*n.b.: I'm reasonably certain that sooner or later equipment and electrical costs will become a very tiny part of the mining calculation. Probably sooner. Technology marches on, and as Bitcoin gains market share smarter and better equipment designers will enter the fray. I expect massive - and I mean massive - hashing power to be available within the next two or three years, for the cost of a decent desktop computer.
In the mid-term, I'm convinced that the most expensive part of mining will be space/property costs.
Honestly, I don't see that as an argument against a fixed block reward.
It is true that hashing power will go up massively (moore's law and all that, plus there is the specialization using ASICs), but the reward is fixed (50 BTC every x minutes). It doesn't matter how much or how little hashing power is added to the network, the rate of coin generation and therefore its effect on the global market would still be capped.
I'd say the main benefit is that it would remove some of the uncertainty around transaction fees being enough. Now we are left to wonder if transaction fees will be enough to compensate for the declining block rewards. If not, it will cause hashing power to drop (because of the cost of electricity, et cetera) and if it drops too much a 51% attack might become feasible and more profitable.