Why the EU ETS needs reforming: an empirical analysis of the impact on company investments

Abstract

The European Union's Emissions Trading Scheme (EU ETS) is so far the largest emissions trading system in the world. A rigorous ex post empirical analysis of the scheme is presented. The effect of the scheme on firms' investment decisions in carbon-reducing technologies is analysed by using detailed firm-level data from Swedish industry. Based on difference-in-difference estimation as well as a before–after difference estimation, the results reveal that the EU ETS has not had a significant effect on firms’ decisions to invest in carbon-mitigating technologies. However, although the EU ETS appears to have no direct effect on investments, it is too early to dismiss the system. Consideration is given to how the EU ETS can realize its potential to become an effective tool in the EU climate and energy policy portfolio.

Policy relevance

A thorough analysis and discussion considers the ability of the EU ETS to create strong incentives for investment in carbon-reducing measures. The empirical results (using detailed firm-level data from Swedish industry) add to earlier findings in the literature showing the limitations of the EU ETS to influence investments and innovation. This is a critical and pressing issue for policy makers. With even modest reforms such as the back-loading of allowances meeting strong resistance from some Member States, the future of the EU ETS is rightly put in question. A key question is whether the EU ETS can and should be reformed in a way so that it can have a real impact on investments, or whether other policy instruments should take an increasing role for long-term transformation of the energy system.

Keywords:

• carbon
• clean technology
• climate change
• EU ETS
• firm behaviour
• investment
• technological adoption

Funding

Financial support from the Mistra Foundation through the research program Mistra Indigo (Instrument Design for Global Climate Mitigation) is gratefully acknowledged. The article has also benefited from valuable comments from Andreea Mitrut and three anonymous reviewers.

Notes

1. Phase I, 2005–2007; Phase II, 2008–2012.

2. The present authors follow the EU ETS Directive (2003) for deciding which sectors to include in the ETS. According to the directive, the trading scheme covers all refineries and coke ovens, production and processing of ferrous metals, the mineral industry, the pulp and paper industry, and all energy combustion units with a rated thermal input exceeding 20 MW. Hence, the sectors in the analysis defined as EU ETS sectors are the pulp and paper industry; refinery and coke oven industry; manufacture of mineral products; manufacture of basic metals; and energy and heating.

3. The five control sectors are manufacture of wood and wood products; manufacture of rubber and plastic products; manufacture of machinery and equipment n.e.c.; manufacture of fabricated metal products, except machinery and equipment; and manufacture of motor vehicles, trailers, and semi-trailers.

4. The percentages are rounded.

5. It should be emphasized that the data provided by Statistics Sweden do not explicitly display carbon abatement technology, but this is something that has been derived based on the written descriptions provided by firms. Although the detailed data comprise one of the strengths of this article, it should be noted that even though we have aimed to include only carbon-mitigating investments (using best practice), there is a potential risk that some investments not aimed at carbon mitigation were also included. In particular, this could be the case for investments defined as energy efficiency measures; e.g. we cannot distinguish between energy efficiency measures aimed at reducing fossil fuel use and measures reducing biomass use in the production process. The former should be defined as a carbon-mitigating investment and the latter should not. Hence, there could be a potential overestimation of carbon-mitigating investments in our data, but because this effect should occur both in the sample of EU ETS firms as well as in the sample of control firms, we do not expect any qualitative effect on the results.

6. The threshold of €1 million is based on a thorough descriptive analysis of the data, including both the distribution of the investments and an analysis of investment patterns over the year. Several limits (€500,000 as well as €100,000) were analysed, but, based on the descriptive analysis, we concluded that the limit should be set at €1 million to capture differences in investment behaviour related to the size of the investment.

7. Note that, if a large investment was made, this observation is dropped from the small investment sample. Hence, the dependent variable for the large investment is a binary variable equal to one if an investment larger than or equal to €1 million was made and zero otherwise. For small investments the binary variable is equal to one if an investment less than €1 million was made, zero otherwise, and if a large investment was made the observation is dropped.

8. Note that a firm can make both small investments and large investments and is then classified both as a small investor and a large investor.

9. Accounting for the CO₂ tax exemptions for the industry sector. However, district heating is not exempted from the CO₂ tax and, based on the energy situation 2007 in Sweden where heating consumption accounted for 27% of total heating consumption and electricity consumption, it is assumed that the energy sector, on average, pay 27% of full CO₂ tax.

10. The reason for the price drop was, according to Alberola et al. (2007), the announcement that verified emissions were below allocated allowances. Banking of the allowances between phase I and phase II was not allowed, which was also a major contributor to the price decline.

11. Statistics Sweden does not release data that can allow the identification of individual firms.

12. The actual windfall profits also depend on a firm's ability to pass on the cost of allowances in its product prices. Also, using the average allocation hides the variation that exists among firms.

13. 2002, €1 = SEK9.16; 2003, €1= SEK9.13; 2004, €1 = SEK9.13; 2005, €1= SEK9.28; 2006, €1= SEK9.25; 2007, €1 = SEK9.25; 2008, €1 = SEK9.61; http://www.scb.se/Pages/TableAndChart____32240.aspx

14. Because expenses on R&D are missing for non-ETS firms for year 2001, R&D is not lagged for year 2002 for the non-ETS sectors.

15. The yearly average oil price is collected from the US Energy Information Administration (EIA) and converted from nominal US dollars per barrel to real € per barrel, using average annual exchange rate and average annual US inflation. We also tested with the natural gas price instead of the world market oil price. These two prices follow each other almost one-to-one (correlation of 0.94).

16. That is, the difference between: the difference of the investment decisions of the ETS sectors and non-ETS sectors after the implementation of the scheme and the difference of the investment decisions of the ETS sectors and non-ETS sectors before the implementation of the scheme. This can be written as

17. The results are available from the authors upon request. The procedure GLLAMM (Generalized Linear Latent And Mixed Models) in the statistics software STATA/SE 12.0 was used for the multilevel estimations.

18. ‘If, for more than six consecutive months, the allowance price is more than three times the average price of allowances during the two preceding years on the European carbon market, the Commission shall immediately convene a meeting of the Committee established by Article 9 of Decision No 280/2004/EC’; http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2009:140:0063:0087:EN:PDF