The short-term impacts of COVID-19 outbreak on carbon emissions: causal evidence from China

ABSTRACT

This study precisely examines the short-term impacts of the COVID-19 outbreak on CO2 emissions through a sharp regression discontinuity design based on daily emissions data of 31 provinces in China. I find that the COVID-19 outbreak leads to an 18.4% CO2 emissions reduction in China. The total reduction effects are mainly attributed to the power, domestic aviation, ground transport and industry sectors. Additionally, the reduction effects across different provinces are almost uniform and it takes approximately 9 weeks for China to alleviate it. Furthermore, the positive spillover effects of CO2 emissions weakened after the COVID-19 outbreak due to the strict lockdown policies in China. Overall, these findings offer novel insights into the impacts of major public health emergencies on CO2 emissions and inform future CO2 emissions reduction goal settings in developing countries.

KEYWORDS:

• COVID-19 outbreak
• CO2 emissions
• causal inference
• China

JEL CLASSIFICATION:

• Q01
• Q15
• Q51

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ The website of the Carbon Monitor Dataset is https://cn.carbonmonitor.org, where the CO2 emission data used in this paper is available. The Carbon Monitor Dataset is established for near-real-time monitoring of global CO2 emissions. Accounting methods and detail of this dataset are transparent and available (Liu et al. 2020).

² Regression discontinuity design is a powerful causal inference method that leverages cut-off points to create a quasi-experimental setting and estimate robust local average treatment effects. The discontinuity can be estimated by both parametric and non-parametric methods. This paper emphasizes the results from the non-parametric method. As the dates of the COVID-19 outbreak are arguably exogenous, I exclude fixed effects and control variables in the simplest form. Specifically, referring to the practice of Greenstone et al. (2022), I first run an OLS regression where the dependent variable is the raw CO2 emissions and the independent are control variables and fixed effects, and then apply discontinuity non-parametric estimation for the residualized CO2 emissions.

³ Specifically, I divide the whole sample into two subsamples according to before the first province outbreaks and after the last province outbreaks, then I run the spatial Durbin model for each subsample.

⁴ RD estimates in this paper use the uniform kernel and first-order polynomial, and the bandwidths are selected by minimizing the mean squared error of regression functions on both sides of the cut-off.

⁵ The reduction effects estimated in this paper are comparable to the 17% decrease in daily global CO2 emissions observed by Le Quéré et al. (2020) in April 2020 compared to the 2019 average level.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China [71973094, 71833003 and 72173085]; National Social Science Fund of China [21&ZD077]; Shanghai Jiao Tong University Fund for Modern Financial Research [202204].