One of the reasons we think we should take action about climate change is that the costs of doing something about the problem are lower than the stream of future damages if we fail to act.

Figuring out what damages from climate change will be 100 or more years into the future is difficult. It requires pairing climate models with economic models that simulate a global economy far out into the future. These very long-range forecasts make many economists’ skin crawl. Could we have predicted today’s economy on the eve of World War I?

There are many good and many not so good papers which provide estimates of what future weather and climate will do to a variety of sectors like agriculture, energy, disease, conflict, etc. One of the hardest things to figure out, though, is how much human well-being will be lost from factors that are not traded in markets.

One of my favorite new papers on the topic is by my student, Patrick Baylis, who is entering the shark tank (also known as the job market for economists). Patrick’s job market paper is available here. He cites observations going back hundreds of years, which point out that being hot makes people unhappy. While Max A. might be very grouchy on a hot day and his wife Lori G. might be able to subjectively scale how much more grouchy he is compared to a cold day, this is hardly a reliable measurement and there could be other factors that make Max grouchy on a hot day.


Patrick notes that there is a giant database of expressions of human emotion, which he can download. It’s called Twitter. You should try it sometime. It turns out that this large social media site has an interface which allows you to download tweets, many of which are geocoded.

Figure 1 above shows the density of tweets across the US, which largely mimics the distribution of population. Patrick uses algorithms from computational linguistics to translate the occurrence of certain words into “happiness” scores.

He also looks for the occurrence of emoticons and swear words. He matches those scores to temperatures and a number of other climate indicators and estimates the impacts of different temperatures on revealed happiness. Since he observes the same individual at many points in time, he can make sure that his results are not driven by the fact that happier people may simply be living in more pleasant areas.

His findings are really interesting:

  1. An individual living a day with an average (not maximum!) temperature in the 80-90 degree range, relative to a day in the 60-70 degree range, experiences a drop in happiness similar to a drop in happiness from a Sunday to a Monday. I don’t know about you, but for me that is essentially going off an emotional cliff. The figure below shows this for one of his measures of happiness quite clearly.
  2. dat

  3. He doesn’t observe a similar drop in happiness on cold days. He shows convincing evidence that this is not a selection effect of who lives where. But this might be due to the fact that we can much better fight off cold than we can warmth.
  4. “It’s not the heat, it’s the humidity.” He shows that hot and humid days are much worse than days with just high temperatures, which is probably a finding that residents of Washington D.C. can verify.
  5. He shows that people adapt partially but not fully to hot climates. In other words, people in Texas are more heat tolerant than people in Minnesota, but they are still negatively affected when it gets hot out.
  6. He also shows suggestive evidence that on hot days the frequency of typographical errors skyrockets. This is further evidence that on hot days cognitive ability decreases, which has been shown elsewhere.
  7. When graduate students go on the market, the advice given is often to nudge the boundary in a field, since this is how scientific progress is made. Patrick’s work is more than a nudge: it takes an understudied question in the literature, pulls from work in other disciplines to establish a new methodology to investigate that question, and uses that methodology to document the answer across multiple dimensions.

    But hey, I’m biased. I'm his advisor and all my academic children are above average. At the end of the paper, Patrick implements a preliminary technique to estimate the cost of temperature change, and finds that a one-degree increase in temperature has similar happiness implications to living to an area with $500 lower median income.

    The real reason I think this paper is important is that almost everyone will be exposed to more extreme temperatures – across the globe. If Patrick’s results are valid for other countries as well, what we will see is no benefits from fewer cold days and only costs from hotter days.

    This might be the economics paper that has quantified the most broadly applicable non-monetary costs of climate change. And these are not in the models. Yet. Get to it modelers.

    Cross-posted from the blog of the Energy Institute at Haas (tag line: Research that Informs Business and Social Policy).