Models can overcome a key oversight (em

that both supply and demand affect competition.

This past December, the Federal Energy Regulatory Commission (FERC) issued a policy statement describing important changes in how it will evaluate proposed mergers under the Federal Power Act's public interest standard. These changes should lead to significant improvements (em not only in the evaluation of mergers, but also for other matters that affect market power, %n1%n including industry restructuring and market-based pricing.

It comes as no surprise that the policy statement identifies "effects on competition" as the central issue raised by a merger application. Of greater interest, the statement takes positions on how such effects should be analyzed, of which three stand out.

First, FERC formally adopts the Horizontal Merger Guidelines developed by the Department of Justice (DOJ) and the Federal Trade Commission (FTC) as a framework to analyze competitive effects, bringing its evaluation of market power into the mainstream of antitrust. Combined with heightened interest in the electric industry at the federal antitrust agencies, this action should mean no lack of work for experienced antitrust attorneys and economists.

Second, FERC specifies a "competitive analysis screen" as a first test to discriminate between mergers that raise competitive issues warranting further investigation, and those that do not. This screen describes procedures for several tasks, such as 1) defining relevant product and geographic markets in which market power might be enhanced, 2) identifying potential suppliers in such markets, and 3) computing market shares and concentration indexes for comparison with the Guidelines standards. To satisfy the policy statement, applicants typically must apply the screen and show results under various system conditions and price levels for those groups of customers that might suffer antitrust injury.

Third, as a component of the analytic screen, the FERC adopts a "delivered price test" to determine the geographic scope of competition (see sidebar). Together with information on transmission constraints, this test identifies which generating units to include in computing market shares and concentration for energy delivered to various groups of customers.

This new delivered price test is grounded in sound economic principles. It represents a substantial improvement over the Tier 1/ Tier 2 method that it replaces. Unlike the latter, the delivered price test takes into account the effects of generating costs, transmission losses and tariffs, and other ancillary service costs (em all of which affect the ability of each generating unit to compete in supplying energy to a particular group of customers at a price that would constrain the exercise of market power by owners of other generators.

Nevertheless, the test fails to take account of all factors that affect the geographic scope of competition in energy markets. Competition depends on demand as well as supply conditions. Thus, despite its merits, the FERC's proposed method ignores the role played by the geographic pattern of loads in determining the geographic scope of competition in the supply of energy to any particular group of customers.

Measuring Market Power:

The Role of Opportunity Costs

How should applicants and intervenors go about complying with FERC's required method for determining the geographic scope of competition and for computing market shares and concentration indexes?

One approach would simply carry out the computations as the FERC has described them. Given the data, the task would be straightforward: Write a computer program to execute the delivered price test and compute the shares and concentration indexes required for the FERC's analytic screen.

However, given those same data, applicants and intervenors can and should perform a more refined analysis (em one that makes use of additional data that are available in the electric power industry.

To determine whether a generator that is not exercising market power would be able and willing to supply a particular group of customers at a specific price, it is not enough merely to check that generator's variable costs of production and delivery and the availability of transmission capacity. One must also consider whether sales of energy to the customers in question would require the generator to forego profitable sales of energy to customers located elsewhere. That is, one must consider the generator's opportunity costs. For example, a generator in Ohio that could pass the delivered price test for supply of energy to Maryland at a competitive price of $25/MWh would not in fact be willing to sell energy in Maryland if the energy could instead be sold in Michigan for $30/MWh. Therefore, it makes sense to ask whether there is a way to integrate the geographic pattern of loads into the analysis of competition required by FERC.

In fact, merger applicants should and can do precisely that by using computer simulation models similar to those used in the electric power industry to analyze generation dispatch. These models incorporate data of the type required by the FERC, as well as publicly available data on the geographic pattern of forecast loads. Adaptations of these models that are designed to analyze market power are not only available, but are already in use to analyze mergers, industry restructuring, and market-based pricing. We caution, however, that some of these models are poorly designed and hense produce misleading results.

Analyzing Data:

The Choice of Models

Broadly speaking, four types of computer simulation models that are widely used in the electric power industry might be considered as tools for analyzing market power issues.

1. Dispatch/Transportation Models. This category, which includes the National Power Model that we use at our company, can incorporate data on loads, transmission constraints between control areas, transmission losses and pricing, and the determinants of generator costs (em heat rates, fuel prices, and variable operating and maintenance costs that account for environmental compliance costs.

This class of model retains a major advantage compared to the three other model types described below. Dispatch/transportation models are already available in a form that can be used to analyze market power. They are used in connection with restructuring, merger, and market-based pricing proposals.

These models enable one to analyze, on an hour-by-hour basis and under alternative assumptions, the ability of a utility or set of utilities profitably to raise prices above competitive levels by reducing output, the magnitude of any anticompetitive price increase, the locations of customers adversely affected by the price increase, and the generators that would increase output in response to the price increase provided they would behave competitively. They can provide all the information necessary to implement the FERC's analytic screen, with the added refinement that they take account of the geographic pattern of loads.

Dispatch/transportation models are limited in one respect: They do not explicitly incorporate the electrical properties of the transmission grid. However, a load-flow model can be used to check that implied power flows are consistent with the properties of the transmission grid. Transmission constraints in the dispatch/

transportation model can be modified accordingly.

2. Dispatch/Unit-commitment Models. Models in this group, such as PROMOD, are commonly used in planning generation resources and estimating fuel consumption and operating costs for individual utilities. These models incorporate the most realistic representations of those generator characteristics relevant to unit-commitment and dispatch decisions. However,

dispatch/unit-commitment models are likely to prove less useful

than well-designed, dispatch/

transportation models in analyzing competition among utilities or in addressing market power issues in areas larger than local load pockets.

Dispatch/unit commitment models incorporate detailed proprietary data for individual utilities; a utility is unlikely to have access to all the information required by the models for its competitors' generating units. Furthermore, these models are designed for relatively small numbers of control areas. They may not perform as well for regions large enough for proper analysis of the geographic scope of competition and computation of valid market shares. Finally, these models were not designed to answer questions concerning competition, market shares, or market power. As a result, applicants would find it necessary either to revise the models or to perform development work on post-processing programs before using them to address a number of central issues of market power.

3. Load-flow Models. These models, such as PSS/E and Power World, simulate energy flows over a transmission grid. Their strength comes from the fact that they explicitly account for the flow of energy over all paths on an alternating current transmission grid (that is, loop flows), as well as the properties of the transmission grid that give rise to thermal and voltage constraints on energy transfers. They can also provide information on losses.

Pure load-flow models also exhibit a weakness: They ignore information on generation costs and transmission tariffs. As a result, taken alone these models can provide only limited information regarding market shares and market power. Even so, these models represent the standard source of estimates regarding the transfer capability limits of the transmission system.

4. Load-flow/Dispatch Models. This last category, which includes vehicles such as the Multi-Area Production Simulation (MAPS) and Transmission-Oriented Production Simulation (TOPS) models, integrate load-flow models with

dispatch and are capable of simulating dispatch over large areas while accounting for load flows. Their disadvantage is evident: They require large amounts of data, many of which are not publicly available. For example, the models require data on loads at individual busses instead of the publicly available aggregate loads for utilities. And while this group offers a more detailed modeling of the transmission grid, that feature may not produce better results than obtained using simpler models.

For example, low-voltage lines not included in these models may limit transfer capabilities below the levels implicit in the models. In addition, these models currently do not include operating guides that allow transmission operators to increase transfer capability by opening lines, changing transformer and capacitor configurations, or dispatching plants out of merit order.

Finally, these models were not developed to analyze market power. Substantial development work in interactive solving and post-process programs would be necessary before these models would produce results that could form the basis for the market shares required by the FERC.

Other Model Uses:

Some Practical Examples

We expect that during the next twelve months participants in the electric power industry will make widespread use of computer simulation models, particularly the

dispatch/transportation models, in addressing market power issues at the FERC (em not only because of the new approach articulated in the FERC's statement on merger policy, but also because these models are useful in other contexts.

For example, computer simulation models should prove useful in making presentations on market power before the Justice Department, as the DOJ has already suggested their use in comments on the FERC's merger policy. %n2%n Many states are actively reviewing the competitive effects of mergers and industry restructuring proposals, offering still another forum for computer simulation models. Indeed, our own interest in these models has been driven by state proceedings.

Also, these models can be used for a number of purposes beyond analyzing the geographic scope of competition in computing market shares and concentration. For example, they may support or refute claimed instances of market power not detected by the FERC's analytic screen. The policy statement explicitly raises this possibility: That other evidence may show market power problems that the screen does not reveal.

In the case of any merger that does not pass the analytic screen, one should anticipate a detailed investigation of competitive effects. Well-designed models will naturally play a key role in those investigations. Specifically, computer simulation models will provide certain, albeit incomplete answers to the ultimate question under the Merger Guidelines: Will a particular merger create or enhance market power?

Properly constructed models provide information on the ability of one, two, or more utilities profitably to raise prices above competitive levels by reducing their outputs or sales of energy. These models have an inherent limitation, however. they cannot replace traditional antitrust analysis; instead, they can serve only as a supplement. Thus, while these models may demonstrate that one or more utilities might well possess the capacity to exercise market power by engaging in certain types of behavior (for example, acting as a dominant firm, or as noncooperating oligopolists %n3%n), the models do not deal (or do not deal well) with some other theories of anticompetitive behavior, such as collusion, for example. %n4%n

If a model cannot accommodate a particular type of anticompetitive behavior, it cannot demonstrate whether such behavior should pose a problem. Therefore, studies of market share and concentration will continue to play an important role in any investigation of market power.

Utilities will also use computer simulation models for purposes unrelated to regulatory filings and antitrust litigation. Some utilities possess market power. These

utilities will find that computer simulation models offer a useful tool to evaluate alternative pricing strategies to determine which one will exploit market power most effectively and yield the greatest profits.

For example, suppose a utility that participates in a price-based dispatch pool runs three generating units, A, B and C, with incremental costs of $20, $25, or $30 per megawatt-hour (MWh), respectively. The utility might then consider two alternative pricing strategies. Strategy 1 would dispatch any unit whenever the market price exceeded that unit's incremental cost. Strategy 2 would dispatch unit A when the market price climbed above $22/MWh, and units B and C, respectively, whenever the price exceeded $28 or $34/MWh.

In this case, dispatch/transportation models can be used to determine which pricing strategy is the more profitable. These models can prove useful to utilities without market power (em to predict the pricing behavior of competitors and suppliers, to assist them with their own unit-commitment and dispatch decisions. These models should also prove useful to evaluate investment decisions, the effects of entry by competitors, or even stranded costs. Dispatch/ transportation models are already found to be useful in evaluating pricing and investment strategies for utility fuel suppliers, such as coal producers. t

Mark Frankena is a senior vice president and John Morris is a vice president at Economists Incorporated, Washington, DC (http://www.ei.com), where they specialize in competition analysis in regulated industries, including electric power, natural gas, and telecommunications. Dr. Frankena and Dr. Morris previously were employed at the Bureau of Economics, Federal Trade Commission. They have testified recently before the Federal Energy Regulatory Commission on the competitive effects of the Primergy merger.

The Delivered Price Test

Measuring the Extent of Market Power

When will the FERC consider a generator operating in one location as a competitor in another area? That question is answered by the "delivered price test," as shown by example involving two locations, nodes A and B.

Assume a competitive price of $20 per megawatt-hour (MWh) for energy delivered to customers at node A during a particular hour. In that case, a generator located at node B will not supply energy for delivery to node A, and will have no incentive to reduce its sales to raise the market price at node A above the competitive level, unless the sum of that generator's variable costs of generation plus the variable costs of transmission and ancillary services associated with a transaction from node B to node A equals $20/MWh or less.

Further, a generator located at node B will not have the ability to play a role in constraining a "small but significant" (5%) anticompetitive price increase at node A unless the sum of these same variable costs is $21/MWh or less.*

For these reasons, a generator at node B is not considered to be a supplier in the delivered energy market relevant to customers at node A unless that generator passes the delivered price test (em that is, unless it can deliver energy to node A at a variable cost of $21/MWh or less.

*Those not familiar with antitrust analysis may misinterpret the term "small but significant and non-transitory," as used to describe a price increase of 5 percent. That term does not imply that the antitrust agencies will not challenge a merger that would lead to an anticompetitive price increase of less than 5 percent. To the contrary, Section 1.0 of the Horizontal Merger Guidelines states that "The 'small but significant and non-transitory' increase in price is employed solely as a methodological tool for the analysis of mergers: it is not a tolerance level for price increases."

1"Market power" denotes the ability of one or more firms profitably to raise prices above competitive levels for a significant period of time. 2See also, G. Werden, "Identifying Market Power in Electric Generation," PUBLIC UTILITIES FORTNIGHTLY, Feb. 15, 1996, p. 19. 3Noncooperating oligopolists exercise market power simultaneously without an agreement. Rather, each supplier maximizes its own profits based on an assumption about how other suppliers will behave. For example, in the Cournot model, each supplier assumes that other suppliers will not change their outputs in response to the output chosen by the first supplier. Each supplier then takes into account the effect of its own output on the market price and chooses the output that maximizes its own profits. 4Unlike noncooperating oligopolists, colluding suppliers reach, monitor and enforcement an agreement, although this may be done tacitly without explicit communication or side payments.

Articles found on this page are available to Internet subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.