The Basel Committee has proposed changes to reduce heterogeneity in banks' internal risk-based capital models, including removing some exposures from the internal ratings-based approach, setting floors for key model parameters, and setting an output floor. However, banks argue these floors will reduce risk sensitivity and increase capital charges. Alternatives proposed include harmonizing standards and a constrained internal ratings approach. The Basel Committee will meet in November to finalize the changes.

1. The system of floors for RWA
credit risk:
Issues and Challenges
Nov. 21, 2016
Hassan Omidi Firouzi
1

2. Proposed changes
by the Basel
Committee:
Removal of IRB for
exposures with
infrequent defaults;
Set IRB parameter
input floors;
Set output floors.
In March 2016, the BCBS published a consultation regulatory document setting a new
set of regulatory measures in order to reduce the heterogeneity of banks’ capital
charge using IRB models.
2

3. Following the introduction of Basel II, the banks are allowed to use IRB for credit risk and thus
regulatory capital charge calculation.
Types of IRB:
• Advanced IRB (A-IRB): Key risk parameters, PD, LGD, EAD, can be fully estimated using bank’s
internal models.
• Foundation IRB (F-IRB): Bank only estimates PD and the rest of parameters are prescribed by
the Basel Committee.
Using IRB approach results in heterogeneity of risk charges across
different regions as well as various nations.
But why?
3

4. RWA density = RWAs/(unweighted total assets). We refer to [4] for further discussion on
this.
IRB approach adoption over time
4

5. Reasons for heterogeneity of risk model outputs:
Variations in IRB
modelling
techniques;
Input data
differences;
Differences in
national/regional
standards and
local
implementation.
The role of the
validation process
in explaining
cross-country
discrepancies [1];
Up to 75% of the
variability in risk
weights for credit
risk is driven by
differences in
underlying risk
(e.g., in the
relative share of
different asset
classes) [2].
Reaction of the
regulatory body
to these
variations so far
is:
•Introducing the
leverage ration
within Basel III.
5

6. Deep into the BCBS proposals:
 Low default portfolios (LDP): LDPs are portfolios where risk-modelling techniques
are less advanced, the quantity and quality of relevant data may be limited, and
banks do not have an information advantage compared to external sources [4].
Mainly:
1) The BCBS proposes to remove IRB for loans to banks and financial
institutions with total assets more than €50bn;
2) In the case of corporates belonging to groups with revenues above
€200m, only the Foundation IRB approach would be allowed;
3) Other proposals have been proposed by the BCBS regarding the use of
IRB for CVA calculation.
6

7. Deep into the BCBS proposals:
Input floors [4]: For the sake of comparability between different internal risk models,
the BCBS proposes to set up input floors when estimating key parameters in credit risk
tracking. They are:
1) The PD for corporate and retail portfolios cannot fall below 0,05%, meaning
that at least 5 out of 10,000 obligors are expected to default every year
(0,10% for revolving credit cards);
2) The LGDs for unsecured exposures (where no collateral is available) are
being floored at 25% for corporates, 30% for retail exposures and 50% for
credit cards. Floors for secured exposures depend on the type of collateral
provided, ranging from 0% (in the case of eligible financial instruments) to
20% (for physical capital, such as machinery or inventories), with retail
mortgages set at 10%;
3) Banks will not be allowed to use internal EAD estimations for LDPs and for
non-revolving exposures.
7

8. Deep into the BCBS proposals:
Output floors [4]: The BCBS proposes to set a floor for the minimum regulatory capital
arising from IRB approach.
This floor is still under calibration but the BCBS expects that this floor will be a
percentage of the SA (standardized approach), something between 60%-90% of SA.
Yet to be determined.
But one-size-fits-all floors makes a right balance between risk sensitivity, simplicity
and comparability of models’ outputs?
8

9. Issues regarding these proposals (the fallacy of floors):
It turns out that the banks around the globe (mainly in Europe) are not happy with
these new resolutions; They believe these prescribed rules are not the real remedies to
deal with the heterogeneity in IRB models’ outputs. The main shortcomings of floors
highlighted by academics and practitioners are the following [3, 4]:
•Floors will result in reducing IRB model’s risk-sensitivity as they will be directly connected to SA;
•Increase in aggregate capital charges;
•By imposing floors across the board, supervisors would give banks an additional “risk budget”, which they
would use to absorb the extra costs generated by higher regulatory capital; hence, portfolios where the
actual risk content is below the floor would gradually shift towards more lucrative (and riskier) exposures;
•Output floors would refer to a new (and untested) standardized approach that the Basel Committee is still
developing: the latter should first be implemented, and put to work for some time, before it is used as a
benchmark to discipline internal models;
•Internal ratings have become a common language, whereby different structures and roles within a bank can
communicate horizontally as well as escalate risk vertically, up to the board of directors. Imposing floors
would lack credibility in the eyes of the banks’ top management;
•The new standardized approach (SA) would still relies heavily on external ratings, one may wonder whether
using it to constrain the IRB approach is fully consistent with the EU legislators’ desire to reduce mechanistic
regulatory reliance on credit rating agencies;
•Regulators may be unable or unwilling to set the right level of the IRB output floors.
9

10. We refer to [4] for further discussion on this.
Risk budget demonstration
10

11. Suggested alternative solutions
The following alternative solutions have been pointed out by the industry aiming at reducing the
heterogeneity of models’ outputs while keeping the right balance between risk sensitivity,
simplicity and comparability of models’ outputs [3, 5]:
 Harmonization of definitional issues and standards for modelling. A large part of RWA variation is driven
by diversity in both bank and supervisory practices.
 Constrained Internal Ratings Based approach instead of shift to SA.
 Banks propose to estimate parameters based on pooled data across institutions, implying a much higher
degree of harmonization.
 Promote smart supervision, model risk validation, and enhanced disclosure on risk methodologies.
Basel Committee will meet on 28-29 November 2016 (Santiago, Chile) to possibly finalize the
proposed changes upon reaching an agreement. EU is the main body that will be against these
changes.
11

12. References
1. Arroyo, J., Colomer, I., García-Baena, R., González-Mosquera, L. Comparing risk-weighted assets: the importance of
supervisory validation processes. Estabilidad Financiera 22, 9–29, 2012.
2. Basel Committee on Banking Supervision (BCBS). Regulatory Consistency Assessment Programme (RCAP) analysis of risk-
weighted assets for credit risk in the banking book. Bank for International Settlements, Basel, 2013.
3. Haselmann, R., Wahrenburg, M. Banks' internal rating models - time for a change? The system of floors as proposed by the
Basel Committee. European Parliament, November 9, 2016.
4. Resti, A. Banks' internal rating models - time for a change? The system of floors as proposed by the Basel Committee.
European Parliament, November 9, 2016.
5. Laurent, J.P., Omidi Firouzi, H. The best and the worst of VaR in a Basel III context. Available at:
http://laurent.jeanpaul.free.fr/making%20the%20best%20out%20of%20VaR%20in%20a%20Basel%20III%20context%205%20
September%202016.pdf, Sept. 5, 2016.
12