Wednesday, July 20, 2011

Anticipating Future Hot Summers and Getting Ready to Adapt

Heidi Cullen's OP-ED in today's NY Times serves a useful purpose in reminding us that this summer's heat is likely to be a typical outcome in the near future.   She does cross the line when she states; "In the United States, in any given year, routine weather events like a hot day or a heavy downpour can cost the economy as much as $485 billion in crop losses, construction delays and travel disruptions, a recent study by the National Center for Atmospheric Research found. In other words, that extra 1.5 degrees might be more than we can afford."

Taking this fact to be "true" today does not mean that it will be true in the future.  She appears to take the NCAR's study as a law of physics.  Using Google, I found  the technical paper that she is citing.  To quote these nerds; "To estimate the sensitivity of the U.S. economy to weather variability, we used non-linear regression analysis to model the relationships between sectoral GSP and economic inputs of
capital, labor, and energy and a set of weather indicators."   As reported in their Table 1, they find that agriculture's value of output is lower when cooling degree days are higher (i.e it is hot outside) and when rainfall is more volatile. Climate change will increase the likelihood of both of these scenarios.

Cullen actually misquotes the authors. Here is their quote;

"Holding technology and economic inputs constant (i.e., setting them at
their 1996–2000 averages), we then used parameter estimates from these 11 empirical models
with 70 years of historical weather data to identify states more sensitive to weather impacts and
rank the sectors by their degree of sensitivity to weather variability. We calculated the aggregate
dollar amount of variation in U.S. economic activity associated with weather variability could
vary by 3.4% or $485 billion a year of 2008 gross domestic product."

Now, if you bother to read their paper and look at their Table 1, you will see some strange results. They claim that manufacturing output is negatively related to the volatility of rainfall. I don't believe this.  They also find a big negative effect for finance and utilities! This looks weird and merits an explanation.  I agree with one writer that hydro electric utilities will be affected by rainfall variation but hydro is a relatively small share (less than 10%) of U.S power generation.  Also note that in their Table 1, that the construction industry's output increases with summer heat (CDD) and rainfall volatility. Does that make sense?

The authors report their facts; "A primary finding of this study is that every sector is statistically significantly sensitive to at least one measure of weather variability, and two sectors—FIRE and wholesale trade—show
sensitivity to all four measures of weather variability." but they are a little bit lazy as they don't provide any explanation for whether this finding is credible or not.

Their claims that weather variability affect the U.S economy over the 70 years from 1930 to 2000 differ from the NBER results of Dell, Jones and Olken available here.

But, let's assume that the $485 billion loss that Cullen reports in the past is correct.  The whole point of my Climatopolis project is that past loss estimates vastly overstate the impact of future costs because of adaptation.  Thanks to people such as Heidi Cullen, we know that we face the challenge of climate change and that this summer is no fluke. Forward looking people will make a host of investment choices across many margins including where they live, how they live and products they buy.  Farmers will make a host of new choices that collectively shield us from the extent of damage that Cullen cites.

For more details about my thinking read this.

UPDATE:  I'm receiving some angry email about this post. I am a fan of Ms. Cullen but her quote is incorrect. If you read the direct quote I reproduce above she appears to say that the economy is rolling along and then weather shocks occur and this causes a drop of $485 billion in output.  Keep in mind that our economy's total value is about 14 trillion (corrected!).  But, the authors of the study actually say that when you compare GNP at the "best" weather (for maximizing output for each sector) versus the "worst" weather that this yields an estimate of $485 billion.   She is thus running a strange experiment of taking us from great weather to the worst weather rather than evaluating what happens when we have average weather and then overnight the worst weather (for maximizing GNP) emerges.

To all the critics out there, please read the research team's Table 1. You will be amazed by some of the strange coefficient estimates they report.  The impact of climate on agricultural output is smaller than the impact of climate on electric utility output!  Construction's output rises when weather is hot and rain volatile.  Could this be correct?   Please explain to me what the causal mechanism could be.