threshold in Credit Support Annex

[spenserzhou]

Can anyone tell me how to determine threshold in a CSA?

I understand it depends on the ratings of counterparties, or sometimes even the types of collaterals. However, is there a methodology, like VaR or anything else, which helps to quantify the requirements for threshold?

Thanks

 

[hamu4ok]

Can anyone tell me how to determine threshold in a CSA?

I understand it depends on the ratings of counterparties, or sometimes even the types of collaterals. However, is there a methodology, like VaR or anything else, which helps to quantify the requirements for threshold?

Thanks

threshold is a trade-off between having some exposure uncollaterized (i.e. under more counterparty risk) and having less burden with collateral management (less operational, legal risk).
Collateral is a thing meant to be actively managed - valued properly, safeguarded, monitored, frequent margin calls would certainly cost the company money, resources and threat of risk (not receiving collateral on time, human errors, misvaluation, fraud). If you have zero threshold, then every change in MtM causing your position into negative would require you to pay-in a collateral, positive position you would be expecting to receive collateral from your counterparty. The remargin period is usually several days, up to 10-20 days, so with zero threshold, you might have ten days of positive change in MtM, and each of ten days your CP would have to send the collateral and you would be "waiting" for it< first batch you receive in say 10 days, 2nd batch might delay for 12 days, but 3rd batch receive earlier than 2nd batch, etc ---> that's gonna be a whole mess to sort out properly.

This is why threshold in agreed, say USD 100K, if the exposure is under 100K there is no need to send collateral, only excess of this amount would have to collaterized.

As I said, you get a little bit more of counterparty risk (in terms of uncollaterized portion of exposure = threshold), but at the same time your operational workload is substantially less (less operational risk, legal risk) - it is a trade-off.

 

[bpdulog]

I have a few questions related to the following section:

"In the case of a two-way CSA the behavior is not completely monotonic with respect to an increasing threshold such that a (two-way) threshold of $1m appears slightly more beneficial than a zero-threshold CSA. It is interesting to explain this effect in a bit more detail.

This non-monotonic behavior in the two-way CSA case is related to EE being less than NEE while 95% PFE is greater than 5% PFE. Recall that we are dealing with a set of four trades, three of which have a positive sensitivity to overall interest rates. In the zero-threshold case, there are many scenarios where the institution must post a relatively small amount of collateral due to a negative drift (relating to the NEE being greater than the EE). This tends to weaken the benefit of the collateralization. On the other hand, with a small threshold, many of these scenarios do not result in collateral posting and the ability to mitigate the paths around the 95% PFE, where interest rates are high, outweighs the need to post collateral for the paths around the (smaller) 5% PFE."

1. Is EE always less than NEE, or is it just a function of this specific base case scenario with 4 trades? They are referring to a BCVA in this section, correct?
2. What is causing EE to be less than NEE?

 

[bpdulog]

Hi,

I will add one other question:

If EE ends up being more than NEE w/ a 2 way CSA, do we get a reverse effect similar to the red line below? Or is it a steeper line but same shape as the one way CSA?

 

[bpdulog]

Hi,

Does anyone know of the CVA will display a similar blip for a low threshold vs 0 threshold if EE>NEE??

 

[bpdulog]

Hi,

I will add one other question:

If EE ends up being more than NEE w/ a 2 way CSA, do we get a reverse effect similar to the red line below? Or is it a steeper line but same shape as the one way CSA?

View attachment 1097

Greetings all,



I updated the graph from above for better clarity and added a green line that shows the hump above the one way CSA -does anyone know the impact if EE > NEE instead? It would be interesting to model but the Jon Gregory spreadsheets don't have a threshold variable

In addition, I am confused about the following section:

"In the zero-threshold case, there are many scenarios where the institution must post a relatively small amount of collateral due to a negative drift (relating to the NEE being greater than the EE). This tends to weaken the benefit of the collateralisation. On the other hand, with a small threshold, many of these scenarios do not result in collateral posting and the ability to mitigate the paths around the 95% PFE, where interest rates are high, outweighs the need to post collateral for the paths around the (smaller) 5% PFE.

So smaller % PFE results in less collateral posting which makes sense. But what is meant by mitigating the paths?

Let us say instead of having 3 payer swaps, they are now the receiver. Does this now mean that CVA is a downward sloping line as threshold increases?