The email contained just one word -- "Amen" -- and a link to a New York Times op-ed, titled "The Problem With Energy Efficiency."

As I read it, I wondered why someone would feel the need to parlay the occasion of the Nobel Prize being awarded to LED innovators into an opportunity to say something overtly negative like "LED and other ultraefficient lighting technologies are unlikely to reduce global energy consumption or reduce carbon emissions."

Then I read the byline and realized the op-ed was penned by The Breakthrough Institute (TBI), and it all became clear. That statement, and the entire op-ed, takes a real phenomenon and blows it up into an undeserved attack on a powerful energy savings innovation: solid-state lighting.

BTI has a longstanding pattern of putting out provocative pieces that receive strong criticism from both academics and industry practitioners. I won't dwell on that here, but you can read for yourself herehere, here and here.

From what I can tell from my brief conversations with well-meaning BTI folks, the organization has a noble mission: to promote government spending on deep R&D geared toward identifying long-term clean energy solutions, and to promote rational discussion of climate policy even when there are uncomfortable truths to be wrestled with.

Heck, with that kind of mission, I should love BTI.

But their tactics, which were illustrated by their op-ed on LEDs published last week, are really unfortunate. They include:

1. Taking a worst-case scenario that suggests that a near-term solution might, under some really questionable assumptions, turn out to be useless or even a net negative for the climate.

2. Using that analysis as the basis of some outreach to the general public which declares something like, "Hey, maybe energy efficiency doesn't actually help us at all!" Always with lots of buried caveats and instances of damning with faint praise ("Oh, but there are lots of other reasons to like LEDs, no really, we like them, honest!"). This is often paired with a misleading headline (which they then declare was not their choice) that makes the entire thing sound even more negative.

3. Hiding behind the knowledge that few people will bother to actually look into the details of the analysis, and then responding to all criticisms with "What's your argument, logic, evidence against what we wrote or our report? Substance please." (That's an actual quote from Twitter.)

4. Abdicating responsibility for the chaos left in their wake as anti-progress advocates ("Don't ban the bulb!") co-opt BTI's analysis and use it to attack innovators.

Well, bluff called on number three. I actually did go back and look up the study ("Solid-State Lighting: An Energy-Economics Perspective," Tsao, Saunders et al., 2010) that someone affiliated with BTI told me was the basis for the claim made in the op-ed that the "rebound effect" on lighting may be 100 percent or more. And because that paper relied heavily on the findings of another analysis ("The World's Appetite for Light," Tsao and Waide, 2010), I looked that up as well. What I found was illuminating.

Before we jump in, let's clarify a few things.

First of all, "rebound" is a real phenomenon that can indeed have significant impacts on net energy savings from various mitigation approaches. The idea is simple: if you save money by reducing how much you spend on energy, at least some of that saved money will go into additional consumption of energy, either directly or indirectly. Basically, more money in your wallet equals more consumption. This is not a bad thing from an economic perspective, of course. But if you care about energy-related emissions, this can be a problem. Some of those savings come back around as new consumption. In other words, they rebound.

But the magnitude of that rebound matters a great deal. And arguments between those who say rebound is high for energy efficiency and those who say rebound is low can get pretty fierce at times. As best as I can tell, the only reasonable conclusion one can safely draw from the available studies is that levels of rebound probably vary widely by technology, situation, etc. And it would also seem likely that many policymakers and advocates don't really recognize the importance of rebound when figuring out how much emphasis to put on various climate policy choices. 

It's important stuff in the right context. But it's not my context. My context is the LED entrepreneur, the home energy service provider, and the politicians who can do a lot to help or hurt LED adoption in the near term. People trying to make an immediate, positive impact (or the people fighting them). Those are the folks I deal with, rather than academics.

And so when I see a blanket statement in the New York Times that makes a fairly aggressive and absolute argument that LEDs don't save energy, I need to understand it.

So I looked at the studies from Tsao & Saunders and Tsao & Waide. And let me be clear: I don't intend to disparage the researchers at all; I am simply exploring the limitations of the analyses. Here's what I found.

The Tsao & Saunders paper is a very clever (and I don't mean that disparagingly) application of a "constant" that the Tsao & Waide study had turned up. Tsao & Waide looked at a data set made up of a composite of a number of small data sets, focusing on the consumption of lighting according to various surveys and estimations. One major part of their data set is a group of estimations of lighting consumption going back to the 1700s in Britain. Tsao & Waide acknowledge that they "consider those [data sets regarding] contemporary consumption of light to be much more accurate than those for historical consumption of light." Those contemporary data sets are based on surveys of various country-level light consumption studies conducted by the likes of IEA and Navigant.

Tsao & Waide looked at the GDP per person of each relevant country at the time of each lighting survey, and compared that to lighting-related spending. What they found is a fascinatingly straight line -- put together over that 300-year span, it would appear that lighting spending remains very constant, coming in at around 0.72 percent of GDP.

That's pretty amazing, actually. But does it make sense? That no matter how much money you make, and no matter what your circumstances, you're going to spend an average of 0.72 percent of your take-home pay on lighting? Across lighting technologies ranging from candles to kerosene to incandescents to LEDs? If true, this is a remarkably consistent basic truth about humanity.

But statistically speaking, if you dig into the study's actual data set, you find that much of that really predictable bit is driven by the early, less-reliable historical estimations. If lighting spending as a portion of overall budget doesn't swing wildly from, say, 0.2 percent to 40.0 percent across the data set, while GDP per person grows from $1,800/year to $30,000/year, you're going to find a "tight fit."

Meanwhile, in those admittedly more accurate contemporary data sets? The results are all over the map.

U.K. (2000): 0.324 percent implied share of GDP/person/year.

U.S. (2001): 0.644 percent.

China (2006): 0.381 percent.

China (2005): 0.341 percent.

Former Soviet Union states (2000): 0.933 percent.

OECD (2005): 0.598 percent.

Japan and Korea (2005): 0.72 percent

Australia and New Zealand (2005): 0.531 percent.

World non-grid-tied (1999): 1.165 percent.

So let's ponder these for a minute. First of all, there is little evidence that there's one cosmic constant level of lighting spending per GDP. In fact, there's a 3X swing here. Second, it's hard to even find any kind of relationship between GDP level and proclivity for light, since among the lowest spenders are the U.K., Australia and New Zealand, and China, and off-grid lighting ranks the highest. Remember, this is not a measure of the cost of the light; this is a measure of how much people spend for it. 

I look at this and I find very little evidence to support any firm conclusion that people will always spend a certain portion of their budget on light. It's just as plausible to look at this data and conclude that people with limited access to electricity are forced to pay more than they want to, and that in many advanced economies, consumers would be happy to spend less than they do today.

But based upon this 0.72 percent "constant," Tsao & Saunders then embark upon a lengthy exercise which I will attempt to boil down into the following oversimplified version.

No matter how much money you make, it turns out you will spend 0.72 percent of it on ice cream. Regardless of your health, your financial situation, your culture, or your tastes, you will buy the amount of ice cream you can afford with that immutable budget. Someone is cutting the price of ice cream in half. Therefore, you will buy twice as much ice cream. (Obviously, substitute lighting for ice cream.)

Now that we can see that real-world data shows us this constant isn't to be relied upon, however, why wouldn't we expect that people in emerging economies would be happy to see their spending on lighting go down by two-thirds and then choose spend the saved money on additional food, shelter and other items? I mean, isn't that what Tsao & Waide's data itself suggests might happen, since that group is apparently starting much higher than the "norm" they think they've identified?

So that's one fundamental problem with the analysis underlying BTI's assertion that LEDs are unlikely to make a positive impact. They rely upon a study that relies upon an assumption that is itself unreliable.

Let's look at some of the others, shall we? 

Capital costs

Of course, "lighting spending" is made up of more than just the energy consumption component; there's also the upfront capital cost. The ratio between the two is going to vary widely. And if the LED spending ratio is more skewed toward the upfront cost, that means less spending would be directed to lighting-driven energy consumption, even if the 0.72 percent magic constant holds true.

Tsao & Saunders make this problem go away. They go through an analysis which ends up suggesting that this ratio is also amazingly constant across all time, space and technology. You can see their rationale in the paper, and it's not completely unfounded. They look at cost curves and the like and conclude that their admitted approximation for the ratio will still hold true in the long term.

But it just doesn't pass the sniff test, does it? Anyone involved in the LED lighting industry knows that selling right now is all about total cost of ownership arguments to overcome the upfront cost barrier of more expensive LEDs. The energy savings more than make up for it, but it's a different ratio than what buyers were used to in the past.

Forecasting cost curves a few decades into the future, the studies' authors and their defenders say this different ratio will revert back to that magic "capital vs. ongoing cost" ratio that apparently holds true across centuries of lighting technologies. OK, then.


Do we always need more and more and more light in the living room? Of course not. So Tsao & Saunders acknowledge that some economies might become "saturated" with light and start to see their rebound levels drop. But the authors suggest we're a long way from this point, even in developed economies.

So let me point you to this 2014 study from Germany, looking at lighting retrofits in 2,000 households. They found a rebound level of 6 percent. That's not a typo -- the calculated rebound in a real-world survey of lighting retrofits was only 6.0 percent.

The astute reader will realize the limitations of this study, of course. It's a short-term study. It's just residential. It's just retrofits. It's just Germany. So I'm not suggesting actual global lighting efficiency rebound is 6 percent.

But this is pretty darn strong evidence that rebound can be very low in regions where incomes are high and the built environment is already built out. 

Time lags

Perhaps these studies are correct, and over the span of decades the rebound effect for lighting is actually very high. In the meantime, however, there would still be major climate benefits captured. Because climate change isn't about a snapshot of 2030 or 2050 emissions levels and sources. It's about cumulative emissions. Think capital cost ratios will revert? In the meantime, we've got at least a decade of lower energy spending while we wait for capital costs to indeed fall. Think the near-term results found in Germany will erode strongly over time? In the meantime, we've got a long period (what's the lifespan of an average residential building?) with significant savings accruing. 

And with the dynamics of climate change, you have to put a big discount rate on those savings. After all, every decade we can forestall catastrophic climate change is another decade for BTI's desired deep R&D to bear fruit and solve everything! So even if you do still believe the Tsao & Saunders analysis and what it implies for long-term rebound, we still capture a lot of significant and meaningful energy savings in the near term by switching to LEDs. 


The most salient point in the Tsao & Saunders paper, at least for me, is where they note that in emerging economies, the rebound is expected to be strongest, because these economies are far from saturated and have the strongest GDP growth profile. The 1.5 billion people without lighting will eventually become lighting consumers, and the cheaper it is, the more they'll obtain as they seek to catch up with standards currently enjoyed in more advanced economies. This is true.

But it can also be directly addressed more easily in emerging economies than in more advanced economies. Current forecasts do suggest that the power for such lighting won't be clean energy -- that's true. But LEDs matched with distributed PV are very powerful. It's a well-matched application, as many have attested.

Of course, this doesn't fit the BTI worldview, so they have to attack this idea as well. But let's address this real issue head-on: I would urge the LED lighting industry to take this as a challenge to make even more of an impact on the world by working together to promote solar lighting for the developing world. It would be a really powerful statement and directly address the biggest potential source of rebound.


Of course, the entire Tsao & Saunders analysis is built on an extrapolation of historical lighting trends into the future. This may appear to be a reasonable approach -- but intelligence and controls are radically changing the lighting industry.

Companies like Digital Lumens are seeing lighting-specific energy savings of 90 percent at installation sites. About half of that savings comes from the physical lighting element switch, which is what the Tsao & Saunders analysis focuses on. But the other half of the savings comes from the use of controls to shine light only where it's needed, exactly when it's needed. LEDs, as solid-state devices, are very controllable, with no warmup time required and very good dimmability. So LEDs plus controls are incredibly powerful. And they will likely lead to a future that is very different from historic lighting usage patterns.

Take another look back at the discussion of capital cost ratios. If the usage profile of a light radically changes because of embedded controls that mean it's turned on a lot less frequently, won't that impact the ratio? If lighting controls allow us to only use light when it's needed, won't that accelerate saturation as we avoid some redundant lighting systems? 

The Tsao & Saunders paper doesn't consider controls at all. But embedded controls are going to dramatically change the way lights operate in the future, in ways that potentially create an inflection point that undermines the basic premise of extrapolation from historical usage patterns.

Final thoughts

So there you have it. BTI put out a needlessly antagonistic op-ed belittling the potential of LEDs. When challenged about it, they say they have data and everyone else is ignoring it. But when you actually look at the analysis, the data is a small set which doesn't even necessarily support their argument.

Plus, there are other reasons why the analysis is potentially incomplete: real-world evidence of saturation effects dramatically opposes their findings; they omit potential shifts in usage from controls and technology leapfrogging; and they omit very real benefits stemming from time lags between any rebounds during which significant savings can accrue.

It's truly unfortunate. The point BTI is ostensibly trying to make is a valid one: rebound effects are important and may not be given due consideration by the long-term scenario planners involved in international climate policy discussions. Rebound is real, and it may be significant in the domain of lighting in particular. It's true. 

But there's very little evidence that it's anywhere near the "total rebound" level implied in the statements made in that New York Times op-ed. "Significant" does not equal 100 percent, and to take these papers and use them as the basis for that claim is misleading. The Tsao & Saunders and Tsao & Waide analyses are interesting and useful, but clearly limited in how they should be applied. BTI is applying them in the public sphere in ways that are entirely inappropriate.

And any argument to be made about the relative value of one climate-mitigation solution versus another one is pretty pointless to me anyway, when we desperately need an "everything, ASAP" strategy right now.

Why some like BTI feel the need to overreach with their own analysis to make broad attacks on LEDs (and by extension, those working on LEDs) in the popular press is beyond me. After all, why did that email come to me saying "Amen"? Because that person, like many others, was indeed misled by such rhetorical certitude. And that's a shame. 

So let me be clear: LED lighting is one powerful tool (among many others) for addressing climate change. And the people working hard to accelerate adoption of LED lighting should be very proud of what they are doing. We can do even more, too. But to say that LEDs are "unlikely" to make a positive impact on climate change is just not true.