FEDERAL TRADE COMMISSION
Hearings on the Changing Nature of Competition in a Global and Innovation-Driven Age

(November 7, 1995)
Revised January, 1996

PROOF OF EFFICIENCIES IN MERGERS AND JOINT VENTURES:
TESTING EX ANTE CLAIMS AGAINST EX POST EVIDENCE

Joseph F. Brodley
Boston University Law School and
Department of Economics

FINAL STATEMENT

TABLE OF CONTENTS

PROOF OF EFFICIENCIES IN MERGERS AND JOINT VENTURES:

TESTING EX ANTE CLAIMS AGAINST EX POST EVIDENCE

Joseph F. Brodley(1)

Efficiencies claims in mergers and joint ventures confront the dilemma that some anticompetitive transactions may achieve important production and innovation efficiencies. Yet under existing procedures enforcement agencies must resolve efficiency claims in advance of the transaction, at a time when the agency cannot effectively test expansive and unverified efficiency projections by interested parties. Resolution of this dilemma requires procedural innovation.

More specifically, proof of efficiencies in mergers and joint ventures requires a two-stage procedure: an ex ante or threshold stage in which the FTC (or Justice Department) would screen transactions to determine if they are likely to create substantial efficiencies, and an ex post stage in which the enforcement agency would examine actual efficiency outcomes. Thus, the first stage inquiry would focus on whether there is an identified market imperfection that justifies consideration of an efficiencies defense and whether the proposed transaction is an appropriate means to achieve the specified efficiencies. The second stage subjects the transaction to an after-the-fact audit to determine whether efficiencies have in fact been achieved, and if not, administers appropriate relief to restore competitive conditions.

The proposed procedure would require verification of efficiency claims. It also serves as a separating mechanism, discouraging inflated efficiency assertions without barring valid claims. Firms that assert unfounded efficiencies claims would know that they face divestiture or other costly remedies if efficiencies are not forthcoming, and would likely be deterred from making extravagant ex ante claims. The two-stage procedure is also information-conserving in that the agency can apply an under inclusive legal rule at the ex ante stage when information is uncertain; it can then undertake more stringent review at the ex post stage when actual results are available. A two-stage procedure has been used in Germany, and has been proposed by three antitrust scholars (Pitofsky, 1992; Scherer, 1990; Brodley, 1987).(2)

This paper attempts to develop a framework for a two-stage, efficiency review. Part I outlines the general conditions for use of the procedure. Part II suggests procedures for the Stage One inquiry. Part III covers the Stage Two review. Part IV discusses possible objections.

I. Requirements for Use of Two-Stage Procedure

The proposed two-stage procedure would, at least initially, govern only the exercise of enforcement agency discretion in determining whether to challenge an anticompetitive merger or joint venture (hereinafter jointly referred to as "COLLABORATION"). The essential precondition for use of the procedure would be an initial determination by the enforcement agency that the proposed collaboration is anticompetitive; and that in the absence of a fully justified efficiencies defense, confirmed by ex post evidence, the agency will refuse to clear the transaction. In addition, the following additional conditions would have to be satisfied.

A. Substantial Cost Saving or Quality Improvement

The efficiency gain must be substantial. A numerical standard, while somewhat arbitrary, is necessary to avoid otherwise intractable balancing issues and to provide greater certainty in business planning. Thus, the efficiency gain ought to equal at least five percent of total cost or unit price, or substantially improve the quality of the product. Use of a five percent threshold follows the Merger Guidelines, which define a price increase of less than five percent as involving insignificant pricing power for market definition purposes. Similar reasoning justifies use of a five percent threshold for efficiencies in production markets. In innovation markets a higher threshold may be appropriate, however, in view of the difficulty of predicting innovation outcomes and the larger anticompetitive dangers from collusion in innovation markets.

B. Time Between Ex Ante and Ex Post Proceeding

The ex post proceeding should normally be held between three and five years after the ex ante determination. Efficiency realization generally will require a longer time period than that used in competitive analysis of mergers, but an upper bound for efficiencies review is needed to prevent indefinite postponement of the ex post review. In addition, the longer the time period, the more difficult it becomes to trace efficiency gains to any particular cause or event. However, a single time period, for example three years, seems inflexible in view of the variety of industries and transactions that will be involved. Thus, the agency would specify the review period when it grants ex ante clearance, not to exceed five years.

C. Types of Efficiencies

An efficiencies defense should recognize both productive and innovation economies. These encompass scale economies, including network economies, economies of scope, transactional and information economies, and innovation economies involving the invention, development and diffusion of new technology. However, an efficiencies defense should exclude purely pecuniary or capital-raising economies, and should admit managerial economies only on a very strong showing.

1. Scale Economies

Authorities uniformly agree that if an economies defense is to be recognized, scale economies should be included. Scale economies include network economies, since the more people or firms in a network, the lower the costs per unit or higher the quality of the output. While scale economies are available from integration of production among smaller firms, most larger firms have already reached sufficient size to achieve scale economies (Scherer & Ross, 1990). Thus, scale economies will not frequently justify a horizontal merger between large firms, but may well justify a joint venture by market powerful firms into a market in which the participants are not presently engaged, or a vertical merger.

2. Other Productive Economies

Other productive economies, such as scope economies, based on the efficiencies available from producing multiple products, and transactional economies should also be included in an efficiencies defense. This conclusion received general support at the hearings (see testimony of Arquit, Brodley, Griffin, Jones, Muris, Rill, Salop and Sanderson). However, it is contrary to the views of some writers, who would limit consideration to scale economies (Areeda, 1980; Pitofsky 1992). While non-scale productive economies present greater problems of measurability, they also appear to provide a larger source of efficiency gain, at least for collaboration between large firms. The greater difficulty of proof of non-scale economies should not lead to over inclusiveness since the burden of proof would be on the parties.

3. Innovation Economies

Innovation economies should be included in an efficiencies defense because they provide the greatest potential enhancement of social wealth. Indeed, innovation efficiency or technological progress provides the single most important factor in the growth of real output in the industrial world, representing over half the total gain in United States output over extensive time periods (Gilbert & Sunshine, 1995; Brodley, 1987). Moreover, the benefits of innovation are eventually passed on to consumers through the natural diffusion of technical knowledge (Roberts & Salop, 1994). On the other hand, innovation economies are the least measurable type of efficiency. But it is this very characteristic that underscores the need for the two-stage procedure.

A two-stage procedure is especially necessary for innovation economies. If efficiencies review is limited to ex ante determination, the antitrust decision maker faces the heroic task of predicting whether a new combination of inputs will produce knowledge that does not yet exist. It is precisely in this situation that an ex post audit can make recognition of innovation efficiencies prudent. If despite ex ante promise, efficiencies are not forthcoming, the transaction can be restructured to remove antitrust risk.

4. Pecuniary, Capital-Raising, and Managerial Economies

Pecuniary and capital-raising economies should not be included in an efficiencies defense. The efficiencies justification should be restricted to the economies that make the largest contribution to increased social wealth -- production and innovation advance. Pecuniary economies, such as tax avoidance savings or bargaining advantages available to large buyers, reduce private costs, but not social costs.(3) Capital-raising economies should not be recognized, as capital markets sometimes discriminate against small firms for reasons that do not rest on increased risk factors or other costs (Scherer & Ross, 1990). Managerial economies should be recognized only in extraordinary cases since their existence assumes a shortage of available managerial talent or experience, which appears unlikely in most U.S. markets.

D. Pass-On to Consumers

An efficiencies defense should not be conditioned on an immediate pass-on of cost savings to consumers. Production and innovation economies produce large social benefits even when not immediately passed on to consumers. They produce real resource savings with valuable spill overs into other markets, compounding social wealth, far exceeding the gains from allocative efficiencies (Brodley, 1987). Thus, as early as 1916, Judge Learned Hand recognized that the long-run interest of consumers may differ from their short-run desire for "an immediate fall in prices."(4) Indeed, as Chairman (then Professor) Pitofsky has pointed out, complete (and immediate) pass-on occurs only in a fully competitive market, in which case no efficiency defense would be needed (Pitofsky, 1992).

Moreover, eventual pass-on of benefits to consumers is in fact likely to occur for at least two reasons. First, under the procedures I am proposing, the duration of efficiency-justified market power would normally be limited, thus assuring that consumers would eventually share in the benefits of the efficiencies defense. Second, as previously noted, in the case of innovation efficiencies, diffusion of innovation to rivals assures an eventual pass-on of benefits to consumers (Roberts & Salop, 1994).

E. Least Restrictive Alternative

The merger or other collaboration must be the least restrictive means reasonably available to achieve the efficiencies (and in that sense be specific to the merger or other collaboration). In operational terms this means that any feasible alternative would involve significantly higher costs. The least restrictive alternative requirement is necessary to prevent the unnecessary sacrifice of competition and because courts and agencies cannot meaningfully balance efficiency gains against anticompetitive losses except in polar cases.

Courts and agencies are generally unable to balance efficiency and anticompetitive effects because in any close case a balancing analysis requires estimates of magnitudes and probabilities. Enforcers lack information to make such estimates reliably, particularly in advance of a transaction. Thus, while courts often embrace balancing in principle, they rarely, if ever actually attempt it.(5)

Under these circumstances the least restrictive alternative serves as a next best alternative to balancing. The least restrictive alternative clearly provides a more manageable inquiry than balancing because it requires only an incremental analysis. The decision maker need simply compare roughly similar alternatives to determine whether proposed collaboration is significantly more restrictive than necessary. This avoids the need for global balancing, and to considerable degree reflects actual judicial practice.(6)

The least restrictive alternative provides an acceptable, if rough proxy for weighing benefits against costs. When a less restrictive alternative is reasonably available, blocking the transaction causes little efficiencies loss because the parties can turn to the slightly more costly (and less restrictive) alternative. Thus, the incremental benefit from the transaction will generally be small even if the efficiencies gain is large.

On the other hand, if there is no alternative means for achieving the projected efficiencies, the incremental benefit from the transaction is likely to be much larger because barring the transaction will totally prevent realization of the efficiencies. It is, of course, possible that the efficiencies gain is modest even when no less restrictive alternative exists. Under these circumstances the least restrictive alternative is overly permissive (assuming the transaction involves a significant collusive loss). Thus, the least restrictive alternative approach provides an imperfect substitute for weighing. But more is not realistically possible within the legal process.(7)

F. Competitive Effect or Affirmative Defense?

Is the efficiencies determination part of an overall assessment of competitive effects or should efficiencies be determined separately after a prima facie case of liability is established? The answer is that both kinds of analysis are needed. Under current practice enforcement agencies consider efficiencies and collusive effects in a single unitary analysis of competitive effects. The analysis is inherently qualitative, involving a rough trade off between efficiency benefits and anticompetitive losses. If the transaction appears motivated primarily by efficiency goals, and if perceived efficiency effects clearly and powerfully outweigh perceived anticompetitive tendencies, the agency is likely to clear the transaction (barring a less restrictive alternative). Where both efficiency and collusive tendencies are present, the agency will normally refuse to grant clearance. The proposed two-stage efficiencies procedure would involve no change in these enforcement review standards, and hence the initial efficiencies review would continue to be part of a single unitary analysis of competitive effects.

Thus, the two-stage efficiencies review would operate only after the agency has initially refused to clear a merger or joint venture. Under the proposed two-stage procedure efficiencies become an affirmative defense, but with differing requirements of proof at the ex ante and ex post stages. At the ex ante stage the parties must demonstrate that the proposed collaboration is plausibly and convincingly likely to create substantial efficiencies. As discussed in part II below, exact quantification of efficiencies would not be necessary at the ex ante stage, but the parties would be required to submit reasonable projections and estimates.

At the ex post stage quantification of efficiency gains is feasible. Thus, respondents would have to make an affirmative showing of efficiencies realization. Even at the ex post stage, however, an effort to conduct a tradeoff analysis, weighing efficiency gains against collusive losses leads to highly complex proceedings and unpredictable outcomes. Thus, the ex post efficiency justification would require proof only that the efficiency gain achieved the specified magnitude, for example five percent of total costs. In that sense the second stage efficiency defense is an absolute defense.

II. Stage One: Ex Ante Procedure

The first stage review subjects the efficiency claim to several threshold conditions to identify claims appropriate for an efficiencies defense. These conditions are designed to assure that conditionally approved collaboration will likely achieve substantial social efficiencies, causing the least intrusion on existing competition consistent with achieving efficiencies, preserving potential competition from related markets, and not foreclosing an effective ex post remedy if efficiencies fail to be realized. The failure to meet any one of these preconditions disqualifies the transaction from possible efficiency justification (and of course precludes ex post review). Since the criteria differ in some respects as between production and innovation markets, I discuss them separately. The section concludes with a discussion of standards of proof.

A. Threshold Requirements

1. Production Markets

In production markets an efficiencies defense would be available only if each of the following preconditions are satisfied.

Precondition 1: A Substantial and Non-Transitory Market Imperfection Blocks Realization of Significant Efficiencies.

The indispensable first condition for a market failure defense in production markets is existence of a market imperfection or structural impediment that prevents the competitive market from achieving significant efficiencies. A substantial structural impediment may be defined as a condition requiring the creation of market power through merger, joint venture or other collaboration to achieve social efficiencies. The classic example is a natural monopoly, where realization of optimal scale economies requires consolidation of the market into a single firm. Unless such a structural impediment exists, there is no reason to permit collaboration that creates or perpetuates market power.

The imperfection must block a real social efficiency. A restraint on competition cannot be justified on the basis of private wealth enhancement alone, but must also increase social wealth. In addition, the structural impediment must be non-transitory and durable in the sense that it cannot be overcome short of collaboration that creates market power.

Precondition 2: The Proposed Collaboration will Plausibly Overcome the Identified Market Impediment by Providing Anticipated Cost Savings of at Least Five Percent or Substantially Enhance Product Quality or Variety.

The second precondition requires that the proposed collaboration plausibly overcome the market impediment that alone justifies consideration of an efficiencies defense. A presumptively justifiable efficiency claim is defined as a cost saving of at least five percent of total cost or unit price, or a substantial enhancement of product quality or variety. At the ex ante stage the proponents of the efficiency defense cannot realistically prove the magnitude of the cost saving or the value of the product improvement. But they can be expected to make a plausible and reasonably convincing showing that the proposed collaboration is likely to produce a substantial efficiency gain. Thus, the second precondition requires that the restraint be efficacious. But the restraint should also be parsimonious, in that the intrusion on competition should be limited; and the following three preconditions make that consideration explicit.

Precondition 3: The Proposed Collaboration Maintains Competition at Other Levels of the Productive Process,and Does Not Permanently Eliminate Competition at Any Productive Level, Except under Conditions of Strict Necessity.

It does not follow that because a correctable market imperfection exists at one level of production, competition must be reduced at all other levels. Indeed, it is essential that competition be preserved to the maximum extent possible at other productive levels. This follows both from our general preference for achieving efficiency through competitive processes, and because retention of competition at other productive levels enhances the likelihood of future entry at the competitively-restrained level, thereby protecting the consumer interest in the long run.

Similarly, even at the level at which the collaboration creates market power, the restraint should not be perpetual, except under conditions of strict market necessity. Instead the restraint should last for only such time as required to allow less anticompetitive mechanisms to develop. For example, if competitors with market power are allowed to form a joint venture because no single firm possesses the necessary productive capabilities, that does not by itself justify making the joint venture permanent. Rather, the duration of the joint venture should be limited to a period that allows the participants to develop the capabilities to undertake profitable individual production. Limited duration prevents a permanent peace treaty between competitors, and forces each participant to look toward the eventual day when it will face its rivals with cooperative ties broken.

Thus, permanent restraints on competition should be permitted only if the participants can show market necessity in the sense that absent the perpetual linkage, the market could not exist at all, or could exist only in drastically shrunken form. Even in that case it may be necessary that there be some public regulatory presence to constrain cartel behavior.(8)

Precondition 4: The Proposed Collaboration Does Not Create Undue Market Power or Single Firm Dominance.

Some mergers and joint ventures contain such high collusive risks that they ought never to be permitted even for a limited period. But analysts will differ on where the line should be drawn. In the European Community the cut- off for an efficiencies-justified merger is a market share of 40-45 percent based on the EC definition of a dominant firm (Testimony of Joseph P. Griffin). By contrast, Chairman Pitofsky, following the Merger Guidelines, suggested in his 1992 article that the cut-off should be a single firm market share of less than thirty-five percent and a concentration ratio not to exceed HHI 1800 (Pitofsky, 1992).

The Merger Guidelines standard is probably appropriate for mergers in production markets, although it would appear to allow relatively few efficiency justifications under current permissive enforcement standards. Joint ventures merit a somewhat more permissive standard since they present less anticompetitive risk, particularly those of limited scope or duration. A higher market share is also clearly appropriate for collaboration in innovation markets, where concentrations above HHI levels are sometimes justified, market shares are often difficult to estimate, and alternative measures of concentration may be necessary (see discussion of innovation collaboration, infra).

Precondition 5: The Proposed Collaboration is the Least Restrictive Means Reasonably Available to Achieve the Efficiencies. The proposed restraint should do no more damage to competition and the consumer interest than is reasonably necessary to overcome the market imperfection. This consideration, encompassed in the traditional judicial language that the restraint be a "least restrictive alternative," means in operational terms that any feasible alternative would involve significantly higher costs. As we have seen, the least restrictive alternative precondition is necessary because courts and agencies cannot effectively balance efficiency gains against anticompetitive losses except in extreme cases; and thus the least restrictive alternative test provides a rough substitute for balancing that requires only incremental analysis between feasible alternatives.

Some writers have objected to the "least restrictive alternative" test as being overly stringent because astute lawyers with the advantage of hindsight can usually suggest some alternative that appears less restrictive than the action actually taken. But after sight is not the proper viewpoint since the "least restrictive alternative" requirement should apply ex ante and not ex post. Thus, the issue should simply be whether at the time of the ex ante hearing the proposed collaboration is more restrictive of competition than other effective and reasonably available alternatives.

Precondition 6: An Effective Ex Post Remedy is Feasible.

A two stage efficiencies review will not be meaningful unless a failure to demonstrate efficiencies in the ex post proceeding leads to effective relief. But some transactions are so difficult to unwind, or the costs of monitoring firm behavior ex post so high, as to preclude an effective remedy. In that event an efficiencies defense should not be allowed.

Enforceability of the ex post remedy is facilitated by the consent order procedure suggested by Chairman Pitofsky in his 1992 article. Under this procedure the ex ante approval of the collaboration would be conditioned on the parties' consent to divest or otherwise restructure their collaboration if the antitrust agency subsequently found that promised efficiencies were not forthcoming. Where divestiture or similar restructuring is not practical, alternative remedies may sometimes suffice. These include (1) a "Crown Jewel" procedure under which the parties agree to divest themselves of an even more valuable property if the relief specified in the ex ante order is not carried out (see Yao & Dahdouk, 1993); (2) a penalty payment equal to the shortfall in the claimed cost reduction that induced ex ante approval (testimony of Kevin O'Connor and Steven Salop); (3) reduction in the price of the product to below the pre-merger level (testimony of Steven Salop).

Whatever its form, the ex post remedy would be imposed only with the respondents' consent. In addition, the antitrust agency might properly require firms that seek the benefits of the two-stage procedure to themselves propose how the transaction can be restructured to restore competition. An advance consent requirement may block some transactions for which efficiency claims are made, but consumers should not be asked to bear the costs of anticompetitive collaboration unless effective means exist to restore competition when efficiency gains do not occur. (The full list of production market preconditions is set forth in Appendix A)

2. Innovation Markets

The preconditions for an efficiencies defense in innovation markets are similar to those in production markets -- with three important exceptions. First, there need be no showing of a specific market imperfection in view of the more general market failure that occurs in innovation markets due to imperfect appropriability, beneficial social externalities, and perhaps other factors (Spence, 1984).(9) Thus, the respondents need establish only that the proposed transaction is a necessary and sufficient means to achieve significant innovation. Hence the first precondition for an efficiencies defense in innovation markets is as follows:

Precondition 1: The Proposed Collaboration is Likely to Produce Significant Innovation, which is Otherwise Not Likely to Occur

The second possible difference in the preconditions for innovation markets involves the cost saving required to justify an efficiencies defense. It can be argued that the five percent threshold used for production markets should be increased. First, expected innovation outcomes from collaboration are more difficult to predict. Second, anticompetitive risk appears greater since collusion in innovation markets may reduce innovation itself, causing a welfare negative compounding effect. On the other hand, the expected diffusion of innovation limits the period of collusive risk, thus reducing the percentage cost saving needed to offset any given collusive price rise (Roberts & Salop, 1994). On balance, a higher cost saving threshold for innovation efficiencies can be justified, but should not exceed 10 percent.

The third difference in the preconditions for innovation markets involves the standard of undue market power. This cannot be set at any specific concentration ratio in view of the difficulty of defining innovation markets. Thus, the market concentration ratios in the Merger Guidelines terms are not suitable. However, following the precedent of other government articulations, we can establish an upper bound in terms of the number of non-collaborating firms or joint venture groups remaining in the innovation market, for example three or four.(10) Thus, the suggested "market power" precondition for innovation markets is as follows:

Precondition 4: The Proposed Collaboration Does Not Create Single Firm Dominance in the Innovation Market or Reduce the Number of Actual or Potential Research Efforts to Less than Four

The inability to specify exact quantitative standards for innovation collaboration emphasizes the importance of the remaining preconditions, particularly the precondition that preserves effective competition at other productive levels and which precludes permanent suppression of competition at any productive level except under conditions of strict market necessity. The complete list of preconditions in innovation markets is set forth in Appendix B.

C. Proof of Efficiencies

Proof of efficiencies is necessary in the ex ante proceeding to demonstrate that the proposed collaboration is plausibly and convincingly likely to create substantial efficiencies. Such proof would be based primarily on evidence of internal plans, and cost and engineering studies, particularly those that the parties have relied on in planning the transaction -- as distinct from studies prepared with a view to litigation.

In proving efficiencies, the parties need not undertake costly investigations beyond what they require for their own internal decision making. However, if their pre-merger planning studies fail to establish a prima facie case of efficiencies, they can submit additional evidence, including (1) economic results of past mergers or joint ventures within the industry (Kwoka & Warren-Boulton, 1986), (2) cross-industry studies showing that similar transactions under comparable conditions have led to relative market success as compared with rivals, measured by growth of sales or lower costs (Yao & Dahdouh, 1993), and (3) stock market event studies, indicating increased market price for the combined value of the acquiring and acquired firms, following announcement of a merger or joint venture. While stock market event studies have severe limitations due to the "noise" or complexity of factors influencing the market price, they may provide at least an upper bound for the projected ex ante efficiency gain from a merger (see Kwoka & Warren-Boulton, 1986).

D. Burden of Proof

The respondents would have the burden of presenting a prima facie case on all issues with the exception of the least restrictive alternative. On that issue the burden would be on the enforcement agency to propose any assertedly less restrictive alternatives likely to produce comparable efficiencies. Respondents would then have the burden to show that the alternatives so specified would involve significantly higher costs or were otherwise not reasonably available.

In evaluating an efficiencies defense respondents generally bear the burden of proof because they possess superior information about their own efficiencies. However, on the issue of a less restrictive alternative it would be oppressive to require the participants to prove that the transaction is preferable to any conceivable alternative. The participants need only show that it is significantly less costly than any plausible alternative proposed by the government.

III. Stage Two: Ex Post Procedure

The Stage Two procedure subjects the ex ante efficiencies claims to the test of experience. Ex post review would occur after a sufficient time has passed to allow the projected efficiencies to be realized, which I have suggested should be a period of between three and five years. The parties will clearly have the incentive to cooperate in the ex post inquiry because a negative finding by the enforcement agency will require divestiture or other structural relief.

Despite the advantage of the ex post perspective, efficiency assessment presents formidable problems. To begin, there is a base-line issue since the production costs of firms tend to fall over time. Thus, the inquiry must focus not simply on whether the firm's costs have declined, but on whether they have fallen relative to other firms in the industry. In addition, when efficiencies are realized within a single division of a larger firm, difficult cost allocation issues arise. However, several tests are available to evaluate ex post efficiencies, which enable the enforcement agency to conduct a meaningful ex post review.

A. Proof of Efficiencies

Efficiencies may occur at the plant level or the firm level. Proof of efficiencies is much simpler at the plant level, where a single economic test may suffice. Proof at the firm level will typically require the use of several tests, applied sequentially.

1. Plant Level Economies

A merger or joint venture achieves economies at the plant level when it reduces the firm's costs or increases the value of its output in individual plants. Plant level savings, which typically involve economies of scale or scope, can be measured by engineering studies and statistical cost analysis. A clear showing under either measure should normally suffice to establish a prima facie case.

Engineering studies provide probably the best method for identifying and measuring scale economies. Such studies involve estimation of costs for each machine or process, on-site interviewing of workers, consideration of alternative plant designs, and integration of accumulated data through statistical and mathematical models (Scherer & Ross, 1990). While engineering studies are costly, they provide the optimal means to identify scale economies, and can also be useful in measuring economies of scope.

Statistical cost analysis uses multiple regressions to measure the effects of specific inputs and other factors on costs and output volume. Such analysis can identify scale and scope economies resulting from productive changes made at the plant level by comparing outcomes at the production-modified plant with the firm's other plants.(11)

Thus, both engineering studies and statistical cost analysis can effectively assess plant level economies. However, most plant economies are realized at relatively modest scale, well below Merger Guideline cut-offs. The real need for an efficiencies defense exists at the firm level, and here the two tests are less definitive, particularly statistical cost analysis.

2. Firm Level Economies

Assessment of economies at the firm level requires multiple tests, applied sequentially. The tests fall into three categories: (1) input-based tests, (2) comparative tests, and (3) confirming tests. Input-based tests screen transactions to identify whether the collaborating parties provided the specified inputs they claimed would lead to enhanced efficiency. Comparative tests measure whether the collaborating firms gained efficiencies that their non-collaborating rivals were unable to achieve. Confirming tests measure efficiencies directly, thereby verifying that the comparative advantage gained by the collaborating firms represented real efficiencies and not some extraneous factor. Together the several tests provide a more confident appraisal of firm level economies than any single test.

a. Input-Based Tests

Input-based tests attempt to measure efficiencies indirectly by ascertaining whether the parties have applied the inputs necessary to produced desired outputs. For example, in assessing R&D output effects, an input-based test would infer increased R&D output from the fact that the parties had substantially increased R&D expenditures, combined complementary technologies, created a new research organization, or taken other steps deemed likely to increase R&D. Input-based tests have the advantage of advance predictability and high verifiability, but also have drawbacks.

The problem with input tests is that they focus the parties' compliance efforts not on achieving efficiency outcomes, but on satisfying specified input standards. Moreover, the designation of input criteria is subject to possible manipulation by the parties, who may propose input standards that they can meet at low cost, or perhaps standards that will not succeed at all, but will gain enforcement agency clearance. The agency need not accept such standards, but in formulating input criteria the parties have an informational advantage in knowing whether specified inputs are likely to produce efficiency outcomes.

Input standards are useful, however, in screening cases where particular inputs are indispensable to achieve efficiencies. For example, if the parties propose a hospital merger to secure economies of consolidation, a failure to merge operations of the collaborating hospitals would provide a simple means of determining that efficiencies have not been achieved. Thus, where particular inputs are required to produce claimed efficiencies, an input-based test provides an effective threshold filter. If the parties have failed to provide the specified inputs, the agency might apply a rebuttable presumption that the ex post efficiency showing has not been satisfied.(12)

b. Comparative Tests

Comparative tests, in contrast to input-based tests, attempt to identify efficiency outcomes. Thus, comparative tests seek to determine whether the conditionally approved collaboration produced efficiency gains for the collaborating firms that their rivals did not achieve. The prime comparative test is the survivor test, which measures the ability of a firm to survive and grow relative to other firms in the industry. An alternative, but weaker, test of comparative advantage is the stock market event study, which measures the relative increase in the price of a firms' shares following a defined event -- here the provisionally-approved collaboration.

(1) Survivor Test

The survivor test, originated by Professor Stigler, measures economies by the ability of a firm to maintain relative market share (Stigler, 1968). Applied to ex post efficiency assessment, efficiencies would be proved by the ability of a firm benefiting from a claimed efficiency to survive and grow relative to its competitors. Thus, if a merger or joint venture achieves substantial efficiencies, one would expect that rivals following a different strategies would not "survive" in the sense of maintaining (or increasing) relative market share.

For example, in the General Motors-Toyota joint venture, where a key efficiencies claim was that joint production with Toyota would enable General Motors to master Japanese productive methods in its other plants, one would expect that these plants would have surpassed their U.S. rivals, which lacked access to Toyota's superior expertise. If, as some reports indicate, GM plants continued to lag their rivals, this would refute the efficiency claim based on diffusion of technology. On the other hand, if GM achieved similar improved performance in its other plants, or some of them, e.g. the Saturn plant, this would tend to confirm the efficiency claim.

The survivor test has attractive properties for adjudicating efficiencies in antitrust cases, but also serious limitations. On the positive side, the test is easier for courts and agencies to apply than other tests. Indeed, courts already make considerable use of survival-type analysis.(13) In addition, the survivor test provides clearer guidance for business than other tests since a firm will generally know whether its market share is growing relative to other firms, although it may not know the individual market shares of its rivals. Finally, the survivor test has good consumer welfare effects when used in an ex post efficiencies review since the test encourages firms to increase market share through competitive pricing, and thereby "pass" the survivor test. Thus, the test promotes the transfer of efficiency benefits to consumers.

On the other hand, the survival test, particularly if not supplemented by other tests, has limitations. The test has produced sometimes curious results, as when a wide range of plant sizes pass the test (which Stigler explained as indicating that firms had constant costs above a certain output level), or when only the smallest and largest firms survive (Scherer, 1990). Survival may result not from productive efficiencies, but from monopolizing or predatory strategies, pecuniary economies, or the protection afforded by a cartel price umbrella. Application of the test may be frustrated if all members of the industry adopt a similar practice, in which case no firm can show relative advantage.

Nevertheless, as applied to the ex post proceeding these anomalies of the survivor test appear manageable. The fact that the test is unable to identify efficiencies resulting from industry-wide collaboration would not matter since such collaboration is ruled out by the ex ante precondition that the collaboration not create undue market power. Where a cartel price umbrella assures that all firms "survive" despite the existence of substantial efficiencies among the collaborating firms, the survival test produces a "false negative," failing to identify a real efficiency. But surprisingly, this result may be acceptable as a matter of antitrust policy since it forces the collaborating firms to price at competitive levels, thereby passing on efficiencies benefits to consumers.

When collaboration produces significant pecuniary economies, the survival test may be over inclusive since the test does not distinguish between private and social efficiencies. But this result also appears acceptable. Courts and agencies are likely to have difficulty in distinguishing between pecuniary and socially productive economies. For example, what might appear as a use of strategic buying power, may in fact reflect lower costs in serving the buyer. Under these conditions over inclusion of potential pecuniary economies within an efficiencies defense may be preferable to under inclusion in view of the large potential gains from efficiencies. Reinforcing this preference, the cost of over inclusion is reduced by the ex ante preconditions limiting efficiency-justified collaboration in time and scope.(14)

In sum, the survival test provides a useful, although not exclusive, method for identifying efficiencies. It is particularly helpful in eliminating unfounded efficiencies claims. Thus, a loss of relative market share following a merger or other collaboration provides a strong indication of either failure to achieve efficiencies or intensified collusion, and both justify rejection of an efficiencies defense.

(2) Stock Market Tests

Stock market tests measure the profitability of a merger (or other event) based on the subsequent behavior of the stock price relative to the shares of comparable firms. A stock market test provides a supplemental indicator of comparative advantage, but confronts serious difficulties if used by itself. The biggest limitation is the difficulty of attributing causality to an event that has occurred years before the inquiry. While one can more confidently attribute causality to price effects occurring immediately after announcement of a merger or joint venture, immediate stock effects provide a highly imperfect measure of efficiency benefits which may take years to realize. Indeed, the uncertainty of future efficiencies projections is a prime reason for using the two-stage procedure. By the time of the ex post inquiry three to five years later, efficiency effects may have become clear to public investors, but by that time other price-influencing events will have occurred which it may not be possible to filter out by statistical means.(15) In addition, stock market prices reflect the value of the whole firm, whereas efficiencies may be realized only within a single division.(16)

Thus, stock market tests contain too much "noise" to justify efficiency inferences when considered by themselves. However, a large increase in the stock market value of the collaborating firms relative to rivals supplements a positive survivor test finding of efficiencies.

c. Confirming Tests

Confirming tests, such as engineering studies, provide direct measures of efficiencies gains, thereby validating the inferences drawn from the input-based and comparative tests. Confirming tests also provide independent indicators of efficiency, but as applied to economies at the firm level, their findings are less decisive than in plant level cases. Confirming tests consist of engineering studies, statistical cost analysis, and (under carefully limited circumstances) profit tests.

(1) Engineering Studies

Engineering studies, based on expert engineering assessment of plant operation and design, can effectively identify many firm level economies. Such studies may measure cost and output effects of inventory controls, purchasing, warehousing, resource management system and "just-in-time" supply management. In the ex post efficiencies review engineering studies can identify whether the changes that assertedly increased efficiencies and led the collaborating firms to grow faster than their rival were based on real cost reductions. An important constraint on the use of engineering tests has been their high cost (Scherer & Ross, 1990), but in an antitrust litigation context these costs may be manageable.

(2) Statistical Cost Analysis

Statistical cost analysis, which uses multiple regressions to identify the effects of specific inputs on costs or output volume, confronts problems of data availability and small number of observations when applied to firm level economies. Conducted at the firm level, statistical cost analysis typically involves a cross-sectional comparison between comparable firms within an industry. However, the needed data for industry cross-sectional studies are difficult to gather outside the regulated industries, where regulatory agencies require detailed cost reporting under uniform accounting standards (Scherer & Ross, 1990). Cross-sectional comparisons are also limited by the difficulty of assembling enough separate observations of comparable firms to reach statistically significant results.

An alternative, more tractable approach might involve a time series study limited to the merged firm or joint venture, showing cost reductions over time following the conditionally approved collaboration. If the findings of such a study are congruent with the survivor test findings over the same time period, this would provide confirming evidence of efficiencies.

(3) Profit Tests

While profit-based tests, particularly those based on accounting profits, confront severe problems, under some conditions profit tests may supply limited confirming evidence of efficiencies. In addition to the drawbacks discussed for stock market tests, accounting profits are subject to distortions based on the variety of ways that companies can allocate costs and compute profits, e.g. allocation of general overhead, depreciation, and attribution of cost responsibility to specific products and time periods (Fisher, 1983). Thus, accounting profit tests are manipulatable, and are particularly unreliable at the divisional level where profit depends on internal decisions on transfer pricing.

Nevertheless, under certain circumstances accounting profit tests may provide supplemental evidence of efficiencies realization. A time series study showing the trend of profits within the firm may reliably indicate the trend of the firm's profits since a firm cannot arbitrarily change its accounting system. Hence, a significant increase in profit following conditionally approved collaboration, provides some confirming evidence of efficiencies if consistent with findings of relative growth under the survivor test.

B. Interim Anticompetitive Effects

The ex post review should include an assessment of interim anticompetitive effects resulting from the conditionally approved collaboration. This does not mean that the agency should conduct a roving inquiry into the antitrust behavior of the participants. The agency should determine only whether the approved efficiency-creating transaction has caused anticompetitive effects not strictly necessary to the approved collaboration. (17)

It may be objected that the investigation of interim anticompetitive effects is unnecessarily burdensome to the parties. But the inquiry need not be onerous if it is strictly confined to effects stemming from the approved collaboration. The agency can further minimize the burden on the parties by staffing the ex post investigation (so far as possible) with individuals who participated in the ex ante proceeding, thus avoiding the need to reeducate staff with consequent burden on the parties. In addition, the parties themselves can reduce costs (and encourage compliance) by requiring corporate personnel involved in the collaboration to maintain adequate records. Finally, the burden of proof would be on the government to show any interim anticompetitive effects resulting from the collaboration.

Further, it may also be objected that good antitrust behavior during the interim period provides no reliable indication of competitive behavior after conclusion of the ex post inquiry, when the parties are no longer under close scrutiny. But two considerations reduce the force of this objection. First, deterrence of possible collusion during an initial three to five year period is not a trivial gain. Assurance of an initial period of procompetitive behavior produces not only immediate benefits, but may set in motion competitive developments with long run effects, such as competitive entry into product markets or related R&D markets. Second, even if the parties have collusive designs, the delay in achieving collusive return substantially reduces the probability that collusion will occur. The expected profit from collusion must be discounted both by the delay in its realization and the probability of detection, itself increased by the interim antitrust scrutiny. Thus, the interim inquiry into anticompetitive effects produces real benefits.(18)

C. Burden of Proof

The respondents should have the burden of proving that the claimed efficiencies have been substantially achieved. It is reasonable to place the burden on respondents since they have the best information on their own costs and operations. The antitrust agency, however, should assume the burden of proving any interim anticompetitive effects. This simply applies the normal legal presumption that firms are assumed to have complied with the law. There would, of course, be no issue of less restrictive alternative at the ex post stage.

IV. Possible Objections to Proposed Two-Stage Procedure

Critics may object to the two-stage procedure on several grounds. The critics may assert first, that effective ex post relief is not feasible; second, the prospect of ex post relief creates transaction-inhibiting uncertainty; third, the procedure involves excessive regulation that is beyond enforcement agency capability; and fourth, the agency cannot credibly commit to effective ex post review.

A. Effective Ex Post Relief Not Feasible

The most insistent objection to the ex post efficiency review is that it is not feasible to "unscramble" a merger several years after the fact when the merged assets have been fused into a new entity (see testimony of Augustine, Eddy, Lande and Sanderson).

There are several answers to this objection. First, the only relief the agency will impose is one to which the parties have expressly consented, and if effective ex post relief involves unsustainable costs, the parties would presumably not have agreed to the conditioned merger.

Second, the agency can force the parties to think even more carefully whether they are willing to accept effective relief by utilizing the FTC's "Crown Jewels" procedure or an alternative method of performance bonding. Under the FTC's "Crown Jewels" procedure the agency clears a merger on the condition that the parties agree to divest themselves of a more valuable "Crown Jewel" property if the relief specified in the settlement order is not carried out (see Yao & Dahdouk, 1993). Under alternative performance bonding procedures, which have been used by state attorneys general, the parties would agree to a penalty if efficiencies are not realized, equal to the shortfall between the claimed future efficiencies gain and the actual cost reduction achieved (see O'Connor testimony).

Third, spin-offs and partial sales of assets are a frequent corporate transaction, now far more feasible than in earlier years when the unscrambling assets problem was identified in the merger literature (see Scherer, 1990).

Fourth, even if the scrambling of assets objection retains some force generally, the objection is much weaker as applied to intellectual property acquisitions, where effective relief may be readily accomplished thorough licensing provisions, and as applied to joint ventures, where the common enterprise usually operates as a separate business and profit center.

Finally, if the agency resolutely establishes an enforcement policy requiring effective ex post relief, it will not often face the problem of costly divestiture because parties will screen their transactions to minimize such risk.

B. Transaction-inhibiting Uncertainty

At least one witness thought that the conditional approval procedure might inhibit desirable collaboration by creating uncertainty during the provisional period (see Augustine testimony). This objection seems to assume that mergers that now receive agency clearance will become subject to the new procedure. But this ought not to be the case. The conditional approval procedure should be used only after the agency first refuses to clear the transaction, applying present enforcement standards.

Thus, the conditional approval procedure creates no uncertainty as to any transaction that would now be cleared. Any uncertainty arises only from the fact that the procedure gives firms an option to utilize a procedure that they do not now have. To be sure, there will be some uncertainty as to the value of that option and whether the parties will face an ex post remedy. But it seems curious to object to a procedure that creates a valuable right because its ultimate worth is uncertain. Perhaps the critics fear that the two-stage procedure will lead to a tougher ex ante review standard, but the agency can make it clear that this will not occur.

Finally, even if we assume that the two-stage procedure creates some degree of undesirable uncertainty, its inhibiting effect on efficient collaboration appears benign. The likelihood that ex ante-approved collaboration will face future challenge varies inversely with the probability and magnitude of the potential efficiency gain. Thus, the larger and more likely the efficiency gain, the lower the risk that the collaborating firms will face an ex post remedy. It follows that any inhibiting effect will fall on the least promising collaborations, where potential efficiency losses will be small.

C. Excessive Regulation Beyond Agency Capability

Some witnesses challenged the ex post review as excessively regulatory and beyond the capability of an antitrust enforcement agency (see testimony of Addy, Lande and Sanderson). In fact, the ex post procedure is essentially adjudicatory, not regulatory. The ex post proceeding does not attempt to regulate how the parties should conduct their affairs, but simply verifies their ex ante efficiencies claims and administers appropriate relief, similar if not identical, to the kind it normally imposes. If the agency finds that the parties have achieved the asserted efficiencies, the case is closed. If the agency finds that such efficiencies have not been achieved, the agency must restore competitive conditions, but it will do so by a divestiture or similar one-shot order that involves no ongoing regulation of the industry. Further, the relief imposed is by consent, the parties having already agreed to accept a remedy if efficiencies are not forthcoming, and under my proposal, they would also have consented to the specific terms of the remedy. Moreover, as emphasized above, the collaborating firms need not undergo a two-stage efficiencies determination, but can chose to adhere to the present practice of single stage merger review.

Nor does the ex post review require the agency to do what is beyond its capability. Efficiencies claims are made under present review procedures and require evaluation, although they may not be quantified. While the proposed two-stage review would require quantification, the burden would fall on the parties to produce convincing evidence of efficiencies magnitudes. The agency must of course evaluate such claims and rebut them as necessary; and this may require consultation with outside experts, e.g. manufacturing engineers. But the agency currently evaluates efficiency claims and consults outside experts when their advice is required. None of this should be beyond the capability of a specialized agency with skilled economic and legal staff. The only possibly intractable task -- the balancing of efficiency gains against collusive losses -- is one the agency need not undertake since, as proposed here, the efficiencies defense at the ex post stage would be an absolute defense. Thus, the ex post review is neither excessively regulatory nor beyond the agency's capabilities.

D. Enforcement Agency Commitment

Perhaps the most serious objection to the conditional review procedure is that the enforcement agency lacks credible means to commit itself to effective ex post relief. Thus, respondents may claim efficiencies they know they cannot achieve, believing that when the time comes for ex post review, they will be able to avoid divestiture or other significant relief. The parties might anticipate that three to five years after the initial transaction, they will be able to persuade the agency, now likely to be led by different individuals and perhaps under a different Administration, that a variety of intervening events make it undesirable to upset an ongoing collaboration. This objection is distinct from the objection that effective relief is not feasible. Here relief is feasible, but the agency is unwilling to require it.

The problem of institutional commitment confronts all regulatory bodies that seek to make long range policies. As applied to the two-stage procedure, the problem can be managed in several ways. First, the antitrust agency's ability to adhere to its commitment is strengthened by the fact that the respondents have agreed in advance to the relief to be administered. Thus, the agency need impose no new remedy, but only hold the respondents to a course of action which they have already accepted -- an easier position for an agency to maintain. Second, enforcement agency personnel who participated in the ex ante proceeding should to the extent feasible, be assigned to the ex post inquiry. Third, the agency can further assure continuity of personnel by shortening the time period between the ex ante and ex post proceedings to the minimum necessary to allow opportunity for realization of efficiencies. Fourth, joint ventures should be favored over mergers whenever a joint venture appears reasonably sufficient to realize projected efficiencies, since an effective remedy is much more easily achieved in a joint venture. Finally, a future agency may, at least to some degree, be constrained from wholesale reversal of an ongoing long term enforcement program by the concern that its own ability to make credible commitments will be undermined.

APPENDIX A

Preconditions for Production Markets

Precondition 1: A substantial and non-transitory market imperfection blocks realization of significant efficiencies.

Precondition 2: The proposed collaboration will plausibly overcome the identified market impediment by providing anticipated cost savings of at least five percent or substantially enhance product quality or variety.

Precondition 3: The proposed collaboration maintains competition at other levels of the productive process, and does not permanently eliminate competition at any productive level, except under conditions of strict market necessity.

Precondition 4: The proposed collaboration does not create undue market power or single firm dominance.

Precondition 5: The proposed collaboration is the least restrictive means reasonably available to achieve the efficiencies

Precondition 6: An effective ex post remedy is feasible.

APPENDIX B

Preconditions for Innovation Markets

Precondition 1: The proposed collaboration is likely to produce significant innovation, which is otherwise not likely to occur.

Precondition 2: The proposed collaboration will plausibly overcome the identified market impediment by providing anticipated cost savings of at least ten percent, or substantially enhance product quality or variety.

Precondition 3: The proposed collaboration maintains competition at other levels of the productive process, and does not permanently eliminate competition at any productive level, except under conditions of strict market necessity.

Precondition 4: The proposed collaboration does not create single firm dominance in the innovation market or reduce the number of actual or potential research efforts to less than four. Precondition 5: The proposed collaboration is the least restrictive means reasonably available to achieve the efficiencies

Precondition 6: An effective ex post remedy is feasible.

REFERENCES

Areeda, Phillip & Turner, Donald F., 4 Antitrust Law (1980)

Ausubel, Lawrence M., "The Failure of Competition in the Credit Card Market," 81 Am. Econ. Rev. 50 (1991)

Baxter, William F., "The Definition and Measurement of Market Power in Industries Characterized by Rapidly Developing and Changing Technologies," 53 Antitrust L.J. 717 (1984)

Bleeke, Joel & Ernst, David, "Is Your Strategic Alliance Really a Sale?" Harvard Business Review (Jan-Feb, 1995), 97

Brodley, Joseph F., "Antitrust Law and Innovation Cooperation," 4 J. of Econ. Perspectives (1990), 97

Brodley, Joseph F., "The Economic Goals of Antitrust: Efficiency, Consumer Welfare, and Technological Progress," 62 N.Y. Univ. L. Rev. (1987) 1020

Brodley, Joseph F., "Industrial Deconcentration and Legal Feasibility: The Efficiencies Defense," 9 J. of Econ. Issues (1975) 365, 373

Cohen, Marcus, "Music, Radio Broadcasting and the Sherman Act," 29 Georgetown L.J. 407 (1941)

Fisher, Franklin M., "R & D Cooperation and Competition," in Brookings Papers on Economic Activity (1990), 192-94

Fisher, Franklin M. & McGowan, John J., "On the Misuse of Accounting Rates of Return to Infer Monopoly Profits," 73 Am. Econ. Rev. 82 (1983)

Gilbert, Richard J. & Sunshine, Steven C., "Incorporating Dynamic Efficiency Concerns in Merger Analysis: The use of Innovation Markets," 63 Antitrust L.J. 569 (1995)

Katton, Joseph, "Efficiencies and Merger Analysis," 62 Antitrust L.J. 513 (1994).

Katz, Michael I. & Ordover, Janusz A., "R & D Cooperation and Competition," in Brookings Papers on Economic Activity (1990) 137

Kwoka, John E. Jr., & Warren-Boulton, Frederick R., "Efficiencies, Failing Firms, and Alternatives to Merger: A Policy Synthesis," 31 Antitrust Bull. (1986) 431

Pitofsky, Robert, "Proposals for Revised United States Merger Enforcement in a Global Economy", 81 Georgetown L.J. 195 (1992)

Roberts, Gary L. & Salop, Steven C., "Efficiency Benefits in Dynamic Merger Analysis," Harvard Law School Seminar in Law and Economics (1994) (preliminary draft)

Scherer, F.M. "R & D Cooperation and Competition," in Brookings Papers on Economic Activity (1990)

Scherer, F. M. & Ross, David, Industrial Market Structure and Economic Performance (Houghton Mifflin, 1990)

Spence, Michael A., "Cost Reduction, Competition, and Industry Performance," 52 Econometrica 101 (1984)

Teece, David J., "Information Sharing, Innovation, and Antitrust," 62 Antitrust L. J. (1994) 465

Stigler, George J., The Organization of Industry (1968)

von Hippel, Eric, The Sources of Innovation (Oxford University Press 1988)

Yao, Dennis A. & DeSanti, Susan S., "Innovation Issues Under the 1992 Merger Guidelines" 61 Antitrust L. J. 505 (1993)

Yao, Dennis A. & Thomas, N. Dahdouh, "Information Problems in Merger Decisions Making Their Impact on Development of an Efficiencies Defense," 62 Antitrust L.J. 23 (1993)

ENDNOTES:

(1) Professor of Law and Economics and Kenison Distinguished Scholar of Law, Boston University. I am indebted to Dennis Yao for valuable comments and discussions, but the views expressed are my own.

(2) Full citation of references can be found at the end of the paper.

(3) Roberts and Salop argue that some pecuniary economies may lead to social economies by reducing monopolistic input prices, which can distort the allocation of resources by diverting consumption to higher cost competitive inputs (Roberts & Salop, 1994). But determination of the welfare effects of input substitution raises complex issues that make even ex post outcomes difficult to assess.

(4) U.S. v. Corn Products Refining Co., 234 F. 964 (S.D.N.Y. 1916).

(5) A brief LEXIS search in mid-1994 identified 15 cases in which courts evaluated less restrictive alternatives and none where courts weighed efficiencies benefits against anticompetitive effects.

(6) See United States v. Ivaco, Inc., 704 F. Supp. 1409, 1425 (W.D. Mich. 1989) (court suggests imaginative menu of less restrictive alternatives, thereby avoiding direct efficiencies assessment).

(7) One of the few attempts at serious balancing took place under the Public Utility Holding Company Act of 1935, which contained an explicit efficiencies defense to the divestiture "death sentence" the Act imposed on complex utility holding companies. While the SEC attempted in some of its decisions to weigh efficiency loss against competitive gain, its opinions reflected no intellectually convincing or coherent analysis (Brodley, 1975).

(8) Thus, in the BMI case, which ultimately upheld a market-dominant joint venture, an antitrust consent decree required non-exclusive licensing and provided for judicial review of unreasonable licensing fees. See Broadcast Music, Inc. v. Columbia Broadcasting System, 441 U.S. 1 (1979).

(9) Increased innovation appears most likely from transactions that (1) increase the profitability of innovation, especially intellectual property protection in cases where little or no such protection exists (Roberts & Salop, 1994), (2) combine complementary R&D assets (which may occur both horizontally and vertically (Teece, 1994), (3) eliminate unproductive duplication of innovation effort (Gilbert & Sunshine, 1995), (4) intensify diffusion of innovation, and less probably (5) generate scale economies in innovation (Baxter, 1984; Fisher, 1990; Roberts & Salop, 1994).

(10) See Senate-House Conference Report, National Cooperative Research Act of 1984, 98th Cong., 2d Sess (1984), p. 10 (anticompetitive effects unlikely in R&D joint ventures when there are "several other -- perhaps four comparable R&D efforts [or potential efforts];" Intellectual Property Guidelines §4.3 (enforcement agencies will not challenge licensing arrangements not facially anticompetitive where four entities or technologies exist independent of the parties).

(11) For example, statistical cost analysis may be used to assess the effects of capital stock (identifying technological efficiency of new plant), number of products (identifying possible economies of scope), and cumulative outcome volume (identifying possible learning curve economies).

(12) The presumption is rebuttable, rather than conclusive, because changed conditions might make it senseless to apply the proposed inputs, or the parties may have achieved comparable efficiencies by other means.

(13) For example in BMI the Supreme Court's conclusion that ASCAP's blanket license was a market necessity was aided by the fact that when a second licensing organization (BMI) arose to compete with ASCAP, the new licensing agency also adopted a blanket license. Broadcast Music, Inc. v. Columbia Broadcasting System, supra. The clear implication drawn by the Court was that a different arrangement would not have "survived" as a viable market alternative. A collusive explanation for BMI's parallel behavior can be ruled out because BMI was user-created, and in fact reduced the license fee (see Cohen, 1941 on origin of BMI).

(14) Nor is use of the survivor approach precluded by the risk that collaborating firms will use monopolizing or predatory tactics to exclude rivals or reduce their market share, and thereby satisfy the test. To be sure, the survival test creates an incentive for firms to under price their rivals, but as noted earlier, short of predation, that is a desirable property. Thus, any danger of predatory manipulation of the test is probably more than balanced by the gains from inducing competitive pricing. Further, if conduct is in fact predatory, competitors would presumably complain to the antitrust agency prior to the ex post review.

(15) For example, share prices following a merger or joint venture may increase for reasons other than productive or innovation efficiency, such as increased market power, reappraisal of previously undervalued assets or pecuniary economies. Similarly, the share prices of rivals may fall as result of welfare-reducing strategic and predatory tactics or higher costs due to internal policies of smaller firms to draw high managerial salaries to reduce double taxation (Kwoka & Warren-Boulton, 1986; Scherer & Ross 1990).

(16) However, if the assets of a division are sold to an arm's length buyer, the asset price provides a good measure of the division's market value. This can then be compared with arm's length sales of similar assets (if any) to provide a more accurate test of relative advantage created by the collaboration. Cf. Ausubel, 1991 (greater value of bank credit card operations shown by asset sales prices three times higher than asset sales prices for non-credit card bank assets).

(17) For example, have the parties to an innovation joint venture established cartel punishment mechanisms that inhibit competitive behavior, such as competitive entry into product markets or rivalry in other R&D markets? Is there evidence of spill over collusion from the approved collaboration?

(18) An interim inquiry into anticompetitive effects may also indicate that the collaboration no longer presents antitrust risk, as appeared in the 1993 GM-Toyota joint venture proceeding. In such a case it probably does not make sense to insist on drastic relief even if efficiencies have not been realized. While the agency should have discretion to permit the collaboration to continue under these circumstances, the agency should avoid any suggestion that the ex post proceeding allows a relitigation of the anticompetitive aspects of the collaboration.


Last Modified: Monday, June 25, 2007