Coase and Cap-and-Trade: Evidence on the Independence Property from the European Carbon Market

By Aleksandar Zaklan

Cap-and-trade has become a popular approach to mitigating environmental externalities. It promises cost-effectiveness: by combining a cap with the trading mechanism abatement takes place where it is most economical, minimizing aggregate costs for a given amount of abatement. Based on a seminal insight by Coase, under certain conditions agents' equilibrium solutions for the amount of the externality produced are independent from the allocation of allowances. Changes in allocations are limited to having distributional effects. Dependence between allocations and emissions would be a symptom of underlying friction in the allowance market.

The independence property is important in real-world settings, as policy makers often use free allocation in political bargaining surrounding cap-and-trade schemes. For example, in the EU Emissions Trading System (EU ETS) most of the cap was freely allocated during its initial phase, and close to half of the cap is still distributed for free today to mitigate carbon leakage risk. A cap-and-trade scheme in which emissions are independent from allocations is attractive for policy makers, as they can focus on whether free allocation is desirable from an equity perspective. Empirical evidence on whether the independence property holds in operational cap-and-trade schemes can therefore shed light on whether wide-spread use of free allocation should be avoided. Yet despite its importance the existing empirical evidence is extremely scarce.

The paper by BCCP Fellow Aleksandar Zaklan contributes to filling an important gap in the literature by testing whether emissions are independent from allowance allocations for firms regulated under the world's largest multilateral cap-and-trade scheme, the EU ETS. The analysis focuses on power producers, responsible for the largest share of the EU ETS's greenhouse gas emissions and arguably the most sophisticated sector under the EU ETS with respect to the allowance trade. However, not all power producers may be able to participate in the allowance trade equally. Financially or organizationally constrained emitters, likely smaller firms, may be less able to internalize trading costs or overcome behavioral biases affecting their emission decisions. At least some producers may exhibit dependence between allocations and emissions, which may affect the independence property at the sector level.  

The regression analysis pursues a difference-in-differences strategy using a panel of power producers from coal and gas during the period 2009 to 2017. The identification strategy exploits a change in allocation policy: in the majority of EU member states power producers lost a large share of their freely allocated allowances in 2013 and must purchase them through primary auctions or on the secondary market. However, while power generators generally faced full auctioning starting in 2013, a special provision - the so-called 10c rule - preserved a significant amount of free allocation for producers in eight EU member states, providing substantial exogenous variation in allocation levels. The analysis thus compares emission decisions of producers who lost most of their free allocation (treatment group) with those who continued receiving free allocation under the 10c rule (control group).

The main results show that emission decisions by power producers under the EU ETS are in line with the Coasean prediction. The independence property cannot be rejected for the power sector as a whole and for larger emitters representing by far the largest share of power sector emissions. For these firms a change in the level of free allocation does not cause a significant change in emissions. Emitters respond to the post-treatment allowance shortfall by increasing allowance purchases. However, there is suggestive evidence that the independence property fails for small emitters. An explanation could be trading costs affecting small emitters' emission decisions or, alternatively, resource constraints leading to poorer decisions by small emitters, in line with previous research. Behavioral bias may thus link abatement effort to the level of free allocation. However, as small emitters represent a very small share of the power sector's total emissions, distorted emission decisions among this group of firms do not significantly affect the independence property at the sector level.

The results suggest that policy makers may use free allocation as a tool in the political bargaining process without significantly distorting the cap-and-trade program, as long as the setting is comparable to the one in this analysis. Of course, how the cap is allocated has important distributional implications. The finding for small emitters motivates further investigation of the mechanisms behind this result, to understand whether policy makers should focus on the cost side or on attempting to reduce behavioral bias.

The full paper "Coase and Cap-and-Trade: Evidence on the Independence Property from the European Carbon Market" is published in American Economic Journal: Economic Policy.

This text is jointly published by BCCP News and BSE Insights.