I sent a private response to Roma, which was a more wordy version of
 the following points.
 
 - LISTSERV is a distributed system, figuring out the "complete"
   resources used to support a list would be extraordinarily complex.
   Thankfully, it's also completely unneccesary. :)
 
 - Any model that estimates a well defined subset of the total resources
   will probably be acceptable.
 
 - One reasonable model is to estimate the local resources that will be
   dedicated to the list (both machine and human).
 
 - Since they site in question doesn't charge for machine resources,
   use some rates derived from a sample of other people's rates.
 
 I wanted to post this to let people know why Roma might ask you for
 your chargeback rates. :)
 
 -jj