Nuclear and fossil fuel steam generation of electricity: differences and similarities.

• Author: Kamerschen, David R.

1. Introduction

   Nuclear power, produced in about 112 plants, accounts for roughly one-fifth of U. S. electricity. Because of cost overruns, technical problems, loss of public and political confidence and backing, falling fuel prices, and slower than anticipated growth in energy demand, this figure is significantly below what many expected [17, 269]. In fact, no U.S. utility has ordered a nuclear plant in over 15 years. Recent environmental legislation, however, may improve nuclear power's outlook. While nuclear electric power is clearly an important alternative, the basic production technology and costs, and the efficiency and equity of the nuclear rate structure, have received little rigorous statistical investigation.

   The costs and efficiency of electric utilities with nuclear plants have not been studied because of perceived differences in the underlying production function or measurement difficulties. A need to preserve technological homogeneity is often suggested as a reason to not include nuclear power. However, most studies have pooled fossil-fuel plants of different vintages, fuel types, capacities, regions, and load characteristics, with little more than a disclaimer concerning aggregation difficulties.(1) It is not clear that nuclear steam electric generation is significantly different from conventional steam production methods, and pooling to measure total firm steam generation costs is a more serious aggregation problem.

   Several recent cost studies deal with the uniqueness of nuclear power, but arrive at different conclusions. Naughton [22] rejected the hypothesis of separability of nuclear and fossil-fuel costs for a 1980 cross-section of 78 privately owned electric utilities, where 10% of the steam generation was nuclear and the remainder was fossil-fuel. Naughton's emphasis, however, was on the efficiency of the rate structure, not on generation characteristics. We compare Naughton's results with a study of rate structure efficiency, using our sample, in Thompson, Kamerschen, and Danielsen, [29] and section IV of this paper.

   Krautmann and Solow [18] evaluated the cost effectiveness of nuclear power and estimated economies of scale, using a pooled cross-section of 43 plants over the period 1976-78. By using a short-run variable-cost model, they found long-run average-cost curves were concave, contrary to the evidence on fossil-fuel generation found in Nerlove [24] and Christensen and Greene [9]. The Krautmann and Solow [18] study however, does not provide strong evidence about scale economies. Whether nuclear power should be separated from conventional fossil-fuel methods and what will be its implications for electric generation cost estimation and efficiency need additional comprehensive analysis.

   Using data from a 1985 cross-section of 40 privately owned electric utilities of which approximately 36% of steam generation was nuclear,(2) we estimate the cost functions of nuclear and fossil-fuel generation at the firm-level. Statistical tests and other relevant measures are used to determine whether nuclear power should be treated as a sample separate from conventional steam generation.

   Section II of our paper discusses the development of the statistical model. Section III contains the estimation results and the results of the comparison tests on fossil-fuel and nuclear steam cost functions. Section IV discusses the efficiency of the rate structure for nuclear utilities. Section V contains a summary and conclusion.

2. The Statistical Model

   The general form of the generation cost equation used for both nuclear and fossil-fuel steam generation does not vary greatly from the traditional neoclassical cost model according to Cowing and Smith [10].

   Arguably, the labor and capital input demand equations for nuclear power contain variables related to the unique technical and regulatory characteristics of this type of generation. Plant age, for example, may be more critical for nuclear plant costs than for other technologies for several reasons. First, the supposed benefits of nuclear power are its low operating and fuel costs. As a plant ages, for example, its downtime increases and more expensive "replacement power" must be provided to meet the demand. The firm has an economic incentive to spend more on maintenance (mostly labor) to avoid this cost. Second, nuclear power is still a relatively new and complex technology. As a plant ages, operation and maintenance workers gain experience. For example, operators can anticipate downtime and can coordinate it with fuel replacement to reduce maintenance and operations costs. As a result of these two conflicting forces, increased costs from deterioration, and decreased costs from the "experience curve,"(3) the effects of plant age could be important. The net effect, however, cannot be hypothesized, a priori.

   The Nuclear Regulatory Commission (NRC) regulates every aspect of a nuclear plant's operation. NRC penalties and fines, largely for safety violations, are a small cost relative to operations costs (usually less than 1%). However, fines may indicate a rising trend in current capital and labor costs necessary to resolve or avoid regulatory violations and the adverse publicity they can cause in the future. It is hypothesized that the greater the value of regulatory-imposed fines, the higher the costs, ceteris paribus.

   These special characteristics, and other technical constraints of nuclear power, such as limited input substitution, could affect the content and form of the cost function and are, therefore, subject to statistical inquiry. This study utilizes the long-run version of the translog cost function for both fossil-fuel and nuclear power, which assumes the firm is in long-run equilibrium. It is implicit in the long-run cost function that all inputs have adjusted fully to their equilibrium value, given current market prices.

   Researchers differ, however, about how capital in the electric utility industry should be treated. Generally, it comes down to the level of aggregation that is employed. At the plant level, a reasonable argument can be made that the majority of capital is quasi-fixed and, therefore, may not have adjusted to its long-run equilibrium value at any given time. However, at the firm level, considerable discretion is generally available over a variety of plants of different ages and technologies. This issue is discussed in Nelson 1231. Nuclear power may represent a special case.

   Several studies, including Naughton [221 and Krautmann and Solow [18], have estimated nuclear power costs with a short-run, variable-cost function, presumably on the assumption that nuclear power represents a quasi-fixed investment. Naughton used firm-level data, since his emphasis was on the efficiency of the tariff structure. Krautmann and Solow used nuclear reactor and plant data in their study of scale economies in nuclear plants.

   This study, emphasizing firm behavior, implicitly assumes fossil-fuel and nuclear power are similar in their response to capital prices. The multi-firm ownership of many nuclear plants makes distributing the capital expense a complex issue. Even for one plant, the assumption of a fixed capital expense may not be entirely valid. Recent expert testimony on depreciation expenses for a nuclear power plant in this sample explored 4800 separate capital systems that comprise the plant, each of which depreciates, is retired, or is replaced at a different rate. New technology and capital costs apparently play an important role in this process.(4)

   For fossil-fuel and nuclear steam production, the following general translog cost function was estimated, based on the model developed by Christensen, Jorgenson, and Lau [7], and Christensen, Jorgenson, and Lau [8], and originally applied to electric utility costs by Christensen and Greene [9]:

   [Mathematical Expression Omitted]

   where

   C = Total cost of generation,

   Q = Output of electricity,

   Pi = Price of the ith...