Skip navigation

As part of my work, I rely on statistical analysis to guide my thinking and broader analysis.  Lately Ive seen correlations go haywire, as can be expected in times of stress.  Just to give a flavor of the industries that have been hit: oil rigs in the US and Canada, furniture manufacturers, RV manufacturers, auto manufacturers, even GDP (mainly consumption and residential investment).  For a credit analyst, the all time most worrying of these is the path of recovery rates, which are tightly tied to default rates, as default rates rise to unprecedented levels.  These trends and their correlation to other indicators no longer follow historical patterns.

As a result, our models can only be used if we extrapolate trends, which is very risky.  So as I regress variable on variable on variable in an attempt to at least define the contours of a pattern or a trend, I find myself forecasting/predicting beyond the reach of applicable historic data. If anyone has an idea how to handle this, then please share it.  I have some ideas, such as adjusting predicted outcomes by the lower of the error terms (assuming that tings cant REALLY get as bad as as the model usually predicts), but I would love to hear from the blogoshphere.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: