Review of Industrial Organization 19: 165–180, 2001.
© 2001 Kluwer Academic Publishers. Printed in the Netherlands.
Economies and Diseconomies: Estimating
Electricity Cost Functions
MICHAEL T. MALONEY
Clemson University, Department of Economics, Clemson SC 29634, U.S.A.
Abstract. This paper presents estimates of the variable cost function of electricity generation.
The cost function is estimated using a two dimensional deﬁnition of capacity utilization. Because
electricity cannot be conveniently stored, generation facilities follow the load across demand cycles.
Capacity utilization can be captured empirically in two ways. One is generation relative to capacity
when a unit is connected to the system; the other is the percent of time the unit is disconnected. The
estimated cost function shows that both dimensions affect average cost, which generally declines as
capacity utilization increases.
Key words: Capacity utilization, cost estimation, electric generation.
Estimation of the cost function for electricity generation has always intrigued eco-
Numerous papers have explored virtually every nuance of the problem.
One reason for this is the fact that relatively good plant and ﬁrm level data have
historically been available from reports that regulated utilities are required to ﬁle
with the federal government and before state regulatory commissions. Also, the
electric industry is big, and therefore, worthy of attention by economists. Industry
revenues in 1998 were $217 billion.
Simple, single equation models have been ﬁt and more complicated multi-
equation estimating techniques were pioneered with application to electricity
production. Technological progress has always been a focus; both simple and soph-
isticated models have been used to examine the effect of technological progress on
Thanks go to Skip Sauer, David Dismukes, and Roger Betancourt for helpful comments.
Cowing and Smith (1978) review the literature of the sixties and seventies. Following that, see
Stewart (1979), Stevenson (1980), Joskow and Schmalensee (1983), Betancourt (1986), Betancourt
and Edwards (1987), and Nelson (1989).