We use an iterative approach. An initial institution is chosen and the results from one to a few sessions are examined. We use our learning from the experimental results to inform changes that will fine tune and improve the auction rules for the next set of experimental sessions. We tested three mechanisms (institutions) : proportional (fixed price, firms supply requests for capacity), clock (price and capacity determined by clock) and hybrid (Clock pricing with proportional allocation of capacity). For the Y configuration, bidding could be the pipeline paths AB and AC, or bidding could be on the pipeline segments A, B, and C.
The proportional mechanism is a multi-round sealed bid auction. In the proportional mechanism, each participant is awarded a share of capacity equal to his share of the total bid amount if and only if his bid is accepted. Each period (or auction) consists of three rounds, each subject \(i\) submits a single capacity request \(q_i\) for either the small or large pipeline. A bid for the small pipeline can be allocated to either the large or small pipeline. A bid for the large pipeline can only be allocated to the large pipeline. The allocation price for the large pipeline is e$30 per unit of capacity allocated. The allocation price (unit construction cost) for the small pipeline is e$40 per unit of capacity allocated. The large pipeline is built if \(\sum _i q_i \ge 40\), the small pipeline is built if \(\sum _{i(small)} q_i \ge 20\). If both constraints are met, then the large pipeline is allocated, i.e., built.
If the large pipeline is chosen to be built and \(s_L = \sum _i q_i > 40\), then each bidder \(i\) is awarded \(s_i = q_i/s_L\), a proportional share of the pipeline capacity, and each bidder \(i\) pays \(s_i * 30\). If the small pipeline is chosen to be built and \(s_S = \sum _i q_i > 20\), then each bidder \(i\) is awarded \(s_i = q_i/s_S\), a proportional share of the pipeline capacity, and each bidder \(i\) pays \(s_i * 40\).
We tested three possible update rules : bids on the capacity quantity can only decrease, bids on the capacity quantity can only increase and no constraint on change of the capacity quantity. All three rules were applied in each round after the first,
Brief description of clock here (or is that above?).
There are a number of design questions that need to be answered when implementing a clock auction. One important question asks “what is the rule for stopping the clock on a specific item, and what are the conditions, if any, for restarting the clock on an item?”
Another asks “what information is supplied to the bidders after a click of the clock?” Do we allow a situation where the sum of bids is greater than the cost, but there are individual bids that are below the average (or incremental) cost. Do we always attempt to maximize revealed surplus, or is the larger pipeline always preferred to a smaller pipeline if cost is covered.
One of our clock auction variations is as follows. For the clock mechanism a price is initially set for each of the segments or paths. A single price for the SP configuration, and two or three prices for the YP configuration. Subjects submit a quantity (\(q\)) for each offered segment or Opt Out. If a subject Opts Out, their last bid is retained for the allocation, and they are not allowed to change or resubmit their bid. After all subjects have either Opted Out or submitted a bid the auction either ends or the price of the offered segments is increased. The price for a segment is increased only if demand for the segment changes.
The auction ends when all the subjects Opt Out or when there is no change in the capacity demanded (this is the excess demand rule). The clock stops on an individual segment when demand is greater than capacity for the large pipeline. Design A starts with a price less than e$30. Design Y starts with prices less than e$15. The final allocation is based on subjects Opt Out bid and determined by the pipeline that maximizes surplus. The allocated price is the lowest accepted price; which could be below where clock stopped. If the Hybrid Rule is used, then capacity is determined by proportional allocation.
The experiments have been grouped into nine sets by value sets, mechanisms and rule sets. The list is displayed in Table 1.