Skip to content

Want to Schedule Your Electricity Use to Reduce Pollution? Here’s How

Some of us occasionally feel the urge to turn off the kitchen/porch/office light as our small step towards addressing global climate change. A Berkeley PhD student – Gavin McCormick – has started a nonprofit to provide information on exactly how our actions impact pollution.

Gavin’s organization, WattTime, is analyzing data from around the US to provide information like the map below. You can hover over an area to get almost live information on CO2 emissions.


It turns out that it’s not straightforward to generate the information behind maps like this – far from it.

Consider two different scenarios. In both, I’m assuming that you’re doing something that you don’t usually do – turning off the porch light earlier, for instance – and that other consumers do not change what they do.

To understand what Gavin is doing, we first need a basic understanding of electricity grid operations. Many readers no doubt understand this better than I do, and this is a highly stylized description, but it works for these purposes.

Wherever you live, there’s a grid operator charged with balancing the electricity system and ensuring that there’s enough power generated at any given time to meet demand. The operator also needs to ensure that the plants are not generating too much as that could damage equipment.


The grid operator communicates with power plant operators in the region about how much they’re producing and with the grid operators in adjacent areas about imports and exports. Much of the communication is automated and, for instance, the grid operator’s “request” for less electricity when you turn off your light would likely be communicated through something called Automatic Generation Control (AGC).

Go back to turning off the porch light. If you live in Ohio and it’s a regular night, it’s possible that the AGC system will instruct a coal plant to reduce production. That’s great if you’re trying to reduce GHGs or other pollutants, as coal plants are as dirty as they come.

Turning off your light in West Texas may not be as helpful for reducing greenhouse gas emissions, especially if it’s windy. In this case, you well might be asking the grid operator to back off on the output from wind plants.

In fact, many wind producers receive subsidies from the federal government that increase in the amount of electricity they produce. This is called the production tax credit, and it was recently $23/MWh. So, wind producers actually want you to keep your lights on in the middle of a windy night. In fact, prices can be negative (though rarely below -$23/MWh), meaning that the wind turbine producers are willing to pay consumers to take their power at that point in time.

West Texas Market Clearing Prices for 2010

One thing that’s great about Gavin’s site is that he’s striving to use a better methodology than what’s out there now. There are other sites that simply report the average emissions in a given hour. So, for the Ohioan turning off the light, the other sites are averaging across zero-GHG-emissions nuclear power plants and the coal plants.

This calculation will provide a misleadingly low estimate of the impact of your actions, though, since the nuclear power plants won’t change their output when you turn off your light. They are not marginal, in the language of economists. So, it doesn’t make sense to account for their emissions, as your actions don’t affect them. While average GHG emissions vary by a small amount throughout the day, the marginal plant can vary a lot – from particularly dirty coal to emissions-free wind. The map above still reflects averages, but WattTime maps reflecting marginal emissions are coming very soon.

My example about a single porch light is actually below WattTime’s aim, as a 100W porch light is 1/5,000,000th of the output of a typical 500 MW plant. They’re really targeting the larger decisions by, for example, designing smart plugs for electric cars and industrial load controllers.

If Gavin’s company takes off, which we’re all cheering for, the calculations get a lot more complicated. If his company accounts for a large and predictable share of electricity demand, grid operators might anticipate that fewer people will turn on their porch lights when polluting plants are marginal, and adjust their decisions about which plants to turn on for the day. Ultimately, planners might anticipate this reaction and adjust the type of power plants they build. Gavin and crew are working on incorporating longer run decisions, so stay tuned for Version 2.0 of their site.

Information will help us make better decisions, and I’m delighted that there are innovators like Gavin out there who are perfecting the information that consumers get, and designing cool websites and apps to put the information in front of us.



Catherine Wolfram View All

Catherine Wolfram is Associate Dean for Academic Affairs and the Cora Jane Flood Professor of Business Administration at the Haas School of Business, University of California, Berkeley. ​She is the Program Director of the National Bureau of Economic Research's Environment and Energy Economics Program, Faculty Director of The E2e Project, a research organization focused on energy efficiency and a research affiliate at the Energy Institute at Haas. She is also an affiliated faculty member of in the Agriculture and Resource Economics department and the Energy and Resources Group at Berkeley.

Wolfram has published extensively on the economics of energy markets. Her work has analyzed rural electrification programs in the developing world, energy efficiency programs in the US, the effects of environmental regulation on energy markets and the impact of privatization and restructuring in the US and UK. She is currently implementing several randomized controlled trials to evaluate energy programs in the U.S., Ghana, and Kenya.

She received a PhD in Economics from MIT in 1996 and an AB from Harvard in 1989. Before joining the faculty at UC Berkeley, she was an Assistant Professor of Economics at Harvard.

23 thoughts on “Want to Schedule Your Electricity Use to Reduce Pollution? Here’s How Leave a comment

  1. Sorry to resurrect an dead thread, but over breakfast it occurred to me that if you model a cap not as a static system, but as a multi-period game with administratively set caps, and if you assume that the electorate is willing to accept a program that binds “a little, but not too much,” (that is, the price does not go “too high”), and if you believe that the carbon reduction that achievable by end users using WattTime would be difficult for a profit-seeking entity to aggregate and monetize, then I think you will see that even under a cap, programs such as WattTime can save carbon.

    Basically, WattTime can shift the supply curve for carbon, resulting in lower prices, or, over time, allowing regulators do reduce the cap without pushing the price over some unacceptable level.

  2. Thank you Catherine for this thoughtful post! Love these comments too. I see three strains, hope you don’t mind my offering my two cents:
    (1) Jack Ellis – People “should” and will make decisions using only price. Totally agree with David here, and I just gave my perspective on the new WattTime blog (
    (2) mcubedecon and Tom Kreutz – Power markets are so complex, surely WattTime must be calculating consumer impact wrong. Sounds to me like you’re both implicitly criticizing a very specific modeling technique that, actually, we agree would be wrong! I’ll try to post next week in more detail about how we DO do it. I’d love to follow up with you then to seek out your thoughts.
    (3) Ben Hobbs – In theory, in regions with binding caps nothing consumers can do actually affects total pollution anymore. I agree this complicates things! But should environmentalists just give up and abandon all conservation, renewables, and efficiency? I say no. As Catherine notes, AB32 and RGGI until recently were not even binding. In fact judging from the EUETS, I think we may head back there before long. But moreover, Ben as you note yourself, the cap is actually very much still an ongoing political fight ( and the lobbyists like my own former employer ( fight almost entirely based on abatement costs. The different caps are not at all truly exogenous.
    Thanks for your interest.

  3. Having worked on this problem myself, I’ve come to the conclusion that it’s specious logic to assign marginal emissions to a given electricity consumer. True, if my switching on a light causes a coal plant on the margin to ramp up its output (and vice versa), it’s tempting to assign coal plant emissions to my consumption. However, I believe that the only coherent methodology is to assign the average emissions of all currently operating plants to all users at each point in time. It isn’t possible or sensible to assign individual users with particular generators based on when they started drawing power. (The early bird get’s the nuke, while the last guy gets the marginal started drawing power. (The early bird get’s the nuke, while the last guy gets the marginal combined cycle, etc.) It’s easy to come up with scenarios that reveal the flaws in this kind of thinking.

  4. One further point: for WATTime to be useful, you have to assume that the utilities don’t offer TOU or similar rates and there is not a trend toward utilities offering such rates. The facts are that there’s a strong push towards those types of rates (we’ve worked for EDF on this issue), and the consumers who would use the WATTime system are the same ones who will choose the TVR programs, like what they already have in Connecticut for example. So the program sees to duplicate what’s already available or has a limited shelf life.

  5. I wish Gavin luck but he’s tilting at windmills (no pun intended?). Once poorly conceived carbon reduction policies are layered on top of already (in some cases needlessly) complex electricity “market” operations, it’s nearly impossible to establish a reliable link between individual actions like scheduling electricity consumption and the impact on aggregate carbon emissions.

  6. Emissions caps complicate this analysis.
    If power sector emissions are capped (as SO2 and NOx are, at least nominally, in the east, and CO2 is in California and RGGI), and the cap is binding, marginal emissions are zero. This is because a plant that generates more now because you turn on your porch light must therefore buy allowances from (or sell fewer allowances to) another plant, who therefore must reduce emissions. Total emissions can’t change, by definition of the cap. (Well, one can draw on a bank, etc., but if the cap remains binding over time, the total emissions over that time can’t increase). The timing and location of emissions will change, perhaps to a more harmful pattern or perhaps to a less harmful pattern. (New Yorkers reducing their SO2 emissions might result in a shift upwind to Ohio, where the impacts of emissions are worse.)
    However, by reducing emissions in this hour by turning off that porch light, there will be epsilon more allowances made available, which lowers the allowance price by a smidgen. If lots of us do that, then low allowance prices might make it easier for policy makers to tighten the cap.

    • Your point (and Jack Ellis’) illustrates a problem in academia–too often graduate student efforts to be “relevant” leave out key aspects that can create misleading results. If this calculator gets to the public, they’ll just become more misinformed. There are some great examples of where graduate student work was useful and insightful in the electricity industry, such as Wolfram’s withholding analysis and Ito’s tiered pricing critique. But usually understanding the electricity industry requires much more experience than can be garnered in graduate school.

      • I agree with Ben Hobbs’ point entirely, but disagree with mcubedecon. The public will not be misinformed about CO2, for two reasons:

        – For CO2, in California, and RGGI until recently, the cap has not been binding.
        – Even if caps become binding, demand responses, engendered by tools like Gavin’s, may provide a low-cost means of reducing emissions. Retail electricity prices do not reflect marginal generating costs, so a tool like Gavin’s strikes me as highly relevant and could lower compliance costs.

        • @Catherine: But now the California and RGGI caps ARE, or will be soon, binding. Why set up a system that won’t be telling the truth soon?

          Also, just as importantly, short-run generation market prices are now diverging from the true marginal resources as renewables come on line. This has become an increasing complaint from independent generators and new technology developers. What may look like the “marginal” energy source may in fact not be the true marginal source.

          And this is particularly true in the hydro dominated West. I see the WattTime site uses as an example an “electricity consumers can shift their demand as they like towards more renewable fuels such as hydropower…” Problem: hydropower is a limited resource that we are already exploiting fully and more importantly we use it to effectively store low cost energy to use in later periods. The CAISO market may show Feather River hydro as the marginal energy source on a hot August afternoon as that its the primary ancillary services provider in the PG&E area. How is this site going to educate consumers about the complexities of hydro-thermal integration? I’m not seeing that this site catches the true nuances of a very complex machine.

          So I stand by my comment that graduate students too often are not well versed enough in the lore of an industry to capture all of the necessary elements to produce effective tools. That’s why experience is so important, and too often overlooked by policy makers.

          • I see your point (that WattTime would not be telling the truth if a cap is binding) if the message they communicate is that consumers can reduce pollution, as I’ve done in the title. But, if the message they send is that consumers can help achieve our carbon goals, that is the truth, and a useful one. This is my second bullet point.

          • I”m missing the distinction–the caps are on carbon emissions, not NOx (unlike RECLAIM and NE OTC caps), so “pollution” = “carbon” here. I think what consumers can do is affect the costs of GHG reduction compliance, so the TVR (time variance rate to encompass all flavors of time-based rates) is a preferred signal. However the price signal is broken in the CAISO market for a variety of issues, most importantly that utilities are able to both bid into the markets and to recover capital costs in rate base, so they dump their excess power into the market, often below apparent cost.

      • To the extent consumers are faced with time-of-use electricity consumption decisions, they ought to be based on a single attribute, and that attribute should be price. As my wife astutely pointed out many years ago, consumers are not going to be sitting in front of their computer screens looking at electricity prices, and the same holds true for hourly emissions. Instead, to the extent consumers make usage decisions based on price, it will be with the aid of intelligent agents. In order for those agents to make rational decisions, they’ll need current and forward prices. Adding emissions to the decision-making logic is unlikely to be workable.

      • “To the extent consumers are faced with time-of-use electricity consumption decisions, they ought to be based on a single attribute, and that attribute should be price. ”

        Fine, except Gavin’s website does not appear to be about what is best, but what is possible. Retail rates block real-time prices and utilities and PUCs are showing no enthusiasm for changing that. If an end user can costlessly redispatch their load to a time of lower carbon emissions, why should they not be able to? We all get that under a binding cap, this won’t reduce net emissions, but it frees up emissions for more economic activity under the cap. That’s still a good thing.

        As I said in another comment which seemed to get swallowed, a lot of folks in the power industry may have the horse and cart backwards. The *political* reality is that real-time prices are not a precondition for price-responsive load. More likely real-time prices will follow when there is already non-trivial existing amount of automated responsive load. And since money is obviously not going to be the driver in that, it will have to be something else that enough end users care about.

        • I disagree about the trend on TOU/RTP prices. CA, NY, CT and NJ are examples of states that already offer TOU pricing and are moving further along (and the CA Statewide Pilot showed that TOU prices induce responses that are a reasonable approximation of RTP prices). IL is considering a program, and I’m sure I’m not aware of the other states considering such steps. Plus there are other places that offer voluntary TOU pricing. The people using WATTime will be the same as those who will enroll in voluntary programs.

      • The California Statewide Pilot is not a useful guide. It’s a 10 year old 18 month study that could not possibly take into account widespread automatic response to real time price signals because such automation did not exist. Moreover, it still does not exist at the residential and small commercial level.

        Also, the notion that TOU approximates RTP might have been valid then, might even be valid today (probably not), but almost certainly won’t be valid ten years hence unless utilities continually update their TOU multipliers /and/ periods. Should be interesting to hearing from the ratepayers with rooftop PV when updated TOU rates provide for negative prices for their noontime output.

        So far TOU is mostly (entirely?) opt-in, and the reality is that it is harder to opt in to TOU than it is to press go on a new dishwasher that happens to have an “eco mode” button that does something you do not understand — do not need to understand — to reduce carbon intensity.

        • The recent SMUD study shows that automated responses can work with TOU prices quite effectively. The incremental savings from RTP are miniscule. Expected negative midday prices (which may be quite deceptive due to CAISO market failures) can be put into TOU rate schedules as well. Putting consumers in the position of having to respond to minute to minute changes rather than being able to anticipate load patterns that are fairly representative to long term system conditions can be counterproductive.

          As for signing up for TOU rates, that’s a pretty simple task in California with smart meters everywhere. Making one call is much easier than having to choose each day to program your dishwasher.

          Finally, buying new “smart” appliances is the wealthy person’s response–most households can’t justify new appliances on the savings they might gain. Plus almost half of California households are renters–they can’t even choose to change out their appliances.

      • I find this a very interesting topic, so please forgive me for responding at length.


        My point was not about what can be put into TOU rates, but what will be put into TOU rates. Moreover, my point was not about forcing someone to adjust to minute by minute emissions, but enabling them to do so. Changing your rate is harder than you think. You need to know what an electric rate is, that alternatives exist, why you might want a different one, how they work, and what risk you might be taking on in changing. And then you need to call, where it is not unlikely that the person on the other end of the line will subtly discourage you from changing. This is indeed harder than just making an in-store purchasing decision to get a machine that will do “something green” on your behalf.

        I also see nobody here advocating for getting all new appliances. But appliances need to be replaced from time to time and if someone could put together a non-profit infrastructure to make new ones a bit smarter, what’s the problem? (by the way, for renters, principal/agent issues in appliance selection can also apply to tariff selection when the tenant does not get their own bill.)


        I don’t see this as a debate about something vs. TOU or RTP at all. Everyone who reads this blog is in favor of some kind of time-based pricing. And if TOU is available and TOU captures most of the potential benefit of RTP, that’s great, though I suspect that if TOU really is as good as RTP it is because the CAISO market generates very muted price variation (for a lot of reasons, not all of them good). And let’s not forget the vast majority of load around the world where TOU is but a gleam in a utility commissioner’s eye.


        I agree with mcubedecon’s point about the difficulty of determining what is marginal. In fact, in the RTM in particular, it takes Holmesian forensics to figure out why prices gyrate certain ways. Things like redispatch of resources a few intervals in advance of an expected ramp that then maybe doesn’t even occur can generate strange prices indeed. But I hardly see that as a reason not to try.

        In fact, that prices might not be great proxies for carbon is, if anything, an argument against Jack Ellis’s notion that price should be the only variable of consideration. Perhaps it is hard for some to believe that there are consumers that actually care more about carbon than price per se?

        Methodologies that integrate CIMIS and other data can work out carbon
        intensity and correlate it with system conditions on a statistical basis. That can then be used to inform inform real-time models of carbon intensity, so I don’t think this is an intractable problem at all.


        As a practical matter, any device that is going to respond to a TOU schedule or price signal is going to need that schedule or signal. One way is to build a system where a user must enter and update this information manually. AFAICT, in California, the state of the art for this process is to navigate to your utility’s tariff page, find your tariff, open a pdf and copy down the numbers you see there.

        The alternative is to provide a service with a machine interface that allows devices to pull down this info themselves, and this sounds like what WattTime is trying to build. (Smart meters can do this, too, but they mostly don’t. PG&E and SDG&E do not push rate info over their Zigbee smart meter HANs, and SCE only pushes an estimated rate and
        does not support TOU!) I’m sure they’ll fix this one day.

        I do agree that such a service needs to not only provide a point estimate of the current marginal carbon (or price) but also a reasonable short term forecast of the same.


        Regarding graduate students, experience, and “industry lore,” well, I think policy makers actually respect experience quite a bit, as they should, but there is plenty of need for graduate students and their untested ideas. Experienced energy professionals have their own limitations that graduate students generally do not share. Experience often comes with rigidity, reticence, and difficulty seeing a system much different than the present. What’s more, energy folks tend to get a bit conditioned to institutional decision-making, unwilling to appreciate thinking that does not emerge from formal processes.

        Taking power system data and presenting it to users who may decide to act on some particular aspect of it makes an end run around utilities, regulators, and thousands of person-years worth of meetings, projects, battles, public comments, false starts, restarts, etc. It’s actually pretty cool to try /even/ if it doesn’t work. Lots of regulatory processes create systems that don’t work, too. 🙂

        • @David: I disagree with the barriers to changing your rates. I’ve done it in the past–it was extremely easy. The fact is the customers who are going to sign up for WATTime are the same sophisticated customers who are going to find changing to TOU rates easy. The WATTime audience is going to be fairly well off–look at the statistics of who has regular Internet access.

          Which brings us the bigger question of what really is going to have broad based effects on reducing GHG emissions. Small “act locally, think globally” actions that are not replicable across the entire population and even the entire globe are truly a waste of our time. Working hard to save a single ton of CO2 emissions is simply not worth the effort if that effort could be leveraged to create much larger savings down the road. Borenstein has written here about how California may have lost its vision of creating a framework that can be extended across the U.S. because it focuses too hard on capturing every ton of reduction. WATTime strikes me as a tool focused on a small segment of the consumer base that actually steers away pressure on the utilities to fix their rates system so that the entire customer base sees the correct price signals. WATTime becomes a “feel good” action that doesn’t have the long term consequences that really are necessary to get real change.

          As for appliance interfaces, manufacturers need a standard communication protocol. Trying to end run the utilities is a dead end that could even be counterproductive.

          Any graduate student can set up an alternative system. Probably I should complain that experienced academics shouldn’t be endorsing the system unless they’ve fully vetted it and it delivers information that accurately reflects the outcomes we see.

%d bloggers like this: