Hurricane forecasting and other difficult jobs

It should surprise no one that forecasting the number of hurricanes in a season is a difficult task.  The report (after the break) illustrates the challenges and the costs associated with inevitable forecasting errors.  I will also discuss the implications of this for EBPM.




Hurricane forecasting problems

This a classic problem for agencies.  They are asked to forecast and then criticized if they don't nail the prediction.  No one can predict the number of hurricanes we will experience.  Over the long run, the predictions have been pretty good.  When they are off, though, they get hammered.  This is then trotted out to discredit the organization and, somehow, the entire debate over climate change.

In some ways, this is the dark side of EBPM.  Making data-based policy often involves specific predictions -- more specific than traditional policy dialogue.  When one says there may be hurricanes this year, it is hard to be wrong.  When you say there will be seven, it is hard to be exactly right.  One of the chief challenges of EBPM will be managing expectations.  We need to make people aware that all processes are stochastic (they include random components that make measures go up and down a little bit) and exact predictions are difficult.  We will be wrong.  In the long run, we will be better off with data.