Low airport fees in the US are largely the result of deferred maintenance on facilities. The "heavy lifting" in terms of modernizing the airport facilities in Canada has been largely performed in the past decade of high AIFs, and expansion projects are now few and far between. Meanwhile, a whole plethora of US airports, particularly ones which do not charge significant AIFs, are in quite dilapidated condition and will likely be forced to increase their rates quite dramatically in the future. I think FT'ers are well aware of some prime examples.

Having said that, the whole idea of allowing carriers to sell connecting flights between two points in the 'other' country is relevant for mostly low-fare situations -- ie: grabbing any revenue they possibly can over and above the non-avoidable cost of flying a seat. Or very high-end traffic, ie: selling seats into a market when everything is full with the domestic carriers.

For instance, in Canada, Thanksgiving is in early October. It is in late November in the US. The planes run absolutely full in Canada domestically on the morning after Canadian Thanksgiving, but the transborder flights are quite empty. If AC could sell the 50 seats on YVR-SFO-YYZ that otherwise probably would go empty that day, that's more usable capacity to the market without putting a single additional flight onto the network. UA, AA, etc., could do the same. Likewise, Canadian connections would become useful during US Thanksgiving, usually a fairly boring low-load November days in Canada.