How sure are you that your models are correct?

Karl Smith makes a really good point:

Now imagine that you withheld a payroll tax cut or food stamp relief or any other program on the basis of fear about long term budgets. Depending on your macro estimates somewhere between millions and hundreds of millions of people suffered for this. 

What did you get in return for their suffering? 

Absolutely nothing. Nothing. Nothing. 

Every time you ask a real living person to suffer for some future goal you have to know that you are betting their well-being on your being right about the future. 

How sure are you that you are right? 

Austerity costs with probability one. Attempting to effect long term growth is always a gamble.

I do think that this point is worth remembering in policy discussions.  Models of distant time periods (say 30t o 50 years in the future) are subject to dramatic changes in assumptions.  Could the people living in 1890 (who had never seen a plane) have imagined what 1940 (and the air war of the Battle of Britain) would be like?

This is not to say that we should be reckless.   But policies like austerity in a time of high unemployment have immediate and real costs.  To presume that one is really preparing for the future one should be really, really confident that one can predict it . . .

EDIT: As a clarification, this is much more salient for things like Health Care costs where things like technological progress could completely change the growth curve and less of an issue for Global Warming where we have an observable and deterministic physical process.