Relationship between CO2 emissions and energy consumption sub-types under impact of AI-related patents and energy-related R&D investments: Evidence from the USA by novel quantile-based methods

Abstract

The importance of AI and R&D investments has become increasingly salient in the context of rising carbon dioxide (CO2) emissions. So, this study examines how CO2 emissions relate to energy consumption (EC) sub-types and whether AI-related patents (AIP) and energy-related R&D investments (ERD) moderate the relationship. In this vein, the study focuses on the USA, uses EC sub-types as explanatory variables, considers the moderating role of AIP and ERD, and applies novel quantile-based methods on data from 1981/Q2 to 2020/Q4. The results indicate that (i) oil and coal EC are associated with higher CO2 emissions across quantiles in both bivariate and multivariate models; (ii) while gas EC increases CO2 emissions across all quantiles in bivariate and multivariate cases, there is a decreasing impact at lower quantiles with ERD moderation; (iii) nuclear EC increases CO2 emissions across all quantiles in bivariate case, whereas the impact changes under the moderating impacts of AIP and ERD; (iv) renewable EC decreases CO2 emissions across all quantiles in bivariate case, while the reducing impact is almost same under the moderating impacts of AIP and ERD; (v) AIP has a much stronger moderating impact than ERD on relationship between CO2 emissions and EC sub-types; (vi) there are generally causal impacts across quantiles, except for some lower, middle, and higher ones, where the causal impact varies across the variables pairs. Accordingly, the study outlines policy options consistent with the distributional patterns observed.