You are on page 1of 18

Big Data:

How to Turn Big Data Into


Great Information
Paulo Celio Di Cellio Dias
Luiz Henrique Outi Kauffmann
Guilherme Barreto Fernandes

N 86
January 2014

Business Cycles Benchmarking Basel III Downturn LGD:


and Credit Risk Parameters A More Conservative Approach
Joo Ricardo M. G. Costa Filho Carlos Antonio Campos Nogueira for Economic Decline Periods
Frederico Araujo Turolla 26 Mauro Ribeiro de Oliveira Jnior
19 Armando Chinelatto Neto
42
3

06

Big Data: How to Turn Big Data The concept of Big Data set a new threshold for the
Into Great Information data analysis process. Current technology enables
creating a massive volume of information based
Paulo Celio Di Cellio Dias on the behavior of a business firm or an individual,
Luiz Henrique Outi Kauffmann such as credit card transactions and cash/checking
Guilherme Barreto Fernandes account flows.

19

Business Cycles and Credit The paper uses econometric procedures to determine
the business cycle. The authors examine trends and
Joo Ricardo M. G. Costa Filho analyze their findings, generating a series of periods
Frederico Araujo Turolla for the Brazilian economys recent past, as well as an
estimate of potential GDP growth.

26

Benchmarking Basel III Risk Parameters The differences in regulatory capital allocated by Fi-
nancial Institutions (FIs) that have already migrated to
Carlos Antonio Campos Nogueira internal credit risk rating models are significant, even
for portfolios with similar risk mixes. The effects have
been negative for fair competition, given the impact of
the cost of capital on spread levels.

42

Downturn LGD: A More Conservative A relevant statistical correlation between the portfolios
Approach for Economic Decline Periods rate of default (RD) and its loss given default (LGD) was
identified in an exposure that the IRB approach cate-
Mauro Ribeiro de Oliveira Jnior gorizes as Home Mortgage for a major Brazilian finan-
Armando Chinelatto Neto cial institution.
26

Benchmarking Basel
III Risk Parameters

Carlos Antonio Campos Nogueira


27

Introduction

The issue of risk weight disparities:

The differences in regulatory capital allocated by Finan-


cial Institutions (FIs) that have already migrated to internal credit
risk models, particularly in Europe, have been significant, even
for portfolios with similar break-downs (risk mixes). Such discre-
pancies have obvious negative impacts on a fair competition re-
gime, given the impact of the cost of capital on the spread of va-
rious financial institutions.

Keywords: PD, LPD, IRB, RWA, Basel II, Basel III, Bench-
marking.

Financial institutions that made the Where:


migration to internal ratings based (IRB) I - Ki = credit risk weight factor asso-
credit risk models calculate the risk wei- ciated with the obligor or homogeneous risk
ghts of their assets (credit portfolios) based pool "i";
on formulas provided in the Basel Accord II - EADi = amount of the Exposure at
(June 2006 version) and reproduced locally Default parameter associated with exposure
in the Central Bank of Brazils (BACEN) re- i in connection with obligor or homogeneous
gulations, more specifically in Circular Let- risk pool "i"; and
ter N. 3.648, whose article 34 sets forth the III - F = factor defined in article 4, Re-
expression used to calculate the capital as- solution No. 4.193 of 2013.
sociated with the credit risk of FIs operating
under IRB, which is termed RWACIRB: A factor F of 8% as originally defi-
ned in the Accord is the one adopted by Eu-
Art. 34. The monthly amount of
ropean banks. IN Brazil, the factor varies
R
WACIRB must correspond to the result of the
over time (assuming maturing risk manage-
formula:
ment across the financial Industry in gene-

RWACIRB = K x EAD
i i
ral), according to the aforementioned article
4 of Resolution No. 4.193, which reads:
F
28

Art. 4 The minimum RW require- III PD = Probability of Default


ment corresponds to the application of the IV LGD = Loss Given Default
factor "F" to the RWA amount, where "F" V R = Correlation factor
equals:
Where R=0,04.
I - 11% (eleven percent), from Octo-
ber 1st, 2013 to December 31st, 2015; The formula is very similar for
portfolios in other exposure categories
II 9.875% (nine point eight seven and sub-categories, and the R correla-
five percent), from January 1st, 2016 to De- tion factor may be a constant, as in the
cember 31st, 2016; above example, or a function of PD.
Clearly, both the risk-weight fac-
III 9.25% (nine point twenty-five tors (K or, more generically, RW, for
percent), from January 1st, 2017 to Decem- Risk Weight") and consequently the
ber 31st, 2017; risk-weighted assets (RWAs) are direct
functions of the PD and LGD risk para-
IV 8.625% (eight point six two five meters. Therefore, the aforementioned
percent), from January 1st, 2018 to Decem- discrepancies in RW have their roots in
ber 31st, 2018; and the estimations of those parameters.
A benchmarking study was held
V - 8% (eight percent), from Janua- over the past three years by independent
ry 1st, 2019. organizations and international regula-
tors, particularly from Europe, led by the
The term RWA stands for Risk EBF (European Banking Federation), the
Weighted Asset, and the factor K is the FSA, the Banco de Espaa, the IMF, and
risk weight (RW) proper. The weight culminating with a recent Basel Commit-
factors formula varies according to the tee on Bank Supervision (BCBS) publi-
exposure classes to which the portfolios cation (bcbs256, July 2013). These stu-
migrating to IRB belong. For a credit card dies, despite certain differences in ap-
portfolio (qualified retail revolving sub-ca- proach (some of them focus on whole-
tegory), for example, the formula for K is: sale, sovereigns and banks, while others
address mortgage portfolios), conclu-
de that many of the regions banks have
{
K= LGDxN [ N-1(PD) + R x N-1 (0,999)
1-R
]-PDxLGD} been using such diverging RW estima-
tions that some allocate two to four ti-
Where: mes more capital than their competitors
I N = function of the normal accu- for very similar credit portfolios. Table 1
mulated distribution and the graph 1, developed respectively
II N -1 = reverse of the normal accu- by Barclays Capital and Banca DItalia,
mulated distribution function illustrate the high dispersion of RWs.
29

The RWA/TA graphic 1 (total risk- The graphic 2, developed by Ban-


-weighted assets divided by total assets, or co de Espaa (through its publication Es-
risk-weighting density) clearly shows a tabilidad Financiera, # 21) shows the va-
high variation between 15.1% and 78.1% of riation by exposure category and sub-ca-
the average RW. tegory.
30

One question that may emerge at BCBS survey), which eliminates any in-
an early stage of the analysis is that stu- fluence from obligor-specific risk diffe-
dies involving banks in several different rences (although differences in risk ari-
countries may be deeply affected by di- sing from the exposures mix remain).
fferences in legislation and business cul- In addition to the results of these
ture. But even after differences in cross- benchmarking studies, a top-down analy-
-border laws, standards and practices are sis (based on information disclosed as per
eliminated and a comparison is made in- Pillar III) of RW figures for the same por-
volving a single countrys financial institu- tfolio of a given bank has in some cases
tions, the differences persist, as seen in a been showing very wide changes betwe-
benchmarking study by the FSA (prede- en two consecutive years, going as high
cessor to the PRA) for British banks, so- as 20%.
wing significant differences (high disper-
sion) among PD values for corporation, Given the issue,
bank and sovereign portfolios, as shown what are its causes?
graphic 3. The causes, as noted earlier, in-
clude changes in the risk mix of exposu-
res to a single obligor, or to different obli-
gors in the same pool (for retail). This
involves differences in legal support
in various countries/regions to collec-
tion proceedings and collateral seizure.
Differences also exist between banks in
terms of the IRB migration status (rollout
stage). At some, more than 30% of por-
tfolios may still be treated according to
the standard approach (which increa-
ses average RW), while others stand at
less than 15%. However, according to
the BCBS survey, close to 25% of the di-
fferences lie associated with banks in-
ternal practices for the assembly of their
internal risk rating systems (IRRSs), whi-
The FSA study, as well as a more ch reflect mainly on the risk parameters
recent one by the BCBS, was based on PD, LGD and EAD, which, as seen, are
the HPE (Hypothetical Portfolio Bench- the determinant variables in RW and
marking Exercise) method, according to RWA calculations (along with EADs).
which a hypothetical portfolio (corpora- The graph 4 derives from the
tions, banks and sovereigns) is sent to EBF study on mortgage portfolios and
several banks with a request for a reply illustrates some of the elements with
providing risk-parameter information (PD the most influence on the portion of the
in both cases, plus LGD and EAD in the variation associated with banks inter-
31

nal practices, as well as those connec- time series used to calculate the PD and
ted with the portfolios intrinsic risk mix LGD risk parameters is more clearly
(Bars 1, 2, 4, 9, 10, 11, 12, 15). illustrated in the graphic 5, which also co-
One of the elements noted in the mes from the EBF mortgage portfolio sur-
graph 4 the extent and depth of the data vey and shows the great variation of the

Graphic 4: Qualitative Assessment of the Factors with the Most Impact on RWA Variation
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
1. Characteristics of
mortage markets

2. Legal framework for


mortages in the EU

3. Definition of
downturn LGD

4. Characteriscs of
housing markets

5. Criteria for cycle


adjustment in PD

6. LGD model
approach

7. Length and thickness


of data series

8. Definition
of defaut

9. Relevance of
the mortage market

10. Granularity of LGD


collateral segmentation

11.Treatment of
refinancing operations

12. Differences between


expected loss and provisions

13. Model approval and


model change policies

14. Portfolio segmentation


and master scale in PD

15. Imposed requirements on


scorecard and PD models

16. Use test acceptance


requirements
Source: EBF

Graphic 5: Historic Data Used to Calculate PD Across Banks

1986 1991 1996 2001 2006 2011


Source: EBF
32

time span covered by the data used in PD with the risk profiles of exposures and
models across different European banks. obligors.
The BCBS survey of corporation, However, for retail portfolios,
sovereign and bank portfolios indicates the contribution from LGD is even grea-
the following elements as the main cau- ter than that from PD in terms of explai-
ses of PD variations: ning the high variation of RWs, as seen
> Definition of the full economic cycle, in table 2 next, found in another sec-
which is fundamental to obtaining Long tion of the BCBS survey and also disclo-
Run PD (LRPD) and Downturn LGD; sed in BCBS256. In this case, the study
> Historic coverage of the database used performed a multivariate analysis of the
by the models; data gathered by the Capital Monitoring
> Different strategies selected to com- Group (CMG).
pensate for insufficient data in the time The parameters EAD and M (Ma-
series. For example, corporate portfo- turity) also contribute, the latter applying
lios frequently use external data, while to the banking segment only. Table 3,
banks prefer adjustment to PD calcula- next, is based on BCBSs HPE benchma-
tions for mortgage portfolios. rking study and succinctly shows the im-
> Definition of default. Some banks use portance of PD, LGD and M for RW varia-
different criteria, within the limits set tions, segregated by different exposure
by the Accord, to acknowledge default segments (corporation, retail, sovereigns
events. and banks): the relative importance of
> Varying methodologies used to perform PD and LGD differ from the previous
adjustments in obtaining Long Run PD table, which is based on a larger sam-
(LRPD). This cause is associated with ple. That is, by focusing only on obligors
the previously mentioned definition of with low default frequencies in the cor-
the full economic cycle. poration, banks and sovereigns catego-
> Aggregating data from different sour- ries, the importance of the intrinsic risk
ces to form the Referential Data Set on the obligor dimension as manifested
(RDS). Banks differ as to the relative in PD shows a different trend than in the
weights applied to external data sour- preceding table, with LGD contributing
ces with varying levels of association more than PD to variations in RW (remo-

Table 2: Relative significance of IRB parameters in explaining variations in RWA

PD LGD M
Corporations *** *
Retail ** *** n/a
Sovereigns ** *
Banks ** ** *

*** Highly significant ** Moderately significant * Some impact Impact not observed
33

Table 3 : Relative impact of variations in PD and LGD on RWs

Sovereigns Banks Corporations Corporations


(high end) (medium-sized)
With PD and LGD variations 1.10 0.43 0.22 0.23
PD variation removed 0.57 0.31 0.16 0.20
LGD variation removed 0.31 0.24 0.14 0.16

Standard deviation of RWs, computed for the banks taking part in the HPE. Numbers based on the deviations of
each banks individual medians relative to the overall median. For sovereigns, for example, the standard deviation
of the relative deviation of the medians is 110%.

ving LGD variation leaves a lower residu- rable risk parameters that would reduce
al standard deviation than removal of the their RWs and, consequently, the alloca-
PD element). ted capital.
Reactions varied, covering a
wide of radicalism. At the more revolu-
Having acknowledged the tionary end lie suggestions to comple-
problem and identified its tely abandon capital allocation based on
causes, how to proceed? internal RW calculations, retiring the IRB
Now that the Brazilian financial approach and adopting criteria based on
arena (and major banks in particular) is at fixed RW percentages (a return to the
an advanced stage in the preparation of standard approach) or on simpler indica-
IRB candidacies, the inevitable concern tors, giving up RWs and proposing allo-
arises that similar variations ma occur do- cation based on the leverage rate only
mestically, which may discourage the full (as Andrew Haldane suggests in The
use of internal credit risk rating models. dog and the frisbee).
The question is how to address However, experience shows that
the problem before those effects emerge. the pure and simple adoption of fixed
Worldwide, and particularly in Europe, RWs would be a methodological return
variations in RWA have generated con- to Basel I principles, with all the drawba-
cerns of all kinds and among all players. cks that this would imply, essentially en-
Investors worried about the soundness couraging banks to abandon risk mana-
of the capital allocated by the institutions gement and adversely selecting risky as-
in which they invested their funds. Banks sets with greater potential gains, as such
worried about the possibility of over-allo- a choice would not affect allocated capi-
cating capital (and becoming less compe- tal and its cost.
titive) or under-allocating it (incurring un- On the other hand, adopting cri-
covered risks). Regulators wonder if some teria based on a single indicator, such as
banks (including some of global syste- the leverage rate, would cause the same
mic importance) might be massaging problem by leading banks to work their
their models to obtain excessively favo- portfolio mixes so as to optimize that in-
34

dicator, even with high intrinsic risks at 102) provides basic guidelines on how to
systemically critical exposures, complete- implement benchmarking studies. Ac-
ly subverting the intended goal of adop- cording to the document, two main kinds
ting the proposed indicator (the so-called of benchmarking exist:
Goodharts law effect). One of them (called Type 1 in this
At the least radical end of the article) is used to identify outlier obligors,
spectrum lies the BCBS itself, whose comparing the PDs (or LGDs and EADs)
aforementioned bcbs256 proposes (p. of similar pools obtained from several FIs
10) carrying out benchmarkings to defi- (application is more appropriate for Pool
ne tolerance ranges for the variation of Data). In this case, the goal is not to test
PD, LGD and FCC (a constituent of EAD). the quality of the IRRSs, but the poten-
The proposal seems well balanced and tial inaccurate categorization of specific
capable of leading to more converging obligors (an input for the adjustment pro-
RWs (at least in terms of different prac- cess in pool formation).
tices, as opposed to the portfolios risk The other type (Type 2) uses ex-
mix). At the same time, it retains the sti- ternal ratings, scores or PDs (or LGDs
mulus for banks to keep up their invest- and EADs) for comparison with the pa-
ments and efforts toward improving in- rameters calculated in-house by internal
ternal models and managing the risk of risk rating systems (or IRRSs, the formal
their credit portfolios. name for the models and their scores-
and ratings-mapping methodologies) of
the FI in question. The goal, in this case,
But how to run the is to effectively check the quality of the
benchmarking studies? models used by FIs IRRSs.
Although the observation of RW In type 1, the benchmarks are
variations is relatively recent, the notion built by aggregating data from several
of using benchmarks is not novel within FIs. This may be done by an independent
the Basel II and III context. In fact, para- and trustworthy public- or private-sector
graph 532 of the Accord provides for the entity, including supervisors.
practice. BACEN circular letter No. 3.648, In type 2, the ratings, scores or
which sets forth the rules for IRB applica- PDs are usually provided by third par-
tions, translates the same view, as a re- ties, including those used by supervi-
commendation, in Article 153. It is true sors, which in turn develop their own,
that, in these cases, benchmarking is set occasionally with third-party assistance
within the context of the internal valida- (using data collected by those third par-
tion resources of each bank adopting IRB, ties or by the supervisors themselves).
rather than motivated by an attempt to re- Although type 1 is clearly what
duce RW variations across the various we have in mind as a tool to reduce RW
institutions present in the marketplace. variation, the techniques overlap and
The end goal, however, is the same: cali- we will therefore briefly approach both
brating risk parameter estimations. types, focusing on examples of PD ben-
BCBS Working Paper 14 (pp. 96- chmarking. We will also focus on ben-
35

chmarking examples for retail portfolios Cycle (TTC) type;


(proper and low-end SME, treated as po- > Make comparisons across similar po-
ols of retail obligors and exposures). Copo- ols, consolidated from data obtained
rations (high-end), banks and sovereigns from various FIs, without identifying the
methods have already been covered, par- obligors (see the discussion of the use
ticularly for portfolios with a low number of computer encryption techniques,
of default events (LDPs), using methods ahead). This will involve using cluste-
such as the one discussed by Markus Ri- ring techniques based on the contents
cke and Georg von Pfstl in Quantitative of certain variables that identify obligor
Validation of Rating Models for Low De- pools as opposed to customers (e.g.:
fault Portfolios through Benchmarking, geography, profession, income);
based on the use of the correlation coeffi- > Converging to a consistent rule for sco-
cient between score/rating scales. re/rating mapping based on a master
scale;
About type-1 benchmarking: > While obtaining outlier events, the
In this mode, the main challenges analysis must be supported by discri-
include: mination techniques in order to separa-
> Obtaining samples (individual and po- te the effects of methodology differen-
oled PDs) from various similar institu- ces from rating deviations.
tions so as to form a homogeneous base The steps involved in carrying
(peer group) that will serve for compa- out the benchmarking are:
rison purposes (benchmarking). 1. Obtaining the longest possible series
> Converging to a consistent mapping of of PDs (and DFs) individual and ave-
ratings to a master scale. rage and their independent variables
> While obtaining outlier occurrences, for several pools (to increase the inter-
the analysis must be supported by dis- section of retention periods for diffe-
crimination techniques to separate the rent FIs);
effects of different methodologies from 2. Obtaining pool formation rules whe-
classification deviations. rever possible (and disclosed by the
Because the main goal of type-1 banks).
benchmarking is to identify outliers, it is 3. Asking banks to exclude the edges
crucial to obtain data from several insti- (out of the diagonal) from backtesting
tutions. To do so, the appeal to FIs of pro- dispersion graphs, thereby obtaining
viding information must increase, and more homogeneous sub-pool data.
their fears of doing so must be reduced. 4. Consolidating and applying clustering
In addition, excessive complexity must be techniques to form the peer groups
avoided, as most FIs are still at the initial pools.
stages of model construction, particular- 5. Mapping each FIs average PDs against
ly in terms of their LGD and EAD models. a master scale of ratings.
Therefore, the focus must lie on: 6. Comparing each peer group pool (ins-
> Comparing only PD ratings, prefera- titutions with similar risk structures
ble through IRRSs of the Through-the obligors and exposure types) with each
36

individual FIs corresponding pool, and ting overlaps. The best method is to se-
reporting classification errors (% of lect the interval with the biggest possible
outliers in the dispersion matrix). overlap, as illustrated in the graphic 6.
7. The clustering must be periodically re- Obviously, the detailed execution
peated. of the techniques described above lies
Step five, which is perhaps the outside the scope of an article of an in-
most complex of all, involves techniques troductory and conceptual nature such
to enable mapping across different insti- as this, but the guidelines are essentially
tutions scales, which may pose a challen- those discussed here.
ge given the differences in number of ra- Next follows an analysis of the
ting levels, PD intervals within each level, distribution of individual obligors PDs
and the delicate treatment of the resul- from the pools located within the over-

Graphic 6 : Rating Scales A and B

Rating Scale A
Rating A1 A2 A3 A4 A5 A6 D
Average PD 0.10% 0.25% 0.70% 1% 5% 10% 20%

Rating Scale B
Rating B1 B2 B3 B4 B5 B6 B7 B8 B9 D
Average PD 0.05% 0.08% 0.15% 0% 0% 1% 2% 4% 12% 20%

Relative impact of variations in PD and LGD on RWs

A1 A2 A3 A4 A5 A6 D
0.1% 0.25% 0.7% 1.0% 5.0% 10.0% 20.0%

G1 G2 G3 G4 G5 G6 D
0.1% 0.25% 1.0% 2.0% 5.0% 12.0% 20.0%

B1 B2 B3 B4 B5 B6 B7 B8 B9 D

0.05% 0.08% 0.15% 0.2% 0.4% 0.8% 2.0% 4.0% 12.0% 20.0%
37

laps, which give an idea of how much the tially overlaps anothers, as illustrated in
FIs diverge or converge, as illustrated in the graphic 8).
the graphic 7.
A comparative analysis of the ave- About type-2 benchmarking:
rages within each overlap interval must The challenges in this form of
be accompanied by a dispersion analysis benchmarking include:
based on standard errors to identify po- > Selecting the source that will provide
tential phase shifts (where a FIs pool par- the benchmark: similarity among the

Graphic 7

12

78

10
Ratings

10

PDs
79

11
8

83

Graphic 8
30

25

20

15

10

5
S1
0
1,90%

1,75%

1,60%

1,40%

1,35%

1,20%

1,15%

1,10%

1,05%

1,00%
38

assumptions used in different IRRSs is > Developing a pre-analysis of the diffe-


important, as are the definitions of late rences.
payment, default and loss. For ratings The comparison must be repea-
scales, the underlying philosophy (PIT ted periodically and the dispersions evo-
or TTC) is also important. For PDs, whe- lution over time must be observed and
ther they are dynamic or static. explained.
> Defining a rule to match (map) the inter- The results are plotted on a dis-
nal models estimations against the ex- persion matrix that, in general terms,
ternal benchmark (particularly necessa- will show some deviation from the dia-
ry for pooled PDs). gonal (which would be the only comple-
To perform this kind of benchma- ted part in the matrix if a methods PDs
rking, the recommended general steps were exactly the same as those obtained
are: by means of the other). Typically, this will
> Concerning the peer groups formed for produce a deviation like the one shown
type 1, the comparison may be drawn next graphic 9.
against the suppliers own models (po- The benchmarking studies de-
tentially with the inclusion of proprietary veloped by the BCBS and FSA using the
independent variables); HPE approach are both type 1, but res-
> Obtaining the FIs stressed PDs for the tricted to corporations, banks and so-
pools that made up the peer groups; vereigns portfolios, which are typically
> Comparing the PDs generated by the LDPs. This involves a small volume of
suppliers model with those of the FIs, data that can be processed in any IT
generating dispersion matrices; structure, even a laptop computer.

Graphic 9

Lowest Risk
Internal PDs

Highest Risk

External PDs
39

For retail, however, benchmarking gor and exposure belong, but without re-
involves vast data volumes and the de- vealing them (preserving each banks
mand on IT resources places such efforts methodology, such as mean, median,
beyond the reach of most supervisors, whi- etc). Finally, the banks also send each
le business firms (e.g. credit bureaux) are obligors or exposures PD, LGD and FCC,
far better equipped and prepared to ga- unencrypted in this case, as the Bench-
ther, process and analyze high volumes of marker will need to group and compared
data, and may greatly assist in this initiati- them.
ve. However, one problem that may emer- 3. The Benchmarker receives the data and,
ge from this possibility is the potential re- without being able to identify either obli-
luctance from banks to send information gors or operations (not having been pro-
to third parties, even if it only includes the vided with the Security Managers sym-
identification of the obligor and exposure metrical encryption key), groups, sum-
and the respective PD, LGD and EAD. marizes and sends the analyses back to
One way to navigate this problem the banks (comparing each one against
and ensure banking secrecy is to use a the average of the peer group of compa-
third party with notarial authentication po- red banks). In addition (or alternatively),
wers (which this article will refer to as se- the Benchmarker may send the supervi-
curity manager) and computer encryption sor dispersion graphs of the risk parame-
techniques, as follows: ters together with the respective analyses.
1. Before gathering the data for benchma- With such a solution, information
rking purposes, the security manager is segregated in such a manner as to pre-
provides the participating banks with serve secrecy and confidentiality, since:
symmetrical algorithm encryption keys The Security Manager only issues
(the same key for all banks, referred to symmetrical encryption keys for AES- or
as C), or hashing algorithm seeds, digi- Triple DES-type algorithms, for example.
tally signed with the security managers Alternatively hashing algorithms (MD5 or
digital certificate. SHA-2) may be used, for which the mana-
2. The banks use this key (or seed) to en- ger would only send the banks the applica-
crypt obligor identities (Taxpayer IDs of ble root (or seed). Therefore, the Security
the consumers and SMEs treated as re- Manager receives no information on obli-
tail customers) and exposures (con- gors, exposures, or even data associated
tracts). They then send the data to the with each banks strategy (the intermedia-
organization charged with gathering te values of PD. LGD or EAD used to defi-
the data and analyzing the results of the ne obligor and exposure pools).
benchmarking study, which we will call The Benchmarker receives suffi-
Benchmarker. Together with the iden- cient information to perform comparisons
tification of the obligor or exposure, the and analyses, but, lacking access to the
banks also send the encrypted interme- symmetrical encryption key (or hashing
diate PD, LGD and FCC (which is part of seed), is unable to identify/recognize obli-
EAD calculations) for each pool, in order gors (Taxpayer IDs), or even the interme-
to characterize the pool to which the obli- diate pool PDS, LGDs and FCCs (which re-
40

flect banks strategies). marking has become an imperative marke-


The graphic 10 below illustrates tplace need in the sense of supporting and
the overall process: providing inputs to the various financial
institutions models calibration, making the
Conclusion financial market more fair and competitive
Although originally conceived sim- while maintaining the use of internal credit
ply as an internal validation tool for banks risk models, according to the rules and gui-
applying for IRB, Risk-Parameter Bench- ding principles set forth in Basel III.

Graphic 10

Security Manager
Encryption
key C
Obligor and exposure IDs + pool PDs, LGDs
and FCCs, all encrypted using C. Unencrypted
(plain text) individual PDs, LGDs and FCCs

Banks Benchmarker

Bank x peer average


comparsions and analysis
Dispersion graphs
and analysis of peer group

Supervisor
Author

Carlos Antonio Campos Nogueira


Has a degree in Physics (PUC - RJ) and a Masters Degree in Astrophysics and Elementary Particles (Centro Bra-
sileiro de Pesquisas Fsicas - CBPF/CNPQ). A practice leader in the IT consulting area, he has more than 30 ye-
ars continuous experience, providing services to more than 40 companies in close to 70 projects and conduc-
ting Basel II-related projects for the past seven years. Managing and founding partner of IntelliSearch. E-mail
can@intellisearch.com.br
41
References

European Banking Federation - Study on Internal Rating Based (IRB) models in Europe (source for graphics 4 e
5). http://www.ebf-fbe.eu/uploads/Study%20on%20Internal%20Rating%20Based%20(IRB)%20models%20in%20
Europe.pdf

Lesl, Vanessa Le and Avramova, Sofiya - International Monetary Fund (IMF) - Revisiting Risk-Weighted As-
sets Why Do RWAs Differ Across Countries and What Can Be Done About It?, Maro de 2012 (source for table 1
and graphic 3). http://www.imf.org/external/pubs/ft/wp/2012/wp1290.pdf

Working Paper N. 14, Studies on the Validation of Internal Rating Systems Basel Committee on Banking Su-
pervision (BCBS), Maio de 2005. (source for graphic 4). http://www.bis.org/publ/bcbs_wp14.pdf

Ricke, Markus and Pfstl; Georg von- Financial Stability Report Quantitative Validation of Rating Models for
Low Default Portfolios through Benchmarking. http://www.oenb.at/en/img/fsr_14_special_topics_04_tcm16-76882.
pdf

Analysis of risk-weighted assets for credit risk in the banking book - Regulatory Consistency Assessment Pro-
gramme (RCAP) Basel Committee on Banking Supervision (BCBS), Julho de 2013. (source for tables 2 e 3). http://
www.bis.org/publ/bcbs256.pdf

Ledo; Mayte Estabilidad Financiera n21 Banco de Espaa - Towards More Consistent, Albeit Diverse, Risk-
-Weighted Assets Across Banks. (source for graphic 2). http://www.bde.es/f/webbde/Secciones/Publicaciones/In-
formesBoletinesRevistas/RevistaEstabilidadFinanciera/11/ref0321%20.pdf

Cannata; Francesco, Casellina; Simone and Guidi; Gregorio, set. 2012, Inside the labyrinth of Basel risk-
-weighted assets: how not to get lost, Banca DItalia, Occasional Papers, n 132 (source for graphic 1). http://
www.bancaditalia.it/pubblicazioni/econo/quest_ecofin_2/qef132/QEF_132.pdf

Circular n 3.648 Banco Central do Brasil Bacen http://www.bcb.gov.br/pre/normativos/circ/2013/pdf/


circ_3648_v1_O.pdf

International Convergence of Capital Measurement and Capital Standards, A Revised Framework Comprehen-
sive Version Junho de 2006 http://www.bis.org/publ/bcbs128.pdf