Antitrust Limits on Merger Decision Markets?

Josh Wright —  5 February 2008

At Overcoming Bias, Robin Hanson points to the absence of decision markets evaluating competitive conditions in the post-merger world as evidence that “these companies are just not serious about finding the highest value applications of prediction markets.” Here’s a description of the markets that Robin has in mind:

Decision markets could say whether this merger is good for shareholders, by estimating the combined stock price given a merger, and given no merger. Similarly, decision markets could say whether this merger is good for these firms’ customers, by estimating the price and/or quantity of web ads given a merger, and given no merger. This might help convince regulators to approve the merger.

I certainly agree that these types of internal decision markets might be relevant to both firms contemplating mergers and to antitrust regulators. I’m also very hawkish on the use of prediction markets within firms and to inform policy as a general matter. But I’m not sure that Robin is right that the problem in the merger context is that firms do not take prediction markets seriously enough. At least, thats not the only problem. One immediate and important problem is that these markets might also help convince regulators to reject the merger based on estimates of post-merger prices. The type of information is likely to be a more accurate form of information that government agencies already seek from the merging parties, usually in the form of internal documents, analyses, and forecasts, to analyze the likely competitive effects of a proposed merger.

The antitrust concern is obvious: any information created via internal prediction markets, including information that suggests that prices will increase, will surely be used by the agencies to argue that the merger will violate the Clayton Act. I do wonder whether these types of concerns provide a substantial additional expected costs to the establishment of decision markets on antitrust events, e.g. exactly the types of events where prediction markets might be extremely useful to firms. My hunch is that if antitrust counsel were sitting in the room when firms are contemplating launching a prediction market on a merger or other antitrust event, he or she might strongly oppose the idea for fear of explaining the “hot documents” that it might produce.

The above discussion assumes that we are talking about “internal” prediction markets. And I do believe that the types of antitrust concerns raised there are real and possibly constrain adoption of these markets for antitrust events in the real world. However, if we allow for prediction markets where folks outside the firm (e.g. competitors) can also participate, the law professor can’t help but raise a scenario fit for an antitrust exam hypothetical:

Firm A launches a prediction market on antitrust event X (lets say X = raise prices 5% next month). In response, or perhaps simultaneously, rival Firm B also launches a market in which antitrust event Y (Y = raise prices 5% next month) is actionable and Firm A managers and employees can participate. One can also think of more precise and problematic contracts in terms of the prospects for collusion, e.g. “Firm A’s price on January 1, 2008 is greater than or equal to $5.50.” But lets leave that contract aside for the moeBoth firms increase prices by 5% at the start of next month. The DOJ sues alleging that A and B have reached an agreement to raise prices in violation of Section 1 of the Sherman Act. What result? What about if the FTC sues under Section 5 of the FTC Act as a stand-alone offense? Of course, this example is not really about prediction markets. Its about collusion.

Some economists who favor the use of prediction markets have argued for safe harbors for their use from gambling regulations. That’s the easy case since the social value of gambling laws is not likely to be low or perhaps negative whereas there is little dispute about the potential social value in prediction markets. I wonder if these economists would favor a similar safe harbor from antitrust regulations? I can already hear Geoff pointing out that antitrust regulations do not have much social value either. So what about a safe harbor from hard core cartel offenses under Section 1? Or a “safe harbor” preventing antitrust regulators from using internal prediction markets as evidence in court to prove the likely competitive effects of a merger?

4 responses to Antitrust Limits on Merger Decision Markets?


    I mean those firms who would like to avoid costly regulatory outcomes. Your answer appears to assume a confidence in both courts (and regulators) that I think is characteristically absent from antitrust counsel. A market evaluation that predicts no impact on price might be used against a firm to demonstrate that there are no cost-savings from the merger and that it violates the antitrust laws. Or a market evaluation that predicts a price increase for reasons unrelated to the acquisition of market power might be used to argue that the merger violates the antitrust law.

    You are right that those firms who suspect that the market would predict a signficant price decrease that would vindicate their claims before regulators or courts would be eager for such an evaluation. But this is just one of the possible outcomes. And outcomes with any price increase or even price neutrality *might* be used against the firm as evidence of a violation of the Clayton Act. Anyway, isn’t one of the benefits of using market evaluations for internal firm decision-making precisely their use in situations where the firm doesn’t have a strong prior about the impact of the merger on prices?


    If by “risk averse” you mean those who suspect that the merger will in fact raise prices, then I can understand their reluctance to let markets evaluate this claim. Those who think markets will vindicate their claims should be more eager to get market evaluations.


    Robin: Thanks for the comment.

    Yes, markets forecasting price increases would really be considered evidence that post-merger prices would increase. This doesn’t mean necessarily via collusion, but also unilateral price effects. The agencies predict post-merger prices in advance of the merger and challenge those they believe will result in higher prices. They generally do not have the luxury of observing the actual realized prices (though there are some exceptions to this rule).

    As to your hypothetical: “If the markets had forecast price decreases and the price increased anyway, would that also be considered additional evidence of collusion, beyond the price increase?”

    Again, there are no post-merger realizations to use as evidence when the merger is analyzed or litigated. Still, it is true that if the parties had forecasts of price decreases in advance of the merger, they could use this as evidence that the merger should be approved. So the markets could be used either way by the parties — as can any non-market related forecasts and documents which contain predictions about post-merger prices.

    The question should really be whether prediction markets like these would really be more successful in convincing agencies to either challenge or approve the merger depending on their prediction. If one believes the benefits of prediction markets are that great in terms of enhanced accuracy, and I do, than the answer should to that question should be “yes.”

    As such, I can see risk averse antitrust counsel advising clients to avoid the use of prediction markets in the same way that antitrust compliance programs attempt to steer firms clear of producing the sort of hot documents that also make steering antitrust events through the regulatory process more difficult.

    Note, I think my post makes clear that I do not believe this is a good turn of events. But if the question is “why not use prediction markets,” I do wonder whether this sort of issue is part of the answer.


    Would markets forecasting price increases really be considered any more evidence of collusion that the actual price increases themselves? Hmm. If the markets had forecast price decreases and the price increased anyway, would that also be considered additional evidence of collusion, beyond the price increase? You can’t have it both ways. Btw, I write about insider trading laws and prediction markets here.