Understanding the economic impact of datacentre deployment and how to engage critically with claims around jobs

[mrchrisadams]

Hello fellow Edge Ryders,

I have a pretty technical question about understanding the economic impact of deploying datacentres around the world, and I’m hoping some of the replies will help point me in the right direction to understand how claims around job creation are made in studies like the one below.

This report here, where the Data Center Coalition commissioned PWC to study the impact of datacentres in the US economy makes some eye opening claims about their economic impact of datacentres. Here are the headline stats they lead with:

Full report is available on the data center coalition website - here’s the direct download link to avoid needing to give your email address

Anyway, there’s an interesting comment here on LinkedIn:

Back of the napkin math, but that’s about 2% of total US GDP over that time period - which would make it twice as big as the entire agricultural sector, and half the size of all of construction. That seems unlikely, unless I’m missing something.

Source: Comment on Linkedin by Sebastian Moss

These figures seem to be based on a combination of a economic model, IMPLAN, and some public data.

So my question is this - how are these made, and how are policy makers able to critique them when they’re used?

I’ve ended up in the unexpected position where I’m doing a talk at the IEA in Paris in October, about datacentres, energy systems and climate, and while I feel relatively comfortable talking about grids and tech, when these figures are thrown around, I’m less confident.

I can see why policy makers and local governments would lap these economic figures up, but the main responses I see from people skeptical of their impact seem to be focussed on local resistance, without ever addressing the economic uplift claims.

If the hive mind here has any recommendations on how to interrogate claims like this, or even respond to them, I’d be grateful, and this seems to be an area of renewed discourse around digitalisation, particularly in Europe.

[alberto]

Uh, @mrchrisadams … this seems to be not very informative. Little better than junk, in fact. Let’s see:

• Total employment impact = direct employment impact x 6. This seems wild. 30 seconds of web search returns this estimate for the economic impact of the Biden’s COVID relief package (source): USD 1.9 trillion stimulus package yields one trillion per quarter x four quarters + 0.5 trillion per quarter x four quarters, back-of-the envelope. That means ~ 2 trillion direct expenditure yield ~ 6 trillion in total expenditure. Your estimation of the multiplier is about double what the data say for the Biden stimulus package, assuming the impact on jobs is similar to that on GDP.

• Employment is not necessarily additional. Biden’s stimulus package is, by definition, additional: without a political decision, this money would not have been spent. Not so for corporate investments. If companies had not built data centres, they might have spent some (or even all) of that same money in other investments. In that case, the impact on employment would be accordingly reduced, possibly to zero if every dollar spent on data centres displaces investment in other assets.

Hope that helps.

[mrchrisadams]

This is really helpful @alberto , thank you!

I was worried about getting bogged down in a discussion about the details of whatever model they used, but keeping at this level, from an argumentation POV is much easier for others to follow along, and much more inclusive in general.

I really appreciate the stimulus package as a comparator for sanity checking the high level assumptions - it feels like an elegant reframing of the core issue