The Efficiency Paradox

In a previous post, I pointed out the highly counterintuitive fact that increasing the energy efficiency of any technology leads to wider use of that technology and eventually greater overall usage of energy by that technology. As a result of this effect, attempts to foster energy conservation by using more energy-efficient technologies backfire.

I think I have figured out the economics of why this happens.

When we think about conservation-through-efficiency we usually think of it in terms of the energy consumption of the old technology versus the energy consumption of the new technology. Obviously, if the new technology uses less energy then the old then substituting the new for the old saves energy. Saving energy almost always means saving money. Yet that common sense view doesn’t capture the true economics of the situation.

The true cost of anything has nothing to do with money. Instead, the true cost of any action is the opportunity cost. Increasing the energy efficiency of any technology lowers the opportunity cost and raises the marginal utility of the technology. Increasing the marginal utility leads to a wider use of the technology, which eventually swamps the initial energy savings.

Take lightbulbs (remember, though, that this effect occurs with all technology). Marginal utility does not concern itself with the opportunity cost of the first lightbulb or fixture that someone installs somewhere, but rather the last. It’s the cost of installing one more fixture that matters. Say a site manager considers whether to add additional lighting for security and accident safety. He calculates each new fixture costs $10 per year to operate but that his savings from reduced insurance will only be $8 a year. Economically, he won’t decide to install the new lights. Adding the new lights won’t pay off in improved safety. He does not base this decision on his current cost of lighting but rather on the benefit that each new light brings. Then an energy efficiency improvement drops the cost of operating the lights to $5 per year. Now the site manager can justify installing more lights even though overall his total cost for lighting may well rise. Each new light saves him 3 bucks a year in insurance payments. The savings in insurance, not the savings in his energy bill, drives the new installation.

The efficiency paradox occurs because increased energy efficiency makes it less costly to use a technology for newer or wider purposes. Worse, the greater the improvements in relative efficiency the more rapidly the technology spreads and the more energy it consumes. Only if all other factors remained constant and the new technology merely replaced the old one one-for-one would increased energy efficiency actually save energy. Technological history suggests that that never happens.

If improved energy efficiency actually saved energy then the evolution of our technology would look vastly different. Today, we could make a vehicle with all the performance characteristics of a Model-T that got 100+ miles to the gallon. Instead, we used each new leap in energy efficiency in the internal combustion engine to make the car perform new tasks. We use cars to carry more, more quickly, farther, with greater safety and less pollution. As a result, per capita consumption of energy for internal-combustion-engine based transportation rose steadily over the technology’s entire history. Today, we could make a computer comparable to a desktop circa 1980 that ran off mere body heat. Instead, we build more and more power-hungry systems and cram them into every conceivable niche. Increased energy efficiency reduces the tradeoff cost of using a technology in ways or scales not done before.

Any attempt to conserve energy through efficiency will fail in the end unless we somehow forcibly restrict the use of the new technology only to direct substitution of the older technology. Doing so, however, will have far worse consequences in the long run.

17 thoughts on “The Efficiency Paradox”

  1. Yes. Another term for efficiency increase is productivity increase. The people arguing with you are framing the issue in terms of cost rather than productivity. If cost were the main issue we could satisfy our lighting needs with a couple of LED flashlights per household, which would meet or exceed the 19th-Century standard.

    The concept of conservation as a moral and economic good does not get nearly the skeptical attention it deserves. Wealth derives from human creativity, not from natural resources or other objects.

  2. Take lightbulbs (remember, though, that this effect occurs with all technology). Marginal utility does not concern itself with the opportunity cost of the first lightbulb or fixture that someone installs somewhere, but rather the last. It’s the cost of installing one more fixture that matters. Say a site manager considers whether to add additional lighting for security and accident safety. He calculates each new fixture costs $10 per year to operate but that his savings from reduced insurance will only be $8 a year. Economically, he won’t decide to install the new lights. Adding the new lights won’t pay off in improved safety. He does not base this decision on his current cost of lighting but rather on the benefit that each new light brings. Then an energy efficiency improvement drops the cost of operating the lights to $5 per year. Now the site manager can justify installing more lights even though overall his total cost for lighting may well rise. Each new light saves him 3 bucks a year in insurance payments. The savings in insurance, not the savings in his energy bill, drives the new installation.

    Now let’s take our very same plant manager. He’s realized the insurance savings and has adequate, perfectly safe lighting. Additional lighting would not futher enhance safety and no further insurance savings are available to him. Now a technology comes forth that will cut the energy cost of his lighting from $5/yr to $2/yr.

    He will switch to the new technology for the energy cost saving but will not add more lighting since there is no benefit to be gained from additional lighting.

    If he is operating thousands of lights and, therefore, saving thousands of dollars/yr in energy cost he may, in fact, make an economic decision to purchase some machinery that consumes electricity but was previously economically unjustified (or unavailable) but now affordable due to the reduced energy costs of his lighting.

    If that is the case then he will consume some or all of the “saved” energy. If it is not the case he will use the operational cost savings as profit and/or reduce the price of his product to improve his competitive position.

    Let’s have another look at our manager. His lights cost him $10/yr to operate. His insurance company tells him if he had more, or better, lighting he could save $X on insurance because it would enhance workplace safety. He calculates the cost of the additional lighting, it is > $X, so he doesn’t add or improve his lighting. Clever insurance salesman or jeopardized worker drops a dime to OSHA who drops in and explains to Mr. Manager that his site does not meet minimum lighting requirements for workplace safety and if he does not rectify the deficiency he will be subject to fines or closure.

    He bears the cost of improving the lighting, attains safety compliance, and calls his insurance company to start receiving his reduced rate. Now the lightbulb salesman walks in and explains to him how his $10/yr lighting cost can be reduced to $5/yr. Mr. Manager puts in the new light technology and reduces his operating cost $5/yr. He doesn’t install additional lighting to bring his lighting cost back up to $10/yr because he has all the lighting he needs.

  3. Shannon,

    Apologies for the previous “anonymous” comment. Your overall macro level economic argument is solid. Increased efficiencies free up resources to be used for other purposes which will, ultimately, consume the resources freed up by the increased efficiencies.

    Lighting, however, is a particularly inapt example of this. It can work the way you claim but it need not. Joe Barowner, for example, may find that he can employ more efficient technology to reduce his costs for lighting sufficiently to cover the cost of the larger, glitzer outdoor lighting he’s been wanting to put in because he believes it will draw additional business.

    If, on the other hand, Joe Barowner is satisfied with his lighting he will still employ the new technology purely to save cost.

    Let’s have a look at your local Little League. The league has grown such that the number of games that need to be played exceeds the field-hours available. You must either add fields or make the existing fields available for more hours. Economic analysis indicates that lighting the fields would add sufficient field-hours at less expense than adding additional fields. The phone call is place to some local lighting contractors for bids. Each and every one of the contractors, presuming they are professional, will put forth a proposal that lights the fields to at least the minimum standards necessary. Fail to meet the standards and there will be no insurance, at any cost, and therefore no additional field-hours. Meet the standards and lighting is sufficient. Installing the most efficient technology will have little or no bearing upon the “amount” of light installed. If sufficient efficiency gains are available it is possible that they will free up enough resources (money) to light another field and further increase field-hours available. But if there is no need for additional lighted fields there will be no need to increase lighting and efficiency gains will reduce operating cost without increasing total consumption of electricity.

    Lighting is installed to the level of “sufficiency”. Sufficiency is defined either by regulatory standards or by personal preference. Once sufficiency is attained there in no need for additional lighting unless regulations or personal preference changes. Technology which improves efficiency is unlikely to substantially alter either regulatory standards or perjsonal preference. Commercial/industrial lighting, or public safety lighting, have long been governed by regulatory standards. Household lighting has long been inexpensive enough that homeowners generally satisfy their personal preferences.

    BTW, I think there’s a hole in your efficient use argument. Efficiency gains in consumption drive usage up indirectly. Efficiency gains in production, on the other hand, can drive usage more directly. If we can produce electricity more efficiently (sell it cheaper) we can drive additional consumption more directly than if we only consume electricity more efficiently.

    If electrical power is unavailable or very expensive we wouldn’t have all the electrical appliances and gadgets we are so dependent upon or fond of. The increasing electrical efficiency of those appliances and gadgets indirectly helps keep the cost of electricity low but out use of the appliances and gadgets is rarely driven by the cost of electricity to run them. Electricity is cheap therefore we think nothing of buying more stuff that uses it.

    Do you suppose a single TV salesman has ever had this happen… potential customer walks in and looks at TVs. After determining which few of the many models are most attractive to him, Customer asks Salesman how much electricity the TVs use.

    “Oh, Model A consumes twice as much per hour as Model B,” says Salesman.

    “Oh, well,” says Customer, “I prefer Model A but give me two of the Model B instead. I’m purchasing to meet my electricity budget.”

  4. Shannon,

    Sorry to monopolize and go on bloviating, butt…

    Marginal utility does not concern itself with the opportunity cost of the first lightbulb or fixture that someone installs somewhere, but rather the last.

    This is not always true. In the Network Equipment Provider world, for example, when dealing with “emerging markets” it is the initial “opportunity cost” (the cost for Subscriber #1) that frequently drives business rather than the “marginal” cost of “Subscriber #1,000,000”.

    If a particular NEP’s product has excellent marginal costs for Subscriber #1,000,000 but high opportunity cost for Subscriber #1 that NEP is unlikely to win the business. The NEP with low opportunity cost for Subscriber #1 will win despite higher marginal cost for Subscriber #1,000,000. This is due to limited capital availability for startup. If you can’t afford Subscriber #1 then it makes no difference how low the marginal cost of Subscriber #1,000,000 is.

    Let’s say Tommy Tinkerer knows how to build a better Widget and Manny Embeaye knows the widget business. Manny tells Tommy that they can sell these New & Improved Widgets – in 5 years they’ll be selling 100,000 widgets a year fersure fersure. But they’ll only sell 2,000 the first year, 10,000 the second year, and so until they’re finally profitable in the 5th year when they’re selling widgets hand over fist.

    Widget making machines have been dropping in price. 5 years ago they cost millions each, today only a million, and in 5 years they’ll cost half a million. But widget making machines can produce only 10,000 widgets a year. In five years Tommy and Manny will need ten of them. Today they need only one. They can’t wait for the widget making machines to cost only half as much five years from now. They need to find $1M to buy machine #1 or they’ll never need machine #’s 2 through 10.

  5. Knucklehead,

    Increased efficiencies free up resources to be used for other purposes which will, ultimately, consume the resources freed up by the increased efficiencies.

    Actually, my argument is that increasing the efficiency of a particular technology causes increased overall consumption by that very same technology. Energy uses doesn’t shift from one task to another. Instead we employ the same technology for more task.

    Lighting, however, is a particularly inapt example of this

    It maybe far less obvious because the easily observable changes occur over longer periods of time but the same effect applies.

    Once sufficiency is attained there in no need for additional lighting unless regulations or personal preference changes.

    I don’t think that sufficiency is every reached. As the efficiency increases, once extravagant uses become ordinary and eventually trivial. What constituted safe and standard lighting in 1950 would get you sued for negligence today. Once a “well lit street” had one light on the street corner illuminating the intersection. Then they had lights down the sidewalk interspersed with strips of darkness and no significant light on the street. Now we expect a “well lit street” to be bathed in light from end to end and from sidewalk to sidewalk. Ditto for parking lots, parks and other places in which lighting was an extravagance just a few decades ago.

    There are other changes which also consume energy. We burn a lot of electricity in lighting now in order to produce a broader spectrum of light than in the past. We may pump out the same lumens but we improve the quality of the light itself. That takes power.

    Do you suppose a single TV salesman has ever had this happen…

    It did back in the day when televisions were new an their relative power consumption was higher. Major appliances like air conditioners, washer/dryers and water heaters certainly do compete on energy use.

    Basic lighting these days is truly “to cheap to meter.” I have no no idea how much money I spend for lighting but I do know how much I spend for climate control. I do know that I use more energy for lighting both in lumens and in quality than my family used when I was a child. I also use more lights in more places such as outdoors and in my garage.

  6. Shannon,

    I’m way overboard with this light bulb thing butt…

    I don’t think that sufficiency is every reached. As the efficiency increases, once extravagant uses become ordinary and eventually trivial.

    This just doesn’t hold for lighting. In industrial uses too much lighting can be as unsafe as too little. In the ball field example too much lighting it too much lighting – as bad a safety hazzard as too little.

    For personal preference among householders people prone to extravagant lighting will have extravagant lighting regardless of the cost of operation of a particular bulb. If you like spotlights all over your yard you’ll have them. If your wife wants a dozen bulb chandelier over her dining room table she’s not gonna make the decision to get it just because LED lights are now available in candelabra format and use only 10% as much electricity as incandescent lights. And she certainly isn’t going to hang one over your bed just because it’s cheap to operate.

    If she doesn’t like candelabra chandeliers she’s not gonna have one just because it’s cheap. If you don’t like spotlights in your bushes you won’t have them just because they’re cheap.

    Enough light is the right light. We’re not going to all turn our homes into Las Vegas wedding chapels because operating costs of lights drops radically. We’re not all going to turn our homes into year round versions of the Griswald’s Christmas just because light bulbs get dirt cheap to run. Not now, not a hundred years from now.

    An extravagance is a necessity after you’ve had it for 90 days does not apply to lighting ;)

    I’ll shut up about it now, I promise.

  7. Knucklehead,

    Don’t apologize every point you make is well founded and well argued…

    …but…

    …history simply doesn’t support your conclusions. Looking back over the last 100 years, we have progressively used more and more lighting as lighting as become more and more efficient at turning electricity in photons. We generate more CO2 (our real concern at this point) per capita for lighting than previous generations even though we have much greater efficiency. I have had the opportunity to see many older buildings and houses with the original fittings from before WWII and they did use half the number of bulbs and fixtures we do. Moreover the actual output of the lighting was far less both in lumens and in breadth of spectrum.

    I think our real point of divergence lays in our respective time scales. In the case of lighting you seem to think in terms of energy savings in the immediate short-term of months or years. I think in terms of decades. New technologies diffuse into the system relatively slowly and not in sudden lumps. By the time it becomes widespread a new efficient technology will have altered people expectations enough that it loses all its energy savings.

    My real point has little to do with lighting and more to do with the entire idea of conservation via efficiency. This idea exerts tremendous political appeal and we have based significant public policy on it since the 70’s and it simply does not work. Technology and energy use evolves in the exact opposite way. I fear that if we do have real energy problems we don’t have an accurate popularly accepted model of how our energy use really evolves. Any public policy we make based on a flawed model will likely have more negatives than positives.

  8. Maybe I’m just being dense, but I’m not so sure that knuckle doesn’t have something in his sufficieny arguement. I agree with you that modern homes use much more lighting than older homes. You’re right that modern cars are much more powerful than their ancestors. IN both cases we’ve sacrificed potential efficiency gains by converting them into performance. But, in both cases, I think we are reaching a point of diminishing returns. I simply do not need any more light in my house. Indeed my house is equipped with many lights that I rarely ever turn on. Mid-century homes were dimly lit. My house is, when I need, bright as day. I think the same is true with cars, although I think the performance curve may have a little further to rise before every idiot has more horsepower than he can handle. But we’re not far off.

  9. Shannon,

    If your argument is that people will use more of something when it costs less than when it costs more I am completely on board. Unfortunately your agument seems to be that people will spend exactly as much on lighting every month no matter how much lighting the money provides.

    The difference is price elasticity. My grocery spending is very inelastic. I buy pretty much the same stuff every week regardless of the cost. Aspargus costs twice as much off season as not but I like it so I buy it regardless. My demand for Asparagus is inelastic. My weekly grocery spending varies not by how many pounds of meat and veggies I eat but by how much they cost that week. When Hoof-and-mouth disease strikes Europe my baby back ribs get more expensive.

    As you have been pointing out repeatedly the desire for lighting is much more elastic. Unlike my desire for asparagus people currently use less than they like because they are stopped by the price. Where you jump off the tracks is by asserting that everyone spends exactly as much on lighting as they choose to do at this very moment and if lighting gets cheaper people will use exactly as much more as they need to in order to maintain spending. This is silly.

    My consumption of groceries is capped by my desire. When asparagus is local and cheap I don’t eat twice as much. I do appreciate the efficiency because I can spend that money on something else I desire be it consumption or savings. The fact that I spent $4 on a pound of asparagus last week doesn’t compel me to spend $4 for two pounds of asparagus this week.

    Be happy that people will benefit from cheaper lighting by getting more of what they want – be it more lighting or more air conditioning or maybe baseball tickets. The money they save has to go _somewhere_ of course. For the love of God please stop saying that if light was free we would turn the Earth into the Sun.

  10. Per capita energy use in Japan is far below that of the U.S. and believe me, there are plenty of lights in the night sky.

    I’m guessing it’s mostly because fuel efficient mass transit is widely available and, secondly, because central heating is rare in homes. Instead, more space heaters are used. If you’ve lived without central heating, you will surely join me in damning space heaters: smelly and definitely inefficient, if what you’re looking for is comfort. But, overall energy-use wise, space heaters are far more efficient.

    Another twist is that Japanese homes have far less insulation than your typical suburban McMansion.

  11. In all the arguements over lighting, everyone is forgetting another marginal impact. When lighting is cheap enough, it becomes more common to simply leave lights on all the time rather than shut them off when not in use. In my office they will light pretty much the entire floor if one person is working. I’ve actually taken to installing CFs in fixtures that the kids are prone to leave on just so I don’t care as much. We had a rash of burglaries in the neighborhood a few years ago – the police suggestion was to leave porch lights on at night. In my case and others, that meant putting a cheap CF in the porch light and leaving it on 24 hours a day. Eventually, this will saturate, but the key learning is that it always takes longer than you would think.

  12. Be happy that people will benefit from cheaper lighting…

    I am happy. I have just finished replacing all the bulbs in my own home with CFLs. Increasingly effecient technology is what makes us as comfortable as we are and offers hope for a better tomorrow.

    What worries me though is that as a matter of public policy we believe that increasing efficiency ultimately leads to overall reductions in energy use. It does not. The opposite occurs. We use more energy and we use more energy total with the improved technology itself. Basing policy on a flawed model will cause major problems.

    In the late 70’s, California stopped authorizing any new power plants based on the idea that increasing efficiency would keep electricity demand flat or decreasing. Whoops. Today, we avoid hard choices in our long term energy policy because we fall for myths like conservation-via-effeciency. If we do face a situation wherein we must curtail our CO2 output considerably, using technologies like CFLs will cause us to produce more, not less CO2.

  13. Your reply to my comment is a perfect example of where we part ways. In two paragraphs you first say something true and then assert something unrelated.

    Agree:
    What worries me though is that as a matter of public policy we believe that increasing efficiency ultimately leads to overall reductions in energy use. It does not.
    and
    In the late 70’s, California stopped authorizing any new power plants based on the idea that increasing efficiency would keep electricity demand flat or decreasing. Whoops. Today, we avoid hard choices in our long term energy policy because we fall for myths like conservation-via-effeciency.

    Very true. The resources conserved will end up being applied somewhere else. I don’t think anyone with an economic bone in their body would disagree.

    Huh?
    The opposite occurs. We use more energy and we use more energy total with the improved technology itself. Basing policy on a flawed model will cause major problems.
    and
    If we do face a situation wherein we must curtail our CO2 output considerably, using technologies like CFLs will cause us to produce more, not less CO2.

    Doubling the efficiency of lighting will allow people to use more lighting but it doesn’t cause people to use twice as much. You keep saying that people will always spend as much or more on electricty dedicated to lighting no matter how cheap lighting gets. The demand for lighting isn’t infinitely elastic or even very elastic. People desire some darkness after all.

    Be happy that you are half right. People have unlimited desires so conservation in one area allows them to fulfill desires in another area (yes some part of that may be more outdoor lighting). Electricity is fungible so you are right and the greens are wrong – squeeze the amount of electricity used on the lighting end and the resources oozes out the other end in the form on air conditioning or aluminum or whatever.

  14. Jack Diederich,

    Doubling the efficiency of lighting will allow people to use more lighting but it doesn’t cause people to use twice as much.

    I know that it sound illogical but if you look at the long term history of electrical lighting that is exactly what happens. We can construct all the well-reasoned arguments we want but the empirical evidence says that as the energy efficiency of lighting increases so does our total consumption of electricity for lighting itself.

    I think its important to remember that not all lighting is equal. In the last 30 years or so we have moved away from a concern of producing mere lumens and more towards producing lumens over a broad spectrum which requires more electricity than a narrow spectrum. Even if we don’t use more electricity today for basic illumination than we did 30 years ago we do use more to produce higher quality light.

    I didn’t call it a paradox for nothing.

  15. It depends on where you sit on the usage curve. I think that the curve that describes lighting use is pretty much described by the equation y = 1-1/10^x. That curve ramps up rapidly until it reaches an inflection point, after which, the slope decreases markedly. That curve has a maximum value of 1, which y appoaches asymptotically. In Tatyana’s model, that asymptote is the amount of lighting that the human eye feels comfortable with. How close are we to that level in every place that a human might want light? We’re obviously not at the asymptote, so cheaper lighting will spur more lights – but how many?

    In my opinion, we are probably somewhat just past the inflection point, so that use will increase, but at a much slower pace than in the past. I don’t think that I will double the number of lights in my house, but with the advent of those cheaper LED lights, I’ve put in night lights in the hallways where there was no use before, but at minimal electricity use. I’ve also replaced all of the outside and garage lights with the fluorescent ones, and some of the inside lights (my wife likes the softer incandescent ones for indoor use). But I’m probably not the marginal case.

    The other question that impact energy use, as opposed to lighting use, is the efficiency ratio. Most of the fluorescents I’ve seen run at 25 – 33% of the energy use of a corresponding incandescent. I doubt that the number of lights in place will double as a result of efficiency, because I think that we are past the inflection point and closer to the asymptote than to the origin of that curve. However, the number of lights will go up. Let’s say it goes up 20 – 30%. The other question is: will people opt for higher wattage lights if the cost is lower? Let’s say they bump up that wattage (like I did in my garage) so that the savings per existing socket is about 50%, rather than 25%. That means that the Green predictions that lighting energy use will drop by 66 – 75% will be wrong, and we’ll see a drop of perhaps 20% in electricity consumption by lighting.

  16. John,

    It depends on where you sit on the usage curve.

    I am highly sympathetic to the saturation argument but I think we have an abysmal track record of projecting real-world saturation. What seems like an unthinkably extravagant use today, will seem like a required use in the future. Look at cars. Imagine describing an SUV to an automotive engineer in 1925. He would think you were insane. A car with air conditioning? That’s insane. Just roll down a window.

    I did manage to find one source from the Energy Information Administration that shows that per capita consumption of energy for lighting increased from 6.59 kBtu in 2001 to 6.83 kBtu in 2004. The report says:

    Energy consumption for space heating, water heating, and refrigeration per household decreases over time, while energy use for lighting and all other applications grows, despite continuing increases in their energy efficiency (Figure 36).

    It looks like lighting maybe even more sensitive to the effeciency paradox than I suspected.

Comments are closed.