Not sure I buy this GDP to BTU cost justification, as GDP is largely a bullshit stat in the first place. That said, presumably factoring in negative marginal productivity of debt is "an exercise left to the reader" if you wanna keep following that line of thinking.
I'd say the more accurate comparison would be comparing the net cost of the financial system versus the net cost of operating the global BTC mining network. Good luck with that, but I suspect the current financial system isn't particularly energy efficient either.
Considering how many goobers make their living off of seignorage and the money multiplier effect, supporting these leeches and the malinvestments they enable would make that cost potentially astronomical. In a way, this was the point this author was trying to make, albeit with what I would consider a poor economic metric for doing so (which surprises me, usually the institute is a lot better at resisting the temptation to use econometrics as a justification for arguments).
The most efficient system in this regard is almost always going to be the ones which are:
- Fastest, as processor time scales to energy usage.
- Lowest maintenance costs. Purely digital currency of almost any kind clearly wins here, as transport & manufacture costs of both paper and specie are guaranteed to be far far more for the same amount of value transferred via digital technologies, despite the "inefficiency" of BTC's PoW algo.
Lowest wasted economic output due to malinvestment. Any "Hard" money always wins in this category, and thus far running secure code based on global miner consensus seems to be working out at least as well as the old Global Standard if not better in that regard. This is where the global fiat ponz really falls flat on its' ass.
Anyways, if people actually cared about efficiency, they'd be going ham on BCH and other related coins with speed optimizations which are still based on proof of work. BCH wins here, not BTC.