Ad Hoc Commentary – CVA, DVA, FVA and sovereign debt crisis
Ever since Lehman failed, counterparty credit risk became the main topic among regulators, legislators, academics, lawyers and accountants that are working on the banking sector. Now we have an alphabet soup of xVA concepts: CVA (credit valuation adjustment), DVA (debit valuation adjustment), FVA (funding valuation adjustment), IMVA (initial margin valuation adjustment), CollVA (collateral valuation adjustment), KVA (capital valuation adjustment) and so on. CVA, DVA and FVA are the most important. The other xVAs are simply obfuscating a very simple concept.
The state of the art today is:
1. For pricing, you use CVA and FVA
2. For accounting, you use CVA, DVA and FVA.
To be precise IFRS 13 only requires you to report CVA and DVA. JP Morgan under the leadership of Jamie Dimon reports FVA in addition to CVA and DVA. You will see why this makes perfect sense in a moment.
Let us first go back to first principles. When one enters into a derivatives transaction, one implicitly grants one’s counterparty an option to default (the quantification of this is the CVA) and, at the same time, one also receives an option to default (the quantification of this is the DVA). In other words: one sells credit protection on one’s counterparty (CVA); and one buys credit protection on oneself (DVA).
After the failure of Lehman, market actors began to treat plain vanilla interest-rate (IR) swaps as defaultable IR swaps. The right way to price and risk manage this defaultable IR swap is to use a credit/IR hybrid model that takes into account the existing portfolio with the counterparty. We can call this the macro credit/IR hybrid model. We contrast this with mainstream micro IR models that use IR only to price IR swaps. The micro models are oblivious to credit and the pre-existing portfolio (i.e. diversification benefits of the portfolio). As you can imagine, to get the quants to rewrite the micro models into macro models is a mathematical headache. Even if you have done that, running the macro models in the front office is going to take computational power and time. It is not surprising that banks decided to keep the micro models, and create an xVA desk to manage the counterparty risks separately.
You might note that credit and IR are not independent economic variables. You either have the correct hybrid macro model and one desk manages it, or you don’t it. Creating an extra xVA desk on top of the traditional desks is at best a makeshift effort because the model is elusive. Any result from the makeshift arrangement is thus sub-optimal. This reality discredits people who complain that regulators are double charging banks (via capital requirements) for xVA desks’ market hedges. In reality, depending on the covariance matrix, the current regulations could be undercharging or overcharging. Nobody knows for sure until someone comes around and quantify it using a true hybrid credit/IR model.
Now that we have created some intuition, we should be able to agree that a better price for the derivative is pre_Lehman + CVA – DVA. Since every bank is different, we will get an explosion of prices. This is expected to those familiar with the supply-demand curve in Economics 101. Every producer on the supply curve has a different cost of production. The ideal market clearing price should be pre_Lehman + CVA – DVA_of_a_typical_bank. That is why it is typical for sales and traders to argue DVA/CVA to win deals. What they are essentially doing is to converge to the market clearing price.
However, market uses pre_Lehman + CVA – FVA when pricing. To understand why, we remind ourselves that DVA is based on market observable credit spreads. You might be a market fundamentalist and question the cost of funding that is handed to you by your treasurer versus the credit spreads you observe on the market. But ultimately, that is usually a futile argument for those in the trenches. So, those on sales and trading desk generally use FVA instead of DVA. Thus, we reach at the state-of-the-art for pricing: pre_Lehman + CVA – FVA.
Let us now move on to accounting rules. IFRS 13 states that we need to report CVA and DVA but are silent on FVA. The problem with reporting pre_Lehman + CVA – DVA is the DVA. By including DVA, you are effectively being rewarded for the deterioration of your own credit quality. That is fine if you plan to liquidate tomorrow and cease to be a going concern. However, since accounting books should reflect the organization as a going concern, you need to somehow get rid of the DVA. Since you must report DVA, the way to get rid of DVA is to use FVA. Thus, we reach at the state-of-the-art for accounting books: pre_Lehman + CVA – DVA + FVA.
The other xVAs are not very important. Briefly, IMVA is just the cost of funding initial margin. CollVA is a consequence of outdated CSA leading to the phenomenon of cheapest collateral to deliver. KVA is just a method to pass the cost of new regulatory capital charge back to the sales and trading desk.
Now we know what xVA is, there are two thoughts yours truly want to leave you with:
A. xVA will only increase the cost of doing business
B. xVA will likely not help in a sovereign debt crisis
(A) is obvious since mitigating risk is tantamount to buying insurance. It is worth remembering that risks are never destroyed. For example, we can mitigate counterparty risk (CVA) by increasing funding risk (FVA). Simply put, risk can be mutated but not destroyed.
(B) will be the unintended consequence of the success in reducing counterparty risk. There will likely be more cash held as collateral. However, in a sovereign debt crisis, when confidences in the full-faith of sovereigns are being tested, it is perhaps debatable on how good cash collaterals are. Cash is after all an IOU that everyone trusts, and this trust emanates from sovereigns.
Good luck in the markets.