Tải bản đầy đủ
3 The eligibility premium in times of a liquidity crisis: the 'sub-prime turmoil' of 2007

3 The eligibility premium in times of a liquidity crisis: the 'sub-prime turmoil' of 2007

Tải bản đầy đủ


Bindseil, U. and Papadia, F.


MRO weighted average

Rate (%)








Figure 7.4.

Evolution of MRO weighted average, 1 Week repo, and 1 Week unsecured interbank rates in 2007.


LTRO weighted average rate



Rate (%)




2/ /20
14 07
2/ /20
28 07
3/ /20
14 07
3/ /20
28 07
4/ /20
11 07
4/ /20
25 07
5/ 00
9/ 7
5/ 20
23 07
6/ 00
6/ 7
6/ 20
20 07
7/ 00
4/ 7
7/ 20
18 07
8/ 00
1/ 7
8/ 20
15 07
8/ /20
29 07
9/ /20
12 07
9/ /20
2 0
10 6/2 7
/1 00
10 0/2 7
/2 00
4 7
11 /20
/7 07
11 /2
/2 00
1 7
12 /20
/ 0
12 5/2 7
/1 00
9/ 7



Figure 7.5.

Evolution of LTRO weighted average, 3M repo, and 3M unsecured interbank rates in 2007.

Until including July, weighted average LTRO rate was very close to, albeit
slightly higher (2 basis points on average) than, the EUREPO rate as seen
above. The EURIBOR, on its turn, was also close to the weighted average
LTRO but somewhat higher (on average 7 basis points). Since the beginning
of August, as seen above, the spread between EURIBOR rate and the
EUREPO has grown dramatically (on average to 64 basis points) and
the weighted average LTRO has tended to follow much more closely the
EURIBOR than the EUREPO, so much that its spread to the latter increased


Risk management and market impact of credit operations

Table 7.7 Spreads containing information on the GC and Eurosystem collateral eligibility
premia – before and during the 2007 turmoil

Jan – July 2007
August – December 2007



1 week

3 months

1 week

3 months






weighted average OMO rates. Source: ECB.
Source: ISDA. 2006. ‘ISDA Margin Survey 2006’, Memorandum, Table 3.1.

to 50 basis points. This behaviour manifests, even more clearly than in the
case of the one week MRO, a very aggressive bidding by commercial banks
at Eurosystem operations, facilitated by the ability to use a much wider
range of collateral in these operations as compared with private repurchase
transactions. Indeed, it is surprising that the secured operations with the
Eurosystem are conducted at rates which are closer to those of unsecured
operations than to those prevailing in private secured operations.
Table 7.7 summarizes again all spread measures during the pre-turmoil
and turmoil period. Overall, the second half of 2007 episode shows that
indeed, eligibility premia for being acceptable collateral, either in interbank
operations or for central bank operations, soar considerably in the case of
financial market turmoil and implied liquidity fears.
3.4 Effects of eligibility on issuance
The preceding analysis has maintained as simplifying assumption that the
amounts of securities of different types are given. However, issuance activity
should react to yield effects of eligibility decisions. First, there may be a
substitution effect, with debtors seeking to fund themselves in the cheapest
way; thus, eligible instruments should substitute, over time, ineligible
instruments. Second, agents may decide to issue, in the aggregate, more debt
since the lower the financing costs, the greater the willingness to issue debt
should be. While the substitution effect could, at least in theory, be significant even for an eligibility premium of only a few basis points, the second
effect would require more substantial yields differentials to be relevant. Here,
it suffices to note that the assumption (maintained so far) of a zero elasticity
of issuance to yield changes caused by eligibility decisions biases any estimate
of the eligibility premium to the upside, particularly in the long term. In the


Bindseil, U. and Papadia, F.

extreme case of infinite elasticity, the only consequence of a changing
eligibility premium would be on the amounts issued, not on yields.

4. Conclusions
This chapter has presented an analytical approach to the establishment of a
central bank collateral framework to protect against credit losses. A collateral framework should ensure that the residual risks from credit exposures (e.g. lending) are in line with the central bank credit risk tolerance. At
the same time, such a framework should remain reasonably simple. If a
central bank accepts different types of collateral, it should apply differentiated risk mitigation measures, to ensure that the risk remaining after the
application of these measures complies with its risk tolerance, whatever
asset from its list of eligible collateral is used. The differentiation of risk
control measures should also help avoid that counterparties provide in a
very disproportionate way one particular type of collateral. This could
happen, in particular, if too lax risk control measures were applied to one
given asset, thus making its use as collateral too attractive, in particular if
compared to standard market practice. Once the necessary risk mitigation
measures have been defined for each type of asset, the central bank can rank
each asset type according to its costs and benefits and then set a cut-off
point which takes into account collateral demand.
The chapter stresses, however, that the collateral framework needs to
strike a balance between precision and flexibility for counterparties in
choosing collateral. In addition, any framework needs to maintain a degree
of simplicity, which implies that it is to be seen as an approximation to a
theoretically optimal design. Its actual features have to be periodically
reviewed, and if necessary modified, in the light of experience, in particular
in the light of the actual exposures and use of the different types of collateral
and resulting concentration risks.
If the collateral framework and associated risk mitigation measures follow
the above-outlined methodology, aiming at socially optimal configurations,
one should not denote an effect on asset prices as distortion. This also
implies that market neutrality is not necessarily an objective of a central
bank collateral framework: effects on market equilibria are acceptable as far
as they move towards optimality.
The chapter concentrates on one particular effect of the Eurosystem
collateral framework on market equilibrium, namely on the ‘eligibility


Risk management and market impact of credit operations

premium’, i.e. the reduction of the yield of a given asset with respect to
another asset, which is similar in all other respects but eligibility as collateral
with the central bank. First, it shows how the proposed model allows to
understand the origin and nature of the eligibility premium. Second,
it carries out an empirical analysis to get an idea of the size of such a
premium. While the size of the eligibility premium is likely to change over
time, in the case of the euro area, the broad range and large amount of
eligible collateral makes the eligibility premium small under normal circumstances. Some empirical measures, the limitations of which need to be
stressed, consistently indicate an average level of the eligibility premium not
higher than 5 basis points. However, this premium will of course be different for different assets and possibly also for different counterparties.
More importantly, the eligibility premium rises with an increase in the
demand for collateral, as occurring particularly in the case of a financial
crisis, as illustrated by the financial market turmoil during 2007. An increase
in the eligibility premium should also be observed in case the supply of
available collateral were to shrink.
Independently from the conclusion reached about the complex empirical
issue of the eligibility premium, there are good reasons why a central bank
should accept a wider range of collateral than private market participants:
First, central bank collateral serves monetary policy implementation and
payment systems, the smooth functioning of which is socially valuable.
While in the inter-bank market uncollateralized operations are always an
alternative, central banks can, for the reasons spelled out in the introduction, only lend against collateral. A scarcity of collateral, which could particularly arise in periods of financial tensions, could have very negative
consequences and needs to be avoided, even at the price of having ‘too
much’ collateral in normal times. Second, as a consequence of the size of
central bank operations, it may be efficient to set up specific handling, credit
assessment or risk mitigation structures which the private sector would find
more costly to set up for inter-bank operations. Finally, there is no guarantee that the market can establish efficient collateralization conventions,
since the establishment of these conventions involves positive network
externalities (see e.g. Katz and Shapiro 1985 for a general presentation of
network externality issues). Indeed, the central bank, as a large public
player, could positively influence market conventions. For instance, trade
bills became the dominant financial instrument in the inter-bank market in
the eighteenth, nineteenth and early twentieth century in the United
Kingdom and parts of Europe (see e.g. King 1936; Reichsbank 1910)


Bindseil, U. and Papadia, F.

because central banks accepted them for discounting. The last two points
can be summarized by noting that the central bank is likely to have a special
collateral-related ‘technology’ compared with private market participants,
either because economies of scale or because of its ability to exploit network
externalities. This in turn confirms the point that it can positively impact
market equilibria, as argued above.


Risk mitigation measures and credit
risk assessment in central bank
policy operations
Fernando Gonza´lez and Phillipe Molitor

1. Introduction
Central banks implement monetary policy using a variety of financial
instruments. These instruments include repurchase transactions, outright
transactions, central bank debt certificates, foreign exchange swaps and
the collection of fixed-term deposits. Out of these instruments, repurchase
transactions are the most important tool used by central banks in the
conduct of monetary policy. Currently the Eurosystem alone provides
liquidity to the euro banking system through repurchase transactions with
a total outstanding value of around half a trillion euro.
Repurchase transactions, also called ‘reverse transactions’ or ‘repos’,
consist of the provision of funds against the guarantee of collateral for a
limited and pre-specified period of time. The transaction can be divided
into two legs, the cash and the collateral leg.
The cash leg is akin to a classical lending operation. The lender transfers
an amount of cash to a borrower at the initiation of a transaction. The
borrower commits to pay the cash amount lent plus a compensation (i.e.
interest) back to the lender at maturity.
By the nature of lending, any lender bears credit risk, namely the risk that
the borrower will fail to comply with its commitments to return the borrowed cash and/or provide the required compensation (i.e. interest) at the
maturity of the transaction. Several tools are available to the lender to
mitigate this risk.
First, counterparty risk can be reduced by conducting operations only
with counterparties of a high credit quality, so that the probability of a
default is small. In a central banking context, the set of institutions having
access to monetary policy operations is generally specified with the goal of


Gonza´lez, F. and Molitor, P.

guaranteeing equal treatment to financial institutions while also ensuring
that they fulfil certain operational and prudential requirements.
Second, and reflecting the same idea, counterparty risk can also be
reduced by implementing a system of limits linking the exposure to each
counterparty to its credit quality, so that the potential loss is kept at low
levels. For central banks, however, such a system is generally incompatible
with an efficient and transparent tender procedure for allotting liquidity.
Finally, counterparty risk can be mitigated by requiring the borrower to
provide adequate collateral. This approach mitigates financial risks without
limiting the number of counterparties or interfering with the allotment
procedure. It is a common approach chosen by major central banks when
conducting repurchase operations. When combined with the appropriate
risk management tools, collateralization can reduce the overall risk to negligible levels.
The collateral leg of a repurchase transaction consists, hence, of providing
collateral amounting at least in value to the cash borrowed to the lender,
which is returned by the borrower upon receiving back the cash lent and the
compensation at maturity of the transaction.
The lender in a collateralized reverse transaction may still incur a financial
loss. However, this would require more than one adverse event to occur at
the same time. This could happen as follows: the borrower would first
default on his obligation to the lender, resulting in the lender taking possession of the collateral. Assuming that at the time of the default the value of
the collateral covered the value of the liquidity provided through the reverse
transaction, financial risk could arise from the following two possible
 Credit risk associated with the collateral. The issuer of the security or the
debtor of the claim accepted as collateral could also default, resulting in a
‘double default’. The probability of such a combination of defaults can be
considered negligible if eligible assets satisfy high credit quality standards
and if the lender does not accept assets issued by the borrower or entities
having close financial links to the borrower.
 Market and liquidity risk. This would arise if the value of the collateral
fell in the period between the counterparty’s default and the realization of
the collateral. In the time between the last valuation of the collateral and
the realization of the collateral in the market, the collateral price could
decrease to the extent that only a fraction of the claim could be recovered
by the borrower. Market risk may be defined in this context as the risk of
financial loss due to a fall of the market value of collateral caused by


Risk mitigation measures and credit risk assessment

exogenous factors. Liquidity risk may be defined as the risk of financial
loss arising from difficulties in liquidating a position quickly without this
having a negative impact on the price of the asset. Market and liquidity
risk can also be reduced considerably by following best practices in the
valuation of assets and the risk control measures applied.
The collateral leg is hence intended to mitigate the credit or default risk of
the counterparty borrowing the cash and therefore plays a crucial role in
this type of operations. In case of default of the counterparty, the collateral
taker which in the context of this book is the central bank, can sell the
collateral received and make good any loss incurred in the failed repo
transaction. When the collateral received is default-risk free, as for example
with government bonds, collateralization transforms credit risk (i.e. the risk
of default of the counterparty) into market and liquidity risk (i.e. the risk of
incurring an adverse price movement in the collateral position and the risk
of impacting the price due to the liquidation of a large position over a short
period of time).
Figure 8.1 provides a visual summary of a reverse transaction and the
risks involved.
Two main risk factors need to be considered in the risk management of
collateral underlying repo operations. First, the credit quality of the collateral needs to be sufficiently high so as to give enough reassurance that the
Credit risk ≥ 0
Market risk > 0
Liquidity risk > 0

Asset B
T=0: repurchase
T=τ: return


T= τ: compensation

Collateral provider

Central Bank
Collateral receiver

T= τ: return
T=0: repurchase

Credit risk ≥ 0

Credit risk = 0
Market risk = 0
Liquidity risk = 0

Asset A

Figure 8.1

Risks involved in central bank repurchase transactions. T is a time indicator that is equal to zero at
the starting date and equal to s at the maturity date of the credit operation.


Gonza´lez, F. and Molitor, P.

collateral would not quickly deteriorate into a state of default after the
default of the counterparty. In this regard, it is also crucial that the collateral
quality would be independent from that of the counterparty (i.e. no close
links). To assess the credit quality of the collateral, central banks tend to rely
on external ratings as issued by rating agencies or internal credit quality
assessments as produced by in-house credit systems. This chapter will
review the main sources of credit quality assessments used by central banks
in the assessment of collateral and the main parameters that a central bank
needs to define in its credit assessment framework such as the minimum
credit quality threshold (e.g. a minimum rating threshold) and a performance monitoring of the credit assessment sources employed. Second,
the intrinsic market risk of the collateral should be controlled. As discussed
above, in case of default of the counterparty the collateral taker will sell the
collateral. This sale is exposed to market risk or the risk of experiencing
an adverse price movement. This chapter provides a review of different
methods and practices that have been used to manage the intrinsic market
risk of collateral in such repurchase or repo agreements. In general terms,
such practices can rely on three main pillars: marking to market which helps
reduce the level of risk by revaluing more or less frequently the collateral
using market prices,1 haircuts which help reduce the level of financial risk by
reducing the collateral value by a certain percentage and limits which help
reduce the level of collateral concentration by issuer, sector or asset class. In
this chapter we consider all of these techniques in the establishment of an
adequate central bank risk control framework. Given the central role of
haircuts in any risk control framework, we put considerable emphasis on
haircut determination.
Any risk control framework of collateral should be consistent with some
basic intuitions concerning the financial asset risk that it is trying to mitigate. For example, it should support the perception that a higher haircut level should be required to cover for riskier collateral. In addition, the
lower the marking-to-market frequency, the higher the haircuts need to be.
Higher haircut levels should also be required in case the time to capture the
assets in case of default of the counterparty or the time span needed before
actual liquidation of the assets in case of default of the counterparty
increases (Cossin et al. 2003, 9). Liquidity risk or the risk of incurring a loss
in the liquidation due to illiquidity of the assets should directly impact the

If the collateral value is below that of the loan and beyond a determined trigger level, the counterparty will be
required to provide additional collateral. If the opposite happens the amount of collateral can be decreased.


Risk mitigation measures and credit risk assessment

level of haircuts. Finally, higher credit risk of the collateral received should
also produce higher haircuts.
Despite the central role of collateral in current financial markets and in
particular central bank monetary policy operations, little academic work
exists on risk mitigation measures and risk control determination. Current
industry practice is moving towards a more systematic approach in the
derivation of haircuts by applying the Value-at-Risk approach to collateral
risks but some reliance on ad hoc rule-based methods still persists. On the
whole, despite recent advances in financial modelling of risks, the discussion among academics and practitioners on the precise framework of risk
mitigation of collateral is still in its infancy (see for example Cossin et al.
2003; ISDA 2006). This chapter should also be seen in this light; a comprehensive and unique way of how to mitigate risk in collateralized transactions has yet to emerge. What exists now is a plethora of methods for risk
control determination that are used based on context and user sophistication. The chapter reviews some of these risk mitigation determination
methods of which some are used by the Eurosystem.
This chapter is organized as follows. Section 2 describes how central
banks could assess first credit quality of issuers of collateral assets and the
main elements of a credit assessment framework. Section 3 discusses the
basic set-up of a central bank as a collateral taker in a repurchase transaction
where marking-to-market policy is specified. In Section 4 we discuss various
methods for haircut determination, focusing on asset classes normally used
by central banks as eligible collateral (i.e. fixed-income assets), and review
how to incorporate credit risk and liquidity risk in haircuts. Section 5 briefly
discusses the use of limits as a risk mitigation tool for minimizing collateral
concentration risks and Section 6 concludes.

2. Assessment of collateral credit quality
2.1 Scope and elements
To ensure that accepted collateral fulfils sufficient credit quality standards,
central banks tend to rely on external or internal credit quality assessments.
While many central banks today rely exclusively on ratings by rating
agencies, there are also central bank internal credit quality assessment systems in operation. Historically, the latter was the standard. This section
reviews the main credit quality assessment systems at the disposal of central


Gonza´lez, F. and Molitor, P.

banks to assess the credit quality of collateral used in monetary policy operations. These are external credit rating agencies, in-house credit assessment
systems, counterparties’ internal rating systems and third-party credit scoring
assessment systems.
Before any credit quality assessment is taken into account, the central
bank must stipulate a minimum acceptable level of credit quality below
which collateral assets would not be accepted. Typically, this minimum level
or credit quality threshold is given in the form of a rating level as issued by
any of the major international rating agencies. For example, the minimum
threshold for credit quality could be set at a ‘single A’ credit rating.2
Expressing the minimum credit quality level in the form of a letter rating is
convenient because its meaning and information content is well understood
by market participants. However, not all collateral assets carry a rating from
one of the major rating agencies. An additional credit quality metric is
needed, especially when the central bank accepts collateral issued by a wide
set of entities not necessarily rated by the main rating agencies.
The probability of default (PD) over one year is such a metric. It
expresses the likelihood of an issuer or debtor defaulting over a specified
period of time, normally a year. Its meaning is similar to that of a rating,
which takes into account the probability of default as well as other credit
risk factors such as recovery in case of default. Both measures, ratings and
probability of default, although not entirely equivalent, are highly correlated, especially for high levels of credit quality.
The Eurosystem Credit Assessment Framework (ECAF), which is the set of
standards and procedures to define credit quality of collateral used by the
Eurosystem in its monetary policy operations, uses both metrics interchangeably. In this respect, a ‘translation’ from ratings to probability of default
levels is required (see Coppens et al. 2007, 12). In the case of the Eurosystem,
a PD value of 0.10 per cent at a one-year horizon is considered to be equivalent
to a ‘single A’ rating, which is the minimum level of rating accepted by the
Eurosystem. These minimum levels of credit quality should be monitored
and confirmed regularly by the decision-making bodies of the central bank
so as to reflect the risk appetite of the institution when accepting collateral.
2.1.1 Rating agencies
The core business of public rating agencies such as Standard & Poor’s,
Moody’s and Fitch is the analysis of credit quality of issuers of debt

This means a minimum long-term rating of A- by Fitch or Standard & Poor’s, or A3 by Moody’s.