Financial institutions face implementation of a new accounting requirement that was issued in June of 216 by the Financial Accounting Standards Board (FASB), Financial Instruments – Credit Losses (Topic 326) commonly referred to as “CECL.” This new standard will become effective in 2020 for SEC filers and 2021 for all other entities – but compliance requires significant review and potential change in many aspects of governance, risk management, credit models and other aspects of operations, so banks must prepare well before the implementation date to be ready by then. CECL, or current expected credit losses, represents a major change in how banks will be expected to estimate losses in the allowance for loan and lease losses (ALLL). This presentation, provided at a Kansas Bankers Association meeting in November 2016, gives an overview of CECL and how to prepare for compliance with it.
1. Preparing for CECL
Nov. 9, 2016
Mark Schmelzle, Senior Vice President, Assurance
Allen, Gibbs & Houlik, L.C.
2. Learning Objectives
Review the background
on the CECL model
Discuss steps to
prepare for CECL
Analyze
implementation
guidance and review
examples
3.
4. Background
Proposed accounting pronouncement
issued June 2016 Financial Instruments
– Credit Losses (Topic 326)
For calendar year public SEC filers, the
guidance is effective Jan. 1, 2020.
For all other calendar year entities, the
guidance is effective on Jan. 1, 2021.
Early adoption is permitted.
Modified retrospective application with
cumulative-effect adjustment recognized
in first period of adoption.
5. CECL: How did we get here?
1976
• FAS 5:
Incurred Loss
Probable
1998-
2007
• Earnings
management
2007-
2010
• Financial
crisis
Present
• CECL:
Expected loss
life of loan
6. Scope
Lease
receivables
recognized by
lessor
Trade
receivables
that result
from
revenue
transactions
All debt
instruments
excepts AFS
debt securities
and fair value
through NI
Financial
guarantee
contracts
Loan
commitments
AFS debt securities /
fair value through NI
Equity securities
Equity method
investments
Derivatives
Related party loans
and receivables
between entities under
common control
Pledge receivables of a
not-for-profit entity
In Scope Out of Scope
7. Critical CECL Differences
Factor Incurred loss model CECL
Loss recognition
threshold
Probable that loss has been
incurred at the balance
sheet date and can be
estimated
No threshold
Adjust historical
loss experience
Conditions up to the
measurement date not
reflected in the historical
experience
Conditions throughout the life of the loan
portfolio
Loss recognition
pattern
Losses recognized in
income in the period in
which they happen
Expected losses recognized generally upon
initial recognition of the loan (originated or
purchased) and as expectations change for
future period for the life of the loan
Change in facts
and conditions
supporting loss
estimate
Losses recognized in
income in the period in
which they happen
Recognized in earnings for all expected
losses over life of the loan portfolio
Expected
renewals and
rollovers
Not relevant Expected prepayments should be
considered but expected extensions,
renewal and modifications should not be
considered unless the lender reasonably
expects to execute a TDR
8. Critical CECL Differences (cont.)
Factor Incurred loss model CECL
Content of ALLL Remaining amount of
incurred losses not yet
confirmed and charged off
All expected losses that have not yet been
charged off for the loan portfolio
Losses inherent
in purchased
loans
Loans are booked at their
purchase value (fair value) –
no initial allowance needed
or allowed for such losses;
only for subsequent losses
Loans generally booked at face with an
accretable discount for cash flows expected
to be collected and either:
1. An allowance for expected losses
created in purchase accounting (for
purchased credit impaired loans)
2. An allowance for expected losses
recognized through a charge to earnings
(non-PCI loans)
Loans held for
sale
Carried at lower of cost or
market (LCM)
Same
Losses on loans
identified as
impaired
Estimated based on
expected cash flows,
estimated collateral value or
expected loan sale value
Same
9. Implementation Considerations
Segmentation requirements
Credit losses will be
evaluated on a
collective/pool basis
Financial assets should be
evaluated for impairment
individually if common
characteristics do not exist
Assets cannot be included
in individual and collective
assessments
10. Segmentation Considerations
• Internal and external -
credit score or credit
rating
• Risk rating or
classification
• Financial asset type
• Collateral type
• Size
• Terms
• Geographical location
• Industry of the borrower
• Vintage
• Patterns with the credit
• Reasonable and
supportable forecast
periods
11. Operational Implications
Implementation of standardized
processes & integration of tools,
systems & processes
Lending practices in data
management to ensure data
quality & integrity
Delineation of roles and
responsibilities for personnel
involved
Separation of roles of lending
personnel & risk grading
personnel
Internal
Controls
Interaction
and
Relationships
Between
Departments
and Functions
Data
Governance
Framework
12. Data Requirements
Data quality / limitations
will influence choices
made during
implementation
Vendor data if used in
modeling needs to be
relevant to the portfolio’s
loss history and that
relevance needs to be
demonstrated
13. Audit and Regulatory Considerations
• How is the life of a loan being estimated?
• Does the approach for estimating a loan’s life lend
itself to being independently tested?
Life of loan
• How does the institution segment its portfolio?
• How is the segmentation documented?
• Does the segmentation approach capture risk
changes in the portfolio using a variety of attributes
to assess risk (ratings, types, geography)?
Portfolio
Segmentation
• How is data captured and can it be reproduced?
• How is data maintained?
• What processes have been done to determine
accuracy and validity of the data being used?
Quantitative
Methodologies
14. Audit and Regulatory Considerations
• How well documented and transparent is the
qualitative methodology?
• How does qualitative data align with regulatory
guidance?
• Are other qualitative factors used?
Qualitative
Methodologies
• What methods and assumptions are being used to
develop reasonable and supportable forecasts?
• What are the inputs to determine a loss forecast?
• How relevant are the inputs in the loss forecast to
the portfolio’s loss experience?
• How does the bank determine its forecast horizon?
Reasonable and
Supportable
Forecasts
CECL will likely involve new quantitative analysis and documentation, the
implementation and usage of such analysis will require more judgment and
thus great scrutiny of that judgment.
15. Implementation Analysis
CECL Implementation Cycle
Educate
Organize &
Govern
Quantify
Automate &
Report
Range of needs
• Executive Education
• Industry Events
• Internal Training
• Gap Analysis
• Framework design
• Credit risk models
• Cash flow
generation
• Economic forecasts
• Calculation engine
• Workflow
• Management
reporting
• Operational training
• Staff coursework
• Point of view sessions
• Project planning
• Program management
• Benchmark data
• Qualitative adjustments
• GL postings
• Forecasting and
planning
• Disclosures
16. Differing Methods May be Applied
Discounted cash flow methods
Probability of default
Loss-rate approach
Vintage-year
Roll-rate method (migration analysis)
Regression analysis
17. Discounted Cash Flows
• Similar to past model
• Removes the wording “best
estimate”
• Must consider some risk of
loss
• Likely will need new data to
support cash flow
expectations – credit ratings,
loss information for similar
assets and credit quality.
18. Probability of Default
Considers attributes within a pool of assets and
the institution must demonstrate the strong
predictive power of the model.
More complex than other models
Some of the suggested drivers
to determine default include; risk rating, past-due
status, credit scores, loan-to-value.
19. Loss-rate Approach
Computes a lifetime charge-off rate based
on historical performance of a pool
Must estimate and support the life of pool
when determining lifetime charge-off rate
One quantitative adjustment is made based
on typically several supported factors
20. Loss-rate Example
Fact Pattern
Bank A provides 10-year
amortizing loans to customers.
Bank A manages those loans
on a collective basis based on
similar risk characteristics.
The loans within the portfolio
were originated during the last
10 years.
The portfolio has an amortized
cost basis of $3 million.
21. Loss-rate Example
Fact Pattern
Bank A believes that its most recent 10-year period is a reasonable
period on which to base its expected credit-loss-rate calculation after
considering the underwriting standards and contractual terms for loans
that existed over the historical period in comparison with the current
portfolio.
Bank A’s historical lifetime credit loss rate (that is, a rate based on the
sum of all credit losses for a similar pool) for the most recent 10-year
period is 1.5 percent.
Prepayment has been factored in, which the bank expects to remain
unchanged.
Community Bank A considered whether any adjustments to historical
loss information were needed, before considering adjustments for
current conditions and reasonable and supportable forecasts, but
determined none were necessary.
22. Loss-rate Example
Fact Pattern
Bank A considered significant factors that could affect the
expected collectibility of the amortized cost basis of the portfolio
and determined that the primary factors are real estate values and
unemployment rates.
Bank A observed that real estate values in the community have
decreased and the unemployment rate in the community has
increased as of the current reporting period date.
Based on current conditions and reasonable and supportable
forecasts, Bank A expects that there will be an additional
decrease in real estate values over the next one to two years, and
unemployment rates are expected to increase further over the
next one to two years.
23. Loss-rate Example
Applying Reasonable and Supportable Forecast
Bank A estimates a 10-basis-point increase in credit losses
incremental to the 1.5 percent historical lifetime loss rate due to
the expected decrease in real estate values
A 5-basis-point increase in credit losses was added due to
expected deterioration in unemployment rates.
Management estimates the incremental 15-basis-point increase
based on its knowledge of historical loss information during past
years in which there were similar trends in real estate values and
unemployment rates.
24. Loss-rate Example
Determining the Allowance
Management is unable to support its estimate of expectations for
real estate values and unemployment rates beyond the
reasonable and supportable forecast period.
Reversion techniques should be used for the loss-rate method
for period beyond supportable to life of the pool.
The expected loss rate to apply to the amortized cost basis of
the loan portfolio would be 1.65 percent, the sum of the historical
loss rate of 1.5 percent and the adjustment for the current
conditions and reasonable and supportable forecast of 15 basis
points. The allowance for expected credit losses at the report
date would be $49,500 (3,000,000 * 1.65%).
25. Problems and Challenges with Loss-rate Method
Uses the full 10-year history of a 10-year
old vintage
• Do you have full lifetime history for a
“similar” vintage to each of your
active vintages?
Assumes the lender has scenarios for
relevant macroeconomic factors for one
to two years
• Do you know which factors drive
each of your portfolios and have
scenarios for them?
26. Problems and Challenges with Loss-rate Method
Assumes management has a
model that predicts increased
losses from adverse
macroeconomic conditions.
• Do you have a macroeconomic
model for each of your portfolios?
The cumulative lifetime loss
estimate is applied to the entire
active portfolio, regardless of loan
age.
• May give you results you do not
want, a loan with 6 months left on
a 10-year term is being assessed
a cumulative loss rate as if it had
10-years remaining.
27. Vintage Year Basis
Vintage analysis is based
on loss curves that include
expectations of losses at
each point in the life of a
financial asset. Allowance is
represented by the
remaining area under the
loss curve.
28. Vintage Year Example
Fact Pattern
Bank C is a lending institution that provides
financing to customers purchasing new or used
farm equipment throughout the local area.
Bank C originates approximately the same
amount of loans each year.
The four-year amortizing loans it originates are
secured by collateral that provides a relatively
consistent range of loan-to-collateral-value
ratios at origination.
If a borrower becomes 90 days past due, Bank
C repossesses the underlying farm equipment
collateral for sale at auction.
29. Vintage Year Example
Fact Pattern (cont.)
Bank C tracks those loans on the basis of the calendar year of
origination.
Below is the pattern of credit loss information based on the amount of
amortized cost basis in each vintage that was written off as a result of
credit losses (nonshaded information).
Year 1 Year 2 Year 3 Year 4 Total Expected
20X1 50$ 120$ 140$ 30$ 340$ -$
20X2 40$ 120$ 140$ 40$ 340$ -$
20X3 40$ 110$ 150$ 30$ 330$ -$
20X4 60$ 110$ 150$ 40$ 360$ -$
20X5 50$ 130$ 170$ 50$ 400$ -$
20X6 70$ 150$ 180$ 60$ 460$ 60$
20X7 80$ 140$ 190$ 70$ 480$ 260$
20X8 70$ 150$ 200$ 80$ 500$ 430$
20X9 70$ 160$ 200$ 80$ 510$ 510$
Year of Origination
Loss Experience in Years Following Origination
30. Vintage Year Example
Historical Observations
The majority of losses historically emerge in Year 2 and 3
of the loans.
Historical loss experience has worsened since 20X3.
Loss experience for loans originated in 20X6 has already
equaled the loss experience for loans originated in 20X5.
31. Vintage Year Example
Current Conditions (reasonable and supportable)
There is an oversupply of used farm equipment in
the resale market that is expected to continue.
Severe weather in recent years has increased the
cost of crop insurance; this trend is expected to
continue.
Bank C adjusts historical loss information as noted in
the shaded area of the chart. From here, the bank
arrives at expected loss amounts.
32. Chart
Year 1 Year 2 Year 3 Year 4 Total Expected
20X1 50$ 120$ 140$ 30$ 340$ -$
20X2 40$ 120$ 140$ 40$ 340$ -$
20X3 40$ 110$ 150$ 30$ 330$ -$
20X4 60$ 110$ 150$ 40$ 360$ -$
20X5 50$ 130$ 170$ 50$ 400$ -$
20X6 70$ 150$ 180$ 60$ 460$ 60$
20X7 80$ 140$ 190$ 70$ 480$ 260$
20X8 70$ 150$ 200$ 80$ 500$ 430$
20X9 70$ 160$ 200$ 80$ 510$ 510$
Year of Origination
Loss Experience in Years Following Origination
33. Roll Rate Method
• Also referred to as
migration analysis
• Based on determining
a prediction of credit
losses based on
segmentation
(delinquency or risk
rating, for example)
• Determines the
percent of assets that
will “roll” to a more
severe risk
34. Regression
Uses economic data such
as unemployment rates,
bankruptcy rates to
estimate a relationship
between the data and
expected losses
Considered one of the
more complex models
Requires knowledge of
statistics
35. CECL Not All Bad…
More timely recognition of
credit losses
More timely reporting of
credit losses
Greater transparency about
portfolios or assets
Improved ability to
understand changes in
credit quality from period to
period
Better understanding of
portfolios
PCD asset accounting
improved
Benefits
36. How Can You Prepare?
Educate – consider models that will
work with data you have available.
Identify pools that make sense.
Identify data points that correlate
with your loan types.
Start capturing this data to identify
trends.
What type of loan suffered the loss and
when?
In what year was the loan originated?
Many say this change will raise the ALLL
30-50%.