November 23rd, 2008
There is plenty of blame to go around – like most man-made disasters, the causes are many and interlocking, and certainly no one intended the result.
One of the generally agreed-upon ingredients was the lack of transparency in extremely complex, yet widely traded, derivatives – mortgage-backed securities, SIVs, CDOs, CMOs, CDSes, etc. The opacity of these instruments made it extremely difficult to evaluate their underlying risk and value, which in turn made them hard to price and trade in volatile markets.
Which is a shame, because in principle the trading of derivatives, even (or especially) complex ones, offers significant economic benefits to society – these will be lost if the current panic results in this market being regulated out of existence.
But the transparency problem does need to be solved.
My friend Fred Hapgood mentioned that this ought to be the kind of problem that computers are good at solving. He’s right.
According to Walsh’s article:
SIVs and CMOs were pools of standard MBSs sliced up (often into 64 equal parts) and reassembled according to their riskiness, geographic concentration or other standards. In some cases, Wall Street whiz kids would break a 30 year mortgage into individual securities based on each of 360 monthly payments; then they would bundle thousands of the one-month payments into larger securities offerings. Finally, CDOs were pools of CDSs sliced up and reassembled to reflect default risks more precisely.
It’s easy to understand why these instruments are opaque – their risk and likely return depend on that of multitudes of other derivatives, which in turn depend on hundreds or thousands of individual mortgages and loans, sliced and diced every which way.
In principle these factors can be fully evaluated (after all, they ultimately depend on the creditworthiness of individual borrowers, something the financial world has always coped with). In practice evaluation is impossible because it can only be accomplished by pouring thru thousands of pages of documentation describing the assets and conditions represented by a derivative, and then doing the same for each of the other derivatives of which the derivative is comprised, etc., traversing the entire tree of rights back to the individual borrowers.
As Fred said, this is exactly the sort of thing computers are good at.
So, my proposal is that an appropriate regulatory body should define an XML schema to exactly describe the chain of rights, conditions and assets represented by any traded derivative. Issuers would be required to legally define the derivative in terms of the XML schema, which would be publicly posted in machine-readable form.
The schema would allow derivatives composed of other derivatives to be simply described as a set of links to (or copying of) the XML descriptions of the constituent instruments, with appropriate operators for expressing conditions, time-dependency, etc.
I imagine the regulator would require these XML descriptions to be posted in such a way that a potential trader could find them online in a uniform way – for example, an exchange might maintain a single database with links to the XML descriptions of all traded instruments.
The rest is details.
A standardized XML schema and the single rule that issuers must describe their offerings using the schema should allow automated analysis of arbitrarily complex instruments, all the way down to each individual loan or other asset. Each trader could do their own analysis, with their own assumptions and rules – the XML data would supply only the definition of the instrument to be analyzed.
It seems to me that this would solve the transparency problem in a robust and extensible way. Full information would be provided in a usable form before trading.
Since a portfolio is just an aggregation of assets, entire portfolios could be analyzed in the same way – by buyers, lenders, shareholders, regulators, and analysts. Markets would then be able to sort out, by the usual means, which kinds of derivatives make sense, who is solvent, who is taking risks, and how much.
I’d love to hear criticism of this idea.