I have been on the run since my C-CAF presentation in Silicon Valley September 25th. This week I attended a Board Meeting for Social Compact in Washington DC, and today I spent 45 minutes on the phone with a BusinessWeek journalist. We shared thinking on the subprime crisis and discussed ways in which technology can help unpack the problem and put better safeguards in place to head off similar problems, or as Mark Twain put it:
"History doesn’t necessarily repeat itself, but it does rhyme."
Crises of the magnitude we are now witnessing are avoidable. A key aspect is to wrap common sense around data-driven solutions so that underlying assumptions are transparent and model inputs are comprehensively formulated and model outputs are thoroughly validated both quantitatively and qualitatively. The comprehensive part is necessary for proper context to view the problem, and it necessarily entails some detailed consideration of combinations of factors. I contend that credit modelers have been delegating to much of the job to models and have side-stepped the exercise of working through all of the possible cases that can occur. OK, this sounds like some additional work that is less fun than tweaking sophisticated mathematical formulas, but judging from the headlines these days it is a worthwhile exercise that should become mandatory as new regulations emerge. Also, we must achieve consistency in risk assessment throughout the lending value chain from origination, to servicing, to portfolio performance monitoring, to collections, to packaging for sale to investors, and so on. A more comprehensive framework affords a closer fit to the business reality. The result of addressing these concerns is a Comprehensive Credit Assessment Framework (C-CAF) that is more accurate, more consistent, and more transparent than the siloed loan processes that are in place today.
My co-author, Dr. Zhang, and I are now immersed in our latest project with John Wiley Publishing. It is a book that describes the C-CAF in layman’s terms and provides some great examples of how it can address current shortcomings in this area. The manuscript is due in mid-November, so I apologize in advance for not posting as frequently as I would like. You may see some guest blogs in the coming weeks and I intend to return to more regular postings by Thanksgiving!