Heterogeneity in carbon intensity patterns: A subsampling approach

Abstract

Carbon intensity, defined as carbon dioxide (CO2) emissions per unit of gross domestic product (GDP), is a critical metric for assessing the effectiveness of climate policy across nations. This paper presents an analysis of the persistence and stationarity of carbon intensity data across 176 countries and 44 regions from 1990 to 2014, employing subsampling confidence intervals. Subsampling is a robust statistical technique that performs well with finite samples and requires minimal assumptions about the data. Our findings categorize countries into three distinct groups based on their carbon intensity patterns: convergent, persistent, and divergent. We observe that climate mitigation policies in countries with a convergent pattern tend to have only temporary effectiveness, whereas in countries with a divergent pattern, such policies can lead to permanent changes. Additionally, using unsupervised learning methods, we delve into the underlying factors influencing these classifications. This study is particularly significant for understanding the long-term impacts of climate policies, offering valuable insights for policymakers and international bodies. By identifying and analyzing these distinct patterns, our research contributes to the strategic planning and implementation of more effective and sustainable climate policies globally, aligning with the goals of international agreements like the Paris Accord.