Strategic bidding in multi‐unit auctions with capacity constrained bidders: the New York capacity market

• Published date: 01 October 2015
• Author: Sebastian Schwenen
• Date: 01 October 2015
• DOI: http://doi.org/10.1111/1756-2171.12104

RAND Journal of Economics

Vol.46, No. 4, Winter 2015

pp. 730–750

Strategic bidding in multi-unit auctions with

capacity constrained bidders: the New York

capacity market

Sebastian Schwenen∗

This article employs a simple model to describe bidding behavior in multi-unit uniform price

procurement auctions when firms are capacity constrained. Using data from the New York City

procurementauctions for power generating capacity,I find that firms use simple bidding strategies

to coordinate on an equilibrium that extracts high rents for all bidders. I show theoretically and

empirically that the largest bidder submits the auction clearing bid. All other bidders submit

inframarginal bids that are low enough to not be profitably undercut. Inframarginal bidders

decrease their bids as the pivotal firm’s capacities and its profits of undercutting increase.

1. Introduction

The volume of goods traded through auctions in the economy has been drasticallyincreasing

overthe last decades. This increased use of auctions raises the need to better understand and predict

economic behavior in bid-based selling mechanisms. To address this challenge, an increasing

strand of literature tests and expands existing auction models. As electricity is a completely

homogeneous good and produced by a small number of firms, restructured power markets have

become a major field of applied auction analysis.

Multi-unit auctions are the main auction format used in electricity markets. This article

contributes to the literature by demonstrating that a simple model of multi-unit uniform price

auctions is consistent with observed bidding behavior from capacity auctions in electricity mar-

kets. von der Fehr,and Harbord (1993), Le Coq (2002), Crampes and Creti (2005), Fabra, von der

Fehr,and Harbord (2006), and more recently Fabra, von der Fehr, and de Frutos (2011) developed

a multi-unit auction framework in which capacity constrained bidders with constant marginal

costs compete in electricity auctions. I focus on a modified version of Fabra, von der Fehr, and

Harbord (2006) and apply the auction model to data from the New York Independent System

∗German Institute for Economic Research DIW Berlin and Technichal Universityof Munich; sschwenen@diw.de.

This article was part of my dissertation at the Copenhagen Business School, Department of Economics. I would like

to thank Anette Boom, Richard Green, Chlo´

e Le Coq, Peter Møllgaard, Clinton Levitt,Michael Weichselbaumer, Anne

Neumann, Sophia Ruester, Alberto Lamadrid and seminar participants at the IFN Stockholm and at the 2012 EARIE

conference in Rome for their valuable suggestions. Two anonymous reviewersprovided excellent and helpful comments.

All errors remain my own. Research grants from the Danish Social Science Research Council (09-061934) and the

German Ministry for Economic Affairs and Energy (03MAP274) are gratefully acknowledged.

730 C2015, The RAND Corporation.

SEBASTIAN SCHWENEN / 731

Operator (NYISO) auctions for installed power generating capacity (short, “ICAP” or capacity

market). Results show that that these models are sufficient to predict economic bidding behavior

in multi-unit auctions when bidders are capacity constrained.

By tailoring a multi-unit auction model to the NYISO capacity market, this article reveals

how market concentration and strategic firm behavior can undermine the regulatory intention

of capacity markets. Thereby, this article also adds to the discussion on supply security and

electricity market design. In the period studied between 2003 and 2008, power generating firms

in the NYISO capacity market coordinated on an equilibrium play that was extracting the highest

possible rents for the supply side. The capacity market always cleared at the highest price

possible and thus set incorrect price signals for profitability and market entry of new generating

resources.

The economic theory of multi-unit auctions dates back to the share auction framework by

Wilson (1979). In Wilson’smodel, bidders have uncertainty in their valuation, respectively,in their

costs within a procurement setting. Klemperer and Meyer (1989) analyzefir ms’ bidding strategies

with perfect information on costs but introduce demand uncertainty. In both environments, firms

maximize profits by submitting bid functions that increase over their offered quantity. Green and

Newbery (1992) were the first to tailor a multi-unit auction model to electricity markets, relyingon

the Klemperer and Meyer approach. Newbery and Green designed the model to describe smooth

upward sloping equilibrium supply functions in the UK spot market for electricity and confirmed

the model’s predictions. Wolfram (1998) showed how larger suppliers in the UK electricity spot

market submit higher strategic bids than smaller competitors, as they obtain revenues for higher

clearing prices on relatively more operating units and thus have higher incentives to increase

their bids. More recent structural empirical works by Wolak (2003), Sioshansi and Oren (2007),

and Hortac¸su and Puller (2008) provide additional support by including forward markets and

introducing nonparametric tests. So far, empirical findings for simple multi-unit auction models

in the style of Fabra, von der Fehr, and Harbord (2006) are not documented, which is partly due

to the stylized nature of these models. In contrast to models with smooth supply functions, von

der Fehr, and Harbord (1993) and Fabra, von der Fehr, and Harbord (2006) find that the size of

generating units matters for equilibrium bids and show how capacity constraints of rival firms

incentivize one pivotal firm to increase the clearing price at the margin. Auctions for installed

generation capacity in electricity markets take place in an environment very close to the one

assumed in Fabra, von der Fehr, and Harbord (2006) and are ideal to deliver empirical insights

on the predictions of such models.

The remainder is organized as follows. Section 2 illustrates the design of the New York

electricity and capacity market. Section 3 introduces a model for multi-unit uniform price pro-

curement auctions with capacity constrained firms that reflects the market design illustrated in

Section 2. The empirical approach and the data are discussed in Section 4. Section 5 presents the

empirical findings. I compare optimal bids generated by the model to observed bids in the auction,

assess deviations from the model, and show estimates of the best response functions. Section 6

concludes on the empirical findings and draws policy recommendations based on counterfactual

auction outcomes.

2. The New York capacity market

This section sketches the market design of the New York capacity market. The New York

power market consists of an electricity market and a capacity market. In most if not all other

markets, pricing the commodity only is sufficient to promote long-run investment; neither a

demand nor a regulatory instrument exists for pricing production capacity.

In electricity markets, the existence of dominant firms and the absence of a robust real-time

demand response require that in times of shortage, the market price is set administratively. When

this price cap is set just above marginal costs and significantly below the value of lost load to

consumers (which is the case in many major US electricity markets), electricity prices are a weak

C

The RAND Corporation 2015.