David bio photo

David
García-León

Marie Skłodowska-Curie Research Fellow (Ca' Foscari University and Euro-Mediterranean Center on Climate Change, Venice).


Curriculum Vitae
Email Google Scholar Research Gate Orcid Github Twitter RSS

In light of the projected increase in mean temperatures due to the action of anthropogenic global warming, it is crucial to know up to which extent the economic system is exposed to environmental variables and how their changing nature affect economic performance.

In particular, it is urgent to determine whether a relationship between weather and total income exists and quantify its sign and magnitude as well as assess whether projected increases in temperatures will undermine the ability of countries to grow.

In this study we explore the relationship between weather and economic activity in assorted European regions. To do so, we construct a novel, regional weather and economic data set spanning from years 1990 to 2012 covering the five largest countries in Europe (Germany, France, UK, Italy, and Spain). Looking at its cross-sectional dimension, we would be able to identify the long-term (level) effect of weather on income thanks to a Ricardian-style regression approach. Moving then to its longitudinal dimension, treating the whole data set as a panel, we benefit from the exogenous year-to-year variations in weather to estimate the short-term impact of weather fluctuations in yearly changes of activity, that is, growth.

Data

Finding and gathering a sufficiently detailed European-wide data set has been by far the most challenging task of this paper. Unlike the US, there does not exist in Europe an agency responsible for assembling weather data coming from weather stations from Member States. Indeed, some gridded meteorological datasets covering Europe, like WFDEI, are publicly available but we have opted in this study to use real weather station data and avoid many typical problems associated to gridded datasets, like he creation of a fictitious correlation between weather measures that could bias our conclusions.

Our strategy amounts to retrieve data from national statistical offices and national climate agencies in order to construct a wide-region database as longitudinally largest as possible.

Mean yearly temperatures (°C) [left] and precipitations (mm) [right] in NUTS 2 regions. Year 2000.


Overall, our dataset covers 168 NUTS2 regions during the period 1990-20121. Economic figures have been directly collected from national statistical offices. Their sectoral breakdown is also available, which will enable us to disentangle the main effects in relation to field of activity.


Hence, as far as we know, this work represents the first attempt to develop an integral exercise relating weather and economic variables carried out for the Eastern hemisphere.

Methods & Main Findings

Our methodology is primarily based on the work by Dell et al. (2009, 2012). We follow their spirit to develop an integral case study for Europe.

Long-term relationship

The theoretical background in which this section is embodied is the Ricardian (or hedonic) method applied to climatic variables that stems from the original work by Mendelsohn et al (1994). We estimate the cross-sectional relationship between climate variables|mean temperature and precipitation|,geographic variables and per capita income, i.e.,

where X represents a vector of specific geographic variables, such as elevation and distance to the sea. We estimate the above equation for the whole sample of NUTS regions using OLS. Standard errors are calculated clustering observations by larger NUTS level.


In general, we find qualitatively similar results to Dell et al.. Specifically, we distinguish a negative, significant, tempered effect of temperatures on income in our sample. More precisely, an additional degree is attached to a decrease of 1.6-2.2% of personal income. This negative response is amplified in poor regions.

Short-term relationship

By making use of the longitudinal dimension of our dataset in order to try comprehend the dynamic effects of weather variation in economic activity. Our main identification strategy uses year-to-year fluctuations in temperature and precipitation to identify changes in economic performance. We can then use panel data econometric techniques to inform whether temperature impacts regional growth rates or simply the level of income. Our standard distributed-lag approach

where region fixed effects and regional time fixed effects are taken into account. Also, errors are clustered simultaneously by region and region-year.


We find that an increase of average temperatures of 1°C today hampers the growth potential of regions almost by 0.06 pp, which in accumulated terms represents an overall effect in the long-run slightly larger than the one estimated in the previous section. As of poor regions, the net effect of a 1°C rise in temperature is to decrease growth rates in poor regions by 0.084 percentage points, where again poor regions are a bit more penalised than rich regions.

Conclusion

Take-away messages: We find that one additional degree is associated with a decline in the level of per capita GDP of 1.6% to 2.2%. In line with other authors we find that this effect is exacerbated in poor regions. By studying the short-term dynamics we demonstrate how global warming undermines the ability to grow of the latter. These results supply new evidence in favour of the negative effects of rising temperatures in certain areas of developed economies. Our results have been recently reinforced with the findings of Deryugina and Hsiang (2014) and Colacito et al. (2014) for the US economy.

In light of this, policy makers should account for regional heterogeneity when large-scale environmental policies are formulated. This heterogeneity should also be borne in mind when the interactions between climate and economy are modelled, for instance, in IAM models.

Future Avenues of Research: Since climate change is usually accompanied by extreme weather events, the existence of weather non-linear effects in our economies should be explored. This suggests a new avenue of research that will be covered in future projects.

  1. Data for some countries is only available from 1995.