How Electricity Statistics Are Collected

Electricity statistics come from two very different places. National statistics offices and energy ministries collect them from power plants, usually by survey, and publish monthly or annual figures that are revised for months afterwards. Grid operators measure them with meters, second by second, and publish within hours. The two describe the same electricity system and rarely produce the same number, because they count different things, cover different plants and are corrected on different timetables.

The statistical route

Most countries build electricity statistics the way they build any other economic statistic. Plant operators file returns, typically monthly, stating how much they generated, how much fuel they burned and how much they sold. A national body aggregates those returns, estimates whatever is missing, and publishes.

That process has several features worth knowing about.

Coverage is decided, not automatic. A survey has a threshold. Plants below it are estimated rather than measured, and small generation is now a large category: rooftop solar on hundreds of thousands of houses, farm scale wind, factory generators. Countries with strong metering of small systems capture this well. Others estimate it from installed capacity and assumed output, which introduces a bias that can persist for years.

Self consumption is a judgement call. A steelworks or paper mill that generates its own power may never touch the public grid. Whether that electricity is counted at all, and whether it counts as generation, varies between countries and sometimes between datasets within one country.

Late data is normal. A first estimate for a month may be published four to eight weeks after the month ends, and the annual figure may still be moving two years later.

International compilations sit on top of this. They collect each country’s published figures, map them onto one set of fuel categories, fill gaps with estimates, and restate everything in common units. That step adds consistency and, unavoidably, another layer of judgement.

The operator route

Grid operators are not trying to produce statistics. They are trying to run a network, and their data is a by product of that. Meters at substations and power plants report to control room systems every few seconds, and those readings are aggregated into hourly or sub hourly totals.

Operator data is fast and detailed, with generation by fuel, demand and flows across boundaries, often within an hour or two of real time. It is also provisional. Meters fail, telemetry drops out, plants are mapped to the wrong category, and a figure published this afternoon may be replaced tomorrow. Operators publish revision policies for exactly this reason.

Operator data also has a hard boundary: it covers what the operator can see. Generation behind a customer’s meter, above all rooftop solar, usually appears as reduced demand rather than as generation, which is why operator demand curves in sunny regions dip at midday in a way that looks like people using less electricity.

Gross against net generation

This is the single most common reason two honest sources differ.

A power station consumes electricity to run itself: pumps, fans, cooling, pollution controls. Gross generation is what comes out of the generator terminals. Net generation is what is left after the plant’s own use. The gap is typically a few percent for a gas turbine and larger for a coal plant with flue gas treatment.

Statistical agencies often report gross. Grid operators almost always report net, because net is what actually reaches the network. Comparing a gross figure from one source with a net figure from another produces a discrepancy of several percent that looks like an error and is not.

The same trap exists on the demand side. Terms such as demand, consumption, load and final consumption can each include or exclude transmission and distribution losses, storage charging, and electricity used by the energy sector itself.

Why sources disagree, in summary

• Gross against net generation.
• Different coverage of small, distributed and self consumed generation.
• Different treatment of storage, which may be netted off, reported separately, or counted twice.
• Different fuel categories, especially for waste, biomass, blast furnace gases and hybrid plants.
• Different vintages, with one source revised and the other not.
• Different calendars, where a statistical year may not be a calendar year.

A gap of a few percent between two reputable sources for the same country and year is ordinary. A gap of tens of percent usually means the two figures are not the same measure.

How this site handles it

World Power Monitor keeps the two routes apart on purpose. Compiled national statistics, used for every country page, come from one compiler and carry the year or month they describe. Grid operator readings, used for the US grid zones, come from the operator’s own published data and carry a timestamp. The two are never averaged, blended or combined inside a single figure, because doing so would produce a number that is not a measurement of anything.

Each page therefore shows one source per section, names it, and states the period. Where the same country appears in both layers, the numbers will differ, and that difference is information rather than a defect.

The full description of which dataset feeds which page, how revisions are handled and how fuel categories are mapped is in our methodology. Every dataset, its publisher and its licence are listed on our sources page.

Reading any electricity statistic sceptically

Three questions settle most confusion. What period does this cover, and has it been revised? Is it gross or net, and does it include small and self consumed generation? Who measured it, a meter or a survey? If a chart or a headline cannot answer those, treat the number as an indication rather than a fact.