Current Research

Massetti, E., and R. Mendelsohn. “The Effect of Extreme Temperatures and Adaptation across Eastern American Farms” Revise and Resubmit.

Previous Ricardian models of climate change impacts on agriculture have been criticized because they rely on mean temperatures and do not explicitly include extreme temperatures. This paper compares results using the entire distribution of daily temperature versus just the mean seasonal or growing season temperature in a Ricardian model. Including all temperatures does not increase measured long run damage. The warmest and coldest temperatures cause only modest harm to farmland values suggesting farmers have adapted to these extremes. The paper shows a few examples of these adaptations by farmers.


Massetti, E. and R. Mendelsohn. "The Elusive Effect of Temperature on Crops." In preparation.

We find that the duration over which temperature is measured affects the shape of the yield function. With hourly bins the yield function resembles an inverted hockey stick. With daily and monthly temperature bins, the yield function is hill-shaped and fairly symmetrical. The three specifications give statistically non-distinguishable, but very different representations of how temperature affects yields. Duration of temperature measurement plays instead no role in predicting the effect of uniform warming on yields, which is roughly linear in temperature. The literature has focused on forecasting accuracy instead of functional form specification and structural identification. While the predicted impact of warming may be robust, it is questionable that the effect of temperature on yields is highly non-linear.


Massetti, E. “Learning from Big Climate Data.” In preparation. Slides.

We apply “statistical learning” methods to estimate how climate and weather affect US agriculture using the largest dataset of climate variables ever used in the literature. The goal of this paper is to provide a more precise estimation of the relationship between agricultural productivity and climate. This will lead to a clearer understanding of the potential impacts of climate change on US agriculture. The paper will also tests several statistical learning methods in the literature that studies the effect of climate on the economy. The analysis can be replicated in other sectors and in other countries, provided a rich dataset of climate variable is available.


Massetti, E. "Can the Long-Difference Method Reveal Adaptation to Climate Change?"

Burke and Emerick (2016) find that farmers in U.S. counties that have experienced a positive warming trend over 1982-2002 have not reduced the vulnerability of crops to extreme temperatures. The authors suggests that this is evidence of the limited role that adaptation may play against future climate change. However, a careful analysis of climate data reveals that the observed trends could not be anticipated by farmers because they were driven by short-term, mostly mean-reverting, random weather variations. Not much can be learned about adaptation to climate change from Burke and Emerick’s analysis.


Massetti, E. Chaos in Climate Change Impact Estimates.

GCMs embed chaotic weather dynamics. In chaos theory this is known as the butterfly effect: very small perturbations of initial conditions trigger very large weather deviations from equilibrium. We use a unique dataset with 33 ensemble members of a large GCM to study how chaotic dynamics affect climate change impact estimates. The range between the ensemble members is striking and is currently ignored in the climate change impacts literature. The literature uses one random draw out of an unknown distribution of scenarios to project local climate change impacts. Some studies use scenarios from more than one GCM, but the this does not change the fact that each GCM run has unknown probability and averaging across scenarios from different GCMs does nothing to address the problem of internal variability.