Since the mid 1600's, tulips have come to symbolize economic bubbles. Excess demand for the floral beauties led Dutchmen to pay a hefty price, resulting in the tulip mania of the early 17th century.
One wonders if a June 24 article by New York Times reporter Gretchen Morgenson hadn't been inspired by this tale of yore. Entitled "When Models Misbehave," this prize-winning business columnist describes the challenges of assessing securities that trade in relatively illiquid markets. In the absence of ready buyers and sellers, traders mark to model, making assumptions about the future behavior of inputs such as interest rates. Unfortunately, problems may arise if the underlying assumptions make no sense. Consider the notion that past is prologue. Referring to the sub-prime debacle currently plaguing several large financial institutions, Morgenson describes 2006 and 2007 lending practices as overly generous and likely to tighten. Correctly recognizing that future supply-demand conditions for credit might change leads to an altogether different model outcome.
Lack of independence or "the fantasy that a firm's principals prefer" is another concern. Blind acceptance of model-generated outputs as gospel could mean a subsequent, and arguably tainted, outlay of serious money to other trades.
Morgenson has a good point.
It's easy to be lulled into false security with computer-generated numbers. Unfortunately, bad values beget bad economics. A computational flaw, unstable or inappropriate model and/or low-integrity data could end up costing investors millions of dollars. Trading decisions based on garbage are expensive mistakes.
Good model-building is a start. Validating, testing, revising and testing anew should follow. Heady stuff but anything else might be considered remiss. Importantly, it's up to investors to query asset managers about what's inside the black box.
Anecdotally, I'm not sure there is enough of this rigorous oversight happening right now. As an accredited appraiser, I'm disturbed by the laxity of investigation about valuation and the related process of risk management.
With new accounting rules on their way, we'll talk much more about models and model risk in future posts. In the meantime, click here if you'd like us to send you information about valuation and modeling.