Dear SOCNETters,

I have a simple question for which I would like to hear your views. I got a referee who asks us for a “statistical test to discriminate whether these distributions (of financial exposures, in our case) have power-law or exponential tails”.

What test would you use? Note that whether the tail distribution is power law or exponential is not the main hypothesis of my paper. It’s more descriptive, so the test should be something very simple.

I know that testing power laws has been a long-standing discussion, and here at SOCNET there was a discussion after Stumpf & Porter’s Science Perspective article in 2012. But we are now 3 years further, so I just wondered what you would do if you are asked for a power-law vs exponential tail test that is simple and statistically sound. Any recommendation would be appreciated.

Kind regards,

Marco

_____