A final report summarizing research on auction design for RGGI was submitted to the New York State Energy Research and Development Authority (NYSERDA) in November 2007. Subsequently several constructive comments and suggestions have been made by interested parties. This addendum to the November 2007 report contains reactions to some of these comments, including some new findings based on research done subsequent to submission of the Final Report.

The main issues addressed in this addendum are (1) the choice of auction design (clock versus sealed bid) and how this affects the efficiency of the auction and the ability of parties to collude, (2) variations on design for clock auctions, and (3) auctions that combine different vintages of allowances. To summarize the main results:

Addendum to the October 2007 Auction Design report
April 2008

• In laboratory auctions with communication among participants, successful collusion is more effective in clock auctions than in discriminatory and uniform price auctions.
• An analysis of the 'chat' (instant message communications between bidders prior to submitting bids) indicates that clock auctions facilitate collusion by allowing bidders to focus on a single dimension (quantity reductions).
• The effects of this collusion are reflected in clock prices at or near reserve price levels, with subsequent trading at much higher prices in the spot markets.

• Results of our new experiments, conducted subsequent to the Final Report, indicate that the provision of information about the quantity of demand after each round of a clock auction does not improve price discovery of these auctions.
• This type of demand reduction in clock auctions echoes the striking results reported by Goeree, Offerman, and Sloof (2006) in a much simpler environment with full information provided about quantities demanded during the auction.
• The New England and New York ISO proposal that allowance owners be able to offer allowances for sale in the RGGI auction has definite advantages. We have different suggestions about how this might be implemented. The uncertainty of supply that results can help reduce the potential of collusion.

• Since RGGI allowances are bankable, a bid for a later vintage could be treated as a request to purchase either a later vintage or an earlier vintage, whichever is less expensive. Interpreting bids in this manner prevents a price inversion in which the uniform price for the later vintage is higher than the price for the earlier vintage, although theory suggests this price inversion is inefficient and would not occur in the secondary market. This addendum describes a simple procedure for combined vintage auctions that implements this idea.