Demand and weather, 2001–2026

How much each degree of heat or cold moves Argentina’s electricity demand, how that changed in 25 years, and where. Built only from open public data.

NoteResumen en castellano

Cuánto mueve la demanda eléctrica argentina cada grado de calor o de frío, cómo cambió en 25 años y dónde, solo con datos abiertos y temperaturas de las estaciones del Servicio Meteorológico Nacional.

  • La red se volvió unas cinco veces más sensible al calor y cuatro veces más sensible al frío que en 2001–2005.
  • Los hogares explican tres cuartos de la respuesta al calor y casi toda la del frío; en ellos, el frío pesa casi lo mismo que el calor, aun en un país con mucha red de gas.
  • Comercio e industria responden al calor pero casi nada al frío; los grandes usuarios industriales, a ninguno de los dos.
  • El salto más grande de la respuesta al calor llegó en la segunda mitad de los 2010.
  • En 17 de las 19 distribuidoras el consumo se volcó hacia el verano entre 2014–2016 y 2023–2025.
  • Desde 2022, veranos más calurosos que la normal 1991–2020 sumaron entre 1,5 y 2,6 TWh por año.

Today’s numbers

A steeper U

Each point is one month: its average demand relative to the average of its year (so that growth from one year to the next does not count), against the month’s mean temperature. The U is the grid’s response to heating and cooling. Compare how steep it was in the first years of the series and in the last ones.

Lines: straight-line fits on the cold side and on the warm side of each period, for reading only; the MW per degree come from the model below. 2020 is left out because of the pandemic.

MW per degree, period by period

The extra demand for each degree of daily temperature below the comfort band (cold) or above it (heat), estimated separately for each period, with 95 % intervals.

Residential, commercial and industrial, and large users are published separately from 2005; the total, from 2001. Large users are the large industrial and commercial customers of the wholesale market: they serve as a check, since their consumption should not depend on the weather.

Where: winter against summer, distributor by distributor

Average daily consumption of each winter (June–August) divided by that of the preceding summer (December–February), for the 19 distribution companies published separately. Below 1, the distributor uses more energy in summer than in winter. No model is involved: it is a ratio of consumption.

Weather’s share of each year

How much more (or less) energy each year used because its weather differed from the climate normal of 1991–2020, using the MW per degree of its period. Above zero: hotter summers or colder winters than normal.

My reading

New · 2026-10-10

The grid became far more sensitive to the weather. In 2001–2005 each degree of heat added about 180 MW to Argentina’s demand; in 2021–2026, about 940 MW, five times more. Each degree of cold went from about 160 to 610 MW, almost four times more. As a share of base demand, heat went from under 2 % to about 7.5 % per degree. Demand itself did not grow nearly as much, so this is a change in what electricity is used for, not only in how much is used.

It is a household story. Households carry three quarters of the heat response and practically all of the cold one. Within households, cold costs almost as much as heat: about 600 MW per degree against 715. In a country where most of the population has piped gas, that suggests electric heating (split air conditioners in heat-pump mode, electric heaters) is now a large part of winter demand. Commerce and industry respond to heat (air-conditioned shops, offices and plants) but barely to cold. Large industrial users respond to neither, as expected: their consumption follows production, not the thermometer.

The jump in heat came in the late 2010s. The heat response grew steadily through the 2000s, but its largest step came between 2011–2015 and 2016–2019, when it rose by about 60 %. That is the period in which the summer peak overtook the winter one for good (see the peak demand panel).

The whole country moved towards summer, not only the hot north. In 17 of the 19 distributors published separately, winter consumption fell relative to summer between 2014–2016 and 2023–2025. In Formosa, Chaco and Santiago del Estero, winter now uses barely 53–57 % of summer’s energy. Buenos Aires and the City, the largest market, stayed where they were, with winter slightly above summer; Santa Cruz remains strongly winter-driven; Misiones is the only one that moved towards winter.

Recent summers were hot, and part of the recent growth is weather. Cooling degree-days have been well above the 1991–2020 normal since 2022. Using the MW per degree of each period, that added between 1.5 and 2.6 TWh a year, 1 to 2 % of demand. Corrected for weather, demand grew 1.4 % between 2022 and 2025, against 1.8 % without the correction.

Monthly and hourly estimates

These panels use monthly data; the first study used hourly demand and daily temperatures. The two approaches agree on the comfort band and on the heat response, and differ on the cold response, which the monthly model estimates higher.

  • The temperature source moves the result by about 5 to 10 %. Station measurements and reanalysis grids do not record exactly the same temperature, especially on the hottest days. This explains most of the difference in the heat response.
  • Monthly data overstate the cold response by about 15 %. In a winter month, cold arrives together with other seasonal effects that also raise demand (longer nights, holidays, habits). A daily model separates them by comparing cold and mild days within the same month; a monthly model cannot do it completely, even with a term for the hours of darkness.
  • What this panel is for. Its strength is comparison: between periods, sectors and regions, with the same method and data throughout. For the precise MW per degree of the recent years, the hourly study is the reference.

Method and sources

  • Model. Monthly energy explained by a level for each year, the number of non-working days, the hours without daylight (lighting rises in winter because nights are longer, not because it is cold), and two slopes: one per degree-day below and one per degree-day above a comfort band. Degree-days are computed with the daily mean temperature, so the monthly model is the exact sum of a daily one. The band is estimated with the whole period (2001–2026) and kept fixed so that periods and sectors can be compared. 2020 is excluded. Intervals: bootstrap of months within each period (500 resamples).
  • Temperature. Daily mean temperature of the Servicio Meteorológico Nacional station of each of the 22 provinces of the interconnected grid (21 of them within 70 km of the provincial capital; Formosa uses Presidencia Roque Sáenz Peña), weighted by the number of households of each province (INDEC, 2022 census). The few missing days (24 out of almost 13,000) are interpolated. NASA POWER is used as a check.
  • Climate normal. 1991–2020, the current reference period of the World Meteorological Organization, from the same stations.
  • Distributors. Ratio of average daily consumption in June–August to that of the preceding December–February; Buenos Aires and the City combine Edenor, Edesur, Edelap and EDEA.
  • This page does not offer the data for download. To obtain the data, go to the sources: