I’ve been thinking through the ideological background to the various discussions I’ve been having about methods for pricing complex securities these last few weeks and wanted to put some thoughts down, perhaps to have a subsequent discussion but also to sort, in my own mind at least, some of the issues. I’m thinking this through as I go so apologies if it’s not robust…
I think key to the discussion is that the market is trying to find efficient ways of valuing instruments in illiquid and dark markets. Moreover we aren’t typically trying to read the future per se but we are trying to determine what the value of the instrument would be if they were to exit today. The adjunct of this is that we need a view on future performance in order to determine what we think any prospective bidder is likely to think of the asset being sold.
However, the anchor here is its exit value today rather than its exit value in the future. In the last couple of days I’ve been comparing the types of analysis I would have performed when structuring CDO/CMBS/RMBS etc in my previous role with those types of analysis used to value these instruments. They differ in two key respects.
Firstly the analysis carried out in structuring these types of transaction is a ratings based approach. Always Monte Carlo but Ratings based nonetheless. The implication being that risk is horizontally tranched in order to capture parts of the risk profile for differing levels of reward for investors. By definition different tranches are expected to pay out under their own specific stress scenarios – AAA are harsher and more punitive than BBB for instance with the proviso that the rest of the structure can collapse in a AAA scenario but the AAA investors will get all their money back. i.e. to get AAA on a tranche that tranche has to survive all the stresses one would reasonably expect it to suffer under a multitude of possible outcomes. When assessing value on these bases the approach is again to anchor analysis around the ratings. The circularity here is obvious but still unrecognised (and by that I mean people see the emperor has no clothes but no one’s prepared to start laughing…yet) by most of the market and especially the regulators. Indeed I have argued strongly in the past in various official forums that we are in a state of regulatory capture via the Basel II accord which enforces a ratings based approach. Even if that approach, by definition, leads us to the problems we have now by conspicuously incentivising agents to work the rules (e.g. banks working the AIRB/IRB/Standardised approaches in the Basel II rules to manage their core capital ratios as thinly as possible).
However, the point was not that we determined some value using these approaches but rather that we defined our assets according to a market recognised benchmark. The market would then price to the benchmark. The pain has come, in part, because the benchmarks have been found to be wanting as fundamentals have deteriorated – in two ways. There was an implicit assumption that at worst the environment was in equilibrium – that is it was stable and wouldn’t get any worse even if it didn’t continue to improve indefinitely. The second is that even though historic numbers covering previous periods of stress were used to determine the benchmark they were, because of the probabilistic approach, given an appropriate weight depending on how likely it was considered that such events would occur again. The subtlety here, often missed, is that the probability of occurrence was essentially derived from translating an event relaxation time (the time between successive periods of stress) into a probability that at any particular point that stress would be experienced.
Somewhat differently the traders worked to an alternate mindset. We sold instruments designed to give a certain return. Traders bought based not upon potential capital returns but upon income streams for a set risk profile (our aforementioned ratings based view). Now the traders I know in London used similar methods for analysing instruments as I used to structure them – Monte Carlo based approaches for getting sign off from credit (short hand approaches like DMM/DCFs for getting a feel for the instrument). The better ones also factored in correlation assumptions which are typically entirely absent (or at best extremely crude) from structuring models. However, regardless of their sophistication they were buying for two reasons:
1. To meet investment criteria which are/were normally ratings based. It was generally accepted that the idea of a rating giving an indication of likely full payout on capital for the older was established within structured finance.
2. The investment would generate an acceptable return when set against a particular rating
For them a valuation was important but not the primary driver for buying (at least during periods of stability – this has changed now both as those caught long deteriorating investments are in trouble and those with cash are wondering when the best time to make a capital based investment will be.
For all these parties though it was the time the instrument was being held that was the key feature of their behaviour. These types of instrument would be traded as margins changed (and if there was a significant change in their principal value) but generally they were held and considered based upon their longer term worth. In marginally volatile markets where Fair Value isn’t pogoing like a kangaroo one could damp any concerns one might have about potential sensitivity to discontinuous perspectives regarding credit quality.
This differs significantly from the essential nature of what valuation is for. It is attempting, in most cases, to provide a value of an instrument for reporting purposes as if it was being exited today. Investment criteria, strategy and sentiment are irrelevant in so far as they are not quantifiable. What it cares about (per accounting standards) is what a party would be able to sell the instrument for if they had to do so. Now setting discussions about distressed markets, forced sales, illiquidity and market inefficiency aside a number of issues raise themselves in light of the type of approach structurers have.
Fair Value assessment is not fundamentally concerned with the future value of an instrument just in the risk neutral value of the instrument today. So whilst the future value is an important component of the analysis it isn’t the driving force as such. It’s a second order concern. If we think of it in terms of Boolean logic, it is one of the second order inputs that feeds into the first order result in our hierarchy. This means the methods we may consider suitable for determining fair value may well differ from those the market would employ if it was seeking to structure or invest. (i.e. this is the difference between buy side and sell side and presumably one of the reasons accounting standards define Fair Value as the amount sold for not the amount it could be bought for).
It strikes me that two approaches could be characterised thusly:
1. Ratings based approach – this looks to the market to determine what the market currently thinks suitable returns are for the ratings benchmark.
2. Asset based approach – this looks to the underlying assets and attempts to determine what the market believes will happen to the underlying assets. This is fully decoupled from the ratings of the tranched Notes.
Both seek to build up an idea of appropriate default rates and recoveries – one referencing the benchmark, the other taking its lead from current implied market expectations regarding the underlying assets. The two clearly overlap. However I’d liken it to trees planted in the same soil which have grown together over time.
In this sense I think both can capture current market expectations of value from a Fair Value perspective and this is what we’re asked to do by the majority of clients. However, reading this article from Bloomberg made me think through where and how the market is likely to change structurally.
Now the thrust of this article is that AIG is looking like it will have to write down an additional $30bn of assets – mainly relating to CDS contracts. However they made, their defence, the following statement: “Our methods have been thoroughly vetted and externally evaluated,” Lewis said.
However, regardless of this it’s becoming apparent that these vetting procedures (and the external evaluators) have failed to capture the real risks – indeed they have consistently valued the instruments too highly. Criticising the methods AIG used, Tanya Styblo Beder said “It’s not that all of the models are wrong,’ said Tanya Styblo Beder, chairman of risk-management adviser SBCC Group in New York. “The problem is that people made simplifying assumptions so the calculations were manageable and then had no warning labels to help the users understand the ramifications of these assumptions.”
And finally in a paper titled “The Economics of Structured Finance” that will be published in the Journal of Economic Perspectives, a group of Harvard Business School professors says the greatest problem in the market for CDOs and other structured securities isn’t the decline in value of the underlying assets because of the credit crunch. It is that the securities were overpriced from the start because the models failed to assess the risks, the professors said.
Now, the market is doing as international accounting standards require, and we do so using cutting edge technology. I have no problem with our integrity or expertise in relationship to the work we do.
What I’m thinking about is the road ahead. It is a plausible scenario that ratings will continue to be downgraded and as such will offer a declining credibility within the market (but not disappearing as long as the state of regulatory capture pertains) such that valuations performed one day could collapse on another as ratings are downgraded (e.g. consider CDOs which were Lehman originated which were downgraded overnight from AAA to D). The downgrades imply a deterioration in the instruments which is not necessarily related to the fundamental performance of the asset.
What is likely to happen here? The market is likely to remain confused, as both approaches to valuing these types instruments are valid under accounting rules and hence are used by members of community as authoritative measures of value.
A new route to valuation needs to be constructed. One that allows us to effectvely capture the risks we face.
No comments:
Post a Comment