12,006		-		-
One-to-four family		563		-		16
Acquisition, development, and construction		-		-		-
Other		26,613		-		676
Total nonaccrual loans with no related allowance	$	39,182	$	—	$	692
Nonaccrual loans with an allowance recorded:
Multi-family	$	4,068	$	444	$	28
Commercial real estate		667		53		27
One-to-four family		1,143		199		10
Acquisition, development, and construction		-		-		-
Other		291		20		10
Total nonaccrual loans with an allowance recorded	$	6,169	$	716	$	75
Total nonaccrual loans:
Multi-family	$	4,068	$	444	$	28
Commercial real estate		12,673		53		27
One-to-four family		1,706		199		26
Acquisition, development, and construction		-		-		-
Other		26,904		20		686
Total nonaccrual loans	$	45,351	$	716	$	767

The following table presents additional information about the Company’s impaired loans at December 31, 2019:

(in thousands)	Recorded Investment		Unpaid Principal Balance		Related Allowance		Average Recorded Investment		Interest Income Recognized
Impaired loans with no related allowance:
Multi-family	$	3,577	$	6,790	$	—	$	4,336	$	266
Commercial real estate		14,717		19,832		—		6,140		371
One-to-four family		584		602		—		811		21
Acquisition, development, and construction		389		1,289		—		3,508		364
Other		37,669		114,636		—		39,598		2,494
Total impaired loans with no related allowance	$	56,936	$	143,149	$	—	$	54,393	$	3,516
Impaired loans with an allowance recorded:
Multi-family	$	—	$	—	$	—	$	—	$	—
Commercial real estate		—		—		—		—		—
One-to-four family		—		—		—		—		—
Acquisition, development, and construction		—		—		—		—		—
Other		1,445		4,173		116		4,111		13
Total impaired loans with an allowance recorded	$	1,445	$	4,173	$	116	$	4,111	$	13
Total impaired loans:
Multi-family	$	3,577	$	6,790	$	—	$	4,336	$	266
Commercial real estate		14,717		19,832		—		6,140		371
One-to-four family		584		602		—		811		21
Acquisition, development, and construction		389		1,289		—		3,508		364
Other		39,114		118,809		116		43,709		2,507
Total impaired loans	$	58,381	$	147,322	$	116	$	58,504	$	3,529