JPM, during the investor presentation, explained the adoption of FVA:
- FVA, which represents a spread over LIBOR, has the effect of “present valuing” market funding costs into the value of derivatives today, rather than accruing the cost over the life of the derivatives
- Does not change the expected or actual cash flows
- FVA is dependent on the size and duration of underlying exposures, as well as market funding rate
- The adjustment this quarter is largely related to uncollateralized derivatives receivables, as:
- Collateralised derivatives already reflect the cost or benefit of collateral posted in valuations
- Existing DVA for liabilities already reflects credit spreads, which are a significant component of funding spreads that drive FVA
- Current quarter reflects a one-time adjustment to the current portfolio
- The P&L volatility of the combined FVA/DVA going forward is expected to be lower than in the past
All of these points are consistent with the definition of FVA which Quantifi presented in the previous articles. To recap, FVA arises when the bank has an unsecured trade with a counterparty and hedges it, via a secured trade, with a riskless counterparty. In which case if PV of the trade is positive, PV of the hedge is negative, and to cover margin/collateral call the bank has to borrow cash at its funding rate LIBOR+s where ‘s’ is a funding spread. The trade, and therefore funding terminates if the bank or a counterparty defaults. FVA is the expected value of the funding cost. It can be expressed as expectation of the banks’ funding spread applied to positive PV (discounted to today) until deal maturity, or early termination due to bank or counterparty default.
The difference with FVA is that CVA and DVA have symmetric characteristics in that banks’ CVA is counterparty’s DVA and vice versa – this allows both parties of the trade to agree on bilateral CVA (CVA-DVA).
The aforementioned connection of FVA with DVA is very important. In terms of calculation, FVA is considered a hybrid of CVA and DVA. Similar to CVA, FVA depends on positive exposure and is a cost, i.e. has negative value. Moreover, like DVA, FVA is proportional to the banks’ “own” spread (hence, default probability.) The difference with FVA is that CVA and DVA have symmetrical characteristics in that banks’ CVA is counterpartys’ DVA and vice versa – this allows both parties of the trade to agree on bilateral CVA (CVA-DVA). However, if we assume that Bank borrows at LIBOR+s but lends at LIBOR, FVA has no symmetrical part.
During the investor presentation, JPMs’ Chief Financial Officer, Marianne Lake, illustrated how FVA of such a magnitude can be obtained. When derivatives receivables is net of cash and collateral is $50bn, by applying average duration of 5 years and spread of 50bps, results in a $1.25bn loss. Note here that going forward; JPM will not report a total FVA number but a quarterly FVA gain or loss, which strongly depends on change of funding spread from the end of previous quarter. In this sense it is similar to how DVA is reported, and in this earnings report FVA and DVA are often grouped together.
An interesting point made in the JPM statement was that future FVA/DVA volatility is expected to be significantly lower than in the past quarters. From a reported DVA quarterly loss of $536m, plus the fact that JPMs’ 5y CDS spreads narrowed from 93 bps on 9/30/2013 to 70 bps on 12/31/2013, it follows that spread DV01 for DVA is $536m/23=$23.3m. At the same time, if we assume that FVA was calculated based on a 5y CDS spread, DV01 for FVA is $1,500m/70 =$21.4m. This implies that if JPM’s 5y CDS spread widens by 1bp by the end of next quarter, DVA profit will be $23.3m and FVA loss will be $21.4m.
Collectively, the DVA-FVA gain would only be in the region of around $2m, this represents a significant reduction in volatility of bank earnings due to change in their credit.
Hedging and the move towards XVA
Looking back, significant spread widening led to controversial bank earnings in the third quarter of 2011, when DVA gain for Morgan Stanley stood at $3.4bn and for JPM $1.9bn. Banks are actively attempting to hedge DVA volatility, however it is has proved difficult given that they cannot trade their own spread. But one can see that combining FVA with DVA significantly reduces volatility and alleviates necessity of such a hedge. This is therefore an important argument toward centralized handling of so-called XVA, which comprises of CVA, DVA, FVA, and other adjustments.
Banks are actively attempting to hedge DVA volatility, however it is has proved difficult given that they cannot trade their own spread. But one can see that combining FVA with DVA significantly reduces volatility and alleviates necessity of such a hedge
An ongoing industry debate centres on whether traders include FVA in price and pass it to the counterparty or whether it should be absorbed by the bank. By reporting FVA as a cost, JPM is definitely favouring the latter. Apparently, after total FVA is calculated, it is allocated to particular trades and charged to specific desks.
To avoid considerable FVA costs, banks must attempt to reduce their borrowing rate also in addition to choosing trades that minimize funding cost. In any case, it is imperative to have a margin profitability calculator which compares trade P&L (quoted price after subtraction of risk-neutral price and bilateral CVA) with FVA and other incurred costs, including regulatory capital charges. Following this analysis, only then can a decision to execute any particular trade be made.