He doesn't take care, and so his latest work is gleefully used by Andrew Bolt to prove that activities directed towards 'global-cooling' are a waste of time. What results is an insight into the workings of noble-lie theory.
It is nevertheless possible that ‘Earth Hour’ may have had a significant effect in a more localized area than New South Wales. The Energy Australia estimates were based on Sydney CBD only, and so may not be directly comparable.
Keep in mind that the majority of the 57% of sydneysiders who tuned-in and 'switched-off' would have necessarily been in the suburbs, because the CBD doesn't support 2.2 million people, except on NYE, and it certainly didn't that night.
Taking the point estimate of the EarthHour Dummy variable in the full specification after controlling for EarthDay fixed effects, the estimated impact of half of Sydney apparently sitting in darkness using no appliances was to reduce statewide electricity use by around 2.1%.
Here's Solomon's conclusion:
This is a useful, albeit rough, example of the potential economic cost of some currently debated policies that seek to reduce Australian greenhouse gas emissions by 60% by the year 2050.
All it tells us is that a 10.1% drop in electricity usage observed within the Sydney CBD and not the suburbs, results in a 2.1% drop in electricity being consumed in the state of New South Wales.
Is that useful? Solomon's right about it being rough, though.
Italics and bolding mine.
Global Warning Climate Change