To repeat: It's mostly not the must-have subscription journals' content that is going into the Gold journals. The must-have journals are doing fine, and their prices are going up, not down.
Wrong comparison (doubly wrong!):
1. It's the non-hybrid Gold content that you should be comparing to subscription content, if you want to believe Gold is driving down the price of subscriptions rather than just adding to them (double-payment).
2. It's hybrid Gold that allows double-dipping (which I'm afraid you also don't quite understand).
Hybrid Gold can certainly be monitored:
A. Total yearly hybrid-journal revenue per article is (total subscription revenue)/(total number of articles)
B. The yearly number of those articles that are Gold can also be counted.
It's only not double-dipping if the amount that it increases annually is based on the article increase minus the hybrid-Gold article revenue. (And even then, the institutions that are paying for the hybrid Fools-Gold are simply subsidizing every other institution's Fools-Gold rebate, while they get back only 1/Nth of their own. Not a saving till most articles are hybrid Fools-Gold (as publishers no doubt wish they will be!)
Not clear why you want averages here. Within a given hybrid journal (which is the only place this comparison makes much sense) the figures you need are: subscription-only revenue per article, APCs per article, total APCs per year and total article increase per year. Knowing that, you would know whether and how much double-dipping was being done.
One thing you can be sure of: total publisher revenue per article will not (be allowed to) decrease.
So forget about savings.