Managing Profitability Under CECL Through Loan Pricing (Part 2)

Part 2: Applying the Conceptual Loan Pricing Framework to CECL
As organizations implement CECL, a key question is how CECL estimates should factor into loan origination and pricing decisions. In Part 1 of this series, we illustrated the mechanics of pricing with a hypothetical loan. In Part 2, we use residential loan data to model how, under CECL, pricing decisions may need to change as economic scenarios and expected portfolio losses change.
Under some CECL approaches, loss reserves could be more front-loaded, driving a higher Provision Expense in earlier periods of a loan’s life. Because Provision Expense is a component of a loan’s discounted pro forma Income Statement, the Net Present Value (NPV) of future earnings would be lower. Our analysis back-tests the effect of three different loss reserve approaches on NPV and loan level profitability.
Defining Provision Expense
Provision Expense includes 1) losses realized in a reporting period and 2) the period-over-period change in a loss reserve balance. Realized losses are booked when a loan’s loss can be individually identified and then estimated. This is also referred to as a “write-down”. Under the incurred loss model, reserves1 include future losses that likely exist in the book but can only be estimated collectively, rather than individually. An increase in reserves is known as a “build” and a decrease a “release”. Because the reserve is booked to the Balance Sheet, its build or release is booked to the Income Statement under Provision Expense.
FI Blog CECL Chart 1
The Data and Assumptions
We analyze public Fannie Mae data spanning vintages 2000 through 2012 to gauge the effect CECL may have on loan level NPV and profitability. This data is grouped into three pools to approximate distinct stages of the most recent credit cycle: 2000-2003, 2004-2008, 2009-2012.
We then apply Part 1 conceptual pricing framework to these datasets. We combine actual loan terms, actual loan performance, and simple pricing assumptions to best isolate our independent variable, Lifetime Expected Credit Loss, and our dependent variable, NPV Impact of CECL Adoption.
FI Blog CECL Chart 2
As a refresher, Financial Institutions consider many factors when determining price. As in Part 1, we again briefly discuss below those critical to NPV. (For conceptual definitions of these, refer to Part 1.) Loans originated with a positive NPV contribute to an institution’s profitability.
FI Blog CECL Chart 3
Capital Structure: We align debt and equity ratios with our Part 1 framework and levels commonly seen within mortgage books. These assumptions are consistent across vintages.
FI Blog CECL Chart 4
Cost of Capital: For simplicity, we assume a cost of debt that generates a consistent Net Interest Margin (NIM) of 1.25%. NIM represents the difference between interest revenue and interest expense. This gives us a more isolated NPV comparison between vintages with widely different borrowing costs and during periods of limited data availability. We align cost of equity and tax rate with our Part 1 framework.
FI Blog CECL Chart 5
Operating Costs: Mortgage lenders and servicers typically earn 0.25% of outstanding balance annually as compensation for servicing loans. We assume servicing costs of 0.20%, leaving the lender a reasonable profit margin of 0.05% for these activities. Again, for simplicity, we remove origination expenses, which are often subject to a separate accounting treatment and introduce unnecessary complexity into the analysis. Operating cost assumptions are consistent across vintages.
FI Blog CECL Chart 6
Note Rate: Note Rate drives interest revenue within the pro forma income statement. These data points are easily observed on each loan. Averages are presented below.
FI Blog CECL Chart 7
Expected Life: To capture the effect of prepayments, we remove active loans from each pool. Overall, 15% of loans from vintages 2000-2012 have not reached a zero balance and are excluded from the analysis. A loan reaches a zero balance after either payment in full or liquidation following a default.
FI Blog CECL Chart 8
Expected Loss: For our Expected Credit Loss (ECL), we assume a perfect loss forecast at time of pricing and substitute actual loss for expected loss. This allows us to isolate our analysis on the timing and size of three different reserve frameworks.
FI Blog CECL Chart 9
Our Three Reserve Frameworks
An allowance1 under the incurred loss model reserves for losses that likely exist in the book but cannot be individually identified and estimated via a write-down. When collectively estimating losses for a reserve, institutions define a Loss Emergence Period. This period represents the amount of time that passes from when a loss is incurred (sudden unemployment, natural disaster) to when a loss can be individually identified and estimated. While we effectively identify and estimate losses on these loans through back-testing, we evaluate two different loss emergence periods as if we are only able to collectively estimate losses.

  1. 12 Month Loss Emergence Period under the Incurred Loss Model: On average it takes the institution 12 months to individually identify and estimate loan level losses. Each period’s loss reserve equals the next 12 months of expected realized credit loss. In addition to losses in the current month, the month-over-month build or release of reserve is booked to Provision Expense.
  2. 24 Month Loss Emergence Period under the Incurred Loss Model: On average it takes the institution 24 months to individually identify and estimate loan level losses. Each period’s loss reserve equals the next 24 months of expected realized credit loss. In addition to losses in the current month, the month-over-month build or release of reserve is booked to Provision Expense.
  3. Current Expected Credit Loss: Each period’s loss reserve equals the total lifetime expected credit loss of the portfolio, discounted at the portfolio’s note rate. In addition to losses in the current month, the month-over-month build or release of reserve is recorded to Provision Expense. Our data availability permits a discounted cash flow (DCF) approach for CECL.

As summarized previously, these three pools experience a range of lifetime credit losses from 10 basis points to 191. These different levels of lifetime losses result in a wide range of average reserves.
FI Blog CECL Chart 10
The three reserve frameworks also drive different NPVs. Within each of the three scenarios, we only change the timing of the loss reserves. Thus, the range in NPV for each pool is entirely attributable to the loss reserve framework. Since the pools vary in size, we show basis points rather than dollars. Originating a loan or portfolio at a positive NPV contributes to institutional profitability.
FI Blog CECL Chart 11
We find a negative correlation between this portfolio’s level of lifetime ECL and the NPV impact of CECL adoption. The coefficient of this relationship is approximately -0.19, suggesting that for a similar portfolio and a similar CECL approach, NPV at time of pricing could worsen by 19 basis points for every 100 basis points of lifetime expected credit loss. Said differently, all else being equal, a financial institution with this portfolio should raise its loan price by 20 basis points for every 100 basis points of expected credit loss to maintain its current status of loan level profitability after adopting CECL.
Each data point below represents an individual vintage. Profitability and NPV are most at risk in a distressed credit environment.
FI Blog CECL Chart 12
As we showed in Part 1 of our series, to gauge profitability a bank should discount the loan’s total equity flows by its internal hurdle rate to arrive at the NPV. In Part 2 we illustrated an inverse relationship between expected credit loss and loan portfolio NPV impact from CECL adoption. Your financial institution’s CECL working group should therefore engage with business lines’ Pricing and Financial Planning teams to work through the detailed mechanics of how expected credit loss impacts loan pricing and what processes you may need to solidify and manage this linkage.
To get started managing loan profitability under CECL, we recommend:

  1. Review your current loan pricing methodology. Work with your pricing teams to gain a deeper understanding of the key factors used for loan pricing. To what extent is Provision Expense a key factor? How are current loss rates utilized for pricing?
  2. Conduct an in-depth portfolio assessment. As we showed in the previous chart, establishing the sensitivity of the relationship between expected credit loss and profitability is one key to understanding your pricing methodology under CECL. You will need to understand your portfolio at a detailed level to establish this relationship, including loss rates changes to NPV as a key component of your pricing methodology.
  3. (Re)evaluate data and modeling methods. You may need to revisit your ECL modeling methods and the underlying data needed to support your portfolio assessment. Pool level analysis may not be sufficiently detailed to gain a true understanding of the ECL/NPV relationship. Additional loan level data may need to be acquired and managed to do the analysis.

To learn more about the pitfalls your organization may face when implementing CECL, download a copy of FI’s White Paper titled “Four Pitfalls to Avoid During CECL Implementation.” For more information on FI’s CECL services check out our CECL homepage.
1An institution’s total loan loss reserve tends to include three different accounting treatments which generally cover the following populations: FAS 5 (collective impairment), FAS 114 (troubled debt restructurings) and SOP 03-3 (acquired loans). We focus on FAS 5 in this analysis, as new originations typically fall outside the scope of FAS 114 or SOP 03-3.