*/By Rick Nason, PhD, CFA
Partner, RSD Solutions Inc/*
When I am wearing my academic hat – as opposed to my consulting hat – I
get so exasperated by my academic colleagues who willfully and purposefully
ignore a model or an idea, or conversely adopt a model or an idea, solely
based on the data available.
As an example I had a colleague say to me today, "yes, we considered that
but we could only get 20 years of data and for our model to be statistically
valid we need at least twice as much data so we dropped that variable."
I think every academic has heard (or said, or thought) something to that
effect in their career. The reality is that you do need data to get
results. However if you need so much data for statistical significance,
then perhaps you should consider if you are measuring a model or just
measuring noise. That is what I love about risk managers (and regulators)
– they never make that mistake!
1 comment:
That falls under the "never confuse activity with accomplishment" rule. Being so focused on the act of "doing analysis" we lose sight of why we are analyzing in the first place.
Andy
Post a Comment