In several world regions, climate change is predicted to negatively affect crop productivity. The recent statistical yield literature emphasizes the importance of flexibly accounting for the distribution of growing-season temperature to better represent the effects of warming on crop yields. We estimate a flexible statistical yield model using a long panel from France to investigate the impacts of temperature and precipitation changes on wheat and barley yields. Winter varieties appear sensitive to extreme cold after planting. All yields respond negatively to an increase in spring-summer temperatures and are a decreasing function of precipitation about historical precipitation levels. Crop yields are predicted to be negatively affected by climate change under a wide range of climate models and emissions scenarios. Under warming scenario RCP8.5 and holding growing areas and technology constant, our model ensemble predicts a 21.0% decline in winter wheat yield, a 17.3% decline in winter barley yield, and a 33.6% decline in spring barley yield by the end of the century. Uncertainty from climate projections dominates uncertainty from the statistical model. Finally, our model predicts that continuing technology trends would counterbalance most of the effects of climate change.

The panel data approach with fixed effects has emerged as the preferred method to uncover the effects of climate change on economically relevant outcomes using historical weather data. While the panel method has been criticized for its purported inability to account for long-run adaptation, it has been argued that including nonlinearities in explanatory weather variables makes crosssectional variation in climate enter coefficient identification, suggesting that the estimates obtained from a nonlinear, fixed-effects panel model at least partially reflect long-run adaptation. We formalize this argument in the context of the popular quadratic specification and show that (i) skewness in the historical weather data conditional on location is an essential driver of the bias in the panel estimates relative to the underlying long-run values, and can result in bias in either direction, (ii) in the absence of such skewness, the panel estimates are a convex combination of the short-run and long-run coefficients, and (iii) the panel estimates reflect the long-run values whenever the cross-sectional variation in climate "dominates" the location-specific weather fluctuations, in a sense that we make explicit. We use our framework to revisit impact estimates from nonlinear panel approaches published in the last decade. We find that for large countrywide or global panels, estimates of the effect of temperature primarily represent the long-run response, due to the large cross-sectional variation within these panels. In contrast, our calculations suggest that estimates of the effect of precipitation on outcomes reflect a more even combination of long-and short-run responses.

No abstract

Globally, over 400 million tons of biomass are burned in agricultural fires for management purposes each year, substantially affecting air quality (Korontzi et al.,

The results of ballot referenda have often been used to infer the preferences of voters for various types of policy. However, voters may be influenced by the appearance of competing ballot referenda. We propose a simple theoretical framework for considering the substitutability or complementarity of various land use policies and apply this framework to a novel dataset of 603 landrelated municipal ballot measures in California. We find that the appearance of a competing anti-growth measure decreases support for other anti-growth measures, while the appearance of pro-growth measures do not affect the likelihood of passage for anti-growth or pro-growth measures.