Emissions
A cold winter changes how much gas a country burns for heat, not what a regulated tariff charges. Tested on what countries emitted, the weather explains much of the year-to-year swing.
Correlation, heating demand vs CO₂
Of the CO₂ swing explained by weather
Year-pairs, pandemic excluded
Leaving out the two pandemic years, the panel's change in CO₂ follows its change in heating demand with a correlation of 0.52 across 15 year-pairs, so the weather alone accounts for 27% of the year-to-year swing in what these countries emit. On the fitted slope a winter 10% colder than the one before adds about 2.1% to the year's CO₂ across the panel.
The chart shows it without the statistics. The warm winter of 2014 took heating demand down by about a sixth and the panel's emissions fell with it; the colder 2015 pushed demand back up by an eighth and emissions rose. The years that break the pattern are years something else happened, and they are worth reading as exceptions: in 2019 the panel's coal emissions fell by about a seventh while heating demand barely moved.
Country by country, 25 of the 29 countries emitting 10 Mt or more move with their winters, and 10 of them at a correlation of 0.5 or better, where a winter 10% colder than the last typically adds 2.3% to national CO₂.
The bottom of the chart is where there is little to heat, as in Spain and Portugal; where heat is already electric and the electricity is hydro, as in Norway; or where growth swamps the winter, as in Türkiye, whose emissions have nearly doubled since 2005. The table below adds the gas line on its own, which is the fuel a winter mostly moves.
Ten or so year-pairs is a short series, and any one country's correlation carries wide error: at that length, about 0.63 is where a correlation stops being distinguishable from chance at the conventional 5% level. The evidence is the pattern rather than any single bar. Nearly every country leans the same way, and the lean is strongest where winters are cold and heating burns gas.
So what. When somebody attributes an emissions movement to policy, efficiency or fuel switching, "it was a milder year" is the cheapest competing explanation and usually the one nobody checks. For emissions it is often most of the answer: a country or a company whose footprint is mostly heating fuel can report a cut of a few percent after a mild winter and have reported the weather. Normalising a year-on-year comparison for degree days, or fitting it the way this section does, is routine in energy management, and an emissions figure needs it for the same reason.
Who acts: whoever reports year-on-year emissions progress, and whoever assures it. Cost of getting it wrong: crediting a warm winter to a programme, then explaining the "reversal" the next time the winter is cold.