As we already discussed, financial institutions use more and more Credit Support Annex (CSA) to protect themselves against a potential default of their counterparties and more generally to mitigate the exposure arising from derivative contracts done with their counterparties.

Indeed, as we saw in a previous article Credit/Debit Valuation Adjustment highly rely on calculation of  Expected Positive/Negative Exposures.

We already introduced the main drivers of collateral contracts that can be summarized as follows, but so far we can already split the characteristics into two categories. What is triggering the collateral  (which is discussed in this article) and what could be posted as collateral (which will examined in a second post later on)


Amongst the triggers of collateral we have:

– level of thresholds applied to the counterparty and to ourself

– level of minimum transfer amount

– existence of initial margin or independent amounts

– who is posting the collateral (unilateral or bilateral)


Regarding the collateral materiality we can draw the following features:

– currency of collateral (can the collateral giver post one currency among a basket according to how cheap is it to deliver one currency vs. another ? What happens when the currency posted is non-deliverable ?)

– type of collateral (cash, security, credit letter or both)

– existence of spread over overnight rate


However, before dwelling into these various details, we have to do a small digression to introduce briefly the netting contract.

Netting contract consists in an agreement that allows derivatives’ cash flows to be gathered in a netting set instead of being considered separately from each other. Thus, if several derivatives have cash flows that can be netted together, we will consider the mark to market of the netting set instead of dealing with the mark to market of each derivative product taken independently.
The main effect is that it could strongly reduce the exposures as it possess the subadditivity feature.
The netting contract is not a collateral contract but both netting and collateral are usually found together as part of risk mitigation in interbank market.
However, netting contract is usually found alone for corporate counterparties as it is much more difficult for them to borrow on a regular basis significant amounts of cash or even securities to post in case of margin calls.


  • Thresholds

The threshold corresponds to the level of exposure (positive or negative) above which the collateral will be posted. The higher the threshold, the higher the residual exposure.
As a consequence the greater the credit quality of one entity with regards to its counterparty, the higher the threshold. Thresholds can be contractually defined according to the credit rating of each counterparty and thus they can be adjusted if credit ratings were to improve or to worsen. These adjustments that are contractually defined into the CSA contract are called rating triggers.
Since financial institutions can have very different credit qualities we can frequently observe threshold asymetry. This is typically the case when a private bank is dealing with a supra national institution whose credit quality is considered close to sovereign.
We can generally express exposures with threshold such as :

EPE= E^{Q} [MtM^{+} -(MtM^{+}-Th_{ctp}) ^{+}] ^{+}

ENE= E^{Q} [-MtM^{-} -(-Th_{self}-MtM^{-}) ^{+}] ^{+}


Where Th_{ctp}, is the threshold applied when the counterparty has to post, and Th_{self} is the threshold applied when the calculating entity has to post.


Several remarks arise from these formulas:
We consider here that each collateral payment is perfectly offsetting the Mark to Market which is not necessarily the case in real world as disputes over MtM value can occur sporadically – especially when underlying instruments are exotic derivatives with no ´market price ´. Also collateral amounts posted can suffer from delays and therefore won’t perfectly offset the MtM, indeed it is impossible and unrealistic to have a real time reconciliation between MtM and collateral posted so there is generally at least a one business day delay.
We can see that under a 0 threshold contract EPE/ENE are virtually equal to 0. Also, when the threshold is infinite EPE/ENE are equal to MtM discounted under risk-neutral probability or equivalent to Call/Put with MtM as underlying asset and 0 strike.
When the threshold is different from 0 or infinite, the EPE/ENE can be considered as a Call/Put spread with MtM as underlying and whose two strikes  are 0 and threshold.


Let’s take the example of a bilateral collateral contract with 10M Eur thresholds applied to swap trades between two counterparties.
From the calculating counterparty’s perspective if the MtM of the swaps’ netting set is +20M EUR at some moment then the other counterparty has to deliver 10M EUR as part of the margin call.
Renegociating a collateral contract with a lower threshold incurs a gain (resp. loss) of CVA (resp. DVA) for the xVA desk. If a contract was to switch from infinite threshold to 0, the whole exposures would fall down to 0, which would make the counterparty without any risk and therefore the initial charge would turn into a profit (loss) if the initial CVA was greater (smaller) than the initial DVA.
Put differently the xVA desk is short the threshold applied originally to the counterparty and is long the threshold to himself.


  • Minimum Transfer Amount (MTA)

The MTA represents the MtM amount above which the collateral has to be posted. It prevents from unnecessary collateral transfers for negligible amounts and therefore additonal workload for back office department in charge of cash or security transfers.
MTA is similar to thresholds in terms of effect on the exposures since it leaves both counterparties exposed to any MtM amount below the MTA level.
Let’s take the previous example of the collateral contract with 10M EUR thresholds between the two counterparties and say there is a MTA level of 1M for both. If the MtM goes to +11MEUR from the calculating counterparty’s perspective then there would be no collateral posted by the other while there would be 1MEUR if we had only thresholds. However if the MtM goes to +11.1M EUR then the other counterparty would have to post 1.1MEUR.
Exposures can be rewritten in terms of MTA:

EPE= E^{Q} [MtM^{+} -(MtM^{+}-Th_{ctp})_{1(MtM^{+}-Th_{ctp})>MTA_{ctp}}  ^{+}] ^{+}

ENE= E^{Q} [-MtM^{-} -(-Th_{self}-MtM^{-})_{1(-Th_{self}-MtM^{-})>MTA_{selt}} ^{+}] ^{+}


  • Initial Margin/Independent Amounts

Initial margin and independent amounts are additional sources of collateralization. They represent the amount of collateral supplied by one or the two counterparties to account for potential under-collateralization (IM is typically posted in cash while IA can be posted in cash or bonds).

This amount can be viewed as a initial buffer for risk mitigation.

Initial Margin is typically part of the clearing process and is required by clearing houses as a percentage of a new trade’s notional (like 1 %).

In fall 2016, Dodd-Frank regulation corpus implies that the US and Japanese major banks have to post bilateral Regulatory Initial Margin (RIM) to account for potential under-collateralization (European banks should be concerned in 2017 due to the EMIR version of RIM).

The RIM only concerns new operations among specific types of trades (for instance all kinds of physical exchanges of principal or physical settlements are exempted from RIM).

RIM value is not arbitrary and can actually rely on multiple factors. The standard method promoted by ISDA (Standard Initial Margin Model – SIMM) involves the calculation of a 11 days parametric VaR (99%). The latter is supposed to cover the Margining Period of Risk.

IM and IA are not necessarily constant over the time buckets, for instance RIM is likely to be amortized according to the profile of underlying sensitivities. IA can depend on the maturities of the instruments they are supposed to cover so they can be amortized over the exposures’ time buckets.


Let’s take an example,  a european bank deals a swap with a sovereign entity such as the OKB (Oesterreichische Kontrollbank) in Austria. The swap exhibits a very risky exposure profile from the sovereign entity’s point of view in case the bank was to default. So since the sovereign entity has a much better credit quality and rating than the bank, it could propose to be supplied with independent amounts consistent with the exposure profile.

  • What does this mean in terms of exposure profile ?

Two main cases have to be taken into account, depending on whether the IM or IA are segregated with an independent third party custodian.

Segregated IA can be actually demanded by a counterparty, also by definition RIM has to be posted on a segregated account.

Let’s suppose that in our example, the IA are not segregated, the bank will have an additional exposure at several time buckets which correspond to the forward independent amounts vector (i.e. the MtM+ would suffer from an upward translation equals to the level of independent amounts for each time bucket).

Indeed, if the sovereign entity was to default it would leave unpaid the MtM+ but also the IA posted by the bank at the contract inception. Conversely, the exposures from the sovereign entity point of view would be reduced by the IA vector.

In case the IM/IA are segregated, they would not generate any exposures in terms of CVA/DVA, however as they cannot be rehypothecated they represent an amount to be funded and thus should have an impact in terms of funding exposures (used for the Funding Valuation Adjustment calculation).


The previous case would be represented as follows in terms of exposures (IA is considered as a negative amount since it is posted to the counterparty):





Please note that we assumed the CSA was 0 Threshold/MTA for the bank and infinite thresholds/MTA for the sovereign (the latter is never posting collateral whereas the former does as soon as the MtM becomes negative).


  • Asymmetric Collateral

When one entity has a much better credit quality than its counterparty it can be reflected in the collateral contract that has been negociated and which counterparty will have to post collateral or not.
Once again in case of collateral contract between a private bank and a sovereign counterparty, it can happen that only the private bank has to post the collateral. In terms of exposures it is equivalent to have an infinite threshold amount.