Accounts receivable balances are predominately composed of amounts currently due from customers (e.g., for wireless services and monthly device lease payments), device insurance administrators, wholesale partners, other carriers and third-party retail channels.

We estimate credit losses associated with our accounts receivable portfolio segment using an expected credit loss model, which utilizes an aging schedule methodology based on historical information and adjusted for asset-specific considerations, current economic conditions and reasonable and supportable forecasts.

Our approach considers a number of factors, including our overall historical credit losses, net of recoveries, timely payment experience as well as current collection trends such as write-off frequency and severity. We also consider other qualitative factors such as macro-economic conditions, including the expected economic impacts of the Pandemic.

We consider the need to adjust our estimate of credit losses for reasonable and supportable forecasts of future economic conditions. To do so, we monitor professional forecasts of changes in real U.S. gross domestic product and forecasts of consumer credit behavior for comparable credit exposures. We also periodically evaluate other economic indicators such as unemployment rates to assess their level of correlation with our historical credit loss statistics.

Based upon customer credit profiles at the time of customer origination, we classify the EIP receivables segment into two customer classes of “Prime” and “Subprime.” Prime customer receivables are those with lower credit risk and Subprime customer receivables are those with higher credit risk. Customers may be required to make a down payment on their equipment purchases if their assessed credit risk exceeds established underwriting thresholds. In addition, certain customers within the Subprime category may be required to pay a deposit.

To determine a customer’s credit profile and assist in determining their credit class, we use a proprietary credit scoring model that measures the credit quality of a customer using several factors, such as credit bureau information, consumer credit risk scores and service and device plan characteristics.

Installment receivables acquired in the Merger are included in EIP receivables. We applied our proprietary credit scoring model to the customers acquired in the Merger with an outstanding EIP receivable balance. Based on tenure, consumer credit risk score and credit profile, these acquired customers were classified into our customer classes of Prime or Subprime. For EIP receivables acquired in the Merger, the difference between the fair value and UPB of the receivable at the acquisition date is accreted to interest income over the contractual life of the receivable using the effective interest method. EIP receivables had a combined weighted-average effective interest rate of 5.6% and 6.7% as of December 31, 2021 and 2020, respectively.

The following table summarizes the EIP receivables, including imputed discounts and related allowance for credit losses:


(in millions)	December 31, 2021		December 31, 2020


EIP receivables, gross	$	8,207	$	6,213
Unamortized imputed discount	(378)		(325)

EIP receivables, net of unamortized imputed discount	7,829		5,888
Allowance for credit losses	(252)		(280)

EIP receivables, net of allowance for credit losses and imputed discount	$	7,577	$	5,608
Classified on the consolidated balance sheets as:
Equipment installment plan receivables, net of allowance for credit losses and imputed discount	$	4,748	$	3,577
Equipment installment plan receivables due after one year, net of allowance for credit losses and imputed discount	2,829		2,031

EIP receivables, net of allowance for credit losses and imputed discount	$	7,577	$	5,608

Many of our loss estimation techniques rely on delinquency-based models; therefore, delinquency is an important indicator of credit quality in the establishment of our allowance for credit losses for EIP receivables. We manage our EIP receivables portfolio segment using delinquency and customer credit class as key credit quality indicators. The following table presents the


	Originated in 2021				Originated in 2020				Originated prior to 2020				Total EIP Receivables, net of unamortized imputed discounts
(in millions)	Prime		Subprime		Prime		Subprime		Prime		Subprime		Prime		Subprime		Grand total
Current - 30 days past due	$	3,894	$	2,419	$	878	$	464	$	22	$	6	$	4,794	$	2,889	$	7,683
31 - 60 days past due	24		37		7		10		—		—		31		47		78
61 - 90 days past due	8		15		3		5		—		—		11		20		31
More than 90 days past due	7		15		4		8		1		2		12		25		37
EIP receivables, net of unamortized imputed discount	$	3,933	$	2,486	$	892	$	487	$	23	$	8	$	4,848	$	2,981	$	7,829

We estimate credit losses on our EIP receivables segment applying an expected credit loss model, which relies on historical loss data adjusted for current conditions to calculate default probabilities or an estimate for the frequency of customer default. Our assessment of default probabilities includes receivables delinquency status, historical loss experience, how long the receivables have been outstanding, customer credit ratings as well as customer tenure. We multiply these estimated default probabilities by our estimated loss given default, which is the estimated amount or severity of the default loss after adjusting for estimated recoveries.

As we do for our accounts receivable portfolio segment, we consider the need to adjust our estimate of credit losses on EIP receivables for reasonable and supportable forecasts of economic conditions through monitoring external professional forecasts and periodic internal statistical analyses, including the expected economic impacts of the Pandemic.

Activity for the years ended December 31, 2021 and 2020, in the allowance for credit losses and unamortized imputed discount balances for the accounts receivable and EIP receivables segments were as follows:


	December 31, 2021						December 31, 2020						December 31, 2019
(in millions)	Accounts Receivable Allowance		EIP Receivables Allowance		Total		Accounts Receivable Allowance		EIP Receivables Allowance		Total		Accounts Receivable Allowance		EIP Receivables Allowance		Total

Allowance for credit losses and imputed discount, beginning of period	$	194	$	605	$	799	$	61	$	399	$	460	$	67	$	449	$	516
Beginning balance adjustment due to implementation of the new credit loss standard	—		—		—		—		91		91		—		—		—

Bad debt expense	231		221		452		338		264		602		77		230		307
Write-offs, net of recoveries	(279)		(248)		(527)		(205)		(175)		(380)		(83)		(249)		(332)
Change in imputed discount on short-term and long-term EIP receivables	N/A		187		187		N/A		171		171		N/A		136		136
Impact on the imputed discount from sales of EIP receivables	N/A		(135)		(135)		N/A		(145)		(145)		N/A		(167)		(167)

Allowance for credit losses and imputed discount, end of period	$	146	$	630	$	776	$	194	$	605	$	799	$	61	$	399	$	460

We do not have material, unmitigated off-balance-sheet credit exposures as of December 31, 2021. In connection with the sales of certain service and EIP accounts receivable pursuant to the sale arrangements, we have deferred purchase price assets included on our Consolidated Balance Sheets measured at fair value that are based on a discounted cash flow model using Level 3 inputs, including customer default rates and credit worthiness, dilutions and recoveries. See Note 4 – Sales of Certain Receivables for further information.