A recent article in The Wall Street Journal caught my attention, but its news received no other attention that I saw. The article, “Americans Are No Longer Gluttons for Electricity—Thank the LED Bulb” (sorry, it may be behind a paywall), pointed out that average per-residence electricity use in the US has declined since 2010 according to statistics collected by the US Department of Energy.

But wait, there's the inevitable caveat: “Last year, driven by a hot summer and cold winter, the numbers ticked up to an average of 10.97 megawatt-hours per residence, but the overall trend remained down.” Data points that don’t fit the explanation are almost always due to the weather!

Is the downturn from 2010 through 2017 the base of a longer-term trend, and 2018 was the aberration? Or maybe it’s the other way around, and analysts are seeing what they want to see in the numbers? One thing seems pretty clear: the switch from incandescent bulbs to LEDs has had a definite impact on residential consumption. Still, lighting only accounts for about 10-20% of home energy use so saving, say, 80% of that 10-20% by using LEDs reduces consumption by only a few percentage points—certainly meaningful, but not at all the whole story.

Surely, there are contrary consumption trends going on that will both increase and decrease electricity usage. The article points out that in the early 1950s, air conditioners were just starting to appear in homes, and only around 9% of homes had a television. Today, 87% of homes are air-conditioned, and nearly all of them have at least one television with 39% having three or more. On the proverbial other hand, all of these units are far more efficient: just compare a 1950s 14-inch tube-type TV (~500 W) to our much-larger flat-screen TV today (~200 W). Still, we do have lots more “stuff” and even if each item is fairly low-power for the functions it provides, it all adds up.

I also wonder about the accuracy (not precision) of these numbers. First, as an engineer, I always find “average” to be a handy but very risky number to use, because it inherently smooths out a lot of important top-level granularity. Further, while I assume the data-analysis folks have various correction factors they routinely use (just as is done for other large-scale statistics including gross domestic product), these are often based on historical trends, which may no longer be valid.

Does the energy number include those who are totally off the grid, a small but perhaps significant number? What about those who generate their own power (via solar panels or wind) and then feed the excess back to the grid—do we know how much power they used?

There are at least two ways of looking at this data credibility. Perhaps the numerous smaller errors all cancel out in the end, and the final result is close to correct (but how would we know that?). Or perhaps they tend to mostly err to one side thus creating a skew in conclusions which, like it or not, is used for projections and policy setting. There will be “experts” and pundits on both sides, some saying, “This is great news, we’re making progress” and those who say, “This isn’t good enough, we need to do a better job.”

Even more worrisome to me is that we tend to extrapolate these numbers 5 and even 10 years into the future. We all know that such projections are error-prone, especially when the numbers at the starting base are close, such that small errors there can result in large errors in those projections. One of my many peeves is when pundits and lazy reporters resort to the “at this rate, such-and-such will happen in X years” type of analysis, which implies that this rate will continue for the entire period, that there are no nonlinearities, saturations, or limits, when we know those considerations exist in the real world.

What’s your view on the changes in per-residence electricity consumption over the next 2, 5, and even 10 years? Will improved energy efficiency in everything from small wall chargers to lighting, insulation, and HVAC outweigh the increase in the number of devices? Or will the sheer numbers of these devices and products overwhelm the energy savings? And what will be the relative impact of more EVs on residential electricity consumption?

Bill Schweber is an EE who has written three textbooks, hundreds of technical articles, opinion columns, and product features.