Afaik energy use being correlated with economic growth depends on how much of the economy is based on energy-intensive industries, like manufacturing.

The US, as many other developed countries, has been shifting out of manufacturing economies to service economies for a while now.

Case in point: Nothing at Disney stipulates a correlation between their "economic output" and their energy consumption.

Meanwhile, a steel/aluminum plant in China will have a pretty direct correlation between economic output and energy input.

> Case in point: Nothing at Disney stipulates a correlation between their "economic output" and their energy consumption.

Not convinced. Whether movie creation or Disneyland, it all seems O(n) to energy use, except the constant factor is much smaller than for a steel plant, so we don't notice the correlation yet. Computers need electricity, man-hours ultimately have an energy cost, you need space and energy to entertain all the Disneyland visitors, etc.

I think I've found the answer to my question in other comments, though. The article focuses on US economy, but given the scale of imports, it doesn't really make sense to treat US economy in isolation. Just like I suspect the "dematerialization" is an artifact of not counting inputs to outsourced manufacturing of imported goods, I now suspect the decorrelation of energy use and economic growth is just an artifact of not counting the energy embodied in imported goods.

As I understand it, a service-based economy is more energy-intensive. A service based economy requires a pre-existing industrial base; services activity are actually managing actual stuff streams, and streams about these streams. Plus, you need heating, transportation, internet etc for all of your office workers, which requires energy.

The absence of obvious correlation doesn't remove the need. Just like there would be no advanced economy without a working agricultural infrastructure to provide food to everyone...

Doesn't that really depend on what the services are based?

Driving food around in a gig economy is just another logistics jobs, driving refined metals around between different refining processes is also just another logistics job, but one with more "energy-intensive cargo".

All that tech, the cars, the infrastructure, they need those base materials you can't just "service into existence" as you can do with IP and copyrights.

Sure, all that tech needs electricity to run, but the really big emission and energy footprint they create during their creation, not during their use.

Just a hypothesis but I would guess it's because services and productivity improvements are driving economic growth rather than products.

I bet reoccurring service fees are accounting for an increasing proportion of spending to the average American. The traditional electricity, water, phone, cable, and internet services have largely increased in costs while also expanding to include house cleaning and sub services like Netflix, Amazon Prime, Hulu, Spotify, Patreon, Adobe, Dropbox, Curiosity Stream, etc. And that's not even including other sub services that offer physical goods at inflated prices over retail (butcherbox, stitchfix, wanc, birchbox).

There's just so much shit to subscribe to anymore.

Edit: I forgot, digital-only video games and movie rentals too.

That doesn't explain the fertilizer usage very well, and from what I understand the story with ag is improvements in efficiency.

Agriculture is mostly down to glyphosate resistance, i.e. GMOs.

If you can spray a field with glyphosate, you don't need to till at the start of the season. Tilling has huge externalities, both the carbon emitted from the powerful plow used, and also the runoff from the disturbed soil. That runoff, which can cause all sorts of contamination, algal blooms, etc., is also leeching needed nutrients from the soil.

If you avoid tilling, you ultimately need to use less fertilizer.

Coupled with decreasing acreage for a given yield, that might be an ok trade off.

The article says that imports were accounted for.

The accounting seems to be done per category. Manufacturing is a wasteful process, so when you import finished goods instead of manufacturing them yourself and account on basis of mass, a lot of used mass is missing.

Now ordinarily, I'd assume it's a trivial thing that obviously must have been corrected for. But recently I read that apparently, US's disappearing manufacturing was hidden in statistics for a long time, because some people miscounted semiconductor industry as if it was manufacturing appliances. So I guess mistakes like that happen.

Only imports of raw material. Not finished goods. And don't forget all the resources embodied in the actual means of production! That's a lot of heavy machinery.