(EAD) X Loss Given Default (LGD)

Insolvency Practice and the Credit Process
Philip D. Sherman
Senior Adviser in Asia
The Risk Management Association
November 2004
1
Outline
• Banks must take risks . . . They need to have a process to manage the
risks and the inevitable losses . . . and a capital cushion
• The insolvency community deals with the risk failures, whether of
process problems or due to the environment or borrower or both
• Banks need to tap insolvency knowledge to improve their process and
decision-making so that they can
– minimize exposure at the time of default
– recover as much as possible of that amount
• We can look at process and capital in terms of the modern credit risk
equation:
Expected Loss = Probability of Default X Exposure at Default X
Loss Given Default
. . . with focus on the last two terms
2
The “Credit Equation” Summarizes the Methodology
for Risk Measurement
Expected Loss (EL) =
Probability of Default (PD/ PoD) X Exposure at Default (EAD) X Loss Given Default (LGD)
plus correction factors for tenor and correlations.
EAD and LGD loom large in the arithmetic
3
LGD Is Particularly Important in Asia
•Problems of client information mean PD is hard to estimate . . .
•Collateral . . . which is a focus of LGD . . . Is a the center of the process
•LGD must therefore be managed very well
Collateral
Refinancing
Restructuring of
Business &
Management
4
Bank Credit Processes Lay Out the Drill for Credit
Portfolio Maintenance
Credit Policy &
Port-folio Management
Portfolio Acquisition
Marketing
Credit
Application
Credit
Approval
Documentation
& Disbursement
Monitoring
Workout
Credit Organization and Culture
The Basel Committee requires
•An Orderly, professional and imaginative process
•The measurement of risk and the use of measurement in all credit activities
5
The Credit Equation Ties Closely to the Credit Process
Process
Element
Credit Policy
Credit
Equation
Term
Uses all
Marketing
Based on
policy
Credit
Application
PD
Credit
Approval
PD
Documentation
&
Disbursement
Monitoring
Workout
LGD
EAD, PD
LGD
Feedback
•Process
•Credit Specifics
6
Capital Is Allocated Against “Unexpected” Losses
7
The Credit Equation is Used to Determine Credit
Capital by Way of “Risk Weights
Standardized Approach Mandates Capital in Accordance with Rated or Unrelated Risks,
incorporating all equation elements
Rating
AAA to
AA-
A+ to A-
Risk
Weight
20%
50%
Pillar I
Capital
1.6%
4%
BBB+ to
BB100%
8%
Below BB-
Unrated
150%
100%
12%
8%
The Internal Ratings Based Approach (IRB) produces a range of requirements based on
Expected Loss, with a mandated LGD (Foundation) or a calculated LGD ( Advanced)
Rating
AAA to
AA-
A+ to A-
BBB+ to
BB-
Below BB-
Unrated
Standard
Capital
1.6%
4%
8%
12%
8%
IRB
Capital*
1-2%
2-4%
4-11%
11-30%
NA
*Mandated LGD assumption in Foundation Process
8
Banks Build the Equation into Their
Credit Management System
•Some banks already used Expected Loss
•A much larger group uses “Risk Ratings” and some LGD assumption
•“Facility Ratings are another way to manage LGD and potentially EAD
•Ratings translate into action guidelines in risk management:
Low Facility
Rating, e.g.
High LGD
High facility
Rating, e.g.
Low LGD
Manage Risk,
Monitor client,
Collateral
Low EL -–
Manage returns
High Risk
Rating, i.e.,
Low PD
High EL Avoid
Manage Risk,
Monitor client,
collateral
Low Risk
Rating, i.e.
High PD
9
Basel Sets Requirements for Collateral Management. . .
•Legal enforceability
•Objective market value
•Frequent revaluation
•First claim
•Clear credit policy for collateral
•Appropriate liquidation analysis in
credit approvals
•Distinct operational unit to manage
collateral
•Adequate insurance
•Property monitoring, e.g., to
ensure taxes paid
•Environmental liability risk
management
10
. . . and LGD Estimation
Basel LGD Estimate Process
•Estimates have to be used for management
• “Track record” in using data for at least three years.
•Assessment of principal drivers supported by analysis
•Adequate time frame
•Historical experience/empirical evidence
•Adequate statistical base and analysis, with consistent
application and with appropriate validation procedures
•External comparisons
Collateral Analysis
•Dependence between borrower and
•Stress testing
collateral value
•Comprehensive view of “loss”
•Currency mismatch
•Consistency of default definition
•Conservative valuation/estimation of
workout period
•Key characteristics of borrower/facility/product
•Conservatism!!!!
•Country and industry factors
•Clear, consistent collateral policy
•Legal factors including insolvency regime
keeping in mind creditworthiness of
obligor
•Procedure for overrides of grades
•Robust
collateral management systems
•Collateral analysis
•Concentration monitoring/action
•Policies on appropriateness of collateral,
liquidation potential and revaluation
procedures
11
Basel Defines “Default” and “Loss”
Default
Per paragraph 272 of the final Accord document, a default occurs when one or more
of the following conditions obtain:
•It is determined the obligor is unlikely to pay its debt obligations
(principal, interest, fees) in full
•A credit loss even associated with any obligation of the obligor, such as
charge-off, specific provision, interest or fees
•The obligor is more than 90 days past due on any credit obligation
•The obligor has filed for bankruptcy of similar protection from creditors
Loss
“This should include the discount effects, funding costs and direct and
indirect costs association with collecting on the instrument in the de
termination of loss. Banks should not simply measure the loss recorded
in accounting records, although they should be able to compare the two”
(Basel Accord, Para 339).
12
LGD Studies Are Thin and U. S. Orientated
•Shortage of data and focus on bond markets and syndicated loans
•Many definitions and calculation algorithms, particularly “default” and “loss,”
are far from being agreed or implemented.
•Individual bank universes of defaults are much narrower
•Difference between studies of losses “when all is said and done” and losses
as measured by security prices subsequent to default
•Both EAD and LGD are very reflective of the bank credit process/willpower
•In Asia, there are wide differences by country.
13
S & P Has Developed “Loss Stats” Date
Type
General recovery
data unsurprising,
but 2003 a bad year
Available
Recovery
%
1988-2001
Recovery
%
19982002
Recovery
%
2003
Bank
84
74
72
Senior Secured Bond
69
46
29
Senior Unsecured Bond
52
37
21
Senior Subordinated Bond
35
21
NA
Subordinated Bond
30
15
NA
Junior Subordinated Bond
17
3
NA
Ultimate
Recovery %
Debt below (i.e., how senior?)
16
Collateral
13
Debt above
10
Aggregate default rate
7
Industry factor
7
GDP
0
Recovery heavily correlated to
structural financing factors
14
Fitch Produces Similar General Results, but Clarifies a Skewed
Distribution
Type
Loan Recovery %
Bond Recovery%
Senior Secured
72
62
Senior Unsecured
52
42
15
Moody’s Results Are Congruent but Lower
Rating Agency LGD Models
•
•
S & P Loss Stats
Moody’s LossCalc
• Fitch’s DART
16
Other Studies
•
Edward Altman of NYU’s Stern School has delved into LGD and
collateral management issues. papers.stern.nyu/~ealtman/
–
•
Altman and colleagues link recovery and default rates, which is currently not part of Basel
thinking.
A Bank of Italy group produced “Italian Banks Workout Activity: Costs,
Timing and Recovery Rates,” which was presented at third FAIR.
Principal conclusions:
–
–
–
–
–
–
Private agreements were much more important in case settlement than bankruptcy
proceedings/foreclosure
Recovery timing ranged from 6-7 years for bankruptcy/composition proceedings to around
two years for private agreements. Foreclosures take longer than proceedings based on
pledged securities. Foreclosure takes longer in southern and central Italy (PDS note: which
have environments closer to Asia than the north) than in northern Italy.
Annual recovery cost estimated at 1.2% of NPL’s
Average recovery “by 1999” (it was not clear the period covered but obviously fairly long
since the recovery periods were up to 264 months!) was 37% with considerable dispersion
amongst banks. (PDS: Bank differences would seem to be a key element for further study)
Short recovery periods led to higher recovery rates
Consumer recoveries were better than enterprise recovering; “producer families” were above
the mean
17
An RMA Paper Digs into Causes
Define:
LGD = (Charge-off - charge-off recovery) / Outstanding balance at default
α: The beta distribution’s center parameter and can be derived from equations below
β: The beta distribution’s shape parameter and can be derived from equations below
Min: Minimum of all cases
Max: Maximum of all cases
α and β are then derived from the following equations:
where μ, δ2 are population mean and variance respectively.
2. We then transform LGD from a beta to a normal distribution suitable
for use in OLS regressions, using the definitions of α and β as calculated above.
18
RMA Conducts Cooperative LGD Studies
Data Model
•Facility and Customer Number
•Type of Company
•Syndication indicator
•Country
•Facility risk rating
•Obligor risk rating
•Authorized limit
•Amount outstanding at default
•Spread index and per cent
•Industry
•Collateral
•Collateral value and evaluation frequency
•Unfunded risk protection information
•Credit mitigation produce
•Facility type
•Seniority
•Facility purpose
•Credit event
•Cash flow information
•Expense information
•Resolution event
Loss Calculations
There are very extensive breakdowns with in
categories. Methodology to calculate actual
economic losses is defined in some detail
as well as is the event of default, which for
purpose of this study includes:
•Past due
•Unlikely to pay
•Non accrual
•Credit loss
•Facility sale
•Distressed restructuring
•Bank-filed bankruptcy
•Obligor-filed bankruptcy
•Unknown (which please
minimize)
19
Asia Begins to Implement Basel
Country
Response on Pillar I
Singapore
The three banks will use IRB
Hong Kong
Big banks to use IRB, as they will anyway, flexibility for smaller
banks
Malaysia
Implementation over time. Standardized - 2008 and IRB
(voluntarily) - 2010
Indonesia
Basel-like framework introduced but details to come
Thailand
Appears will adopt Basel II but details still to come
Philippines
Standardized in 2007, IRB possibly by 2010
PRC/India/Sri
Lanka
Not hostile to Basel II but sticking to Basel I for now
20
RoC’s JCIC Pioneers Data Collection
Cash Flow
Recovery
Seniority
Collateral Type
Direct Costs
Legal/Trustee Fees
Accountant
Indirect Costs
Management fee
Duration of workout
Behavior
Opportunity cost
Moral hazard
Adverse selection
APR rule
Choice of auctioning
vs. restructuring.
Singapore banks work with S & P, results not in yet
21
Specific Asian Issues in LGD Study
•Poor historical data —it has been lost, was never created,
is difficult to locate and extract
•Mergers mean data is either lost or non-comparable
•Data on paper, not digitized
•Definition difficulties (which however can be resolved up to a point)
•Problems in defining what needs to be measured – some elements,
e.g., management willpower – are more measured in results which
are still hard to compare
22
Asia’s NPL Estimates Appear Too Low
Country
2002
Av.
NPL%
2003
Est.
Singapore
8
7-7
“Reliability ahead of region.”
Hong Kong
5
5-6
Same as Singapore
Malaysia
16
13-14
“Numbers fairly reliable” 16% on three months basis, disclosed by regulator for whole
industry although bank standard is six months.
Thailand
30
28-9
“Impaired assets 75% above official levels.” which were 16.5%. Loose rules on loan
rescheduling to avoid default, which is at 90 days.
Philippines
32
30
Similar to Thailand. includes large volume of real estate collateral owned by banks, not
easily saleable
Taiwan
13
10
“Impaired assets twice official levels” of 6.4%. Interest can be allowed to run for 180
days
20
46
incl. IBRA
15-17
“Impaired assets 150% above official levels” of 8%. Most bad loans were shipped off
to IBRA and the low number industry number reflects the low level of lending
23
30
Indonesia
India
Comments
Indian NPL rules set at 180 days although change planned for 2004.
Source: S & P Asia-Pacific Banking Outlook 2004
23
Asian LGD’s Vary A Lot by Country, but Are High
Relative to the U. S.
Country
Est.
LGD
Hong Kong
50
India
70
Indonesia
85
Malaysia
55
Philippines
75
Singapore
25
Taiwan
60
Thailand
70
Source: S & P Asia-Pacific Banking Outlook 2004
24
Many Factors Affect LGD Management
•
External
– Economic Environment
– Legal Environment
– Financial Markets
•
“Banking”
–
–
–
–
–
•
Structure of Facilities
Structure of Borrowers
Structure of Lenders
Documentation
Collateral
Management
–
–
–
–
Monitoring
Valuation & Collateral Management
Personnel and staffing
Decision-making and Willpower
25
An Informal “Expert Panel” Generally Confirms the List
EAD/LGD Factors
Importance for
LGD in Asia,
1998-2004
Too many or too diverse banks or too unstructured banking groups
10
No truly lead bank group
7
Local legal environment did not support collection/restructuring efforts
13
Deal documents missing
5
Deal documents poorly drawn
9
Security documents missing
5
Securities not properly registered
6
Security documents poorly drawn
7
Unrealistic collateral valuations
10
Security value is the same as the enterprise value, e.g., machinery, building project
5
Collateral deteriorated in liquidation value but this was not recognized
8
Collateral monitoring, especially property, did not take account of the risk of "boom" prices
6
Collateral concentrations
5
Deal structure too aggressive in terms of local law and regulations
8
Deal structure too complicated
8
Group financial structure too complex and hard to understand/control
10
Inadequate general credit monitoring allowed advances which could have been avoided
7
26
Recommendations for Action in Asia
1.
2.
3.
4.
5.
6.
7.
8.
Banks to be required to focus on LGD, data gathering
Regulators to set LGD coefficients
Regulators should use Basel definitions and standards
Regulatory papers to deal with LGD in depth, collateral and
recovery process
FAIR to recommend a Basel Committee treatment of “recovery”
Asian regulators to develop “Asian” treatment for collateral
Regulators and banks to collaborate to improved legal environment
FAIR to increase bank involvement in its activities
27
Philip D. Sherman
Senior Adviser in Asia
The Risk Management Association
[email protected] or [email protected]
Telephone: 65-6836-1297
Mobile: 65-9788-5001
28