Thread by @SteveCicala

My ongoing work on electricity consumption as an economic indicator in the @WSJ by @russellgold. I will keep this thread going with updates...

wsj.com/articles/plung…

Here is raw electricity consumption in Northern Italy by hour of week. The top curve is the week before any regional quarantines were instituted. The middle curve is last week. The bottom is this week. This is all raw data with no temperature adjustment.

The temperature adjustments are important because of heating and cooling. In this first week of closures there was about a 3% drop in consumption, but it didn't really look like anything in the raw data.

That has very much changed in the last week.

Here is temperature-adjusted, percent change in consumption since 21 February, 2020--a Friday. Not a lot of change over Saturday/Sunday, but it starts to show on Monday. The hump you see a few days later is the next weekend, just before major shelter-in-place orders were issued.

I'm posting these now because we're making major policy decisions to support the economy without really knowing what is happening in this fast-moving situation.

There was a similar discussion underway in the fall of 2008. The economy had been declining for almost a year, but the recession had not yet been called. It takes a while to pull in all of the relevant data, and ensure it's not a brief, transitory shock.

Here is what we would have known in real time via electricity consumption, plotted against the first preliminary and final employment numbers from @BLS_gov. The start of the recession was officially dated as Dec 2007--in Dec 2008.

Here is the same figure with a 2-month smoothing bandwidth.

Because hiring and firing is costly, it makes sense that firms would ramp down production, perhaps move up maintenance schedules *before* laying people off. On the other end, firms are trying to ramp up with skeleton crews before hiring new people.

I'll follow up here with current estimates from the US, which really only started to institute serious measures (but not widespread testing...) on Friday. There's plenty of refinement to do, so it's tricky to not over-react to what might be a transitory blip.

In the meantime, a special thanks to @ivanhigueram who has been working with me around the clock to get this out, and the folks at @wattTime, who have made pulling consumption data a lot easier since I first started doing it!

An update on the picture emerging from Europe:

Italy has continued to deteriorate with the more stringent shut-downs implemented: Now 30% below early February (including temperature and seasonal controls)

Other widely-reported shut-downs are visible in the data. Here is Spain, down ~15%:

I'd love to report the results from France, but they haven't updated since mid-March. So here's Belgium instead, headed toward -20%:

Germany has a noisier picture, but is down 10%.

Poland is also down 10%.

As is the Czech Republic and most Eastern European countries.

Austria down 15%:

The Nordic countries do not appear to be showing too much impact (though 2.5% is half of the fall for the Great Recession in the US). Here's Sweden:

Perhaps a reporting change in Finland? An important caveat is that this data is subject to change with revisions.

In any case, putting all of the EU (more or less) together and weighting by consumption in early-mid February, there has been about a 13% decline as of 25 March 2020.