Executive summary
The financial crisis that began in August 2007 has revealed important gaps in statistics on
credit risk transfer (CRT) instruments. In particular, information on structural changes in
global CRT markets and on the transfer and ultimate distribution of credit risk has not been
sufficiently comprehensive or timely.
This report explores how data on CRT collected under the auspices of the CGFS could be
enhanced. One main focus was to be on expanding the coverage of credit default swap
(CDS) instruments to gain a better understanding of the structural changes in global CRT
markets, as well as obtaining better information on the transfer and ultimate distribution of
credit risk.
The proposed extended CDS reporting template takes into account the usefulness of new
data for analysis and the need to minimise the burden on reporting agents. This was
achieved via a two-stage merits and costs consultation process. A questionnaire was first
sent to member central bank and official sector analysts to evaluate the benefits of a set of
possible improvements to CRT statistics. On the basis of the results of this evaluation, the
proposed changes were streamlined and sent to reporting agents for another round of
consultation. Based on the outcome of this exercise, this report proposes the following short-
term and longer-term changes to the existing CDS reporting.
On the basis of their high degree of usefulness to analysts and low reporting costs, two items
have been identified as candidates for quick implementation, possibly to be first implemented
in the 2010 BIS Triennial Survey of Foreign Exchange and OTC Derivatives Markets:
• a new counterparty field of central counterparties (CCPs) – a priority item; and
• index CDS as a new “reference entity” – an encouraged item.
With a view to improving the consistency of data across reporting countries, a list of qualified
CCPs will be issued to reporting agents. Separately, in order to improve the identification of
counterparties, reporting agents will also be asked to record contracts with hedge funds
using the European Union’s definition of hedge funds as a reference.
To allow reporters enough time to prepare for more complex changes, an extended template
incorporating the recommendations listed below will be proposed to the CGFS for full
implementation by June 2011, which would allow the first set of new data to be published in
October that year:
• regional counterparty breakdowns to be recorded of the total outstanding amounts
bought and sold for all CDS contracts, and a list of counterparties and their
geographical location to be included in the new guidelines;
• CDS on asset-backed securities (ABS) to be introduced as a new reference entity
under the subcategory of portfolio or structured products, with implementation
subject to further work on what types of ABS should be included and a clear
definition being made available to reporters;
• in the spirit of the reporting of other non-CDS derivative instruments, net market
values based on the BIS guidelines for regular credit default swap reporting to be
added; and
• reporting agents to be asked to also report the total amounts of synthetic
collateralised debt obligations (CDOs) being bought and sold (ie without any
geographical or counterparty breakdowns).
The report also reviews the potential for using the US Depository Trust and Clearing
Corporation (DTCC) global CDS data to supplement BIS data for the purpose of monitoring
market developments. Initial results suggested that DTCC data captured a significant part of
global markets between reporting dealers but not with non-dealers. Given that the DTCC is in
CGFS – Credit risk transfer statistics
1
the process of improving its records on non-dealers’ transactions, the report recommends
that further comparison exercises be conducted for end-June and end-December 2009 BIS
data. A review of central bank needs for additional breakdowns could also be communicated
to the DTCC by the end of 2009.
Apart from DTCC data, the report also discusses linkages between BIS consolidated banking
statistics, the BIS Triennial Survey and semiannual OTC data. It finds that the BIS
consolidated banking data could be used to gauge a country’s overall derivatives exposures
to foreign counterparties. Furthermore, the dataset could also help gauge credit risk
exposures vis-à-vis other countries or regions. Given that the BIS Triennial Survey has a
larger reporting population than the semiannual survey, the BIS could explore whether the
Triennial Survey could assist in identifying changes in the market, such as a possible greater
involvement of insurance corporations, so as to consider in due time whether a more regular
monitoring would be useful.
The report was approved by the Committee on the Global Financial System at its meeting on
26 June 2009. The recommendations of the Working Group were endorsed and are being
implemented within the schedule outlined in Section 5.
2
CGFS – Credit risk transfer statistics
1. Introduction
The financial crisis that began in August 2007 has revealed important gaps in statistics on
credit risk transfer (CRT) instruments. In particular, information on structural changes in
global CRT markets and on the transfer and ultimate distribution of credit risk has not been
sufficiently comprehensive or timely. The Committee of the Global Financial System decided
in September 2008 to establish a Working Group chaired by Jean-Marc Israël of the ECB to
review CRT statistics (see Annex 1 for the mandate of the Group).
The Working Group was asked to explore how data on CRT collected under the auspices of
the CGFS could be enhanced. One main focus was to be on expanding the coverage of
credit default swap (CDS) instruments to gain a better understanding of the structural
changes in global CRT markets, as well as obtaining better information on the transfer and
ultimate distribution of credit risk. This included examining ways to improve information on
counterparty risk and exposures to various reference entities, and expanding the reporting to
collect details on increasingly popular instruments such as index CDS contracts.
In assessing the usefulness of possible revisions to CRT statistics, the Group was asked to
take into account the reporting burden and the relationship with other statistics. This was
achieved via a two-stage consultation process. First, the Group surveyed member central
bank and official sector analysts about their “wish list” of possible improvements to CRT
statistics. Based on the results, the proposed changes were put forward to reporting agents
for further consultations. The input received helped the Group draw up its recommendations.
Furthermore, to avoid duplication, the Group also considered existing data and initiatives to
collect data on CRT under way at other official and private institutions, and evaluated the
potential usefulness of these alternative data sources in the monitoring of CRT market
developments.
This report is organised as follows. Section 2 discusses the results of the merits and costs
exercise on reviewing CRT statistics reporting; and recommends possible changes to the
current reporting template. Section 3 compares the CDS data published by the Depository
Trust and Clearing Corporation (DTCC) with the BIS data to see whether the weekly
available DTCC data can supplement BIS data for the purpose of monitoring the
developments in CDS markets. In Section 4, linkages of the semiannual over-the-counter
(OTC) derivatives statistics with the BIS consolidated banking statistics and the Triennial
Central Bank Survey are discussed. Section 5 summarises the recommendations.
2. Review of CDS statistics reporting
In reviewing the reporting of CDS statistics as an important focus of CRT, the Working Group
sought to facilitate better analysis of the credit derivatives markets by making proposals to
improve data transparency, and at the same time, not overburdening reporting banks with
data requests. The Working Group thus conducted a merits and costs exercise with analysts
and respondent banks to help identify gaps in statistics on credit risk transfer instruments
and areas for possible improvement. The exercise was organised as a two-stage process.
First, a questionnaire was sent to users in central banks and other official institutions to
evaluate the benefits of some proposed enhancements to the current CDS statistics
CGFS – Credit risk transfer statistics
3
reporting.
1
The questionnaire comprised a set of qualitative and quantitative questions
(Annex 2), the quantitative ones asking users to rank on a scale of 1 to 3 the usefulness of
the proposed changes (Table 1).
Second, based on the feedback received from users, the proposed enhancements were
streamlined. A further questionnaire with the new proposed enhancements was then sent
through Working Group members to their reporting agents for cost evaluation (Annex 3).
Reporters were asked to provide an estimate of the implied costs, on a scale of 1 to 3, of
both development and running costs (Table 1).
2
This section discusses the outcome of the
exercise and proposes some changes to the current reporting template (see Annex 4 for
more detailed responses to the questionnaires).
Table 1
Merits and costs, and sampling populations
1 2 3
Merits to users Limited importance Fairly important Crucial
Costs to reporters Low cost Fairly costly Expensive
2.1 Geographical breakdown of CDS transactions
A geographical breakdown of CDS transactions by counterparty and/or reference entity
would allow analysts to identify how much credit risk is being transferred between countries
and regions as well as the concentration of risks across countries. The Working Group
proposed five options to record these counterparty and reference entity geographical
breakdowns by “domestic versus foreign” or by region/country (Table 2).
3
These options
apply only to the notional amounts outstanding of all CDS contracts bought and sold.
Users found option 4, with regional counterparty and domestic versus foreign reference entity
breakdowns, to be the most useful with an average score of 2.1 (ie very important, Table 3).
Next came options 3 (with regional counterparty breakdown) and 5 (with regional
counterparty and regional reference entity breakdowns). The first two options, which record
domestic versus foreign breakdowns, were considered by users to be the least useful.
1
The questionnaire was completed by users at 10 central banks – Reserve Bank of Australia, European Central
Bank, Bank of France, Deutsche Bundesbank, Bank of Italy, Bank of Japan, Bank of Korea, Bank of Spain,
Swiss National Bank and Federal Reserve Board – and at the IMF and BIS.
2
Reporting agents in 10 countries – Australia, France, Germany, Italy, Japan, Korea, Spain, Switzerland, the
United States and the United Kingdom – took part in the survey.
3
The presence of only a few reporting dealers in most countries other than Japan and the United States means
that adopting a country breakdown might potentially reveal some confidential information about individual
banks’ operations. The regional breakdown was proposed to address this confidentiality issue.
4
CGFS – Credit risk transfer statistics
Table 2
Geographical breakdown options
Option Counterparty breakdown Reference entity breakdown
1 Domestic versus foreign None
2 Domestic versus foreign Domestic versus foreign
3 Region
1
None
4 Region
1
Domestic versus foreign
5 Region
1
Region
1
1
Includes: Japan, the United States, western Europe (the EU 15 countries prior to 2004 and Switzerland),
Latin America, other Asian countries and all other countries.
Table 3
Geographical breakdown: merits and costs
1
Average costs
Setup up Running
Option
Average
merits
Work-load IT Work-load IT
1 1.3 1.6 1.5 1.4 1.4
2 1.5 2.2 2.2 1.9 1.6
3 1.7 2.0 2.0 1.8 1.8
4 2.1 2.4 2.4 2.2 2.1
5 1.7 2.4 2.4 2.3 2.1
1
Simple average of summary responses. See Table 1 for the scale of scores.
Reporting banks on average considered options 4 and 5 the most costly in terms of both
development and running, followed by option 2. Option 3 was thought to be less costly to
develop and run than option 2 but more costly than option 1. According to some reporters,
reference entity data are in general fairly costly to compile, and providing a geographical
breakdown would be challenging. This might explain the relatively low estimated costs for
options 1 and 3.
4
Furthermore, the Depository Trust and Clearing Corporation (DTCC) data
could be used to extract geographical information on protection bought and sold on single-
name reference entities (see Section 3).
4
However, a few reporters thought that the costs of introducing a regional counterparty breakdown outweighed
its merits. Furthermore, should option 3 or any options with an extended geographical counterparty breakdown
be adopted, it would be important to provide a list of these counterparties to facilitate their classification by
reporters.
CGFS – Credit risk transfer statistics
5
In the light of the merits/cost benefit assessment, the Group proposes to adopt option 3, ie
to expand the current template to record a regional counterparty breakdown of notional
amounts outstanding of all CDS contracts bought and sold; a list of counterparties and
their geographical location should be drawn up. Users’ call for a geographical breakdown
by reference entity might instead be met using data from the DTCC.
2.2 Counterparty breakdown
In view of the increasingly important role of central counterparties (CCPs) in the CDS market,
the Working Group proposed including a new counterparty field for positions with CCPs. This
proposed item was considered by users as close to “crucial” with an average score of 2.7
(Table 4), while a majority of reporting agents regarded the addition as not particularly costly.
Before introducing CCPs into the new template, however, it will be clarified whether CCPs
are to be recorded as sole counterparties in CDS trades or whether the “direct” counterparty
as well as CCPs are to be recorded.
Another proposal on the counterparty breakdown was to split securities firms and banks –
treated as a single group in the current template – into two separate counterparties.
Some users thought that this might improve the understanding of the specific role of the
banking sector in the CDS market; others argued that banks and securities firms should be
treated differently as they come under different regulatory frameworks. However, on average,
the merits of implementing this were ranked as less than “fairly important” whilst incurring
fairly significant setup costs (Table 4).
The Group recommends the introduction of CCPs as a new counterparty filed in the CDS
reporting template.
The Group agreed not to propose separating securities firms and banks.
Table 4
Counterparty breakdown
1
Average Average costs
merits Setup Running
A new counterparty field for central counterparties 2.6 1.8 1.5
Separating securities firms from banks 1.6 2.0 1.8
1
Simple average of summary responses. See Table 1 for the scale of scores.
2.3 Counterparty definition
To enhance the comparability of data across reporting countries, the Working Group
identified two potential areas for improvement that are related to the counterparty definitions.
First, what can be classified as hedge funds? In general, most reporters welcomed any
initiatives to improve the reporting guidelines and definitions of reporting. Some non-EU
reporters noted that the EU definition of hedge funds laid down in Guideline ECB/2007/9
(see Annex 2) would be useful as a reference and are willing to refer to it in future reporting
on a best efforts basis. Some reporting agents added that they would greatly appreciate a list
of hedge funds being attached to the reporting forms.
6
CGFS – Credit risk transfer statistics
The second issue is how to accurately record transactions with insurance companies. Market
sources reveal that insurance companies are important participants in CDS markets, yet the
BIS OTC derivatives statistics indicate otherwise. One possible explanation is that in some
countries insurance companies are not allowed to engage in derivatives transactions directly
and instead do so through affiliates. Reporters were asked whether it would be feasible to
“look through” these affiliates’ CDS positions and report them as positions with insurance
companies as counterparty. A majority of reporters noted that it would be a very difficult task
and costly to implement and therefore would recommend not to adopt this reporting practice,
at least as long as there are doubts over the actual extent of this sector’s involvement in the
CDS market. With a view to monitoring this possible involvement, the Working Group
proposed including a related question in the BIS Triennial Survey.
The Group agreed to use the EU definition of hedge funds in the reporting guidelines as a
reference, possibly accompanied by a list of hedge funds in the reporting countries.
Regarding insurance companies, the Group decided not to put forward the proposal to
“look through” transactions conducted by affiliates due to the difficulty cited by reporting
agents. Further work may be needed to develop the notion of counterparty from that of a
monolithic entity into a concept that differentiates between the legal entity that engages in
the transaction and the ultimate obligor.
2.4 Index CDS
The rapidly growing importance of index products in the CDS market in recent years
suggests that the segment might warrant closer monitoring. Currently, index products are
recorded under multi-name instruments. The Working Group proposed five options to
enhance the reporting of index CDS. The first four options treat index CDS as a subset of
multi-name instruments with various levels of detail. The first option would be to record only
the total notional amounts bought and sold for all index products (A in Table 5). The second
and third options propose recording those amounts for all counterparties (B) and all
reference entities (C), respectively. The fourth option involves recording all counterparty and
reference entity breakdowns of index CDS (D). Finally, given the potential difficulties of
classifying index CDS contracts by rating, by maturity and by sector, the Group suggested
adding index CDS as a new “reference entity sector” (E) as an alternative to treating them as
a subset of multi-name instruments.
Table 5
Index CDS options
Index products Total Reference entity
bought and sold By rating By maturity By sector
All index products A C C C E
By counterparty B D D D E
On average, users ranked the recording of counterparty breakdowns of index CDS (B) or the
recording of index CDS instruments in an additional reference sector (E) as having the
highest merit (Table 6). While reporting agents considered the latter option as slightly more
CGFS – Credit risk transfer statistics
7
costly in setup terms, it would incur lower running costs.
5
Furthermore, DTCC data already
provide counterparty breakdowns of index CDS instruments.
Table 6
Index CDS: merits and costs
1
Average costs
Setup up Running
Option
Average
merits
Work-load IT Work-load IT
Table 5 – A 1.9 1.6 1.6 1.3 1.2
Table 5 – B 2.1 1.9 2.1 1.9 1.8
Table 5 – C 1.4 2.4 2.4 2.1 2.0
Table 5 – D 1.7 2.5 2.6 2.2 2.1
Table 5 – E 2.1 2.1 1.7
1
Simple average of summary responses. See Table 1 for the scale of scores.
On the basis of these considerations, the Group agreed to recommend the recording of
index CDS in a new “reference entity sector” in the extended reporting template.
2.5 Asset-backed securities
Another market segment that has grown rapidly in recent years is CDS on securitised
products such as CDS on asset-backed securities (ABS) and mortgage-backed securities
(MBS) and CDS on collateralised debt obligations (CDS on CDOs).
6
The Working Group
proposed adding a subcategory of CDS on securitised products (“ABS”) under the portfolio or
structured products in the extended template. While this proposed new item received
considerable support from users (with an average score of 2.2), it incurs relatively high setup
cost (Table 7). Some reporters noted that identifying such trades on a consistent basis is
difficult as it requires considerable effort to examine the details of underlying securitised
instruments, particularly CDOs. Meanwhile, some users noted that CDS on securitised
products with pure asset-backed securities as underlying credit (ABS and MBS) could be
viewed as an indicator to gauge exposures to “households”.
5
This partly reflects the fact that this option requires reporters to record only the total outstanding amounts
bought and sold, but not for single-name and multi-name instruments separately, which may make it easier to
handle.
6
The standard documentation of the International Swaps and Derivatives Association (ISDA) for CDS on
securitised products is currently available in two ISDA forms of CDS designed for “pay as you go” settlements:
“CDS on ABS and MBS” and “CDS on CDOs”.
8
CGFS – Credit risk transfer statistics
Given the complexity of implementation and potential merits, the Working Group proposes
to introduce “ABS” as a new reference entity under the subcategory of portfolio or
structured products, with implementation subject to further work on what types of ABS
should be included and a clear definition being made available to reporters. If further work
were to suggest that the costs have outweighed the merits, this item would be withdrawn.
In the meantime, the definition and reporting instructions should be further elaborated, if
possible, by end-2009.
Table 7
Asset-backed securities
1
Average Average costs
merits Setup Running
ABS as a new reference entity 2.1 2.3 2.0
1
Simple average of summary responses. See Table 1 for the scale of scores.
2.6 Net market values
The Working Group also proposed adding a new field for net market values alongside the
gross market values in the current reporting template. Two methods of deriving net values
are considered. The first follows the BIS guideline on semiannual OTC derivatives, it calls for
the market value of claims and liabilities to be netted when they are claims on and liabilities
to the same counterparty and both the reporting institutions and the counterparty have a
valid, legally enforceable netting agreement. According to the users, this BIS definition would
be a useful measure to gauge counterparty credit exposure. A second approach focuses on
the credit risk of particular reference entities. For example, the DTCC publishes net notional
amounts outstanding of top 1,000 reference entities. The DTCC definition of netting the sum
of the notional values of protection bought by net buyers with respect to any single reference
entity could be borrowed to derive the net market values of particular reference entities.
Overall, both of these options were ranked as “fairly costly” by reporting agents although the
BIS definition was thought to be less burdensome in terms of both setup and running
(Table 8). However, the responses varied considerably. Some reporters said that the netting
of market values according to the BIS guideline is already in their systems as the BIS
definition is consistent with local accounting and regulatory rules.
7
But others said that they
collect only gross values at present, so gathering the desired information would be a costly
exercise. In other cases, reporters have been recording the net present values associated
with every trade in their system but not applying netting by counterparty.
7
In some cases, reporters calculate net values by the same counterparty by netting all positions in financial and
credit derivatives.
CGFS – Credit risk transfer statistics
9
The Working Group recommends adding the BIS definition of net market values. This
would be in the spirit of the reporting of other BIS semiannual OTC derivatives statistics.
Because, in practice, counterparty netting applies at the level of a given master
agreement and not at the level of the instrument (type of derivatives contract), it was
noted that this statistic will only be a rough proxy for the values that would actually be
settled in a netting event. The issue of netting methodology might warrant further work in
the near future.
Table 8
Net market values
1
Average costs
Setup Running
Applying to counterparty (BIS definition) 2.0 2.0
Applying to reference entity (DTCC definition) 2.4 2.3
1
Simple average of summary responses. See Table 1 for the scale of scores.
2.7 Other credit derivatives
Apart from the CDS market, the Working Group also suggested collecting the total notional
amounts of contracts bought and sold for four other credit derivatives instruments: synthetic
CDOs, forwards, swaps and OTC options. Among these four instruments, users found the
additional reporting on synthetic CDOs to be the most useful with an average score of 2.1
(Table 9). However, that form of reporting also had the highest estimated setup costs due to
the complex structures of these instruments.
Table 9
Other credit derivatives: merits and costs
1
Average costs
Setup up Running
Option
Average
merits
Work-load IT Work-load IT
Synthetic CDOs 2.1 2.1 2.1 1.8 1.7
Forwards 1.7 2.0 2.0 1.8 1.7
Swaps 1.7 1.8 1.8 1.6 1.5
OTC options 1.8 1.9 1.9 1.7 1.6
1
Simple average of summary responses. See Table 1 for the scale of scores.
10
CGFS – Credit risk transfer statistics
