Why We Don’t See Greater Use of Merger Simulations

Oliver Budzinski and Victoriia Noskova focus on of their publication why merger simulations are usually not extra extensively utilized by competitors authorities and in entrance of the courts to foretell future results of mergers regardless of developments in availability of information, AI and computational energy. The institutional setting is an important issue for computational antitrust instruments to be accepted and utilized by competitors authorities.

Merger simulation is a instrument of computational antitrust that applies superior economics and econometric strategies to quantitatively predict the consequences of a merger between corporations based mostly on assumptions of pre-merger conditions.

Choices to permit, block, or modify any notified merger in nearly all merger management jurisdictions of the worldwide financial system are made based mostly on an evaluation of anticipated post-merger results. The principle conventional instruments to evaluate these results are two forms of professional opinions: trade witnesses and consultants outdoors of the trade (e.g. teachers). These opinions are kind of influenced by people’ subjective views and interpretations and may very well be pushed by the incentives construction of an professional, particularly within the case of trade witnesses. Thus, the decision-making physique (be it a court docket like within the U.S. or an administrative physique like within the E.U.) is usually confronted with opposing assessments.

A part of the curiosity in merger simulation fashions is predicated on the anticipation {that a} data-based, quantitative method gives a extra impartial and extra goal prediction in regards to the merger’s results decreasing the anomaly of merger management controversies, main to raised choices for competitors and welfare.

Whatever the number of sorts, most merger simulation fashions have widespread underlying assumptions and may very well be described with the next process: (1) constructing a mannequin of the market with collection of a form of demand and a type of competitors; (2) calibration based mostly on previous (pre-merger) knowledge with the intention to estimate the accuracy of the assumptions of the competitors mannequin and the demand operate; (3) last full simulation based mostly on earlier steps. As an consequence, a brand new post-merger state of affairs is predicted. Simulations often purpose to foretell results on market costs and portions in addition to on shoppers’ and producers’ rents, however may embody predictions of product selection and different related components of competitors.

DoMerger Simulation Fashions Present Higher Predictions?

We contemplate how developments in computational energy and strategies might affect the evaluation of the relevance of merger simulation fashions compared to a couple of decade in the past. We tackle 4 areas of reflection that needs to be thought of when pondering the position of merger simulations in antitrust apply beginning with the flexibility to offer higher predictions of merger results. 

Do merger simulation fashions provide extra correct predictions than witnesses and consultants? Certainly, elaborated simulation fashions might present an excellent match of the pre-merger market growth (because of calibration with real-world market knowledge). Whereas relying on modeling and data-analyzing competences in addition to on knowledge availability, merger simulation fashions do fulfill the promise to cut back the scope of doable biases in direction of private believes and opinions.

On the similar time, various issues limiting the facility of merger simulation fashions needs to be thought of. Though there’s growth in modelling methods and computational powers, the diploma of complexity from real-world markets usually nonetheless can’t be absolutely captured. An instance can be the modelling of extra complicated market environments like digital ecosystems. As well as, predictions of post-merger results are at all times based mostly on assumptions that the basic construction of competitors stays the identical after the merger. Solely if that is so, one can count on an impressively high-quality prediction.

Nonetheless, some circumstances reminiscent of mega-mergers and mergers that considerably reshape the aggressive panorama are more likely to introduce structural breaks, altering the dynamics of competitors. Whereas in whole these circumstances could also be few, they play an particularly essential position in merger management choices and in defending the aggressive course of as a result of these are often the essential circumstances during which authorities contemplate intervention (whereas nearly all of mergers stays unchallenged in apply). Thus, the accuracy of merger simulation predictions considerably decreases when structural breaks happen. Because the chance of structural breaks in addition to deviations from historic patterns is rising over time, merger simulation fashions are simpler in predicting short-run results than mid- to long-run results.

Moreover, the unpredictable nature of improvements and exterior markets’ shocks contribute to this sample. Usually, this limitation will not be distinctive to merger simulation fashions and applies to all forms of predictive, forward-looking proof. Because the future will not be predetermined and is simply created as occasions happen, even probably the most superior computational prediction strategies can’t provide absolutely correct numerical predictions. Recognizing this limitation turns into vital in sure institutional settings.

With respect to development of computational energy together with synthetic intelligence (AI), a few of the abovementioned limitations may very well be lessened (see dialogue in publication) though not absolutely eroded and enhance the hypothetical effectiveness of merger simulations.

Institutional Settings of Antitrust Coverage

A assessment of essential and controversial merger circumstances reveals merger simulation fashions are used sometimes, which means that establishments are disregarding potential enhancements of this instrument. Thus, a greater understanding is required of the institutional circumstances for capturing the advantages of merger simulation fashions.

From the angle of institutional-economic idea, hypothetical effectiveness will not be sufficient – the instrument ought to match into the foundations and practices that form the precise strategy of merger management, i.e. its institutional framework. The principle difficulty for the precise effectiveness of merger simulation fashions is how computationally-produced proof is perceived and accepted by authorities and courts, and the way it intersects with the establishments of the executive and authorized process.

An absence of match with establishments is the primary barrier to the profitable use of merger simulations and computational-quantitative proof on the whole. This under-use of the instrument has already contributed to the enforcement hole in merger management the place doubtlessly anticompetitive mergers have been allowed. On the intense facet, the state of affairs has gained recognition and has led to requires stronger merger management in direction of all forms of mergers, and even to discussions on corporations’ breakups and reversing earlier choices like Fb-Instagram-WhatsApp mergers.

The rationale for the shortage of match is that the present design of procedural guidelines and practices will not be appropriate for immanent traits of proof produced by refined quantitative instruments, together with merger simulation fashions. Among the many causes is the complexity of computational devices, which is incongruent with the everyday information, and incentives of judges and courts, making it essential to incorporate testimonies of consultants on the standard of merger simulation fashions as a instrument for merger assessment.

This ceaselessly results in a battle of consultants from competitors authorities and from corporations, penalties of that are pushed by the allocation of the burden of proof: if judges can’t determine the extra correct amongst competing merger simulation fashions offered by the events, they’re incentivized to rule in opposition to the facet that carries the burden of proof, as this facet could be dominated to having not supplied adequate proof. This de facto will increase the usual of proof utilized in direction of computational predictive instruments, though unintentionally and extra as a facet impact slightly than a deliberate determination from the competitors coverage makers.

Furthermore, the predictive nature of merger simulation fashions might create a false sense of precision and a false sense of certainty since they do give a selected numerical consequence. Such consequence might simply be challenged by a counter-party and create doubts in trustworthiness of the outcomes, e.g. by displaying how minor modifications of parameters might result in variations in outcomes. This will likely undermine the belief of judges and juries within the merger simulation fashions even regardless of the instructions of the consequences stay sturdy and unchallenged.

Counter-intuitively, the development of computational energy (together with AI) might have an reverse impact on precise effectiveness of merger simulation fashions and contribute to extend the shortage of  institutional take-up by multiplying the issues raised above (based mostly on the precise experiences with merger simulation fashions in a number of high-profile circumstances like Volvo/Scania, Nuon/Reliant, Oracle/PeopleSoft, most not too long ago AT&T/Time Warner).

Prices and Advantages to Antitrust Authorities

As a normal rule, an efficient instrument needs to be characterised by its advantages exceeding its prices of software. Nonetheless, there’s not but clear proof relating to the prices and advantages of merger simulation fashions. Efficiency of merger simulations is related to particular forms of prices which may very well be important for competitors authorities, e.g. prices related with work with essential knowledge, prices of using employees with experience in superior theoretical economics and econometrics, and so forth. Some prices may very well be decreased if guarantees maintain true for computational energy to grow to be extra out there, present extra help in administration of merger legislation and scale back its prices. Nonetheless, such instruments will stay appropriate just for large and well-equipped antitrust authorities for the close to future.

Merger Simulations for Earlier Choices

Different potentials for merger simulation fashions are their software in ex-post analysis, which may very well be reviewed as a coverage studying instrument. Right here, post-merger knowledge may very well be used to grasp whether or not the precise results met the expectations of the previous merger management determination. The information created this manner can be utilized for shaping higher merger management guidelines and for drafting higher choices sooner or later. The info availability stays the limiting issue for ex-post instruments, however extra refined computational strategies might contribute to overcoming it.


In our paper, we determine and tackle 4 areas that contribute to understanding present tendencies within the software of merger simulation fashions. From an industrial economics perspective, a rise within the software of merger simulation fashions is clearly favorable, as a result of guarantees this instrument brings for an development of merger management based mostly on impartial and extra goal and exact data-driven predictions.

Nonetheless, the empirical dynamics of merger simulation fashions reveals that their use solely elevated up till the mid-2010s, and stagnated and even decreased afterward. An institutional economics perspective gives explanation why that is the case: due to the shortage of institutional match. The financial coverage perspective that considers incentives inside competitors authorities and legislation courts is slightly underdeveloped on this regard.

Nonetheless, the everyday incentives for authorities and courts don’t contribute to an extension within the software of merger simulation fashions and present ambivalent outcomes of computational enhancements related to extra prices for many who apply them. The least controversial software for merger simulation fashions to date is an ex-post space of analyses, which is often ignored however is related for coverage studying.

In abstract, our important conclusion is: establishments matter! The success of a computational instrument in competitors coverage proceedings, each administrative and court-based, relies upon extra on the institutional design of the proceedings slightly than on the (nonetheless fascinating) developments of the strategies themselves.

Articles symbolize the opinions of their writers, not essentially these of the College of Chicago, the Sales space Faculty of Enterprise, or its college.