Averages, CFLs and Mercury

I’ve been having a useful exchange back and forth with @NobleIdeas about the lightbulb law that is set to take effect in January 2012. (Here’s my first post and @NobleIdeas’ reply. I have another note on this in the works.)

One of the recurring themes of this discussion is the mercury content of compact fluorescent lightbulbs (CFLs). The standard argument is that widespread use of CFLs, despite their mercury content, will lower the ambient mercury level. The reason is that that most environmental mercury in the US is a result of burning coal in power plants, and the amount of mercury from electricity generation that is saved by switching to a CFL is higher than the potential amount of mercury released in the environment (note that I say “potential” because CFL recycling programs will avoid mercury release). This argument is spelled out in @NobleIdeas’ post, and further elaborated in this well-written EPA FAQ.

The purpose of this post is not to argue against this conjecture: I totally agree that, in total, switching from incandescent bulbs to CFLs will result in less environmentally released mercury. However, as is often the case, averages can hide important details, and I’d like to look a little deeper at CFL mercury.

Diversity of Electricity Sources

Electricity sources vary across the United States by surprising amounts. For example, in the EPA’s eGrid database you can see the diversity in electricity sources at a state level, as well as the resulting variation in emissions (the Summary Tables are a great place to start). Although this captures much of the variation, there is disparity from town to town within states as well, or even down to a household as individuals pursue their own clean energy agendas.

As you can imagine, with such a wide disparity in energy sources, “average results” that are true across the US in aggregate may not be true locally. For example, in this post I showed why electric vehicles may actually result in more GHG emissions than gas-powered vehicles in some areas of the country.

State Electrical Sources and Mercury Emissions

Since airborne mercury emissions are a result of burning coal, mercury reduction from switching to CFLs would be lowest in a states that burn the least coal. Among those state is CA, where eGRID2006 reports that only 1.2% of its electricity coming from coal, resulting in .0014 pounds of Mercury per GWh of electricity. This is roughly 1/20th of the national average of .0269 lbs/GWh. [Note: I used eGRID2006 for this analysis since the more recent reports have, for some reason, not included mercury emissions. However, the electricity source mix for the states mentioned below seems consistent to the latest version].

Doing unit conversions and plugging these values into the computations on the EPA FAQ, we see that mercury emissions from electricity has dropped dramatically for both types of bulbs, and the net mercury released by a CFL is now almost twice that of an incandescent bulb since the mercury released at disposal starts to dominate (.50mg v .28mg of mercury).

What does that mean? It means that as California residents switch to CFLs, if they don’t recycle the bulbs they will actually cause an increase in mercury in the environment. This is also true for 5 other states, including Arkansas, Idaho, Maine, New Hampshire and Rhode Island, all of which also have very low uses of coal as a source of electricity. In Montana and Oregon the switch to CFLs will have negligible difference on mercury output.

Does this finding change the statement that switching to CFLs will reduce mercury output for the nation overall? No, that continues to be true. However, if we only cared about mercury in the environment, then the right thing to do would be to switch to CFLs in the states not listed above, and leave the states above on incandescent bulbs.

(Of course, this last statement is also based on averages, and the true optimization would be to assess the mercury at each household and make a determination of when to switch.)


Averaging data across the country lets us make important statements about the impact of decisions and policies. However, averages can mask wide regional, state or local variation in the actual data, and statements made about the aggregate national behavior may not be true when applied at more local levels.

One important area is in electricity, where there are major differences in electricity sources across the country. As we’ve seen with electric cars and mercury emissions, these differences can mean that statements that we take as fact at a national level may not be true at a local level.