Peter Eavis has a Dealbook post which, sadly, reiterates many of the common misunderstandings about derivatives valuation. Let’s start from first principles, and try to understand what is really going on here.
Why are we trying to fair value a derivatives book? First, to create a reliable P/L which accurately reflects the value attributable to security holders. We want the right earnings so that current equity holders are properly compensated for the risk they are taking, for instance. Second, earnings volatility is the paradigmatic definition of risk, so we want earnings to accurately reflect the swings in the value of a derivatives portfolio.
The concept of fair value therefore has at its heart the paradigm that valuation should reflect where something should be sold. Now in practice large portfolios are often sold in toto rather than instrument by instrument, so a relevant question for a portfolio which can be sold this way is ‘how would a bidder value it?’ The answer, typically, is that they would take the mid market fair value then apply a spread to the risks (e.g. a vol point or two on each vega bucket), so a reasonable way to establish fair value, often, is to value at mid then take an appropriate bid/offer reserve. (That last part is important – mid alone isn’t where you can get out, so simply valuing at mid is in violation of the accounting standard and if your auditor lets you get away with that, fire them and get one who won’t.)
We now have two problems:
- How do you establish where mid market is?
- How do you decide if this paradigm works for your portfolio?
Neither of these are particularly difficult questions. Where there is a liquid market of buyers and sellers, then you use market prices. Where there is a liquid market in related instruments, you use those prices to calibrate an interpolator model. Where there isn’t either of those, then perhaps you can use quotes rather than real trade prices. Or if that fails, you make something up. In the latter two cases, though, you will typically need a valuation adjustment to reflect likely uncertainty in your price. Take your best guess, but then take a reserve to reflect how wrong that guess might be*.
The method will fail if your portfolio is a large part of the market, or would take a long time to liquidate. In this case the principle of valuing where you could close out the book would suggest taking an extra reserve† to reflect the price change you would cause if you tried to sell the whole portfolio. Just because you are buying and selling 1% of the position each day does not mean that the prices those trades happen at are reflective of where you could get out of the entire position.
In the Whale farrago, JP (it seems) neither took a prudent bid/offer valuation adjustment nor took an adjustment to reflect the size of their position. This has nothing to do with derivatives being murky and everything to do with not complying with the basic idea of marking to where you can get out of the portfolio.
The usual line peddled at this point is that none of this would be possible if all derivatives were traded on exchange. That’s false. Many exchanges are replete with illiquid contracts where the last published trade price is not reflective of where the current market would be were anyone to try to trade. (Just trying looking at pretty much any far from the money single stock listed equity option, or any commodity/energy contract away from a few benchmarks.) Exchanges are not a replacement for good product control teams trying, daily, to test prices: indeed, if their prices are used without thought, they can be far more dangerous than letting the traders tell you where the market is, then diligently checking.
Financial instrument valuation involves a lot of grunt work. Multiple data sources, experienced individuals, prudent reserves/valuation adjustments and skepticism are all required to do a good job. That’s true of exchange-traded instruments and OTC ones. The estimation of fair value is an important discipline, but it is vital not to lose sight of the fact that it is, despite all this work, an informed guess. There is no platonic ideal of the right price out there waiting to be discovered, especially not for any really big position whether in securities or derivatives. We can rightly blame JP for not doing a good job in forming its estimate, but we should also understand that perfection is unattainable. If really want to know where you could sell a position in any financial instrument the only way to find out is to sell it.
*You do it that way rather than using a ‘prudent’ (i.e. wrong) mark first because you want the price and its volatility to be the best guess (especially if you are hedging), and second because you want to flag to management and owners the uncertainty in that price.
†One of the many changes in accounting standards that have made things worse in recent years is that these size-based price adjustments are often disallowed in US GAAP. What were the FASB thinking of?
Update. Wot he said, too. Especially the bit about loans. Indeed, this qualifies for quote of the day status:
Compared to, like, banking, JPMorgan’s CIO portfolio was a model of transparent valuation, even with the fraud.