clock menu more-arrow no yes

Hot days are worse than cold ones, and this economics research proves it

Could be worse!
Could be worse!
Scott Eisen/Getty Images

Everybody talks about the weather. But now someone has subjected it to rigorous mathematical analysis to produce important conclusions about economic growth. Specifically Tatyana Deryugina and Solomon Hsiang find that hot days are bad for the economy — not just in poor countries with an overwhelmingly agricultural workforce, but in the United States of America. Here in the US, "productivity of individual days declines roughly 1.7% for each 1.8°F increase in temperature above 59°F." That means that "a weekday above 30°C (86°F) costs an average county $20 per person."

Temperature over the weekend has little impact on the economy, but on weekdays it's a big deal. Hot days kill growth both because of negative impacts on agriculture and because it seems that, despite air conditioning, people simply work less efficiently when it's hot. It sounds like a bit of a frivolous research topic, but it has broad implications for everything ranging from climate change to the macro-history of the world to housing policy to my classic debate with Vox editor-in-chief Ezra Klein about the merits of extreme cold versus extreme heat.

Hot weather is bad for the economy

The impact of a single hot day on a county's economy is rather small (if you're in to methodology, skip to the end). There are a lot of days in the year, and there's more to life than the weather. But the impact is clear and it is negative. The negative impact shows up if you look at the whole sample since 1970, and if you break the sample down into decade-long spans it is present in each and every decade. That's a decent indicator that the statistical analysis is robust. It also shows that the detrimental impact of hot weather has not been substantially altered by changes in the American economy over time.

And while any given hot day has a small impact, since hot days are reasonably common the aggregate impact of hot weather on the American economy is pretty big. They estimate that "if counties could choose daily temperatures to maximize output, rather than accepting their geographically determined endowment ... annual income growth would rise by 1.7 percentage points."

Over time that's quite a lot.

The significance of Deryugina & Hsiang's findings

This is important research across several dimensions:

  1. Ezra Klein is wrong: Crucially, the researchers have debunked Klein's advocacy for super-hot days as superior to super-cold days. With the Northeast Corridor stuck in a hellish chill this week, it is critical to remember that things could be worse. Extreme cold is unpleasant, but as a society we can live through it and even thrive. Extreme heat saps our will to live in a more fundamental way, crushing the economy and driving long-term immiseration.
  2. Climate change is even scarier: Standard economic analyses of climate change generally assume large negative impacts on poor countries, whose economies have little ability to adapt to changes in weather patterns. They also assume rich countries could experience negative impacts related to flooding, hurricanes, and other disasters. But the finding that an increase in the number of random Tuesdays on which Chicago experiences weather in the 80s rather than the 70s indicates that climate change is even more alarming than previously thought. More hot weather — and not even crazy hot weather above 100°, just regular old hot days — turns out to be economically costly, even before considering secondary climate impacts.
  3. World history may be pretty simple: It's a well-known stylized fact that hot countries are poorer than cold ones. Europe is richer than Africa. Sweden is richer than Italy. Morocco and South Africa are richer than the Democratic Republic of Congo. Argentina is richer than Colombia. Simply attributing this pattern to the weather seems almost too silly. So scholars either generally assume complicated indirect impacts, through which climate alters institutional development (see, e.g., Acemoglu & Robinson) or attribute it to a giant coincidence. But maybe the basic hotness of hot countries really is holding them back.

America has troubling population trends: Because the coastal US lacks affordable housing due to restrictive zoning, and because people like to complain about cold weather, there is a strong trend in the United States for people to move to hot cities. The populations of the Texas Triangle metro areas and of sunbelt cities from Phoenix to Atlanta are growing quickly. Having people move to hot cities is similar in its economic impact to the country getting hotter — which is to say it's bad. Short of evangelizing for people to move to Minneapolis, it's hard to force people to avoid the sweltering, output-destroying temperatures of the sunbelt. But changes to zoning policy in temperate-to-cold metro areas where people do want to live (Bay Area, Seattle, Northeast Corridor) could do a lot to help.

Deryugina & Hsiang researched the impact of daily temperature

They took National Climatic Data Center temperature measurements and mapped them onto US counties, to create a huge database of county-level high and low temperatures.

We can't measure is the level of economic output on any given day. Instead they used Bureau of Economic Analysis data on yearly economic output for each county. Operating under the assumption that "annual income is a linear combination of many daily incomes that are similarly affected by temperature" they then do a statistical analysis to show how a county's annual income is impacted by the number of hot days it experiences over the course of a year.

The resulting analysis isn't just the naive observation that hotter counties are, on average, poorer than richer ones (though that is true). Instead, they have a regression analysis "augmented to account for autocorrelation, precipitation, lagged effects of weather, unobserved heterogeneity across counties, and non-linear time trends." Because there are a lot of counties in the United States the data is sufficiently rich to provide very suggestive information about the specific causal impact of hot weather on counties.