1. Can you share your views regarding survival probability adjusting FVA versus including FVA in the replacement cost and thus embed it in the CVA calculation? Should we calculate FVA and include survival probabilities or not?
Market practice is quite mixed between either not doing this and doing this (usually with counterparty survival only). The choice should depend on whether or not FVA is included in the settlement valuation in the event of default. This in turn is difficult but most obviously a default valuation would be a two-way collateralised “interdealer” type. This suggests including survival probabilities since FVA disappears at default. One other related point to note here is that in pricing, some banks may lower funding curves for low risk counterparties.
2. What are your thoughts around a number of the additional xVAs such as HVA, RVA, CCVA, etc? And to what extent are these actually valuation adjustments versus some sort of overhead and perhaps treated as an accrual cost (is it really part of the exit price)? In particular for MVA regarding it being an overhead cost.
There are xVAs which are not real in the sense that they are captured elsewhere. For example, Tax Valuation Adjustment (TVA) would most obviously be accounted for in KVA since the revenue in the return on capital formula should include corporation tax considerations. Then there are xVAs which are real but are really overheads and don’t perhaps need to be treated specifically (e.g. HVA). The divide is not always clear – some would say MVA is an overhead but others would argue that it is a real valuation adjustment just like CVA and FVA.
3. Why, in your opinion, is it better to consider FBA and not DVA? Why not the other way around?
Yes. Because funding costs (e.g., an unsecured asset or posting collateral) naturally net with funding benefits (e.g., an unsecured liability or receiving collateral). Whilst FBA can be possibly monetised (e.g., by lending out cash or meeting funding requirements on derivative assets), DVA cannot (since you cannot – for example – sell CDS protection on yourself).
4. Could you explain again why valuation adjustments are done at portfolio level? Why don't/can't we just do CVA and/or FVA at trade or counterparty level?
Taking a simple example. Suppose you do a trade (e.g., receive fixed interest rate swap) with a bank and then unwind this by doing the opposite trade (pay fixed interest rate swap) with the same bank. The net value of the two trades is zero and in default there should be no impact. So, the total CVA is zero. But the CVA of each trade individually will not be. The netting effect of the two trades is why the total portfolio CVA is less negative than the sum of all the individual CVAs.
5. Do you think it is more correct to calculate FVA at the level of netting set or funding set? Do you know what the market practice might be?
Definitely not netting set because you never fund a netting set individually. It should be calculated at the funding set level, whatever that is (e.g., the entire financial markets division of a bank) and I think this is more or less becoming market standard. But note that symmetric FVA means that it can be calculated at the trade level since these are additive to the total FVA (unlike CVA in the last question).
6. For a portfolio involving initial margin, can we ignore CVA and FVA, and consider only MVA?
More or less. A portfolio with initial margin will almost certainly have variation margin with a zero threshold posted on a daily basis. So, ignoring intraday liquidity, there is no FVA. There will be a small CVA due to the margin period of risk but due to initial margin this may be negligible. But CVA (not FVA) probably needs to be quantified and not just assumed to be zero.