According to the Intergovernmental Panel on Climate Change (IPCC), the study of climate change, and particularly global warming (GW), involves a careful analysis of the following four sequential issues or questions: (i) What type of GW exists? (ii) causes of GW (is GW caused by human activities?); (iii) economic effects of GW; and (iv) economic policies to mitigate these effects. Obviously, to determine the type of GW is crucial for the next issues in the chain. The purpose of this paper is to offer a complete answer to the first question by analyzing all the characteristics of the temperature distribution and not only the average (think on the time evolution of the income distribution). We investigate these characteristics by introducing a novel methodology that converts them into time series objects and therefore we can apply all the existing tools of Time Series Econometrics.
By doing that, we can easily analyze the trend behavior of different quantiles and different dispersion measures of the temperature distribution and not only the evolution of the mean, as it is done in most of the standard literature. According to our definition of Global Warming (GW is defined as the existence of an increasing trend in some of the characteristics measuring the central tendency or position (quantiles) of the global temperature distribution) this trend analysis is crucial in order to determine the existence and type of GW. The paper proposes a simple and robust test to determine the existence of a trend in a given temperature distributional characteristic Ct (quantiles, dispersion measures, mean, etc.): Regress Ct on a linear trend and test if the slope is equal zero. If this hypothesis is rejected then there is a trend (there is warming) although we will not know exactly what type of trend (linear, quadratic, logarithm, etc.)
In order to show the generality of our results, we implement two applications: one with N time series observations for each year t, and another with N cross-sectional observations, also for each year t. In both of them, any temperature distributional characteristic can be easily estimated year by year. The first application studies the trend behavior of the distributional characteristics of temperature in Central England from January 1, 1772, to October 31, 2017. In the second application, we analyze global temperatures across different stations in the Northern and Southern Hemispheres for the period 1880–2015. The two applications lead to similar trend results, which can be summarized as follows. First, there exists a trend in most of the characteristics considered (existence of warming). The trend in the lower quantiles is stronger than those in the mean and upper quantiles of the temperature distribution (climate heterogeneity). Second, dispersion measures such as the interquartile range (iqr), standard deviation (std), and range (max–min) show a negative trend. Therefore, we conclude that GW (and Local Warming) is not only a phenomenon of an increase in the average temperature, but also of a larger increase in lower temperatures, leading to decreased dispersion. Ignoring these could have serious consequences for climate analyses. For instance, an acceleration in global ice melting causes an increase in the sea level which in turn increases coastal erosion and elevates storm surge as warming ocean temperatures create more frequent and intense coastal storms. This climate risk has an effect on the (iii) issue by increasing the economic risk.
The existence of climate heterogeneity (lower and upper quantiles correspond to different regions) indicates that the mitigation policies should consist of a common component for the whole Globe and an idiosyncratic element focused on the given region of interest.
Based on our results, this paper ends by suggesting that future international climate agreements should consider recommendations for the whole temperature distribution and not only for the average as the present agreements do.
The paper, titled “Trends in Distributional Characteristics: Existence of Global Warming“, is published in the Journal of Econometrics.