We hear a lot about "complexity" as a contributing factor to the financial crises. Steven Schwarcz (Duke) pins blame on the complexity of financial products like asset-backed securities.
We need to be precise about what "complexity" means. In a paper (not quite ready for ssrn, but which I'll post in December), I outline three types of complexity for financial products:
1. Contractual complexity: this form of complexity measures how intricate the contractual terms of a financial product are. For example, how complex are the terms governing subordination and how many tranches of securities are being issued. I don't think contractual complexity poses as big a problem for sophisticated investors.
2. Derivative complexity: this form of complexity is created by the fact that asset-back securities and derivatives derive their value from underlying assets. But those underlying assets may in turn derive their value from other assets -- mortgage-backed securities are used to create CDOs which are used to back CDO squareds. Rinse, lather, repeat ad infinitum.
The problem with derivative complexity is that small errors in calculating the risk associated with underlying assets are magnified at each step of the securitization and credit derivative chain. So that a small error in calculating the risk of default on mortgages can translate into huge errors in pricing CDO squareds. The risk of error is magnified when there are unexpected correlations in losses on assets. For a great paper on this topic, see here. The problem is that, given multiple layers of securitization and hedging, it is now extremely difficult for any investor or swap counterparty to identify the ultimate assets that underlay their financial instrument.
Derivative complexity is a huge problem for financial products.
3. Systemic complexity: This form of complexity captures the idea that the value of financial instruments like asset-backed securities depends not just on the value of underlying assets, but the value of substitute financial instruments in the marketplace. These values, in turn, depend on market interest rates and on the trading behavior of multiple financial institutions. As I noted in an earlier post, homogeneity among the trading strategies of many firms can pose serious systemic consequences.
One approach to forms of complexity would be to put regulatory limits on the complexity of financial products. I'm not sure (1) how these regulations would work effectively, (2) if we understand the cost of prohibiting additional risk spreading, and (3) if this is politically feasible.
So I explore alternatives of using technology to help investors - particularly but not exclusively the sophisticated investors who purchase asset-back securities - better understand these forms of complexity.
One proposal is to require that assets going into a securitization have data tags affixed to them to allow investors to trace back and find the ultimate underlying assets that back their securities. This builds off the SEC's XBRL initiative, which requires issuers to affix data tags to their electronic SEC filings to allow investors to download information directly into spreadsheet and other analytic software. Data tagging assets has already been proposed, but I'm not sure if policymakers understand why this is so important.
Technology can also help us improve securities disclosure beyond asset-backed securities. I also examine whether we can make all securities disclosure more interactive. For example, financial disclosure ranging from Value-at-risk models (which attempt to quantify the financial risk a company faces) to valuations of inventory contain embedded assumptions. Interactive disclosure would allow investors to change assumptions and see how financial numbers would change.
We can also learn quite a bit from software about re-designing the "look and feel" of all electronic disclosure to help investors sort through volumes of disclosure. Consider how much easier computers with icons and drop down menus are (courtesy of the graphic user interface) are now then when we had to type in command prompts into DOS. Redoing the "menu design" of disclosure also builds off ideas from behavior law and economics about structuring the menu of choices available to consumers to improve investor decision-making in the face of cognitive biases.
I've already posted about open source risk models.
This is a work-in-progress and feedback is welcome. For one critique of this paper, see here.
TrackBack URL for this entry:
Links to weblogs that reference If Complexity is the Problem, is Technology the Cure?: