Friday, June 27, 2008

Modeling in a complex world - watch your parameters

George Pelham-Box, one of the most influential statisticians of the 20th century, once remarked that "essentially all models are wrong, but some are useful".
Working with models provides enough support for his strong statement. Models can only capture the relationships that are actually included in the parameters. One-off events or events that are not included in the sample will not be able to be explained by the model. The new period of stagflation will be a strong test for any model developed in the last 25 years. Current models may not be able to track the actual behavior in the economy. Watching for a deterioration of current relationships may be the most important tasks for a systematic modeler.