I can't believe no one has mentioned the biggest hole yet in your argument:
Of the 168 (or whatever the real number is) Americans who broke 14:20 in the 5k last year, how many do you think had broken 14:20 in previous years? Just because there are 168 people who run sub-14:20 in a year doesn't mean there are 168 NEW people who have run that kind of time for the first time.
Considering that, your emistate of 120 per year looks ridiculously high.