The department has released a 2010 study that experimentally tests different design features of primary auctions to allocate Australian Emissions Units (AEU). The findings of the study remain relevant to the consideration of auction design for the carbon pricing mechanism.
The department has released a position paper on the detailed auction design that will inform the development of the legislative instrument for auctioning carbon units.
Executive summary
This study experimentally tests different design features of primary auctions to allocate Australian Emissions Units (AEU) to the entities covered by the proposed emissions trading scheme. On the basis of our experimental results and the theoretical and empirical academic literature on permit auction design, we recommend an appropriate auction format.
The key objectives of the Government in relation to the permit allocation auctions are to promote allocative efficiency and efficient price discovery, as well as to raise auction revenue, with the former objectives given priority over the latter. In the proposed Australian scheme – called Carbon Pollution Reduction Scheme (CPRS) – each permit will have a date stamp (vintage), indicating the year in which it becomes applicable. Those permits can be banked (used later than date stamp) and to a limited extent also borrowed (used earlier than date stamp). It is planned to auction both current and future vintages.
Using experimental methods, this study addresses the following main questions:
- Should the auction follow a sealed-bid approach or should it be implemented as an ascending clock auction with proxy-bidding?
- If a clock auction is used, should the aggregate demand be revealed during the auction?
- Should multiple vintages be auctioned off simultaneously or sequentially?
To answer these questions, we developed an experimental design consisting of two dimensions: auction format (sealed bid, clock auction without revelation of aggregate demand, clock auction with revelation of aggregate demand) and market environment (1 vintage, 2 vintages auctioned sequentially, 2 vintages auctioned simultaneously, 2 vintages auctioned simultaneously with subsequent secondary market). In a 3x4 factorial design we tested all cells in those two dimensions, resulting in a total of 12 experimental treatments. An additional treatment tested the robustness of the eventually recommended design (see below) with a larger number of traders.
From January to March 2010 we conducted experiments at the University of New South Wales (UNSW) and at Karlsruhe University (KU). Our analysis relies on 1,120 participants in 75 experiment sessions, participating in a total of 447 auctions.
With respect to Questions 1 and 2, all three auction formats perform rather similarly in our experiments with respect to efficiency, auction revenues, and bidder profits. This is true in all single-vintage and two-vintage market environments tested, and it is consistent with the existing experimental literature. However, there is some indication that an open clock format yields a lower volatility of final auction prices than a closed clock or a sealed bid format.
With respect to Question 3, we find no support for the conjecture that a simultaneous auction of two vintages yields higher efficiency than auctioning them off sequentially. To the contrary, we obtain evidence that a sequential auction yields more efficient allocations, higher auction revenues, and better price signals.
Given the lack of evidence of superiority of one of the auction formats with respect to efficiency, other characteristics of an auction such as price volatility or simplicity take a more prominent role in applied auction design. Based on a prudential weighing of theoretical considerations, empirical evidence on previous permit auctions and our experimental findings we make the following recommendations for the CPRS auction design:
The CPRS auctions should:
- Apply the format of a clock auction with proxy-bidding, and
- Reveal aggregate demand after each round, and
- Be conducted sequentially if more than one vintage is auctioned at an auction event, with the earliest vintage auctioned first.
In the longer term, once liquid secondary markets have evolved, we recommend that switching to a sealed bid format be considered. We also recommend reviewing the performance of sequential auctioning of multiple vintages and considering a change to a simultaneous auction if necessary. We suggest that an appropriate time for such a review may be after the end of the year of the first vintage covered by the CPRS.