Experimental Market Design and Methodology: The Alaska Gas Pipeline

1 Introduction

The problem we investigate is the construction of a gas pipeline that is to be shared initially by several incumbent firms. The current method of allocation is based on an adversarial regulatory process in which the price of access to any user is derived from historical costs. Our goal is to use experimental test bedding to explore more flexible solutions to the problem of access that is responsive to changing economic and technological conditions. The abstract methodology of science does not apply to engineering design problems of this kind.

The experimental design as we envision it is partitioned into several stages corresponding roughly to the sequential horizon for construction, production, and resource change. We will allow firms, both incumbents and new entrants to adjust their use of the pipeline to alterations in the mix of depletion, new exploration and discovery over the economic life of the pipeline and their supporting facilities.

Stage one of the design determines construction pipeline capacity, financing and the percent of capacity to be owned by each co-tenant. The construction decision consists of both the size and the path of the proposed pipeline. Total capacity, financing and their allocation among the co-tenants are interdependent and need to be determined simultaneously based on individual firm willingness to pay. There are several possible institutions for this determination, such as serial cost sharing and proportional cost sharing, each having various means of implementation.

In an application, expected future growth in demand combines with scale economies to infer that the efficient initial pipeline is constructed with excess capacity. Not all that capacity needs to be implemented with a corresponding level of pumping capacity. Such economic considerations lead to the concept of initial pipeline reserve capacity, to be implemented when needed at a subsequent time by adding pumping capacity. Hence, incumbents, depending on their initial holdings of gas reserves, and plans for further exploration, are motivated to bid for either (1) initial pipeline and pumping capacity, or (2) reserve capacity, or for both. We illustrate these various concepts in Figure 1. We accommodate these elements in our discussion of the different stages in the development plan, but the experiments we report are more modest, consistent with our step-by-step learning and adjustment methodology.

Stage two concerns exploration for new gas and the exchange of capacity rights at auction, if needed, between those with depleting reserves and those who find new gas.

Stage three determines whether there is a need to expand pipeline operating (pumping) capacity, and who pays for and is awarded the new capacity shares.

We make no separate provision for the construction of gas processing facility capacity for conditioning the gas for shipment. To reduce the complexity of the initial series of experiments and allow learning to proceed in step-by-step fashion, we will subsume processing into pipeline operating capacity by implicitly assuming that processing throughput capacity is proportional to pipeline throughput pumping capacity, and that pumping capacity shares are the same as processing shares. But the principles articulated below can be extended naturally to a design in which there is a separate determination of the sharing of rights in facilities and the exchange of those rights as needed. A full such treatment would require investigation by a team that includes technical and managerial industry practitioners.

Hence, this study is best described as a proof-of-concept, pre-implementation investigation. An example can be found in rassentiSmithCcombinatorial1982 on markets for airport access rights; examples of actual implementations include emissions trading (CA SOx and NOx reference; VA auction of NOX; Sears logistics auctions; for a description of electric power design implementations (rassenti2002using)).

The experiments we report here will describe the first stage of a study of the use of alternative auction procedures for awarding capacity rights to a pipeline joint venture in accordance with individual bidders’ willingness to pay which taken together determines the initial fixed investment cost. In this first stage, we compare and contrast the performance of a traditional sealed bid auction with a combinatorial clock auction in which an ascending price clock quotes the bid price on each round and bidders simply indicate when they are no longer active as the clock ticks up. The auction results determine capacity on each leg of a pipeline that allows for a telescoped size on different legs.

Later stages of the research deal with auctions (1) to transfer capacity among incumbent holders of initial rights or to new entrants willing to pay for it; (2) for pumping capacity expansion; and (3) for complementary exploration of lease rights to new gas. The latter auctions permit rights to explore and rights of pipeline access to be transferred in package combinations depending on the bids tendered. These later stages, however, must be part of the planning horizon and generally consistent with the first stage.

Combinatorial auctions have been successfully applied in industry to auction logistic services, and emission permits. Recently the methods have been extended to use clocks to greatly simplify combinatorial bidding. Although the advantages of using combinatorial clock procedures to action items that reflect complex properties of substitution and complementarities in use value have been examined, we will intentionally begin with other more traditional auction formats for the gas pipeline problem and make step-wise comparisons as we proceed. Hence, we do not—nor do we expect it of our readers—take for granted any claimed benefits of combinatorial auction methods in this application. We build on principles and learning from other contexts while exploring this new context.