Efficiency: What It Means in Physics, Biology, Economics and Computing

Efficiency is one of those words that everybody uses and that turns out to mean several precise, different things depending on who is speaking. An engineer means a ratio of energy out to energy in. An ecologist means the share of energy that passes from plants to the animals that eat them. An economist might mean an allocation of resources in which nobody can be made better off without making someone else worse off. A computer scientist means how the running time of an algorithm grows with the size of its input. A statistician means how much information an estimate squeezes out of the data.

What these meanings have in common is captured in the general definition used by the Wikipedia article on the subject: efficiency is the measurable ability to avoid wasting materials, energy, effort, money and time while performing a task. In a more general sense, it is the ability to do things well and without waste. In its most quantitative form, it is a ratio of useful output to total input.

This guide works through that core idea and then follows it into each of the major fields that have built their own version of it. Along the way it explains the hard limits on efficiency set by physics, gives worked examples with real figures from government and scientific sources, traces the history of the "efficiency movement" that made the word a political slogan in the early twentieth century, and looks at the surprising ways in which greater efficiency can lead to greater total consumption.

The core definition

Efficiency as a ratio

The Wikipedia article expresses efficiency mathematically as

r = P / C

where P is the amount of useful output ("product") and C is the amount of resources consumed. When P and C are measured in the same units, as with energy in and energy out, r is a pure number, often given as a percentage. When they are measured in different units, the ratio is still useful but is usually given a different name. Fuel economy in miles per gallon, or the light output of a lamp in lumens per watt, are ratios of this second kind, often called "efficacy" or "productivity" rather than efficiency.

Three questions have to be answered before any efficiency figure means anything:

1. What counts as useful output? For a light bulb, it is light, not heat. For an electric heater, heat is the whole point.
2. What counts as input? A power plant's efficiency can be calculated on the energy content of the fuel, but there are two conventions for fuel energy content (discussed below), and they give different answers.
3. Where is the boundary drawn? The efficiency of a car's engine is very different from the efficiency of the whole car, which differs again from the efficiency of the whole system from oil well to wheels.

Most confusing claims about efficiency come from leaving at least one of these unstated.

Efficiency versus effectiveness

Efficiency is often confused with effectiveness. The Wikipedia article draws the distinction this way: efficiency is about "doing things right," and effectiveness is about "getting things done." A process can be effective without being efficient, for instance by relying on longer working hours or cheap energy and labour to hit its targets.

The management writer Peter Drucker is widely credited with popularising this distinction. His 1967 book The Effective Executive contrasts efficiency, the ability to do things right, with effectiveness, the ability to get the right things done, and argues that for knowledge workers effectiveness matters more. (Quotation researcher Barry Popik notes that the pithy one-line version of this contrast that circulates as a Drucker quotation may be a paraphrase of what Drucker said in lectures and in the book, rather than his exact words.)

The distinction matters because efficiency is always efficiency at something. An organisation can become extremely efficient at a task that should not be done at all. The political scientist Deborah Stone, in Policy Paradox: The Art of Political Decision Making, makes a related point that the Wikipedia article quotes: efficiency is "not a goal in itself." It is a way of getting more of whatever goal one actually values, and different people value different outputs.

Efficiency in physics and engineering

The first law: nothing exceeds 100 percent

In physics, energy efficiency is the ratio of useful energy (or work) delivered to the energy supplied. Because energy is conserved, the useful output can never exceed the input. As the Wikipedia article puts it, the efficiency of a heat engine can never exceed 100 percent. All the energy that does not end up as useful output still exists; it has gone somewhere less useful, usually as low-temperature heat.

This is why efficiency is the natural language for discussing energy losses. A device that is 25 percent efficient wastes three quarters of what it takes in.

The second law: heat engines face a stricter limit

Heat engines, which turn heat into mechanical work, face a much tighter ceiling than 100 percent. The Carnot limit, named after the French engineer Sadi Carnot, states that the maximum efficiency of any heat engine operating between a hot reservoir at absolute temperature T_H and a cold reservoir at T_C is

η_max = 1 − T_C / T_H

Both temperatures must be absolute (kelvin or Rankine). The Wikipedia article on thermal efficiency gives an example: for hot gasoline combustion at 1,089 K and ambient air at 294 K, the maximum is 1 − 294/1,089 ≈ 0.730, or 73.0 percent.

Worked example (hypothetical). A power station boiler produces steam at 823 K (550 °C) and rejects heat to a river at 293 K (20 °C). The Carnot limit is 1 − 293/823 ≈ 0.644, or about 64 percent. No steam plant working between those temperatures could do better, however well engineered, and real plants fall well short because of friction, finite heat-transfer rates and other irreversibilities.

The Carnot limit explains why engineers push for ever-higher combustion and steam temperatures, and why combined-cycle plants, which use the hot exhaust of a gas turbine to raise steam for a second turbine, reach higher efficiencies than either cycle alone.

Real efficiencies of heat engines

The Wikipedia article on thermal efficiency gives representative values:

The gap between the ideal Otto cycle figures and the roughly 25 percent of a real gasoline engine is a measure of how much is lost to friction, heat transfer through cylinder walls, incomplete combustion, pumping losses and running at part load.

Power plants and heat rate

Electric utilities in the United States usually describe efficiency as "heat rate": the amount of fuel energy, in British thermal units (Btu), needed to generate one kilowatt-hour (kWh) of electricity delivered to the grid. The US Energy Information Administration (EIA) explains the conversion. One kilowatt-hour contains 3,412 Btu, so

efficiency = 3,412 / heat rate

The EIA gives two examples: a heat rate of 10,500 Btu/kWh corresponds to about 33 percent efficiency, and a heat rate of 7,500 Btu/kWh to about 45 percent.

Worked example. 3,412 / 10,500 = 0.325, or 32.5 percent; 3,412 / 7,500 = 0.455, or 45.5 percent. The EIA rounds these to 33 and 45 percent. A plant that improves its heat rate from 10,500 to 7,500 Btu/kWh burns 1 − 7,500/10,500 ≈ 28.6 percent less fuel for the same electricity.

The EIA also notes that heat rate is calculated on net generation, the electricity delivered to the transmission line after subtracting what the plant uses itself for fuel handling, pumps, cooling and pollution controls. That is an example of the boundary question in practice.

The fuel-value trap: HHV and LHV

When fuel burns, it produces water vapour. If that vapour is condensed, it gives up extra heat (its latent heat of vaporisation). Fuel energy content can therefore be quoted as a higher heating value (HHV), which includes that latent heat, or a lower heating value (LHV), which excludes it. The Wikipedia article on thermal efficiency is blunt about the consequence: "Not stating whether an efficiency is HHV or LHV renders such numbers very misleading." The same machine can appear several percentage points more efficient on an LHV basis than on an HHV basis.

Heat pumps: when "efficiency" exceeds 1

Heat pumps, refrigerators and air conditioners do not turn heat into work; they use work to move heat. Their performance is expressed as a coefficient of performance (COP): the useful heat delivered (or removed) divided by the work input. Because they move heat rather than create it, a COP above 1 is normal and does not violate any law.

The thermodynamic limits, as given in the Wikipedia article on thermal efficiency, are

COP_heating ≤ T_H / (T_H − T_C) COP_cooling ≤ T_C / (T_H − T_C)

and COP_heating = COP_cooling + 1.

Worked example (hypothetical). A heat pump draws heat from outdoor air at 0 °C (273 K) and delivers it to a heating system at 35 °C (308 K). The ideal heating COP is 308 / (308 − 273) = 308 / 35 = 8.8. Real heat pumps achieve much less than this ideal, but the calculation shows why heat pumps can deliver several units of heat per unit of electricity, and why their performance drops as the temperature difference widens on cold days.

The comparison with an electric resistance heater is instructive. The Wikipedia article notes that a resistance heater is close to 100 percent efficient at turning electricity into heat. A heat pump can deliver more heat than that from the same electricity, not by being "more than 100 percent efficient" at conversion, but by moving existing heat from outdoors.

Cars: where the energy goes

The US Department of Energy and Environmental Protection Agency's fueleconomy.gov site breaks down where the energy goes in conventional and electric vehicles.

For gasoline vehicles:

For all-electric vehicles, based mainly on a 2012 Nissan Leaf:

These tables illustrate the boundary question vividly. Measured from the energy in the tank or battery to the wheels, electric cars are several times more efficient than gasoline cars. But the electric car's numbers start at the wall socket. A full comparison would also count how efficiently the electricity was generated and transmitted, and how efficiently the gasoline was refined and delivered, which changes the totals considerably.

Chained efficiencies

When energy passes through several stages, the overall efficiency is the product of the stage efficiencies, not their average.

Worked example (hypothetical). Electricity generated at 45 percent efficiency (the EIA's 7,500 Btu/kWh example) passes through a grid assumed, for illustration, to deliver 95 percent of it, then charges a car with 10 percent charging losses (90 percent), and the car delivers 89 percent of the battery energy to the wheels (within the EPA's combined range including regenerative braking). The overall efficiency from fuel to wheels is

0.45 × 0.95 × 0.90 × 0.89 ≈ 0.342, or about 34 percent.

Even though every stage after the power plant is 89 percent or better, the result is dominated by the weakest link. The general lesson is that improving the least efficient stage usually offers the largest gains.

Lighting

Lighting is a classic case where the useful output is a small fraction of the input. An archived US Department of Energy Energy Saver page states that residential LEDs, especially ENERGY STAR rated products, "use at least 75% less energy, and last 25 times longer, than incandescent lighting." Strictly, lighting is measured in efficacy (lumens of light per watt of electricity) rather than as a percentage efficiency, because light output is weighted by how sensitive the human eye is to each colour. But the principle is the same: an incandescent bulb turns most of its input into heat, and an LED turns far more into visible light.

Other physical efficiencies

The Wikipedia article lists a number of specialised efficiency measures used in physics and engineering. In brief:

Efficiency in living systems

Photosynthesis

Plants are, by engineering standards, inefficient solar collectors. The Wikipedia article on photosynthetic efficiency gives these figures for the fraction of solar energy converted to biomass:

The article contrasts this with photovoltaics: a typical solar installation averages about 22 watts per square metre, roughly 10 percent of average insolation, several times the photosynthetic figures above. It also notes that the two outputs differ in kind: solar panels produce electricity, while plants produce stored chemical energy that can become food or liquid fuel.

Ecological efficiency and the "10 percent rule"

Ecological efficiency is the efficiency with which energy moves from one trophic level (feeding level) of an ecosystem to the next: from plants to herbivores, herbivores to carnivores, and so on. The Wikipedia article on the subject describes the "ten percent law": net production at one level is typically about 10 percent of the level below it. Most of the rest is lost to respiration, to waste, and to organisms that die without being eaten.

The idea is credited to the ecologist Raymond Lindeman, writing in 1942, though he did not call it a law. Lindeman reported ecological efficiencies ranging from about 0.1 percent to 37.5 percent. Later work found that marine food chains tend to be more efficient than terrestrial ones: about 20 percent between phytoplankton and herbivorous zooplankton, and up to about 15 percent between higher marine levels, compared with roughly 10 percent between terrestrial levels. Marine plankton is more digestible, and marine animals spend less energy on temperature regulation, movement and parental care.

Worked example (illustrative, using the 10 percent rule). If the plants in a field fix 10,000 kilocalories of net energy, the herbivores that eat them build about 1,000 kilocalories into their bodies, the predators that eat those herbivores about 100, and a top predator about 10. After three transfers, only 0.1 percent of the original energy remains. This is why food chains rarely have more than a handful of links, and why large predators are scarce compared with the plants that ultimately feed them.

Efficiency in economics

Economists use "efficiency" in several distinct senses. Most describe how well resources are allocated, rather than how well a single machine converts inputs to outputs.

Pareto efficiency

The most fundamental concept is named after Vilfredo Pareto (1848–1923), an Italian civil engineer and economist. A Pareto improvement is a change that makes at least one person better off without making anyone worse off. An allocation is Pareto efficient (or Pareto optimal) when, in the Wikipedia article's words, "all possible Pareto improvements have already been made."

Pareto efficiency is a weak condition in one important sense: it says nothing about fairness. The Wikipedia article stresses that an economy in which a wealthy few hold most of the resources can still be Pareto efficient. In its example, splitting a pie in half between two of three people is Pareto efficient, even though the third person gets nothing, because any change would make one of the first two worse off.

The first fundamental theorem of welfare economics, first demonstrated mathematically by Kenneth Arrow and Gérard Debreu, states that under certain conditions a competitive market produces a Pareto-efficient outcome. Those conditions are demanding: markets must exist for all goods, there must be no externalities, markets must be perfectly competitive, and participants must have perfect information. Much of applied economics is about what happens when these conditions fail.

Kaldor-Hicks efficiency

Because almost every real policy change creates some losers, the Pareto criterion is rarely satisfied in practice. Nicholas Kaldor and John Hicks, who both published their proposals in 1939, offered a looser test. Under Kaldor-Hicks efficiency, a change is an improvement if the winners gain enough that they could, in principle, compensate the losers and still come out ahead, whether or not compensation is actually paid.

The Wikipedia article on Kaldor-Hicks efficiency explains that this idea underlies cost-benefit analysis: a project such as a new airport goes ahead when its total benefits exceed its total costs. The criterion has well-known weaknesses. It is unclear why hypothetical compensation should matter if no compensation is paid. Tibor Scitovsky showed that the Kaldor test alone can rank A above B and B above A at the same time, and the combined test may not be transitive.

A catalogue of inefficiencies

The Wikipedia article on efficiency lists several kinds of economic inefficiency:

These categories can conflict. A firm can be productively efficient (making its product at minimum cost) while the economy as a whole is allocatively inefficient (making too much of that product relative to what people want).

Other economic uses

• Market efficiency. In finance, an efficient market is one in which prices reflect available information, so that it is hard to earn above-average returns consistently without taking above-average risk.
• Efficiency wages. The theory that employers may pay above the market-clearing wage because higher pay raises productivity, reduces turnover or improves morale.
• Business efficiency. Measures of how well a firm converts inputs such as labour and capital into outputs, the subject of productivity measurement texts such as Sickles and Zelenyuk's Measurement of Productivity and Efficiency (Cambridge University Press, 2019), which the Wikipedia article cites.

The Efficiency Movement: when the word became a cause

In the United States, "efficiency" was more than a technical term in the early twentieth century. It was a political and social movement. The Wikipedia article on the Efficiency Movement says it flourished in the United States from 1890 to 1932.

Key figures

Efficiency in government

The movement extended from factories into government. The Wikipedia article lists, among others, the Chicago Efficiency Division (1910–1916), the Chicago Bureau of Public Efficiency (1910–1932) and the Massachusetts Commission on Economy and Efficiency (1912). The idea was that public administration, like manufacturing, could be improved by expert study and the elimination of waste.

Decline

According to the same article, the movement was somewhat discredited after 1929, when Democrats blamed the Great Depression on Hoover. The underlying techniques survived in industrial engineering, operations research and management science, but the idea that efficiency was a sufficient political programme did not.

The history is a useful reminder of Deborah Stone's point. Taylor's methods raised output per worker, but the question of who benefited from that higher output, and at what cost to the workers whose every motion was being timed, was a matter of fierce dispute. Efficiency answered "how," not "for whom."

Efficiency in computing

Algorithmic efficiency

In computer science, the efficiency of an algorithm describes how much time (number of steps) and memory it needs as the size of its input grows. Rather than measuring exact running times, which depend on the hardware, computer scientists usually describe how the cost scales, using "big-O" notation.

Worked example (illustrative). Consider sorting a list of one million items. A simple algorithm whose cost grows with the square of the input size (n²) needs on the order of (10⁶)² = 10¹² basic steps. An algorithm whose cost grows as n × log₂ n needs on the order of 10⁶ × log₂(10⁶) ≈ 10⁶ × 19.93 ≈ 2 × 10⁷ steps. The second algorithm needs roughly 50,000 times fewer steps. For small lists the difference hardly matters; for large ones, it is the difference between a fraction of a second and hours. This is why algorithmic efficiency usually matters far more than small speedups in hardware.

Other computing measures

The Wikipedia article also lists the efficiency factor and storage efficiency, the latter describing how much useful data a storage system holds relative to the raw capacity it consumes. In each case the logic is the same: useful output over resources consumed.

Efficiency in statistics

In statistics, efficiency measures the quality of an estimator: how precisely it estimates a quantity from a given amount of data. The Wikipedia article on statistical efficiency defines the efficiency of an unbiased estimator T of a parameter as the ratio of the smallest possible variance (set by the Cramér-Rao bound, which depends on the Fisher information) to the estimator's actual variance. An efficient estimator achieves that smallest variance, so its efficiency is 1, or 100 percent.

Worked example. For data drawn from a normal distribution, the sample mean reaches the Cramér-Rao bound and is 100 percent efficient. The sample median, for large samples, has a variance of about π/(2N) (for unit variance), which gives an asymptotic efficiency of 2/π ≈ 64 percent. In practical terms, the median needs roughly 1/0.64 ≈ 1.57 times as many observations to estimate the centre as precisely as the mean does, which is equivalent to the article's statement that the median's variance is about 57 percent higher.

This is a good case study in why efficiency is not the only virtue. The median is more robust to outliers, so if there is doubt about whether the data really are normal, the less efficient estimator may be the better choice. Efficiency is measured against a model; if the model is wrong, the "efficient" choice can be worse.

The Jevons paradox: when efficiency increases consumption

It seems obvious that making something more efficient should reduce how much of it is used. In 1865, the English economist William Stanley Jevons argued in The Coal Question that the opposite can be true. He observed that England's coal consumption had soared after James Watt's steam engine, which was more efficient than Thomas Newcomen's earlier design, made coal a cheaper source of power. Jevons wrote: "It is a confusion of ideas to suppose that the economical use of fuel is equivalent to diminished consumption. The very contrary is the truth."

The rebound effect

Modern economists describe the mechanism as the rebound effect. An efficiency gain lowers the effective cost of whatever the resource provides (miles of travel, hours of light, tonnes of steel), which increases demand for it. Some of the expected savings are "taken back" by greater use. The Wikipedia article on the Jevons paradox states that "the Jevons paradox occurs when the rebound effect is greater than 100%, exceeding the original efficiency gains."

In the 1980s, Daniel Khazzoom and Leonard Brookes revived the argument for energy use across whole economies. In 1992, Harry Saunders named the hypothesis that efficiency gains "work to increase (rather than decrease) energy consumption" the Khazzoom-Brookes postulate.

How big is the rebound?

The evidence suggests that full Jevons-style backfire is not the usual case. The Wikipedia article reports that at the microeconomic level "the rebound effect is usually less than 100%," citing studies by Greening and colleagues and by Small and Van Dender, and that direct rebound is usually small in mature markets such as oil in developed countries. It also notes that Gillingham and colleagues argue the rebound effect is overplayed, and that other environmental economists see limited empirical evidence for widespread Jevons paradox effects. The size of the rebound depends on how sensitive demand is to price.

Illustrative example (from the Wikipedia diagrams). Suppose cars become 20 percent more fuel-efficient. If demand for travel is elastic, travel might rise by 40 percent and fuel use would increase. If demand is inelastic, travel might rise by only 10 percent and fuel use would fall, though by less than the full 20 percent.

The practical upshot is that efficiency improvements usually save resources, but rarely as much as a naive calculation predicts. Where the goal is to reduce total consumption, efficiency policies are often combined with pricing, standards or caps.

Energy efficiency as global policy

Energy efficiency has become a central plank of energy and climate policy. The International Energy Agency (IEA) tracks progress using global primary energy intensity, the amount of energy used per unit of economic output. In its Energy Efficiency 2024 report, the IEA projected that global energy intensity would improve by roughly 1 percent in 2024, the same pace as 2023 and about half the 2010–2019 average.

At the COP28 climate conference, governments agreed to double the global rate of energy efficiency improvement, from about 2 percent in 2022 to 4 percent a year by 2030. The IEA said the world was not yet on track to achieve this. It also reported that governments representing more than 70 percent of global energy demand had implemented new or updated efficiency policies in 2024.

Note how this measure works: energy intensity is energy per unit of GDP, so it improves both when technologies become more efficient and when an economy shifts toward less energy-intensive activities. It is an efficiency ratio in the broad sense, with all the boundary questions that implies.

Measuring and improving efficiency in practice

Across fields, people who measure efficiency well tend to follow a similar discipline.

1. Define the useful output precisely. Is a lamp's output light in all directions, or light on the desk? Is a vehicle's output passenger-miles or vehicle-miles?
2. State the input and its basis. For fuel, say whether heating values are higher or lower. For electricity, say whether figures start at the power plant, the grid, or the wall socket.
3. Draw the boundary explicitly. Engine, vehicle, or well-to-wheels? Factory, firm, or supply chain?
4. Use a baseline and the same conditions. Efficiency often depends on operating point: a car engine's efficiency at highway cruising differs from its efficiency in city traffic, as the fueleconomy.gov tables show.
5. Find the weakest link. In chained systems, overall efficiency is the product of stages, so the least efficient stage often offers the biggest improvement.
6. Compare against the right limit. A heat engine should be compared with its Carnot limit, not with 100 percent. A crop's photosynthetic efficiency should be compared with the theoretical limit for usable sunlight.
7. Watch for rebound. If cheaper service leads to more use, total savings will be smaller than per-unit savings.

Common misconceptions

"Efficiency and effectiveness are the same thing." They are not. Efficiency is about the ratio of output to input; effectiveness is about achieving the right goal. A process can be efficient at the wrong task, or effective but wasteful.

"Anything below 100 percent efficiency is badly designed." For heat engines, the relevant ceiling is the Carnot limit, which can be far below 100 percent. A power plant at 45 percent may be very well engineered given its temperatures.

"Heat pumps break the laws of physics by being more than 100 percent efficient." Heat pumps move heat rather than create it, so their coefficient of performance can exceed 1 without violating energy conservation.

"Electric cars are 90 percent efficient, so they use a tenth of the energy of a 9 percent efficient system." Only if both figures share the same boundary. The EPA's electric-vehicle figures start at the charger; a full comparison must include generation and transmission.

"More efficiency always means less consumption." The rebound effect and, in extreme cases, the Jevons paradox mean efficiency gains can be partly or wholly offset by increased use.

"An efficient economy is a fair economy." Pareto efficiency is compatible with extreme inequality, as the Wikipedia pie example shows. Fairness is a separate question.

"The most efficient estimator is always best." Statistical efficiency is defined relative to a model. A less efficient but more robust method, such as the median, can be better when the model is uncertain.

Frequently asked questions

What is the simplest definition of efficiency? Useful output divided by total input. When both are in the same units, the result is a number between 0 and 1, often expressed as a percentage.

Can efficiency ever exceed 100 percent? Not for energy conversion, because energy is conserved. Figures above 100 percent usually mean either a different kind of ratio (such as a heat pump's coefficient of performance) or an inconsistent accounting basis (such as using lower heating values for fuel while recovering latent heat from water vapour).

What is the Carnot efficiency? The maximum possible efficiency of a heat engine operating between two temperatures, equal to 1 − T_C/T_H with temperatures in kelvin.

How efficient are power plants? In the EIA's examples, a heat rate of 10,500 Btu/kWh corresponds to about 33 percent efficiency and 7,500 Btu/kWh to about 45 percent. Combined-cycle plants approach 60 percent, according to the Wikipedia article on thermal efficiency.

How much of a gasoline car's fuel energy moves the car? According to fueleconomy.gov, about 18 to 25 percent in combined driving, with 14 to 20 percent in the city and 22 to 30 percent on the highway.

What is the 10 percent rule in ecology? The rough observation that each trophic level retains about 10 percent of the energy of the level below. Lindeman, who introduced the idea in 1942, reported values from about 0.1 to 37.5 percent, so 10 percent is a rule of thumb, not a law.

What is Pareto efficiency? An allocation in which no one can be made better off without making someone else worse off. It says nothing about whether the allocation is fair.

What is the Jevons paradox? The observation, first made by William Stanley Jevons in 1865 about coal, that efficiency gains can increase total consumption of a resource by making it cheaper to use. It occurs when the rebound effect exceeds 100 percent.

What does "efficiency" mean for an algorithm? How its time and memory requirements grow as the input grows, usually expressed in big-O notation.

What does "efficiency" mean in statistics? How close an estimator's variance comes to the smallest variance theoretically possible. The sample median of normally distributed data has an efficiency of about 64 percent relative to the sample mean.

Why did the "efficiency movement" fade? According to the Wikipedia article on the movement, it was somewhat discredited after 1929, when the Great Depression was blamed on Herbert Hoover, its most prominent national figure.

Summary

Efficiency, at its simplest, is useful output divided by input. That one idea has been made precise in very different ways across disciplines. Physics fixes hard ceilings: no conversion can exceed 100 percent, and heat engines face the stricter Carnot limit. Engineering measures how close real machines come, from cars that move only a fifth or so of their fuel energy to the wheels to combined-cycle plants approaching 60 percent. Biology shows that sunlight-to-biomass conversion runs at a few percent at best, and that energy thins out rapidly up a food chain. Economics asks whether resources are allocated so that no one can be made better off without someone being made worse off, and it reminds us that such an allocation can still be grossly unequal. Computing and statistics measure efficiency in steps and in variance. Across all of them, the same cautions apply: define the output, the input and the boundary; compare with the right limit; and remember, as Jevons did, that doing more with less can also mean doing more.

Sources

• Efficiency (Wikipedia)
• Thermal efficiency (Wikipedia)
• US Energy Information Administration, How efficient are power plants? (FAQ)
• fueleconomy.gov, Where the Energy Goes: Gasoline Vehicles
• fueleconomy.gov, Where the Energy Goes: All-Electric Cars
• US Department of Energy, Energy Saver: LED Lighting (archived 2016)
• International Energy Agency, Much faster progress on energy efficiency is needed to meet global 2030 goal (2024)
• Photosynthetic efficiency (Wikipedia)
• Ecological efficiency (Wikipedia)
• Pareto efficiency (Wikipedia)
• Kaldor-Hicks efficiency (Wikipedia)
• Efficiency Movement (Wikipedia)
• Efficiency (statistics) (Wikipedia))
• Jevons paradox (Wikipedia)
• Barry Popik, "Management is doing things right; leadership is doing the right things"