CECL CONSIDERATIONS FOR CREDIT CARDS

Transcription

1 WHITE PAPER CECL CONSIDERATIONS FOR CREDIT CARDS A Special Case June 20, 2018 Written by Vikas Sharma Practice Lead, Banking Analytics Manish Jain Vice President, Analytics Varun Aggarwal Senior Engagement Manager, Analytics Contributors Jyoti Thakral Senior Manager, Analytics Prudhvi Pandurangarao Arisetti Senior Consultant, Analytics lookdeeper@exlservice.com

2 CECL CONSIDERATIONS FOR CREDIT CARDS A Special Case Among all asset classes, credit cards pose a set of unique challenges for financial institutions in measuring Current Expected Credit Loss (CECL). In this paper, we discuss these specific challenges and key considerations for the benefit of banking and financial services industry. For a high-level primer on CECL, refer to this previous paper. Nearly 45% of U.S. households carry credit card debt, with the total debt exceeding USD $830 billion by the end of Most financial institutions in the US, including community banks and credit unions, offer credit cards. Without doubt, credit cards are an important asset class for U.S. financial institutions from both a volume and value perspective. Current Approach: U.S. GAAP As per the current US GAAP standards, ALLL for credit card debt is primarily estimated based on two guidelines - FAS 5 and FAS 114: FAS 114 is applicable for loans which are considered to be impaired based on current information and events, meaning it is probable that the creditor will be unable to collect all due amounts. ALLL for such a loan is generally estimated based on the present value of the expected future cash flows over the life of the loan discounted at the loan s effective interest rate. Loss reserves for these loans typically form a small component of total ALLL. FAS 5 is applicable for the non-impaired credit card loans. These loans are segmented into homogenous groups based on risk characteristics and evaluated collectively. ALLL for such loans is estimated for a fixed forecast window (usually 12 months) based on incurred loss methodologies, and constitute the major portion of ALLL. In addition to FAS 5 and FAS 114 allowances, ALLL estimate has another component that covers volatility for the forecast of expected credit losses. This volatility may arise due to fluctuations in macroeconomic conditions, differences in asset performance expectations and policy changes. While estimating losses as per FAS 5 guideline under GAAP, financial institutions need to account for expected losses to be incurred in a fixed forecast window (typically 12 months) from non-impaired loans. These future losses can be traced back to: The outstanding balance at the time of observation Future draws such as purchases, balance transfers, cash advances New accounts acquired during the forecast window Major Changes Under CECL The Financial Accounting Standard Board (FASB) has issued CECL standard on loss impairment. The new standard increases complexity for all financial assets in general and credit cards in particular. A few changes that have major implications on ALLL estimation for credit cards. EXLSERVICE.COM 2

3 A. Shift from Fixed Forecast Window to Life Of Loan Excerpt from FASB s ASU No : : An entity shall estimate expected credit losses over the contractual term of the financial asset(s). This implies that instead of estimating loss reserves for a fixed time horizon (next 12 months), expected losses are to be estimated over the entire life of the loan. Determining the expected life of a credit card receivable is inherently complicated since borrowers have payment options to make minimum, partial or full payments. B. No Allowance Required for Current Undrawn Amount Excerpt from FASB s ASU No : : Bank M has a significant credit card portfolio, including funded balances on existing cards and unfunded commitments, or available credit, on credit cards. Bank M s card holder agreements stipulate that the available credit may be unconditionally cancelled at any time : When determining the allowance for credit losses, Bank M estimates the expected credit losses over the remaining lives of the funded credit card loans. Bank M does not record an allowance for unfunded commitments on the unfunded credit cards because it has the ability to unconditionally cancel the available lines of credit. Even though Bank M has had a past practice of extending credit on credit cards before it has detected a borrower s default event, it does not have a present contractual obligation to extend credit. Therefore, an allowance for unfunded commitments should not be established because credit risk on commitments that are unconditionally cancellable by the issuer are not considered to be a liability. The above guidelines imply that for revolving lines which can be unconditionally cancellable by the issuer, such as most of credit cards, issuers should only estimate losses on funded balances. Institutions should not account for losses originating due to future draws which is significantly different from current GAAP guidelines. C. No Provision Needed for New Originations in Next 12 Months Excerpt from FASB s ASU No : : When developing an estimate of expected credit losses on financial asset(s), an entity shall consider available information relevant to assessing the collectability of cash flows. This information may include internal information, external information, or a combination of both relating to past events, current conditions, and reasonable and supportable forecasts. The idea of using available information along with no allowance for any unfunded commitment to estimate expected losses over contractual term of loan implies that there is no requirement to include estimation of losses from new originations. EXLSERVICE.COM 3

4 Challenges for Credit Card Loans and Feasible Solutions As revolving lines of credit, credit cards have many more complexities when compared to non-revolving lending products like mortgages, auto loans, student loans and personal loans. This section highlights four key challenges and presents their possible solutions. A. Portfolio Segmentation Challenge For estimation of loss reserves, it is critical to identify various dimensions that impact credit cardholder s payment behavior such as intent, capacity and time of payment. Apart from external factors such as macroeconomic conditions and industry trends, loan level and customer level attributes play a significant role. In comparison to an overall model, segmented level models provide an additional lens and typically achieve higher accuracy. FASB has recommended looking at groups of loans with similar risk characteristics. However, no standardized schema of portfolio segmentation is issued, leaving it up to financial institutions to decide their starting point. When borrowers have options to revolve balances, roll forward and roll back, segmentation problem becomes more severe. Solution Consider the following dimensions before making a segmentation decision: Zero vs. positive balance: As a CECL model does not require considering any allowance for unconditionally cancellable loan commitments, zero balance accounts should be excluded. Current vs. delinquent: Delinquent accounts behave quite differently and are much more likely to result in losses as compared to current accounts. Transactors vs. revolvers: Though the definition of transactions and revolvers is subjective and vary across financial institutions, transactors vs. revolvers is an important dimension because of the difference in payment pattern. Revolvers, by definition, have a much longer life of the loan. Transactors, on the other hand, are most likely to pay off the loan in next month. Credit rating: Customers with a lower credit score or rating are more likely to charge-off. Low vs. high tenure: New accounts have limited history for predicting expected losses and may need a separate model or a completely different modeling methodology. EXLSERVICE.COM 4

5 B. Life of Loan Estimation Challenge Defining the effective life for revolving loans like credit cards is a daunting task for all institutions. Though there might be some historical analysis, research and literature available to calculate effective lives of a mortgage, auto loan or personal loan products, most banks would not have analyzed credit card receivables to estimate the remaining life of an outstanding balance as required by CECL. Analysis of payment patterns over time for closed-end loans like mortgages, auto loans, student loans, and other term loan products enables financial institutions to identify an estimated life for those loans and integrate the results into the lifetime loss estimation methods applicable under new accounting standard. Moreover, vintages of closed-end loans can be observed over time to determine an estimated life and lifetime loss experience. However, these approaches are not directly applicable for a credit card receivable balance given the revolving nature. Credit card vintages are rather looked upon as accounts age without direct relation with expected cash flows. Possible Solution Since credit card balances fluctuates due to future drawdowns and need not get reduced to zero for active accounts, it is recommended to design an amortization plan for the current balance. Furthermore, there is merit in developing such a plan by segments. C. Amortization Plan Challenge It is important to develop an understanding of the specific amortization pattern in estimating the life of the loan. To calculate the balance liquidation curve, banks will have to estimate future payments and attribute a portion of it to current outstanding balance. Though the Credit Card Accountability Responsibility and Disclosure Act of 2009 (the CARD Act) will govern how the payments are applied to outstanding amounts, simulating it is significantly complex. One of the important clauses in the CARD Act requires the application of a credit cardholder s payments on the basis of a hierarchy under which the components of the cardholder s balance with higher APRs are generally paid off before the components with lower APRs. Section 104 of the CARD Act: Upon receipt of a payment from a cardholder, the card issuer shall apply amounts in excess of the minimum payment amount first to the card balance bearing the highest rate of interest, and then to each successive balance bearing the next highest rate of interest, until the payment is exhausted. Most financial institutions may not have granular data to attribute historical payments by different balance components to create a clean view to replicate the payment allocation process as per CARD Act. In addition, the balance may move across different categories such as promotional balance, go-to balance, risk balance and protected balance, which complicates the analysis. EXLSERVICE.COM 5

6 Possible Solution There are multiple approaches that can be explored for amortization. These approaches differ in terms of complexity, data requirements and the impact they will have on estimated reserve. First-In First-Out (FIFO): FIFO approach applies all future payments to the current balance first. Though this may not align with CARD Act guidance, it is fairly easy to implement and is likely to be followed by several financial institutions. CARD Act replication: In this approach, banks not only need to predict future payments, but also future draws across different balance segments, such as purchases, cash advances, and promotional balances, and at different APR levels. This will enable them to simulate the way payments will be applied as per the Card Act. However, this is a quite complex approach and will require historical granular data for balances and payments by different segments. Hybrid approach: Banks can also use different variations of FIFO approach by attributing a portion of future payments instead of full payments to observation balance in order to capture the impact of future customer behavior on the payment amount. This approach will have a longer tail than FIFO approach and will result in a higher reserve. D. Event Definition for Model Development Challenge Under the current US GAAP approach, it is straightforward to define the dependent variables for developing the models. The current models are based on charge-off events and the balance at the time of charge-off within a fixed performance window. However, the concept of lifetime loss estimation under CECL and the complications with respect to revolving lines has made it difficult. Banks may need to segregate accounts which are charging off on current outstanding balance vs. future draws. Possible Solution While developing the probability of default models, there are four types of credit card accounts that may be observed during available performance window: Type A: Accounts that get charged off before paying off the current balance. Such accounts are bad and should be treated as events for predicting charge-off likelihood. Type B: Accounts that pay off the current balance but eventually get charged-off. Such accounts are charged off on unfunded commitments only, and hence should be treated as good accounts Type C: Accounts that pay off the current balance and remain active. Such accounts are good accounts. Type D: Accounts that remain active without paying off the current balance. Such accounts are right to be censored, and are technically good accounts for available performance window. However, if the performance window is too short and proportion of outstanding balance is high, reasonable assumptions should be made for extrapolating payment patterns. Conclusion Credit cards present a unique challenge since issuers will have to estimate loss from current outstanding balance ignoring future drawdowns, a study which probably was never done in the past. The most important part of the exercise will be to calculate balance liquidation curves. There are multiple options which can be explored. The final approach can be chosen based on the output accuracy, the complexity of the approach and ease of implementation. End Note 1. Federal Reserve Bank of New York Consumer Credit Panel EXLSERVICE.COM 6

7 EXLSERVICE.COM 7

8 EXL (NASDAQ: EXLS) is a leading operations management and analytics company that designs and enables agile, customer-centric operating models to help clients improve their revenue growth and profitability. Our delivery model provides market-leading business outcomes using EXL s proprietary Business EXLerator Framework, cutting-edge analytics, digital transformation and domain expertise. At EXL, we look deeper to help companies improve global operations, enhance data-driven insights, increase customer satisfaction, and manage risk and compliance. EXL serves the insurance, healthcare, banking and financial services, utilities, travel, transportation and logistics industries. Headquartered in New York, New York, EXL has more than 27,000 professionals in locations throughout the United States, Europe, Asia (primarily India and Philippines), South America, Australia and South Africa ExlService Holdings, Inc. All Rights Reserved. For more information, see GLOBAL HEADQUARTERS 280 Park Avenue, 38th Floor New York, New York T F United States United Kingdom Czech Republic Romania Bulgaria India Philippines Colombia South Africa EXLSERVICE.COM lookdeeper@exlservice.com