Netting, Close-out and Related Aspects

[Vicky26]

The opposite signs of MTM is due to fact that each party owes some amount to another. Say A & B, where A defaults. A owed 20 to B and B owed 10 to A. Then MTM values for B are +20 and -10.
If correlation is high, then both MTM values will go up/down together. And thus Net value (+10 here) will move together. But, if correlation is low then one MTM moves up/down and other moves down/up. Say +20 gets to +21 and -10 gets to -9. Then the net value increases from +10 to +12. Which kind of exaggerates the net value in case of default.
I could not understand how low correlation benefits netting in the way it does in diversification.

 

[Clay Carter]

Hi @Vicky26! You're thinking about this correctly in terms of the mechanics, but what the slide is trying to convey is a bit more subtle. Let me walk through it.

The slide says netting only helps when the MtM values have opposite signs. That's the whole point. Netting collapses multiple exposures into a single net claim. If A owes B 20 and B owes A 10, the non-netted gross exposure for B is 20 (you only count what's owed to you, the positive side). With netting, it's only 10. That's the benefit.

Now here's where correlation comes in. Ask yourself: when does netting produce the biggest reduction in exposure?

It produces the biggest reduction when one trade is deeply in-the-money and another is deeply out-of-the-money at the same time, because that's when the offset is largest. For that to happen consistently, the MtM values need to move in opposite directions relative to each other, which is what negative (or low) correlation means in this context.

Here is how I think about it:

With high positive correlation, all MtM values tend to be positive together or negative together. The non-defaulting party either has a lot of gross exposure on everything, or nothing. The netting doesn't really help because there's rarely a large offsetting negative trade to cancel out the positive.

• With low or negative correlation, one trade is likely up while another is likely down. That produces the opposite signs that make netting valuable. The net exposure is far lower than the gross.

Your example shows it nicely. When correlation is low and one goes to +21 while the other goes to -9, the net is +12. That is still well below the gross positive exposure of +21. Without netting, B would be a creditor for 21. With netting, only 12. So netting is still doing its job.

I think the confusion is that you are comparing the net figure over time, watching it rise, and thinking "that's bad." But the benchmark for netting is not the previous net, it's the gross exposure. The benefit of netting is always measured as: gross positive exposure minus net exposure. And that difference tends to be larger when correlation is low, because you're more likely to have large offsetting positions at any given moment.

It's the same logic as portfolio diversification in spirit: low correlation means the positions don't all move the same way, and that creates room for offsets to work their magic.

Kindly,
CC

 

[Vicky26]

So when correlation is low, +20 goes to +21 and -10 goes to -9 then net exposure is 12 which was 10 initially.
In case of positive correlation, +20 goes to +21 and -10 goes to -11 then new exposure remains 10....same as initial 10 exposure.
So in such case positive correlation is better in atleast keeping the exposure same. Then how is negative correlation better in case of netting.

 

[Clay Carter]

So when correlation is low, +20 goes to +21 and -10 goes to -9 then net exposure is 12 which was 10 initially.
In case of positive correlation, +20 goes to +21 and -10 goes to -11 then new exposure remains 10....same as initial 10 exposure.
So in such case positive correlation is better in atleast keeping the exposure same. Then how is negative correlation better in case of netting.

@Vicky26 Let's sort it out by looking at the direction of the two deltas rather than just the end result.

In your first case, +20 moves to +21 and -10 moves to -9. Look at the direction of change on each trade. One went up by 1, and the other also went up by 1 (it became less negative, which is still a move in the same direction, upward). Both trades moved the same way at the same time. That is what positive correlation looks like, not low correlation. When two trades are positively correlated, a move in the underlying market pushes both MtMs the same direction together.

In your second case, +20 moves to +21 while -10 moves to -11. Here one trade moved up and the other moved down. That is opposite movement, which is the signature of negative or low correlation. When the underlying market moves, one trade gains while the other loses, so they offset each other.

So the labels in the original post got flipped. The scenario where net exposure stayed flat at 10 is actually the low or negative correlation case. The scenario where net exposure grew from 10 to 12 is the positive correlation case.

Once you relabel it that way, everything lines up with the slide. Under negative correlation, the two MtMs move in opposite directions, so gains on one trade are absorbed by losses on the other, and the net position barely moves. That is the stability that netting is supposed to buy you. Under positive correlation, both MtMs move the same direction together, so instead of offsetting each other they reinforce each other, and the net exposure swings further in whichever direction the market goes. That is a worse outcome from a credit risk standpoint, because your net exposure becomes more volatile and more capable of spiking upward before you ever get to a default event.

Think of the extreme cases to build intuition. If two trades are perfectly negatively correlated, like a swap and its exact mirror image, whatever one gains the other loses in equal measure, so the net exposure barely ever moves off wherever it started, often close to zero. If two trades are perfectly positively correlated, like two identical copies of the same trade, they rise and fall together, so netting gives you nothing, you are just carrying double the exposure of one trade, moving in lockstep.

So to answer your question, positive correlation is not actually better here. It only looked that way because the scenario that stayed flat at 10 had been mislabeled as positive correlation when it was really the negative correlation case. Negative or low correlation is the one that keeps net exposure contained, and that is the entire reason it is considered more favorable for netting.