Assessing the effects of climate change policy on the volatility of carbon prices in reference to the Great Recession

Abstract

This paper examines the effects of having or not having a climate change policy on the behaviour of carbon price volatility before, within and after the 2008/09 global recession. Our Markov-regime switching model analysis shows that the voluntary carbon market at the Chicago Climate Exchange was in a high-volatile regime within, and two years before, this recession. The mandatory carbon market in the European Climate Exchange was relatively in stable and low-volatile regime over these periods, except at the end of the recession. The voluntary market exhibited more price volatility features than the mandatory one. After the recession, both markets experienced high probabilities of being at low-volatile regimes. Our results suggest that high-volatile regimes were not caused by the recession per se. However, statistical tests show that there were distinct low-and high-volatile regimes during the recession period, indicating that the recession aggravated price volatility of both markets.

Keywords:

• volatility
• carbon markets
• Great Recession

Acknowledgements

I would like to thank Prof Ken Willis, the editor of this Journal, and the unanimous reviewers for providing constructive comments.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. Although the voluntary CCX's cap-and-trade operation closed its door in January 2011, a subsidiary called Carbon Financial Instruments was established for exchange trading in allowance for those who participate in the programme (Gronewold 2011).

2. Additional review of volatility studies related to carbon price can be found in Sousa, Aguiar-Conraria, and Soares (2014).

3. Even though this is a restrictive assumption, it is widely used in applied research works for a number of cases. For instance, Klaassen (2002) describes that two regimes are sufficient in the case of forecasting volatility. Hamilton (1994) provides more general treatment for N regimes Markov-chain processes.

4. The longer version of this paper includes estimation results from different families of GARCH models. Model selection criteria, such as Akaike Information Criterion and Bayesian Information Criterion, reveal the superior estimating power of Exponential GARCH with general error distribution.

5. Engel and Granger (1987) and Hamilton (1994) provide detailed specification about formulating the cointegration processes and estimating Vector Autoregressive and vector error correction model. In this paper, only the results from these formulations are presented.